(Lockwood and Moulton 1994). Only a proportion
of introduced species succeeds in establishing and
maintaining populations over lengthy periods. In
tests of data for introductions of passeriform birds
to the Hawaiian islands, Tahiti, and Bermuda, it has
been found that the surviving species exhibit what
is termed morphological overdispersion, i.e. they
are more different from one another in characteris-
tics that are believed to relate to niche separation
than chance alone would predict (Lockwood and
Moulton 1994). Similar results have been reported
for game birds (galliforms) for the Hawaiian
islands and New Zealand (Moulton et al. 2001).
These results appear to support the hypothesis that
interspecific competition shaped the composition
of these anthropogenic communities.
Like other island assembly rules the morpholog-
ical overdispersion pattern require careful scrutiny
to ensure that it is not an artefact of, for example: (1)
the biogeographic region of origin of the intro-
duced species, (2) the ecological or taxonomic struc-
ture of the introduced species, or (3) the numbers of
introduced birds (e.g. see Blackburn and Duncan
2001; Cassey et al. 2005a). Of these factors, the intro-
duction effort has been highlighted as crucial in a
number of data sets (Cassey et al. 2005a,b). On the
other hand, in their analysis of game bird introduc-
tions, Moulton et al. (2001) argue that evidence for
the introduction effort hypothesis in their data set is
weak, citing examples of game birds failing because
they were introduced into inappropriate environ-
ments in rather indiscriminate fashion: in such cir-
cumstances mere weight of numbers may not
ensure survival. Hence, although it is certainly the
case that greater numbers of released birds
provides a greater likelihood of initial success of a
species establishing, they argue that the emergent
morphological pattern of the assembled community
of exotic species shows evidence of competitive
structuring.
Just when you think it might be safe to conclude
in favour of competitive effects, once again the case
has been disputed, by reference to the data derived
from the New Zealand game-bird introductions.
Duncan and Blackburn (2002) slay the beautiful
hypothesis with the traditional ugly fact. They
argue that competition among morphologically
similar species could not have been responsible for
the failure of most game-bird introductions to New
Zealand because most species were released at
widely separated locations or at different times,
and would never have encountered other morpho-
logically similar introduced game birds, or if they
did would not have done so at the necessary densi-
ties to influence the outcome. They conclude that
the significant pattern of morphological overdis-
persion in this instance must therefore result from
some other cause, such as, for example, greater
effort being expended on introductions of morpho-
logically distinct species. Interactions with intro-
duced predators, and with other anthropogenic
changes in New Zealand further complicates attri-
bution of the overdispersion pattern (Duncan et al.
2003). Accepting the criticisms of the New Zealand
requires that the Hawaiian data stand alone, and
taken in isolation the Hawaiian galliform data fail
to show overdispersion, and so, according to
Duncan and Blackburn (2002) the case falls. From
this we may conclude that the most predictable fea-
ture of island assembly rules is that any claim of
evidence of competitive effects will be contested: at
last some evidence of competition, even if only
amongst biologists!5.2 Nestedness
If island biotas are in essence randomly drawn
from a regional species pool, as per the simplest
form of the EMIB, then an archipelago of islands
should exhibit differences in composition from one
island to the next and the degree of overlap should
be predictable on the basis of a ‘null’ model.
Diamond’s assembly rules seek to describe depar-
tures from such an expectation, another form of
departure is where island biotas exhibit nestedness.
Anested distributiondescribes the situation where
smaller insular species assemblages constitute sub-
sets of the species found at all other sites possessing
a larger number of species (Patterson and Atmar
1986). Diamond (1975a) intended his assembly
rules approach to be deployed on groups of ecolog-
ically related species, or guilds, whereas the analy-
ses of nestedness (like incidence functions) provide
descriptive tools that can be applied to either126 COMMUNITY ASSEMBLY AND DYNAMICS