finch species on remote islands such as Hawaii and
the Galápagos are highly divergent in terms of beak
sizes (and seed sizes utilized) compared with
finches on continents and, plausibly, this can be
related to an absence of competitor lineages on the
islands (Schluter 1988).
Increases in niche breadth within an island line-
age are not always morphologically apparent. The
ancestor of the Darwin’s finch of Cocos Island,
Pinaroloxias inornata, colonized an island where
‘empty’ niche space and the lack of interspecific
competition provided conditions for ecological
release. This endemic finch has diversified behav-
iourally and, while showing little morphological
variation, exhibits a stunning array of stable indi-
vidual feeding behaviours spanning the range nor-
mally occupied by several families of birds (Werner
and Sherry 1987). This intraspecific variability
appears instead to originate and be maintained
year-round behaviourally, possibly via observa-
tional learning.
Density compensation
Density compensationrefers to higher than nor-
mal (i.e. than mainland) densities of a species on an
island, linked to the lower overall richness of the
island assemblage, and often suggested to indicate
a form of competitive release. It was recognized by
Crowell (1962) in a comparative study of birds on
Bermuda with those of similar habitats on the
North American mainland. This density adjust-
ment may occur to an apparently excessive degree,
in which case it is termed density overcompensa-
tion(Wright 1980). The occurrence and degree of
density compensation is variable, and a single
species may exhibit higher, lower, or comparable
densities on different islands as compared with
the mainland. This is, on reflection, unremarkable,
as particular species experience different condi-
tions on different islands, e.g. in relation to pres-
ence/absence of predators, the presence/absence
of competitors, or the suitability of habitats.
To examine the generality of density compensa-
tion effects, we should consider entire guilds or
entire assemblages rather than just subsets of
species. Remote islands are typically considered
species poor for their area, and one corollary of this
is that the species–abundance distribution is
expected to be attenuated, with fewer very rare
species able to sustain a presence than would be
found in an average equivalent area of mainland.
Where colonizing species find ‘empty niche space’
and undergo ecological release, this effect will be all
the greater. It follows that the average population
density of species occurring on islands should be
greater than for a mainland assemblage, and that
this effect should be more marked the lower the
richness per unit area (Fig. 7.1).
As discussed in the island ecology section, varia-
tion in island species richness can best be accounted
for by bringing additional variables, particularly
climatic and habitat factors, into the regression
models. Given which, we ought to take account sta-
tistically of the same variables in testing if overall
density levels differ from those that would be antic-
ipated as a function of the overall resource base of
an island and the accompanying richness of species
thus supported. In short, density compensation
across an entire assemblage is expected as a simple
corollary of lower richness, and this ‘null model’
should first be considered before reaching for other
factors in explanation (Williamson 1981).
Even so, there is a considerable literature on
density compensation, focusing on a variety of
explanations for apparent cases of overcompensa-
tion. Two prominent ideas are: (1) a species being
able to find more food per unit effort and thus
attaining a higher abundance within the same
habitat without a niche shift; and (2) niche
enlargement, such that a species utilizes a wider
range of habitats, foraging strata, foraging tech-
niques, or dietary components (MacArthur et al.
1972). In turn, these mechanisms may be under-
pinned by the compositional character of the
island assemblage, for example, the absence of top
predators, or of large-bodied competitors. In the
latter case, a similar bird biomass might be
distributed amongst more numerous smaller-
bodied bird species.
Blondel and Aronson (1999) provide an example
(Table 7.1) of increased density through both mech-
anisms (1) and (2) among tits (Parusspecies) in
Corsica. The island has three species, and the nearby174 ARRIVAL AND CHANGE