Nobody Set Out To Do This

Start with the part that is not in dispute. A humpback whale feeds in cold water at high latitude through the summer, then crosses an ocean to give birth in warm shallow water where there is almost nothing to eat — and does not eat.

The technical term is capital breeder. Roman and colleagues put it plainly: most baleen whales acquire much of the energy required for migration and reproduction in a relatively short period on the feeding grounds, then spend it. They typically fast in winter and through the crossing. Whalers noticed this before biologists measured it — whales killed at the end of the breeding season often weighed significantly less than those killed on the feeding grounds.

The distances are the longest of any mammal. Southern Hemisphere humpbacks have been recorded on one-way journeys of about 8,300 km. Gray whales travel more than 11,000 km between feeding grounds off Russia and breeding areas along Baja California.

So: a very large animal, running on stored fat, crossing an ocean it cannot refuel in, to a destination it will find with a precision we will spend the next seven spreads failing to explain.

The exceptions are real and we are keeping them. Not every baleen whale does this. Some North Atlantic fin whales stay on the feeding grounds year-round on the acoustic record; some North Atlantic right whales do too. Humpback and blue whales in the Gulf of California feed in winter. Roman and colleagues name partial migration and non-migratory populations directly, and say the spatial dynamics of several baleen whales are relatively unknown or unexplored. There is no standard whale here either.

In 2011 Travis Horton and six colleagues published satellite tracks for sixteen humpback whales, tagged off Brazil, Rarotonga and New Caledonia between 2003 and 2010, and looked at how straight the lines were.

They found twenty-eight constant-course segments. Every one of them held its heading to better than five degrees at the ninety-five per cent confidence level. Fifteen of the twenty-eight held better than one degree.

These were not brief straight bits inside a wander. The constant-course segments made up between 14 and 95 per cent of an individual whale’s total track distance, and more than half of the distance travelled in nine of the sixteen tracks. Holding a line is not something these whales occasionally do. It is most of what the crossing is.

What one degree buys, and this arithmetic is ours. A course held to one degree over the longest segment — 2,230 km — puts you about 39 kilometres to one side of where you aimed. Over the same distance a ten-degree error puts you nearly 400 km out, which in the Southern Ocean is the difference between a feeding ground and open water. Horton and colleagues do not print this conversion; we did it from their numbers so the degree means something.

A straight line on a chart is not the same as a straight line in water. Water moves.

Horton and colleagues took sea-surface current directions and speeds from Argo drifter buoys, at the time and place of each whale, and asked what the currents were doing to them. The answer is the most interesting sentence in the paper after the conclusion: the currents deflected the whales’ headings by as little as under one degree and as much as twenty-five degrees — and did not deflect their tracks.

Read that slowly. The whale is being pushed sideways by up to twenty-five degrees and arrives on the line anyway. It is not holding a heading. It is holding a track, and adjusting the heading continuously to keep it. That is a harder thing, and it requires an outside reference to check against — you cannot correct for a drift you cannot detect.

The segments also crossed varied sea-floor depths and a range of weather, including a tropical storm off the coast of Brazil in January 2004. Whatever the reference is, cloud did not take it away.

This is the spread that rules out the easy answer. They just swim in a straight line would be a perfectly good explanation for a fish in still water. These are not still-water tracks. A body that arrives on the line through a storm and a current is running a correction loop against something.

Only three of the sixteen tagged whales kept transmitting all the way to the high-latitude feeding grounds. Here is where they finished.

All three ended their migrations within about a hundred kilometres of 58°S, 23°W — a point in the South Atlantic, south-west of nothing in particular — despite migrating in different months, of different years, along distinctly different routes.

Three is a small number and the paper does not pretend otherwise; it is three because the tags stopped. But it is the observation that separates two very different claims. Holding a straight line is a compass problem. Arriving at a specific patch of ocean from three different directions in three different seasons is a map problem — you have to know where you are, not only which way you are pointing.

Horton and colleagues say their findings are compatible with goal orientation using a map-and-compass system. What they will not say is which map.

Our own shelf has the compass half of this three times over. A bird reads the field through something in its eye that is still being worked out; a turtle imprints on the magnetic signature of one beach; a salmon does something similar to find one stream. Each of those has a candidate mechanism, hedged. This one has a measurement and no mechanism, which is why it is here and not in with them.

The oldest answer in animal navigation is a sun compass. Horton and colleagues calculated, for every whale position, the azimuth of the Sun at sunrise and sunset and the altitude of its transit, and checked.

Across and between individual segments the Sun’s azimuth differed by roughly one to ten degrees, and its altitude by roughly three to twenty-six. That is plenty of variation to test against. Two things fell out, and between them they close the door:

Whales from the same area followed similar headings under different Sun positions. And whales followed different headings under similar Sun positions.

The authors leave a door open, and it is worth keeping open honestly: if the Sun was being used, its position must have been read against a moving reference — some internal sense of where you are and what time it is, against which a solar azimuth would mean something different in each place. That is not a compass. That is a compass plus a clock plus a map, and the paper does not claim to have found any of the three.

And the storm from spread four is sitting here too. Sun compasses need a sky. At least one of these constant-course segments crossed a tropical storm without bending.

The other old answer is a magnetic compass, and this collection has already spent it three times, on birds, turtles and salmon. Here it comes apart.

Magnetic inclination varied by as much as twenty-two degrees and as little as one across individual constant-course segments. Declination varied by as much as twelve degrees and as little as half a degree. If a whale were steering by either, the headings should track them. They do not.

The cleanest case is a pair of tagging sites. Whales leaving Rarotonga departed on westerly to north-westerly headings. Whales leaving New Caledonia departed on southerly to south-easterly headings — across a difference in magnetic declination of less than one degree. Near-identical field, opposite directions.

What the authors are careful not to say is that magnetism is irrelevant. Their sentence is that whales did not navigate using only a simple magnetic inclination compass, and that spatial information derived from the field could still be informing the movements in ways a scatter plot of headings would not reveal. That is a real distinction and we are keeping it: what failed here is the simplest version, not the whole family.

Why this is not a fourth magnetoreception zine. The bird, the turtle and the salmon each have a candidate mechanism doing real work. The humpback has a precision nobody has matched to a sense. That gap is the subject.

In 2025 Debashis Chatterjee and Prithwish Ghosh came back to the problem with the right tools, and the right tools are stranger than they sound.

Direction needs its own arithmetic. Ordinary statistics cannot handle it. The average of 350° and 10° is not 180°; it is 0°, because a circle has no ends. Angles have to be modelled on the circle they live on, which is what a von Mises–Fisher distribution is for, and what circular–circular regression does when both the thing you are predicting and the thing you are predicting it from are angles. The circular–circular regression they use is their own; they present it as novel, and we are repeating that as their claim rather than as a survey of the literature we have not done.

Applied to blue and humpback whales off the west coast of North America, it found a statistically significant non-zero association between the whales’ directional changes and the environmental cues — and one correlate standing above the rest. Not the Sun. Not the magnetic field. The correlation between solar position and geomagnetic change.

Which is, precisely, an answer to Horton’s objection. Neither cue works in isolation; the interaction of the two carries the strongest signal. That is fourteen years of progress and it is real.

And then the authors write this, in their own discussion, about their own result:

They go further: some parameters, while statistically associated with the direction a whale turns, may not individually exert a biologically meaningful influence. A significant p-value against a cue is not a sense organ. They say so themselves, and it is the best thing in the paper.

Before the regression, Chatterjee and Ghosh had to answer a smaller question: how many different directional behaviours are in this data at all?

A mixture model asks exactly that. You fit one distribution, then two, then more, and let an information criterion tell you where adding another stops paying for itself. For blue whales the answer came out at nine. For humpbacks, in the paper’s own discussion, at twenty-nine.

Twenty-nine is not noise around an average heading. It is the finding that there is no average heading — that describing how a humpback changes direction takes twenty-nine distinct answers, and the model got worse when it was made to pick fewer.

This is the shape our field guides keep arriving at from twelve different directions: there is no standard star, no standard nervous system, no standard shark, no standard nest, no standard way to rest. Here it arrives as a model-selection result. Somebody asked a dataset how many kinds of behaviour were in it, and the dataset said twenty-nine, and stop trying to average us.

An inconsistency we are not going to smooth over. The paper’s abstract says nine mixtures for both species. Its discussion says nine for blue and twenty-nine for humpback, twice, in two separate passages. We cannot reconcile these from the published text, so we are printing both and telling you the discrepancy is the paper’s, not ours. The argument on this spread survives either number: nine is already not one.

Everything so far has been about a line: how straight it is, what holds it, and how little we can say about that. Now change what the line is made of.

The whale on that track is not empty. It is a fasting body burning through a summer’s worth of stored high-latitude food, and everything it burns has to come out somewhere. Urine. Sloughing skin. The faeces of a nursing calf. Placentas. Carcasses. All of it released thousands of kilometres from where it was eaten.

Helen Edgar reached this exact hinge in Always Migrating, on the spread where the salmon carries the ocean into the forest: “Neither salmon nor whale sets out to feed a forest or an ocean. But each unique way of moving through the world is, at the same time, an act of shaping it — and of being reshaped by it in turn.” She reads it through Barad’s intra-action — the idea that a migrating body is not a separate thing that then has effects, but something that comes into being through the relating itself.

This zine is the companion that goes and weighs it.

What she covered and what is left. Her spread names the whale pump: nutrients moved upward, from depth back into sunlit water, within a feeding ground. That is a different flux from the one on the next five spreads. This one runs sideways — thousands of kilometres, from a rich ocean to a poor one — and it was not measured at global scale until 2025.

Joe Roman and eight colleagues put a number on it in 2025, for four relatively well-studied species that migrate in the traditional pattern: gray, humpback, and North Atlantic, North Pacific and southern right whales.

They call it the great whale conveyor belt, and their assessment of it is flat and remarkable: to their knowledge, baleen whales provide the largest long-distance nutrient subsidy on the planet — the highest quantities of carbon and nitrogen ever recorded moved by animals across thousands of kilometres.

The nitrogen matters more than the tonnage. It goes from high-latitude water that has plenty to tropical water that is oligotrophic — nutrient-poor by definition, which is why it is so clear and so blue. Whale nitrogen arriving there is allochthonous: it did not come from that system. It came from an ocean thousands of kilometres away, in a body.

Read the middle row as the honest one. Not every whale in a population migrates every year, and the urea output of males and non-breeding females is poorly known — the authors say so. But mothers with calves are obligated to make the crossing. So the mothers-and-calves figure is a floor nobody has to argue about, and it is still two-thirds of the nitrogen.

Tonnage alone would not do much spread thinly over an ocean. The reason this flux registers at all is geometry.

Roman and colleagues mapped feeding ranges against calving ranges and took the ratio. Whales feed across enormous areas over a whole summer, moving between patches. Then they aggregate — into bays, lagoons, a strip of coast — to give birth, because a small sheltered place is safer from killer whales and cheaper to hold a calf in.

A North Pacific gray whale gathers over an area up to forty-six times larger than the one it unloads into. That concentration is the mechanism. The same nitrogen released evenly across the Bering and Chukchi Seas would be a rounding error; released into a few bays along Baja California, it is a season’s fertiliser.

Nothing about the whale changed between those two states. What changed is how much room the nutrient had to be in. A quantity that is negligible spread out and decisive concentrated is not a different quantity. It is the same one, in a different room.

A caution the authors print themselves. These polygons come partly from opportunistic sightings and satellite-tagged individuals, supplemented from the published literature. The ratios shift depending on how you define a calving ground, and the paper says refining these areas is outstanding work. Take 46:1 as the shape of the thing, not a constant.

Global estimates are estimates. There is one place where both the whale demography and the abiotic nutrient budget have been studied well enough to put side by side, and the comparison is the strongest thing in the paper.

In the Hawaiian Islands Humpback Whale National Marine Sanctuary, the current humpback population releases about 3,142 kg of nitrogen a day through urea, or about 3,715 kg a day counting detrital nitrogen. Against that, the Hawaii Ocean Time-series gives a vertical nitrogen flux from upwelling and convective mixing of about 2,419 kg a day in April.

During the breeding season, in that sanctuary, the whales out-deliver the ocean’s own physics. Not by a rounding margin — by somewhere between a quarter and three quarters again as much. The animals are a comparable term to upwelling in a nutrient budget.

And here is the part that does not read like ecology. Most of the biomass is carcasses. Placentas. Dead neonates. Adult whales that did not finish the crossing. The transport runs on birth and on death — on the two events the migration exists to make possible and the one it cannot prevent.

Tiger sharks and other large sharks turn up to feed on the carcasses. In Australia and Brazil, sharks pup nearshore during the whales’ breeding season. Somebody else’s calving ground has become the reason this is a good place to have your own.

Scale check, from the authors. The quantity of nitrogen is comparable to what subtropical seabirds move — a well-studied process. What is not comparable is the distance. Seabirds move nutrients tens or hundreds of kilometres, mostly within high latitudes. Whales move it thousands, and from rich water to poor.

Before whaling, this flux was roughly three times larger — 7,822 tons of nitrogen a year against 3,784. What industrial whaling removed was not only a great many animals. It was a supply line, and it was removed before anyone knew it was there.

That fact is exactly where this piece could go wrong, so:

Why the first refusal is first. An argument from usefulness holds only as long as the usefulness does. Build the case for a life on what it delivers, and you have agreed in advance to the terms on which it can be taken away — which is the same trade offered to disabled people every time our worth is priced in contribution. We know how that argument ends, because we have been on the receiving end of it.