Tropical Forest Community Ecology

Patterns of Herbivory and Defense in Tropical Forests 71

but, for the purpose of the comparisons of this
chapter, we filtered the information as follows.
We selected only those studies from which specific
data regarding plant phenology (i.e., evergreen or
deciduous in TDF) and adaptation to growth in
specific microhabitats (i.e., slow in shaded under-
story or rapid in forest gaps of TRF) could be
obtained to separate species accordingly. In addi-
tion, regarding more practical aspects, given that
the predominant information available in the lit-
erature consists of standin glevels of herbivory,
we restricted our analyses to studies reporting
herbivory as a discrete measure of accumulated
herbivore attack throughout the lifespan of leaves
(see discussions in Lowman 1992 and Filipet al.
1995). This meant that unfortunately we were
not able to use some important studies based
on herbivory rates for comparisons of herbivory,
although for some of those studies we used
other data presented therein (e.g., plant defense).
Regarding defensive characteristics (total pheno-
lics, tannins, and toughness), our comparisons
are restricted to data correspondin gto mature
leaves. The most detailed study available for this
aspect (Coley 1983) clearly shows that inter-
specific variation can be better predicted by the
analysis of mature leaves. An important insight
regarding variation in herbivory is the hetero-
geneityassociatedwithplantontogeny(Boegeand
Marquis 2006) and it would have been impor-
tant to explore to what extent the data based on
juvenile/mature plants (as we have done here)
are consistent or not with the data correspond-
in gto seedlin gs. Unfortunately, data on seedlin g
herbivory were available only for TRF plants,
includin ga substantial number of species (de la
Cruz and Dirzo 1987, see also Table 5.1), but we
found no equivalent or even partial information
for TDF species. This is an aspect that warrants
further assessment. Finally, our comparisons are
based on data correspondin gto each individual
species, instead of averaging values for a given
study (i.e., site) when it involved several taxa.
In those few cases in which a given species was
studied several times in a single site, we aver-
aged the values. Table 5.1 provides all sources of
information used in this study; unpublished data
reported therein are available from the authors
upon request.

DOE SEMPIRICAL EVIDENCE

MATCH PREDICTIONS?

Herbivory

A first approach was to compare overall TDF

species versus TRF species (Figure 5.2, upper

panel). This comparison shows that the two

groups of plants are statistically indistinguish-

able (Mann–Whitney’sZ=1.69, d.f.=149,

P = 0.09), not supportin gour predicted out-

come. However, when we separated the TRF

and TDF data into evergreen and deciduous

species of TDF on the one hand and slow-

growth and rapid-growth species from TRF

on the other (Figure 5.2, lower panel), the

Leaf area eaten (%)

0

10

20

30

40

0

10

20

30

Evergreen

Dry forests Rain forests

**** ****

****

Deciduous Slow growth Rapid growth

Figure 5.2 Levels of herbivory for TDF species (open

boxes) and TRF species (shaded boxes), comparin gthe

total conglomerate of species of each forest type (upper

panels) and the species separated accordin gto their

phenology in TDFs (lower panel, left) and growth habit

in TRFs (lower panel, right). Data represent medians,

quartiles (25–75% and 5–95%), and extreme values.

∗∗∗∗P<0.0001.