evolutionarily conditioned by such storm events
(Whittaker 1995; Walker et al. 1996).
Not all extreme weather phenomena take the
form of storms. ENSO phenomena are large-scale
interannual events that are the product of varia-
tions in air pressure and associated rainfall patterns
between the Indonesian and South Pacific regions,
coupled with ocean-current and temperature
variation. In the Asia–Pacific region, ENSO events
are associated with heightened amplitude of inter-
annual climatic variation—that is, more intense
droughts and wet periods. As an illustration, the
island of Aututaki (Cook Islands) received
3258 mm rainfall in December 1937, which is in
excess of the average annual total (Nunn 1994).
That events of the magnitude of ENSO phenomena
have a detectable ecological impact, even on popu-
lations conditioned evolutionarily to the island way
of life, was shown by the significant increase in
numbers of four endemic Galápagos bird species
(three finches and a mockingbird) as a consequence
of the increased food supply resulting from the
exceptionally high rainfall during 1982–1983 (Gibbs
and Grant 1987). Analyses of historical data for
hurricanes show that the frequency of climatic phe-
nomena such as hurricanes and ENSO events
varies over decadal time scales, further complicat-
ing analysis. This goes to illustrate that, despite the
reddened spectrum of climatic variation, short-
term fluctuations deserve the attention of ecologists
and biogeographers (Stoddart and Walsh 1992;
Benton and Spencer 1995; Schmitt and Whittaker
1998).
There is a premise that, because of the compara-
tive simplicity of geology and climate of many
oceanic islands, their landscapes should attain a
condition of dynamic equilibrium. Nunn (1994)
argues that this notion cannot be supported as a
generality:
... as is evident from the often catastrophic impacts of
certain irregular climatic phenomena on oceanic island
landscapes, the degree to which particular environments
can accommodate the impacts of particular uncommon or
extreme events without fundamental alteration, while
variable, is generally low. For those oceanic island land-
scapes in a state of dynamic equilibrium, the effects of
such events may be to cause a threshold of landscape
development to be crossed.... The effects of irregular
climatic phenomena are so variable, so site-specific, that it
is pointless to attempt a broad generalization. (pp. 159–160)
These observations might prompt us to ask
ourselves, first, with what portion of the magnitude–
frequency spectrum of climatic variation might
oceanic island biotas best be considered to be in
dynamic equilibrium and, second, might not extreme
climatic events have caused ecological or evolution-
ary thresholds to be crossed in the same manner as
described above for their landscapes?Disturbance from volcanism and mega-landslides
In contrast to the geological stability of continental
fragments or continental islands, oceanic islands
are characterized by the repetitive incidence of geo-
logical catastrophes, amongst which volcanic erup-
tions and gravitational flank collapses or landslides
(with the tsunamis they generate affecting nearby
islands) are the most frequent.Volcanic eruptions
Volcanic eruptions are the main constructive forces
of oceanic islands: they bring them into existence,
and subsequently make them larger and higher,
whereas the processes of erosion (by rain, wind, and
sea) and subsidence provide the opposing forces
that reduce them to sea-level or below (e.g. Figures
2.11, 2.15). However, the same volcanic processes
that enlarge islands simultaneously damage or
destroy island ecosystems. The impacts can be dif-
fuse, whereby volcanic ash is deposited in thin lay-
ers over a wide area, or can be intense and entirely
destructive. In some cases, destructive volcanic ash
flows can be deposited over the entire island, more
or less completely eliminating the existing biota.
This is known as island sterilization, and implies
that colonization has to begin from nothing. It is, of
course, very difficult to demonstrate complete ster-
ilization, even for historical events such as the 1883
Krakatau event (Backer 1929; Docters van Leeuwen
1936; Whittaker et al. 1989; Thornton 1996). The chal-
lenge is even greater for events as distant as Gran
Canaria’sc.3.5 Ma event (Marrero and Francisco-
Ortega 2001a,b), although in this case it can be
inferred from phylogenetic evidence that someNATURAL DISTURBANCE ON ISLANDS 41