Download Abiotic and Biotic Factors Affecting the Distribution and Abundance

Abiotic and Biotic Factors Affecting the Distribution and
Abundance of Soybean Aphid in Central North America
by
Christine Anne Bahlai
A Thesis
presented to
The University of Guelph
In partial fulfilment of requirements
for the degree of
Doctor of Philosophy
in
Environmental Biology
Guelph, Ontario, Canada
© Christine Bahlai, April, 2012
ABSTRACT
Abiotic and Biotic Factors Affecting the Distribution and
Abundance of Soybean Aphid in Central North America
Christine A. Bahlai
University of Guelph, 2012
Advisors:
Rebecca H. Hallett
Arthur W. Schaafsma
Soybean aphid, Aphis glycines Matsumura, is an important pest of North
American soybean. This dissertation identifies and addresses knowledge gaps, and
integrates existing knowledge regarding distribution and abundance of this species.
Early summer soybean colonization patterns by A. glycines were examined
relative to landscape parameters, including density of overwintering hosts (buckthorn).
An information-theoretic model selection approach was used to determine which
landscape parameters were most influential in the distribution of colonizing aphids.
Though buckthorn abundance best explained aphid colonization and population density, a
density-dependent effect was observed. When aphid populations were low, more aphids
were found in the vicinity of buckthorn, when aphid populations were higher, more
aphids were found farther from buckthorn.
Suction trap captures of migrating populations of A. glycines from 2005-2009
from a suction trap network covering much of central North America were examined. A
model selection approach was used to determine the environmental triggers of summer
and fall aphid flights, and spatial analysis and modeled wind trajectories were used to
examine patterns in the abundance of alates. Two alate activity peaks were observed in
fall. In summer, formation of alates was a function of field infestation.
A tritrophic population model was built using DYMEX, a mechanistic lifecycle
based modeling software package. The model incorporated soybean, A. glycines, and
three natural enemy species, interacting based on phenological, physiological and
functional response data available in the literature. The model was validated using
Ontario field data, and several simulations were performed and are discussed.
An evaluation of proposed control strategies for efficacy and impact on natural
enemies and the environment was conducted. Two novel concepts are presented: the
natural enemy unit, a standardization of the impact of predator guild on prey populations
by the number of prey an individual predator can eat, and the selectivity index, where the
selectivity of a pesticide is a function of the change in ratio of natural enemy units to prey
before and after treatment. The selectivity index was inversely correlated with the
Environmental Impact Quotient (EIQ), a theoretical measure of impact, validating EIQ's
field applicability.
Dedication
This dissertation is dedicated to the memory of my grandmother, “Baba” Pasha
Karnjenko Bahlai, whose hard work, dedication to her family, perseverance against all
odds, not to mention her talent for growing her own food, serve as a constant inspiration
to me.
iv
ACKNOWLEDGEMENTS
There are so many people without which this thesis would not have been possible.
I’d like to thank the Natural Sciences and Engineering Research Council of
Canada for the generous fellowship that supported me through this degree. Additional
scholarships were generously provided by the Keefer family trust; the Mary Edmunds
Williams trust, the family of Fred W. Presant, and the University of Guelph. This
research was funded by a grant to Rebecca Hallett and Art Schaafsma from the
Agriculture and Agri-Food Canada Pest Management Centre’s Pesticide Risk Reduction
Program.
To my advisory committee: Rebecca Hallett, Art Schaafsma, Christina Difonzo,
Terry Gillespie and Jonathan Newman. Thank you all for your support, advice,
mentorship, inspiration, motivation and patience over these last five years. You’re all
wonderful people and I’m so grateful to have been able to work with you.
Rebecca, you deserve a special acknowledgement. You have been my mentor and
my example for the type of person I want to become. Thank you, thank you, thank you.
For everything.
To my family, friends, colleagues, and caretakers. Thank you for standing by me
during this rough time (among other rough times). I’m sorry I’m so mean while I’m
writing.
To name a few of the brave: Michael Bahlai, Jillian Bainard, Rob Baldwin,
Tracey Baute, Neil Carter, David Cheung, Gabrielle Duval, Hannah Fraser, Jean Gerrath,
Angela Gradish, Richard Harrington, David Hooker, Lillian Iantorno, Mohammed Jadi,
Darren Kriticos, Paul Kron, Doug Landis, Debbie MacDonald, Andrea Miller, Ang
Orshinski, Todd Phibbs, Jonathan Schmidt, Joy Roberts, Cynthia Scott-Dupree, Mark
Sears, Momtaz Talukdar, David Voegtlin, Virginia Warren, Linda Wing, Ross Weiss,
Andrew Welsman, thank you for offering advice/ offering an ear/ offering a hand/ taking
care of me/ keeping me alive.
To my laboratory conspecifics Adam Brunke, Lauren Des Marteaux, Braden
Evans, Andrew Frewin, Erik Glemser, Jamie Heal, Cara McCreary, Jocelyn Smith,
Yingen Xue, it has been my pleasure to work with such a brilliant group of wonderful
people. Your friendship will be cherished always.
To my husband, John. Thank you for being there, as a sounding board, for talking
me out of the wrong things and talking me into the right things. I can’t believe how long
you’ve listened to me talk about my project. I love you.
To my daughter, Penelope. Thank you for your patience, and for letting Mommy
work sometimes. No more bedtime stories about aphids. I love you.
v
TABLE OF CONTENTS
Abstract ............................................................................................................................... ii
Dedication .......................................................................................................................... iv
Acknowledgements ............................................................................................................. v
List of tables ....................................................................................................................... ix
List of figures ................................................................................................................... xiii
Chapter 1: Soybean aphid in North America: an introduction and review ......................... 1
1.0 Introduction ............................................................................................................................ 1
1.1 Background ............................................................................................................................ 1
1.1.1 Spring migrations of A. glycines and patterns in soybean field colonization ................. 6
1.1.2 Migration and dispersal by A. glycines in summer and fall ............................................ 8
1.1.3 Integrating the literature: the need for a full-lifecycle phenological model for Aphis
glycines .................................................................................................................................. 11
1.1.4 Responding to economically damaging outbreaks of A. glycines ................................. 13
1.2 General objectives ................................................................................................................ 16
Chapter 2: Modeling distribution and abundance of soybean aphid in soybean fields using
measurements from the surrounding landscape ................................................................ 19
2.0 Abstract ................................................................................................................................ 19
2.1 Introduction .......................................................................................................................... 20
2.2 Materials and Methods......................................................................................................... 23
2.2.1 Study site and landscape parameters ............................................................................. 23
2.2.2 Aphid monitoring .......................................................................................................... 23
2.2.3 Models........................................................................................................................... 26
2.3 Results .................................................................................................................................. 27
2.3.1 Aphid presence and colonization .................................................................................. 27
2.3.2 Aphid density ................................................................................................................ 29
2.4 Discussion ............................................................................................................................ 30
2.4.1 Statistical methods ........................................................................................................ 30
2.4.2 Factors affecting aphid distribution .............................................................................. 31
2.5 Acknowledgements .............................................................................................................. 36
vi
Chapter 3: Factors inducing migratory forms of soybean aphid and an examination of
North American spatial dynamics of this species in the context of migratory behaviour. 37
3.0 Abstract ................................................................................................................................ 37
3.1 Introduction .......................................................................................................................... 38
3.2 Methods ............................................................................................................................... 43
3.2.1 Analysis......................................................................................................................... 46
3.2.2 Determination of migration status of captured aphids .................................................. 46
3.2.3 Response to environmental variables ............................................................................ 47
3.2.4 Geospatial analysis and flight trajectories..................................................................... 50
3.3 Results .................................................................................................................................. 51
3.3.1 Determination of migration status of captured aphids .................................................. 51
3.3.2 Response to environmental variables ............................................................................ 52
3.3.3 Geospatial analysis and flight trajectories..................................................................... 64
3.4 Discussion ............................................................................................................................ 71
3.5 Acknowledgements .............................................................................................................. 79
Chapter 4: A mechanistic model for a tritrophic interaction involving multiple natural
enemies: using natural enemy units to quantify the impact of a guild.............................. 81
4.0 Abstract ................................................................................................................................ 81
4.1 Introduction .......................................................................................................................... 82
4.2 Model specification.............................................................................................................. 85
4.2.1 Model calibration ........................................................................................................ 107
4.2.2 Model validation ......................................................................................................... 109
4.2.3 Whole season simulations ........................................................................................... 113
4.3 Discussion .......................................................................................................................... 113
4.4 Acknowledgements ............................................................................................................ 121
Chapter 5: Choosing organic pesticides over synthetic pesticides may not effectively
mitigate environmental risk in soybeans......................................................................... 123
5.0 Abstract .............................................................................................................................. 123
5.1 Introduction ........................................................................................................................ 124
5.2 Materials and Methods ....................................................................................................... 127
5.2.1 Selection of insecticides for inclusion in experiments ................................................ 127
5.2.2 Determination of direct contact toxicity to natural enemies ....................................... 127
5.2.3 Determination of field efficacy and selectivity ........................................................... 129
5.2.4 Field Selectivity calculation ........................................................................................ 130
vii
5.2.5 Environmental Impact assessment .............................................................................. 131
5.3 Results ................................................................................................................................ 132
5.4 Discussion .......................................................................................................................... 136
5.5 Acknowledgements ............................................................................................................ 139
Chapter 6: General Discussion and Conclusions ............................................................ 140
6.1 Discussion .......................................................................................................................... 140
6.1.1 Distribution of A. glycines colonizing soybean........................................................... 140
6.1.2 Alatoid morph production and regulation ................................................................... 142
6.1.3 Quantifying biocontrol services: the Natural Enemy Unit .......................................... 145
6.1.4 Development of a tritrophic population model ........................................................... 146
6.1.5 Managing A. glycines: developing evidence-based practice ....................................... 147
6.2 Future directions ................................................................................................................ 149
6.2.1 An assessment of population regulation in a naturalized population versus a recent
introduction of an invasive crop pest. .................................................................................. 149
6.2.2 An assessment of population regulation in a recently introduced crop pest with a large
adopted geographic range. ................................................................................................... 150
6.2.3 An evaluation of how environmental parameters are used in phenological and
population models. ............................................................................................................... 150
6.3 Conclusions ........................................................................................................................ 151
Appendix 1: Covariance structure of environmental parameters used in Chapter 3 ...... 183
Appendix 2: Supplemental information regarding pesticide selection ........................... 184
viii
LIST OF TABLES
Table 2-1 Location of sites used in survey of soybean aphid populations and landscape
parameters. Soybean fields were located in Southwestern Ontario, Canada……….…....24
Table 2-2: Parameters used in analysis for predicting aphid populations using landscape
parameters. Aphid presence, colonization and density were treated as dependant variables
and the remaining landscape parameters were used as independent variables in all
models. Means reported are combined averages for the two study years………………..25
Table 2-3: Competing models to explain aphid population measures (presence Ap,
colonization Ac and density Ad) using landscape parameters and the respective Akaike’s
information criterion (AIC) for models tested. A random number generator (0-100) was
used to provide a ‘baseline’ AIC for each aphid population measure and sampling year
and average AIC generated by 50 random models are outlined in dashed boxes. AICs for
the three best ranked single-parameter models for each dataset are outlined by bold
boxes. If the best-ranked model for a given dataset was no more than two units below the
average AIC for the ‘random’ models, the models were not used for further analysis.
Abbreviations used are described in Table 2………………………………….…………28
Table 2-4: Regression coefficients for linear models for aphid density Ad as a function of
field perimeter to area ratio P / A, facing hedgerow to area ratio Hf / A, or estimated
number of buckthorn to field area ratio Bd x Hf / A respectively…………….………….30
Table 3-1: Locations of suction traps in the North American soybean aphid suction trap
network that were included in this study…………………………………….…………..44
Table 3-2: Description of environmental parameters and response variables used in
models…………………………………………………………………………...………47
Table 3-3: AIC values for Zero-inflated Poisson regressions of suction trap captures
versus field populations of A. glycines at different temporal delays between field
population using categorical field infestation and suction trap count of alate aphids.
Julian date was used as a predictor of excess zeros……………………………………..51
Table 3-4: Model selection results for models predicting summer soybean aphid flights
using environmental parameters. Predictors used were raw data and models took the form
Flight event (=0 or 1) = predictor (binomial models), Total aphid captures = predictor
(Poisson models) or Total aphid captures = predictor, with Julian date as a predictor of
excess zeros (zero-inflated Poisson models). For each model, AIC and rank amongst that
model type are given. Models which are statistically significant at α=0.05 are marked
with an asterisk. Statistics that could not be computed are marked with a dash. NA marks
non-applicable models (ie: quadratic models using categorical variables)…………… 53
Table 3-5: Model selection results for models predicting fall soybean aphid flights using
environmental parameters. Predictors used were raw data and models took the form
ix
Flight event (=0 or 1) = predictor (binomial models), Total aphid captures = predictor
(Poisson models) or Total aphid captures = predictor, with Julian date as a predictor of
excess zeros (zero-inflated Poisson models). For each model, AIC and rank amongst that
model type are given. Models which are statistically significant at α=0.05 are marked
with an asterisk. Statistics that could not be computed are marked with a dash…….….54
Table 3-6: Model selection results for models predicting male soybean aphid flights
using environmental parameters. Predictors used were raw data and models took the form
Flight event (=0 or 1) = predictor (binomial models), Total aphid captures = predictor
(Poisson models) or Total aphid captures = predictor, with Julian date as a predictor of
excess zeros (zero-inflated Poisson models). For each model, AIC and rank amongst that
model type are given. Models which are statistically significant at α=0.05 are marked
with an asterisk. Statistics that could not be computed are marked with a dash….……..55
Table 3-7: Model selection results for models predicting summer soybean aphid flights
using residuals from parameter data to account for co-linearity with location, or location
and time. Predictors used were residuals based on linear models taking the form
parameter = (spatial and temporal predictors) and models took the form Flight event =
residual predictor +spatial and temporal predictors (binomial models), Total aphid
captures = residual predictor +spatial and temporal predictors (Poisson models) For each
model, AIC and rank amongst that model type are given. Models which are statistically
significant at α=0.05 are marked with an asterisk. NA marks non-applicable models (ie:
quadratic models using categorical variables)…………………………………………..56
Table 3-8: Model selection results for models predicting fall soybean aphid flights using
residuals from parameter data to account for co-linearity with location, or location and
time. Predictors used were residuals based on linear models taking the form parameter =
(spatial and temporal predictors) and models took the form Flight event = residual
predictor +spatial and temporal predictors (binomial models), Total aphid captures =
residual predictor +spatial and temporal predictors (Poisson models) For each model,
AIC and rank amongst that model type are given. Models which are statistically
significant at α=0.05 are marked with an asterisk……………………………………….57
Table 3-9: Model selection results for models predicting male soybean aphid flights
using residuals from environmental parameter data to account for co-linearity with
location, or location and time. Predictors used were residuals based on linear models
taking the form parameter = (spatial and temporal predictors) and models took the form
Flight event = residual predictor +spatial and temporal predictors (binomial models),
Total aphid captures = residual predictor +spatial and temporal predictors (Poisson
models) For each model, AIC and rank amongst that model type are given. Models
which are statistically significant at α=0.05 are marked with an asterisk……………….58
Table 3-10: Model coefficients and computed optimal values (± standard error SE) of
environmental parameters for predicting aphid flight events and total aphid captures in
summer, fall, and for male aphids. Coefficients for linear (x) and quadratic (x2) are
x
given. Models with non-significant regression coefficients (at α=0.05) are marked NS and
optimum values are not computed……………………………………………….………59
Table 3-11: Model selection results, model coefficients and computed optimal values (±
standard error SE) of environmental parameters for predicting aphid flight events and
total aphid captures for two aphid flight events in fall. Coefficients for linear (x) and
quadratic (x2) are given. Models with statistically significant regression coefficients (at
α=0.05) are marked with an asterisk. Optimum values are not computed for models with
non-significant coefficients. Percent variation (% var, 100SE/mean) is given for the
Poisson models…………………………………………………………….…………….60
Table 4-1: Description of functions governing developmental processes in ‘soybean’
submodel. All processes are continuous (i.e. applied to the relevant life stage at each
time step), unless noted as an establishment process. Establishment processes were
applied to a given cohort once, upon entry into the relevant lifestage. See Fig. 4-2 for a
schematic of the soybean lifecycle ……………………………….……………………..96
Table 4-2: Description of functions governing developmental processes in ‘coccinellid’
submodel. All processes are continuous (i.e. applied to the relevant life stage at each time
step), unless noted as an establishment process. Establishment processes were applied to
a given cohort once, upon entry into the relevant lifestage. See Fig. 4-3 for a schematic of
the coccinellid lifecycle………………………………………………………………….97
Table 4-3: Description of functions governing developmental processes in ‘wasp’
submodel. All processes are continuous (i.e. applied to the relevant life stage at each time
step), unless noted as an establishment process. Establishment processes were applied to
a given cohort once, upon entry into the relevant lifestage. See Fig.4-3 for a schematic of
the wasp lifecycle………………………………………………..……………………….99
Table 4-4: Description of functions governing developmental processes in ‘orius’
submodel. All processes are continuous (i.e. applied to the relevant life stage at each time
step), unless noted as an establishment process. Establishment processes were applied to
a given cohort once, upon entry into the relevant lifestage. See Fig. 4-3 for a schematic of
the orius lifecycle………………………………………………………………………101
Table 4-5: Description of functions governing developmental processes in ‘aphid’
submodel. All processes are continuous (i.e. applied to the relevant life stage at each time
step), unless noted as an establishment process. Establishment processes were applied to
a given cohort once, upon entry into the relevant lifestage. See Fig. 4-3 for a schematic of
the aphid lifecycle. For each life stage, host plant where the aphid originated is given in
brackets. Note that because DymexTM limits the number of stage transfer possibilities for
a given life stage to two, ‘nymph: environmental conditioning (soybean)’ is a dummy life
stage to allow nymphs (soybean) to become apterae (soybean), alates (soybean) or
apterae conditioned to produce sexuals (soybean)……………..………………………103
Table 5-1: Insecticides evaluated for use in control of the soybean aphid…………….128
xi
Table 5-2: Toxicity ratings used to calculate Environmental Impact Quotient for
Beauveria bassiana, which does not have a published EIQ value……….…………….132
Table 5.3: Relative direct contact mortality of natural enemies treated with six
insecticides at field rate…………………………………………………………………133
xii
LIST OF FIGURES
Figure 1-1: Lifecycle of soybean aphid, Aphis glycines, in temperate climates…….........3
Figure 3-1. Captures of A. glycines in suction traps by A) Julian date, and B)
Photoperiod, for 2005-2009 growing seasons at 47 trapping sites in central North
America…………………………………………………………………………………..48
Figure 3-2. Three dimensional scatter plots of log of suction trap captures of A. glycines
vs Julian date and A) Photoperiod and B) Categorical field infestation (for summer trap
captures only). Traps that did not capture any aphids during a given sampling week are
excluded from the figure for clarity……………………………………………...………48
.
Figure 3-3. Fall captures of female and male A. glycines in suction traps by photoperiod.
A smoothing line (LOESS) with a span of 0.3 gives the local weighted regression of the
female aphid captures. Traps that did not capture any aphids during a given sampling
week are excluded from the figure for clarity. Timing of the early activity peak (E) and
late activity peak (L) are marked with arrows………………………………………..….49
Figure 3-4. Density of flying aphids observed in central North America during the week
of August 11, the calculated ‘optimal’ summer flight date, in 2005-2009. Density
surfaces were based on suction trap captures and estimated using ordinary Kriging. Traps
that did not report any data during the sampling week were excluded from Kriging
analysis and are not shown on the figure. IA= Iowa, IL=Illinois, IN=Indiana,
KS=Kansas, KY=Kentucky, MI=Michigan, MN=Minnesota, ON=Ontario,
WI=Wisconsin…………………………………………………………………………...65
Figure 3-5. Density of flying aphids observed in central North America during the week
of September 22, the calculated ‘optimal’ early fall flight date, in 2005-2009. Density
surfaces were based on suction trap captures and estimated using ordinary Kriging. Traps
that did not report any data during the sampling week were excluded from analysis and
are not shown on the figure. See Fig. 3-4 for explanation of abbreviations…………….66
Figure 3-6. Density of flying aphids observed in central North America during the week
of September 29, the calculated ‘optimal’ late fall flight date, in 2005-2009. Density
surfaces were based on suction trap captures and estimated using ordinary Kriging. Traps
that did not report any data during the sampling week were excluded from analysis and
are not shown on the figure. See Fig. 3-4 for explanation of abbreviations……………..67
Figure 3-7. Density of male aphids observed in central North America during the week
of September 29, the calculated ‘optimal’ flight date, in 2008-2009. Density surfaces
were based on suction trap captures and estimated using ordinary Kriging. Traps that did
not report any data during the sampling week were excluded from analysis and are not
shown on the figure. No males were observed in 2005 or 2006, and male populations
were insufficient for Kriging analysis in 2007. See Fig. 3-4 for explanation of
abbreviations…………………………………………………………………..…………68
xiii
Figure 3-8. Using wind trajectories to determine origin of flying aphids during the week
of August 1-8, 2009. Soybean fields at Arva, ON and Manhattan, KS both had abrupt
increases in aphid populations during this week, suggesting the increase may be due to an
immigration event. A) Field aphid density as estimated by Kriging on Aug 1, 2009, over
the study area; B) Field aphid density as estimated by Kriging on Aug 8, 2009, over the
study area; C) Aphid flight density estimated by Kriging for Aug 2-8 and 24-hour
backwards wind trajectories arriving at 400m above Arva and Manhattan at noon on
each day Aug 2-8; D) Aphid flight density estimated by Kriging for Aug 2-8 and 24-hour
backwards wind trajectories arriving at 1200m above Arva and Manhattan at noon on
each day Aug 2-8. Traps that did not report any data during the sampling week were
excluded from Kriging analysis and are not shown on the figure……………….………69
Figure 3-9. Using wind trajectories to determine origin of flying aphids during the week
of August 8-15, 2009. Soybean fields at Arva, ON had an abrupt increase in aphid
populations during this week, suggesting the increase may be due to an immigration
event; but fields at Manhattan, KS, did not have have an abrupt growth in aphid
populations during this week. A) Field aphid density as estimated by Kriging on Aug 8,
2009, over the study area; B) Field aphid density as estimated by Kriging on Aug 15,
2009, over the study area; C) Aphid flight density estimated by Kriging for Aug 9-15
and 24-hour backwards wind trajectories arriving at 400m above Arva and Manhattan at
noon on each day Aug 9-15 ; D) Aphid flight density estimated by Kriging for Aug 9-15
and 24-hour backwards wind trajectories arriving at 1200m above Arva and Manhattan
at noon on each day Aug 9-15. Traps that did not report any data during the sampling
week were excluded from Kriging analysis and are not shown on the figure…….……..70
Figure. 4-1. Schematic of biotic model components. The model consists of five lifecycle
submodels ‘soybean,’ ‘aphid,’ ‘coccinellid,’ ‘orius,’ and ‘wasp’ interacting with each
other through external calculations of soybean aphid density measures and the natural
enemy unit, as well as with environmental conditions…………………………….…….86
Figure. 4-2. Schematic of ‘soybean’ submodel. Stages vulnerable to feeding by soybean
aphid are shaded. Labels below life stages correspond to soybean developmental
stages…………………………...………………………………………………...…..…..87
Figure. 4-3. Schematic of natural enemy submodels. Predatory life stages (shaded) were
used in the computation of Natural Enemy Units (NEUs) acting on aphid populations.
Because adult parasitic wasps are rarely observed in the field, wasp mummies were used
in the computation of observable NEUs, which were used to validate the model with field
data………………………………………………………………………….……………87
Figure. 4-4. Schematic of ‘aphid’ submodel. For clarity, aphid life stages are always
referred to followed by the host on which they originated (in brackets). Note that
‘environmental conditioning’ is a dummy life stage because the number of paths a
lifecycle can take when leaving a given life stage is limited to two by software…….….88
xiv
Figure. 4-5. Model performance at Alvinston site in 2007. A) Predicted (▲) and
observed (○) aphid populations by Julian day; B) predicted vs. observed aphid-per-plant
populations , C) predicted (▲)and observed (○) NEUs by day of year; and D) predicted
vs. observed NEU-per-plant populations . Regression lines were constrained to have a
zero intercept……………………………………………………………………………110
Figure. 4-6. Model performance at Shetland site in 2007. A) Predicted (▲) and observed
(○) aphid populations by Julian day; B) predicted vs. observed aphid-per-plant
populations , C) predicted (▲)and observed (○) NEUs by day of year; and D) predicted
vs. observed NEU-per-plant populations . Regression lines were constrained to have a
zero intercept……………………………………………………………………………111
Figure. 4-7. Model performance at Arva site in 2009. A) Predicted (▲) and observed (○)
aphid populations by Julian day; B) predicted vs. observed aphid-per-plant populations,
C) predicted (▲)and observed (○) NEUs by day of year; and D) predicted vs. observed
NEU-per-plant populations. Regression lines were constrained to have a zero
intercept............................................................................................................................112
Figure. 4-8. Abundance of soybean aphid morphs by date as predicted by the model. The
model was initiated on 5 January, using 2007 weather data from an Environment Canada
weather station near London, ON, with 1000 ‘spring eggs’, 500 soybean plants planted on
20 May, and ‘background’ natural enemies (as described in text). All aphid life stages are
given on this figure except for nymphs occurring on buckthorn and soybean. Winged
morphs are given in red. Arrow indicates location of possible second peak of gynoparae
activity………………………………….……………………………………………….114
Figure. 4-9. Aphid density (in vulnerable aphids per plant) over the growing season for
three soybean planting dates, two natural enemy treatments, and weather data from two
different growing seasons, as predicted by the model. Each panel consists of predicted
aphid population densities for the planting dates 6 May, 20 May and 4 June, for A) 2007
weather and high NEUs (background NEUs, as described in the text, increased by an
order of magnitude); B) 2007 weather and low NEUs (background NEUs only); C) 2009
weather and high NEUs; and D) 2009 weather and low NEUs. Weather data for 2007 and
2009 were obtained from an Environment Canada weather station near London,
ON……………………………………………………………………………………....115
Figure. 4-10. Abundance of diapause eggs of soybean aphid on December 27 (end of
simulation) as a function of planting date, natural enemy abundance and growing season,
as predicted by model. Each panel consists of predicted aphid diapause egg abundances
for the planting dates 6 May, 20 May and 4 June at low NEUs (i.e. background NEUs, as
described in the text) and high NEUs (i.e. background NEUs increased by an order of
magnitude) after a given growing season. A) 2007; B) 2009. Weather data for 2007 and
2009 were obtained from an Environment Canada weather station near London,
ON………………………………………………………………………………...…….116
xv
Figure 5-1. a) Observed field efficacy of six insecticides for soybean aphid control.
Aphid count data were Henderson-Tilton adjusted(Henderson and Tilton 1955) and
subjected to a mixed model ANOVA by post-treatment sampling period with year of
experiment, block, pass of tractor, site, and interaction terms between block and pass,
block and site, and pass and site incorporated into the model.
b) Observed field selectivity of six insecticides for aphid control. Field selectivity was
determined using the natural enemy-to-aphid ratio in treatment plots, for exact calculation
see Materials and Methods. Observed efficacy and selectivity within sampling period
marked by the same letter are not significantly different at α=0.05 (LSD)………….....135
Figure 5-2: Least-square mean yield in fields treated with six insecticides. Data were
subjected to a mixed model ANOVA with block, site, treatment incorporated into the
model. Observed yields marked by the same letter are not significantly different at
α=0.05 (LSD)…………………………………………………………………………..136
Figure 5-3. Relationship between observed field selectivity and the inverse of
Environmental Impact Quotient at field rates. Field selectivities presented as least square
means (± SE) of field selectivities observed at four sites in 2009. Equation of regression
line is Field selectivity = (3.3±1.7)/EIQ+(0.3±3.1)+site effect, with F93 = 4.23, p =
0.0035…………………………………………………………………………………..137
xvi
CHAPTER 1
Soybean aphid in North America: an introduction and review
1.0 Introduction
Invasive arthropods can dramatically impact the productivity of agricultural
systems. Following the introduction an establishment of an invasive pest in a new
geographic region, a large amount of research activity is required to gain understanding
of behavior, ecology, and phenology of the pest in its new range. Soybean aphid (Aphis
glycines Matsumura)(Hemiptera: Aphididae) is one such pest: in the decade since its
introduction to North America in 2000 (Ragsdale et al. 2004), this species has been the
subject of study by numerous research groups, and a vast array of literature is now
available documenting aspects of the biology of aphids in North America. Several major
reviews of A. glycines biology and ecology have been performed over the intervening
years (i.e.:Ragsdale et al. 2004, Wu et al. 2004, Heimpel et al. 2010, Ragsdale et al. 2011,
Tilmon et al. 2011). Thus, the purpose of this review is not to summarize existing
literature, but to identify and address existing gaps in the understanding of A. glycines
biology in North America in order to develop management recommendations based on
available data and constraints.
1.1 Background
Soybean aphid is a severe pest of cultivated soybean (Glycines max (L.) Merr.) in
North America (Ragsdale et al. 2004). Native to eastern Asia, A. glycines has since
spread through much of central North America (Ragsdale et al. 2004). A. glycines can
1
reduce yield in soybeans by feeding and they vector diseases such as soybean mosaic
virus (Wu et al. 2004). A. glycines first arrived in Ontario and Quebec in 2001 (Hunt et
al. 2003). At the outset of this study in 2007, management of this species consisted of
broad spectrum insecticides applied before A. glycines reached economically damaging
levels (OMAFRA 2005).
In its native range in China, A. glycines reaches economically damaging
populations only sporadically, and only rarely requires chemical control in soybean (Liu
et al. 2004). An abundance of natural enemies is credited with keeping A. glycines below
economic thresholds in this part of the world (Liu et al. 2004). In North America,
outbreak populations of A. glycines are frequent. Since its initial outbreak in North
America, it has been observed that A. glycines undergoes a cyclical abundance:
widespread outbreak populations have been observed in odd-numbered years (2001,
2003, 2005, and 2007) and only sporadic outbreaks in even numbered years (Bahlai 2007,
Welsman 2007, Bahlai and Sears 2009, Rhainds et al. 2010c). Outbreaks in even
numbered-years are usually restricted to the north-western portion of the North
American range (i.e.: Iowa, Wisconsin and Minnesota) (USDA 2011). Mechanisms
underlying this cyclical abundance are not well understood, but may be driven by natural
enemy abundance, weather, crop phenology and management practices, or, most likely, a
combination of these factors (Ragsdale et al. 2004, Bahlai and Sears 2009, Heimpel et al.
2010).
In addition to being of economic concern in North America, soybean aphid is an
invasive species with a very complicated lifecycle, involving both parthenogenic and
sexual reproduction and host alternation (Ragsdale et al. 2004), making the species an
2
interesting case study in terms of invasion biology and climate-phenology interactions
(Fig 1-1).
In temperate climates, A. glycines has an obligate association with buckthorns,
woody shrubs in the genus Rhamnus, and a few closely related species within the
Rhamnaceae. These shrubs act as the primary host, that is, where mating and
overwintering occur. In spring, eggs of A. glycines undergo temperature-dependent
development, hatch, and, after several generations, emigrate from overwintering hosts to
soybean (Bahlai et al. 2007, Welsman et al. 2007). Towards the end of the soybean
Spring
Winter
Rhamnus spp.
Glycines spp.
Fall
Summer
Figure 1-1: Lifecycle of soybean aphid, Aphis glycines, in temperate climates.
3
growing season, environmental cues cause aphid populations to produce gynoparae and
males, and these morphs migrate to buckthorn (Voegtlin et al. 2005). Gynoparae, upon
arrival at overwintering hosts, produce oviparae, which mate with the males, and then
deposit eggs at the bud-twig interface on the shrubs.
Soybean aphid is part of a multi-trophic level invasive species system: in North
America, the aphid resides on non-native crops (soybean), and overwinters on a nonnative, invasive shrub, common buckthorn (Rhamnus cathartica L.) (Ragsdale et al.
2004), and its primary predators are native to Europe and Asia (Majerus and Kearns
1989). A recent review by Heimpel et al. (2010) explored the extensive ecological
implications of this ‘invasional meltdown.’ Much of the success in establishment of A.
glycines can be attributed to the previous establishment of R. cathartica in North
America. Interestingly, R. cathartica is not present within soybean growing regions in
Asia; there A. glycines overwinters on several other species of buckthorn, but
predominantly on Rhamnus davurica Pall. (Wu et al. 2004). Though A. glycines is able to
overwinter on a number of buckthorn species, aphids occurring on R. cathartica have
greater longevity and ovipositional rates than those observed on other suitable hosts (Yoo
et al. 2005). R. cathartica is widespread through much of the northern portion of North
America’s soybean growing region, and occurs most commonly, and in high abundance,
in agricultural hedgerows and adjacent woodlots (Knight et al. 2007, Kurylo et al. 2007,
McCay and McCay 2009). Overwintering hosts in Asia are neither as abundant nor as
widespread as R. cathartica is observed to be within the North American soybean
growing region (Heimpel et al. 2010).
4
Natural enemies are important regulators of populations of A. glycines (Wang et
al. 1991, Rongcai et al. 1994, Van Den Berg et al. 1997, Fox et al. 2004, Liu et al. 2004,
Rutledge et al. 2004, Fox et al. 2005, Nielsen and Hajek 2005, Costamagna and Landis
2006, Desneux et al. 2006, Mignault et al. 2006, Brosius et al. 2007, Butler and O'Neil
2007b, a, Costamagna and Landis 2007, Costamagna et al. 2007a, Donaldson and Gratton
2007, Donaldson et al. 2007, Kaiser et al. 2007, Miao et al. 2007, Costamagna et al.
2008, Noma and Brewer 2008, Bahlai and Sears 2009, Xue et al. 2009, Zhang et al.
2009b, Frewin et al. 2010, Hallett et al. In prep). A large portion of the scientific
literature on A. glycines involves biological control of this species. Conservation
biocontrol, that is ‘modification of the environment or existing practices to protect and
enhance specific natural enemies or other organisms to reduce the effect of pests’
(Eilenberg et al. 2001) is the primary focus of much of this work. Though much of the
resident natural enemy community acting on A. glycines consists of introduced species,
most of these species became established prior to or concurrently with the aphid, and
were either introduced as part of classical biological control programs for other species
(e.g. multicoloured Asian ladybeetle Harmonia axyridis (Koch 2003)) or are of unknown
origin (e.g. parasitic wasp Aphelinus certus (Frewin et al. 2010)). Some researchers have
pursued classical biological control of A. glycines, and several species of parasitic wasps
have been released in the Midwestern United States, with varying degrees of biological
control success (Heimpel et al. 2004, Wyckhuys et al. 2009). However, classical
biological control programs are rare, in practice, because of high costs, unpredictable
efficacy rates, and potential non-target impacts of releasing non-native organisms into an
5
ecosystem (Coll 2009), thus, this avenue for control of A. glycines has not been pursued
in Canada.
Though conservation biocontrol may be impaired by intraguild predation when
natural enemy communities are diverse (Straub et al. 2008), in most systems, including
the A. glycines natural enemy complex, this has been shown to have minimal effect on
overall biocontrol efficacy (Straub et al. 2008, Xue et al. 2012). Conservation biological
control is an important regulator of pest populations, yet studies that can directly show
yield benefits or quantify the ecosystem services provided by biocontrol agents are rare
(Jonsson et al. 2008).
1.1.1 Spring migrations of A. glycines and patterns in soybean field colonization
Ragsdale et al. (2004) first remarked on an apparent phenological disjunction
between A. glycines and soybean. They observed that, in spring, migratory morphs
(alates) are often observed on buckthorn shrubs several weeks prior to the planting of
soybean in the vicinity, and A. glycines could not be detected at all on buckthorn by the
time soybean had emerged. A subsequent study by Welsman et al. (2007), however, was
able to detect small numbers of A. glycines on buckthorn within a week of its first
occurrence on soybean. However, neither study was able to detect edge effects
associated with the direct movement of A .glycines from buckthorn hedgerows to
adjacently planted soybean, and, indeed, at the outset of this project, no relationship
between overwintering host density and risk of colonization within a given soybean field
had been documented (Heimpel et al. 2010). If summer hosts are available in the vicinity
of overwintering sites, it seems an unnecessary risk for aphids to attempt longer-distance
6
migrations. However, the observations of Ragsdale et al. (2004) and Welsman et al.
(2007) suggest that, though R. cathartica is instrumental to the success of A. glycines in
North America, it is possible that the effect of the spatial distribution of R. cathartica on
that of the aphid is minimal at a local scale. Indeed, if A. glycines populations on
buckthorn produce alates for several weeks prior to the planting of soybean, it is possible
that spring migrants could travel some distance from overwintering sites before settling
on a suitable summer host, which could obscure patterns associated with those migrants
occurring on buckthorn later, and moving to more locally-planted soybean. Thus, in order
to detect patterns in colonization of soybean fields by A. glycines, existing data should be
examined with analytical methods that are robust to multiple sources of variation.
Though numerous studies (e.g. Bahlai et al. 2007, Welsman et al. 2007) of
soybean aphid interactions with environmental factors have been completed in both
North America and Asia, most of the previously published literature relies on nullhypothesis significance testing (NHST). Bayesian and likelihood statistical methods
focus on model specification and relative fit, and can be used to approach primary data in
ways that incorporate previous work; many biological statisticians provide impassioned
arguments on why these methods should routinely be used rather than NHST (see Taper
and Lele (2004) for a compilation of these arguments). These methods are becoming
more commonly used in the biological sciences, however, agricultural ecology largely
remains a field where NHST is championed over other methods: the vast majority of
studies present their results accompanied by F, df and p-values, and frame their
discussion around what is significant or not, based on a decision rule championed by
convention. In situations where variation (noise) can come from multiple unknown or
7
un-quantified sources such as the case with colonization patterns of A. glycines, it is
unlikely statistically significant results documenting these patterns will ever be obtained
without the accumulation of very large datasets (Anderson et al. 2000).
Information theory is a branch of likelihood statistics where the remaining
information (that is, unexplained noise in a dataset after a model is applied) is quantified
(Akaike 1974). Akaike (1974) introduced An Information Criterion (known colloquially
as Akaike’s Information Criterion or simply AIC), as a method to rank multiple
competing models by their relative degree of model fit. AIC is a unitless statistic
computed using the likelihood ratio or residual sum of squares (as a measure of model fit)
and the number of parameters used in the model (to aid in the selection of a parsimonious
model with the fewest necessary parameters) (Burnham and Anderson 2002). AIC values
are meaningless in isolation: they are a statistic used to rank performance of competing
models, and a smaller AIC value indicates a model fits the available data better than a
model producing a larger AIC (Akaike 1974).
The objective of Chapter 2 of this dissertation is to use information-theoretic
model selection methods to elucidate patterns in colonization of soybean fields by A.
glycines. The hypothesis of this study is that colonization patterns related to landscape
parameters can be detected using a model selection approach, where a NHST approach
has failed to detect these relationships.
1.1.2 Migration and dispersal by A. glycines in summer and fall
Aphids will migrate for two primary reasons: to move between overwintering and
summer hosts in spring and fall; and to move between summer hosts when conditions
8
become unfavourable in the currently exploited host patch (Moran 1992). Summer
migrations of A. glycines are extremely important factors in the population dynamics of
this species in soybean (Rhainds et al. 2010b). On soybean, A. glycines population
distributions are often extremely patchy, particularly early in the growing season, and
that, coupled with high rates of population growth, can lead to non-uniform resource
exploitation by the aphids (Onstad et al. 2005). Genetic evidence suggests that soybean
fields are usually colonized by a few closely related, and probably locally occurring, A.
glycines clonal lines early in the growing season. Genetic diversity then increases over
the growing season, and a general west-to-east pattern of increasing diversity is observed,
suggesting that A. glycines is undergoing long-distance flights in this direction over its
range (Michel et al. 2009). As fall approaches, gynoparae (aphid morphs that produce
oviparae) and male aphids are produced, and these morphs migrate to overwintering
hosts. Gynoparae and viviparous alates of A. glycines cannot be morphologically
distinguished (Voegtlin et al. 2004b), but these morphs differ in behavior and ecology
and are induced by different environmental cues (Bonnemaison 1951, Moran 1992).
Despite obvious advantages of being able to move between hosts, migration is
both risky and costly for aphids. Though the apparent risk of mortality for insects flying
at higher altitudes is quite low (around 2%) (Taylor 1960), it is likely that, because of the
mixing effects of wind, only a small proportion of aphids taking flight in this way will
land on a suitable host plant and ultimately go on to reproduce, with as little as one per
thousand surviving (Taylor 1977). Winged aphids that survive migration are typically
much less fecund and have delayed reproduction as compared to their apterous
conspecifics (Zera and Denno 1997). A study on A. glycines found that aphids which had
9
engaged in longer flights were significantly less fecund than those that had only engaged
in short flights, were significantly shorter-lived, and had offspring with lower fecundity
as well (Zhang et al. 2009a). It has been argued that long distance flights are exceptional
because it is advantageous for an aphid to find suitable hosts in the shortest amount of
time (Loxdale et al. 1993), but in the case of A. glycines, longer distance flights may be
an advantage to escape intraspecific competition because of widespread infestations in a
given area. Indeed, it has been estimated that fewer than 5% of aphids emigrate from
soybean when population densities are below 4000 aphids per plant (Donaldson et al.
2007), suggesting that when large emigration events of A. glycines occur, the local area is
already well exploited, and aphids will have to travel a longer distance to escape
competition from conspecifics. In laboratory experiments, Lu and Chen (1993) found that
crowding of apterae increased production of alates in the subsequent generation of A.
glycines, that this effect was more pronounced on older (presumably less nutritious)
leaves, and that higher constant temperatures had an inhibitory effect on alate production.
Subsequent work found that maturity of the secondary host only slightly impacted on the
phenology of A. glycines, suggesting that plant cues, at least from the secondary host,
play a minimal role in dictating the timing of lifecycle events for this species, though
earlier-maturing varieties of soybean may impede some production of gynoparae
(Rhainds et al. 2010a).
A network of suction traps suited to sample migrating and dispersing aphid
populations has been in place in central North America since 2001 (Schmidt et al. 2012),
and our collaborators within southwest Ontario have been participating in this study since
2003. Each week, samples of insects caught by these traps are collected, separated, and
10
all aphids captured are identified to species. Suction trap captures provide a snapshot of
the abundance of migratory aphids present in an area in a given week. These data
provide a unique opportunity to examine the environmental cues which govern alatoid
morph formation and flight behaviors of A. glycines in North America, as well as to
examine spatial patterns in morph occurrence.
The objective of Chapter 3 of this dissertation is to compare abundance of A.
glycines morphs sampled by these traps to weather data collected from nearby weather
stations in order to elucidate patterns in aphid migration and morph occurrence in North
America. As in the colonization study, AICs will be used to rank multiple competing
models. Two hypotheses will be tested in this study: 1) Differing environmental cues are
involved in the induction of alate viviparous versus gynoparous morphs of A. glycines
and thus these morphs can be distinguished from each other by their occurrence relative
to these cues; and 2) A. glycines alates move in a west-to-east direction when migrating.
1.1.3 Integrating the literature: the need for a full-lifecycle phenological model for
Aphis glycines
A considerable body of research examining aspects of the population ecology and
phenology of A. glycines exists in the scientific literature, and thus, the chief challenge in
understanding biology of this species is the sheer number of factors that can be shown to
influence its biology. In addition to studies remarking on general aspects of seasonality
of A. glycines (e.g.:Inoue 1981, Zhang and Zhong 1982, Hirano 1996, Ragsdale et al.
2004, Wu et al. 2004), numerous studies have examined particular aspects of the aphid’s
ecology under specific sets of conditions, often unique to that particular study or research
group. Researchers have examined behaviour, phenology, rate of population growth and
11
survivorship of soybean aphid on winter and summer hosts with respect to environmental
conditions (Ito 1953, Chen et al. 1984, Tian et al. 1990, Hirano et al. 1996, McCornack et
al. 2004, Venette and Ragsdale 2004, Onstad et al. 2005, Bahlai et al. 2007, Welsman et
al. 2007, Zhang et al. 2008, Zhang et al. 2009a) and environmental control of morph
production (Lu and Chen 1993, Hodgson et al. 2005). Biotic influences on populations
of A. glycines are also well-studied; previous work has characterized the communities of
natural enemies on A. glycines populations, and quantified impact of natural enemies on
aphid biology and population growth (Wang et al. 1991, Rongcai et al. 1994, Van Den
Berg et al. 1997, Fox et al. 2004, Liu et al. 2004, Rutledge et al. 2004, Fox et al. 2005,
Nielsen and Hajek 2005, Costamagna and Landis 2006, Desneux et al. 2006, Mignault et
al. 2006, Brosius et al. 2007, Butler and O'Neil 2007b, a, Costamagna and Landis 2007,
Costamagna et al. 2007a, Donaldson and Gratton 2007, Donaldson et al. 2007, Kaiser et
al. 2007, Miao et al. 2007, Costamagna et al. 2008, Noma and Brewer 2008, Bahlai and
Sears 2009, Xue et al. 2009, Zhang et al. 2009b, Frewin et al. 2010, Hallett et al. In prep),
and the effect of host quality, species, cultivar and phenology on A. glycines behavior and
population dynamics has been examined (Chung et al. 1980, Hill et al. 2004, Yoo et al.
2005, Rutledge and O'Neil 2006, Costamagna et al. 2007b, Diaz-Montano et al. 2007,
Hesler and Dashiell 2007, Clark et al. 2009, Rhainds et al. 2010a).
It is thus warranted to use the existing literature to develop a whole-lifecycle
population model for A. glycines integrating environmental cues, natural enemies, and
hosts. Abiotic and biotic factors affecting A. glycines populations are both numerous and
interacting, and yet most of the previous literature only focuses on a few factors within a
given study. The population dynamics of A. glycines are both top-down and bottom-up
12
regulated (Costamagna and Landis 2006, Costamagna et al. 2007b). Oviposition rates
(Rutledge et al. 2004) and voracity (Xue et al. 2009, Frewin et al. 2010) amongst natural
enemy species are density-dependent, and all these parameters interact with
environmental conditions. An understanding of these interactions will require the
implementation of a series of complex, interactive models.
Dymex (Hearne Scientific Software, Chicago IL) is a software package which
allows for the development of deterministic population models based on available
biological data (CSIRO 2007). This package can be used to develop mechanistic models
to predict population dynamics and phenology, as well as to help identify areas of the
biology and ecology of an organism which require further study.
The objective of Chapter 4 of this dissertation is to develop a tritrophic population
and phenology model incorporating A. glycines, a phenology model for soybean, and
phenology and population models for two predatory insects and one parasitic wasp. This
model will be used to explain population dynamics in the field and to test hypotheses
about factors governing population dynamics of A. glycines.
1.1.4 Responding to economically damaging outbreaks of A. glycines
Heavy infestations of A. glycines can significantly decrease the yield of soybean
crops (Ragsdale et al. 2004). Management strategies are in place to mitigate these yield
losses, but each of these strategies is associated with ecosystem impacts. Choosing when
and which insecticidal agents should be used in response to population outbreaks of A.
glycines has been a subject of considerable interest in the literature (e.g. Ragsdale et al.
2007, Johnson et al. 2008, Kraiss and Cullen 2008b, a, Johnson et al. 2009, Myers et al.
13
2009, Ohnesorg et al. 2009, Zhang and Swinton 2009). The goal of most of these studies
was to develop cost-effective, environmentally-sound management responses which
minimize impact on natural enemies, as natural enemies are often able to provide
effective regulation of A. glycines populations without cost to growers or the
environment. Some studies also tested methods of control that could be used in organic
production systems (Kraiss and Cullen 2008b, a). Most of this previous research was
performed in the United States, and because of differing regulatory and agronomic
practices, it is essential that similar work be performed in Canada, taking these factors
into account. At the outset of this study in 2007, the government-recommended action
threshold for applying insecticides to control A. glycines was 250 aphids per plant and
increasing, and only two insecticides, λ- cyhalothrin and dimethoate were registered for
use (OMAFRA 2005). Additionally, the registration of dimethoate was nearing review,
and it was possible that the registration for this product would not be renewed in Canada
(Health Canada Pest Management Agency 2009). Our group recently completed a study
where we developed a practical dynamic action threshold which dramatically reduces
pesticide use by incorporating the impact of the resident natural enemy community on
aphid population growth into management decisions (Hallett et al. In review). In addition
to optimizing timing of management, it is also essential to evaluate several novel
products, including those which could be used in organic cropping systems, in order to
develop the most sustainable management practices possible.
In order to evaluate the efficacy and environmental impact of novel insecticidal
products thoroughly and effectively, multiple approaches to evaluation are required.
Measuring impact of insecticides on natural enemy communities is a key component to
14
pesticide risk assessment. Natural enemies are likely to be at high risk for exposure, as
they are often present when insecticides are applied, and are more likely than other
organisms to be susceptible, as many important natural enemy species, like the targeted
pest species, are arthropods. Additionally, if natural enemy communities are adversely
affected by a pesticide, the ability of these insects to control remaining populations of
pest insects will be impaired. Thus, environmental assessment of pesticide impact should
start with examination of the impact of these products on natural enemies, in both
laboratory and field settings.
A variety of methods exist for estimating the net ecosystem impact a pesticide has
on the environment (Levitan et al. 1995). Kovach et al. (1992) present a methodology for
calculating Environmental Impact Quotients (EIQ). EIQs provide a relative measure of
ecological impacts of pesticides. These values incorporate ratings of dermal toxicity
(DT), chronic toxicity (C), systemicity (SY), fish toxicity (F), leaching potential (L),
surface loss potential (R), bird toxicity (D), soil half-life (S), bee toxicity (Z), beneficial
arthropod toxicity (B), and plant-surface half life (P). Calculation of the EIQ for a
specific pesticide takes the form:
In the case of missing values, the average rating of the other variables is
substituted for that variable. Many commonly used pest control products have EIQ
parameters available in the scientific literature (Kovach et al. 2009). Specific EIQ values
are based on equivalent weights of active ingredient, and must be transformed to EIQ
15
field use ratings (EIQFURs) before they can be used to calculate the relative
environmental impacts of various pesticides (Kovach et al. 1992). EIQFURs are obtained
by multiplying the percent active ingredient (%AI), the application rate, and the number
of applications, using the following calculation:
Though the bulk of this dissertation focuses on developing an understanding of
the biology of soybean aphid outbreaks, Chapter 5 evaluates some potential chemical
management responses. The objective of this chapter is to evaluate the efficacy and
potential impact of four novel insecticides (two synthetic products and two naturally
derived products that could be approved for use in organic cropping systems) and
develop recommendations for the use of these products. 1
1.2 General objectives
An understanding of the physiological ecology of migration, host selection and
location by A. glycines, and dynamics that occur with natural enemies, hosts and the
environment at each point in the aphid lifecycle is not only essential to A. glycines
management; such information will also contribute to the overall body of knowledge of
aphid physiology, ecology and behavior.
The aim of this project is to gain a greater understanding of factors controlling
the phenology of A. glycines, to develop models for predicting A. glycines abundance and
distribution incorporating relevant abiotic and biotic factors, and to assess environmental
viability of management strategies in the context of this knowledge. Integrated pest
1
This dissertation will focus on Canadian management recommendations, particularly those within the
province of Ontario, but efforts will be made to generalize conclusions to the entire North American range
of A. glycines where possible.
16
management strategies incorporating A. glycines ecology and novel products are required
for an environmentally and economically sustainable approach to controlling this pest.
Factors controlling the population ecology and phenology of A. glycinesc are complex:
attention must be given to climate and micro-climate inputs (temperature, rainfall,
humidity), geographic and biogeographic factors (wind and the proximity of
overwintering hosts), biological factors (predators, parasitoids), and human inputs
(management strategies for A. glycines and other pests, agronomic practices, cultivar
choice). The impacts of each of these factors at all stages in the A. glycines life cycle
have the potential to affect abundance of this species and thus should be accounted for.
My work will serve to verify and expand upon previous work to predict future
infestations of A. glycines, assess environmental and economic costs associated with
existing and experimental management strategies, and develop recommendations for
soybean growers based on this research.
There are two central research questions in this project:
1. What causes outbreaks of soybean aphid?
To answer this, impacts of naturally occurring biocontrol agents (predators &
parasitoids) on A. glycines populations and occurrence of damage must be
determined , and the role of environmental factors (temperature, day length) in
the abundance and phenology of A. glycines (including factors affecting alate
production and migratory behavior) examined.
2. What can be done to manage outbreaks of soybean aphid to greatest
economic advantage and with least environmental impact?
17
This question can be answered by evaluating the effect of conventional
management strategies and management strategies incorporating reduced risk
chemical and bio-pesticides on aphid and natural enemy abundance, by
calculating Environmental Impact Quotients for conventional and experimental
A. glycines management strategies, and evaluating the efficacy of each of these
methods.
18
CHAPTER 2
Modeling distribution and abundance of soybean aphid in
soybean fields using measurements from the surrounding
landscape2
2.0 Abstract
Soybean aphid (Aphis glycines Matsumura) is a severe pest of soybean in central
North America. Outbreaks of the aphid in Ontario are often spotty in distribution, with
some geographical areas affected severely and others with few or no aphid populations
occurring in soybean for the duration of the season.
A. glycines spend summers on soybean and overwinter on buckthorn, a shrub that
is widespread in southern Ontario, and is commonly found in agricultural hedgerows and
at the margins of woodlots. A. glycines likely use both short distance migratory flights
from buckthorn and longer distance dispersal flights in the search for acceptable summer
hosts. This study aims to model colonization of soybean fields by A. glycines engaged in
early-season migration from overwintering hosts. Akaike’s Information Criterion (AIC)
was used to rank numerous competing linear and probit models using field parameters to
predict aphid presence, colonization, and density. The variable which best modeled
aphid density in soybean fields in the early season was the ratio of buckthorn density to
field area, though a dramatic differences in relationships between the parameters were
2
Originally published as Bahlai, C.A., S. Sikkema, R.H. Hallett, J. Newman and A.W. Schaafsma. 2010.
Modeling distribution and abundance of soybean aphid in soybean fields using measurements from the
surrounding landscape. Environ. Entomol. 39: 50-56. Author contributions: CB conducted the analyses,
interpreted the data and wrote the paper, SS collected data, conducted initial analyses; JN suggested
statistical methods, edited the MS; RH interpreted the data, edited the MS, AS conceived the study, edited
the MS.
19
observed between study years. This study has important applications in predicting areas
that are at elevated risk of developing economically damaging populations of soybean
aphid and which may act as sources for further infestation.
KEYWORDS: Aphis glycines, Rhamnus cathartica, hedgerow, model selection
2.1 Introduction
Soybean aphid (Aphis glycines Matsumura) is an invasive pest of cultivated
soybean (Glycine max (L.) Merr.) in North America (Ragsdale et al. 2004). Originally
occurring in soybean growing regions throughout Asia (Wu et al. 2004), Aphis glycines
was first documented in North America in 2000 (Alleman et al. 2002). Since its initial
colonization, the aphid frequently reaches economically damaging levels in soybean
fields in the midwest and Great Lakes regions of the continent (Hunt et al. 2003,
Ragsdale et al. 2004, Venette and Ragsdale 2004). Population dynamics of A. glycines in
a given soybean field have been difficult to predict. Ragdsale et al. (2004) first remarked
upon an apparent two-year cycle in outbreak populations of A. glycines: widespread
economic outbreaks were observed in 2001 and 2003, but in 2002 very few North
American soybean fields were colonized by A. glycines, and in those that were, the aphid
did not reach high densities. In southwestern Ontario, where our study was performed,
this general trend has continued to date (Bahlai 2007, Welsman 2007). However,
agricultural extension personnel have remarked upon exceptions to this ‘odd year’ rule:
localized aphid outbreaks are often observed in small geographical areas during ‘even’
(low) years, with even-year outbreaks sometimes only affecting a single soybean field (T.
Baute, personal communication).
20
Aphis glycines is a heteroecious aphid which overwinters on woody hosts, most
typically on buckthorn shrubs of the genus Rhamnus and several other closely related
shrubs (Voegtlin et al. 2004a, Voegtlin et al. 2005). R. cathartica L., common buckthorn,
is considered the most important overwintering host of A. glycines in North America
because of its wide distribution and its high density in many soybean growing regions
(Voegtlin et al. 2005). R. cathartica thrives in habitats which are frequently disturbed and
offer intermediate light levels (Kurylo et al. 2007), and is often observed in margins of
woodlots and in agricultural hedgerows (Bahlai et al. 2008, Welsman et al. 2007).
Egg hatch of A. glycines is temperature dependant and usually occurs between the
middle and end of April in southwestern Ontario (Bahlai et al. 2007). After hatching, A.
glycines undergo several parthenogenic generations on R. cathartica before a generation
of winged (alate) aphids are produced (Liu et al. 2004). Apterous A. glycines can be
found on buckthorn until the first week of June, at which time alates occur and numbers
sharply decline on the overwintering host (Welsman et al. 2007). In ‘odd’ (high) years,
this sharp decline corresponds with the initial detection of small numbers of A. glycines
in soybean fields by the second and third weeks of June (Bahlai 2007, Welsman et al.
2007). In ‘even’ (low) years, A. glycines is seldom detected in soybean fields before the
middle of July.
Aphis glycines likely uses both short distance migratory flights from buckthorn
and longer distance dispersal flights aided by weather patterns in the search for
acceptable summer hosts (Zhang et al. 2008). Observed colonization patterns within
soybean fields are very patchy and not apparently correlated with hedgerows in early
summer: winged adults likely move within the field, feeding and depositing nymphs at
21
multiple locations (Ragsdale et al. 2004). If fields are colonized later in the summer,
however, higher populations of A. glycines are usually observed near windbreaks
(Ragsdale et al. 2004). Given that they act as both windbreaks and overwintering host
habitats, agricultural hedgerows and associated landscape parameters likely play an
important role in the colonization of soybean fields by A. glycines. Yet, the ecological
mechanisms of this association have not been directly elucidated. Many variables affect
population dynamics of A. glycines. The effects of natural enemies, weather, plant health,
and migratory populations from other locales can be difficult to quantify, especially when
aphid populations are low; but nonetheless, these factors impact the number of A.
glycines individuals observed in a given soybean field. Without directly quantifying these
variables, detecting patterns in aphid colonization associated with other parameters (such
as landscape variables) can be challenging. Analyses relying on null-hypothesis
significance tests may not yield a statistically significant result in such variable datasets,
and thus, existing scientifically significant patterns may be ignored (Taper and Lele
2004). In systems like these, it is desirable, instead, to approach the analysis of data using
likelihood statistics to rank models relative to each other, rather than relative to an
arbitrary significance level (Taper 2004). An (Akaike’s) information criterion (AIC) is a
statistical tool which employs the likelihood function and allows the performance of large
sets of competing models to be compared relative to each other simultaneously (Akaike
1974). This study develops and compares models demonstrating associations between
colonization and density of mid-summer aphid populations in soybean fields and various
landscape parameters. This information will be used to identify fields that are most likely
22
to be colonized by A. glycines in ‘low’ (even) years, and fields that may act as sources for
further infestation in ‘high’ (odd) years.
2.2 Materials and Methods
2.2.1 Study site and landscape parameters
Twenty-five soybean fields in an area spanning approximately 125 km (east-west)
by 70 km (north-south) in the southwestern Ontario soybean growing region were
selected in 2005 (Table 2-1). The study site included fields with very low and very high
buckthorn densities in the vicinity and a variety of field shapes and areas (ranges of all
measured parameters are provided in Table 2-2). At each site, the following landscape
parameters were measured: buckthorn density B, field area A, field perimeter P, length of
field perimeter with facing hedgerow Hf, and facing hedgerow within a 4 km radius of
the sampling site in the field H4. Buckthorn density was determined by counting the
number of shrubs immediately adjacent to the sampling site in a span of 80 m of
hedgerow. P, Hf and H4 were measured by walking the spans and using a handheld GPS
unit to measure distance traveled. Field area was calculated using field perimeter
measurements.
2.2.2 Aphid monitoring
During the week of July 15 in 2005, all 25 sites were scouted for aphid
populations. Single plants were sampled at sites 5, 25, 45, 65 and 85 m into the field
along four transects spaced 20 m from each other and originating at a field edge adjacent
23
Table 2-1 Location of sites used in survey of soybean aphid populations and landscape parameters.
Soybean fields were located in Southwestern Ontario, Canada.
o
o
Site #
N
W
2005 sampling
2006 sampling
43.107
81.325
Yes
1
43.082
81.294
Yes
Yes
2
43.112
81.270
Yes
Yes
3
43.080
81.248
Yes
4
43.066
81.268
Yes
Yes
5
43.106
81.297
Yes
Yes
6
43.089
81.287
Yes
7
43.039
81.347
Yes
8
42.929
81.498
Yes
Yes
9
43.283
80.863
Yes
Yes
10
43.324
80.911
Yes
11
43.358
80.798
Yes
12
43.340
80.794
Yes
Yes
13
43.345
80.911
Yes
Yes
14
42.844
81.886
Yes
Yes
15
42.868
81.876
Yes
Yes
16
42.865
81.839
Yes
Yes
17
42.868
81.856
Yes
18
42.977
81.974
Yes
Yes
19
42.762
81.902
Yes
20
43.274
80.791
Yes
Yes
21
43.281
80.624
Yes
22
43.131
80.837
Yes
Yes
23
43.151
80.658
Yes
24
42.868
81.847
Yes
Yes
25
to a hedgerow, for a total of 20 plants per field. Each plant was destructively sampled,
and all A. glycines on sampled plants were counted. In 2006, 15 of the fields were either
re-planted to soybean or had a field immediately adjacent to the 2005 sites planted to
soybean, and these fields were sampled during the week of July 15. Sampling procedures
were similar to that used in 2005 except at each sampling point, three plants were
destructively sampled, for a total of 60 plants per field.
24
Table 2-2: Parameters used in analysis for predicting aphid populations using landscape parameters. Aphid
presence, colonization and density were treated as dependant variables and the remaining landscape
parameters were used as independent variables in all models. Means reported are combined averages for the
two study years.
Observed range
Variable
Abbreviation
Measurement
Possible values
(mean +/- SD)
Aphid presence
Ap
At least one aphid observed
in field
0, 1
0–1
(0.80± 0.41)
Aphid colonization
Ac
A. glycines observed on at
least 2 plants in field
0, 1
0–1
(0.58±0.50)
Aphid density
Ad
Total number of A. glycines
observed in a field divided
by total number of plants
sampled
Any value ≥0
0 – 26.5 /plant
(4±8 /plant)
Buckthorn presence
Bp
0, 1
0–1
(0.65±0.48)
Buckthorn density
Bd
At least one buckthorn
shrub observed adjacent to
field
Total number of buckthorn
shrubs observed in 80m of
hedgerow, divided
by 80
Any value ≥0
0 - 0.33 plants /m
(0.07±0.10 /m)
Field perimeter
P
Total length of perimeter
around field
Any value >0
796 – 2815 m
(1777±545 m)
Field area
A
Total field area
Any value >0
28338 – 293995 m2
(141560±76737 m2)
(14.1±7.7 ha)
Facing hedgerow
Hf
Total length of field
perimeter composed of
hedgerow or woodlot
Any value ≥0
105 – 1435 m
(489±374 m)
Hedgerow within 4
km
H4
Total length of hedgerow
within 4km of sampling site
Any value ≥0
2968 – 8126 m
(5077±1536 m)
Estimated number
of buckthorn shrubs
facing field
Estimated number
of buckthorn shrubs
within 4km
Perimeter to area
ratio
Facing hedgerow to
area ratio
Bd x Hf
-
Any value ≥0
0 – 280
(40±71)
Bd x H4
-
Any value ≥0
0 – 2438
(373±545)
P/A
-
Any value ≥0
Hf / A
-
Any value ≥0
0.0078 - 0.0280 /m
(0.015±0.005 /m)
0.0008 - 0.0095 /m
(0.004±0.002 /m)
Hedgerow within 4
km to area ratio
H4 /A
-
Any value ≥0
0.0191 - 0.1206 /m
(0.05±0.03/m)
Estimated number
of buckthorn shrubs
facing field to area
ratio
Estimated number
of buckthorn shrubs
within 4km to area
ratio
(Bd x Hf) / A
-
Any value ≥0
0 - 0.0019 /m2
(0.0004±0.0006 /m2)
(Bd x H4) / A
-
Any value ≥0
0 - 0.0199 /m2
(0.004±0.006 /m2)
25
2.2.3 Models
Data were analyzed separately by year. Parameters used in the models are
described in Table 2. Models tested are listed in Table 3. Probit and linear models were
used to describe aphid presence, colonization and density using landscape parameters.
Aphid presence was defined as at least one aphid observed during sampling of a field.
Aphid colonization was defined any number of aphids observed on at least two of the
sampled plants in a field. All collected data were used to test aphid presence and
colonization models. Data were filtered to include only fields where aphid colonization
had occurred to test models for aphid density (15 fields in 2005, 8 in 2006). In addition
to landscape parameters, a random number generator (range 0 to 100) was used to create
fifty dummy independent variable datasets to model aphid presence, colonization and
density, so these models could be used as a point of comparison.
Probit models are best used for bivariate responses (i.e.: presence/absence), so this
function was used to model aphid presence and colonization. Probit analyses were
performed in SAS v. 9.1 (SAS Institute, Cary, NC) using the probit link of PROC
LOGISTIC. Linear functions were used to model aphid density. These analyses were
performed in SAS using PROC MIXED.
In SAS, AIC is calculated by default and provided in the output of both PROC
LOGISTIC and PROC MIXED. Models with the best performance were identified using
the minimum AIC estimation method (MAICE) (Akaike 1974). Models with no more
than two units difference between their calculated AICs are considered to be equivalent
in performance (Burnham and Anderson 2002). Single-parameter models were ranked by
their respective AICs, and for each year and each aphid population measure, the single-
26
parameter models with the three lowest AICs were selected for further analyses, provided
these models also out-performed the random number generator (i.e.: had an AIC more
than two units less than the average AIC of the random data models). The selected
parameters were squared and were combined by adding them together to generate new,
more complex models, and the performance of these models were evaluated as above and
ranked using the AICs generated.
2.3 Results
2.3.1 Aphid presence and colonization
Aphid presence (at least one aphid observed in the field) was observed in 80% of
fields surveyed in both 2005 and 2006 (20/25 and 12/15, respectively). Aphid
colonization (aphids ovserved on at least two sampled plants) was observed in 60% of
fields surveys in 2005 (15/25 fields) and 53% of the fields sampled in 2006 (8/15 fields).
Average aphid density in fields where aphid colonization was observed was 11.6 ± 9.6
aphids per plant in 2005 and 0.41 ± 0.38 aphids per plant in 2006. Average area of fields
sampled was similar in both sample years (14.5± 7.8 ha in 2005, 13.6± 7.6 ha in 2006), as
was field perimeter (1783±554 m in 2005, 1766± 549 m in 2006).
Many of the models tested provided improvements over random for explaining
aphid colonization, that is, generated an AIC more than two units less than the random
number models (Table 2-3). None of the models could explain aphid presence better than
a random number generator, thus, the remainder of this section will focus upon results
pertaining to aphid colonization. Models favored by our data varied between years of
27
Table 2-3: Competing models to explain aphid population measures (presence Ap, colonization Ac and
density Ad) using landscape parameters and the respective Akaike’s information criterion (AIC) for models
tested. A random number generator (0-100) was used to provide a ‘baseline’ AIC for each aphid population
measure and sampling year and average AIC generated by 50 random models are outlined in dashed boxes.
AICs for the three best ranked single-parameter models for each dataset are outlined by bold boxes. If the
best-ranked model for a given dataset was no more than two units below the average AIC for the ‘random’
models, the models were not used for further analysis. Abbreviations used are described in Table 2.
Akaike’s Information Criterion (AIC)
A
Ac
Ad
Model
p
2005
2006
2005
2006
2005
2006
27.9
17.8
36.5
23.9
95.3
10.3
Average AIC (50 random number models)
Landscape parameters
28.7
18.0
37.6
21.9
Bp
28.7
19.0
32.8
23.4
91.5
7.5
Bd
29.0
19.0
37.6
22.3
109.0
23.8
P
28.9
18.9
37.6
22.7
118.3
34.9
A
28.9
19.0
37.6
23.5
110.1
24.0
Hf
27.0
18.7
35.0
22.8
113.3
26.1
H4
27.5
19.0
31.9
23.8
103.7
15.9
Bd x H f
28.8
19.0
31.6
23.1
107.8
22.3
Bd x H4
28.8
19.0
37.3
24.6
85.2
2.5
P/A
28.9
19.0
37.5
24.3
85.8
1.3
Hf / A
29.0
18.7
37.6
24.5
87.0
5.4
H4 /A
28.7
18.8
35.2
24.6
82.5
-1.4
Bd x H f / A
28.9
18.7
34.9
24.3
86.4
2.9
Bd x H4/ A
Additive linear models
Bp + (Bd x Hf)
Bp + (Bd x H4)
(Bd x Hf) + (Bd x H4)
(P / A) + (Hf / A)
26.6
29.5
27.7
-
19.9
20.0
20.9
-
29.2
22.3
33.4
-
23.6
23.3
25.1
-
69.7
(P / A) + (Bd x Hf / A)
-
-
-
-
66.5
(Hf / A) + (Bd x Hf / A)
-
-
-
-
66.4
(Hf / A) + (Bd x Hf / A) + (P / A)
-
-
-
-
50.7
28.7
28.6
28.1
-
18.0
19.4
21.0
-
37.6
32.9
22.3
-
21.9
25.3
21.4
-
78.8
76.4
-
-
-
-
68.5
-8.7
12.5
12.8
23.6
Higher order polynomial models
Bp 2
(Bd x Hf) 2
(Bd x H4) 2
(P / A) 2
(Hf / A) 2
(Bd x Hf / A) 2
28
-4.2
-7.6
13.8
study. In 2005, the ‘high’ aphid year in our period of study, the best single parameter to
explain aphid colonization was estimated buckthorn within 4km (Bd x H4).
In 2006, the ‘low’ aphid year in our study, colonization by A. glycines in soybean
fields was best explained by the simple presence or absence of buckthorn (Bp), followed
by field perimeter (P) and field area (A). A marginal improvement over the buckthorn
presence model was observed by using a higher order polynomial model taking into
account the length of hedgerow within 4 km of the study site and estimated number of
buckthorn within 4 km.
2.3.2 Aphid density
Models for aphid density were ranked similarly and consistently by datasets from
both years. Generally, models incorporating field area as a normalization factor had
improved performance over models which did not. The best single parameter for
predicting aphid density was the estimated number of buckthorn shrubs facing the field
per unit area of field (Bd x Hf / A), but the field perimeter to area ratio (P / A) and
hedgerow length to area ratio (Hf / A) were also high-ranked parameters. The best linear
model tested was the additive model including all three of these factors. Models
comprised of the square of each of these terms also outranked the linear models. Models
involving higher order polynomials (to x7) were tested, but are not reported here because
although these models offered improvements in fit according to their respective AICs,
these observed ‘improvements’ were likely a result of overfitting. Although the three
models which best predicted aphid density were consistently ranked between the two
study years (Table 2-3), the relationship (i.e.: the sign of the observed regression
29
Table 2-4: Regression coefficients for linear models for aphid density Ad
as a function of field perimeter to area ratio P / A, facing hedgerow to
area ratio Hf / A, or estimated number of buckthorn to field area ratio Bd
x Hf / A respectively.
Parameter
Year
Slope
Intercept
P/A
2005
2006
-608 ± 298
-18 ± 25
20 ± 4
0.6 ± 0.4
Hf / A
2005
2006
-923 ± 814
29 ± 51
14 ± 4
0.2 ± 0.2
Bd x H f / A
2005
2006
-4668 ± 4284
64 ± 231
11 ± 2
0.3 ± 0.2
coefficients) changed between the two study years for the two top-ranked parameters Bd
x Hf / A and Hf / A (Table 2-4). In 2005, a negative correlation was observed between
aphid density and both Bd x Hf / A and Hf / A, while in 2006, a positive correlation was
observed between aphid density and these parameters.
2.4 Discussion
2.4.1 Statistical methods
Though consistent improvement in model AICs were observed by increasing the
complexity of aphid density models, this result is likely of little biological relevance.
Firstly, it has been argued that AIC has a tendency to favor models which over-fit data,
despite its inclusion of a penalty term for overly complex models (Taper 2004).
Secondly, this study was designed to determine how landscape parameters affect aphid
distribution and not to elucidate the exact models of interaction. Populations of A.
glycines are affected by numerous additional parameters, many of which also likely
interact with landscape, thus it is probable that greater improvements in models would be
30
observed by directly accounting for these factors, rather than landscape parameter-only
models of increasing complexity. Finally, since the parameters used in many of the
multiple-parameter models are auto-correlated, model parsimony may be affected (Zuur
et al. 2007).
2.4.2 Factors affecting aphid distribution
Aphid presence and colonization measures can be used to gain insight into which
soybean fields are most likely to develop economically damaging aphid infestations later
in the season. These two parameters were defined differently and modeled separately to
determine which could be more reliably modeled. AIC values for aphid presence models
did not vary more than two units from the average of the random models, whereas aphid
colonization models produced AICs which indicated larger differences between models
and more significant improvements over the random models. Aphid colonization is likely
a more reliable measure than aphid presence with which to predict subsequent aphid
infestation.
The best predictor of Ac, colonization by A. glycines in a given field, in a low
aphid year is the presence or absence of overwintering hosts. In high aphid years, this
relationship is not observed, possibly due to increased dispersal flights in response to
high aphid densities early in the season, or interactions with natural enemies. Even still,
an association between aphid colonization and the estimated number of buckthorn within
4km of study site was observed in high years. Interestingly, aphid density Ad was shown
to be negatively correlated with increasing density of buckthorn or increasing hedgerow
face per unit field area in our ‘high’ aphid year. This effect may be due to interactions
31
between our measured landscape parameters and natural enemy populations, and patterns
of aphid dispersal.
A. glycines relies upon buckthorn in hedgerow habitats for overwintering, but
these same habitats may favor colonization and population growth of aphid natural
enemies. In general, habitats with greater diversity such as agricultural landscapes with
abundant hedgerows favor natural enemy populations because they provide alternate food
sources, shelter, and varied microclimates (Landis et al. 2000). Aphids occurring in
wheat fields adjacent to hedgerows are more effectively controlled by Coccinella
septempunctata (Bianchi and van der Werf 2003). Similarly, important soybean aphid
predators, particularly Harmonia axyridis, are more abundant in habitats with more forest
and hedgerow habitat (Gardiner et al. 2009a, b). Thus, if abundances of both natural
enemies and A. glycines are associated with similar landscapes, the dynamics between
these species in hedgerows have the potential to dominate over landscape effects in
determining aphid density. The degree of natural enemy impact on aphid population
density varies from year to year, further confounding the patterns emerging between
aphid density and landscape. Landis et al. (2008) quantified the biocontrol service
rendered by natural enemies in soybean fields 2005 and 2006 in a study area
geographically adjacent to our study site, and found there was a 12-fold decrease in
natural enemy impact on soybean aphid in 2006 as compared to 2005. If larger numbers
of A. glycines were present on buckthorn in the hedgerows in the spring of 2005 in our
study region, these populations could support greater abundances of natural enemies in
the hedgerows. Even if fewer natural enemies were present at the beginning of the year,
they would have an opportunity to proliferate while feeding on relatively abundant aphids
32
in hedgerows, potentially causing localized depletions of aphids. Colonization success of
A. glycines would be improved for individuals dispersing farther from overwintering sites
with high densities of natural enemies, rather than those which colonize adjacent
soybean. This may help to explain the negative correlations observed in 2005 between
aphid densities and the increasing density of buckthorn or increasing hedgerow face per
unit field area; natural enemies likely exhibited numerical responses to higher densities of
overwintering aphid populations in hedgerow habitats in the spring. In 2006, when
overwintering aphid populations were less abundant (Welsman et al. 2007), it is likely
that natural enemies did not have the same opportunity to aggregate in the hedgerows in
spring. Aphids colonizing soybean fields directly adjacent to overwintering sites would
be less impacted by predation, and thus a positive correlation between aphid density and
the two parameters is observed.
This variable and confounding effect of natural enemies on aphid density has
likely contributed to the lack of clear association between hedgerows, overwintering
hosts, and aphid density remarked upon in previous studies. However, it is possible that
natural enemies cannot account for this switch between positive and negative correlations
observed between the two study years. When population densities are high on
overwintering hosts, even in the absence of higher densities of natural enemies, migrating
alates of A. glycines might be triggered by population cues to fly greater distances from
the overwintering site when searching for summer hosts to colonize.
Ragsdale et al. (2004) suggested that the July colonization of Ontario soybean
fields is due to the movement of A. glycines from other soybean fields, likely from some
distance away, rather than direct movement from overwintering hosts. If this was always
33
the case, we would expect to find either that early summer aphid colonization was not
related to presence of buckthorn, or that aphid colonization was strongly related to the
presence of hedgerows alone, due to windbreak effects. Although these two measures are
not mutually exclusive (hedgerows act as habitats for buckthorn, thus an area with more
hedgerow has the potential to have more buckthorn at it), our study found that aphid
density was better modeled by buckthorn density than hedgerow-length. This result
suggests several possibilities: 1) in their northern range, small populations of A. glycines
may remain undetected on overwintering hosts longer into the summer than in southerly
regions; 2) small populations of A. glycines occur undetected in soybean near to
overwintering hosts, until they reach detection limits in mid-July; or 3) A. glycines
arriving in dispersal flights from other geographic areas are more likely to select fields
with suitable overwintering habitat nearby for colonization.
While the first two possibilities seem the most likely, the latter possibility
warrants further investigation. Soybean aphid oviparae occurring on buckthorn in autumn
produce a sex pheromone to attract males to their location (Zhu et al. 2006a). Alate
soybean aphids respond to soybean-produced volatiles (Zhu and Park 2005), and thus it is
likely that buckthorn-produced volatiles could influence their behaviour as well.
Preferential selection of soybean fields with overwintering hosts nearby could be a result
of semiochemical cues associated with buckthorn or as a result of signaling from
conspecifics already colonizing these fields. Several other species of aphid are known to
employ aggregation pheromones to help maintain populations at moderate densities; it is
thought that these aggregations help aphid populations to dilute individual risk of
predation or parasitism (Wertheim et al. 2005).
34
Using landscape parameters to identify fields at greatest risk of becoming
colonized by A. glycines and developing high aphid densities is a promising method to
improve scouting efficiency within a given geographic area. Landscape parameters would
not change dramatically from year to year, and thus could be used to identify fields where
scouting efforts should be focused. Fields identified as high risk may also be ideal
candidates for prophylactic use of neonicotinoid-treated soybean seed. Though these seed
treatments usually do not provide protection into late summer when aphid populations
tend to reach economically damaging levels (Johnson et al. 2008), seed treatments could
suppress early season aphid population growth and prevent these first-colonized fields
from acting as source habitats for movement of aphids into surrounding areas.
The most important landscape factor in determining if a field is likely to be
colonized by A. glycines in a low aphid year is the simple presence of buckthorn, the
overwintering host. Thus, the ongoing attempts to eradicate common buckthorn from
agricultural hedgerows and woodlots (Pergams and Norton 2006, Delanoy and Archibold
2007) can be supported on the basis of their expected beneficial impact on soybean aphid
populations. R. cathartica is a widely-distributed invasive species in North America
(Kurylo et al. 2007). Numerous ecosystem and agroecosystem impacts are associated
with the invasion of this shrub (Delanoy and Archibold 2007, Knight et al. 2007). If
buckthorn was locally eradicated, a field previously at ‘high risk’ for soybean aphid
colonization would become lower risk, and the need for insecticide applications targeted
against soybean aphid may be reduced or eliminated.
35
2.5 Acknowledgements
The authors would like to thank Andrew Welsman for his assistance, Doug
Landis for his helpful comments on a presentation of this work, and Mark Sears and
Angela Gradish for their comments on this manuscript. We also acknowledge the
financial support of the University of Guelph- Ontario Ministry of Agriculture Food and
Rural Affairs Sustainable Production Program, the Ontario Soybean Growers,
Agriculture and Agri-Food Canada, the Natural Science and Engineering Research
Council of Canada and the Keefer Family Trust.
36
CHAPTER 3
Factors inducing migratory forms of soybean aphid and an
examination of North American spatial dynamics of this
species in the context of migratory behaviour.3
3.0 Abstract
Soybean aphid (Aphis glycines) is a severe invasive pest of North American
soybeans. Since 2005 a suction trap network spanning central North America has
monitored aerial populations of this species. Captures of female A. glycines in summer
and fall, and males in fall were subjected to binomial, Poisson, and Zero-Inflated Poisson
regressions with corresponding environmental parameters such as photoperiod and
weather variables, and then models were ranked by AIC to determine which parameters
best explained occurrence and density of these aphid groups. Best-fit models were used
to compute ‘optimal’ values of each parameter, when relevant. Aerial aphid populations
at key points in their lifecycle were interpolated by kriging trap captures over the study
area, and the response surface was visualized using GIS software. Backwards wind
trajectories were computed for two sites over two sampling weeks in summer 2009 to see
if these trajectories would coincide with apparent movements and colonization events by
flying aphids. In summer, occurrence and density of flying A. glycines was best explained
by the level of aphid infestation in fields local to the trapping site. In fall, an apparent
bimodal distribution in aphid flight activity was detected. An early fall peak in activity
3
Manuscript in preparation for Ecological Entomology. Full author list is yet to be determined as more
than 10 research groups contributed data to the suction trap network ; but CB will serve as lead author for
compiling environmental data, analysis and writing the MS.
37
was best explained by degree day accumulations, but a late fall peak was best explained
by photoperiod, and the latter peak coincided with the peak in male activity. Kriging
interpolation indicated widely variable aphid flight activity among years and locations in
both summer and fall. Wind trajectories coincided well with A. glycines population
increases in fields, but did not reliably predict immigration events. Suction traps are
useful for determining when emigration of A. glycines occurs from an area, and wind
trajectories can explain the direction of aphid movement, but caution must be taken in
using this methodology for forecasting immigration events.
KEYWORDS: migration, Akaike information criterion, model selection, kriging,
trajectory
3.1 Introduction
Soybean aphid (Aphis glycines Matsumura) is a severe invasive pest of North
American soybean (Ragsdale et al. 2004). A. glycines is a heteroecious, holocyclic aphid,
that is, it reproduces sexually once per year, and this sexual reproduction occurs on its
primary (or overwintering) host (Ragsdale et al. 2004). This life history pattern is
common throughout the range of A. glycines (Zhang and Zhong 1982). In North America,
overwintering hosts are woody shrubs in the family Rhamnaceae, especially on Rhamnus
cathartica L., common buckthorn. The overwintering range of soybean aphid in North
America is thus geographically limited to places where R. cathartica occurs, which is
primarily north of the 41st parallel (Voegtlin et al. 2004a, Voegtlin et al. 2005, Welsman
et al. 2007). Alatoid A. glyines occur at three points in the aphid’s lifecycle: as spring
colonists moving from the overwintering host to soybean; between soybean fields in the
summer; and as sexual morphs migrating back to overwintering hosts in autumn
38
(Ragsdale et al. 2004, Wu et al. 2004). The development of alate A. glycines in spring is
temperature- and density-dependent, and early summer soybean aphid infestations in
soybean are related to the local presence and density of R. cathartica in the northern
portion of their range (Chen et al. 1984, Bahlai et al. 2007, Bahlai et al. 2010a). Flights
from the overwintering host to colonize the summer host are thought to be relatively local
and short in duration (Schmidt et al. 2012). Nutritional cues from host plants are also
presumed to impact the timing of spring migrations from R. cathartica, (Welsman et al.
2007). Many host-alternating aphid species produce alates in response to decreasing
nitrogen content of the maturing leaves of their primary host (Harrison 1980).
Migration plays an important role in the population dynamics of this species. Like
other economically important species of aphid (e.g.: green peach aphid Myzus persicae
(Zhu et al. 2006b) ) clonal lines of A. glycines are not limited to the soybean fields they
initially colonize after migrating from the primary host. Progeny of the colonists may also
go on to disperse great distances from their overwintering range, making it difficult to
predict economically damaging infestations (Michel et al. 2009). Emigration events and
associated mortalities can lead to population crashes during the growing season for
several aphid species, including A. glycines (Karley et al. 2004, Rhainds et al. 2010a).
Aphis glycines is well-known for its large migratory populations. In years when
the aphid is abundant, this species can dominate the complex of aphid alates captured in a
variety of crops other than soybean (Nault et al. 2004). Genotypic diversity of aphids
sampled from soybean fields increases throughout the growing season, suggesting that
early season colonists are relatively closely related and likely come from nearby sources,
while later-season immigrants are less related and likely originate from more distant
39
sources (Michel et al. 2009). Under optimal conditions, alate viviparous A. glycines can
fly up to 16 km at a speed of up to 3 km/h, in the absence of wind (Zhang et al. 2008).
However these data cannot be extrapolated to in-field behaviour of viviparous alates nor
to migratory spring and fall morphs because of the increased complexity of natural
conditions and the differences in morph physiology and behaviour. Environmental
conditions responsible for inducing the production of sexual morphs of A. glycines in
autumn have not been determined, but in Indiana, sexual morphs are typically first
observed moving to the primary host around 15 September, i.e. at a photoperiod of
approx. 12.5 h. (Rhainds et al. 2010a).
Examining environmental cues associated with triggering alate and sexual morphs
and flight events of A. glycines is crucial to understanding and predicting economic
outbreaks of this species, but the factors governing the initiation of flight physiology and
behaviour are complex and interacting. Factors governing aphid polymorphism4 can vary
widely, even in closely related species (Lambers 1966). Maternal morph and the cues a
mother is exposed to typically govern the morph of the offspring, with short or declining
photoperiod governing the production of sexual forms by virginoparae, crowding of
apterous virginoparae leading to alate offspring, and alate mothers typically only
producing apterous daughters (Mousseau and Dingle 1991). However, maternal effects
do not govern all morph determination in aphid populations. Michaud (2001) found that
alate development in brown citrus aphid, Toxoptera citricida (Kirkaldy) was determined
by the population density experienced by nymphs, and that nymphs were not irreversibly
4
Some authors (e.g.: Mousseau and Dingle 1991) draw distinctions between polymorphism, which they
define as varied morphologies between varied genotypes, and polyphenisms, as variations in morphologies
within a single clonal line. Since this paper deals with the production of various morphs at the population
level, the term polymorphism is used throughout.
40
committed to becoming alates until the third instar. Similarly, Hardie (1981) found that
apterization of young nymphs induced by long days could be reversed by exposing
nymphs to short photoperiods prior to their fourth day of life.
Many environmental and population cues affect aphid development and
phenology. Additionally, virginoparous and sexuparous morphs within a single species
often do not display consistent responses to cues (Taylor 1977, Loxdale et al. 1993).
Zhou et al (1995) found that temperature was a better predictor than latitude and
longitude for captures of migratory aphids in suction traps for five aphid species in Great
Britain, although the periods during which temperature had the greatest effect on
phenology varied by species. Worner et al. (1995) used host plant phenology and
weather to predict spring phenology of damson-hop aphid Phorodon humuli and found
that the performance of models incorporating different parameters varied by site.
Similarly, regional forecasting using date-based phenological models to predict flights of
two species of cereal aphid (Sitobion avenae and Metopolophium dirhodum) was limited
by site-specific effects (Clark et al. 1992). The production of sexual forms of pea aphid
(Acyrthosiphon pisum) was triggered by declining photoperiod, and the production of
males required shorter photoperiods than production of female sexual morphs (Via 1992).
Bonnemaison (1951) found that population density and that plant cues were most
important in the formation of virginoparous alates in four species, and photoperiod ,
moderated by host plant cues, induced aphid populations to produce sexuals, but
thresholds of response varied among species. Since many factors affect aphid phenology
and it is likely that different cues govern the timing of different aspects of the A. glycines
lifecycle, it is important to examine the relative importance of these factors at various
41
times of the year. Other aphids are known to travel great distances in migratory flight.
Arriving alates of A. pisum were estimated to have traveled 300km (Smith and MacKay
1989), and in some cases, aphids are known to survive wind-mitigated flights exceeding
1300 km (Elton 1925).
Although it is not known what altitude A. glycines typically flies, long distance
flights of M. persicae are associated with low-level jet streams (600-1800m above sea
level) (Zhu et al. 2006b), though wind-trajectories at the planetary boundary layer (300500m) were used to track migrations of A. pisum and these were well-corroborated with
the photoperiodic responses of these aphids (Smith and MacKay 1989). Genetic work
suggests that migrating A. glycines in North America generally follow a west to east
trajectory, as Michigan and Ontario populations differ significantly in diversity
compared to populations in the western portion of its range (Michel et al. 2009). In
laboratory studies, alate viviparous A. glycines had optimal flight performance with
regards to distance and duration at 12-24 hours after eclosion, and preferred temperatures
of 16 to 28oC and relative humidity of 45 to 90% for flying (Zhang et al. 2008). Alate A.
glycines flew faster at higher temperatures, but had longer flight durations at 24oC (Zhang
et al. 2008).
Suction traps are commonly used to monitor aerial insect populations, particularly
aphids, as a means of providing pest warnings to the agricultural industry (Woiwod et al.
1984). This study utilizes existing data from a suction trap network designed to monitor
the aerial A. glycines in the midwestern United States (North Central Regional Soybean
Aphid Suction Trap Network http://www.ncipmc.org/traps/). Previous work using this
data set found suction traps had potential as monitoring tools, but their application was
42
limited by the spatial variability of aphid populations (Rhainds et al. 2010b). More
recently, Schmidt et al. (2012) provided a comprehensive summary of the spatial trends
observed in aphid trap captures on a per sampling week, per year basis for captures from
this network from 2005-2008. However, this study did not incorporate environmental or
field infestation data, nor did it infer how these data might affect aphid phenology.
In this paper we examine aspects of soybean aphid phenology in North America
pertaining to the factors affecting the production of winged aphids, both the summer
asexual forms and fall sexual morphs. We also explore the implications these data have
for understanding the biology of this aphid throughout its range by examining dynamics
of this species in central North America relative to environmental cues.
3.2 Methods
Forty seven suction traps were established throughout central North America
(Table 3-1; trap locations), in a study area spanning approximately 1400 km west to east,
and 1200 km north to south. The majority of traps in the network were based on a design
by Allison and Pike (1988) and captures from these traps were also used in Rhainds et al.
(2010b) and Schmidt et al (2012). A detailed description of the suction trap network in
2005-2008 can be obtained in Schmidt et al. (2012). These traps sample air 6.7 m above
ground and sample air at a rate of approximately 570 m3/h. The five traps situated in
43
Table 3-1: Locations of suction traps in the North American soybean aphid suction trap
network that were included in this study
Region
IN
MN
IA
WI
WI
ON
MI
SD
IL
MO
MN
IL
IL
IN
MI
ON
IL
WI
ON
ON
MI
MN
WI
KY
KS
IA
IL
IL
MI
IL
MN
IA
IN
IL
WI
IN
MO
IN
KY
ON
MN
IN
WI
IA
IL
Site
ACRE
Albert Lea
Ames
Antigo
Arlington
Arva
Bean & Beet
Brookings
Brownstown
Columbia
Crookston
Dekalb
Dixon spring
DPAC
East Lansing
Elora
Freeport
Hancock
Harrow
Huron
Kellogg
Lamberton
Lancaster
Lexington
Manhattan
McNay
Metamora
Monmouth
Monroe
Morris-IL
Morris-MN
Nashua
NEPAC
Perry
Pioneer
Pit
Portageville
PPAC
Princeton
Ridgetown
Rosemount
SEPAC
Seymour
Sutherland
Urbana
Lat
40.47
43.76
42.02
45.14
43.30
43.08
43.30
44.31
38.95
38.94
47.81
41.84
37.43
40.25
42.72
43.64
42.28
44.11
42.03
43.32
42.32
44.24
42.83
38.09
39.14
40.97
40.77
40.94
41.92
41.37
45.59
42.93
41.10
39.81
44.76
40.45
36.41
41.44
37.10
42.45
44.72
39.04
44.52
42.92
40.10
Long
-86.99
-93.20
-93.77
-89.15
-89.33
-81.20
-84.13
-96.78
-88.96
-92.32
-96.48
-88.85
-88.67
-85.15
-84.46
-80.41
-89.70
-89.54
-82.90
-81.50
-85.38
-95.32
-90.79
-84.53
-96.64
-93.42
-89.34
-90.72
-83.39
-88.43
-95.90
-92.57
-85.40
-90.82
-91.56
-86.93
-89.70
-86.93
-87.86
-81.88
-93.10
-85.53
-88.33
-95.54
-88.23
44
Ontario employed a slightly different trap design and sampling occurred at 7.6 m above
ground level with an estimated air sampling rate of 702 m3/h. Traps in Ontario consisted
of a 90 cm tall PVC pipe with a 20 cm exterior diameter, suspended on a recycled freestanding steel television tower fitted with a sliding track mechanism so the device could
be lowered to collect samples. Data used in this study consisted of the total number of A.
glycines female alates and males counted weekly from individual suction traps between
May to October, 2005 - 2009. The final dataset had 210 location years and 3839 unique
observations; individual observations with missing data were excluded from relevant
analyses. Similar traps used in Europe are thought to be representative of aerial aphid
populations within a radius of > 100 km, but ≤500 km, of the trap because of the
randomization effects of wind (Cocu et al. 2005, Shortall et al. 2009).
Field infestation levels corresponding to suction trap counts were obtained from
the USDA PIPE website (http://tinyurl.com/7po3azb). Aphid infestations are reported on
a categorical, per plant count basis, where 0= no aphids observed, 1= >0-5 aphids per
plant, 2= 6-39 aphids, 3=40-149 aphids, 4=150-249 aphids, 5=250-499 aphids and 6=
≥500 aphids per plant. Data were selected for inclusion in this study if a field was
monitored weekly and was within 50km of a suction trap. If multiple fields were
monitored within 50 km of a trap, only the field with highest aphid count was included.
Weather data (absolute maximum, absolute minimum, and average temperature
and total precipitation over the sampling period,) was obtained from Environment
Canada National Climate Archive (http://www.climate.weatheroffice.gc.ca/), Weather
Underground (http://www.wunderground.com/history/), and National Weather Service
(http://www.weather.gov/climate/). Stations were within 100 km of the sampling site.
45
Day length was calculated for each site on the sampling date (Forsythe et al. 1995). A
summary of environmental parameters and response variables collected is given in Table
3-2. The covariance structure between environmental variables is given in Appendix 1.
3.2.1 Analysis
Initial data exploration (Fig.3-1, 3-2) revealed two well-resolved peaks in aphid
flight activity separated by a period of minimum activity at a photoperiod of approx. 13
h, so data were subdivided into two sets, ‘summer,’ defined as sampling dates with a
corresponding photoperiod ≥13h, and ‘fall’ with corresponding photoperiods <13h. In
most of the study region, 13h photoperiod occurred during the first week of September.
Male aphids were only observed in the fall, and their capture data were analysed
separately.
3.2.2 Determination of migration status of captured aphids
In order to determine whether trap captures represented aphids immigrating to or
emigrating from the trap vicinity, counts of alates captured in suction traps from the
summer dataset were modeled using soybean field infestation from one and two weeks
before and after the suction trap sample was collected. If suction trap observations
represent immigrants, then field infestation after suction trap collection should be a better
predictor of suction samples; if suction trap captures represent emigrants leaving a given
area, field infestation prior to capture should be a better predictor. AIC (Akaike’s
Information Criterion) (Burnham and Anderson 2002) was used to rank zero-inflatedPoisson (ZIP) models (Jackman 2012) (using Julian date as a predictor of excess zeros)
which best predicted suction trap samples. Field categorical data taken the week the
sample was taken, field categorical data from one week and two weeks before, and
46
Table 3-2: Description of environmental parameters and response variables used in models
Parameter
Description
Possible values
Flight event
The presence of aphids in a trap
0, 1
Suction trap captures
Number of female soybean aphids observed in a
trap
Number of male soybean aphids observed in a
trap
Degree day accumulation from July 1, in units
o
C*days, using 10oC as base temperature
Any integer ≥ 0
Categorical field infestation level based on a 0-6
scale from highest reporting field within 50 km
of trap site on date of sampling.
Day of year, based on date suction trap was
collected
Latitude at suction trap location, in decimal
degrees
Longitude at suction trap location, in decimal
degrees
Maximum temperature observed at the nearest
reporting weather station within 50 km during
seven days preceding sample collection, in oC
Mean temperature observed at the nearest
reporting weather station within 50 km for the
seven days preceding sample collection, in oC
Minimum temperature observed at the nearest
reporting weather station within 50 km during
seven days preceding sample collection, in oC
Day length at trapping site on day of sampling
Integers 0 to 6
Total precipitation, in mm, observed at the
nearest reporting weather station within 50 km
for the seven days preceding sample collection
Any value ≥ 0
Male captures
Degree day accumulation
Field Infestation
Julian date
Latitude
Longitude
Maximum temperature
Mean temperature
Minimum temperature
Photoperiod
Total precipitation
Any integer ≥ 0
Any value ≥ 0
Observed range
(Mean ± SD)
0 to 1
(0.5±0.5)
0 to 18280
(67±496)
0 to 548
(0.4±9.8)
0 to 1864
(674±370)
0 to 6
(1.8±1.7)
Integers 0 to 366
o
Any value -180 to
+180o
Any value -180o to
+180o
Any value ≥ -273oC
132 to 308
(230±41)
36.4 to 47.8
(41.9±2.3)
-96.8 to -80.4
(-89.1±4.1)
6 to 52
(28.4±4.9)
Any value ≥ -273oC
1 to 30
(19.0±4.9)
Any value ≥ -273oC
-14 to 22
(9.1 ±5.4)
0-24 h
9.9 to 15.8
(13.3±1.4)
0 to 455
(17.7±25.8)
from one week and two week after. All statistical analyses were completed using R
2.13.0 (R Project for Statistical Computing, http://www.r-project.org/).
3.2.3 Response to environmental variables
For each of the summer, fall and male datasets, the relationship between
environmental variables and alate aphid counts was modeled using either binomial
models (i.e. whether alates were observed or not observed in a given sampling period) or
Poisson and ZIP models (for counts of total alates observed in a given sampling period)
(Zuur et al. 2009, O'Hara and Kotze 2010). Because multiple candidate predictors were
used to create models for most analyses, an information-theoretic model-selection
approach, specifically AIC-based parameter ranking, was used to rank predictors.
47
Figure 3-1. Captures of A. glycines in suction traps by A) Julian date, and B) Photoperiod, for
2005-2009 growing seasons at 47 trapping sites in central North America.
Figure3-2. Three dimensional scatter plots of log of suction trap captures of A. glyines vs Julian
date and A) Photoperiod and B) Categorical field infestation (for summer trap captures only).
Traps that did not capture any aphids during a given sampling week are excluded from the figure
for clarity.
.
48
Figure 3-3. Fall captures of female and male A. glycines in suction traps by photoperiod. A
smoothing line (LOESS) with a span of 0.3 gives the local weighted regression of the female aphid
captures. Traps that did not capture any aphids during a given sampling week were excluded from
the figure for clarity. Timings of the early activity peak (E) and late activity peak (L) are marked
with arrows.
Partial linear regression was applied to remove spatial and temporal influence among
environmental predictors and determine the ranking of these parameters outside of season
and location effects (Zuur et al. 2007). ‘Optimum’ values (i.e. values of independent
variables at the point of model maxima) of relevant environmental parameters were
determined by computing the value of that parameter at the maximum as described by the
quadratic regression equation, where significant.
Visual inspection of a plot of suction trap captures by photoperiod (Fig. 3-3)
suggested a bimodal distribution may be responsible for the poorer-than-expected
performance of photoperiod as a predictor of aphid phenological events in fall (see
results). Subsequently, bimodality was detected by the likelihood ratio test for bimodality
(Likelihood ratio =204.38, Bimodality=TRUE) in the residuals from the ZIP model for
trap captures as predicted by the square of photoperiod (Holzmann and Vollmer 2008).
49
Bimodality was also detected using Hartigan and Hartigan’s (1985) dip test (n>200,
dip=0.0133, p<0.05) when site effects were incorporated into the model to account for
the zero bias of the data. Thus, an additional analysis was performed, dividing the ‘fall’
dataset at the apparent minimum between the two modes, into ‘early fall’ where
photoperiod was between 12-13h, and ‘late fall’ where photoperiod was ≤ 12h.
3.2.4 Geospatial analysis and flight trajectories
Spatial analysis was performed in ArcMap 9.3 (ESRI, http://www.esri.com/). The
density of flying soybean aphids at peak flight times (in summer, fall and for males) over
the sampling area for each sampling year was interpolated using ordinary Kriging and
splines. After initial analysis, Kriging was chosen in favour of splines to model aphid
flight density because of fewer anomalous high points in resultant models, an observation
that is, in general, consistent with the literature (Dubrule 1984).
Wind trajectories were modeled using the HYSPLIT web application (Hybrid
Single Particle Lagrangian Integrated Trajectory Model, National Oceanic and
Atmospheric Administration, http://ready.arl.noaa.gov/HYSPLIT.php). Backwards wind
trajectories were modeled for the week ending in August 8, and the week ending in
August 15, 2009 for sampling sites at Arva, ON (43.08oN, 81.20oW), and Manhattan,
KS (39.14oN, 96.64oW). At these sites aphid populations increased by two categorical
units during the sampling week of August 8, suggesting an aphid immigration event may
have occurred. Trajectories from the week of August 15 were modeled as a control, as
aphid field populations also increased by two categorical units during this week at Arva,
but no such increase was observed at Manhattan. Two trajectory altitudes representing
the midpoint of the plantary boundary layer and low-level jet stream altitudes ranges
50
were used, as other aphid species are known to fly at these altitudes (Smith and MacKay
1989, Zhu et al. 2006b). Backwards trajectories were modeled for air packets arriving at
the site at noon for each day within the sampling week, at 400 and 1200 m above ground
level, using the vertical velocity model and the GDAS meteorological data files within
HYSPLIT. Trajectories were modeled for a duration of 24 hours, the typical length of
time A. glycines is able to maintain optimal flight performance (Zhang et al. 2008).
Trajectory data were then plotted in ArcGIS and compared to interpolated surfaces
displaying the beginning and end of the sampling week. The wind trajectories modeled in
this study were initiated at noon on the simulation date, and simulations ran for 24 hours.
3.3 Results
3.3.1 Determination of migration status of captured aphids
The best predictor of suction trap counts in summer was categorical field
infestation one week before suction samples, and generally field assessments performed
before the suction trap assessment were better predictors of suction trap count than those
taken after (Table 3-3). Thus, it is most likely that suction trap captures primarily
represent emigrants from an area and therefore captured alates are most likely to have
experienced environmental conditions of the trap location.
Table 3-3: AIC values for Zero-inflated Poisson regressions of suction trap captures versus field
populations of A. glycines at different temporal delays between field population using categorical
field infestation and suction trap count of alate aphids. Julian date was used as a predictor of
excess zeros. For all models tested, p<0.05.
Timing of field population rating relative to trap count
Categorical field infestation
2 weeks before
1 week before
103901
92939
Same week
1 week after
2 weeks after
93502
101771
109467
51
3.3.2 Response to environmental variables
Preliminary analysis of the environmental variable data indicated that ZIP models
employing Julian date as a predictor of excess zeros in general outperformed Poisson
models with regards to model fit and conformation to assumptions of analysis for the raw
data from all three datasets (Tables 3-4 to 3-6). However, both model types ranked
parameters similarly and, in some cases, regression parameters could not be computed, so
Poisson models were used to compute optimum or threshold values of the various
applicable parameters. Field infestation level was not used to model fall suction trap
captures or captures of male aphids because soybean fields are not consistently monitored
for aphid populations once soybean begins to senesce.
Because of co-linearity between many parameters and spatial and temporal
measures partial linear regression was used to compute the rankings of certain parameters
after removing the influence of spatial (latitude, longitude) and temporal (Julian date)
effects (Tables 3-7 to 3-9).
In general, quadratic models performed better than their single-parameter
counterparts, indicating many parameters had optimum or threshold values. In cases
where biological precedent and statistical significance were found, computed ‘optimal’
values are presented in Tables 3-10 and 3-11.
52
Table 3-4: Model selection results for models predicting summer soybean aphid flights using
environmental parameters. Predictors used were raw data and models took the form Flight event
(=0 or 1) = predictor (binomial models), Total aphid captures = predictor (Poisson models) or
Total aphid captures = predictor, with Julian date as a predictor of excess zeros (zero-inflated
Poisson models). For each model, AIC and rank amongst that model type are given. Models which
are statistically significant at α=0.05 are marked with an asterisk. Statistics that could not be
computed are marked with a dash. NA marks non-applicable models (ie: quadratic models using
categorical variables).
Parameter
Degree day
accumulation
Field
infestation
Julian date
Latitude
Longitude
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Binomial models
Poisson models
Model
structure
linear
AIC
Rank
AIC
Rank
1315
10
*
239777
13
*
174015
12
*
quadratic
1287
5
*
207394
5
*
154089
2
-
linear
1088
1
*
137007
1
*
119940
1
*
quadratic
NA
*
NA
linear
1240
3
*
229233
9
*
-
-
-
quadratic
1196
2
*
199265
3
*
-
-
-
linear
1302
8
*
226899
8
*
170331
7
*
quadratic
1289
6
*
216176
7
*
164233
6
-
linear
1291
7
*
207098
5
*
158273
4
*
quadratic
1286
4
*
198379
2
*
154247
3
-
linear
1395
18
240554
16
*
173710
9
*
quadratic
1396
19
238855
11
*
172287
8
-
linear
1389
14
240987
17
*
174530
17
*
quadratic
1391
17
240425
15
*
174163
14
*
linear
1383
12
241342
19
*
173866
12
*
quadratic
1384
13
241003
18
*
173760
11
*
linear
1327
11
*
237446
10
*
174484
16
*
quadratic
1312
9
*
215419
6
*
160580
5
*
linear
1390
16
*
240240
14
*
174199
14
*
quadratic
1390
15
239025
12
*
-
-
-
NA
*
*
53
Zero-inflated Poisson
models
AIC
Rank
Table 3-5: Model selection results for models predicting fall soybean aphid flights using
environmental parameters. Predictors used were raw data and models took the form Flight event
(=0 or 1) = predictor (binomial models), Total aphid captures = predictor (Poisson models) or Total
aphid captures = predictor, with Julian date as a predictor of excess zeros (zero-inflated Poisson
models). For each model, AIC and rank amongst that model type are given. Models which are
statistically significant at α=0.05 are marked with an asterisk. Statistics that could not be
computed are marked with a dash.
Parameter
Degree day
accumulation
Julian date
Latitude
Longitude
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Binomial models
Poisson models
Model
structure
linear
AIC
Rank
AIC
Rank
553
14
350546
4
*
273967
2
*
quadratic
554
17
315754
1
*
224248
1
-
linear
543
9
387885
8
*
-
-
-
quadratic
544
10
343251
3
*
-
-
-
linear
538
8
*
415454
18
*
336670
12
*
quadratic
526
1
*
380067
7
*
319122
7
*
linear
552
13
414984
16
*
342725
15
*
quadratic
554
16
393211
10
*
-
-
-
linear
536
5
396204
11
*
333432
11
*
quadratic
538
8
352167
5
*
297762
4
*
linear
530
3
*
415145
17
*
342042
14
*
quadratic
527
2
*
372591
6
*
307596
5
*
linear
535
4
*
410323
15
*
326698
9
*
quadratic
537
6
400768
12
*
316404
6
*
linear
545
11
388628
9
*
323519
8
*
quadratic
546
12
342599
2
*
289634
3
-
linear
553
15
409844
14
*
338938
13
*
quadratic
555
18
402647
13
*
327279
10
*
*
*
*
54
Zero-inflated poisson
models
AIC
Rank
Table 3-6: Model selection results for models predicting male soybean aphid flights using
environmental parameters. Predictors used were raw data and models took the form Flight event
(=0 or 1) = predictor (binomial models), Total aphid captures = predictor (Poisson models) or Total
aphid captures = predictor, with Julian date as a predictor of excess zeros (zero-inflated Poisson
models). For each model, AIC and rank amongst that model type are given. Models which are
statistically significant at α=0.05 are marked with an asterisk. Statistics that could not be computed
are marked with a dash.
Parameter
Degree day
accumulation
Julian date
Latitude
Longitude
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Binomial models
Poisson models
Model
structure
linear
AIC
Rank
AIC
Rank
125
12
1675
7
*
279
4
*
quadratic
122
8
1564
3
*
267
3
-
linear
111
3
*
1678
8
*
-
-
-
quadratic
88
1
*
824
1
*
-
-
-
linear
125
16
1843
16
*
489
7
*
quadratic
123
9
1622
5
*
484
6
*
linear
125
14
1859
17
*
562
15
quadratic
127
17
1789
12
*
508
11
linear
125
15
1860
18
561
14
quadratic
120
7
1754
11
*
563
16
linear
125
11
1825
14
*
496
8
*
quadratic
114
5
1622
6
*
498
9
*
linear
124
10
1753
10
*
245
2
*
quadratic
114
6
*
1566
4
*
219
1
*
linear
111
4
*
1683
9
*
554
13
*
quadratic
88
2
*
825
2
*
426
5
-
linear
125
14
1841
15
*
553
12
*
quadratic
127
18
1796
13
*
507
10
*
*
55
Zero-inflated Poisson
models
AIC
Rank
-
Table 3-7: Model selection results for models predicting summer soybean aphid flights using
residuals from parameter data to account for co-linearity with location, or location and time.
Predictors used were residuals based on linear models taking the form parameter = (spatial and
temporal predictors) and models took the form Flight event = residual predictor +spatial and
temporal predictors (binomial models), Total aphid captures = residual predictor +spatial and
temporal predictors (Poisson models) For each model, AIC and rank amongst that model type are
given. Models which are statistically significant at α=0.05 are marked with an asterisk. NA marks
non-applicable models (ie: quadratic models using categorical variables).
Inputs
Parameter
Residuals accounting for
spatial effects
Degree day
accumulation
Field
infestation
Julian date
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Residuals accounting for
spatial and temporal
effects
Degree day
accumulation
Field
infestation
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Model
structure
linear
Binomial models
AIC
Rank
Poisson models
AIC
Rank
1124
7
*
202828
7
*
quadratic
1099
6
*
170177
4
*
linear
1032
3
*
117315
1
*
quadratic
NA
*
NA
linear
1084
5
*
192754
5
*
quadratic
984
2
*
162497
2
*
linear
1237
8
204943
13
*
quadratic
1238
13
204579
11
*
linear
1237
9
205184
15
*
quadratic
1238
12
205173
14
*
linear
1241
14
203213
9
*
quadratic
1242
15
203211
8
linear
1078
4
*
196039
6
*
quadratic
977
1
*
165922
3
*
linear
1238
11
204804
12
*
quadratic
1237
10
203885
10
*
linear
1049
11
184660
5
*
quadratic
1048
10
184600
4
*
linear
949
1
*
116528
1
*
quadratic
NA
linear
1032
4
*
192728
13
*
quadratic
1034
5
192457
12
*
linear
1037
6
189454
9
*
quadratic
1039
7
189350
8
*
linear
1050
12
187620
7
*
quadratic
1052
13
187524
6
*
linear
1004
3
*
180748
3
*
quadratic
972
2
*
174146
2
*
linear
1048
9
191877
11
*
quadratic
1047
9
191132
10
*
56
*
NA
*
Table 3-8: Model selection results for models predicting fall soybean aphid flights using residuals
from parameter data to account for co-linearity with location, or location and time. Predictors used
were residuals based on linear models taking the form parameter = (spatial and temporal
predictors) and models took the form Flight event = residual predictor +spatial and temporal
predictors (binomial models), Total aphid captures = residual predictor +spatial and temporal
predictors (Poisson models) For each model, AIC and rank amongst that model type are given.
Models which are statistically significant at α=0.05 are marked with an asterisk.
Inputs
Parameter
Residuals accounting for
spatial effects
Degree day
accumulation
Julian date
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Residuals accounting for
spatial and temporal
effects
Degree day
accumulation
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Binomial models
Poisson models
Model
structure
linear
AIC
Rank
AIC
Rank
539
12
*
291250
2
*
quadratic
536
10
*
275233
1
*
linear
533
4
*
386481
8
*
quadratic
534
7
*
340088
3
*
linear
534
6
*
382716
6
*
quadratic
532
1
361613
5
*
linear
532
2
411849
14
*
quadratic
534
8
385866
7
*
linear
535
9
407305
12
*
quadratic
537
11
404776
11
*
linear
533
3
387348
9
*
quadratic
533
5
342405
4
*
linear
541
13
407524
13
*
quadratic
543
14
400293
10
*
linear
535
10
290303
2
*
quadratic
502
1
*
282704
1
*
linear
531
2
*
370981
8
*
quadratic
531
3
331796
6
*
linear
532
5
383449
11
*
quadratic
533
6
345223
7
*
linear
534
8
330840
5
*
quadratic
535
11
324772
4
*
linear
535
9
385756
12
*
quadratic
531
4
322956
3
*
linear
533
7
378920
10
*
quadratic
535
12
371773
9
*
57
*
*
*
*
Table 3-9: Model selection results for models predicting male soybean aphid flights using residuals
from environmental parameter data to account for co-linearity with location, or location and time.
Predictors used were residuals based on linear models taking the form parameter = (spatial and
temporal predictors) and models took the form Flight event = residual predictor +spatial and
temporal predictors (binomial models), Total aphid captures = residual predictor +spatial and
temporal predictors (Poisson models) For each model, AIC and rank amongst that model type are
given. Models which are statistically significant at α=0.05 are marked with an asterisk.
Inputs
Parameter
Residuals accounting for
spatial effects
Degree day
accumulation
Julian date
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
Binomial models
Poisson models
Model
structure
linear
AIC
Rank
AIC
Rank
128
9
1601
5
*
quadratic
128
11
1576
4
*
linear
114
3
*
1641
6
*
quadratic
91
1
*
786
1
*
linear
129
13
1826
13
*
quadratic
125
7
1773
11
*
linear
128
11
1828
14
*
quadratic
117
5
1648
8
*
linear
128
8
1753
9
*
quadratic
121
6
*
1502
3
*
linear
114
4
*
1645
7
*
quadratic
92
2
*
797
2
*
linear
129
12
1820
12
*
quadratic
131
14
1772
10
*
linear
114
7
1537
6
*
quadratic
116
11
1539
7
*
linear
115
8
1643
12
quadratic
111
5
1580
8
*
linear
100
4
1434
5
*
quadratic
99
3
1397
4
*
linear
95
1
1058
2
*
quadratic
96
2
925
1
*
linear
116
10
1642
11
*
quadratic
112
6
1311
3
*
linear
115
9
1613
10
*
quadratic
117
12
1583
9
*
*
linear
Residuals accounting for
spatial and temporal
effects
Degree day
accumulation
Maximum
temperature
Mean
temperature
Minimum
temperature
Photoperiod
Total
precipitation
58
*
*
59
Male
Fall
Summer
-1.60E-05
-4.50E-01
-4.10E-01
-5.60E-02
-1.00E-01
-1.70E-01
-1.00E-01
-2.20E+02
-2.30E-03
NS
-2.00E-01
NS
NS
NS
NS
NS
-8.50E+01
NS
Julian date
Latitude
Longitude
Maximum temp.
Mean temp.
Minimum temp.
Photoperiod
Tot. precipitation
1.10E+02
2.00E+03
6.30E-02
2.80E+0
1
4.80E+00
6.60E+02
3.40E+01
1.30E+00
11.9
269.2
42.8
11.1
244.6
32.8
3424.9
9.7
-2.20E-02
-2.30E-01
-5.00E-02
-9.50E-02
-1.20E-01
-1.90E-02
-1.00E+01
-1.30E-03
3.30E-05
-1.80E-05
Degree day
accumulation
1.50E-02
1049.7
13.5
97.8
31.2
105.6
NS
-5.60E-02
NS
NS
NS
NS
NS
NS
4.80E-03
5.30E+00
229
45.3
-98.2
Julian date
Latitude
Longitude
Maximum temp.
Mean temp.
Minimum temp.
Photoperiod
Tot. precipitation
2.20E-06
-4.10E-03
2.10E+01
1.90E-01
NS
2.00E-06
1.30E-01
9.30E-01
6.80E-01
Tot.
precipitation
day
Degree
accumulation
8.50E-01
4.00E+00
-1.90E+00
2.90E-04
1.10E-02
3.80E-03
-1.90E-03
-4.40E-02
-9.60E-03
NS
NS
NS
-7.70E-01
2.60E-02
4.40E-02
9.20E-03
1.30E-02
2.40E-02
1.30E-02
1.20E+01
5.60E-04
1.80E-06
1.60E-04
1.70E-03
4.40E-04
6.80E-04
8.50E-04
2.50E-04
7.60E-02
2.40E-05
1.20E-06
1.10E-07
3.50E-05
1.50E-03
4.80E-04
4.60E-04
6.20E-04
3.80E-04
2.10E-02
1.90E-07
2.50E+02
3.30E+01
-9.90E+00
5.50E+00
7.00E+00
2.90E+00
5.10E+03
6.70E-02
4.50E-02
1.20E+01
1.90E+01
-9.00E+00
4.70E+00
4.20E+00
6.70E-01
2.50E+02
3.80E-02
-3.40E-03
5.50E-02
2.10E+00
9.90E+00
-7.20E+00
1.20E+00
1.30E+00
4.40E-01
7.10E+01
1.80E-02
Most aphids captured (Poisson
2
models)
x
SE
x
-4.60E-03
-1.10E-01
-3.80E-02
-1.90E-02
-2.70E-02
-1.20E-02
-2.50E+00
SE
Julian date
Latitude
Longitude
Maximum temp.
Mean temp.
Minimum temp.
Photoperiod
optimum
-2.30E-05
SE
NS
x
Degree day
accumulation
SE
x2
Flight events (binomial models)
Parameter X
Aphid dataset
1.40E+01
3.50E+00
1.60E+00
7.10E-01
9.50E-01
2.80E-01
3.00E+02
1.90E-02
4.30E-03
8.60E-02
1.40E-01
7.90E-02
3.50E-02
2.90E-02
4.70E-03
1.80E+00
9.30E-04
3.00E-04
2.60E-04
1.60E-02
1.30E-01
8.80E-02
2.80E-02
2.70E-02
9.80E-03
5.70E-01
1.90E-04
SE
272.3
40.1
-88.1
27.4
20.5
14.1
11.9
14.4
1393.5
267.2
40.7
-90.1
24.6
17.4
18.1
12
14.5
51.8
1487.3
223.7
44.2
-94.4
30.6
24.3
17.8
13.9
390.5
optimum
44
12
40.9
9.8
8
4.5
1.9
10.7
414.3
5.6
0.9
2.2
0.5
0.4
0.5
0.2
0.9
10.1
22.2
4.8
1.7
3.3
2.1
1.5
1.3
0.3
10.1
SE
X
X
X
X+ location+date
X+ location+date
X+ location+date
X
X+ location+date
X+ location+date
X
X
X
X+ location+date
X+ location+date
X+ location+date
X
X+ location+date
X+ location+date
X
X
X
X+ location+date
X+ location+date
X+ location+date
X
X+ location+date
Model
parameters
Table 3-10: Model coefficients and computed optimal values (± standard error SE) of environmental parameters for predicting aphid
flight events and total aphid captures in summer, fall, and for male aphids. Coefficients for linear (x) and quadratic (x 2) are given.
Models with non-significant regression coefficients (at α=0.05) are marked NS and optimum values are not computed.
60
341
335
337
344
334
191
197
182
200
198
Degree day
accumulation
Julian date
Latitude
Longitude
Photoperiod
Degree day
accumulation
Julian date
Latitude
Longitude
Photoperiod
Late
fall
AIC
*
*
*
*
rank
2
3
1
5
4
4
2
3
5
1
-1.00E-01
7.90E+00
9.10E-06
1.20E-02
x2
3.30E-02
2.40E+00
3.50E-06
4.10E-03
SE
Flight events (binomial models)
Early
fall
Parameter X
Aphid dataset
2.70E+00
6.00E+01
2.00E+02
8.70E+00
6.90E-03
2.10E+00
SE
-1.80E025.90E+00
x
41.7
12.6
997
251
optimum
36.
9
10.
7
10
79
24
9
SE
222388
147873
227523
239824
152978
83146
89618
111814
108655
89657
AIC
*
*
*
*
*
*
*
*
*
*
3
1
4
5
2
1
2
5
4
3
rank
-1.60E-05
-4.00E-01
-3.40E-01
-8.40E-02
-1.90E+02
-1.60E-05
-2.10E-02
-1.20E-01
-4.30E-02
-1.10E+01
x2
1.40E-07
1.50E-03
2.60E-03
7.30E-04
7.10E-01
2.10E-07
6.30E-04
2.20E-03
7.40E-04
3.10E-01
SE
Most aphids captured (Poisson models)
x
4.10E-02
2.10E+02
2.80E+01
-1.50E+01
4.40E+03
4.30E-02
1.10E+01
9.60E+00
-7.90E+00
2.60E+02
3.20E-04
7.90E-01
2.10E-01
1.30E-01
1.70E+01
4.80E-04
3.30E-01
1.80E-01
1.40E-01
7.70E+00
SE
1317
269
40.3
-89
11.9
1305
265.8
41.5
-92.3
12.1
optimum
31
2.8
0.9
2.2
0.1
44
22.
5
2.2
4.5
1
SE
2.3
1
2.2
2.5
1.1
3.4
8.5
5.4
4.9
8.3
%
var
Table 3-11: Model selection results, model coefficients and computed optimal values (± standard error SE) of environmental parameters for
predicting aphid flight events and total aphid captures for two aphid flight events in fall. Coefficients for linear (x) and quadratic (x2) are given.
Models with statistically significant regression coefficients (at α=0.05) are marked with an asterisk. Optimum values are not computed for
models with non-significant coefficients. Percent variation (% var, 100*SE/mean) is given for the Poisson models.
Summer data
In binomial models predicting the likelihood of a flight event, weather parameters
rarely yielded statistically significant models, but significant relationships between the
density of flying aphids and weather parameters were observed in most cases (Tables 3-4
and 3-7). Field infestation best explained the occurrence and density of flying aphids
(Table 3-4). Julian date was the next best predictor for flight events (optimum at day
229, August 16) (Tables 3-4, 3-10). Additionally, spatial parameters, particularly
longitude, were strong predictors of both the occurrence and density of flying aphids. The
‘optimal’ longitude for aphid flights computed was 98.2oW, which is outside the study
area, suggesting that this value is an artifact of analysis and that the likelihood of
observing summer aphid flights was highest on the western margins of the study area.
The optimal latitude for summer aphid flights was 45.3oN, approximately the latitude of
Minneapolis, MN. The highest density of flying aphids was predicted to be at 44.2oN,
94.4oW, in the south eastern corner of Brown County, MN.
When spatial effects were accounted for in the model, photoperiod became the
best predictor for whether aphid flights were to occur, but density of flying aphids was
still best predicted by field infestation (Table 3-7). When both spatial and temporal
effects were accounted for in the model, field infestation level was the best predictor for
both aphid flight events and density (Table 3-7).
Optimum values for various weather parameters were computed, where
statistically significant relationships were obtained (Table 3-10), but should be
interpreted with caution because these parameters had poorer performance explaining
61
aphid flights than those outlined above. These optima and should only be used to refine
models that incorporate the more predictive parameters identified previously.
Fall data
Latitude and mean temperature were the best performance parameters for
predicting fall aphid flights (Table 3-5). However, a relatively small range of AIC values
was obtained in this analysis, and very little difference between AIC values was observed
when spatial and spatio-temporal effects were removed from models, and only degreeday accumulation improved model performance in the latter case (Table 3-8). This lack
of variation in AIC values suggests aphid flight events in fall are not well explained by
any of the parameters tested.
Degree-day accumulation outranked photoperiod as a predictor of optimum aphid
flight density (Table 3-5). This effect was observed consistently in models accounting for
spatial and spatio-temporal effects as well (Table 3-8).
Because of bimodality in the fall data when examined by photoperiod, the dataset
was divided into two subsets corresponding to ‘early fall’ and ‘late fall’ activity peaks.
The late fall peak appeared to coincide with the detection of male aphids (Fig. 3-3) and
parameters fit data from the early fall and late fall subsets differently (Table 3-11). In
early fall, the best predictor of whether an aphid flight would occur was photoperiod
(with an ‘optimum’ predicted photoperiod of 12.6h ±10.7h) (Table 3-11). However, the
AIC for the photoperiod model was only one unit lower than the one for Julian date and
thus model performance is considered equivalent, the optimum value for this parameter
also had a similar large standard error ( day 251±249). In late fall, the best predictor for
aphid flight occurrence was latitude, with an optimum value of 41.7±36.9oN, or the
62
approximate latitude of Des Moines, IA. None of the other parameters yielded
statistically significant regression parameters.
The models for aphid density were statistically significant for all parameters
tested for both early and late fall. In early fall, degree day accumulation was as the best
predictor of flying aphid density, followed by Julian date and photoperiod, but in late fall,
Julian date and photoperiod outranked degree day accumulation as predictors of flight
density. Latitude and longitude were also significant predictors of aphid density, and the
density of aphids was estimated to be highest at 41.5oN, 92.3oW (i.e. north-western
Keokuk county, IA) in early fall, and 40.3oN, 89.0oW (i.e. southern McLean county, IL)
in late fall. The two density peaks occurred about three days apart, a difference in
photoperiod of approx. 0.2h, and though the standard errors overlapped between the two
peaks for all variables tested, the proportional error (reported as % variation in Table 311) was much smaller for the late fall density peak.
Male data
Julian date and photoperiod were the best predictors of both the occurrence and
density of male soybean aphid captured in suction traps (Table 3-6). This relationship
held in models where spatial effects were accounted for, but minimum temperature
during the sampling week became the top-ranked predictor when both spatial and
temporal effects were accounted for in the models. Males were most likely to be
observed and had the highest density at a 11.9h photoperiod, coinciding exactly with the
‘late fall’ peak density of female aphids. The location where the highest density of males
was predicted to occur was 40.1oN, 88.1oW, just west of Champaign, IL. The longest
63
photoperiod at which a male was trapped was 12.6 h, corresponding to a date of 11
September at the site it was captured.
3.3.3 Geospatial analysis and flight trajectories
Density of flying aphids was interpolated by year from trap captures from the
sampling date closest to the ‘optimum’ flight date for summer (Fig. 3-4), early fall (Fig.
3-5), late fall (Fig. 3-6), and males (Fig.3-7). Male density could only be interpolated for
2008 and 2009 because males were not observed in 2005-2006, and numbers were too
low to perform Kriging analysis in 2007.
During the week ending on Aug. 8, 2009, field infestations at Arva and Manhattan
underwent abrupt growth events (Fig. 3-8A-B). Backwards wind trajectories were well
corroborated with apparent aphid population movement during the week for both
Manhattan and Arva. A trajectory arriving 400m above ground level at Manhattan on
August 5 was shown to have originated 24 hours earlier in an area with a high density of
flying aphids in northern Iowa, which may explain the apparent jump in field aphid
populations in the vicinity of Manhattan during that week (Fig 3-8C). Similarly, wind
trajectories arriving at 400m over Arva on Aug. 3, or 1200m over Arva Aug. 3, 6 and 7,
originated in areas with high populations of flying aphids (Fig. 3-8C-D).
64
Figure 3-4. Density of flying aphids observed in central North America during the week of August
11, the calculated ‘optimal’ summer flight date, in 2005-2009. Density surfaces were based on
suction trap captures and estimated using ordinary Kriging. Traps that did not report any data
during the sampling week were excluded from Kriging analysis and are not shown on the figure.
IA= Iowa, IL=Illinois, IN=Indiana, KS=Kansas, KY=Kentucky, MI=Michigan, MN=Minnesota,
ON=Ontario, WI=Wisconsin
65
Figure 3-5. Density of flying aphids observed in central North America during the week of
September 22, the calculated ‘optimal’ early fall flight date, in 2005-2009. Density surfaces were
based on suction trap captures and estimated using ordinary Kriging. Traps that did not report
any data during the sampling week were excluded from Kriging analysis and are not shown on the
figure. See Fig. 3-4 for explanation of abbreviations.
66
Figure 3-6. Density of flying aphids observed in central North America during the week of
September 29, the calculated ‘optimal’ late fall flight date, in 2005-2009. Density surfaces were
based on suction trap captures and estimated using ordinary Kriging. Traps that did not report
any data during the sampling week were excluded from Kriging analysis and are not shown on the
figure. See Fig. 3-4 for explanation of abbreviations.
67
Figure 3-7. Density of male aphids observed in central North America during the week of
September 29, the calculated ‘optimal’ flight date, in 2008-2009. Density surfaces were based on
suction trap captures and estimated using ordinary Kriging. Traps that did not report any data
during the sampling week were excluded from Kriging analysis and are not shown on the figure.
No males were observed in 2005 or 2006, and male populations were insufficient for Kriging
analysis in 2007. See Fig. 3-4 for explanation of abbreviations.
68
Figure 3-8. Using wind trajectories to determine origin of flying aphids during the week of August
1-8, 2009. Soybean fields at Arva, ON and Manhattan, KS both had abrupt increases in aphid
populations during this week, suggesting the increase may be due to an immigration event. A)
Field aphid density as estimated by Kriging on Aug 1, 2009, over the study area; B) Field aphid
density as estimated by Kriging on Aug 8, 2009, over the study area; C) Aphid flight density
estimated by Kriging for Aug 2-8 and 24-hour backwards wind trajectories arriving at 400m
above Arva and Manhattan at noon on each day Aug 2-8; D) Aphid flight density estimated by
Kriging for Aug 2-8 and 24-hour backwards wind trajectories arriving at 1200m above Arva and
Manhattan at noon on each day Aug 2-8. Traps that did not report any data during the sampling
week were excluded from Kriging analysis and are not shown on the figure.
69
Figure 3-9. Using wind trajectories to determine origin of flying aphids during the week of August
8-15, 2009. Soybean fields at Arva, ON had an abrupt increase in aphid populations during this
week, suggesting the increase may be due to an immigration event; but fields at Manhattan, KS,
did not have have an abrupt growth in aphid populations during this week. A) Field aphid density
as estimated by Kriging on Aug 8, 2009, over the study area; B) Field aphid density as estimated
by Kriging on Aug 15, 2009, over the study area; C) Aphid flight density estimated by Kriging for
Aug 9-15 and 24-hour backwards wind trajectories arriving at 400m above Arva and Manhattan
at noon on each day Aug 9-15 ; D) Aphid flight density estimated by Kriging for Aug 9-15 and 24hour backwards wind trajectories arriving at 1200m above Arva and Manhattan at noon on each
day Aug 9-15. Traps that did not report any data during the sampling week were excluded from
Kriging analysis and are not shown on the figure.
70
In the week ending in Aug. 15, 2009, field infestation at Arva underwent another
abrupt growth event, but aphid infestation at Manhattan remained relatively constant
(Fig. 3-9A-B). During this week, a trajectory ending 400m over Manhattan on Aug. 11
originated in an area with high flying aphid activity, near Omaha, NB, but no jump in
field populations in the vicinity was observed, despite conditions being very favourable
for aphid flight activity at the point of origin (24oC mean daily temperature, no
precipitation and an average wind speed of 7km/h) (Weather Underground
http://www.wunderground.com/history/). A trajectory arriving 1200m over Arva on Aug.
10, however, corroborates with the jump in field infestation at this site during this week.
3.4 Discussion
This study examines the production and dynamics of alates of A. glycines in
summer and fall, but not spring migrations of this species. Molecular work suggests that
A. glycines undergoes a genetic bottleneck in the spring on the overwintering host, and
relatively few clonal lines survive to colonize the summer host, leading to low local
genetic diversity of aphids in soybean early in the growing season (Michel et al. 2009).
Traps in the North American suction trap network have inconsistently captured A.
glycines before early June, suggesting that this bottleneck results in such low numbers of
spring migrants that suction traps are not an effective tool for studying the spring
migration phenology of this aphid (Rhainds et al. 2010b, Schmidt et al. 2012).
This study found that captures of A. glycines at suction traps most likely represent
aphids leaving the general vicinity of the trap. Rhainds et al. (2010b) attempted to use
local captures at suction traps to forecast aphid populations in a given area, but found that
71
local and seasonal variability limited this application. We have provided an additional
explanation for the limited utility of suction traps for local forecasting: captures at suction
traps predominantly represent emigrants from an area and thus populations at a given trap
must be regarded as a source population and not as a representation of immigrating
aphids. This supports an observation made by Schmidt et al. (2012), that, in general,
increased numbers of alates are trapped by this network at more northerly latitudes,
where the overwintering host is present at higher densities, and thus alates are likely
captured as they leave source habitats (i.e. soybean fields first colonized after spring
migrations from buckthorn).
Suction traps have been useful for local forecasting for other aphids, however. For
example, suction trap captures of Rhopalosiphum padi and Myzus persicae in New
Zealand were well-correlated temporally with densities of each species in their respective
host crops (Teulon et al. 2004). Even if a given suction trap is not useful for forecasting
aphid infestations at a given location, suction trap data can provide insight into the
density, and thus relative importance, of migratory aphids to aphid population dynamics
in a given growing season (Klueken et al. 2009).
It is unsurprising that field infestation level was the best predictor of the density
of alates of A. glycines in summer. Population density as a trigger for alate production is
apparently dependent on tactile stimulation due to crowding, that is, repeated direct
physical contact between aphids. Thus, behavioural mechanisms affecting apterous aphid
distribution on a given host plant will also affect the production of alates (Harrison
1980). Alate production in A. glycines is density dependent in soybean, but density is also
locally regulated by behavioural mechanisms amongst apterae and nymphs. When the
72
resources of a colonized leaflet have been well-exploited by conspecifics, A. glycines
individuals move to other, less populated leaflets on a plant and to directly adjacent
plants (Ito 1953). As population density increases, however, it becomes advantageous for
A. glycines to produce alates and engage in migration to colonize other locales.
Contrary to our result, an earlier study (Hodgson et al. 2005) linked production of
alates in A. glycines to photoperiod and plant growth stage, but the study did not
distinguish between autumn gynoparae and summer asexual morphs, as they cannot be
morphologically distinguished (Voegtlin et al. 2004b). Hodgson et al. (2005) did not
directly account for population density of aphids, but did note that production of alatoid
nymphs was approx. 0-6% when aphids were just reaching detectable levels in soybean in
early summer, and steadily increased over the growing season. Thus, their finding that
plant stage and photoperiod govern alate production, though valid, is likely a result of colinearity of these parameters and increasing density of aphids in soybean fields over the
growing season (Fig. 3-2b illustrates this phenomenon).
The apparent bimodal distribution of A. glycines flight activity observed in fall
and the different ranking of parameters associated with the two activity peaks may have a
biological explanation. The early fall activity peak is likely composed of gynoparous
alates migrating to buckthorn to deposit oviparae prior to the production of males.
Gynoparous A. glycines have been induced in the laboratory after approximately 4 weeks
of exposure to a short (10L:14D) photoperiod and corresponding temperature regime
(20:12°C), and males were produced after gynoparae under these same conditions
(though the length of time between the production of gynoparae and males was
unspecified) (Zhu et al. 2006a). The early fall activity peak observed in the present study
73
occurred at a photoperiod of 12.1 ± 1.0 h, but if the conditions inducing the gynoparae
occurred up to four weeks prior as suggested by the methodology of Zhu et al. (2006a),
the critical photoperiod for sexual morph induction would be around 13.3 h and
decreasing. Development of subsequent generations after induction of gynoparae is likely
thermally dependant, which would explain why degree day accumulation, rather than
photoperiod or Julian date, is the top-ranked explanatory variable for this activity peak.
Once oviparae are established on the overwintering hosts, they release sex
pheromones which males use in mate location, however, surviving alate females also
respond to these pheromones (Zhu et al. 2006a). Both males and gynoparous females had
electrophysiological responses to pheromones produced by oviparous A. glycines, and
traps baited with pheromones caught more males and gynoparous alates than unbaited
traps (Zhu et al. 2006a). The late fall activity peak in female aphids corresponded exactly
to the peak activity of males by photoperiod, though not by date, suggesting that
photoperiod-mitigated release of sex pheromones by oviparae may govern flight
behaviour in both sexes at this time of year. This secondary peak of females may be a
bet-hedging strategy: one group of females initially colonizes overwintering hosts and
awaits males, the second group follows males and may produce a second group of
oviparae as the first oviparae are producing eggs. Both male and female aphids can mate
multiply (Doherty and Hales 2002) so this strategy may maximize opportunities for
sexual reproduction.
A critical photoperiod of approx. 13.3h of daylight for the production of sexual
morphs has interesting implications with respect to aphid phenology in other portions of
its range, particularly equatorial regions. Aphid species are typically referred to as
74
anholocyclic when, as a species, they are incapable of producing sexual morphs and eggs,
however, some authors may also use this term to describe populations that do not lay
eggs because of ecological circumstances, such as lack of a suitable overwintering host or
appropriate environmental cues (Lambers 1966). Holocyclic aphid species may reproduce
parthenogenetically indefinitely, if they are not exposed to cues inducing sexual forms
and do not encounter any lethal environmental conditions (Bonnemaison 1951). Because
sexuparous production is photoperiod-dependent in A. glycines, populations of this
species in equatorial regions will not produce sexuals and thus will be functionally
anholocyclic. Indeed, soybean aphid reaches economically damaging numbers on
soybean in Sumatra, but has not been observed on woody overwintering hosts in that
region, and production of alate A. glycines in Indonesian soybean is likely to be a
response to population density (Van Den Berg et al. 1997).
For many of the models for summer, fall and male migrants alike, Julian date and
photoperiod were ranked very closely as explanatory parameters, with Julian date
sometimes outperforming photoperiod as a predictor. This result was unexpected, given
previous work emphasizing photoperiod as a driving factor of the phenology of aphids
(e.g. Smith and MacKay 1989, Via 1992, Hodgson et al. 2005), and is likely an artifact of
sampling resolution. Because samples were taken only every seven days, and on the same
date throughout the sampling area, subtle differences in photoperiod between sites were
unable to account for any more variation than otherwise explained by date.
Most environmental parameters tested were shown to have statistically significant
relationships with the production or density of flying aphid populations. In some cases,
this trend could simply be a result of co-linearity between parameters that was not
75
entirely eliminated, even in models designed to account for these factors. However, in
some cases, these associations may be explained in the context of the biology of A.
glycines. The association between rain and alate density, for instance, in summer, fall and
male populations (Table 3-10) may be an artifact of sampling resolution. Alates likely
will not take off during rain, and thus, a larger cohort of aphids may mature during, and
then take flight after a period of unfavourable weather, and the week-long sampling
period used for this study would lose these details. Alternately, rain events can be highly
localized and may not affect the exact location within the sampling area from which
alates are originating (Klueken et al. 2009).
A. glycines flight activity was widely variable, both spatially and in magnitude,
from year to year (Fig. 3-4). If aphid activity was high, it usually peaked in the vicinity of
northern Iowa and southern Minnesota, areas where field infestations of the aphid usually
reach economic levels every year, according to scouting data for field infestation used in
this study. Because these areas are well within the overwintering range of A. glycines it
would be expected that field infestations of A. glycines reach high numbers in this area.
High numbers in this region effectively function as a source population for aphids
infesting the larger summer range of the aphid, where it regularly occurs as far south as
Manhattan KS (39.1oN). However, the lack of detection of A. glycines flight activity
north of the 41st parallel in the more eastern portion of the range was unexpected. Only
in 2005 was high density aphid flight detected in summer in Michigan and the
westernmost portion of southwestern Ontario, though overwintering hosts occur in these
areas in high densities, and large populations of aphids have been observed to overwinter
in these areas (Bahlai et al. 2007, Welsman et al. 2007, Bahlai et al. 2010a). A number of
76
factors may account for this discrepancy such as differences in climate, natural enemy
abundance of community composition, and agricultural practice between these two
regions. The Great Lakes modify the continental climate in central North America, and
this effect is more pronounced in areas such as Michigan and southwestern Ontario which
are surrounded by more water (Scott and Huff 1996). This lake effect could result in
conditions that are less favourable for rapid development and reproduction of A. glycines,
and less favourable flight conditions because of more frequent precipitation events. More
likely, the natural enemy complex of A. glycines in southwestern Ontario and Michigan
may differ from more western regions in both composition and aphid suppression
efficacy. In southern Ontario, a widely occurring parasitic wasp Aphelinus certus has
been shown to be extremely effective in suppressing population growth of A. glycines
(Frewin et al. 2010), and thus may prevent aphids from reaching population densities on
summer hosts required to induce large-scale emigration events. Though present in other
regions, A. certus is only noted for its biocontrol efficacy in the eastern portion of the
range of A. glycines (Ragsdale et al. 2011). Finally, cropping practices may differ
between the two regions, enhancing the effect of natural enemies in suppressing
populations of the aphid (Gardiner et al. 2009b).
Abundance of gynoparous A. glycines in fall (Figs. 3-5, 3-6) were as variable,
year-to year, as summer populations, and spatial distributions were more variable. In
every year except 2005, flight activity of A. glycines was more widespread in late fall
than in early fall. In 2009, flying A. glycines associated with both activity peaks were
extremely abundant, with many traps capturing thousands of aphids in a given week.
Lower numbers were observed in the eastern portion of the range (MI and ON, and
77
northeastern KY and IL) for the first activity peak, and only a few traps on the margins of
the study area captured low numbers by the late fall peak. The trap that captured the most
aphids (> 18,000 during the week of Sept 25, 2009) was at Urbana, IL (40.10oN,
88.23oW), and also captured a large number of male aphids (Fig. 3-7) at this time,
indicating that the late fall peak in gynoparous alate activity coincides spatially, as well
as temporally, with flights of male A. glycines.
Male aphids, though present, were not nearly as abundant as gynoparous females
during the late fall activity peak (note the change in scale in Fig. 3-7). No males were
captured by the suction trap network in 2005 and 2006; and in 2007, captures were
insufficient for interpolation analysis.
Although backwards wind trajectories were well-corroborated with increases in
aphid populations at a given site and thus, potential migration events from source
populations (Figs. 3-8, 3-9), we caution the use of this technique to forecast aphid
immigration events. The occurrence of a trajectory linking an area of high aphid density
to one of lower does not necessarily mean a migration event has occurred (Fig. 3-9). Use
of wind trajectories to forecast aphid migration events is limited by the temporal
resolution of samples. Because numerous trajectories occur over a one week trapping
period, the use of cumulative captures over a week may mask temporal synchrony
between aphid flight activity and a given wind vector. Additionally, each sample may
represent multiple flight events (Schmidt et al. 2012). The use of meteorological data
such as wind trajectories is somewhat speculative because it does not address the
possibility of aphids originating from a more local source (Loxdale et al. 1993). In order
for this method to be applied to forecasting migration events of A. glycines, an increased
78
temporal resolution of suction sampling, combined with sampling of fields where
trajectories indicate immigration events are likely to occur would be required, and
molecular analysis of source and sink populations, such as those performed by Michel et
al. (2009) could then determine if (and which) wind trajectories were involved in the
movement of aphids. Similarly, a probabilistic model to forecast immigration events
could be developed by expanding this work to include more site combinations, examining
backwards trajectories for each sampling week, and examining the proportion of
corroborating trajectories that result in apparent immigration events.
Although, in general, our findings are limited in regards to directly forecasting
when and where infestations of A. glycines will appear, this study provides a new
understanding of the phenology and ecology of this species, particularly with respect to
flights of gynoparae to overwintering hosts. A thorough understanding of soybean aphid
phenology is beneficial in the study of environmental change ecology. Because aphid
development is rapid, multivoltine and highly temperature dependent, phenological
changes in aphid populations will likely be apparent earlier than changes other insect taxa
(Harrington et al. 2007). Continued monitoring of aerial populations of A. glycines in
North America is warranted: additional spatial and phenological trends may become
apparent with the accumulation of additional data.
3.5 Acknowledgements
The authors would like to thank all those who contributed data to the North
Central Suction trap network and scouted aphids for the USDA Pipe website. Todd
79
Phibbs (U of Guelph Ridgetown) provided technical support and maintenance of the
Ontario suction trap network. CB was funded by a PGS-D3 postgraduate fellowship
from the Natural Sciences and Engineering Council of Canada.
80
CHAPTER 4
A mechanistic model for a tritrophic interaction involving
multiple natural enemies: using natural enemy units to quantify
the impact of a guild5
4.0 Abstract
When natural enemy communities are diverse, it becomes challenging to quantify
the impact a guild has on a prey species. Individual natural enemy species differ with
respect to prey consumption rates and foraging behaviours and may occur at different
times in the lifecycle of a prey species in response to environmental cues, densities, or the
availability of alternate prey. In order to standardize the impact of natural enemy guilds
on prey species, the Natural Enemy Unit (NEU) was developed, where NEU is defined as
the number of individuals of a predatory species that can kill 100 individual prey in 24h.
Soybean aphid (Aphis glycines) is a severe pest of soybean in North America with a
diverse natural enemy guild. In this chapter, a mechanistic population and phenology
model is developed for this species, incorporating environmental cues, host plant cues
and natural enemy dynamics interfaced through a NEU calculation. Use of the NEU
simplifies the incorporation of additional natural enemy species to the model, improving
the ability of this model to be generalized to multiple locations.
KEYWORDS: Aphis glycines, Aphelinus certus, Coccinella septempunctata,
Orius insidiosus, DYMEX, deterministic model.
5
Manuscript in preparation for Ecological Modeling, by C Bahlai, R. Weiss, and R. Hallett. CB designed,
implemented, the model and wrote the paper, RW helped design the underlying model structure and
provided calibration support, RH was involved in model conception, and editing and MS editing.
81
4.1 Introduction
Natural enemies are important regulators of the population growth of pest species,
and a high diversity of natural enemies typically enhances biological control. When a
natural enemy guild consists of multiple predator and parasitoid species, an increased
proportion of prey insects are typically consumed by comparison with systems where
only one natural enemy species is present (Aquilino et al. 2005). Straub and Snyder
(2008) found that diversity in natural enemy communities enhanced aphid suppression in
two cropping systems; they attributed this enhancement to interspecific differences in
foraging patterns allowing prey resources to be partitioned more effectively within the
tropic level. Additionally, they found that as diversity decreased, individuals of a given
predatory species would often allocate less time to foraging.
When natural enemy communities are diverse, however, it can be more difficult to
empirically determine the net impact of the guild on the prey species. Individual natural
enemy species differ with respect to prey consumption rates and foraging behaviours
(Xue et al. 2009, Frewin et al. 2010). Predatory species may occur at different times in
the lifecycle of a prey species because of differential responses to environmental cues,
densities, or the availability of alternate prey items (Yoo and O'Neil 2009). In order to
standardize the impact of natural enemy guilds on prey species, the Natural Enemy Unit
(NEU) was developed (Chapter 5; Bahlai et al. 2010b).6 One NEU is defined as the
number of individuals of a predatory species that can kill 100 individual prey in 24h,
6
Although Chapter 5 occurs later in this thesis than the current chapter, the work for Chapter 5 was
performed prior to the work presented in the current chapter, and has already been published (Bahlai et al.
2010b).
82
assuming saturating density of prey, and this measure was originally used to quantify the
net impact of a pesticide application on resident biocontrol services.
Soybean aphid (Aphis glycines) is a severe pest of soybean in North America
(Ragsdale et al. 2004). Since its introduction to North America in 2000, a diverse natural
enemy guild including both predators and parasitoids have adopted A. glycines as prey
(Fox et al. 2004, Rutledge et al. 2004, Fox et al. 2005, Nielsen and Hajek 2005, Desneux
et al. 2006, Mignault et al. 2006, Gardiner and Landis 2007, Kaiser et al. 2007,
Costamagna et al. 2008, Bahlai and Sears 2009, Frewin et al. 2010, Hallett et al. In prep).
The species composition of the natural enemy guild varies among locations, but in the
eastern portion of the North American range of A. glycines, several coccinellids
(Coccinella septempunctata L. and Harmonia axyridis Pallas), the predatory bug Orius
insidiosus (Say) , and the parasitic wasp Aphelinus certus Yasnosh, are consistently
observed in field surveys when A. glycines is present (Hallett et al. In prep).
Aphis glycines undergoes a very complicated lifecycle. The aphid is heteroecious,
that is, host alternating, and holocyclic, that is, producing viviparous morphs through
most of the spring and summer and a single mating generation in fall. Morph
determination and fecundity of aphids is dependent on density, photoperiod, temperature
and host plant quality, and all of these factors may interact with each other to moderate
their relative influence (De Barro 1992). The strength of density dependence in aphid
population dynamics and morph production is, in general, poorly understood (Newman
et al. 2003). Models describing the phenology and population ecology of aphids must
incorporate these factors in order to develop realistic predictions.
83
Population dynamics of A. glycines have been previously by Onstad et al. (2005),
however, this model was derived entirely empirically, did not account for natural enemy
populations, and did not incorporate environmental cues nor phenological cues from host
plants. As cautioned by its authors, the applicability of this model is limited to A.
glycines occurring in fields in southern Illinois, and only in late July through August.
Thus, development of a model for the population ecology and phenology of A. glycines,
which predicts the dynamics and life history of this organism over its entire lifecycle and
which can be generalized to multiple geographic areas, is warranted. A mechanistic
model for A. glycines would allow the integration of the existing literature to develop
insights into its biology in North America.
DYMEXTM (Hearne Scientific Software Pty Ltd, South Yarra, Australia), is a
mechanistic, lifecycle-based population modeling software package. Mechanistic
population models constructed in DYMEX have been developed for a variety of
applications, such as to model the adult emergence events of a crucifer pest (Hallett et al.
2009), the dynamics between a pathogen and its host crop (White et al. 2004), insectpathogen-crop dynamics under climate change conditions (Griffiths et al. 2010), to
predict establishment and spread of an invasive fruit fly with respect to availability of
foodplant resources (Garcia Adeva et al. 2012), and to examine the effect of
environmental conditions on the feeding efficiency and potential for non-target effects of
an introduced herbivorous biocontrol insect (Kriticos et al. 2009).
This chapter describes a tritrophic model for A. glycines, soybean, and three
natural enemy taxa. Mortality of A. glycines due to predation by natural enemies is
modeled using the NEU, facilitating the adjustment of the model to include additional
84
natural enemy species. This model is used to forecast dynamics of these species under
varied environmental and agronomic conditions.
4.2 Model specification
The model uses a one-day time step in computing all parameters, and requires
user input of meteorological data and latitude. The model consists of five interacting
species lifecycle sub-models, with inputs from external modules calculating
environmental parameters, NEUs and density dependent factors (such as fecundity and
consumption rates of natural enemies) (Fig. 4-1). The lifecycle sub-models included
‘soybean,’ ‘aphid,’ ‘coccinellid,’ ‘orius,’ and ‘wasp.’ Schematics of lifecycle sub-models
specifying all life stages for soybean, the natural enemies, and aphids are given in Figs. 42, 4-3, and 4-4 respectively. The ‘soybean’ model is highly user customizable because of
wide variation in phenology between varieties and maturity groups of soybean, but the
default values used in the model are representative of the average phenology of varieties
typically used in Ontario (Cara McCreary, University of Guelph Soybean Breeding
Program, personal communication). The natural enemy guild was broken into three
groupings to represent the dominant members of this group. The ‘coccinellid’ model is
based on the phenology of Coccinella septempunctata, because the environmental control
of the phenology of this species is well-documented in the literature, though some aspects
of Harmonia axyridis phenology are incorporated. The ‘orius’ model is based on the
biology of the predatory bug Orius insidiosus. The ‘wasp’ model is based on the parasitic
wasp Aphelinus certus. These four species dominate the natural enemy guild in the
eastern North American range of A. glycines (Hallett et al. In prep), but the model can be
85
altered to incorporate additional natural enemy species occurring at different locations.
The parameterization of the lifecycle submodels are given in Tables 4-1 to 4-5, and a
description of the computation of all model inputs is given below.
Figure. 4-1. Schematic of biotic model components. The model consists of five lifecycle submodels
‘soybean,’ ‘aphid,’ ‘coccinellid,’ ‘orius,’ and ‘wasp’ interacting with each other through external
calculations of soybean aphid density measures and the natural enemy unit, as well as with
environmental conditions.
86
Figure. 4-2. Schematic of ‘soybean’ submodel. Stages vulnerable to feeding by soybean aphid are
shaded. Labels below life stages correspond to soybean developmental stages as described by
Pedersen (2009).
Figure. 4-3. Schematic of natural enemy submodels. Predatory life stages (shaded) were used in the
computation of Natural Enemy Units (NEUs) acting on aphid populations. Because adult parasitic
wasps are rarely observed in the field, wasp mummies were used in the computation of observable
NEUs, which were used to validate the model with field data.
87
Figure. 4-4. Schematic of ‘aphid’ submodel. For clarity, aphid life stages are always referred to
followed by the host on which they originated (in brackets). Note that ‘environmental
conditioning’ is a dummy life stage because the number of paths a lifecycle can take when leaving a
given life stage is limited to two by software.
Each lifecycle submodel consisted of the life stages of a given organism, with lifestage
specific parameters governing mortality, development and either stage transfers (i.e.: egg
to larva, larva to pupa, etc.) or fecundity and progeny production rate (i.e.: adult to egg or
nymph). Unless otherwise noted, lifecycle submodels used cohort-based stage transfers
using a step function , i.e.
where SH is the step height and Th is the threshold. Stage transfers used a step
height of 0.75 (i.e.: once a condition is met, 75% of the population will transfer to the
88
next life stage at each time step; if the condition continues to be satisfied at the next time
step, 75% of the remaining population will make the stage transfer, and so on), thus:
where X(t) is the population in a given stage at time t. This approach introduces
some variation in response of simulated populations to environmental conditions, i.e. not
all individuals in a cohort will transfer on the same day. Daily progeny production for all
insects in reproductive life stages was also modeled using a step function (equation 4-1),
with the threshold occurring at the physiological or chronological age at which
oviposition is first observed, and a step height equal to the maximum number of progeny
which can be produced in one day. Unless the literature indicated otherwise, a
development threshold of 10oC was used for all organisms to accrue physiological age. A
sex ratio of 1:1 was assumed for all natural enemy species for reproductive purposes, and
it was assumed populations of apterae conditioned to produce sexual morphs produced
males and gynoparae at a 1:1 rate.
Several mathematical functions were used to describe various aspects of
organismal biology. These will be described in general below, and constants used in the
model are specified in the lifecycle parameterization tables (4-1 to 4-5). A linear
function,
where m is the slope and b is the intercept, was used to incorporate functions or
values computed elsewhere in the model.
A linear-above-threshold function (LAT),
89
where T0 is the lower threshold, was used to model biological parameters
occurring above a threshold such as heat stress and temperature dependant development.
A linear-below-threshold function (LBT),
where T1 is the upper threshold, was used to model aspects of biology occurring
below a threshold, such as cold stress.
A linear-between-threshold function (LBtwT),
was used to model ecological parameters affected by inputs between two
thresholds, such as density dependence in progeny production.
For situations requiring the combination of parameters (ie. multiple sources of
mortality or multiple factors affecting progeny production) acting simultaneously, one of
two combination rules had to be applied. A product combination rule,
was used to combine factors 1, 2, ..., n in most cases (e.g. multiplying a maximum
daily progeny production rate by a function describing the effect of an environmental
condition on progeny production). A compliment product rule,
was used to combine mortality factors 1, 2, ..., n, as survivorship (not mortality) is
the relevant measure when combining individual mortality factors because within the
90
Dymex framework it is assumed that each subsequent mortality condition acts on the
survivors of the previous condition (Maywald et al. 2007).
Environmental parameters were computed based on user-input meteorological and
location data. A ‘Daylength’ module computed the daily photoperiod based on latitude
and day of year, and a separate module computed scotoperiod (=24h - photoperiod) . A
‘Circadian’ module was used to compute the daily temperature cycle (daily cycle) using
daily maximum and minimum values and a composite sine+exponential daily
temperature module, computed in 24 hourly increments. Average daily temperature
(average temperature) was computed from the numerical average of the daily maximum
and minimum temperatures. Seven day running mean daily temperature, seven day
running mean minimum temperature, and seven day running mean maximum
temperature were computed for each time step using the numerical mean of the value
for mean, minimum and maximum temperature, respectively, for the current time step
and the six previous.
In order to incorporate density-dependent factors and relevant density outputs that
could be related to field conditions into the model, a number of computations had to be
performed outside the individual lifecycle sub-modules. For each time step, these
computations were performed prior to calculations within the lifecycle sub-models, thus
it is relevant to note that outputs associated with these calculations are based on the
populations occurring on the day before the time step being reported. For the first
calculation, all of these functions are set to their default value of 0, and after one
iteration, they are updated to include relevant outputs from the lifecycle sub-models. The
total vulnerable aphids, that is, the number of non-alate aphids occurring on both
91
buckthorn and soybean that are vulnerable to predation, was defined as the sum of all
nymphs, apterae and oviparae occurring on buckthorn, and nymphs, apterae and apterae
conditioned to produce sexuals occuring on soybean. Alates and gynoparae were
excluded from this calculation because it was assumed that these morphs were less
vulnerable to predation, due to their increased mobility, and eggs were excluded because
many causes of egg mortality are not well understood. Total vulnerable aphids on
soybean was defined as the sum of all nymphs, apterae and apterae conditioned to
produce sexuals occurring on soybean. Total soybean was defined as the total number
of soybean plants in all life stages, while total vulnerable soybean was defined as the
total number of plants in vulnerable life stages (vegetative, reproductive, and pod stages).
Proportion mature soybean was defined as the number of mature soybean plants
divided by total soybean; proportion vulnerable soybean was defined as total
vulnerable soybean divided by total soybean. Vulnerable aphids per soybean plant was
computed by dividing total vulnerable aphids on soybean by total soybean. For the
purposes of model calibration and validation, an additional variable four day running
mean aphids per soybean plant was also generated to minimize the effect of short-term
population fluctuations predicted by the model.
The natural enemy unit (NEU)is defined as :
where N is the total number of natural enemy species, ni is the total number of
individuals of natural enemy species i observed on 10 plants, and Vi is the average
voracity of natural enemy species i, that is, the number of pest insects it can kill in 24 h
92
divided by 100 (Chapter 5). Voracities used by Hallett et al. (in prep) and in Chapter 5
were used to weight the individual species in the model NEU computation:
NEUs were computed in two ways: total NEUs and observable NEUs, because although
parasitic wasp adults are the stage which attacks and kills aphids, mummies (i.e. dead
aphids in which larval parasitic wasps develop) are the stage most frequently recorded in
field surveys. Total NEUs was calculated using the number of wasp adults, and
observable NEUs was calculated using the number of mummies. NEUs per plant, a
variable directly comparable to field survey data, was computed by dividing observable
NEUs by total soybean. As with vulnerable aphids per plant, an additional variable four
day running mean NEU per soybean plant was also generated using observable NEUs
to minimize the effect of short-term population fluctuations on model calibration. All
subsequent calculations involving NEU refer to total NEUs, because this is the relevant
measure for mortality and density-dependent calculations. Vulnerable aphids per NEU
was computed by dividing total vulnerable aphids by total NEUs.
Absolute mortality due to predation, that is, the total number of aphids consumed
by the natural enemy guild in a given time step, was calculated under the assumption that
the voracity of a natural enemy is a function of the total available prey in the ecosystem.
The maximum possible predation by the natural enemy guild is 100 aphids per NEU, or
100 x total NEUs, but when there are few prey per NEU or prey and NEU are both low,
the mortality due to predation will be lower than the theoretical maximum. This equation
is based on the functional response work of Xue et al. (2009), Frewin et al. (2010) and
Hallett et al (in prep), in which individual natural enemy species all exhibited a type II
functional response. A Type II two functional response is usually modeled using a
93
rectangular hyperbola function, however, this was found to perform poorly at very low
prey densities. Instead a Type III functional response, which can be approximated using a
sigmoid function, was used. A Type III functional response behaves very much like a
Type II functional response at higher prey densities but allows for decreased foraging
efficiency at lower prey densities, which can be approximated using a sigmoid function.
A sigmoid function was used to describe estimated mortality due to predation:
but this expression was found to have poor performance at low values of vulnerable
aphids to NEU or vulnerable aphids per soybean plant, so this expression was nested into
another expression designed to constrain mortality under these conditions. This
expression was called absolute mortality due to predation and takes the form of several
nested if-then-else statements which can be expressed as:
94
if total NEUs ≤ 0, mortality due to predation is zero, else:
then if vulnerable aphids per soybean plant >25
then if the maximum possible predation 100× total NEU < 90% total
vulnerable aphids,
then absolute mortality due to predation is 100× total NEUs,
else absolute mortality due to predation is limited to 90% of the
total vulnerable aphids
else if estimated mortality due to predation < total vulnerable aphids ×
vulnerable aphids per soybean plant ÷ 100
then absolute mortality due to predation is equal to estimated
mortality due to predation
else absolute mortality due to predation is limited to total
vulnerable aphids × vulnerable aphids per soybean plant ÷ 100
Proportional predation mortality risk, that is, the risk of mortality due to
predation to a given life stage i, was computed as follows:
where Ni is the total number of aphids in vulnerable life stage i during that time
step. This calculation assumes no preference for a given morph or life stage by natural
enemies, and was computed for nymphs, apterae and oviparae occurring on buckthorn,
and nymphs, apterae and apterae conditioned to produce sexuals occurring on soybean.
95
Because parasitic wasps are directly dependent on vulnerable aphids for oviposition
sites, wasp egg survivorship had to be constrained by the availability of unparasitized
aphids. Thus, a wasp egg establishment mortality function was created:
Table 4-1: Description of functions governing developmental processes in ‘soybean’ submodel. All
processes are continuous (i.e. applied to the relevant life stage at each time step), unless noted as an
establishment process. Establishment processes were applied to a given cohort once, upon entry into
the relevant lifestage. See Fig. 4-2 for a schematic of the soybean lifecycle.
Function Driving
Process
Life stage
used1
variable
Parameters1
Notes
To = 10oC, m=1
A 10oC development
threshold was used for all
soybean life stages .
Development
Seed, vegetative,
reproductive, pod
LAT
Daily cycle
Stage transfer
Seed to vegetative
Step
Chronological Th=10 d, SH=0.75 Threshold based on average
age
time from planting to
emergence for cultivars used
in southwestern Ontario. 2
Reproductive to pod
Step
Pod to maturity
Step
Vegetative to
reproductive
Step
Chronological
age
Chronological
age
Physiological
age
1
Th=43, SH=0.75 As above.
Th=18, SH=0.75 As above.
Th=587 DD,
SH=0.75
Soybean flowering is both
photoperiod and temperature
dependent, but because
photoperiodic response
varies so greatly between
cultivars, a threshold of 587
DD, based on the average
degree-day requirements for
several cultivars (Kumar et
al. 2008), was used to
initialize model.
LAT = linear-above-threshold, equation 4-4 in text; To= lower threshold, m=slope. For step
functions, SH= step height, Th= threshold.
2
Personal communication, C. McCreary, University of Guelph Soybean breeding program
96
Table 4-2: Description of functions governing developmental processes in ‘coccinellid’ submodel. All
processes are continuous (i.e. applied to the relevant life stage at each time step), unless noted as an
establishment process. Establishment processes were applied to a given cohort once, upon entry into
the relevant lifestage. See Fig. 4-3 for a schematic of the coccinellid lifecycle.
Function
Driving
Process
Life stage
used1
variable
Parameters1 Notes
Development
Stage transfer
Cold stress
mortality
Random
mortality
Egg
LAT
Daily cycle
To = 6.8oC, m=1 Development threshold for C.
septempunctata eggs (Obrycki
and Tauber 1981).
Larva
LAT
Daily cycle
To = 12.6oC, m=1 Average development
threshold for four larval instars
of C. septempunctata
(Obrycki and Tauber 1981).
Pupa
LAT
Daily cycle
To = 12.1 oC,
m=1
Development threshold for C.
septempunctata pupae
(Obrycki and Tauber 1981).
Egg to larva
Step
Physiological Th=50.4DD,
age
SH=0.75
Degree day requirements for
egg hatch of C.
septempunctata (Obrycki and
Tauber 1981).
Larva to pupa
Step
Physiological Th=104.9DD,
age
SH=0.75
Total degree day requirements
all four larval instars of C.
septempunctata (Obrycki and
Tauber 1981).
Pupa to adult
Step
Physiological Th=50.7DD,
age
SH=0.75
Degree day requirements for
pupal eclosion of C.
septempunctata (Obrycki and
Tauber 1981).
Egg, larva, pupa
LBT
7 day running T1 = 4 oC, m=mean
0.111
minimum
temperature
Coccinellids overwinter as
adults, so assume temperatures
under 4oC are sub optimal for
immature stages, and that
complete mortality occurs at
-5oC.
Adult
LBT
7 day running T1 = 4 oC,
mean
m=-0.0204
minimum
temperature
Assume that though conditions
under 4oC are sub optimal,
adults will have greater
tolerance for cold conditions
than larvae, and that complete
mortality does not occur until
adults are exposed to -40 oC.
Egg
Constant
-
Eggs have high rates of sibling
cannibalism and a portion of
each brood are unfertilized and
do not develop. C.
eptempunctata had an egg
mortality rate of 21.4% (Banks
1956).
0.214
97
Process
Life stage
Function
used1
Driving
variable
Random
mortality
Larva, pupa
Step
Chronological Th=2d, SH=0.03 Up to 6% mortality was
age
observed for coccinellid larvae
and pupae over 48 hours in the
field (Schellhorn and Andow
1999).
Adult
Step
Chronological Th=1d, SH=0.01 Assume 1% random mortality
age
per day for adults.
Mortality due to Egg
soybean
senescence
Linear
Proportion
mature
soybean
m=1, b=0
As soybean senesces, eggs will
fall off with leaves and fewer
oviposition sites will be
available.
Fecundity
Adult
Constant
-
32
Total fecundity for C.
septempunctata was up to 65
eggs per female (Banks 1956).
Assume a 1:1 sex ratio.
Progeny
production
Adult
Step
Chronological Th=7.4d, SH=9 H. axyridis had a preage
oviposition interval of 7.4d and
laid, on average, 18 eggs per
day (Lanzoni et al. 2004).
Assume 1:1 sex ratio.
Adult
LBtwT
Vulnerable
To =0, T1=50,
aphids per
m=0.02
soybean plant
Assume coccinellids will not
lay eggs in the absence of
food, and assume progeny
production rate is proportional
to number of vulnerable aphids
present in ecosystem, with
maximum progeny production
rate reached at 50 aphids per
soybean plant.
Adult
Step
Proportion
vulnerable
soybean
Th=0.5, SH=1
Assume vulnerable soybean
plants are needed to provide
oviposition sites for
coccinellids.
Adult
LBtwT
Daily cycle
To =13.3, T1=30, Threshold for ovarian
m=0.05999
development in adult C.
septempunctata is 13.3oC, with
an optimum reached at
approximately 30oC (Phoofolo
et al. 1995). Assume progeny
production is linearly
proportional to ovarian
development rate.
1
Parameters1
Notes
LAT = linear-above-threshold, equation 4-4 in text; To= lower threshold, m=slope. LBT= Linearbelow-threshold, equation 4-5 in text, T1= upper threshold, m=slope. LBtwT= Linear-betweenthresholds, equation 4-6 in text, To= lower threshold, T1= upper threshold, m=slope. for step
functions, SH= step height, Th= threshold; for general step functions, SH o= step height before,
SH1=step height after, Th=threshold
98
Table 4-3: Description of functions governing developmental processes in ‘wasp’ submodel. All
processes are continuous (i.e. applied to the relevant life stage at each time step), unless noted as an
establishment process. Establishment processes were applied to a given cohort once, upon entry into
the relevant lifestage. See Fig.4-3 for a schematic of the wasp lifecycle.
Function
Driving
Process
Life stage
used1
variable
Parameters1 Notes
Development
Egg
LAT
Daily cycle
To = 9.1, m=1
Mummy
LAT
Daily cycle
To =11.6oC, m=1 Thermal threshold for A. certus
mummy development is
11.6oC (Frewin et al. 2010).
Egg to mummy
Step
Physiological Th=96DD,
age
SH=0.75
Degree day accumulation
required for mummy formation
in A. certus is 96 DD (Frewin
et al. 2010).
Mummy to adult
Step
Physiological Th=90DD,
age
SH=0.75
Degree day accumulation
required for adult eclosion
from mummy in A. certus is 90
DD (Frewin et al. 2010).
Cold stress
mortality
Egg, mummy
LBT
7 day running T1 = 10oC, m=mean
0.1
minimum
temperature
It is unknown how A. certus
overwinters, so for this model,
it is assumed the overwintering
form is the adult and cold
stress only affects immature
forms. The development rate
of A. certus immature drops
considerably below 10oC, and
presumably cannot survive at
temperatures below 0oC (A.
Frewin, personal
communication).
Heat stress
mortality
Egg, mummy, adult
LAT
7 day running To = 32 oC,
mean
m=0.056
maximum
temperature
In laboratory colonies, all life
stages of A. certus do poorly at
temperatures above 32oC (A.
Frewin, personal
communication). Assume
100% mortality at 50oC.
Random
mortality
(establishment
process)
Egg
Direct
Wasp egg
establishment
mortality
Computed using equation (413) in text. Constrains
survivorship of wasp eggs so
that new eggs will not survive,
if more than 90% of the
vulnerable aphid population is
parasitized.
Mummy
Constant
-
Approx. 12% of mummies
never produced adult A. certus
wasps (Frewin et al. 2010).
Stage transfer
0.12
99
Thermal threshold for egg
development of A. certus is
9.1oC (Frewin et al. 2010).
Function
used1
Driving
variable
Old age mortality Adult
General step
Chronological SHo=0.01,
age
SH1=0.90,
Th=10d
Adult A. certus live 10-14 days
(A. Frewin, personal
communication). Assume
random mortality of 1% per
day until a wasp is 10d old,
then 90% mortality per day
thereafter.
Fecundity
Adult
Constant
-
Determined empirically in
calibration process. The upper
ceiling on total fecundity of A.
certus is unknown (A. Frewin,
personal communication).
Progeny
production
Adult
Step
Chronological Th=1d, SH=10
age
Functional response
experiments for A. certus were
initiated 24 h (1d) after
eclosion, and females
produced, on average, 20 eggs
per day (Frewin et al. 2010).
Assuming a 1:1 sex ratio, 10
eggs per adult can be produced
daily.
Adult
LBtwT
Vulnerable
To =0, T1 =50,
aphids per
m=0.02
soybean plant
Wasps cannot lay eggs in the
absence of food; assume
progeny production rate is
proportional to number of
vulnerable aphids present in
ecosystem, with maximum
progeny production rate
reached at 500 aphids per
soybean plant.
Adult
LBtwT
Daily cycle
Process
Life stage
1
Parameters1
100
Notes
To =15, T1 =30, A. certus oviposition tends to
m=0.067
occur between 15 and 30oC
and is temperature dependent
(A. Frewin, personal
communication).
LAT = linear-above-threshold, equation 4-4 in text; To= lower threshold, m=slope. LBT= Linearbelow-threshold, equation 4-5 in text, T1= upper threshold, m=slope. LBtwT= Linear-betweenthresholds, equation 4-6 in text, To= lower threshold, T1= upper threshold, m=slope. for step
functions, SH= step height, Th= threshold; for general step functions, SH o= step height before,
SH1=step height after, Th=threshold
100
Table 4-4: Description of functions governing developmental processes in ‘orius’ submodel. All
processes are continuous (i.e. applied to the relevant life stage at each time step), unless noted as an
establishment process. Establishment processes were applied to a given cohort once, upon entry into
the relevant lifestage. See Fig. 4-3 for a schematic of the orius lifecycle.
Function Driving
Process
Life stage
used1
variable
Parameters1 Notes
Development
Egg
LAT
Daily cycle
To = 10.2, m=1
Nymph
LAT
Daily cycle
To =13.7oC, m=1 Development threshold
computed using data presented
in Isenhour and Yeargan
(1981).
Egg to nymph
Step
Physiological Th=73DD,
age
SH=0.75
Degree day requirements
computed using data presented
in Isenhour and Yeargan
(1981).
Nymph to adult
Step
Physiological Th=145DD,
age
SH=0.75
Degree day requirements
computed using data presented
in Isenhour and Yeargan
(1981).
Cold stress
mortality
Egg, nymph
LBT
7 day running T1 = 10oC, m=mean
0.1
minimum
temperature
O. insidiosus overwinters as an
adult so it is assumed
immature forms will be
affected by cold stress.
McCaffrey and Horsburgh
(1986) present a lower
estimate of developmental
threshold for O. insidiosus, so
it is assumed cold stress
accumulates below this
temperature and , as in the
wasp model, complete
mortality occurs at 0oC.
Heat stress
mortality
Egg, nymph, adult
LAT
7 day running To = 32 oC,
mean
m=0.056
maximum
temperature
Though no records exist of O.
insidiosus suffering from heat
stress in temperate climates,
several studies examine O.
insidiosus biology at constant
temperatures and do not report
data from temperatures above
32oC(Isenhour and Yeargan
1981, McCaffrey and
Horsburgh 1986). thus
assumed that O. insidiosus
begins to suffer from heat
stress at 32oC and complete
mortality occurs at 50oC, as in
the wasp submodel.
Stage transfer
101
Development threshold
computed using data presented
in Isenhour and Yeargan
(1981).
Process
Life stage
Function
used1
Driving
variable
Parameters1
Notes
Random
mortality
(establishment
process)
Nymph
Constant
-
0.03
Approximately 3% of
immature O. insidiosus do not
reach adulthood due to random
mortality (Kiman and Yeargan
1985).
Old age mortality Adult
General step Chronological SHo=0.03,
age
SH1=0.95,
Th=40d
Female O. insidiosus lived up
to 40 days (Kiman and
Yeargan 1985). Assume 3%
random mortality before and
95% mortality after 40th day of
life.
Fecundity
Adult
Constant
-
A maximum of 100 eggs were
laid per female when fed an
optimal diet (Kiman and
Yeargan 1985). Assume 1:1
sex ratio, resulting in a net
fecundity of 50 eggs per adult.
Progeny
production
Adult
Step
Chronological Th=1d, SH=1
age
Approximately two eggs per
day were produced per female
(Kiman and Yeargan 1985), so
a net of one egg per day can be
produced by each adult.
Adult
LBtwT
Vulnerable
To =0, T1 =1,
aphids per
m=1
soybean plant
Assume Orius will not lay
eggs in the absence of food,
but because of lower fecundity
rate compared to coccinellids
and wasps, maximum progeny
production rate is reached at 1
aphid per soybean plant.
Adult
LBtwT
Daily cycle
To =15, T1 =30, As in the wasp model, assume
m=0.067
progeny production by Orius is
thermally-dependent between
15 and 30oC.
Adult
General step Scotoperiod
SHo=1, SH1=0, Adult female O. insidious
Th=10.05h
collected in the Guelph area
had entered reproductive
diapause after Aug 15
(Schmidt et al. 1995),
corresponding to a scotoperiod
of 10.05h.
1
50
LAT = linear-above-threshold, equation 4-4 in text; To= lower threshold, m=slope. LBT= Linearbelow-threshold, equation 4-5 in text, T1= upper threshold, m=slope. LBtwT= Linear-betweenthresholds, equation 4-6 in text, To= lower threshold, T1= upper threshold, m=slope. for step
functions, SH= step height, Th= threshold; for general step functions, SH o= step height before,
SH1=step height after, Th=threshold
102
Table 4-5: Description of functions governing developmental processes in ‘aphid’ submodel. All
processes are continuous (i.e. applied to the relevant life stage at each time step), unless noted as an
establishment process. Establishment processes were applied to a given cohort once, upon entry into
the relevant lifestage. See Fig. 4-3 for a schematic of the aphid lifecycle. For each life stage, host plant
where the aphid originated is given in brackets. Note that because Dymex TM limits the number of
stage transfer possibilities for a given life stage to two, ‘nymph: environmental conditioning
(soybean)’ is a dummy life stage to allow nymphs (soybean) to become apterae (soybean), alates
(soybean) or apterae conditioned to produce sexuals (soybean)
Function Driving
Process
Life stage (host)
used1
variable
Parameters1 Notes
Development
Stage transfer
LAT
Daily cycle
To =10oC, m=1
Nymph (buckthorn),
LAT
nymph (soybean), apterae
conditioned to produce
sexuals (soybean),
gynoparae (soybean),
oviparae (buckthorn)
Daily cycle
To =9.5oC, m=1 Development threshold for A.
glycines nymphs (Hirano et al.
1996).
Diapause egg (buckthorn) Step
to spring egg (buckthorn)
Chronological Th=120d,
age
SH=0.75
Eggs of A. glycines which
were laid before early
November did not hatch when
exposed to warm conditions
until late February of the
following year, suggesting an
obligate chilling period of
approx. 120d (Bahlai et al.
2007).
Spring egg (buckthorn) to Step
nymph (buckthorn)
Physiological Th=54DD,
age
SH=0.75
A degree day accumulation of
54 DD was required for egg
hatch of A. glycines (Bahlai et
al. 2007).
Nymph (buckthorn) to
apterae (buckthorn)
Physiological Th=57.1DD,
age
SH=0.75
A degree day accumulation of
57.1 DD was required for
nymphs A. glycines to mature
(Hirano et al. 1996).
Nymph (buckthorn) to alate Step
(buckthorn)
Physiological
age and
Vegetative
soybean: total
number
Nymph (soybean) to
nymph: environmental
conditioning (soybean)
Physiological Th=57.1,
age
SH=0.75
As above, but DD
requirements rounded down to
allow this condition to be met
prior to the condition
governing transition to apterae
(buckthorn). Second condition
allows alates to be produced
only after soybean has begun
to emerge from ground.
.
A degree day accumulation of
57.1 DD was required for
nymphs A. glycines to mature
(Hirano et al. 1996).
Spring egg (buckthorn)
Step
Step
103
Th=57 DD,
SH=0.75
and
Th=1, SH=0.95
Development threshold for A.
glycines eggs (Bahlai et al.
2007).
Function Driving
used1
variable
Parameters1
Notes
Th=57, SH=0.75
and
SHo=0.05,
SH1=0.99,
Th=4000
As above, but DD
requirements rounded down to
allow this condition to be met
prior to the condition
governing transition to nymph:
environmental conditioning
(soybean). Second condition
makes formation of alates
density-dependent. When
aphid density is less than 4000
aphids per plant, <5% of
nymphs become alates
(Donaldson et al. 2007).
Process
Life stage (host)
Stage transfer
Nymph (soybean) to alate Step
(soybean)
and
general
step
Physiological
age and
Vulnerable
aphids per
soybean plant
Nymph: environmental
Step
conditioning (soybean) to
apterae (soybean)
Chronological Th=1d, SH=0.95 Dummy stage to allow
age
conditions for apterae
conditioned to produce sexuals
production to be met.
Nymph: environmental
Step
conditioning (soybean) to
apterae conditioned to
produce sexuals
Scotoperiod
Cold stress
mortality
Nymph (buckthorn),
LBT
apterae (buckthorn), alate
(buckthorn), nymphs
(soybean), alate (soybean),
apterae (soybean), apterae
conditioned to produce
sexuals (soybean),
gynoparae (soybean),
oviparae (buckthorn)
7 day running T1 = 10oC, m=mean
0.1
minimum
temperature
Heat stress
mortality
Nymph (buckthorn),
LAT
apterae (buckthorn), alate
(buckthorn), oviparae
(buckthorn)
7 day running To = 27oC,
mean
m=0.043
maximum
temperature
Nymph (soybean), alate
(soybean), apterae
(soybean), apterae
conditioned to produce
sexuals (soybean)
gynoparae (soybean)
7 day running To = 32oC,
mean
m=0.056
maximum
temperature
Density
dependent
mortality
LAT
Nymph (soybean), apterae LAT
(soybean)
Th=10.7h,
SH=0.25
Vulnerable
To = 1000,
aphids per
m=0.0000345
soybean plant
104
Sexual morphs are likely
triggered at a photoperiod of
13.3h and decreasing (=
scotoperiod of 10.7h and
increasing) (Chapter 3).
Assume that aphids are
tolerant to brief periods of
freezing temperatures but
begin to do poorly when
minimum temperatures are
below 10oC and complete
mortality occurs when
minimum daily temperatures
do not exceed 0oC, on average,
for 7 days.
Buckthorn-dwelling morphs of
A. glycines began to die when
exposed to 27oC constant
temperatures (CB, personal
observation). Assume 100%
mortality occurs at 50oC.
Assume soybean-dwelling
morphs have greater heat
tolerance than buckthorndwelling morphs and that they
begin to accumulate heat stress
above 32oC. Assume total
mortality occurs at 50oC.
Typical plant capacity is
around 20000 to 30000 aphids
per soybean plant (C. DiFonzo,
personal communication).
Assume aphids begin to be
affected by crowding at 1000
aphids per plant, and 100%
mortality is reached at 30000
aphids per plant.
Process
Life stage (host)
Function Driving
used1
variable
Parameters1
Notes
Mortality due to Nymphs (soybean),
soybean
alate(soybean), apterae
phenology
(soybean), apterae
conditioned to produce
sexuals (soybean)
Step
Mature
Th=1, SH=0.25 As soybean begins to reach
soybean: total
maturity, assume resident
number
aphids will be negatively
affected.
Mortality due to Nymphs (soybean),
soybean
alate(soybean), apterae
senescence
(soybean), apterae
conditioned to produce
sexuals (soybean)
Step
Proportion
mature
soybean
Th=0.95, SH=1 When soybean reaches full
maturity, all resident aphids
will die.
-
0.70
Random
mortality
(establishment
process)
Diapause egg (buckthorn) Constant
Approx. 70% mortality is
observed in overwintering eggs
(Welsman et al. 2007).
Other mortality Apterae (buckthorn)
factors
Step
Vegetative
Th=1, SH=0.75 Assume once soybean begins
soybean: total
reach reproductive stage,
number
buckthorn is a sub-optimal
host and apterae (buckthorn)
begin to die out.
Apterae (buckthorn)
Step
Day of year
Th=166, SH=1
By June 15 (=166 Julian day),
no A. glycines were observed
on buckthorn (Welsman et al.
2007).
Gynoparae (soybean)
Step
Day of year
Th=300, SH=1
Gynoparae are not permitted to
survive the winter, even if cold
conditions never occur.
Oviparae (buckthorn)
Step
Day of year
Th=310, SH=1
Oviparae are not permitted to
survive the winter, even if cold
conditions never occur
Proportional
predation
mortalityi
m=0.75, b=0
Computed using equation (412). This function accounts for
predation mortality risk at each
life stage i, and must be
multiplied by the total number
of individuals occurring in life
stage i. A linear function,
rather than a direct function,
was used in order to allow user
customization of the degree of
impact a natural enemy guild
has on aphid populations. A
slope of 0.75 was empirically
determined suggesting that
mortality due to predation is
slightly less than predicted.
Predation
mortality
Nymph (buckthorn),
Linear
apterae (buckthorn), nymph
(soybean), apterae
(soybean), apterae
conditioned to produce
sexuals(soybean), oviparae
(buckthorn)
Old age mortality Apterae (buckthorn),
Step
apterae (soybean), apterae
conditioned to produce
sexuals (soybean), oviparae
(buckthorn)
Chronological Th=15d,
age
SH=0.75
105
Adult aphids that have not
flown live approx. 15 d (Zhang
et al. 2009a).
Process
Life stage (host)
Function Driving
used1
variable
Parameters1
Notes
Old age mortality Alate (buckthorn), alate
(soybean), gynoparae
(soybean)
Step
Chronological Th=10d,
age
SH=0.75
Adult aphids that have
engaged in flight live approx.
10 d (Zhang et al. 2009a).
Fecundity
Apterae (buckthorn)
apterae (soybean)
Constant
-
61
Mean fecundity of female A.
glycines at 25oC was 61
progeny (McCornack et al.
2004).
Alate (buckthorn), alate
(soybean), gynoparae
(soybean)
Constant
-
15
A. glycines that had flown had
a reduced total fecundity of 15
progeny per female (Zhang et
al. 2009a).
Alate (buckthorn), alate
(soybean), gynoparae
(soybean)
Constant
-
0.001
Taylor (1974) suggested that
only one in one thousand alate
aphids found a suitable host for
larviposition after leaving their
native host patch.
Apterae conditioned to
Constant
produce sexuals (soybean)
-
30
Mean fecundity of female A.
glycines at 25oC was 61
progeny (McCornack et al.
2004). Assuming apterae
conditioned to produce sexuals
produce gynoparae and males
at 1:1 ratio, approx. 30
gynparae are produced by each
aptera on average.
Oviparae (buckthorn)
-
5
A. glycines produces 4 to 5
eggs per ovipara when
occurring on common
buckthorn, R. cathartica (Yoo
et al. 2005).
Progeny
production
Constant
Apterae (buckthorn), alate Step
(buckthorn), alate
(soybean), apterae
(soybean), apterae
conditioned to produce
sexuals (soybean)
Chronological Th=1d, SH=7
age
McCornack et al (2004) found
a reproductive period of 9d,
and a total fecundity of 61 for
A. glycines, suggesting that
aphids can produce approx 7
progeny per day.
Apterae conditioned to
Step
produce sexuals (soybean),
gynparae (soybean)
Physiological Th=57 DD,
age
SH=7
As above, but including
physiological time to develop
from nymphs (Hirano et al.
1996), as nymphal stages of
these two morphs are not
explicitly considered in the
model.
Oviparae (buckthorn)
Physiological Th=57 DD,
age
SH=0.5
Estimate progeny production
rate as one every two days per
ovipara. Function includes
physiological time to develop
from nymphs (Hirano et al.
1996), as nymphal stages of
oviparae are not explicitly
considered in the model.
Step
106
Process
Life stage (host)
Function Driving
used1
variable
Progeny
production
Apterae (soybean)
LBT
Apterae (buckthorn), alate LBtwT
(buckthorn), alate
(soybean), apterae
(soybean), apterae
conditioned to produce
sexuals (soybean),
gynoparae (soybean),
oviparae (buckthorn)
Oviparae (buckthorn)
Step
Parameters1
Notes
Vulnerable
T1 =20000, m=- Assume fecundity is density
aphids per
0.00005
dependent, and that typical
soybean plant
plant capacity is around 20000
to 30000 aphids per soybean
plant (C. DiFonzo, personal
communication). Assume full
production at 0 aphids per
plant, and practically none at
20000.
Daily cycle
To =10, T1 =26, Progeny production is
m=0.0625
temperature dependent for all
reproductive life stages.
Assume the lower threshold
for progeny production is
10oC, the lower development
threshold, and peaks at 26oC,
the optimum temperature for
development amongst summer
morphs (McCornack et al.
2004, Bahlai et al. 2007).
Day of year
Th=274, SH=1
In surveys, eggs of A. glycines
have never been observed
before Oct 1 (=274 day of
year) (CB, personal
observation).
1
LAT = linear-above-threshold, equation 4-4 in text; To= lower threshold, m=slope. LBT= Linearbelow-threshold, equation 4-5 in text, T1= upper threshold, m=slope. LBtwT= Linear-betweenthresholds, equation 4-6 in text, To= lower threshold, T1= upper threshold, m=slope. for step
functions, SH= step height, Th= threshold; for general step functions, SH o= step height before,
SH1=step height after, Th=threshold
4.2.1 Model calibration
Because population data for A. glycines and its natural enemy complex are largely
confined to the soybean growing season, with only relative measures of population
density and phenology occurring through much of the aphid’s lifecycle, quantitative
calibration of this model is restricted to that time period. Challenges in validation are
common for process-based simulation models such as these because of a lack of
quantitative, whole-season scouting data (Kriticos et al. 2003), and thus it must be
107
cautioned that interpretation of the results of this model outside the soybean growing
season should be limited to qualitative assertions.
In order to fine tune the parameters used in the model, we used field scouting data
for aphids and their natural enemies obtained in 2007 from fields near Alvinston (42.8oN,
81.9oW) and Shetland, ON (42.7oN, 82.0 oW) (Hallett et al. In prep), and weather data
(maximum and minimum daily temperature and total precipitation) obtained from the
Environment Canada National Climate Archive
(http://www.climate.weatheroffice.gc.ca/). Scouting data consisted of whole-plant counts
of aphids and natural enemies, performed weekly on plants from these two observation
fields. The model was initialized using 500 soybean plants, planted 20 May, and soybean
phenology was tuned to match observed phenology in the field at each site. Aphid and
natural enemy observation data were scaled to aphids or natural enemies per 500 soybean
plants to match the scale of the model. Aphid and natural enemy lifecycle sub-models
were initialized based on the first observation of a given taxon in the field, but it was also
assumed that all taxa had some low level of activity beneath the limits of detection of
scouting (one coccinellid adult, two wasp adults, and one orius adult were programmed to
arrive at the 500 soybean patch every day for the duration of the simulation; this will
henceforth be referred to as ‘background NEUs’; 15 aphid nymphs (soybean) were
deposited in the system each day as well to account for nymphs being produced by alate
moving in from other locales). Because evacuated mummies of A. certus often remain on
a plant after adult eclosion (A. Frewin personal communication), and are often counted in
surveys and given the weekly sampling resolution of the input data, the initialization
stage of the wasp module was partitioned between adult wasp and mummies at an
108
empirically derived, site specific proportion totaling to the number of mummies recorded
on the date they were first observed in the field. Proportions were determined by
simulation and chosen by which best approximated the population growth of wasps
observed in the subsequent sampling week in field data.
Four day running means of aphids and NEU per soybean plant from the model
output were compared to scouting data, and model parameterization was adjusted, as
appropriate, to improve model fit of field data. Four day running means were used to
minimize the effect of fluctuations in predicted values on model calibration. Parameters
adjusted in the calibration process are noted in the model specification tables.
Performance of the model at the Alvinston and Shetland sites after calibration is
given in Figs. 4-5 and 4-6.
4.2.2 Model validation
Aphid and natural enemy scouting data from an observation soybean field near
Arva, ON (43.1oN, 81.3oW), in 2009 were obtained from the Ontario Ministry of
Agriculture, Food and Rural Affairs (C. McCreary, personal communication). These data,
collected using a similar sampling procedure as that for the calibration sites, were used to
initialize the calibrated model to examine its performance at a different site, in a different
growing season, using data collected by another research group. Appropriate weather
data were obtained from Environment Canada, as in the calibration experiments. O.
insidiosus individuals were never observed at this site and so they were excluded from
the ‘background NEUs’ initialization. Performance of the model at this site is given in
Fig. 4-7.
109
110
Slope= 1.31,
F1,8=461.4, p<0.05,
Adj. R2=0.981
Figure 4-5. Model performance at Alvinston calibration site in 2007. A) Predicted (▲) and observed (○) aphid populations by
Julian day; B) predicted vs. observed aphid-per-plant populations , C) predicted (▲)and observed (○) NEUs by Julian day; and
D) predicted vs. observed NEU-per-plant populations . Regression lines were constrained to have a zero intercept.
Slope= 0.98,
F1,8=24.5, p<0.05,
Adj. R2=0.723
111
Slope= 1.90,
F1,7=457.5, p<0.05,
Adj. R2=0.983
Figure 4-6. Model performance at Shetland calibration site in 2007. A) Predicted (▲) and observed (○) aphid populations by
Julian day; B) predicted vs. observed aphid-per-plant populations , C) predicted (▲)and observed (○) NEUs by Julian day; and
D) predicted vs. observed NEU-per-plant populations . Regression lines were constrained to have a zero intercept.
Slope= 1.06,
F1,7=36.9, p<0.05, Adj.
R2=0.818
112
Slope= 0.70,
F1,6=14.8, p<0.05, Adj.
R2=0.663
Figure 4-7. Model performance at Arva site in 2009, used for model validation . A) Predicted (▲) and observed ( ○) aphid
populations by Julian day; B) predicted vs. observed aphid-per-plant populations , C) predicted (▲) and observed ( ○) NEUs by
Julian day; and D) predicted vs. observed NEU-per-plant populations . Regression lines were constrained to have a zero intercept.
Slope= 1.42,
F1,6=144.1, p<0.05,
Adj. R2=0.953
4.2.3 Whole season simulations
After validation, the model was initialized for full-season runs starting on May 1
with 1000 spring eggs on buckthorn, and using a 500-soybean initialization for the habitat
patch, and background NEUs as described in the model calibration section. Soybean
lifecycle parameters were set to their defaults to represent average soybean phenology,
and May 20 was used as a default planting date. A series of simulations was then
performed to determine the effect of growing season (using 2007 and 2009 weather data),
natural enemy density (using background NEUs and 10x background NEUs) and planting
date (using soybean planting dates of May 6, May 20, and June 4) on predicted
populations of aphids on soybean over the growing season and diapause aphid eggs
occurring in winter. Abundance of aphid morphs as predicted by model simulation for the
2007 growing season is presented in Fig. 4-8. Density of A glycines over the growing
season and overwintering egg populations of A. glycines, as predicted by the model for
both growing seasons and as a function of planting date and natural enemy abundance are
given in Figs. 4-9 and 4-10 respectively.
4.3 Discussion
The calibration process resulted in a model that performed reasonably well in
predicting aphid and NEU population growth at both sites (Figs. 4-5 and 4-6). In general,
the model under-predicted NEU density observed at the end of the growing season, which
could be explained by two factors. Firstly, it is possible that density of parasitic wasps is
over-estimated late in the growing season in field surveys. When an adult wasp emerges
113
Figure. 4-8. Abundance of soybean aphid morphs by date as predicted by the model. The model
was initiated on 5 January, using 2007 weather data from an Environment Canada weather station
near London, ON, with 1000 ‘spring eggs’, 500 soybean plants planted on 20 May, and
‘background’ natural enemies (as described in text). All aphid life stages are given on this figure
except for nymphs occurring on buckthorn and soybean. Winged morphs are given in red. Arrow
indicates location of possible second peak of gynoparae activity.
114
115
Figure. 4-9. Aphid density (in vulnerable aphids per plant) over the growing season for three soybean planting dates, two natural enemy
treatments, and weather data from two different growing seasons, as predicted by the model. The model was initiated on 5 January of each
simulation year, and weather data was obtained from from an Environment Canada weather station near London, ON, with 1000 ‘spring eggs’,
and 500 soybean plants Each panel consists of predicted aphid population densities for the planting dates 6 May, 20 May and 4 June, for A)
2007 weather and high NEUs (background NEUs, as described in the text, increased by an order of magnitude); B) 2007 weather and low NEUs
(background NEUs only); C) 2009 weather and high NEUs; and D) 2009 weather and low NEUs. Weather data for 2007 and 2009 were obtained
from an Environment Canada weather station near London, ON.
Figure. 4-10. Abundance of diapause eggs of soybean aphid on December 27 (end of simulation) as
a function of planting date, natural enemy abundance and growing season, as predicted by model.
The model was initiated on 5 January of each simulation year, and weather data was obtained from
from an Environment Canada weather station near London, ON, with 1000 ‘spring eggs’, and 500
soybean plants. Each panel consists of predicted aphid diapause egg abundances for the planting
dates 6 May, 20 May and 4 June at low NEUs (i.e. background NEUs, as described in the text) and
high NEUs (i.e. background NEUs increased by an order of magnitude) after a given growing
season. A) 2007; B) 2009. Weather data for 2007 and 2009 were obtained from an Environment
Canada weather station near London, ON.
116
from the aphid mummy, the aphid mummy may remain on the plant for some time, which
means scouting data later in the growing season may represent a degree of cumulative
counts of parasitized aphids, rather than a time step cohort. Secondly, as aphid density
increases, it is probable that natural enemies occurring in adjacent habitats would move
into soybean fields to feed, so the natural enemy complex at the end of the growing
season likely represents both resident and immigrant populations of these taxa.
The ability of the model to predict both aphid and natural enemy populations
would be enhanced by allowing individuals of all taxa to migrate in and out of a habitat
patch in response to appropriate conditions. Currently, the model does not specifically
account for immigration of aphids, nor their natural enemies, and yet, these events are
likely both common and very influential on population dynamics of these species
(Chapter 3). Future versions of the DYMEX framework will allow spatially-explicit
dispersal patterns to be incorporated into the model (Parry et al. 2011). This development
will enhance model applicability in highly dispersive species like A. glycines by
incorporating spatial dynamics of the species.
The model was reasonably well correlated with the population growth of natural
enemies at the validation site, however, it predicted that populations of A. glycines would
reach much higher numbers than observed (Fig. 4-7). The poor performance of the aphid
model is likely due, in part, to field specific differences in natural enemy abundances,
observer effects leading to systematic under-estimation of natural enemy abundance at
this site, and/or mis-estimation of aphid density. At this site, no O. insidiosus individuals
were recorded over the growing season, yet at sites monitored by our group in the vicinity
that year, O. insidiosus was abundant (Chapter 5, Hallett et al. In prep). Also, in these
117
data, as aphid numbers reached >250/plant, aphid densities were estimated rather than
counted, leading to greater potential for observer bias. Another factor which may have
affected model performance at this site is the likely presence of additional natural enemy
species not included in the present model. In our concurrent research trials occurring in
the vicinity, we observed several additional natural enemy species including predatory fly
larvae (Syrphids, Aphidoletes midges) and lacewing larvae feeding on A. glycines
(Chapter 5, Hallett et al. In prep). Future revisions of this model should include these taxa
to increase precision.
At our calibration sites, it was determined that the proportional predation
mortality likely over-estimated the impact of the natural enemy community on population
growth of A. glycines. Thus, an empirically determined correction factor of 0.75 was
applied when proportional predation mortality was incorporated into the aphid submodel.
This correction factor may reflect the degree of search effort employed in our surveys;
Hallett et al. (In prep) involved considerable energy expenditure on the parts of
technicians to characterize the natural enemy community of A. glycines and may not
reflect survey data that is collected under less controlled scouting conditions. Thus the
correction factor may not need to be applied under all circumstances: for instance, at the
validation site, where we suspect natural enemy counts were systematically underestimated compared to our surveys for the reasons described above.
Deterministic population models such as this one typically have poorer
performance at low population numbers because they ignore demographic stochasticity
(Hardman 1976). In general, stochastic population models are better at predicting
population fluctuations in tritrophic systems than are deterministic population models
118
(Ives and Jansen 1998). Soybean aphids often have patchy distributions in fields (Huang
et al. 1992, Su et al. 1996) and natural enemies may follow similar patterns (Wang et al.
1991), though random distribution is usually observed when aphid populations reach high
densities (Shusen et al. 1994). Nevertheless, the large standard deviation in average aphid
and natural enemy populations at our calibration and validation sites suggest patchy
distribution within soybean fields still occurs through much of the growing season. This
lack of uniform distribution complicates model calibration but may be resolved by
increased sampling. Similarly, model calibration could be improved by increased
temporal resolution in sampling: the model predicts fluctuations in both aphids and NEUs
occurring at periods shorter than one week, and thus the model was calibrated using a
four-day running mean of predicted values compared to field conditions to minimize the
effect of these fluctuations.
The model was used to examine abundance of aphid morphs over the course of
the growing season which suggested a possible a secondary peak of gynoparae occurring
later in the fall (Fig. 4-8). A secondary peak of gynoparae flight activity was observed
under field conditions (Chapter 3), and it was suggested that each peak corresponded to a
different environmental cue, with the early-fall peak most closely linked with degree day
accumulation, and the second more closely linked with photoperiod. This differential
response to cues by gynoparae was not built into the model, and apterae conditioned to
produce sexuals, the morphs that produce gynoparae, follow a similar bimodal activity
distribution earlier in the season, suggesting that conditions leading to the bimodal
activity distribution of gynoparae occur at least one generation before gynoparae are
produced. It is possible that a single unfavourable weather event led to a brief period of
119
suppressed activity for all morphs of A. glycines, and this effect could be passed on to
subsequent generations. A similar pattern was observed for oviparae, which may support
this hypothesis. To test this, simulations should be performed to generate data that can be
used to compare gynoparae production over several growing seasons to data from the
suction trap network.
The model predicts natural enemies have a very important role in overall aphid
suppression, but the effect of plant phenology on aphid phenology seems to be variable,
depending on the growing season (Fig 4-9). In 2007, late growing season population
dynamics of A. glycines were dramatically affected by planting date, but in 2009, very
little variation was predicted to occur in late-season soybeans. Fewer eggs are produced
in all simulations with higher levels of natural enemy suppression (Fig 4-10). Planting
date of soybean had a dramatic impact on the predicted number of aphid eggs produced at
the end of the season in simulations using 2007 weather data, with fewer eggs produced
in simulations using earlier planting dates (Fig. 4-10A) but this effect was not observed
with 2009 weather data (Fig. 4-10B). This result suggests that relative influence of plant
phenology on the phenology and population ecology of A. glycines is highly interactive
with environmental conditions. These responses are difficult to observe in purely
empirical research, because it is impossible to de-couple the effects of environment and
host plant phenology in the field.
Though our model performed reasonably well at predicting population dynamics
of A. glycines and its natural enemies, the number of parameters used in the model limits
its use for prediction of aphid dynamics in a specific region. Complex simulation models
accounting for a wide range of parameters are often more realistic than simpler models of
120
population dynamics, and often can be used to more accurately reflect field conditions
(Onstad 1988). However, model precision and applicability may be compromised by an
over-parameterized model (Cox et al. 2006), indeed, several authors have made extensive
arguments for the simplification of models in biological systems (Peck 2004, PilkeyJarvis and Pilkey 2008). Models of biological processes including a large number of
parameters can give a perception of precision, however, each of these parameters will
have some degree of imprecision or variation, and models are not typically robust to
withstand large or unusual disturbances (Pilkey-Jarvis and Pilkey 2008). Nevertheless,
models such as ours are useful for qualitative predictions in isolated populations (e.g.
relative overwintering populations of aphid eggs as a function of planting date).
The model described in this paper represents an integration of the available
literature on A. glycines and its dominant natural enemy taxa in eastern North America.
Future iterations of this model should include additional natural enemy taxa to increase
generalizability, and should incorporate factors accounting for the movement of all
species into and out of a given habitat patch, and all parameters used in the model should
be subjected to rigorous sensitivity analyses to maximize parsimony
4.4 Acknowledgements
The authors would like to thank Darren Kriticos (CSIRO Australia) for advice and
comments offered during the drafting of this model, Cara McCreary and Tracey Baute (U
of Guelph Soybean Breeding Program and Ontario Ministry of Agriculture, Food and
Rural Affairs) for help parameterizing the soybean model, and Andrew Frewin (U of
121
Guelph) for extensive conversation about the biology of A. certus. CB was funded by a
Natural Sciences and Engineering Research Council of Canada PGS-D3 fellowship.
122
CHAPTER 5
Choosing organic pesticides over synthetic pesticides may not
effectively mitigate environmental risk in soybeans 7
5.0 Abstract
Background: Selection of pesticides with small ecological footprints is a key factor in
developing sustainable agricultural systems. Policy guiding the selection of pesticides
often emphasizes natural products and organic-certified pesticides to increase
sustainability, because of the prevailing public opinion that natural products are
uniformly safer, and thus more environmentally friendly, than synthetic chemicals.
Methodology/Principal Findings: We report the results of a study examining the
environmental impact of several new synthetic and certified organic insecticides under
consideration as reduced-risk insecticides for soybean aphid (Aphis glycines) control,
using established and novel methodologies to directly quantify pesticide impact in terms
of biocontrol services. We found that in addition to reduced efficacy against aphids
compared to novel synthetic insecticides, organic approved insecticides had a similar or
even greater negative impact on several natural enemy species in lab studies, were more
detrimental to biological control organisms in field experiments, and had higher
Environmental Impact Quotients at field use rates.
Conclusions/Significance: These data bring into caution the widely held assumption
7
Originally published as: Bahlai , C.A., Y. Xue, C. McCreary, A.W. Schaafsma and R.H. Hallett. 2010.
Choosing organic pesticides over synthetic pesticides may not effectively mitigate environmental risk in
soybeans. PLoS ONE 5(6): e11250. doi:10.1371/journal.pone.0011250. CB conceived and designed
experiments conducted trials, analysed data, developed the NEU concept, wrote the MS,YX conducted
trials and conducted voracity experiments underlying NEU concept, CM conducted trials, AS conceived
and designed experiments, RH conceived and designed experiments and wrote the paper.
123
that organic pesticides are more environmentally benign than synthetic ones. All
pesticides must be evaluated using an empirically-based risk assessment, because
generalizations based on chemical origin do not hold true in all cases.
5.1 Introduction
A public call for sustainability in agriculture has resulted in numerous government
initiatives to develop environmentally friendly agricultural practices (European
Commission 1991, Lynch et al. 1996, Ministry of Science and Technology of the
People´s Republic of China 2001, Agriculture and Agri-Food Canada 2003, Jones 2004,
U.K. Department for Environment Food and Rural Affairs and the Forestry Commission
2005). In 2003, the Canadian government initiated the Pesticide Risk Reduction Program
to provide infrastructure for the development and implementation of reduced-risk
approaches for managing pests in crops (Agriculture and Agri-Food Canada 2003). This
program, similar to ones in the UK (U.K. Department for Environment Food and Rural
Affairs and the Forestry Commission 2005) and USA (Jones 2004), sought to reduce
environmental risk associated with older chemical insecticides by replacing them with
low risk alternatives. Though generalizations about the relative safety of natural and
synthetic chemicals have been questioned in the past (Ames et al. 1990), these
sustainability programs often continue to emphasize the development of organic and
natural insecticides for pest control. These programs make the assumption that natural
insecticides present less risk to the environment than synthetic insecticides, aligning with
public opinion (James 1990) and influential scientific papers purporting greater
sustainability of organic practice (Reganold et al. 2001).
The sustainability of agricultural practices is a subject of ongoing debate in the
literature (Trewavas 2001, Elliot and Mumford 2002, Shepherd et al. 2003, Lynch 2009).
124
Many studies have compared organic, conventional and integrated pest management
(IPM) production systems as a whole, but even within a commodity system, the
conclusions reached in these studies are widely divergent. A 1999 study of New Zealand
apple production suggested an integrated approach was more sustainable (Suckling et al.
1999), but a 2001 study of the same system in Washington favoured an organic
management approach (Reganold et al. 2001). Differing outcomes may be attributed
partially to differing geography, climate and pest complexes at the two locations, but it is
likely that differences in assessment methodology and the inconsistencies between
specific practices classed as organic or conventional at each location were also influential
in obtaining the observed results. Comparing organic, conventional and integrated
agriculture is not as simple as it may initially appear (Shepherd et al. 2003): each system
is characterized by a suite of practices which are ideologically, rather than empirically
defined (Elliot and Mumford 2002), these systems are not mutually exclusive from each
other (Reganold et al. 2001, Elliot and Mumford 2002), and vary from region to region
depending on regulations (Suckling et al. 1999). Because of these variations,
generalizations about the overall sustainability of one system over another are never
universal (Trewavas 2001). Pest management practices are often specifically highlighted
in the sustainability of organic versus conventional agriculture debate, but much of the
debate is fuelled by a fundamental misconception that organic farms do not use pesticides
(Avery 2006). In fact, organic farms, like conventional farms, have access to a suite of
pesticides (Avery 2006, Thompson and Kreutzweiser 2006); the primary difference is
that organic regulations prohibit all synthetic (i.e.: human-made) chemicals but allow a
vast array of mineral and botanical pesticides (Canadian General Standards Board 2008),
whereas conventional pesticides can be both naturally and synthetically derived and are
regulated individually, on a per active ingredient, per formulation basis (OMAFRA
2005).
125
Generalizations about the relative sustainability of one suite of practices over
another are dangerous when integrated into policy: government regulations based on
faulty assumptions about agricultural systems are expensive and do not effectively reduce
the environmental risks they are designed to mitigate (Kleijn et al. 2001). It is thus more
productive, and more broadly applicable, to evaluate a given tactic for environmental
sustainability on its individual properties and build policy based on results of these
individual evaluations (Thompson and Kreutzweiser 2006).
Many national and international initiatives exist to develop environmentally
sustainable strategies for managing outbreaks of soybean aphid, including Agriculture
and Agri-Food Canada’s (AAFC) Pesticide Risk Reduction Program (Agriculture and
Agri-Food Canada 2003). Soybean aphid is a severe pest of cultivated soybean in North
America (Ragsdale et al. 2004), and approximately 1.2 million hectares of soybean are
cultivated each year in Canada alone (Statistics Canada 2009). Since its introduction to
North America 10 years ago (Ragsdale et al. 2004), numerous studies have examined the
role of biological control agents in managing populations of aphids (Heimpel et al. 2004,
Rutledge et al. 2004, Desneux et al. 2006, Costamagna et al. 2007a, Xue et al. 2009), but
foliar insecticides remain necessary when populations of aphids exceed economic
thresholds. The need for reduced risk pesticides in this system is profound: only two
foliar insecticides are currently registered for soybean aphid control in Canada
(OMAFRA 2005), one of which is currently under review for re-registration (Health
Canada Pest Management Agency 2009). A broader suite of insecticides with varied
mechanisms of action are needed to ensure effective insecticide resistance management
can occur (Brattsten et al. 1986). The objective of this study is to evaluate several novel
synthetic and organic-approved pesticides for control of soybean aphid, and place the
results in the context of Canadian agro-environmental policy.
126
5.2 Materials and Methods
5.2.1 Selection of insecticides for inclusion in experiments
In May 2008, the Pest Management Centre at AAFC provided us with a list of 14
potential insecticides for inclusion in our experiments. We reviewed each insecticide and
eliminated those which had the same mode of action as any other insecticide registered
for use against soybean aphid in Canada, and then contacted the suppliers to assess the
economic feasibility of using these insecticides in field crops. Two novel synthetic and
two organic insecticides were identified to be tested for management of soybean aphid,
and the two registered insecticides were included in the experiment as conventional
controls. Experimental application rates for novel insecticides were developed in
consensus with supplier companies (Table 5-1). Table A-1 (Appendix 2) provides a
complete list of insecticides considered for inclusion in this experiment, and the rationale
for products selected.
5.2.2 Determination of direct contact toxicity to natural enemies
Adults and larvae of multicoloured Asian ladybeetle Harmonia axyridis and adults
of insidious flower bug Orius insidiosus were treated with formulated insecticides at the
equivalent of 0.5, 1 and 2x field rate using an airbrush spray tower. The untreated control
consisted of 1 mL of distilled water. Groups of insects (8-10) were anesthetized using
CO2 then placed in a 50 mm glass Petri plate lined with a piece of 47 mm qualitative
filter paper, treated using the spray tower, and then placed in post-treatment containers.
Each insecticide-concentration combination was repeated four times. The spray tower
was rinsed with acetone, then distilled water, between each application.
127
Orius insidiosus assays: Orius insidiosus adults were obtained from commercial
suppliers (BioBest Biological Systems Canada and MGS Horticultural Inc.). Repetitions
Table 5-1: Insecticides evaluated for use in control of the soybean aphid.
Active ingredient Trade name
(ai)
(Supplier)
Category
Mode of action
Conventional
(synthetic)
Cyhalothrin-λ
Matador 120E®
(Syngenta)
Neurotoxin- sodium
channels
Conventional
(synthetic)
Dimethoate
Lagon 480®
(Cheminova)
Neurotoxinacetylcholine
esterase inhibitor
Novel
(synthetic)
Spirotetramat
Movento® (Bayer) Fatty acid
biosynthesis inhibitor
Novel
(synthetic)
Flonicamid
Beleaf® (FMC)
Novel (organic)
Mineral oil
Novel (organic)
Beauveria
bassiana
Rate
per ha
%ai
EIQEIQ
*
FUR**
13.1
83 mL
47.2
0.4
43.55
1000 mL
33.5
12.5
22.4
196 mL
34.2
1.3
Neurotoxinpotassium channels
50
196 g
8.7
0.8
Superior 70 oil®
(UAP)
Oxygen exchange
99
11,000 mL
30.1
280.2
Botanigard®
(Laverlam)
Entomopathogenic
fungus
22
1,000 g
16.7
3.3
*
per unit weight environmental impact quotient (EIQ)
predicted EIQ-field use rating (EIQ-FUR) for a single application of the insecticide, converted to lbs/ac, as
convention dictates.
**
of 10 adult O. insidiosus were treated, and then placed, post-treatment, in 10 cm plastic
Petri plates lined with filter paper moistened with distilled water, and containing 1-2
washed baby spinach leaves, and an excess of frozen Ephistia eggs (BioBest Biological
Systems Canada) for food. Mortality was recorded at 18, 24 and 48h post treatment.
Harmonia axyridis adult assays: Harmonia axyridis were obtained from
aggregations on buildings in Guelph, Ontario, Canada, and were reared in laboratory
cultures using procedures described by Xue et al. (2009). Repetitions of 10 adult H.
axyridis were treated, and then placed in 10 cm plastic Petri plates lined with filter paper
moistened with distilled water, and containing several barley leaves infested with birdcherry oat aphid (Aphid Banker System; Plant Products, Brampton, Ontario, Canada), and
128
an excess of frozen Ephistia eggs (BioBest Biological Systems Canada) for food.
Mortality was recorded every 24h for 168 h (7 d).
Harmonia axyridis larvae assays: Second and third instar H. axyridis were
obtained from the laboratory culture described above. Assays were performed as adult
assays above, except repetitions consisted of 8 individuals and instead of being placed
together in a Petri plate, were placed individually into cells of a rearing tray (BIO-RT-32,
C-D International, Inc.) with Ephistia eggs and aphid-infested barley to avoid
cannibalism.
Statistical analysis of bioassay data: Mortality data was normalized using the
Henderson-Tilton adjustment (Henderson and Tilton 1955), and subjected to a mixed
model ANOVA accounting for concentration (relative to field rate), treatment, and
assessment time. Assessment time was treated as a repeated measure in the analysis.
5.2.3 Determination of field efficacy and selectivity
In 2009, four soybean fields in southwestern Ontario with aphid populations
approaching the action threshold of 250 aphids per plant were identified in collaboration
with government extension personnel in July and August, 2009. After obtaining
permission from landowners, sites were assessed once weekly until aphid populations
exceeded 250 aphids per plant. Upon reaching this threshold, field experiments were
initiated. In our initial screening trial in 2008, treatments were applied to a single site
with a moderate density of aphids (~120 aphids per plant), due to low aphid populations
across our region during that year.
Field experiments employed a RCBD consisting of four blocks of 15 3.7x15.2m
beds, with 3 untreated controls per block (one for each tractor pass required), our six
129
insecticides and six other products or formulations not reported in this study. Insecticides
were applied using a Teejet Duo nozzle configuration with spray tips #TT11002 at a
height of 50cm above the canopy. Spray pressure at the nozzle was 276 kPa and the
tractor travelled at a ground speed of 9.7km/h. Fluid delivery rate was maintained at 187
L/ha for all treatments. 2-3 soybean plants were destructively sampled from each bed at
each assessment, and assessments were completed 1) immediately before treatment, 2)
one week after treatment and 3) two weeks after treatment. Total numbers of aphids,
ladybeetles, lacewings, parasitized aphid mummies, syrphid larvae, and flower bugs were
assessed on each plant.
Aphid counts were transformed using Henderson-Tilton adjustments to account for
population changes in the control between time of treatment and time of assessment, then
subjected to a mixed model ANOVA accounting for site, year, tractor pass, replicate, and
treatment. A post-hoc LSD was used to determine differences between individual
treatments, and treatment groups (organic and synthetic) were compared using
CONTRAST statements under GLM.
5.2.4 Field Selectivity Calculation
Field selectivity of each insecticide was estimated by calculating the change in the
ratio of natural enemies to aphids in each plot, and subjecting these data to a mixed
model ANOVA as above. We defined field selectivity as the relative change in the
natural-enemy-to-pest population ratio observed after treatment. We standardized the
counts of natural enemies of different species by defining a Natural Enemy Unit (NEU),
where 1 NEU is the number of predators or parasitoids required to kill 100 pest insects in
24h. Thus,
N
NEU total   niVi
i 1
130
(5-1)
where N is the total number of natural enemy species, ni is the total number of
individuals of natural enemy species i observed on 10 plants, and Vi is the average
voracity of natural enemy species i, that is, the number of pest insects it can kill in 24 h
divided by 100. Using functional response data obtained by Xue et al.(2009), we defined
our soybean aphid ecosystem specific calculation as:
NEU total  1 nladybeetles  0.08  nmummies  0.15  nsyrphids  0.08  nOrius  0.35  nlacewings (5-2)
where nladybeetles is the total number of adult and larvae of ladybeetles of Harmonia
axyridis or Coccinella septempunctata, nmummies is the total number of parasitized aphids,
nsyrphids is the total number of Syrphidae larvae, nOrius is the total number of Orius spp.,
and nlacewings is the total number of Chrysopidae observed on 10 soybean plants.
Field selectivity was defined as the ratio of NEU/Aphids (NEU/A) after treatment
to NEU/A before treatment, normalized by the control, as in the Henderson-Tilton
adjustment (Henderson and Tilton 1955), and took the form:
 ( NEU / A) post treatment 
 ( NEU / A) pre treatment 
Selectivit y  

(5  3)


 ( NEU / A) pre treatment  treated  ( NEU / A) post treatment  control
This selectivity index results in values < 1 if a treatment kills more natural enemies
than target pests, and values >1 if a treatment kills more target pests than natural enemies.
Larger numbers will indicate a more target-selective pesticide. The selectivity index
assumes the applied treatment has at least some efficacy against the target pest.
5.2.5 Environmental Impact Assessment
EIQs were estimated using established methodology incorporating data from
MSDS sheets provided by the supplier of the insecticides, an EIQ-field use rating was
calculated for each insecticide, using the assumption that one application at field rate per
season would provide equivalent aphid control (Kovach et al. 1992, 2009). See Table 5-2
131
Table 5-2: Toxicity ratings used to calculate Environmental Impact Quotient for Beauveria bassiana,
which does not have a published EIQ value.
Variables from EIQ Equation*
Toxicity to
beneficials
B
Plant surface
half life
P
Chronic health
effects
C
Mode of action
SY
Soil residue half
life
S
Runoff potential
R
Leaching
potential
L
Bee toxicity
Z
Fish toxicity
ingredient
F
Bird toxicity
Active
D
Dermal toxicity
DT
1
1
1
3
1
1
3
1
1
1
5
Beauveria
bassiana
* Ratings were developed in accordance with methodology presented in Kovach et al.(Kovach et al.
1992)
for values used in the calculation of EIQ for Beauveria bassiana, which does not have an
existing published EIQ value.
5.3 Results
Working with AAFC, we identified four novel products to evaluate as potential
reduced risk insecticides to include in integrated pest management programs for soybean
aphid (Table 5-1). Two of these insecticides contained synthetic active ingredients, the
other two are natural insecticides permitted for use in certified organic crops in Canada
(Canadian General Standards Board 2008). We included formulations of the two
currently registered insecticides in the experiments as conventional controls.
We completed laboratory assays to estimate the direct contact toxicity of these
insecticides to several natural enemy species when applied at field rates (Table 5-3). We
used two of the soybean aphid’s primary predator species in this study, multicoloured
Asian ladybeetle Harmonia axyridis and insidious flower bug Orius insidiosus (Desneux
et al. 2006, Xue et al. 2009).
132
Table 5.3: Relative direct contact mortality of natural enemies treated with six insecticides at field
rate.
Relative H-T adjusted % mortality*
Treatment
Harmonia axyridis adults
Harmonia axyridis larvae
Orius insidiosus adults
Cyhalothrin-λ
34.9b
48.2b
99.1a
Dimethoate
70.7a
99.6a
77.2b
Spirotetramat
2.0de
5.3de
20.7e
Flonicamid
0.8e
10.9c
39.5d
Mineral oil
2.6d
6.3d
60.7c
Beauveria bassiana
10.9c
2.7e
59.5c
Untreated control
-0.1e
-0.1f
-0.2f
*Insecticides were applied at 0.5, 1 and 2x field rate using an airbrush sprayer. Mortality was assessed at 18,
24 and 48 h post treatment for O. insidiosus, and every 24 h for 7d for H. axyridis adults and larvae.
Mortality data were Henderson-Tilton adjusted (1955) and subjected to a mixed model ANOVA by species
and life stage, with relative rate incorporated into the model, and assessment time treated as a repeated
measure. Observed mortality within a species and life stage followed by the same letter are not significantly
different at α=0.05 (LSD).
There were significant differences in mortality by treatment applied for all insect
groups F6,657=325.25, P<.0001 for ladybeetle adults; F6,993= 1069.34, P<.0001 for
ladybeetle larvae; F6,277=228.11, P<.0001for flower bug adults), but generally, the two
currently registered insecticides were most toxic to natural enemies under laboratory
conditions. The other four insecticides were much less toxic to the ladybeetle, though it
was found that one of the organic insecticides, Beauveria bassiana, was slightly more
toxic to adults, and one novel synthetic, flonicamid, was slightly more toxic to larvae
than the remaining novel insecticides. The four novel pesticides all caused some
mortality to the insidious flower bug, but the two organic insecticides had significantly
higher toxicity than the two novel synthetic insecticides.
133
We conducted a two year, five site study to examine the performance of these
insecticides against aphids, and selectivity with respect to natural enemies under field
conditions (Fig. 5-1). In addition to efficacy, it is desirable for an insecticide to have a
high selectivity for its target pests in order to minimize environmental impact, and to
conserve biological control services provided by other organisms residing in the treated
area. All synthetic insecticides had similar efficacy one week after treatment (F6,148=7.48,
P <0.0001), though dimethoate efficacy was reduced in the second assessment week (Fig.
5-1a), and yield in plots treated with synthetic insecticides did not differ significantly
(F6,90= 3.51, P=0.0036) (Fig. 5-2). The two organic insecticides had lower efficacy than
the synthetic insecticides at one week (F1,148=25.16, P <0.0001) and two weeks (F1,121=
17.48, P <0.0001 ) post-treatment and did not offer significant yield protection over the
untreated control (Fig. 5-2). Field selectivity was highest amongst synthetic insecticides,
and lowest amongst organic insecticides included in this experiment (F1,119=9.00,
P=0.0033), and although dimethoate had the numerically lowest selectivity amongst the
synthetic insecticides, it was still numerically more selective than the organic
insecticides.
Net environmental impact of applying each insecticide at given rates was estimated
using an Environmental Impact Quotient analysis.(Kovach et al. 1992) The per-unit-EIQ
was highest for cyhalothrin-λ, a conventional synthetic insecticide (Table 5-1), but the
EIQ-field use ratings were highest amongst the older synthetic, dimethoate, and the two
organic insecticides. The high EIQ-field use rating of dimethoate was due to both a high
application rate and a relatively high per-unit EIQ. The EIQ-field use rating for the
mineral oil insecticide, though, was more than an order of magnitude higher than that of
dimethoate, due to its relatively high per-unit-EIQ and its extremely high application rate.
The remaining four insecticides had relatively low EIQ-field use ratings compared with
mineral oil and dimethoate.
134
Figure 5-1. a) Observed field efficacy of six insecticides for soybean aphid control. Aphid count data
were Henderson-Tilton adjusted (Henderson and Tilton 1955) and subjected to a mixed model
ANOVA by post-treatment sampling period with year of experiment, block, pass of tractor, site, and
interaction terms between block and pass, block and site, and pass and site incorporated into the
model.
b) Observed field selectivity of six insecticides for aphid control. Field selectivity was determined
using the natural enemy-to-aphid ratio in treatment plots, for exact calculation see Materials and
Methods. Observed efficacy and selectivity within sampling period marked by the same letter are
not significantly different at α=0.05 (LSD).
135
Figure 5-2: Least-square mean yield in fields treated with six insecticides. Data were subjected to a
mixed model ANOVA with block, site, treatment incorporated into the model. Observed yields
marked by the same letter are not significantly different at α=0.05 (LSD).
5.4 Discussion
EIQ allows relative impact of various control strategies within a crop to be ranked;
it is a standard method for indexing the total environmental impact of an application of a
given pesticide. EIQ relies on data which is commonly available on MSDS sheets,
incorporates the application rate of a pesticide, and is not site or pest-specific, so it
provides a less biased estimation than other pesticide ranking systems used to quantify
environmental impact (Levitan et al. 1995, Avery 2006). Because EIQ is based on a
rating system and does not rely on field obtained data, some authors have criticized its
use (Elliot and Mumford 2002).
136
Figure 5-3. Relationship between observed field selectivity and the inverse of Environmental
Impact Quotient at field rates. Field selectivities presented as least square means (± SE) of field
selectivities observed at four sites in 2009. Equation of regression line is Field selectivity =
(3.3±1.7)/EIQ+(0.3±3.1)+site effect, with F93 = 4.23, p = 0.0035.
However, we found a clear inverse relationship between field selectivity and EIQ
for insecticides tested in this study when applied at field rates (Fig. 5.3), suggesting that
EIQ rankings are relevant predictors of at least some in-field parameters for
environmental impact, and our results strongly support the continued use of EIQ for
ranking pesticide impact. Responses of natural enemy communities are strong indicators
of ecological impact of an insecticide, because they are arthropods, like the targets, and
are thus likely to be biologically similar to the target of the insecticide, and because they
are often found alongside the pest at the time of an insecticide application, heightening
their exposure compared to other non-target organisms.
Looking at the issue empirically, our results show that with regards to
environmental impact, target selectivity and efficacy, the novel synthetic insecticides we
tested have better performance than organic insecticides; suggesting that certain organic
137
management practices are not more environmentally sustainable than conventional ones.
It has been purported that organic systems are not just better for the environment, but are
more economically sustainable because of the price premiums associated with organic
food (Reganold et al. 2001). Consumers are often willing to pay more for products they
believe are produced in the most sustainable way possible, but we have shown that the
organic methods available are not always the most sustainable choice. Carefully designed
integrated pest management systems are likely the best strategy for minimizing
environmental impact of agriculture: where certified organic systems may reject the
technology with the smallest environmental impact based on ideology (Trewavas 2001),
IPM maintains the flexibility to incorporate any strategy empirically determined to have
the smallest impact. In fact, it has been argued that studies which have concluded that
IPM has a greater impact than organic management (e.g.: Reganold et al. 2001) have
simply tested a poorly designed IPM strategy in which the efficacy and impact of
individual tactics included in the program were not effectively examined (Elliot and
Mumford 2002), did not accurately reflect IPM practice, or employed biased methods of
evaluation (Avery 2006). Though IPM practice does not typically come with price
premiums associated with the production of organic food, IPM strategies are still
commonly used by many conventional farmers (Olson et al. 2008), and given increased
consumer awareness of the benefits of IPM practice, adoption rates are likely to rise.
It is for these reasons that we reject the organic-conventional dichotomy and
emphasize that, in order to optimize environmental sustainability, individual tactics must
be evaluated for their environmental impact in the context of an integrated approach, and
that policy decisions must be based on empirical data and objective risk-benefit analysis,
not arbitrary classifications.
138
5.5 Acknowledgements
We sincerely thank L. Des Marteaux, D. Makynen, J. Smith,T. Phibbs, D.
Hooker, A. Gradish and T. Baute for technical assistance on this project, R. Norris and B.
Stirling for the use of their respective farms, C. Scott-Dupree for use of the airbrush
spray tower, M.K. Sears and J.A. Newman for providing comments on this manuscript,
and C. Petzoldt and D. Marvin for providing support with the EIQ method. We would
also like to thank Syngenta, Bayer, FMC, UAP, and Laverlam for providing insecticides
for our experiments, and BioBest Canada and MGS Horticultural for providing insects
139
CHAPTER 6
General Discussion and Conclusions
6.1 Discussion
The aim of this project was to elucidate the factors controlling the phenology of
A. glycines, to develop models for predicting A. glycines abundance and distribution
which incorporate biotic and abiotic factors, and to assess environmental viability of
management strategies in the context of this knowledge.
6.1.1 Distribution of A. glycines colonizing soybean
In Chapter 2, the distribution and abundance of A. glycines colonizing soybean
fields early in the growing season were examined relative to several landscape
parameters, including the distribution of overwintering hosts. Though no relationship
between landscape, overwintering hosts, and colonization by A. glycines had been
documented previously (Heimpel et al. 2010), using an information-theoretic approach,
this study was able to reveal one and detected a change in pattern in this relationship
between the two study years. The key factor affecting the density of colonizing A.
glycines amongst factors examined in this study was the density of buckthorn in the
hedgerow facing the field, normalized by the area of the field in which the aphids were
observed, but, interestingly, the sign of the interaction changed between the two study
years. In the ‘low’ aphid year, 2006, when A. glycines populations did not typically reach
economically damaging levels in our study region, abundance of A. glycines was
positively correlated with buckthorn density normalized by field area. In 2005, the ‘high’
140
aphid year, abundance of colonizing A. glycines was negatively correlated with buckthorn
density normalized by field area, and yet, this was still the best predictor amongst the
landscape parameters tested. This effect suggests that other factors not accounted for
directly in this study interact with the density of A. glycines, and may include densitydependence on the duration of flight by migratory A. glycines moving from overwintering
hosts, or localized depletion of spring morphs of A. glycines by natural enemies which
exhibit numerical or functional responses in areas where overwintering hosts are
abundant.
An understanding of which fields are most likely to be colonized by A. glycines is
extremely desirable from a management perspective, because the decision to use
prophylactic seed treatments must be made before soybean is planted, and thus, long
before A. glycines occurs. This study provides insight into which factors may affect the
likelihood of a given soybean field to become infested by A. glycines, though the
apparent density-dependent change in relationship between landscape factors and
colonization by A. glycines makes universal recommendations impossible at this time.
Additional work, incorporating similar data from other locales into the analysis, may
clarify these patterns further; the continued use of an information-theoretic approach may
be able to elucidate patterns in existing data from similar studies where no effect has been
observed.
Information theory and ecoinformatics are a suite of tools that are largely underutilized in agricultural ecology, and yet much of the observational data generated by
research and monitoring in integrated pest management lends itself well to these
approaches (Rosenheim et al. 2011). Manipulative studies may provide researchers with
141
greater confidence in their research results, but there are many situations where
manipulative studies are prohibitively costly, sensitive to external perturbation, or
impossible to implement or control. Observational studies offer increased scalability, an
ability to screen multiple factors affecting a desired variable and to generate hypotheses,
and the ability to integrate data generated by multiple sources (Rosenheim et al. 2011).
6.1.2 Alatoid morph production and regulation
It is for the reasons outlined above that a similar approach to analysis as used for
Chapter 2 was used in Chapter 3, which integrated five years of suction trap capture data
of alate A. glycines from the American suction trap network (Schmidt et al. 2012), an
Ontario network, and field surveys of A. glycines by extension personnel from across the
North American soybean growing region. The objective of Chapter 3 was to compare
abundance of alate A. glycines sampled by these traps to weather data to elucidate
patterns in alate morph phenology and distribution in North America. It was found that in
summer, alate production in A. glycines was most directly related to the level of
infestation (i.e. the relative population density of A. glycines) in soybean fields, with
environmental parameters playing a more important role in the production of winged
sexual morphs in the fall. Unexpectedly, two activity peaks associated with gynoparae
were observed in fall; a broad peak in early fall, and a later narrow peak, coinciding
directly with the activity peak for male aphids. The early fall activity peak was best
predicted by degree-day accumulation, and the late fall activity peak was best predicted
by Julian date or photoperiod. The association between the early fall peak and degree day
accumulation suggests that host plant cues may induce this first wave of migrants leaving
soybean for buckthorn: soybean phenology is both thermally- and photoperiod142
dependent (Kumar et al. 2008), and thus degree-day accumulation may coincide with
some physiological state in soybean used by A. glycines as a cue to initiate buckthorn
colonization. It was not possible to directly examine the role of host plant phenology in
the production of these morphs as these data were not available. The close association
between Julian date, photoperiod and degree day accumulation (all parameters that will
co-vary with soybean phenology late in the growing season), and the occurrence of
sexual morphs, suggests host plant phenology may play a role in regulating production of
gynoparae and males of A. glycines.
The relative contributions of various parameters in the regulation of morph
production in aphids is difficult to discern in observational studies, particularly because
many of these parameters will co-vary. Morph determination of aphids is dependent on
density, photoperiod, temperature and host plant quality and phenological cues, and all of
these factors may interact with each other to moderate their relative influence (De Barro
1992). Additionally, factors affecting morph determination may vary by species: for
instance, alate production in Cinara pinea (Mordv.) is thought to be entirely regulated by
host plant cues (Kidd and Tozer 1984); in Therioaphis maculate (Buckton), crowding is
the primary cue for the induction of nonsexual alates (Toba et al. 1967); but in
Rhopalosiphum padi L., multiple, interacting cues are involved in morph determination
(De Barro 1992). Some authors have rejected the notion of population density as the most
important factor inducing alates in aphids in favour of nutritional cues (Müller et al.
2001), but others consistently report crowding, even when accounting for other,
interacting cues, as the most important parameter in morph determination in nonsexual
morphs (De Barro 1992)
143
Documenting the effect of plant quality or phenology on morph production is
experimentally challenging: because aphids occur on host plants that are also responding
to environmental cues which affect their phenology, manipulating environmental
conditions will also manipulate host physiology, so observed responses in the aphid may
be as a result of environment or host cues. Similarly, it is impossible to eliminate the
effect of environmental cues in studies examining the effect of host quality on aphid
morph regulation when working with intact plants. To a certain degree, this problem
could be overcome with various host treatments (i.e. rearing host plants under controlled
conditions and only exposing them to experimental conditions when aphids are
introduced, incorporating multiple planting dates or maturity groups of host plants into a
study to introduce variability in response to environmental conditions into a study), but
ideally, host plant effects would be eliminated altogether by rearing aphids on artificial
diet. Though A. glycines has not been reared successfully on an artificial diet, other
species of aphid can complete development and, indeed, thrive, on these preparations
(Mittler and Dadd 1962). Alate production in Myzus persicae (Sulzer) varied when reared
on diets with manipulated nutritional content, however, alate production was lower in
mothers fed nutritionally unbalanced diets and highest on diets that favoured fecundity
and high developmental rates amongst the alatoid nymphs produced (Mittler and Kleinjan
1970). Thus, alate production is not simply a response to declining nutritional content in
host plants (or, at higher population densities, increased competition with intraspecifics
leading to decreased access to a host plant’s nutritional content), but nutritional state
interacts with aphid population density and environmental cues, as well.
144
In turn, density dependence in alate formation may play an important role in the
observed population dynamics of A. glycines in soybean fields. Density-dependent
dispersal behaviours by nonsexual alates have been implicated in mid-season population
crashes for several aphid species (Karley et al. 2004, Mashanova et al. 2008). However,
the effect of density dependence on population regulation of A. glycines is poorly
understood; indeed, it has been argued that alate morph production is rather low (~5%) at
densities up to 4000 aphids per plant, and that population regulation below this density is
likely achieved by predation by natural enemies (Donaldson et al. 2007) .
6.1.3 Quantifying biocontrol services: the Natural Enemy Unit
In Chapters 4 and 5, the Natural Enemy Unit (NEU), a novel method for
quantifying the biocontrol services of a resident natural enemy guild was used for two
different applications. The NEU standardizes the impact of a given natural enemy
species by its maximum relative voracity, and multiplies this voracity factor by the
abundance of each species resulting in a single normalized measure of biocontrol
capacity in terms of the number of prey insects that can be consumed in 24 hours. In
addition to the applications described in Chapters 4 and 5 of this dissertation, the NEU
has been used as a method of predicting whether or not natural enemy populations are
sufficient to maintain aphid populations below economic levels in the field (Hallett et al.
In prep).
In Chapter 4, the NEU was used to interface natural enemy population submodels
with a model for A. glycines phenology and population ecology. Density dependent
factors were applied post-hoc to the NEU to allow net biocontrol services to mimic
roughly a predator with a type II functional response, when aphid populations did not
145
occur in excess. This approach effectively modeled dynamics between A. glycines and its
natural enemy guild, allowing the model to be revised in the future to include additional
natural enemy submodels with minimal structural changes to the overall model.
The NEU was used in Chapter 5 as part of a computation to determine the
selectivity of an insecticide as a function of the net impact on resident biocontrol
services. The Selectivity Index is computed by calculating the ratio of NEU to aphids
post-treatment over the ratio of NEU to aphids pre-treatment in a treated plot, and
dividing this by the same ratios in a control plot as a normalization factor. This index
correlated well with the inverse of the Environmental Impact Quotient Field Use Rating
(EIQFUR) for each pesticide, suggesting that the EIQFUR, as a theoretical measure,
translated well to field conditions.
6.1.4 Development of a tritrophic population model
The objective of Chapter 4 of this dissertation was to develop a tritrophic
population and phenology model incorporating A. glycines, soybean phenology, and
phenology and population models for representative members of the natural enemy guild
of A. glycines. Using the large amount of available literature, a mechanistic, cohortbased model was developed, incorporating density-, temperature-, and photoperioddependent processes, the NEU, and interactions with host phenology. The model contains
numerous user-customizable parameters and inputs, allowing multiple ‘what-if’ scenarios
to be tested; scenarios examined in this dissertation included how soybean planting date
affected overwintering populations of eggs of A. glycines, and how natural enemies affect
population dynamics of A. glycines occurring on soybean in the summer.
146
A key limitation to the applicability of this model to the field is its behavior as a
closed system. Unlike in nature, natural enemy population growth was constrained within
the model by the reproductive capacity of each species, and unless explicitly specified by
the user, no natural immigration or emigration events will occur in response to
availability of prey. This limits the ability of the model to predict population dynamics in
fields where immigration or emigration events of any of the modeled taxa occur, and thus
care must be taken in the initialization of the model to capture these events whenever
possible.
6.1.5 Managing A. glycines: developing evidence-based practice
In Chapter 5, four novel insecticidal products for A. glycines control were
evaluated. Two of these products, flonicamid and spirotetramat, were synthetic
chemistries intended for use in conventional agricultural systems; the remaining two,
mineral oil and Beauvaria bassiana, were naturally derived and thus could be used in
organic systems (Canadian General Standards Board 2008). Two natural enemy species
in laboratory studies, and the foliar natural enemy community in field studies were used
to inform an assessment of the selectivity and environmental impact of the tested
insecticidal products because resident natural enemy populations are extremely important
regulators of the population growth of A. glycines. Negative pesticide impacts on the
resident natural enemy community will leave crops vulnerable to further attack by A.
glycines in the event of pesticide failure or an immigration event of this species occurring
after the insecticide application. Additionally, because the foliar natural enemy
community is largely composed of arthropods that are usually present at the time of
147
insecticide application, their similar physiology and increased exposure to insecticides
make these species more likely to be affected by insecticide use.
Our results suggest that the two novel synthetic products had greater efficacy than
the two naturally-derived products, but perhaps more surprisingly, had considerably more
favourable environmental profiles than the naturally-derived products. This result is
discussed within the context of agri-environmental policy, and challenges the commonly
held notion that organic practice is uniformly better for the environment (James 1990). It
is my belief that the binomial classification of organic versus conventional should be
discarded in favour of pluralism, and instead that individual agricultural practices be
evaluated using objective, evidence-based standards to guide decision-making, as others
have argued before (Coats 1994, Suckling et al. 1999, Elliot and Mumford 2002).
Implementation of evidence-based practices for management of A. glycines, like any
agricultural pest, is not without its challenges, however. As in agriculture, many
industries struggle to implement evidence-based practice (e.g. nursing, Rosswurm and
Larrabee 1999) while many practitioners within the industry champion intuitive-based
practice based on personal history and experience (e.g.: dentistry, Chambers 2010).
Challenges involved in gaining public acceptance of evidence-based practice in
agriculture include poor public understanding of risk (James 1990), perceived health
benefits of eating food produced using organic techniques (Dangour et al. 2009),
chemophobia (Entin 2011), and poor understanding of the relative benefit of
incorporating conventional practices in yield and food safety (Cooper and Dobson 2007).
Ironically, the consequences of poor understanding of relative risk potentially can be
more hazardous than the risk a consumer takes action to avoid (Ropeik 2004), and so it is
148
a matter of public health for proponents of evidence-based agriculture to actively strive to
enable better public understanding (Entin 2011).
The University of Guelph soybean aphid research group has developed a scouting
tool which employs an understanding of natural enemy-aphid population dynamics to
inform scouting decisions (Hallett et al. 2011, Hallett et al. In prep). This tool, along with
selection of the optimal insecticide when application is warranted (Chapter 5) and
judicious use of seed treatments (Chapter 2), will help to minimize the risks involved
with the management of A. glycines at no net cost to the farmer.
6.2 Future directions
The work presented in this dissertation provides the foundation for future work in
several areas which I intend to pursue in my next position.
6.2.1 An assessment of population regulation in a naturalized population versus a
recent introduction of an invasive crop pest.
A. glycines was introduced to North America approximately 10 years ago. In
Japan, where the aphid has been documented for approximately 2500 years, A. glycines is
considered a pest of soybean, but, unlike in North America, only rarely reaches
economically damaging levels. This discrepancy between North American and Japanese
aphid populations is intriguing for both the obvious reasons of pest management but also
from a perspective of invasive species biology. Regulators of A .glycines phenology and
population ecology are now relatively well studied in North America, however, outbreaks
of the species remain difficult to predict. I hope to explore the relationship between
Japanese and North American aphid populations using the population models I developed
149
in this dissertation (Chapter 4) and aphid population data collected in Japanese soybean
fields. This work will be completed in collaboration with scientists from the National
Agricultural Research Center of Japan in Tsukuba.
6.2.2 An assessment of population regulation in a recently introduced crop pest with
a large adopted geographic range.
In this dissertation, I developed the Natural Enemy Unit as a method to quantify
the predator/parasitoid guild for soybean aphid, in the context of pesticide impact on that
guild (Chapter 5). In Chapter 4, I used the NEU to quantify the net impact of the
predator/parasitoid community in a population model for A. glycines. I believe this
concept can be expanded to incorporate members of the natural enemy guild from
throughout the range of A. glycines and then used to explain differences in population
dynamics at different locations. I will continue to use this model system to explore
applications of the NEU concept by using NEU to quantify the relative influence of
predator guilds of differing composition and abundance and under differing
environmental conditions at a range of geographical locations. This technique can be used
to explain the various patterns in population dynamics observed for these species
throughout their North American ranges.
6.2.3 An evaluation of how environmental parameters are used in phenological and
population models.
Insect phenology and population ecology are functions of both current and past
conditions. Because environmental cues are continuous and, typically, insect sampling is
either cumulative (e.g. trap collections: all insects caught in the trap over the sampling
period represent the sample) or instantaneous (e.g. sweep netting performed
150
intermittently: only the insects active at the moment of sampling are collected), a
disconnect exists between the resolution of sampling and the resolution of available
environmental data. This challenge became particularly evident when I examined
phenology of alate A. glycines captured in suction traps relative to environmental
conditions (Chapter 3). Often, the solution is either to average environmental conditions
over the sampling period, as was done in Chapter 3, or to use the instantaneous
environmental conditions at the time of sampling. These approaches, however, ignore or
minimize factors of potentially great influence, such as the amplitude of variability in
environmental conditions, extreme weather events, and microclimate. I aim to conduct a
detailed critical review of the literature to examine the use of abiotic factors in population
and phenological models, with emphasis on techniques used to address insectenvironment interactions, and will develop recommendations for optimal practices in this
area. This review will improve our ability to address limitations in experimental design
and aid in interpretation of studies examining insect-environment interactions.
6.3 Conclusions
An understanding of the physiological ecology of migration, host selection and
location by A. glycines, and dynamics that occur with natural enemies, hosts and the
environment at each point in the aphid lifecycle is essential to A. glycines management.
A. glycines is a well-studied organism within its North American range, it provides a
unique opportunity to integrate previously collected data and published literature, which
allows broader scale implications and applications of findings to be examined over a
larger geographical range. Because of the dispersive ability and long-range migrations of
151
A. glycines, and their occurrence throughout North American soybean growing regions,
individual studies using local data are limited in their applicability. Large scale studies
integrating the work of multiple research groups, flexible approaches to analyses, and a
willingness to approach control measures from an evidence-based perspective are
essential to the long-term management of this species.
152
BIBLIOGRAPHY
Agriculture and Agri-Food Canada. 2003. Pesticide risk reduction and minor use
programs: improving ways to manage pests with new technology. Government of
Canada.
Akaike, H. 1974. A new look at the statistical model identification. IEEE Transactions
on Automatic Control 19: 716-723.
Alleman, R. J., C. R. Grau, and D. B. Hogg. 2002. Soybean aphid host range and virus
transmission efficiency, Proceedings of the Wisconsin Fertilizer Agline Pest
Management Conference, Madison.
Allison, D., and K. S. Pike. 1988. An inexpensive suction trap and its use in an aphid
monitoring network. Journal of Agricultural Entomology 5: 103-107.
Ames, B. N., M. Profet, and L. S. Gold. 1990. Nature's chemicals and synthetic
chemicals: comparative toxicology. Proceedings of the National Academy of
Sciences, USA 87: 7782-7786.
Anderson, D. R., K. P. Burnham, and W. L. Thompson. 2000. Null hypothesis testing:
problems, prevalence, and an alternative. The Journal of Wildlife Management
64: 912-923.
Aquilino, K. M., B. J. Cardinale, and A. R. Ives. 2005. Reciprocal effects of host plant
and natural enemy diversity on herbivore suppression: an empirical study of a
model tritrophic system. Oikos 108: 275-282.
153
Avery, A. A. 2006. Organic pesticide use: What we know and don't know about use,
toxicity, and environmental impacts, pp. 58-77, Crop Protection Products for
Organic Agriculture. American Chemical Society, Washington, DC.
Bahlai, C. A. 2007. Ecological interactions of Harmonia axyridis and Aphis glycines in
Ontario agroecosystems. Master's thesis. University of Guelph, Guelph.
Bahlai, C. A., and M. K. Sears. 2009. Population dynamics of Harmonia axyridis and
Aphis glycines in Niagara Peninsula soybean fields and vineyards. Journal of the
Entomological Society of Ontario 140: 27-39.
Bahlai, C. A., J. A. Welsman, A. W. Schaafsma, and M. K. Sears. 2007. Development
of soybean aphid (Homoptera: Aphididae) on its primary overwintering host,
Rhamnus cathartica. Environmental Entomology 36: 998-1006.
Bahlai, C. A., S. Sikkema, R. H. Hallett, J. Newman, and A. W. Schaafsma. 2010a.
Modeling distribution and abundance of soybean aphid in soybean fields using
measurements from the surrounding landscape. Environmental Entomology 39:
50-56.
Bahlai, C. A., Y. Xue, C. M. McCreary, A. W. Schaafsma, and R. H. Hallett. 2010b.
Choosing organic pesticides over synthetic pesticides may not effectively mitigate
environmental risk in soybeans. PLoS ONE 5: e11250.
Banks, C. J. 1956. Observations on the behaviour and mortality in Coccinellidae before
dispersal from the egg shells. Proceedings of the Royal Entomological Society of
London. Series A, General Entomology 31: 56-60.
Bianchi, F. J. J. A., and W. van der Werf. 2003. The effect of the area and
configuration of hibernation sites on the control of aphids by Coccinella
154
septempunctata (Coleoptera: Coccinellidae) in agricultural landscapes: a
simulation study. Environmental Entomology 32: 1290-1304.
Bonnemaison, L. 1951. Contribution à l’étude des facteurs provoquant l’appartion des
forms ailées et sexuées ches Aphidinae. Ph.D. dissertation. Université de Paris,
Paris.
Brattsten, L. B., C. W. J. Holyoke, J. R. Leeper, and K. F. Raffa. 1986. Insecticide
resistance: Challenge to pest management and basic research. Science 231: 12551260.
Brosius, T. R., L. G. Higley, and T. E. Hunt. 2007. Population dynamics of soybean
aphid and biotic mortality at the edge of its range. Journal of Economic
Entomology 100: 1268-1275.
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodal inference:
a practical information-theoretic approach. Springer Science + Business Media,
LLC, New York.
Butler, C. D., and R. J. O'Neil. 2007a. Life history characteristics of Orius insidiosus
(Say) fed diets of soybean aphid, Aphis glycines Matsumura and soybean thrips,
Neohydatothrips variabilis (Beach). Biological Control 40: 339-346.
Butler, C. D., and R. J. O'Neil. 2007b. Life history characteristics of Orius insidiosus
(Say) fed Aphis glycines Matsumura. Biological Control 40: 333-338.
Canadian General Standards Board. 2008. Organic productions systems permitted
substances list. Government of Canada.
Chambers, D. W. 2010. Evidence-based dentistry. Journal of the American College of
Dentists 77: 68-80.
155
Chen, R., S. Wang, X. Bao, E. Xu, and W. Xie. 1984. Prediction of aphid infestation by
the number of overwintering soybean aphid eggs. (Translation). Journal of Jilin
Agricultural Sciences 34: 56-61.
Chung, K. H., S. H. Kwon, and Y. I. Lee. 1980. Studies on the density of soybean
aphids in different cultivars, planting dates and spacing. (Translation). Journal of
the Korean Society of Crop Science 25: 35-40.
Clark, S. J., G. M. Tatchell, J. N. Perry, and I. P. Woiwod. 1992. Comparative
phenologies of two migrant cereal aphid species. Journal of Applied Ecology 29:
571-580.
Clark, T. L., B. Puttler, and W. C. Bailey. 2009. Is horsenettle, Solanum carolinense L.
(Solanaceae), an alternate host for soybean aphid, Aphis glycines Matsumura
(Hemiptera: Aphididae)? Journal of the Kansas Entomological Society 79: 380383.
Coats, J. R. 1994. Risks from natural versus synthetic insecticides. Annual Review of
Entomology 39: 489-515.
Cocu, N., K. Conrad, R. Harrington, and M. D. A. Rounsevell. 2005. Analysis of
spatial patterns at a geographical scale over north-western Europe from pointreferenced aphid count data. Bulletin of Entomological Research 95: 47-56.
Coll, M. 2009. Conservation biological control and the management of biological control
services: are they the same? Phytoparasitica 37: 205-208.
Cooper, J., and H. Dobson. 2007. The benefits of pesticides to mankind and the
environment. Crop Protection 26: 1337-1348.
156
Costamagna, A. C., and D. A. Landis. 2006. Predators exert top-down control of
soybean aphid across a gradient of agricultural management systems. Ecological
Applications 16: 1619-1628.
Costamagna, A. C., and D. A. Landis. 2007. Quantifying predation on soybean aphid
through direct field observations. Biological Control 42: 16-24.
Costamagna, A. C., D. A. Landis, and C. D. Difonzo. 2007a. Suppression of soybean
aphid by generalist predators results in a trophic cascade in soybeans. Ecological
Applications 17: 441-451.
Costamagna, A. C., D. A. Landis, and M. J. Brewer. 2008. The role of natural enemy
guilds in Aphis glycines suppression. Biological Control 45: 368-379.
Costamagna, A. C., W. v. d. Werf, F. J. J. A. Bianchi, and D. A. Landis. 2007b. An
exponential growth model with decreasing r captures bottom-up effects on the
population growth of Aphis glycines Matsumura (Hemiptera: Aphididae).
Agricultural and Forest Entomology 9: 297-305.
Cox, G. M., J. M. Gibbons, A. T. A. Wood, J. Craigon, S. J. Ramsden, and N. M. J.
Crout. 2006. Towards the systematic simplification of mechanistic models.
Ecological Modelling 198: 240-246.
CSIRO. 2007. Modeling natural systems. Available online:
http://www.csiro.au/solutions/DymexSoftware.html
Dangour, A. D., S. K. Dodhia, A. Hayter, E. Allen, K. Lock, and R. Uauy. 2009.
Nutritional quality of organic foods: a systematic review. American Journal of
Clinical Nutrition: ajcn.2009.28041.
157
De Barro, P. 1992. The role of temperature, photoperiod, crowding and plant quality on
the production of alate viviparous females of the bird cherry-oat aphid,
Rhopalosiphum padi. Entomologia Experimentalis et Applicata 65: 205-214.
Delanoy, L., and O. Archibold. 2007. Efficacy of control measures for European
buckthorn ( Rhamnus cathartica L.) in Saskatchewan. Environmental
Management 40: 709-718.
Desneux, N., R. J. O'Neil, and H. J. S. Yoo. 2006. Suppression of population growth of
the soybean aphid, Aphis glycines Matsumura, by predators: the identification of a
key predator and the effects of prey dispersion, predator abundance, and
temperature. Environmental Entomology 35: 1342-1349.
Diaz-Montano, J., J. C. Reese, J. Louis, L. R. Campbell, and W. T. Schapaugh.
2007. Feeding behavior by the soybean aphid (Hemiptera: Aphididae) on resistant
and susceptible soybean genotypes. Journal of Economic Entomology 100: 984989.
Doherty, H. M., and D. F. Hales. 2002. Mating success and mating behaviour of the
aphid, Myzus persicae (Hemiptera: Aphididae) European Journal of Entomology
99: 23-27.
Donaldson, J. R., and C. Gratton. 2007. Antagonistic effects of soybean viruses on
soybean aphid performance. Environmental Entomology 36: 918-925.
Donaldson, J. R., S. W. Myers, and C. Gratton. 2007. Density-dependent responses of
soybean aphid (Aphis glycines Matsumura) populations to generalist predators in
mid to late season soybean fields. Biological Control 43: 111-118.
158
Dubrule, O. 1984. Comparing splines and kriging. Computers & Geosciences 10: 327338.
Eilenberg, J., A. Hajek, and C. Lomer. 2001. Suggestions for unifying the terminology
in biological control. Biocontrol 46: 387-400.
Elliot, S. L., and J. D. Mumford. 2002. Organic, integrated and conventional apple
production: why not consider the middle ground? Crop Protection 21: 427-429.
Elton, C. S. 1925. IX. The Dispersal of Insects to Spitsbergen. Transactions of the Royal
Entomological Society of London 73: 289-299.
Entin, J. 2011. Scared to death: how chemophobia threatens public health. American
Council on Science and Public Health, New York.
European Commission. 1991. Council directive of 15 July 1991 concerning the placing
of plant protection products on the market. European Union.
Forsythe, W. C., E. J. Rykiel, R. S. Stahl, H. Wu, and R. M. Schoolfield. 1995. A
model comparison for daylength as a function of latitude and day of year.
Ecological Modelling 80: 87-95.
Fox, T. B., D. A. Landis, F. F. Cardoso, and C. D. Difonzo. 2004. Predators suppress
Aphis glycines Matsumura population growth in soybean. Environmental
Entomology 33: 608-618.
Fox, T. B., D. A. Landis, F. F. Cardoso, and C. D. Difonzo. 2005. Impact of predation
on establishment of the soybean aphid, Aphis glycines in soybean, Glycine max.
BioControl 50: 545-563.
Frewin, A. J., Y. Xue, J. A. Welsman, B. A. Broadbent, A. W. Schaafsma, and R. H.
Hallett. 2010. Development and parasitism by Aphelinus certus (Hymenoptera:
159
Aphelinidae), a parasitoid of Aphis glycines (Hemiptera: Aphididae).
Environmental Entomology 39: 1570-1578.
Garcia Adeva, J. J., J. H. Botha, and M. Reynolds. 2012. A simulation modelling
approach to forecast establishment and spread of Bactrocera fruit flies. Ecological
Modelling 227: 93-108.
Gardiner, M. M., and D. A. Landis. 2007. Impact of intraguild predation by adult
Harmonia axyridis (Coleoptera: Coccinellidae) on Aphis glycines (Hemiptera:
Aphididae) biological control in cage studies. Biological Control 40: 386-395.
Gardiner, M. M., D. A. Landis, C. Gratton, N. Schmidt, M. O. Neal, E. Mueller, J.
Chacon, G. E. Heimpel, and C. D. DiFonzo. 2009a. Landscape composition
influences patterns of native and exotic lady beetle abundance. Diversity and
Distributions 15: 554-564.
Gardiner, M. M., D. A. Landis, C. Gratton, C. D. DiFonzo, M. O'Neal, J. M.
Chacon, M. T. Wayo, N. P. Schmidt, E. E. Mueller, and G. E. Heimpel.
2009b. Landscape diversity enhances biological control of an introduced crop
pest in the north-central USA. Ecological Applications 19: 143-154.
Gardiner, M. M., D. A. Landis, C. Gratton, C. D. DiFonzo, M. O'Neal, J. M.
Chacon, M. T. Wayo, N. P. Schmidt, E. E. Mueller, and G. E. Heimpel.
2009c. Landscape diversity enhances the biological control of an introduced crop
pest in the north-central USA. Ecological Applications 19: 143-154.
Griffiths, W., J.-P. Aurambout, H. Parry, P. Trebicki, D. Kriticos, G. O’Leary, K.
Finlay, P. D. Barro, and J. Luck. 2010. Insect-pathogen-crop dynamics and
their importance to plant biosecurity under future climates: Barley yellow dwarf
160
virus and wheat – a case study, Proceedings of 15th Agronomy Conference 2010,
Lincoln, New Zealand.
Hallett, R., T. Baute, and C. Bahlai. 2011. The Aphid Advisor app for Blackberry
smartphones. Available online: http://www.aphidapp.com/about.aspx
Hallett, R. H., S. A. Goodfellow, R. M. Weiss, and O. Olfert. 2009. MidgEmerge, a
new predictive tool, indicates the presence of multiple emergence phenotypes of
the overwintered generation of swede midge. Entomologia Experimentalis et
Applicata 130: 81-97.
Hallett, R. H., C. A. Bahlai, Y. Xue, and A. W. Schaafsma. In prep. Incorporating
Natural Enemy Units into a dynamic action threshold for the soybean aphid,
Aphis glycines (Hemiptera: Aphididae)
Hardie, J. 1981. Juvenile hormone and photoperiodically controlled polymorphism in
Aphis fabae: Postnatal effects on presumptive gynoparae. Journal of Insect
Physiology 27: 347-355.
Hardman, J. M. 1976. Deterministic and stochastic models simulating the growth of
insect populations over a range of temperatures under Malthusian conditions. The
Canadian Entomologist 108: 907-924.
Harrington, R., S. J. Clark, S. J. Welham, P. J. Verrier, C. H. Denholm, M. Hullé,
D. Maurice, M. D. Rounsevell, N. Cocu, and European Union Examine
Consortium. 2007. Environmental change and the phenology of European
aphids. Global Change Biology 13: 1550-1564.
Harrison, R. G. 1980. Dispersal polymorphisms in insects. Annual Review of Ecology
and Systematics 11: 95-118.
161
Hartigan, J. A., and P. M. Hartigan. 1985. The dip test of unimodality. The Annals of
Statistics 13: 70-84.
Health Canada Pest Management Agency. 2009. PMRA Re-evaluation Workplan
(April 2009 to March 2010). Government of Canada.
Heimpel, G., L. Frelich, D. Landis, K. Hopper, K. Hoelmer, Z. Sezen, M. Asplen,
and K. Wu. 2010. European buckthorn and Asian soybean aphid as components
of an extensive invasional meltdown in North America. Biological Invasions 12:
2913-2931.
Heimpel, G. E., D. W. Ragsdale, R. Venette, K. R. Hopper, R. J. Neil, C. E.
Rutledge, and Z. Wu. 2004. Prospects for importation biological control of the
soybean aphid: anticipating potential costs and benefits. Annals of the
Entomological Society of America 97: 249-258.
Henderson, C. F., and E. W. Tilton. 1955. Tests with Acaricides against the brown
wheat mite. Journal of Economic Entomology 48: 157-161.
Hesler, L. S., and K. E. Dashiell. 2007. Resistance to Aphis glycines (Hemiptera:
Aphididae) in various soybean lines under controlled laboratory conditions.
Journal of Economic Entomology 100: 1464-1469.
Hill, C. B., Y. Li, and G. L. Hartman. 2004. Resistance of Glycine species and various
cultivated legumes to the soybean aphid (Homoptera: Aphididae). Journal of
Economic Entomology 97: 1071-1077.
Hirano, K. 1996. Ecological characteristics and causes of the occurance of the soybean
aphid, Aphis glycines. (Translation). Shokubutsu Boeki 50: 17-21.
162
Hirano, K., K. Honda, and S. Miyai. 1996. Effects of temperature on development,
longevity and reproduction of the soybean aphid, Aphis glycines (Homoptera:
Aphididae). Applied Entomology and Zoology 31: 178-180.
Hodgson, E. W., R. C. Venette, M. Abrahamson, and D. W. Ragsdale. 2005. Alate
production of soybean aphid (Homoptera: Aphididae) in Minnesota.
Environmental Entomology 34: 1456-1463.
Holzmann, H., and S. Vollmer. 2008. A likelihood ratio test for bimodality in twocomponent mixtures with application to regional income distribution in the EU.
AStA Advances in Statistical Analysis 92: 57-69.
Huang, F., X. Ding, X. Wang, and Z. Huang. 1992. Studies on the spatial distribution
pattern of soybean aphid and sampling techniques. Journal of Shenyang
Agricultural University 23: 81-87.
Hunt, D., R. Foottit, D. Gagnier, and T. Baute. 2003. First Canadian records of Aphis
glycines (Hemiptera: Aphididae). Canadian Entomologist 135: 879-881.
Inoue, H. 1981. Major species of aphids and their seasonal occurrence on soybean in
Chikugo. (Translation). Proceedings of the Association for Plant Protection of
Kyushu 27: 109-111.
Isenhour, D. J., and K. V. Yeargan. 1981. Effect of temperature on the development of
Orius insidiosus, with notes on laboratory rearing. Annals of the Entomological
Society of America 74: 114-116.
Ito, Y. 1953. Studies on the population increase and the movements of soybean aphid,
Aphis glycines Matsumura (Translation). Oyo-Kontyu 8: 141-148.
163
Ives, A. R., and V. A. A. Jansen. 1998. Complex dynamics in stochastic tritrophic
models. Ecology 79: 1039-1052.
Jackman, S. 2012. Package ‘pscl’ for R. computer program, version 1.04.1.
James, K. H. 1990. Risk perceptions and food choice: An exploratory analysis of
organic- versus conventional-produce buyers. Risk Analysis 10: 367-374.
Johnson, K. D., M. E. Neal, J. D. Bradshaw, and M. E. Rice. 2008. Is preventative,
concurrent management of the soybean aphid (Hemiptera: Aphididae) and bean
leaf beetle (Coleoptera: Chrysomelidae) possible? Journal of Economic
Entomology 101: 801-809.
Johnson, K. D., M. E. O'Neal, D. W. Ragsdale, C. D. Difonzo, S. M. Swinton, P. M.
Dixon, B. D. Potter, E. W. Hodgson, and A. C. Costamagna. 2009. Probability
of cost-effective management of soybean aphid (Hemiptera: Aphididae) in North
America. Journal of Economic Entomology 102: 2101-2108.
Jones, E. 2004. Grants awarded to develop pesticide risk reduction programs. United
States Environmental Protection Agency, Press release 10/14/04.
Jonsson, M., S. D. Wratten, D. A. Landis, and G. M. Gurr. 2008. Recent advances in
conservation biological control of arthropods by arthropods. Biological Control
45: 172-175.
Kaiser, M. E., T. Noma, M. J. Brewer, K. S. Pike, J. R. Vockeroth, and S. D.
Gaimari. 2007. Hymenopteran parasitoids and dipteran predators found using
soybean aphid after its midwestern United States invasion. Annals of the
Entomological Society of America 100: 196-205.
164
Karley, A. J., W. E. Parker, J. W. Pitchford, and A. E. Douglas. 2004. The midseason crash in aphid populations: why and how does it occur? Ecological
Entomology 29: 383-388.
Kidd, N. A. C., and D. J. Tozer. 1984. Host plant and crowding effects in the induction
of alatae in the large pine aphid, Cinara pinea. Entomologia Experimentalis et
Applicata 35: 37-42.
Kiman, Z. B., and K. V. Yeargan. 1985. Development and reproduction of the predator
Orius insidiosus (Hemiptera: Anthocoridae) reared on diets of selected plant
material and arthropod prey. Annals of the Entomological Society of America 78:
464-467.
Kleijn, D., F. Berendse, R. Smit, and N. Gilissen. 2001. Agri-environment schemes do
not effectively protect biodiversity in Dutch agricultural landscapes. Nature 413:
723-725.
Klueken, A. M., B. Hau, B. Ulber, and H. M. Poehling. 2009. Forecasting migration of
cereal aphids (Hemiptera: Aphididae) in autumn and spring. Journal of Applied
Entomology 133: 328-344.
Knight, K., J. Kurylo, A. Endress, J. Stewart, and P. Reich. 2007. Ecology and
ecosystem impacts of common buckthorn ( Rhamnus cathartica ): a review.
Biological Invasions 9: 925-937.
Koch, R. L. 2003. The multicolored Asian lady beetle, Harmonia axyridis: A review of
its biology, uses in biological control, and non-target impacts. Journal of Insect
Science 3: 1-16.
165
Kovach, J., C. Petzolt, J. Degnil, and J. Tette. 1992. A method to measure the
environmental impact of pesticides. New York’s Food and Life Sciences Bulletin
139: 1-8.
Kovach, J., C. Petzolt, J. Degnil, and J. Tette. 2009. A method to measure the
environmental impact of pesticides: Table 2, List of pesticides. . Available online:
http://www.nysipm.cornell.edu/publications/eiq/default.asp?metatags_Action=Fin
d%28%27PID%27,%274%27%29#table2
Kraiss, H., and E. M. Cullen. 2008a. Insect growth regulator effects of azadirachtin and
neem oil on survivorship, development and fecundity of Aphis glycines
(Homoptera: Aphididae) and its predator, Harmonia axyridis (Coleoptera:
Coccinellidae). Pest Management Science 64: 660-668.
Kraiss, H., and E. M. Cullen. 2008b. Efficacy and nontarget effects of reduced-risk
insecticides on Aphis glycines (Hemiptera: Aphididae) and its biological control
agent Harmonia axyridis (Coleoptera: Coccinellidae). Journal of Economic
Entomology 101: 391-398.
Kriticos, D. J., M. S. Watt, T. M. Withers, A. Leriche, and M. C. Watson. 2009. A
process-based population dynamics model to explore target and non-target
impacts of a biological control agent. Ecological Modelling 220: 2035-2050.
Kriticos, D. J., J. R. Brown, G. F. Maywald, I. D. Radford, D. Mike Nicholas, R. W.
Sutherst, and S. W. Adkins. 2003. SPAnDX: a process-based population
dynamics model to explore management and climate change impacts on an
invasive alien plant, Acacia nilotica. Ecological Modelling 163: 187-208.
166
Kumar, A., V. Pandey, A. M. Shekh, and M. Kumar. 2008. Growth and yield response
of soybean (Glycine max L.) in relation to temperature, photoperiod and sunshine
duration at Anand, Gujarat, India. American-Eurasian Journal of Agronomy 1:
45-50.
Kurylo, J. S., K. S. Knight, J. R. Stewart, and A. G. Endress. 2007. Rhamnus
cathartica: native and naturalized distribution and habitat preferences. Journal of
Torrey Botanical Society 134: 420-430.
Lambers, D. H. R. 1966. Polymorphism in Aphididae. Annual Review of Entomology
11: 47-78.
Landis, D. A., S. D. Wratten, and G. M. Gurr. 2000. Habitat management to conserve
natural enemies of arthropod pests in agriculture. Annual Review of Entomology
45: 175-201.
Landis, D. A., M. M. Gardiner, W. van der Werf, and S. M. Swinton. 2008.
Increasing corn for biofuel production reduces biocontrol services in agricultural
landscapes. Proceedings of the National Academy of Sciences 105: 20552-20557.
Lanzoni, A., G. Accinelli, G. G. Bazzocchi, and G. Burgio. 2004. Biological traits and
life table of the exotic Harmonia axyridis compared with Hippodamia variegata,
and Adalia bipunctata (Col., Coccinellidae). Journal of Applied Entomology 128:
298-306.
Levitan, L., I. Merwin, and J. Kovach. 1995. Assessing the relative environmental
impacts of agricultural pesticides: the quest for a holistic method. Agriculture,
Ecosystems & Environment 55: 153-168.
167
Liu, J., K. Wu, K. R. Hopper, and K. Zhao. 2004. Population dynamics of Aphis
glycines (Homoptera: Aphididae) and its natural enemies in soybean in northern
China. Annals of the Entomological Society of America 97: 235-239.
Loxdale, H. D., J. I. M. Hardie, S. Halbert, R. Foottit, N. A. C. Kidd, and C. I.
Carter. 1993. The relative importance of short- and long- range movement of
flying aphids. Biological Reviews 68: 291-311.
Lu, L. H., and R. L. Chen. 1993. Analysis of factors inducing alatae in Aphis glycines.
(Translation). Acta Entomologica Sinica 36: 143-149.
Lynch, D. 2009. Environmental impacts of organic agriculture: A Canadian perspective.
Canadian Journal of Plant Science 89: 621-628.
Lynch, S., C. Greene, and C. Kramer-LeBlanc. 1996. Proceedings of the third national
IPM symposium/workshop: Broadening support for 21st century IPM. U.S.
Department of Agriculture, Economic Research Service, Natural Resources and
Environment Division.
Majerus, M., and P. Kearns. 1989. Ladybirds. The Richmond Publishing Co. Ltd,
Slough.
Mashanova, A., A. Gange, and V. Jansen. 2008. Density-dependent dispersal may
explain the mid-season crash in some aphid populations. Population Ecology 50:
285-292.
Maywald, G. F., D. J. Kriticos, R. W. Sutherst, and W. Bottomley. 2007. Dymex
Model Builder v.3 user guide. Hearn Scientific Software, Melbourne.
168
McCaffrey, J. P., and R. L. Horsburgh. 1986. Biology of Orius insidiosus
(Heteroptera: Anthocoridae): a predator in Virginia apple orchards.
Environmental Entomology 15: 984-988.
McCay, T., and D. McCay. 2009. Processes regulating the invasion of European
buckthorn (Rhamnus cathartica) in three habitats of the northeastern United
States. Biological Invasions 11: 1835-1844.
McCornack, B. P., D. W. Ragsdale, and R. C. Venette. 2004. Demography of soybean
aphid (Homoptera: Aphididae) at summer temperatures. Journal of Economic
Entomology 97: 854-861.
Miao, J., K. Wu, K. R. Hopper, and G. Li. 2007. Population dynamics of Aphis
glycines (Homoptera: Aphididae) and impact of natural enemies in northern
China. Environmental Entomology 36: 840-848.
Michaud, J. P. 2001. Colony density and wing development in Toxoptera citricida
(Homoptera: Aphididae). Environmental Entomology 30: 1047-1051.
Michel, A. P., W. Zhang, J. Kyo Jung, S.-T. Kang, and M. A. Rouf Mian. 2009.
Population genetic structure of Aphis glycines. Environmental Entomology 38:
1301-1311.
Mignault, M.-P., M. Roy, and J. Brodeur. 2006. Soybean aphid predators in Québec
and the suitability of Aphis glycines as prey for three Coccinellidae. BioControl
51: 89-106.
Ministry of Science and Technology of the People´s Republic of China. 2001.
National High-tech R&D Program (863 Program)
169
Mittler, T. E., and R. H. Dadd. 1962. Artificial feeding and rearing of the aphid, Myzus
persicae (Sulzer), on a completely defined synthetic diet. Nature 195: 404-404.
Mittler, T. E., and J. E. Kleinjan. 1970. Effect of artificial diet composition on wingproduction by the aphid Myzus persicae. Journal of Insect Physiology 16: 833850.
Moran, N. A. 1992. The evolution of aphid life cycles. Annual Review of Entomology
37: 321-348.
Mousseau, T. A., and H. Dingle. 1991. Maternal effects in insect life histories. Annual
Review of Entomology 36: 511-534.
Müller, C. B., I. S. Williams, and J. Hardie. 2001. The role of nutrition, crowding and
interspecific interactions in the development of winged aphids. Ecological
Entomology 26: 330-340.
Myers, S. W., D. B. Hogg, and J. L. Wedberg. 2009. Determining the optimal timing of
foliar insecticide applications for control of soybean aphid (Hemiptera:
Aphididae) on soybean. Journal of Economic Entomology 98: 2006-2012.
Nault, B. A., D. A. Shah, H. R. Dillard, and A. C. McFaul. 2004. Seasonal and spatial
dynamics of alate aphid dispersal in snap bean fields in proximity to alfalfa and
implications for virus management. Environmental Entomology 33: 1593-1601.
Newman, J. A., D. J. Gibson, A. J. Parsons, and J. H. M. Thornley. 2003. How
predictable are aphid population responses to elevated CO2? Journal of Animal
Ecology 72: 556-566.
Nielsen, C., and A. E. Hajek. 2005. Control of invasive soybean aphid, Aphis glycines
(Hemiptera: Aphididae), populations by existing natural enemies in New York
170
State, with emphasis on entomopathogenic fungi. Environmental Entomology 34:
1036-1047.
Noma, T., and M. J. Brewer. 2008. Seasonal abundance of resident parasitoids and
predatory flies and corresponding soybean aphid densities, with comments on
classical biological control of soybean aphid in the midwest. Journal of Economic
Entomology 101: 278-287.
O'Hara, R. B., and D. J. Kotze. 2010. Do not log-transform count data. Methods in
Ecology and Evolution 1: 118-122.
Obrycki, J. J., and M. J. Tauber. 1981. Phenology of three coccinellid species: thermal
requirements for development. Annals of the Entomological Society of America
74: 31-36.
Ohnesorg, W. J., K. D. Johnson, and M. E. O'Neal. 2009. Impact of reduced-risk
Insecticides on soybean aphid and associated natural enemies. Journal of
Economic Entomology 102: 1816-1826.
Olson, K. D., T. Badibanga, and C. DiFonzo. 2008. Farmers' awareness and use of IPM
for soybean aphid control: report of survey results for the 2004, 2005, 2006, and
2007 crop years. . University of Minnesota, Department of Applied Economics.
OMAFRA. 2005. Field crop protection guide 2005-2006. Ontario Ministry of
Agriculture, Food and Rural Affairs.
Onstad, D. W. 1988. Population-dynamics theory: The roles of analytical, simulation,
and supercomputer models. Ecological Modelling 43: 111-124.
171
Onstad, D. W., S. Fang, and D. J. Voegtlin. 2005. Forecasting seasonal population
growth of Aphis glycines (Hemiptera: Aphididae) in soybean in Illinois. Journal of
Economic Entomology 98: 1157-1162.
Parry, H. R., J.-P. Aurambout, and D. J. Kriticos. 2011. Having your cake and eating
it: A modelling framework to combine process-based population dynamics and
dispersal simulation, pp. 2535-2541, 19th International Congress on Modelling
and Simulation, Perth, Australia.
Peck, S. L. 2004. Simulation as experiment: a philosophical reassessment for biological
modeling. Trends in Ecology & Evolution 19: 530-534.
Pedersen, P. 2009. Soybean growth and development. PM 1945. Iowa State University
Extension, Ames.
Pergams, O. R. W., and J. E. Norton. 2006. Treating a single stem can kill the whole
shrub: a scientific assessment of buckthorn control methods. Natural Areas
Journal 26: 300-309.
Phoofolo, M. W., J. J. Obrycki, and E. S. Krafsur. 1995. Temperature-dependent
ovarian development in Coccinella septempunctata (Coleoptera: Coccinellidae).
Annals of the Entomological Society of America 88: 72-79.
Pilkey-Jarvis, L., and O. H. Pilkey. 2008. Useless arithmetic: Ten points to ponder
when using mathematical models in environmental decision making. Public
Administration Review 68: 470-479.
Ragsdale, D. W., D. J. Voegtlin, and R. J. O'Neil. 2004. Soybean aphid biology in
North America. Annals of the Entomological Society of America 97: 204-208.
172
Ragsdale, D. W., D. A. Landis, J. Brodeur, G. E. Heimpel, and N. Desneux. 2011.
Ecology and management of the soybean aphid in North America. Annual Review
of Entomology 56: 375-399.
Ragsdale, D. W., B. P. McCornack, R. C. Venette, B. D. Potter, I. V. MacRae, E. W.
Hodgson, M. E. Neal, K. D. Johnson, R. J. Neil, C. D. DiFonzo, T. E. Hunt, P.
A. Glogoza, and E. M. Cullen. 2007. Economic threshold for soybean aphid
(Hemiptera: Aphididae). Journal of Economic Entomology 100: 1258-1267.
Reganold, J. P., J. D. Glover, P. K. Andrews, and H. R. Hinman. 2001. Sustainability
of three apple production systems. Nature 410: 926-930.
Rhainds, M., H. J. S. Yoo, L. Bledsoe, C. S. Sadof, S. Yaninek, and R. J. O'Neil.
2010a. Impact of developmental maturity of soybean on the seasonal abundance
of soybean aphid (Hemiptera: Aphididae). Environmental Entomology 39: 484491.
Rhainds, M., H. J. S. Yoo, K. L. Steffey, D. J. Voegtlin, C. S. Sadof, S. Yaninek, and
R. J. O'Neil. 2010b. Potential of suction traps as a monitoring tool for Aphis
glycines (Hemiptera: Aphididae) in soybean fields. Journal of Economic
Entomology 103: 186-189.
Rhainds, M., H. J. S. Yoo, P. Kindlmann, D. Voegtlin, D. Castillo, C. Rutledge, C.
Sadof, S. Yaninek, and R. J. O'Neil. 2010c. Two-year oscillation cycle in
abundance of soybean aphid in Indiana. Agricultural and Forest Entomology 12:
251-257.
173
Rongcai, Y., Y. Ming, and W. Guizhu. 1994. Study on control of soybean aphid by
Harmonia (Leis) axyridis. (Translation). Ji Lin Agricultural Science 44: 30-32,
57.
Ropeik, D. 2004. The consequences of fear. EMBO Rep 5: S56-S60.
Rosenheim, J. A., S. Parsa, A. A. Forbes, W. A. Krimmel, Y. H. Law, M. Segoli, M.
Segoli, F. S. Sivakoff, T. Zaviezo, and K. Gross. 2011. Ecoinformatics for
integrated pest management: expanding the applied insect ecologist's tool-kit.
Journal of Economic Entomology 104: 331-342.
Rosswurm, M. A., and J. H. Larrabee. 1999. A model for change to evidence-based
practice. Journal of Nursing Scholarship 31: 317-322.
Rutledge, C. E., and R. J. O'Neil. 2006. Soybean plant stage and population growth of
soybean aphid. Journal of Economic Entomology 99: 60-66.
Rutledge, C. E., R. J. Neil, T. B. Fox, and D. A. Landis. 2004. Soybean aphid predators
and their use in integrated pest management. Annals of the Entomological Society
of America 97: 240-248.
Schellhorn, N. A., and D. A. Andow. 1999. Mortality of Coccinellid (Coleoptera:
Coccinellidae) larvae and pupae when prey become scarce. Environmental
Entomology 28: 1092-1100.
Schmidt, J. M., P. C. Richards, H. Nadel, and G. Ferguson. 1995. A rearing method
for the production of large numbers of the insidious flower bug, Orius insidiosus
(Say) (Hemiptera:Anthocoridae) The Canadian Entomologist 127: 445-447.
Schmidt, N. P., M. E. O'Neal, P. F. Anderson, D. Lagos, D. Voegtlin, W. Bailey, P.
Caragea, E. Cullen, C. DiFonzo, K. Elliott, C. Gratton, D. Johnson, C. H.
174
Krupke, B. McCornack, R. O'Neil, D. W. Ragsdale, K. J. Tilmon, and J.
Whitworth. 2012. Spatial distribution of Aphis glycines (Hemiptera: Aphididae):
a summary of the suction trap network. Journal of Economic Entomology 105:
259-271.
Scott, R. W., and F. A. Huff. 1996. Impacts of the Great Lakes on regional climate
conditions. Journal of Great Lakes Research 22: 845-863.
Shepherd, M., B. Pearce, B. Cormack, L. Philipps, S. Cuttle, A. Bhogal, P. Costigan,
and R. Unwin. 2003. An assessment of the environmental impacts of organic
farming. United Kingdom Department for Environment, Food and Rural Affairs.
Shortall, C. R., A. Moore, E. Smith, M. J. Hall, I. P. Woiwod, and R. Harrington.
2009. Long-term changes in the abundance of flying insects. Insect Conservation
and Diversity 2: 251-260.
Shusen, S., Y. Boren, L. Dianshen, and Y. Yanjie. 1994. Study on space dynamics of a
natural population of Aphis glycines Matsumura. (Translation). Journal of Jilin
Agriculture University 16: 75-79.
Smith, M. A. H., and P. A. MacKay. 1989. Seasonal variation in the photoperiodic
responses of a pea aphid population: evidence for long-distance movements
between populations. Oecologia 81: 160-165.
Statistics Canada. 2009. Cereals and oilseeds review. Ministry of Industry, Government
of Canada.
Straub, C. S., and W. E. Snyder. 2008. Increasing enemy biodiversity strengthens
herbivore suppression on two plant species. Ecology 89: 1605-1615.
175
Straub, C. S., D. L. Finke, and W. E. Snyder. 2008. Are the conservation of natural
enemy biodiversity and biological control compatible goals? Biological Control
45: 225-237.
Su, J., K. Hao, and X. Shi. 1996. Spatial distrubution and sampling technique of Aphis
glycines Matsumura. (Translation). Journal of Nanjing Agricultural University 13:
55-58.
Suckling, D. M., J. T. S. Walker, and C. H. Wearing. 1999. Ecological impact of three
pest management systems in New Zealand apple orchards. Agriculture,
Ecosystems & Environment 73: 129-140.
Taper, M. L. 2004. Model identification from many candidates, pp. 488-524. In M. L.
Taper and S. R. Lele [eds.], The nature of scientific evidence: statistical,
philosophical, and empirical considerations. University of Chicago Press,
Chicago.
Taper, M. L., and S. R. Lele [eds.]. 2004. The nature of scientific evidence: statistical,
philosophical and empirical considerations. University of Chicago Press, Chicago.
Taylor, L. R. 1960. Mortality and viability of insect migrants high in the air. Nature 186:
410-410.
Taylor, L. R. 1974. Insect migration, flight periodicity and the boundary layer. Journal
of Animal Ecology 43: 225-238.
Taylor, L. R. 1977. Migration and the spatial dynamics of an aphid, Myzus persicae.
Journal of Animal Ecology 46: 411-423.
176
Teulon, D. A. J., M. A. W. Stufkens, and J. D. Fletcher. 2004. Crop infestation by
aphids is related to flight activity detected with 7.5 metre high suction traps. New
Zealand Plant Protection 57: 227-232.
Thompson, D. G., and D. P. Kreutzweiser. 2006. A review of the environmental fate
and effects of natural "reduced-risk" pesticides in Canada, pp. 245-274, Crop
Protection Products for Organic Agriculture. American Chemical Society,
Washington, DC.
Tian, Z., S. Zhao, and C. Hu. 1990. Study of the prediction of developmental stage and
population size of soybean aphid in Northern Liaoning, China. (Translation).
Plant Protection (Zhiwu Baohu) 16: 19-21.
Tilmon, K. J., E. W. Hodgson, M. E. O'Neal, and D. W. Ragsdale. 2011. Biology of
the soybean aphid, Aphis glycines (Hemiptera: Aphididae) in the United States.
Journal of Integrated Pest Management 2: A1-A7.
Toba, H. H., J. D. Paschke, and S. Friedman. 1967. Crowding as the primary factor in
the production of the agamic alate form of Therioaphis maculata (Homoptera:
Aphididae). Journal of Insect Physiology 13: 381-396.
Trewavas, A. 2001. Urban myths of organic farming. Nature 410: 409-410.
U.K. Department for Environment Food and Rural Affairs and the Forestry
Commission. 2005. Departmental report 2005. United Kingdom Department for
Environment, Food and Rural Affairs.
USDA. 2011. Soybean Aphid Integrated Pest Management Platform for Extension and
Education (IPM-PIPE). Available online: http://sba.ipmpipe.org/cgibin/sbr/public.cgi?host=All%20Legumes/Kudzu&pest=soybean_aphid
177
Van Den Berg, H., D. Ankasah, A. Muhammad, R. Rusli, H. A. Widayanto, H. B.
Wirasto, and I. Yully. 1997. Evaluating the role of predation in population
fluctuations of the soybean aphid Aphis glycines in farmer's fields in Indonesia.
Journal of Applied Ecology 34: 971-984.
Venette, R. C., and D. W. Ragsdale. 2004. Assessing the invasion by soybean aphid
(Homoptera: Aphididae): where will it end? Annals of the Entomological Society
of America 97: 219-226.
Via, S. 1992. Inducing the sexual forms and hatching the eggs of pea aphids.
Entomologia Experimentalis et Applicata 65: 119-127.
Voegtlin, D. J., R. J. Neil, and W. R. Graves. 2004a. Tests of suitability of
overwintering hosts of Aphis glycines: identification of a new host association
with Rhamnus alnifolia L'Heritier. Annals of the Entomological Society of
America 97: 233-234.
Voegtlin, D. J., S. E. Halbert, and G. Qiao. 2004b. A guide to separating Aphis
glycines Matsumura and morphologically similar species that share its hosts.
Annals of the Entomological Society of America 97: 227-232.
Voegtlin, D. J., R. J. O'Neil, W. R. Graves, D. Lagos, and H. J. S. Yoo. 2005.
Potential winter hosts of soybean aphid. Annals of the Entomological Society of
America 98: 690-693.
Wang, X., X. Ding, and F. Huang. 1991. Studies on the spatial distribution of aphideating ladybirds in soybean fields. (Translation). Journal of Shenyang
Agricultural University 22: 13-16.
178
Welsman, J. A. 2007. Ecology and control of the soybean aphid, Aphis glycines
Matsumura (Homoptera: Aphididae). dissertation. University of Guelph, Guelph.
Welsman, J. A., C. A. Bahlai, M. K. Sears, and A. W. Schaafsma. 2007. Decline of
soybean aphid (Homoptera: Aphididae) egg populations from autumn to spring on
the primary host, Rhamnus cathartica. Environmental Entomology 36: 541548(8).
Wertheim, B., E.-J. A. van Baalen, M. Dicke, and L. E. M. Vet. 2005. Pheromonemediated aggregation in nonsocial arthropods: an evolutionary ecological
perspective. Annual Review of Entomology 50: 321-346.
White, N. A., S. Chakraborty, and G. Murray. 2004. A linked process-based model to
study the interaction between Puccinia striiformis and wheat, 4th International
Crop Science Congress, Brisbane, Australia.
Woiwod, I. P., G. M. Tatchell, and A. M. Barrett. 1984. A system for the rapid
collection, analysis and dissemination of aphid-monitoring data from suction
traps. Crop Protection 3: 273-288.
Worner, S. P., G. M. Tatchell, and I. P. Woiwod. 1995. Predicting spring migration of
the damson-hop aphid Phorodon humuli (Homoptera: Aphididae) from historical
records of host-plant flowering phenology and weather. Journal of Applied
Ecology 32: 17-28.
Wu, Z., D. Schenk-Hamlin, W. Zhan, D. W. Ragsdale, and G. E. Heimpel. 2004. The
soybean aphid in China: a historical review. Annals of the Entomological Society
of America 97: 209-218.
179
Wyckhuys, K., R. Koch, R. Kula, and G. Heimpel. 2009. Potential exposure of a
classical biological control agent of the soybean aphid, Aphis glycines , on nontarget aphids in North America. Biological Invasions 11: 857-871.
Xue, Y., C. A. Bahlai, A. Frewin, M. K. Sears, A. W. Schaafsma, and R. H. Hallett.
2009. Predation by Coccinella septempunctata and Harmonia axyridis
(Coleoptera: Coccinellidae) on Aphis glycines (Homoptera: Aphididae).
Environmental Entomology 38: 708-714.
Xue, Y., C. Bahlai, A. Frewin, C. McCreary, L. Des Marteaux, A. Schaafsma, and R.
Hallett. 2012. Intraguild predation of the aphid parasitoid Aphelinus certus by
Coccinella septempunctata and Harmonia axyridis. Biocontrol: 1-8.
Yoo, H. J. S., and R. J. O'Neil. 2009. Temporal relationships between the generalist
predator, Orius insidiosus, and its two major prey in soybean. Biological Control
48: 168-180.
Yoo, H. J. S., R. J. O'Neil, D. J. Voegtlin, and W. R. Graves. 2005. Host plant
suitability of Rhamnaceae for soybean aphid (Homoptera: Aphididae). Annals of
the Entomological Society of America 98: 926-930.
Zera, A. J., and R. F. Denno. 1997. Physiology and ecology of dispersal polymorphism
in insects. Annual Review of Entomology 42: 207-230.
Zhang, G., and T. Zhong. 1982. Research on the life cycle patterns of several aphids.
Sinozoologia 2: 7-17.
Zhang, W., and S. M. Swinton. 2009. Incorporating natural enemies in an economic
threshold for dynamically optimal pest management. Ecological Modelling 220:
1315-1324.
180
Zhang, Y., W. U. Kongming, K. A. G. Wyckhuys, and G. E. Heimpel. 2009a. Tradeoffs between flight and fecundity in the soybean aphid (Hemiptera: Aphididae).
Journal of Economic Entomology 102: 133-138.
Zhang, Y., K. Wu, K. A. G. Wyckhuys, and G. E. Heimpel. 2009b. Effect of
parasitism on flight behavior of the soybean aphid, Aphis glycines. Biological
Control 51: 475-479.
Zhang, Y., L. Wang, K. Wu, K. A. G. Wyckhuys, and G. E. Heimpel. 2008. Flight
performance of the soybean aphid, Aphis glycines (Hemiptera: Aphididae) under
different temperature and humidity regimens. Environmental Entomology 37:
301-306.
Zhou, X., R. Harrington, I. P. Woiwod, J. N. Perry, J. S. Bale, and S. J. Clark. 1995.
Effects of temperature on aphid phenology. Global Change Biology 1: 303-313.
Zhu, J., and K.-C. Park. 2005. Methyl salicylate, a soybean aphid-induced plant volatile
attractive to the predator Coccinella septempunctata. Journal of Chemical
Ecology 31: 1733-1746.
Zhu, J., A. Zhang, K. C. Park, T. Baker, B. Lang, R. Jurenka, J. J. Obrycki, W. R.
Graves, J. A. Pickett, D. Smiley, K. R. Chauhan, and J. A. Klun. 2006a. Sex
pheromone of the soybean aphid, Aphis glycines Matsumura, and its potential use
in semiochemical-based control. Environmental Entomology 35: 249-257.
Zhu, M., E. B. Radcliffe, D. W. Ragsdale, I. V. MacRae, and M. W. Seeley. 2006b.
Low-level jet streams associated with spring aphid migration and current season
spread of potato viruses in the U.S. northern Great Plains. Agricultural and Forest
Meteorology 138: 192-202.
181
Zuur, A. F., E. N. Ieno, and G. M. Smith. 2007. Analysing ecological data. Springer,
New York.
Zuur, A. F., E. N. Ieno, and C. S. Elphick. 2009. A protocol for data exploration to
avoid common statistical problems. Methods in Ecology and Evolution 1: 3-14.
182
APPENDIX 1
Covariance structure of environmental parameters used in
Chapter 3
Figure A-1. Scatter plot matrix to examine covariance structure between parameters used in models
for captures of soybean aphid in suction traps in central North American, 2005-2009. Parameters are
1) Suction trap captures, 2) Degree day accumulation, 3) Field infestation, 4) Julian date 5) Latitude,
6) Longitude, 7) Maximum temperature, 8) Mean temperature, 9) Minimum temperature, 10)
Photoperiod, and 11) Precipitation, as described in Table 3-2.
183
APPENDIX 2
Supplemental information regarding pesticide selection
Table A-1: Complete list of insecticides under consideration provided by
Agriculture and Agri-Food Canada (AAFC)
Active
ingredient
(trade name,
supplier)
Mechanism of
action
Comments from AAFC*
Evaluation
Include in
trial?
HGW86 (N/A,
Dupont)
Unknown
Presented by the company at
the Minor Use Meeting
No interest from
supplier
No
Pymetrozine
(Fulfil,®
Syngenta)
Affects
neuromuscular
junctions
Reviewed as Reduced Risk
(RR) pesticide for aphid
control in US. Registered in
Canada for aphids in
potatoes but not soybeans
No interest from
supplier
No
Spirotetramat
(Movento®,
Bayer)
Fatty acid
biosynthesis inhibitor
Reviewed as RR product by
IR-4/US. Potential for
registration in legume
vegetable crop group,
soybean, etc
Novel MOA in
soybeans, interest
from supplier
Yes
Beauveria
bassiana
(Botanigard®,
Laverlam)
Entomopathogenic
fungus
Biopesticide
Novel MOA in
soybeans, interest
from supplier, can
be used in
organic-certified
crops
Yes
Imidacloprid
(Admire®,
Bayer)
Acetylcholine agonist
Reviewed as OP
replacement by IR-4/US
Same class of
insecticide as seed
treatments
registered in
soybean.
No
mineral oil
(Superior 70
Oil®, UAP)
Oxygen exchange
-
Novel MOA in
soybeans, interest
from supplier, can
be used in organic
certified crops
Yes
MOI-201plant
extract
(Marrone
Organic
Innovations)
Unknown
Botanical, see attached
presentation
No interest from
supplier
No
184
Active
ingredient
(trade name,
supplier)
Mechanism of
action
Comments from AAFC*
Evaluation
Include in
trial?
Flonicamid
(Beleaf ®,
ISK)
Neurotoxin- affects
potassium channels
Reviewed as OP
replacement by IR-4/US
Novel MOA in
soybeans, interest
from supplier
Yes
Clothianidin
(Poncho®,
Bayer)
Acetylcholine agonist
Reviewed as RR and OP
replacement product by IR4/US
Same class of
insecticide as seed
treatments
registered in
soybean.
No
Acetamiprid
(Assail®,
Nisso America
Inc.)
Acetylcholine agonist
Reviewed as RR and OP
replacement product by IR4/US
Same class of
insecticide as seed
treatments
registered in
soybean.
No
Thiamethoxa
m (Actara®,
Syngenta)
Acetylcholine agonist
Reviewed as OP
replacement by IR-4/US
Same class of
insecticide as seed
treatments
registered in
soybean.
No
Cinnamaldehy
de
(Proguard®,
Proguard Inc.)
Unknown
Natural product IR-4/US
registered in soybean in US
Product has been
discontinued
No
Spinetoram
(Delegate®,
Dow
AgroSciences)
Acetylcholine agonist
(but not at
neonicotinoid site)
Reviewed as RR product by
IR-4/US.
No interest from
supplier
No
Chrysoperla
carnea
(Kagetaro®,
Arysta
Lifesciences)
Predator
Biopesticide IR-4/US
(pepper & strawberries)
Not compatible
with foliar spray
technologies
No
* OP = organophosphorus insecticide, RR= reduced risk pesticide, IR-4/US= “Interregional project #
4”- an ongoing pesticide risk reduction program commissioned by the United States Department of
Agriculture.
185