32 Paul V.A. Fineetal.
biome area should match that of species rich-
ness. However, despite considerable disagreement
on the methods used to define biome bound-
ari es and calculat eth ear ea of th eworld’s biom es
(Terborgh 1973, Rosenzweig 1992, 1995, Rohde
1997, Rosenzweig and Sandlin 1997, Gaston
2000, Fin e2001, Hawkins and Port er 2001), a
continent’s biome areas do not generally increase
in siz etowards th e equator, matching th elati-
tudinal gradient of increasing species diversity.
Moreover, attempts to match biome area with
the species richness of trees (Fine 2001) or birds
(Hawkins and Porter 2001) did not confirm the
prediction that the area of a biome should corre-
late with the total number of species that have
distributions within it. Because of these incon-
sistencies, the GAH has not been embraced by
most biologists concerned with understanding
the latitudinal diversity gradient. Indeed, in a
recent review of hypotheses to explain the latitu-
dinal gradient, the GAH is only briefly mentioned,
and ultimately rejected as unimportant (Schemske
2002).
Should th eGAH b edismiss ed? H er ew emak e
th ecas ethat doing so would b epr ematur eb ecaus e
it has not been adequately tested. First, due to
the difficulty of defining and determining the spa-
tial extent of biomes, and to the lack of reliable
data on global distributions of most organisms,
th etwo studi es that sought to match biom ear ea
with species totals were not conducted with com-
plete datasets: Fine (2001) considered only North
American trees north of Mexico and Hawkins
and Porter (2001) excluded tropical birds from
their study. Second, the mechanisms by which
geographic area is thought to affect species rich-
ness involve speciation and extinction, processes
that operate over large time scales. When com-
paring the sizes of a continent’s biomes with
species richness, one must consider how the
biomes may have changed in area during the time
that its lineages have been undergoing speciation
and extinction (McGlone 1996, Penningtonet al.
2004, Ricklefs 2004). The GAH was advanced
to explain diversity patterns at the largest spatial
scales. Its mechanistic basis lies in differential rates
of speciation and extinction, both of which occur
over time scales of 10^5 –10^6 years (Magallón and
Sanderson 2001, Whittakeret al. 2001, Ricklefs
2003, 2004). B elow, w e explor ea n ew way to
incorporat etim einto an evaluation of th eGAH.
In this chapter, we first review the theory and
evidence for population genetic mechanisms by
which geographic area may influence the species
diversity of a biome. Second, we discuss the diffi-
culties inherent in testing the GAH, and present
recommendations for defining biomes, calculat-
ing their areas, and evaluating the number of
species found within them. Next, we present data
on tree diversity in forest biomes, and evaluate
th eGAH in th econt ext of curr ent biom ear ea
for the world’s moist forests. Then, we correlate
current tree diversity of biomes with estimates of
the areas of past forest biomes integrated over the
last 55 million years. Finally, focusing on trees,
w ediscuss our r esults within th econt ext of two
alternative hypotheses which have been advanced
to explain the latitudinal gradient of species diver-
sity: the tropical conservatism hypothesis and the
species–energy hypothesis.MECHANISMS
Why are larger areas predicted to include more
species? Specifically, what role does area play in
the population-level processes influencing rates
of speciation and extinction? These questions are
important in determining whether large, diverse
areas like the tropics act as cradles of biodiversity
from which new species arise, or as museums that
preserve existing species from extinction (Stebbins
1974, Moritzet al. 2000).
In the cradle-versus-museum debate (reviewed
by Chown and Gaston 2000), an important
premise has been that larger areas allow species
to have larger ranges, and much attention has
been given to how range size relates to speciation
and extinction rates. The expectation of a positive,
peaked, or negative relationship between range
size and the probability of speciation seems to
depend on which parameters (extrinsic or intrin-
sic) are emphasized. For example, consider the
extrinsic effect of geographic barriers, such as
mountain ranges, rivers, etc. If large ranges are
more likely to be subdivided by such barriers,
disrupting gene flow and causing allopatric spe-
ciation, w emight pr edict a positiv er elationship