was taken to imply additional combinational rules
preventing apparently reasonable combinations
occurring with the frequency to be expected by
chance. As most changes to an island’s avifauna are
likely to be produced by single species additions or
losses, recombinations requiring two or more gains
and/or losses simultaneously (to avoid a ‘forbid-
den’ combination) might be anticipated to have low
transition probabilities. Hence, some combina-
tions might be viable, but difficult to assemble from
the ‘permitted’ combinations.
Diamond undertook similar statistical analyses
for the guilds of gleaning flycatchers, myzomelid
sunbirds, and fruit-pigeons. The niche relation-
ships among the species of PtilinopusandDucula
fruit-pigeons provide a classic example of resource
segregation. Diamond first established the form of
community structure within the ‘mainland’ low-
land forests of New Guinea. The larger pigeons
forage preferentially on bigger fruits, providing
one element of resource segregation. If large and
small species feed within individual trees, the
lighter pigeons are able to feed on the smaller,
peripheral branches, better able to support their
weight. Thus niche separation can be maintained
within a guild of co-occurring species. Yet, within
a single locality in New Guinea, no more than
eight species will be encountered, forming a
graded size series, each species weighing approxi-
mately 1.5 times the next smaller species. There
are 8 size levels filled by combinations drawn
from 18 species. On satellite islands off New
Guinea, subsets are drawn in such a way that on
smaller or more remote islands, size levels are
emptied as the guild is impoverished according to
a consistent pattern. Level 1 (the smallest birds)
empties first, followed by levels 2 and 5, followed
by level 8. It is intuitively sensible that the small-
est birds should be excluded first as they will be
restricted to the smallest fruit types, for which
they would have to compete with slightly larger
pigeons that can also make use of larger fruits.
This form of structuring could also be identified
on those islands regarded as being supersaturated,
or undersaturated, suggesting that the time
required to adopt such structure is shorter than
that required for equilibration of species number
on islands. This observation resonates with subse-
quent analyses of ecosystem build-up within the
flora and at least some elements of the fauna of
Krakatau (e.g. Whittaker et al. 1989; Zann and
Darjono 1992; Thornton 1996).
The above constitutes the bulk of Diamond’s
species assembly theory as it concerns us here.
However, he also applied the ideas to non-insular
communities, a context in which many of the
recent developments in the formulation and test-
ing of assembly rules have come (e.g. Wilson and
Roxburgh 1994; Weiher and Keddy 1995; Wilson
and Whittaker 1995). Diamond’s studies incorpo-
rated distributional data at the island level in the
first instance, but also involved: data on habitats
in which particular species were observed or
trapped; body weights; feeding relationships;
observations of bird movements between islands;
and historical records of species colonizations and
extinctions, in cases providing apparent evidence
of the unsustainability of certain combinations of
species.Criticisms, ‘null’ models, and responses
Considering how much evidence for prehistoric anthro-
pogenic extinctions of island birds has emerged over the
past decade, biogeographers today should hesitate to
regard any distributional attribute of modern South
Pacific birds as being uninfluenced by human activity.
(Steadman 1997b, p. 750)
Diamond’s papers provide a coherent theory of
the biogeography of the New Guinea island birds.
This island assembly theory could be seen as a
development of several of the lines of argument
to be found in MacArthur and Wilson’s (1967) The
theory of island biogeography. Yet, the approach,
and particularly the assembly rules, drew criti-
cism from Daniel Simberloff and colleagues
(Simberloff 1978; Connor and Simberloff 1979).
The problems were sufficiently troublesome, or
intractable, as to hamper subsequent progress,
and the approach has arguably failed to receive
the attention that its intrinsic interest warrants
(Weiher and Keddy 1995).
One element of the logic employed by Diamond
that might be criticized is the implicit acceptanceISLAND ASSEMBLY THEORY 113