The cormorant: what we actually know

1The animal#

Names and classification#

The great cormorant is Phalacrocorax carbo, described by Linnaeus in 1758 as Pelecanus carbo. The genus Phalacrocorax was introduced by Brisson in 1760, from the Greek phalakros, bald, and korax, raven. Carbo is Latin for charcoal.

The family has been substantially reorganised since 2014. Kennedy and Spencer, working on molecular data, found the old genus Phalacrocorax to be paraphyletic, meaning it grouped together birds that don't share a single common ancestor to the exclusion of others. Seven genera are now generally recognised: Microcarbo, Poikilocarbo, Phalacrocorax, Urile, Gulosus, Nannopterum and Leucocarbo. Phalacrocorax proper now holds twelve species.

The European shag moved out of Phalacrocorax in that reorganisation and now sits in its own genus as Gulosus aristotelis, a change the IOC adopted in 2021. The split is deep. Gulosus diverged from the Nannopterum and Leucocarbo clade somewhere between nine and eleven million years ago, and Phalacrocorax from Urile between about nine and ten, so a shag isn't a smaller cormorant in any close sense. In build, the shag has twelve tail feathers to the cormorant's fourteen, a thinner bill, a breeding crest and a green sheen. It runs 68 to 78 cm long with a 95 to 110 cm wingspan.

Nothing in the naming separates "cormorant" from "shag" as categories. The great cormorant is called the black shag in New Zealand.

Six subspecies are recognised worldwide. Two occur in Britain, and they're dealt with below.

Size and appearance#

Published size ranges for the species mix subspecies and continents, and British birds sit below the top of them. The species range is 70 to 102 cm long with a 121 to 160 cm wingspan. The commonly quoted mass range of 1.5 to 5.3 kg is wide for the same reason, and the upper figure rests on a weak source.

The figures to use are the ones separated by subspecies and sex. For P. c. sinensis in Germany, Dunning recorded 36 males at a mean of 2.28 kg and 17 females at 1.94 kg. For P. c. carbo on Prince Edward Island, eleven males averaged 3.68 kg and eleven females 2.94 kg. Koffijberg and Van Eerden put carbo at about 10% larger than sinensis in linear measurements. The BTO's working figure for British birds is 2.3 kg.

North American reference works giving 2,600 to 3,700 g describe carbo on that side of the Atlantic and run heavy for birds here.

A breeding adult is blackish, with a white throat, yellowish facial skin, a square white patch on the thigh and white plumes on the head and neck. Juveniles are brownish with a whitish throat and belly, and some immatures are nearly white underneath. Adult plumage is reached at three to four years. The white thigh patch and head plumes are seasonal. No primary source gives a precise moult timetable for British birds.

Eggs are pale blue or green under a chalky white outer layer, measured at 63 by 41 mm by one source and 66 by 40 mm by the BTO.

The two subspecies#

P. c. carbo is the Atlantic form, long established on British coasts. P. c. sinensis is the continental form, described by Staunton in 1796, and it founded its first British inland colony at Abberton Reservoir in Essex in 1981.

The gular pouch angle, measured where the feathered edge of the throat meets the bill, is the only field character with any reliability, and published thresholds differ between the sources that give them. They agree on the structure of the problem: there's a carbo range, a sinensis range, and an overlap band in which birds can't be assigned, covering roughly one bird in ten. Confident assignment usually needs the bird in the hand. Plumage, size and sheen don't work.

The distinction is also eroding. Microsatellite work by Goostrey and colleagues in 1998, and mitochondrial sequencing by Winney and colleagues in 2001 across 334 birds from western Europe, both found mixed ancestry at English inland colonies. The two reports that reviewed this evidence concluded that the distinction between the subspecies is becoming blurred.

The genetics are messier than either side of the argument tends to allow. Winney's mitochondrial work found Abberton to be roughly 65 to 70% carbo-related in maternal line, while Besthorpe was 74 to 82% continental. British inland colonies aren't uniformly sinensis. The authors proposed that male sinensis colonised and bred with carbo-type females, called the hypothesis tentative, and noted that mitochondrial DNA traces only the maternal line. The two marker sets don't fully agree, and no modern genomic restudy could be found.

Across six colonies surveyed in 1998, the proportion of carbo rose with colony age, with a rank correlation of 0.98. Colonies founded by sinensis acquire more carbo as they mature.

The documented biological differences, from Natural England's favourable conservation status definition:

The plumage question#

The claim that cormorant feathers aren't waterproof is on the RSPB's own species page, where it's also given as the reason the birds move faster underwater. The primary literature contradicts both halves, and has done since 2005.

Grémillet and colleagues examined four subspecies from tropical to polar latitudes and found that each body feather has a loose, instantly wet outer section and a highly waterproof central portion. All four retained an insulating air layer, thinner than in other diving birds but present. The plumage is partially wettable, which isn't the same as not waterproof.

An earlier study by the same group calculated the minimum insulating air layer at 1.62 mm in males and 1.90 mm in females, and recorded birds holding a constant body temperature through dive series lasting up to 158 minutes in water at 12°C, across a sample of seven cormorants and three shags. Those are calculated minima rather than direct measurements, and the 2005 paper gives no millimetre figure of its own.

The benefit is buoyancy, not speed. Ribak and colleagues measured live birds through simulated foraging bouts and found retained water rose with time in the water, but birds limited dive bouts to under eighteen minutes, so the added mass never exceeded 6% of body mass. At that maximum it's calculated to reduce the dry bird's buoyancy by 18%. Carcass measurements give only the upper bound, because live birds slow the water's penetration.

The preen gland isn't deficient. Sellers found glands of normal size and birds actively applying oil. Stangier and colleagues measured the gland at 3.46 g and found its structure resembles that of other aquatic birds. Holste and colleagues measured water contact angles on feathers at about 128° for cormorant against about 136° for Muscovy duck, a modest difference rather than a categorical one, and found that stripping the waxes with pentane changed the cormorant figure by 1.8°, which wasn't significant. The partial wettability is structural, built into the open outer vane of the feather, not a grooming or secretion failure.

Why the wings are spread#

Sellers' 1995 study on the River Severn and at Thurso is the fullest observational work on the question, and it's freely available. He watched 126 birds, with full foraging detail on 33, and tested five explanations.

Wing-spreading followed 93% of exits from the water. Of 101 birds, 55 spent under four minutes at it, 8% went over ten minutes, and the longest was 18.5 minutes.

Foraging success made no detectable difference, with a Fisher exact test returning p = 0.525. That removes the main support for the idea that the posture signals success to other birds, and for the idea that it helps swallow fish, though a null on 33 birds rules nothing out absolutely.

Wind speed had a strong negative effect, p < 0.002, with the median number of spreading bouts falling from four in calm air to two at force 4. Birds immersed for under five minutes averaged 1.47 minutes of spreading, against 4.75 minutes for longer immersion. Continuous rain almost stopped the behaviour. Sunshine had no detectable effect, which is the finding that works against sunbathing.

White and colleagues ran the direct experiment on the food-warming idea in 2008, comparing fed and unfed birds after swimming and diving. Duration correlated only with time spent on the water. Ingesting food made no difference.

Sellers concluded from this that the function is drying the plumage, to save the metabolic cost of carrying and evaporating the water, and that thermoregulation was unlikely given how brief the behaviour is and how low the temperatures were. The refinement worth having is that it's probably not the wings. Stangier and colleagues, and Rijke independently, argue that the posture mainly exposes the body to sun and wind, speeding evaporation from the contour plumage rather than the flight feathers. Sellers had allowed for this. The accurate phrase is drying the plumage.

One older explanation should be set aside. Rijke's wettability index was tested directly by Elowson in 1984 across fourteen water-bird species using scanning electron microscopy. Of eighteen ramus comparisons only five were significant, and none of those were in breast or back feathers, which is where Rijke's own conclusion rested. Observed contact angles differed from predicted by 14 to 33°, and reed cormorant angles came out larger than mallard's, the opposite of the prediction. Rijke restated his position in 2023, so it belongs in the record as one author's continuing view rather than as settled. Muzio and Rubega's 2024 systematic review is the most recent assessment, and its conclusion is that the field still overlooks barbules as a class.

2How it lives#

Diving#

Great cormorants seldom dive deeper than 10 m. The handbook maximum given by Cramp and Simmons is 9.5 m, and that is a typical ceiling rather than an absolute one: tracking in Greenland recorded mean dive depths of 8.0 to 18.0 m, where the birds feed on the bottom and dive depth follows water depth.

The figures of 30, 45 and 80 m that circulate belong to other species. Imperial shags have been filmed at 80 m.

For dive durations the best British data come from Wright's doctoral work on wintering birds at Loch Leven. Median dive duration was 25 to 27 seconds and median surface interval six to nine seconds, with significant differences between individuals. Birds made up to 495 dives in a single foraging trip and one to five trips a day, with individual trips running as long as 395 minutes. Dive duration and surface interval both fell within a trip. Neither varied with water depth, nor between solitary and flock feeding.

Activity is concentrated early. Corrected for day length, 58% of birds were active during the first tenth of the day, falling to 17% by the fourth. Averaged across daylight hours, around 30% were active at any given moment. Monthly activity rose from under 10% in September to 43% in January, falling to 12% by March.

Eyesight#

Cormorants are usually described as visual pursuit predators. The acuity measurements sit awkwardly with that.

White and colleagues tested five great cormorants underwater across a range of light levels, contrasts and distances. At 1.4 lux, 82% contrast and 2.12 m, mean acuity was 11.8 arcmin. That is comparable to an unaided human underwater and roughly sixty times worse than an eagle's.

On that basis the authors calculated that a 10 cm fish is probably only detectable within about a metre, and proposed brief short-distance pursuit, rapid neck extension or prey-flushing instead. Their own phrase for it is more like herons than hawks. The model is an upper bound; turbidity and camouflage reduce it further.

Poor acuity doesn't mean the birds need good light. In Greenland, 46% of December dive bouts took place at night, with illuminance below 1 lux at around 10 m, which fits close-quarter capture rather than visual pursuit.

Foraging in groups#

Lekuona watched birds across five winters in northern Spain and southwest France and recorded three strategies: line fishing, zig-zag fishing, and driving shoals towards banks. Fish size varied with both strategy and the bird's position in the group, with larger fish taken by zig-zag fishing and by birds in the middle. Birds in larger groups had higher foraging success and shorter dives, while aggression and food theft rose with group size.

Breeding#

Colonies are occupied from mid-March to mid-September, with earlier arrival and earlier laying inland. The clutch is typically three to four eggs, with an observed range of three to six, and one brood. Incubation runs 28 to 31 days and is shared. Fledging takes 48 to 52 days. First breeding is typically at three years.

Chicks stay in the nest for roughly the fledging period of 48 to 52 days and continue to be fed after leaving it, though no source found here puts a reliable figure on how long. Each parent feeds them by regurgitation once a day. Hatchlings are naked and blackish-purple, growing thick dark brown down within a week.

Inland nests are more substantial than coastal ones, built of sticks and lined with leaves, grasses and water plants, and they whitewash with guano as the season progresses, which makes nest trees conspicuous.

Coastal birds breed within a narrow window with chicks hatching at similar times. Inland birds have a long season with a wide spread, which reduces competition for food when chicks are large.

Birds are very faithful to their natal colonies. Of Abberton-hatched birds, around 7% dispersed to breed elsewhere between 1993 and 1996, rising to 12% in 1997 and 18% in 1998. Pairs reuse nest sites year after year and build on surviving structures, so nests become large.

Apparent second broods usually aren't. Craik and Bregnballe reported a west Scotland colony in 2008 where two synchronised waves fledged 47 young from 18 nests in June and 28 young from 14 nests in September. They concluded it was most likely newly arrived birds that had failed at a nearby colony, because the alternative would mean 78% of pairs raising successful second broods.

Lifespan#

The British longevity record is 21 years, 6 months and 21 days, set in 1984. The European record is considerably longer: a bird ringed in Iceland was found dead at 32 years and 1 month, and a Danish ring was read in the field at 27 years and 2 months. Typical life expectancy for a bird that reaches breeding age is eleven years.

The BTO's annual survival figures for Britain are 0.88 for adults and 0.58 for first-year birds. Those are identical to the sinensis values in the table above, so they should not be read as an independent pooled estimate across both subspecies. Broken down by subspecies, first-year survival is 44% in carbo and 58% in sinensis, which puts first-year mortality at around 56% and 42%.

Causes of death are poorly recorded. The BTO gives none. Natural England records an average of 2,614 cormorants shot under licence each winter in England between 2015/16 and 2018/19, which it describes as removing about 7% of the English wintering population annually. That percentage is calculated against its own 2017/18 estimate of 37,257 birds, and section 3 sets out why that estimate is contested. In the Baltic, HELCOM's compilation put the annual shot total at 10,000 to 20,000 across the region, though that figure predates the national series quoted in section 6, and at least 1,700 cormorants are reported to have died of H5N1 in the Baltic region across the summers of 2021 and 2022, though that figure traces only to a popular account rather than to an agency or peer-reviewed source. The harsh winter of 2009/10 caused declines across all Baltic regions. Net entrapment occurs and is documented by age class in Bregnballe and Frederiksen's work. Beyond that, no primary source quantifies cause-specific first-year mortality, and starvation is an inference rather than a finding.

Movement and roosting#

Carbo is sedentary or partially migratory. Sinensis is migratory.

Coastal breeders may stay near their colonies year-round, though many disperse, some as far as northern France and Iberia, and immatures travel further than adults. Birds from inland English colonies move further than coastal breeders, with many ring recoveries in the Netherlands and France and smaller numbers from Iberia and Tunisia.

Ringing shows foreign birds at English inland colonies in the breeding season: sixteen ringed in the Netherlands, six in Denmark and one in Sweden. British carbo also move inland to breed, with three birds from St Margaret's Island in Pembrokeshire and one from the Farne Islands recorded at inland colonies. The report that compiled this cautioned that because few birds have been colour-ringed at English and Welsh coastal colonies, the influence of British carbo is likely to be considerably greater than those numbers suggest.

A behavioural shift is documented, though its date isn't agreed. Ringing data show coastal-breeding carbo making increasing use of inland waters in winter from about the mid-1970s, according to Natural England; Newson and colleagues date it to about the mid-1960s. Either way, part of the inland winter increase is redistribution of native birds rather than continental immigration.

Colour-ring resightings from Abberton show juveniles dispersing within two months of fledging, averaging over 100 km from the colony. Mean resighting distance stays at 100 to 150 km until February, after which birds move back towards the colony. Second-years and adults return between January and March, adults earlier. Most first-years don't return in their first breeding season.

On roosting, the terminology matters and is widely muddled. A bird sitting about on a post or a tree during the day, after feeding, isn't a roost. A roost is the communal overnight site, and those are what the January night-roost counts, the standard survey method across Europe, actually count.

Wright monitored four radio-tracked birds at Loch Leven over 420 hours of night observation and found signals stayed steady, indicating birds remained stationary on roosts. He concluded that cormorants there feed only during daylight. That sits oddly against the Greenland finding of 46% nocturnal dive bouts in December, and no source addresses the discrepancy directly. Latitude and prey are the likely explanation, but it hasn't been tested.

Published median and maximum individual movement distances, named flyways into Britain, and quantified winter roost fidelity are all missing from the accessible literature. The sources that would settle them weren't obtainable.

3How many there are#

The headline figures are indices rather than counts, and the two behave differently.

UK wintering numbers#

The JNCC Official Statistic on wintering waterbird population trends, published 16 April 2026 with data to June 2025, puts the UK cormorant trend at +46% over 25 years (1998/99 to 2023/24) and +23% over 10 years (2013/14 to 2023/24). By country over the same periods: England +64% and +24%, Wales +27% and −5%, Scotland −21% and +9%, Northern Ireland −12% and +16%.

There's a separate England-only Official Statistic on provisional cormorant indices, published 23 July 2026 and updated annually in late July. It gives +58% over 25 years (1999/2000 to 2024/25) and +19% over 10 years (2014/15 to 2024/25), with site coverage at 94% of the expected total. The next update is due in spring 2027. JNCC's own description of the recent England picture is a levelling off.

They're indices, set to 100 in the latest year, not counts. And JNCC publishes no confidence interval for any cormorant trend figure and states that statistical significance of trends isn't currently assessed. Anyone treating a year-on-year index movement as a measured change in bird numbers is reading more into it than the publisher claims.

The UK index has also flattened. Taking the unsmoothed UK series, it has sat between 99 and 103 every year since 2019/20. The England series is less settled over the same span, dropping to 96 in 2021/22.

For an absolute figure, the most recent official estimate is APEP4, published in 2020: over 62,000 birds in Great Britain and over 64,500 in the UK, on a 2012/13 to 2016/17 basis. Both are published as minima rather than central estimates. No confidence interval is published for cormorant, and there's no newer published UK wintering total, so the estimate is ten to fourteen winters old.

UK breeding numbers, which move the other way#

Seabirds Count, the 2015 to 2021 census, counted 8,829 apparently occupied nests in the UK, against 9,258 in Seabird 2000. That is a fall of 429, or 4.6%. By country: England 3,333 (up 203), Scotland 3,458 (down 219), Wales 1,477 (down 311), Northern Ireland 561 (down 102). Across Britain, Ireland, the Isle of Man and the Channel Islands the total went from 14,000 to 13,330, a fall of 4.8%.

So the UK wintering index is up and the UK breeding count is slightly down. Both are true, they aren't in conflict, and quoting either as "the UK cormorant trend" without saying which is misleading.

Welsh breeding trends should be treated with caution. The Seabird Monitoring Programme reports a 23-year increase for Wales while Seabirds Count reports a 17% decline since Seabird 2000, and the SMP attributes the discrepancy to the small number of sites monitored there. The SMP's own table and narrative figures for Wales don't agree with each other.

The inland story#

The modern inland phase began at Abberton Reservoir, Essex, in 1981, with nine pairs. It reached 310 pairs within seven years. Breeding has since been recorded at 89 inland sites in England, with 2,362 pairs at 48 occupied sites in 2012. All England figures in this paragraph are England only; the UK inland total in the 2012 Western Palearctic census is higher, because it adds the single Welsh and four Scottish inland colonies. There were about 2,200 pairs in 2005. Coastal breeding in England was about 1,430 pairs in 1999 to 2002, so by 2005 the inland population already exceeded the coastal one.

The inland population then stabilised at around 2,300 pairs, with declines at older colonies including Abberton, Paxton Pits, Besthorpe, Walthamstow and Rutland Water offset by new colonies, including a range expansion into the southwest.

Two method notes on that census, both from the authors. Around 4% of colony counts were missing, 51 from a possible 1,240, and were filled by interpolation assuming constant proportional annual change. Where several counts existed for a site-year, the largest was used, which biases upward.

In 2012, England had 48 inland colonies, all believed mixed; Wales had one; Scotland four; Northern Ireland none. The largest single colony was Walthamstow Reservoirs at 226 nests, a tenth of the inland population. Thirty-two of the 48 colonies held fewer than 50 nests.

The subspecies split, and why the figures are weaker than they look#

The 2012 Western Palearctic census gives the UK about 6,500 coastal carbo pairs, 955 inland carbo and 2,809 inland sinensis. The report itself states that UK breeding numbers by subspecies couldn't be determined with certainty, because some colonies held both and the proportions were unknown for all of them.

Those inland figures aren't measured. They come from applying a relationship between colony age and sinensis percentage, observed at six inland colonies in the late 1990s, projected forward fourteen years, giving roughly 66% sinensis for the inland breeding population in England. They're a model output with no stated error. Note that the 66% applies to England while the 955 and 2,809 are UK figures, and the two don't reconcile arithmetically in the published document, which is a further reason to treat the split as indicative.

For wintering birds there's no split at all. Natural England states that because of the difficulty of separating the subspecies in the field, it doesn't know the extent to which each contributes to the wintering population in England. Anyone citing a specific sinensis share of the UK wintering flock is going beyond the published evidence.

England holds around 4% of the European breeding population of carbo, under 1% of sinensis, and about 6% of the European wintering population. The 4% is against England's coastal carbo colonies rather than its total nest count, and the 6% rests on a European wintering total that, as the next part of this section sets out, hasn't been counted recently.

European numbers#

The 2012 breeding census put the Western Palearctic at 406,000 to 421,000 breeding pairs, excluding parts of Russia and western Kazakhstan: about 42,500 carbo and about 371,000 sinensis, of which 214,800 pairs were in the EU-28. By comparison, the 2006 census of a narrower area gave about 284,500 pairs, 52,143 carbo and 232,311 sinensis. Different geographies and methods, so the two should not be presented as a time series.

The 2012/13 survey involved more than 1,200 people across 38 areas. Winter roost censuses were held in 2003, 2009 and January 2013. No published continental total from the January 2013 winter roost count could be located.

The widely quoted figure of more than 1.5 million cormorants in Europe isn't a count. It comes from the 2012/13 census extrapolated by an assumed conversion of 4.5 birds per counted nest in autumn, a conversion the source document itself calls non-trivial and dependent on age structure. That document also states that even the best counts carry considerable uncertainty, and that the European total is estimated to have increased since 2014 but hasn't been counted recently.

The Baltic, which holds the largest share, hasn't simply risen. Censuses gave 156,552 pairs in 2006, 165,500 to 167,000 in 2009 and 159,115 in 2012, which HELCOM describes as stabilisation at 155,000 to 170,000. Denmark fell across that series from 38,014 to 27,237 and Sweden from 44,000 to 40,598. Danish national monitoring records 30,281 pairs in 2024, down 2.6%, and stability at around 30,000 to 33,000 since 2014, against a peak near 40,000 in the 1996 to 2005 period.

How the UK winter figures are actually made#

WeBS counts are volunteer look-see counts, monthly. Because the same sites aren't covered every month in every year, an expected count is imputed for unvisited sites based on the pattern across months, years and other sites. Missing values use the Underhill index; trends are smoothed with generalised additive models. Only sites with at least half the possible visits are used.

The load-bearing assumption, in the methods document's own words, is that the pattern of change across the covered sites is representative of the pattern at country level.

And the methods document says where that assumption is weakest. Estuarine species are well covered, because over 90% of British estuaries are counted monthly. But for more widespread species, a large proportion of birds occur at small inland sites and habitats not well monitored by WeBS core counts, and the selection of such sites follows no formal sampling pattern, so it's unclear whether they're a representative sample of the country. Cormorant is exactly that kind of species.

Indices are also recomputed annually and change retrospectively, so a cited index value isn't stable over time.

The one place this was tested against a direct survey#

Natural Resources Wales commissioned a dedicated winter census of cormorant and goosander in Wales for 2020/21, delivered by BTO Cymru.

The extrapolated figure, derived by applying the Welsh share of Atlas records to the APEP4 Great Britain total, is 5,332 to 5,642 birds. The direct census produced 2,894.

The extrapolation was 1.84 to 1.95 times the surveyed estimate. NRW states neither ratio; that arithmetic is ours, from the two figures above.

The reason is spatial. For cormorant in Wales, estuarine birds monitored by WeBS represented 52% of the wintering population. A further 42% were on rivers above the upper limit of WeBS counts, invisible to the scheme. Stillwaters added the remaining 6%.

The survey behind that figure covered 956 km of river in December across 103 sections on ten rivers, around 500 km of repeat surveys, and around 900 km of tributaries in January and February, with NRW reporting coverage of 92% of the 2,482 km surveyable length across the whole exercise. It recorded 564 cormorants directly, with a mean group size of 1.7 and 81% of records being single birds. The 2,894 total is modelled throughout. The only directly observed number is 564 birds. Rivers and stillwaters were modelled from the surveyed sections, estuaries from historic WeBS data, and unsurveyed rivers extrapolated from the river models. NRW calls its totals conservative estimates. So the comparison above is between two estimates, one of them far better grounded, rather than between an estimate and a count.

England has had its own downward revision. Defra's 2013 review reported that the BTO had derived revised time series using an improved method, and that there were approximately 20% fewer cormorants in England than previously thought. The English wintering estimate for 2010/11 became 24,758 birds, about 14,700 of them inland. For 2002/03 the revised figure was 21,005, against roughly 23,000 and 30,000 from two earlier sources for the same year.

Two official series for the same population, and they disagree#

Natural England's favourable conservation status definition carries its own England wintering series: 21,899 birds in 1988/89 (range 20,055 to 24,125), about 29,000 across 1996 to 2000, 30,965 in 2004/05, stable at about 30,000 from 2005/06, 32,786 in 2014/15 (range 31,355 to 34,931) and 37,257 in 2017/18. The 2017/18 figure isn't a survey result. It was extrapolated from the 2014/15 estimate using population change from WeBS data, and the document says so.

Natural England's series tracks the published England index reasonably well. Scaling its 2014/15 figure of 32,786 forward by the index gives about 38,300 for 2017/18 against the 37,257 it reports, so the extrapolation is at least internally consistent.

Defra's revised figure is the one that sits outside it. Scaling Natural England's 2014/15 estimate back by the index gives about 29,200 for 2010/11, against Defra's 24,758. That puts Defra's figure about 18% below the scaled one, or about 15% if the gap is expressed against the larger number. On the smoothed series the difference is nearer 14%. However it's expressed, it isn't noise: Defra's revision was a deliberate correction for unrepresentative inland coverage, so the two series embody different judgements about how many cormorants WeBS is missing.

Every percentage in the licencing debate, including the share of the English wintering population removed under licence and the share the policy ceiling represents, is calculated against one of these figures. Which one is used moves the answer by something close to a fifth.

The only published UK cormorant winter totals carrying confidence limits come from BTO Research Report 612, commissioned by Fera and Defra precisely because the standard index was unrepresentative for inland birds and carried no confidence limits. That series, for England and Wales combined, peaks at 26,220 birds (22,358 to 28,737) in winter 2003 and gives 22,458 (19,236 to 25,002) in 2009. The inland intervals are very wide: the 1988 inland estimate spans 6,234 to 14,624, a factor of 2.3.

Seasonality#

Colonies are occupied from mid-March to mid-September. WeBS indexes cormorant from September to March.

Precise arrival and departure dates for continental birds are not published. Natural England says only that some continental birds move into England during the winter, with numbers further boosted by birds from coastal colonies in Wales and Scotland. The WeBS methods document notes the indexing months were chosen partly for robustness to changes in the timing of arrivals and departures under climate change, which is an acknowledgement that the timing is shifting and not fixed.

4What it eats#

Daily food intake#

No single figure is correct, and the body that assembled the evidence declined to give one.

The Cormorant Research Group's consensus compilation standardises published estimates using a nitrogen-corrected assimilation efficiency of 77.65%. That efficiency is itself extrapolated from work on the double-crested cormorant, a different species, which is worth knowing before treating any of the standardised values as precise. The results, by subspecies, stage and method:

The compilation's conclusion is that only a range is possible, and for adult sinensis it puts the likely range at 522 to 707 g a day. The two nest-balance figures cover adults plus nestlings and aren't comparable with the rest.

The one full bioenergetics model, built for wintering birds at Loch Leven from radio-tracking, captive metabolic measurements and published data, gives 672 g a day, with a predicted maximum range of 441 to 1,095. A later review tabulates the same Loch Leven figure with a range of 540 to 803, so the wider bracket comes from the primary paper and the narrower one from secondary citation. Its authors describe the result as similar to great cormorants breeding in temperate conditions and to other aquatic birds of similar size, and framed the tool as producing average, minimum and maximum values for local conditions rather than a universal number.

The figure used operationally in European management is 500 g a day, described as the mean of a seasonal range of 200 to 700, with the greatest need in May and June when birds have young. Compensation schemes in some European countries work from a figure of this kind, though the specific formulae couldn't be traced to an obtainable source.

Two features of this evidence base are worth stating directly.

First, the highest figures come from the weakest method. Predictive equations based solely on body mass, derived from cold-water flapping seabirds, produce the 707, 790 and 843 values. The compilation also states that pellets should not be used to estimate daily food intake, that stomach contents can't give it, that direct observation can't give it, and that captive birds almost invariably underestimate it. The compilation's own warning to policy makers is that necessarily wide confidence limits apply to some of the figures it produces.

Second, authors making comparisons with available fish tend to use values at or towards the upper end of 25 to 30% of body mass a day. That is a deliberately conservative-for-the-fishery choice, not a central estimate, and it propagates into a lot of published consumption figures.

On voracity, the Loch Leven modelling authors describe their result as similar to great cormorants breeding in temperate conditions and to other aquatic birds of similar size. A wider claim circulates that cormorant intake is lower than that of comparable seabirds, attributed to the REDCAFE final report, but that volume couldn't be obtained and the claim isn't in the parts of REDCAFE that are available.

Diet composition#

The cormorant is almost entirely piscivorous and consistently takes relatively few species. The Cormorant Research Group's compilation recognised at least 115 fish species in the food of European cormorants, while noting that in most areas only a few are staple food. Within a habitat, different studies show similar prey spectra despite different methods, and that consistency is among the better-replicated findings in the diet literature.

For England and Wales there's one large dataset, though it's a selected sample rather than a representative one, being carcasses of birds shot under licence at sites where damage was suspected. Cefas examined 1,412 cormorant carcasses from birds shot under licence between 1993/94 and 2003/04, covering both subspecies. Carcasses came 72% from stillwater fisheries, 23% from river fisheries and 5% from fish farms.

From 6,749 prey items across 32 species:

• Roach was the single most common prey at 47% of items by number and 27% by mass.
• Five species, roach, perch, bream, three-spined stickleback and minnow, made up almost 80% of items.
• Salmonids were just over 5% of items by number, but brown and rainbow trout together made 37% of the diet by mass, reflecting their larger size.
• Salmon was just under 1% of items and 0.3% by mass.
• Rainbow trout appeared only in birds shot at put-and-take trout fisheries.

Mean weights recorded: three-spined stickleback 1.4 g, minnow 2.5 g, perch 17.8 g, roach 21.6 g, bream 34.4 g, grayling 173.3 g, brown trout 201.1 g, rainbow trout 450.5 g.

On Welsh rivers, 111 birds were examined, 36 from upper salmonid catchments and 77 from the lower Wye. Brown trout was the main prey by mass on upland rivers; roach and dace predominated on the lower Wye. Salmon was under 1% of diet by weight on upland rivers and 5% on average on the lower Wye, with the proportion rising in spring in line with the smolt run.

In Scotland, stomach contents from 391 cormorants across 21 river systems and seven lochs between 1991 and 1996 found brown trout, salmon, minnow, three-spined stickleback and eel the most widespread prey, with patterns consistent with opportunist generalists taking the commonest and largest prey available. There's a latitudinal gradient: salmon was under 1% of winter and spring diet by mass in southern Scottish rivers but 18% in northern rivers, reflecting lower fish community diversity in the north.

Stable isotope work found that the cormorants sampled at inland English fisheries had been feeding on freshwater prey, with the move to freshwater happening over several months and birds then staying on freshwater through autumn and winter.

Eel deserves separate mention. Eel is now relatively rare in cormorant diet, which the synthesis attributes to the collapse of eel populations across Europe. Much of the older literature showing eel as a key prey item predates that collapse.

One methodological finding is worth knowing before trusting any roost-based diet sample. Strontium isotope provenancing of otoliths from pellets at a Bavarian lake found that 36% of pellets contained prey assigned to waterbodies other than the adjacent lake, and that birds used different foraging sites within a single day. A diet sample at a roost doesn't necessarily describe the water next to it.

Size of fish taken#

Cormorants can consume relatively large fish, exceptionally up to around 1.5 kg, and a mended salmon kelt of 1.53 kg was recorded in a Scottish river bird. In the England and Wales dataset, median fork length for many common species was at or below 10 cm, with 15 to 35 cm for brown trout, rainbow trout and grayling. The sample is dominated by small cyprinids and sticklebacks, so those medians describe that prey community rather than cormorant diet everywhere. Individual fish over 1 kg appear on rare occasions.

Diet and fish size vary with bird size. Males are larger than females, and carbo larger than sinensis.

The direction of size selectivity is system-specific and doesn't generalise, which is one of the clearest cases in this literature where a single statement would misrepresent the evidence:

• In a Danish lake, across roach, bream and perch at 120 to 367 mm, larger individuals had higher predation rates for all three species, with perch most vulnerable, raising the possibility of age and size truncation.
• For down-migrating sea trout smolts in a Danish stream, body length and predation risk were negatively correlated, with the benefit of size becoming clear above roughly 150 mm.
• For adult anadromous brown trout, individuals between 35 and 43 cm were most likely to be taken, with smaller and larger fish less likely. At least 15.4% of tagged adult sea trout were predated, most of them in the river, and females more often than males. The authors note this challenges the assumption that cormorants don't prey on reproductively mature salmonids.

Injury to fish that escape#

An underwater video study of six great cormorants over 158 pursuits, and of nine double-crested cormorants over 518 pursuits, found injury without ingestion extremely rare. The primary paper couldn't be obtained during checking, so the counts below come from a secondary summary of it. In the double-crested birds, 2 of 518 pursuits ended with a fish injured but not eaten, which is 0.4%. In 241 other unsuccessful pursuits the bird never touched the fish. For great cormorants, 2 of 70 unsuccessful pursuits involved a fish seized and lost, and both weighed over 500 g. The authors concluded that current knowledge doesn't support the claim that cormorants injure more fish than they ingest. Their stated limitation matters: the results apply to small prey of 15 to 22 cm at low densities, with tame or captive birds in tanks.

Against that, sampling of harvested fish from five Czech ponds found the proportion of wounded fish ranging from under 1% at two ponds to 21% and 47.4% at two others, with the high figures coming from ponds with far higher bird presence. Wounds were 1 to 5 mm deep, perforating skin and muscle, and condition coefficient was significantly lower in wounded fish in four of five comparisons. On secondary infection, the authors found common ectoparasites on wounded fish but reported no outstanding infestations compared with standard spring carp levels, a qualification that is usually dropped when the study is cited. Mortality wasn't directly measured.

The two findings may be reconcilable. The studies differ in prey size relative to the bird's gape and in stocking density, and both would be expected to matter: a fish too large to swallow and a pond too crowded to leave are the conditions under which repeated unsuccessful attacks would be likely. No study has tested that directly, so it remains a plausible explanation rather than a demonstrated one.

The figure of 0.3 to 0.4 kg of fish wounded per cormorant per day, which circulates in the compensation literature, isn't a direct measurement and shouldn't be presented as one.

For a wild river comparison, a Swiss study is reported as finding that 10 to 16% of surviving grayling in any one winter bore marks of a recent cormorant attack, with no evidence that this harmed population dynamics. The paper itself couldn't be accessed to confirm the percentages, and its wider conclusions were disputed at the time.

5What the evidence says about impact on fish#

The global meta-analysis#

The 2021 global meta-analysis of cormorant predation effects on fish populations is the broadest attempt to pool the evidence.

It screened 603 publications. Only 27 tested fish population parameters against cormorant predation using significance-based hypothesis tests. 22 were poolable.

The pooled effect was −0.169, with a 95% confidence interval of −0.505 to 0.167 and p = .256. Negative, but not statistically significant.

There was a significant difference among prey taxa, p = .006, with Cyprinidae and Percidae appearing most vulnerable, possibly as a result of prey selection. Salmonidae featured very little, so the meta-analysis doesn't bear much on the salmonid question in either direction.

That result doesn't say there's no effect. It says that after 603 publications, only 22 studies were rigorous enough to pool, and the pooled evidence doesn't demonstrate one. Any claim of a settled strong effect, or a settled null effect, overstates what has been shown.

The authors' own recommendation is that research should focus on effects on fish populations rather than diet analysis alone, and they note the scarcity of studies in natural ecosystems, probably because of the difficulty of isolating cormorant predation from every other pressure.

Studies showing severe local impact#

In Danish lowland rivers, cold winters and low coastal prey availability pushed birds onto rivers, and their arrival coincided with a sharp fish decline. PIT-tag recovery indicated cormorants ate an estimated 30% of wild trout and 72% of wild grayling tagged in a small river. In a medium-sized river, 79% of radio-tagged adult grayling were lost over one winter. The authors suggest predation was sufficient to explain the collapse of grayling and brown trout populations in many Danish streams.

Supporting electrofishing on a 2 km stretch of one Danish river recorded grayling falling from 412 individuals in 2009 to 6 in 2010, after the cold winter of 2009/10 drove cormorants inland.

At Loch Leven, Stewart and colleagues estimated consumption over seven months at over 80,000 brown trout and 5,000 rainbow, against angler catches of 5,828 and 12,815. That estimate is built by the count-times-intake-times-diet method described later in this section, at the same water where bird counts were later shown to understate visiting numbers by roughly tenfold, so it carries the uncertainty of that method in both directions.

In a Danish lake, cormorants were major predators of roach, bream and perch, with perch most vulnerable and the possibility of size truncation of the population.

On the Swedish Baltic coast, a study that propagated its own uncertainty estimated cormorant consumption at about 44% of catches in one archipelago and 10% in another, with consumption of harvestable-sized fish at about 14% and 5% of total human harvest. For flounder it estimated a 35 to 41% decrease in human harvest. Its own 95% intervals span about an order of magnitude, and the authors describe the variation they assumed as arbitrary.

A seven-year study in South Tyrol, a 13,000 km² region with no lakes, reported that with 32 cormorants present in winter conditions were acceptable, while with 65 the fish populations were in danger. That is a single region and the figures come from a conference presentation rather than a primary paper, but it illustrates how low the carrying capacity of a river system for this bird may be.

Studies showing little or no impact#

In two north-eastern Swiss rivers with some of the highest recorded cormorant densities in the country, predation intensity was positively correlated with yield in the largest grayling population, and no evidence was found for a negative effect on fish population dynamics. Growth, age structure and age at maturity tracked the strength of the recruiting age-2 cohort, not predation. Young-grayling biomass predicted cormorant predation intensity, which is the pattern expected if the birds track fish abundance rather than drive it.

That study and the Danish work examined the same species in comparable systems and reached opposite conclusions. The designs differ in a way that matters. The Swiss work used fishery yield time series as a population proxy and tested for correlation; the Danish work used individually tagged fish and recovered tags from colonies. Yield data are confounded by angling effort, regulations and cohort strength, and are insensitive to losses. Tagging gives a direct per-individual fate but introduces capture, handling and tagging effects and can't be scaled to the population without a pre-predation baseline. Neither design settles the question, and they fail in opposite directions.

For Scottish salmonids, two successive reviews concluded that there's substantial evidence that piscivorous birds can take large numbers of fish at site level, but a lack of evidence that such predation affects fish species at population level or causes direct economic losses. The second review added that this can't be interpreted as a lack of the existence of an effect, rather that it's very difficult to measure.

Smolt predation#

The most-cited figure in the European debate comes from a Danish review of 24 studies, mostly in the River Skjern estuary. In all 24, cormorant predation was estimated at over 20% of the smolt run, with a mean of 47% and a range of 23 to 88%.

The mean carries a caveat the original doesn't draw. Its authors characterised the variation across species, years and systems as modest, and concluded from it that cormorant predation on smolts is generally an important factor in lowland rivers. The arithmetic objection is separate: the mean is unweighted, and sample sizes run from 17 tagged fish to 64,500. The two coded-wire studies with the largest samples, 64,500 and 10,000 fish, gave the lowest figures at 23% and 31%. The 88% and 67% values come from samples of 42 and 50. An n-weighted mean would be substantially lower than 47%.

The context also limits transfer. Most Danish salmonid rivers and coastal areas sit within 25 km of colonies holding hundreds or thousands of breeding pairs, and the estuary at the centre of the work held a colony of 1,200 to 2,500 pairs.

There's one unresolved methodological problem running through almost all of it. The studies involving wild smolts required capture, handling and tagging during migration. The original authors acknowledge the question of whether this increases predation probability, and argue it wasn't a major problem because different methods gave similar estimates. The UK synthesis pushes back directly, calling that defence perhaps unlikely given the widely differing sizes and attachment methods of the tags used, and stating that further work is clearly needed.

The only quantified UK cormorant smolt estimates both come from the River Bush in Northern Ireland. The first, published in 1988 on 1986 fieldwork, worked from bird counts, published intake figures and stomach analysis of ten shot birds. It estimated cormorant predation during May at 51 to 66% of the wild smolt run and 13 to 28% of hatchery releases. The UK synthesis assesses those estimates as likely atypical, because the river sits near one of the largest cormorant colonies in these islands, and notes the small sample. A later study on the same river reported lower predation, apparently influenced by the abundance of cyprinid and percid prey at alternative feeding sites.

Across the literature the spread runs from 1 to 2% (goosander, northern Norway, where sandeels were abundant) to 88% (cormorant, Denmark). The reasons for a roughly fiftyfold range are alternative prey availability, colony proximity, method, tagging effects, sample size, and whether the predator studied was cormorant or goosander.

For Wales, no quantified smolt loss estimate could be found. The synthesis notes that little is known about cormorant or goosander diet during the smolt run, likely April to May, which is precisely the period of greatest concern.

Compensation, and where it stops working#

The general biology here is widely agreed. In many fish populations including Atlantic salmon and trout, strong compensatory processes operate and are strongest at the earliest life stages. Fry and small parr normally exceed carrying capacity and natural mortality is high, so fish eaten by predators tend to be replaced by others that would have died of something else.

Recruitment of older life stages in freshwater, and of salmon during the marine phase, is considered density-independent and not subject to compensation. As compensation weakens with age, the relative importance of predation increases. That is why the smolt stage is the crux.

Two situations are identified where predators are likely to have strong negative effects. The first is predation on older parr, smolts and marine stages, where mortality is more likely additive. The second is populations already at low abundance: if predators take a particular prey even when it's rare, which is typical of opportunists, they have a disproportionate impact on depleted populations.

Compensation theory is usually invoked to argue that predation doesn't matter much. The same theory predicts larger proportional effects in depleted stocks. The UK synthesis applies it directly, observing that most Welsh salmon and sea trout populations are in a depleted state with insufficient returning adults for optimal production, so there's likely to be lower potential for compensation and the impact of anything that removes fish will proportionally be greater.

The same reasoning cuts the other way on healthy stocks and on early life stages, where compensation is strongest and predation losses are most likely to be replaced. And it applies to every source of removal, not only to birds, so it's an argument for the weight of each pressure rather than for the primacy of any one of them.

For put-and-take fisheries, compensation in the biological sense isn't really relevant. Losses are made up by additional stocking, so the compensation is economic rather than biological.

Where additive and compensatory have been tested head to head, the answers conflicted. Two studies on Columbia River steelhead examining the same colonies reached opposite conclusions, one finding avian predation compensatory with predation rates accounting for under 1% of variation in survival, the other finding tern predation super-additive at the smolt stage. A review of the dispute found important differences in definitions and underlying models that could partly account for it. Whether predation is scored additive or compensatory depends heavily on how additivity is defined, which covariates are in the model, which life stages are evaluated and over what period.

Why consumption arithmetic goes wrong#

On the River Wye, eleven goosanders shot in July and August 1996 had an average diet of 63% juvenile salmon by mass. Estimated losses to birds, based on counts of feeding birds, exceeded the estimated standing stock of salmon by almost seven times. The authors couldn't assess impact as a result. Similar discrepancies were reported in Scotland, where at certain times and places the calculated fish consumption implied the birds would remove every fish from the river within a short period. That didn't happen.

The listed error sources were bird density, bird diet, daily food intake taken as an upper limit from the literature, and salmon standing crop not measured directly but taken from four other rivers, likely underestimated, and not accounting for seasonal movement within the catchment. The core problem identified was that birds take most fish from mainstem pools while fish density estimates come mainly from shallow riffle habitat on narrower streams.

Bird counting compounds it. Wright found counts at one time of day up to 40% higher than counts at a different time the same day, and that at Loch Leven the number of cormorants visiting the fishery was probably ten times the mean number counted at any one time. Frederiksen and colleagues found a roost used by 66% more birds over a season than the peak count recorded there. And being counted at a site doesn't mean a bird takes all, or any, of its food there.

So the standard method, bird count multiplied by daily intake multiplied by diet proportion, has compounding error in every term, and there's at least one documented case where the output exceeded the available fish by almost seven times.

National and European totals#

The European figure in circulation is 270,000 tonnes of fish a year, attributed to a European cormorant population of more than 1.5 million birds.

The source supplies the inputs, so the arithmetic can be checked: 1.5 million birds × 500 g a day × 365 days = 273,750 tonnes. Two uncertain quantities, a population and a daily intake, multiplied together, with no species, seasonal, regional or age-structure resolution and no confidence interval. Both inputs carry large declared uncertainty, the population resting on the 2012/13 census count of sinensis pairs extrapolated by an assumed 4.5 birds per nest. Using the stated daily intake bounds of 200 to 700 g alone, the total would span roughly 109,500 to 383,250 tonnes.

There's no robust published national UK estimate of total fish biomass removed by cormorants. The only official UK tonnage figure found is for eel: 29 to 43 tonnes consumed in England and Wales during the birds' breeding season, derived from breeding bird numbers in the Environment Agency's Eel Management Plans. It covers the breeding period only and uses the same three-term chain described above.

For comparison on how such estimates should be bracketed, the best-specified regional study available, on the Swedish Baltic coast, ran 100,000 Monte Carlo simulations and reported that its 95% intervals spanned about an order of magnitude. Its authors described the coefficient of variation they assumed as arbitrary. That is the calibration point to hold against any national figure quoted without brackets.

European eel#

The European eel is critically endangered on the IUCN Red List. Against the 1960 to 1979 reference level, the recruitment index is 98.6% lower in the North Sea series and 94% lower elsewhere, with provisional 2019 values at 1.4% and 6.0% of reference. Yellow and silver eel data indicate a mean decline of 50 to 80% over three generations, 39 years.

The EU eel regulation requires member states to allow 40% of adult silver eels to escape to spawn, relative to the best estimate of pristine escapement, and explicitly lists addressing predators among the measures.

The IUCN assessment itself cites cormorant predation as material, noting eel consumption by cormorants is similar in quantity to fishery landings in the Baltic and in some Swedish lakes.

Direct tagging evidence exists and should be cited with its own authors' caveat attached. In a Danish estuary in 2003, 10,000 tagged eels gave an estimated 4,426 eaten, about 44%, with a 95% confidence interval of 1,968 to 8,745. Across three years, nearly 100,000 fish were tagged and over 10,000 pellets collected, but only 112 tags were recovered, and the authors concluded the method didn't provide high-quality documentation. The same study produced a flounder estimate of 194%, which the authors state is nonsense, attributing it to unrepresentative pellet sampling. The 44% and the 194% come from the same paper, and the second should always accompany the first.

On the Swedish Baltic coast, estimated impact on silver eel escapement at two sites equalled roughly 168% and 7% of human catches by biomass. The two sites in the same study differ by a factor of twenty-four, which is the clearest illustration available of how site-specific these figures are.

A distinction worth holding: because eel stocks have collapsed, absolute eel consumption by cormorants has fallen while proportional pressure on what remains can be high. Both "cormorants barely eat eels now" and "cormorants are a leading source of eel mortality in fresh water" can be true at once.

6The law#

Protection in the UK#

The great cormorant is an ordinarily protected wild bird under section 1 of the Wildlife and Countryside Act 1981. It isn't on Schedule 1 and not on any general licence.

Section 1(1) makes it an offence to intentionally kill, injure or take any wild bird, to take, damage or destroy a nest while it's in use or being built, or to take or destroy an egg. Section 1(2) covers possession of a live or dead bird, any part or derivative, or an egg. The disturbance offence in section 1(5) applies only to Schedule 1 birds, so it doesn't apply to cormorant.

On penalties, section 21(1) provides for up to six months' imprisonment, a fine not exceeding level 5, or both. Section 21(5) treats each bird as a separate offence where more than one is involved, and section 21(6)(a) requires the court to order forfeiture of the bird.

Scotland applies the same Act with a materially lower threshold. Section 1(1) there reads "intentionally or recklessly", inserted by the Nature Conservation (Scotland) Act 2004 and in force from 1 October 2004. Scottish penalties are far higher: on summary conviction up to twelve months and a fine up to £40,000, and on indictment up to five years.

Northern Ireland operates under article 4(1) of the Wildlife (Northern Ireland) Order 1985, also on an "intentionally or recklessly" threshold. It adds an offence with no direct England and Wales equivalent: obstructing or preventing a bird from using its nest.

The A06 licence in England#

The A06 is an individual licence under section 16 of the Wildlife and Countryside Act 1981, not a class licence and not a general licence. Cormorant appears on no general or class licence, and Natural England's published register shows only individual site and area licences, each named to a holder and a location.

The statutory ground is section 16(1)(k), for preventing serious damage to fisheries or inland waters. Every cormorant row in Natural England's published register carries that purpose string.

Two statutory constraints do most of the work. Section 16(1A)(a) prevents a grant unless the authority is satisfied there's no other satisfactory solution. Section 16(5) allows licences to be modified or revoked at any time and caps validity at two years, or five for a licence granted by Natural England.

The default appropriate authority for a 16(1)(k) licence relating to fish is the agriculture minister, with the Marine Management Organisation for the restricted English inshore region and Natural Resources Wales for Wales. Natural England issues inland fisheries A06 licences under delegated arrangements, and the delegation instrument itself isn't readily traceable in published sources, so a bare statement that Natural England is the section 16 appropriate authority should carry that qualification.

Under the current guidance, last updated 3 August 2026, the licence covers cormorant, goosander, merganser and grey heron, is issued by Natural England's wildlife licencing team and takes up to 30 working days.

An applicant must supply a record of non-lethal action covering at least one month before application, evidence that damage has occurred or is likely, the species, numbers, locations and dates proposed, the names and addresses of everyone authorised to act, and a declaration of wildlife convictions for all of them.

The cormorant window runs 1 September to 15 April, extendable into May where a smolt run is demonstrated. Grey heron runs 16 September to 28 February, and goosander and merganser 16 September to 31 March with the same May extension.

There's a reporting duty within 14 days of licence expiry, whether or not any action was taken. Failure can lead to revocation and refusal of future applications, and renewal is requested through the same return.

Natural England's cormorant policy has long been described as applying three tests: that serious damage is occurring or likely and provable by circumstantial evidence, that non-lethal measures have been tried and failed or are impracticable, and that shooting can reasonably be expected to reduce numbers or stop them increasing. The current guidance page carries the no-other-satisfactory-solution requirement and the damage-evidence requirement but not the third test in those terms, so the three-test formulation should be read as policy description rather than as current published wording.

The method condition on the register is almost invariably shooting to kill to reinforce scaring. Lethal control is conditioned as an adjunct to scaring rather than as a measure in its own right.

Area-based licencing began as a trial in 2014/15, with one licence per area setting a total for that area. Contemporary descriptions put an area at roughly 100 to 150 km² and ten to fifteen fisheries, with up to six authorised shooters per site. The current guidance page gives none of those figures.

The other three nations and Ireland#

Scotland. NatureScot licences on the same section 16(1)(k) basis, by two routes. Fish farm licences are site-specific, normally issued 1 September to 30 April, applied for at least six weeks ahead, and require stock-loss records and regular bird counts, at least daily where staff are on site daily. Inland fisheries require a full annual application; licences aren't renewed automatically.

Wales. Natural Resources Wales is the appropriate authority by statute. The window is narrower than England's: NRW states it can only issue licences between 1 October and 17 March inclusive. Decisions take up to 40 working days and there's no charge. Applicants must supply a bird count, evidence of serious damage including lost income, recorded losses, catch records, stock replacement costs or lost ticket and membership income, and the non-lethal methods used with their effectiveness. The condition is that non-lethal methods must have been tried and had limited or no effect.

Northern Ireland. DAERA licences under article 18(3)(d) of the 1985 Order, on the most restrictive terms in these islands. The applicant must operate a fishery large enough that exclusion by fencing, netting or tank covers is impossible. No licences are issued for May, June or July. No more than two licences a year are issued for the same fishery, and a licence is valid for up to 30 days after stocking. Unlicenced shooting attracts a fine up to £5,000 and up to six months.

Republic of Ireland. NPWS operates a permit to stop damage to flora and fauna by protected wild birds, under section 42 of the Wildlife Act 1976, functioning as an Article 9 derogation. The NPWS page doesn't list cormorant among covered species and gives no evidence requirements, window, conditions or reporting duty, so parity with the UK routes should not be assumed.

What the licence figures actually show#

Natural England publishes a licence-level register. Aggregating the cormorant rows by licence start year, and excluding amendment rows to avoid double counting:

Licence counts here are rows in the register rather than unique application references. On a unique-reference basis the bird totals are identical but the licence counts differ slightly, running 350, 396, 420, 399, 409, 414, 341, 347, 304, 423, 419 and 404 across the same years. Including amendment rows, 2025 rises to 3,615 birds across 419 or 420 licences depending on the same choice. Every convention crosses 3,000.

Area licences are a minority of volume throughout. In 2025, counted on the amendments-included basis, they accounted for 24 licences and 717 birds against 396 site licences and 2,898 birds. The largest single cormorant authorisations in 2025 were 90 birds in Wiltshire, Hampshire and Dorset, 70 in Lancashire and 64 in Hampshire.

Refusals aren't published. The register contains granted licences only, and a search of all 18,216 rows for refusal, withdrawal or rejection markers returns nothing. Natural England doesn't publish a disclosure log of its replies to information requests, publishing annual lists of request titles only, and no FOI or EIR response giving cormorant refusal counts could be found in the public domain. There's therefore no published refusal rate for cormorant licences in England, in any year.

Scotland publishes what England doesn't, including birds actually taken. The dataset begins in June 2019, so the first row is a part year:

Scottish refusals aren't in that dataset either. No cormorant licence statistics are published at all for Wales, Northern Ireland or the Republic of Ireland.

Licenced isn't killed#

England published birds-shot figures alongside licenced figures until 2011/12:

Defra's summary of that period is that an average of 71% of licenced birds were shot. The table's own columns give 73%, so the published summary and the published table disagree by two points.

No annual birds-shot series has been published for England since 2011/12. One four-year average exists: Natural England's figure of 2,614 shot each winter between 2015/16 and 2018/19. Set against the mean licenced total for those licence years, that implies around 95% of licenced birds being shot, well above Defra's 71% for the previous decade. Natural England's own position is that it publishes only the numbers covered on licences, that the actual numbers affected are significantly lower, and that publishing the latter isn't possible. Nothing published explains the difference, and licence years and winters aren't quite the same period, so the comparison should be treated as a question rather than a finding.

The ceiling#

The English limit is Defra policy, not statute. The 2004 policy review set it at 2,000 birds a year with scope for up to 3,000 for a short period, characterised at the time as about 11% and 18% of the English overwintering population respectively. The two percentages don't share a denominator (2,000 at 11% implies about 18,200 birds, 3,000 at 18% about 16,700), but both point to a wintering population then taken as a little over 20,000. Natural England's current estimate is 37,257, nearly double it, and although the 2013 review considered the ceiling and recommended no increase, that was before the current population estimate, and no public document shows it being reconsidered against the newer figure. Defra's 2013 evidence summary calls it a threshold set by policy and stresses it isn't a formal quota, since the number licenced is based on birds visiting the fishery and present in the local area.

The 2013 review recommended no increase, concluding that licenced shooting at around existing levels was unlikely to result in a significant decline in the cormorant population, and noting that a national cull would be inconsistent with the Birds Directive.

The cap is now generally cited as 3,000, and the Angling Trust reported in 2020 that the scheme remains in place with a national cap of 3,000 in any one year. The date the ceiling was formally raised isn't documented in any Defra or Natural England source that could be found. One advocacy briefing attributes it to 2006. That gap is worth noting rather than filling.

Against a cited cap of 3,000, the 2025 licenced total of 3,460, or 3,615 including amendments, sits above it. So do 2023 at 3,149 and 2019 at 3,015, so 2025 is the largest overshoot rather than the only one. Two caveats belong with that observation: licenced isn't killed, and a licence year spans two calendar years.

No national numerical cap was found in published sources for Scotland, Wales or Northern Ireland.

The European framework#

The great cormorant appears in no annex of the Birds Directive. It isn't in Annex I, not Annex II, not Annex III. The only cormorants in Annex I are Phalacrocorax aristotelis desmarestii and Phalacrocorax pygmeus. The great cormorant was removed from Annex I by Directive 97/49/EC of 29 July 1997, on the basis that it had reached favourable conservation status.

The consequence is structural. The bird is generally protected under Article 5 but isn't huntable under Annex II and not subject to special conservation measures under Annex I. Control is possible only by Article 9 derogation, which is why every national regime in Europe looks like licencing rather than a hunting season.

Article 5 prohibits deliberate killing or capture by any method, deliberate destruction of or damage to nests and eggs, removal of nests, taking and keeping eggs, and deliberate disturbance particularly during breeding and rearing.

Article 9(1) permits derogation where there's no other satisfactory solution, on grounds including preventing serious damage to crops, livestock, forests, fisheries and water. Article 9(2) requires each derogation to specify the species, the means authorised, the conditions of risk and the circumstances of time and place, the deciding authority, and the controls to be carried out. Article 9(3) requires an annual report from each member state to the Commission.

On scale, the European framework document records nearly 10,000 Article 9 derogations for great cormorant across 2015 to 2023, second only to house sparrow, with 86% of the cormorant's own derogations sought for deliberate killing, more than for any other species. Twenty-two member states derogated, twenty-three counting the UK.

Those figures come from the European Environment Agency's derogations dashboard and carry two limitations the framework document states itself. The dashboard's own published temporal coverage is 2015 to 2019 rather than 2015 to 2023, and UK data covers only the shorter period. More importantly, France does not report, and France is the largest culler in Europe, so the totals exclude its biggest contributor.

Both the framework document and the Commission agree that the number of birds actually killed is unknown and likely considerably fewer than the number permitted.

BirdLife's analysis of the Article 9(3) reporting database estimates 353,636 to 437,927 great cormorants killed under derogations across the EU between 2009 and 2017, with a reported minimum of 276,840 excluding ringing, research and conservation. It gives no per-country breakdown, and reporting completeness varied badly: France had three of nine annual reports accessible, Germany six of nine. The scale of the gap shows in the arithmetic. That estimate averages 39,000 to 49,000 birds a year, while the national figures later in this section give France, Germany and Denmark alone more than 60,000 in recent years.

The European management plan, which has recently moved#

The European Parliament called for a European cormorant management plan in a resolution of 4 December 2008, asking the Commission to fund a population estimation model, define serious damage under Article 9(1)(a), issue derogation guidance and submit a coordinated plan. Parliament repeated the call in 2018 and again in October 2022.

For roughly fifteen years nothing was adopted. Two things have since happened.

The Commission has issued the guidance. Commission Notice C/2026/4094, published 31 July 2026, is non-binding guidance on Articles 5 and 9 of the Birds Directive, and its Annex II is specifically on applying Article 9 derogations to the great cormorant, addressing serious damage to fisheries and agriculture. That annex confirms the Annex I removal described above, footnoting it to Commission Directive 97/49/EC of 29 July 1997, and names the only Annex I cormorants as the Mediterranean shag and the pygmy cormorant.

And a framework plan now exists. The Commission asked EIFAAC, the FAO's European Inland Fisheries and Aquaculture Advisory Commission, for management advice. A stakeholder consultation on 25 April 2025 drew 114 participants from 27 countries, a Brussels conference followed on 3 June, and a final draft went to the Commission in early July 2025. The framework carries the reference EIFAAC/2025/3(Rev.1) and is dated August 2025, though the European Parliament's briefing dates it to November 2025. EIFAAC presented the framework to the European Parliament's fisheries committee on 17 March 2026, seeking endorsement. The Commission confirmed continued engagement, pointed to its Birds Directive guidelines, and raised the option of a single-species management plan under the African-Eurasian Migratory Waterbirds Agreement. The framework has been presented but not endorsed.

Separately, a Council information note of 22 May 2026, presented by Czechia and supported by nine other member states, went to the Agriculture and Fisheries Council as an other-business item, calling for immediate work on a European management plan built on the EIFAAC framework. Ten MEPs across four political groups had sent an open letter on 2 July 2025 calling for a coordinated strategy.

What other countries do#

Figures are patchy and mix shooting, egg oiling and nest destruction, which aren't interchangeable.

Denmark is the best documented. Breeding pairs ran 32,645 in 2020 and 30,281 in 2024, the latter down 2.6% on 2023, with the series stable at around 30,000 to 33,000 since 2014 against a peak near 40,000 in 1996 to 2005. Colony management in 2024 covered 3,959 nests in 13 colonies, about 13.1% of all nests, mostly egg oiling. Birds shot averaged around 3,200 a season from 2007/08 to 2016/17, rose to around 5,390 from 2017/18 to 2022/23, and reached 9,059 in 2023/24 on the preliminary figure, since revised to 9,231, with 8,342 provisional for 2024/25. The national report notes many shot birds originate outside Denmark.

France operates the clearest cap and quota system. The ministerial order of 19 September 2022 set departmental ceilings for fish-farm ponds across 58 departments with a national annual ceiling of 27,892 birds, 24% of the 2021 estimated population, and banned open-water shooting nationwide. The preceding order had allowed 50,283 birds across 86 departments including open waters. Around 74,000 birds were shot across the winters of 2019/20 and 2020/21 combined. The order of 3 September 2025 sets ceilings for 2025 to 2028 at 26,566 a year and 79,698 across the three years, again for fish farms only, across 96 departments of which 66 carry a non-zero ceiling.

Germany reported 19,351 birds killed in 2023 and 16,288 in 2020, against a population of 23,000 to 28,000 breeding pairs described as stable. Control is devolved to the Länder. The European Parliament's January 2026 briefing states that every German state has adopted a cormorant ordinance, though a Bundestag research paper puts it at all the area states bar Hessen and the Saarland, which is eleven of thirteen and leaves the three city-states out. Bavaria permits shooting from 16 August to 14 March within 200 m of waters and outside protected areas, averaging around 8,000 birds a winter since 1996 with a range of 3,686 to 11,061. Germany's are the largest annual kill totals reported to a national parliament, though France's ceilings and its reported take across 2019/20 and 2020/21 are higher.

Poland grants permits through regional environmental protection directorates. Figures circulate for permits issued and birds killed in 2012 but no primary source for them could be found, so they aren't repeated here. On population, HELCOM records 1,800 pairs in 1959 falling to 1,100 by 1970, then 1,300 in 1972 and 1,400 in 1973, and 25,800 in 2006, 27,100 in 2009 and 26,600 in 2012, so the series peaks in 2009 and is falling by 2012.

Sweden licences through county boards, with 2,000 to 9,000 birds shot a year since the mid-1990s and an average of 3,208 eggs oiled a year in the Stockholm archipelago between 2001 and 2010.

Finland issues very few derogations, and the gap between permitted and taken is the widest on record anywhere: 204 birds shot of 990 permitted across 2010 to 2012, and 10 of 700 permitted across 2015 to 2017.

The Czech Republic removed the great cormorant from its list of specially protected species in 2013, with compensation claims for damage valid only to 31 March that year. Delisting didn't make the bird freely shootable. A Czech practitioner account sets out the conditions that still apply: a derogation from the nature protection or gamekeeping authority, a hunting licence, no more than 20% of a flock, a season of 1 August to 30 April, and nothing inside protected bird areas. The same account gives annual figures of 6,050 in 2012 falling to 4,420 in 2019, and seasonal figures rising from 5,138 in 2018/19 to 17,421 in 2022/23, though those are tied to a shooting-support scheme rather than presented as national totals.

Italy licences regionally. A figure of around 14,000 birds shot between winter 1994/95 and 2004/05 circulates from a 2005 conference poster, but no accessible copy of it could be found, so it shouldn't be relied on. The one current Italian data point located is Emilia-Romagna capping culls at 10% of the previous winter's population. Italy is otherwise undocumented for the modern period.

The Netherlands operates small-scale provincial derogations. Figures circulate but no national total from a primary source could be found.

Legal challenge#

France produced the most significant cormorant-specific ruling, and it's narrower than most accounts of it suggest. On 8 July 2024, in case 468607 joined with 472238 and 474705, the Conseil d'État annulled the 2022 order in so far as it set no departmental ceiling for destruction on open waters in the département du Doubs. The national federation's first application was withdrawn, but its second, against the implicit refusal to amend the order, succeeded and was annulled with costs; the local angling associations' claim also succeeded. The finding was an error of assessment in setting no ceiling where no study had ruled out the risk to protected fish, not a manifest error, which is a materially more deferential standard, and not a general holding that open-water shooting can't be restricted. Ministers were given four months from notification to issue an amending order. A framework order followed on 24 February 2025, around three and a half months past that deadline, allowing prefects to set departmental ceilings of up to 20% of the departmental wintering population, or 30% where no fish-farm ceiling applies.

Earlier, on 5 April 2022, the administrative tribunal at Grenoble annulled a 2019 prefectoral order authorising destruction of 88 birds in Savoie as disproportionate to the objective pursued, one of many annulled that year.

Poland. A 2013 Supreme Administrative Court decision on a cormorant permit is reported in the Polish legal press, but the judgment itself couldn't be located, so its holding isn't described here.

No cormorant-specific judicial review in England could be found. The nearest useful precedent is R (McMorn) v Natural England [2015] EWHC 3297 (Admin), where the court quashed a refused buzzard licence on the basis that Natural England had set an evidential threshold beyond what the Directive and the Act require, and had no lawful basis for treating buzzards differently from other common species causing serious damage. Cormorants are not the subject of the case, but they run through its comparator reasoning: the judgment discusses Defra's cormorant policy, its three tests including inference of damage from circumstantial evidence, the national cap, and licences granted at a trout fishery. The reasoning bears directly on consistency of treatment between species.

The 2019 and 2020 general licence episode wasn't a cormorant case. After Natural England revoked general licences for other species, the Angling Trust asked for cormorants to be added to a general licence under section 16(1)(k). Ministers declined on 16 November 2020, and the individual A06 scheme with its national cap remained. The published sources give no official ministerial reasoning.

No cormorant-specific infringement proceeding or formal complaint at EU level could be found on the public record.

7Control and deterrence#

What shooting does#

The evidence here is better than in most of this subject, and it supports a distinction that both sides tend to collapse. Local short-term displacement is demonstrated. Sustained local reduction and population-level reduction aren't, which is a statement about what has been shown rather than about what is happening.

Defra-funded trials ran thirteen six-week experiments over two winters at river and stillwater fisheries in England and Wales, comparing control sites, lethal shooting of up to twenty birds over two weeks, and non-lethal shooting with blanks. Control sites rose 25% and 40% while lethal and non-lethal sites combined fell 45% and 41%, with the reduction estimated at over 50%. Shooting worked better at smaller sites.

There was no significant difference between lethal and non-lethal shooting. The belief that killing birds enhances the scaring effect is, in Defra's own words, generally accepted but equivocal for cormorants. It has remained unresolved for more than twenty years.

And numbers returned to pre-treatment levels within two weeks at two sites and six weeks at four others. Long-term habituation to shooting has never been investigated.

At population level, a BTO study found no evidence that licenced control reduces local cormorant population size in winter: killing in one winter showed no evidence of lowering counts at local sites the following winter. Some results linked population growth to higher-intensity control, which the authors suggest may mean either that control increases dispersal and creates apparent growth, or that licences are issued reactively after numbers rise. They note birds may move to unmonitored areas, and that whether control reduces predation at fisheries, and its cost-effectiveness, remain unanswered.

In Denmark, three years of work on two large fjords found that coordinated shooting could make cormorants leave earlier in autumn, but that large effort without coordination in time and space, or not aimed at roosts, didn't achieve the desired reduction, and reductions weren't sustained in later years.

In Bavaria in 1996/97, over 6,000 birds were shot without reducing the average winter population.

Practitioner experience points the other way and should be reported as what it is. Angling Trust fishery management advisers told NRW that it's rare for a fishery to reduce predation to acceptable levels without some licenced shooting, that shooting to scare reduces visits when done regularly and timed to arrival, and that coordinated shooting on the last birds in a flock prevents short-distance relocation along a river. That is field experience, not controlled study, and it concerns effect at a fishery rather than effect on a population. The two claims aren't in conflict.

Non-lethal methods#

Habituation is the central problem. Defra's position is that with many deterrents the impact is likely to diminish as habituation occurs with any scaring technique not reinforced by a demonstration of real danger. Auditory deterrents habituate over weeks or less. Fixed visual deterrents work up to about 200 m and typically for days to weeks unless moved and combined.

Deterrence works best before or as birds arrive, before they learn the site is good foraging.

Success depends on birds having a better alternative, which means managing one fishery can require knowledge of a much wider area, and a method that works at one site may simply export the problem.

Full netting is the one method described in the literature as reliable exclusion, and the one birds don't habituate to in the ordinary sense, because it's a physical barrier rather than a deterrent. It isn't absolute: cormorants have been recorded learning to walk in under nets at pond aquaculture. It has been applied to water bodies of several hectares and to linear waterways of several kilometres. Mesh guidance runs 5 to 7 cm, dark, thick and non-monofilament, though larger mesh of 15 to 50 cm can suffice for cormorants. Maintenance isn't optional: thin, light-coloured, large-mesh monofilament netting caused more entanglement and death, and torn netting can turn a pond into a trap where birds take more fish than at an unprotected water.

Fish refuges carry the strongest controlled result found in this literature. A Defra-funded cross-over trial on two adjacent ponds in Berkshire over the winters of 2003 and 2004, using twelve mesh-covered cages with shade cloth and saplings inside, found cormorant foraging success about 67% lower, fish loss about 79% lower and about 77% fewer visits. The caveats are substantial: two ponds, one location, two winters, and Defra's own summary notes that field results for fish stocks were more equivocal. Mesh of around 10 cm excludes cormorants. Refuges deteriorate and need maintenance, suit smaller coarse species at smaller stillwaters, and are explicitly not suitable for rivers, with cost and practicality ruling out larger stillwaters. Among licencees who used them, 65% considered them successful and only 3% had tried them and found them ineffective; non-use was driven by unsuitability at 44% and interference with angling at 17%.

Lasers have the most consistent published results of the scaring tools. A UK trial with a red 5 mW laser scared most cormorants from a night roost within twenty minutes, and repeated evenings caused temporary desertion. Green lasers at 20 to 30 mW are effective over one to two kilometres. Above 50 mW they are no more effective and more dangerous. They're target-selective, so non-target species are unaffected, which is a real advantage over acoustics. Limitations: some birds fail to leave when facing away from the light, they're ineffective in daylight, mist or fog, and birds usually leave briefly and return, so consecutive evenings are needed. On safety, one commercial laser was assessed as safe if not aimed at an unprotected eye within 155 m, a distance much reduced through binoculars. Hand-held pointers can be as effective as expensive laser guns.

Of the visual deterrents, human disturbance is rated the most effective by every source consulted here that ranks them. It works when birds associate people with danger, and timing matters because cormorants feed most actively just after dawn. The clearest demonstration of how quickly they adapt comes from an Israeli study where birds shifted feeding from early morning to early afternoon, when workers took lunch. The cost is labour, and one European source records a human cost too, with farmers reporting stress and health problems from watch-keeping.

Scarecrows and effigies perform poorly and inconsistently. Of fourteen US hatchery managers, one reported high success and six no effect. In an Italian survey of thirteen freshwater farms, four reported no effect, three success lasting days and two benefits lasting weeks or months. Birds learn effigies aren't a threat, and some come to associate them with good feeding. Raptor models work at some sites and not others, with a plausible mechanism: the avoidance response is partly learned, and white-tailed eagles are absent from many cormorant areas.

Gas cannons and pyrotechnics habituate quickly if not varied. Birds have been seen perching on cannons. Pyrotechnics are useful at small roosts but failed outright at a large roost of 2,500 to 3,000 birds, with no clear or lasting effects.

Distress calls are short-term at best, and the product category has a problem at its root: commercial cormorant distress calls trialled in Italy turned out to be nestling begging calls. Adult cormorants rarely produce distress calls and repeated attempts to record them failed.

Airborne ultrasonic devices. Most birds don't hear above 20 kHz, so no mechanism has been put forward that would make these work, no trial found here reports deterrence, and ultrasound attenuates quickly in air. The underwater acoustics described below operate at ultrasonic frequencies but on a different principle, in a different medium, against a diving bird.

Underwater acoustics based on killer whale calls worked at 90 kHz, but only on ponds below about 300 kg per hectare of fish density, and birds habituated at extensive aquaculture.

Wires and lines work on the principle that cormorants need 8 to 12 m of open water to take off, so wires impede landing, feeding and departure. Trials in several countries showed they can substantially reduce impacts, but results have been very variable and some trials showed no benefit. The only published spacing figure found is around 2 m for cormorants, from a single government factsheet. Wires don't fully exclude birds and efficacy may fall as birds learn to avoid them.

Dogs divide the sources. One government-linked review says efficacy against cormorants isn't known; another rates border collies among the best options, with no habituation. The evidence is thin either way.

Conditioned taste aversion produced the longest effect recorded for any non-lethal method. Captive birds that ate fish dosed with carbachol avoided brown trout for seven months without reinforcement, the full duration of the trial, though they still ate other species. It's captive-only and judged difficult to apply in the field, and it has had less attention than its result would suggest.

GenusWave TAST is a marine mammal deterrent. The company's materials describe deterring marine mammals, and where species are named they're seals and other pinnipeds. It works on the mammalian acoustic startle reflex. No published evidence for its use against cormorants or any other bird could be found. Any claim that TAST is a cormorant deterrent is unsupported by anything in the public record.

Stocking strategy has limited reach. Field visits found stocking fish over 700 g and delaying introductions worked at one reservoir, and at another carp grew too large for cormorants to take. 56% of sites reported stocking changes. But the approach applies only to put-and-take fisheries, not to natural or river fisheries, which is where most of the conflict sits. Note also that high stocking is itself part of the causal chain: Defra recorded that particularly high stocking levels are felt to have contributed to the problem by attracting cormorants to some areas.

Roughly how long each method lasts#

8Contested claims#

"It's a sea bird that doesn't belong on fresh water"#

This is the weakest claim in the debate, and it's worth being precise about who makes it. It isn't the Angling Trust's formal position. Their 2019 impact report says carbo is traditionally thought of as a coastal bird and that sinensis hunts largely inland, and notes inland colonies have existed since 1981. Their 2024 briefing attributes growth to natural range expansion. The sea-bird framing circulates in angling media and forums rather than in the briefings.

Inland tree-nesting in East Anglia is documented in the 1540s. Until the 1940s inland breeding was recorded at five English sites across Suffolk, Norfolk, Kent, Dorset and Cumbria, with breeding at several halted probably by human persecution. Inland breeding occurred in Wales for centuries, and in Scotland and Ireland. A leaflet co-sponsored by the Environment Agency, English Nature, the RSPB and angling bodies states that cormorants have always been found inland in Britain and that numbers have been controlled since medieval times. Sinensis is more generally confined to brackish and fresh water, so a freshwater habit is the subspecies norm rather than an aberration.

Where the evidence runs out is on scale. The pre-1940s records are few and scattered. They establish that inland occurrence wasn't unknown; they can't establish that historical inland abundance resembled today's. The honest statement is that inland occurrence is ancient and the current scale is new.

"They moved inland because the sea emptied"#

No peer-reviewed source identifying marine stock collapse as a driver could be found, and the claim doesn't appear in any formal Angling Trust, Defra or Environment Agency document consulted here.

The documented drivers are sufficient without it. Legal protection arrived in the Netherlands in 1965, Denmark in 1971, the EC Birds Directive in 1979 and the Wildlife and Countryside Act in 1981, after which continental numbers rose from 5,800 pairs in eight colonies in 1978 to 61,720 pairs in 116 colonies by 2005 across the Netherlands and Denmark. Organochlorine pesticides had suppressed breeding through the 1950s and 1960s and the Baltic into the mid-1970s. Reservoir and gravel pit creation from the 1960s added inland habitat. Colony saturation drove dispersal, with Abberton-hatched birds dispersing at 7% in the mid-1990s rising to 18% by 1998. Milder winters reduced ice cover and increased northern overwintering.

There's a real nuance that is different from the claim. Over-fishing is linked to fewer large predatory fish and more small prey, which favours cormorants. That is a food-web composition argument, not an emptied sea. And carbo does move inland in winter for a documented reason: coastal fish move to deeper water and weather limits foraging. That is normal seasonal behaviour.

The claim is unsupported. Nothing tests it directly, so it's unfalsified rather than disproven, but the evidenced mechanisms are sufficient without it.

"Sinensis is non-native"#

Wrong as a matter of taxonomy and law. Sinensis is a subspecies of a native species, described in 1796, long resident in continental Europe. Neither subspecies appears on any invasive species listing consulted. The species is Least Concern on the IUCN Red List and green-listed on Birds of Conservation Concern in the UK.

What is genuinely recent is sinensis breeding in Britain, from 1981, founded mainly by birds from the Netherlands and Denmark.

The heavy qualification is that British inland colonies are genetically mixed and increasingly carbo, and that inland nesting isn't a sinensis-only trait, since carbo shows an increasing trend for inland breeding.

"Shooting doesn't work because others replace them"#

Covered in section 7. The short version is that the claim holds for populations and is unproven for fisheries, and the people arguing past each other are usually answering different questions. Local displacement is demonstrated. Sustained local reduction isn't. Population-level reduction isn't. Whether displacement reduces fish losses at the fishery hasn't been established either way, and no study found here addresses whether it reduces losses across a catchment.

One qualifier is often dropped. Defra's own assessment notes that the UK experiences only modest immigration and emigration compared with mainland Europe, which weakens the pure-replacement story for the UK specifically relative to, say, Bavaria.

"Numbers are at an all-time high"#

Not supportable as a statement about a natural baseline, because no pre-industrial baseline exists.

The north-west European population was around 5,000 breeding pairs in the early 1960s and around 13,500 by 1981. The species came close to extinction in Europe in the first half of the twentieth century, and the Netherlands fell to around 600 to 800 pairs in the late 1960s. Current numbers are far above that trough. But the trough was produced by persecution and organochlorines, so "all-time high" measures against an artificially suppressed baseline rather than a natural one.

"Declining" is equally unsupportable for UK wintering birds. The defensible statement is that numbers are substantially increased from a persecution and pesticide-suppressed twentieth-century low, with recent levelling off in England, regional declines in Wales and Scotland on different time windows, UK breeding numbers slightly down since Seabird 2000, and Baltic numbers stabilised since 2006 with Denmark and Sweden falling.

"Cormorants are being persecuted"#

The strongest form of this argument is historical and well documented. Numbers have been controlled in Britain since medieval times, inland breeding at several of the five pre-1940s English sites was halted probably by persecution, and European populations were low through the nineteenth and twentieth centuries from habitat loss and persecution. That history is what produced the low baseline against which current numbers are judged.

On current wildlife crime, the evidence is genuinely thin and this should be stated rather than filled in. Defra's position is that no official figures or research exist on the scale of unlawful killing, and the 2013 review listed it as a named evidence gap. The RSPB's 2011 State of the UK's Birds reported numbers rising from the mid-1980s to the mid-1990s and falling thereafter, naming shooting, licenced and unlicenced, as a probable contributory factor. That decline doesn't appear in the index series in section 3, and the attribution is in any case an attribution rather than a count. Wildlife and Countryside Link's most recent wildlife crime report carries no mention of cormorants or fish-eating birds, and no cormorant category at all. Wildlife crime isn't a notifiable offence, so there are no central national records. No documented UK dataset of illegal cormorant roost destruction could be found.

The absence of data isn't evidence of absence, given non-notifiable status and Defra's own identification of the gap. But no documented incident series exists, and an article that implies one is overstating the record.

"Habitat improvement and refuges solve it"#

Refuges have the best controlled evidence of any single non-lethal method, at around 79% lower fish loss, but from one two-pond trial with equivocal stock-level field results, and they're explicitly unsuitable for rivers and impractical at larger stillwaters. "Solve the problem" overstates that. "Almost totally ineffective" contradicts both the controlled trial and the 65% licencee success rate.

Cost is a real gap. No published per-hectare or per-site cost figures for refuges, netting or wires could be found anywhere in this literature.

"Numbers rose and the fish declined, so one caused the other"#

Predators track prey abundance. In the Swiss study, young-grayling biomass predicted cormorant predation intensity rather than the other way round. In the Baltic, cormorants shot in one sound were in good condition with almost only stickleback in their stomachs, suggesting attraction to abundant prey rather than causation of its abundance.

Coincident timing isn't causation, and it's used as though it were by people arguing in both directions.

9What isn't known#

A list of genuine gaps. Most have been named as gaps by a government body, a peer-reviewed source or the people who produce the data; a few are simply places where no source could be found, and those are marked as such.

Cost. Almost no published cost figures exist for any management method. No per-hectare refuge, netting or wire costs could be located. Defra lists cost-effectiveness as inadequately evaluated.

Whether killing enhances scaring. Unresolved since the Defra-funded trials found no significant difference between lethal and non-lethal shooting. Still unanswered.

Long-term habituation to shooting. Named as uninvestigated by the trials that measured the short-term effect.

Catchment-scale effect. No study found here establishes reduced fish losses across a catchment from any intervention.

Where displaced birds go, and how long a site stays unattractive after scaring.

Compensatory recruitment. No empirical UK study. The European framework document carries no discussion of compensatory mortality or of breeding response to culling.

The scale of unlawful killing. No official figures or research, and no cormorant category in wildlife crime reporting.

Refusal rates for cormorant licences in England. Not published in any year.

Birds actually shot in England after 2011/12. No published series.

Cormorant licence statistics for Wales, Northern Ireland and the Republic of Ireland. None published.

The date England's licencing ceiling was formally raised from 2,000 to 3,000.

The UK wintering split between the two subspecies. Natural England states explicitly that it's unknown.

Arrival and departure dates for continental birds in Britain, and named flyways, median and maximum individual movement distances and quantified winter roost fidelity. Colony resighting data give mean distances, but the sources that would settle the rest couldn't be obtained.

Any UK wintering population total more recent than a 2012/13 to 2016/17 basis.

Which of the two official England wintering series is right. Indexed to a common year, Defra's 2013 revision sits about 15% below Natural England's series. The two embody different judgements about how much of the inland population WeBS misses, no published document reconciles them, and every percentage in the licencing debate depends on which is used.

Confidence intervals for the published cormorant indices. None are produced, and statistical significance of trends isn't assessed.

Gut passage time in this species. Not quantified in the accessible literature. The figures for how fast a cormorant digests a fish that circulate in angling writing have no primary support that could be found.

A pan-European winter roost total from the January 2013 census. The count happened. No published total could be located.

How many cormorants are actually killed under derogation across Europe each year. Derogation counts exist but France doesn't report into them, and France is the largest culler. The one EU-wide kill estimate, BirdLife's 353,636 to 437,927 across 2009 to 2017, comes with reporting gaps the analysis itself documents, and both the Commission and the European framework document say the real figure is unknown.

Annual national totals for Poland, Italy and the Netherlands. Figures circulate for each but no primary source for them could be found.

Impact magnitude itself. After 603 publications, only 22 were rigorous enough to pool, and the pooled effect wasn't statistically significant. The absence of a demonstrated population-level effect isn't evidence that there's none. The authors who looked attribute it to the difficulty of isolating predation from every other pressure.

10Sources#

Population and monitoring#

• JNCC, Official statistic on wintering waterbird population trends, 16 April 2026
• JNCC, Official statistic on provisional Cormorant population indices for England, 23 July 2026
• Frost et al. (2026), Waterbirds in the UK 2024/25, BTO/RSPB/JNCC/NatureScot
• BTO, Wetland Bird Survey: Survey Methods, Analysis and Interpretation, v3.0, April 2026
• Woodward et al. (2020), APEP4, British Birds 113(3): 69-104
• JNCC, Seabirds Count comparative dataset (2015-2021)
• Underhill (1989) and Underhill & Prys-Jones (1994), the imputation method behind the WeBS indices
• Seabird Monitoring Programme Report 1986-2021
• Chamberlain, Austin, Green, Hulme & Burton (2013), Bird Study 60(3), and BTO Research Report 612 (2012)
• NRW Evidence Report 592, A winter census (2020/21) of cormorant and goosander in Wales
• NRW Evidence Report 593, Spring 2021 survey of Cormorant and Goosander on the River Usk
• Newson et al. (2013), British Birds 106: 737-743; BTO Research Report 433 (2006)
• Bregnballe et al. (eds.) (2014), Breeding numbers of Great Cormorants in the Western Palearctic, 2012-2013, DCE Scientific Report 99
• Aarhus University DCE Technical Reports 328 and 360 (Danish national monitoring)
• HELCOM, Population development of the Great Cormorant

Biology#

• Grémillet, Chauvin, Wilson, Le Maho & Wanless (2005), Journal of Avian Biology 36: 57-63
• Ribak, Weihs & Arad (2005), Journal of Avian Biology 36: 89-95; (2008) Journal of Experimental Biology 211: 3009-3019
• Grémillet, Tuschy & Kierspel (1998), Functional Ecology 12: 386-394 (calculated insulating air layer)
• Stangier et al. (2023) and Holste et al. (2023), Journal of Ornithology 164
• Sellers (1995), Ardea 83(1): 27-36
• White, Martin & Butler (2008), Journal of Avian Biology 39: 576-578
• White, Day, Butler & Martin (2007), PLoS ONE 2(7): e639
• Elowson (1984), The Auk 101: 371-383
• Muzio & Rubega (2024), Journal of Avian Biology 2024(11-12)
• Grémillet et al. (2005), Biology Letters 1(4): 469-471
• Wright (2002), PhD thesis, University of Glasgow
• Kennedy & Spencer (2014), Molecular Phylogenetics and Evolution 79: 249-257
• Winney et al. (2001), Heredity 86: 45-53; Goostrey, Carss, Noble & Piertney (1998), Molecular Ecology 7: 329-338
• EURING list of longevity records for European birds (2017)
• Natural England, Definition of Favourable Conservation Status for Great Cormorant (RP2952)
• Cramp & Simmons (1977), Birds of the Western Palearctic
• Dunning (1992), CRC Handbook of Avian Body Masses; Koffijberg & Van Eerden (1995) on subspecies size
• Lekuona (1999), Ardeola 46(1): 13-21 (group foraging)
• Craik & Bregnballe (2008), Seabird 21: 93-97 (apparent second broods)
• Oehm et al. (2022) on pellet provenancing; Skov et al. (2014) and Källo et al. (2020, 2023) on size selectivity
• Rijke (1968, 2023) on feather structure, and Elowson's test of it
• Bregnballe & Frederiksen (2006), Wildlife Biology 12: 143-150 (net entrapment)
• Grémillet et al. (1996), Journal of Field Ornithology 67(4): 637-648 (nest-balance intake)
• Stewart et al. (2005), Loch Leven consumption
• Wright (2003) and Frederiksen et al. (2003), on bird-count error

Diet and impact#

• Carss (1997), Cormorant Research Group consensus on diet and food intake
• Carss & Russell (2022), NRW Evidence Report 591
• Grémillet, Wright, Lauder, Carss & Wanless (2003), Journal of Applied Ecology 40: 266-277
• Russell, Cook, Ives & Davison (2022), Ardea 109
• Marquiss et al. (1998); Bearhop et al. (1999), Journal of Applied Ecology 36: 75-84
• Ovegård, Jepsen, Bergenius Nord & Petersson (2021), Fish and Fisheries 22(3): 605-622
• Jepsen, Ravn & Pedersen (2018), Hydrobiologia 820: 189-199
• Suter (1995), Journal of Applied Ecology 32(1): 29-46
• Skov et al. (2014), Journal of Limnology 73(1)
• Källo et al. (2020), ICES Journal of Marine Science 77(2); (2023) Ecology of Freshwater Fish 32
• Kennedy & Greer (1988), Aquaculture and Fisheries Management 19
• Jepsen, Klenke, Sonnesen & Bregnballe (2010), Marine and Freshwater Research 61(3)
• Östman et al. (2013), PLOS ONE 8(12): e83763
• Hansson et al. (2018), ICES Journal of Marine Science 75(3), with the Heikinheimo exchange
• Grémillet, Enstipp, Boudiffa & Liu (2006), Marine Biology 148(5)
• Kortan, Adámek, Flajšhans & Piačková (2008), Knowledge and Management of Aquatic Ecosystems 389
• Feltham et al. (1999), River Wye
• IUCN Red List assessment, Anguilla anguilla (2020)
• Oehm et al. (2022), Journal of Wildlife Management 86: e22248 (strontium provenancing of pellet otoliths)
• Haeseker et al. (2020), Payton et al. (2020) and Carothers et al. (2021), the Columbia River additive-or-compensatory dispute
• Jepsen, Cowx & Van Anrooy (2025), EIFAAC conference presentation (South Tyrol and Danish electrofishing figures)
• Marquiss & Carss (1994), prey size ranges by habitat

Management and law#

• Wildlife and Countryside Act 1981, sections 1, 16 and 21
• Wildlife (Northern Ireland) Order 1985, articles 4 and 18
• Directive 2009/147/EC (Birds Directive); Directive 97/49/EC
• Commission Notice C/2026/4094, 31 July 2026, Annex II
• European Parliament resolution of 4 December 2008 (2008/2177(INI)); resolutions of 2018 and 2022
• EIFAAC, Framework for a European Management Plan for the Great Cormorant, EIFAAC/2025/3(Rev.1), version August 2025
• European Parliament Research Service briefing on cormorant predation, January 2026
• Council information note 9449/26, 22 May 2026
• European Environment Agency, derogations and exceptions dashboard
• Aarhus University DCE Technical Report 360 (Denmark 2025)
• Conseil d'État 468607, 8 July 2024; Tribunal administratif de Grenoble 1908141, 5 April 2022
• Natural England, Licences for the control of birds 2014 to 2025
• NatureScot, Lethal Control Data June 2019 to December 2025
• Defra (2004) policy review setting the English licencing ceiling, and Defra (2013), Impacts of predation by fish-eating birds on inland fisheries in England, review and evidence summary
• Nature Conservation (Scotland) Act 2004; Wildlife Act 1976 (Republic of Ireland), section 42
• NatureScot, NRW, DAERA and NPWS licencing guidance pages
• Council Regulation (EC) No 1100/2007 on the recovery of the European eel; Environment Agency Eel Management Plans
• Ministerial decision of 16 November 2020 declining to add cormorants to a general licence
• Natural England guidance, Fish-eating birds: licence to protect your fishery (A06 and A07), updated 3 August 2026
• RSPB species account for cormorant; RSPB, State of the UK's Birds 2011
• Birds of Conservation Concern; the IOC World Bird List
• Angling Trust cormorant impact report (2019), briefing (March 2024) and account of the 2020 ministerial decision
• Cormorants: The Facts, Moran Committee leaflet co-sponsored by the Environment Agency, English Nature, the RSPB and angling bodies
• Wildlife and Countryside Link, Wildlife Crime 2023
• Kratochvíl, Czech practitioner account of cormorant management and shooting figures
• GenusWave published materials
• NRW, Controlling fish-eating birds on a coarse or trout fishery
• Parrott et al. (2003); McKay et al. (1999)
• Chamberlain et al. (2013), licenced control and local population size
• Bregnballe et al. (2015), Pest Management Science 71(2)
• Russell et al. (2008), Fisheries Management and Ecology 15: 189-198
• NRW Evidence Report 594, appraisal of lethal and non-lethal control
• INTERCAFE Cormorant Toolbox; REDCAFE Final Report (2003)
• R (McMorn) v Natural England [2015] EWHC 3297 (Admin)
• BirdLife International, analysis of Article 9(3) derogation reporting
• Arrêtés of 19 September 2022 and 3 September 2025 (France); Bundestag answers and Bayerisches Landesamt für Umwelt (Germany); HELCOM (Sweden, Estonia, Baltic); ELY Centre material (Finland); regional press and court reports (Poland, Czechia); Volponi, Andreotti & Tomasini (2005) conference poster (Italy)
• McKay et al. (1998) on conditioned taste aversion; NRW Deterring birds factsheet

Last reviewed October 2026. Where a figure here is superseded by a new official release, the release date is given in the text so it can be checked.