The effects of plant quality on caterpillar growth and defense against
natural enemies
P. D. Coley, M. L. Bateman and T. A. Kursar
Coley, P. D., Bateman, M. L. and Kursar, T. A. 2006. The effects of plant quality on
caterpillar growth and defense against natural enemies. 
 Oikos 00: 00
00.
A survey of 85 species of Lepidoptera feeding on 40 hosts on Barro Colorado Island,
Panama showed that growth and defensive traits of caterpillars were correlated with the
nutritional and defensive traits of their hosts. Growth rates were faster on young than
mature leaves, reflecting the higher nitrogen and water content of the former. Growth
was also positively correlated with leaf expansion rate, partially because of higher
nitrogen and water contents of fast-expanding young leaves. Specialists grew faster than
generalists, but both responded positively to nutritional quality. There was no effect of
lepidopteran family on growth. In analyses where the effects of nitrogen and water were
removed, the residuals for growth rate were greater for young than for mature leaves
and were positively correlated with expansion rates of young leaves. This suggests that
traits other than nutrition were also important. As young, expanding leaves cannot use
toughness as a defense, one possible explanation for the differences in growth is
differences in chemical defenses. Growth rate residuals for both specialists and
generalists were higher for the more poorly defended fast-expanders, but the effect
was greatest for generalists, perhaps because generalists were more sensitive to
secondary metabolites. We predicted that slow growth for caterpillars would increase
their risk to natural enemies and would select for higher defenses. Generalists had more
defensive traits than specialists and were less preferred in feeding trials with ants.
Similarly, species feeding on mature leaves were the most defended and those feeding on
fast-expanding young leaves were the least defended and most preferred by ants. Thus
the effects of plant secondary metabolites and nutrients dictate herbivore growth rates,
which in turn influence their susceptibility to the third trophic level and the importance
of defenses.
P. D. Coley (coley@biology.utah.edu) and T. A. Kursar, Dept of Biology, Univ. of Utah,
Salt Lake City, UT 84112-0840, USA. 
 PDC, TAK and M. L. Bateman, Smithsonian
Tropical Research Institute, Balboa, Panama.
Herbivores face many challenges with respect to diet
selection as leaves have an array of chemical and physical
defenses (Rosenthal and Berenbaum 1991) and are very
low in protein relative to seeds or fruits (Mattson 1980).
The mechanisms by which herbivores address the
challenges of toxic secondary metabolites, high tough-
ness and low nutritional value will affect their growth,
and, in turn, their fitness. For lepidopteran herbivores,
fast growth can lead to larger body size at pupation, and
an increase in the quality and quantity of egg production
(Haukioja and Neuvonen 1985, Ohmart et al. 1985,
Awmack and Leather 2002). The length of the larval
period should also influence the risk of predation. For
example, more slowly growing larvae may be under
stronger selection for defense against natural enemies
(Feeny 1976, Benrey and Denno 1997, Loader and
Damman 1991). It has been argued that plant traits
that slow larval growth function as plant defenses by
increasing the larva?s risk to predation or parasitism.
However, the slow-growth high-mortality hypothesis
Accepted 26 May 2006
Subject Editor: Stig Larsson
Copyright # OIKOS 2006
ISSN 0030-1299
OIKOS 00: 00
00, 2006
DOI: 10.1111/j.2006.0030-1299.14928.x
OIKOS 00:0 (2006) Online Early (OE): 1-OE
(Benrey and Denno 1997) has received mixed support
(Ha?ggstro?m and Larsson 1995, Lill and Marquis 2001,
Medina et al. 2005, Kursar et al., in press). In this study
we examine how diet quality may constrain an herbi-
vore?s ability to maximize growth and to minimize
vulnerability to the third trophic level. We examine this
in a tropical forest where both the extent of plant
defenses (Coley and Aide 1991) and the impact of the
third trophic level (Dyer and Coley 2002) may be higher
than in the temperate zone.
For an insect herbivore, two of the most important
nutritional components of leaves are nitrogen and water.
Both have been shown to affect growth and diet choice in
laboratory and field studies with herbivorous insects
(Mattson 1980, Rausher 1981, Scriber and Slansky 1981,
Raupp and Denno 1983, Osier and Lindroth 2001,
Holton et al. 2003). Typically, herbivores prefer young
expanding leaves, because they have higher nutritional
value and lower toughness than mature leaves (Marquis
and Braker 1994, Coley and Barone 1996). This pre-
ference is particularly marked in tropical rainforests,
where more than 75% of the lifetime damage for a leaf
occurs during the short window when it is expanding
(Coley and Aide 1991).
Although young leaves generally have higher herbiv-
ory than mature, there is a 6-fold difference among
species in rates of damage to young leaves (Kursar and
Coley 2003). Species differ substantially with respect to
the rate of young leaf expansion, nutrients, and second-
ary metabolites (Coley and Kursar 1996), all traits which
influence herbivory on young leaves. In tropical species,
the most rapidly expanding leaves double in area every
day (Kursar and Coley 1992). Due to the physiological
constraint of allocating resources to growth, rapidly
expanding leaves tend to be high in nutrients and low in
defensive compounds (Orians and Janzen 1974, Kursar
and Coley 1991). Rapid expansion should reduce the
leaves? period of vulnerability to herbivores thereby
reducing damage (Feeny 1976, McKey 1979, Aide and
London?o 1989). However, because of the higher nitrogen
and lower chemical defenses, fast-expanders actually
suffer high rates of herbivory. In contrast, plants with
slow-expanding young leaves have less nitrogen and their
leaves are chemically well defended (Kursar and Coley
2003). In summary, the rate of expansion of young leaves
may correlate with diet quality, including the combined
effects of nutrition and defense, and consequently with
caterpillar growth rate.
Rates of predation and parasitism can be high in the
tropics (Jeanne 1979, Cornell and Hawkins 1995,
Hawkins et al. 1997, Stireman et al. 2005) and cater-
pillars have evolved a variety of defensive strategies
including rapid growth, accumulation of toxic chemicals,
construction of shelters, spines and hairs, cryptic colora-
tion and defensive or evasive behaviors (Eisner 1970,
Stamp and Casey 1993, Gross 1993, Gentry and Dyer
2002). Is the great variation among caterpillar species in
defenses related to the differences among food quality of
plant hosts (Barbosa 1988, Gauld and Gaston 1994,
Dyer 1995)?
We predicted that both caterpillar growth rates and
defense against natural enemies would be correlated with
diet quality. In general, caterpillars feeding on fast-
expanding leaves should benefit from the combination of
high nutrients and low chemical defenses. These cater-
pillars should grow the fastest. Because fast-growing
caterpillars have a short window of vulnerability to
natural enemies, they should also invest the least in
defenses. Caterpillars feeding on slow-expanding young
leaves will encounter intermediate levels of nutrients and
higher levels of toxic compounds. Thus we predict that
they should have intermediate growth rates and defense.
Mature leaf-feeders will have the lowest nutrients and
also must overcome high toughness and secondary
metabolites. These caterpillars should grow the slowest
and invest the most in anti-predator defenses. These
predictions apply to both specialist and generalist
caterpillars, but they may be more pronounced for
generalist caterpillars. Growth rates of specialist cater-
pillars should be most influenced by the nutritional
quality of leaves, as they are presumably adapted to the
secondary metabolites of their hosts. Generalists should
also be affected by nutrition, but might have a more
marked negative response to plant chemical defenses.
To determine how host plant defense and nutrition
affect herbivore growth rates and defenses against
natural enemies, we examined a community of cater-
pillars and their host plants from a tropical moist forest
in Panama. We compared the growth rates and defensive
traits of 85 species of caterpillar feeding on 40 species of
plants in 24 families that differed in the nutritional
quality, defenses and expansion rates of their leaves.
Methods
Study site
We conducted our study in a moist tropical lowland
forest on Barro Colorado Island (BCI), Panama (98N,
808W). The island is maintained and protected by the
Smithsonian Tropical Research Institute and is part of a
larger forested corridor that extends from the Atlantic to
the Pacific coasts. BCI experiences a marked dry season
that is usually four months long, and the vegetation is
classified as tropical, moist forest (Holdridge et al. 1971,
Croat 1978, Leigh 1999).
Plant species
We opportunistically collected caterpillars from 40
woody plant species of the island?s diverse understory
2-OE OIKOS 00:0 (2006)
of shade-tolerant plants (Appendix 1). These relatively
common plant species represent a variety of life histories.
The plant species included shrubs, juvenile lianas and
immature trees with growth strategies that differed
widely, even within plant genera. Leaf development
times ranged from very rapid to very slow. While many
of our species produced leaves continuously, others
flushed synchronously.
For each species, we analyzed several characteristics
that are relevant to leaf development and herbivore
attack: synchrony of leaf production, expansion rates of
young leaves, as well as nitrogen, and water contents of
both young and mature leaves. Not all data were
collected on all plant species. To measure the young
leaf expansion rate, we marked freshly emerged leaves
and measured their area with plastic grids every 48 h
until the leaves stopped expanding. We calculated
the daily percent increase in size from 15
80% of full
size during the expansion phase using the following
equation:
Expansion rate as percent per day
100[e(ln (area2=area1)=time)1]
where ??area1?? and ??area2?? are leaf areas at two different
measurements and ??time?? equals the number of days
between measurements. Values of 100% per day indicate
that the leaves doubled in size daily. We categorized a
plant species as a slow expander if its leaf expansion
rate was less than 28% per day and as a fast expander if
its leaf expansion rate was more than 28% per day.
Nitrogen and water content are highest in young leaves
and then diminish as leaves age (Coley 1983, Kursar and
Coley 1991, 1992). To measure these parameters, we
collected young leaves that were between 10% and 20%
expanded and mature leaves that were fully expanded
and fully toughened. Three replicates were run for each
species and age. Leaves were dried at 558C to obtain the
dry weight. Nitrogen content of the dried ground leaves
was determined with an isotope ratio mass spectrometer
(delta S, Finnigan MAT, San Jose, California, USA).
Lepidopteran collection and rearing
We reared a total of 400 individuals from 85 species
of caterpillars from October 1996 to June 1999
(Appendix 2). Our collection periods included both the
dry and wet seasons. For each caterpillar, we recorded
the plant species and leaf age on which it was feeding,
and photographed specimens. Using larvae and adults,
we identified the lepidopterans to the lowest taxon
possible: most to family and some to species. Voucher
specimens are stored on BCI, and some duplicate
specimens are with experts for identification. Using our

/10 years of field notes we classified the caterpillars as
specialists or generalists. Thus this classification is
primarily based on local diet breadth. When possible,
we corroborated our classifications using published
caterpillar databases and field guides. Following pre-
vious workers, we called a caterpillar a generalist if it fed
on more than two families of plants (Janzen 1984,
DeVries 1987, Marquis 1991, Barone 1998). The specia-
list caterpillars usually fed on plants from one family,
and most restricted their feeding to a single genus.
We reared all caterpillars individually in closed plastic
containers at ambient temperature in a screened and
shaded porch. Thus, temperature conditions were similar
for all growth measurements. We fed them leaves of the
same species and age as those on which they were
initially found. Each generalist species was collected
from a range of host plants, but individuals were reared
on leaves of the same species on which the caterpillar was
found. Leaves were replaced with fresh ones at least
every other day. A moistened paper towel in each cup
helped keep leaves fresh. Caterpillars were collected
opportunistically, and were generally found while in the
early instars.
To calculate relative growth rates, we weighed cater-
pillars every 24
48 h. For each healthy individual, we
averaged its weight change between measurements and
divided it by the midpoint weight to get an individual
relative growth rate (g g1 d1). Depending on the
developmental time, we obtained from 3 to 10 measure-
ments per caterpillar before it pupated. The negative
weight gains just before pupation were dropped. We
found no effect of initial weight on subsequent growth
rates (r2/0.000, p/0.3, n/145; Kursar et al., unpubl.).
We averaged relative growth rates for all the individuals
of the same caterpillar species feeding on the same leaf
age of the same species of plant to get a mean relative
growth rate (RGR) for each caterpillar/host combina-
tion.
Caterpillar defenses
Caterpillars were scored for traits that could serve as
defenses against natural enemies as has been done in
other studies (Gentry and Dyer 2002). We scored the 85
species of caterpillars that were reared for growth
measurements, plus an additional 11 morphospecies
that were not reared. Caterpillars were scored from
0 to 2 for ??color??, with scores increasing from cryptic
species, to colorful to warningly colored. ??Gregarious-
ness?? was scored as 0 for solitary feeders and 1 for
gregarious feeders. These two traits may indicate greater
chemical defense. Values for ??hairs?? and ??spines?? were 0
for none, 1 for sparse and 2 for dense. If caterpillars
made shelters from cut, webbed or folded leaves, they
received a score of 1 for the ??shelter?? category.
??Behavior?? was scored as 0 if caterpillars hid when
not feeding or as 1 if they remained conspicuous. This
OIKOS 00:0 (2006) 3-OE
was evaluated by watching each caterpillar in the field.
Other behavioral defenses, such as evasive tactics when
attacked, were not measured.
Statistics
We analyzed our data with SAS (SAS Institute Inc.,
2000). Growth rates were compared among leaf classes
and among specialist and generalist herbivores with
ANOVA (Proc GLM). Simple and multiple regressions
were used to examine the effects of expansion, nitrogen
and water on growth (Proc GLM and Proc REG). A
principle components analysis (Proc PRINCOMP) and
a discriminant analysis (Proc DISCRIM) were used to
separate caterpillar species based on water, nitrogen and
expansion rates of the host. ANCOVA was used to
examine the growth residuals for different leaf classes
after the effects of water and nitrogen had been removed
(Proc GLM). The relationship between expansion and
growth was examined in a partial regression with water
and nitrogen removed (Proc REG). Differences between
anti-predator defenses of specialists and generalists
were compared with a t-test, corrected for multiple
comparisons. Defense differences for caterpillars feeding
on different leaf classes were compared with an ANOVA,
also corrected for multiple comparisons.
Results
Leaf nutritional quality for young and mature leaves
We compared the nutritional quality of young and
mature leaves of 37 species of plants. Young leaves had
approximately 80% more nitrogen than mature leaves
(3.6% vs 1.95%; pB/0.001). Young leaf nitrogen content
varied among species by a factor of 4.3 (from 1.73% to
7.38% of dry weight), while mature leaf nitrogen content
ranged from 1.10% to 3.75%. For all species except one,
young leaves had higher nitrogen content than the
mature leaves. Young leaves also had a higher water
content than mature leaves (77.1% of fresh weight vs
62.5%, pB/0.0001). Young leaf water content ranged
from 64% to 92% of fresh weight while mature leaf water
content ranged from 44% to 84%. Nitrogen content was
positively correlated with water content for both young
and mature leaves considered together (r2/0.40, pB/
0.001), mature only (r2/0.22, pB/0.005, n/37), and
young only (r2/0.20, p/0.005, n/38).
Caterpillar growth rates on young and mature leaves
Do the large differences in nutritional quality between
young and mature leaves influence caterpillar growth
rates? Caterpillar growth rates ranged by 14-fold in our
community survey of 85 lepidopteran species (0.05 to
0.72 to g g1 d1). There was no effect of caterpillar
family, but significant effects of both leaf age and diet
breadth (Table 1). Growth was significantly faster on
young leaves for both generalists and specialists (Fig. 1;
ANOVA F3,94/25.4, pB/0.001). Eleven species of cater-
pillar were grown on both young and mature leaves, and
in 10 cases, growth was faster on young leaves (paired t-
test, t/2.2 pB/0.05, n/11), with average growth on
young leaves being 2.2 times higher.
The extent of diet specialization also influenced
growth rates (Fig. 1; ANOVA F3,94/9.58, p/0.0026).
Specialists grew ten times faster than generalists on
mature leaves and nearly two times faster than general-
ists on young leaves. These differences between specia-
lists and generalists are not explained by differences in
traits of their host plants. In a principle components
analysis separating caterpillars species according to the
water, nitrogen and expansion rates of their hosts, the
first two axes explained 87% of the variance yet failed to
separate generalists and specialists. In a discriminant
analysis using the same host traits, only 44% of the
generalists were correctly classified and 86% of the
specialists.
The differences in caterpillar growth rates on young
and mature leaves are partly explained by differences in
nitrogen and water. For specialists, there was a moderate,
but significant, positive relationship between nitrogen
Table 1. ANOVA for effects of herbivore family, leaf age (young
vs mature) and diet breadth (specialist vs generalist) on relative
growth rates of caterpillars g g1 d1.
Variables DF Mean square F value p value
Diet 1 0.201 12.5 B/0.001
Leaf age 1 0.368 22.9 B/0.001
Family 18 0.018 1.1 0.373
Error 63 1.902
Fig. 1. Relative growth rates (g g1 d1) of specialist and
generalist lepidopteran caterpillars feeding on young and
mature leaves of 37 different plant species. The number of
morphospecies in each category is presented, values with
different letters are significantly different at pB/0.05 (ANOVA
F3,94/13.3, pB/0.001).
4-OE OIKOS 00:0 (2006)
content of the host leaves and RGR (Fig. 2a; r2/0.17,
p/0.0003, n/69). There was a similar positive relation-
ship for water (r2/0.08, p/0.0052, n/82). For general-
ists, the effects of both nitrogen and water on growth
were larger (Fig. 2b; nitrogen: r2/0.68, p/0.007, n/8;
water: r2/0.73, n/11, pB/0.0005). A multiple regres-
sion using both water and nitrogen was significant
for both specialists and generalists (r2/0.16, p/0.001,
n/68, r2/0.96, pB/0.001, n/7, respectively).
Caterpillar growth rates on young leaves with
different leaf expansion rates
Expansion rates of young leaves differed substantially
among species, with leaf expansion rates of 106% per day
(leaves doubling in size in less than a day) for the fastest
species (Talisia princeps ) and expansion rates of 3.6%
per day (doubling time of 20 days) for the slowest species
(Psychotria limonensis ). Expansion rate was positively
correlated with leaf nitrogen content (Fig. 3; r2/0.42,
pB/0.001, n/24) but not with water. Caterpillar growth
rates varied 10-fold among young-leaf feeders and were
positively correlated with expansion rates of young
leaves. For specialists, this relationship was significant
(r2/0.13, p/0.003, n/60 caterpillars) and for general-
ists it was not (r2/0.11, p/0.24, n/7). In a multiple
regression including water, nitrogen and expansion rate,
the standardized partial coefficients were 0.02, 0.10 and
0.12 respectively, showing that nitrogen and expansion
explained most of the variance in growth (r2 /0.20, pB/
0.0001).
Effects of nutrients and defenses on growth
The differences in leaf traits between mature, fast- and
slow-expanding leaves is reflected in the growth rates of
caterpillars (Fig. 4a). The differences in growth for
larvae feeding on different leaf functional groups is
significant for both specialists (ANOVA, F2,82/18.3,
pB/0.001; Fig. 4a) and generalists (ANOVA, F2,11/
4.38, p/0.047, not shown). When we removed the
effects of nitrogen and water, the residual growth rates
of specialists were still highest on fast-expanding young
leaves and slowest on mature leaves (Fig. 4b, ANOVA
F2,78/8.39, pB/0.001). For generalists, the trend was the
same, but the differences were not significant (ANOVA
F2,8/1.61, p/0.28, not shown).
In the case of young leaves, when the effects of
nitrogen and water were removed, the growth residuals
were also still positively correlated with expansion rates
of leaves. For specialists the explanatory power was
modest (r2/0.10, pB/0.01, n/60), but for generalists,
there was a strong relationship (r2/0.52, pB/0.05,
n/7).
Are plant traits correlated with herbivore defenses
against natural enemies?
Because generalists grow more slowly than specialists
(Fig. 1), we hypothesized that their extended larval
period should select for greater defenses against natural
enemies (Table 2). Three of the comparisons were
Fig. 2. Effects of leaf nitrogen content on relative growth rates
(g g1 d1) of lepidopteran caterpillars. For specialist cater-
pillars r2/0.18, pB/0.001, n/69 morphospecies and for
generalist caterpillars r2/0.68, pB/0.007, n/8 morphospecies.
Fig. 3. Correlation between expansion rate of young leaves (%
increase in size per day) and nitrogen content (% dry weight) for
24 species (r2/0.44, pB/0.001, slope/0.021).
OIKOS 00:0 (2006) 5-OE
significant at the level appropriate for repeated analyses
(pB/0.01), and trends are evident for the other two.
Generalists were more brightly or warningly colored and
were more often gregarious, two correlates of effective
chemical defense (Table 2A). They also had more hairs
and spines. However, they never made protective shelters
and did not hide when resting, perhaps because they
have effective physical and chemical defenses and have to
feed for such an extended period.
A similar pattern for more defense in slow-growing
caterpillars was seen when comparing species that fed on
different leaf classes: mature leaves, slow-expanding or
fast-expanding young leaves. For specialists, warning
color, conspicuous behavior, gregariousness and hairs
were high for species feeding on mature leaves, inter-
mediate on slow-expanding young leaves and lowest on
fast-expanding young leaves (Table 2B). Although
mature leaf feeders rarely build houses, species feeding
on slow-expanding leaves were more likely to live in
shelters than those feeding on fast-expanding leaves.
Generalists were not analyzed because of the small
sample size.
We tested palatability of caterpillars to ants in feeding
trials using Ectatomma ruidum , the most common insect
predator found on plants on BCI (Coley et al. 2005).
Ectatomma more frequently rejected caterpillars that fed
on mature leaves and more readily carried off species
that fed on fast-expanding young leaves (Fig. 5).
Discussion
Does plant nutrient content affect caterpillar
growth?
There were large differences in caterpillar growth rates
across hosts, with RGR differing 14-fold among species.
In general, growth rates on young leaves were double
those on mature leaves, and within young leaves, an
increase in leaf expansion rate was correlated with an
increase in caterpillar growth rate. These differences in
growth are partially explained by nutrition, as nitrogen
content, and to a lesser extent water content, were
positively correlated with growth. Thus, despite the
presence of physical and chemical defenses, the signature
of nutrition is readily evident in caterpillar growth rates.
Fig. 4. (A) Relative growth rates (g g1 d1) of specialist
caterpillar species on fast-expanding young leaves (n/21),
slow-expanding young leaves (n/39) and mature leaves (n/
23; ANOVA, F2,82/18.3, pB/0.0001). Different letters indicate
that values are significantly different at pB/0.05. (B) Residuals
for relative growth rate with the effects of water and nitrogen
content removed (ANCOVA, F2,78/8.4, pB/0.005, fast n/21,
slow n/38, mature n/20).
Table 2. Traits of caterpillars that could serve as defenses against natural enemies. For ??color??, scores increase from cryptic species,
to colorful to warningly colored (0
2). For ??gregariousness?? solitary feeders received a score of 0 and gregarious feeders a score of
1. Values for ??hairs?? and ??spines?? ranged from 0 to 2. If caterpillars made shelters, they received a score of 1 for the ??shelter??
category. Behavior received a 0 if it was hiding when not feeding, versus a 1 if it was resting in an obvious position. Generalist and
specialist scores were compared with a t-test, scores for caterpillar species feeding on different leaf classes were compared with an
ANOVA. ??N?? is the number of morphospecies of caterpillars, values are the mean scores with standard errors in parentheses. To
account for multiple comparisons, p values should be B/0.01.
N Color Gregarious Hairs Spines Shelter Behavior
A. Diet specialization
Generalists 20 1.19 (0.13) 0.29 (0.10) 1.24 (0.22) 0.33 (0.14) 0.00 (0.00) 0.91 (0.07)
Specialists 86 0.38 (0.07) 0.08 (0.03) 0.17 (0.06) 0.13 (0.04) 0.39 (0.05) 0.29 (0.05)
t-test (p value) B/0.001 0.032 B/0.001 0.09 
 B/0.001
B. Leaf classes (specialists only)
Mature leaves 25 0.65 (0.14) 0.15 (0.07) 0.35 (0.13) 0.27 (0.12) 0.19 (0.08) 0.58 (0.10)
Slow-expanders 39 0.28 (0.10) 0.08 (0.04) 0.10 (0.06) 0.05 (0.04) 0.51 (0.08) 0.18 (0.06)
Fast-expanders 22 0.23 (0.11) 0.00 (0.00) 0.09 (0.09) 0.09 (0.06) 0.41 (0.11) 0.14 (0.08)
ANOVA (p value) 0.03 0.15 0.12 0.08 0.03 B/0.001
6-OE OIKOS 00:0 (2006)
Do plant defenses affect caterpillar growth?
We did not measure plant defenses directly, but indirect
evidence suggests that both toughness and secondary
metabolites affect caterpillar growth. In a multiple
regression with water, nitrogen and expansion rates,
nitrogen and expansion contributed most to growth
differences. When the effects of nitrogen and water
were removed, there were still correlations of larval
growth with leaf age classes and with leaf expansion
rates, suggesting that other traits must be influencing
growth. One possibility is other nutritional factors that
we did not measure, such as phosphorous, minor
minerals or non-structural carbohydrates. These traits
vary among species and tissues and can affect host
use and performance (Awmack and Leather 2002,
Perkins et al. 2004). Alternative explanations for growth
differences among leaf classes are physical and chemical
defenses.
Mature leaves are very tough, and toughness is one
of the most effective defenses (Coley 1983). The slowest
residual growth rates were associated with mature
leaves, mostly a result of high toughness. Young leaves
are not tough, so differences in caterpillar growth rates
among species with different expansion rates more
likely reflect differences in secondary metabolites (Coley
and Barone 1996, Kursar and Coley 2003, Coley et al.
2005). When the effects of nitrogen and water were
removed, the residuals for growth were smallest on
species with slowly expanding leaves, indicating that
slow leaf expansion is associated with a plant trait that
slows caterpillar growth. Previous studies of 24 species
demonstrated that extracts from slow-expanding young
leaves were less preferred by insect herbivores in choice
tests and reduced growth more in fungal pathogens
than extracts from that fast-expanders (Kursar and
Coley 2003). These bioassays indicated that slow-
expanding species have more deterrent or more toxic
secondary metabolites. The present study, based on
herbivore growth rates, is an independent test of these
ideas and provides additional evidence consistent with
more effective secondary metabolites in species with
slow-expanding young leaves. Even though these corre-
lations are informative, much variation in caterpillar
growth rate is unexplained. Ultimately, a good under-
standing of the role of plant chemistry will require
direct analysis of plant chemical defenses and herbivore
responses.
How do specialist and generalist caterpillars differ?
Specialist herbivores grew significantly faster than
generalists. We found no evidence that specialists
choose a subset of plants which would shape their
growth rates. Using three plant traits indicative of diet
quality (water, nitrogen and leaf expansion rate), PCA
and discriminant analyses were unable to separate
specialists and generalists. The growth differences
were also apparently not due to phylogenetic biases as
herbivore family had no significant effects (Table 1).
Furthermore, the generalist species were from six
different families, and all but one (Saturniidae) had
specialist representatives. We suggest that specialists
grew faster because they were less sensitive to host-
plant defenses or could make more efficient use of host
nutrients. Thus, rather than being shaped by phylogeny
or the diet quality provided by the host, the slow
growth of generalists may be due to a trade off with the
herbivores? broad diet breadth.
Both specialists and generalists showed positive
responses to increased nitrogen and water, demonstrat-
ing the importance of nutritional traits. However, we
predicted that generalists would be more sensitive to
secondary metabolites as they might lack specialized
detoxification or sequestration mechanisms (Futuyma
and Moreno 1988, Berenbaum and Zangerl 1994,
Mackenzie 1996, Cornell and Hawkins 2003). This is
best examined in young leaf-feeders as confounding
effects of toughness are avoided. For young leaves,
a partial regression removing the effects of water
and nitrogen showed a much stronger relationship
between caterpillar growth and leaf expansion rate for
generalists (r2/0.52, pB/0.05, n/7) than for specia-
lists (r2/0.10, pB/0.01, n/60). Although we cannot
Fig. 5. Rejection and acceptance rates by the predatory ant,
Ectatomma ruidum for caterpillars specialized on different leaf
classes: fast-expanding young leaves, slow-expanding young
leaves and mature leaves. Ant behaviors were scored as rejection
when they contacted the caterpillar but immediately rejected it
and as acceptance if they carried it to the nest. For each of the
36 caterpillar morphospecies tested (n/12 for fast, n/12 for
slow and n/12 for mature), approximately 9 individuals were
presented to ants and the proportion rejected or accepted was
calculated. Rejection and acceptance values do not add to 100%
because there were other classes of behaviors (such as biting or
carrying the caterpillar for a short distance). For comparisons
within rejection or acceptance rates, values with different letters
are significantly different at pB/0.05 using Duncan?s multiple
range test (ANOVA, reject F2,36/2.61, p/0.09, accept F2,36/
2.72, p/0.08).
OIKOS 00:0 (2006) 7-OE
eliminate effects of other unmeasured nutritional
differences, we interpret this result to indicate that
secondary metabolites present a greater challenge to
generalists and may have a bigger negative effect on
their growth. However, even among specialists, the
growth rate residuals were significantly correlated with
leaf expansion rate, suggesting better defended plant
species present a bigger cost of detoxification for all
Lepidoptera. Thus bottom-up effects of plant quality
due to nutritional and defensive differences were
evident for both specialist and generalist herbivores.
These results also indicate that the more responsive,
and therefore more appropriate, Lepidoptera for
bioassays of plant chemical activity would be the
generalists.
Are plant traits correlated with herbivore defenses
against natural enemies?
We find that the differences in plant nutrition and
defense place predictable constraints on the life history
traits of herbivores feeding on them, including herbi-
vore growth and defense against natural enemies. The
nutritional value of the host plant is correlated with the
growth rate of the herbivores. Differences in larval
growth rates influence the amount of time a larva is at
risk of predation, so slow-growing caterpillars should
be under strong selection to evolve defenses against
natural enemies (Price et al. 1980, Loader and Dam-
man 1991, Benrey and Denno 1997). Our results
support this hypothesis, with much higher defense
scores for generalists than specialists and for mature
versus young leaf feeders. The differences in defensive
traits were also confirmed in feeding trials with ants.
The generalists were from eight different families, so it
is unlikely that defense differences were driven by
phylogenetic biases.
Although our data indicate that slow larval growth
rates of generalists are correlated with greater defense,
these results conflict with other studies in which
specialists were less palatable to predators (Bernays
and Cornelius 1989, Dyer 1995). In Dyer?s work with
tropical lepidopterans, specialists were more able to
sequester host plant chemicals and had reduced preda-
tion (Dyer and Floyd 1993, Dyer 1997, Gentry and
Dyer 2002). While we noted that some specialist
caterpillars had aposomatic coloring or spines, suggest-
ing effective defense, we also found 14 species of
specialists in 11 families that were green and constructed
shelters. In our study, the generalists were better
defended by all the measured traits, including indicators
of chemical defense, however these could be produced
autogenously rather than sequestered from the host.
More extensive studies of the physical and chemical
defenses of generalists would shed light on whether
sequestration determines host choice.
What is the relative importance of top-down vs
bottom-up effects on caterpillar traits?
These results could be viewed as supporting bottom-up
effects, with diet quality shaping herbivore traits and
hence their susceptibility to the third trophic level.
However, an alternative explanation is that the evolution
of caterpillar defenses influences host use. For example, a
caterpillar with effective intrinsic defenses, such as
spines, hairs or chemicals it synthesizes would be less
constrained in diet than a caterpillar that relies on
sequestering host secondary metabolites. Hence, the
evolution of intrinsic defenses could predispose the
evolution of generalist feeding. In either case, the greater
investment in defensive traits for species with longer
larval developmental times suggests that predation is a
strong selective factor and that top-down effects could
be important in the ecology and evolution of tropical
lepidopteran larvae (Dyer and Coley 2002).
The relative importance of top-down versus bottom-
up effects should determine herbivore populations. The
low abundance of mature leaf feeders and generalists is a
notable characteristic of tropical rainforests despite the
obvious abundance of potential food (Barone 1998,
Novotny et al. 2002, 2004). We hypothesize that the
very slow growth rates of these herbivores result in
strong top-down effects, which may severely limit their
success (Hairston et al. 1960, Price et al. 1980). Thus, for
mature leaf feeders, bottom-up effects of poor food
quality interact with predators and parasitoids to
determine herbivore populations. In contrast, the fast-
growing herbivores that feed on young, expanding leaves,
grow rapidly and are more abundant, suggesting the
primacy of bottom-up effects. These herbivores may be
more analogous to temperate species found on spring
flushes. These temperate and tropical young leaf feeders
are not only confronted with plant nutrition and defense,
but they must also be adapted to plant phenological
traits that result in a very short window of food
availability.
It should be noted that the results in this study are
from a single site. Our field observations in Central
African Republic and the Republic of Congo indicate
that the caterpillar community has relatively few young
leaf specialists (pers. obs.). Instead, the forest is domi-
nated by caterpillars that are well defended species and
fairly abundant. Hence the lepidopteran community
composition can best be understood by considering
how diet quality, an herbivore?s susceptibility to enemies,
and herbivore defenses play out in different ecosystems
(Singer and Stireman 2005).
8-OE OIKOS 00:0 (2006)
Conclusions
This study shows that growth and defensive traits of
lepidopteran herbivores are constrained by the nutri-
tional and defensive traits of their hosts. This is mediated
by the effect of leaf toughness, secondary metabolites
and nutrients on herbivore growth rates which in turn
influences their susceptibility to the third trophic level.
Thus the combination of bottom up factors such as
nutritional quality and defenses of plants, and top-down
effects of natural enemies all interact to shape diet
breadth and abundance of caterpillars.
Acknowledgements 
 We acknowledge financial support from
the NSF (DEB 0108150 and DEB 0234936 to PDC and TAK).
We are grateful to Edmundo Ayarza, Terry Brncic, Lia Florey,
Rachel Goeriz, Marie Massa and Mathew Smith for help with
field work, to Brett Wolfe and Mary Jane Epps for data
organization, to the Autoridad Nacional del Ambiente of
Panama, Oris Acevedo and the late Daniel Millan of the
Smithsonian for facilitating research in Panama, and to Egbert
Leigh for stimulating discussions.
References
Aide, T. M. and London?o, E. C. 1989. The effects of rapid leaf
expansion on the growth and survivorship of a lepidopteran
herbivore. 
 Oikos 55: 66
70.
Awmack, C. S. and Leather, S. R. 2002. Host plant quality and
fecundity in herbivorous insects.
Annu. Rev. 
 Entomol.
47: 817
844.
Barbosa, P. 1988. Natural enemies and herbivore
plant inter-
actions: influence of plant allelochemicals and host specifi-
city. 
 In: Barbosa, P. and Letourneau, D. K. (eds), Novel
aspects of insect-plant interactions. John Wiley and Sons,
pp. 201
229.
Barone, J. A. 1998. Host-specificity of folivorous insects in a
moist tropical forest. 
 J. Anim. Ecol. 67: 400
409.
Benrey, B. and Denno, R. F. 1997. The slow-growth-
high-mortality hypothesis: a test using the cabbage butterfly.

 Ecology 78: 987
999.
Berenbaum, M. R. and Zangerl, A. R. 1994. Costs of inducible
defense: protein limitation, growth and detoxification in
parsnip webworms. 
 Ecology 75: 2311
2317.
Bernays, E. A. and Cornelius, M. L. 1989. Generalist caterpillar
prey are more palatable than specialists for the generalist
predator Iridomyrmex humilis. 
 Oecologia 79: 427
430.
Coley, P. D. 1983. Herbivory and defense characteristics of tree
species in a lowland tropical forest. 
 Ecol. Monogr. 53:
209
233.
Coley, P. D. and Aide, T. M. 1991. Comparison of herbivory
and plant defenses in temperate and tropical broad-leaved
forests. 
 In: Price, P. W., Lewinsohn, T. M., Fernandes, W.
W. et al. (eds), Plant
animal interactions: evolutionary
ecology in tropical and temperate regions. John Wiley and
Sons, pp. 25
49.
Coley, P. D. and Barone, J. A. 1996. Herbivory and plant
defenses in tropical forests. 
 Annu. Rev. Ecol. Syst. 27:
305
335.
Coley, P. D. and Kursar, T. A. 1996. Anti-herbivore defenses of
young tropical leaves: Physiological constraints and ecolo-
gical tradeoffs. 
 In: Mulkey, S. S., Chazdon, R. and Smith,
A. P. (eds), Tropical forest plant ecophysiology. Chapman
and Hall, pp. 305
336.
Coley, P. D., Lokvam, J., Rudolph, K. et al. 2005. Divergent
defensive strategies of young leaves in two species of Inga .

 Ecology 86: 2633
2643.
Cornell, H. V. and Hawkins, B. A. 1995. Survival patterns and
mortality sources of herbivorous insects: some demographic
trends. 
 Am. Nat. 145: 563
593.
Cornell, H. V. and Hawkins, B. A. 2003. Herbivore responses to
plant secondary compounds: a test of phytochemical
coevolution theory. 
 Am. Nat. 161: 507
522.
Croat, T. B. 1978. The flora of Barro Colorado Island.

 Stanford Univ. Press.
DeVries, P. J. 1987. The butterflies of Costa Rica and their
natural history: Papillionidae, Pieridae, Nymphalidae.

 Princeton Univ. Press.
Dyer, L. A. 1995. Tasty generalists and nasty specialists? A
comparative study of antipredator mechanisms in tropical
lepidopteran larvae. 
 Ecology 76: 1483
1496.
Dyer, L. A. 1997. Effectiveness of caterpillar defenses against
three species of invertebrate predators. 
 J. Res. Lepidoptera
34: 48
68.
Dyer, L. A. and Floyd., T. 1993. Determinants of predation
on phytophagous insects: the importance of diet breadth.

 Oecologia 96: 575
582.
Dyer, L. A. and Coley, P. D. 2002. Tritrophic interactions in
tropical and temperate communities. 
 In: Tscharntke, T.
and Hawkins, B. (eds), Multitrophic level interactions.
Cambridge Univ. Press, pp. 67
88.
Eisner, T. 1970. Chemical defense against predation in arthro-
pods. 
 In: Sondeheimer, E. and Simeone, J. B. (eds),
Chemical ecology. Academic Press, pp. 157
217.
Feeny, P. P. 1976. Plant apparency and chemical defense.

 Recent Adv. Phytochem. 10: 1
40.
Futuyma, D. and Moreno, G. 1988. The evolution of ecological
specialization. 
 Annu. Rev. Ecol. Syst. 19: 207
233.
Gauld, I. and Gaston, K. 1994. The taste of enemy-free space:
parasitoids and nasty hosts. 
 In: Hawkins, B. and Sheehan,
W. (eds), Parasitoid community ecology. Oxford Univ. Press,
pp. 279
299.
Gentry, G. L. and Dyer, L. A. 2002. On the conditional nature
of Neotropical caterpillar defenses against their natural
enemies. 
 Ecology 83: 3108
3119.
Gross, P. 1993. Insect behavioral and morphological defenses
against parasitoids. 
 Annu. Rev. Entomol. 38: 251
273.
Ha?ggstro?m, H. and Larsson, S. 1995. Slow larval growth on a
suboptimal willow results in high predation mortality in the
leaf beetle Galerucella lineola . 
 Oecologia 104: 308
315.
Hairston, N. G., Smith, F. E. and Slobodkin, L. B. 1960.
Community structure, population control, and competition.

 Am. Nat. 94: 421
425.
Haukioja, E. and Neuvonen, S. 1985. The relationship between
size and reproductive potential in male and female Epirrita
autumnata (Lep., Geometridae). 
 Ecol. Entomol. 10: 267

270.
Hawkins, B., Cornell, H. and Hochberg, M. 1997. Predators,
parasitoids, and pathogens as mortality agents in phytopha-
gous insect populations. 
 Ecology 78: 2145
2152.
Holdridge, L., Grenke, W., Hatheway, W. et al. 1971. Forest
environments in tropical life zones. 
 Pergamon Press.
Holton, M. K., Lindroth, R. L. and Nordheim, E. V. 2003.
Foliar quality influences tree-herbivore-parasitoid interac-
tions: effects of elevated CO2, O3, and plant genotype.

 Oecologia 137: 233
244.
Janzen, D. 1984. Two ways to be a tropical big moth: Santa
Rosa saturniids and sphingids. 
 Oxford Surv. Evol. Biol. 1:
85
140.
Jeanne, R. L. 1979. A latitudinal gradient in rates of ant
predation. 
 Ecology 60: 1211
1224.
Kursar, T. A. and Coley, P. D. 1991. Nitrogen content and
expansion rate of young leaves of rainforest species:
implications for herbivory. 
 Biotropica 123: 141
150.
Kursar, T. A. and Coley, P. D. 1992. Delayed greening in
tropical leaves: an anti-herbivore defense? 
 Biotropica 24:
256
262.
Kursar, T. A. and Coley, P. D. 2003. Convergence in
defense syndromes of young leaves in tropical rainforests.

 Biochem. Syst. Ecol. 21: 929
949.
OIKOS 00:0 (2006) 9-OE
Kursar, T. A., Wolfe, B. T., Epps, M. J. et al. 2006. The effects of
plant quality, food availability, competition and parasitism
on performance and diet electivity in tropical Lepidoptera
feeding on Inga (Fabaceae). 
 Ecology (in press).
Leigh, E. G. 1999. Tropical forest ecology: a view from Barro
Colorado Island. 
 Oxford Univ. Press.
Lill, J. T. and Marquis, R. J. 2001. The effects of leaf quality
on herbivore performance and attack from natural enemies.

 Oecologia 126: 418
428.
Loader, C. and Damman, H. 1991. Nitrogen content of
food plants and vulnerability of Pieris rapae to natural
enemies. 
 Ecology 72: 1586
1590.
Mackenzie, A. 1996. A tradeoff for host plant utilization in the
black bean aphid. 
 Evolution 50: 155
162.
Marquis, R. J. 1991. Herbivore fauna of Piper (Piperaceae)
in a Costa Rican wet forest: diversity, specificity and impact.

 In: Price, P. W., Lewinsohn, T. M., Fernandes, G. W. et al.
(eds), Plant
animal interactions: evolutionary ecology in
tropical and temperate regions. Wiley & Sons, pp. 197
208.
Marquis, R. J. and Braker, H. E. 1994. Plant-herbivore
interactions: diversity, specificity and impact. 
 In: McDade,
L. A., Bawa, K. S., Hespenheide, H. A. et al. (eds), La Selva,
ecology and natural history of Neotropical rain forest. Univ.
of Chicago Press, pp. 261
281.
Mattson, W. J. 1980. Herbivory in relation to plant nitrogen
content. 
 Annu. Rev. Ecol. Syst. 11: 119
161.
McKey, D. 1979. The distribution of secondary plant metabo-
lites. 
 In: Rosenthal, J. P. and Janzen, D. (eds), Herbivores:
their interactions with secondary plant metabolites. Aca-
demic Press, pp. 55
133.
Medina, R. F., Barbosa, P. and Waddell, K. 2005. Parasitism
levels in Orgyia leucostigma feeding on two tree species:
implications for the slow-growth- high-mortality hypothesis.

 Entomol. Exp. Appl. 115: 193
197.
Novotny, V., Miller, S. E., Basset, Y. et al. 2002. Predictably
simple: assemblages of caterpillars (Lepidoptera) feeding on
rainforest trees in Papua, New Guinea. 
 Proc. R. Soc.
Lond. B 269: 2337
2344.
Novotny, V., Basset, Y., Miller, S. E. et al. 2004. Local species
richness of leaf-chewing insects feeding on woody plants
from one hectare of a lowland rainforest. 
 Conserv. Biol.
18: 227
237.
Ohmart, C. P., Stewart, L. G. and Thomas, J. R. 1985. Effects
of food quality, particularly nitrogen concentrations, of
Eucalyptus blakelyi foliage on the growth of Paropsis
atomaria larvae (Coleoptera: Chrysomelidae). 
 Oecologia
65: 543
549.
Orians, G. H. and Janzen, D. H. 1974. Why are embryos so
tasty? 
 Am. Nat. 108: 581
592.
Osier, T. L. and Lindroth, R. L. 2001. Effects of geno-
type, nutrient availability and defoliation on aspen phyto-
chemistry and insect performance. 
 J. Chem. Ecol. 27:
1289
1313.
Perkins, M. C., Woods, H. A., Harrison, J. F. et al. 2004.
Dietary phosphorus affects the growth of larval Manduca
sexta . 
 Arch. Insect Biochem. Physiol. 55: 153
168.
Price, P. W., Bouton, C., Gross, P. et al. 1980. Interactions
among three trophic levels: influences of plants on inter-
actions between insect herbivores and natural enemies.

 Annu. Rev. Ecol. Syst. 11: 41
65.
Raupp, M. J. and Denno, R. F. 1983. Leaf age as a predictor of
herbivore distribution and abundance. 
 In: Denno, R. F.
and McClure, M. S. (eds), Variable plants and herbivores in
natural and managed systems. Academic Press, pp. 91
124.
Rausher, M. D. 1981. Host plant selection by Battus philenor
butterflies: the roles of predation, nutrition, and plant
chemistry. 
 Ecol. Monogr. 51: 1
20.
Rosenthal, G. A. and Berenbaum, M. R. 1991. Herbivores: their
interactions with secondary plant metabolites, 2nd ed.

 Academic Press.
SAS. 2002. Version 8.1, SAS Institute Inc., Cary, NC, USA.
Scriber, J. M. and Slansky, F. J. 1981. The nutritional ecology of
immature insects. 
 Annu. Rev. Entomol. 26: 183
211.
Singer, M. S. and Stireman, J. O., III. 2005. The tri-trophic
niche concept and adaptive radiation of phytophagous
insects. 
 Ecol. Lett. 8: 1247
1255.
Stamp, N. E. and Casey, T. M.. 1993. Caterpillars: ecological
and evolutionary constraints on foraging. 
 Chapman and
Hall.
Stireman, J. O., III., Dyer, L. A., Janzen, D. H. et al. 2005.
Climatic unpredictability and parasitism of caterpillars:
implications of global warming. 
 Proc. Natl Acad. Sci.
USA 102: 17384
17387.
Appendix 1 and 2 are available as O14928 at:
www.oikos.ekol.lu.se
10-OE OIKOS 00:0 (2006)