species richness within the Bismarck Islands:
A-tramps 3–9 islands; B-tramps 10–14 islands;
C-tramps 15–19 islands; and D-tramps 20–35
islands. The smallest or most species-poor islands
generally have representatives of both tramp and
supertramp categories.
The conclusion reached from the incidence func-
tions was that the island avifaunas represented
highly non-random selections of the species pool.
Rather, it appeared that species distributions were
related to factors such as: competition; the absence
of suitable habitat; island sizes being smaller than
minimum territory requirements; seasonal or
patchy food supply; historical factors such as land-
bridge connections; disturbance and changes in car-
rying capacity; and interactions of these factors
with population fluctuations.
The dynamics of island assembly
In addition to being found predominantly on the
smaller islands, the supertramps were also charac-
teristic of the recently recolonized volcanic islands
(Diamond 1974), suggesting that these species are
excellent colonists but poor competitors. The more
widespread tramp species, being found in rela-
tively high frequencies on all but the smallest
islands, must, in contrast, be both capable colonists
and good competitors. The species restricted to
large islands might be absent from others because
of poor dispersal or some form of intrinsic unsuit-
ability to small islands. As this high-Sset includes
species of larger body size (e.g. herons), and species
of large home ranges and strong dispersal potential
(e.g. large hawks), the latter appeared important, at
least for some of the species involved. Diamond
concluded that the high-Sand A-tramp categories
actually included a fairly heterogeneous mix of
species, some limited by dispersal and some by
other factors. About 56% of the high-Sspecies are
endemic to the Bismarcks at the semispecies or
species level, such that the patterns under discus-
sion retain at least some signal from evolutionary
timescales.
Some species distributions were interpretable in
relation to habitat requirements. For example, as
only five islands in the data set exceed 3500 feet
(1067 m) elevation, the 13 montane bird species are
restricted to these 5 islands. Of the 5 islands, one is
very small and another was recently defaunated by
volcanic explosion: 12 of the 13 montane birds are
lacking from both of these islands. Similar fits to
habitat could be found for other Bismarck bird
groups. The endemic species, which are mostly also
high-Sspecies, are mostly lowland forest or moun-
tain species. By contrast, the tramp species confined
to scarcer lowland habitats were not found to have
differentiated beyond subspecies level. This was
thought to be indicative of the greater frequency of
dispersal needed to maintain their lineage in a scat-
tered system of habitat patches across the archipel-
ago. The supertramp species differ again, in that
where they occur on an island, they tend to occur in
a greater variety of habitats, and Diamond (1975a,
p. 381) characterized their ecology as ‘Breed, dis-
perse, tolerate anything, specialize in nothing.’
Thus, the lack of supertramps on an island is a
product of competitive exclusion and not of disper-
sal constraints. The gradient between the super-
tramps and the high-S species can be viewed
as equivalent to that between r-selected and
K-selected species, or pioneering and late succes-
sional plant species. In each case, the gradient rep-
resents an approximation of a more complex reality,
with varying degrees of explanatory power from
one case study to another (Begon et al. 1986; and for
a critical evaluation of r–Kselection see Caswell
1989).
As will become apparent in Chapter 9, there are
parallels between Diamond’s ideas and E. O.
Wilson’s (1959, 1961) earlier ideas of taxon cycles.
Both theories seek to explain distributional patterns
by means of competitive effects, habitat relation-
ships, and evolutionary considerations. Although
we have placed them in separate sections of the
book, they are both evolutionary ecological models,
both are dynamic models, and both invoke a key role
for competition. They differ principally in that the
taxon cycle model focuses more on evolutionary
change within the taxon as a biogeographical
process, while the assembly rules ideas are con-
cerned with identifying and explaining composi-
tional regularities across a series of islands, invoking
ecological forces such as ecological succession.110 COMMUNITY ASSEMBLY AND DYNAMICS