72 Rodolfo Dirzo and Karina Boege
contrast between deciduous versus slow-growth
species is highly significant: the median dam-
age of TDF deciduous species is about two-
fold greater than that of slow-growth TRF
species (Mann–Whitney’sZ=−3.88, d.f.=96,
P=0.0001). This result supports the expected
pattern: TDF species, with episodic leaf availabil-
ity to herbivores and presumably lower invest-
ment in defenses, are considerably more damaged
than slow-growing species of TRF with con-
tinuous foliage availability (see bold arrows in
Figure 5.1).
Itisconceivablethatthedifferencesinherbivory
between the plants of the two forest types could be
associated with a host of environmental variables
beyond those related to our hypothesis of water
availability/phenology. A more controlled test of
our prediction would be to compare levels of her-
bivory between deciduous and evergreen species
of dry forests (see discussion above). Some of the
data in Table 5.1 permit such comparison, par-
ticularly based on data from Mesoamerican TDF
(Costa Rica and Mexico). Such comparison yielded
highly significant differences (Figure 5.2, lower
panel, left): TDF deciduous species had a median
herbivory value 2.8 times as large as that of the
evergreen species (Mann–Whitney’sZ=−4.19,
d.f.=79,P<0.0001). These results are consis-
tent with, and slightly sharper than, those of the
comparisonbetweendeciduousTDFspeciesversus
slow-growthTRFspecies.Thus,ingeneral,thelev-
els of herbivory of the deciduous species of the dry
forest mirror the original expectation, while lev-
els of herbivory of the evergreen species resemble
those of the slow-growing plants of the less sea-
sonal TRF (see ascendin gdia gonal arrow for TDF
in Figures 5.1 and 5.2).
Separatin gslow- growin gversus fast- growin g
species in TRF provides results that are consistent
with previous findings (Coley 1983, Dirzo 1987)
and with the resource availability hypothesis
(Janzen 1974, Coleyet al.1985): species of rapid
growth sustained levels of herbivory 3.4 times
as large as those found in the slow-growing
species (Mann–Whitney’sZ=6.62, d.f.=68,
P<0.0001). Given such contrast, rain forest
species of rapid growth tend to resemble the
response of deciduous TDF species (see ascending
broken arrow in Figure 5.1), although their levels
of herbivory are significantly different (Mann–
Whitney’sZ=4.4, d.f.=94,P<0.0001).
Plant defense
Data were available to compare three attributes
related to plant defense: total phenolics, con-
densed tannins, and leaf toughness (Table 5.1,
Figure 5.3). However, data for total phenolics and
condensed tannins in the case of TDF plants,
although available for a large number of species
(64), correspond to a single though comprehen-
sive study in a single locality (Janzen and Water-
man 1984), while data for these two variables
for TRF plants are derived from eight independent
studies and a large number of species (152). (The
measurement of phenolic compounds in these
studies is largely based on the use of quebracho
standards; see Appelet al. 2001 for a discussion
on the limitations to evaluatin gphenolic com-
pounds usin gsuch standards.) Likewise, data for
leaf toughness are derived from three studies for
each of the two forest types, involvin g42 species
for TDF and 69 species for TRF (see Table 5.1).
Overall, considerin gall types of plants within
dry and rain forests, total phenolics concentra-
tion was only marginally greater in TRF plants
(P=0.06; Figure 5.3, upper panel, left). How-
ever, when we separate the species according
to their phenology in the case of TDF plants
and their growth rate strategy in the case of
TRF plants, significant differences are readily
observed (Kruskal–Wallisχ^2 =12.00, d.f.=3,
P=0.007; Figure 5.3, lower panel, left). The
comparisons show that slow-growing species of
TRF have a two times greater concentration of
total phenolics when compared with deciduous
species from TDF (Mann–Whitney’s Z=−3.2,
d.f.=119,P=0.001). Again, an extended com-
parison between evergreen and deciduous species
of the dry forest shows a consistent pattern: ever-
green species had 60% greater concentrations of
total phenolics, though differences are marginally
significant (P=0.06).
Condensed tannins had a similar concentra-
tion when considerin gthe overall group of
species from both forest types (Figure 5.3, upper
panel, center). Furthermore, a separation of the