The Sixth Extinction: An Unnatural History

nest is blown out of a tree in a hurricane. The following year, all the chicks
turn out to be males, and the year after that, the nest is raided by a snake.
Species X is now headed toward local extinction. If the island is home to
two breeding pairs, the odds that both will suffer such a string of fatal bad
luck is lower, and if it’s home to twenty pairs, it’s a great deal lower. But
low odds in the long run can still be deadly. The process might be
compared to a coin toss. It’s unlikely that a coin is going to come up heads
ten times in a row the first ten (or twenty or hundred) times it is flipped.
However, if it’s tossed often enough, even an unlikely sequence is likely to
occur. The rules of probability are so robust that empirical evidence of the
risks of small population size is hardly necessary; nevertheless, it’s
available. In the nineteen-fifties and sixties, bird-watchers kept
meticulous records of every pair that bred on Bardsey Island, off Wales,
from common house sparrows and oystercatchers to much rarer plovers
and curlews. In the nineteen-eighties, these records were analyzed by
Jared Diamond, who at that time was working as an ornithologist,
specializing in the birds of New Guinea. Diamond found that the odds that
any particular species had gone missing from the island could be plotted
along a curve whose slope declined exponentially as the number of pairs
increased. Thus, he wrote, the main predictor of local extinction was
“small population size.”
Small populations, of course, aren’t confined to islands. A pond may
have a small population of frogs, a meadow a small population of voles.
And in the ordinary course of events, local extinctions occur all the time.
But when such an extinction follows from a run of bad luck, the site is
likely to be recolonized by members of other, more fortunate populations
wandering in from somewhere else. What distinguishes islands—and
explains the phenomenon of relaxation—is that recolonization is so
difficult, in many cases, effectively impossible. (While a land-bridge island
may support a small remnant population of, say, tigers, if that population
winks out, new tigers presumably aren’t going to paddle over.) The same
holds true for any sort of habitat fragment. Depending on what surrounds