communities simply cannot be detected on a
one-on-one basis (if it ever can), and returning to
the analogy, the jigsaw is continually being
disturbed and reformulated in new ways, while
perhaps never quite being completed.
To conclude, the island ecological rates for
Krakatau contain a lot of sampling noise, through
which clear structural features are evident, reveal-
ing island recolonization to be in essence a special
case of succession. Rates of turnover have, in addi-
tion, been affected by the physical environmental
dynamics of the system. Hierarchical features
across trophic levels are evident and it has also
been argued that, even within a single taxon, not all
species additions have equivalent effects. The bird
data from Anak Krakatau indicate that although
the addition of another insectivore or frugivore
may not perturb the existing community structure
significantly, the addition of the first predatory
birds caused a major community perturbation
(Thornton 1996). Similarly, the appearance of the
first animal-dispersed fruits, and the first birds, and
the first bats, post-1883 must also have constituted
important events. A more recent example is the
crossing of a threshold population size of fig trees
on Anak Krakatau, in providing for resident popu-
lations of fig wasps and of frugivores (Compton
et al. 1994). Once again, we see that different taxa
reveal a variety of trajectories of their colonization
curves, and greater and lesser degrees of turnover.
5.5 Concluding observations
This chapter has reviewed a range of forms of struc-
ture in the composition and dynamics of island
biotas. We have paid particular attention to succes-
sional processes, as exemplified first within the
New Guinea birds studies of Jared Diamond and
colleagues, and second, through the Krakatau case
study. Notwithstanding the debates concerning the
statistical tests of such patterns, the evidence for
non-random patterns from a variety of empirical
studies is overwhelming (Schoener and Schoener
1983; Weiher and Keddy 1995). Competition is
merely one of the forces involved, but it does have
a role, even if it remains difficult to demonstrate.
Progress has been made, as the tools of analysis
of island assembly patterns, incidence functions
and nestedness have been debated, refined and
evaluated. In the very nature of the response vari-
ables, this body of work encompasses more com-
plexity than the previous chapter and in some ways
has been harder to advance. By encompassing more
complexity, it is correspondingly harder to find
generality. Yet, theories of island assembly may in
the end be of as much or perhaps more practical
value to conservationists (Chapter 10; Worthen
1996) than, for example, are attempts to refine geo-
metric models of the relationship between coastline
configuration and immigration rate. We hope thus
to have shown how the study of ecological bio-
geography can shed some light on the species num-
bers games of the previous chapter. The aim of the
next chapter is to provide a synthesis drawing
some of these ideas together in a re-assessment of
progress in island ecology.5.6 Summary
Island biotas are not simply random draws from
regional species pools. Instead, they typically
exhibit compositional structure: some species,
species combinations, or species types, are found
more frequently, and some less frequently, than
might be expected by chance. A comprehensive
island assembly theory was formulated by Jared
Diamond, in his studies of birds on islands in the
Bismarck and other groups around New Guinea.
His analyses were founded on several forms of dis-
tributional patterns. Incidence functions describe
the frequency of a species as a function of island
species richness (or sometimes island size). Those
species found only on the most species-rich islands,
which were also the larger, land-bridge islands,
were termed high-Sspecies. Supertramps were, in
contrast, found only on small, remote, species-poor
islands, including islands recovering from past dis-
turbance. Between these two extremes are the ‘lesser
tramps’, found on varying numbers of the islands.
Chequerboard distributions, were also observed,
whereby taxonomically and ecologically related
species have mutually exclusive but interdigitating
distributions across a series of islands. Within par-
ticular guilds of birds, Diamond found evidence forSUMMARY 143