PLANT-INSECT INTERACTIONS
White-Tailed Deer Alter Specialist and Generalist Insect Herbivory
Through Plant Traits
ERIC M. LIND,1,2 EMILY P. MYRON,1,3 JENNIFER GIACCAI,4 AND JOHN D. PARKER1
Environ. Entomol. 41(6): 1409Ð1416 (2012); DOI: http://dx.doi.org/10.1603/EN12094
ABSTRACT Within a plant species, leaf traits can vary across environmental, genetic, spatial, and
temporal gradients, even showing drastic differences within individuals. Herbivory can also induce
variation in leaf morphology, defensive structure, and chemistry including nutritional content. Indi-
rect effects ofprior insectherbivoryon laterherbivoreshavebeenwell documented, but the induction
of trait changes after vertebrate herbivory has been little explored. Here, we examined how browsing
of spicebush (Lindera benzoin L.), a dominant understory shrub in eastern mesic forests, by white-
taileddeer (Odocoileus virginianusL.) alteredplant quality and subsequent foliar herbivoryby insects.
Browsing history explained!10% of overall leaf trait variation; regenerated leaves had greater water
contentandspeciÞc leaf area(P"0.009),butwere lower innitrogenandgreater incarbon(P#0.001),
than leaves on unbrowsed plants. However, browsing did not shift terpene chemistry as revealed by
GC-MS. In the lab, caterpillars of the specialist spicebush swallowtail (Papilio troilus L.) preferred
(P " 0.02) and grew 20% faster (P " 0.02) on foliage from browsed plants; whereas total herbivory
in the Þeld, including generalist insect herbivory, was twice as high on unbrowsed plants (P" 0.016).
These results suggest that the ecological impacts of deer in forest understories can have cascading
impacts on arthropod communities by changing the suitability of host-plants to insect herbivores.
KEY WORDS indirect interactions, induced defenses, leaf traits, terpenes, vertebrate herbivory
Leaf traits often determine host-plant suitability for
foliar herbivores. Preference and performance of her-
bivorous insects, for example, can hinge on leaf nu-
tritional content, especially the relative concentration
of carbon and nitrogen (Mattson 1980, Awmack and
Leather 2002, Behmer 2009). Other leaf traits can also
inßuence suitability, including leaf morphological
traits like toughness, thickness, and trichome density
(Agrawal and Fishbein 2006); chemical defenses
(Broadway and Duffey 1988); and chemical signaling
to higher trophic levels (Kessler and Baldwin 2001).
Given the overwhelming importance of plant tissue
traits in mediating herbivory, a major challenge for
understanding insect community structure is the spa-
tial and temporal heterogeneity of plant tissue quality.
Spatially, soil conditions (Wright et al. 2010) and light
environment (Osier and Jennings 2007, Niesenbaum
and Kluger 2006, Mooney et al. 2009) can alter plant
nutrient content and defensive compounds, resulting
in substantial variation in plant palatability in space.
Even variation among leaves on an individual plant
can affect insect herbivore growth and survival (Ro-
slin et al. 2006) and oviposition choice (Kessler and
Baldwin 2002). Temporally, the seasonality of leaf
quality (Schultz et al. 1982) and ontogenetic devel-
opment of a plant (Boege and Marquis 2005) can also
have signiÞcant impacts on insect preference and per-
formance.
Onekey sourceof leaf trait variation thatvariesboth
spatially and temporally is induction, or changes ini-
tiated after herbivory (reviewed in Karban 2011).
Typically, induction of defensive chemicals reduces
future susceptibility of the plant, resulting in detri-
mental effects on subsequent herbivores, and is thus a
net beneÞt to plant Þtness (Agrawal 1998). In other
cases, induction of nutrients toward a wound can ac-
tually increase future herbivory by making leaf tissue
more palatable (Kaplan et al. 2010). Importantly, ev-
idence suggests that these induced traits can have
asymmetric impacts on different herbivores, leading
to trait-mediated indirect interactions among herbi-
vores co-occurring on the same host plant (Van Zandt
and Agrawal 2004, Ohgushi 2005).
These types of indirect interactions via induced
defenses often are studied among insects co-occurring
on a host-plant, but indirect interactions among her-
bivores need not co-occur in space or time. For in-
stance, vertebrateherbivores canhave large effects on
plant abundance and community composition (Mc-
Shea et al. 2007), and studies have shown subsequent
impacts on the distribution and performance of plant-
1 Corresponding author: Eric M. Lind, Smithsonian Environmental
Research Center, 647 Contees Wharf Rd., Edgewater, MD 21037
(e-mail: elind@umn.edu).
2 Current address:DepartmentofEcology,Evolution andBehavior,
UniversityofMinnesota, 100EcologyBldg., 1987UpperBufordCircle,
St. Paul, MN 55108.
3 Nicholas School of the Environment, Duke University, Durham,
NC 27708.
4 SmithsonianMuseumConservation Institute, Suitland,MD20746.
0046-225X/12/1409Ð1416$04.00/0 ! 2012 Entomological Society of America
associated insects, including beaver browsing on cot-
tonwood (Martinsen et al. 1998), moose browsing on
willows (Den Herder et al. 2004), and deer browsing
onavarietyofwoodyplants (Barrett andStiling2007).
These indirect effects often have focused on chemical
defenses (e.g., higher concentrations of toxins), but
nutritional changes because of regrowing tissuemight
be an important component of indirect effects medi-
ated by induction.
The objective of this study was to quantify plant-
mediated indirect interactions between early season
(AprilÐMay) vertebrate and later-season (JuneÐJuly)
insect herbivores on a shared host plant. We were
especially interested in the potential impacts ofwhite-
tailed deer, an herbivore that has undergone a recent,
dramatic increase in abundance (Coˆte´ et al. 2004).
Here, we examined whether extensive deer browsing
on a common understory shrub, spicebush (Lindera
benzoin L.), indirectly altered insect communities by
inducing changes in host plant tissue. We asked the
following speciÞc questions: 1) Does deer browsing
induce chemical and physical changes to leaf tissue?
2)Do inducedchanges result inmoreor lessherbivory
bygeneralist chewing insectherbivores in theÞeld?3)
Do deer-induced changes in leaf quality result in dif-
ferential preference and performance by a specialist
insect herbivore?
Materials and Methods
StudyArea andOverview.Researchwas conducted
at the Smithsonian Environmental Research Center
(SERC), a 2650-acre research preserve located on the
western shore of the Chesapeake Bay (38.8860$ N,
%76.5500$ W), MD. Forests are primarily secondary
mesic forest growth 75Ð120 yr of age after agricultural
abandonment (Parker et al. 2010). The forest under-
story at SERC contains &200 native and non-native
plant species, but spicebush (L. benzoin) is the most
common understory shrub. Within a 16-ha mapped
Forest Dynamics Plot at SERC, spicebush comprises
41% of all woody plant stems &1 cm in diamter at
breast height, occurring at a density of 850 stems ha%1
(J.D.P., unpublished data). White-tailed deer (Odo-
coileus virginiana) occur in estimated densities of !4
km%2 on and around the SERC property (J.D.P., un-
published visual counts from helicopter surveys with
infrared imaging), which is in agreement with esti-
mated historical densities (McCabe and McCabe
1997). At SERC, deer browse spicebush primarily as
spicebush ßowers are senescing and leaves are ßush-
ing, leaving distinctive tearing marks on leaves and
stems (E.M.L., unpublished data).
We conducted studies over two growing seasons to
quantify the frequency and intensity of deer browse
on spicebush, its impacts on leaf traits including ter-
pene chemistry, and the response of specialist and
generalist insects. In summer 2008, we sampled a
500-m2 spicebush patch for browsing, as well as
marked two separate 500-m transects of spicebush
plants fromwhichwe sampled leaves for trait analysis,
labherbivory assays, and insect herbivory rates. In late
winter of 2009, we fenced a separate set of spicebush
plants to prevent browse by deer, and then, in the
summer of 2009, we collected leaves from these and
paired browsed plants for analysis of terpene chem-
istry. Finally, we used data gathered on browsing rates
from a separate study in the summers of 2010Ð2011 to
bolster our estimates of deer browse rates on spice-
bush. All studies were conducted within closed can-
opy secondary forest within 2-km of the geographic
coordinates above (see previous paragraph).
Frequency and Intensity of Deer Browsing on
Spicebush.Toquantify theextentand intensityofdeer
browse on spicebush plants at SERC, and thus the
potential for indirect impacts on insect herbivores, we
conducted two surveys of spicebush in the forest un-
derstory. In June of 2008, we established a 50-m by
10-m grid with nodes spaced every 5 m. At each node,
the nearest spicebush plant was measured for height
of tallest growing stemand largest diameter of canopy.
We then counted each live stem, noting which, if any,
had evidence of deer browsing. The frequency of deer
browsing on a plant was determined by dividing the
number of stems having at least one leaf or meristem
with evidence of deer herbivory by the total number
of stems on that plant. To determine the spatial extent
of this herbivory, in 2010, we recorded the same met-
ric of deer browsing on all spicebush plants#2 m tall
found inN" 489 1-m2 quadrats located systematically
&15 m apart in a nearby 16-ha mapped Forest Dy-
namics Plot. In 2011, we revisited 186 of these spice-
bush plants and scored evidence of any browsing to
assess whether browsing in 1 yr was predictive of
browsing in the next year, which might suggest that
deer were targeting particular plants.
Consequences of Browsing on Plant Leaf Traits. To
determine whether deer browsing had indirect im-
pacts on foliar insects, we quantiÞed insect herbivory
and leaf chemical andmorphological traits onbrowsed
versus unbrowsedplants.WeÞrst identiÞed spicebush
plants with evidence of current-year browse (freshly
injured stems and leaves, typically 30Ð50% of growing
stems), in early June 2008, along two transects!500m
in length located in closed secondary forest canopy.
We found and tagged a total of 120 browsed plants
along both transects combined. Selected plants were
fully accessible todeer (!2mtall).We then found the
nearest unbrowsed (lacking signs of browsing in the
current season) spicebush plant, approximately
matched for size, within 2Ð5 m of each of the tagged
plants.
To quantify how deer browsing altered host-plant
quality, we measured a series of leaf traits on a subset
of thesepairedplants in late Juneof 2008.Weremoved
by hand and collected 10 fully expanded, terminal
leaves fromeachof thebrowsedandunbrowsedplants
from 30 haphazardly chosen pairs. Leaves from
browsed plants included only those that grew after
earlier browsing during the same growing season.
Leaveswereplaced inplasticbags inacooler, and then
taken within 2 hr to the lab where we immediately
measured leaf traits. We quantiÞed leaf toughness as
the grams of force required to pierce a fully expanded
1410 ENVIRONMENTAL ENTOMOLOGY Vol. 41, no. 6
leaf below the inßorescence by using a force gauge
penetrometer, type 516 (Chatillon, Largo, FL). We
calculated speciÞc leaf area (SLA) as the area of a leaf
(centimeters squared as measuredwith a LP-80 hand-
held leaf area meter, LiCor, Lincoln, NE) divided by
its dry mass (after 24 h in a drying oven at 60$C). We
estimatedpercentwater content as ([freshmass%dry
mass]/fresh mass)*100. We analyzed a subset of
leaves from each plant for percentage carbon (C),
nitrogen (N), and phosphorus (P). We Þrst Þlled a
plastic 2-ml centrifuge tube with dried leaf tissue and
a stainless steel ball, and ran samples for 5min in a ball
mill grinder until samples were a Þne powder. We
calculated percent C and N by using a!2-mg sample
of ground plant powder in an EAI CE-440 elemental
analyzer (Exeter Analytics, Coventry, United King-
dom). We determined percent P by placing a known
mass (!2mg)ofdried, ground leafmaterial in amufße
furnace at 550$C for 2 hr (Miller 1998), followed by
colorimetric analysis on a microplate spectrophotom-
eter (PowerWave XS; Biotek,Winooski, VT) by using
the ammonium molybdate method (Clesceri et al.
1998).
Measuring leaf quality for herbivores requires syn-
thesizing the multivariate, collinear nature of many
leaf traits (Cornelissen et al. 2003, Wright et al. 2004).
Because wewere interested in quantifying the overall
multivariate change in leaf characteristics, although
not assuming the functional response of insect herbi-
vores, we used principal components analysis (PCA,
Legendre and Legendre 1998) to identify indepen-
dent axes of variance in leaf traits (Wright et al. 2004).
These axes scores then were used independently as
response variables in linear models, with browsing
history as a Þxed factor and with pair as a random
factor to ask whether trait combinations differed by
recent browsing.We also tested the amount of overall
variance in leaf traits explained by current-year
browsing by using redundancy analysis (Legendre
and Legendre 1998), which constrains one axis of a
PCA to align with the treatment of interest. All leaf
trait values were standardized to a common scale
(converted to z-scores) before analysis.
To assess changes in plant secondary chemistry in-
duced by deer browsing, we used 1.22 m (4-ft)
chicken wire fencing to exclude deer from 20 small
(#2 m) spicebush plants in separate patches, distinct
from those used in the leaf trait analysis. Caging pre-
ceded spring leaf ßush and the onset of browsing,
although some plants were in ßower. In July of 2009,
wecollected!50gof leaves fromeachprotectedplant
and from a nearby, similarly-sized plant that had ev-
idenceof current-yearbrowse. Immediatelyuponhar-
vesting, we cut the leaves with shears directly into a
vialÞlledwithdichloromethaneandkept samples cool
until placing them in a freezer (%20$C). In February
of 2010, we analyzed the leaf extract samples by using
gas chromatography-mass spectrophotometry (GC-
MS), using themethods of Fine et al. (2006). A known
standard not naturally occurring in the plant tissue
(tetradecane)was addedas an internal control toeach
sample at a concentration of 1 "l/ml. Samples were
analyzed using an Agilent 6890NGC and Agilent 5975
InertMS detector (Agilent Technologies, SantaClara,
CA).Twomicroliters of sample solutionwere injected
into a 250$C-inlet by using splitless injection. Samples
were run on an HP-5MS column (30 m by 0.5 "m by
250 "m) programmed from 30 to 250$C at 10$C/min
holding at theÞnal temperature for 3min at a constant
ßow of 1 ml/min. We compared observed peaks with
databases of identiÞed compounds in the National
Institute of Standards and TechnologyÕs NIST11 li-
brary by using AMDIS software version 2.70 and with
known concentrations of commercial terpene refer-
ence compounds (limonene, pinene, and caryophyl-
lene oxide). The abundance of compounds was esti-
mated by the area of detected peaks scaled by the
tetradecane standard. We used PCA and randomiza-
tion tests to partition variance in the abundance of
detected compounds and test for differences between
browsed and unbrowsed plants.
Impacts of Deer Browsing on Foliar Insect Her-
bivory. To quantify how patterns of insect herbivory
changed after deer browsing, we collected all leaves
along a major branch (n" 20Ð40 leaves from branch
originating in amain stem) from 25 pairs of previously
marked plants (not used in the trait collections). We
ßattened the leaves in a plant press until dry, and then
photographed each sheet of leaves with a scale bar.
These leaves then were scanned and analyzed using
image processing software (SigmaScan, Systat Soft-
ware, San Jose, CA) to quantify herbivory as the area
of leaf missing. The percent of leaf area missing was
analyzed using a linear mixed effects model with
browsing as a Þxed factor, the area of the intact leaf as
a random covariate and with error nested by leaf
within plant within pair.
Impacts of Deer Browsing on a Specialist Insect
Herbivore.We also examinedwhether deer browsing
would affect an herbivore that specializes on spice-
bush. The spicebush swallowtail butterßy, Papilio troi-
lusL. (Lepidoptera: Papilionidae), is foundcommonly
in deciduous lowland forests in eastern North Amer-
ica, where it feeds primarily on spicebush and the
confamilial sassafras (Sassafras albidumNutt.Nees) in
the Lauraceae (Wagner 2005). At SERC, spicebush
swallowtails are uni- to bivoltine, with adults Þrst ap-
pearing in ßight in late June and caterpillars appearing
shortly thereafter,where they feedon the leavesof the
plant on which they were laid. Sassafras is relatively
uncommon at SERC (J.D.P. et al., unpublished data),
and thusmostP. troilus caterpillars are foundon spice-
bush.
To determine the potential impact of deer browsing
on P. troilus,we examinedwhether P. troilus exhibited
preference and performance differences between
browsed versus unbrowsed spicebush plant tissues. In
early July of 2008, we collected Þrst- to third-instars
from shelters on spicebush plants outside of the ex-
perimental transect areas and brought them to the lab.
Before use in feeding assays, we kept individual cat-
erpillars in plastic containers (six oz deli containers,
Solo Cup Co., Lake Forest, IL) with moistened Þlter
December 2012 LIND ET AL.: TRAIT-MEDIATED EFFECTS OF DEER ON INSECT HERBIVORES 1411
paper and fresh leaves (Þeld-collected from un-
browsed spicebush plants) changed every other day.
To test the preference of caterpillars for browsed
versus unbrowsed spicebush leaves, we conducted
choice feeding assays in a growth chamber (constant
temperature of 25$C,with a photoperiod of 16:8 [L:D]
h). Before each assay, we starved caterpillars for!12
h and recorded their fresh mass to the nearest milli-
gram. Then, we offered each individual (n" 30) one
fully expanded, entire leaf from each of a browsed
and unbrowsed plant in a pair and allowed it to feed
ad libitum for 24 h. Before and after the trial, we
recorded the area of each leafwith the leaf areameter.
Weusedanalysis of covariance to test for the inßuence
of browsing (Þxed effect) on the amount of leaf area
consumed, using the initial area of the leaf and initial
mass of the caterpillar as covariates.
We estimated the performance of P. troilus cater-
pillars on deer-browsed and unbrowsed spicebush
plants by randomly assigning caterpillars to be reared
solely on leaves from browsed (n" 18) or unbrowsed
(n" 17) spicebush plants. Before beginning the trial,
we recorded themass and instar of caterpillars we had
collected in the Þeld and then placed each in a deli
container with moistened Þlter paper and one or two
fully expanded leaves selected from previously
marked plant pairs not used in the trait or herbivory
trials. Leaves were collected from 20 to 30 plants at a
time andmixed so that caterpillars experienced leaves
from multiple host plants, but only one browsing his-
tory, across their development. Caterpillars were kept
in a growth chamber that we set at conditions as
described above. Every second or third day, we
changed leavesandrecorded the instar andmassof the
caterpillar. Relative growth rate for each caterpillar
was calculatedas [ln(mass2/mass1)/(time2% time1)],
which gives an estimate of growth in d%1. Because
these Þeld-collected caterpillars entered the experi-
ment at different points (from Þrst to third instars at
the beginning of the trial), we took the mean of the
relative growth rate values across the rearing trial for
each caterpillar as one measure of performance. We
recorded pupalmass on the second day after pupation
as a second measure of performance. We used linear
models toanalyzeeachperformancemetric separately
with browsing history of the plant as the Þxed factor;
in the pupal mass analysis the initial mass of the cat-
erpillar was used as a random covariate.
We conducted statistical analyses in R version 2.12
(R Development Core Team 2010).
Results
White-taileddeer browsing of spicebush is frequent
and widespread at SERC. In the small-scale gridded
survey, a majority (53 ' 2.2%, mean ' SE) of spice-
bushplants#2m tall showedevidence of current year
browsing by deer. In the survey across 16 ha, 890
spicebush plants were recorded in 265 uniquemeters-
squared quadrats; of these, browsing of spicebush was
recorded on 30.3% (N" 270) of plant individuals and
in 40% (N" 106) of quadrats. Moreover, browsing on
186 of these spicebush plants in 2010 was not predic-
tive of browsing in 2011 (#2 test, P " 0.97). Thus, we
attribute any trait differences in browsed plants to
current-year deer browsinghistory rather thanbrows-
ing in previous years.
Deer browsing caused a signiÞcant shift in multi-
variate space of leaf traits, explaining !10% of ob-
served variation in seven variables (constrained axis
eigenvalue" 0.683, permutation test P" 0.005). Prin-
cipal components analysis determined independent
axes of variance, with the Þrst three explaining a com-
bined 64% of variance in leaf traits. The Þrst axis
separated plants with a high percentage of H2O and
SLA fromplantswith tough leaves (Fig. 1); the second
axis separated plants with high percentage P from
those with larger leaves (greater leaf area); the third
axis separated plants with greater percentage N from
thosewith high percentageC(Fig. 1). In independent
linear models (t-tests), browsed plants were signiÞ-
cantly higher along the water-SLA axis (Fig. 1;
F1,50" 7.3885, P" 0.009). Plants did not differ signif-
icantly along the size-phosphorus axis (PC2) despite
a trend toward greater percentageP inbrowsed leaves
(F1,50" 2.6688, P" 0.11). Plants differed signiÞcantly
along thenitrogen-carbon axis,withunbrowsedplants
having greater percentage N and browsed plants hav-
ing greater percentage C (Fig. 1; F1,50 " 12.464, P #
0.001).
Using GC-MS, we detected 12 identiÞable terpene
compounds in spicebush leaves, including monoter-
Fig. 1. Principal components analysis of leaf traits shows
shift in traits of spicebush leaves after browsing by deer.
Filled circles represent browsed plants, open circles repre-
sent unbrowsedplants, plotted according to their PCA scores
of two PCA axes (PC1 and PC3, explaining 29 and 18% of leaf
trait variance, respectively). Variables are plotted according
to their mean value in the two PCA axes: leaf area; SLA,
speciÞc leaf area; toughness, leaf toughness; percentageH2O,
percent water; percentage C, percent carbon; percentage
N, percent nitrogen; percentage P, percent phosphorus.
Browsed and unbrowsed plants differed signiÞcantly in PC1
and PC3 scores, although not in PC2 scores (18% variance
explained, not shown).
1412 ENVIRONMENTAL ENTOMOLOGY Vol. 41, no. 6
penes, and sesquiterpenes such as caryophyllene
(Table 1). Three monoterpenes (eucalyptol, $-lina-
lool, and citronellal) appeared only in plants with a
history of current year browse, although not in sufÞ-
cient occurrence and concentration to detect a sig-
niÞcant difference. Analyzing the overall terpene pro-
Þle in the PCA (Fig. 2), the majority of plants are
clustered when plotted against the Þrst two PC axes,
which explain 50% of the variance in terpene chem-
istry. More browsed than unbrowsed plants appear to
have divergent terpene proÞles in these axes (Fig. 2),
but the permutation test revealed little of the variance
was explained by current-year browse (3.5% variance
explained, P " 0.21).
In the Þeld, there was signiÞcantly greater total
insect folivory on unbrowsed plants (Fig. 3; F1,1 "
6.762, P" 0.016). Visual inspection of invertebrates on
browsed and unbrowsed spicebush plants indicated
that a diversity of potential herbivores including spe-
cialist and generalist caterpillars, leaf beetles, and or-
thopterans were the most common insect herbivores,
likely to be responsible for the differential damage.
In contrast to patterns of total herbivory in theÞeld,
the specialist caterpillar P. troilus in the lab signiÞ-
cantly preferred leaves from plants that had been
browsed (Fig. 4a and F1,1" 5.5697, P" 0.02). Rearing
experiments demonstrated that swallowtail caterpil-
lars also performed better on leaves from browsed
plants, with a mean relative growth rate signiÞcantly
greater for caterpillars reared on these leaves (Fig. 4b
and F1,31" 5.947, P" 0.02). Pupal mass did not differ
by browsing treatment (Fig. 4c and F1,16 " 0.2287,
P " 0.63).
Discussion
Widespread and intense deer browsing induced sig-
niÞcant changes in spicebush leaf traits, including
shifts in morphology and nutrient concentration. In
the Þeld we also observed greater total insect her-
bivory on unbrowsed plants, but a specialist insect
Table 1. Mean standardized fraction of GC-MS peak area of distinct terpene compounds of spicebush leaves from plants browsed (n!
17) or unbrowsed (n ! 19) by white-tailed deer
ID Chemical RTa Browsed (SE) Unbrowsed (SE) t-statistic P value
1 5-Hepten-2-one, 6-methyl- 8.69 0.00806 (0.00183) 0.0108 (0.002) %1.02 0.31
2 Eucalyptol 9.58 7.87e-05 (5.51e-05) 0 (0) 1.43 0.17
3 beta-linalool 10.58 0.000148 (0.000112) 0 (0) 1.32 0.21
4 citronellal 11.43 0.000603 (0.000298) 0.000132 (8.23e-05) 1.52 0.14
5 carvol 12.93 1.85e-05 (1.85e-05) 0 (0) 1 0.33
6 Caryophyllene 15.53 0.0114 (0.00154) 0.0106 (0.0016) 0.383 0.7
7 trans-nerolidol 17.08 0.00243 (0.000794) 0.00181 (0.000604) 0.619 0.54
8 UnidentiÞed sesquiterpene 22.17 0.115 (0.0139) 0.132 (0.0166) %0.769 0.45
9 3,7,11,15-Tetramethyl-2-hexadecen-1-ol 20.1 0.0347 (0.00808) 0.038 (0.00836) %0.285 0.78
10 UnidentiÞed sesquiterpene 20.35 0.0048 (0.00266) 0.000424 (0.000424) 1.62 0.12
11 UnidentiÞed sesquiterpene 20.55 0.00352 (0.00152) 0.00821 (0.00262) %1.55 0.13
12 UnidentiÞed sesquiterpene 21.76 9.76e-05 (9.76e-05) 0 (0) 1 0.33
a RT: retention time in GC-MS sample.
Fig. 2. Principal components analysis of abundance of 12
terpene compunds (gray numbers, see Table 1 for identity)
detected in spicebush leaves by GC-MS. Filled circles rep-
resent browsed plants, open circles represent unbrowsed
plants, plotted according to their PCAscores of twoPCAaxes
(PC1 and PC2, explaining 29 and 21% of leaf terpene vari-
ance, respectively). Permutation tests show no signiÞcant
effect of current-year browsing history on terpene chemistry
(P " 0.21).
Fig. 3. Unbrowsed plants in the Þeld sustain more leaf
herbivory from chewing insects than browsed plants. Mean
fractionconsumedper leaf type isplottedaspoints for clarity,
although the statistical model analyzed area consumed with
total leaf area as a covariate. Bars indicate standard errors,
and P value is of model estimate of Þxed effect of browsing
on amount of leaf tissue consumed.
December 2012 LIND ET AL.: TRAIT-MEDIATED EFFECTS OF DEER ON INSECT HERBIVORES 1413
showed preference for browsed regrowth in lab trials.
Although previous studies have implicated induction
in chemical defenses as predictors of future insect
herbivory, we did not detect signiÞcant shifts in de-
fensive terpene compounds.
The changes to forest plant communities caused by
white-tailed deer have been addressed by a large body
of literature (e.g., McShea and Rappole 2000, Coˆte´ et
al. 2004), but the potential for deer browsing to be
reßected inchanges to insect communities is relatively
unstudied (but see Barrett and Stiling 2007, Bressette
et al. 2012). Spicebush is frequently and heavily
browsed by white-tailed deer (Results, and Liang and
Seagle 2002, Bressette et al. 2012), and spicebush itself
supports a diversity of plant-associated insects and
their avian predators (Niesenbaum 1992, Skoczylas et
al. 2007). Thus, induced changes to spicebush leaves
could have pervasive impacts on insect communities.
In our study, browsing altered spicebush leaf mor-
phology and nutrient content. Leaves on browsed
plants were thinner, less tough, and had higher water
content compared with leaves from nearby un-
browsed plantsÑchanges that are generally favorable
to insect herbivory. However, leaves from browsed
plants also had lower percentage N and greater per-
centage CÑchanges that can make leaves less attrac-
tive to herbivores.
Plant defense theory (reviewed in Stamp2003)pos-
its a fundamental tradeoff between allocation of re-
sources to growth and to defense against herbivory.
Thus, we might expect spicebush leaves to be more
well-defended after browsing, or to be faster growing
(more nutrient-rich, and thus more palatable) after
browsing, but not both. Our Þndings of higher SLA
and water content (Fig. 1) suggest a growth response
by the plant, although leaves of browsed plants had
less percentage N and more percentage C (Fig. 1).
Because this observed higher percentageC could rep-
resent an investment in carbon-based defenses, we
tested whether browsing induced higher levels of ter-
penes and terpenoids, carbon-based compounds
found in high concentrations in spicebush (Tucker et
al. 1994) that are known to deter deer browsing (Dun-
can et al. 2001). We hypothesized that plant terpene
chemistry might also be important to the spicebush
caterpillar P. troilus because adults use plant defenses
as cues to Þnding a suitable host plant (Carter and
Feeny 1999), and because P. troilus caterpillars exude
terpenoids from their defensive osmeteria as fourth
instars (Omura et al. 2006).
Wedidnot seea strongdifference inoverall terpene
proÞle among browsed versus unbrowsed plants, al-
though certain browsed individuals appeared more
chemically unique than others (Fig. 2). This diver-
gence was based on the presence of three monoter-
penoids and one unidentiÞed sesquiterpene unique to
plants that had been browsed; unbrowsed plants had
no compounds that were not also found in at least one
browsed plant (Table 1). These results suggest that
deer herbivorymay induce terpene synthesis in spice-
bush, such as that found in pines after defoliation by
leaf-cutter ants (Barnola et al. 1994). However, de-
tailed studies of the impact of Sitka black-tailed deer
(Odocoileus hemionus sitkensis Merriam) on western
redcedar (Thuja plicata Donn ex D.Don) could not
isolate induction of terpenes in Þeld (VourcÕh et al.
2002) or in greenhouse experiments (VourcÕh et al.
2003), suggesting that terpenoid induction by verte-
brates may not be widespread. Our ability to docu-
ment induced terpene chemistry, and particularly any
within-plant tradeoffs between defensive terpene
chemistry and leaf nutrient content, may be limited
because our terpene measurements were conducted
Fig. 4. Preference and performance of the specialist her-
bivore P. troilus (Lepidoptera: Papilionidae) on leaves col-
lected from recently browsed versus unbrowsed spicebush
(L. benzoin) plants. Third- or fourth-instar caterpillars pref-
erentially consumed (A) and had greater relative growth
rates (B) but not pupal mass (C) on regrown leaves from
browsed versus unbrowsed plants. Bars indicate standard
errors, and P value is from model estimate of Þxed effect of
browsing on amount of leaf tissue consumed. In (A), mean
fractionconsumedper leaf type isplottedaspoints for clarity,
although the statistical model analyzed area consumed with
total leaf area as a covariate. In (B) and (C), P values are of
linearmodel estimates of browsing impact on relative growth
rate and pupal mass, respectively.
1414 ENVIRONMENTAL ENTOMOLOGY Vol. 41, no. 6
separately from our estimation of other leaf traits.
Finally, the terpenoids that we identiÞed in spicebush
foliage contained no known compounds that overlap
with P. troilus osmeterial secretions (Omura et al.
2006), suggesting de novo synthesis rather than se-
questration of plant defensive compounds.
The reduced herbivory that we observed on leaves
from browsed plants in the Þeld (Fig. 3) is consistent
with a decrease in leaf percentage N, as N is often
limiting to insects and thus a strong predictor of insect
preference (Mattson 1980). In contrast, Barrett and
Stiling (2007) found that herbivorybyKeydeer (Odo-
coileus virginianus clavium Barbour & G.M. Allen)
caused leaf percentage N to increase, with concomi-
tant increases in insect herbivory. Although contra-
dictory in the overall direction of the impact, both
results suggest a strong inßuence of nutrient content
as an induced trait that predicts future herbivory.
Response to induced leaf trait changes also differedby
abroadcategoryofherbivore specialization.Although
we did not identify the chewing herbivores responsi-
ble for the leaf injury observed in the Þeld, con-
temporaneous visual surveys found a diversity of
generalist herbivores (Orthoptera, Coleoptera, and
Lepidoptera), in addition to the locally known spe-
cialist (P. troilus) and the oligophagous caterpillar,
Epimecis hortaria F. (Lepidoptera: Geometridae).
Thus, we consider the Þeld herbivory to represent a
generalized response to leaf traits induced by deer
browse, implying thatdeercouldhave largelynegative
plant-mediated impacts on generalist insect herbi-
vores of spicebush.
In contrast, the specialist P. troilus responded pos-
itively in terms of larval preference and growth rate
performance to leaves from plants browsed by deer
(Fig. 4). Faster larval development is hypothesized to
lead to greater survival through escape from natural
enemies (Benrey and Denno 1997), raising the pos-
sibility of facilitation of this and other similarly-re-
sponding specialists as part of an indirect interaction
web triggered by deer herbivory on spicebush (Oh-
gushi 2005).
The mechanisms of the preference for, and better
performance on, browsed plants by the specialist cat-
erpillar are not clear. Leaves from browsed plants
were thinner and had greater average water content,
but less percentage N, than leaves from unbrowsed
plants. We did not assess the broader suite of consti-
tutive carbon-based molecules such as Þber, tannins,
and phenolic acids, but these digestibility-reducing
compounds could play a role in increasing available
nitrogen even if total percentage N by mass is lower
in browsed leaves.
In conclusion, we found that white-tailed deer her-
bivory induced signiÞcant changes in spicebush leaf
chemistry that had opposing effects on insect her-
bivory in the Þeld (a mixed group of specialists and
generalists) versus a specialist caterpillar in the lab.
Such induced changes suggest the potential for wide-
spread but relatively unstudied indirect effects medi-
ated by this keystone herbivore.
Acknowledgments
We thank Lauren Richie, Christian Latimer, Christine
Cochrane, and Anne Chamberlain for help in the Þeld and
lab, Candy Feller for helpful discussions on spicebush her-
bivory at SERC, and two anonymous reviewers for helpful
comments on an earlier version of themanuscript. This work
was supported in part by an NSF REU site grant (NSF-DBI
0851303).
References Cited
Agrawal, A. A. 1998. Induced responses to herbivory and
increased plant performance. Science 279: 1201Ð1202.
Agrawal, A. A., and M. Fishbein. 2006. Plant defense syn-
dromes. Ecology 87: S132ÐS149.
Awmack, C. S., and S. R. Leather. 2002. Host plant quality
and fecundity in herbivorous insects. Annu. Rev. Ento-
mol. 47: 817Ð844.
Barnola, L. F., M. Hasegawa, and A. Ceden˜o. 1994. Mono-
and sesquiterpene variation in Pinus caribaeaneedles and
its relationship toAtta laevigataherbivory. Biochem. Syst.
Ecol. 22: 437Ð445.
Barrett, M. A., and P. Stiling. 2007. Relationships among
Key deer, insect herbivores, and plant quality. Ecol. Res.
22: 268Ð273.
Behmer, S. T. 2009. Insect herbivore nutrient regulation.
Annu. Rev. Entomol. 54: 165Ð187.
Benrey, B., and R. F. Denno. 1997. The slow-growth-high-
mortality hypothesis: a test using the cabbage butterßy.
Ecology 78: 987Ð999.
Boege, K., and R. J. Marquis. 2005. Facing herbivory as you
grow up: the ontogeny of resistance in plants. Trends
Ecol. Evol. 20: 441Ð448.
Bressette, J. W., H. Beck, and V. B. Beauchamp. 2012. Be-
yond the browse line: complex cascade effects mediated
by white-tailed deer. Oikos (in press).
Broadway, R. M., and S. S. Duffey. 1988. The effect of plant
protein-quality on insect digestive physiology and the
toxicity of plant proteinase-inhibitors. J. Insect Physiol.
34: 1111Ð1117.
Carter, M., and P. Feeny. 1999. Host-plant chemistry inßu-
ences oviposition choice of the spicebush swallowtail
butterßy. J. Chem. Ecol. 25: 1999Ð2009.
Clesceri, L. S., A. E. Greenberg, and A. D. Eaton. 1998.
Standard methods for the examination of water and
wastewater. American Public Health Association, New
York.
Cornelissen, J., S. Lavorel, E.Garnier, S.Diaz,N.Buchmann,
D. Gurvich, P. Reich, H. ter Steege, H. Morgan, M. van
der Heijden, et al. 2003. A handbook of protocols for
standardised and easy measurement of plant functional
traits worldwide. Aust. J. Bot. 51: 335Ð380.
Coˆte´, S. D., T. P. Rooney, J. P. Tremblay, C. Dussault, and
D. M. Waller. 2004. Ecological impacts of deer over-
abundance. Annu. Rev. Ecol. Syst. 35: 113Ð147.
Den Herder, M., R. Virtanen, and H. Roininen. 2004. Ef-
fects of reindeer browsing on tundra willow and its as-
sociated insect herbivores. J. Appl. Ecol. 41: 870Ð879.
Duncan, A. J., S. E.Hartley,M. Thurlow, S. Young, andB.W.
Staines. 2001. Clonal variation in monoterpene concen-
trations in Sitka spruce (Picea sitchensis) saplings and its
effect on their susceptibility to browsing damage by red
deer (Cervus elaphus). For. Ecol. Manag. 148: 259Ð269.
Fine, P.V.A., Z. J.Miller, I.Mesones, S. Irazuzta,H.M.Appel,
M.H.H. Stevens, I. Saaksjarvi, L. C. Schultz, and P. D.
Coley. 2006. The growth-defense trade-off and habitat
December 2012 LIND ET AL.: TRAIT-MEDIATED EFFECTS OF DEER ON INSECT HERBIVORES 1415
specializationbyplants inAmazonian forests. Ecology 87:
S150ÐS162.
Kaplan, I., S. Sardanelli, B. J. Rehill, and R. F. Denno. 2010.
Toward a mechanistic understanding of competition in
vascular-feeding herbivores: an empirical test of the sink
competition hypothesis. Oecologia 166: 627Ð636.
Karban, R. 2011. The ecology and evolution of induced re-
sistance against herbivores. Funct. Ecol. 25: 339Ð347.
Kessler, A., and I. Baldwin. 2001. Defensive function of her-
bivore-inducedplant volatile emissions innature. Science
291: 2141Ð2144.
Kessler, A., and I. Baldwin. 2002. Manduca quinquemacula-
taÕs optimization of intra-plant oviposition to predation,
food quality, and thermal constraints. Ecology 83: 2346Ð
2354.
Legendre, P., and L. Legendre. 1998. Numerical ecology.
Elsevier, Amsterdam, the Netherlands.
Liang, S. Y., and S. W. Seagle. 2002. Browsing and micro-
habitat effects on riparian forest woody seedling demog-
raphy. Ecology 83: 212Ð227.
Martinsen, G. D., E. M. Driebe, and T. G. Whitham. 1998.
Indirect interactions mediated by changing plant chem-
istry: beaver browsing beneÞts beetles. Ecology 79: 192Ð
200.
Mattson, W. 1980. Herbivory in relation to plant nitrogen
content. Annu. Rev. Ecol. Syst. 11: 119Ð161.
McCabe, T. R., and R. E. McCabe. 1997. Recounting
whitetales past, pp. 11Ð26. InW. J. McShea, H. B. Under-
wood, and J. H. Rappole (eds.), The science of overabun-
dance: deer ecology and populationmanagement. Smith-
sonian Institution Press, Washington, DC.
McShea,W. J., and J. H. Rappole. 2000. Managing the abun-
dance and diversity of breeding bird populations through
manipulation of deer populations. Conserv. Biol. 14:
1161Ð1170.
McShea,W. J.,W.M.Healy, P.Devers, T. Fearer, F.H.Koch,
D. Stauffer, and J. Waldon. 2007. Forestry matters: de-
cline of oaks will impact wildlife in hardwood forests. J.
Wildl. Manag. 71: 1717Ð1728.
Miller, R. O. 1998. High-temperature oxidation: dry ashing,
pp. 53Ð56. In Y. P. Kalra (ed.), Handbook of reference
methods for plant analysis. CRC, Boca Raton, FL.
Mooney, E. H., E. J. Tiedeken, N. Z. Muth, and R. A. Niesen-
baum. 2009. Differential induced response to generalist
and specialist herbivores byLindera benzoin (Lauraceae)
in sun and shade. Oikos 118: 1181Ð1189.
Niesenbaum, R. 1992. The effects of light environment on
herbivory and growth in the dioecious shrub Lindera
benzoin (Lauraceae). Am. Midl. Nat. 128: 270Ð275.
Niesenbaum, R. A., and E. C. Kluger. 2006. When studying
the effects of light on herbivory, should one consider
temperature? The case of Epimecis hortaria F. (Lepidop-
tera: Geometridae) feeding on Lindera benzoin L. (Lau-
raceae). Environ. Entomol. 35: 600Ð606.
Ohgushi, T. 2005. Indirect interaction webs: herbivore-in-
duced effects through trait change in plants. Annu. Rev.
Ecol. Syst. 36: 81Ð105.
Omura, H., K. Honda, and P. Feeny. 2006. From terpenoids
to aliphatic acids: further evidence for late-instar switch
in osmeterial defense as a characteristic trait of swallow-
tail butterßies in the tribe Papilionini. J. Chem. Ecol. 32:
1999Ð2012.
Osier, T. L., and S. M. Jennings. 2007. Variability in host-
plant quality for the larvae of a polyphagous insect foli-
vore inmidseason: the impact of light on three deciduous
sapling species. Entomol. Exp. Appl. 123: 159Ð166.
Parker, J. D., L. J. Richie, E. M. Lind, and K. O. Maloney.
2010. Land use history alters the relationship between
native and exotic plants: the rich donÕt always get richer.
Biol. Invasions 12: 1557Ð1571.
R Development Core Team. 2010. R: a language and envi-
ronment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. (http://www.R-
project.org).
Roslin, T., S. Gripenberg, J. P. Salminen, M. Karonen, R. B.
O’Hara, K. Pihlaja, and P. Pulkkinen. 2006. Seeing the
trees for the leaves - Oaks as mosaics for a host-speciÞc
moth. Oikos 113: 106Ð120.
Schultz, J., P. Nothnagle, and I. Baldwin. 1982. Seasonal and
individual variation in leaf quality of two northern hard-
woods tree species. Am. J. Bot. 69: 753Ð759.
Skoczylas, D. R., N. Z. Muth, and R. A. Niesenbaum. 2007.
Contribution of insectivorous avifauna to top down con-
trol of Lindera benzoin herbivores at forest edge and
interior habitats. Acta Oecol. 32: 337Ð342.
Stamp, N. 2003. Out of the quagmire of plant defense hy-
potheses. Q. Rev. Biol. 78: 23Ð55.
Tucker, A.O.,M. J.Maciarello, P.W. Burbage, andG. Sturtz.
1994. Spicebush [Lindera benzoin (L.) Blume var. ben-
zoin, Lauraceae]: a tea, spice, and medicine. Econ. Bot.
48: 333Ð336.
Wagner, D. L. 2005. Caterpillars of eastern North America:
a guide to identiÞcation and natural history. Princeton
University Press, Princeton, NJ.
Wright, D. M., G. J. Jordan, W. G. Lee, R. P. Duncan, D. M.
Forsyth, andD. A. Coomes. 2010. Do leaves of plants on
phosphorus-impoverished soils contain high concentra-
tions of phenolic defence compounds? Funct. Ecol. 24:
52Ð61.
Wright, I. J., P. B. Reich, M. Westoby, D. D. Ackerly, Z.
Baruch, F. Bongers, J. Cavender-Bares, T. Chapin, J.H.C.
Cornelissen,M.Diemer, et al. 2004. Theworldwide leaf
economics spectrum. Nature 428: 821Ð827.
VanZandt, P. A., andA.A. Agrawal. 2004. Community-wide
impacts of herbivore-induced plant responses in milk-
weed (Asclepias syriaca). Ecology 85: 2616Ð2629.
Vourc’h, G., B. Vila,D.Gillon, J. Escarre´, F. Guibal,H. Fritz,
T. P. Clausen, and J.-L. Martin. 2002. Disentangling the
causes of damage variation by deer browsing on young
Thuja plicata. Oikos 98: 271Ð283.
Vourc’h,G., J. Russell,D.Gillon, and J.Martin. 2003. Short-
term effect of defoliation on terpene content in Thuja
plicata. Ecoscience 10: 161Ð167.
Received 28 March 2012; accepted 13 September 2012.
1416 ENVIRONMENTAL ENTOMOLOGY Vol. 41, no. 6