International Association for Ecology
Quantitative Defense Theory and Patterns of Feeding by Oak Insects
Author(s): Stanley H. Faeth
Source: Oecologia, Vol. 68, No. 1 (1985), pp. 34-40
Published by: Springer in cooperation with International Association for Ecology
Stable URL: http://www.jstor.org/stable/4217793
Accessed: 07/05/2010 21:20
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=springer.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and
extend access to Oecologia.
http://www.jstor.org
Oecologia
1985
Springer-Verlag
Oecologia(Berlin)(1985)68:34-40
©
Quantitative defense theory and patterns of feeding by oak insects
Stanley H. Faeth
Departmentof Zoology, Arizona State University, Tempe, AZ 85287, USA
Summary. Patterns of herbivory over a two year period
on Quercusemoryi(Fagaceae) were correlatedwith seasonal
and yearly changes in tannin and protein content. Quantitative defense theory predicts that tannin and protein content
in apparent plants should be negatively and positively correlated, respectively,with degree of herbivory. Most herbivory occurred early in the growing season, but the pattern
varied between the two years. Tannin and protein content
sometimes varied negatively and sometimes positively with
degree of herbivory; they did not consistently covary with
herbivory. Protein content was positively correlated with
herbivory in 1981-1982 but not in 1982-1983. Condensed
tannin content was negatively correlated with herbivory in
1981-1982 but not in 1982-1983. Hydrolysable tannin content was positively correlatedwith herbivoryactivity. Multiple regression analyses indicated these phytochemical variables explained either no significant variation in herbivory
(1982-1983) or did so in a fashion opposite (1981-1982)
to the predictions of the theory of quantitative defense.
Feeding by oak insects was not solely a function of
seasonal changes in quantitativedefenses and nutrients.Obviously, population dynamics of the insects are sensitive
to factors other than phytochemistry of the trees and I
discuss other factors that can influence patterns of herbivory.
Feeny (1970) proposed that most lepidopterousinsects feed
on Quercus robur early in the spring because leaf quality
declines as the growing season progresses. He concluded
that both number of species and abundances decline in a
season because tannin content and leaf toughness increase,
while protein content decreases as leaves mature (Feeny
1968; Feeny and Bostock 1968; Feeny 1970). Based primarily on this work, Feeny (1975, 1976) and others (Rhoades
and Cates 1976) postulated a theory that linked general
types of chemical defense with "apparency" of plants.
Long-lived plants, such as trees, which are likely to be discovered by herbivores, evolved "quantitative" defenses,
while short-lived, transitory plants evolved "qualitative"
defenses. Quantitative defenses act in a stoichiometricfashion. For example, tannins supposedly bind plant proteins
and insect proteolytic enzymes and thus increasingamounts
render protein less available to insect herbivores.
These theories were ascendant for almost a decade:
"predictable", "apparent" plants should respond evolutionarily to herbivore pressureswith quantitative defenses.
Tannins were considered as the ideal quantitative defense.
Most of the evidence rested upon the fact that tannins are
ubiquitous and abundant among apparent, woody plants.
Experimentalevidence for the defensive nature and mode
of action of tannis was exiguous or nonexistent (Zucker
1983). Recently, the idea that tannis are generalized defenses and are always detrimental to herbivorous insects
has been challenged (Bernays 1978, 1981; Bernays et al.
1980; Bernaysand Woodhead 1982; Fox 1981; Berenbaum
1983; Coley 1983; Zucker 1983).
Here, patternsof herbivoryand seasonal changes in tannins and protein content of Quercusemoryi are examined
relative to the theory of quantitativedefense in "apparent"
plants.
Material and methods
Sampling
Six trees (- 3 m in height) of Quercus emoryi (Fagaceae)
were randomly selected from approximately 50 similarlysized trees that were located within 150 m radius of each
other in a riparian zone surrounding Parker Creek in the
Sierra Ancha ExperimentalForest (USDA), Gila, Co. AZ.
Leaves of Q. emoryi persist for one year from budbreak
in mid-April to early May.
The six study trees were monitored for numberof leaves
damaged by insects, and seasonal changes in hydrolysable
and condensed tannin, protein, and water content. Samples
of 100-200 leaves were randomly collected from each tree
monthly from April 1981 (budbreak) to November 1981,
and from April 1982 to November 1982. Leaves were also
sampled in early April 1981 (before budbreak)and in February and April 1983 (before budbreak).Leaf samples thus
represent the entire growing seasons of 1981-1982 and
1982-1983.
Estimates of damaged leaves
Sampled leaves were transported to the laboratory where
they were segregated into insect-damagedand undamaged
categories. Leaves were considered damaged if > 10% of
leaf area was consumed (determined by a Licor leaf area
meter). This level of damage was chosen because damage
35
,t**
less than this amount was not always discernibleas resulting
from insects (e.g., small necrotic areaswere caused by fungi,
bacteria, and physical damage). Mean cumulative percent
of leaves damaged for all trees was calculatedfor 1981-1982
and 1982-1983.
Using percent of leaves damaged as an indicator of insect herbivory assumes that once leaves were damaged, no
further herbivory occurred on these leaves. In other words,
herbivory late in the season occurred on leaves previously
undamaged by earlier herbivores. To test this assumption,
I regressed average mass (leaf mass and area are highly
correlatedin Q. emoryi, r2=0.965, P<0.001, Bultman and
Faeth, unpublished data) of damaged leaves for each tree
in each season with date of leaf sample. None of the 12
regressions(6 trees x 2 seasons) had slopes that were significantly different from zero (P>0.30). Since average damage
did not increase with time, this suggests that insect herbivores did not repeatedly consume leaves that were already
damaged by earlier insects.
Phytochemicalanalyses
Insect-damagedand intact leaves were analyzed separately
for each tree at each sample date for protein and hydrolysable and condensed tannin content. Condensed tannins
analyses included segregation into monomeric fractions of
phenols, polymeric fractions (condensed tannins), and total
vanillin positive fractions (monomers plus polymers). Only
analyses of condensed tannins are presentedhere; total vanillin positive shows similar trends because monomeric fractions were low and varied little. Randomly sampled leaves
(same leaves used in Materials and methods: Sampling)
from each tree were immediately frozen in dry ice in the
field, later lyophilized in the laboratory, and then ground
into a fine powder. Ground samples were stored in jars
at 5° C with a dessicant until analyses could be performed.
Each analysis for each tree from each date was done in
triplicate. Details of protein and tannin analyses are found
in Faeth (1986). Brief summariesare provided here.
Protein content of ground, lyophilized leaf samples was
determined by the BioRad Method (Richmond, CA).
Monomeric phenols, condensed tannins, and total vanillin
positive were determined by a modified acidified vanillin
method (Broadhurstand Jones 1978). Monomeric phenols
and total vanillin positive were first determined, and then
condensed tannins were determined by subtracting
monomers from total vanillin positive for each trial. Hydrolysable tannins were determinedby a modified iodate technique (Bate-Smith 1977) and expressedas percent dry mass
tannic acid equivalents as compared to a tannic acid standard curve prepared for each run. For protein and tannin
analyses, all extraction solutions, reagent concentrations,
temperatures,and time of reactions were empiricallydetermined to standardizeand optimize reactivity.
Water content of leaves was measuredin 1981-1982 and
1982-1983 for newly-emergedleaves, 1 month, and 1 year
old leaves.
Statistical analyses
Repeated measures ANOVA (Winer 1971) was used to test
for differencesin protein and hydrolysable and condensed
tannin content between intact and herbivore-damaged
leaves and for effects of seasonality on phytochemistryfor
study trees in 1981-82 and 1982-83.
~~~~~~~~~~~4,
4
4
l
l4
4
UJ
40-
l iil4'I
4
11
4
o
20-
I ~a
~1
-_ A i1
30-
a
10-
-j
0
;
*
9
o
I
u)
MAY
1981
I
I
I
SEPT
I
I
I
I
JAN
1982
I
I
I
I
MAY
I
I
I
I
SEPT
I
I
I
I
JAN
1983
I
I
I
I
MAY
Fig. 1. Seasonal changes in cumulative percentageof leaves damaged by insects in 1981-1982 and 1982-1983. Each point represents
the mean cumulativepercent of leaves damaged for 6 study trees,
with bars representingstandarderrorsof means at each date
Protein and hydrolysableand condensed tannin content
were regressedseparatelywith cumulative increases in herbivory during 1981-82 and 1982-83 to test the predictions
of Feeny (1975, 1976) and Rhoades and Cates (1976) that
herbivory should be positively correlatedwith protein content and negatively correlatedwith tannin content. Sample
dates when no herbivory occurred on trees were not used
in regression analyses, since the focus of this study is relationship of timing of herbivory and phytochemistry.Moreover, inclusion of zero herbivory points does not qualitatively alter results. Finally, multiple regression (separately
for each year) was used to determine the relative amount
of variance in patterns of herbivory explained by the independent variables of protein and tannin content. All assumptions of normality and homoscedasticity of variances
were tested.
Results
Seasonalpatterns of herbivory
Most herbivory occurred early in the growing season
(Fig. 1), and the patterns of herbivory varied between
1981-1982 and 1982-1983. In the 1981-1982 growing season, virtually all leaf chewing occurred quickly, from late
April to late May (Fig. 1). In 1982-1983, herbivory was
prolonged, occurring from late April to October (Fig. 1);
the two major episodes of herbivory were in spring (late
April-lateMay) and fall (earlySeptemberto late November).
In both years, the primary herbivores were Lepidoptera
(Arctiidae, Noctuidae, Lasiocampidae, Phycitidae, Geometridae, Stenomidae, Tortricidae), Coleoptera (Chrysomelidae and Cucurlionidae), Orthoptera (Acrididae and Tettigoniidae), and Hymenoptera (Tenthredinidae).Sucking insects were not included in this study because the original
concept of quantitativedefense theory was developed from
seasonal patterns of leaf chewers (Feeny 1970, 1975). However, like leaf chewers, sucking insects are also most numerous early in the growing season. Herbivory by vertebrates
did not occur on the study trees.
Seasonal and yearly changes in phytochemistry
In both years, protein content of leaves on the 6 study
trees varied significantlywith time during the growing sea-
36
8'
8-
6-
6
a-
4.
z
4
X
-p
MY
\I
SEPT
JA
MA
V
__.s
2
2
v
.
.
MAY'
1981
.
'SEPT
30- l-
I
20-
0
-
0
3
.
.
I
.
.
JAN
1982
-
*----
v
MAY
.
.
.
.'
.'
.
SEPT'
MAY
1982
.
' '.
JAN
1983
MAY
UNDAMAGED
DAMAGED
INSECT
30-
20-
I
W
g_:
<
=-------
10-
10
."
e-,
_ .»
.
*
- - - -_
_. _
-_
_
°
MAY
1981
JAN '
1982
'SEPT
-----
_
Fig. 2. Seasonal changes in percent relative
protein content of undamaged and insect
damaged leaves in 1981-1982 and 19821983. Points representmeans of 6 trees.
Standarderrors for each point are in
Appendix I
6-
MAY
UNDAMAGED
*.-----
INSECT
DAMAGED
0
.
.
SEPT
MAY
1982
.
.
JAN
1983
.
'MAY
Fig. 3. Seasonal changes in percent
hydrolysabletannin content of undamaged
and insect damaged leaves in 1981-1982 and
1982-1983. Points representmeans of 6
trees. Standarderrors for each point are in
Appendix I
6-
3-
en 4-
4-
z
I
w
2-
0
-
,)sl
-
- - -
- -
-
----
. -_ *
.
-_
--
2-
O*
,/~~~~~~~~~
,Y
0
at0
!
I
MAY'
1981
i'
I
I
SEPT
I II
I
JAN'
1982
I
I
MAY
0
T i
MAY
1982
i
I
I I
SEPT'
i'
i i i
'JAN
1983
son (Fig. 2; 1981-1982, F=22.89, df=6, 30, P<0.001;
1982-1983, F=13.43, df=8, 40, P<0.001). In 1981-1982,
protein content was initially high at budbreak, declined rapidly, and then increased gradually over the growing season.
However, in 1982-83, protein levels were more variable
through the season (Fig. 2). In both years, protein content
of leaves damaged by herbivores was significantly lower
than that of intact leaves (Fig. 2, 1981-1982, F= 17.03, df=
1, 5, P<0.01; 1982-1983, F=25.15, df=1, 5, P<0.01);
herbivory by insects was associated with a decline in nutritional content of leaves.
Hydrolysable tannin content of leaves was initially high
in both years at the onset of budbreak, comprising> 18%
of dry weight of leaves (Fig. 3). Hydrolysable tannin content then declined rapidly (within 30 days) and remained
at relatively constant levels for the remainder of the growing
season. Hydrolysable tannin content varied significantly
with time (1981-1982, F=3.48,
df=6,
30, P<0.01;
1982-1983, F=9.51, df=8, 40, P<0.001) only when the
first sample date was included. If this sample was deleted,
the effect of time became nonsignificant (P>0.05). This
early decline may simply be due to a dilution effect; hydrolysable tannins were probably synthesized before or at budbreak, and concentrations declined only because other leaf
constituents (e.g. lignin, cellulose) were added during leaf
expansion.
This phenomenon may also explain the apparent de-
T
I
I
MAY
Fig. 4. Seasonal changes in percent condensed
tannin content of undamagedand insect
damaged in 1981-1982 and 1982-1983. Points
representmeans of 6 trees. Standarderrors for
each point are in Appendix I
crease in hydrolysabletannin on leaves consumed by herbivores (Fig. 3). This trend was evident in both years (Fig. 3)
but only significantin 1982-83 (1981-1982, F= 1.01, df= 1,
5, P>0.30; 1982-1983, F=25.87, df=l, 5, P<0.01). Decline in hydrolysabletannin content may be part and parcel
of either increases in wound-repaircompounds (Yang and
Pratt 1978) or inducible defenses in damaged leaves, since
hydrolysable tannin content did not vary with time once
leaves were expanded.
Condensed tannin content varied significantlywith time
in each growing season (Fig. 4; 1981-1982, F= 3.36, df=6,
30, P<0.05; 1982-1983, F=14.84, df=8, 40, P<0.001).
Condensed tannin content of insect-damaged leaves was
greater than in intact leaves in both years (Fig. 4;
1981-1982, F=6.67, df=l, 5, P<0.05; 1982-1983, F=
16.35, df= 1, 5, P<0.01). The differencesin condensed tannin content were maintained between insect-damaged and
intact leaves throughout each growing season, despite the
fact little herbivoryoccurredafter the first half of the growing season (Fig. 1).
Water content of undamaged leaves in 1981-1982 and
1982-1983, respectively, was as follows: budbreak 67%,
68%, 1 month after budbreak 54%, 55%; 6 months after
budbreak45%, 46%, and just before abscission 44%, 45%.
Water content of insect-damagedleaves just before abscission was similar to undamaged leaves: 41.6% and 42.7%
in the growing seasons, respectively.
37
100 90-
v 802 7003
>
*0
*
60-
50U 40- 30-
S,
2010-
*
*
V*a
0o
.,*
% PROTEIN
(DRY WT
1981
%
(DRY WT.)
1981
PROTEIN
@g
-
. *.
aa0
s
2
O
6
4
10
8
Fig. 5. Relationship of percent of leaves damaged
by insects in sample periods with relative percent
protein content of leaves in 1981-1982 and 19821983. Each point representspercent of leaves
damaged for an individual tree and corresponding
protein content at a sample date. See text for
estimates, significance,correlationcoefficients, and
R2 of regressions
% PROTEIN (DRY WT) 1982
100Q
90-
C 80: 70C
60: 50w 40o- 30*
aS 20-
I%
0o
·
6*
10-
I
· .
l;
Ir
HYDROLYZABLE
l
TANNINS
I
rb
I
ne
(DRY
1W
981
% HYDROLYZABLE
TANNINS(DRYWT.) 1981
I
w
30
0
5
10
15
20
25
30
Fig. 6. Relationship of percent of leaves damaged
by insects in sample periods with hydrolysable
tannin content of leaves in 1981-1982 and 19821983. Each point representspercent of leaves
damaged for an individual tree and corresponding
hydrolysabletannin content at a sample date. See
text for estimates, significance,correlation
coefficients, and R2 of regressions
% HYDROLYZABLE
TANNINS (DRY WT.) 1982
100 I
-
90-
Fig. 7. Relationship of percent of leaves damaged
by insects in sample periods with condensed tannin
content of leaves in 1981-1982 and 1982-1983.
Each point representspercent of leaves damaged
for an individual tree and correspondingcondensed
tannin content at a sample date. See text for
estimates, significance,correlationcoefficients, and
R2 of regressions
w 80E 70a 60.
u3 50.
* .
40-
*.
30.
S
*
.
20100.
*
*
*·6¶ .
u%
C
0
_
1
D T
% CONDENSED
S (Y
W
TANNINS (DRY WT.)
1
1981
0. 6
2
*
3
4
% CONDENSED TANNINS (DRY WT.) I982
Relationship of leaf phytochemistry with herbivory
Protein content. Amount herbivory (cumulative percent of
damaged leaves) by leaf chewers was significantlyand positively correlated(P< 0.001) with seasonal mean protein levels in leaves in 1981-82 (Fig. 5, regression equation: y=
17.67X-26.75, R2=0.632). However, in 1982-1983,
amount of herbivory was not significantly correlated (P>
0.20) with protein content (Fig. 5).
Hydrolysabletannincontent. Amount of herbivorywas significantly and positively correlated (P< 0.001) with seasonal hydrolysable tannin content in 1981-82 (Fig. 6, regression equation: y = 3.89X-27.84, R2 = 0.882). However, in 1982-1983, this relationship of herbivory with hydrolysable tannin content (Fig. 6) was not significant (P>
0.20).
Condensedtannincontent.Amount of herbivorywas significantly and negatively correlated (P<0.05) with condensed
tannin content in 1981-1982 (Fig. 7, regression equation:
y= -24.95 X+ 57.67, R2 = 0.213). However, because of the
large amount of variation about the regression line, condensed tannin content explainedrelativelylittle of the variation (21.3%) in number of leaves damaged. In 1982-1983,
this relationship(Fig. 7) was no longer significant(P> 0.05)
but was still negative (r, correlation coefficient= -0.27).
Multiple regression analyses
Stepwise multiple regression analyses were performed for
each year, with amount of herbivory during each sample
period as the dependentvariable, and protein, hydrolysable
and condensed tannin as independentvariables.I used mul-
tiple regression because it is evident from the individual
regressions that some independent variables are likely to
be collinear.
Multiple regressionfor herbivory in 1981-1982 showed
hydrolysable tannin content was entered on the first step,
and this variable explained 88.2% of the variation in herbivory (F= 104.46 d.f. = 1, 14, P < 0.01). However, this relationship was positive. Other phytochemical variables entered after hydrolysabletannin content did not significantly
increase amount of variation explained. Examination of residuals, residual plots, and residual statistics showed no
violation of assumptions used in multiple regressionanalyses.
Multiple regressionanalyses for 1982-83 indicated condensed tannin content entered the regressionequation first,
but did not significantlyaccount for variation in herbivory
(F=2.35, d.f.=1, 23, P>0.10). Other phytochemical variables entering the equation after condensed tannin content
did not significantlyincreaseamount of variationexplained.
Examination of residuals revealed no violation of assumptions.
Discussion
Herbivory patterns were variable from year to year. About
30% of available leaves were damaged by herbivores, with
differentseasonal patternsbetweenthe two growing seasons
(Fig. 1). Composition of the insect fauna did not noticeably
vary between years. However, little leaf area was removed
(<5% of total area). Edwards and Wratten (1983) found
similarpatternsin severalplant species, i.e., few leaves completely consumed but a large proportion of leaves with some
38
damage. Virtually all herbivory in 1981-82 occurs within
30 days of budbreak, while in 1982-83, the majority of
herbivory occurs between budbreak in late April October.
In fact, the largest increase occurred between late August
and mid-October.
Seasonal trends in phytochemistry do not conform to
Feeny's (1970) results that protein content increases and
tannin content declines over the growing season. Feeny and
Bostock (1968) reported that hydrolysable tannin content
remains fairly constant throughout the growing season.
Hydrolysable tannin content in Q. emoryi was highest at
budbreak (Fig. 3), and then rapidly declines to remain at
constant levels for the remainder of the growing season.
Some tropical tree species also have higher concentrations
of phenolic compounds in new leaves than old leaves (Milton 1979; Oates et al. 1980). The decline in Q. emoryi was
likely a dilution effect resulting from addition of other leaf
constituents as the leaf constituents as the leaf expands;
nevertheless, insects feeding just after budbreak must still
contend with a relative high hydrolysable tannin content
per unit of leaf ingested.
Condensed tannins do not increasein a monotonic fashion demonstrated by Feeny (1970) for Q. robur and for
other perennials (Lawton 1976; Dement and Mooney
1974). Condensed tannins increased only in the first 2 or
3 months, then leveled off for the remainderof the season
in both years.
Feeny (1970) showed protein levels decline monotonically through the growing season, and felt that protein levels are at least partly responsible for early feeding by oak
insects. Protein levels of Q. emoryi were high at budbreak
in 1981-1982, declined rapidly and then increased monotonically for the rest of growing season until protein content
at the latter part of the season approached levels at budbreak (Fig. 2). In 1982-1983, changes in protein content
were much more variable, with highest levels found at mid
and late season (Fig. 2). Discrepancies in seasonal trends
in protein between Q. robus and Q. emoryi may partially
result from different assay methods. Kjeldahl (Feeny 1970)
is an indirect measurementof protein, since protein is estimated from organic nitrogen content which includes nonproteinaceous organic nitrogen (Bradstreet 1965; Feeny
1970). BioRad (this study) is a direct measure of number
of peptide bonds. When BioRad is compared to Kjeldahl
for Emory oak, there are large discrepanciesin relativeprotein, especially early in the growing season (Faeth unpublished data). Clearly, however, protein content, which is
considered a key nutrient for insects (McNeill and Southwood 1978; Mattson 1980; Scriberand Slansky 1981), did
not decrease monotonically through the season, and was
not consistently related to seasonal herbivory patterns as
predicted by the theory of quantitative defense.
Inducible chemical changes in herbivore-damaged leaves
Condensed tannin content was markedly and significantly
greater in damaged leaves than intact leaves in both years
(Fig. 4). Protein and hydrolysabletannin content were generally lower in damaged than in intact leaves, but it is not
clear if these decreases were simply caused by a relative
increase in condensed tannin. The increase in condensed
tannins suggests they are induced by Q. emoryi in response
to herbivory. It could be argued that insects on oaks are
simply selectively feeding on leaves with lower protein or
hydrolysable tannin content or with increased condensed
tannin content. However, other experiments (Faeth 1986)
have clearly shown that these same chemical changes occur
after insect or mechanicaldamage to leaves.
Induced response to herbivory has been documented
in other plants (Haukioja and Niemela 1979; Ryan 1979;
Rhoades 1979; Carroll and Hoffman 1980; Bryant 1981;
Schultz and Baldwin 1982; Baldwin and Schultz 1983).
Schultz and Baldwin (1983) reportedhigherlevels of hydrolysable and condensed tannins in leaves of red oak trees
that were previouslydefoliated or currentlybeing defoliated
by gypsy moths. This response was short-lived, peaking
about 20 days after defoliation. Contrary to these results,
hydrolysable tannins did not increase in damaged leaves,
although sampling periods may have been too infrequent
to detect short term increases.Also, induction in condensed
tannin content in damaged leaves persisted throughout the
growing season in both years (Fig. 4). This persistence of
elevated condensed tannins in damaged leaves occurreddespite the fact that the same leaves were not repeatedlydamaged throughout the year.
Herbivoryand chemicaldefense
Regression analyses of amount of herbivory with phytochemical changes in Q. emoryi do not generally support
the predictions of the theory of quantitative defense in apparent plants (Feeny 1975, 1976; Rhoades and Cates 1975).
While protein content was positively correlatedwith herbivory in 1981-1982 as predictedby the theory, this relationship was not significant in 1982-83. Similarly, the theory
of quantitative defense predicts a negative correlation in
timing of herbivory with tannin levels. This prediction is
fulfilled in 1981-1982 for condensed tannin content, but
not in 1982-1983. Moreover, while the regression of condensed tannin and amount of herbivory was significantly
negative in 1981-1982, it explained only a small amount
of variability (21%). Finally, hydrolysable tannin content
was significantly and positively correlated with herbivory
in 1981-1982, a result directly contrary to the predictions
of quantitativedefense theory.
In general, overall changes in leaf chemistry were poor
predictorsof seasonal patterns in herbivory. In 1981-1982,
hydrolysabletannin content explainedmost of the variation
(88%) in herbivory,but this was a positive correlation and
contradicts predictions of the theory. Other phytochemical
variablesdid not significantlyincreasethe amount of variation explained. In 1982-83, no single chemical variable or
combination of variables explained significant variation in
herbivoryintensity.
The generalityof quantitative defensive theory must be
seriously questioned. Traditionally-held quantitative defenses such as tannins are not consistently good indicators
of deterrence to herbivores: in fact, tannin content may
vary with herbivoryin a fashion opposite to their purported
role as defensive chemicals. Karban and Ricklefs (1984)
found that lepidopterandiversitiesand abundancesare not
correlatedwith either tannin or nutrientcontent of various
tree species. Coley (1983) also did not find much support
for the theory of quantitative defense theory in examining
the relationshipof herbivory to phenolic and tannin levels
in tropical trees. Coley (1983) did find that leaf toughness
and fiber content were negatively correlated, while water
content was positively related with herbivory. Leaf tough-
39
ness was low and water content of Q. emoryi was high
(67%) in 1981-1982 when most herbivory occurred (Fig. 1).
But in 1982-1983, a large fraction of herbivory occurred
in the fall when water content had declined to near lowest
levels and leaf toughness should have been high.
Tannins can be harmful or deterrent to individual insect
species on specific plants. Zucker (1983) argued various
tannins are likely very specific in their mode of action
against herbivores and pathogens, based on their stereochemical heterogeneity. Bernays (1981) argued against the
idea of tannins as quantitative, generalized defenses based
on conflicting effects of tannins on insects in bioassay studies. The original notions of tannins as quantitative defenses in apparent plants is also suspect based solely on
seasonal and yearly patterns of tannins and herbivory, upon
which the theory was originally based. Phytochemical changes alone are inadequate to explain herbivory patterns,
especially if variability in herbivory and chemistry from
year to year is considered.
Alternativehypotheses
Despite some year-to-yearvariability in herbivory patterns
(Fig. 1), it is clear that insect herbivores consume a large
fraction of leaves in the spring and sometimes in the fall.
My results indicate phytochemical changes alone cannot
explain this pattern. What then are possible explanations
for the pattern of spring and occasional fall feeding of insects on oak that was originally shown by Feeny (1970)
and documented here? Holmes et al. (1979) suggested intensity of bird predation in midsummer may force lepidopterous insects to feed early in spring or in fall on northeastern
trees. Two other hypotheses may explain the patterns of
feeding by leaf-chewing insects on Q. emoryi:
the spring to avoid leaves previously damaged by other
herbivores.Since herbivoresseem to be generallyunresponsive to phytochemical changes that occur seasonally, it is
doubtful that damaged leaves are avoided because of direct
toxic effects of increased tannins. However, predators
(Heinrich and Collins 1983) and parasitoids (Vinson 1975;
Faeth 1985, 1986) use physical or chemical changes in damaged leaves to locate insect hosts. Risks from intensified
attack by predators and parasites associated with damaged
leaves may have selected for early feeding by insects.
Certainly, other factors may also play a role in patterns
of insect herbivoryon long-lived plants (e.g., leaf toughness,
water content, or variable predator and parasite attack).
I am not claiming that seasonal changes in phytochemistry
and nutrientsin long-lived plants have no effect on seasonal
patterns of herbivory, only that these changes do not, in
themselves, explain herbivory patterns. Responses of each
insect species to specific chemical and phenological changes
in "apparent" plants may account for some variation in
herbivory. However, sufficient evidence has accumulated
such that quantitative defenses alone cannot be considered
the only explanation for these patterns.
Acknowledgments.I thank M. Axelrod, G.C. Faeth, T.G. Faeth,
T.J. Faeth, N.J. Tiller, and W.T. Tiller for assistance in the field.
M. Axelrod, A. Gieschen, and C. Schilling diligently performed
many of the phytochemical analyses in the laboratory. I am
especiallygratefulto W.V. Zuckerfor his help in developing tannin
and protein assay procedures. M. Auerbach, T.L. Bultman, P.C.
Coley, J. Schultz, D. Simberloff, D.R. Strong, Jr. and M. Taper
made many helpful comments on the manuscript.The U.S. Forest
Serviceand the Centerfor EnvironmentalStudies (ASU) generously provided access and accomodations at the SierraAncha Experimental Station. This researchwas supported by a Faculty GrantIn-Aid from ASU and NSF grants BSR-8110832 and BSR8415616.
1. Early feeding may be partially linked with temperature
requirements and life histories of chewing insects and Q.
emoryi. In the study area, temperaturesincreased rapidly
after May to highs of - 38° C, which may prohibit development or decrease fecundity of leaf-chewing insects. In
1981-1983, most feeding was concentrated just after budburst from May to June, but before hot, dry conditions
in June through mid-July (Fig. 1). In 1982-1983, most feeding again occurred in the same interval, and also from Sep-
tember to November, when temperatures subsided. This
second interval of feeding may reflect feeding by diapaused
larvae or adults, or successful later generations of spring
feeders. It is interesting to note that temperatures were
above normal and precipitation below normal in the summer of 1981, while in 1982 both temperatures and precipitation were at normal levels (National Weather Service). Temperature and precipitation may partially explain the lack
of late summer and fall feeding in 1981 if these factors
caused low survivorship during the early summer months.
Additional support for this hypothesis is that leaf-mining
insects on Q. emoryi, which are bufferedfrom external temperatures and changes in humidity since they are enclosed
in leaf tissues (Hering 1951; Faeth et al. 1981), begin feeding
in late June and continue through fall and winter months
(Faeth 1985, 1986).
2. Patterns of herbivory on Q. emoryi may be related to
chemical and physical changes that occur in damaged leaves
(Figs. 2-4). Many herbivores may concentrate feeding in
AppendixI. Standard errors of means for Figs. 2, 3, and 4. Standard errors are not shown on graphs because of some overlap
of S.E. bars
Date
Protein
Hydrolysable
tannin
Condensed
tannin
intact
damaged
intact
damaged
intact
damaged
4-30-81
5-31-81
7-9-81
8-2-81
9-5-81
10-10-81
11-7-81
4-11-82
0.75
0.42
0.93
0.65
0.52
0.62
0.71
0.24
0.81
0.78
1.01
0.63
0.59
0.70
0.87
0.31
2.57
1.30
2.36
2.31
0.98
1.37
2.40
1.21
1.41
2.03
1.74
1.47
2.57
1.84
3.01
0.26
0.43
1.32
1.37
1.42
1.19
1.25
1.23
0.36
1.24
1.39
1.63
1.45
1.48
1.43
4-30-82
5-19-82
6-16-82
7-14-82
8-18-82
9-11-82
10-7-82
11-19-82
2-26-82
11-26-83
0.62
0.86
1.36
0.82
0.77
0.59
1.05
0.76
1.05
0.56
0.38
0.73
0.74
0.50
0.96
0.51
0.83
0.96
1.44
0.37
4.73
1.85
1.95
1.67
1.38
1.73
1.83
1.31
1.49
6.30
1.76
1.59
2.43
1.56
1.57
1.99
1.31
2.00
0.40
0.75
0.75
0.81
0.86
0.57
0.87
0.66
0.76
0.45
0.54
0.86
1.76
1.42
0.98
0.59
1.11
1.51
40
References
BaldwinIT, Schultz JC (1983) Rapid changes in tree leaf chemistry
induced by damage: evidence for communication between
plants. Science 221:277-279
Bate-Smith EC (1977) Astringent tannins of Acer species. Phytochemistry 16:1421-1426
BerenbaumMR (1983) Effects of tannins on growth and digestion
in two species of papilionids. Entomol Exper et Appi
34:245-250
Bernays EA (1978) Tannins: an alternative viewpoint. Entomol
Exper et Appl 24:244-253
BernaysEA (1981) Plant tannins and insect herbivores:an appraisal. Ecol Entomol 6:353-360
Bernays EA, ChamberlainDJ, McCarthyP (1980) The differential
effects of ingestedtannic acid on differentspeciesof Acridoidea.
Entomol Exper et Appi 28:158-166
BernaysEA, Woodhead S (1982) Plant phenols utilizedas nutrients
by a phytophagous insect. Science 216:201-202
BradstreetRB (1965) The Kjeldahl method for organic nitrogen.
Academic Press, NY
Broadhurst RB, Jones WT (1978) Analysis of cindensed tannins
using acidified vanillin. Sci Food Agric 29:788-794
Bryant JP (1981) Phytochemical deterrence of snowshoe hare
browsing adventitious shoots of four Alaskan trees. Science
213:889-890
Carroll CR, Hoffman CA (1980) Chemicalfeeding deterrentmobilized in response to insect herbivory and counteradaption by
Epilachnatredecimnota.Science 209:414-416
Coley PD (1983) Herbivory and defensive characteristicsof tree
species in a lowland tropical forest. Ecol Mono 53:209-233
Dement WA, Mooney HA (1974) Seasonal variationin production
of tannins and cyanogenic glycosides in the chaparral shrub,
Heteromelesarbutifolia.Oecologia (Berlin) 15:65-76
EdwardsPJ, WrattenSD (1983) Wound induceddefencesin plants
and their consequencesfor patternsof insect grazing.Oecologia
(Berlin) 59:88-93
Faeth SH (1985) Host leaf selection by leaf-mininginsect, Stilbosis
juvantis: interactions among three trophic levels. Ecology
66:870-875
Faeth SH (1986) Indirect interactions between temporally-separatedherbivoresmediatedby changes in the host plant. Ecology
(in press)
Faeth SH, Mopper S, Simberloff D (1981) Abundancesand diversiy of leaf-mining insects on three oak host species: effects of
host-plant phenology and nitrogen content of leaves. Oikos
37:238-251
Feeny PP (1968) Effect of oak leaf tannins on larval growth of
the winter moth Operophtera brumata. Insect Physiol
14:805-817
Feeny PP (1970) Seasonal changes in oak leaf tannins and nutrients
as a cause of springfeeding by wintermoth caterpillars.Ecology
51:565-581
Feeny PP (1975) Biochemicalcoevolution between plants and their
insect herbivores.In: LE Gilbert, PH Raven (eds) Coevolution
of animals and plants: Symposium V. First InternationalCongress System. and Evol Biol. University of Texas Press, Austin,
Texas pp 3-19
Feeny PP (1976) Plant apparency and chemical defense. In: JW
Wallace, RL Mansell (eds) Biochemical interactions between
plants and insects. Recent Adv Phytochem 10:1-40
Feeny PP, Bostock H (1968) Seasonalchanges in the tannincontent
of oak leaves. Phytochem7:871-880
Fox LR (1981) Defense and dynamics in plant-herbivoresystems.
Amer Zool 21:853-864
Haukioja E, Niemela PE (1979) Birch leaves as a resource for
herbivores: seasonal occurrenceof increased resistance in foliage after mechanical damage of adjacent leaves. Oecologia
(Berlin)39:151-159
Heinrich B, Collins SL (1983) Caterpillar leaf damage and the
game of hide-and-seekwith birds. Ecology 64:592-602
Hering EM (1951) Biology of the leaf miners.Junk, Gravenhages,
The Netherlands
Holmes RT, Schultz JC, Nothnagle PJ (1979) Bird predation on
forest insects: an exclosureexperiment.Science 206:462-463
Karban R, Ricklefs RE (1984) Leaf traits and species richness
and abundance of Lepidopteranlarvae on deciduous trees in
southernOntario. Oikos 43:165-170
Lawton JH (1976) The structureof the arthropod community on
bracken. Botan J Linn Soc 73:187-216
Mattson WJ (1980) Herbivoryin relationto plant nitrogencontent.
Ann Rev Ecol System 11:119-161
McNeill S, Southwood TRE (1978) Role of nitrogenin the development of insect-plant relationships.In: JB Harborne (ed) Biochemical aspects of plant and animal coevolution. Academic
Press, NY, pp 77-98
Milton K (1979) Factors influencing leaf choice by howler monkeys: a test of some hypotheses of food selection by generalist
herbivores.Amer Nat 114:362-378
Oates JF, Waterman PG, Choo GM (1980) Food selection by
the south Indian leaf-monkey, Presbytisjohnii, in relation to
leaf chemistry.Oecologia (Berlin)45:45-56
Rhoades DF (1979) Evolution of chemical defense against herbivores. In: GA Rosenthal, DH Janzen (eds) Herbivores.Their
interactionswith secondaryplant metabolites.Academic Press,
NY, pp 3-54
Rhoades DF, Cates RG (1976) Toward a general theory of plant
antiherbivorechemistry.Rec Adv Phytochem 10:168-213
Ryan CA (1979) Proteinase inhibitors. In: GA Rosenthal, DH
Janzen (eds) Herbivores. Their interactions with secondary
plant metabolites.Academic Press, NY, pp 599-618
SchultzJC, BaldwinIT (1982) Oak leaf quality declinesin response
to defoliation by gypsy moth larvae. Science 217:149-151
ScriberM, Slansky F (1981) The nutritionalecology of immature
insects. Ann Rev Entomol 26:183-211
Vinson SB (1975) Biochemicalcoevolution betweenparasitoidsand
theirhosts. In: PW Price(ed) Evolutionarystrategiesof parasitic insects and mites. Plenum Press, NY, pp 14-48
Winer BJ (1971) Statistical principles in experimental design.
McGraw-Hill,NY
Yang SF, Pratt HK (1978) The physiology of ethylene in wounded
plant tissues. In: G Kahl (ed) Biochemistryof wounded plant
tissues. de Gruyter, Berlin,pp 595-622
Zucker WV (1983) Tannins: does structure determine function?
An ecological perspective.Amer Nat 121:335-365
Received March 20, 1985