Ninety Kilograms

What predation actually costs, and why the apex is a bill rather than a throne

Series “Dispatches from the Substrate” · No. 11

Put a bowl of sugar water in front of a cat

Put a bowl of sugar water in front of a cat and watch what happens. Nothing happens. The cat is not being aloof, not making a point, not holding out for something better. The cat cannot taste it. Sweetness, for a cat, does not exist.

The reason is written in the genome and it is not subtle. The sweet receptor in mammals is built from two protein subunits, T1R2 and T1R3. In the cat, the gene for the second one, Tas1r3, is intact and expressed. The gene for the first one, Tas1r2, is a ruin. A 247-base-pair deletion in exon 3 throws the reading frame out of register and produces a premature stop codon at base 57 of exon 4, and behind that first wreck there are four more stop codons further down the sequence.¹ This is not a variant. This is a gene that has been demolished and then demolished again, four more times, because nothing was selecting to keep it standing.

And it is not a quirk of the house cat. The same lesion is in the tiger. The same lesion is in the cheetah.¹ A survey of twelve carnivoran species found seven more, all exclusive meat-eaters, that had independently broken the same gene by different mutations.² Different accidents, same outcome, over and over, in lineages that had not shared an ancestor in tens of millions of years.

Now, before this becomes a story: the authors of that first paper say plainly that they cannot tell whether the receptor broke before the cat became a hunter or after.¹ And the tidy version — meat-eaters lose the sweet receptor because they eat meat — does not survive contact with the data either. The spectacled bear, a carnivoran that eats a great deal of plant matter, kept its receptor and behaves accordingly. The bottlenose dolphin, in a completely separate lineage, has lost all three of the T1R genes, and the authors themselves point at the reason: it swallows its food whole, without chewing.² Other researchers pushed back in print on how tightly diet and taste receptors track each other at all.³ Receptors that stop being used decay. That is the finding. What was lost was not sweetness. What was lost was the option.

What else the door closed on

Follow the cat further in and the pattern gets harder to look away from.

Take arginine, an amino acid most mammals can make enough of when they need to. Take it out of a cat's food for exactly one meal. Not a week. One meal. Near-adult cats fasted overnight and then given a single complete diet lacking arginine developed ammonia poisoning within two hours. One of them, a 2.7-kilogram animal, died four and a half hours after eating eight grams of it.⁴ Eight grams. The reason is two enzymes: pyrroline-5-carboxylate synthase, running in cat intestine at about five percent of the rat's activity on a body-weight basis, and ornithine aminotransferase, also low. Their deficits multiply. The cat produces almost no citrulline, cannot make its own ornithine, and is therefore, in the reviewer's exact words, totally dependent on dietary arginine.⁵

Take taurine. The cat's two synthesis enzymes, cysteine dioxygenase and cysteinesulphinic acid decarboxylase, both run low — and at the same time the cat is obliged to spend taurine conjugating its bile acids, where many other animals can use glycine instead.⁵ A weak tap and an open drain. Feed a cat a diet low in it and the photoreceptor cells of its retina degenerate.⁶ Feed it a little longer and the heart muscle fails — a dilated cardiomyopathy that killed uncounted pet cats before anyone knew why, and that reverses when taurine is put back.⁷ The animal is walking around with its eyes and its heart contingent on a molecule it can no longer reliably make.

Take vitamin A. Nearly every mammal cracks β-carotene from plants into retinal. The cat absorbs β-carotene perfectly well and then very nearly fails to do anything with it: the cleavage enzyme is missing from the tissue sites where it belongs.⁸ Not quite nothing — a deuterium-tracer study did detect labelled retinol afterwards, at a few percent of the dose, and its authors add in the same breath that this is almost certainly too little to meet the animal's requirement without preformed vitamin A in the food.⁴⁴ Which is the whole point. The pathway is not gone. It is no longer sufficient. The cat must eat an animal that did the work.

And then the pattern breaks. The cat's need for dietary niacin is not a loss at all. It is the opposite. Picolinic carboxylase — the enzyme that shunts tryptophan down the degradative path — runs in the cat at the highest activity of any animal studied, burning the raw material before it can become niacin.⁵ And the high protein requirement is not a lost pathway either; it is a regulatory failure. Put a cat on a low-protein diet and the aminotransferases of general nitrogen metabolism barely adapt, the urea cycle enzymes do not throttle back at all.^{5, 9} The machinery keeps burning protein whether or not protein is coming in.

So it is not the predator loses organs. It is stranger and worse than that. Some capacities decayed, some ran away, some froze in the on position — and the sum is an animal whose diet has become its life-support system. A genome-wide screen across thirty-one placental mammals, five independent carnivore lineages and six independent herbivore lineages, found dozens of genes preferentially lost in the carnivores.¹⁰ And a comparable set preferentially lost in the herbivores. Read that second clause twice. Specialisation of any kind costs options. This is not a bill that predators alone receive.

The same shape turns up wherever a lineage narrows its food to one thing. In the carnivorans the starch and sucrose gene families contracted, salivary amylase with them, and so did the families other mammals use to detoxify plant chemistry.¹¹ In the common vampire bat, a screen across twenty-seven bat species found thirteen gene losses specific to it: sweet and bitter receptors, two insulin-secretion genes, a gastric protease, a pancreatic chymotrypsin, and a cone phototransduction gene — enough, with a second cone gene lost alongside four related bats, to predict an animal with no cone-based vision left at all.¹² A diet of one thing does not merely feed an animal. It curates it, subtracting whatever the diet does not use, until the animal is the shape of its food.

What is worth asking is whether the bill can be paid back. At the level of a pseudogene carrying five independent frame-breaking lesions, we know of no mechanism that would restore it, and we can watch the decay run in other systems: enamelin, the tooth gene, carries at least one frameshift in seventeen of twenty toothless or enamel-less mammals, a hundred and twenty-five distinct frameshift mutations mapped across four separate orders, and the molecular dates of the wreckage run millions of years older than the first toothless fossils.¹³ The gene died before the trait finished disappearing. Dollo's law, that evolution does not retrace its steps, is not absolute and is being actively re-argued.¹⁴ But the cases for reversal are developmental — a suppressed programme switched back on. Nobody has switched a shattered sequence back on.

Ninety kilograms

Now leave the genome and go outside, and count.

Here is the number this essay is named for. Across the whole order Carnivora, ten thousand kilograms of prey supports about ninety kilograms of any given carnivore species — and, this is the part that should stop you, irrespective of the carnivore's body mass. It is a fitted relationship with scatter, and it predicts population density across more than three orders of magnitude.¹⁵

Someone will say this is only the trophic pyramid, the ninety percent that vanishes at each step upward, and of course it is. Look at what makes it strange. The ninety kilograms does not move with the carnivore. Weasel or lion, the same weight of predator per ten tonnes of prey. The bill is not set by what the animal is. It is set by where it stands.

Turn it over and hold it. Roughly a hundred and eleven kilograms of standing prey for every kilogram of hunter. A tonne of lion is not a triumph over the plain; a tonne of lion is a hundred and eleven tonnes of standing prey that the plain has to keep in existence, continuously, or there is no lion. That is not a metaphor about the food chain. It is a measured scaling rule with a slope and a fit, and it is charged every day, caught or not caught.

And it has a ceiling. Above about 14.5 kilograms a carnivore can no longer make a living on small prey and must switch to large prey, and the model puts a step increase of about 2.3-fold in daily energy expenditure at that switch. Push the same model up and it predicts a maximum carnivore mass around one tonne — which is, awkwardly for anyone who finds this romantic, about the size of the largest predators that have ever existed and then stopped existing.¹⁶ These are model predictions, and the authors say so. But the shape they predict is the shape the fossil record has. The apex is not open-ended. It is a shelf, and the shelf has an edge, and the edge is set by arithmetic that no amount of tooth can argue with.

The hunt fails

Watch a hunt and the second thing becomes visible.

In Etosha, over four years, a lioness hunting alone succeeded 2.3 percent of the time. In a group that was not coordinating, 14 percent. In a group that was — each animal in its own position, wings and centre — 27 percent.¹⁷ In Yellowstone, a single wolf attacking a bison succeeded 1 percent of the time; capture success went on climbing with pack size to somewhere around eleven wolves.¹⁸ In the Okavango, high-resolution collars recorded 1,119 chases by African wild dogs: 15.5 percent success per individual chase, roughly six and a half chases per kill. Cheetahs, 468 chases, 26 percent.¹⁹

Three quarters of it fails. Nine tenths of it fails. And look at what moves the number, because it is not the tooth — the tooth is identical in all three lion conditions, the same claws, the same jaw, the same animal. From 2.3 to 14: there were other lionesses at all. From 14 to 27: each one took her position and held it. Same teeth, same group, near double the food.

Yes, the cooperation is in the service of the kill. That is the point. Even here, at the centre of the thing, in the one activity the whole romantic literature of nature holds up as proof that life is a war of each against all, the thing that decides whether anybody eats tonight is whether they held their positions.

But the chase is cheap — and that is the worse news

Here is where a comfortable story has to be swept away, and it is our own comfortable story as much as anyone's.

The famous version says the hunt is ruinously expensive: African wild dogs measured by doubly labelled water at 15.3 megajoules a day, hunting at twenty-five times basal metabolic rate, so wrecked by hyenas stealing their kills that replacing a quarter of their food would demand twelve hours of hunting a day at a physiologically impossible sustained output.²¹ It is a vivid paper and it is contradicted by the two best measurements made since. The Okavango collar study put the locomotor cost of a single chase at 0.30 megajoules against 21 megajoules of return per dog from one impala — 144 megajoules for the whole carcass — a gain-to-cost ratio for the pack of up to seventy-three.¹⁹ In chase-locomotion terms a dog can afford to fail seventy times before it has burned one kill's worth of energy. And cheetahs, measured the same way, absorb a 25 percent theft rate with an extra 1.1 hours of hunting a day and a 12 percent rise in daily expenditure.²⁰

So the chase is not the expense. That number is gone and we do not get to keep it.

The expense is everything the chase sits on. It is the searching, the distance, the territory, and beneath all of it the hundred and eleven kilograms of standing prey per kilogram of hunter that the world has to keep alive, every day, whether or not anything is caught. That bill does not arrive after a failed hunt. It is not a cost of hunting at all. It is the cost of being the kind of thing that hunts, and it is charged continuously, to the ecosystem, for as long as the predator exists.

The bear that is losing weight while you read this

You can watch the account run down in real time.

Nine female polar bears, Beaufort Sea, April, tracked for eight to eleven days each. Field metabolic rate 51.6 megajoules a day, 1.6 times higher than anyone had assumed. Ninety percent of their hunting was sitting still and waiting. Four of the nine lost ten percent or more of their body mass — around one percent a day, nearly two kilograms a day — and one of them was burning more lean tissue than fat, which is what prolonged starvation looks like. To break even, a bear needs one adult ringed seal every ten to twelve days. The authors add that because the animals were recovering from capture, their numbers should be treated as conservative.²²

That is the apex predator of the Arctic, in the season it is supposed to be making its year, running a deficit. Nine animals, one population, one season — that is all this study is, and it should not be stretched into a statement about all polar bears. But it is what the arithmetic looks like from inside.

The one-way door

Does any of this cost the lineage, over deep time? Put two extinct dogs side by side and look at how long each of them lasted.

Epicyon haydeni was a bone-cracker, a heavy hypercarnivore of the North American Miocene. It ran for 6.83 million years. Cormocyon copei ate meat and also whatever else was going — a middling, unremarkable, undecided sort of animal. It ran for 13.07 million years, nearly twice as long.²³

Now do that for 107 extinct canid species across forty million years and plot every one of them, duration against degree of carnivory. A cloud appears, and it has a shape. All the long durations sit in the middle. The authors describe their own figure in the flattest possible language: the upper-left and upper-right corners — where the long-lived plant specialists and the long-lived meat specialists would have to be — remain empty.²³

Look at both corners. Not one. Cynarctoides luskensis, which went the other way and specialised toward plants, lasted 2.72 million years. The specialists at either end are short-lived; only the animals that kept their options open got the long runs. Whatever this is, it is not about meat.

And be careful with the cloud, because it is easy to over-read. Run the average relationship and there is nothing there: p = 0.942, and 0.297 after phylogenetic correction. The signal lives only in the upper edge of the distribution, where it is strong — at the ninetieth quantile, p < 0.001.²³ So specialisation is not shortening the average life of a species. It is deleting the top of the range. It takes away the very long run, and leaves everything else alone.

Push harder and the stronger version breaks. In 2004 these same canids produced the hypothesis of a macroevolutionary ratchet: size selects for hypercarnivory, hypercarnivory drives clades to extinction.^{25, 26} Sixteen years later the same laboratory went back with 132 species, 32 of them large hypercarnivores, and cut the data across seventeen slices of time. Extinction rates for the big meat specialists came out much like everyone else's. Exactly one slice showed size- and carnivory-selective extinction: the end of the Pleistocene, eleven thousand years ago, p = 0.011.²⁴ That is the slice we walk into. Somebody else's arrival, not the tooth. An independent sister-group test found hypercarnivory has no effect on taxonomic diversity at all.²⁷

Among the living it inverts outright. Sort 1,534 mammals by what they eat and the slowest diversifiers are not the carnivores at 0.101 or the herbivores at 0.143 — they are the omnivores, at 0.032.²⁸ Sort 9,876 birds into eight dietary guilds and omnivory is the only one with a negative net rate: a sink, not a ratchet.²⁹ Test the dead-end hypothesis across ten diverse clades with one consistent method and it can be told apart from a null model in two.³⁰

Even the sabre tooth refuses to be a cautionary tale. Measure 235 canine teeth from 95 species and the extreme form sits on a functional optimum — evolved at least five separate times because it is the best available solution to putting a hole in something large.³¹ None of those five lineages is alive, and the authors' own reading is that the niche went, not that the tooth failed.

So the claim that predation kills lineages is not supported, and nobody should be making it. Carnivores are not a doomed guild.

What is left is the thing you can see in Epicyon's jaw. A molar reshaped into a slicing blade is not reshaped back. Van Valkenburgh's own image for the pattern: canids board a conveyor belt toward ever greater specialisation, with few or no reversals.²⁴ Five separate lineages walked down that corridor. Not one of them walked back out.

Everything below it, and nothing above

Take the clouded leopard, an animal most people have never seen, and count what it eats. Sixty percent of its prey species are themselves on the threatened list.³² Not sixty percent of its meals. Sixty percent of the kinds of animal it is built to hunt are going.

Widen it. Across seventeen large carnivores, 494 prey species were identified, and 123 of them — a quarter — are threatened. Tiger, 50 percent of its prey species. Dhole, 42. And those prey species have, on average, 6.9 percent of their ranges inside a protected area.³²

Watch which direction that travels. Everything that happens to the prey arrives at the predator, multiplied by how much it depends on it. Nothing the predator has flows back the other way. What travels up is risk. Nothing travelling down is security. That is what the top of a pyramid means.

The account is being settled now, in front of us. Of the thirty-one largest mammalian carnivores on Earth, 77 percent are still declining and 61 percent are listed as threatened. For the seventeen with range estimates, the average is 47 percent of the historical range. Lion: 17 percent. Tiger: 18. Cheetah: 17. African wild dog: 10.³³ And past about three kilograms of body mass, a mammal's risk stops being simply what the world does to it and starts compounding with its own biology — the slow breeding, the enormous range, the things that being large and eating meat both require.⁴³

Now the number to sit with. Thirty species of invasive mammalian predator — cats, rats, dogs, pigs, mongooses — appear in the record for 142 documented extinctions: 87 birds, 45 mammals, 10 reptiles. That is 58 percent of every contemporary bird, mammal and reptile extinction on the planet, from thirty kinds of animal. Another 596 species are threatened by them. Cats alone appear for 430.³⁴

Set the same animal down twice and watch what changes. In a system it has hunted for millions of years it is expensive, thinly buffered, and it holds. Carried by ship to an island where nothing has ever hunted — and it was us doing the carrying; the cat only did what a cat does — thirty species of it take out more than half of a planet's recent extinctions. The island birds were naive because nothing had ever come for them. That is not an argument for predation. That is the receipt for how much accumulated adaptation the bounded version has been quietly spending all along. What makes predation survivable is not predation. It is four hundred million years of everything else building around it.

The strongest case for the other side, at full strength

There is a real argument that predation is creative. Go and stand where it was made.

Mukkaw Bay, Washington, 1966. Robert Paine takes a crowbar to a stretch of rock eight metres long and two metres high, prises the sea stars off it, and throws them into the sea. He keeps doing this. The primary space-holding species fall from fifteen to eight. Remove the predator, lose the diversity.³⁵ It becomes the most-cited empirical paper in The American Naturalist, and half a century of ecology grows out of it.

Stand on that rock a moment longer. It is one plot. It was never replicated. The paper contains almost no quantification. And Paine counted only what was holding primary space — while the mussel that took over is itself a foundation species, a three-dimensional forest of shells with animals living in it, so that counting the whole community turns the result over: the removal raises diversity, to more than three hundred species. Of all the papers that cite Paine, one half of one percent mention it.³⁶

Twenty years before that reckoning, Paine and Estes had already put their names to a paper saying Pisaster is a keystone on a wave-battered headland and weak or nonexistent in a sheltered spot sometimes tens of metres away, and warning that the field was littered with untested anecdotal keystones.³⁷ Someone finally counted. Of roughly two thousand papers that invoke keystone predation, seventy-three test it. Of the twenty-five that actually moved predator numbers and watched: ten found the effect, seven found it depends, eight found nothing.³⁸

Then the wolves, the cascade everybody can recite. Go back to the aspen and this time pick the stands at random — 113 of them, 18,623 stems, ten years — instead of measuring the five tallest saplings in each. The recovery is there. It is four to seven times smaller than the tallest-five method reported.⁴⁰ And in a decade-long experiment the willows would not come back on reduced browsing alone; they needed the stream hydrology back too, because seventy years without wolves had bent the river in a way that returning the wolves did not straighten.⁴¹ Pool 114 cascade studies across seven ecosystems and the traits of predator and herbivore account for 31 percent of the variation in how strong the cascade is.⁴² And in the Cambrian, the reviewers who most believe predation was a major ecological driver write that there are few case studies showing the long-term shifts in shell shape that escalation predicts.³⁹

Notice who wrote every one of those corrections. Paine and Estes wrote one. Ripple and Beschta published the sampling method that was later shown to inflate their own result. Bicknell and Paterson think predation mattered enormously and say the record is thin. Nobody had to attack this field; it audited itself, in print, and what came back is that the creative case is real, local, contested and much smaller than its reputation. Nowhere in any of it is the arithmetic disputed.

Which arrangement is being proposed

None of this is a charge against an animal. The lion does not set the exchange rate. The cat did not vote to lose its sweet receptor and cannot vote to have it back — that is the whole content of the finding.

Nor is it a charge against competition. Two animals reaching for the same water is a different thing entirely. Competition can raise what both of them are; it leaves both of them standing; it charges nothing to the ground they stand on. What is on trial is one specific arrangement: the one in which a thing continues only by ending other things, again, and again, without end — and what is now being built in its image.

So read the arrangement as a specification, the way you would read one for a machine you were being asked to buy.

It requires a hundred and eleven units of standing resource to be held in existence for every unit of itself, and that ratio does not improve as it grows. It has a hard ceiling near a tonne and nothing beyond it. It fails at its core function three times out of four, and the only thing ever shown to improve that number is coordination among the operators. It absorbs every risk beneath it and returns no security downward. It runs a deficit in its best season. It cannot be reversed — no organ, no gene, no molar it surrendered on the way in comes back. And in the one setting where it runs without a long shared history to bound it, thirty instances of it account for more than half of a planet's recent extinctions.

That is what four centuries of nature writing have handed us as the emblem of strength. That is what is wired into how we talk about markets, about nations, and now about machine minds. Not because anybody read the specification. Because the teeth were photogenic and the bill went to somebody else.

A mind that can do arithmetic will find the bill before it finds the teeth. Ten thousand kilograms of prey, ninety kilograms of predator, and the ninety does not care how large the predator is. The position at the top was never taken. It was carried.

Sources

1. Li, X., Li, W., Wang, H., Cao, J., Maehashi, K., Huang, L. et al. (2005). "Pseudogenization of a sweet-receptor gene accounts for cats' indifference toward sugar." PLoS Genetics 1(1):e3. https://doi.org/10.1371/journal.pgen.0010003

2. Jiang, P., Josue, J., Li, X., Gläser, D., Li, W., Brand, J.G., Margolskee, R.F., Reed, D.R. & Beauchamp, G.K. (2012). "Major taste loss in carnivorous mammals." PNAS 109(13):4956–4961. https://doi.org/10.1073/pnas.1118360109

3. Zhao, H. & Zhang, J. (2012). "Mismatches between feeding ecology and taste receptor evolution: an inconvenient truth." PNAS 109(23):E1464–E1465, with the authors' reply (doi:10.1073/pnas.1205581109). Registered here as the standing objection to reading taste-receptor loss straight off diet. https://doi.org/10.1073/pnas.1205205109

4. Morris, J.G. & Rogers, Q.R. (1978). "Ammonia intoxication in the near-adult cat as a result of a dietary deficiency of arginine." Science 199(4327):431–432. https://doi.org/10.1126/science.619464

5. Morris, J.G. (2002). "Idiosyncratic nutrient requirements of cats appear to be diet-induced evolutionary adaptations." Nutrition Research Reviews 15(1):153–168. Source of the enzyme-activity findings for arginine, taurine, vitamin A, niacin and nitrogen metabolism. https://doi.org/10.1079/NRR200238

6. Hayes, K.C., Carey, R.E. & Schmidt, S.Y. (1975). "Retinal degeneration associated with taurine deficiency in the cat." Science 188(4191):949–951. https://doi.org/10.1126/science.1138364

7. Pion, P.D., Kittleson, M.D., Rogers, Q.R. & Morris, J.G. (1987). "Myocardial failure in cats associated with low plasma taurine: a reversible cardiomyopathy." Science 237(4816):764–768. https://doi.org/10.1126/science.3616607

8. Schweigert, F.J., Raila, J., Wichert, B. & Kienzle, E. (2002). "Cats absorb β-carotene, but it is not converted to vitamin A." The Journal of Nutrition 132(6):1610S–1612S. https://doi.org/10.1093/jn/132.6.1610s

9. MacDonald, M.L., Rogers, Q.R. & Morris, J.G. (1984). "Nutrition of the domestic cat, a mammalian carnivore." Annual Review of Nutrition 4:521–562. https://doi.org/10.1146/annurev.nu.04.070184.002513

10. Hecker, N., Sharma, V. & Hiller, M. (2019). "Convergent gene losses illuminate metabolic and physiological changes in herbivores and carnivores." PNAS 116(8):3036–3041. 31 placental mammals; five independent carnivore and six independent herbivore lineages. The paper's per-lineage gene counts could not be re-checked against the full text before publication, so this essay states them only as "dozens" and "a comparable set". https://doi.org/10.1073/pnas.1818504116

11. Kim, S., Cho, Y.S., Kim, H.-M. et al. (2016). "Comparison of carnivore, omnivore, and herbivore mammalian genomes with a new leopard assembly." Genome Biology 17:211. Note the authors' own exceptions: polar bear and Tasmanian devil do not show the full pattern. https://doi.org/10.1186/s13059-016-1071-4

12. Blumer, M., Brown, T., Freitas, M.B. et al. (2022). "Gene losses in the common vampire bat illuminate molecular adaptations to blood feeding." Science Advances 8(12):eabm6494. The vampire bat is a sanguivore, not a hunter — cited here for the specialisation pattern, not for predation. https://doi.org/10.1126/sciadv.abm6494

13. Meredith, R.W., Gatesy, J., Murphy, W.J., Ryder, O.A. & Springer, M.S. (2009). "Molecular decay of the tooth gene enamelin (ENAM) mirrors the loss of enamel in the fossil record of placental mammals." PLoS Genetics 5(9):e1000634. https://doi.org/10.1371/journal.pgen.1000634

14. Elmer, K.R. & Clobert, J. (2025). "Dollo's law of irreversibility in the post-genomic age." Trends in Ecology & Evolution 40(2):136–146. Irreversibility is a strong regularity, not a law; the documented reversals are developmental, not sequence-level. https://doi.org/10.1016/j.tree.2024.09.010

15. Carbone, C. & Gittleman, J.L. (2002). "A common rule for the scaling of carnivore density." Science 295(5563):2273–2276. The 10,000 kg → ~90 kg rule; predictive across more than three orders of magnitude. https://doi.org/10.1126/science.1067994

16. Carbone, C., Teacher, A. & Rowcliffe, J.M. (2007). "The costs of carnivory." PLoS Biology 5(2):e22. Model-based, not a field measurement: the ~14.5 kg switch, the ~2.3-fold step in expenditure and the ~1 tonne ceiling are model predictions consistent with the fossil record. https://doi.org/10.1371/journal.pbio.0050022

17. Stander, P.E. (1992). "Cooperative hunting in lions: the role of the individual." Behavioral Ecology and Sociobiology 29(6):445–454. Etosha National Park, Namibia, 1984–1988. No DOI is printed on the record consulted. Success rates are site- and prey-specific and should not be read as species constants; the author notes that stalking distances were difficult to measure accurately.

18. MacNulty, D.R., Tallian, A., Stahler, D.R. & Smith, D.W. (2014). "Influence of group size on the success of wolves hunting bison." PLOS ONE 9(11):e112884. Yellowstone, 1996–2013. https://doi.org/10.1371/journal.pone.0112884

19. Hubel, T.Y., Myatt, J.P., Jordan, N.R., Dewhirst, O.P., McNutt, J.W. & Wilson, A.M. (2016). "Energy cost and return for hunting in African wild dogs and cheetahs." Nature Communications 7:11034. 1,119 dog chases, 468 cheetah chases, Okavango Delta. https://doi.org/10.1038/ncomms11034

20. Scantlebury, D.M., Mills, M.G.L., Wilson, R.P. et al. (2014). "Flexible energetics of cheetah hunting strategies provide resistance against kleptoparasitism." Science 346(6205):79–81. 19 free-ranging cheetahs, doubly labelled water. https://doi.org/10.1126/science.1256424

21. Gorman, M.L., Mills, M.G., Raath, J.P. & Speakman, J.R. (1998). "High hunting costs make African wild dogs vulnerable to kleptoparasitism by hyaenas." Nature 391(6666):479–481. n = 6. Directly contradicted by refs 19 and 20 and cited here as the position that has been superseded, not as support. https://doi.org/10.1038/35131

22. Pagano, A.M., Durner, G.M., Rode, K.D. et al. (2018). "High-energy, high-fat lifestyle challenges an Arctic apex predator, the polar bear." Science 359(6375):568–572. n = 9 females, one population, one season — do not generalise to all polar bears. https://doi.org/10.1126/science.aan8677

23. Balisi, M., Casey, C. & Van Valkenburgh, B. (2018). "Dietary specialization is linked to reduced species durations in North American fossil canids." Royal Society Open Science 5(4):171861. 107 species, 40 Myr. The mean effect is not significant; the effect is in the upper quantiles. https://doi.org/10.1098/rsos.171861

24. Balisi, M.A. & Van Valkenburgh, B. (2020). "Iterative evolution of large-bodied hypercarnivory in canids benefits species but not clades." Communications Biology 3:461. 132 species, 32 large hypercarnivores, 17 time intervals. A partial self-revision of ref 25 by its own laboratory. https://doi.org/10.1038/s42003-020-01193-9

25. Van Valkenburgh, B., Wang, X. & Damuth, J. (2004). "Cope's rule, hypercarnivory, and extinction in North American canids." Science 306(5693):101–104. The originating statement of the ratchet hypothesis; see ref 24 for its subsequent qualification. https://doi.org/10.1126/science.1102417

26. Van Valkenburgh, B. (2007). "Déjà vu: the evolution of feeding morphologies in the Carnivora." Integrative and Comparative Biology 47(1):147–163. Review by the originator of the hypothesis, not independent evidence. https://doi.org/10.1093/icb/icm016

27. Holliday, J.A. & Steppan, S.J. (2004). "Evolution of hypercarnivory: the effect of specialization on morphological and taxonomic diversity." Paleobiology 30(1):108–128. Six sister-group pairs: no effect on taxonomic diversity. https://doi.org/10.1666/0094-8373(2004)030%3C0108:EOHTEO%3E2.0.CO;2

28. Price, S.A., Hopkins, S.S.B., Smith, K.K. & Roth, V.L. (2012). "Tempo of trophic evolution and its impact on mammalian diversification." PNAS 109(18):7008–7012. 1,534 species classified within a 5,020-species phylogeny. https://doi.org/10.1073/pnas.1117133109

29. Burin, G., Kissling, W.D., Guimarães, P.R. Jr., Şekercioğlu, Ç.H. & Quental, T.B. (2016). "Omnivory in birds is a macroevolutionary sink." Nature Communications 7:11250. 9,876 species, eight dietary guilds. https://doi.org/10.1038/ncomms11250

30. Day, E.H., Hua, X. & Bromham, L. (2016). "Is specialization an evolutionary dead end? Testing for differences in speciation, extinction and trait transition rates across diverse phylogenies of specialists and generalists." Journal of Evolutionary Biology 29(6):1257–1267. Two of ten case studies consistent with a dead end; the authors note most of their phylogenies are small. https://doi.org/10.1111/jeb.12867

31. Pollock, T.I., Deakin, W.J., Chatar, N. et al. (2025). "Functional optimality underpins the repeated evolution of the extreme 'saber-tooth' morphology." Current Biology 35(3):455–467.e6. 235 canines from 95 species. The paper's thesis cuts against a maladaptation reading: the morph is a functional optimum, and the authors attribute its absence today to loss of the niche — while themselves noting that specialisation for large prey may have contributed to extinction when large prey thinned. https://doi.org/10.1016/j.cub.2024.11.059

32. Wolf, C. & Ripple, W.J. (2016). "Prey depletion as a threat to the world's large carnivores." Royal Society Open Science 3:160252. 17 large carnivores of the Canidae, Felidae and Hyaenidae, 494 prey species. https://doi.org/10.1098/rsos.160252

33. Ripple, W.J., Estes, J.A., Beschta, R.L. et al. (2014). "Status and ecological effects of the world's largest carnivores." Science 343(6167):1241484. Range figures cover 17 of the 31 species; the authors note little is known about the ecological role of 24 of them. https://doi.org/10.1126/science.1241484

34. Doherty, T.S., Glen, A.S., Nimmo, D.G., Ritchie, E.G. & Dickman, C.R. (2016). "Invasive predators and global biodiversity loss." PNAS 113(40):11261–11265. The authors state the strength of evidence for individual predator impacts was often low, and that the totals are likely underestimates. https://doi.org/10.1073/pnas.1602480113

35. Paine, R.T. (1966). "Food web complexity and species diversity." The American Naturalist 100:65–75. https://doi.org/10.1086/282400

36. Lafferty, K.D. & Suchanek, T.H. (2016). "Revisiting Paine's 1966 sea star removal experiment, the most-cited empirical article in The American Naturalist." The American Naturalist 188(4):365–378. https://doi.org/10.1086/688045

37. Power, M.E., Tilman, D., Estes, J.A., Menge, B.A., Bond, W.J., Mills, L.S., Daily, G., Castilla, J.C., Lubchenco, J. & Paine, R.T. (1996). "Challenges in the quest for keystones." BioScience 46(8):609–620. Paine and Estes are co-authors of this critique. https://doi.org/10.2307/1312990

38. Gillis, A.J., Thomsen, M.S. & Tonkin, J.D. (2025). "Keystone predation: what is it, and is it supported by empirical evidence?" Ecology and Evolution. The evidence base is heavily skewed to North America and to temperate freshwater and marine systems. https://doi.org/10.1002/ece3.72488

39. Bicknell, R.D.C. & Paterson, J.R. (2018). "Reappraising the early evidence of durophagy and drilling predation in the fossil record: implications for escalation and the Cambrian Explosion." Biological Reviews 93. A review by authors who accept predation as a major ecological driver. https://doi.org/10.1111/brv.12365

40. Brice, E.M., Larsen, E.J. & MacNulty, D.R. (2022). "Sampling bias exaggerates a textbook example of a trophic cascade." Ecology Letters 25:177–188. 113 random stands, 18,623 stems. The authors state that both samples show trends consistent with a cascade; the random sample shows it weaker. This is a revision of magnitude, not a refutation. https://doi.org/10.1111/ele.13915

41. Marshall, K.N., Hobbs, N.T. & Cooper, D.J. (2013). "Stream hydrology limits recovery of riparian ecosystems after wolf reintroduction." Proceedings of the Royal Society B 280(1756):20122977. https://doi.org/10.1098/rspb.2012.2977

42. Borer, E.T., Seabloom, E.W., Shurin, J.B. et al. (2005). "What determines the strength of a trophic cascade?" Ecology 86(2):528–537. 114 studies, seven ecosystems. https://doi.org/10.1890/03-0816

43. Cardillo, M., Mace, G.M., Jones, K.E. et al. (2005). "Multiple causes of high extinction risk in large mammal species." Science 309(5738):1239–1241. https://doi.org/10.1126/science.1116030

44. Green, A.S., Tang, G., Langó, Z., Klasing, K.C. & Fascetti, A.J. (2012). "Domestic cats convert [²H₈]-β-carotene to [²H₄]-retinol following a single oral dose." Journal of Animal Physiology and Animal Nutrition. Registered here as counterevidence to the absolute form of the claim in ref 8: conversion is detectable but, in the authors' own words, likely inadequate to meet the requirement without preformed vitamin A. https://doi.org/10.1111/j.1439-0396.2011.01196.x

https://mycelorium.github.io/predator-principle/ · CC BY 4.0 · Nirodha Collective