Narwhals enter the fjord as soon as the ice opens, and leave before it closes again

Arctic migrations are readily described as set by the calendar. A study published open access on 21 May 2026 in Scientific Reports shows that in the narwhal, it is the state of the ice that gives the order, to within a few days. And its conclusion about the future is the opposite of what one would expect: the disappearance of the sea ice would not drive narwhals out of their fjords, it would hold them there.

Three and a half years of listening under the ice

Julie Sofie K. Larsen and her colleagues, from the Greenland Climate Research Centre and Aarhus University, worked in Inglefield Bredning, a glacial fjord in north-west Greenland. It is about a hundred kilometres long, up to twenty-six wide, and more than nine hundred metres deep. Fast ice closes it from December to May or June. In summer it holds a subpopulation estimated at about 8,300 narwhals.

Six autonomous acoustic recorders were moored there between 2022 and 2025, suspended between 95 and 575 metres deep, from the edge of the North Water polynya to the middle of the fjord. They produced close to 10,020 hours of effective recordings.

The logistics are worth mentioning, because they explain what the study contributes. In order to have data for the start of summer, the instruments were deployed in spring, through more than a metre of sea ice, in collaboration with Inuit hunters from the region, by dog sledge, with drills and traditional tools. The autumn recoveries were made in small boats. That is what made it possible to cover the period nobody had been recording until then.

How clicks are turned into presence

The narwhal hunts by echolocation, emitting highly directional clicks. Detection software picks them out of the continuous signal, then a classifier retains those whose peak frequency lies between 20 and 80 kilohertz and whose duration stays below one millisecond, the range characteristic of the species.

Two indicators are drawn from this. The number of clicks per day measures general presence. The number of hours containing buzzes, defined as any second counting at least fifty clicks, serves as a feeding indicator: a predator closing on its prey speeds up its click rate until it produces a continuous buzzing. Our page on passive acoustic monitoring describes these methods.

One important calculation frames the interpretation: the detection range was estimated at about 5 kilometres. Adding the 700 metres a narwhal can travel under the ice before having to breathe gives 5.7 kilometres. Yet detections occurred while the ice edge was about 6 kilometres away, which indicates that the animals were moving actively under the sea ice to reach openings.

What the figures say

The authors classified each day according to four ice stages, then modelled the number of clicks and buzzes as a function of that stage.

Complete ice (spring)	3,048 clicks per day on average
Ice edge less than 5 km from the mooring	activity sharply up
Ice retreating	50,434 clicks per day, the maximum
Ice-free fjord	18,400 clicks per day
Autumn, ice forming	38,118 clicks per day
Autumn, complete ice	3,402 clicks per day

The gap is a factor of sixteen between complete ice and retreating ice. It is one of the clearest signals that can be obtained in field ecology.

Why retreating ice, and not open water

The peak occurs neither under closed sea ice nor in open water, but during the retreat phase. Two explanations are put forward, and they are not mutually exclusive.

The first is protection. The killer whale is the narwhal’s main predator, and it hunts poorly in dense ice, its tall dorsal fin hampering its movement between the floes. A partially covered area offers a refuge the predator does not venture into. Our page on killer whales and our video on their hunting at the ice edge show the limits of that exercise.

The second is feeding, and the argument is precise. Juvenile polar cod live clinging to the underside of the sea ice from the larval stage up to a size of ten to twenty-five centimetres, after which they move to deep water. And that is exactly the size class found in narwhal stomachs. The ice edge is therefore not only a shelter: it is a larder. Our pages on diets and climate change set out this dependence on ice at the bottom of the chain.

The drop in open water is not a drop in activity

This is the point a quick reading of the abstract misses. When the fjord clears completely, detections fall at the moorings. The authors do not conclude that the narwhals become less active, but that they have moved elsewhere, deeper into the fjord, towards the glacier fronts.

These fronts are zones of high productivity: melting there drives an upwelling of nutrient-rich deep water into the lit layer, which concentrates prey. Local hunters’ observations and earlier work indicate that narwhals gather there as soon as the fjord is accessible.

In other words, a drop in signal on a hydrophone does not mean a drop in presence in the area. This is a structural limit of passive listening, which the authors accept and which is a reminder of why it combines poorly with an estimate of population size.

A telling asymmetry between spring and autumn

The finest result of the study rests on a comparison between the two transition seasons.

In spring, the narwhals enter immediately after the opening, as soon as the passage is practicable. In autumn, by contrast, activity at the moorings is already very low when the ice reaches the stations. So they do not stay until the last moment: they leave before access becomes constrained.

The explanation proposed is the risk of entrapment. A narwhal shut in by a rapid refreeze in a bay with no opening dies, and such mortality episodes are documented. A selection pressure therefore operates in favour of those that leave early.

Detections do not stop everywhere for all that. At the outermost mooring, in the polynya, they are almost continuous all year round. Some of the animals stay at the entrance to the fjord, waiting for the first opening.

What the melt would actually do

One expects an Arctic species to lose its habitat when the ice disappears. The authors describe a different, and more devious, mechanism.

If the fjord becomes accessible all year round, the narwhals will no longer have any reason to leave it. The constraint that pushes them out to the open sea in winter today would disappear, and their residence in the fjord would lengthen.

Yet it is on the open sea, on the shared wintering grounds, that animals from different fjords meet, and it is probably there that breeding takes place. Removing the seasonal movement therefore amounts to removing the genetic mixing between subpopulations.

The concern is all the more serious in that the narwhal already shows low genetic diversity, established by the sequencing of its genome. A species with little diversity that fragments into isolated groups loses its capacity to adapt at the very moment when its environment is changing fastest. Our pages on cetacean demographics and conservation statuses set this kind of risk in context.

A second consequence is flagged, on a longer timescale. Glaciers that end in the sea sustain the productivity of the fjord; if they retreat until they ground on land, that input ceases and the food resource degrades. A fjord accessible all year round could therefore become a less nourishing fjord.

The limits

Four reservations frame these results, and the authors set them out themselves.

1. Acoustics does not follow individuals. It measures a presence within hydrophone range, not journeys. Telling apart animals that stay from animals that take turns would require tagging.

2. A single fjord system. Nothing guarantees that the relationship takes the same form elsewhere in the range, from Arctic Canada to Svalbard. Work carried out in the Fram Strait suggests, moreover, that some populations remain resident all year round.

3. Feeding depth is not measured. The link with juvenile polar cod rests on a match of timing and prey size, not on a direct observation.

4. Three years do not make a trend. The study establishes a relationship between ice and presence, not a change over time. Detecting a shift in timing would require a much longer series.

What this work owes to the hunters of Qaanaaq

One last point deserves to be noted, because it is rarely visible in a publication. The moorings could not have been deployed in spring without the Inuit hunters of Avanersuaq, who provided the sledges, the knowledge of the state of the ice and the ability to drill through more than a metre of thickness.

Their observations also enter into the argument: it was they who reported having seen narwhals swimming under the consolidated ice edge while feeding, an observation cited as a reference in the paper on the same footing as a publication.

It is a concrete illustration of what Arctic research owes to local knowledge, and of what monitoring conducted by satellite alone would lose. Our pages on tracking technologies, the acoustic profile and how to act set these approaches in context, and our video on the narwhal’s tusk returns to the anatomy of the species.

Source: Larsen J. S. K., Ladegaard M., Madsen P. T., Olsen S. M. and Simon M., “Sea ice dynamics structure narwhal presence and seasonal movements in a Northwest Greenland fjord system“, Scientific Reports, volume 16, article 23172, 21 May 2026, open access.