Tropical Forest Community Ecology

22 Jérôme Chave

0.01 0.1 1 10 100 1000 10 000

0

0.01

0.02

0.03

0.04

0.05

0.01 0.1 1 10 100 1000 10 000

F(

r)

Mean from pairs of single

hectares

Within 50 ha plot

Theory (s=40m)

Panama

0

0.01

0.02

0.03

0.04

0.05

Distance (km)

F(

r)

Mean from pairs of single

hectares

Within 25 ha plot

Theory (s=55m)

Ecuador

Figure 2.4 The probabilityFthat

randomly selected pairs of trees are the

same species, as a function of distancer,

on a semilogarithmic scale, in Panama

(top) and Ecuador (bottom), and a best fit

of the dispersal model to the data for

r>100 m. Modified from Conditet al.

(2002).

Amazonmaybeexplainedbydifferencesinhabitat
heterogeneity. They quantified the amount of
floristic variation attributable to space and envi-
ronmental variation in the Panama dataset and
found that most of the floristic turnover was
caused by the environment, in particular rainfall.
Ruokolainen andTuomisto (2002), then Legendre
et al. (2005) had noticed flaws in Duivenvoorden
et al.’s approach and we recently reanalyzed the
Panama dataset usin gboth ordination and dis-
tance matrix approaches (Chustet al. 2006). We
found that environment alone (i.e., rainfall, topog-
raphy, and soil properties) explained 10–12% of
the floristic variation, space alone (logarithmically
transformed geographical distance) 22–27%, and
the interaction between the two 13–18%, depend-
in gon the statistical method we employed. The

unexplained fraction varied between 46 and 49%.

Phillipset al. (2003) performed a similar variation

partitionin gapproach for a network of 88 perma-

nent plots around the city of Puerto Maldonado,

southeast Peru (ca. 50 km east of Terborghet al.’s

study plots) and found that about 10% of the

tree floristic variation could be explained by space

and 40% by abiotic habitat conditions. A pre-

liminary comparison of the two studies therefore

shows that floristic similarity is better explained

by distance in Panama than in the western Ama-

zon, as suggested in Conditet al. (2002), even

when habitat differences are accounted for. This

conclusion is further supported by an indepen-

dent smaller-scale study in a Mexican tropical

forest by Balvaneraet al. (2002), who reported

that the fractions of deviance explained by either