high disturbance versus low disturbance, dry versus
wet, and lowland versus upland habitats. The A. muel-
leri complex prefers Cecropia trees in low-light, low-
disturbance habitats such as small light gaps in ma-
ture forest. The A. alfari complex prefers Cecropia
trees in high-light, high-disturbance habitats such as
roadsides, abandoned pastures, and river margins.
Azteca coeruleipennis prefers trees growing in high-
light conditions in dry forest. Cecropia trees in cloud
forests are nonmyrmecophytic (Janzen 1973).
Complex community structure in the Cecropia-
Azteca system was first described for the Monteverde
area (Longino 1989a). A 20-km transect crossing the
Continental Divide crosses four distinct assemblages
of Cecropia and Azteca. Populations of C. peltata are
found on the dry Pacific slope below 1000 m and
are inhabited by colonies of A. coeruleipennis and
A. alfari complex. Cecropia obtusifolia occurs in
a band from 1000 to 1400 m and is inhabited by
A. xanthochroa and A. constructor. Cecropia poly-
phlebia occurs from 1400 m through the cloud for-
est to 1100 m on the Atlantic slope and is nonmyrme-
cophytic (Janzen 1973). Cecropia insignis occurs at
1100 m and below on the Atlantic slope and is inhab-
ited by A. xanthochroa and A. constructor.
The nonmyrmecophytic C. polyphlebia is possibly
derived from myrmecophytic ancestors (Janzen 1973).
In Monteverde, it is not sharply differentiated from
the C. obtusifolia just below it (J. Longino, pers. obs.).
Trees in a contact zone between 1400 and 1500 m
elevation show a range of phenotypes intermediate
between the typical forms of each species. In addition
to intermediate forms of foliage and fruit characters,
the pads where Miillerian bodies are produced gradu-
ally become less distinct and increasingly covered
with long hairs. Colonies of Azteca in mature trees
gradually drop out in this transition zone. Queens of
A. xanthochroa and A. constructor attempt to colo-
nize C. polyphlebia at higher elevations (queens can
be found in the saplings well up into the cloud for-
est) but never successfully establish colonies.
When myrmecophytic-ant interactions are consid-
ered as spatially variable communities of interacting
species, basic ecological questions of distribution and
coexistence can be asked. What mechanism drives the
habitat segregation between the A. alfari and A. muel-
leri complexes? Alternative hypotheses based on light
availability and disturbance regime have been pro-
posed (Davidson and McKey 1993b). The light avail-
ability hypothesis (Davidson and Fisher 1991) pro-
poses that A. alfari complex species outcompete A.
muelleri complex species in saplings but also require
the high rates of resource input that a high-light en-
vironment provides. Azteca muelleri complex species
may be able to survive with lower resource input rates
and can thus occur in habitats where A. alfari com-
plex colonies cannot. The disturbance hypothesis
(Longino 1991b) proposes that A. alfari complex spe-
cies are more abundant in frequently disturbed habi-
tats because they can reproduce early in small trees.
Azteca muelleri complex colonies would be at a dis-
advantage if they required large, long-lived trees to
achieve a colony size sufficient for reproduction. In
low-disturbance habitats, trees with A. alfari complex
colonies may soon succumb due to the poor defen-
sive behavior of the ants. In contrast, trees with A.
muelleri complex colonies may flourish and reach a
large size, and the ant colonies produce large num-
bers of new colonizing queens over a long time span.
These hypotheses are not mutually exclusive, and
both processes could co-occur.
Monteverde is an excellent site to investigate these
hypotheses. Compressed productivity gradients are
caused by light availability, temperature (elevation),
and water (wet-dry transition). Common garden ex-
periments could be carried out with multiply colo-
nized saplings planted in different habitats to exam-
ine predictions of queen survivorship and competitive
ability as a function of productivity. The disturbance
hypothesis could be examined by demographic stud-
ies of tree survivorship, colony growth, and colony
reproduction as a function of ant species. Document-
ing the range of phenotypes between C. obtusifolia
and C. polyphlebia would allow quantification of the
myrmecophytic traits (Miillerian body production,
reduction of stem pith, development of thin spots in
the stem wall) necessary for ant colony survival and
how it varies among ant species and along productiv-
ity gradients. No hypotheses have been advanced to
explain the coexistence of A. xanthochroa and A.
constructor, two relatively similar species that are
both common in C. obtusifolia trees in Monteverde.
These questions await further research on the rich
myrmecophyte community of the Monteverde area.293 Plant-Animal Interactions