![Example of (A) RGB digital image, (B) binary image of color segmentation of saturation layer of an HSI image in the range of [0.03, 1], and (C) saturation layer after color segmentation. The label of the filter paper (in panel A) was digitally removed prior to color segmentation.](https://www.researchgate.net/profile/Dawn-Olson-2/publication/272020674/figure/fig1/AS:295104746147841@1447369862399/Example-of-A-RGB-digital-image-B-binary-image-of-color-segmentation-of-saturation_Q320.jpg)
Content uploaded by Dawn Olson
Author content
All content in this area was uploaded by Dawn Olson on Feb 09, 2015
Content may be subject to copyright.
ARTHROPOD BIOLOGY
Do Counts of Salivary Sheath Flanges Predict Food
Consumption in Herbivorous Stink Bugs (Hemiptera:
Pentatomidae)?
ADAM R. ZEILINGER,
1,2,3
DAWN M. OLSON,
4
TANIA RAYGOZA,
5
AND DAVID A. ANDOW
6
Ann. Entomol. Soc. Am. 1–8 (2015); DOI: 10.1093/aesa/sau011
ABSTRACT Counts of salivary sheaths and salivary flanges have been widely used in studies of feeding
behavior and crop damage of pestiferous stink bugs (Hemiptera: Pentatomidae) and other sheath-
feeding Hemiptera. While salivary flanges can effectively predict crop damage by stink bugs, previous
studies have assumed that food consumption (e.g., ingestion) and preference can also be inferred from
flange data. Yet this assumption has remained untested. We investigated the relationship between the
number of stink bug salivary flanges and consumption of cotton bolls for two important agricultural pest
species: Nezara viridula (L.) and Euschistus servus (Say). We inferred food consumption rates from mea-
sures of relative growth rate and excreta quantity. To measure excreta, we quantified the color intensity,
or chromaticity, of excreta using digital image analysis. We found a positive relationship between growth
rate and the number of flanges for fifth instars of E. servus. However, we found no relationship between
growth or excretion and the number of flanges for all stages of N. viridula and for E. servus adults. Our
results indicate that counts of salivary flanges should not be used to infer food consumption or preference
in studies on N. viridula and E. servus adults, but can be used in studies of E. servus nymphs. Species-
and stage-specific differences in the relationship between consumption and salivary flanges suggests
distinct feeding strategies among species and stages; such differences may be potentially important in
determining crop damage from pestiferous stink bugs.
KEY WORDS stylet, feeding preference, Pentatomidae, salivary sheath, haustellate
Introduction
Among the diverse feeding strategies of the Hemiptera,
salivary-sheath feeding appears to be particularly com-
mon among the phloem- and seed-feeding herbivorous
guilds (Miles 1972). Although the function of salivary
sheaths in feeding remains unclear, their use is com-
mon among the Sternorrhyncha, Auchenorrhyncha,
and the Pentatomomorpha families of the Heteroptera
(Miles 1972,Morgan et al. 2013).
For pentatomid stink bug agricultural pests, the
number of salivary sheaths and salivary flanges—the
gelling saliva deposited and visible on the exterior sur-
face of a food item—are good predictors of the loss of
crop yield or quality from stink bug feeding (Bowling
1979,1980;Viator et al. 1983;Barbour and Bacheler
1990;Bundy et al. 2000). Damage from stink bug
feeding is generally more closely related to the number
of stylet probing events (i.e., probe frequency) than the
total amount of plant fluids consumed (Kawamoto
et al. 1987;sensu stricto Backus 2000). This may be
due to the mechanisms by which stink bug feeding re-
duces crop yield and quality. Damage is caused through
three mechanisms: mechanical damage by probing with
their stylets (Depieri and Panizzi 2011), oxidative dam-
age by the injection of digestive enzymes present in
their saliva (Depieri and Panizzi 2011,Peiffer and
Felton 2014), and the introduction of microbial patho-
gens into the fruiting structure (Hollay et al. 1987;
Medrano et al. 2007,2009). Thus, the number of sali-
vary sheaths or flanges has been a good predictor of
crop damage from stink bug feeding.
Some authors have also used salivary sheaths or
flanges of herbivorous pentatomid species to infer feed-
ing activity and preference (Bowling 1979,1980;Kester
et al. 1984;Lye and Story 1988;Simmons and Yeargan
1988;Panizzi et al. 1995). Such inferences are based on
the assumption of a positive relationship between the
number of flanges and food consumption. However,
the existence of such a relationship in the herbivorous
Pentatomidae has remained untested.
Flanges are a record of the number of stylet penetra-
tions or probes, not necessarily the amount of plant
biomass consumed or ingested. Further, there is some
evidence that the relationship is not a simple one.
1
Conservation Biology Program, Department of Entomology, Uni-
versity of Minnesota, St. Paul, MN 55108.
2
Present Address: Berkeley Initiative for Global Change Biology,
130 Mulford Hall #3114, University of California, Berkeley.
3
Corresponding author, e-mail: arz@berkeley.edu.
4
Crop Protection, Research, and Management Unit, USDA-ARS,
Tifton, GA 31794.
5
Department of Entomology, University of California Riverside,
Riverside, CA 92507.
6
Department of Entomology and Center for Community Genetics,
University of Minnesota, St. Paul, MN 55108.
Published by Oxford University Press on behalf of Entomological Society of America 2015.
This work is written by US Government employees and is in the public domain in the US
Pentatomid- and aphid-resistant crop varieties reduced
herbivore development but had no effect on the num-
ber of salivary flanges produced (Kester et al. 1984,Ni
and Quisenberry 1997). The coreid bug Clavigralla
scutellaris (Westwood) produces consistently fewer sali-
vary deposits inside pea pods than on the exterior sur-
face of the pod (i.e., flange), indicating that food is not
ingested from every flange (Mitchell et al. 2004).
Panizzi et al. (1995) showed a poor relationship be-
tween mean number of salivary flanges and feeding
duration across nymphal stages of Nezara viridula (L.)
and suggested that the relationship between consump-
tion rate and number of flanges depends on an individ-
ual’s level of starvation. Likewise, in a two-choice
preference experiment, Zeilinger (2011) found a posi-
tive relationship between time spent on a cotton plant
and number of flanges for Euschistus servus (Say) but
an inconsistent relationship for N. viridula. Quantifying
the relationship between food consumption and flange
counts would contribute to a basic understanding of var-
iation in pentatomid feeding strategies, and contribute
to pest management by improving measurements of
feeding and preference.
In this study, we sought to investigate the relationship
between the number of salivary flanges and food con-
sumption or ingestion for two important stink bug pests
in southeastern United States: N. viridula and E. servus
(McPherson and McPherson 2000).Asproxiesoffood
consumption rate, we measured relative growth rate
and quantity of excreta from stink bug’s feeding on cot-
ton and tested for relationships between these variables
and the number of flanges. Relative growth rate should
be positively related to food consumption rate regard-
less of the quality of that food item. Growth rates are
also relevant to population dynamics of herbivores
(O’Connor et al. 2011). Following Powles et al. (1972),
we also assume that the relationship between excreta
quantity—specifically the mass of opaque solutes in ex-
creta—and food consumption rate is positive in stink
bugs. While the relationship between excreta quantity
and food quality may not always be positive (Stadler
et al. 1998), a positive relationship between excreta
quantity and food consumption rate should be indepen-
dent of food quality. What’s more, cotton is a sufficient
food resource to provide consistently positive growth
rates and successful reproduction for N. viridula and
E. servus (Herbert and Toews 2011,2012;Zeilinger
et al. 2011). If a clear relationship exists between the
number of salivary flanges and food consumption rate,
then salivary sheaths and sheath flanges could be valu-
able indicators not only of stink bug crop damage but
also of stink bug food consumption and preference.
On the other hand, if no relationship exists, then
researchers should not use counts of flanges to make
inferences on stink bug consumption or preference.
Materials and Methods
Cotton Plants and Insect Rearing. We investi-
gated the relationship between the number of stink
bug salivary flanges and relative growth rates in a
field-cage experiment, and the relationship between
salivary sheaths and excreta in a laboratory experiment.
All work was conducted at the Coastal Plain
Experiment Station, U.S. Department of Agriculture–-
Agricultural Research Service (USDA-ARS), Tifton,
GA during the summer of 2008.
In the field, we planted four blocks of transgenic
glyphosate-tolerant, non-Bt cotton (variety DP-494RR),
with each block planted two weeks apart; the first block
was planted on 29 April. Planting date was randomly
assigned to blocks. Two blocks were 30 rows by 20
row-meters and two blocks were 24 rows by 20 row-
meters. We used standard cotton production practices
as described in Zeilinger et al. (2011). Laboratory colo-
nies were started from insects collected on early season
hosts in spring of 2008 near the Tifton Experiment Sta-
tion; colonies were maintained as described in Zeilinger
et al. (2011).
Growth Rate Study. To measure stink bug relative
growth rate and the number of salivary flanges, we
caged a single stink bug on a single, first position,
undamaged, developing cotton boll that was 8 d old
since flowering in the field. We also noted that selected
plants showed no evidence of prior herbivory, indicat-
ing little or no induction of plant defensive compounds
that may have affected boll quality (Hagenbucher et al.
2013). Treatments consisted of one of two stink bug
species, N. viridula and E. servus, and three age–sex
combinations, fifth-instar nymph, adult female, and
adult male, in a two-way factorial design (six treatments
total). The fifth instars were 1–2d old since molting
from the fourth instar, and adults were 5–6 d since
molting. We used fifth-instar nymphs and adults
because they feed the most of all stages and cause the
greatest amounts of damage to cotton and other crops
(Tod d 198 9,Greene et al. 1999). Newly molted fifth-
instar nymphs and adults that are 6d past molting pro-
duce the greatest number of salivary flanges within
their respective stages (Simmons and Yeargan 1988).
Treatments were randomly assigned to cotton bolls.
The duration of the trials varied between 71 and 74h.
We calculated relative growth rate by measuring the
change in mass of each stink bug before and after the
trials and standardizing it by the initial mass (Farrar
et al. 1989). Stink bugs were starved for 12h prior to
being weighed. We measured boll diameter (at the wid-
est point), boll length (from base to tip), and counted
the number of salivary flanges produced by stink bug
feeding. We calculated boll surface area from the boll
diameter and length using the equation for the surface
area of a prolate spheroid. To aid in counting flanges,
we stained the green carpel walls of the bolls in an acid
fuschin solution according the methods described in Ni
and Quisenberry (1997) and Zeilinger et al. (2011).
Excreta Study. For the excreta study, treatments
consisted of the same stink bug species used in the
Growth Rate Study, but only 6-d-old adult females and
males were used. Stink bugs were starved 12h prior to
the start of the experiment. At the start, one stink bug
was placed in a 125-ml plastic cup with a single boll for
3 d. The stem of the boll protruded from the bottom of
the cup through a small hole and was inserted into a
piece of soaked florist foam to maintain moisture in the
2A
NNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
boll. Inside the cup, a circular piece of filter paper
(55 mm in diameter; Whatman Ltd., Maidstone, UK)
was placed under the boll to collect the excreta of the
stink bug. Prior to placing the boll in the cup, the bracts
were removed so that all the excreta would drop onto
the filter paper. Stink bugs often avoid excreting onto
food items (A.R.Z., personal observations), so nearly all
of the excreta was collected on the filter papers.
At the start of the laboratory trials, quarter-size bolls
were collected from plants in the field plots described
above. The bolls were randomly collected from the
plots, excluding the first 3m of edge. During the
experiment, the filter paper was changed daily. Excreta
appeared as stains on the filter paper (Fig. 1A). On the
3rd day, the boll was removed from the cup and the fil-
ter paper was replaced, with only the stink bug remain-
ing in the cup. After an additional 24h, the filter paper
was collected and the stink bug removed. For each
boll, we measured diameter, height, and number of
salivary flanges, as described in the Growth Rate Study.
To measure the amount of excreta produced
by each stink bug, we digitally scanned all filter
papers using an HP Scanjet 4890 (Hewlett-Packard,
Palo Alto, CA) with the following specifications: 24-bit
color, 300 dpi, with an “unsharp” filter, brightness ¼
31, contrast ¼10, and color saturation ¼16. For each
scanned filter paper, we converted the RGB (Red,
Green, Blue) image to an HSI (Hue, Saturation, Inten-
sity) image. The saturation component of an HSI image
provides information, in the range of [0, 1], on the
intensity of color, or chromaticity. Saturation values
closer to unity represent a higher concentration of
dried excreta (more intense color), whereas lower satu-
ration values represent pixels with more white or lack
of color (Gonza´lez et al. 2004). Saturation is linearly
related to chromaticity as represented in an RGB color
space (Smith 1978), but unlike the RGB color space,
saturation in the HSI color space represents the chro-
maticity of all color present (Gonza´ lez et al. 2004).
We segmented the saturation component image
based on pixel value; values inclusive of 0.03 and 1 rep-
resented excreta; values less than 0.03 were filter paper
without excreta, the background of the scanner, and
other noise. Preliminary studies showed that the lower
threshold of 0.03 provided the optimal threshold for
selecting signal (excreta) from noise (background;
Fig. 1B and C). We calculated mean saturation value
by dividing the sum of the saturation of all pixels within
the range [0.03, 1] by the number of pixels within that
range. Our mean saturation value measurements were
similar to the “modified saturation value” measure-
ments of Smith et al. (1995) for secondary metabolite
concentrations in plant root hairs. All digital image
processing was conducted in MATLAB R2010a (Math-
Works, Natick, MA); conversion of RGB to HSI image
space was done using the rgb2hsi function (Gonza´lez
et al. 2004).
Calibration of Saturation Value. To understand
the relationship between excreta quantity and mean
saturation value, we calibrated saturation values using
known concentrations of stink bug excreta. We col-
lected excreta from E. servus and N. viridula adults
feeding on cotton bolls for three days, air dried the
samples, weighed the resulting excreta solutes, and
then performed a serial dilution of the solutes in DI
(de-ionized) water. We dropped 30 ll of each dilution
in a single drop onto clean filter papers, then scanned
and analyzed the digital image as described above. For
each drop, we calculated mean saturation value and
estimated the fit of linear calibration curves between
mean saturation value and excreta solute mass. Mean
saturation was arcsine-square-root transformed for stat-
istical analysis because it was in the range [0, 1] (Sokal
and Rohlf 1981). Using the coefficients from these cali-
bration curves, we estimated total excreta mass pro-
duced by each stink bug.
Statistical Analysis. For the growth rate study, we
tested the differences in relative growth rate between
species and stage-sex combinations using ANOVA and
pair-wise comparisons using Tukey’s HSD. We also fit
an ANCOVA model with relative growth rate as
response variable, the cube root of boll surface area as
acovariate, and all interactions between stink bug spe-
cies, stage-sex combination, and number of salivary
Fig. 1. Example of (A) RGB digital image, (B) binary image of color segmentation of saturation layer of an HSI image in
the range of [0.03, 1], and (C) saturation layer after color segmentation. The label of the filter paper (in panel A) was digitally
removed prior to color segmentation.
2015 ZEILINGER ET AL.: SALIVARY FLANGES OF HERBIVOROUS PENTATOMIDAE 3
flanges. Finally, for each species and stage-sex combi-
nation, we investigated relationships between relative
growth rate (as response variable) and boll surface area
and number of flanges (as explanatory variables) with
ordinary least-squares (OLS) multiple linear
regressions. Results from robust regression analyses
(not shown) supported the OLS regression results. To
meet the assumptions of normality and constant error
variance, we square-root transformed growth rates and
number of flanges. Salivary flange data were scaled to
number of flanges d
1
to match the temporal scale of
relative growth rate data and to standardize the units of
the parameters.
For the excreta study, excreta mass was estimated
from calibration curves based on the mean saturation
values in stink bug trials. We tested for differences in
excreta produced between stink bug species and sexes
with Type III ANOVA due an imbalanced design. We
fit an ANCOVA model with excreta mass as response
variable, the cube root of boll surface area as a covari-
ate, and all interactions between stink bug species, sex,
and the number of salivary sheaths. Flange counts and
excreta mass were square-root transformed. A general-
ized linear model with a quasi-Poisson error distribu-
tion corroborated the results of the ANCOVA (results
not shown). All analyses were conducted in R 3.1.0 (R
Core Team 2014).
Results
Growth Rate Study. Stink bug relative growth
rates did not differ between species (Tabl e 1;
E. servus ¼0.34 60.042 d
1
[mean 6SE], N¼69;
N. viridula ¼0.23 60.032 d
1
,N¼68). Relative
growth rates, however, differed among stage-sex
combinations with the highest growth rates found in
fifth-instar nymphs (Table 1;females¼0.12 60.03,
N¼47; males ¼0.08 60.02, N¼43; fifth-instar
nymphs ¼0.38 60.03, N¼47). The number of salivary
flanges differed between species, with a greater
number of sheaths produced by N. viridula
(F
1,123
¼11.45, P¼0.001; E. servus ¼33.54 66.00;
N. viridula ¼49.04 65.58). Fifth-instar nymphs of
both species produced significantly more flanges than
adult females and males (F
2, 123
¼13.48, P<0.001;
females ¼17.77 62.81; males ¼24.29 63.61; fifth-
instar nymphs¼40.52 65.16); no difference was found
between males and females (Tukey’s HSD, P¼0.101).
Over all species and stage-sex combinations, there
was a significant positive relationship between the
number of flanges and relative growth rate (Table 1).
This relationship did not differ significantly by species
or stage-sex, although the interaction of species by
stage-sex was nearly significant. However, least-squares
regression indicated that variation in the number of
flanges explained a significant portion of variation in
relative growth rates only for E. servus nymphs (Fig.
2). In this case, relative growth rate was positively
related to the number of flanges.
Excreta Study. Calibration curves relating excreta
mass to mean saturation value indicated a linear rela-
tionship that differed between the two stink bug spe-
cies (Fig. 3). Using the linear model coefficients from
these calibration curves, we estimated excreta mass
produced in each stink bug trial. Excreta produced by
E. servus was significantly greater than that produced
by N. viridula (Tab le 2;E. servus:0.6860.04 mg,
N¼37; N. viridula: 0.53 60.11 mg, N¼27), whereas
there was no difference in excreta produced between
sexes (Table 2 ;females¼0.65 60.08 mg, N¼35;
males ¼0.58 60.06, N¼29). The number of flanges
did not differ significantly between species (F
1,
59
¼0.25, P¼0.618; E. servus:65.89267.873;
N. viridula:64.1165.84) or sexes (F
1, 59
¼2.00;
P¼0.162; females: 72.83 67.16; males: 55.8667.11).
In addition, excreta mass did not explain a significant
portion of the variation in the number of flanges for
any species–sex combination (Tab le 2;Fig. 4).
Discussion
Salivary flange data have been used often to make
inferences on the feeding behavior of herbivorous
pentatomids. However, the relationship between the
number of salivary flanges and food consumption or
ingestion has remained unresolved. To contribute to an
understanding of what can be inferred from stink bug
salivary flange data, we sought to relate variation in the
number of flanges to variation in growth rates and
quantity of excreta, both used as indicators of food con-
sumption. Relative growth rates should be partly influ-
enced by food consumption rates and are relevant to
population dynamics of herbivores (O’Connor et al.
2011). Likewise, rates of defecation should be partly
influenced by food consumption (Powles et al. 1972).
While excreta quantity in herbivorous hemipterans can
be dependent on developmental stage and food quality
(Stadler et al. 1998), we controlled for both variables in
our study.
In the first study, E. servus had greater relative
growth rates whereas N. viridula produced more sali-
vary flanges, which corroborates our earlier results
(Zeilingeretal.2011). Fifth-instar nymphs of both
Table 1. Results of ANCOVA relating the number of salivary
sheaths to relative growth rate
Term df SS MS Fstatistic P-value
Boll surface area 1 0.19 0.19 15.41 <0.001***
Species 1 0.01 0.01 0.81 0.371
Stage-sex 2 1.74 0.87 69.97 <0.001***
Sheaths 1 0.07 0.07 5.99 0.016 *
Species Stage-sex 2 0.07 0.03 2.78 0.066
Species Sheaths 1 0.01 0.01 0.57 0.450
Sex Sheaths 2 0.01 0.01 0.43 0.650
Species Stage-sex Sheaths 2 0.03 0.02 1.25 0.291
Error 116 1.45 0.01
df, degrees of freedom; SS, sum of squares; MS, mean square.
Sheaths and relative growth rate were square-root transformed;
boll surface area was cube-root transformed, according to Box-Cox
transformation.
The Stage-sex effect included three levels: adult males, adult
females, and fifth-instar nymphs.
*P<0.05; ***P<0.001.
4ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
species had higher growth rates and produced more
flanges than adults. Greene et al. (1999) found that
fifth-instar N. viridula caused more damage to cotton
bolls than adults. We found a significant, positive rela-
tionship between growth rate and the number of
flanges only for E. servus fifth-instar nymphs. The lack
of a relationship for adults may have been due to their
very low growth rates. At the same time, we found
no relationship between relative growth rates and the
number of flanges for N. viridula nymphs even though
they had greater growth rates than conspecific adults.
We also investigated the relationship between salivary
flanges and food consumption using excreta, in part,
because variation in relative growth rates of adult stink
bugs was small. Specifically, we assumed that the quan-
tity of solutes in stink bug excreta should be positively
Fig. 3. Calibration curves of stink bug excreta solute
mass with mean saturation value. The relationship between
excreta mass and mean saturation was approximately linear
for E. servus (triangles, dashed line) and N. viridula (circles,
solid line). *P<0.05, ***P<0.001.
Table 2. Results of ANCOVA relating the number of salivary
sheaths to excreta mass
Term df SS MS Fstatistic P-value
Boll surface area 1 0.05 0.05 0.68 0.414
Species 1 0.36 0.36 5.11 0.028*
Sex 1 0.03 0.03 0.38 0.540
Sheaths 1 0.02 0.02 0.22 0.643
Species Sex 1 0.00 0.00 0.03 0.867
Species Sheaths 1 0.01 0.01 0.08 0.783
Sex Sheaths 1 0.00 0.00 0.06 0.807
Species Sex Sheaths 1 0.05 0.05 0.68 0.414
Error 54 3.81 0.07
df, degrees of freedom; SS, sum of squares; MS, mean square.
Sheaths and excreta mass were square-root transformed; boll sur-
face area was cube-root transformed, according to Box-Cox
transformation.
*P<0.05.
Fig. 2. Variation in relative growth rates and number of salivary sheaths (d
1
) for both E. servus (A–C) and N. viridula
(D–F) and for adult females (A, D), fifth-instar nymphs (B, E), and adult males (C, F). Relative growth rate explained a
significant amount of variation in salivary sheaths only for E. servus nymphs. R
2
values result from multiple linear regression
models with number of salivary sheaths and boll surface area as explanatory variables.
2015 ZEILINGER ET AL.: SALIVARY FLANGES OF HERBIVOROUS PENTATOMIDAE 5
related to food consumption. We estimated the quantity
of excreted solutes using mean saturation value from
digital image analysis. Saturation values measure the
intensity of color—or chromaticity—within a digital
image (Smith 1978). Our calibration curves showed that
this was a reliable and linear measure of excreted solute
quantity. Importantly, this measure ignores the volume
of water in excreta. Because N. viridula and E. servus
are cell-rupture feeders, as are most herbivorous penta-
tomids, their excreta should be more color saturated
than phloem or xylem feeders. Excreta mass was greater
for E. servus than for N. viridula, whereas there were
no differences in the number of salivary flanges
between species. That E. servus had greater growth
rates and excreta mass than N. viridula suggests that
they may have greater food consumption rates. None-
theless, we found no relationship between excreta mass
and the number of flanges for either species or sex.
Our two experiments differed markedly in the pat-
terns of feeding behavior described. Relative growth
rates differed between stage-sex combinations whereas
excreta mass did not, although this is likely due to the
inclusion of fifth-instar nymphs in the growth rate
experiment. At the same time, excreta mass differed
between species whereas relative growth rate did not.
Excreta may be an effective measure of consumption
rate where growth rates are low (i.e., adult insects) and
directly measuring food consumption is difficult (i.e.,
piercing-sucking insects). Both species produced more
flanges in the excreta study than in the growth rate
study. Because stink bugs in the excreta study fed on
bolls removed from plants, it is possible that these bolls
were of lower quality than bolls remaining on the plant,
thereby resulting in increased probing behavior of both
species.
Previous authors have used the number of salivary
flanges to infer food consumption and feeding prefer-
ence (Bowling 1980,Kester et al. 1984,Lye and Story
1988,Simmons and Yeargan 1988,Panizzi et al. 1995).
In the present study, we were unable to find a relation-
ship between the number of flanges and food
consumption for three stages of N. viridula and for
adult E. servus, using two different measures of
consumption. In an earlier feeding preference study,
we also found that the relationship between N. viridula
tenure time on cotton boll choices and the number of
flanges was inconsistent; in some cases the relationship
was positive whereas in other cases no relationship
Fig. 4. Variation in excreta mass, calculated from calibration curves, and number of salivary sheaths for female (A, B) and
male (C, D) adults of each species, E. servus (A, C) and N. viridula (B, D). Number of salivary sheaths did not explain a
significant portion of the variation in excreta mass for any species–sex combination.
6ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
existed (Zeilinger 2011). Based on these results, the
number of salivary flanges should not be used to
infer food consumption or feeding preference for
N. viridula.
While we found no relationship between food
consumption rate and salivary flanges for N. viridula
and E. servus adults, we found a positive relationship
for E. servus nymphs. Previous work has highlighted
differences in the feeding behavior of these two
species and other related species. Zeilinger et al. (2011)
found that E. servus growth rates were reduced by
caterpillar damage on cotton plants whereas N. viridula
growth rates were mostly unaffected. In contrast, cater-
pillar damage reduced the number of flanges from
N. viridula while it had no effect on flange counts from
E. servus.Thus,E. servus may have responded to cater-
pillar damage by consuming less, whereas consumption
by N. viridula may not have changed, as they may have
consumed the same amount of plant fluids from fewer
feeding holes. Depieri and Panizzi (2011) studied the
feeding activity of N. viridula and Euschistus heros (F.)
on soybeans in Brazil. They found that N. viridula
spent more time feeding and penetrated deeper into
seed tissue than E. heros. Moreover, the relationship
between feeding duration and soybean damage differed
between the two species. Importantly, Depieri and
Panizzi (2011) showed diverse feeding behaviors among
four different herbivorous stink bug species. Based on
the available evidence, it appears that N. viridula tends
to feed for longer periods of time within each feeding
puncture relative to E. servus. Potentially contradictory
evidence, on the other hand, suggests that N. viridula
also tends to show greater movement among cotton
bolls compared with E. servus (Zeilinger 2011,Huang
and Toews 2012). Regardless, differences in movement
and feeding behavior indicate that the per capita
damage to crop plants may also differ between these
species and among herbivorous stink bug species more
generally (Depieri and Panizzi 2011).
Further work will be needed to understand whether
our results for E. servus nymphs—with a positive rela-
tionship between growth rates and salivary flanges—or
our results for N. viridula—with no relationship—is
the more general case for pentatomids and other
sheath-feeding Hemiptera. Further work will also be
needed to investigate potential effects of different host
plants and food quality on the consumption–sheath
relationship. In the meantime, salivary flange data col-
lected from E. servus nymphs can be used to predict
food consumption but should not be used to make
inferences on N. viridula feeding; data from other pen-
tatomids should be interpreted with care.
Acknowledgments
This study was partially supported by an National Research
Initiative (NRI) grant 2008-02409 from the U.S. Department
of Agriculture to D.A.A., D.M.O., and John Ruberson, an
Integrative Graduate Education and Research Traineeship
(IGERT) grant 0653827 from U.S. National Science Founda-
tion to the University of Minnesota, a Thesis Research Grant
and a Doctoral Dissertation Fellowship from the Graduate
School, University of Minnesota, to A.R.Z., and grants from
the Dayton-Wilkie Fund of the Bell Museum of Natural
History, University of Minnesota, to A.R.Z. We thank
A. Hornbuckle and M. Smith for assisting the experiments;
T. Brown, E. Rosengren, J. Skelton, and N. Lynch for assist-
ing in digital image analysis; S. Acton for advising on the
digital image analysis method; M. Daugherty for providing
access to MATLAB and advice on statistical analyses; and
P. Oikawa for advising on statistical analyses.
References Cited
Backus,E.A.2000.Our own jabberwocky: clarifying the termi-
nology of certain piercing-sucking behaviors of homopterans,
pp. 1–13. In G. P. Walker and E. A. Backus (eds.), Principles
and applications of electronic monitoring and other tech-
niques in the study of homopteran feeding behavior, Thomas
Say Publications in Entomology. Entomological Society of
America, Lanham, MD.
Barbour, K. S., and J. S. Bacheler. 1990. Reduction in yield
and quality of cotton damaged by green stink bug (Hemi-
ptera: Pentatomidae). J. Econ. Entomol. 83: 842–845.
Bowling, C. C. 1979. The stylet sheath as an indicator of feeding
activity of the rice stink bug. J. Econ. Entomol. 72: 259–260.
Bowling, C. C. 1980. The stylet sheath as an indicator of feeding
activity by the southern green stink bug on soybeans. J. Econ.
Entomol. 73: 1–3.
Bundy,C.S.,R.M.McPherson,andG.A.Herzog.2000.An
examination of the external and internal signs of cotton boll
damage by stink bugs (Heteroptera: Pentatomidae).
J. Entomol. Sci. 35: 402–410.
Depieri, R., and A. Panizzi. 2011. Duration of feeding and su-
perficial and in-depth damage to soybean seed by selected
species of stink bugs (Heteroptera: Pentatomidae). Neotrop.
Entomol. 40: 197–203.
Farrar,R.,J.Barbour,andG.G.Kennedy.1989.Quantify-
ing food consumption in insects. Ann Entomol Soc Am. 82:
593–598.
Gonza´lez,R.C.,R.E.Woods,andS.L.Eddins.2004.
Digital Image Processing using MATLAB. Pearson Prentice
Hall, Upper Saddle River, NJ.
Greene, J. K., S. G. Turnipseed, M. J. Sullivan, and G. A.
Herzog. 1999. Boll damage by southern green stink bug
(Hemiptera: Pentatomidae) and tarnished plant bug (Hemi-
ptera:Miridae) caged on transgenic Bacillus thuringiensis cot-
ton. J. Econ. Entomol. 92: 941–944.
Hagenbucher,S.,D.M.Olson,J.R.Ruberson,F.L.Wa¨ckers,
and J. Romeis. 2013. Resistance mechanisms against arthro-
podherbivoresincottonandtheirinteractionswithnatural
enemies. Critical Rev. Plant Sci. 32: 458–482.
Herbert, J. J., and M. D. Toews. 2011. Seasonal abundance
and population structure of brown stink bug (Hemiptera:
Pentatomidae) in farmscapes containing corn, cotton, peanut,
and soybean. Ann. Entomol. Soc. Am. 104: 909–918.
Herbert, J. J., and M. D. Toews. 2012. Seasonal abundance
and population structure of Chinavia hilaris and Nezara
viridula (Hemiptera: Pentatomidae) in Georgia farmscapes
containing corn, cotton, peanut, and soybean. Ann. Entomol.
Soc. Am. 105: 582–591.
Hollay, M. E., C. M. Smith, and J. F. Robinson. 1987. Struc-
ture and formation of feeding sheaths of rice stink bug (Het-
eroptera: Pentatomidae) on rice grains and their association
with fungi. Ann. Entomol. Soc. Am. 80: 212–216.
Huang, T.-I., and M. D. Toews. 2012. Feeding preference
and movement of Nezara viridula and Euschistus servus
(Hemiptera: Pentatomidae) on individual cotton plants.
J. Econ. Entomol. 105: 847–853.
2015 ZEILINGER ET AL.: SALIVARY FLANGES OF HERBIVOROUS PENTATOMIDAE 7
Kawamoto, H., N. Ohkubo, and K. Kiritani. 1987. Modeling
of soybean pod feeding behavior of stink bugs. Appl. Ento-
mol. Zool. 22: 482–492.
Kester,K.M.,C.M.Smith,andD.F.Gilman.1984.Mecha-
nisms of resistance in soybean (Glycine max [L] Merrill)
genotype Pi171444 to the southern green stink bug, Nezara
viridula (L) (Hemiptera, Pentatomidae). Environ. Entomol.
13: 1208–1215.
Lye, B. H., and R. N. Story. 1988. Feeding preference of the
southern green stink bug (Hemiptera: Pentatomidae) on to-
mato fruit. J. Econ. Entomol. 81: 522–526.
McPherson, J. E., and R. M. McPherson. 2000. Stink Bugs of
Economic Importance in America North of Mexico. CRC
Press, Boca Raton, FL.
Medrano,E.G.,J.F.Esquivel,andA.A.Bell.2007.Tr ans -
mission of cotton seed and boll rotting bacteria by the south-
ern green stink bug (Nezara viridula L.). J. Appl. Microbiol.
103: 436–444.
Medrano,E.G.,J.F.Esquivel,R.L.Nichols,andA.A.Bell.
2009. Temporal analysis of cotton boll symptoms resulting
from southern green stink bug feeding and transmission of a
bacterial pathogen. J. Econ. Entomol. 102: 36–42.
Miles, P. W. 1972. The saliva of Hemiptera. Adv. Insect Physiol.
9: 183–255.
Mitchell,P.L.,R.Gupta,A.K.Singh,andP.Kumar.2004.
Behavioral and developmental effects of neem extracts on
Clavigralla scutellaris (Hemiptera: Heteroptera: Coreidae)
and its egg parasitoid, Gryon fulviventre (Hymenoptera: Sce-
lionidae). J. Econ. Entomol. 97: 916–923.
Morgan,J.K.,G.A.Luzio,E.-D.Ammar,W.B.Hunter,
D.G.Hall,andR.G.Shatters.2013.Formation of stylet
sheaths in aere (in air) from eight species of phytophagous
Hemipterans from six families (Suborders: Auchenorrhyncha
and Sternorrhyncha). PLoS ONE 8: e62444.
Ni,X.,andS.S.Quisenberry.1997.Distribution of Russian
wheat aphid (Homoptera: Aphididae) salivary sheaths in re-
sistant and susceptible wheat leaves. J. Econ. Entomol. 90:
848–853.
O’Connor, M. I., G. Benjamin, and C. J. Brown. 2011. Theo-
retical predictions for how temperature affects the dynamics
of interacting herbivores and plants. Am. Nat. 178: 626–638.
Panizzi,A.R.,C.C.Niva,andE.Hirose.1995.Feeding pref-
erence by stink bugs (Heteroptera: Pentatomidae) for seeds
within soybean pods. J. Entomol. Sci. 30: 333–341.
Peiffer, M., and G. W. Felton. 2014. Insights into the saliva of
the brown marmorated stink bug Halyomorpha halys (Hemi-
ptera: Pentatomidae). PLoS ONE 9: e88483.
Powles, M. A., P. A. Janssens, and D. Gilmour. 1972. Urea
formation in the green vegetable bug Nezara viridula.
J. Insect Physiol. 18: 2343–2358.
R Core Team. 2014. R: A language and environment for
statistical computing. R Foundation for Statistical Comput-
ing, Vienna, Austria.
Simmons,A.M.,andK.V.Yeargan.1988.Feeding fre-
quency and feeding duration of the green stink bug (Hemi-
ptera: Pentatomidae) on soybean. J. Econ. Entomol. 81:
812–815.
Smith, A. R. 1978. Color gamut transform pairs, pp. 12–19. In
ACM Siggraph Computer Graphics. ACM.
Smith, M.A.L., J. F. Reid, A. C. Hansen, Z. Li, and D. L.
Madhavi. 1995. Non-destructive machine vision analysis of
pigment-producing cell cultures. J. Biotech. 40: 1–11.
Sokal, R. R., and F. J. Rohlf. 1981. Biometry.W.H.Freeman
and Company, New York, NY.
Stadler, B., B. Michalzik, and T. Mu¨ lle r. 1998. Linking aphid
ecology with nutrient fluxes in a coniferous forest. Ecology
79: 1514–1525.
Todd,J.W.1989.Ecology and behavior of Nezara viridula.
Ann. Rev. Entomol. 34: 273–292.
Viator, H. P., A. Pantoja, and C. M. Smith. 1983. Damage to
wheat seed quality and yield by the rice stink bug and south-
ern green stink bug (Hemiptera: Pentatomidae). J. Econ.
Entomol. 76: 1410–1413.
Zeilinger, A. R. 2011. The role of competitive release in stink
bug outbreaks associated with transgenic Bt cotton (Ph.D dis-
sertation), University of Minnesota.
Zeilinger,A.R.,D.M.Olson,andD.A.Andow.2011.Com-
petition between stink bug and heliothine caterpillar pests on
cotton at within-plant spatial scales. Entomol. Exp. Appl. 141:
59–70.
Received 26 May 2014; accepted 7 November 2014.
8ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA
