11. Patterns ofTree Species Diversity
167
levels of species diversity that exist. Wills et al. (1997) using a different method of
analysis but the same data set as Hubbell (1997a), has shown higher density dependence. It should be expected that, the more equable the environment, the greater the
hegemony of biotically over abiotically mediated selection. For species in similar
population densities, biotically induced density-dependent mortality should increase
in intensity with decline in rainfall seasonality. Yet Wills and Condit (in press)
found that density dependent mortality is substantially lower in the aseasonal climate at Pasoh, Malaysia, than at Barro Colorado Island, Panama with its three
month dry season. Could it be that drought-related, or generalized browsing-induced juvenile tree mortality is density dependent and stronger in seasonal climates? Evidence for this is so far lacking.
The CTFS tree demography plots are large enough to describe the canonical
distribution of species abundances (Preston 1962). They indicate that the differing
species richness of tropical forests along the rainfall seasonality gradient is not so
attributable to differing numbers of rare species (e.g., Ashton 1984, 1998b) as to
differing numbers in the peak abundance classes (Fig. 1). The number of very abundant species also does not greatly vary. There is a shift in the peak of species abundances caused by the differences in stand density. Preston predicted that the canonical distribution of species abundances should be bell-shaped. The number of
species in low population densities in samples from less seasonal climates is greater
than expected; but although the number of rare species hardly changes with increased seasonality within the evergreen forests, that number increasingly conforms
to Preston's prediction as the peak of the abundance curve shifts to the left. These
results imply that both density dependence and factors influencing population extinction such as Allee effects differ little in intensity between these forests; and that
the key to the maintenance of species richness must be sought among the species
majority, those of moderate population density. In forests of the aseasonal tropics
but of differing richness, however, numbers of rare species and those of peak abundance covary (Fig. 2).
In a regenerating temperate woodland, we have observed an enormous diversity of patterns of tree population mortality, ranging from apparently random but
continuing mortality of the powerfully defended Rhus toxidodendron L by a pathogen, to the clustered but periodic mortality among mature trees of several genera
occasioned by gypsy moth (Porthetria dispar) infestation, neither of which patterns
would easily be identified by current methods of monitoring. The case of Shorea
albida Sym., a dipterocarp which is the sole species in the closed canopy through
thousands of square kilometers of peat swamp forests in northwest Borneo, should
be a warning. Sharply circumscribed areas as great as 10 km 2 were defoliated and
killed during the sixties by an unidentified hairy caterpillar, but attacks later ceased.
These forests may have survived as pure stands for several thousand years.
Density dependence may also be induced by limitations on fecundity resulting
from low carrying capacity of species-specific dispersal agents (Ashton 1998b).
Most tree species in rain forests appear to have polylectic dispersal syndromes, but
there are exceptions. An interesting example is Durio, Bombacaceae, which con-
167
levels of species diversity that exist. Wills et al. (1997) using a different method of
analysis but the same data set as Hubbell (1997a), has shown higher density dependence. It should be expected that, the more equable the environment, the greater the
hegemony of biotically over abiotically mediated selection. For species in similar
population densities, biotically induced density-dependent mortality should increase
in intensity with decline in rainfall seasonality. Yet Wills and Condit (in press)
found that density dependent mortality is substantially lower in the aseasonal climate at Pasoh, Malaysia, than at Barro Colorado Island, Panama with its three
month dry season. Could it be that drought-related, or generalized browsing-induced juvenile tree mortality is density dependent and stronger in seasonal climates? Evidence for this is so far lacking.
The CTFS tree demography plots are large enough to describe the canonical
distribution of species abundances (Preston 1962). They indicate that the differing
species richness of tropical forests along the rainfall seasonality gradient is not so
attributable to differing numbers of rare species (e.g., Ashton 1984, 1998b) as to
differing numbers in the peak abundance classes (Fig. 1). The number of very abundant species also does not greatly vary. There is a shift in the peak of species abundances caused by the differences in stand density. Preston predicted that the canonical distribution of species abundances should be bell-shaped. The number of
species in low population densities in samples from less seasonal climates is greater
than expected; but although the number of rare species hardly changes with increased seasonality within the evergreen forests, that number increasingly conforms
to Preston's prediction as the peak of the abundance curve shifts to the left. These
results imply that both density dependence and factors influencing population extinction such as Allee effects differ little in intensity between these forests; and that
the key to the maintenance of species richness must be sought among the species
majority, those of moderate population density. In forests of the aseasonal tropics
but of differing richness, however, numbers of rare species and those of peak abundance covary (Fig. 2).
In a regenerating temperate woodland, we have observed an enormous diversity of patterns of tree population mortality, ranging from apparently random but
continuing mortality of the powerfully defended Rhus toxidodendron L by a pathogen, to the clustered but periodic mortality among mature trees of several genera
occasioned by gypsy moth (Porthetria dispar) infestation, neither of which patterns
would easily be identified by current methods of monitoring. The case of Shorea
albida Sym., a dipterocarp which is the sole species in the closed canopy through
thousands of square kilometers of peat swamp forests in northwest Borneo, should
be a warning. Sharply circumscribed areas as great as 10 km 2 were defoliated and
killed during the sixties by an unidentified hairy caterpillar, but attacks later ceased.
These forests may have survived as pure stands for several thousand years.
Density dependence may also be induced by limitations on fecundity resulting
from low carrying capacity of species-specific dispersal agents (Ashton 1998b).
Most tree species in rain forests appear to have polylectic dispersal syndromes, but
there are exceptions. An interesting example is Durio, Bombacaceae, which con-