Tropical Forest Ecology: Sterile or Virgin for Theoreticians? 131DiversityIn some everwet equatorial forests, a 25 ha square
contains over 800 species of tree≥10 cm dbh; a
similar square of seasonal tropical forest has about
200 species, far more than in a similar area of
temperate-zone forest (Table 8.3). Why are there
so many kinds of tropical tree?Testing Hubbell’s neutral theoryDo all these species of tropical tree differ in
ways that allow them to coexist? To answer, first
consider how fast a tree species can spread in
a neutral world (Hubbell 2001) where it makes
no difference what species a tree belongs to. Let
time be measured in tree generations: if trees have
annual death ratem, thentyears corresponds
toT=mttree generations. Consider a neutral
species that begins withnmature trees at time 0.
Let each tree alive at timethav eprobabilitymdt
of dying by timet+dtand equal probability of
producing a seed by then that instantly becomes a
mature tree, independently of the fates of all other
trees of its species. Using the methods of branch-
ing processes, this neutral theory predicts that if
the species still survivesTntree generations
later, the probability that this species has overkT
reproductive trees then ise−k(Leigh 2007; see
also Fisher 1930, p. 80).
Some tree species have spread much faster than
chanc ewould allow. About 20 million y ears ago,
the tree speciesSymphonia globulifera(Guttiferae)
first appeared in the Neotropics, after dispers-
ing across th eAtlantic from Africa (Dicket al.
2003). If the death rate of this tree species
is 2% year−^1 , its descendants crossed 400,000
tree generations ago. Now this species averages
about two trees≥10 cm dbh ha−^1 all through
Amazonia, and it has spread into Central America.
This species must have over 10 million reproduc-
tiv eadults in th eN eotropics (L eighet al. 2004).
An initially rare neutral species lucky enough
to surviv eso long would hav eprobabilitye−^25 ,
less than 10−^10 , of having so many reproduc-
tive trees after 400,000 tree generations. Similarly,
th eg enusOcotea(Lauraceae) first appeared in
South America about 20 million years ago, after
dispersing across the sea from North America
(Chanderbaliet al. 2001, p. 139). NowOcotea
averages about five trees≥10 cm dbh ha−^1 all
through Amazonia, representing hundreds of
species: this clade, too, has spread far faster than
chance would allow. Many other tree species
have multiplied extensively in tropical forests after
dispersing across oceans (Pennington and Dick
2004).
Although a d ecisiv eadvantag espr eadSym-
phoniaandOcoteathroughout Amazonia, they
form only a small minority of Amazonia’s trees.
Other tree species must differ enough from
these invaders to avoid competitive displacement
(Leighet al. 2004). These species coexist because
no one species can do all things well (MacArthur
1961). Trees, like other organisms, face trade-offs:
enhancing one ability usually entails sacrifices
in others.
Another way to show that tree diversity reflects
differences that allow different species to coex-
ist is to show that natural selection driven by
trade-offs causes tree speciation. Tree speciation
is usually allopatric (Coyn eand Orr 2004, L eigh
et al. 2004). Nonetheless, if a tree species is
divided into two completely isolated populations,
millions of years may elapse before they become
mutually intersterile (Ehrendorfer 1982). Closely
related species isolated by having different pol-
linators or flowering times are often completely
fertile when crossed artificially (Gentry 1989,
Kay and Sch emsk e2003). In on esympatric pair
of larg eh erbs,Costus, plants of each species
will not accept pollen from a nearby plant of
the other, but can be fertilized by pollen from
plants of th eoth er 400 kmaway (Kay2002),
as if reproductive isolation were favored by selec-
tion (presumably driven by reduced fitness of
hybrids for either parent’s way of life). Selection
against hybrids arises if a peripheral population
occupies a habitat requiring adaptations unsuit-
abl efor th epar ental habitat (St ebbins 1982).
Gillett (1962) suggested that rainforest trees
had congeners in savanna, a habitat demand-
ing a very different physiology (Hoffmannet al.
2004), because novel anti-herbivore defenses
allow invasions of other, already occupied, habi-
tats. Tree speciation associated with transitions
from wet forest to dry forest or savanna (when
dry habitats w er e expanding at th eb eginning