136 Egbert G. Leigh, Jr
in dry forest canopy, especially in the season of leaf
flush (van Bael and Brawn 2005), in accord with
the prediction that, where predators are present,
insect abundance should not increase with plant
productivity.
How much do tropical forests depend on preda-
tors for protection from herbivores? Birds help
defend other tropical forests from insects. On Barro
Colorado Island, over a third of the foliage eaten by
insects that birds eat, and this appears true at other
sites as well (Leigh 1999, pp. 167–168). More
direct evidence comes from fragmentation of for-
est by reservoirs (whose impact is not confounded
with the effects of fire, cows, or invading pioneer
trees), which offers forest ecologists their near-
est equivalent to the marine ecologist’s exclusion
experiments (Leighet al. 2002). Fragmentation
causes extinctions, especially of predators, releas-
ing prey populations, with effects that cascade
through the fragments’ communities (Terborgh
et al. 2001). When the Guri reservoir fragmented
dry forest in Venezuela, leaf-cutter ants exploded
on newly isolated islets (Terborghet al. 1997).
Islets less than 1 ha carried up to six mature
leaf-cutter colonies ha−^1 , whereas islands greater
than 80 ha, with a full complement of mam-
mals, carried two colonies per 3 ha (Rao 2000).
As leaf-cutters on large islands have access to
more preferred trees (Terborghet al. 2006), they
can hardly be limited by seasonal shortage of
food: th ey must b econtroll ed by som epr edator
or pathogen (Terborghet al. 2001). Army ants
and armadillos, now extinct from small islets,
limit recruitment (Rao 2000) and perhaps sur-
vival (Swartz 1998), of leaf-cutter ants on large
islands and th emainland. Th e explosion of l eaf-
cutters on small islets severely reduced recruit-
ment of trees, especially canopy trees, and favored
well-defended species (Terborghetal. 2001, 2006,
Raoetal. 2002).
In sum, this crude theory suggests that employ-
ing animals as pollinators and seed dispersers,
which allows plants to escape specialist pests
through being rare, could enable a more diverse
and productive flowering forest to replace a less
productive, less diverse, better-defended, wind-
pollinated gymnosperm forest. Extending this
theory shows how the higher productivity of flow-
ering forest enables predators to assume a greater
role in its anti-herbivore defense. This theory
brings hom eth el esson that w ecannot und erstand
tropical for est if w eignor eanimals.CONCLUSIONS
In sum, mathematical theory related to forest
structur eand production has accomplish ed vari-
ous useful “odd jobs.” Simple theory suggests how
soil quality affects forest dynamics, the hydraulic
architecture and wood density of its trees, char-
acteristics of their leaves, and the apportionment
of resources between above- and below-ground
activities, without altering forest production very
much. Very crude theory shows how a tree
species’s physiology and life history are related
to its response to the trade-off between growing
fast in bright light and surviving in shade. The-
ory also shows how forest productivity is reduced
by how trees compete for light and nutrients.
Despite the efforts of Enquist (2002), however, a
comprehensive theory of tree shapes and forest
structure and production seems far off.
Mathematical theory in ecology has largely
been concerned with the maintenance of species
diversity (MacArthur 1972). How species arise
has attracted much less interest, even though the
process of speciation might reveal much about
how species coexist.
Testing a null hypothesis derived from Hubbell’s
neutral theory of forest ecology suggests that
w ecannot und erstand why th er ear eso many
kinds of tropical trees without knowing how
the differences among tree species allow them to
coexist.
Ar etr e es so much mor ediv ers ein th etropics,
where no winter depresses populations of pests
and pathogens, because each species there is kept
rare enough by specialist pests and pathogens to
make room for many others (Janzen 1970, Con-
nell 1971)? This hypothesis is plausible. Crude
theory, and some evidence, suggests that the more
diverse a forest, the less its species suffer from spe-
cialist pests and the faster they can grow. Testing
the pest pressure hypothesis of tree diversity is cru-
cial to understanding how tree diversity affects
tropical forest productivity, but it has been diffi-
cult to put to a decisive test. Because a tree’s fate