ArticlePDF Available

Abstract and Figures

1. Herbivory often induces systemic plant responses that affect the host choice of subsequent herbivores, either deterring or attracting them, with implications for the performance of both herbivore and host plant. Combining measures of herbivore movement and consumption can efficiently provide insights into the induced plant responses that are most important for determining choice behaviour. 2. The preferences of two frugivorous stink bug species, Nezara viridula and Euschistus servus between cotton plants left undamaged or damaged by Helicoverpa zea and Heliothis virescens larvae were investigated. A novel consumer movement model was used to investigate if attraction rates or leaving rates determined preferences. Stink bug consumption rates were measured using salivary sheath flanges. Finally, the systemic induction of selected phenolic‐based and terpenoid secondary metabolites were measured from heliothine herbivory on developing cotton bolls, to investigate if they explained stink bug feeding responses. 3. Heliothine herbivory did not affect the N. viridula feeding preference. However, we found opposing effects of H. zea and H. virescens herbivory on the behaviour of E. servus . Avoidance of H. zea ‐damaged plants is not obviously related to phenolic or terpenoid induction in cotton bolls; whereas a preference for H. virescens ‐damaged plants may be related to reductions in chlorogenic acid in boll carpel walls. 4. The present results highlight the inferential power of measuring both consumer movement and consumption in preference experiments and combining behavioural responses with phytochemical responses. Furthermore, while plant‐mediated interactions among herbivorous insects are well studied, interactions among frugivorous species specifically have been poorly documented.
Content may be subject to copyright.
Ecological Entomology (2015), DOI: 10.1111/een.12221
Behavioural and chemical mechanisms of
plant-mediated deterrence and attraction among
frugivorous insects
ADAM R. ZEILINGER,
1DAWN M. OLSON,
2DAN MACLEAN,3
NAOKI MORI,
4RYU NAKATA4and DAVID A. ANDOW51Conservation
Biology Program, Department of Entomology, University of Minnesota, St. Paul, Minnesota, U.S.A., 2Crop Protection, Research, and
Management Unit, USDA-ARS, Tifton, Georgia, U.S.A., 3Department of Horticulture, University of Georgia, Tifton, Georgia,
U.S.A., 4Division of Applied Life Sciences, Graduate School of Agriculture, Kyoto University, Kyoto, Japan and 5Department of
Entomology and Center for Community Genetics, University of Minnesota, St. Paul, Minnesota, U.S.A.
Abstract. 1. Herbivory often induces systemic plant responses that affect the host
choice of subsequent herbivores, either deterring or attracting them, with implications
for the performance of both herbivore and host plant. Combining measures of herbivore
movement and consumption can efciently provide insights into the induced plant
responses that are most important for determining choice behaviour.
2. The preferences of two frugivorous stink bug species, Nezara viridula and
Euschistus servus between cotton plants left undamaged or damaged by Helicoverpa
zea and Heliothis virescens larvae were investigated. A novel consumer movement
model was used to investigate if attraction rates or leaving rates determined preferences.
Stink bug consumption rates were measured using salivary sheath anges. Finally, the
systemic induction of selected phenolic-based and terpenoid secondary metabolites were
measured from heliothine herbivory on developing cotton bolls, to investigate if they
explained stink bug feeding responses.
3. Heliothine herbivory did not affect the N. viridula feeding preference. However,
we found opposing effects of H. zea and H. virescens herbivory on the behaviour of
E. servus. Avoidance of H. zea-damaged plants is not obviously related to phenolic
or terpenoid induction in cotton bolls; whereas a preference for H. virescens-damaged
plants may be related to reductions in chlorogenic acid in boll carpel walls.
4. The present results highlight the inferential power of measuring both consumer
movement and consumption in preference experiments and combining behavioural
responses with phytochemical responses. Furthermore, while plant-mediated interac-
tions among herbivorous insects are well studied, interactions among frugivorous species
specically have been poorly documented.
Key words. Competition, consumer choice, facilitation, gossypol, induced resistance,
induced susceptibility, Noctuidae, Pentatomidae.
Correspondence: Adam R. Zeilinger, Berkeley Initiative for Global
Change Biology and Department of Environmental Science, Policy,
and Management, University of California Berkeley, 130 Mulford Hall
#3114, Berkeley, CA 94720-3114, U.S.A. E-mail: arz@berkeley.edu
Conicts of interest: The authors declare that they have no conicts
of interests in the work described in this manuscript.
Introduction
Insect herbivores exhibit strong and distinct intra-specic prefer-
ences among host plants. These preferences are related to many
factors, including plant size and form, nutritional quality, and
morphological and chemical host plant defences (Bernays &
Chapman, 1994). Induced plant responses to previous herbivory
can affect feeding preference by subsequent herbivores (Karban
& Baldwin, 1997; Halitschke et al., 2008), and a meta-analysis
Published 2015. This article is a U.S. Government work and is in the public domain in the USA. 1
2Adam R. Zeilinger et al.
suggested that, in general, herbivorous insects tend to avoid
feeding on host plants damaged by heterospecics, although the
overall effect is weak (Kaplan & Denno, 2007). Examples of sig-
nicant induced preferences can, nonetheless, be found across a
variety of systems (Agrawal, 1999; Long et al., 2007; Anderson
et al., 2011).
Feeding preference for herbivores is most often inferred
from measures of relative consumption (Rieske & Raffa, 1998;
Agrawal, 1999, 2000; Fisher et al., 1999; Beckerman, 2000;
Long et al., 2007; Anderson et al., 2011; Gutbrodt et al., 2011).
The primary exception is aphids, where measures of location are
used (Gianoli, 2000; Messina et al., 2002; Zytynska & Preziosi,
2013). Measures of consumption relate directly to resource qual-
ity but provide little information about the attractiveness of
the choices (Nicotri, 1980; Schoonhoven et al., 2005; Zeilinger
et al., 2014). In contrast, measures of location or movement
can provide insight into both resource quality and attractiveness
(Zeilinger et al., 2014). Taken together, measures of consump-
tion and movement may provide complementary insights into
the nature of consumer choice.
The probability that a herbivore is associated with a given
host choice is the balance of the herbivore’s attraction rate
to and leaving rate from that choice (Zeilinger et al., 2014).
Whether the preference is determined by attraction or by leaving
may provide clues to the most relevant causes. Differences in
attraction rates between two choices must be mediated by visual,
olfactory, or (possibly) electrical eld cues that can be detected
from a distance, whereas differences between leaving rates could
be mediated by a wider range of cues (Bernays & Chapman,
1994; Schoonhoven et al., 2005; Clarke et al., 2013).
We studied the plant-mediated interactions between two herbi-
vore guilds on cotton, Gossypium hirsutum: sucking frugivores,
Nezara viridula L. and Euschistus servus Say (Heteroptera:
Pentatomidae), and chewing folio-frugivores, Helicoverpa zea
(Boddie) and Heliothis virescens Fabricius (Lepidoptera: Noc-
tuidae). Specically, we investigated effects of heliothine dam-
age on the feeding preference of stink bugs using whole plants.
All four species are polyphagous and important pests of cot-
ton in southeastern U.S.A. (Fitt, 1989; McPherson & McPher-
son, 2000). The stink bug species are primarily seed feeders
(Panizzi & Slansky, 1991; McPherson & McPherson, 2000),
whereas the heliothine larvae feed on numerous plant tissues
but prefer reproductive structures (Fitt, 1989). Little informa-
tion exists on the effects of induced plant responses on fru-
givores and seed herbivores in general (Preisser & Bastow,
2006; Whitehead & Poveda, 2011), and on seed-sucking frugi-
vores in particular (Slansky & Panizzi, 1987; Kaplan & Denno,
2007).
In previous no-choice experiments, we found evidence for
asymmetrical, species-specic competitive interactions between
the same stink bug and heliothine species, with the strongest
effect between H. zea and E. servus (Zeilinger et al., 2011).
Based on those results, we hypothesise that E. servus nymphs
will avoid feeding on plants previously damaged by H. zea but
that H. virescens will have no effect. Little is known about
which resistance traits in cotton plants inuence herbivorous
stink bug performance or behaviour. Nonetheless, based on
previous literature, we hypothesise that phenolic compounds or
the terpenoid gossypol could be important (Stamp et al., 1997;
Bi et al., 1997; Evangelista et al., 2011; Hagenbucher et al.,
2013).
Here we investigate stink bug feeding preference
to heliothine-damaged and undamaged plants using a
bio-statistical consumer movement model (Zeilinger et al.,
2014). Additionally, to link stink bug host selection to
induced plant responses (Pareja et al., 2009), we measured
the concentrations of gossypol in seeds and selected phenolic
secondary compounds chlorogenic acid, gallic acid, rutin,
and anthocyanins in the carpel walls of cotton bolls from
heliothine-damaged and undamaged plants.
Materials and methods
Cotton plants
Non-transgenic DP491 cotton plants were grown in a green-
house at the USDA-Agricultural Research Station (USDA-ARS)
in Tifton, GA, U.S.A., maintained at 26 ±5C,andanLD
13:11 h photoperiod. During the summers of 2009 and 2010,
plants were grown in 2.3-litres pots, with potting soil made of
80% (by volume) Miracle-Gro Potting Mix®(21-7-14 N-P-K)
and 20% peat moss. In 2009, plants were fertilised at the rate
of 1.34 g N l1soil; in 2010, plants were fertilised at the rate of
1gNl
1soil, using slow-release Osmocote®(14-14-14 N-P-K)
for fertiliser. In 2009, we sprayed plants with the insecticides
acetamaprid (Assail 30G®, 0.5 mg a.i. per plant) or spirome-
sifen (Oberon 2®, 1.1 mg a.i. per plant) if we detected infesta-
tions of spider mites (Tetranychus sp.), aphids (Aphis gossypii),
or whiteies (Bemisia sp.) in the greenhouse. In 2010, we
sprayed plants with a rotating twice-monthly spray schedule of
acetamiprid, spiromesifen, and dicofol (Kelthane MF4®, 4.3 mg
a.i. per plant). Neonicotinoids (e.g. acetamaprid) are not toxic to
adult and fth instar nymphs of stink bugs 4days after applica-
tion (Kamminga et al., 2009), and do not appear to affect jas-
monic acid and salicylic acid levels that are critical for plant
defence signalling; nonetheless they do affect other phytohor-
mones and have the potential to inuence plant-induced resis-
tance (Szczepaniec et al., 2013). The effects of spiromesifen,
dicofol, and acetamaprid specically on stink bug nymphs and
plant induction are unknown. Regardless, insectiticides were
applied to all plants equally thereby controlling for any sys-
tematic bias. Plants were not sprayed with any insecticides at
least 1 week prior to their use in experiments. Any plants with
detectable damage from or infestations of greenhouse pests were
not used in the experiments. At 6 weeks old, all plants were
treated with the growth regulator Compact®(Mepiquat chlo-
ride; 0.5 μl a.i. per plant).
Insects
Colonies of stink bugs and heliothines were maintained as
described in Zeilinger et al. (2011).
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
Herbivore-induced deterrence and attraction 3
Fig. 1. Diagram of the experimental setup. The caterpillar and the stink bug were physically separated and the bolls presented to the stink bug remained
attached to the cotton plant that allowed us to make inferences about plant-mediated interactions.
Stink bug feeding preference experiment
Two cotton plants were paired according to the age of
the cotton bolls from the time of owering, and these bolls
were presented to the stink bugs (Fig. 1). All plants were
7–8 weeks old and had about six bolls per plant. Treatments
(caterpillar-damaged, undamaged) were randomly assigned to
each plant within a pair, and stink bug species and helioth-
ine species were randomly assigned to plant pairs. When the
experimental bolls were 7 days old, the damage-treatment plant
received a single fourth instar heliothine larva (7– 8 days old) of
either species in a 15 cm2ne-mesh cage on the rst-position
boll on the branch immediately below the experimental boll
(Fig. 1), and allowed to feed for 36 h. We injured the boll on the
branch immediately below the bolls presented to the stink bugs
because systematic-induced responses in cotton are strongest
above the site of damage (Hagenbucher et al., 2013).
Thirty-six hours after the heliothine damage began we caged
one newly-emerged fth instar stink bug nymph, (1– 2 days
since molting) of either species with the two experimental
bolls. In 2010, we measured the height and diameter of the
experimental bolls before caging the stink bug. We conducted
trials with each combination of stink bug species and heliothine
species in each year, except trials with N. viridula and H.
virescens were conducted only in 2010. The stink bug cage
was a 30 ×12 cm2cylindrical cage made of 1 mm mesh netting.
After being chilled for 10 min at 6 C, the stink bug was
placed in the cage equidistant from the two bolls, which
were 20 cm apart (Fig. 1). Stink bugs were starved for 12 h
prior to the start of the trials. In 2009, we obtained the
following sample sizes: E. servus +H. zea =15, E. servus +H.
virescens =12, and N. viridula +H. zea =19. In 2010, the
sample sizes were: E. servus +H. zea =15, E. servus +H.
virescens =14, N. viridula +H. zea =13, and N. viridula +H.
virescens =22.
We started all trials between 22.00 and 23.00hours because
N. viridula, and presumably E. servus, feed most actively at
night (Shearer & Jones, 1996). The location of the stink bug
was observed for 10 continuous minutes immediately after the
stink bug was placed in the cage, then 30 min, 1, 12, 18, 24, and
36 h thereafter. At each observation time, stink bug location was
recorded as in the neutral space (no choice), on the damaged
plant, or on the undamaged plant.
At the end of the trial, we calculated the nal boll volume
using the equation for the volume of a spheroid. We also counted
stink bug salivary sheath anges as described in Zeilinger et al.
(2011). Salivary sheath anges are produced by and left as a
record of stink bug feeding (Miles, 1972) and are positively
associated with feeding activity in E. servus but not in N. viridula
nymphs (Zeilinger et al., 2015). For the heliothine-damaged
boll, we also estimated the amount of lint and seeds consumed in
10% increments. In 2010, we also measured the initial and nal
boll volume by water displacement, from which we estimated
the per cent of the initial boll volume consumed by the larva.
Analysis of cotton secondary metabolites
Non-transgenic DP491 cotton plants were grown in a green-
house as described in the preference experiments, in 2011 and
2012 for phenolic and terpenoid chemical analyses, respec-
tively. Plants were damaged with H. zea and H. virescens
fourth instar larvae as described in the preference experiments,
and treatments were: undamaged plants, H. virescens-damaged
plants, and H. zea-damaged plants. We excised the undamaged
rst-position boll (equivalent to the experimental bolls in the
preference experiment) for analysis.
Cotton boll carpel wall phenolic compounds
In 2011, we tested for systemic induction of selected phenolic
compounds in the three treatments with ve replicates of
each. In each treatment, we assessed concentrations of the
phenolic-based compounds: anthocyanins, chlorogenic acid,
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
4Adam R. Zeilinger et al.
Table 1 . Description of model variants used in model selection procedure.
Model no. Model description d.f. Parameters estimated
M1 One pand one 𝜇parameter 2 p,,𝜇,
M2a pby choice 3 pd,,p
u,,𝜇,
M2b 𝜇by choice 3 p,,𝜇d,,𝜇u,
M2c pby year 3 p,09,p
,10,𝜇,
M2d 𝜇by year 3 p,,𝜇,09,𝜇,10
M3a pand 𝜇by choice 4 pd,,p
u,,𝜇d,,𝜇u,
M3b pand 𝜇by year 4 p,09,p
,10,𝜇,09 ,𝜇,10
M4a pby choice and year, 𝜇by choice 6 pd,09,p
d,10,p
u,09,p
u,10,𝜇d,,𝜇u,
M4b pby choice, 𝜇by choice and year 6 pd,,p
u,,𝜇d,09,𝜇u,09 ,𝜇d,10,𝜇u,10
M4c pby choice and year, 𝜇by year 6 pd,09 ,p
d,10,p
u,09,p
u,10,𝜇,09 ,𝜇,10
M4d pby year, 𝜇by choice and year 6 p,09 ,p
,10,𝜇d,09 ,𝜇u,09,𝜇d,10 ,𝜇u,10
M5 Fully parameterised 8 pd,09,p
d,10,p
u,09,p
u,10,𝜇d,09 ,𝜇u,09,𝜇d,10 ,𝜇u,10
For each pi,j or 𝜇i,j parameter, the isubscript refers to the plant choice estimated: dfor damaged plant, ufor undamaged plant, or for parameters
estimated across plants. Likewise, the jsubscript refers to the year estimated: 09 for 2009, 10 for 2010, or for parameters estimated across years. For
trials with N. viridula and H. virescens, only models M1, M2a, M2b, and M3a were estimated.
gallic acid, and rutin. We separated the carpel wall from the
lint of the excised boll, and placed the carpel wall in liquid
nitrogen. Subsequently, carpel walls were stored at 80 Cfor
24 h, freeze-dried for 24 h, crushed to a powder with a mortar and
pestle and analysed by high-performance liquid chromatography
(HPLC) (Appendix S1).
Cotton seed gossypol and derivative compounds
In 2012, we tested for systemic induction of gossypol and
its derivatives in cotton seed in the three treatments with ve
replicates of each. We separated cotton seeds from the lint of
the excised boll and placed the seeds in liquid nitrogen. Sub-
sequently, seeds were processed as described for the phenolic
compounds and analysed under liquid chromatography–mass
spectrometry (LC/MS) (Appendix S1).
Statistical analysis
Stink bug movement. We tested stink bug preference by
estimating attraction and leaving rates from a continuous-time
consumer preference model described in Zeilinger et al. (2014):
dPd
dt =−𝜇dPd+pdPn
dPu
dt =−𝜇uPu+puPn
Pn=1PdPu(1)
where Pd,Pu,and Pnare the probabilities that a stink bug is
located on the heliothine-damaged plant, the undamaged plant,
and in the neutral space between plants, respectively; pdand pu
are stink bug attraction rates to the damaged and undamaged
plants, respectively; and 𝜇dand 𝜇uare stink bug leaving
rates from the damaged and undamaged plants, respectively.
Attraction and leaving rate parameters were estimated using
maximum likelihood estimation and the analytical solution of
system 1.
We were interested in testing differences in preference
between choices and years. For each heliothine-stink bug
species combination, we estimated parameters for a series
of model variants, which differed in allowing parameters to
depend on plant treatment, year, both plant treatment and
year, or neither, in a manner analogous to a two-way 
with main effects and interaction. For example, we tested for
differences in attraction rates between the two plant choices by
comparing two variants of model 1: one in which plant-specic
attraction rates were allowed to vary, pdpu, and one in which
one attraction rate was t for both plants, pd=pu. Similarly,
we tested differences between years by comparing models in
which year-specic parameters were estimated and in which
parameters were estimated with data combined over years.
This model selection scheme included 12 total model variants
(Table 1). We tested the model assumptions of time-constancy of
parameters and independent consecutive choices as described
in (Zeilinger et al., 2014). We ranked model variants using
Aikake’s Information Criterion corrected for a small sample
size (AICc).
To provide an overall description of stink bug preference, we
estimated the expected long-term probability that stink bugs will
be located on the two choice plants (Zeilinger et al., 2014). We
calculated the equilibrium for each stink bug-heliothine-year
combination using the attraction and leaving rate parameter
estimates and variances in a Monte Carlo simulation, assuming
normality (Bolker, 2008) and using 10000 simulations.
Boll consumption. First, we used  to test for differences
between stink bug species, heliothine species, and years in initial
stink bug boll volume (2010 only), nal stink bug boll volume,
consumption by heliothines, and the number of salivary sheath
anges (both years).
Second, we tested for differences between the damaged
and undamaged plants in initial and nal boll volumes, and
the number of salivary sheath anges using paired t-tests
and a preference index from Kogan and Goeden (1970) to
reduce potential confounding of between-choice variation and
among-trial variation (Lockwood, 1998).
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
Herbivore-induced deterrence and attraction 5
Finally, to investigate consumption, we used  to
investigate the relationship between the number of salivary
sheath anges (response variable) and the total stink bug tenure
time, the percent of boll volume consumed by heliothine
larvae, years, and the interaction between tenure time and years
(explanatory variables). Total tenure time is the estimated total
amount of time during the trials that each stink bug spent on
the boll of the heliothine-damaged plant. To calculate tenure
time, we assumed that any switching by stink bugs that was
not directly observed occurred at the median between two
observation points.
Phenolics. We calculated the concentrations (mg g1
freeze-dried boll capsule wall) of the phenolic compounds
using standard curves for each compound and the mass of the
freeze-dried boll capsule walls (Appendix S1). We tested for
differences in the treatments using a , followed by
univariate s for each compound separately. In all models,
explanatory variables included week, as a partial incomplete
block effect (Oehlert, 2000), and damage treatment.
Gossypol and its derivatives. MS peaks for gossypol were
identied using a standard (Alfa Aesar, >98% purity), and
deoxyhemigossypol and hemigossypol were identied using
molecular weight and registered (Pub Chem) references. As with
the phenolics, we tested for differences in the treatments using
 followed by univariate s for each compound
separately.
Statistical programs. All analyses were conducted in R
3.1.1 (R Core Team, 2014). Maximum likelihood estima-
tion of consumer preference model parameters proceeded as
described in Zeilinger et al. (2014) except that we used the
mle2 function because it allows the specication of linear
models between variables in the data through the ‘parameters’
option – which we used to estimate year-specic parameters
(Bolker, 2008; Bolker & R Core Team, 2014). We conducted
s with Type III sums of squares with the Anova function
(car package). R scripts for likelihood and optimisation func-
tions of the 12 consumer movement model variants and AICc
model selection are provided in Appendix S2 and on-line at
https://github.com/arzeilinger/Consumer-Choice-Model (under
the le name ‘multi-year-models.R’).
Results
Stink bug movement
Our data on E. servus host choice met both assumptions of the
consumer movement model, whereas data on N. viridula only
met the assumption of time-constancy of attraction and leaving
rates (Appendix S3)
The most appropriate models for our data depended on the
stink bug species and heliothine species (Table 2). For trials with
E. servus and H. zea, seven models had ΔAICc<7(Table2).
Parameter estimates and 95% condence intervals averaged
from the seven best models indicated that movement rates by
E. servus were similar between years (Fig. 2). In both years,
attraction rates were similar between choices but leaving rates
from H. zea-damaged plants were signicantly greater than from
undamaged plants. Based on the model-averaged parameter
estimates, E. servus are, over the long term, signicantly
more likely to be associated with undamaged plants than
H. zea-damaged plants (Fig. 3).
Attraction rates of E. servus were similar between undam-
aged and H. virescens-damaged plants (Fig. 2). Leaving rates
from undamaged plants were signicantly greater than from
H. virescens-damaged plants. However, leaving rates were very
low overall. The slight differences in movement rates between
choices did not translate into signicant differences in equilibrial
probabilities (Fig. 3). While E. servus are slightly more likely to
be associated with H. virescens-damaged plants than undamaged
plants, the high uncertainty suggests little effect of H. virescens
on E. servus movement (Fig. 3).
Nezara viridula movement rates in 2009 were overall about
twice those in 2010 (Fig. 2). At the same time, both attrac-
tion and leaving rates were similar between choices. As a
result, N. viridula were equally likely to be associated with H.
zea-damaged plants than undamaged plants (Fig. 3). Results
were similar for trials with N. viridula and H. virescens, with
no evidence of preference (Table 2, Figs 2 and 3)
Boll consumption
Based on the paired t-tests, we found no signicant differences
between the damaged and undamaged plants in the initial or nal
volume of bolls presented to stink bugs (Table 3). The initial boll
volume was 5.28 ±0.27 cm3(mean ±SEM) and 5.58 ±0.28 cm3
for bolls on damaged and undamaged plants, respectively. The
nal volume for bolls on damaged and undamaged plants was
8.54 ±0.34 and 8.33 ±0.30 cm3, respectively. Helicoverpa zea
caterpillars consumed 54.3 ±4.0% of the boll volume, and H.
virescens caterpillars consumed 47.9 ±4.6% of the boll volume.
Based on  results, there were no signicant differences
in heliothine consumption between stink bug species, heliothine
species, or years (Table S2, Appendix S3). Collectively, these
tests indicate that differences in stink bug movement between
choices and between years were not due to variation in boll size
or consumption rates by the heliothines.
Euschistus servus nymphs produced 25.1 ±3.6 and 22.5 ±3.5
salivary sheath anges on damaged plants and undamaged
plants, respectively; N. viridula nymphs produced 21.4 ±2.6 and
19.4 ±2.7 sheath anges on damaged plants and undamaged
plants, respectively. We found no differences in the number of
sheath anges between stink bug species, heliothine species, or
years (Table S2, Appendix S3). Based on  results, vari-
ation in the number of sheath anges was signicantly and pos-
itively associated with stink bug tenure time, except for trials
with N. viridula and H. zea (Table 4). In contrast, the numbers
of sheath anges were not associated with heliothine damage. In
2009, E. servus produced marginally signicantly more sheath
anges on undamaged plants relative to H. zea-damaged plants,
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
6Adam R. Zeilinger et al.
Table 2 . Results from the AICcmodel selection procedure.
Stink bug Heliothine Model Model named.f. AICcΔAICc
E. servus H. zea M2b 𝜇by choice 3 110.0 0.0*
M2a p by choice 3 112.9 2.9*
M3a p and 𝜇by choice 4 113.7 3.7*
M2c p by year 3 114.4 4.4*
M1 One p and one 𝜇parameter 2 115.0 5.0*
M2d 𝜇by year 3 115.3 5.3*
M4a p by choice and year, 𝜇by choice 6 116.8 6.8*
M3b p and 𝜇by year 4 117.6 7.6
M4c p by choice and year, 𝜇by year 6 118.9 8.9
M4d p by year, 𝜇by choice and year 6 120.2 10.2
M4b p by choice, 𝜇by choice and year 6 123.1 13.1
M5 Fully parameterised 8 131.1 21.1
H. virescens M2c p by year 3 77.1 0.0*
M2a p by choice 3 78.6 1.5*
M1 One p and one 𝜇parameter 2 79.4 2.4*
M2b 𝜇by choice 3 80.4 3.3*
M3b p and 𝜇by year 4 81.0 3.9*
M3a p and 𝜇by choice 4 82.5 5.4*
M2d 𝜇by year 3 82.9 5.8*
M4c p by choice and year, 𝜇by year 6 85.5 8.5
M4a p by choice and year, 𝜇by choice 6 87.0 9.9
M4d p by year, 𝜇by choice and year 6 90.0 12.9
M4b p by choice, 𝜇by choice and year 6 95.2 18.1
M5 Fully parameterised 8 110.1 33.0
N. viridula H. zea M3b p and 𝜇by year 4 97.5 0.0*
M2c p by year 3 97.7 0.3*
M1 One p and one 𝜇parameter 2 98.4 0.9*
M2a p by choice 3 100.7 3.2*
M2b 𝜇by choice 3 101.3 3.8*
M2d 𝜇by year 3 101.4 3.9*
M3a p and 𝜇by choice 4 103.7 6.2*
M4c p by choice and year, 𝜇by year 6 106.2 8.7
M4d p by year, 𝜇by choice and year 6 106.7 9.2
M4a p by choice and year, 𝜇by choice 6 109.6 12.2
M4b p by choice, 𝜇by choice and year 6 113.3 15.9
M5 Fully parameterised 8 121.0 23.6
H. virescens M1 One p and one 𝜇parameter 2 46.7 0.0*
M2b 𝜇by choice 3 49.1 2.3*
M2a p by choice 3 49.4 2.7*
M3a p and 𝜇by choice 4 50.9 4.2*
*Models with ΔAICc<7 were considered good models and selected for averaging (Burnham et al., 2011).
Models are described in greater detail, including the parameters estimated, in Table 1.
Smaller ΔAICcvalues indicate a better t to the data relative to other models.
and signicantly more sheath anges on H. virescens-damaged
plants relative to undamaged plants (Table 3). Heliothine dam-
age had no effect on the number of sheath anges from N.
viridula (Table 3).
Induction of phenolic compounds
Concentrations of the four phenolic-based compounds
anthocyanins, chlorogenic acid, gallic acid, and rutin – were
signicantly different among H. zea-damaged, H. virescens-
damaged, and undamaged plants (Wilks’ 𝜆=0.286, F8,24 =
2.613, P=0.0327). The concentrations of anthocyanins, chloro-
genic acid, and gallic acid tended to be lower in bolls of plants
damaged by heliothine larvae than in bolls from undamaged
plants (Fig. 4). Univariate s for each compound indi-
cated that only chlorogenic acid was signicantly lower in bolls
from plants damaged by H. virescens compared with bolls from
undamaged plants (Table 5, Fig. 4).
Induction of gossypol and its derivatives
Six compounds were induced in cotton boll seeds by both
H. zea and H. virescens herbivory. Three of these compounds,
based on chemical standards and compounds registered in Pub
Chem, were gossypol, hemigossypol, and deoxyhemigossypol.
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
Herbivore-induced deterrence and attraction 7
Fig. 2. Model-averaged parameter estimates ±95% CIs of attraction
rates (a, c) and leaving rates (b, d) in 2009 (a, b) and 2010 (c, d) for
each combination of stink bug and heliothine species. All rates are in
units of stink bug individuals per hour. ‘Es-Hz’ indicates E. servus-H.
zea trials (lled circle); ‘Es-Hv’ indicates E. servus-H. virescens trials
(open circle); ‘Nv-Hz’ indicates N. viridula-H. zea trials (lled triangle);
and ‘Nv-Hv’ indicates N. viridula-H. virescens trials (open triangle).
Leaving rates for the E. servus-H. virescens combination in 2009 are
marginally signicantly different [damaged plants: 0.016 (0.004, 0.027)
estimate (±95% CI); undamaged plants: 0.011 (0.006, 0.015)]. Note the
different y-axis scales among panels. Variances are estimated using the
normal approximation method (Bolker, 2008). Parameter estimates and
CI are averaged from all models with asterisks (*) shown in Table 2.
The three other compounds are unknown; there are 108 com-
pounds registered (Pub Chem) that have structural similarity
to gossypol. None had molecular weights corresponding to the
other three compounds.
Our subsequent analysis focused on the identiable com-
pounds: gossypol, deoxyhemigossypol, and hemigossypol.
Peak areas of gossypol and its derivatives differed sig-
nicantly among undamaged controls, H. zea-damaged,
and H. virescens-damaged plants (Wilks’ 𝜆=5.55 ×107,
F12,14 =1565, P<0.0001). Univariate s indicated that
peak areas of each compound differed signicantly among treat-
ments: deoxyhemigossypol (N =5, F2,12 =22 179, P<0.0001);
gossypol (N =5, F2,12 =18 126, P<0.0001); hemigossypol
(N =5, F2,12 =15 513, P<0.0001). We found no peaks for
gossypol or its derivatives in undamaged plants (Fig. 5). Both
H. zea and H. virescens signicantly induced gossypol and its
derivatives relative to undamaged control plants and H. virescens
induction was signicantly greater than H. zea (Fig. 5).
Discussion
We investigated select behavioural and chemical mechanisms
underlying feeding preference of frugivorous stink bug species
Fig. 3. Estimated medians ±95% CIs of equilibrial probabilities (P1*,
P2*) that stink bugs are located on heliothine-damaged or undamaged
cotton plants in (a) 2009 and (b) 2010. Estimates were derived from
10 000 Monte Carlo simulations using model-averaged parameter esti-
mates shown in Fig. 2 and their variances. The horizontal dashed lines
indicate probabilities (Pj*)of0.5.
for heliothine-damaged and undamaged cotton plants. We esti-
mated choice-specic attraction and leaving rates using a
bio-statistical consumer movement model, and compared these
rates to heliothine-induced changes in phenolic and gossypol
production in developing cotton bolls.
While our insecticide applications targeting greenhouse pests
may have inuenced stink bug preference, the possible effects
may also be minor. Neonicotinoids can alter phytohormone lev-
els in cotton; however, jasmonic acid and salicylic acid lev-
els appear to be unaffected (Szczepaniec et al., 2013). Because
insecticides were applied more frequently in 2010 than in 2009,
we might expect insecticide effects to translate into signicant
year effects in our analyses. Only N. viridula showed signi-
cant differences in movement between years, with lower move-
ment in 2010. This year effect may be due to acetamiprid
applications, as N. viridula appears to be more susceptible
than E. servus to acetamiprid (Tillman & Mullinix, 2004; Till-
man, 2006). Nonetheless, the patterns of interactions reported
here – signicant deterrence of E. servus and no preference of
N. viridula are consistent with previous studies conducted in
eld plots free of foliar insecticides (Zeilinger et al., 2011).
We found strong species-specic interactions between stink
bug feeding preferences between heliothine-damaged and
undamaged cotton plants. Importantly, based on our 
and t-tests, stink bug preferences were not due to variation
in boll size or consumption rates between heliothine species.
Nezara viridula nymphs did not show any clear preferences
between heliothine-damaged and undamaged plants. However,
E. servus showed opposing responses to H. zea and H. virescens
herbivory.
Herbivory by H. zea clearly increased E. servus leaving
rates and marginally reduced the number of salivary sheath
anges. In contrast, herbivory by H. virescens marginally
reduced leaving rates and clearly increased E. servus con-
sumption as measured by salivary sheath anges (Zeilinger
et al., 2015). It appears that E. servus preference for and con-
sumption of bolls on herbivore-damaged plants are related.
However, different mechanisms may be responsible for the
different responses to H. zea and H. virescens herbivory:
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
8Adam R. Zeilinger et al.
Table 3 . Statistics for paired t-tests for comparisons between damaged and undamaged plants.
Response variable Stink bug Heliothine Year d.f. MeanLower CI Upper CI
Initial stink bug boll volume E. servus H. zea 2010 14 0.94 0.83 1.06
E. servus H. virescens 2010 13 0.97 0.89 1.05
N. viridula H. zea 2010 11 1.01 0.93 1.10
N. viridula H. virescens 2010 20 0.97 0.85 1.09
Final stink bug boll volume E. servus H. zea 2009 14 1 0.89 1.11
2010 14 0.95 0.79 1.11
H. virescens 2009 11 1 0.90 1.10
2010 14 0.98 0.88 1.07
N. viridula H. zea 2009 18 1.03 0.94 1.12
2010 11 1.01 0.87 1.16
H. virescens 2010 21 1.02 0.84 1.20
Salivary sheath anges E. servus H. zea 2009 14 0.52 0.03 1.01
2010 14 0.91 0.38 1.53
H. virescens 2009 11 1.76 1.28 2.23*
2010 14 1.24 0.70 1.77
N. viridula H. zea 2009 18 1.09 0.60 1.58
2010 11 0.93 0.38 1.49
H. virescens 2010 21 1.18 0.90 1.46
*98.3% CIs do not overlap 1. Critical value (𝛼=0.017) adjusted according to Bonferroni’s correction with k =3 test.
Means of Kogan and Goeden’s (1970) preference index (PI). Values of 0 PI<1 indicate that the response is greater on the undamaged plant than the
damaged plant, whereas values of 1 >PI 2 indicate the opposite.
Table 4 . Statistics for  of the number of salivary sheath anges (response variable)and time spent on boll, the heliothine consumption score,
and year (explanatory variables)
.
Stink bug Heliothine Explanatory variable Estimate t statistic P-value R2d.f.§
E. servus H. zea Intercept 2.35 2.72 0.012* 0.68 4,26
Year 0.78 1.61 0.119
Tenure time 0.22 6.85 <0.001***
Consumption score 0.01 0.83 0.415
Year ×tenure time 0.02 0.67 0.512
H. virescens Intercept 0.96 0.68 0.505 0.45 4,20
Year 0.43 0.39 0.698
Tenure time 0.13 3.61 0.002**
Consumption score 0.01 0.91 0.374
Year ×tenure time 0.02 0.42 0.680
N. viridula H. zea Intercept 2.77 2.83 0.010* 0.21 4,22
Year 0.49 0.77 0.449
Tenure time 0.06 1.67 0.109
Consumption score <0.01 0.29 0.774
Year ×tenure time 0.01 0.26 0.800
H. virescens Intercept 3.23 4.24 <0.001** 0.40 2,18
Tenure time 0.12 3.44 0.003**
Consumption score <0.01 0.26 0.797
*P-value <0.05; **P-value <0.01; ***P-value <0.001.
Salivary sheath ange data were square-root transformed to meet assumptions of normality and constant error variance.
All data analysed were from the heliothine-damaged plants only.
§The rst number indicates the numerator degrees of freedom and second number indicates the denominator d.f.
H. zea had a stronger relative effect on host choice whereas
H. virescens had a stronger effect on consumption. Without
information on both movement and consumption, our infer-
ences on stink bug preference would have been erroneous, weak,
or both.
Euschistus servus host choice was inuenced entirely by
differential leaving rates. Herbivore leaving rates could be
inuenced by a wide range of cues (Bernays & Chapman,
1994; Schoonhoven et al., 2005). While visual and olfactory
cues cannot be ruled out, tactile or gustatory (i.e. tissue-bound)
cues may be more important. Plant volatile organic compounds
(VOCs) generally appear to be important cues for predaceous
and herbivorous stink bugs (Pavis & Renou, 1990; van Loon
et al., 2000; Weissbecker et al., 2000). However, given that stink
bugs respond to low concentrations of VOCs (Pavis & Renou,
1990) and as attraction rates remained equivalent, it seems
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
Herbivore-induced deterrence and attraction 9
Fig. 4. The mean ±SEM concentrations (mg g1dried boll carpel
wall) of phenolic compounds in boll carpel walls in three treatments:
undamaged plants (black bars), H. virescens-damaged plants (grey bars),
and H. zea-damaged plants (white bars). Treatment means and SEMs
are adjusted for block (week) means and back-transformed. The same
letter over bars within the same phenolic species indicates that those bars
are not signicantly different according to Tukey’s HSD for multiple
comparisons.
Table 5 . Statistics for univariate s of concentrations of
phenolic-based compounds in boll carpel walls (response variable) and
week and herbivory treatment (explanatory variables).
Phenolic species Effect d.f. FP-value
Anthocyanin Week 1 1.509 0.238
Treatment2 1.419 0.272
Error 15
Chlorogenic acid Week 1 8.425 0.010*
Treatment 2 6.813 0.008*
Error 15
Gallic acid Week 1 0.093 0.765
Treatment 2 3.447 0.059
Error 15
Rutin Week 1 0.155 0.700
Treatment 2 0.217 0.807
Error 15
*P-value <0.0125. Critical value (𝛼) adjusted according to Bonferroni’s
correction with k =4 tests.
Treatments included undamaged plants, H. virescens-damaged plants,
and H. zea-damaged plants.
unlikely that VOCs are responsible for the differential leaving
rates.
We suggest that differential leaving rates and changes in con-
sumption may be associated with gustatory cues from induced
tissue-bound secondary metabolites. If different phytochemi-
cal mechanisms explain E. servus responses to H. zea than
to H. virescens herbivory – as is suggested by our choice
and consumption results – then we hypothesise that reductions
in chlorogenic acid concentrations in boll carpel walls may
Fig. 5. The mean ±SEM peak area [ion current (m/z)] for deoxy-
hemigossypol (D), gossypol (G), and hemigossypol (H) systemically
induced in seeds from developing cotton bolls from plants damaged by
H. zea,H. virescens, or left undamaged. Asterisks (*) indicate that peak
areas were different among all treatments according Tukey’s HSD for
multiple comparisons. Peak areas for all three compounds were 0 ±0.0
for undamaged control plants.
explain E. servus enhanced consumption on bolls from H.
virescens-damaged plants. Chlorogenic acid is known to be a
powerful anti-herbivore compound (Bi et al., 1997) and toxic to
predatory stink bugs (Stamp et al., 1997). However, chlorogenic
acid is unlikely to explain E. servus responses to H. zea dam-
age, as we found a negligible change in systemic chlorogenic
acid levels in H. zea-damaged plants. Additionally, while we
found strong induction of gossypol from heliothine herbivory,
these patterns are inconsistent with E. servus responses. Inter-
estingly, both heliothines induced gossypol in cotton seeds, even
though H. zea suppresses induction of these compounds in cot-
ton squares and leaves (Bi et al., 1997; Olson et al., 2008). To our
knowledge, this is the rst report that H. zea induced elevated
gossypol levels in cotton tissues. Furthermore, while we were
unable to clearly identify the precise plant secondary metabo-
lites responsible for E. servus preference and consumption, our
behavioural and chemical results support previous reports that
H. zea and H. virescens induce distinct, species-specic, and
ecologically relevant responses in cotton plants (De Moraes
et al., 1998).
We detected both competitive and facilitative interactions
between stink bug nymphs and caterpillar herbivores of cot-
ton. While less common than competition, facilitation is not a
rare outcome of interactions among herbivorous insects. Eleven
per cent of interactions examined by Kaplan and Denno (2007)
were facilitative, with 84% of facilitative interactions mediated
by induced plant responses. Furthermore, plant-mediated inter-
actions within the same system can vary between competition
and facilitation depending on multiple factors, including plant
genotype (Cronin & Abrahamson, 1999), leaf phenology and
nutrition (Rieske & Raffa, 1998), and differential elicitation of
defence signalling pathways by plant pathogens (Thaler et al.,
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
10 Adam R. Zeilinger et al.
2010). The apparent prevalence of plant-mediated competition
and facilitation within the same system warrants further investi-
gation into the biochemical mechanisms as well as further theory
development.
Based on our Monte Carlo simulations, we expect differences
in leaving rates of E. servus to translate into signicantly
more nymphs associated with undamaged plants than with
H. zea-damaged plants. The differences in leaving rates between
H. virescens-damaged and undamaged plants are insufcient
to inuence signicantly the expected E. servus population
density. Our results reported here and in Zeilinger et al. (2011)
indicate that changes in density of H. zea larvae in a cotton eld
are predicted to affect the spatial distribution and abundance
of E. servus in cotton, as suggested for other inter-specic
interactions among herbivores (Underwood, 2004). Heliothine
damage in cotton elds in the southeastern U.S. can be extensive
and severe (Greene et al., 2011), and even under moderate
levels of heliothine damage, E. servus responses to heliothine
herbivory could have population-level impacts. For example,
herbivore acceptance of a high-quality host can be disrupted
merely by the proximity of low-quality hosts (Bernays, 1999;
Underwood et al., 2011).
Acknowledgements
This work was supported by the National Research Initiative of
the US Department of Agriculture, National Institute of Food
and Agriculture (grant number 2008-35302-04709 to DAA and
DMO); the US National Science Foundation IGERT program
(grant number 0653827 to the University of Minnesota); and
a Thesis Research Grant, a Doctoral Dissertation Fellowship,
and grants from the Dayton-Wilkie Fund of the Bell Museum
of Natural History, University of Minnesota (to A.R.Z.). We
would also like to thank A. Hornbuckle, M. Smith, A. Zittrouer,
T. Brown, and J. Ransom for assistance in performing the
experiments; C. Roberts, C. Blanco, and M. Toews for providing
insects; T. Potter for assisting with terpenoid chemical analysis;
P. Oikawa for illustrating our experimental design; and G.
Heimpel, M. Daugherty, J. Thaler, D. Hare, two anonymous
reviewers, and an associate editor for helpful comments on early
drafts of this paper. We also thank the R Help on-line community
for assistance with R programming. A. R. Zeilinger designed
and conducted the preference experiment, with assistance from
D. A. Andow and D. M. Olson. D. M. Olson and D. MacLean
designed and conducted the chemical analysis of phenolic
compounds. D. M. Olson, N. Mori, and R. Nakata designed and
conducted the analysis of terpenoid analysis. A. R. Zeilinger and
D. A. Andow designed the statistical analysis. A. R. Zeilinger
wrote the manuscript with assistance from all co-authors.
Supporting Information
Additional Supporting Information may be found in the online
version of this article under the DOI reference:
10.1111/een.12221
Appendix S1. HPLC and LC/MS methods for secondary
metabolites in cotton bolls.
Appendix S2. R script for maximum likelihood estimation
and AIC model selection.
Appendix S3. Additional analyses: testing model assump-
tions and  results.
References
Agrawal, A.A. (1999) Induced responses to herbivory in wild
radish: effects on several herbivores and plant tness. Ecology,
80, 1713– 1723.
Agrawal, A.A. (2000) Benets and costs of induced plant defense for
Lepidium virginicum (Brassicaceae). Ecology,81, 1804– 1813.
Anderson, P., Sadek, M.M. & Wäckers, F.L. (2011) Root herbivory
affects oviposition and feeding behavior of a foliar herbivore. Behav-
ioral Ecology,22, 1272– 1277.
Beckerman, A.P. (2000) Counterintuitive outcomes of interspecic
competition between two grasshopper species along a resource
gradient. Ecology,81, 948– 957.
Bernays, E.A. (1999) When host choice is a problem for a generalist
herbivore: experiments with the whitey, Bemisia tabaci.Ecological
Entomology,24, 260– 267.
Bernays, E.A. & Chapman, R.F. (1994) Host-Plant Selection by Phy-
tophagous Insects.Contemporary Topics in Entomology. Chapman &
Hall, New York, New York.
Bi, J.L., Murphy, J.B. & Felton, G.W. (1997) Antinutritive and oxidative
components as mechanisms of induced resistance in cotton to Heli-
coverpa zea.Journal of Chemical Ecology,23, 97– 117.
Bolker, B.M. (2008) Ecological Models and Data in R. Princeton
University Press, Princeton, New Jersey.
Bolker, B. & R Core Team. (2014) bbmle: Tools for General Maximum
Likelihood Estimation. R package, version 1.0.17.
Burnham, K.P., Anderson, D.R. & Huyvaert, K.P. (2011) AIC model
selection and multimodel inference in behavioral ecology: some
background, observations, and comparisons. Behavioral Ecology and
Sociobiology,65, 23– 35.
Clarke, D., Whitney, H., Sutton, G. & Robert, D. (2013) Detection and
learning of oral electric elds by bumblebees. Science,340, 66– 69.
Cronin, J.T. & Abrahamson, W.G. (1999) Host-plant genotype and
other herbivores inuence goldenrod stem galler preference and
performance. Oecologia,121, 392– 404.
De Moraes, C.M., Lewis, W.J., Pare, P.W., Alborn, H.T. & Tumlinson,
J.H. (1998) Herbivore-infested plants selectively attract parasitoids.
Nature,393, 570– 573.
Evangelista, W.S. Jr., Santos, R.L., Torres, J.B. & Zanuncio, J.C. (2011)
Effect of gossypol on survival and reproduction of the zoophy-
tophagous stinkbug Podisus nigrispinus (Dallas). Revista Brasileira
de Entomologia,55, 267– 271.
Fisher, A.E.I., Hartley, S.E. & Young, M. (1999) Behavioural responses
of the leaf-chewing guild to the presence of Eriocrania mines on silver
birch (Betula pendula). Ecological Entomology,24, 156– 162.
Fitt, G.P. (1989) The ecology of Heliothis species in relation to
agroecosystems. Annual Review of Entomology,34, 17– 53.
Gianoli, E. (2000) Competition in cereal aphids (Homoptera: Aphididae)
on wheat plants. Environmental Entomology,29, 213–219.
Greene, J.K., Robinson, D.M., Carter, K.M. & Devinney, G.M. (2011)
Performance of new and existing Bt cotton technologies when
inundated with heavy/natural populations of bollworm in SC-2010.
Proceedings of the Beltwide Cotton Conference. Presented at the
Beltwide Cotton Conference. Cotton, Inc., Atlanta, Georgia, pp.
993– 999.
Gutbrodt, B., Mody, K. & Dorn, S. (2011) Drought changes plant chem-
istry and causes contrasting responses in lepidopteran herbivores.
Oikos,120, 1732– 1740.
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
Herbivore-induced deterrence and attraction 11
Hagenbucher, S., Olson, D.M., Ruberson, J.R., Wäckers, F.L. & Romeis,
J. (2013) Resistance mechanisms against arthropod herbivores in
cotton and their interactions with natural enemies. Critical Reviews
in Plant Sciences,32, 458– 482.
Halitschke, R., Stenberg, J.A., Kessler, D., Kessler, A. & Baldwin,
I.T. (2008) Shared signals –“alarm calls” from plants increase
apparency to herbivores and their enemies in nature. Ecology Letters,
11, 24– 34.
Kamminga, K.L., Herbert, D.A. Jr., Kuhar, T.P., Malone, S. & Koppel,
A. (2009) Efcacy of insecticides against Acrosternum hilare and
Euschistus servus (Hemiptera: Pentatomidae) in Virginia and North
Carolina. Journal of Entomological Science,44,1.
Kaplan, I. & Denno, R.F. (2007) Interspecic interactions in phy-
tophagous insects revisited: a quantitative assessment of competition
theory. Ecology Letters,10, 977–994.
Karban, R. & Baldwin, I.T. (1997) Induced Responses to Herbivory.
University of Chicago Press, Chicago, Illinois.
Kogan, M. & Goeden, R.D. (1970) The host-plant range of Lema
trilineata daturaphila (Coleoptera: Chrysomelidae). Annals of the
Entomological Society of America,63, 1175– 1180.
Lockwood, J.R. III (1998) On the statistical analysis of multiple-choice
feeding preference experiments. Oecologia,116, 475– 481.
Long, J.D., Hamilton, R.S. & Mitchell, J.L. (2007) Asymmetric compe-
tition via induced resistance: specialist herbivores indirectly suppress
generalist preference and populations. Ecology,88, 1232– 1240.
van Loon, J.J., de Vos, E.W. & Dicke, M. (2000) Orientation behaviour
of the predatory hemipteran Perillus bioculatus to plant and prey
odours. Entomologia Experimentalis et Applicata,96, 51– 58.
McPherson, J.E. & McPherson, R.M. (2000) Stink Bugs of Economic
Importance in America North of Mexico. CRC Press, Boca Raton,
Florida.
Messina, F.J., Taylor, R. & Karren, M.E. (2002) Divergent responses
of two cereal aphids to previous infestation of their host plant.
Entomologia Experimentalis et Applicata,103, 43– 50.
Miles, P.W. (1972) The saliva of Hemiptera. Advances in Insect Physi-
ology,9, 183– 255.
Nicotri, M. (1980) Factors involved in herbivore food preference.
Journal of Experimental Marine Biology and Ecology,42, 13– 26.
Oehlert, G.W. (2000) A First Course in Design and Analysis of
Experiments. WH Freeman, New York, New York.
Panizzi, A.R. & Slansky, F. Jr. (1991) Suitability of selected legumes and
the effect of nymphal and adult nutrition in the southern green stink
bug (Hemiptera: Heteroptera: Pentatomidae). Journal of Economic
Entomology,84, 103– 113.
Pareja, M., Mohib, A., Birkett, M.A., Dufour, S. & Glinwood, R.T.
(2009) Multivariate statistics coupled to generalized linear mod-
els reveal complex use of chemical cues by a parasitoid. Animal
Behaviour,77, 901–909.
Pavis, C. & Renou, M. (1990) Etude de la forme des reponses électroan-
tenographiques chez un hétéroptère pentatomidae [Nezara viridula
(L.)] =Study of the shape of the electroantennogram responses in
Nezara viridula (L.)(Heteroptera, Pentatomidae). Comptes Rendus de
l’Académie des Sciences. Série 3, Sciences de la Vie,310, 521–526.
Preisser, E.L. & Bastow, J.L. (2006) Plant damage from and defenses
against “cryptic” herbivory: a guild perspective. Journal of Plant
Interactions,1, 197– 210.
R Core Team (2014) R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria.
Rieske, L.K. & Raffa, K.F. (1998) Interactions among insect herbivore
guilds: inuence of thrips bud injury on foliar chemistry and suitabil-
ity to gypsy moths. Journal of Chemical Ecology,24, 501– 523.
Schoonhoven, L.M., van Loon, J.J.A. & Dicke, M. (2005) Insect-Plant
Biology. Oxford University Press, Oxford, U.K.
Shearer, P.W. & Jones, V.P. (1996) Diel feeding pattern of adult female
southern green stink bug (Hemiptera: Pentatomidae). Environmental
Entomology,25, 599– 602.
Slansky Jr, F. & Panizzi, A.R. (1987) Nutritional ecology of
seed-sucking insects. Nutritional Ecology of Insects, Mites, Spi-
ders, and Related Invertebrates (ed. by Slansky Jr, F.), pp. 283 320.
Wiley, New York, New York.
Stamp, N.E., Yang, Y. & Osier, T.L. (1997) Response of an insect
predator to prey fed multiple allelochemicals under representative
thermal regimes. Ecology,78, 203– 214.
Szczepaniec, A., Raupp, M.J., Parker, R.D., Kerns, D. & Eubanks, M.D.
(2013) Neonicotinoid insecticides alter induced defenses and increase
susceptibility to spider mites in distantly related crop plants. PLoS
One,8, e62620.
Thaler, J.S., Agrawal, A.A. & Halitschke, R. (2010) Salicylate-mediated
interactions between pathogens and herbivores. Ecology,91,
1075– 1082.
Tillman, P.G. (2006) Susceptibility of pest Nezara viridula (Het-
eroptera: Pentatomidae) and parasitoid Trichopoda pennipes (Diptera:
Tachinidae) to selected insecticides. Journal of Economic Entomol-
ogy,99, 648–657.
Tillman, P.G. & Mullinix, B.G. (2004) Comparison of susceptibility
of pest Euschistus servus and predator Podisus maculiventris (Het-
eroptera: Pentatomidae) to selected insecticides. Journal of Economic
Entomology,97, 800– 806.
Underwood, N. (2004) Variance and skew of the distribution of
plant quality inuence herbivore population dynamics. Ecology,85,
686– 693.
Underwood, N., Halpern, S. & Klein, C. (2011) Effect of host-plant
genotype and neighboring plants on strawberry aphid movement in
the greenhouse and eld. American Midland Naturalist,165, 38–49.
Weissbecker, B., van Loon, J.J., Posthumus, M.A., Bouwmeester, H.J.
& Dicke, M. (2000) Identication of volatile potato sesquiterpenoids
and their olfactory detection by the two-spotted stinkbug Perillus
bioculatus.Journal of Chemical Ecology,26, 1433– 1445.
Whitehead, S.R. & Poveda, K. (2011) Herbivore-induced changes in
fruit-frugivore interactions. Journal of Ecology,99, 964–969.
Zeilinger, A.R., Olson, D.M. & Andow, D.A. (2011) Competition
between stink bug and heliothine caterpillar pests on cotton at
within-plant spatial scales. Entomologia Experimentalis et Applicata,
141, 59– 70.
Zeilinger, A.R., Olson, D.M. & Andow, D.A. (2014) A likelihood-based
biostatistical model for analyzing consumer movement in simultane-
ous choice experiments. Environmental Entomology,43, 977– 988.
Zeilinger, A.R., Olson, D.M., Raygoza, T. & Andow, D.A. (2015) Do
counts of salivary sheath anges predict food consumption in herbivo-
rous stink bugs (Hemiptera: Pentatomidae)? Annals of the Entomolog-
ical Society of America,108, 109– 116. DOI: 10.1093/aesa/sau011.
Zytynska, S.E. & Preziosi, R.F. (2013) Host preference of plant geno-
types is altered by intraspecic competition in a phytophagous insect.
Arthropod-Plant Interactions,7, 349– 357.
Accepted 30 April 2015
Published 2015. This article is a U.S. Government work and is in the public domain in the USA., Ecological Entomology, doi: 10.1111/een.12221
... In addition, there is mounting evidence that the strong reduction of lepidopteran populations in Bt cotton and, associated therewith, altered interspecific interactions among species also benefits non-target herbivores 15 . Stink bugs as well as cotton aphids, Aphis gossypii (Hemiptera: Aphididae) can benefit from the release of either direct interference competition or plant-mediated indirect competition with Bt-sensitive Lepidoptera [15][16][17][18][19] . There is evidence that plant-mediated indirect competition in cotton is partly driven by inducible defensive compounds. ...
... Squares as a food source had a positive effect on L. hesperus performance when compared with bolls despite the fact that squares contained much higher concentrations of gossypol (Fig. 3). That cotton terpenoids might not be responsible for plant-mediated indirect competition between caterpillars and sucking bugs was also suggested by Zeilinger et al. 19 . They found that the boll-feeding stink bug E. servus avoided cotton plants damaged by caterpillars of Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae) while it was attracted to plants damaged by Heliothis virescens Fabricius (Lepidoptera: Noctuidae). ...
... Given the large array of different cotton defenses against caterpillars, it is most likely that other potentially inducible defense mechanisms, such as chlorogenic acid, condensed tannins, or other phenolic compounds might explain the negative effects of cotton induction on L. hesperus performance 19,21,40 . Although C:N ratios in plants had no effect on L. hesperus performance, we cannot rule out that other changes in cottons nutritional quality affected L. hesperus as it has been reported that caterpillar damage can affect amino acid composition, water content or the oxidative status in cotton 40,41 . ...
Article
Full-text available
Bt cotton was genetically modified to produce insecticidal proteins targeting Lepidopteran pests and is therefore only minimally affected by caterpillar damage. This could lead to reduced levels of inherent, systemically inducible defensive compounds in Bt cotton which might benefit other important cotton herbivores such as plant bugs. We studied the effects of plant defense induction on the performance of the plant bug Lygus hesperus by caging nymphs on different food sources (bolls/squares) of Bt and non-Bt cotton which were either undamaged, damaged by Bt tolerant caterpillars, or treated with jasmonic acid (JA). Terpenoid induction patterns of JA-treated and L. hesperus-damaged plants were characterized for different plant structures and artificial diet assays using purified terpenoids (gossypol/heliocide H1/4) were conducted. Nymphs were negatively affected if kept on plants damaged by caterpillars or sprayed with JA. Performance of nymphs was increased if they fed on squares and by the Bt-trait which had a positive effect on boll quality as food. In general, JA-sprayed plants (but not L. hesperus infested plants) showed increased levels of terpenoids in the plant structures analyzed, which was especially pronounced in Bt cotton. Nymphs were not negatively affected by terpenoids in artificial diet assays indicating that other inducible cotton responses are responsible for the found negative effects on L. hesperus. Overall, genetically engineered plant defenses can benefit plant bugs by releasing them from plant-mediated indirect competition with lepidopterans which might contribute to increasing numbers of hemipterans in Bt cotton.
... If so, competitive release may play an important role in stink bug outbreaks in Bt cotton. Herbivory by H. zea larvae on non-Bt cotton plants reduces nymphal growth rates of E. servus by 36% at within-plant scales and deters feeding by E. servus nymphs at the whole-plant scale (Zeilinger et al. 2011Zeilinger et al. , 2015 ). Moreover, the speciesspecific and multi-scale nature of these interactions indicate that they are mediated by induced resistance in the shared host plant (Zeilinger et al. 2015). ...
... Herbivory by H. zea larvae on non-Bt cotton plants reduces nymphal growth rates of E. servus by 36% at within-plant scales and deters feeding by E. servus nymphs at the whole-plant scale (Zeilinger et al. 2011Zeilinger et al. , 2015 ). Moreover, the speciesspecific and multi-scale nature of these interactions indicate that they are mediated by induced resistance in the shared host plant (Zeilinger et al. 2015). We aim to further test the competitive release hypothesis. ...
... servus and N. viridula), we compiled previously published results on their interactions into a meta-analysis. We calculated the effect size as Hedges's d from competition experiments described in Zeilinger et al. (2011Zeilinger et al. ( , 2015). Each species combination consisted of n = 4 effect sizes, except the N. viridula and H. virescens combination in which n = 3. ...
Article
Full-text available
The adoption of transgenic Bt cotton has, in some cases, led to environmental and economic benefits through reduced insecticide use. However, the distribution of these benefits and associated risks among cotton growers and cotton-growing regions has been uneven due in part to outbreaks of non-target or secondary pests, thereby requiring the continued use of synthetic insecticides. In the southeastern USA, Bt cotton adoption has resulted in increased abundance of and damage from stink bug pests, Euschistus servus and Nezara viridula (Heteroptera: Pentatomidae). While the impact of increased stink bug abundance has been well-documented, the causes have remained unclear. We hypothesize that release from competition with Bt-susceptible target pests may drive stink bug outbreaks in Bt cotton. We first examined the evidence for competitive release of stink bugs through meta-analysis of previous studies. We then experimentally tested if herbivory by Bt-susceptible Helicoverpa zea increases stink bug leaving rates and deters oviposition on non-Bt cotton. Consistent with previous studies, we found differences in leaving rates only for E. servus, but we found that both species strongly avoided ovipositing on H. zea-damaged plants. Considering all available evidence, competitive release of stink bug populations in Bt cotton likely contributes to outbreaks, though the relative importance of competitive release remains an open question. Ecological risk assessments of Bt crops and other transgenic insecticidal crops would benefit from greater understanding of the ecological mechanisms underlying non-target pest outbreaks and greater attention to indirect ecological effects more broadly.
Article
The corn earworm, Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae), is a cosmopolitan pest in the field crop landscape in the southeastern United States. Field corn (Zea mays L.) is the most important midseason host for H. zea where intensive selection pressure occurs for resistance to insecticidal toxins from Bacillus thuringiensis (Bt). Because spatial patterns of H. zea in field corn have not been extensively studied, field corn was sampled for H. zea larvae and injury in 2021 and 2022. Patterns of spatial aggregation were identified in a number of fields in both larval populations and injury. Aggregation of H. zea larvae was less common at R5 than at R2. Associations between the spatial patterns of H. zea and the variability in crop phenology were identified in some fields, with positive associations between plant height and H. zea larvae, indicating that ovipositing H. zea moths avoid areas with reduced plant height and delayed reproductive maturity. Additionally, negative spatial associations between stink bug ear injury and H. zea larvae and their injury were found in a small number of cases, indicating some spatial interactions between the two pest complexes and their injury. Results from these studies provide valuable insight into the spatial patterns of H. zea in field corn. An understanding of the local dispersal and population dynamics of H. zea can be used to help further improve integrated pest management and insect resistance management programs for this major polyphagous pest.
Book
Many scientists have reported an extensive amount of information on the biology, life history, and damage potential of stink bugs. However, this information is scattered among numerous journals, periodicals, and other publications. Stink Bugs of Economic Importance in America North of Mexico brings together the applied and nonapplied literature in one complete and concise format. The book gives you: • Section by section discussions of various economic stink bug species and damage to individual crops • Separate tables of host plants organized by common name, scientific name, and family name • General biology for each economic stink bug species Strategies for the control of destructive species • Keys for identification of stink bug species • Numerous unique line drawings • Over 700 references on stink bug publications Written by two top-notch researchers whose experience is complementary, the book examines these constant pests. The first comprehensive resource on this fascinating and destructive group of insects, Stink Bugs of Economic Importance in America North of Mexico provides you with a reference that you can use in the laboratory or in the field for easy identification of pentatomids.
Article
We briefly outline the information-theoretic (I-T) approaches to valid inference including a review of some simple methods for making formal inference from all the hypotheses in the model set (multimodel inference). The I-T approaches can replace the usual t tests and ANOVA tables that are so inferentially limited, but still commonly used. The I-T methods are easy to compute and understand and provide formal measures of the strength of evidence for both the null and alternative hypotheses, given the data. We give an example to highlight the importance of deriving alternative hypotheses and representing these as probability models. Fifteen technical issues are addressed to clarify various points that have appeared incorrectly in the recent literature. We offer several remarks regarding the future of empirical science and data analysis under an I-T framework.
Article
Publisher Summary Salivary function is especially interesting in Hemiptera because of the effects the saliva has on the living and surviving organisms, on which many of these insects feed. The saliva of Hemiptera is by no means a simple secretion—in addition to the usual salivary functions of moistening food and mixing it with hydrolytic enzymes before ingestion, the saliva of phytophagous species plays an important physiochemical role during the mechanical penetration of plant tissues by the piercing and sucking mouthparts; in accomplishing this task, the saliva may vary in its chemical composition and physical consistency from one moment to the next. Moreover, deposits of solidifying components of the saliva of many species persist in the food plants, modifying the long term effects of feeding by the insects. This chapter compiles the various types of investigation on salivary functions in the Homoptera and Heteroptera, and suggests profitable lines of future investigation based on analogous functions in different taxonomic groups.