98 SPECIES NUMBERS GAMES: THE MACROECOLOGY OF ISLAND BIOTAS
bringing climate-based energy availability into
analyses of island species numbers may improve
model fits.
Studies on gamma (regional) scale patterning of
diversity within large landmasses have shown that
remarkably simple climate-based models can
account for the first-order pattern of species rich-
ness regionally and globally (O’Brien 1993, 1998;
Hawkinset al. 2003, Field et al. 2005). These studies
imply that there is a causal connection over time
and space between climate-based water-energy
dynamics, photosynthesis, subsequent biological
activity, and species richness of a wide variety of
taxa—and that it is worth giving more attention to
climate and energy flows in island systems.
However, there are also good reasons why a simple,
universally applicable formulation may prove
elusive.
On theoretical grounds, it is to be expected that
spatial scale will be important to the applicability of
species–energy theory (Whittaker et al. 2001). We
know that the species richness of two samples will
be dependent on how large an area is sampled.
Comparison of a 1 m^2 patch of chalk grassland with
a whole hillside of the same habitat will inevitably
find the whole hillside to be richer. It follows that in
searching for causal explanations of diversity pat-
terns between differing locations, area should be
held constant. On coarse spatial scales (comparing
areas of tens to hundreds of square kilometres), cli-
mate is one of the more prominent sets of variables
influencing plant species richness. At much finer
scales, factors such as geology, slope, and grazing
regime are likely to be proportionally more impor-
tant. That is to say, different theories of diversity
may hold relevance to different (albeit overlapping)
spatial scales (Wright et al. 1993; Whittaker et al.
2001). Island species–energy theory builds both
energy and area in to the ordinate, as island posi-
tion on the energy axis is corrected for total island
area. Thus it is likely that where island sizes span
several orders of magnitude, deviations from the
best-fit relationship may have different causes for
different portions of the island size continuum.
Wright (1983) argues that species–energy theory
is consistent with several long-established patterns.
For instance, the energy requirements higher up the
trophic pyramid are such that smaller areas may be
unable to support top predators, and large-bodied
animals are often among the first to be lost from
areas newly turned into isolates. Wylie and Currie
(1993) have applied the species–energy approach to
mammals (excluding bats) on land-bridge islands
from across the world. They made use of a slightly
different set of explanatory variables, but the
approach was otherwise very similar. They found,
as Wright had for birds and angiosperms, that
island energy explained more of the variation in
mammal species richness than did area. However,
the improvement was much less marked for
mammals. The range of island sizes in the mam-
mals data set was 0.4 km^2 to 741 300 km^2 , and it is
possible that the greater range, and in particular,
the extension of range down to much smaller
islands, may have given rise to a stronger relation-
ship to area in their data. Latitude was also found
to have a stronger effect on mammal species num-
ber than Wright had found. Of 100 log–log
species–area regressions published in Connor and
McCoy (1979), the average explanatory value of
area was less than 50%. In the three data sets
analysed by the species–energy method, the figures
ranged from 70 to 90%.
It is a curious fact that the early impact of Wright’s
(1983) paper was far more evident in the literature
on species-richness gradients within and across con-
tinents, than in island biogeography. However, a
recent paper by Kalmar and Currie (2006), which
models island bird richness globally, marks a key
step forward in the application of these ideas within
island biogeography. Kalmar and Currie compiled a
data set of land-bird species richness for 346 marine
islands, ranging from 10 ha to 800 000 km^2 , and rep-
resentative of global variation in climate, topogra-
phy, and isolation. Using a fifth-root transformation
of richness data (close to a logarithmic transforma-
tion) and a variety of simple transformations of
explanatory variables, they found that as much as
85–90% of the variation in bird species numbers
on islands globally can be accounted for by con-
temporary environmental variables. Their explana-
tory variables included area, elevation, various
isolation metrics, mean annual temperature, and
annual precipitation, and they also examined