Tropical Forest Community Ecology

Disparity in Tree Species Richness among Biomes 35

TESTING THE GAAH

We chose three biomes defined on the basis
of temperature parameters that are biologically
meaningful for trees. In our view, tropical, tem-
perate, and boreal biomes should not be defined
by latitudinal boundaries; instead, they should be
defined by the physiological boundaries important
for trees (Fine 2001). We define boreal biomes
as areas that experience− 40 ◦C temperatures
in winter. This is the threshold of spontaneous
nucleation of supercooled water, which requires
a specific adaptation in plants to avoid death by
xylem cavitation (Woodward 1987). We define
temperate biomes as those delimited by the− 40 ◦C
isoclin eon th ehigh-latitud ebord er, and by th e
frost isoclin eon th elow-latitud ebord er. Th efrost-
lin eis important b ecaus eth elack of frost tol eranc e
limits tropical trees from expanding into temperate
areas (Sakai and Weiser 1973, Woodward 1987,
Latham and Ricklefs 1993, Fine 2001). Finally,
tropical biomes are defined as areas that never
experience 0◦C temperatures. Extra-tropical trees
are likely limited from crossing into tropical areas
by the trade-off in growth that accompanies frost
tolerance, giving temperate trees a competitive dis-
advantag ein tropical ar eas (MacArthur 1972,
Lo ehl e1992, Fin e2001).
We divided each of six continents into boreal,
temperate (including subtropical), and tropical
moist/wet and dry forests using the World Wildlife
Fund Ecoregions data tables (Olsonet al. 2001).
W ew er enot abl eto obtain estimat es for th e
world’s tropical dry and temperate Mediterranean
tree floras, so we ignored the dry forest areas and
present only moist/wet forest data. Eurasia and
North America each include two separate temper-
ate moist forest biomes, geographically separated
by mor ethan 1000 km and with almost no ov er-
lap in species composition (Petrides and Petrides
1992, Petrides 1998). Therefore, each of the tem-
perate biomes in these continents is treated here
as a separate entity.
We searched for estimates for tree species rich-
ness for 11 biome areas: two boreal areas (North
America and Eurasia), six continental temperate
areas (Europe, East Asia, Eastern North Amer-
ica, Western North America, South America, and
Australia), and three tropical areas (Neotropics,

African Tropics, and Asian Tropics [including

India, Malayan Peninsula, and Borneo]). The

Australian tropics and Papua New Guinea (PNG)

were not included because we could not find

reliable estimates on the species richness of the

tr e eflora of PNG, nor of th eamount of ov er-

lap between PNG and Asia and/or Australia

(to decide whether Australia/PNG warranted a

separate designation from the Asian Tropics).

Estimates for tree diversity per biome area are

admittedly speculative (see Table 3.1 for sources),

especially for tropical forests. Significance of the

relationships between biome area and species

richness was tested by pairwise correlation of the

log-transformed variables.

ESTIMATING HISTORIES OF BIOME

AREAS

T esting th eGAH whil etaking into account

past fluctuations in biome area requires some

knowledge of the tempo of plant diversification.

Magallón and Sanderson (2001) estimated speci-

ation rates for angiosperms and proposed an aver-

age overall rate of 0.0893 net speciation events

per million years (maximum rate of 0.32 events

per million years in the most rapidly diversifying

clade, Asteraceae). From these values, we deduced

that th er ec ent history of biom esiz eon th eord er of

tens of millions of years was a reasonable window

over which fluctuations in biome area could be

expected to have an effect on extant diversity

levels.

To estimat eth esiz eof biom ear eas through

time, we used paleoclimatic and paleovegetation

maps that estimated lowland moist/wet trop-

ical, temperate, and boreal biomes from five

recent sources (Dowsett et al. 1999, Morley

2000, Beerling and Woodward 2001, Willis and

McElwain 2002, and C.R. Scotese’s PALEOMAP

project [www.scotese.com]; see also Parrishet al.

1982, Scot es e2004) (Figur e3.1). Eoc en e,

Oligocene, and Miocene reconstructions were

largely drawn from reconstructions by Willis and

McElwain (2002), Pliocene reconstructions were

almost entirely based on Dowsettet al. (1999),

and mid-Holocene and Last Glacial Maximum

reconstructions were based on Beerling and