Tropical Forest Community Ecology

Testing Explanations for Species Diversity using FDPs 103

any other; its sixth census in 2005 marked a total
of 25 years of investigation.

Historical perspective

Although the BCI FDP, established in 1980,
is often considered to be the first large FDP
(Table 7.1; Losos and Leigh 2004), the first large
area (13.4 ha) of tropical forest was censused
by Hubbell (1979) in tropical dry forest in Costa
Rica. Originally intended for an investigation into
the diet and leaf selection of leaf-cutting ants,
the tree ma pof this 13.4 ha plot was used by
Hubbell (1979) to test the hypothesis that adult
tropical trees have relatively uniform dispersion
as predicted by the Janzen–Connell hypothesis.
Hubbell (1979) also introduced a neutral model
to explain the relative abundances of tree species
in this Costa Rican plot. From this experience
Hubbell recognized the power of large plots to
investigate the patterns of relative abundance and
diversity in tropical forests (Hubbell 2004), and
with his collaborator Robin Foster sought a site
for a permanent plot that could be resampled at
regular intervals. They chose BCI because of the
logistical support available, its well-known flora
(Croat 1978), the relatively flat topography of the
island summit (formerly a hill, that was isolated
by the waters of Lake Gatun during the develop-
ment of the Panama Canal), and the low number
of poisonous snakes (S.P. Hubbell personal com-
munication). It was difficult to obtain funding
to establish the first large FDP (Hubbell 2004),
in large part because many biologists, including
those familiar with the characteristics of tropical
forest, were skeptical of the usefulness of such
a large plot. As the value of the rich detail of
data on species’ relative abundance, diversity, and
forest dynamics provided by the BCI FDP was rec-
ognized (Hubbell and Foster 1986a,b, 1988), the
need for comparative data elsewhere in the tropics
became the motivating factor for the development
of other plots, which began with the Pasoh plot
in Malaysia (Table 7.1). As more plots were estab-
lished, the CTFS network was founded to maintain
consistent methodologies and foster collabora-
tion among researchers from the participating
plots.

Tropical forest community structure

revealed by FDPs

The first census of the FDP on BCI, Panama

revealed 305 species among 235,000 trees and

saplings≥1 cm diameter at breast height (dbh)

(169 species per ha; Hubbell and Foster 1986a).

Tree diversity in the BCI plot is relatively low

when compared with some of the other FDPs

(Table 7.1). The first census of the Pasoh plot

in Peninsular Malaysia (50 ha, completed in

1988) found 814 species≥1 cm dbh (495 species

per ha; Manokaranet al. 2004), while the Lambir

plot, in Sarawak, Malaysia (established in 1990),

had even more species, with 1182 in 52 ha

(618 species per ha; Leeet al. 2004). A relatively

small plot of only 25 ha inYasuní, Ecuador (estab-

lished in 1995) had 1104 species (655 per ha;

Valenciaet al. 2004a): this is the most diverse in

the CTFS FDP network.

Inadditiontototalspeciesrichness,dominance–

diversity curves (Figure 7.1) are another way

to compare tropical diversity among FDPs.

Dominance–diversity curves plot the proportional

abundance (on a log 10 scale) versus the rank

order abundance of an FDP’s tree species, and the

curve nearly always has a long tail. Dominance–

diversity curves demonstrate that most FDPs con-

tain few common and many rare tree species.

For example, the first census of the 50 ha BCI

plot revealed that 22 species were represented by

a single individual (Hubbell and Foster 1986a),

while just under 50% of species were represented

by 99 or fewer individuals – in a total sample of

235,000 trees! In the Pasoh FDP, the equivalent

ratio is 367 of 815 species (45%). If one consid-

ers that there are only about 1000 species of tree

north of the Mexican border in North America

(Hubbell and Foster 1986a), the magnitude of

tropical forest diversity becomes evident. The high

degree of species rarity across large areas of trop-

ical forests, and the need to determine the spatial

distribution and sufficient samples of these rare

species, is the single greatest justification for such

large plots.

In tropical forests growing on islands (Luquillo

and Sinharaja), where diversity is relatively low,

on former mainlands (BCI), and on true main-

lands (Pasoh), the dominance–diversity curves