70 THE BIOGEOGRAPHY OF ISLAND LIFE
particular host plants required by butterfly larvae.
This may limit their potential for rapid evolutionary
change on islands (see above). On the other hand,
the capacity of lepidopterans to reach moderately
remote oceanic islands should not be too lightly dis-
missed, as evidenced by Holloway’s (1996) long-term
light-trap study on Norfolk Island, which recorded
38 species of non-resident Macrolepidoptera over
a 12 year period. Norfolk Island is 676 km from
New Caledonia, 772 km from New Zealand, and
1368 km from the source of most of the migrants,
Australia. Most of the arrivals appear to be corre-
lated with favourable synoptic situations, such as
the passage of frontal systems over the region.
Lizards are also widely distributed in the tropical
Pacific. The proportions of skinks in the three cate-
gories of endemic are remarkably similar to the
equivalent figures for Pacific birds (Table 3.4). In
general, in each of skinks, geckos, birds, and but-
terflies, the proportion of species endemic to an
archipelago is best explained by reference to the
number of large (high) islands and to total land
area. However, scrutiny of the data on an archipel-
ago-by-archipelago basis reveals greater differ-
ences. For instance, 100% of New Caledonia’s
skinks are endemic to the Pacific Ocean islands,
and as many as 93% are endemic to the New
Caledonian islands themselves. The equivalent fig-
ures for birds (including subfossils) are 47% and
33%, respectively (Adler et al. 1995). These data
contrast with those for Hawaii, on which most
birds are endemic, but where there are no endemic
lizards. Adler et al. (1995) suggest that these differ-
ences are explicable in relation to the differing
dispersal abilities of lizards and birds. Birds, being
better dispersers, reached Hawaii relatively early
and have radiated spectacularly, whereas the three
species of skink may only have colonized fairly
recently and possibly with human assistance. New
Caledonia, in contrast, may not be sufficiently iso-
lated to have allowed such a degree of avifaunal
endemism to have developed.
Numbers of birds and butterflies are remarkably
similar on the less remote Pacific archipelagos east
of New Guinea, but on the more remote archipela-
gos the numbers of bird species are far greater than
those of butterflies (Adler et al. 1995). This is
reflected statistically by archipelago area being the
most important geographical variable in explaining
bird species richness, whereas for butterflies isola-
tion is the more important variable. Another pat-
tern noted by Adler et al. (1995) is that in Pacific
butterflies, Pacific birds, and Antillean butterflies
the number of endemic species increases more
steeply with island area than does the number of
more widely distributed species. This observation
supports the idea that a larger island area provides
the persistence and variety of habitats most con-
ducive to radiation on isolated islands (see further
discussion in Chapter 9).
The proportions of Pacific island mammals in
each of the endemism categories falls closer to the
figures for birds and skinks than those for butter-
flies. However, the 5 marsupial species make it noTable 3.4Degree of species endemism among tropical Pacific and Indian Ocean island faunas. (From Adler
1992, 1994; Adler and Dudley 1994; Adler et al. 1995; Carvajal and Adler 2005.)
Group Total number Continental Regional endemics Local endemics
Pacific Ocean butterflies 285 157 (55%) 28 (10%) 100 (35%)
Pacific Ocean skinks 100 21 (21%) 13 (13%) 66 (66%)
Pacific Ocean birds 592 145 (25%) 59 (10%) 388 (65%)
Pacific Ocean mammals 106 42 (40%) 7 (6%) 57 (54%)
Indian Ocean birds 139 60 (43%) 10 (7%) 69 (50%)
Continental, species also occurring on continental land masses;regional endemics, species occurring
on more than one archipelago within the region but not on continents or elsewhere;local endemics,
species restricted to a single archipelago or island. NB: Carvajal and Adler (2005) mistakenly give 292 as the
figure for local endemic Pacific Ocean birds. The correct figure is as given above (G.H. Adler, personal
communication, 2005).