0198566123.pdf

ranges: ordinary islands, 0.05–1.132; habitat
islands, 0.09–0.957; and mainland samples
0.276
to 0.925. It is important to establish the reliability of
MacArthur and Wilson’s generalizations on z, and
whether the apparently errant values reported by
Williamson can be explained away. More generally,
the simple formulation S cAzin practice has pro-
vided a wealth of opportunity for confusion and
controversy and we therefore return to the topic
and update what we have learnt since MacArthur
and Wilson, below.

Species abundance distributions

In most plant and animal communities there are
few species of many individuals and many more
species of few individuals. This is one of the most
robust patterns in community ecology. There are
two foundational theories concerning the distribu-
tion of abundance, they differ in relation to the
abundance of the rarer species. Fisher et al. (1943)
suggested that the largest class of species is of those
that are individually rarest. This gives rise to the
logarithmic series of abundance. Preston (1948,
1962) developed the alternative theory that species
more typically fit a log–normal series of abun-
dance, i.e. that the most numerous species are

those of middling abundance. He argued that

insufficient sampling commonly had given rise to

the apparent fit of the logarithmic model. This can

be illustrated diagrammatically for a hypothetical

community if the abundances of individual species

are plotted on a log scale against number of species

on an arithmetic scale (Fig. 4.3). If the sample is

small, then the sparsest species of the log–normal

distribution will not be sampled, and the abun-

dance distribution will be that shown to the right of

theveil linedrawn at point a (Preston 1948). On

increasing the sampling effort, more of the rarer

species of the system will be sampled, pushing the

veil line to position b. By analogy, a small area

veils’—or rather excludes—the existence of all the

species whose total abundance falls below a critical

minimum.

To put this into a more familiar ecological context,

imagine an area delimited for study within a contin-

uous woodland habitat, containing many birds, but

just one or two pairs of a particular bird species—the

rare end of the distribution. Through chance effects,

such as harsh weather or predation, the species

might be lost locally, but be replaced almost imme-

diately by other individuals from the surrounding

woodland taking over the territory. Repeat this

thought experiment, but for an isolated island: the

82 SPECIES NUMBERS GAMES: THE MACROECOLOGY OF ISLAND BIOTAS

2.6

2.1

1.6

1.1

0.6

0.1

0 1 2 3 4567

Log species number

Log area (km^2 )

logS= 1.54 + 0.088 log A

logS= 0.03 + 0.298 log A

Figure 4.2Species–area relationships for ponerine and cerapachyine ants in Melanesia. Solid dots represent islands; open circles, cumulative
areas of New Guinea up to and including the whole island; triangles, archipelagos (not used in the regression); and the square, all of South-East
Asia. (Adapted from Wilson 1961.)