Ecosystem Decay in Closed Forest Fragments 315herbivore pressure has already resulted in sig-
nificant shifts in tree community composition
on S islands, such that there is now a signifi-
cant positive relationship between the density of
herbivores and the relative abundance of “unpre-
ferred” tree species (r^2 = 0.24, P < 0.05;
Rao et al. 2001, Feeley and Terborgh 2005,
Orihuela-Lopezet al. 2005). These unpreferred
tree species tend to have nutrient-poor/highly
defended leaves and as a result their increased
predominance has led to a decrease in leaf litter
quality (r^2 =0.23,P <0.05). Consequently,
soil nutrient availability is now decreasing on S
islands supporting dense herbivore populations
(for example, there is a significant positive rela-
tionshi pbetween C:N [high C:N indicates low
soil fertility] and the density of howler monkeys,
r^2 =0.28–0.50,P<0.05; Feeley and Terborgh
2005). This decrease in soil nutrient availabil-
ity will likely lead to slower growth rates, fur-
ther stressing these declining forests (Pastor and
Naiman 1992, Pastoret al. 1993).
Unlike howler monkeys, leaf-cutter ants
sequester nutrients underground in their fungal
gardens and refuse chambers. The deepest cham-
bers in mature leaf-cutter colonies can be 5 m
below the surface at a level reached by few roots.
During the rainy season, when the water table
rises, the nutrients concentrated in underground
chambers can be leachedaway,leading one inves-
tigator studying nutrient fluxes in a Venezuelan
rainforest to conclude that “leaf-cutting ants bleed
mineral elements out of a rainforest” (Haines
1983). This may explain why tree growth was
slower on S islands supporting hyperabundant
leaf-cutter ants but lacking howler monkeys than
on large landmasses supporting much lower den-
sities of leaf-cutter ants.
The presence of hyperabundant herbivores can
carry important implications for other major
faunal groups, such as birds. For example,
the density and diversity of birds on small
islands is positively correlated with the density
of howlers (r^2 = 0.59,P < 0.001; Feeley
and Terborgh 2006). In contrast, islands lack-
ing howlers have experienced accelerated rates
of avian extinctions during 10 years of obser-
vation (r^2 = 0.78,P < 0.005), perhaps in
part as an indirect consequence of nutrient
sequestration by hyperabundant leaf-cutter ants
(Feeley 2005a).
Evidence for such bottom-u peffects on bird
communities was limited to S islands. On some
M and L islands, hyperabundant mesopredators
(capuchin monkeys) imposed a negative top-down
effect on bird communities via nest predation
(Terborghet al. 1997a). Olive capuchins on the
Venezuelan mainland utilize a 200 ha home range
(Robinson 1986), but groups of this species per-
sisted on some Lago Guri islands of less than
15 ha. The greatly reduced home ranges of these
trapped capuchins are likely to be searched at
a much higher frequency than the vastly larger
home ranges of unrestricted groups. Accordingly,
breeding bird densities on M islands harboring
capuchins were markedly less than those on
similar-sized islands lacking capuchins and in
some cases were depressed to such a degree that
total numbers of breeding birds were less than
on small predator-free islands despite the order
of magnitude difference in area (Terborghet al.
1997b). In another situation, capuchin densities
approximately doubled on the 190 ha L island
between 1993 and 2003, during which time the
island lost 57% of its avian species (Feeley 2005a).
The disappearance of birds on M and L islands
due to mesopredator release may eventually have
important implications for the vegetation, poten-
tially causing altered rates of pollination, seed
predation, seed dispersal, or even insect herbivory
(Marquis and Whelan 1994, Van Bael et al.
2003).Bottom-up forces prevail in the absence
of top-down regulationCurrent ecological theory predicts that con-
sumers will increase in the absence of top-down
regulation, but does not predict the amount of
increase (Oksanen and Oksanen 2000). One of
the most unexpected findings of our research was
that some species increased by many-fold (up to
100 times in the case of leaf-cutter ants), imply-
ing that the carrying capacity of the Guri forest
for primary consumers is potentially much greater
than suggested by animal densities on the nearby
mainland.