BIODIVERSITY AND CLIMATE 191
Fig. 10. The proportion of mammal species
represented by (a) bats, (b) Carnivora and (c)
rodents, as a function of absolute latitude.
levels may, at least in part, reflect the degree to
which each is completely sampled by the fossil
record.
Discussion
In general, the results presented here (Figs 3, 5,
6, 7 and 9) are consistent with the continental-
scale patterns and trends found by Currie (1991)
for North America, and by Pianka (1981) for
Australia, and appear to support the
species-energy hypothesis. However, Currie's
conclusion, that PET (as a proxy for energy)
correlates best with species diversity patterns, is
not found in this study to apply to all continents,
nor to all taxonomic groups (Table 2, Fig. 15).
The problem, as stated at the beginning of this
paper, is that different organisms are limited by
different environmental factors (and combi-
nations of factors) and to different extents. It is
therefore difficult to imagine why any one
environmental parameter should account for all
of the diversity patterns around the globe
(Gaston 2000). It is also important to distinguish
between the pattern of species diversity, which
requires a causal explanation, and latitudinal
gradients, which are a graphical abstraction of
the data. Latitude per se cannot be a determi-
nant of species richness (Gaston 2000).
The links between climate, biogeography and
diversity are obviously complex and, given the
results presented here, must be considered in
terms of not only absolute numbers of taxa, but
also the physiology and ecology of those taxa.
Reptiles show the simplest and steepest diversity
gradients, which appear to be independent of
regional and hemispheric biases (Fig. 3c). As
ectotherms, reptile survival depends, primarily,
upon absorption of energy from the environ-
ment, above some critical minimum energy
(temperature) level (Fig. 4d), and this is sup-
ported by the correlations found in this study
(Tables 2 and 4). The pattern of amphibian
diversity (Fig. 3d) reflects amphibians' physio-
logical and ecological dependence on both
temperature and water (Tables 2 and 5; Fig. 7).
This is a requirement shared by plants, which for
North America at least, show a similar diversity
distribution to amphibians (Fig. 8).
The pattern of mammalian species diversity is
far more complex (Fig. 3b). There is no simple
monotonic gradient from high to low latitudes
and they do show a hemispheric asymmetry. As
endotherms, mammals' dependence on primary
energy sources is indirect through the filter of
their various feeding strategies. Their diversity
patterns may still ultimately reflect a climate
signal (Frey 1992; Janis 1993; Janis et al 2000),
and consequently the physiological and eco-
logical structure of mammalian faunas can be
used to reconstruct past habitats and thereby
climate (Andrews et al. 1979), but an under-
standing of mammal diversity in terms of energy
requires a detailed understanding of mammalian
ecology. Endothermic herbivores, for example,
should show their highest diversity in regions of
high plant productivity, which is itself a function
of temperature and precipitation (Lottes &
Ziegler 1994). Such a conclusion is supported by
the highest rho values for mammal diversity in
Table 2 correlating with mean annual NDVI.
Interestingly, Janis et al. (2000) have postulated
that the decline in North American ungulate
