Island size rule

Why do large mammals stranded on an island shrink over time while small ones grow?

Sicily and Malta once held elephants about a metre at the shoulder, descended from a mainland species several times that height. Cyprus had hippos the size of a large pig. On Flores there was a dwarf elephant relative — and, in the same deposits, rats the size of rabbits and a stork taller than a man.

Both directions, in one place. If islands were simply harsh, everything should shrink; if they were simply permissive, everything should grow. Something is acting on size relative to what the animal already was. What could sort a lineage's fate by its starting mass?

Take the large-bodied case first, because it is the more forced. What does a big animal need that a small one does not? Not just more food, but more range — a large herbivore needs a wide area per individual, and a population needs enough individuals not to die of bad luck. On the mainland, area is effectively unbounded. On an island it is a hard number.

So an island imposes an arithmetic: total forage divided by the appetite of one animal gives the number of animals, and below some number the population simply fails. A lineage arriving at elephant size on a small island faces a choice it does not make consciously — either the smaller-bodied individuals leave more descendants, or there are no descendants at all.

Now the small-bodied case, which needs a different question: why was the mainland mouse small? Not because small bodies are efficient in the abstract — being small is largely a way of not being eaten, and it is paid for with a punishing metabolic rate, constant feeding, and short life. Islands, especially remote ones, routinely lack mammalian predators altogether, because a viable predator population needs even more area than its prey. Remove the predator and you have removed the thing that was holding the animal down.

Do you see how the two cases share one shape? In both, island conditions withdraw a pressure or impose one, and the animal moves toward a size it was previously kept away from. The large animal is pushed down by area and food; the small animal is released upward now that hiding is no longer the priority. There is a second release for the small animal: with few competitors present, niches that on the mainland belong to bigger-bodied species stand vacant, and a rodent can grow into one.

That is Foster's 1964 observation and Van Valen's 1973 naming of it as the island rule — and the two mechanisms together predict something neither would alone. If large forms shrink and small forms grow, both are moving toward the middle, so the rule is not "islands shrink things" but "islands compress the range of sizes toward some intermediate value". Where exactly that intermediate sits, and whether it corresponds to an energetically optimal mammalian body size, is estimated differently by different analyses.

The rule is a tendency, not a law, and this is the part most worth being careful about.

The strength of the effect varies with the island itself. A small, remote island with no predators shows it most sharply; a large island, or one close enough to the mainland that colonists keep arriving and interbreeding, shows it weakly or not at all, because gene flow keeps resetting the population toward the mainland form. Diet matters too — the pressures that shrink a browsing herbivore are not the ones acting on an insectivore.

More fundamentally, whether the rule is general at all has been genuinely disputed. Analyses in the 2000s argued that the apparent pattern was largely an artefact of which clades were sampled and how sizes were compared, and that within individual groups the trend often vanished. More recent large comparative studies across mammals, birds and reptiles have reported support for it, with the effect strongest on small, remote islands and in particular lineages. The honest position is that the direction of the pattern is well attested in the classic cases and its generality across all taxa is still argued.

One misreading to head off: dwarfing is not degeneration, and gigantism is not flourishing. Both are the same process — a population settling at the size its new circumstances reward — and both leave the animal exquisitely fitted to a place and badly exposed if predators, humans included, ever arrive.

• Why mountaintops, lakes and forest fragments produce the same effects as true islands.
• How insular dwarfs and giants fare when humans arrive, and why so many are known only as fossils.

Every term the collection defines is gathered in the glossary.