Content uploaded by Dawn Olson
Author content
All content in this area was uploaded by Dawn Olson on Jul 10, 2018
Content may be subject to copyright.
1
Published by Oxford University Press on behalf of Entomological Society of America 2018.
This work is written by (a) US Government employee(s) and is in the public domain in the US.
Community and Ecosystem Ecology
Landscape Effects on Solenopsis invicta (Hymenoptera:
Formicidae) and Geocoris spp. (Hemiptera: Geocoridae),
Two Important Omnivorous Arthropod Taxa in Field Crops
Dawn M.Olson,1,7 Adam R.Zeilinger,2 Kristina K.Prescott,3 Alisa W.Coffin,4
John R.Ruberson,5 and David A.Andow6
1Crop Protection, and Management Research Unit, USDA-ARS, Tifton, GA 31794, 2Department of Environmental Science, Policy,
and Management, University of California Berkeley, Berkeley, CA 94720, 3Department of Ecology, Evolution and Behavior, University
of Minnesota, St. Paul, MN 55108, 4Southeast Watershed Research Laboratory, USDA-ARS, Tifton, GA 31794, 5Department of
Entomology, Kansas State University, Manhattan, Kansas 66506, 6Department of Entomology and Center for Community Genetics,
University of Minnesota, St. Paul, MN 55108, and 7Corresponding author, e-mail: Dawn.olson@ars.usda.gov
Subject Editor: Shannon Murphy
Received 19 March 2018; Editorial decision 19 June 2018
Abstract
The economically important brown stink bug, Euschistus servus (Say)(Hemiptera:Pentatomidae), is a native pest
of many crops in southeastern United States and insecticide applications are the prevailing method of population
suppression. To elucidate biological control of E. servus populations, we investigated two egg predators’ (red
imported fire ants, Solenopsis invicta Buren(Hymenoptera:Formicidae), and Geocoris spp.(Hemiptera: Geocoridae))
responses to both local and landscape factors that may have influenced their combined ability to cause mortality
in immature E.servus. We estimated the density of fire ants and Geocoris spp. on four major crop hosts—maize,
peanut, cotton, and soybean—in 16 landscapes over 3 yr in the coastal plain of Georgia, USA. Both Geocoris spp.
and fire ant populations were concentrated on specific crops in this study, maize and soybean for Geocoris spp.
and peanut and cotton for fire ants, but the percentage area of specific crops and woodland and pasture in the
landscape and year also influenced their density in focal fields. The crop specific density of both taxa, the influence
of the percentage area of specific crops and woodland in the landscape, and the variability in density over years
may have been related to variable alternative resources for these omnivores in the habitats. Despite the variability
over years, differential habitat use of fire ants and Georcoris spp. may have contributed to their combined ability to
cause E.servus immature mortality.
Key words: Lasso elastic net analysis, big-eyed bug, brown stink bug
Landscape complexity, typically measured as the amount of natural
or seminatural habitat, often has positive effects on arthropod nat-
ural enemies. Indeed, four recent meta-analyses (Bianchi etal. 2006,
Chaplin-Kramer etal. 2011, Shackelford et al. 2013) show that a
higher proportion of natural or seminatural habitat in a landscape is
associated with a higher abundance and diversity of natural enemies.
The meta-analysis of Shackelford etal. (2013) analyzed the effects of
both local complexity and landscape complexity on various arthro-
pod species; they dened local complexity as measurement of the
distance from a sample of arthropods (e.g., a pan or pitfall trap,
or a transect walk) to eld margins. Further, Houston etal. (2017)
found that landscape diversity and plant vigor had a higher inuence
on biological control than did local diversity. Therefore, analyses of
complexity at both local and landscape spatial scales may be impor-
tant in predicting species response to noncrop habitat.
In a previous study (Olson etal. 2018), we investigated the nite
rate of population increase of the economically important brown
stink bug, Euschistus servus Say (Hemiptera: Pentatomidae) in the
major row crops—peanut, eld and sweet corn, soybean, and cot-
ton—in landscapes of southern Georgia. We found that E. servus
reproduction was partly inuenced by local-scale factors such as soy-
bean and the combined density of all stages and species of Geocoris
spp. Fallén (Hemiptera: Geocoridae) and re ants, Solenopsis invicta
Buren (Hymenoptera: Formicidae), suggesting that these species may
cause signicant mortality to E.servus immature stages. Because of
their presumed importance in reducing E.servus reproduction, here
we investigated the inuence of landscape and local scale-factors
on the density of these predators in the same landscapes previously
studied (Olson et al. 2018) to identify factors that might be mal-
leable to further enhance E.servus predation in the area. While a
Environmental Entomology, XX(X), 2018, 1–7
doi: 10.1093/ee/nvy104
Research
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
complex of generalist predators, which include Geocoris spp. and
S.invicta, exert mortality as a group on stink bugs (Yeargan 1979,
Ragsdale etal. 1981, Stam et al. 1987, Van Den Berg etal. 1995,
Ehler 2002, Tillman 2011), sampling of the entire complex (an add-
itional seven predators) was out of the scope of thisstudy.
S.invicta is an exotic species in North America and is established
in the southern United States and California (Lofgren etal. 1975,
Eubanks 2001). The ultimate distribution in North America may
be limited, as S.invicta lacks the ability to hibernate (Lofgren etal.
1975), and arid environments slows their expansion (Porter and
Tschinkel 1987, LeBrun etal. 2012). Newly mated S.invicta queens
select and thrive in open or heavily disturbed, often human-altered
ecosystems for founding new colonies (LeBrun etal. 2012, Tschinkel
and King 2013, King and Tschinkel 2016, Tschinkel and King 2017).
In addition, when S.invicta penetrates undisturbed areas, their dens-
ity is higher only when moisture is present (LeBrun et al. 2012).
Ants in general occupy many trophic positions within ecosystems
feeding on a variety of items, including plant tissue, nectar, seeds and
both vertebrates and invertebrates (Hölldobler and Wilson 1990,
Allen etal. 2004, Tschinkel 2006). Presence of S.invicta has been
found to increase predation of both pest and benecial arthropods
by 20–30% in elds of cotton, Gossypium hirsutum L, and soy-
bean, Glycine max (L) Merr (Eubanks 2001, Eubanks et al. 2002,
Diaz etal. 2004), and they can have positive, neutral and negative
effects on a variety of arthropod species at the soil surface of cot-
ton agroecosystems (Wickings and Ruberson 2011). Olson and
Ruberson (2012) determined that the dominant egg predator on the
stink bug Nezara viridula L.(Hemiptera: Pentatomidae) differed by
crop: S.invicta was the dominant egg predator in cotton and peanut,
whereas long-horned grasshoppers (Tettigoniidae) were dominant
egg predators in soybean.
Two species of Geocoris, G.punctipes (Say), and G.uliginosus
(Say), are common and widely distributed in eastern North America.
Of these two species, G.punctipes has been more thoroughly studied
than G.uliginosus. This discrepancy may in part reect the tendency
of G. punctipes to climb on plants, whereas G. uliginosus spends
most of its time on or near the ground (Readio and Sweet 1982, but
see Torres and Ruberson 2006b). Geocoris punctipes and G.uligino-
sus pass the winter as adults in reproductive diapause in temperate
regions (Ruberson etal. 2001). Both nymphs and adults of G.punc-
tipes are abundant in several prominent cropping systems includ-
ing cotton and soybean (Naranjo and Stimac 1985, Pfannenstiel
and Yeargan 1998). In addition, Geocoris spp. are prone to remain
in habitats because of their ability to derive nutrients from plants
(Eubanks and Denno 1999). Because of the crop-specic mortality
of S.invicta on stink bug eggs, and the tendency of Geocoris spp.
to remain in habitats, we tested the hypothesis that crop and with-
in-crop eld variables explain more of the variation in re ant and
Geocoris spp. numbers and distribution than do landscape variables.
Materials and Methods
SamplingPlan
We selected two areas, one located in the southwestern and the
other located in the east-central Coastal Plain of Georgia with a
total of 16 landscapes, and sampled E.servus host crops in those
areas (intensively produced maize, peanut, cotton and soybean)
over 3 yr as discussed and illustrated in Olson etal. (2018). Briey,
each eld was sampled using two permanent parallel transects
(spaced 30.5 m apart) running perpendicular to the edge of the
eld. All sampled eld edges were adjacent to woodland. Atotal
of 40 sampling points in 2009 (2 transects × 20 sample points=40
sample points per eld) and 60 sampling points in 2010 and 2011
(2 transects × 15 sample points= 30 sample points per eld per
year=60 sampling points total over both years) were established
along each transect. Thus, there were a total of 100 sample points
per eld over the 3 yr. The rst sample was at 1 m into the crop
from the crop edge for all years and in 2009, the next 19 sam-
ples were spaced at 5 m intervals, whereas in 2010 and 2011,
samples 2 through 10 were spaced at 5 m intervals and the last 5
samples were spaced at 10 m intervals (= 101 m from crop edge
for all elds). Samples alternated weekly from crop rows 1–5 to
the left and the right sides of transects to reduce repetitive plant
and population disturbance. Sampling of maize was done using a
two-person, whole-plant visual count with the samplers on oppo-
site sides of the maize row for a total sampling distance of 1.5 m
(ca 8 plants/sample). Peanut was sampled using a Vortis suction
sampler (15cm diameter inlet, Burkard Manufacturing Company,
Ltd., Hertfordshire, UK) with 12×3-s suctions at 7–8cm from the
soil surface (2121cm2 sampled per sampling station). Cotton and
soybean were sampled using a 1.5-m drop cloth with 10 shakes
per sample. In all crops, about 1.5 linear row-meters of plants
were sampled at each point. Different sampling techniques were
utilized to maximize capture of insects on each of these structur-
ally different crops (Olson and Ruberson, personal observations).
Each crop eld was sampled weekly beginning at the fruiting stage
when E. servus is most abundant and ending with full maturity
or harvest. Maize is an earlier planted crop than the other crops
and was sampled from early June to August, and the other crops
were sampled from mid-July to late September–October. Harvest
or full maturity dates varied within and among crops. There were
a total of 10,810 samples in maize, 17,906 in cotton, 18,050 in
peanut, and 10,930 in soybean. The number of nymphs and adults
of Geocoris spp. (almost entirely G.punctipes and G.uliginosus)
and S. invicta were counted and recorded. Atotal of 181 elds
were sampled.
Landscape Characteristics
For each landscape, we determined the percentage area of E.servus
host crops—maize, peanut, cotton and soybean—and the percent-
age area of seminatural and natural habitat—woodland, pasture
and noncrop stink bug hosts in woodland edges—or ‘green veining’
(GV). There were fourteen landscapes of 4.8×4.8 km (2,330 ha),
and two landscapes of 5.3×8.3 km (4,399 ha), with the latter land-
scape size needed to encompass all required crop species (Olson etal.
2018). We also determined the number of crop elds within 100, 500
and 1,000 m radii from the sampled elds as a measure of cropland
connectivity. These distances were chosen because the stink bug prey
species of the sampled predators have been observed dispersing up
to 1,000 m per day by ight in search of feeding or oviposition sites
(Kiritani etal. 1965).
Local Field Characteristics
Local scale variables included sampled crops, the distance from
the sampling point to the edge of the sampled eld and the perime-
ter-to-area ratio (PA) of sampled crop elds.
Statistical Analysis
First, we investigated which factors explained Geocoris spp. and
S. invicta density and distribution. We included three factors in
this analysis: 3 yr, 16 landscapes, and three groups of four crop
elds in each landscape. The design was a nested analysis of vari-
ance (ANOVA) (SAS Institute Inc. 1998) with years as whole plot
2 Environmental Entomology, 2018, Vol. XX, No. X
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
factors and whole plot error dened by landscapes within years
(error 1). The subplots were the crops (and crop interactions with
year), and groups were the unit of replication (error 2)after remov-
ing all landscape interactions with crops (and landscape by crop
interactions with year). The hierarchical structure of the ANOVA
accommodated the nonindependence of samples within eld. The
natural logarithms of S. invicta and Geocoris spp. counts + 0.50
were analyzed because the variances of the untransformed data
were heteroscedastic.
Second, we determined the set of landscape and local scale var-
iables that best explained variation in Geocoris spp. and S.invicta
density using the elastic net method for variable selection, in
which LASSO is a special case (Hastie et al. 2009). The elastic
net analysis is used to identify the best set of predictor variables,
is insensitive to the ordering of variable inclusion or exclusion, as
occurs in all stepwise methods, and is effective when predictors
are correlated.
For the elastic net analysis, we included the following predic-
tor variates at the landscape scale: the percentage area of GV in
the landscape, the percentage area of specic E.servus crop hosts
in the landscapes (maize, peanut, cotton, soybean), the percentage
area of total crops and the number of crop elds within 100, 500,
and 1,000 m from the sampled eld as a measure of cropland con-
nectivity. The number of crop elds at various distances from the
focal eld was not included in S.invicta analysis as this social spe-
cies has complex foraging behaviors and distances travelled that
are based on social form (mono- vs polygyene colonies), colony
size, territoriality, and quality of food (Tschinkel 2011). The pre-
dictor variates at the local scale included the distance from the
sampling point within the sampled eld to the eld edge, counts of
Geocoris spp. (differentiated by species but pooled for analyses),
the perimeter-to-area ratio of the focal crop elds, counts of re
ants within the focal elds, focal crop species, and year. Factors
with more than two levels were split into multiple binary dummy
variables. All continuous variates were transformed to standard-
ized normal variables.
We used K-fold cross-validation (Hooten and Hobbs 2015) to
choose the best values of the regularization parameters, γ2 and γ1,
based on minimum value of the mean-squared error (MSE) (note:
while other authors have represented the elastic net regulators as λ
and α, we adopted the notation of Hooten and Hobbs (2015) to dif-
ferentiate the elastic net regulators from our estimates of stink bug
reproduction, λ in our previous study). These analyses allowed us to
simplify the model to a LASSO model. We then used the truncated
Gaussian test of Tibshirani etal. (2017) to calculate P-values and
95% condence intervals for each predictor included in the LASSO-
selected bestmodel.
All geospatial manipulations and analyses were conducted
using ArcGIS for Desktop (Version 10.3, Advanced 2014). LASSO
model selection was conducted using the glmnet and glmnetUtils
packages in R 3.0 (Friedman etal. 2010, R Core Team 2016, Ooi
and Microsoft 2017). The truncated Gaussian test was performed
using the ‘selectiveInference’ R package (Tibshirani et al. 2017).
R code for running elastic net cross-validation and truncated
Gaussian tests can be found at https://github.com/arzeilinger/
stink_bug_reproduction_lasso.
Results
There was a signicant effect of year and of crop, and a signicant
interaction between year and crop on Geocoris spp. density (num-
bers per 1.5 linear row-meters of plants) (Table1). Geocoris spp.
density was lower in 2009 than in 2010 and 2011 in all crops and
highest in soybean in 2010 and in maize in 2011 (Fig.1). Crop had
a signicant effect on S.invicta density (Table1). Densities were sig-
nicantly higher in soybean (mean ± SE=0.55±0.01, n=10,930),
than in maize (0.33 ± 0.01, n = 10,810), peanut (0.20 ± 0.004,
n=18,050), and cotton (0.13± 0.003, n=17,906). Densities were
also signicantly higher in maize than in cotton and peanut (Fig.1).
The ratio of nymphs to adults was higher in soybean (2.15) and
maize (2.02) and lower in cotton (0.74) and peanut (0.85).
Year and crop signicantly affected S.invicta density (numbers
per 1.5 linear row-meters of plants), and there was a signicant inter-
action between year and crop (Table2). In maize, S.invicta density
was higher in 2011 than in 2009 and 2010 (Fig.2). For cotton, their
density was higher in 2010 and 2011 than in 2009, and was higher
in cotton than in maize and soybean all 3 yr (Fig.2). Higher densities
of S. invicta were found in 2010 peanut followed by 2009 peanut
and then 2011 peanut. Crop also had a signicant effect on S.invicta
density (Table2). S.invicta density was signicantly higher in peanut
(5.02±0.19, n=18,050), than in cotton (3.42±0.06, n=17,906),
maize (0.45±0.05, n=10,810), and soybean (1.14±0.03, n=0.02,
n =10,930). S. invicta density in soybeans consistently approxi-
mated a mean of 1 ant per sample all 3yr.
For the elastic net analysis of Geocoris spp. densities (number
per 1.5 linear row-meters of plants), cross-validation indicated
that γ2=1 resulted in the smallest mean-squared error (MSE). This
reduces the elastic net analysis to a LASSO procedure and suggests
that multicollinearity among predictor variates was of relatively
minor importance. The best estimate of the regulator γ1 (i.e., that
produced the smallest MSE) was 0.0220. Geocoris spp. density
was negatively associated with the landscape factors: percentage of
GV, number of cotton elds at 100-m and 1,000-m radii, number
of maize elds at 1,000-m radius, and number of peanut elds at
500-m radius. In contrast, densities were positively associated with
the percentage area of cotton elds in the landscape and the number
of maize elds at 100-m radii (Table3). For within-eld variables,
Geocoris spp. counts were positively related to the distance to the
eld edge and negatively related to the perimeter-area ratio of the
focal elds. Geocoris spp. numbers were signicantly lower in 2009
and in cotton elds; and were signicantly greater in 2011 and in
soybean and maize sampled elds (Fig. 1). The percentage of all
crops in the landscape was also included in the best model, though
the magnitude was small and there was no signicant relationship
Table1. ANOVA on the effects of year, crop and their interactions on ln(Geocoris spp. density + 0.5)
Source DF Type III SS MS F-value P-value
Year 2 693.4384402 346.7192201 1,435.22 <0.001
Error 1 5 103.58349 42.773296
Crop 3 383.6853879 127.2284626 526.65 <0.001
Year × crop 6 405.9212387 67.6535398 280.05 <0.001
Error 2 62339 16,839.78059 0.24158
Environmental Entomology, 2018, Vol. XX, No. X 3
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
with Geocoris spp. counts. For the elastic net analysis of S.invicta
density (number per 1.5 linear row-meters of plants), cross-valida-
tion indicated that again γ2=1 resulted in the smallest MSE, and the
best estimate of the γ1 was 0.0282. For landscape variables, S.invicta
density were negatively associated with the percentage area of maize
and all crops combined in the landscape; at the same time, their
density was positively associated with the percentage area of GV
and the percentage area of soybean in the landscape (Table3). Maize
and soybean had strong negative associations, whereas peanut had
a strong positive association with S.invicta density (Table4). Year
2011 had a negative association with S.invicta density (Table4).
Discussion
Geocoris spp. and re ant densities were associated with both local
and landscape factors involving the identity and percentage area of
specic crops and the percentage area of woodland and pasture in
the landscape. This is consistent with the results of a previous study
where both local and landscape scale processes were important in
structuring populations of an omnivorous ground beetle species
(Lucas and Maisonhaute 2015). Two within-eld factors—dis-
tance to the edge of the eld and the perimeter to area ratio of focal
elds—were also related to Geocoris spp. but not re ant density.
Table2. ANOVA on the effects of year, crop, and their interactions on ln(Solenopsis invicta density + 0.5)
Source df Type III SS MS F-value P-value
Year 2 68.98047 34.49023 28.05 <0.001
Error 1 5 2.1562494 0.7599523
Crop 3 11,835.95391 3,945.31797 3,208.61 <0.001
Year × crop 6 565.77047 94.29508 76.69 <0.001
Error 2 62,339 76,615.37243 1.22960
Fig.1. Relationship between year and crop and Geocoris spp. density (number per 1.5 linear row-meters of plants). Means were separated using Tukey’s highest
significant difference at P<0.05 and means with different letters are significantly different.
Fig.2. Relationship between year and crop and Solenopsis invicta density (number per 1.5 linear row-meters of plants). Means were separated using Tukey’s
highest significant difference at P<0.05 and means with different letters are significantly different.
4 Environmental Entomology, 2018, Vol. XX, No. X
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
The number of specic crop elds at distances between 100 m and
1,000 m from the focal eld were also associated with Geocoris spp.
density, indicating that crop connectivity may also be important in
structuring populations of these species.
Geocoris spp. density differed over the 3 yr. Their densities were
highest in 2011 and lowest in 2009. Geocoris spp. density can be
highly variable over years (e.g., Davis 1981) and may be related to
interannual variation in winter temperature ranges that occur in
the study area and may inuence the overwintering dynamics of
Geocoris spp. (Ruberson etal. 2001). In another study of local and
landscape characteristics on ground beetles, Lucas and Maisonhaute
(2015) also found high variability among years and they suggested
that food availability for omnivorous species (prey, seeds, or other
food items) also varies greatly between years thereby accounting for
the yearly differences.
Several studies have identied a complex of predators, includ-
ing Geocoris spp. and S.invicta, of eggs and nymphs of stink bugs
through ELISA techniques, observations and sentinel eggs where
their combined predation of the egg and nymphal stages of stink
bugs can be high (Yeargan 1979, Ragsdale etal. 1981, Stam et al.
1987, Van Den Berg etal. 1995, Ehler 2002, Tillman 2011). Most of
these studies were conducted on N.viridula, but in general, a similar
complex of generalist predators were feeding on stink bug eggs and
nymphs in various crops with the percentage that each species con-
tributed to the complex being variable over crop, season, and year.
Our study also indicates that Geocoris spp. and S.invicta together
are important predators of E. servus, providing some support for
the efcay of a complex of predators in causing stink bug mortal-
ity. In addition, potential reductions of reproduction of E.servus by
Geocoris spp. due to their variability over years may be offset by the
presence of other predator species in the complex, but this would
need to be tested.
The positive relationship between Geocoris spp. density and soy-
bean and maize crop may be due to the prevalence of nonprey food
resources in these elds during the time of this study. Both soybean
and maize are fed upon by G.punctipes (Stoner 1970, Naranjo and
Stimac 1985), and many Geocoris spp. were found in the silks of
maize where there were high densities of pollen. Both maize and
soybean also had the highest ratios of nymphs to adults, suggesting
higher reproduction or higher survival of nymphs in these crops.
This was also found in a previous study (Pfannenstiel and Yeargan
1998). In addition, G.punctipes populations varied among closely
spaced crop species, being more abundant in soybean and tomato
than in maize and tobacco, suggesting crop-specic colonization
preferences for these species (Pfannenstiel and Yeargan 1998).
The focal and nearby cotton elds had negative associations
with Geocoris spp. density, in part because of the relatively high fre-
quency of insecticide applications in these elds (Olson etal. 2018).
However, the percentage area of cotton in the landscape was posi-
tively associated with their density. Cotton was the dominant crop
grown in the study area (Olson etal. 2018) and Geocoris spp. can
be very prevalent and effective predators of various pests of cotton
(Lingren etal. 1968, Ali and Watson 1982, Ruberson etal. 1994).
Table3. Results from LASSO method for variable selection relat-
ing ln(Geocoris spp. density + 0.5) to a set of landscape and in-field
factors
Effect Estimate (± 95% CI)aP-valuea
Soybean 0.16 (0.15, 0.17) <0.001
Year 2009 −0.12 (−0.13, −0.09) <0.001
Percentage of cotton 0.11 (0.09, 0.15) <0.001
Year 2011 0.07 (0.06, 0.08) <0.001
Maize 0.06 (0.05, 0.07) <0.001
GV −0.05 (−0.09, −0.04) <0.001
Number of cotton elds (100 m) −0.04 (−0.05, −0.03) <0.001
Cotton −0.04 (−0.05, −0.03) <0.001
Distance to the edge of the eld 0.04 (0.03, 0.04) <0.001
Number of maize elds (100 m) 0.03 (0.02, 0.04) <0.001
Number of cotton elds (1,000 m) −0.03 (−0.04, −0.01) <0.001
Number of maize elds (1,000 m) −0.02 (−0.03, −0.01) 0.012
Number of peanut elds (500 m) −0.02 (−0.03, −0.01) <0.001
PA −0.01 (−0.02, −0.00) 0.013
Percentage of all crops 0.0 (−0.06, 0.02) 0.605
Percentage of maize 0 ND
Percentage of peanut 0 ND
Percentage of soybean 0 ND
Number of peanut elds (100 m) 0 ND
Number of soybean elds (100 m) 0 ND
Number of all crop elds (100 m) 0 ND
Number of maize elds (500 m) 0 ND
Number of cotton elds (500 m) 0 ND
Number of soybean elds (500 m) 0 ND
Number all crop elds (500 m) 0 ND
Number of peanut elds (1,000 m) 0 ND
Number of soybean elds (1,000 m) 0 ND
Number of all crop elds (1,000 m) 0 ND
Peanut 0 ND
Year 2010 0 ND
Mean re ants 0 ND
Variables are ordered from largest to smallest absolute coefcient estimate.
GV=woodland, pasture and noncrop stink bug hosts in woodland edges;
ND=Not determined; PA=perimeter to area ratio of sampled elds.
aCoefcient estimates from LASSO analysis. 95% condence interval
(± 95% CI) and P-values were calculated from truncated Gaussian test only
for variables included in best model. Variables excluded from model have coef-
cient estimate of 0.
Table4. Results from LASSO method for variable selection relat-
ing ln(Solenopsis invicta density + 0.5) to a set of landscape and
in-field factors
Effect Estimate (± 95% CI)aP-valuea
Maize −0.29 (−0.29, −0.28) <0.001
Soybean −0.21 (−0.22, −0.20) <0.001
Percentage of maize −0.18 (−0.19, −0.17) <0.001
Peanut 0.12 (0.11, 0.13) <0.001
Percentage of all crops −0.07 (−0.08, −0.06) <0.001
GV 0.06 (0.05, 0.07) <0.001
Year 2011 −0.05 (−0.05, −0.04) <0.001
Distance to the edge of the eld 0 ND
PA 0 ND
Percentage of cotton 0 ND
Percentage of peanut 0 ND
Percentage of soybean 0 ND
Cotton 0 ND
Year 2009 0 ND
Year 2010 0 ND
Mean Geocoris spp. 0 ND
Variables are ordered from largest to smallest absolute coefcient estimate.
GV=woodland, pasture and noncrop stink bug hosts in woodland edges;
ND=Not determined; PA=perimeter to area ratio of sampled elds.
aCoefcient estimates from LASSO analysis. 95% condence interval (±
95% CI) and P-values were calculated from truncated Gaussian test only for
variables included in best model. Variables excluded from model have coef-
cient estimate of 0.
Environmental Entomology, 2018, Vol. XX, No. X 5
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
Cotton plants are also important in the ecology of G. punctipes
(Torres and Ruberson 2006a) as this species prefers to oviposit on
the plant terminals and young leaves of cotton which can serve as
ready food supplements for nymphs (Torres and Ruberson 2006b).
There were relatively strong negative relationships between
S. invicta densities and maize and soybean. S. invicta have been
found to suppress the densities of many herbivorous insects in soy-
bean (Eubanks 2001) and stinkbug eggs in maize (Tillman 2011).
The low density of re ants observed on soybean and maize plants
suggests that these plants were less preferred foraging sites possibly
because of low availability of prey and nonprey resources during the
study and/or unfavorable properties of the plants or elds for forag-
ing. S.invicta nests were often present in these elds, suggesting that
these ants may have foraged more on the ground than on the plants
(Wickings and Ruberson 2011), which would have biased sampling
success to suction sampling (peanut) over visual (maize) and shake-
cloth (cotton and soybean) counts. However, efforts were made to
take suctions from areas at least 7 cm above the soil and higher
S.invicta prevalence in peanut over that in cotton and soybean has
also been found in the canopy of peanut utilizing sentinel egg masses
(Olson and Ruberson 2012).
Peanut and cotton had positive associations with S.invicta dens-
ity. This is supported by a previous study that showed the prevalence
of this ant species and its importance in causing egg mortality in
the stink bug Nezara viridula L.(Hemiptera: Pentatomidae) in these
crops (Olson and Ruberson 2012). Most commercial cotton varie-
ties have an abundance of extraoral nectaries that provide carbo-
hydrate resources to S.invicta (Agnew etal. 1982). Further, cotton
can also support large populations of honeydew-producing species,
especially aphids (e.g., Diaz etal. 2004), which also benet S.invicta
populations. Peanut provides a dense ground cover that is amenable
as a microhabitat for ants, especially as the season progresses and
the canopy closes (Kharboutli and Mack 1991), and is amenable to
a variety of prey (Vogt etal. 2001).
There was also a strong negative association between the per-
centage of all crops in the landscape and S.invicta density. This rela-
tionship is likely due to high soil disturbance in most crop elds
because of the widespread use of conventional tillage in the study
area; much higher S. invicta density has been found in conserva-
tion versus conventional tillage elds due to reduced soil disturbance
(Stinner and House 1990).
The percentage area of GV in the landscape had a signicant and
negative relationship with Geocoris spp. density. This is contrary to
the often-positive relationship between noncrop habitat and insect
natural enemies in landscapes (Bianchi etal. 2006, Chaplin-Kramer
et al. 2011, Shackelford et al. 2013). Geocoris spp. are omnivo-
rous (Eubanks and Denno 1999, Tillberg et al. 2006, Torres and
Ruberson 2006a, Resasco et al. 2012), and this ability to use food
resources from different trophic levels may allow them to remain in
a habitat for longer periods of time despite variation in availability
of any one prey (Coll and Guershon 2002). The percentage area of
GV was mainly comprised of woodland (planted and natural pine
and oak species), which may not have provided needed resources for
these species, which may prefer arable land. For example, Schmidt
etal. (2008) found negative associations between the abundance and
diversity of spiders and noncrop habitats for several species which
are known to have strong preferences for arable land. In contrast,
S. invicta density was strongly and positively associated with the
percentage area of GV in the landscapes, which is consistent with
other studies of natural enemy species (Bianchi etal. 2006, Chaplin-
Kramer etal. 2011, Shackelford etal. 2013). S. invicta often nest
in woodland (Cook etal. 1997) where there is relatively little soil
disturbance. Therefore, woodland may provide S.invicta a spatial
and temporal refuge from soil disturbance which may result in
woodland being a source of S.invicta into elds and a positive rela-
tionship between their density and percentage area of woodland in
the landscape.
There were no signicant relationships found between the density of
Geocoris spp. and S.invicta, suggesting that these predators have mini-
mal inuence on each other (but see Eubanks 2001). However, Geocoris
spp. density and reproduction and/or nymphal survival were higher in
maize and soybean than in cotton and peanut, where S.invicta density
was the highest. Site selection for initial colonization may be very impor-
tant in affecting spatial distributions of Geocoris spp. and S.invicta as
predators of E.servus, especially for ants that establish sedentary col-
onies (Tschinkel and King 2017). In addition, subsequent differential
reproduction, susceptibility to insecticides and the presence and activity
of natural enemies of Geocoris spp. and S.invicta can also contribute to
differential spatial distribution of these E.servus predators.
In summary, our hypothesis that densities of the two taxa would
be inuenced more by local than landscape scale characteristics was
only partly supported; the percentage area of specic crops and
woodland and pasture in the landscape also inuenced their density
in focal elds. The variability in the density of both taxa over years
may have been related to variable alternative resources for these spe-
cies in cropland and woodland habitats. Despite the variability over
years, the differential habitat use of S.invicta and Geocoris spp. may
have contributed to their combined ability to cause E.servus imma-
ture mortality. The density of the two taxa we studied were associ-
ated with signicant reduction in E.servus reproduction, suggesting
that additional mortality from other predators in the reported stink
bug complex could be highly signicant in the crops studied.
Acknowledgments
We thank Andy Hornbuckle, Melissa Thompson, and numerous student workers
for their help in the eld. The project 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 D.A.A., D.M.O., and J.R.R.).
ReferencesCited
Agnew, C. W., W. L.Sterling, and D. A.Dean. 1982. Inuence of cotton nec-
tar on red imported re ants and other predators. Environ. Entomol. 11:
629–634.
Ali, A. S.A., and T. F. Watson. 1982. Efcacy of dipel and Geocoris puncti-
pes (Hemiptera: Lygaeidae) against the tobacco budworm (Lepidoptera:
Noctuidae) on cotton. J. Econ. Entomol. 75: 1002–1004.
Allen, C. R., D. M.Epperson, and A. S.Garmestani. 2004. Red imported re
ant impacts on wildlife: a decade of research. Am. Midl. Nat. 152: 88–103.
2014. ArcGIS for desktop.10.3, advanced. Environmental Systems Research
Institute, Inc., Redlands, CA.
Bianchi, F. J.J. A., C. J. H. Booij, and T.Tscharntke. 2006. Sustainable
pest regulation in agricultural landscapes: a review on landscape com-
position: biodiversity and natural pest control. Proc. Royal Soc. B 273:
1715–1727.
Chaplin-Kramer, R., M. E.O’Rourke, E. J.Blitzer, and C. Kremen. 2011. A
meta-analysis of crop pest and natural enemy response to landscape com-
plexity. Ecol. Lett. 14: 922–932.
Coll, M., and M.Guershon. 2002. Omnivory in terrestrial arthropods: mixing
plant and prey diets. Annu. Rev. Entomol. 47: 267–297.
Cook, J. L., S. Johnston, R. E.Gold, and S. B. Vinson. 1997. Distribution
of Caenocholax fenyesi (Strepsiptera: Myrmecolacidae) and the habitats
most likely to contain its stylipized host, Solenopsis invicta (Hymenoptera:
Formicidae). Environ. Entomol. 26: 1258–1262.
Davis, D. L. 1981. Population dynamics of four species of Geocoris in the
peanut agroecosystem. PhD dissertation, Texas A&M Libraries.
6 Environmental Entomology, 2018, Vol. XX, No. X
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
Diaz, R., A.Knutson, and J. S.Bernal. 2004. Effect of red imported re ant
on cotton aphid population density and predation of bollworm and beet
armyworm eggs. J. Econ. Entomol. 97: 222–229.
Ehler, L. E. 2002. An evaluation of some natural enemies of Nezara viridula in
northern California. Biocontrol 47: 309–325.
Eubanks, M. D. 2001. Estimates of the direct and indirect effects of red
imported re ants on biological control in eld crops. Biol. Control 21:
35–43.
Eubanks, M. D., and R. F.Denno. 1999. The ecological consequences of varia-
tion in plants and prey for an omnivorous insect. Ecology 80: 1253–1266.
Eubanks, M. D., S. A.Blackwell, C. J.Parrish, Z. D.Delamar, and H.Hull-
Sanders. 2002. Intraguild predation of benecial arthropods by red
imported re ants in cotton. Environ. Entomol. 31: 1168–1174.
Friedman, J., T.Hastie, and R.Tibshirani. 2010. Regularization paths for gen-
eralized linear models via coordinate descent. J. Stat. Software 33: 1.
Hastie, T., R.Tibshirani, and J. H.Friedman. 2009. The elements of statistical
learning: data mining, inference, and prediction. Springer, New York.
Hölldobler, B., and E. O.Wilson. 1990. The ants. Springer Verlag, Heidelberg.
Hooten, M. B., and T.Hobbs. 2015. A guide to Bayesian model selection for
ecologists. Ecol. Monogr. 85: 3–28.
Houston, W., A. F.Miles, K. M.Daane, and M. A.Altieri. 2017. Landscape
diversity and crop vigor outweigh inuence of local diversication on bio-
logical control of a vineyard. Ecosphere 8: 1–18.
Kharboutli, M. S., and T. P. Mack. 1991. Relative and seasonal abundance
of predaceous arthropods in Alabama peanut elds as indexed by pitfall
traps. J. Econ. Entomol. 84: 1015–1023.
King, J. R., and W. R.Tschinkel. 2016. Experimental evidence that dispersal
drives ant community assembly in human-altered ecosystems. Ecology 97:
236–249.
Kiritani, K., N.Hokyo, K.Kimura, and F.Nakasuji. 1965. Imaginal dispersal
of the southern green stink bug Nezara viridula L., in relation to feeding
and oviposition. Jap. J.Appl. Entomol. 9: 291–297.
Lebrun, E. G., R. M.Plowes, and L. E.Gilbert. 2012. Imported re ants near
the edge of their range: disturbance and moisture determine prevalence
and impact of an invasive social insect. J. Anim. Ecol. 81: 884–895.
Lingren, P. D., R. L.Ridgway, and S. L.Jones. 1968. Consumption by several
common arthropod predators of eggs and larvae of two Heliothis species
that attack cotton. Ann. Entomol. Soc. Am. 61: 613–618.
Lofgren, C. S., W. A.Banks, and B. M.Glancey. 1975. Biology and control of
imported re ants. Annu. Rev. Entomol. 20: 1–30.
Lucas, E., and H. E. Maisonhaute. 2015. Differential responses of granivo-
rous, omnivorous and carnivorous species of ground beetles (Coleoptera:
Carabidae) to local and landscape characteristics in a Canadian landscape.
Acta Soc. Zool. Bohem. 79: 83–94.
Naranjo, S. E., and J. L.Stimac. 1985. Development, survival, and reproduc-
tion of Geocoris punctipes (Hemiptera: Lygaeidae): effects of plant feeding
on soybean and associated weeds. Environ. Entomol. 14: 523–530.
Olson, D. M., and J. R.Ruberson. 2012. Crop-specic mortality of south-
ern green stink-bug eggs in Bt- and non-Bt cotton, soybean and peanut.
Biocontrol Sci. Techn. 22: 1417–1428.
Olson, D. M., K. R.Prescott, A. R.Zeilinger, S.Hou, A. W.Cofn, C. M.Smith,
J. R.Ruberson, and D. A.Andow. 2018. Landscape effects on reproduc-
tion of Euschistus servus (Hemiptera: Pentatomidae), a mobile, polypha-
gous, multivoltine arthropod herbivore. Environ. Entomol. 47: 660–668.
doi: 10.1093/ee/nvy045
Ooi, H., and Microsoft. 2017. glmnetUtils: Utilities for “Glmnet”. R package
version 1.0.2. https://CRAN. R-project.org/package=glmnetUtils
Pfannenstiel, R. S., and K. V. Yeargan. 1998. Association of predaceous
Hemiptera with selected crops. Environ. Entomol. 27: 232–239.
Porter, S. D., and W. R. Tschinkel. 1987. Foraging in Slenopsis invicta
(Hymenoptera: Formicidae): effects of weather and season. Environ.
Entomol. 16: 802–808.
Ragsdale, D. W., A. D.Larson, and L. D.Newsom. 1981. Quantitative assess-
ment of the predators of Nezara viridula eggs and nymphs within a soy-
bean agroecosystem using an ELISA. Environ. Entomol. 10: 402–405.
R Core Team. 2016. R: a language and environment for statistical computing.
R Foundation for Statistical Computing, Vienna, Austria.
Readio, J., and M. H.Sweet. 1982. A review of the Geocorinae of the United
States east of the 100th meridian (Hemiptera: Lygaeidae). Misc. Publ.
Entomol. Soc. Am. 12: 1–91.
Resasco, J., D. J.Levey, and E. I. Damschen. 2012. Habitat corridors alter
relative trophic position of re ants. Ecosphere 3: 100.
Ruberson, J. R., G. A. Herzog, W. R. Lambert, and W. J. Lewis. 1994.
Management of the beet armyworm in cotton: role of natural enemies.
Fla. Entomol. 77: 440–453.
Ruberson, J. R., K. V.Yeargan, and B. L.Newton. 2001. Variation in diapause
responses between geographic populations of the predator Geocoris punc-
tipes (Heteroptera: Geocoridae). Ann. Entomol. Soc. Am. 94: 116–122.
SAS Institute Inc. 1998. SAS/STAT user’s guide, version 6, 4th ed, Vol. 1–2. SAS
Institute Inc., Cary, NC.
Schmidt, M. H., C.Thies, W.Nentwig, and T.Tscharntke. 2008. Contrasting
responses of arable spiders to the landscape matrix at different spatial
scales. J. Biogeogr. 35: 157–166.
Schweiger, O., J. P.Maelfait, J. W.van Wingerden, F.Hendrickx, R.Billeter,
M. I. Speelmans, I. Augenstein, B.Aukeme, S.Aviron, D.Bailey, etal.
2005. Quantifying the impact of environmental factors on arthropod com-
munities in agricultural landscapes across organizational levels and spatial
scales. J. Appl. Ecol. 42: 1129–1139.
Shackelford, G., P. R. Steward, T. G. Benton, W. E. Kunin, S. G. Potts,
J. C. Beismeijer, and S. M. Sait. 2013. Comparison of pollinator and nat-
ural enemies: a meta-analysis of landscape and local effects on abundance
and richness in crops. Biol. Rev. 88: 1002–1021.
Stam, P. A., L. D. Newson, and E. N. Lambremont. 1987. Predation and
food as factors affecting survival of Nezara viridula (L.) (Hemiptera:
Pentatomidae) in a soybean ecosystem. Environ. Entomol. 16: 1211–1216.
Stinner, B. R., and G. J.House. 1990. Arthropods and other invertebrates in
conservation-tillage agriculture. Annu. Rev. Entomol. 35: 299–318.
Stoner, A. 1970. Plant feeding by a predacious insect, Geocoris punctipes. J.
Econ. Entomol. 63: 1911–1915.
Tibshirani, R. J., R. Tibshirani, J. Taylor, J. Loftus, and S. Reid. 2017. selec-
tiveInference: tools for post-selection inference. R package version 1.2.2.
https://CRAN. R-project.org/package=selectiveInference.
Tillberg, C. V., D. P. McCarthy, A. G. Dolezal, and A. V. Suarez. 2006.
Measuring the trophic ecology of ants using stable isotopes. Insectes Soc.
53: 65–69.
Tillman, P. G. 2011. Natural biological control of stink bug (Heteroptera:
Pentatomidae) eggs in corn, peanut and cotton farmscapes in Georgia.
Environ. Entomol. 40: 303–314.
Torres, J. B., and J. R.Ruberson. 2006a. Interactions of Bt-cotton and the
omnivorous big-eyed bug Geocoris punctipes (Say), a key predator in cot-
ton elds. Biol. Control 39: 47–57.
Torres, J. B., and J. R.Ruberson. 2006b. Spatial and temporal dynamics of
oviposition behavior of bollworm and three of its predators in Bt and
non-Bt cotton elds. Entomol. Exp. Appl. 120: 11–22.
Tschinkel, W. R. 2006. The re ants. The Belknap Press of Harvard University
Press, Cambridge, MA.
Tschinkel, W. E. 2011. The organization of foraging in the re ant, Solenopssis
invicta. J. Insect Sci. 11: 1–30.
Tschinkel, W. R., and J. R.King. 2013. The role of habitat in the persistence of
re ant (Solenopsis invicta) populations. PLoS One 8: e78580.
Tschinkel, W. R., and J. R. King. 2017. Ant community and habitat limit
colony establishment by the re ant, Solenopsis invicta. Funct. Ecol. 31:
955–964.
Van Den Berg, H., A.Bagus, K.Hassan, A.Muhammad, and S.Zega. 1995.
Predation and parasitism on eggs of two pod-sucking bugs, Nezara virid-
ula and Piezodorus hybneri in soybean. Int. J.Pest Manag. 41: 134–142.
Vogt, J. T., R. A.Grantham, W. A.Smith, and D. C.Arnold. 2001. Prey of the
red imported re ant (Hymenoptera: Formicidae) in Oklahoma peanuts.
Environ. Entomol. 30: 123–128.
Wickings, K. G., and J. R.Ruberson. 2011. Impact of the red imported re ant
(Hymenoptera: Formicidae) on epigeic arthropods of cotton agroecosys-
tems. Ann. Entomol. Soc. Am. 104: 171–179.
Yeargan, K. V. 1979. Parasitism and predation of stink bug eggs in soybean
and alfalfa elds. Environ. Entomol. 8: 715–719.
Environmental Entomology, 2018, Vol. XX, No. X 7
Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy104/5051074
by U S Dept of Agriculture user
on 10 July 2018
