42 Paul V.A. Fineetal.
biomes more frequently crossed than others, and
if so, is there a relationship with the length of
the boundary line? Does lineage diversification
(and expansion into a new biome) coincide with
increases or decreases in the area of a biome over
time?
Species–energy hypothesis
The relationship between extant log biome area
and log biome species richness is not linear
largely because of the two boreal biome areas
(Figure 3.3). These boreal sites cover dispropor-
tionately large areas, yet are depauperate in tree
species (Table 3.1). Rosenzweig (1992, 1995)
accounted for this discrepancy by adding pro-
ductivity as a corollary to the GAH. Indeed,
leaving aside area, many other proponents of the
“species–energy hypothesis” have linked global
diversity patterns to productivity (or correlates of
productivity: Curri eand Paquin 1987, Adams and
Woodward 1989, Wrightet al. 1993). All of these
studies found strong correlations between tree
diversity and actual evapotranspiration (AET).
What is th em echanism by which productivity
affects speciation and extinction? The common
view is that higher productivity enables more
individuals to inhabit an area, increasing pop-
ulation density and leading to higher specia-
tion and lower extinction rates compared with
areas with low productivity (Gaston 2000). This
argument is strikingly similar to that underly-
ing the GAH. Ricklefs (1999) argued convinc-
ingly that explanations involving environmental
determinism (like productivity) and explanations
involving history and regional effects should be
disentangled. Although productivity (or energy)
may be important at the local level, regional
processes and historical events contribute to
species richness patterns and can override local
effects (Ricklefs and Schluter 1993, Ricklefs
1999).
Boreal biomes have been in existence for only
th epast 4–10 million y ears (Graham 1999,
Willis and McElwain 2001). Because trees require
specific adaptations to survive boreal climates
(Woodward 1987), it would require a radia-
tion several orders of magnitude faster than the
fastest known plant radiation (Hawaiian silver-
swords, Baldwin and Sanderson 1998) for boreal
biomes to have species richness totals similar to
tropical biomes. In other words, while produc-
tivity may slightly mediate the effect of area (by
influencing the number of individuals that can
shar espac ein a biom e), th e eff ect of produc-
tivity on current tree species richness patterns
must be negligible compared with the effect of
th esiz eof biom es through tim e(McGlon e1996,
Fine and Ree 2006). As a thought experiment
(cf. ter Steegeet al. 2000), let us imagine a
world where moist tropical areas were small and
periodically reduced in size, perhaps by extreme
dryn ess, whil elarg e extra-tropical ar eas with cli-
mates similar to today’s boreal biomes stayed the
sam esiz efor t ens of millions of y ears. In such
a world, would w efind highly div ers etropical
rainforest and a low-diversity boreal forest, or the
reverse?CONCLUSION
We evaluated the importance of geographic area
in explaining tree diversity patterns at the largest
spatial scales. Because area is predicted to influ-
encein situspeciation and extinction rates in
a biome, we considered changes in a biome’s
siz eov er tim ep eriods appropriat efor thos epro-
cesses. We tested the GAH with empirical data
on global tree diversity and estimates of biome
extent over three large slices of time within the last
55 million years to the present, finding a signifi-
cant relationship between biome size integrated
over time and current species diversity. Although
oth er explanations may also b evalid for th elat-
itudinal gradient in tree diversity, our analysis
sugg ests a significant rol efor th esiz eand ag eof
a biome area in determining its species richness.
Under this explanation, tropical forests simulta-
neously represent both a museum that preserves
and a cradle that generates new lineages. In addi-
tion, the differential reduction of habitable area in
tropical, temperate, and boreal zones, which likely
caused differential increases in extinction during
th ePl eistoc en ein each of th e11 biom ear eas, may
have been an important factor affecting current
tree diversity patterns.