PATTERNS OF LOSS ACROSS ISLAND TAXA 309were browsers on understorey vegetation. The ter-
restrial omnivores all but disappeared too, along
with the Hawaiian land crabs. Prehistoric extinc-
tions among insectivores (37%) and arboreal frugi-
vores/omnivores (40%) have been proportional to
the losses among passerines generally, but grani-
vores have suffered much worse and nectarivores
much better in comparison. With predators disap-
pearing, habitats shifting, and species richness
falling during the shake-up following Polynesian
settlement, it is likely that some species, even
guilds, became both relatively and absolutely more
widespread and abundant than under conditions
before human contact. James (1995) argues that the
differential effects of extinction on vertebrate feed-
ing guilds in prehistoric times may still be affecting
plant communities, i.e. that the losses must collec-
tively have implications for the functional charac-
teristics of the remaining ecosystems, but such a
legacy is difficult to separate from the other driving
forces for change.
The pre-human avifauna of New Zealand was
probably about twice the size of the native avifauna
today. The most famous losses were of the giant,
flightless moas, the largest of which weighed
around 250 kg (Rudge 1989; Holdaway 1990; Bunce
et al. 2003). The moa exhibited exceptional variation
in body size in relation to sex and habitat, such that
until very recently there were thought to have been
as many as 11 species. Genetic analyses of bones
recovered from sites around New Zealand reveal
there to have been only two distinct genetic line-
ages, D. novaezealandiae in North Island and
D. robustusin South Island (Bunce et al. 2003).
An interesting story is attached to the loss of one
of the smaller New Zealand birds, the Stephens
Island wren, named Traversia lyalliafter the light-
house keeper (Lyall) who discovered it. Stephens
Island is a tiny island in Cook Strait, off South
Island. The wren was unknown to science until in
1894 the lighthouse keeper ’s cat brought in a few
specimens it had killed. The exotic predator met the
island endemic: end of story. At least, this was the
conclusion reached at the time, prompting a corre-
spondent to the Canterbury Pressto suggest that in
future the Marine Department should see to it that
lighthouse keepers sent to such postings should be
prohibited from taking any cats with them, ‘even if
mouse-traps have to be furnished at the cost of the
State’. Although cats have been claimed to have by
far the worst record in exterminating New Zealand
birds (King 1984), we should not place all the blame
on the lighthouse keeper and his cat. Before the
arrival of the Polynesians, the wren was actually a
widespread mainland resident (Holdaway 1989;
Rosenzweig 1995). Its loss from mainland New
Zealand was probably due to a combination of
habitat changes and the introduction of rats by the
Polynesians; the loss from Stephens Island was but
the final step to extinction. Interestingly, it is just
such small offshore islands that provide the
enclaves that may yet save some of New Zealand’s
remaining endemics from extinction as, cleared of
exotic pests, translocated native populations are
able to establish and persist. Here, in microcosm,
can be found some of the essential ingredients of
the diversity disaster of oceanic islands, and a life-
line of hope for the future.
The Polynesians discovered New Zealand
around AD950, having voyaged from the area con-
taining the Marquesas, the Society Islands, and the
South Cooks, more than 3000 km to the east. They
brought with them six species of plants and two
species of mammals. Polynesian settlement led to
the destruction of half of the lowland montane
forests, widespread soil erosion, and the loss or
reduction of much of the vertebrate fauna. During
the phase of their expansion, New Zealand lost its
sea lions and sea elephants, and at least 25 bird
species, including the 2 species of moa which, in the
absence of terrestrial mammals, had evolved to
occupy the browser/grazer role normally occupied
by animals such as kudu, bushbuck, or deer. The
most severe modification occurred between 750
and 500 years ago, when a rapidly increasing
human population overexploited animal popula-
tions and used fire to clear the land. By about
AD1400, the major New Zealand grazing systems
had ceased to exist, their place being taken by
unbrowsed systems: the removal of the moas thus
had significant effects on the structure and species
composition of the vegetation.
Holdaway’s (1990) analyses of subfossil remains
indicate that the pre-human avifauna was