318 Jo hn Terborg hand Kennet hFeeley
influence the presence and/or abundance of non-
volant vertebrates in fragments (Malcolm 1994,
Gasconet al. 1999, Lauranceet al. 2002, For-
manet al. 2003). Thus, comparisons of results
from open systems with Lago Guri are difficult.
But some general principles do emerge.The loss of
top predators predictably releases herbivores and
mesopredators (Soulé and Terborgh 1999). These
groups then impose top-down pressure on plants,
smaller vertebrates, and arthropods. Accelerated
nutrient cycling may enhance productivity, but
excessive browsing by herbivores leads to struc-
tural as well as compositional changes in the
vegetation, while hyperdense mesopredators exert
negative pressures on birds, lizards, amphibians,
snakes, and doubtless some arthropods (Crooks
and Soulé 1999, McShea and Rappole 2000).
Eventually, herbivore pressure will result in a
more herbivore-resistant vegetation that in turn
will suppress herbivore numbers from the bot-
tom u p(Pastor and Naiman 1992, Pastoret al.
1993).
The results from Lago Guri open a window
on the operation of trophic cascades in terres-
trial systems by demonstrating that a perturbation
at one level (predator removal) can propagate to
the producer level via multiple pathways. These
pathways are both direct, such as the effects of
altered densities of pollinators, seed dispersers,
and herbivores on plant recruitment, and indirect,
such as the effects of hyperabundant herbivores
on nutrient cycling, tree growth, and bird diver-
sity (Davidsonet al. 1985). Contemporary theory
(e.g., Oksanen and Oksanen 2000) is primarily
concerned with direct effects exerted through her-
bivory. Our results from Lago Guri reveal a much
deeper reality that finds parallels in the intri-
cate interaction webs of the marine realm (Yodzis
2001). In fact, the very complexity of the Guri
trophic cascade, with its multiplicity of pathways,
implies that plant composition is established and
maintained by a balance of processes represent-
ing the numerous interaction links, both direct
and indirect, between plants and animals. From
a scientific standpoint, these new insights are
both exciting and challenging, but the conser-
vation implications are unsettling. The existence
of multiple pathways, operating with various
interaction strengths and strongly sensitive to
spatial scale as well as the character of the matrix
and the encompassing landscape, implies that
almost any change in the structure/composition
of an animal community may lead to instability in
the composition of the plant community. Changes
in plant community composition, in turn, will
inevitably feed back to the animal community
with consequences we are as yet unable to predict.ACKNOWLEDGMENTS
We thank EDELCA for allowing us to conduct
the research in the Guri impoundment and espe-
cially Luis Balbas for many kinds of advice and
support. The work described here is the result
of a long-term collaborative effort involving the
work of many students and researchers, all of
whom we gratefully acknowledge. K. Glander is
thanked for data on the body mass of howler
monkeys. We are grateful to Walter Carson and
two anonymous reviewers for many helpful sug-
gestions on the manuscript. Financial support
was provided by the John T. and Katherine D.
MacArthur Foundation and the National Science
Foundation (DEB97-09281, DEB01-08107).REFERENCES
Adler, G.H. (1996) The island syndrome in isolated pop-
ulations of a tropical forest rodent.Oecologia108,
694–700.
Aizen, M.A. and Feinsinger, P. (1994) Forest fragmenta-
tion, pollination, and plant reproduction in a Chaco
dry forest, Argentina.Ecology75, 330–351.
Alvarez, E., Balbas, L., Massa, I., and Pacheco, J. (1986)
Aspectos ecológicas del embalse Guri.Interciencia11,
325–333.
Anjos, L.D. and Bocon, R. (1999) Bird communities
in natural forest patches in southern Brazil.Wilson
Bulletin111, 397–414.
Aponte, C., Barreto, G.R., and Terborgh, J. (2003)
Consequences of habitat fragmentation on age struc-
ture and life history in a tortoise population.Biotropica
35, 550–555.
Asquith, N.M. and Mejia-Chang, M. (2005) Mammals,
edge effects, and the loss of tropical forest diversity.
Ecology86, 379–390.