Post on 26-Sep-2020
This is an Accepted Article that has been peer-reviewed and approved for publication in the Insect
Science but has yet to undergo copy-editing and proof correction. Please cite this article as doi:
10.1111/1744-7917.12545.
This article is protected by copyright. All rights reserved.
Author running head: X. Z. Ni et al.
Title running head: Predicting brown stink bug abundance in corn
Correspondence: Xin-Zhi Ni, USDA-ARS, Crop Genetics and Breeding Research Unit, University of
Georgia Tifton Campus, 2747 Davis Road, Bldg. #1,Tifton, GA 31793-0748, USA. Tel: +1 (229) 387-
2340; fax: +1 (229) 387-2321; email: xinzhi.ni@ars.usda.gov
ORIGINAL ARTICLE
Monitoring of brown stink bug (Hemiptera: Pentatomidae) population
dynamics in corn to predict its abundance using weather data
Xin-Zhi Ni1, Ted E. Cottrell2, G. David Buntin3, Xian-Chun Li4, Wei Wang5 and Hong Zhuang6
1USDA-ARS, Crop Genetics and Breeding Research Unit, Tifton, GA 31793, USA; 2USDA-ARS,
Southeastern Fruit and Tree Nut Research Laboratory, Byron, GA 31008, USA; 3Department of
Entomology, University of Georgia, Griffin, GA 30223, USA; 4Department of Entomology, University of
Arizona, Tucson, AZ 85138, USA; 5College of Engineering, China Agricultural University, No. 17
Tsinghua E. Road, Beijing, 100083, China and 6USDA-ARS, Quality and Safety Assessment Research
Unit, Athens, GA, USA
Abstract The brown stink bug (BSB), Euschistus servus (Say) (Hemiptera: Pentatomidae), is a serious
economic pest of corn production in the Southeastern U. S. The BSB population dynamics was
This article is protected by copyright. All rights reserved.
2
monitored for 17 weeks from tasseling to pre-harvest of corn plants (i.e., late May to mid-
September) using pheromone traps in three corn fields from 2005 to 2009. The trap data showed
two peaks in early June and mid-August, respectively. The relationship between trap catch and pre-
growing season weather data was examined using correlation and stepwise multiple factor
regression analyses. Weather indices used for the analyses were accumulated growing degree day
(AGDD), number of days with minimum temperature below 0°C (Subz), accumulated daily maximum
(AMaxT) and minimum temperatures (AMinT) and rainfall (ARain). The weather indices were
calculated with lower (10°C) and upper (35°C) as biological thresholds. The parameters used in
regression analysis were seasonal abundance (or overall mean of BSB adult catch) (BSBm), number
of BSB adults caught at a peak (PeakBSB), and peak week (Peakwk). The BSBm was negatively
related to high temperature (AmaxT or AGDD) consistently, whereas 1stPeakBSB was positively
correlated to both ARain and Subz, irrespective of weather data durations (the first 4, 4.5 and 5
months). In contrast, the 7-month weather data (AGDD7) were negatively correlated to the BSBm
only, but not correlated to the 2nd PeakBSB. The 5-year monitoring study demonstrated that
weather data can be used to predict the BSB abundance at its first peak in tasseling corn fields in the
southeastern U.S. states.
Key words Euschistus servus; first trap catch peak; pheromone trap catch; population dynamics;
stepwise regression modeling; weekly mean
The steady increase in the acreage of transgenic Bacillus thuringiensis (Bt) corn and cotton in the
southeastern coastal plain has reduced insecticide applications in cotton production for the
bollworm, also known as the corn earworm, Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae)
management. The reduction in insecticide applications following the introduction of transgenic Bt
This article is protected by copyright. All rights reserved.
3
crops has resulted in increased population density and damage by piercing-sucking hemipteran pests
of corn (Zea mays L.), cotton (Gossypium hirsutum L.), and soybean [Glycine max (L.) Merrill] in the
Midwest (Koch & Pahs, 2014) and the southeastern U. S. (Bundy & McPherson, 2000; McPherson &
McPherson, 2000; Smith et al., 2009; Herbert & Toews, 2011). The three predominant
phytophagous stink bug species (Hemiptera: Pentatomidae) on corn, cotton, and soybean
production in Georgia are the brown stink bug (BSB), Euschistus servus (Say), the southern green
stink bug, Nezara viridula (L.), and the green stink bug, Chinavia hilare (Say) (Tillman, 2010, 2011;
Herbert & Toews, 2011).
However, limited information is available concerning the variation in seasonal stink bug
population dynamics in corn, cotton and soybean fields across the southeastern U.S., which impedes
the development of an effective integrated pest management program targeting stink bugs in these
ecosystems. The introduction of transgenic crops has created a unique opportunity to greatly
improve IPM programs in these ecosystems by bridging the knowledge gap between genetics and
ecology of both the crops and pests (Ni et al., 2014a). Because plant-feeding pentatomid adults are
highly mobile (or elusive when disturbed), and are polyphagous insects, monitoring their populations
on various host crop plants can be difficult, and costly (Cullen & Zalom, 2000; Kamminga et al.,
2009), which frequently causes a lack of precision for estimating population dynamics and
preventing and/or reducing their damage to crops.
In addition, BSB populations fluctuate greatly from year to year in the southeastern U. S., and trap
catch at population peaks can be critical for the timing of needed insecticide applications. Thus, a
seasonal monitoring study was initiated to describe BSB population dynamics in corn fields from
2005 through 2009 on three farms near Tifton, Georgia. The objectives of the experiment were in
twofold: (i) to document the number of BSB aggregation pheromone trap catch peaks in a corn field
This article is protected by copyright. All rights reserved.
4
from tasseling to pre-harvest; and (ii) to develop a model utilizing weather data (i.e., temperature
and rainfall) to specifically predict the number of BSB caught at the two peaks of trap catches and
the abundance of BSB on corn plants in a growing season.
Materials and methods
Insects and plants
Because no significant difference in damage to soybean was detected among the three species of
stink bugs, including N. viridula, C. hilare, and E. servus (or BSB) (Jones, 1979), and the BSB was
usually the dominant species on corn plants at the Tifton, GA location (Ni et al., 2010), the BSB was
chosen as the subject for the study. Commercial transgenic Bt corn hybrids and the planting dates
for the three corn field on three farms during the five year experimental period were recorded
(Supplementary Table 1). The fields planted for this study on the three research farms (i.e.,
Belflower, Lang-Rigdon, and Gibbs Farms) near Tifton, Georgia were surrounded by either riparian
(or woody) areas at the edge of the farms, fallow fields, or peanut, cotton, or sorghum fields, which
is very similar to the cropping systems as described previously by Ni et al. (2016). Standard
agronomic practices and University of Georgia Extension recommendations (Lee et al., 2007) were
applied to maintain the experimental fields (approximately 0.4 ha). Supplemental irrigation was
applied on an as-needed basis, and no insecticides were applied to the experimental corn fields.
This article is protected by copyright. All rights reserved.
5
Pheromone lure
The pheromone traps used in this study have been described in detail by Mizell and Tedders
(1995) and Cottrell et al. (2000). The aggregation pheromone [i.e., methyl (2E, 4Z)-decadienoate]
and yellow color of the pyramidal trap are the two main chemical and visual cues for attracting the
BSB adults. For each trap, a cattle ear-tag treated with pyrethroid insecticides [(13% piperonyl
butoxide, and 10% λ-cyhalothrin) by Schering-Plough Animal Health Corp., Summit, New Jersey] was
placed in the pheromone trap to kill the insects captured in the trap.
Trap locations and monitoring in a corn field
A total of 10 traps were placed evenly throughout a corn field with three traps on two opposing
sides of the field and four traps through the center of the fields, as shown by Ni et al. (2016).
Because stink bugs prefer to feed on developing fruits/seeds of plants, and low level of stink bug
damage can be seen on seedling stage of corn plants with high level of overwintering population (Ni
et al., 2010; Ni et al., 2016), monitoring of the BSB population in the corn fields started when the
plants were close to flowering or at tasseling (or VT) stage. The corn planting dates varied because of
variation in weather conditions from year to year (Supplementary Table 1). The monitoring period
stopped after 17 weeks at pre-harvest of the corn. The trap catch was monitored twice per wk, and
the total number of BSBs caught per wk was recorded. Because of the sexes of the BSB adults were
not recorded in all years, and very few nymphs were caught in the traps, no analyses were
performed either between two sexes of adults or trap catch of nymphs throughout the season.
Three parameters relating to insect monitoring (as described in Table 1) were: (i) overall mean
number of BSB adults per trap per wk for the 17-week period (BSBm); (ii) the number of BSB caught
This article is protected by copyright. All rights reserved.
6
at the first or second peak (PeakBSB); and (iii) the week when the peak of trap catch occurred
(Peakwk).
Weather data collection and degree day calculation
For the practical adaptation feasibility, a total of five weather-related parameters were used in
data analysis (Table 1). The five parameters were: (1) the number of days with a minimum
temperature below 0°C (Subz) during winter, i.e., from the first to the last day of the frost between
November of the previous year to March of the monitoring year; (2) Accumulated daily maximum
temperature (AMaxT); (3) Accumulated daily minimum temperature (AMinT); (4) Accumulated daily
rainfall (ARain); and (5) accumulated growing degree days (AGDD).
The daily maximum and minimum temperatures and precipitation data from the Tifton location
were retrieved from www.GeorgiaWeather.net for calculation of the AGDD, which was the sum of
daily growing degree day (GDD). The AGDD was calculated using 10°C as lower developmental
threshold, and 35°C as the upper threshold (Murray, 2008). Because the daily maximum (or
minimum) temperature frequently lasts a relatively short period of time (i.e., minutes in some cases)
in a 24 hours period, the adjustment of the developmental threshold is necessary to increase the
precision of the AGDD calculation in corroboration with the biological significance of insect exposure
to weather conditions (Murray, 2008). Thus, the GDD was calculated with the following five
scenarios; (1) when daily maximum and minimum temperatures are below the lower threshold
(10°C), GDD = (lower threshold-lower threshold)/2, or 0; (2) when maximum temperature is above
lower threshold, and minimum temperature is below the lower threshold, GDD = (maximum - lower
threshold)/2-lower threshold; (3) when maximum and minimum temperatures are between the
upper and lower threshold, GDD = (maximum + minimum)/2-lower threshold; (4) when maximum
This article is protected by copyright. All rights reserved.
7
temperature is above and minimum temperature is below the upper threshold, GDD = (upper
threshold+ minimum)/2-lower threshold; and 5) when daily minimum and maximum temperatures
are above the upper threshold, GDD = (upper threshold-lower threshold), or 25 for the current
study.
Because weather data varied from year to year, AGDD was calculated using two biological fix
(biofix) dates (i.e., the first frost day in winter time and the calendar year) to identify the best set of
the AGDD data for the regression model. Within a biofix date, the AGDD was calculated in four
durations, i.e., the first 4, 4.5, 5 and 7 months. While AGDD data from 4, 4.5, and 5 months were
used for predicting the number of brown stink bugs at the first peak, the 7 month data were used to
predict the trap catch at the 2nd peak at pre-harvest. Because variation in weather data usually
occurs between January and April in the southeastern U.S., which is accompanied with the SBS
overwintering period, the AGDD data and other weather-related indices from the first 4, 4.5, and 5
months close to the first trap catch peak (coincide with corn plant tasseling) were utilized for
stepwise multiple factor regression modeling. The use of three durations with a 15 days interval was
to identify the best model(s) with precision in predicting the trap catch at the first peak in the
upcoming growing season of a year.
Experimental design and data analysis
The experiment utilized a split-split-plot design with split in both time (five years) and space (three
farms within each year) according to Cochran and Cox (1957). Each of the three corn fields on
different farms per year were considered as the main plot, and the 17-week of the monitoring
period was considered split in time. The three farms were considered replications. The ten traps
were considered nested within a field, and placed equidistant in each field by assigning three traps
This article is protected by copyright. All rights reserved.
8
on each longitudinal edge of a field and four traps throughout the center of a field (Ni et al., 2016).
The location assignment of the traps was to assess the effect of the field edge on the BSB trap catch.
The weekly mean of the BSB trap catch was subjected to analysis of variance (ANOVA) using a mixed
effects model [PROC MIXED procedure of the SAS software package (SAS Institute, 2012)]. In the
ANOVA, three variables [i.e., year, week, and trap location (edge versus interior)] were used as fixed
factors for the PROC MIXED procedure with a REPEATED statement, whereas the 10 traps nested
within a field were considered random factors. The difference in weekly trap catch within a year, and
among the years, as well as the effect of trap location (i.e., edge versus interior) in a corn field were
also compared. The trap catch data were analyzed after logarithm transformation, because of the
great variation throughout a growing season. All graphs were generated using Sigma Plot® (version
11.0) (SYSTAT, Richmond, CA). The correlation between the trap catch parameters (i.e., BSBm,
peakBSB, and Peakwk) and five weather data parameters (as described in Table 2) were examined
using PROC CORR procedure (SAS Institute, 2012). After the trap catch data had been subjected to
the Shapiro-Wilk test for normality as described by Peng (2009), the stepwise regression analysis
(PROC REG) was performed to assess the relationship between the BSB trap catch and weather data
(SAS Institute, 2012). The regression model was validated using the original five-year data (2005-
2009), correlation between the observed and predicted values were calculated. In addition, the
prediction of the BSB abundance (i.e., 1st peakbsbv, and bsbmv) was performed using the regression
model with the 4.5 months in the most recent 7-year weather data (from 2010 to 2016) at Tifton,
Georgia location.
Results
Seasonal BSB population dynamics
This article is protected by copyright. All rights reserved.
9
The overall pooled trap catch data from 2005 through 2009 (n = 2550) was significantly different
among the 17 weeks (F = 46.51; df = 16, 128; P = 0.0001) (Fig. 1), among the five years (F = 97.21; df
= 4, 32; P = 0.0001) (Fig. 2), and between the edge and interior of a corn field (F = 9.76; df = 1, 8; P =
0.01). The weekly trap catch was also affected by year × week (F = 7.79; df = 64, 512; P = 0.0001),
and year × location (F = 11.61; df = 4, 32; P = 0.01) interactions, whereas the trap catch data were
not affected by week × location, or year × week × location interactions (P > 0.10). The pooled data of
all five years showed that the trap catch at the edge was greater (4.71 ± 0.13, n = 2040) than that in
the interior of a field (3.91 ± 0.23, n = 510). When the trap catch was further analyzed within a year
(n = 510), trap catch differed among years for the 17-week period every year (Figs. 2A-E), but there
was no significant difference in trap catch between corn field edge and interior within a year, except
in 2007 and 2009. Both pooled and annual data showed two peaks of the BSB trap catch in corn
fields from late May to mid-September (Figs. 1 and 2). The first peak occurred at the end of May to
early June while the second peak occurred in mid-August at pre-harvest. In addition, because trap
catch data only recorded the first peak partially in three years (2006-2008) as shown in Figs. 2B, C,
and D, respectively, the peak week was not further examined in regression analysis. Because Ni et al.
(2016) reported no difference between trap locations in relation to surrounding harvested rye field
and pine tree nursery, no further analysis was performed in the current study for trap location in
relation to the surrounding crop fields or ecological habitats.
Temperature, rainfall, and growing degree day data
All variables related to weather data (Supplementary Table 2) were calculated based on the
original weather data collected at the Tifton location weather station. The pre-growing season
weather data were used in either biofix dates for the calculation of accumulated degree days
This article is protected by copyright. All rights reserved.
10
(AGDD). When the calendar year was used as the biofix date, the AGDD of the first 4, 4.5, and 5
months of each year was calculated separately for the regression analysis for the first peak, while
the 7-month data were used for the regression analysis of the 2nd peak (Supplementary Table 2).
While all four durations of AGDD calculated based on the calendar biofix date were presented here,
only one of the four durations (4.5 months) was presented when the first frost day was used as
biofix date. All 8 variables related to trap catch and weather data (Supplementary Table 2) were
used for further correlation and stepwise regression analyses.
Correlation of trap catch with weather data
Because weather data vary among years, as well as from January to mid-May, and weather data
between 1 January to 15 May were also close to the first peak of BSB catch, which was started in late
May and ended in mid-September, the correlation analysis between trap catch (i.e., BSBm,
1stPeakBSB, and 1stPeakwk) and weather data for the first 4.5 months were performed (Table 2).
The number of stink bugs caught at the first peak (1stPeakBSB) was positively correlated to the
number of subzero days (Subz) and rainfall (ARain4.5), but the overall mean (BSBm) were negatively
correlated to the AGDD4.5 and AMaxT4.5, respectively. The 1stPeakwk was not correlated to any of
the weather indices (Table 2), so the Peakwk data were not used in regression modeling. The
correlation analyses between trap catch and other weather data durations (i.e., the 4, 5 and 7
months, respectively) showed similar patterns, and the results are not presented here.
Stepwise multiple factor regression modeling
This article is protected by copyright. All rights reserved.
11
The normality test confirmed that the trap catch dataset was not different from the normal
distribution (Shapiro-Wilk test of normality for trap data was not significant, P values = 0.4–0.95).
The stepwise regression analysis was conducted individually for each of the three trap catch data
parameters against the five weather data parameters as described in Table 1. A total of 12 (4
periods of early season weather data x 3 trap catch data parameters) regression analysis procedures
were conducted. However, the weekly mean of the trap catch (BSBm) can be predicted using the
five-year dataset for all weather data periods (i.e., 4, 4.5, 5, and 7 months), and the number of stink
bugs at the first peak (1stPeakBSB) can be predicted using 4, 4.5, and 5 month weather data. The 7-
month data could only predict the overall weekly mean (BSBm), but not stink bug catch at the
second peak (2ndPeakBSB). In addition, when the first frost day was used as the biofix date, the
goodness-of-fit of the regression model was not as good as the previous seven models using only
calendar-year weather data as the biofix date (Table 3). Thus, only one regression model based on
the AGDD data using the first frost day as the biofix date was developed and presented in Table 3.
The stepwise multiple factor regression analyses utilized 2 to 5 steps (Table 3). The overall weekly
mean (BSBm) was negatively correlated to AMaxT or AGDD7, while the trap catch at the first peak
(1stPeakBSB) was positively correlated to the number of Subz days, and ARain (Table 3).
Validation of the linear regression models
The model validation (or predicted) values (i.e., 1stpeakbsbv, bsbmv) were paired with the
corresponding field observations (i.e., 1stPeakBSB and BSBm) for each year, as shown in
Supplementary Table 3. The model prediction of the 1stpeakbsbv and bsbmv using 4.5-month
weather data showed the least deviation between the observed and predicted values shown by the
bolded values of the means and standard errors in Supplementary Table 3. The correlation
This article is protected by copyright. All rights reserved.
12
coefficients between observed and predicted values of the peak data (1stPeakBSB) were greater (r =
0.997, P = 0.0001, n = 15) than the observed and predicted values of weekly means (BSBm) (r =
0.911, P = 0.0001, n = 15). In addition, the 1stPeakBSB was also positively correlated to BSBm (r =
0.649, P = 0.009, n = 15). The validations of the two models using the 7-month weather data were
not performed because the model is of little value at the end of the growing season, which was not
as useful as the six models using the pre-growing season weather data. Similarly, the model using
the first frost day + 4.5-month weather data was not validated because the goodness-of-fit for the
model was lower (r2 < 0.87) than the other six models associated with the first peak (r2 > 0.89) as
shown in Table 3. Furthermore, the BSB abundance in the most recent 7 years (from 2010 to 2016)
was also calculated using the linear regression models with 4.5-month weather data (Supplementary
Table 4). The correlation of two predict values (i.e., 1stpeakbsbv and bsbmv) were not significant (r =
0.742, P = 0.056, n = 7), which could be caused extremely warm winter for 2011–2012 that led to the
bsbmv value being negative in 2012, as shown in Supplementary Table 4.
Discussion
Since the introduction of synoptic population model by Southwood and Comins (1976), a number of
models that use deterministic and stochastic factors influencing population dynamics of both r- and
K-strategist insect pests have been examined in detail in ecological research of insect pests (Huffaker
et al., 1984). In particular, utilization of the AGDD to understand insect biology and physiology and
then to predict insect population dynamics throughout a growing season of a given crop has been
examined extensively in recent decades on a number of insect pests and their host plants for both
agricultural (Higley et al., 1986; Kingsolver, 1989; Murray, 2008) and forestry ecosystems (van Asch
& Visser, 2007). For pentatomid pests, as a group of K-strategists, Kamminga et al. (2009) described
This article is protected by copyright. All rights reserved.
13
a model using weather data to predict C. hilare abundance. They reported that seasonal flight
activities of C. hilare can be predicted by utilizing an 18-year (1990-2007) dataset of black light catch
in Virginia, U.S.A. They described that two weather parameters (i.e., mean monthly precipitation and
number of days below freezing) in the early season (from January to April) were adequate to predict
the weekly mean of adults caught in a black light trap. In addition, the three peaks of the black light
catch throughout the season with 10%, 50%, and 90% of the seasonal catch were identified to occur
at 153, 501 and 1 066 degree days from 1 January of a year (Kamminga et al., 2009). Nielson et al.
(2013) reported that black light traps can be used to monitor the distribution and abundance of an
invasive species - brown marmorated stink bug, Halyomorpha halys (Stål) (Hemiptera:
Pentatomidae) in New Jersey. A network of more than 70 black light traps throughout New Jersey
on vegetable and fruit farms participated in the IPM scouting program from 2004–2011 were utilized
to monitor H. halys distribution and abundance, in addition to key lepidopteran pests [e.g., the
European corn borer, Ostrinia nubilalis Hübner (Lepidoptera: Crambidae), and Helicoverpa zea
(Boddie) (Lepidoptera; Noctuidae)] in New Jersey (Nielson et al., 2013). They determined that 685
degree days with the lower threshold of 14°C was required for female H. halys maturation, which fell
into the 29th and 32nd Julian weeks in southern and northern New Jersey, respectively.
The current study is one of a series of studies striving to understand the biology and ecology of
BSB population dynamics associated with corn production. In addition to understanding the impact
of BSB feeding damage on grain quality in corn production (Ni et al., 2010), stink bug damage on
corn kernels was correlated to aflatoxin contaminations in a corn field during some years, but not
every year (Ni et al., 2011; Ni et al., 2014c). However, the common smut, Ustilago maydis (Persoon)
Roussel, infections of corn ears were not correlated to the BSB carrying smut spores (Ni et al.,
2014b). Diurnal flight activities of the BSB adults assessed using pheromone traps when corn plants
were at tasseling stage (VT), coincides with the harvest of winter grain crops (e.g., wheat and rye)
This article is protected by copyright. All rights reserved.
14
(Reisig, 2011; Ni et al., 2016). The findings from the current study on the BSB (or E. servus), a
bivoltine insect that overwinters as adults (McPherson & McPherson, 2000), are consistent with the
report on C. hilare abundance in relation to weather data (Kamminga et al., 2009). The cause of
positive correlation of the first peak BSB catch to cold and wet wintery weather (i.e., Subz and ARain)
could be the result of high survivorship of sustained overwintering (diapause) BSB adult populations.
In contrast, the negative correlation between the BSB abundance (BSBm) and temperature (i.e.,
AMaxT or AGDD) could be the result of termination of diapause, which led to high mortality under
brief deep frost conditions subsequently. Temperature and photoperiod have been well
documented for inducing and terminating insect diapause in general (Xu et al., 2014). The cold and
wet conditions under short photoperiod in winter or early spring sustained diapause state of the
overwintered adults, which leads to high survivorship, and later high trap catch of the overwintering
adults at the first peak of the BSB monitoring period. In contrast, an extended period with high
temperature during an insect diapause might terminate diapause, and subsequent freezing
temperature with precipitation in winter or early spring could lead to high mortality of a post-
diapause insect population. Thus, the 1stPeakBSB was high following a cold and wet winter and vice
versa. The six linear regression models for the first peak described in the current study could be
readily used to predict the BSB abundance at corn plant tasseling (flowering) time, which is critically
important for insecticide applications to prevent corn ear and kernel damage (Reisig, 2011), as well
as allowing growers to be ready for assessing BSB abundance in other crops (e.g., cotton, peanut,
and soybean) that flower at a later time in the cropping systems of the southeastern U.S. (Greene et
al., 2001; Herbert & Toews, 2011; Tillman, 2011; Temple et al., 2013). The linear regression models
between pheromone trap catch of the BSB adults at the first peak and pre-growing season (4.5
month) weather data described here could be critically valuable to preserve maize and other crop
yield and quality in all southern U.S. states under warm temperate climate with similar cropping
This article is protected by copyright. All rights reserved.
15
systems (e.g., corn/sorghum, cotton, and peanut/soybean as main field crops, and pecan and peach
as main orchard crops). In a similar manner, the model for the second peak of BSB adult catch and
the 7-month weather data could be further examined and utilized to estimate overwintering
population of this bivoltine pest in the southeastern U.S. states. Such information would be valuable
in integrated management of the brown stink bug and other pests in the agricultural ecosystems in
the southeastern coastal plain region of the U.S.
In conclusion, the regression models developed from the current 5-year study can be utilized to
predict the BSB abundance in corn using pre-growing season weather data by targeting the two
peaks. While the predicted stink bug number at the first peak would be valuable in designing
contingent management strategies in corn and other crops of the same year, the predicted stink bug
abundance at the second peak could be used to estimate the overwintering population for this
bivoltine pest, which is valuable for assessing brown stink bug abundance for the following year. In
particular, the two models developed using the first 4.5 month (or pre-growing season) weather
data of a year, would allow growers to predict BSB damage using trap catch at the first peak
(1stPeakBSB) and seasonal BSB abundance (BSBm) before the BSB infestation actually occurs, which
is not only critically important for timely stink bug management on corn, but also on cotton and
soybean production in the southeastern U.S. states.
Acknowledgments
Mention of trade names or commercial products in this article is solely for the purpose of providing
specific information and does not imply recommendation or endorsement by the U. S. Department
of Agriculture. We thank G. Gunawan, K. Da, R. Powell, Jr., J. C. Mullis, and P. M. Tapp (USDA-ARS,
This article is protected by copyright. All rights reserved.
16
Crop Genetics and Breeding Research Unit, Tifton, GA) for their technical assistance during the
experiment. We thank J. K. Westbrook (USDA-ARS Insect Control and Cotton Disease Research Unit,
College Station, TX) and N. C. Elliott (USDA-ARS Plant Science Research Laboratory, Stillwater, OK) for
their critical reviews of the earlier version of the manuscript. We also thank anonymous reviewers
and the editor for their advice and constructive comments that have strengthened the manuscript.
The research was supported in part by the Georgia Agricultural Commodity Commission for Corn.
Disclosure
All authors declare no conflict of interest.
References
Bundy, C.S. and McPherson, R.M. (2000) Dynamics and seasonal abundance of stink bugs
(Heteroptera: Pentatomidae) in a cotton–soybean ecosystem. Journal of Economic
Entomology, 93, 697–706.
Cochran, W.G. and Cox, G.M. (1957) Factorial experiments with main effects confounded: split-plot
designs. Experimental Designs (eds. W.G. Cochran & G.M. Cox), pp. 293–316. John Wiley and
Sons, Inc. New York, NY.
Cottrell, T.E., Yonce, C.E. and Wood, B.W. (2000) Seasonal occurrence and vertical distribution of
Euschistus servus (Say) and Euschistus tristigmus (Say) (Hemiptera: Pentatomidae) in pecan
orchards. Journal of Entomological Science, 35, 421–431.
This article is protected by copyright. All rights reserved.
17
Cullen, E.M. and Zalom F.G. (2000) Phenology-based field monitoring for consperse stink bug
(Hemiptera: Pentatomidae) in processing tomatoes. Environmental Entomology, 29, 560–
567.
Greene, J.K., Turnipseed, S.G., Sullivan, M.J. and May, O.L. (2001) Treatment thresholds for stink
bugs (Hemiptera: Pentatomidae) in cotton. Journal of Economic Entomology, 94, 403–409.
Herbert, J.J., and Toews, M.D. (2011) Seasonal abundance and population structure of brown stink
bug (Hemiptera: Pentatomidae) in farmscapes containing corn, cotton, peanut, and soybean.
Annals of the Entomological Society of America, 104, 909–918.
Higley, L.G., Pedigo, L.P. and Ostlie, K.R. (1986) DegDay: a program for calculating degree-days, and
assumptions behind the degree-day approach. Environmental Entomology, 15, 999–1016.
Huffaker, C.B., Berryman, A.A. and Laing, J.E. (1984) Natural control of insect populations. Ecological
Entomology (eds. C.B. Huffaker & R.L. Rabb), pp. 359–398. John Wiley and Sons, New York.
Kamminga, K.L., Herbert Jr. D.A., Kuhar, T.P. and Brewster, C.C. (2009) Predicting black light trap
catch and flight activity of Acrosternum hilare (Hemiptera: Pentatomidae) adults.
Environmental Entomology, 38, 1716–1723.
Kingsolver, J.G. (1989) Weather and the population dynamics of insects: integrating
physiological and population ecology. Physiological Zoology, 62, 314–334.
Koch, R.L. and Pahs, T. (2014) Species composition, abundance, and seasonal dynamics of stink bugs
(Hemiptera: Pentatomidae) in Minnesota soybean fields. Environmental Entomology, 43,
883–888.
This article is protected by copyright. All rights reserved.
18
Lee, R.D., Harris, Jr. G.H., Prostko, E., Buntin, G.D., Kemerait, R. and Harris, K.G. et al. (2007) A guide
to corn production in Georgia 2008. Georgia Coop. Extension Service Misc. Publication CSS
01-08, 61 pp.
McPherson, J.E. and McPherson, R.M. (2000) Euschistus spp. Stink Bugs of Economic Importance in
America North of Mexico (eds. J.E. McPherson & R.M. McPherson), pp. 101–127. CRC Press,
Boca Raton, FL.
Mizell III, R.F. and Tedders W.L. (1995) A new monitoring method for detection of the stinkbug
complex in pecan orchards. Proceedings of Southeastern Pecan Growers Association, 88, 36–
40.
Murray, M.S. (2008) Using Degree Days to Time Treatments for Insect Pests. Utah State University
Extension and Utah Plant Pest Diagnostic Laboratory IPM-05-08, April 2008, 5 pp.
Ni, X.Z., Da, K., Buntin, G.D., Cottrell, T.E., Tillman, P.G. Olson, D.M. et al. (2010) Impact of brown
stink bug (Heteroptera: Pentatomidae) feeding on corn grain yield components and quality.
Journal of Economic Entomology, 103, 2072–2079.
Ni, X.Z., Wilson, J.P., Buntin, G.D., Guo, B., Krakowsky, M.D. Lee, R.D.et al. (2011) Spatial patterns of
aflatoxin levels in relation to ear-feeding insect damage in pre-harvest corn. Toxins, 3, 920–
931.
Ni, X.Z., Lei, Z., He, K., Li, X., Li, X. and Xu, W. (2014a) Integrated pest management is the lucrative
bridge connecting the ever emerging knowledge islands of genetics and ecology. Insect
Science, 21, 537–540.
This article is protected by copyright. All rights reserved.
19
Ni, X.Z., Toews, M.D., Buntin, G.D., Carpenter, J.E., Huffaker, A. Schmelz, E.A. et al. (2014b) Influence
of brown stink bug feeding, planting date and sampling time on common smut infection of
maize. Insect Science, 21, 564–571.
Ni, X.Z., Wilson, J.P., Toews, M.D., Buntin, G.D., Lee, R.D. and Li, X. et al. (2014c) Evaluation of spatial
and temporal patterns of insect damage and aflatoxin level in the pre-harvest corn fields to
improve management tactics. Insect Science, 21, 572–583.
Ni, X., Cottrell, T.E., Toews, M.D., Tillman, P.G. and Buntin, G.D. (2016) Diurnal activities of the brown
stink bug (Hemiptera: Pentatomidae) in and near tasseling corn fields. Journal of
Entomological Science, 51, 226–237.
Nielsen, A.L., Holmstrom, K., Hamilton, G.C., Cambridge, J., Ingerson–Mahar, J. (2013) Use of black
light traps to monitor the abundance, spread, and flight behavior of Halyomorpha halys
(Hemiptera: Pentatomidae). Journal of Economic Entomology, 106, 1495–1502.
Peng, C.Y.J. (2009) Comprehensive descriptive analysis and normality test. In: SAGE Research
Methods, pp. 163–207.
Reisig, D.D. (2011) Insecticidal management and movement of the brown stink bug, Euschistus
servus, in corn. Journal of Insect Science, 11, 168.
SAS Institute (2012) SAS/STAT version 9.4 User’s guide. The SAS Institute, Cary, NC.
Smith, J.F., Luttrell, R.G. and Greene, J.K. (2009) Seasonal abundance, species composition, and
population dynamics of stink bugs in production fields of early and late soybean in South
Arkansas. Journal of Economic Entomology, 102, 229–236.
This article is protected by copyright. All rights reserved.
20
Southwood, T.R.E. and Comins, H.N. (1976) A synoptic population model. Journal of Animal Ecology,
45, 949–965.
Temple, J.H., Davis, J.A., Micinski, S., Hardke, J.T., Price, P. and Leonard, B.R. (2013) Species
composition and seasonal abundance of stink bugs (Hemiptera: Pentatomidae) in Louisiana
Soybean. Environmental Entomology, 42, 648–657.
Tillman, P.G. (2010) Composition and abundance of stink bugs (Heteroptera: Pentatomidae) in corn.
Environmental Entomology, 39, 1765–1774.
Tillman, P.G. (2011) Influence of corn on stink bugs (Heteroptera: Pentatomidae) in subsequent
crops. Environmental Entomology, 40, 1159–1176.
van Asch, M. and Visser, M.E. (2007) Phenology of forest caterpillars and their host trees: the
importance of synchrony. Annual Review of Entomology, 52, 37–55.
Xu, L., Ni, X., Wang, Z. and He, K. (2014) Effects of photoperiod and temperature on diapause
induction in Conogethes punctiferalis (Lepidoptera: Pyralidae). Insect Science, 21, 556–563.
Manuscript received June 20, 2017
Final version received September 15, 2017
Accepted September 24, 2017
This article is protected by copyright. All rights reserved.
21
Table 1 Acronyms of trap catch and weather data used for regression analysis
Abbreviation Variable
Trap catch
BSBm Overall weekly mean of brown stink bugs per trap throughout a season
PeakBSB Number of brown stink bugs recorded at the first or second peak
Peakwk The week when the first or second peak of stink bug catch occurred
Weather data
Subz The number of days with frost, or minimum temperature < 0°C throughout the whole
winter
AGDD Accumulated growing degree day
AMinT Accumulated daily minimum temperature in Celsius
AMaxT Accumulated daily maximum temperature in Celsius
ARain Accumulated rainfall (mm)
This is an Accepted Article that has been peer-reviewed and approved for publication in the Insect
Science but has yet to undergo copy-editing and proof correction. Please cite this article as doi:
10.1111/1744-7917.12545.
This article is protected by copyright. All rights reserved.
Table 2 Pearson’s Correlation Coefficients of 8 indices of the five year (n = 5) data for the 4.5-month
weather data from 1 January to 15 May between 2005 and 2009
Subz AGDD4.5 AMaxT4.5 AminT4.5 ARain4.5 1stpeakBSB 1stpeakwk
AGDD4.5 −0.42
0.49
AMaxT4.5 −0.62 0.96
0.26 0.009
AMinT4.5 −0.28 0.68 0.59
0.64 0.21 0.29
ARain4.5 0.62 −0.34 −0.55 0.10
0.26 0.57 0.34 0.88
1stpeakBSB 0.89 −0.45 −0.67 −0.17 0.90
0.04 0.44 0.21 0.78 0.04
1stpeakwk 0.66 −0.79 −0.86 −0.25 0.45 0.60
0.22 0.11 0.06 0.68 0.44 0.29
BSBm 0.54 −0.96 −0.99 −0.58 0.58 0.65 0.80
This article is protected by copyright. All rights reserved.
23
0.35 0.009 0.002 0.30 0.30 0.23 0.11
In each table cell, top value = r, bottom value = P of the MANOVA statement of the SAS
software. The bolded values indicate significant correlation with P < 0.05.
Table 3 Linear regression models based on different durations of weather data.
Biofix date Regression equation† Statistics of stepwise
regression (P < 0.05)
Calendar
year 4 months
BSBm = 49.03 − 0.019×(AMaxT4) r2 = 0.887; 2 steps
1stPeakBSB = −13.20 + 0.825×(Subz) +
0.034×(ARain4) r2 = 0.996; 3 steps
4.5 months BSBm = 57.19 − 0.019×(AMaxT4.5) r2 = 0.973; 2 steps
1stPeakBSB = −13.44 + 0.896×(Subz) +
0.030×(ARain4.5) r2 = 0.994; 3 steps
5 months BSBm = 55.07 − 0.016×(AMaxT5) r2 = 0.914; 4 steps
1stPeakBSB = − 9.70 + 0.683×(Subz) +
0.027×(ARain5) r2 = 0.995; 5 steps
7 months BSBm = 43.82 − 0.025×(AGDD7) r2 = 0.865; 2 steps
First frost
day 4.5 months 1stPeakBSB = −9.86 + 0.045×(ARain) r2 = 0.819; 4 steps
†In regression equation column, BSBm = overall weekly mean of brown stink bugs per trap
throughout the 17-week monitoring period; 1stPeakBSB = number of brown stink bugs at the first
peak; AGDD = accumulated growing degree days of the first 7 months; AMaxT or AMinT =
This article is protected by copyright. All rights reserved.
24
accumulated maximum or minimum daily temperature readings; and ARain = accumulated rainfall
data. Subz = number of days with minimum temperature below 0°C.
Figure Captions
Trap Catch of Brown Stink Bugs (2005-2009, n =150)
Week (AGDD)
1 (9
51.9
3)
2 (1
052.
78)
3 (1
156.
57)
4 (1
273.
66)
5 (1
388.
03)
6 (1
506.
42)
7 (1
625.
39)
8 (1
738.
71)
9 (1
860.
39)
10 (1
982.
29)
11 (2
102.
96)
12 (2
226.
97)
13 (2
351.
22)
14 (2
472.
75)
15 (2
590.
17)
16 (2
704.
77)
17 (2
812.
67)
Num
be
r o
f b
row
n s
tink b
ug
s p
er
tra
p
0
2
4
6
8
10
12
Fig. 1 Weekly mean of trap captured E. servus adults (or BSBm) during the 17-week monitoring
period from 2005 to 2009 with 30 traps per year (10 traps for each of the three fields) (n = 150). The
error bars represent the standard error of the mean. Week (AGDD) denotes the five-year mean of
AGDD (calculated from 1 January of each year) at the beginning of the weekly monitoring period.
This article is protected by copyright. All rights reserved.
25
E) 2009 (n = 30)
Week
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
C) 2007 (n = 30)
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Nu
mb
er
of b
row
n s
tink
bu
gs
pe
r tr
ap
0
5
10
15
20
25
30
D) 2008 (n = 30)
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
B) 2006 (n = 30)
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
300 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
Week: F = 15.44; df = 16,128; P < 0.0001
Trap location: F = 5.74; df = 1,8; P = 0.04
Week*location: F = 0.73; df = 16,128, P = 0.76
A) 2005 (n = 30)
Week: F = 26.01; df = 16,128; P < 0.0001
Trap location: F = 2.25; df = 1,8; P = 0.17
Week*location: F = 1.93; df = 16,128, P = 0.02
Week: F = 7.10; df = 16,128; P < 0.0001
Trap location: F = 18.27; df = 1,8; P < 0.003
Week*location: F = 0.93; df = 16,128, P = 0.53
Week: F = 14.18; df = 16,128; P < 0.0001
Trap location: F = 0.02; df = 1,8; P = 0.90
Week*location: F = 0.81; df = 16,128, P = 0.67
Week: F = 16.30; df = 16,128; P < 0.0001
Trap location: F = 1.05; df = 1,8; P = 0.34
Week*location: F = 0.74; df = 16,128, P = 0.75
Fig. 2 Weekly mean of trap captured E. servus adults during the 17-week monitoring period with 10
traps in each of the three corn fields (n = 30) from 2005 to 2009, respectively. The error bars
represent the standard error of the mean.
本文献由“学霸图书馆-文献云下载”收集自网络,仅供学习交流使用。
学霸图书馆(www.xuebalib.com)是一个“整合众多图书馆数据库资源,
提供一站式文献检索和下载服务”的24 小时在线不限IP
图书馆。
图书馆致力于便利、促进学习与科研,提供最强文献下载服务。
图书馆导航:
图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具