(e.g. sea-dispersed vs bird-dispersed plants) can
differ greatly, which may also serve to shape the form
of species–area comparisons across archipelagos.
For a moderately dispersive group, it might be
assumed that very small distances have no impact
on colonization and thus on species richness. At
the other extreme, beyond a certain limit, in the
order of hundreds of kilometres, the group might
be absent, and thus further degrees of isolation
again have no impact. Similarly, Martin (1981) has
shown that where the range of areas spans small
island sizes, ISAR slopes are liable to be lower
than for larger areas. It is thus not surprising that
comparative studies show a range of forms for
species–area effects between taxa and islands,
with different explanatory variables to the fore
(e.g. Sfenthourakis 1996). The collective value of
such studies is the promise of refinement of, for
instance, the dispersal limits for particular taxa,
and thus of insights into the dynamics of particu-
lar insular systems (e.g. Lomolino and Davis
1997).
What shape is the species–area relationship?
Interest in the form of the species–area relationship
is enduring amongst ecologists. Recently,
Rosenzweig (1995) has argued that there are four
different ‘scales’ of SPARs (the point scale, islands,
intraprovince, interprovince). But, as we show in
Box 4.4, it is not only the scale that varies between
studies, but the spatial organization of the sample
units and the nature of the response variable itself.
Hence, the first thing to recognize in attempts to
understand the form of the species–area relation-
ship is that the literature encompasses apples
(species accumulation curves: SACs) and oranges
(island species–area relationships: ISARs). Whilst
SACs take a variety of forms, the distinction
between them and ISARs is fundamental (Box 4.4,
Grayet al. 2004). To reiterate, an ISAR is simply the
relationship for a set of islands between the number
of species occurring on each island and the area of
each island.
So, what is the shape of the ISAR? A trite answer
would be ‘what shape do you want it to be?’ As set
out above, the conventional approach has been to
find a simple transformation of the data that more
or less linearizes the relationship, as straight-line
fits make the relationship tractable for further
analyses. Tjørve (2003) has perhaps controversially
argued that our approach to model fitting for
SPARs should ‘be based on the recognition of biol-
ogy rather than statistics’ (p. 833), i.e. that we
should examine the fit of theoretical (mechanistic)94 SPECIES NUMBERS GAMES: THE MACROECOLOGY OF ISLAND BIOTAS
Box 4.4 Combining species accumulation curves and island species–area relationships:
apples and oranges?
In Box 4.2 we set out a simple typology of
macroecological tools used in analysing species
diversity phenomena in relation to area. Although
there are several ways in which areas of mainland
may be sampled (graphics a, b, and c in Box 4.2)
in order to construct a curve of cumulative species
number with increased sampling effort (an SAC),
there is only one way to construct an ISAR, and
that is to plot the number of species per island
versus area of that island. The ISAR approach can
also be used for habitat islands, i.e. isolated
patches of a distinct type of habitat that are
surrounded by a strongly contrasting ‘sea’ of
other habitats. In principle, it can also be adopted
for comparing scattered sample plots ofincreasing area from within an unfragmented
continuous ecosystem, but typically, published
analyses use spatially nested sampling systems
instead. Figure 4.2, derived from Wilson (1961), is
a classic example. The mainland New Guinea data
set represents cumulative areas of New Guinea: it
is thus a nested sampling design, and hence
species number is of necessity cumulative.
According to the typology in Box 4.2, this makes
it essentially a SAC on a regional scale, as distinct
from the island data also plotted in Fig. 4.2,
which we would term an ISAR. By plotting an
ISAR for the island data and comparing them with
an SAC for the mainland, might we be comparing
apples and oranges?