Tropical Forest Community Ecology

Symmetric Neutral Theory 155

To test the predictions of neutral theory and

a drifting community, we first had to consider

whether to use the dispersal limitation version of

the theory, or the version with symmetric den-

sity dependence. Recall that on the basis of the

fit to static relative abundance data, these two

versions of the theory cannot be distinguished.

However, on the basis of the dynamics data, there

is a clear winner. Recall from Figure 9.1 that

under symmetric density dependence, we expect

that the ratio of per capita birth rate to per capita

decay rate should exceed unity at small popula-

tion sizes, conferring a growth rate advantage on

species when rare. However, there is no evidence

100

50

0

− 2 − 10

Intrinsic rate of increase, 1982–2005

No. of species

12

Figure 9.4 Distribution of intrinsic rates of increase
among BCI tree species over a 23-year period from
1982 to 2005.

of the necessary rare-species advantage at the

whole plot level (Figure 9.5). Rare and common

species alike have mean values ofb/dthat do not

differ significantly from unity. Thus, we modeled

a drifting neutral community under the original

dispersal limitation mechanism (Hubbell 2001)

and island biogeography theory (MacArthur and

Wilson 1967).

Dispersal limitation and symmetric density

dependence are not mutually exclusive mecha-

nisms, so in principle both can operate simulta-

neously. In fact, we know that density dependence

is pervasive and strong in the BCI tree community

(Hubbell 2008b,c). Harmset al.(2000) demon-

strated very strong density dependence in the

seed-to-seedling transition, as measured by the dif-

ference between the number of seeds collected in a

network of traps and the number of seedlings that

germinate in seedling plots adjacent to the traps.

There are also negative conspecific density effects

on sapling growth and survival (Hubbellet al.

2001, Ahumadaet al. 2004, Uriarteet al.2005).

How do we reconcile these observations with the

results in Figure 9.5? The answer is that all of

these density effects weaken to background within

short distances (most<20 m, to 30–40 m in a few

species; Hubbellet al.2001, Hubbell 2008b) and,

as a result, they do not regulate the populations of

BCI tree species at the scale of 50 ha.

3.0

2.5

2.0

Per capita

b/

d, 1982–20051.5

1.0

0.5

0

1 − 3 − 10 − 30 − 100 − 300 −

Species abundance in 1982

1000 − 3000 −10,000−

Mean ± 1 SD

Figure 9.5 Lack of evidence of density- and frequency dependence in the BCI tree community over the entire

23 years of the study. The mean per capita birth rate/death rate (b/d) ratio±1 standard deviation for species binned

into half log base 10 intervals of abundance. In all abundance categories the confidence limits forb/dbracket unity.