70 Rodolfo Dirzo and Karina Boege
Testin gthese predictions is frau ght with compli-
cations due to the effect of a host of uncontrolled
variables, particularly the interspecific hetero-
geneity in the patterns of defense and herbivory
of plants within each forest type. In particular,
plants of the TRF can be classified into different
relative positions alon ga shade tolerance gradi-
ent: in the extremes of the gradient, they will
be shade tolerant and slow growing or shade
intolerant and fast growing. According to the
resource availability/growth hypothesis (Janzen
1974, Coleyet al. 1985), slow-growth plants
of the mature forest are expected to be bet-
ter defended than rapid-growth plants typical
of forest gaps, and empirical data support this
expectation (Coley 1982). Evidence shows that
the predominant growth strategy among plants
of this type of forest is that of slow growth,
shade tolerance, with relative proportions that
range between 89% (Martínez-Ramos 1994) and
95% (Weldenet al. 1991). Such a distinction is
not applicable to, or is of lower importance, for
TDF plants, where shadin gis not such a crit-
ical aspect of the environment (Mooneyet al.
1985).
On the other hand, in TDFs, two distinct
categories of phenological patterns can readily
be distinguished due to the presence of ripar-
ian habitats. Given their higher water table and
conservation of soil moisture, riparian habitats
are populated by evergreen species (Dirzo and
Domínguez 1995) which, in contrast to the decid-
uous species, retain their foliage throughout the
year. Evergreen species of the TDF, although a
minor fraction of the total flora (e.g., 1.1–9.7%
in a variety of sites from Mesoamerica; Lott and
Atkinson 2002), represent a significant contrast
in phenology that is likely to lead to differences in
patternsof herbivoryanddefense. Forthecompar-
isons we develop in this chapter we attempted to
control for such sources of interspecific variation
by lookin gat the levels of herbivory and defense
separatin gplants accordin gto these strate gies
(slow growth and fast growth in TRF; evergreen
and deciduous in TDF), in addition to the overall,
total-species comparisons (see Figure 5.1).
In addition, we cannot rule out the possibil-
ity that variables other than contrast in water
availability and seasonality will differ between
TDF and TRF, includin gdifferent herbivore
communities, historical factors, etc. Ideally, the
relative effect of these additional, potentially con-
founding, factors should be controlled. A limited,
but useful, mimic of such control can be pro-
vided by the riparian habitats of TDFs. Riparian
habitats intermingled in the predominant vege-
tation of TDF allow for an intra-site comparison
under the same climatic regime, potential herbi-
vore community, and several historical factors to
test our water availability/phenology hypothesis.
Such comparison may have the caveat that ripar-
ian communities represent relatively small habitat
islands for insects. However, they can maintain
residentpopulationsof someinsectspecies,and/or
they can harbor insects of other (non-dormant)
species that migrate to them during the dry sea-
son (see Janzen 1973). Therefore, an additional,
convergent prediction is that the evergreen plant
species of the riparian habitats in TDFs should
have attributes that resemble those of the slow-
growth plants that predominate in TRFs. It fol-
lows, therefore, that in order to test our predictions
for TDF versus TRF plants, it is the deciduous
species of TDFs that should be compared with
the slow-growth species of TRF (bold arrows in
Figure 5.1). Finally, the rapid-growth plant species
of TRF, although not directly relevant for our
water availability/phenology hypothesis, are pre-
dicted to resemble more the responses exhibited
by the deciduous species of TDF (diagonal broken
arrow in Figure 5.1).THE INFORMATION BASE TO TEST
THE PREDICTIONS
We carried out an extensive search of all pub-
lished studies of herbivory or defense or both, in
tropical systems between 1970 and 2004, using
the Web of Science and BIOSIS. In addition, we
included data from our own unpublished stud-
ies on tropical herbivory in the Neotropics, but
particularly in Mexico, for both tropical rain and
dry forests. Specific information on study site
locations, their seasonality, measurements, and
sources of data are given in Table 5.1. The infor-
mation compiled in this table provides sources
of data for the reader interested in this topic,