 

a:

. . . . ‘ «MPH»...
a3ﬁ§1§ . . ‘ t.f . ‘.

...§_:;.~I

Z. ; .

 

 

 

735513“ LIBRARY
3) MlChtgcul State
9 W1 . University

 

 

This is to certify that the
dissertation entitled

LANDSCAPE SCALE AND WITHIN-FIELD INFLUENCES ON
PREDATOR ABUNDANCE AND BIOCONTROL SERVICES
IN SOYBEAN FIELDS.

presented by

MARY MARGARET GARDINER

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. ENTOMOLOGY

 

 

 

Wjo Professor’s Signature
{<1 / all 1/0?

Date

MSU is an affirmative-action, equal-opportunity employer

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K'IProj/Acc&Pres/CIRC/DateDue indd

 

 

LANDSCAPE SCALE AND WITHIN-FIELD INFLUENCES ON PREDATOR
ABUNDANCE AND BIOCONTROL SERVICES IN SOYBEAN FIELDS.

By

Mary Margaret Gardiner

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Entomology

2008

ABSTRACT

LANDSCAPE SCALE AND WITHIN-FIELD INFLUENCES ON PREDATOR
ABUNDANCE AND BIOCONTROL SERVICES IN SOYBEAN FIELDS.

By
Mary Margaret Gardiner

The soybean aphid, Aphis glycines Matumura, is an invasive species ﬁrst detected
in the US. in 2000. This pest has now spread throughout soybean growing regions of the
US. and Canada. Previous work has demonstrated that natural enemies are key drivers
of soybean aphid population dynamics within soybean ﬁelds. The objectives of my
dissertation research were to investigate how landscape diversity and composition
inﬂuenced the species pool of potential biocontrol agents supplied to soybean ﬁelds and
determine how within-ﬁeld interactions between these species affected net biocontrol
services. I investigated these objectives in replicate sites across Iowa, Michigan,
Minnesota, and Wisconsin from 2005-2007. I measured the impact of predators by
experimentally excluding or allowing access to soybean aphid infested plants and
comparing aphid population growth over 14 d. I found that while predators signiﬁcantly
suppressed soybean aphid across the region, there was a great deal of site to site
variation. To determine how the amount of biocontrol services supplied to a soybean
ﬁeld was inﬂuenced by the surrounding landscape, I calculated a biological control
services index (881) based on relative suppression of aphid populations and related it to
landscape diversity and composition at multiple spatial scales surrounding each site. BSI
values increased with landscape diversity, measured as Simpson’s D. Landscapes
dominated by com and soybean ﬁelds provided less biocontrol service to soybean

compared with landscapes with an abundance of crop and non-crop habitats.

The distribution and abundance of natural and agricultural habitats within the
landscape also greatly inﬂuenced the abundance of key predator species including
Coccinellidae, Carabidae, and Araneae. I found variation in the response to landscape
variables of the study organisms based on their trophic level, phenology, and whether
they were native to the north-central US. The activity density of Araneae increased in
soybean ﬁelds in landscapes with an abundance of forests and grasslands. Relationships
with Carabidae activity density varied, with adult-overwinten'ng and predatory species
increasing and larval-overwintering species declining with the abundance of grassland in
the landscape. The coccinellid community also exhibited varied responses to landscape
diversity and composition. Native coccinellids were most abundant in low-diversity
landscapes with an abundance of grassland habitat while exotic coccinellid increased
with the abundance of forested habitats.

I also investigated how within-ﬁeld intraguild interactions affected species
abundance and biocontrol services in soybean ﬁelds. I found that H. axyrz'dis was a
signiﬁcant intraguild predator of both Aphidoletes aphidimyza and Chrysoperla carnea,
however, the direct impact of H. axyridis on A. glycines overcame these negative impacts
on overall biocontrol. I also measured the extent of IGP of native coccinellid eggs in
soybean ﬁelds. I found signiﬁcant IGP occurred and the composition of the potential egg
predators varied by state. In Michigan, exotic coccinellids were the most abundant
predators while in Iowa native species including C. maculata, Hippodamia parenthesis,
and Collops nigriceps dominated the predator community. Thus, in Michigan
landscapes, consumption of native coccinellid eggs favors exotic coccinellids while in

Iowa, this resource is more likely to sustain other native species.

For Steve

iv

ACKNOWLEDGEMENTS

I would like to thank Dr. Doug Landis for being an outstanding advisor. Doug
has provided the support, encouragement, and “opportunities” that have shown me that I
can be a successful researcher. His drive to understand arthropod-mediated ecosystem
services at the landscape scale is a perspective that I hope to contribute to throughout my
career. I would like to thank my committee members, Chris DiFonzo, Jim Miller,
Carolyn Malmstrom, and Mike Brewer all of whom have greatly contributed to my
experience at MSU with their expertise and support throughout the completion of this
degree.

This project was conducted at a regional scale, and I feel that I have three
additional committee members: Dr. Matt O’Neal, Dr. Claudio Gratton, and Dr. George
Heirnpel who have been outstanding mentors to me throughout this degree. I would like
to thank Dr. Matt O’Neal, Nick Schmidt, and Kevin Johnson at Iowa State University for
all of their help ﬁnding ﬁeld sites, collecting and processing data, collecting thousands of
C. maculata eggs, editing, and for hosting Steve and I during our annual Iowa trip. I
thank Dr. Claudio Gratton and Emily Mueller from the University of Wisconsin for
establishing and sampling ﬁeld sites, ground truthing, and editing. I would also like to
thank Dr. Claudio Gratton for his statistical advice and training which have greatly
improved my skills and the quality of my analyses. At the University of Minnesota I
would like to thank Dr. George Heirnpel, Jeremy Chacon and Bruce Potter for

establishing and sampling ﬁeld sites and editing manuscripts. As a team, we were able to

produce three publications with nine + co-authors which speaks to the dedication and
hard work of this group of people.

I would like to thank Chris Sebolt and Mike J ewett for all of their help
establishing and sampling ﬁeld sites in Michigan, they were both a great support to me
and my undergraduate assistants throughout this project. I would also like to thank all of
the students and post-docs in the Landis Lab: Anna Fiedler, Megan Woltz, Rob Ahem,
Jeff Evans, Chris Hamm, and Alejandro Costamagna and the Entomology Department at
MSU for their ﬁiendship and support, we have had many great lunch hour discussions
over the last four years.

This project would not have been possible without the hard work of many
undergraduate students. In particular I would like to acknowledge Mike Wayo,
Aleksandra Kalenak and Nick Longbucco who contributed not only their time and energy
to sampling but also made this project stronger through the critical thought and creativity
they invested. I would also like to thank the staff at all of the research stations were we
conducted our work, in particular to Paul Horny who makes great BBQ. Thank you to
the Bean and Beet Farm (MSU), Kellogg Biological Station (MSU), West Madison
Agricultural Research Station (UW), Arlington Agricultural Research Station (UW),
Hancock Agricultural Research Station (UW), UMore Park (MN), Southwest Research
and Outreach Center (UM), McNay Research and Demonstration Farm (ISU), Northwest
Research and Demonstration Farm (ISU), Northeast Research and Demonstration Farm
(ISU), and the Johnston Farm (ISU).

One of the things that I really enjoyed during my time at MSU was my exposure

to teaching and extension. I would like to thank Joyce Parker and Andy Anderson who

vi

are amazing mentors in undergraduate education. I would like to thank Doug Landis and
Chris DiFonzo for allowing me to work as their teaching assistant and try out new
teaching techniques in their classrooms. In addition, I thank Chris DiFonzo for all of her
mentorship in extension and Anna Fiedler who I have had the opportunity to work with
on several extension activities. I would also like to thank my collaborators Joy Landis,
Chris DiFonzo, Mike Brewer and Takuji Noma for all of their work on our extension
bulletin “Identifying Natural Enemies in Crops and Landscapes.”

I would also like to acknowledge Lauren Bailey who has done an amazing job
running the biodiversity study for the Great Lakes Bioenergy Research Center and Nick
Longbucco who has been instrumental in development of ﬁeld experiments for this
project. Lauren, Nick, Denae Baker and Mike Bigsby have all worked very hard to
establish ﬁeld sites across Michigan this spring. I would not have been able to begin
ﬁeld work on this project this summer as I completed my degree without their hard work.

Finally, one of the great things about moving back to Michigan to attend MSU
was being closer to my family and friends. I would like to thank my parents Mike and
Susan Gardiner and my sister Sarah Gardiner for all of their love and laughter over the
years. I would also like to thank the Mazur and Scott families for their love and support.
Most of all, I thank my husband Steve Scott. Thank you for mapping the Midwest three
times, saving eggs from thunderstorms, and obtaining video of bugs eating other bugs at

3 am. Your love, encouragement, and humor are what keep me going!

vii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ xii
LIST OF FIGURES .............................................................................. xv
KEY TO SYMBOLS AND ABBREVIATIONS ........................................... xxi
CHAPTER 1
INFLUENCE OF LANDSCAPE AND WITHIN-FIELD INTERACTIONS ON
ARTHROPOD FOOD WEBS IN AGROECOSYSTEMS ................................. 1
Introduction .................................................................................. 1
Biology of the soybean aphid .................................................... 3
Impacts of soybean aphid ......................................................... 4
Natural enemies of A. glycines in soybean ..................................... 5
Impact of natural enemies on A. glycines population dynamics ............ 7
Within-ﬁeld interactions and the soybean aphid food web ............... 9
A conceptual framework of IGP and foraging behavior .................. 12
IGP in the A. glycines system ................................................ 14
Impacts of landscape variables on insects in agroecosystems .............. 15
Emotional spatial scale and trophic level ..................................... 19
Inﬂuence of landscape variables depends on farming practice ............ 20
Inﬂuence of the agricultural landscape in the soybean system ............. 21
Summary ........................................................................... 22
CHAPTER 2

IMPACT OF INTRAGUILD PREDATION BY ADULT HARMONIA AXYRIDIS
(COLEOPT ERA: COCCINELLIDAE) ON APHIS GLYCINES (HEMIPTERA:

APHIDIDAE) BIOLOGICAL CONTROL IN CAGE STUDIES ....................... 23
Abstract .................................................................................... 23
Introduction ................................................................................ 24

Intraguild-interactions ........................................................... 25
Impact of IGP on early season biological control ........................... 26
Methods .................................................................................... 27
Rearing ............................................................................ 27
Predator consumption rates ..................................................... 28
Microcosm design .................................................................................... 28
Microcosm experimental procedure ......................................................... 29
Field cage experiment ............................................................ 30
Cage design ........................................................................ 31

viii

Sampling procedure .............................................................. 31

Statistics ........................................................................... 32
Results ....................................................................................... 33
Predator consumption rates ...................................................... 33
Aphidoletes aphidimyza and H. axyridis ....................................... 33
Chrysoperla carnea and H. axyridis ........................................... 34
Field cage experiment ............................................................ 35
Discussion .................................................................................. 37
Microcosm studies ............................................................... 37
Field cages ........................................................................ 39
Impacts of H. axyridis on biological control ................................. 41
Conclusions ........................................................................ 43
CHAPTER 3
LANDSCAPE DIVERSITY ENHANCES BIOLOGICAL CONTROL OF AN
INTRODUCED CROP PEST IN THE NORTH-CENTRAL U.S ...................... 51
Abstract ..................................................................................... 51
Introduction ................................................................................. 52
Methods ..................................................................................... 55
Field sites ......................................................................... 55
Natural enemy impact experimental design .................................. 55
Assessment of cage effects ...................................................... 56
Soybean aphid and natural enemy sampling .................................. 5 7
Landscape analysis ............................................................... 58
Statistical analysis ................................................................ 59
Results ....................................................................................... 61
Aphid populations in 2005 and 2006 .......................................... 61
Predator and parasitoid diversity across the region .......................... 62
Impact of predators on soybean aphid ......................................... 62
Landscape diversity .............................................................. 63
Interpretation of principal components ....................................... 63
Model comparisons .............................................................. 64
Discussion ................................................................................. 64
Impact of biological control ................................................... 65
Ability of the landscape to supply biocontrol services ..................... 66
Response of natural enemies to landscape diversity and composition 67
Landscape spatial scale ......................................................... 68
Implications ...................................................................... 69

ix

CHAPTER 4

LANDSCAPE COMPOSITION INFLUENCES THE ACTIVITY DENSITY OF

CARABIDAE AND ARANAE COMMUNITES IN SOYBEAN FIELDS ............ 82
Abstract .................................................................................... 82
Introduction .............................................................................. 83
Methods .................................................................................... 86

Field sites ........................................................................ 86

Pitfall sampling ........................................................................... 86
Landscape analysis ........................................................................ 87
Statistical analysis ........................................................................ 88
Results .................................................................................... 90
Activity density in soybean ﬁelds ...................................................... 90
Landscape diversity ..................................................................... 90
Principal component analysis of landscape variables .............................. 91
Adult-overwintering species ............................................................ 92
Larval-overwintering species ............................................................ 93
Predatory and herbivorous species ...................................................... 93
Araneae and landscape variables ...................................................... 94
Discussion ................................................................................. 95
Relationships between Carabidae phenology, trophic level, and landscape
composition .............................................................................. 95
Araneae and landscape composition ................................................... 97
Spatial scale and the dispersal of habitat generalists .............................. 98
Implications .............................................................................. 98

CHAPTER 5

LANDSCAPE COMPOSITION INFLUENCES PATTERNS OF NATIVE AND

EXOTIC LADYBEETLE ABUNDANCE ................................................ 128
Abstract .................................................................................... 128
Introduction ............................................................................... 129
Methods .................................................................................... 133

Field sites ........................................................................ 133
Aphid and coccinellid sampling ............................................... 134
Landscape analysis .............................................................. 134
Statistical analysis ................................................................ 136
Results ..................................................................................... 139
Coccinellid diversity and abundance across the region .................... 139
Landscape diversity ............................................................ 139
Principal component analysis of landscape variables ...................... 140
Model ﬁtting of coccinellid abundance, prey, and landscape variables..140
Discussion ................................................................................. 142
Coccinellids and landscape composition .................................... 143

Coccinellidae and landscape scale ............................................ 145

Implications ..................................................................... 1 45
CHAPTER 6
NATIVE COCCINELLIDAE ARE SUBJECT TO SIGNIFICANT EGG
PREDATION IN SOYBEAN FIELDS ...................................................... 157
Abstract .................................................................................... 157
Introduction ............................................................................... 1 5 8
Methods .................................................................................... 161
Field sites ........................................................................ 161
Egg mass collection ............................................................ 161
Measuring predation of frozen egg masses ................................. 162
E g predation experiment ...................................................... 162
Nocturnal egg predation experiment .......................................... 163
Soybean aphid and predator survey .......................................... 163
Landscape analysis .............................................................. 164
Statistical analysis ............................................................... 165
Results .................................................................................... 166
Egg predation experiment ................................................... 166
Nocturnal egg predation experiment ....................................... 166
Aphid and predator populations ............................................. 167
Landscape diversity ............................................................ 167
AICc analysis of egg predation and landscape variables .................. 168
Discussion .............................................................................. 168
Egg predation of C. maculata ................................................ 169
Contribution of nocturnal egg predators .............................................. 170
Local landscape effects on C. maculata egg predation ..................... 171
A new hypothesis for native coccinellid decline ........................... 171
Summary ........................................................................ 1 72
CHAPTER 7
CONCLUSIONS AND FUTURE DIRECTIONS ......................................... 183
Biocontrol services and the landscape scale .......................................... 184
Landscape scale approach to native species conservation .......................... 186
APPENDICES .................................................................................... 189
Appendix A ............................................................................... 190
Appendix B ............................................................................... 198
REFERENCES CITED ........................................................................ 232

xi

LIST OF TABLES

Table 3.1. Study sites with dates of ﬁrst aphid detection, period of predator exclusion
cage study, dates when aphid threshold was reached in the open ﬁeld and exclusion
cages, and plant stages during cage study. Simpson’s D values are shown for the 1.5 km
landscape radius surrounding each soybean ﬁeld ............................................. 71

Table 3.2. Land cover categories used in the calculation of Simpson’s Diversity Index
(D) and as variables included in the principal components analysis ........................ 72

Table 3.3. Summary of model selection statistics for evaluating predation of A. glycines
by generalist natural enemies, calculated as a biocontrol services index (BSI) (see
methods section 2.6). The ﬁrst model listed at each spatial scale is the minimum AICc
model, bold indicates the best overall model. Only models with a A, of 2 or less are
included as competing models .................................................................. 73

Table 3.4. Summary of model selection statistics for evaluating the abundance of
Coccinellidae. The ﬁrst model listed at each spatial scale is the minimum AICc model,

bold indicates the best overall model. Competing models with a A, of 2 or less are

included ............................................................................................. 74
Table 4.1. Land cover categories used in the calculation of Simpson’s Diversity Index

(D) and as variables included in the principal components analysis ....................... 100
Table 4.2. Models compared by AICc analysis at spatial radii of 1-3.5 km ............... 101

Table 4.3. Abundance, trophic level and overwintering stage of Carabidae species found
in soybean ﬁelds in Iowa, Michigan, Minnesota, and Wisconsin .......................... 102

Table 4.4. Carabidae community composition in soybean ﬁelds in Iowa, Michigan,
Minnesota and Wisconsin in 2005-2006 ....................................................... 104

Table 4.5. Spider community composition in soybean ﬁelds in Iowa, Michigan,
Minnesota and Wisconsin in 2005-2006. All spiders were identiﬁed to family, Lycosidae
were identiﬁed to genus and where possible, species ......................................... 106

Table 4.6. Summary of model selection statistics for evaluating the activity density of
Carabidae. The ﬁrst model listed at each spatial scale is the minimum AICc model, bold

indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. 107

Table 4.7. Summary of model selection statistics for evaluating the activity density of
adult-overwintering Carabidae. The ﬁrst model listed at each spatial scale is the

minimum AICc model, bold indicates the best overall model. Only models with a A, of 2
or less are included as competing models ...................................................... 108

xii

Table 4.8. Summary of model selection statistics for evaluating the activity density of
larval-overwintering Carabidae. The ﬁrst model listed at each spatial scale is the

minimum AICc model, bold indicates the best overall model. Only models with a A, of 2
or less are included as competing models ...................................................... 110

Table 4.9. Summary of model selection statistics for evaluating the activity density of
predatory Carabidae. The ﬁrst model listed at each spatial scale is the minimum AICc

model, bold indicates the best overall model. Only models with a A, of 2 or less are
included as competing models .................................................................. 112

Table 4.10. Summary of model selection statistics for evaluating the activity density of
Araneae. The ﬁrst model listed at each spatial scale is the minimum AICc model, bold

indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. 1 13

Table 4.11. Summary of model selection statistics for evaluating the activity density of
Lycosidae. The ﬁrst model listed at each spatial scale is the minimum AICc model, bold

indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. l 15

Table 4.12. Summary of model selection statistics for evaluating the activity density of
Linyphiidae. The ﬁrst model listed at each spatial scale is the minimum AICc model,

bold indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. 1 17

Table 5.1. Land cover categories used in the calculation of Simpson’s Diversity Index
(D) and as variables included in the principal components analysis ....................... 147

Table 5.2. Models compared by AICc analysis at spatial scales of 1.0-3.5 km ......... 148

Table 5.3. Coccinellid community composition in soybean ﬁelds in Iowa, Michigan,
Minnesota and Iowa in 2005-2006 ............................................................. 149

Table 5.4. Summary of model selection statistics for evaluating the abundance of exotic
Coccinellidae. The ﬁrst model listed at each spatial scale is the minimum AICc model,

bold indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. 150

Table 5.5. Summary of model selection statistics for evaluating the abundance of native
Coccinellidae. The ﬁrst model listed at each spatial scale is the minimum AICc model,

bold indicates the best overall model. Only models with a A, of 2 or less are included as
competing models ................................................................................. 151

Table 6.1. Five potential models to explain the intensity of egg predation in Michigan
and Iowa soybean ﬁelds located in landscapes of varying diversity, abundance of forest

xiii

and grassland habitats, and the availability of extraguild prey. The models were
compared using AIC analysis adjusted for a small sample size (AICc). The a priori
hypothesis for testing each model is also provided .......................................... 175

Table 6.2. Percent of the total predator community and mean abundance :t SEM of
predators found in Iowa and Michigan soybean ﬁelds using two sampling measures:
sweep samples and yellow sticky card traps. Sweep samples consisted of a 20—sweep
sample of two rows of soybean plants. Yellow sticky cards were placed just above the
plant canopy and remained in the ﬁeld for 48 h (during the 48 h egg predation
experiment) ....................................................................................... 176

Table 6.3. Summary of model selection statistics for evaluating the intensity of egg
predation in soybean ﬁelds in Iowa and Michigan. The minimum AICc model and

competing models appear in bold ............................................................... 177
Table A1 . 1. Record of deposition of voucher specimens .................................... 190 '
Table B1.1. Land-cover maps .................................................................. 197

xiv

LIST OF FIGURES
Images presented in dissertation are presented in color.

Figure 2.1. Mean number of A. glycines per microcosm in the CONTROL (A. glycines
only), AA (A. glycines + A. aphidimyza), HA (A. glycines + H. axyridis) and BOTH (A.
glycines + A. aphidimyza +H. axyridis) treatments. Larval A. aphidimyza were released
at 0 h, H. axyridis were released at 24 h and removed at 27 h. Only the BOTH and HA
treatments (which included H. axyrz'dis) were counted at 27 h. The Figure inset shows

the change in aphid populations after 3 h of feeding by H. axyridis in the HA and BOTH
treatments. Mean comparisons are based on square-root transformed data (P < 0.05) ...45

Figure 2.2. Mean number of aphids per microcosm in the CONTROL (A. glycines only),
CC (A. glycines + C. carnea), HA (A. glycines +H. axyridis) and BOTH (A. glycines + C.
cornea +H. axyridis) treatments. Larval C. cornea were released at 0 h, H. axyridis were
released at 24 h and removed at 27 h. Only the BOTH and HA treatments (which
included H. axyridis) were counted at 27 h. The ﬁgure insert shows the change in aphid
populations after 3 h of feeding by H. axyridis in the HA and BOTH treatments. Mean
comparisons are based on square-root transformed data (P < 0.05) ........................ 46

Figure 2.3. Mean number of A. glycines per plant in predator exclusion,

medium-mesh (small predators only) and large-mesh (small and large predators)

cage treatments. Asterisk indicates a signiﬁcant difference between square-root
transformed mean A. glycines in the predator exclusion and large-mesh treatments (P <
0.05) .................................................................................................. 47

Figure 2.4. Mean number of lady beetle larvae per plant in the medium-mesh
and large-mesh cage treatments. Asterisk indicates a signiﬁcant difference
between treatments based on square-root transformed data (P < 0.05) ..................... 48

Figure 2.5. Mean number of A. aphidimyza per plant in medium-mesh and
large-mesh cage treatments ........................................................................ 49

Figure 2.6. Mean number of 0. insidiosus per plant in the predator exclusion,
medium-mesh and large-mesh treatments. Asterisk indicates a signiﬁcant

difference between the predator exclusion and large-mesh treatments

based on square-root transformed data (P < 0.05) ............................................. 50

Figure 3.1. Location of 26 soybean ﬁelds sampled in 2005 (ﬁelds 1-10) and 2006 (ﬁelds
11-26). The predator exclusion cage study was completed in all sites. Fields 2, 9, and 25
were excluded from the AICc analysis of BSI and Coccinellidae abundance due to

missing data ......................................................................................... 76

Figure 3.2. Mean number of natural enemies i SEM counted on sticky card traps (7 d
cumulative) and during visual plant observations (illustrates natural enemies present on a

XV

plant at a single point in time) in 2005 and 2006. Syrphidae were not counted on sticky
cards in 2005 ....................................................................................... 77

Figure 3.3. Mean number of A. glycines per plant (:l: SEM) in the cage and open plant

treatments. A signiﬁcant week*treatment interaction was found (F ”4:936, P=0.003).
Comparison of least squares means at 7 (P=0.002) and 14 d (P<0.0001) indicates that
aphid populations were signiﬁcantly higher in the absence of natural enemies . ....78

Figure 3.4. PCA ordination for principal components 1 and 2 of landscape elements
surrounding soybean ﬁelds sampled at a radius of 1.5 km .................................... 79

Figure 3.5. Relationship between the biocontrol services index (BSI) and PC] at a radius
of 1.5 km for 23 ﬁelds sampled for soybean aphid suppression, 2005-06 (y = 0.124x +
0.159). B81 was calculated as the relative difference in aphid number between open and
exclusion cage treatments (see methods section 2.6) .......................................... 80

Figure 3.6. Coccinellidae abundance in 23 soybean ﬁelds sampled for soybean aphid
suppression was best explained by a model containing both PC] and PC2 at a radius of

1.5 km. (y = 0.392x, — 0.361x2 + 2.0638). Partial residual plots illustrate a positive
correlation with PCI and a negative correlation with PC2, indicating that landscapes with
an abundance of forests or grasslands support larger lady beetle populations in soybean
ﬁelds compared with agriculturally-dominated landscapes ................................... 81

Figure 4.1. Map of soybean study sites sampled for Carabidae, Opiliones and Araneae in
2005 (1-13) and 2006 (14-26) using pitfall traps. Sites were selected so that the 3.5 km
landscape radius surrounding each ﬁeld varied in diversity and the abundance of semi-
natural and agricultural habitat .................................................................. 118

Figure 4.2. Activity density by date of (A) total Carabidae (B) C. sodalz's, (C) P.
chalcites, (D) P. permundus, (E) H. pensylvanicus, (F) H. erraticus, (G) P. lucublandus,
(H) S. quadriceps, and (I) P. melanarius collected per pitfall trap for the 26 soybean
ﬁelds sampled in 2005 and 2006. Species are shown in order of overall abundance .. 119

Figure 4.3. Activity density by date of (A) total Araneae (B) Opiliones, and the two most
abundant families of Araneae: (C) Linyphiidae and (D) Lycosidae collected per pitfall
trap for the 26 soybean ﬁelds sampled in 2005 and 2006 .................................... 122

Figure 4.4. Activity density by date of the two most abundant species of Lycosidae (A)
T. terricola and (B) P. saxatilis collected per pitfall trap for the 26 soybean ﬁelds
sampled in 2005 and 2006 ....................................................................... 124

Figure 4.5. PCA ordination for principal components 1 and 2 of landscape elements
surrounding soybean ﬁelds sampled. Data for a radius of 2 km is shown, plots are similar
for all spatial scales. Sites with positive loadings on PCI were correlated with the
variable Forest while negatives loadings on PC] were correlated with the variables Corn
and Soybean. Sites with positive loadings on PC2 were correlated with the variable

xvi

Grassland while negative loadings on PC2 were correlated with the variables Other Crops
and Wheat .......................................................................................... 125

Figure 4.6. Relationship between (A) activity density of Carabidae and PC2 at 1 km, (B)
adult-overwintering Carabidae and PC2 at 1 km, (C) P. chalcites and PC2 at 1.5 km, (D)
larval-overwintering Carabidae and PC2 at 2.5 km, (E) C. sodalis and PC2 at 2.5 km, and
(F) predatory Carabidae and PC2 at 1 km. Carabidae data were collected using pitfall
traps. Means are log (x+l) transformed ....................................................... 126

Figure 4.7. (A and B) Araneae activity density in 26 soybean ﬁelds was best explained
by a model containing both PCI and PC2 at a radius of 1.5 km. Partial residual plots
illustrate a positive correlation with both PCI and PC2. Plots (C) and (D) illustrate the
relationship between Lycosidae activity density and PC2 at 3 km, and the relationship
between Linyphiidae activity density and PCl at 2.5 km respectively .................... 127

Figure 5.1. Location of 33 soybean ﬁelds sampled for coccinellids in 2005 (ﬁelds 1-13)
and 2006 (ﬁelds 14-33) in Iowa, Michigan, Minnesota, and Iowa. Sites were selected so
that the 3.5 km landscape radius surrounding each ﬁeld varied in diversity and the
abundance of natural and agricultural habitat ................................................. 153

Figure 5.2. PCA ordination for principal components 1 and 2 of landscape elements
surrounding soybean ﬁelds sampled at a radius of 1.5 km. Sites with positive loadings on
PCI were correlated with the variable Forest while negatives loadings on PCI were
correlated with the variables Corn and Soybean. Sites with positive loadings on PC2
were correlated with the variable Grassland while negative loadings on PC2 were
correlated with the variables Other Crops and Wheat ....................................... 154

Figure 5.3. Relationship between the mean number of all exotic Coccinellidae (a), and
the two most abundant species: H. axyridis (b) and C. septempunctata (c), versus PCl at
2 km. Coccinellid data were collected using yellow sticky traps. Means are log (x+1)
transformed ........................................................................................ 155

Figure 5.4. Relationship between the mean number of all native Coccinellidae, and the
two most abundant species: H. convergens, and C. maculata versus PC2 (a-c) at 3.5 km
and Simpson’s D (d-i) at 1.5 km. Coccinellid data were collected using yellow sticky
traps. Means are log (x+l) transformed ....................................................... 156

Figure 6.1. Location of 14 soybean ﬁelds in Michigan and Iowa where C. maculata egg
predation was measured during July of 2007. The distance between each location 10 km
or greater. Sites were selected so that the 2 km landscapes surrounding each ﬁeld varied
in diversity and the abundance of natural and agricultural habitat within each state 1 78

Figure 6.2. Mean number of eggs remaining in the exclusion cage treatment (white bars)
and open treatment (black bars) in predation experiments. In the 48 h experiment,
treatments remained in soybean ﬁelds for 48 h, While nocturnal experiment treatments
remained in ﬁelds for 9 h ........................................................................ 179

xvii

Figure 6.3. Relationships between landscape diversity (D), the number of C. maculata
eggs consumed after 48 h of exposure to predators in soybean ﬁelds, and the average
number of egg predators collected per 20-sweep sample of soybean foliage. Landscape
diversity was estimated from 7 land-use categories using the Simpson’s Diversity Index

(D): D = 1/z(p,)2 where p,=proportion of habitat in the ith land-cover category (D
increases as diversity increases). The mean number of predators per 20-sweep sample
was log (x+l) transformed prior to analysis. Untransforrned means are shown here for
interpretation ....................................................................................... 1 80

Figure 6.4. The landscape-mediated IGP hypothesis. Gardiner et al.(in review) found that
native and exotic coccinellid populations in soybean ﬁelds were associated with
landscape elements. Soybean ﬁelds in forested landscapes were dominated by exotic
coccinellids (Michigan) while soybean ﬁelds in landscapes with an abundance of
grassland (Iowa) had greater native coccinellid abundance. We propose that native
coccinellid populations are under greater threat from exotics in soybean ﬁelds in forest
dominated landscapes. Here, exotics are more likely occupy a larger proportion of the
guild of potential intraguild predators and thus consume a greater proportion of native
coccinellid eggs. In grassland dominated landscapes native eggs are more likely to be
utilized by native predators, sustaining their populations. In the above food web circle
size indicates the relative abundance of species within the food web. Solid lines indicate
predator-prey interactions while dotted lines indicate potential IGP. Width of dotted lines

indicates the proposed strength of intraguild interactions ................................... 181
Figure 81.1. .................................................................................... 199
Figure Bl .2. .................................................................................... 200
Figure Bl .3. .................................................................................... 201
Figure Bl .4. .................................................................................... 202
Figure 31.5. .................................................................................... 203
Figure B1.6. .................................................................................... 204
Figure Bl .7. .................................................................................... 205
Figure Bl .8. .................................................................................... 206
Figure Bl .9. .................................................................................... 207
Figure B] .10. ................................................................................. 208
Figure 81.1]. ................................................................................. 209

xviii

Figure Bl.12. ................................................................................. 210

Figure B1.13. ................................................................................. 211
Figure 81.14. ................................................................................. 212
FigureBl.15. ................................................................................. 213
Figure Bl.16. ................................................................................. 214
Figure Bl.17. ................................................................................. 215
Figure Bl.18. ................................................................................. 216
Figure Bl.19. ................................................................................. 217
Figure B1.20. ................................................................................. 218
FigureBl.21. ................................................................................. 219
Figure 8122. ................................................................................. 220
Figure Bl.23. ................................................................................. 221
Figure BI .24. ................................................................................. 222
Figure Bl.25. ................................................................................. 223
Figure B1.26. ................................................................................. 224
Figure B1.27. ................................................................................. 225
Figure 81.28. ................................................................................. 226
Figure 8129. ................................................................................. 227
Figure 8130. .., .............................................................................. 228
Figure 31.3]. ._ ................................................................................ 229
Figure 8132. ................................................................................. 230
Figure 8133. ................................................................................. 231

xix

AICc

BSI

cm

PC]

PC2

SEM

KEY TO SYMBOLS OR ABBREVATION S

Akaike’s Information Criterion, adjusted for small sample size
Biocontrol Services Index

Centimeters

Simpson’s Diversity

Fisher distribution

Hours

Intraguild predation

Number of parameters in model

Meters

Millimeters

Number of observations

Probability under the assumption that the null hypothesis is true
Principal component 1

Principal component 2

Coefﬁcient of determination

Standard error of the mean
Species

Akaike weight

AICc differences

XX

CHAPTER 1

INFLUENCE OF LANDSCAPE AND WITHIN-FIELD INTERACTIONS ON

ARTHROPOD FOOD WEBS IN AGROECOSYSTEMS.

Introduction

Food webs are commonly used to depict feeding relationships within ecological
communities. However, in agroecosystems, knowledge of food web interactions is often
limited, due to the complexity of tri-trophic relationships and the number of species
involved (Bascompte and Melian 2005). In addition, organisms that comprise
agricultural food webs live in a spatially heterogeneous world, where the within-ﬁeld
community is variously connected to the surrounding landscape (Thompson and
Townsend 2003). The arthropod components of food webs are likely to cross spatial
boundaries to arrive in crop ﬁelds ﬁ'om other areas that provide alternative prey or
overwintering habitat. Therefore, landscape variables such as habitat complexity, quality,
and patchiness as well as an organism’s perceived spatial scale can all impact the
structure and ﬁrnction of agricultural food webs.

Invasions of exotic species are likely to profoundly affect the structure of food
webs in agroecosystems. In many cases, biological invasions extend beyond the
introduction of single species, sometimes involving the transplantation of an entire exotic
landscape (Thompson and Townsend 2003). The latter appears to be true in the case of
the soybean aphid, its host plants, and natural enemies. The soybean aphid Aphis
glycines Matsumura, a native of Asia, was ﬁrst detected in the US in Wisconsin in July of

2000. The distribution of soybean aphid now extends throughout the north-central

soybean growing states and into Canada. In the soybean system, species occupying each
trophic level of the food web are introduced into the north-central U.S. Both the primary
and secondary host plants, common buckthom, Rhamnus cathartica L. (Rharnnaceae)
and the cultivated soybean, Glycine max L. (Fabaceae), the soybean aphid, and a
complex lady beetle predators, Harmomia axyridis Pallas, Coccinella septumpuncata L.,
Hippodamia variegata (Goeze), and Propylea quatuordecimpunctata (L.) (Coccinellidae)
are non-native species. These non-native organisms also interact with a host of native
species including Chrysoperla cornea (Stephens) (Chrysopidae), Aphidoletes aphidomyza
(Rondani) (Cecidomyiidae), Hippodamia convergens Guérin-Méneville (Coccinellidae),
Coleomegilla maculata DeGeer (Coccinellidae), and Orius insidiosus Say (Hemiptera).
Studying how landscape variables across the north-central U.S. affect the
functioning of this food web provids a model system to investigate landscape mediated
predator-predator and predator-prey dynamics. Soybean ﬁelds are ephemeral habitats
where disturbances such as tillage, pesticide application, and harvesting require repeated
re-colonization by soybean aphid and its natural enemies. Landscape elements act as
both a source and a sink for these species, shaping within-ﬁeld interactions of soybean
aphid and its natural enemies. In this dissertation, I began by investigating the role H.
axyridis plays in the biological control of soybean aphid (Chapter 2). I was interested in
determining whether early season intraguild predation of the native larval predator
community by this exotic coccinellid resulted in a release of A. glycines. I followed this
study of within-ﬁeld dynamics by determining how the landscape surrounding soybean
ﬁelds inﬂuenced biocontrol services at a regional scale, studying the suppression of A.

glycines supplied by predators in ﬁelds across Iowa, Michigan, Minnesota, and

Wisconsin (Chapter 3). Next, I examined the inﬂuence of landscape composition on two
natural enemy guilds: ground-dwelling species and native and exotic coccinellids
(Chapters 4 and 5). Chapter 4 examined the inﬂuence of landscape composition on the
activity density of Araneae and Carabidae in soybean ﬁelds. I was speciﬁcally interested
in how dispersal capacity, trophic status, and overwintering stage of these species
affected their response to landscape variables. In Chapter 5, I examined the relationship
between the abundance of native and exotic Coccinellidae in soybean ﬁelds and the
surrounding landscape. Speciﬁcally, I was interested in determining if landscape
variables such as the amount of semi-natural habitat present facilitated the success of
exotic populations. Finally, I connect landscape to within-ﬁeld interactions to examine
how the landscape around soybean ﬁelds inﬂuences intraguild predation of native
coccinellids by exotics (Chapter 6). Herein, I review soybean aphid biology, known
natural enemies, potential within-ﬁeld interactions, and landscape dynamics.
Biology of the soybean aphid

Aphis glycines exhibits a heteroecious holocyclic lifecycle (Ragsdale et a1.
2004). From late spring to early fall, soybean aphid feeds and reproduces viviparously on
soybean. Many generations of alate and apterous aphids can occur during the growing
season. In the fall, reduced temperature and photoperiod inﬂuence the production of
winged males and female gynoparae that emigrate in search of Rhamnus hosts. Several
species of Rhamnus are present within the invasive range of this pest. Yoo et a1. (2005)
compared six species of Rhamnus and found that gynoparae prefer R. cathartica and R.
alnifolia L’Her, over R. caroliniana Walter, R. frangula L., R. ﬁangula ‘Asplenifolia’, or

R. frangula ‘Columnaris’. Gynoparae exhibited greater longevity and reproduction on R.

cathartiea and R. alnifolia, and oviparous nymphs produced by gynoparae survived to
maturity only on these hosts. These oviparae will lay the overwintering eggs. In the
spring soybean aphids emerge from eggs and produce two generations on Rhamnus
consisting primarily of Wingless females. A third generation of alate aphids leaves the
overwintering host searching for soybean (Ragsdale et al. 2004).

Temperature is a key regulator of soybean aphid population dynamics.
McComack et al. (2004) reported that 25°C was the optimal developmental threshold for
A. glycines. At higher temperatures, net fecundity, gross fecundity, generation time, and
life expectancy are decreased. Reproductive periods vary with temperature; aphids
reproduce longer and have more offspring at 20° and 25°C than at 30°C and do not
survive above 349°C (McComack et al. 2004).

The paths of soybean aphid colonization of soybean produces a highly patchy
distribution. Newly arrived females give birth to produce small colonies on scattered
plants within a ﬁeld. When soybean is in the vegetative growth stage, A. glycines can
typically be found on partially expanded trifoliates, petioles and stems (Ragsdale et al.
2004). When the plant begins to ﬂower, set pods and develop lateral branches, aphids
become widely dispersed on the plant (Ragsdale et al. 2004). When soybean enters
reproductive growth, its nutritional quality for the aphid changes, aphids on lower,
mature leaves can appear pale and smaller than aphids found feeding on the growing
points in pre-reproductive plants (Ragsdale et al. 2004).

Impacts of soybean aphid
Feeding by A. glycines on the leaves, stems, and pods of the soybean plant

causes reductions in photosynthetic rate, plant growth, and seed yield of soybean plants

(DiFonzo and Hines 2002). This pest can also vector several viruses, including alfalfa
mosaic, soybean dwarf, soybean stunt, soybean mosaic, tobacco ringspot, and bean
yellow mosaic (Clark and Perry 2002, Wang and Ghabrial 2002). Honeydew production
by the soybean aphid promotes the growth of sooty molds further reducing the
photosynthetic capacity of the soybean plant (Hirano et al. 1996). To prevent economic
losses from soybean aphid damage, U.S. growers may make multiple insecticide
applications per year. In 2003, over 42 million acres of soybean in the north-central US.
were infested with soybean aphid, 7 million of which were treated with insecticides
(Ragsdale et al. 2007). Over 1.6 million acres (80% of total acreage) of soybean were
treated for soybean aphid in Michigan in 2005 (Chris DiFonzo, personal communication).
Broad-spectrum organophosphates and pyrethroids are typically used (DiFonzo and
Hines 2002).
Natural enemies of A. glycines in soybean

In its native range the soybean aphid is attacked by more than 30 species of
predators, 8 species of aphidiine and aphelinid parasitoids, and several fungal pathogens
(V andenBerg et al. 1997). In the US, the natural enemy complex is still developing;
however, it is similar in structure with a large number of predatory species, several
parasitoids and a few fungal pathogens.
Predators

In US. soybean ﬁelds, predators supply the majority of the biocontrol (Fox et al.
2004, 2005, Costamagna and Landis 2007a, Costamagna and Landis 2007b, Costamagna
et al. 2007a). Over 22 predatory taxa are known to attack soybean aphid in the north-

central U.S. (Rutledge et al. 2004). This community is comprised of a large ground-

dwelling community dominated by Carabidae species and a diverse community of
foliage—foragers including species in the families: Cantharidae, Coccinellidae,
Charnymaemyiidae, Syrphidae, Anthocoridae, Nabidae, Chrysopidae, and Hemerobiidae
(Rutledge et al. 2004). Of these, the most numerous predators collected by Rutledge et
al. (2004) were the minute pirate bug, 0. insidiosus and the multicolored Asian lady
beetle, H. axyridis. In Indiana, these species accounted for over 85% of the predators
found in the ﬁeld (Rutledge et al. 2004). The most common natural enemies found in
Michigan soybean ﬁelds are: 0. insidiosus, H. axyridis, C. septempunctata, and H.
convergens (Fox et al. 2004, 2005, Costamagna and Landis 2007b, Costamagna et al.
2007a). In addition to these species, Kaiser et al. (2007) have found the cecidomyiid, A.
aphidomyza and the syrphid species Allograpta oblique (Say), Eupeodes americanus
(W iedemann), Syrphus rectus (Osten Sacken), Paragus haemorrhous Meigen, Eupeodes
volucris Osten Sacken, T oxomerus marginatua Say, and Sphaerophorus contigua
attacking soybean aphid.
Parasitoids

In 2000-2002, parasitoids of A. glycines were nearly undetectable in north-central
U.S. soybean production systems (Rutledge et al. 2004); however, in 2003, several
parasitoid species were detected utilizing this new host. Kaiser et a1. (2007) collected
four parasitoid species from sentinel A. glycines on potted soybean plants placed in
soybean ﬁelds at the Kellogg Biological Station, Hickory Corners, MI. These were (in
order of abundance): Lysiphlebus testaceipes (Cresson), Aphidius colemani Viereck,
Aphelinus asychis (Walker), and A. albipodus Hayat & Fatima. Despite the detection of

parasitoids, they continue to play a small role in the conservation biological control of A.

glycines, accounting for less than 4% of the natural enemies found in 2003-2005

(Costamagna and Landis 2007a).

In 2007, releases of a non-native parasitoid Binodoxes communis initiated a
classical biological control program for soybean aphid. Pre—release testing indicated a
very low risk of non-target attack by this parasitoid (Wyckhuys and Heirnpel 2007,
Wyckhuys et al. 2007, Wyckhuys et al. 2008). Releases were made in Minnesota,
Michigan and Iowa. Currently, it is unknown if released wasps have established.
Pathogens

Entomopathic fungi may also play a role in suppressing aphid populations.
Kaiser et al. (2007) commonly observed fungal pathogens attacking pea aphids in alfalfa
and clover ﬁelds but the disease was rarely observed in soybean aphid found in adjacent
ﬁelds. Pandora neoaphidis and Conidiobolus thromboides have also been isolated from
A. glycines.
Impact of natural enemies on A. glycines population dynamics

Understanding how the community of natural enemies present in the soybean
agroecosystem affects A. glycines population dynamics is necessary for successful
biological control. Predators can signiﬁcantly reduce A. glycines adult establishment on
soybean, and are more likely to cause reductions the later soybean aphid arrives in
soybean ﬁelds (Costamagna and Landis 2007a, Costamagna et al. 2007a). The number of
predators required to suppress A. glycines is low; in one experiment by Fox et al. (2004)
total predator populations never exceeded 5 per In2 yet prey populations were suppressed
for half of the trial. This illustrates the potential importance of generalist predators in

suppressing this pest when present at low densities (Fox et al. 2004). Costamagna and

Landis (2007a) compared the impact of top-down regulation of A. glycines by natural
enemies with bottom-up impacts of agricultural management practices. Although no
bottom-up impacts were detected, a strong top-down effect was found, indicating that
generalist predators are important in the regulation of A. glycines (Costamagna and
Landis 2007a).

Physiological and behavioral differences between natural enemies will affect their
role in the biological control of A. glycines. Rutledge et al. (2004) contrasted two
predators commonly found in soybean ﬁelds, 0. insidiosus and H. myridis that differ in
their interactions with A. glycines. Orius insidiosus has a number of traits that suggest it
primarily acts to suppress growing aphid populations (Rutledge et al. 2004). This
generalist predator is present in soybean throughout the season where it increases in
density until soybean plants senesce. Orius insidiosus can survive when aphid
populations are low by feeding on soybean pollen and moisture and minerals from
soybean phloem (lsenhour and Marston 1981, Armer et al. 1998). In contrast, H. axyridis
can act to either prevent or to reduce high aphid densities as one of the ﬁrst coccinellids
to arrive in soybean ﬁelds in the spring (Ragsdale et al. 2004). Harmonia axyridis is
capable of a distinct numerical response in the presence of abundant prey. Populations of
this predator typically peak at or just after the population of aphid prey and exhibit a
strong numerical response to soybean aphid, becoming more common in ﬁelds once

aphids reach 100 or more per plant (Ragsdale et al. 2004).

Within-ﬁeld interactions and the soybean aphid food web
Intraguild predation (IGP)

IGP is deﬁned as predatory interactions between natural enemies occupying the
same trophic level and competing for similar hosts/prey, which can shape the distribution,
abundance, and evolution of intraguild competitors. IGP can lead to spatial and temporal
exclusion of intraguild competitors, competitive coexistence or alternative stable states
(Polis et al. 1997, Polis 1998). Biological theory has evolved considerably from simple
models that state if beneﬁcial insect species are sufﬁciently abundant herbivore
populations will be suppressed, allowing plants to grow until limited by competition
(Hairston et al. 1960). The assumption that the predator and parasitoid community
regulating herbivores can be combined into a single trophic level, having a parallel effect
on pest control is lacking in that it neglects complex, potential negative interactions,
including IGP (Rosenheim et al. 1995, Rosenheim 1998, Colfer and Rosenheim 2001 ,
Meyhofer 2001, Kindlmann and Houdkova 2006). In aphidophagous systems where
several natural enemies compete for a single dominate extraguild prey, interactions such
as IGP are likely to occur (Lucas 2005). Several studies have measured IGP of aphid
parasitoids by aphid predators (Wheeler et al. 1968, Ferguson and Stiling 1996, Colfer
and Rosenheim 2001, Meyhofer 2001, Snyder and Wise 2001, Snyder and Ives 2003,
Snyder et al. 2004a, Kindlmann and Houdkova 2006). When the impacts of IGP on
aphid biological control are evaluated across several systems, three potential outcomes
are found: antagonistic (competition among natural enemies), additive (no competition,

no facilitation), or synergistic interactions (facilitation) (Lucas 2005). These studies

illustrate that IGP can have diverse effects on parasitoid populations and aphid
suppression.
Predator-parasitoid interactions

IGP of parasitoids by predators can disrupt biological control if an intraguild
predator preferentially selects parasitized prey over unparasitized prey, or if the intraguild
predator is inefﬁcient at exploiting herbivores (Colfer and Rosenheim 2001). These
conditions could cause the parasite population to be suppressed while herbivore
populations expand (Ferguson and Stiling 1996, Colfer and Rosenheim 2001, Snyder and
Ives 2003). Alternatively, if an intraguild predator is efﬁcient at exploiting herbivore
prey and selects unparasitized aphids over parasitized aphids or aphid mummies,
herbivore regulation could be either unchanged or enhanced by IGP (Ferguson and
Stiling 1996, Colfer and Rosenheim 2001, Snyder et al. 2004b, Kindlmann and
Houdkova 2006). These differences illustrate the need for system-speciﬁc research to
determine how IGP will affect biological control.

Predator-predator interactions

Generalist predators can be effective biological control agents of native and exotic
pests (Symondson et a1. 2002); however, some interactions between predators can
interfere with pest suppression (Snyder and Wise 1999). Snyder and Wise (1999) deﬁne
interference among predators to include IGP, cannibalism, predator avoidance behavior,
and predator-predator competition. Predators may function from the third trophic level,
suppressing herbivores or from the fourth trophic level suppressing populations of
intermediate predators through IGP and thereby potentially releasing herbivore

populations from their top-down control (Rosenheim and Corbett 2003). Several factors

10

may inﬂuence whether a predator is functioning at the third or fourth trophic level. These
include encounter probabilities, attack probabilities, capture success, and consumption
probabilities may affect prey choice of a predator, habitat structure and physical refuges,
active prey defenses and predator preferences (Shi 1993).

Intraguild interactions between predators have been shown to cause diverse
effects on prey population dynamics and plant productivity (Polis et al. 1989, Polis 1991,
Rosenheim et al. 1995, Rosenheim 1998, Prasad and Snyder 2004, Snyder et al. 2004a).
Venzon et a1. (2002) studied the effects of Orius laevigatus (Fieber) on the population
dynamics of the predatory mite Phytoseiulus persimilis Athias-Henriot and the spider
mite T etranychus urticae Koch on cucumbers. They found that although this predator
killed large numbers of predatory mites and spider mites, the presence of 0. laevigatus
had little to no effect on mite abundance or cucumber plant ﬁ'uit weight (Venzon et al.
2002). Snyder and Wise (2001) examined whether IGP weakens the ability of carabids
and lycosids to control herbivores in gardens of cucumber and squash. In spring
cucumber gardens, the presence of lycosids increased fruit production and decreased
populations of cucumber beetle. Carabids weakly decreased cucumber beetle. In late
season squash gardens, carabids increased fruit production by decreasing densities of
squash bug. Lycosid spiders decreased squash yield by IGP of important hemiptem
predators of the squash bug (Synder and Wise 2001). It appears that late season effects
of carabids and spiders counteracted each other so that the complete guild did not
inﬂuence squash ﬁ'uit production. Snyder et al. (2004a) examined IGP between larvae of
two exotic ladybeetles, C. septempunctata and H. axyridis, and two natives, Coccinella

transversoguttata Brown and H. convergens. When the native ladybeetles were paired

ll

with either exotic species in microcosms they were more often the intraguild prey than
predator. When the two exotics were placed together neither had a clear IGP advantage.
Adding alternative prey did not alter the frequency and patterns of IGP among these
coccinelids (Snyder et al. 2004a). This may be contributing to the speed with which
these exotic lady beetles are displacing natives following invasion (Snyder et al. 2004a).
Prasad and Snyder (2004) studied the effects of predator interference on the biological
control of Delia sp. eggs by a community of ground-active beetles. They found that
smaller staphylinid and carabids beetles commonly preyed upon ﬂy eggs while larger
beetles rarely ate eggs but did feed on the smaller beetles. Despite this IGP, the
community of carabids and staphylinids decreased eggs survival (Prasad and Snyder
2004). A 4-fold increase in beetle community density did not lead to increased activity-
density or egg predation (Prasad and Snyder 2004). Rosenheim et al. (1993) found that
generalist hemiptem predators decreased the survival of lacewing larvae. Experimental
addition of the reduviid Zelus renardii (Kolenati) or Nabis sp. (Nabis altematus Parshley
and N. americorfurus Carayon) generated sufﬁcient lacewing larval mortality to release
aphid (Aphis gossypii) populations from their control (Rosenheim et al. 1993).
A conceptual framework of IGP and foraging behavior

Rosenheim and Corbett (2003) identify encounter frequency as a key factor
contributing to the diet of extreme generalists that attempt to capture and consume any
available prey they encounter. Encounter frequencies are highly inﬂuenced by foraging
behavior. When two predatory strategies are examined, sit-and-wait predation versus
active foraging, they suggest that foraging strategy will inﬂuence the type of prey

selected by a predator. Sedentary prey are consumed by widely foraging “intermediate”

12

predators which may in turn be captured by sit-and-wait “top” predators. Mobile prey
may be consumed by either foraging or sit and wait predators.

Rosenheim and Corbett (2003) tested these ideas by developing a mathematical
model based on a papaya community consisting of a sedentary herbivore, T etranychus
cinnabarinus (Boisduval), which feeds on papaya foliage, two intermediate predators,
Stethorus siphonulus (Kapur) and Phytoseiulus macropilis (Banks), both widely foraging
specialists of spider mites, and the sit and wait omnivorous predator Nesticodes rufipes
(Lucas), a tangle-web spider. They found that foraging predators have the potential to be
effective regulators of herbivore populations, even when they also act as omnivores
consuming other predators. Sit-and-wait predators are unlikely to suppress populations
of sedentary herbivores by foraging heavily on intermediate predators. Sit-and-wait
predators are predicted to function as the top predators, suppressing populations of
intermediate predators and releasing sedentary herbivore populations from their control
(Rosenheim and Corbett 2003).

These above model predictions were conﬁrmed through ﬁeld experiments
(Rosenheim et al. 2004). The widely foraging P. macropilis was capable of generating
strong suppression of T. cinnabarinus populations. Second, the sit-and-wait predator N.
ruﬁpes never produced detectable levels of spider mite suppression. Nesticodes ruﬁpes
consumed both of the foraging predators. Spider mite suppression by S. siphonulus was
vulnerable to disruptive effects of N. ruﬁpes whereas suppression by P. marcropilis was
relatively insensitive to N. ruﬁpes. This difference was attributed to the difference in

body size between these two intermediate predators. Phytoseiulus macropilis is smaller

13

and thus consumes less prey and forages less leading to less ﬁequent encounters with N.
ruﬁpes webs compared with S. siphonulus (Rosenheim et al. 2004).
IGP in the A. glycines system

The model proposed by Rosenheim and Corbett (2003) can be modiﬁed to
incorporate intraguild interactions within the community of generalist predators in
soybean. As with the model system, soybean is fed upon by a sedentary herbivore, the
soybean aphid which in turn is prey for several intermediate predators. In soybean,
however, two groups are discernable among these. Resident intermediate predators are
arthropods which are predatory as larvae, spending their juvenile stage feeding primarily
on the soybean aphid. Important predators in this group include the larvae of ladybeetles,
lacewings, syrphids, aphid predatory midge, and Orius nymphs. A second group of
intermediate predators, the transients, search for prey in several crops within a growing
season, foraging for prey in soybean for shorter durations. These species move into
soybean when soybean aphid is present, feed for some time, then move on, leaving some
resident population of aphids behind. Important species in this group include adult C.
septempunctata and H. axyridis.

A critical time period in the biological control of soybean aphid occurs in the
early season when aphid populations begin to colonize soybean plants. When aphid
populations are low, adult H. axyridis and C. septempunctata move through soybean
ﬁelds feeding within patches of soybean aphid, but are not stimulated to lay eggs. They
will spend short feeding bouts within an aphid patch, removing some of the population
before moving on to other soybean ﬁelds or to search for prey in other crops. During this

period, when ladybeetles are highly mobile IGP events occurring between these transient

14

predators and resident predacious larvae could inﬂuence the success of soybean aphid
biological control.
Impacts of landscape variables on insects in agroecosystems

Agricultural landscapes have changed dramatically with agricultural
intensiﬁcation, resulting in larger more contiguous ﬁelds and non-crop areas reduced in
area and frequency (Bianchi et al. 2006). The diversity and abundance of natural enemies
and the biocontrol services they provide are inﬂuenced by the structure and composition
of the surrounding landscape (Marino and Landis 1996, Colunga-Garcia et a1. 1997,
Bommarco 1998, Elliott et al. 1999, Thies et al. 2003, Schmidt and Tscharntke 2005,
Tscharntke et al. 2005b). Landscape variables such as habitat complexity, habitat quality,
and patchiness as well as an organism’s dispersal capability all inﬂunce the ability of a
landscape to support biological control in agricultural croplands (Elliott et al. 1999, Thies
et al. 2003, Schmidt and Tscharntke 2005). Moreover, agricultural crop ﬁelds are
ephemeral habitats in which anthropogenic disturbances such as tillage, pesticide
application, and harvesting require arthropods to frequently re-colonize crops (W issinger
1997). The surrounding landscape provides the local species pool of arthropods for this
re-colonization and thus may inﬂuence the level of biological control in frequently
disturbed crop ﬁelds (Lee et al. 2001).

In a recent review, Bianchi et al. (2006) found that increased landscape
complexity reduced pest pressure in 45% of the studies. Effects included reductions in
pest density (Ostrnan et al. 2001, den Belder et al. 2002) and reduction in crop injury
(Thies and Tscharntke 1999, Thies et al. 2003). Pest pressure increased with landscape

complexity in 15% of the cases examined (Raffy and Tran 2005, Roschewitz et al. 2005),

15

but the two were unrelated in 40% of the studies (Galecka 1966, Holland and Fahrig
2000, Klug 2003, Thies et al. 2005). It is likely that some of this variation is tied to
differences in the impact of landscape complexity on speciﬁc natural enemy populations.
A more detailed understanding of how natural enemies respond to landscape structure
may help us understand these outcomes. The majority of the work thus far has focused
on response to overall landscape complexity (percent non-crop habitat), habitat
thresholds, edge effects, and habitat dispersal via landscape corridors.
Habitat complexity

The diversity of crop and non-crop habitats within a landscape can greatly affect
communities of herbivores and their natural enemies within an agricultural crop (F ahrig
1992, Marino and Landis 1996, Colunga-Garcia et al. 1997, Bommarco 1998, Elliott et
a1. 1999, Menalled et al. 1999, Thies and Tscharntke 1999, Fahrig 2001, den Belder et al.
2002, Ostrnan 2002, Tscharntke et al. 2002, F ahrig 2003, Menalled et al. 2003, Thies et
al. 2003, Weibull et al. 2003, Schmidt et al. 2004, Tscharntke and Brandl 2004, Schmidt
et al. 2005a, Schmidt et al. 2005b, Tscharntke et al. 2005b, Tscharntke et al. 2005a). For
example, Colunga-Garcia et al. (1997) studied changes in species assemblages of
predatory coccinellids in response to landscape structure with different mosaics of
cultivated (alfalfa, corn, wheat) and uncultivated (deciduous forest, ﬁeld succession)
habitats. Species composition differed between sites with and without deciduous forest
habitat (Colunga-Garcia et al. 1997). Although the deciduous forest habitat had the
highest ladybeetle species richness (Cycloneda munda (Say), Chilocorus stigma (Say),
and Brachiacantha ursine (F.) most abundant in deciduous habitat), there were

exceptions. Coccinella septempunctata was the dominant species in the landscape and

16

was equally abundant all sites, whereas Coleomegilla maculate lengi Timberlake was
more abundant in the site with corn habitat (Colunga-Garcia et al. 1997). F ahrig and
J onsen (1998) found that in a study of 31 alfalfa ﬁelds in an agricultural landscape, more
isolated ﬁelds had overall higher insect species richness. Similarly, Elliott et al. (1998)
used aerial photographs to demonstrate that increases in uncultivated land and in
patchiness surrounding wheat ﬁelds were both associated with increasing abundance and
richness in predator communities in wheat ﬁelds. Bommarco (1998) found that the
fecundity of the generalist predatory carabid, Pterostichus cuperus, was greater in areas
with small ﬁelds, higher perimeter-to-area ratios, and a higher coverage of perennial
crops compared to less spatially complex areas.
Minimal suitable habitat thresholds

Several studies suggest thresholds in landscape structure may exist below which
the search efﬁciency and the ability of natural enemies to aggregate and control pests is
diminished (With and King 1999). Simulation experiments have shown that search
success of natural enemies declined when suitable habitat fell below 20% (With and King
1999). Tenebrionid beetles (Eleodes obsolete) and small crickets (Acheta domestica L.)
also exhibited threshold responses in movement when habitat fell below 20% (Wiens et
al. 1997, With et al. 1999). Parasitism rates also decline in agricultural landscapes when
the non-crop area falls below 20% (Thies and Tscharntke 1999).
Edge eﬂects

Habitat edges that border agricultural ﬁelds could inﬂuence natural enemy
diversity and abundance by providing alternative food resources, shelter, and/or

protection. Natural enemy populations can build up in ﬁeld borders and move into the

17

crop when pest populations begin to build. However, habitat edges may also impede the
movement of arthropods into crops by acting as a physical barrier by providing a more
stable and suitable habitat than the crop (Fortin and Mauffette 2001). A substantial
amount of research has been done to understand the role of ﬁeld edges on the biological
control of pests by natural enemies (Alomar et al. 2002). Overall, these data show that
measuring the activity and richness of predatory populations in ﬁeld boundaries can, but
doesn’t always accurately predict their potential impact on herbivores in neighboring
crops.
Corridors and habitat connectivity

A popular strategy for maintaining populations of plants and animals in
fragmented landscapes is to connect isolated patches with thin strips of habitat called
corridors. Corridors are thought to increase gene ﬂow, promote reestablishment of
locally extinct populations, and increase species diversity within otherwise isolated areas
(Tewksbury et al. 2002). Collinge and Palmer (2002) tested the theory that isolated
habitat patches will experience greater rates of species loss compared with less isolated
habitats. Corridors did not signiﬁcantly inﬂuence overall rates of patch colonization, but
slightly increased the probability of colonization by less vagile species (Collinge 2000).
Collinge (2000) also characterized individual movement pathways of three insect species
in plots with and without corridors. One of the three species signiﬁcantly preferred
corridors, whereas the other two species moved independently of corridors. These results
suggest that conidors have the potential to promote movement of organisms among
habitat patches, but that their function may depend upon species characteristics,

landscape context, patch size, and environmental variation.

18

Functional spatial scale and trophic level

Landscape complexity may not inﬂuence all species equally, or at the same scale
(den Belder et al. 2002, Steffan-Dewenter et a1. 2002, Thies et al. 2003, Schmidt et al.
2005a). Thies et al. (2003) and others have hypothesized that interacting communities
are made up of species with different spatial strategies. The spatial scale experienced by
an organism may depend on its trophic level (Holt et al. 1999). Holt et al. (1999) suggest
that plant species occupy distinct microhabitats whereas herbivores have larger spatial
domains, but are conﬁned to local communities. Predators and parasitoids can utilize
ranges that encompass the habitats of several prey species (Holt et al. 1999). There is no
absolute size for a landscape from an organism-centered perspective; the size of a
landscape varies depending on what constitutes a mosaic of habitat or resource patches
meaningful to that particular organism. Thies et al. (2003) measured the functional
spatial scale at which the rape pollen beetle, Meligethes aeneus F. and its parasitoids are
affected by their landscape. They tested the effects of landscape on trophic-level
interactions using simple (<3% non-crop habitat) to complex (>50% non-crop habitat)
landscapes, analyzed them at 8 spatial scales (concentric circles 0.5-6 km in diameter),
and found that herbivory and parasitism were correlated with percent non-crop area using
a spatial scale of 1.5 km. At all spatial scales, the predictive power of non-crop area was
higher for herbivory than for parasitism. Schmidt et al. (2005) used pitfall traps to collect
spider species in wheat ﬁelds located in landscapes that varied in their proportion of non-
crop habitat. Although the proportion of non-crop habitat did not affect the abundance of
spiders, a relationship was detected between non-crop habitat and species diversity.

Spider species richness in wheat ﬁelds rose with the proportion of non-crop habitat in the

19

surrounding landscape (Schmidt et al. 2005). The most common family of spiders
collected was the Lycosidae. Ballooning species (majority of Linyphiidae, Araneidae,
and Theridiidae), with the exception of Erigone atra and Erigone dentipalpis were poorly
represented, suggesting that ballooning species may respond to a larger spatial scale of
landscape complexity compared with walking species (Schmidt et al. 2005).

Inﬂuence of landscape variables depends on farming practice

A potential complication of studying the effects of landscape complexity in
agricultural systems is the effects production practices can have on arthropod
communities. Weibull et al. (2003) compared the effect of farm management
(conventional farming versus organic) and landscape features on species richness in
cereal ﬁelds, leys (arable ﬁeld temporary with perennial grass and clover mowed 2-4
times a year), and semi-natural pastures. The effects of landscape were more
pronounced on conventional farms, species richness of plants, butterﬂies, and carabids
increased with small-scale landscape heterogeneity or the amount of ﬁeld margin
(perimeter to area ratio). There were no signiﬁcant correlations between species richness
and landscape variables among organic farms. The impact of predators (carabids, rove
beetles, and spiders) on the bird-cherry oat aphid Rhopalosiphum padi (L.) depended on a
high perimeter-to-area ratio and high percentage of perennial crops in effective early
season control of the aphid (W eibull et a1. 2003).
Ostrnan et. al (2001) also measured the impact of naturalenemies on the

population development on the bird cherry-oat aphid on conventional and organic farms.
They found that natural enemies had a greater impact on R. padi establishment on organic

farms than on conventional farms. Irrespective of farming system, landscapes with

20

abundant ﬁeld margins and perennial corps were associated with low R. padi
establishment. After establishment, there was no difference in ground-living enemy
impact on R. padi population growth rate between farming systems, but impact was
greater in landscapes where arable land was contiguous (Ostrnan et al. 2001).

Impacts of the agricultural landscape in the soybean system

The landscape of the north-central U.S. soybean growing region is a mosaic of
crop and non-crop habitats, containing both primary and secondary hosts of A. glycines.
Habitat complexity of a landscape, the spatial arrangement of crop and non-crop habitats,
edge effects, and habitat connectivity within a landscape can greatly inﬂuence soybean
aphid movement between its winter and summer host plants. Soybean ﬁelds are
ephemeral habitats where disturbances such as tillage, pesticide application and
harvesting force repeated re-colonization by soybean aphid and its natural enemies. The
surrounding landscape acts as a source of both. To aid in the development of a successful
biological control program for A. glycines, this study investigated how the proportion of
an agricultural landscape comprised of non-crop habitat inﬂuenced aphid abundance
within soybean ﬁelds.
Summary
The soybean aphid is an invasive pest that was ﬁrst detected in the US. in 2000.

This pest is attacked by a large guild of native and exotic predators throughout its
invaded range. Successful control of soybean aphid requires early season suppression by
predators. To improve management of soybean aphid we must understand the inﬂuence
of landscape composition in supplying predators to crop ﬁelds. In addition, interactions

between species within the soybean aphid food web must be investigated to determine

2]

how these inﬂuence the net biocontrol service provided. This dissertation aims to
highlight the importance of understanding species-speciﬁc relationships with the
landscape and illustrate the importance of conserving natural habitats in promoting

successful pest management in agricultural systems.

22

CHAPTER 2

IMPACT OF INTRAGUILD PREDATION BY ADULT HARMONIA AXYRIDIS
(COLEOPTERA: COCCINELLIDAE) ON APHIS GL YCINES (HEMIPTERA:

APHIDIDAE) BIOLOGICAL CONTROL IN CAGE STUDIES.

Abstract

The soybean aphid, Aphis glycines Matsumura, has become a principal arthropod
pest of soybean in the U. S. since its ﬁrst detection in 2000. This species threatens
soybean production through direct feeding damage and virus transmission. A diverse
guild of insect predators feeds on soybean aphid in Michigan including the exotic
coccinellid Harmonia axyridis, the native gall midge Aphidoletes aphidimyza and the
native lacewing Chrysoperla cornea. In addition to feeding on A. glycines some
members of this guild may also engage in intraguild predation. These interactions may
produce positive, negative, or neutral impacts on A. glycines biological control. I
explored the impact of intraguild predation on soybean aphid population dynamics by
comparing aphid populations in microcosms with either A. aphidimyza larvae or C.
cornea larvae alone, with both a H. axyridis adult and either A. aphidimyza or C. carnea
larvae, and without predators. When H. axyridis was present with larval A. aphidimyza
or C. carnea, the lady beetle acted as an intraguild predator. However, intraguild feeding
did not result in a release of aphid populations compared with microcosms containing
only the intraguild and aphid prey. A similar result was found in ﬁeld cages. Cages
allowing large predators had reduced numbers of A. aphidimyza and C. cornea larvae but

also signiﬁcantly fewer aphids compared with predator exclusion cages. Thus, in both lab

23

and ﬁeld studies the direct impact of H. axyridis on A. glycines overcame its negative
impact as an intraguild predator. Together, these studies indicate that although the exotic
H. axyridis does act as an intraguild predator and may contribute to local declines in A.
aphidimyza and C. cornea, it is also currently important in overall biological control of A.
glycines.

Introduction

In agricultural systems, knowledge of food web interactions is often limited due
to the complexity of tri-trophic relationships and the number of species involved
(Bascompte and Melian 2005). Invasions of exotic species are likely to profoundly affect
the structure of food webs in agroecosystems. The soybean aphid, Aphis glycines
Matsumura, a native of Asia, was ﬁrst detected in the U. S. in Wisconsin in July of 2000.
Its range now extends throughout the north-central soybean growing states and into
Canada. Feeding by A. glycines on the leaves, stems, and pods of the soybean plant
causes reductions in photosynthetic rate, plant growth, and seed yield of soybean plants
(DiFonzo and Hines 2002). This pest can also vector several viruses, including alfalfa
mosaic, soybean dwarf, soybean stunt, soybean mosaic, tobacco ringspot, and bean
yellow mosaic (VandenBerg et al. 1997, Clark and Perry 2002, Wang and Ghabrial
2002)

In this system, the invading community extends beyond soybean aphid, involving
the introduction, both intentional and unintentional, of an entire exotic food web
(Thompson and Townsend 2003). The primary host plants of soybean aphid are
buckthom species in the genus Rhamnus, the most abundant and widespread species is R.

cathartica L. (Rhamnaceae), a native to Asia introduced in the 1880’s as a landscape

24

plant (Yoo et al. 2005). The secondary host plant of A. glycines, cultivated soybean,
Glycine max L. (F abaceae) (also native to Asia), has been cultivated in the US since the
early 1800’s, and is one of the dominant crops in the north-central region. The guild of
predators that attack soybean aphid also contains non-native species, including the
following coccinellids: Harmonia axyridis Pallas, Coccinella septempunctata L. and
Hippodamia variegata (Goeze) (Gardiner and Parsons 2005). These organisms interact
with native predatory species including the coccinellids, Coleomegilla maculata
(DeGeer), Hippodamia convergens Guérin-Méneville, Hippodamia parenthesis (Say), the
lacewing Chrysoperla cornea (Stephens) and Aphidoletes aphidimyza (Rondani)
(Cecidomyiidae) that have adapted to utilizing soybean aphid as prey.
Intraguild interactions

In aphidophagous systems where several predators compete for a single dominant
prey, interactions such as intraguild predation are likely to occur (Lucas 2005). I
Intraguild predation (IGP hereafter), is deﬁned by Lucas et al (1998) as predatory
interactions between predators within a guild. Generalist predators can be effective
biological control agents of native and exotic pests (Symondson et al. 2002); however,
interactions such as IGP can also interfere with pest suppression (Snyder and Wise 1999).
Intraguild interactions between predators have resulted in diverse effects on prey
population dynamics and plant productivity (Polis et al. 1989, Polis 1991, Rosenheim
1998, Prasad and Snyder 2004, Snyder and Clevenger 2004, Snyder et al. 2004a).
Overall, these studies indicate that the impacts of IGP are community speciﬁc. As one of
the most abundant predators found in soybean ﬁelds (Fox et al. 2004, Rutledge et al.

2004, Costamagna and Landis 2007a), H. axyridis has the potential to impact both A.

25

glycines population dynamics and populations of native predators that have adapted to
utilizing soybean aphid as a food source.

There is abundant evidence that H. axyridis acts as an intraguild predator of other
coccinellids and populations of some native species have declined signiﬁcantly after the
establishment of exotics (Elliott et al. 1996, Colunga-Garcia and Gage 1998, Michaud
2002). Although IGP by H. axyridis on other predator groups has received less attention;
this study focuses on the intraguild interactions between H. axjyridis, C. carnea, and A.
aphidimyza. Previous studies have found that IGP by lady beetles on lacewing and A.
aphidimyza larvae did occur in microcosm studies but not under ﬁeld conditions. Lucas
et al. (1998) established that the lady beetle C. maculata was an intraguild predator of
both Chrysoperla ruﬁlabris (Bunneister) and A. aphidimyza in microcosms. In the ﬁeld,
Brown (2003) compared aphid predator composition before and after the introduction of
H. axyridis in apple orchards and found no effect of H. axyridis on A. aphidimyza and
chrysopid populations.

Impact of IGP on early season biological control

A critical time in the biological control of soybean aphid occurs in the early
season when aphids begin to colonize soybean plants. We have observed that when
aphid populations are low and patchy, adult H. axyridis act as “transient predators,”
spending short feeding bouts within an aphid colony, removing some of the aphid
population before moving on to feed elsewhere (Landis unpublished data). Through a
review of mark-recapture studies this behavior was also described by Evans (2003) who
stated that coccinellids often do not remain long in any one location and will forage in

many different habitats throughout their breeding season. When aphid populations are

26

low, adult coccinellids are less likely to lay eggs (Evans 2003); therefore lady beetle
larvae are not yet abundantly present in soybean ﬁelds. My hypothesis was that during
this early season time period H axyridis adults may release soybean aphid populations
ﬁ'om overall control by acting as intraguild predators of smaller “resident predators”,
including larvae of the aphid predatory midge A. aphidimyza and green lacewing C.
carnea. As immature stages these species have limited mobility; individuals in soybean
ﬁelds spend their entire juvenile stage feeding primarily on the soybean aphid on a single
plant or foraging among a few plants within a ﬁeld. If H. axyridis acts as an intraguild
predator and removes the majority of these small resident predaceous larvae, the
remaining aphids could be released from overall control. However, if H. axyridis feeds
primarily on aphids, these species may have an additive negative effect on A. glycines
populations. The objectives of this study were to: 1) investigate how the presence of H.
axyridis impacts populations of the native predators A. aphidimyza and C. cameo in
soybean agroecosystems and 2) determine if IGP among these arthropods impacts
soybean aphid population dynamics.
Methods

Rearing

Soybean plants var. 92B16 (Pioneer Hi-Bred International Inc., Johnston, 1A)
were maintained in a greenhouse at 24°C on a 16:8 h LzD. Soybean aphid was cultured
on soybean in a walk-in growth chamber (Percival Scientiﬁc, Perry, IA) at 24°C:16°C on
a 16:8 h LzD. The predatory midge, A. aphidimyza, was obtained from IPM Laboratories
(Locke, NY) as pupae. Flies were hatched by placing the pupae into a dark growth

chamber for 48 h at 24°C. Adults were released onto soybean aphid-infested soybean

27

plants and allowed to lay eggs. Larvae hatched within 3 d and reached their second instar
within 5 d. The aphid predator C. cornea was obtained from Koppert Biological Systems
(Ann Arbor, MI) as eggs and ﬁrst instar larvae. The eggs and larvae were placed in
rearing boxes on soybean plants infested with A. glycines. A culture of H. axyridis was
initiated by ﬁeld-collecting adult beetles from soybean ﬁelds at the Michigan State
University Beet and Bean F arm (Saginaw, MI); larvae and adult H. axyridis were reared
separately on soybean aphid-infested plants. All predator colonies were maintained at
24°C:16°C on a 16:8 h L:D.
Predator consumption rates

Aphid consumption rates of A. aphidimyza and C. carnea were measured in 60 x
15 mm Petri dish arenas. The bottom of the dish was lined with moist ﬁlter paper and
contained a soybean leaﬂet infested with 30 second instar or adult A. glycines. One
second instar A. aphidimyza or second instar C. carnea was added to each arena and
allowed to forage for 24 h, after which the predator was removed and the number of
remaining aphids recorded. Five replications of each treatment were arranged randomly
on a cafeteria tray and held in a walk-in growth chamber at 24°C:16°C on a 16:8 h L:D.
Microcosm design

Microcosms consisted of a 3.79 L cylindrical-shaped clear plastic container with
three 7 x 13 cm windows of ﬁne mesh netting. The container had a 10 cm opening. The
plastic container was inverted over a 10 cm square plastic pot containing three V1
soybean plants (Teare and Hodges 1994). The opening of the container was held inside

the pot; it was buried in the soil and sealed with tape. Microcosms were held on cafeteria

28

trays in a walk-in growth chamber at 24°C:16°C on a 16:8 h L:D and watered by ﬁlling
the tray with water.
Microcosm experimental procedure

The microcosm experiments were designed to measure the intensity of IGP by the
transient predator H. axyridis on the resident predators A. aphidimyza and C. cornea and
the resulting impact on A. glycines population dynamics. The interaction between H.
axyridis and the two resident larval predators was evaluated in separate experiments. In
both the A. aphidimyza and C. cornea microcosm experiments four treatments were
compared. For A. aphidimyza the treatments were: aphids alone (CONTROL), aphids +
A. aphidimyza (AA), aphids + H. axyridis (HA), and aphids + both predators (H. axyridis
and A. aphidimyza) (BOTH). The treatments for the C. cornea experiment were: aphids
alone (CONTROL), aphids + C. cornea (CC), aphids + H. axyridis (HA), and aphids + H.
axyridis and C. cornea (BOTH).

To begin a microcosm experiment, 10 adult and 5 second instar aphids were
placed onto clean leaf disks. One disk was then attached to a fully expanded leaf on each
of three plants within a pot, for a total of 45 aphids per microcosm. After 48 h, aphids
were counted in each microcosm, and in the A. aphidimyza experiment one second instar
A. aphidimyza was introduced onto each of the three soybean plants per microcosm in the
AA and BOTH treatments. Similarly, in the C. carnea experiment one second instar C.
carnea was introduced onto each of the three soybean plants per microcosm in the CC
and BOTH treatments 48 h after the introduction of A. glycines. This was considered the
beginning of the experiment, time 0 in all analyses. In the A. aphidimyza experiment

aphids averaged (i 1 SEM) 114.47 j; 6.97 (CONTROL), 113.33 : 6.85 (AA), 116.2 1

29

7.11 (HA), and 107.8 1 6.51(BOTH) at time 0 h. In the C. carnea experiment A. glycines
averaged 123.27 : 10.41 (CONTROL), 111.47 3: 8.24 (CC), 129.4 1 7.14 (HA), and
119.8 : 8.57 (BOTH) at 0 h. In both experiments, 24 h after the release of the larval
predators aphids were counted again in all treatments and one adult H. axyridis was
released into the HA and BOTH microcosm treatments. The beetle was allowed to
forage for 3 h, after which it was removed to simulate its transient feeding behavior in the
ﬁeld. During this 3 h H. axyridis had the opportunity to consume A. glycines (HA
treatment), or both aphids and the larval predators (BOTH treatment). After the removal
of H. axyridis, aphids were counted in the HA treatment and aphids and A. aphidimyza or
C. carnea were counted in the BOTH treatment. Aphids were then counted daily in all
treatments for the following two (1 and again every 48 h for an additional two sampling
periods to track aphid population growth in each treatment.
Field cage experiment

To determine how the intraguild interactions examined in the microcosm
experiment compared with interactions in the ﬁeld, where a larger complex of generalist
predators was present, a ﬁeld cage experiment was conducted. In the ﬁeld, H. axyridis
was one of the two large exotic lady beetle predators, with C. septempunctata also
abundant. These two coccinellids are both potential intraguild predators of a large
community of small resident predators, including syrphid ﬂy larvae, coccinellid larvae,
and 0. insidiosus adults and larvae in addition to A. aphidimyza and C. cornea larvae.
Three treatments were designed to measure how IGP by large lady beetle adults on small
predators impacted A. glycines populations. The three treatments were: a predator

exclusion cage, preventing all predators from accessing aphid populations, a medium-

30

mesh cage that allowed small predators to gain access to the cage but excluded large lady
beetle adults, and a large-mesh cage that allowed access by both large lady beetles and
small predators.
Cage design

All cages consisted of 1 m3 PVC frames with mesh enclosures that varied in the
size of the mesh on the sides of the cage. The predator exclusion cage was made entirely
of no-see-um netting (Venture Textiles, Braintree, MA). The medium- and large-mesh
cages had no-see-um netting roofs and cage sides made from either 2 mm mesh (medium-
mesh treatment) or 6 mm mesh (large-mesh treatment) (U.S. Netting, Erie, PA). All
cages were lined at their base with Velcro attached to a plastic barrier buried to secure the
cages at the soil surface and prevent entry of ground-dwelling predators. The cages were
arranged in a completely randomized design with three replicates of each treatment in a
soybean ﬁeld planted in 38.1 cm rows with soybean variety 92B16 (Pioneer Hi-Bred
International Inc., Johnston, IA) at the Entomology Research Farm at Michigan State
University (East Lansing, MI). Each cage contained two rows of plants, with an average
of 25.3 plants per cage. When the cages were established aphid density averaged 22.2
per plant. All predators were hand-removed from caged plants prior to beginning the
experiment. .
Sampling procedure

All cages were sampled weekly for 4 wks. A sample consisted of ﬁve plants that
were randomly selected in each cage. Each plant was examined and the number of

apterous and alate aphids and diversity and abundance of predators was recorded.

31

Statistics

In the microcosm experiments, we compared the change in the number of A.
aphidimyza and C. cornea before and after exposure to H. axyridis using a 2-sample t-
test. A repeated—measures mixed model analyses of variance (ANOVA) was used to
compare differences in aphid abundance between microcosm treatments. Block was
included as a random effects factor and treatment, time, and their interaction as ﬁxed
effects factors. Differences in aphid and predator abundance between treatments were
assessed by comparing least squares means. Aphid counts were square-root transformed
prior to analysis to meet the assumptions of the model.

In the ﬁeld cage experiment, the impact of excluding either large coccinellid
predators (H. axyridis and C. septempunctata) or all predators on aphid populations was
assessed using a repeated measures ANOVA random coefﬁcients model with time
included as a covariate and random effects for subject and subject-speciﬁc slopes on
time. Based on residual analysis aphid data were square-root transformed to meet the
assumptions of the model. Variation in the abundance of predators was also assessed
using a repeated measures ANOVA random coefﬁcients model with time included as a
covariate and random effects for subject and subject-speciﬁc slopes on time. A square-
root transformation was applied to predator means prior to analysis to meet the
assumptions the model. Differences in aphid and predator abundance between treatments
were assessed by comparing least squares means. Both the microcosm and ﬁeld cage
analyses were conducted using the PROC MIXED procedure of SAS version 9.1 (SAS

Institute Inc. 1999).

32

Results
Predator consumption rates

Second instar A. aphidimyza consumed on average 6.2 h: 1.71 second instar and
9.4 i 0.4 adult A. glycines per 24 h, while second instar C. cornea consumed 17 i 4.06
second instar and 3.4 i 0.98 adult aphids per 24 h.

Aphidoletes aphidimyza and H. axyridis

In the A. aphidimyza and H. axyridis microcosm experiment, aphid populations in
the CONTROL and HA treatments, both of which initially contained no predators, rose in
the ﬁrst 24 h while AA and BOTH treatments which contained A. aphidimyza remained
relatively constant during this time period (Figure 2.1). At 27 h, i.e. after introduction
and 3 h of feeding by H. axyridis, aphid populations in both the HA and BOTH
treatments declined. During its 3 h presence, H. axyridis acted as an intraguild predator,
signiﬁcantly reducing numbers of A. aphidimyza larvae (t=6.87, df=14, P<0.0001) in the
BOTH treatment. At time 24 h an average of 2.87 i 0.09 midges were found in the AA
and BOTH treatments. An average of 1.80 i 0.26 midge larvae per microcosm were
removed by H. axyridis during 3 h of foraging, leaving only 1.06 _t 0.23 midges per
microcosm at 27 h in the BOTH treatment.

From 27-168 h overall aphid populations increased but were not consistent in all
treatments as indicated by a signiﬁcant interaction between treatment and time (E6303 =
3.00, P=0.0001). In the AA treatment, A. aphidimyza reduced aphid populations relative
to the control beginning at 48 h and continuing until the end of the experiment (Figure
2.1). In this treatment A. aphidimyza reduced aphid populations in the ﬁrst 48 h of the

experiment (107.8 : 6.51 aphids per microcosm at 0 h to 87.93 i 7.84 aphids per

33

microcosm at 48 h) but after 48 h aphid populations increased, reaching 372.33 1: 30.0
aphids per microcosm by 168 h. Three h of H. axyridis feeding in the HA treatment also
signiﬁcantly reduced aphid abundance, from 140.53 : 9.31 aphids per microcosm at 24 h
to 91.47 i 7.97 per microcosm at 27 h. Following the removal of H. axyridis, aphid
populations in the HA treatment increased and were not signiﬁcantly different from the
AA treatment from 48 h to 120 h. At 168 h the HA treatment had signiﬁcantly higher
aphid populations compared with the AA and BOTH treatments. In the BOTH treatment
aphid population remained constant over the ﬁrst 24 h in the presence of A. aphidimyza.
Three hours of feeding by H. axyridis reduced A. glycines populations in the BOTH
treatment from 107.47 1: 6.06 to 75.86 i 7.6. The BOTH treatment was similar to the AA
treatment throughout the experiment and signiﬁcantly lower than the HA treatment
beginning at 48 h and continuing through 168 h (Figure 2.1).
Chrysoperla cornea and H. axyridis

In the C. cornea and H. axyridis microcosm experiment, aphid populations
increased in the CONTROL and HA treatments and remained relatively constant in the
CC and BOTH treatments (Figure 2.2). As in the A. aphidimyza microcosm experiment,
there was a signiﬁcant treatment by experimental time interaction (Fussos = 16.67,
P<0.0001). Initially, aphid populations increased in all treatments, both in the
CONTROL and HA treatments which did not contain any predators at the beginning of
the experiment and in the CC and BOTH treatments which contained C. cornea larvae.
At time 24 h H. axyridis was added to the HA and BOTH treatments, resulting in a
decline in aphid numbers in these treatments by time 27 h. During its 3 h presence, H.

axyridis acted as an intraguild predator signiﬁcantly reducing numbers of C. carnea

34

larvae (t=2.82 df=14, P=0.014) in the BOTH treatment. At time 24 h an average of 2.26
i 0.12 and 2.6 i 0.13 lacewings per microcosm were found in the RP and BOTH
treatments respectively. The intraguild predator H. axyridis removed an average of 1.07
i 0.28 lacewing larvae within 3 h of foraging, leaving 1.67 i 0.29 lacewing larvae per
microcosm in the BOTH treatment at 27 h. In the CC and BOTH treatments aphid
populations were signiﬁcantly lower than the CONTROL beginning at 48 h and
continuing throughout the experiment (Figure 2.2). In the CC treatment aphid
populations decreased between 48 h and 120 h and then began to increase ending the
experiment at 179.13 : 64.90 aphids per microcosm at 168 h (Figure 2.2). Despite the
IGP, aphid populations in the BOTH treatment were relatively constant from 48 h to 120
h (139.0 1 18.89 aphids per microcosm at 48 h to 141.33 1 34.13 aphids per microcosm
at 120 h) at which time populations began to increase reaching 168.53 i 61.96 aphids per
microcosm at 168 h.
Field cage experiment

Aphid populations increased in all three ﬁeld cage treatments (Figure 2.3),
however the impact of these treatments on aphid populations was not consistent across
the 4 weeks of the experiment, as indicated by a signiﬁcant treatment-by—time interaction
(F6,9=3.79, P=0.036). By wk four, aphid populations were signiﬁcantly lower in the
large-mesh treatment compared with the predator exclusion treatment (Figure 2.3).
Allowing access to C. septempunctata and H. axyridis delayed aphid populations from
reaching the economic threshold of 250 aphids per plant by two wks in the large-mesh
treatment compared with populations in the medium-mesh and predator exclusion

treatments. Adult H. axyridis and C. septempunctata were found only in the large-mesh

35

treatment, indicating that they were effectively excluded from the medium-mesh and
predator exclusion treatments. Populations of large lady beetles were low throughout the
experiment, peaking at 0.4 i 0.24 per plant on July 21. The population composition of
large coccinellids in the large-mesh treatment was 55% C. septempunctata and 44% H.
axyridis. Coccinellid larvae were found in both the medium- and large-mesh treatments
and the difference in their abundance between the two treatments varied across time,
indicated by a marginally signiﬁcant interaction (F3, 12:3.33, P=0.056). Coccinellid larvae
were signiﬁcantly more abundant in the large-mesh treatment on July 21 reaching 1.47 i
0.48 per plant compared with 0.07 i 0.07 per plant in the medium-mesh treatment. Their
populations were similar in the two treatments on the other three sampling dates (Figure
2.4). Harmonia axyridis was the most abundant species of coccinellid larvae found,
averaging 0.05 i 0.28 per plant in the medium-mesh treatment and 0.32 i 0.09 per plant
in the large-mesh treatment. Larvae of C. septempunctata were the second most
abundant species, averaging 0.02 i 0.02 per plant in the medium-mesh treatment and
0.06 _+_- 0.05 per plant in the large-mesh treatment. Lacewing larvae were only detected in
the medium-mesh treatment, at 0.4 i 0.16 larvae per plant on July 21. There was no
difference in the number of A. aphidimyza found in the medium- and large-mesh
treatments on any dates (F,,4_5=0.19, P=0.679) (Figure 2.5). We were unable to restrict 0.
insidiosus from entering the predator exclusion treatment. There was a signiﬁcant
interaction in the abundance of 0. insidiosus between the three treatments and time

(F 633:2.75, P=0.05). Despite removing individuals while sampling the predator

exclusion treatment, 0. insidiosus increased throughout the experiment and was

36

signiﬁcantly higher in this treatment compared with the large-mesh treatment (P=0.027)
by July 29 (Figure 2.6).
Discussion

Generalist predators can exert strong top-down population regulation of A.
glycines in soybean ﬁelds (Costamagna and Landis 2007a). Harmonia axyridis is one of
the most abundant predators found in soybean across the invasive range of A. glycines
(Fox et al. 2004, Rutledge et al. 2004, Costamagna and Landis 2007a), thus
understanding its role in shaping the A. glycines predator guild is of critical importance in
the overall suppression of this invasive species.
Microcosm studies

In microcosm experiments we found that H. axyridis acting as both an intraguild
and a direct predator had differential impacts on both the survival of the intraguild prey
and A. glycines population dynamics. In tests with A. aphidimyza, the presence of this
predator alone slowed, but did not prevent, the eventual increase of A. glycines to high
levels. Based on this result and the aphid consumption rate data we conclude that under
the conditions tested, A. aphidimyza was not efﬁcient enough in preventing the aphid
from reaching outbreak levels. In contrast, only 3 h of feeding by H. axyridis provided
suppression of A. glycines essentially equal to the continuous presence of A. aphidimyza
over the full 168 h. During its 3 h tenure, H. axyridis also acted as a signiﬁcant intraguild
predator of A. aphidimyza. This IGP did not result in a release of aphid populations from
overall control; aphid populations exposed to both predators were not signiﬁcantly
different from aphids exposed only to A. aphidimyza. The overall picture that emerges

from this experiment is that while H. axyridis is an intraguild predator of A. aphidimyza,

37

IGP did not release A. glycines from control as A. aphidimyza is simply not efﬁcient
enough at the densities tested to suppress the aphid on its own.

A somewhat different picture emerges for the C. cornea and H. axyridis
interaction. As in the previous experiment, H. axyridis alone slowed but did not prevent
the eventual increase of A. glycines to high levels. However, in contrast to A.
aphidimyza, C. cornea larvae alone were able to maintain A. glycines populations at low
levels throughout the experiment. IGP of an effective predator could potentially release
aphids from control. While H. axyridis did signiﬁcantly reduce C. cornea, it did not
reduce the levels of this intraguild prey to the same degree as it did A. aphidimyza. This
may be because C. carnea are larger, more mobile, or less preferred intraguild prey than
A. aphidimyza. The overall result was that either C. cornea alone or the combination of
C. cornea and H. axyridis together resulted in experiment-long aphid control.

Taken together, the microcosm experiments illustrate a complex set of
interactions between H. axyridis, A. aphidimyza, C. carnea and A. glycines. First, even a
single visit by a H. axyridis may produce long term impacts on aphid suppression. Three
hours of H. axyridis feeding signiﬁcantly reduced overall A. glycines population growth
in both experiments. In the ﬁeld, a similar result could mean a difference of several days
in when a threshold population is reached or could conceivably even maintain aphid
populations below thresholds, particularly if repeated visits by transient coccinellid
predators occur. Second, intraguild predators can have strong or weak impacts on
intraguild prey depending on their ability to exploit particular guild members and the
intraguild prey’s ability to escape predation. Third, the ability of A. aphidimyza and C.

canned to suppress aphid populations after IGP depends on both the number which

38

survive and their inherent ability to consume prey. Finally, it is likely that relatively
stochastic events such as the exact timing of intraguild interactions and the numbers of all
predators and prey will shiﬂ the outcome of these interactions.
Field cages

In the ﬁeld cage experiment aphid populations were reduced in cages, allowing
colonization by H. axyridis and C. septempunctata. Predation by coccinellids delayed A.
glycines from reaching the 250 per plant threshold for 2 wks compared with the medium-
mesh and predator exclusion treatments. Throughout this investigation numbers of all
predators remained low in all treatments, indicating that small numbers of predators can
have a signiﬁcant impact on A. glycines populations in the ﬁeld. We were unable to
exclude 0. insidiosus from our predator exclusion treatment; its abundance was highest
in this treatment by the end of the experiment. These data indicate that 0. insidiosus may
have beneﬁted ﬁom the exclusion of other predators; however, it appears to have little
impact on aphid biological control when acting alone at the population levels found in
this experiment. While A. aphidimyza larvae were more abundant and lacewing larvae
were only found in the medium-mesh treatment, there was not a signiﬁcant difference in
the number of these intraguild prey between the medium- and large-mesh treatments.
Since IGP did occur in our microcosm experiments, it is likely to also occur at some level
in the ﬁeld. It is possible that the conﬁned arena of a microcosm elevated the extent of
IGP relative to what occurs in soybean ﬁelds. Alternatively, the relatively low and
variable number of A. aphidimyza and C. carnea found in our ﬁeld study may have
reduced our ability to detect differences. Coccinellid larvae were more abundant in the

large-mesh treatment and H. axyridis was the most common species. Since H. axyridis

39

and C. septempunctata were able to deposit eggs inside the large-mesh cage treatment,
larvae would not have to colonize these cages. This could have led to higher populations
in this treatment compared to the medium-mesh where coccinellid larvae would have to
disperse into the cages on their own. Coccinellid larvae are likely to greatly impact both
IGP and aphid population dynamics. Aphid consumption by H. axyridis can vary from
90 to 370 aphids (depending on the species of aphid) during its larval stage (Hukusima
and Kamei 1970 in Koch 2003). Coccinellid larvae are also known to act as intraguild
predators (Lucus et al 1998, Kajita et al. 2000, Snyder et al. 2004 a, b). It is likely that in
this study coccinellid larvae contributed to decreased aphid populations in the large-mesh
treatment and may have engaged in intraguild interactions with other predators.

When evaluating our ﬁeld cage experiment, it is critical to assess the role of cage
design in inﬂuencing the results. Our primary concern was that our predator exclusion
cages prevented dispersal of alate aphids, which could have been responsible for the
elevated aphid populations in this treatment. While a lack of alate dispersal may have
contributed to higher A. glycines populations, there is evidence that this is not the only
factor resulting in lower populations in the large-mesh cage treatment. If dispersal was
driving the difference in aphid abundance, the medium-mesh and large-mesh treatments
should have both had signiﬁcantly fewer aphids, as alates could disperse from these
treatments. Instead, we found that the number of aphids in the medium-mesh which
allowed alate dispersal did not differ from populations in the exclusion treatment where
dispersal was prevented. This indicates that it was the difference between the large-mesh
and medium-mesh treatments that resulted in better aphid suppression. The only

difference between these treatments was the presence or absence of H. axyridis and C.

40

septempunctata. A second concern was that our medium-mesh was small enough to
exclude large lady beetles but still provided equal colonization of small predators
compared with the large-mesh treatment. We evaluated this in a preliminary greenhouse
experiment prior to conducting our ﬁeld experiment. We established two portable hoop-
houses on top of a greenhouse bench and inside each placed one medium-mesh cage and
one large-mesh cage. Inside each cage were 12 heavily-infested soybean plants. The
cages was surrounded on all sides with 2 rows of uninfested soybean plants onto which
we released 0. insidiosus adults, A. aphidimyza adults and C. cornea larvae (100
individuals of each species). The cages were sampled at 48 hr, 96 hr, and 144 hr after the
release of the predators. For both 0. insidiosus (F1, 2:1 .84, P=0.779).and C. cornea (F1,
2=O.07, P=0.817) there was no difference in colonization across the three sampling
periods. As adult A. aphidimyza were released we did not begin to detect larvae until 96
h, when one larva was found in a large-mesh cage. At 144 h there was not a signiﬁcant
difference in colonization by A. aphidimyza in the medium-mesh and large-mesh
treatments (F1, 2:2. 1 3, P=0.281). Our third concern was that abiotic conditions were
equal in all cages. To prevent greater shading in our predator exclusion treatment we
used the exclusion mesh for the roofs of all three cage treatments, so that all treatments
were equally shaded.
Impacts of H. axyridis on biological control

The impact of guild member displacement by an intraguild predator in overall
biological control of aphids is likely to depend on several factors, including the
contribution of each natural enemy to aphid suppression and the severity of the IGP. The

beneﬁt of an intraguild predator such as a lady beetle to overall biological control may be

41

tied to its foraging strategy. Rosenheim and Corbett (2003) modeled how foraging
strategy can impact the outcome of biological control of a sedentary pest. They found
that when the top predator exhibited a sit and wait foraging strategy, actively foraging
intermediate predators were suppressed, releasing sedentary herbivore populations from
overall control. This outcome changed when the top predator was an active forager, as is
the case with H. axyridis. Here the sedentary herbivore was not released from overall
control because both the intermediate and top predators were likely to encounter and
consume it, so any losses of intermediate predators due to IGP had less of a direct effect
on pest suppression (Rosenheim and Corbett 2003). This is what we ﬁnd in the soybean
system, where biological control of the sedentary A. glycines was not impacted by the
removal of A. aphidimyza and C. cornea by the active top predator H. axyridis.

While adult coccinellids are often signiﬁcant intraguild predators (Colunga-
Garcia and Gage 1998, Losey and Denno 1998, Lucas et al. 1998, Phoofolo and Obrycki
1998, Cardinale et al. 2003, Colfer et al. 2003, Snyder et al. 2004b, Snyder et al. 2004a,
Cottrell 2005), inclusion of these predators often results in increased biological control I
(Colfer and Rosenheim 2001, Cardinale et al. 2003, Aquilino et al. 2005). The presence
of H. axyridis has been found to increase biological control of aphid pests in communities
containing other predators as well as predators and parasitoids. Aquilino et al. (2005)
manipulated the diversity of host plant species and predators in microcosms to determine
how top-down and bottom-up diversity affected consumption of pea aphid,
Acyrthosiphon pisum Harris. They found that increasing the diversity of predators
increased aphid consumption. This was attributed to poor performance of Nabis sp. and

C. maculata when each predator was present alone. In contrast, consumption by H.

42

axyridis alone in the majority of treatments was typically equal to or exceeded
consumption in microcosms containing all three predators (Aquilino et al. 2005).
Cardinale et al. (2003) found that when H. axyridis, Nabis sp., and the parasitic wasp
Aphidius ervi Haliday were combined in ﬁeld cages greater pea aphids suppression than
was predicted by the summed impact of each species alone (Cardinale et al. 2003).
Snyder et al. (2004a) found that on caged rose plants infested with Macrosiphum
euphorbiae Thomas, the presence of H. axyridis dampened aphid populations without
impacting the density oprhelinus asychis (Walker) pupae. Similarly, in soybean
Costamagna (2006) found that although high levels of IGP of parasitoids by predators
were detected, no evidence for disruption in the level of parasitism was found. In
contrast, the presence of coccinellids, including H. axyridis, was responsible for strong
suppression of A. glycines and restored soybean biomass and yield to levels similar to
control treatments lacking aphids (Costamagna 2006).
Conclusions

My hypothesis was that early in the season when lady beetles are highly mobile,
IGP events occurring between the exotic coccinellid H. axyridis and larvae of the native
C. cornea and A. aphidimyza could impact the success of soybean aphid biological
control. While the transient top predator H. axyridis did act as an intraguild predator of
both intermediate predators A. aphidimyza and C. cornea, we did not ﬁnd evidence of a
release of soybean aphid due to IGP in either our microcosm or ﬁeld cage experiments.
Based on the results of this study, the presence of H. axyridis may contribute to local
declines in A. aphidimyza and C. cornea; however, biological control of soybean aphid

would not likely be improved by removing H axyridis from the system. When H.

43

axyridis is excluded from aphid colonies in the ﬁeld, populations grow to levels not
signiﬁcantly different from cages excluding all predators. This suggests that intraguild
prey in the soybean system, including A. aphidimyza, C. cornea, and 0. insidiosus, while
important to native biological diversity, are not currently major contributors to overall

biological control of this invasive pest.

44

Figure 2.1.

      

 

 

 

 

 

—o—CONTROL s- AA -----A- HA ------x BOTH
E
700 g, I?
8 R
“a? ‘6
CD
600 E3 g
>5 E e
(uh: 500 200 ‘ :‘E
+ 150
E
g 400
.9
2
7: 300
E
g
B, 200
< " .
100 x
0
0 24 48 72 120 168
Time (h)
TREATMENT 0 24 27 48 72 120 168
CONTROL a a ' a a a a
AA a a - bc bc be c
HA a a a b b b b
BOTH a a a c c c c

Figure 2.1. Mean number of A. glycines per microcosm in the CONTROL (A. glycines
only), AA (A. glycines + A. aphidimyza), HA (A. glycines + H. axyridis) and BOTH (A.
glycines + A. aphidimyza + H. axyridis) treatments. Larval A. aphidimyza were released
at 0 h, H. axyridis were released at 24 h and removed at 27 h. Only the BOTH and HA
treatments (which included H. axyridis) were counted at 27 h. The ﬁgure insert shows
the change in aphid populations after 3 h of feeding by H. axyridis in the HA and BOTH

treatments. Mean comparisons are based on square-root transformed data (P<0.05).

45

Figure 2.2.

—o——CONTROL arm—cc -----. HA --x-- BOTH

1 200

     

 

 

 

 

E
V
9‘, E
8 R
1000 g v
w 8
2 E >
LIJ ‘ °
0‘) < E,
+ 800 I g
s i I
(I)
O
E 600
s,
8
'6 400
E
o:
si
200
0
0 24 48 72 120 168
Time (h)
TREATMENT 0 24 27 48 72 120 168
CONTROL a ab - a a a a
AA a a - b b c 0
HA a b a ab a b b
BOTH a ab 3 b b C C

Figure 2.2. Mean number of aphids per microcosm in the CONTROL (A. glycines only),
CC (A. glycines + C. cornea), HA (A. glycines + H. axyridis) and BOTH (A. glycines +
C. cornea + H. axyridis) treatments. Larval C. carnea were released at 0 h, H. axyridis
were released at 24 h and removed at 27 h. Only the BOTH and HA treatments (which
included H. axyridis) were counted at 27 h. The ﬁgure insert shows the change in aphid
populations after 3 h of feeding by H. axyridis in the HA and BOTH treatments. Mean

comparisons are based on square-root transformed data (P<0.05).

46

Figure 2.3.

—a— Predator Exclusion ---e- -- Medium-Mesh —-— Large-Mesh

 

100000

10000
2
LU
U)

i 1000
C
<_0
E
8

.s 100
0
...>.~
D)
<15

10

1

 

 

7 14 21 29
July 2005

Figure 2.3. Mean number of A. glycines per plant in predator exclusion, medium-mesh
(small predators only) and large-mesh (small and large predators) cage treatments.
Asterisk indicates a signiﬁcant difference between square-root transformed mean A.

glycines in the predator exclusion and large-mesh treatments (P<0.05).

47

Figure 2.4.

- - -e- - - Medium Mesh +Large Mesh

 

g 2.5
‘I' 2 *
E

L“

D.

a 1.5
(U

E

3 1
g

g 0.5
U

3 0

 

 

 

July 2005

Figure 2.4. Mean number of lady beetle larvae per plant in the medium-mesh and large-
mesh cage treatments. Asterisk indicates a signiﬁcant difference between treatments

based on square-root transformed data (P<0.05).

48

3'!
"i

N

u,

- - -e- - - Medium-Mesh ——-—Large—Mesh

3.5

3
1

1.5

0.5

 

 

A. aphidimyza Larvae/Plant + SEM

 

July 2005

Figure 2.5. Mean number of A. aphidimyza per plant in medium-mesh and large-mesh

cage treatments.

49

Figure 2.6.

+ Predator Exclusion —e— Medium-Mesh + Large-Mesh

 

 

3 *
E
LIJ 2.5
U)
+
E 2
.52
O.
\ 15
8
.Q 1 .
8 ‘
,t
7. 0.5
O

0

7 14 21 29

July 2005

Figure 2.6. Mean number of O. insidiosus per plant in the predator exclusion, medium-
mesh and large-mesh treatments. Asterisk indicates a signiﬁcant difference between the
predator exclusion and large-mesh treatments based on square-root transformed data

(P<0.05).

50

CHAPTER 3

LANDSCAPE DIVERSITY ENHANCES BIOLOGICAL CONTROL OF AN

INTRODUCED CROP PEST IN THE NORTH-CENTRAL U.S.

Abstract

Arthropod predators and parasitoids provide valuable ecosystem services in
agricultural crops by suppressing populations of insect herbivores. Many natural enemies
are inﬂuenced by non-crop habitat surrounding agricultural ﬁelds and understanding if,
and at what scales, land use patterns inﬂuence natural enemies is essential to predicting
how landscape alters biological control services. Here I focus on biological control of
soybean aphid, Aphis glycines Matumura, a specialist crop pest recently introduced to the
north-central US. I measured the amount of biological control service supplied to
soybean in 26 replicate ﬁelds across Michigan, Wisconsin, Iowa, and Minnesota across
two years (2005-06). I measured the impact of natural enemies by experimentally
excluding or allowing access to soybean aphid infested plants and comparing aphid
population growth over 14 d. I also monitored aphid and natural enemy populations at
large in each ﬁeld. Predators, principally coccinellid beetles, dominated the natural
enemy community of soybean in both years. In the absence of aphid predators, A.
glycines increased signiﬁcantly, with 5.3 fold higher aphid populations on plants in
exclusion cages versus the open ﬁeld after 14 d. I calculated a biological control services
index (BSI) based on relative suppression of aphid populations and related it to landscape
diversity and composition at multiple spatial scales surrounding each site. I found that

BSI values increased with landscape diversity, measured as Simpson’s D. Landscapes

51

dominated by com and soybean ﬁelds provided less biocontrol service to soybean
compared with landscapes with an abundance of crop and non-crop habitats. The
abundance of Coccinellidae was related to landscape composition, with beetles being
more abundant in landscapes with an abundance of forest and grassland compared with
landscapes dominated by agricultural crops. Landscape diversity and composition at a
scale of 1.5 km surrounding the focal ﬁeld explained the greatest proportion of the
variation in BSI and Coccinellidae abundance. This study indicates that natural enemies
provide a regionally important ecosystem service by suppressing a key soybean pest,
reducing the need for insecticide applications. Furthermore, it suggests that management
to maintain or enhance landscape diversity has the potential to stabilize or increase
biocontrol services.
Introduction

Insects provide vital ecosystem services including decomposition, pollination, and
biological control of crop pests (Losey and Vaughan 2006). In agroecosystems, the
diversity and abundance of natural enemies that provide biological control in crop ﬁelds
are inﬂuenced by the structure and composition of the surrounding landscape (Marino
and Landis 1996, Colunga-Garcia et al. 1997, Bommarco 1998, Elliott et a1. 1999, Thies
et al. 2003, Schmidt and Tscharntke 2005, Tscharntke et al. 2005b). Landscape variables
such as habitat complexity, quality, and patchiness as well as an organism’s dispersal
capability all impact the ability of a landscape to support biological control in agricultural
croplands (Elliott et al. 1999, Thies et al. 2003, Schmidt and Tscharntke 2005).
Moreover, agricultural crop ﬁelds are ephemeral habitats in which anthropogenic

disturbances such as tillage, pesticide application, and harvesting require arthropods to

52

frequently re-colonize crops (Wissinger 1997). The surrounding landscape provides the
local species pool of arthropods for this re-colonization and thus may inﬂuence the level
of biological control in frequently disturbed crop ﬁelds (Lee et al. 2001).

Invasions of exotic species can signiﬁcantly alter the abundance and distribution
of herbivores within a landscape. Landscapes vary in their suitability for invasive
species, and understanding how landscape structure inﬂuences pest suppression is critical
to successfully managing invaders in their non-native range. In the summer of 2000, the
soybean aphid, Aphis glycines Matsumura, was ﬁrst detected in the US, and has now
spread throughout 23 soybean growing states and 3 Canadian provinces (Venette and
Ragsdale 2004). Feeding by A. glycines on the leaves, stems, and pods of the soybean
plant causes reductions in photosynthetic rate (Macedo et al. 2003), plant growth, and
seed yield of soybean plants (Ragsdale et al. 2007). Soybean aphid is consumed by a
diversity of natural enemies that can provide strong top down regulation of its
populations (Costamagna and Landis 2007a) and result in a trophic cascade of increased
crop yields (Costamagna et al. 2007a). While several parasitoids can be found attacking
soybean aphid, generalist predators currently dominate the natural enemy community
(Costamagna et al. in press).

A complex of native and exotic generalist predators attack and feed on soybean
aphid in North America. These include Harmonia axyridis Pallas, Coccinella
septempuncata L., Hippodamia variegata (Goeze), Hippodamia convergens Guérin-
Méneville, and Coleomegilla maculata DeGeer (Coleoptera: Coccinellidae), Orius
insidiosus Say (Hemiptera: Anthocoridae), Chrysoperla cornea (Stephens) (N europtera:

Chrysopidae), Aphidoletes aphidomyza (Rondani) (Diptera: Cecidomyiidae), Leucopis

53

spp. (Diptera: Chamaemyiidae), and Syrphidae (Diptera) (Fox et al. 2004, 2005, Desneux
et al. 2006, Costamagna and Landis 2007a). Generalist predators have been shown to
signiﬁcantly reduce soybean aphid populations (Fox et al. 2005, Desneux et al. 2006,
Costamagna and Landis 2007a), however, the level of suppression is not consistent
between studies. For example, Desneux (2006) reported low overall aphid populations
(less than 10 per plant) varying only 1.1-3.5 fold in the presence/absence of predators. In
contrast, Costamagna and Landis (2007a) reported aphid populations of several thousand
per plant varying 6.8 fold in the presence/absence of predators. These studies were
conducted in landscapes that vary tremendously in complexity, from diverse landscapes
in Michigan (Fox et al. 2004, Costamagna and Landis 2007a) to corn and soybean-
dominated landscapes in Indiana (Desneux et a1. 2006). It is likely that differences in the
efﬁcacy of biological control are due to variation in landscape composition surrounding
soybean ﬁelds across the north central region.

Given strong local variation in the level of soybean aphid suppression, the goal of
this study was to determine if and at what scales, landscape diversity and composition
may explain biological control of A. glycines in agricultural landscapes. My hypothesis
was that landscapes composed of a high proportion of non-crop habitat such as grasslands
and forests would provide overwintering habitat and alternative prey resources to natural
enemies. Such landscapes would in turn increase the abundance of generalist predators
and result in greater suppression of the soybean aphid. My objectives were to 1) measure
the impact of predator assemblages on soybean aphid populations across the north central
US. and 2) determine if the level of biological control detected in soybean ﬁelds was

related to the diversity and composition of the surrounding landscape.

54

Methods

Field sites

Between June and August of 2005 and 2006, I examined the abundance of aphids
and arthropod natural enemies at 26 sites distributed amongst 4 states (Iowa, Michigan,
Minnesota, and Wisconsin) in the north-central US. as part of a larger multi-state study
of soybean aphid management (Figure 3.1). Sites were selected to study how landscape
variation across a large portion of the soybean growing region affected biological control
of soybean aphid. Within each year, a minimum distance of 30 km separated each ﬁeld
site. Across years, I maintained the relative regional distribution of sites, but sampled
different ﬁelds. The average distance between soybean ﬁelds sampled in subsequent
years was 1.6 km (range = 0.2-4.6 km, Figure 3.1). Field size averaged 16.0 ha (range =
13.6-48.1 ha). In each replicate ﬁeld, a randomized complete block design with 4-6
blocks and either 4 (2005) or 5 (2006) treatments was established in 0.4 ha (2005) or 0.2
ha (2006) plots. Multiple experiments occurred within these ﬁelds, and this study was
conducted in plots which were not treated with insecticide.
Natural enemy impact experimental design

Following the methods of Costamagna et al. (2007a), I used cages to exclude
natural enemies from selected plants, contrasting the population growth of aphids in the
presence or absence of these antagonists. When an average population of 10 aphids per
plant was reached in the overall ﬁeld, a cage were erected in the center of each plot (4 per
site), each enclosing a single plant. Each caged plant was paired with an un-caged plant,
1 m away in the same row, hereafter referred to as “open” plants. Aphids were either

added or removed to start each plant with 10 aphids of mixed age classes. In some

55

locations, aphid populations never reached an average level of 10 aphids per plant. In
these sites I manually redistributed aphids from the ﬁeld to the experimental plants to
initiate the study. Cages consisted of a 0.4 m diameter x 1 m tall tomato wire support
covered by a sleeve of ﬁne-mesh white no-see-um netting (openings of 0.65 x 0.17 mm;
Kaplan Simon, Braintree, MA). The mesh sleeve was tied at the top and attached to a
metal “T” post, it was secured at the bottom by burying the edges in the soil. The number
of apterous and alate aphids was recorded weekly on both the open and caged plants
using non-destructive ﬁeld counts 7 and 14 (I post establishment.
Assessment of cage effects

Cage studies must address possible effects on microclimate and test organism
behavior. Using similar cages, Costamagna and Landis (2007a) found that a “sham” cage
treatment (exclusion cages with windows to allow access by natural enemies) and a no-
cage treatments did not differ signiﬁcantly in aphid population growth after 7 or 14 (1.
Another potential concern is that exclusion cages could prevent dispersal of alate aphids
and artiﬁcially elevate aphid populations. To minimize this possibility, I limited our
study to 14 d. After 7 d, alate number was low and similar (1.4 d: 0.1 and 1.9 i 0.1 alate
aphids per plant (mean :l: SEM)) in the exclusion cage and open treatments respectively.
At 14 d the exclusion cage treatment had an average of 20.7 i 1.4 alates per plant while
the open plant had 1.8 i 0.1 alates per plant. To examine the impact of cages with a high
number of alates on our results, I analyzed a sub-set of the data excluding all cages which
had more alates than were found on open plants (10% of cages). Exclusion of these data

from analysis (not shown) did not affect the outcome.

56

Soybean aphid and natural enemy sampling

Destructive whole plant counts were used to estimate A. glycines abundance at
weekly intervals ﬁ'om early June until late August. In each plot, 5 randomly selected
plants were removed from the ground and the number of apterous and alate aphids were
counted on each plant. I present the date of ﬁrst aphid detection and the date when the
economic threshold of 250 aphids per plant (Ragsdale et al. 2007) was reached for each
site (Table 3.1).

During the two weeks of the natural enemy impact experiment, the diversity and
abundance of natural enemies was estimated using two different sampling techniques
(Schmidt et al. 2008b). To measure aerially dispersing natural enemy populations, a
single unbaited yellow sticky card (PHEROCON AM, Great Lakes IPM, Vestaburg, MI)
was placed in each plot at all study sites. In the center of each plot, a metal “T” fence
post was erected with holes every 10 cm vertically. A 0.6 cm diameter dowel was placed
through a hole to suspend a 22.9 x 27.9 cm sticky card just above the plant canopy.
Sticky traps were collected at 7 and 14 d and all adult Coccinellidae, Anthocoridae,
Nabidae, Chrysopidae, Hemerobiidae and Syrphidae (2006 only) were counted. I also
visually inspected plants on day 7 and 14 to measure the diversity and abundance of
natural enemies foraging in soybean. In 2005, ﬁve randomly selected plants were
inspected per plot and all adult and immature Coccinellidae, Orius insidiosus
(Anthocoridae), Nabidae, Chrysopidae, Hemerobiidae, Syrphidae, Aphidoletes
aphidimyza (Cecidomyiidae), Araneae, Aphelinidae, and Braconidae were counted. In
2006, I reduced the number of locations sampled by visual plant counts to one site per

state and inspected 10 randomly selected plants per plot.

57

Landscape analysis

Field geospatial data was collected using a handheld GPS receiver using Wide
Area Augmentation System (WAAS) correction. The spatial coordinate for the center of
each site was used to obtain ortho-rectiﬁed digital aerial imagery. I digitized the habitats
surrounding each study site to a radius of 3.5 km using ARC GIS 9.1. These aerial
images differed temporally, representing images captured between 1998 and 2006. Land
use changes between the image date and study period were recorded by ground-
veriﬁcation in June — August annually, with corrections made during the digitization
process. At the same time I also determined the speciﬁc land cover present within all
areas of each landscape (Table 3.2). Some locations included polygons that were not
visible from a roadway and permission to access private lands could not be obtained.
These polygons were given a value of zero and were excluded from further analysis. The
area of each site that could not be identiﬁed varied from 0-2%. The smallest polygons
identiﬁed included ﬁeld plots on university research farms and small patches of fallow
ﬁeld (< 5 m2) the largest were contiguous urban areas, lakes, and forests (< 11.9 km2).

Landscape diversity was measured using Simpson's Index (D) (Simpson 1949).
Simpson’s Index is typically used to examine the variance of species abundance
distributions, here I applied it to examine variance in the proportion of area covered by
each of 22 landuse categories. This index was calculated for each of the landscapes at

each of the spatial scales examined. The equation for Simpson’s Index (D) is: D =
1/Z(pi)2 where pi=proportion of habitat in the ith land-cover category (D increases as

diversity increases). Using methods modeled after Thies et al. (2003) I measured

58

landscape diversity at 7 spatial scales ranging from 0.5 km to 3.5 km radii (at 0.5 km
intervals) from the ﬁeld center.
Statistical analysis

A repeated-measures mixed model analyses of variance (ANOVA) (PROC
MD(ED in SAS v. 9.1) was used to compare the average aphid populations on the caged
and open plants in the 26 locations sampled. Fixed factors included in the model were:
cage treatment, week, and the cage treatrnent*week interaction. Cage treatment replicate
nested within site was included as a random effect and week was the repeated variable.
Differences in aphid abundance between treatments were assessed by comparing least
squares means. Aphid counts were log transformed prior to analysis to meet the
assumptions of normality and homogeneity of variances (SAS Institute, 1999).

To relate biological control services to landscape variables I calculated the
relative aphid suppression found in each site by expressing the change in aphid numbers
on open and caged plants, as a proportion of aphid abundance in the absence of predators
for any given site. The resulting Biocontrol Services Index (BSI), varied from 0-1, with
values increasing as the level of aphid predation increases.

(pg: 1 (A:p_ A03)
351 = GP

it
Where Ac = the number of aphids on the caged plant on day 14, Ao= the number of

 

aphids on the open plant on day 14, p=plot, and Ir: the number of replicates for a given
site. In one case a negative BS1 value was calculated due to higher aphid populations on
the open plant than in the corresponding exclusion cage treatment. Since this indicated a

lack of effective biological control, this site was given an index value of zero.

59

To evaluate the relationship between predator abundance, BSI, and landscape
variables, I performed a principal components analysis (PCA). The 22 land use
categories were combined into 8 variables included in the PCA analysis: 4 crop variables
(proportion corn, soybean, wheat and other crops), and 4 non-crop variables (proportion
forest, grasslands, urban, and water) (Table 3.2). Principal component axes were
extracted using the correlations among variables and the resulting factors were not
rotated. I restricted our analysis to the ﬁrst two eigenvectors which together explained
between 43.8-56.1% of the variability in landscape data. This was done for each spatial
scale (radii of 0.5—3.5 km). The PCA analysis was completed using J MP version 7 (SAS
Institute 2007).

To assess the inﬂuence of the principal components and landscape diversity (D)
on BSI and coccinellid abundance (the average number of adult Coccinellidae collected
from sticky cards across the two weeks of sampling) I compared all possible models
containing these variables using Akaike’s Information Criterion, adjusted for sample size
(AlCc) (Bumham and Anderson 2002). The AICc analysis was conducted at each spatial
scale of analysis (0.5-3.5 km). At radii of 1 km and larger, eight models were compared:
Intercept only, Diversity (Simpson’s D), PCl, PC2, Diversity + PC], Diversity + PC2,
PCl + PC2, and Diversity + PC] + PC2. At a 0.5 km radius the interpretation of PC2

was not consistent with larger spatial scales and was not included in model analysis. For
_ each analysis, I present the maximum log-likelihood estimate, the Akaike weights 0111,)
and AICc differences (A,). For each response variable I present the model with the

minimum AICc value, which has the best support for the data and any other models with

a AlCc difference of less than 2 (Bumham and Anderson 2002). Models with a A, in this

60

range are termed “competing models” (Ribic and Sample 2001) and are considered to

also have substantial support for the data (Bumham and Anderson 2002). I calculated
adjusted r2 for the minimum AICc model and competing models to evaluate how well the

models explained the variation in the data. I also calculated partial correlations for all
variables in models with more than one predictor. Partial correlations were used to assess

the importance of individual independent variables after adjusting for additional variables
in the model. The AICc analysis and adjusted r2 were determined using R version 2.1.1

(R Development Core Team 2005). Partial correlations were obtained using PROC
CANCORR in SAS v. 9.1 (SAS Institute 1999). Three sites were excluded from this
analysis. In 2005, only the 7 (I count was collected at site 2 and 9, therefore BSI values
could not be calculated for these sites. In 2006, ground-veriﬁcation of the landscape data
was not collected for site 25 (Figure 3.1).

Sites that are geographically close may be more similar than sites that are farther
apart (Legendre 1993). Therefore, before interpreting the results of our AIC analysis for
either BSI or Coccinellidae abundance, I determined if the residuals from the best ﬁt or
competing models were spatially autocorrelated by examining sample variogram plots.
For both response variables I did not detect spatial autocorrelation among the 23 sites.

Results
Aphid populations in 2005 and 2006

In 2005, A. glycines anival in soybean was detected beginning in mid-June in
Michigan, Minnesota, and Wisconsin. Soybean aphid was detected about one month
later in the Iowa sites, beginning on July 7 (Table 3.1). Of the 12 sites sampled, 8

reached the economic threshold of 250 A. glycines per plant in the ﬁeld as a whole. The

61

date at which threshold was reached in the ﬁeld at large varied from mid-July to late-
August (Table 3.1). Populations in exclusion cages exceeded threshold in 11 of 12 sites
on average 11.6 days earlier than the open ﬁeld. In 2006 soybean aphid was detected
beginning in mid-June through mid-July in all sites except site 25 in Iowa where aphids
were not found until August 10. Only two of the 14 sites reached threshold in the ﬁeld at
large while in exclusion cages eight sites exceeded threshold (Table 3.1).
Predator and parasitoid diversity across the region

Coccinellidae, Syrphidae, Chrysopidae, and Anthocoridae were the most
abundant predators collected on sticky cards (Figure 3.2). Nine species of coccinellids
were observed. Across the region, H. axyridis comprised the greatest proportion of the
coccinellid population in both years (70% in 2005, 72% in 2006). Visual plant counts
illustrated that Coccinellidae and Anthocoridae were the most abundant natural enemies
found on soybean plants (Figure 3.2). The most abundant lady beetle species found on
soybean plants was H. axyridis, with adults and larvae comprising 47 and 88% of the
coccinellid community in 2005 and 2006. Anthocoridae consisted solely of adult and
nyrnphal 0. insidiosus.
Impact of predators on soybean aphid

Despite their apparent low abundance, natural enemies had a large impact on
soybean aphid, with signiﬁcantly higher aphid populations occurring on caged versus
open plant treatments. The relationship between these treatments across the two weeks of

the experiment varied as indicated by a signiﬁcant cage treatrnent*week interaction
(F 1,44=9.36, P=0.003, Figure 3.3). However, the open treatment was signiﬁcantly lower

than the exclusion cage both 7 (P=0.002) and 14 (P<0.0001) d after establishment.

62

Landscape diversity

The landscape surrounding each of our ﬁeld sites varied from agriculturally-
dominated to forest and grassland-dominated. Within the 3.5 km landscape radius
surrounding each of the 23 sites, landscape diversity values (D) ranged from 2.43—6.40.
Landscapes with low D values were dominated by corn and soybean ﬁelds while
landscapes with high D values included a diversity of crop and non-crop habitats. The
percentage of the landscape composed of annual cropland ranged from 20-91%.
Landscapes with low to high percentages of annual cropland were found within each state
(Michigan 20-91%, Wisconsin 29-71%, Minnesota 44-73%, and Iowa 27-89%). Soybean
and corn production covered between 4-3 8% and 9-57% of the 3.5 km landscape
respectively. The percentage of the landscape composed of forested habitat ranged from
1 -42%.
Interpretation of principal components

Both principal components were interpreted as measures of landscape
composition. Sites with positive loadings on PCI were correlated with forested land
cover while sites with negative loading values were correlated with corn and soybean.
Positive values of PCI suggest a landscape dominated by forested habitats while negative
values of PC] indicate a landscape dominated by com and soybean agriculture (Figure
3.4). The interpretation of PC2 is scale dependent, with the correlation of the factor
loadings consistent only at scales of 1 km and greater. Here positive scores were
correlated with the variables other crops and wheat indicating that landscapes with high
loadings had a high proportion of locally important fruit, vegetable and ornamental and

small grain crops. Negative values were correlated with the variables grassland and

63

water, indicating landscapes with a high proportion of these more permanent habitat
types. (Figure 3.4).
Model comparisons

Overall, BS1 was best predicted by the Simpson’s D at a spatial scale of 1.5 km
(Table 3.3) having the lowest AICc score of any model at the 7 spatial scales examined.
The Simpson’s D model was also the best predictor of BSI at l, 2, 2.5 and 3 km while the
PCI model had the lowest AICc scores at both 0.5 and 3.5 km. Competing models
included Diversity + PCl , Diversity + PC2, and PCI (Table 3.3). There was a signiﬁcant
positive relationship between BSI and landscape diversity, indicating that the degree of
soybean aphid suppression by natural enemies increased as diversity of crop and non-
crop habitats in the surrounding landscape increased (Figure 3.5).

Overall Coccinellidae abundance was best predicted by the PCI + PC2 model at a
spatial scale of 1.5 km (Table 3.5). This model had the lowest AICc value and contained
43% of Akaike weight. The abundance of coccinellids was positively correlated with
PCI and negatively correlated with PC2 indicating that these predators were most
abundant in soybean ﬁeld in landscapes dominated by forests and to a lesser extent,
grasslands (Figure 3.6). Competing models at 1.5 km included Diversity+PC1+PC2, and
Diversity+PC1. The Diversity and Diversity+PC2 models had the lowest AIC value at
0.5 and 1.0 km respectively while the PCI model had the lowest AICc at spatial scales of
2 km and greater.

Discussion
Biological control of pests is an important yet infrequently measured ecosystem

service (Swinton et al. 2006). In this study, I found that the ability of a landscape to

64

supply biocontrol services as manifested by suppression of the soybean aphid was a
function of the diversity of the agricultural landscape, and that coccinellids, the primary
predator of soybean aphid, responded to the presence of natural habitat within the
landscape.
Impact of biological control

Previous authors have illustrated the importance of natural enemies in the
suppression of aphid populations at the ﬁeld scale (Dennis and Wratten 1991, Grasswitz
and Burts 1995, Rutledge et a1. 2004, Schmidt et al. 2004, Muller et al. 2005, Gardiner
and Landis 2007). Schmidt et al. (2004) found the exclusion of ﬂying natural enemies
led to 94% higher populations of the bird cherry-oat aphid, Rhopalosiphum padi (L.).
Similarly, Fox et al. (2004), Costamagna and Landis (2007a), Costamagna and Landis
(2007a) Desneux (2006), and Donaldson et al. (2007) found that a diverse generalist
natural enemy complex signiﬁcantly reduced soybean aphid populations in individual
soybean ﬁelds while parasitoids are currently not major contributors (Costamagna et al.
in press). Because of its potential for rapid population growth (Costamagna et al. 2007b)
predation early in the season is key to preventing aphids ﬁ'om reaching threshold levels.
Our study is the ﬁrst to demonstrate that such phenomena are regionally important and
result from the actions of natural enemy communities that vary with landscape
composition.

Using exclusion cages, I found that communities of generalist predators are
effective at suppressing A. glycines populations across a wide portion of the north central
U.S. Suppression was found even though soybean production practices, time of ﬁrst

aphid infestation, and plant growth stage all varied widely during the study (Table 3.1 ).

65

This reduction was not always effective in maintaining populations below the economic
threshold, but without predation, aphid populations exceeded thresholds earlier in the
season and more often than in the absence of predators. Since its detection in 2000, the
soybean aphid has exhibited apparently cyclic outbreaks, with high populations in one
year typically followed by low populations the following year. These cyclic outbreaks are
likely attributable to the work of generalist predators which build populations in high
aphid years and prevent outbreak populations the following year. I conducted this study
in a high (2005) and low (2006) soybean aphid year and found that in both cases
predators reduced the proportion of sites that exceeded threshold. Within 14 days,
soybean aphid populations exceeded the threshold of 250 aphids per plant in 92% of
exclusion cages sampled in 2005 and 57% in 2006, while in the presence of predators this
dropped to 67% in 2005 and 14% in 2006, ie. the presence of predators would have
reduced the need for insecticide treatment by 25-43% in 2005 and 2006 respectively.
Remarkably, this study demonstrates that low numbers of natural enemies are required to
achieve this suppression. Through this and other studies (Costamagna and Landis 2007,
Costamagna et al. in press) I now understand that small but consistent predation early in
the season is they key to preventing aphids from reaching threshold levels.
Ability of the landscape to supply biocontrol services

While I observed a signiﬁcant impact of biological control across this 4-state
region, there was also signiﬁcant variation in the amount of aphid suppression among
sites which was strongly inﬂuenced by the diversity of the surrounding landscape.
Agricultural landscapes have changed dramatically with agricultural intensiﬁcation,

resulting in larger more contiguous ﬁelds and non-crop areas reduced in area and

66

frequency (Bianchi et al. 2006). I found that low diversity landscapes dominated by
soybean and corn production had reduced biological control of soybean aphid compared
with diverse landscapes dominated by non-crop habitats. In a 2006 review, Bianchi et al.
(2006) found that increased landscape complexity reduced pest pressure in 45% of the
studies. Effects included reductions in pest density (Ostrnan et al. 2001, den Belder et al.
2002) and reduction in crop injury (Thies and Tscharntke 1999, Thies et al. 2003). Pest
pressure increased with landscape complexity in 15% of the cases examined (Raffy and
Tran 2005, Roschewitz et al. 2005) while the two were unrelated in 40% of the studies
(Galecka 1966, Holland and Fahri g 2000, Klug 2003, Thies et al. 2005). It is likely that
some of this variation is tied to differences in the impact of landscape complexity on
speciﬁc natural enemy populations. A more detailed understanding of how natural
enemies respond to landscape structure may help us understand these outcomes.
Response of natural enemies to landscape diversity and composition

Tscharntke et al. (2005b) addressed the potential impact of agricultural
intensiﬁcation on the disruption of ecosystem services including biological control,
stating that simple landscapes may support lower abundance of natural enemies, reduced
enemy biodiversity perhaps lacking a key species, or by restricting movement of natural
enemies within the landscape. In this study I found that coccinellids were more abundant
in soybean ﬁelds in landscapes with an abundance of forests and grasslands. These
landscapes were more diverse and supported greater biocontrol of soybean aphid as
measured by the 881 compared with landscapes dominated by agricultural habitat.
Coccinellid species have been shown to be key predators of soybean aphid (Costamagna

and Landis 2007a, Gardiner and Landis 2007) as well as other pest aphids. Thus, to

67

promote biological control in agricultural landscapes, it may be desirable to maintain a
diversity of these habitats to favor these natural enemies.

The inﬂuence of landscape diversity on natural enemies has been addressed in
several systems (Bommarco 1998, Elliott et al. 1999, Thies et al. 2003). Predator
diversity and abundance typically increase with landscape diversity, proportion of non-
crop habitat, and degree of connectedness between non-crop habitat patches, while
decreasing with crop ﬁeld size. For example, Elliott et a1. (1999) found that uncultivated
land and habitat patchiness were both associated with increasing abundance and richness
in predator communities in wheat ﬁelds. Similarly, Thies et al. (2003) found that
parasitism of the rape pollen beetle was positively correlated with percent non-crop
habitat.

Landscape spatial scale

In this study, I examined the impact of landscape variables on biological control
at 7 spatial scales, with landscape radii varying from 0.5 km to 3.5 km. Analysis of
landscape features at a radius of 1.5 surrounding focal soybean ﬁelds explained the
highest proportion of the variation in both biological control of soybean aphid (BSI) and
coccinellid abundance. While I know relatively little about the movement of coccinellids
it is possible that a landscape of this size encompasses their ecological neighborhood
(Addicott et al. 1987) containing the diversity of habitats utilized by these species.
During early summer when aphid populations in croplands are low these predators are
likely to be very transient, moving between ﬁelds in search of prey. Landscapes of a
similar grain size have proved important in predicting both herbivory and natural enemy

abundance in other systems. For example, Thies et al. (2003) measured the functional

68

spatial scale at which the rape pollen beetle, Meligethes aeneus F. and its parasitoids were
affected by landscape. They tested the effects of landscape on trophic-level interactions
using simple (<3% non-crop habitat) to complex (>50% non-crop habitat) landscapes,
and analyzed them at 8 spatial scales, (concentric circles 0.5-6 km in diameter) and found
that herbivory and parasitism were most strongly correlated with percent non-crop area at
a diameter of 1.5 km. Schmidt and Tscharntke (2005) found that sheetweb spider
abundance in wheat ﬁelds was correlated with non-crop area at a slightly larger spatial
scale (landscape circles with a radius of 1-3 km).
Implications

This study is the ﬁrst to demonstrate that the biocontrol service provided by
generalist predators in north-central U.S. soybean ﬁelds is dependent on the diversity and
composition of the surrounding landscape. Biological control of A. glycines is reduced in
simple landscapes dominated by com and soybean production. Coccinellidae, an
important family of A. glycines natural enemies, were also found to decline in soybean
ﬁelds located in agriculturally—dominated landscapes. Therefore, maintaining overall
landscape diversity, and especially perennial grassland and forested habitats in the
landscape will favor suppression of soybean aphid. Moreover, these habitats may also
enhance biocontrol services in other crops as the generalist natural enemies found in
soybean ﬁelds are also important antagonists of many other herbivores. Our results
suggest that such non-crop habitats do not need to directly border agricultural ﬁelds to
promote biocontrol services, as I found that occurrence of these habitats within
approximately 1.5 km explained the highest proportion of the variation in both biological

control and coccinellid abundance.

69

Finally, understanding the impact of shifting production practices on biocontrol
services in agricultural landscapes is of critical importance. For example, in many parts
of the north-central U.S. corn and soybean were typically rotated annually. Recently corn
acreage is on the rise in response to increased demand for corn ethanol, with harvested
acres rising by 15 million acres between 2006 and 2007 (NASS 2008). Increasing
pressure to produce biofuel crops on agricultural lands could have a negative effect on
biocontrol services in these landscapes particularly if corn production increases, reducing
landscape diversity and displacing habitats associated with greater predator abundance.
This could have a negative effect on biocontrol services as the pressure to produce

biofuels on cropland increases.

70

Table 3.1. Study sites with dates of ﬁrst aphid detection, period of predator exclusion

cage study, dates when aphid threshold was reached in the open ﬁeld and exclusion

cages, and plant stages during cage study. Simpson’s D values are shown for the 1.5 km

landscape radius surrounding each soybean ﬁeld.

 

 

 

Date
Site Aphid Cage Thresholdl Threshole Plant Simpjon's
State # Year Detection Study (Field) (Exclusion) Stage D
MI 1 2005 6/29 7/6-7/20 7/13 7/20 R2-R3 5.42
MI 2 2005 6/8 7/13-7/20 7/20 7/13 R2-R3 —
MI 3 2005 6/14 6/20-7/5 8/1 7/5 V2-V5 5.09
MI 4 2005 6/14 6/20-7/5 7/18 7/5 V4-R2 6.40
WI 5 2005 6/20 7/11-7/25 — 7/25 R2-R3 3.85
WI 6 2005 6/16 6/30-7/14 — 7/l4 Rl-R2 3.27
WI 7 2005 6/17 6/17-7/1 — 6/17 V5-Rl 3.27
MN 8 2005 6/16 7/14-7/27 7/27 7/27 R2-R3 4.22
MN 9 2005 6/30 7/21-7/28 7/28 — R2-R3 —
IA 10 2005 7/ 7 7/11-7/25 8/29 7/25 R2-R4 3.24
IA 1 1 2005 7/13 7/13-7/28 — 7/28 R2-R3 3.51
IA 12 2005 7/19 7/26-8/10 8/16 8/10 R3-R5 2.31
MI 13 2006 6/14 7/6-7/19 -— 7/19 R1-R2 6.06
MI 14 2006 6/19 7/31-8/14 — 8/14 R2-R3 5.78
MI 15 2006 7/13 7/19-8/2 — 8/2 R3 6.53
MI 16 2006 6/19 8/7-8/21 — 8/21 R3 5.20
WI 17 2006 6/29 7/20—8/3 — — R3-R5 7.06
WI 18 2006 6/9 7/14-7/28 — — V7-R2 2.42
WI 19 2006 7/ 7 7/21-8/4 — 8/4 R2-R4 3 .89
MN 20 2006 6/5 7/17-7/31 -— 7/31 R2-R5 2.98
MN 21 2006 7/5 7/17-7/31 7/25 7/31 R2-R3 3.41
IA 22 2006 6/15 7/25-8/8 — — R4-R5 3.1 1
IA 23 2006 7/19 7/26-8/9 — — R4-R5 3 .61
IA 24 2006 7/10 7/24-8/7 — 8/ 7 R4-R5 2.32
IA 25 2006 8/10 8/10-8/24 8/30 — R4-R6 —
IA 26 2006 7/14 8/11-8/25 - — R5-R6 2.25

 

I Date when overall ﬁeld exceeded the economic threshold of 250 aphids per plant.

#WN

71

Date when plants in exclusion cages exceeded the economic threshold.

Based on Higley and Boethel (1994)
. 2 . . . . .

Simpson’s D for 1.5 km radius, D = l/X(pr) where p1=proportron habrtat 1n the rtlr category

Table 3.2. Land cover categories used in the calculation of Simpson’s Diversity Index

(D) and as variables included in the principal components analysis.

Simpson's Diversity (D) PCA

Cropland Categories

Corn Corn
Soybean Soybean
Wheat Wheat

Sugar Beet Other Cropsl
Dry Bean

Tree Fruit
Sunﬂower
Cabbage

Canola

Potato

Carrot

Squash

Pea

Strawberry

Tilled Field
Ornamental Crops

 

Non-Crop Categories

 

Forage Grassland2
Old Field and Restored

Prairie

Pastureland

Forested Forest
Wetlands and Open Water Water
Urban Urban

 

1 Other Crops contains all Simpson's D cropland
categories except corn, soybean, and wheat.

2 Grassland contains the Simpson's D non-crop
categories Forage, Old Field and Restored Prairie,
and Pastureland

72

.36 V k 02863 .1 .36 v m :0 unwowammm we? Baum—00> a 0.88:05 A. N
bemoan”. 0950-93— ﬁm 05 E «833 we aoEomoaﬂE van NAEVN: M Q Eon? bmﬂoZQ Macmagm H D
_

 

 

 

00.0 00.0 00.0 00.00 0 00.0- +6.30 + 000 E 0.0
00.0 n 60 .000 n o 00.0 00.0 0: 00.: 0 00.0- 6000 + 0.000 + 0m E 0.0
000 0.0.0 00.0 00.2 0 00.0- 0.6.30 + 0m 8. 0.0

000 00.0 00.0 00.00 0 00.0- +030 + 000 :2 0.0

00.0 00.0 00.0 00.: 0 00.7 $030 + 0m E. 0.0

0_.0u0om.00.0uo 00.0 2.0 a: 002 0 02- 0000m++a£+ 00 803.0
00.0 00.0 00.0 00.: 0 03- tam + 0m 8. 0.0

00.0 00.0 00.0 00.0 0 00.0- £0.00 + 0m .5. 0.0

00.0 00.0 00.0 00.: 0 3.. T 10:00 + 0m an 0._

000150.20 no :0 00.0 00: 00.2 0 00.0- 6000+o£+0m 5 0.0
2.0 00.0 00.0 00.: 0 00.0- 060000 + 0m E 0.0

mnouﬂoboo Etna H _3 ._< 002 _M voo:m_0ﬁ_ Eco: A55
03000.03 -03 0 _ Seam

 

.06on wcuomﬁoo me 82:05 000 002 no N we 6 0 5:» 2258 2:0 6on =Sc>o 0005 05 088%”: Eon
.3on 00?. 8:00:58 05 2 280 300% 5000 «0 020: 3on 000m 23. .80. .5580 $5508 83 Cmmv 089: 083.50. #800585

a 0.0 8028—00 $2805 3030: wmﬁcoaow 03 005.03% «do seaweed wcwmsﬁtwo 8m 85:80 cocoa—cm 6on mo Egm .m.m 030.“.

73

 

00.0- n 000 .000 u 60 :0 000 000 00.00 0 :00- 000000 + 06000 + 0m 8. 0.0

:0 00.0 00.0 00.00 0 00.00- 050000 + 000 8. 0.0

00.0 n 60 .000- u a _00 0_ .0 00.0 00.00 0 00.00- 08000 + 5:00 + 0m an 0.0

00. - u 000 .000 u 50 000 _00 00._ 00.00 0 00.00- . 00000000 #05000 + 000 as. 0.0

_ 00.0 00.0 00.0 00.00 0 00.00- 060000 + 0m an 0.0

00.0- u 000 .000 u 50 .000- u o 00.0 000 00.0 00.00 0 00.00- 00000001 005000 + 5000 + 0m E 0.0
00.0. u 000 .000 n 60 00.0 000 00.0 00.00 0 00.:. 00.0000 + 008.10 + 0m 5. 0.0

00.0 n 50 .000- u a 00.0 000 00.0 00.00 0 0000- 006000 18000 + 0m 5 0.0

00.0 00.0 00.0 00.00 0 0.00- 006000 + 000 an 0.0

00.0 n 80 00.0- n Q 00.0 000 00.0 00.00 0 00.00- 006000 + 08000 + 0m an 0.0

00.0- n 000 .000- u 60 .000- u a 00.0 _00 00.0 00.00 0 00.00- 000.0000 + 05000 + 5000 + 0m 8. 0._
. 000- n 00.— .000 u 8.0 000 000 00.0 00.00 0 0000- 00005000 + 005.0000 + 0m .5. 3
00.0- n 000 .000 u 60 00.0 000 00.0 00.00 0 _000- 009000 + 006.10 + 0m E 0._ .
00.0- u 000 .000 n o 00.0 000 00.0 0000 0 00.00- 000000000 +0.00% + 0m E. 0.0

00 :0 000 00.00 0 00.00- 60:00 + 00 no. 0.0

000 u 60 .000 u a 00.0 00.0 00.0 00.00 0 00.00- 6.000 + 9000 + 0m :2 0.0

0_ .0 00.0 00.0 _000 0 00.00- 000000 + 000 an 0.0

08020-80 350 00 8000.03. 03 _0 82 a0 eon-cwwhé 0 0 ass: 050 02000

 

 

63022: 80 002 00 N mo 6 0 503 £358 $82580 .6008 =Sc>o 0.03 05 08865 Eon 40608 00?. 8:80:08 05 00 2000

300% some 00 300: $88 00.0w 2E. 6002:3880 mo 8900550 05 $50296 5.0 0300.80 080.0200 388 we 98805 .0.m 0300.

74

856550 cavacﬁoooo at? 0030:2038 gamma: 0 33:05 00005589 3 0050.00.00/ 0

 

 

 

8d v m 038005 .00 .36 v m :0 Sendai 003 030.05» a 08885.0 N
Domes". 0050690— 50 on... 5 033.00— .«o newton—Dana can GEN: n D 80:3 30.0.0205 0.0509005 u OF
26 00,— 2.00. N 00.00- . 0m 80— 0.0
00.0- n 000 .000 n 60 £0 00 0: 00.00 0 00.00- 000.0000 + 5.0000 + 000 as. 0.0
26 0N.o 00.0 8.00. m 0300- 50100 + 0% Ex 0.0
macaw—2.80 30.0-0m .0 _3 5 002 TM 33:85 332 Anna
0 -03 0 0 _ 0.2000
0300.03. .

00800080 0.0 030.0

75

Figure 3.1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.1. Location of 26 soybean ﬁelds sampled in 2005 (ﬁelds 1-10) and 2006 (ﬁelds
11-26). The predator exclusion cage study was completed in all sites. Fields 2, 9, and 25
were excluded from the AICc analysis of BSI and Coccinellidae abundance due to

missing data.

76

Figure 3.2.

Sticky Card Visual Plant Observation

Araneae

    
   

Braconidae
I 2005

I 2006

00..“ _0-0 -‘

Aphelinidae

 

Cecidomyiida-

 

 

 

Syrphidae
C so idae ’
hry p Hemerobiidae
Nabidae Chrysopidae '.
, ; Nabidae '
Anthocorrdae . . .
_ . Anthocoridae ‘ W ' 1‘ .-
Coccinellidae ‘7 h l C o c cinelli dae'! 1:-
0 1 2 3 0 0.1 0.2 0.3 0.4
Mean per trap i SEM Mean per plant :1: SEM

Figure 3.2. Mean number of natural enemies d: SEM counted on sticky card traps (7 d
cumulative) and during visual plant observations (illustrates natural enemies present on a

plant at a single point in time) in 2005 and 2006. Syrphidae were not counted on sticky

cards in 2005.

77

Figure 3.3.

1200

 

  
  
 

1000 ‘ - Exclusion

2 [:23 Open
5%
\I 800 a
+
E
a.
s 600 "
a.
8
E
4% 400 r
00
‘c'

200 a

  

 

 

 

: ' 1 1
14,-? 3' uni.
L “ 7 ‘ ‘1 ll

- gt. “‘1

Day 7

 

Day 14

Figure 3.3. Mean number of A. glycines per plant (:t SEM) in the cage and open plant
treatments. A signiﬁcant week*treatment interaction was found (F 1,44=9.36, P=0.003).

Comparison of least squares means at 7 (P=0.002) and 14 d (P<0.0001) indicates that

aphid populations were signiﬁcantly higher in the absence of natural enemies.

78

Figure 3.4.

 

 

 

 

0.8
Other Crops
0.6 1 c
O
N 0.4 _ Wheat
’5
D
8
a. 0.2 1
8 ’ Soybean . Forest.
"5 0.0 .
'3 Urban
.g 0 C
an -0.2 0 or“
Water
'0.4 d .
o Grassland
”0.6 I T l I m
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

Principal Component 1

Figure 3.4. PCA ordination for principal components 1 and 2 of landscape elements

surrounding soybean ﬁelds sampled at a radius of 1.5 km.

79

Figure 3.5.

 

 

 

1.0
a 0.8
><
0)
'8
3 0.6
.§
§ 0.4
O
35': 0
g 0 —
m .
0.0 l I l l I J l
1 2 3 4 5 6 7 8

Simpson’s Diversity (D)

Figure 3.5. Relationship between the biocontrol services index (BSI) and PCI at a radius
of 1.5 km for 23 ﬁelds sampled for soybean aphid suppression, 2005-06 (y = 0.124x +
0.159). BSI was calculated as the relative difference in aphid number between open and

exclusion cage treatments (see methods section 2.6).

80

Figure 3.6.

Partial Residuals

 

 

Principal Component 1

 

 

Principal Component 2

Figure 3.6. Coccinellidae abundance in 23 soybean ﬁelds sampled for soybean aphid

suppression was best explained by a model containing both PC] and PC2 at a radius of

1.5 km. (y = 0.392x1 — 0.36lx2 + 2.0638). Partial residual plots illustrate a positive

correlation with PC] and a negative correlation with PC2, indicating that landscapes with

an abundance of forests or grasslands support larger lady beetle populations in soybean

ﬁelds compared with agriculturally-dominated landscapes.

81

CHAPTER 4

LANDSCAPE COMPOSITION INFLUENCES THE ACTIVITY DENSITY OF

CARABIDAE AND ARANAE COMMUNITES IN SOYBEAN FIELDS.

Abstract

The distribution and abundance of semi-natural and agricultural habitats within
the landscape is likely to greatly inﬂuence landscape-scale metapopulation dynamics of
habitat generalists. I investigated the inﬂuence of landscape diversity and composition
on the activity density of Carabidae, Araneae, and Opiliones. Although I hypothesized
each taxa would increase with landscape diversity and the amount of semi-natural habitat,
I found variation in the response of the study organisms. The activity density of Araneae
supported my hypothesis, increasing in soybean ﬁelds in landscapes with an abundance
of forests and grasslands. Relationships with Carabidae activity density varied, with
adult overwintering and predatory species increasing and larval overwintering species
declining with the abundance of grassland in the landscape. Opiliones activity density
was not related to the structure of the surrounding landscape at any of the spatial scales
examined. The spatial scale at which each taxa responded to landscape composition
increased with their capacity for dispersal. Carabidae, which generally disperse by
walking, responded to smaller spatial scales than linyphiid spiders, which frequently
balloon. These data illustrate the importance of understanding species-speciﬁc

relationships to landscape variables.

82

Introduction

Agricultural landscapes support a diverse metacommunity of ground-dwelling
arthropod natural enemies, frequently dominated by Carabidae and Araneae. Many of
these natural enemies are habitat generalists and spatial processes play an important role
in their metacommunity dynamics and inﬂuence interactions between multiple
populations linked via dispersal (Liebold et al. 2004, Schmidt et al. 2008). Whereas
habitat specialists may exist in patches that effectively function as islands, habitat
generalists utilize a mosaic of crop and semi-natural habitats (Schmidt et al. 2008).
Wissinger (1997) suggested that natural enemies that immigrate to agricultural ﬁelds
colonize crops under favorable conditions and retreat to semi-natural habitats when
conditions become hostile. Given this dispersal between habitats within a
metacommunity, the diversity and abundance of these natural enemies available for
biological control in croplands is likely to depend on the large-scale structure of the
landscape matrix.

Several patterns have emerged from the study of landscape inﬂuences on natural
enemy populations. In many cases the diversity, abundance, reproductive success, and
biocontrol service provided by natural enemies increases as landscape diversity and the
proportion of semi-natural habitat increases (Colunga-Garcia et al. 1997, Fahrig and
J onsen 1998, Elliott et al. 1998, Bommarco 1998), whereas pest number and damage
decline (den Belder et al. 2002, Ostman 2002, Thies et al. 2003). For example, spider
species richness and the abundance of linyphiid spiders in croplands increased with the
proportion of non-crop habitat in the landscape (Schmidt and Tscharntke 2005, Schimidt

et al. 2005). Bommarco and Ekbom (2000) report that the fecundity of predatory

83

carabids beetles increased with landscape heterogeneity and with the proportion of
perennial crops in the landscape due to higher arthropod abundance and diversity. In
addition, Gardiner et al. (2008) found that the biocontrol service supplied to soybean
ﬁelds and the abundance of coccinellid predators were positively related to the diversity
of the surrounding landscape.

In addition, thresholds in landscape structure may exist below which the search
efﬁciency and the ability of natural enemies to aggregate and control pests is diminished
(With and King 1999, Wiens et al. 1997, With et al. 1999, Thies and Tscharntke 1999).
Simulation and ﬁeld experiments have shown that biological control by natural enemies
declined when suitable habitat fell below 20% (With and King 1999, Thies and
Tscharntke 1999). For example, Thies et al. (2003) found that in these simple
landscapes, parasitism of rape pollen beetle declined below threshold levels need for
successful biological control.

Finally, landscape complexity does not inﬂuence all species equally, or at the
same scale (Thies et al. 2003, Den Belder et al. 2002, Steffan-Dewenter et al. 2002,
Schmidt et al. 2005). Gardiner et al. (in review) found that native and exotic coccinellid
populations in soybean responded differently to landscape composition. Native species
were more abundant in landscapes with an abundance of grassland while exotic species
were most abundant in forest dominated landscapes. Scale of analysis is also critical in
interpreting landscape patterns (Thies et al. 2003, Gardiner et al. 2008). Species that are
able to ﬂy long distances may survive in a landscape with a smaller number of highly
dispersed fragments of suitable habitat, whereas species with a more limited dispersal

capacity would not. Therefore, the arrangement and size of suitable habitat patches

84

within a landscape are likely to have species-speciﬁc impacts on the activity of habitat
generalists in agricultural croplands.

Across much of the north-central U.S., corn and soybean ﬁelds form the
landscape matrix, which is deﬁned as the most extensive and connected habitat present
(Turner et al. 2001). This matrix contains patches of semi-natural habitats including
woodlots, grasslands, wetlands and open water. Several studies have related the diversity
and abundance of ground beetles and arable spiders in agricultural ﬁelds to the
composition of the surrounding landscape. However, the majority of these studies have
focused on wheat and were conducted in Europe where the matrix structure is much
different from that of intensive agricultural areas of the US.

My objective was to examine how the activity density of habitat generalist
predators is inﬂuenced by the amount of semi-natural habitat patches within an
agricultural matrix in the north-central U.S. Here I examine three taxa: Carabidae,
Araneae, and Opiliones in soybean as part of a larger study on the inﬂuence of landscape
structure on the suppression of soybean aphid by its natural enemies. Due to the intense
disturbance regime present in annual cropping systems, my hypotheses were that: (l) the
activity density of Carabidae, Araneae, and Opiliones would increase with the proportion
of semi-natural habitats in the landscape, (2) the trophic level and overwintering
phenology of a species would affect its response to landscape composition, with
predatory species that overwinter as adults having stronger correlations with semi-natural
habitat, and (3) the spatial scale at which speciﬁc organisms respond to landscape
structure, is positively related to its dispersal capacity. Therefore, those that disperse

mainly by walking would have a smaller ecological neighborhood compared with groups

85

known to frequently utilize ﬂight or ballooning. I relate the patterns found herein to
studies conducted in Europe and in other US. cropping systems.
Methods

Field sites

Between June and August of 2005 and 2006, I examined the activity density of
Carabidae, Opiliones and Araneae in 26 soybean ﬁelds in Iowa, Michigan, Minnesota,
and Wisconsin (Figure 4.1). Sites were selected to study how landscape variation
affected the diversity and abundance of these arthropods, as part of a USDA Risk
Avoidance and Mitigation (RAMP) funded multi-state study of soybean aphid
management. Each year, a minimum distance of 30 km separated each ﬁeld site. Across
years, I maintained the relative regional distribution of our sites, but sampled different
soybean ﬁelds. The average distance between soybean ﬁelds sampled in 2005 and 2006
within a given region was 2.4 km (range = 0.2-14.6 km, Figure 4.1). Field size averaged
16.0 ha (range = 13.6-48.1 ha). In each RAMP ﬁeld (22 sites), a randomized complete
block design with 4-6 blocks and either 4 (2005) or 5 (2006) treatments was established
in 0.4 ha (2005) or 0.2 ha (2006) plots. Multiple experiments occurred within these
ﬁelds; with this study conducted in control plots not treated with insecticide. An
additional 4 commercial production ﬁelds (2 in 2005 and 2 in 2006) were also included in
this study, each containing four 0.4 ha (2005) or 0.2 ha (2006) untreated plots.
Pitfall sampling

Carabidae, Araneae, and Opiliones were collected from each plot using pitfall
traps. One pitfall trap was placed within a soybean row in the center of each plot. The

trap consisted of a 1000 ml plastic cup with a l 1.5 cm diameter opening. Cups were

86

buried at ground level and ﬁlled with approximately 500 ml of a 50% propylene glycol
solution. Each trap was covered with a rain shield constructed from a 0.9 m2 section of

corrugated plastic sheeting and four 0.46 m wire legs. Traps were checked weekly and
all Carabidae, Araneae, and Opiliones were collected and identiﬁed. All Carabidae and
Lycosidae were identiﬁed to species (Lindroth 1968, Kaston 1948). The remaining
Araneae were identiﬁed to family (Ubick 2007). Opiliones were identiﬁed to order.
Landscape analysis

Field geospatial data were collected using a handheld GPS receiver using Wide
Area Augmentation System (WAAS) correction. The spatial coordinate for the center of
each site was used to obtain ortho—rectiﬁed digital aerial imagery. I digitized the habitats
surrounding each study site to a radius of 3.5 km using ARC GIS 9.1. Aerial images
were captured between 1998 and 2006. Land use changes between the image date and
study period were recorded by ground-veriﬁcation in June — August annually, with
corrections made during the digitization process. At the same time, I also determined the
speciﬁc land cover (n = 21 categories) present in all areas of each landscape (Table 4.1).
Some locations included polygons that were not visible from a roadway and permission
to access private lands could not always be obtained. These polygons were given a value
of zero and were excluded from further analysis. The area of each site that could not be

identiﬁed varied from O-2%. The smallest polygons identiﬁed included ﬁeld plots on
university research farms and small patches of fallow ﬁeld (< 5 m2); the largest were
contiguous urban areas, lakes, and forests (< 11.9 kmz).

Landscape diversity was measured using Simpson's Index (D) (Simpson 1949).

Simpson’s Index is typically used to examine the variance of species abundance

87

distributions. Here I applied it to examine variance in the proportion of area covered by
each of 21 land use categories. This index was calculated for each of the landscapes at

each of the spatial scales examined. The equation for Simpson’s Index (D) is: D =
1/Z(pi)2 where pi=proportion of habitat in the ith land-cover category (D increases as

diversity increases). Using methods modeled after Thies et al. (2003) I measured
landscape diversity at 6 spatial scales ranging from 1 km to 3.5 km radii (at 0.5 km
intervals) from the ﬁeld center.

Statistical analysis

To evaluate the relationship between activity density and landscape variables, I
performed a principal components analysis (PCA) on the landscape variables in order to
reduce the dimensions of the data. The 21 observed land cover categories were combined
into seven variables included in the PCA: four crop variables (proportion corn, soybean,
wheat and other crops), and four non-crop variables (proportion forest, grasslands,
wetlands and water, and urban) (Table 4.1). Principal component axes were extracted
using correlations among variables and the resulting factors were not rotated (Mche
and Mefford 1999). I restricted the analysis to the ﬁrst two eigenvectors which explained
between 53.2-57.l% of the variability in landscape data. This was done for each spatial
scale (1.0-3.5 km radii).

To assess the inﬂuence of interpreted principal components and Simpson’s D on
the activity density of Carabidae, Opiliones and Araneae I compared all possible models
containing these variables using Akaike’s Information Criterion, adjusted for a small
sample size (AICc) (Bumham and Anderson 2002). At each spatial scale 7 models were

considered (Table 4.2). For each analysis, I present the maximum log-likelihood

88

estimate, the Akaike weights, which estimate the relative likelihood of a given model
against all other models, and AICc differences (A,). At each spatial scale I present the

model with the minimum AICc value, i.e. with the best support for the data, and any

competing models with a AICc difference of less than two (Ribic and Sample 2001,
Bumham and Anderson 2002). I calculated an adjusted r2 for the minimum AICc model

and competing models to evaluate how well the models explained the variation in the
data. I deﬁne the best overall model as the model with the lowest AICc score across all
spatial scales. I calculated partial correlations for all variables in models with more than
one predictor. Partial correlations were used to assess the importance of individual
independent variables after adjusting for additional variables in the model. The AICc

analysis and adjusted r2 were determined using R version 2.1.1 (R Development Core

Team 2005). Partial correlations were obtained using PROC CANCORR in SAS v. 9.1
(SAS Institute 1999). All Carabidae variables and Opiliones were log (x+1) transformed,
and the mean number of Araneae and Lycosidae were square root (x+0.5) transformed
prior to analysis to meet the assumptions of normality and homogeneity of variances
(SAS Institute, 1999).

Although no site had an overlapping 3.5 km landscape buffer within a given year,
sites did have overlapping buffers across years. Therefore, prior to interpreting the
results of the AICc analysis, I examined potential spatial autocorrelation of the residuals
of the best ﬁt and competing models for each response variable between each site and the
nearest neighboring site using Moran’s I statistic. I did not ﬁnd evidence of spatial
autocorrelation between neighboring sites for any of the best ﬁt models for any response

variable.

89

Results
Activity density in soybean ﬁelds

A total of 29,300 ground beetle individuals comprising 62 species were collected
from 26 soybean ﬁelds in 2005 or 2006 across the 4-state study region (Table 4.3). The
majority of these species were predators or omnivores (91.6%) that overwintered as
larvae (56%) (Table 4.3). Overall, the most abundant species was Cyclotrachelus sodalis
(LeConte), which was most abundant in soybean during late June to early July followed
by Poecilus chalcites (Say) which was most abundant in from mid June to mid July
(Figure 4.2). Other abundant species included (in order of abudance): Pterostichus
permundus (Say), Hatpalus pensylvanicus (DeGeer), Harpalus eraticus Say, Poecilus
lucublandus (Say), Scarites quadriceps Chaudior and Pterostichus melanarius (Illiger)
(Table 4.4 and Figure 4.2). The proportional distribution of species varied across the 4
states sampled. Whereas C. sodalis was the most abundant species in Iowa (51.1% of
total) and Minnesota (58.5%), P. calcites (61.4%) and H. erraticus (46.4%) were the
most abundant species collected in Michigan and Wisconsin respectively. Among all
taxa samped these was considerable variation in abundance from year to year (Figure
4.2).

A total of 3,21 5 Araneae and 9,058 Opiliones were collected in pitfall traps. In
both 2005 and 2006 Araneae activity density peaked in late June. Eight families were
found; Lycosidae was the most abundant followed by Linyphiidae (Table 4.5, Figure
4.3). Nine species of Lycosidae were collected, the most abundant were Trochosa
terricola Thorell and Pardosa saxatilis (Hentz) (Table 4.5, Figure 4.4).

Landscape diversity

9O

The landscape surrounding each of our ﬁeld sites varied from agriculturally-
dominated, to forest and grassland-dominated. Within the 3.5 km landscape radius
surrounding each of the 26 sites, landscape diversity values (D) ranged from 2.43-6.40.
The percentage of the landscape composed of annual cropland ranged from l9-91%.
Landscapes with high and low percentages of annual cropland were sampled in each state
(Michigan 19- 91%, Wisconsin 29-7l%, Minnesota 44-84%, and Iowa 27-89%). At a 3.5
km radius grassland habitat comprised 1-69% (Michigan 3-19%, Wisconsin 4-20%,
Minnesota 6-19%, and Iowa 5-69%) while forested habitat comprised 042% of the
landscape surrounding ﬁeld sites (Michigan 3-32%, Wisconsin 3-42%, Minnesota <1 -
16%, and Iowa 0-6%)

Principal component analysis of landscape variables

Both of the principal components interpreted in this study were measures of
landscape composition (Figure 4.5). Sites with positive loadings on PCI were correlated
with the variable Forest while negatives loadings on PCI were correlated with the
variables Corn and Soybean. Therefore sites with positive values of PCI suggest a
landscape with an abundance of wooded habitat while sites with negative values of PC]
indicate a landscape dominated by com and soybean agriculture (Figure 4.5). For PC2,
sites with positive loadings were correlated with the variable Grassland and negative
loadings were correlated with the variables Other Crops and Wheat. Sites with high
values of PC2 had a high proportion of pastures, old ﬁelds, and restored prairies. Sites
with negative values had a high proportion of locally important ﬁ'uit, vegetable,

ornamental and small grain crops (Figure 4.5). Both principal components indicate the

91

intensity of landscape disturbance, with high values indicating less disturbed grassland
and forested habitats and low values indicating greater agricultural disturbance.
Relationships between activity density and landscape variables

The activity density of Carabidae was best predicted by the PC2 model at all 6
spatial scales examined; the PC2 model at a 1 km radius had the lowest AIC value across
scales (Table 4.6). The PC] + PC2 and intercept only were competing models. A
marginally signiﬁcant negative correlation was detected between Carabidae and PC2 (P =

0.04, r2=0.13), indicating that activity density was reduced in soybean ﬁelds surrounded

by an abundance of grasslands compared to landscapes dominated by regionally
important crops and wheat (Figure 4.6).
Adult-overwintering species

When ground beetles were analyzed by overwintering stage, differences were
found in their response to landscape variables. The activity density of ground beetles
overwintering as adults was best predicted by the PC2 model at all spatial scales except a
radius of 2.5 km where the full model PC1+PC2+Diversity had the lowest AIC value
(Table 4.7). The PC2 model at a radius of 1 km had the lowest AIC value across scales.

Competing models included PC2, PCl + PC2, PC2 + D, and PC] + PC2 + D. Adult-
overwintering beetles were signiﬁcantly negatively correlated with PC2 (P < 0.0001, r2 =

0.57), indicating that their activity density in soybean ﬁelds was greater in landscapes
with an abundance of wheat and regionally important crops compared with landscapes
dominated by grassland (Figure 4.6). The most abundant adult-overwintering species

collected was P. calcites. Like adult—overwintering beetles as a whole, the activity

92

density of P. calcites in soybean declined with the abundance of grassland in the
landscape (P = 0.02, r2 = 0.21) (Figure 4.6).
Larval-overwintering species

The activity density of ground beetles overwintering as larvae was also best
predicted by the PC2 model at all spatial scales; the PC2 model at a radius of 2.5 km had
the lowest overall AlCc value (Table 4.8). Unlike adult-overwintering beetles, the

activity density of larval overwintering species was signiﬁcantly positively correlated
with PC2 (P = 0.01 , r2 = 0.23). The activity density of these species increased in soybean

ﬁelds in landscapes surrounded by an abundance of grassland compared with landscapes
with signiﬁcant amounts of wheat and regional crops (Figure 4.6). A similar pattern was
found for the most abundant species of larval overwintering ground beetle, C. sodalis (P
= 0.01, r2 = 0.23) (Figure 4.6).
Predatory and herbivorous species

The activity density of predatory ground beetle species was best predicted by the
PC2 model at all spatial scales; the PC2 model at a radius of 1 km had the lowest overall
AICc value. Competing models included PC1+D, D, PC2+D and the intercept only
model (Table 4.9). The activity density of predatory species was signiﬁcantly negatively
correlated with PC2 (P = 0.04, r2 = 0.17), indicating that their activity density was greater
in soybean ﬁelds in landscapes with an abundance of regional crops and wheat compared
with grassland dominated landscapes (Figure 4.6). Herbivores were best predicted by the

intercept model at all spatial scales, competing models included PCI and PC2.

93

Araneae and landscape variables

For overall spider activity density, the PC] + PC2 model had the lowest AIC
value at all 6 scales examined. The PCl + PC2 model at a radius of 1.5 km had the
lowest AIC value across scales. Competing models included PC], PC] + D, and PC2 +
D (Table 4.10). Spider activity density increased with both PC] and PC2. This
relationship was signiﬁcant for PCI, indicating increased activity density in soybean
ﬁelds in landscapes surrounded by an abundance of forested habitat (P = 0.01, model
8:035) (Figure 4.7).

When the two most abundant families of spiders, Lycosidae and Linyphiidae were
examined, Linyphiidae had a stronger correlation with landscape variables. The activity
density of Lycosidae was best predicted the PC2 model which had the lowest AICc value
at all spatial scales; the PC2 model at a radius of 3 km had the lowest AICc value overall
(Table 4.11). There was a positive, non-signiﬁcant correlation between PC] and the
activity density of Lycosidae (P = 0.06, r2=0.14) (Figure 4.7). Competing models
included PC2 + D, PC] + PC2, PCl, D, PC2, and the intercept only model. For
Linyphiidae activity density, the PCI model had the lowest AICc value at all spatial
scales, the PCI model at a radius of 3.5 km had the lowest AICc value overall. No
competing models were found for Linyphiidae (Table 4.12). There was a signiﬁcant
positive correlation between the Linyphiidae and PCI (P = 0.0002, r2=0.44), indicating
greater activity density in soybean ﬁelds surrounded by landscapes with an abundance of
forested habitat compared with landscapes dominated by com and soybean ﬁelds (Figure

4.7).

94

Discussion

A diverse community of Carabidae and Araneae were detected in soybean ﬁelds
across Iowa, Michigan, Minnesota and Wisconsin, with 62 species of Carabidae and 8
families of Araneae (9 species of Lycosidae) collected in pitfall traps from 26 sites in
2005 and 2006. I examined three hypotheses related to the response of these ground-
dwelling arthropods to the surrounding landscape. My ﬁrst hypothesis was that the
activity density of Carabidae, Araneae and Opiliones in soybean would increase with the
proportion of semi-natural habitats in the landscape. Rather than a general increase, I
found variation in the response of these taxa to landscape composition. The activity
density of Araneae supported our hypothesis, increasing in soybean ﬁelds in landscapes
with an abundance of forests and grasslands. When Carabidae activity density was
examined I found that the relationship to landscape varied, with adult overwintering and
predatory species increasing and larval overwintering species declining with the
abundance of grassland in the landscape. Opiliones activity density was not related to
the structure of the surrounding landscape at any of the spatial scales examined.
Relationships between Carabidae phenology, trophic level, and landscape composition

Our data did not support our second hypothesis that predatory Carabidae species
that overwinter as adults would be positively correlated with the abundance of semi-
natural habitat in the landscape. Instead, predatory species and those that overwinter as
adults had greater activity density in soybean ﬁelds in agriculturally dominated
landscapes. Carabids that overwintered as larvae had greater activity density in soybean
ﬁelds in landscapes with an abundance of grassland patches. Schmidt et al. (2008) state

that although many studies illustrate an increase in the abundance of habitat generalists

95

with the proportion of semi-natural habitat in the landscape, the metapopulations
dynamics of habitat generalists are often complex. The response of a particular species to
a different habitats in the surrounding landscape may be positive or negative depending
on its success in the focal habitat of study. Schmidt et al. (2008) postulate that species
well adapted to the focal habitat, in this case a soybean ﬁeld, will respond negatively to
different habitats in the surrounding landscape. Alternatively, species that are better
adapted to habitats other than the focal habitat will increase with the amount of this
different habitat in the landscape matrix.

This suggests that Carabidae, particularly predators that overwinter as adults are
more successﬁrl in agricultural systems. Although the majority of the abundant species
collected are considered habitat generalists, the most abundant predatory, adult
overwintering species found, P. chalcites, is thought of as an agricultural specialist
(Larsen et al. 2003) and thus is likely to favor agriculturally dominated landscapes. The
most abundant species of larval overwintering carabid C. sodalis is known to be well
adapted to perennial-dominated habitats including grasslands (Larsen et al. 2003). Thus,
it follows that this species increases in abundance with the presence of grassland patches
in the landscape. Overwintering as an adult gives an advantage of high mobility in
search of overwintering sites. These species would be able to disperse from agricultural
ﬁelds in the fall when conditions deteriorate and recolonize small patches of semi-natural
habitat for overwintering whereas less mobile larvae would be less likely to successfully
locate them.

I found no evidence of a relationship between landscape composition and the

activity density of herbivorous carabids. Overall, herbivores constituted a low percentage

96

of total individuals (8%), the vast majority (94% of herbivores) of which were a single
species H. pensilvanicus. Other studies have found greater herbivore diversity and seed
predation in croplands surrounded by diverse landscapes (Purtauf et al, Menalled et al.
2000). For example, Purtauf et al. (2005) found that the diversity of herbivores in wheat
ﬁelds in central Germany increased with the proportion of non-crop habitat.
Araneae and landscape composition

Araneae had greater activity density in soybean ﬁelds in landscapes with an
abundance of grassland and forested patches. Semi-natural habitat patches also
inﬂuenced the activity density of the two most numerous Araneae families collected;
Lycosidae and Linyphiidae. The activity density of Lycosidae exhibited a weak, positive
correlation with PC2, indicating the abundance of lycosids was greater in soybean ﬁelds
in landscapes with an abundance of grassland patches. Linyphiidae activity density
responded strongly to PCI , increasing in soybean ﬁelds in landscapes with an abudance
of forested patches. Forested habitat was also an important variable in explaining spider
species richness and species composition in oilseed rape ﬁelds (Drapela et al. 2007).
Overall landscape heterogeneity has also been correlated with spider richness and
abundance in wheat ﬁelds (Schmidt and Tscharntke 2005, Schmidt et al. 2005, Clough et
al. 2005). A stronger response to landscape by Linyphiidae compared with Lycosidae
has been documented in other systems. Oberg et al. (2007) found that the activity density
of Lycosidae in cereal ﬁelds was not related to landscape variables whereas the activity
density of Linyphiidae was positively correlated to perennial crops in the surrounding

landscape.

97

Spatial scale and the dispersal of habitat generalists

My ﬁnal hypothesis was that the spatial scale at which an organism responds to
landscape would be a function of its capacity for dispersal, thus those that disperse
mainly by walking would respond to a smaller spatial scale compared with groups known
to frequently utilize ﬂight or ballooning. Carabidae and adult Lycosidae frequently
colonized croplands on the ground (Drapela et al. 2007) thus I expected that these groups
would respond to a smaller landscape scale compared with the Liniphiidae which
frequently balloon throughout their life cycle (Bell et al. 2005). I found support for this
hypothesis with walking dispersers responding at smaller spatial scales compared with
ballooning species.

The negative response in activity density of all Carabidae, adult-overwintering
Carabidae, and predatory Carabidae to grassland abundance was strongest at spatial radi
of 1-1 .5 km. Carabidae overwintering as larvae and C. sodalis responded positively to
grassland abundance at a larger spatial scale of 2.5 km. Among spider variables, the
response of Lycosidae to the abundance of forested habitats was maximized at 3 km,
however AlCc values of the best candidate model at each scale were very similar. The
response of Linyphiidae increased with spatial scale. Since the positive correlation
between Linyphiidae and PC] was strongest at 3.5 km, our largest spatial scale measured
it is possible that these species respond at even larger landscape scales.

Implications

How arthropods respond to landscape composition is a function of habitat use and

dispersal ability. The distribution and abundance of semi-natural and agricultural habitats

within the landscape is likely to greatly inﬂuence landscape-scale metapopulation

98

dynamics of habitat generalists. Agriculturally—important organisms often follow a
bidirectional movement (Gavish-Regev et al. 2008), moving into agricultural ﬁelds when
prey populations emerge and back into semi-natural areas in the fall when crops are
harvested. Carabidae and Araneae known to routinely cross boundaries from semi-
natural habitat patches into agricultural crops (French et al. 2001) but why they do so is
vastly understudied. It is likely that some species exhibit bidirectional movement,
moving into soybean from semi-natural habitats in response to prey availability.
Alternatively, species may be exhibiting negligible movement (Gavish-Regev et al. 2008)
through agricultural habitats, where individuals move through an unsuitable habitat to
colonize other patches of suitable habitat (Gavish-Regev et al. 2008). A reduction in
semi-natural habitat may force species to disperse through soybean to colonize isolated
patches of preferred habitat. Understanding how habitat generalists that vary in their
capacity for dispersal utilize agricultural habitats has implications for biological control
as well as the conservation of biodiversity. Well established patterns exist that illustrate
the inﬂuence of landscape structure on these organisms in several agricultural systems, it
is time to begin building an understanding of the functioning of these organisms at the

metacommunity level.

99

Table 4.1. Land cover categories used in the calculation of Simpson’s Diversity Index

(D) and as variables included in the principal components analysis.

 

Simpson's Diversity
(D) PCA

 

 

Cropland Categories

Corn Corn
Soybean Soybean
Wheat

Sugar Beet Other Cropsl
Orchard

Dry Bean

Sunﬂower

Cabbage

Canola

Potato

Carrot

Squash

Ornamental Crops
Tilled Field
Non-Crop Categories

 

 

Forage Grassland2
Old Field and Restored Prairie
Pastureland

Forested Forest
Wetlands and Open Water Water
Urban Urban

 

l Other Crops contains all Simpson's D
cropland categories except corn, soybean,
and wheat.

2 Grassland contains the Simpson's D non-
crop categories Forage, Old Field and
Restored Prairie, and Pastureland

100

Table 4.2. Models compared by AlCc analysis at spatial radii of l-3.5 km.

 

y=Bo
y=Bo+BIPC1

y=Bo +BIPC2

y=Bo+BID

y=Bo +BIPC1+32PC2
y=BO+BIPCl +32D

y=Bo +BIPC2+32D
y=Bo+BIPCl +32PC2+B3D

lOl

Table 4.3. Abundance, trophic level and overwintering stage of Carabidae species found

in soybean ﬁelds in Iowa, Michigan, Minnesota, and Wisconsin.

 

 

Species N Feedingl Overwintering2
Agonum cupreum 3 P A
Agonum cupripenne 21 P A
Agonum octopunctatum 2 P A
Agonum placidum 224 P A
Amara aenea 93 H A
Amara impunctata l H A
Amara obesa 3 H A
Amara spp. l X X
Anisodactylus harrissii 4 P A
Anisodactylus rusticus 15 P A
Anisodactylus sanctaecrucis 74 P A
Bembidionfemoratum 23 P A
Bembidion nitidum 27 P A
Bembidion obtusum l P A
Bembidion quadrimaculatum 85 P A
Bembidion rapidum 36 P A
Bembidion rupicola 4 P A
Brachinus ovipennis 62 P A
Calathus gregarius 5 O A
Calosoma calidum 1 P A
Chlaenius sericeus 2 P A
Chlaenius platyderus l P A
Chlaenius pusillus 86 P A
Chlaenius tomentosus l P A
Chlaenius tricolor 152 P A
Clivina bipustulata 59 P L
C livina impressefrons 3 l P A
Clivina spp. 2 X X
Cyclotrachelus seximpressus 5 P L
Cyclotrachelus sodalis l l 121 P L
Dicaelus elongatus 43 P A
Dyschirius globulosus l P A
Elaphropus anceps 61 P A
Galeritajanus 4 P A
Geopinus incrassatus 140 P A

102

Table 4.3 (continued)

 

 

Species (continued) N F cedingl Overwintering2
Harpalus aﬂinis 36 P A
Harpalus caliginosus 44 P A
Harpalus erraticus 1310 P A
Harpalus herbaivagus 42 H A
Harpalus pensylvanicus 2267 H L
Lebiafuscata l P A
Lebia spp. 2 X X
Leptotrachelus dorsalis l X X
Poecilus chalcites 7806 P A
Poecilus lucublandus l 254 P A
Pterostichus commutabilis 14 X A
Pterostichusfemoralis 6 X X
Pterostichus melanarius 688 P L
Pterostichus permundus 2380 P L
Scarites quadriceps 847 P A
Stenolophus comma 145 P A
Stenolophus lecontei l P A
Stenolophus ochropezus 5 P A
Stroluphus chropezus 1 P A
Synuchus impunctatus 3 P A
T rechus quadristriatus 27 P L
unknown species 1 2 X X
unknown species 2 3 X X
unknown species 3 l X X
unknown species 4 1 X X
unknown species 5 1 X X
Carabidae 29300

 

I Carabidae are classiﬁed by trophic level as either H for
herbivorous, P for predatory or omnivorous, or X for unknown.

2 Carabidae are classiﬁed by overwintering stage as either A for

adult, H for larva, or X for unknown.

103

Table 4.4. Carabidae community composition in soybean ﬁelds in Iowa, Michigan,

Minnesota and Wisconsin in 2005-2006.

 

 

Species Iowa Michigan Wisconsin Minnesota
Acupalpus partiarius 0 O 0 0
Agonum cupreum 0 0.03 0.04 0
Agonum cupripenne 0.01 0.25 0.07 0
Agonum octopunctatum 0 0.03 0 0
Agonum placidum 0.07 1.49 1.56 1.55
Amara aenea 0 1.23 0.07 0.05
Amara impunctata 0 0.01 0 0
Amara obesa 0.02 0 0 0
Amara spp. 0.01 0 0 0
Anisodactylus harrissii 0.03 0 0 0
Anisodactylus rusticus 0.02 0.08 0.11 0.08
Anisodactylus

sanctaecrucis 0.03 0.76 0.46 0.05
Bembidionfemoratum 0 0.04 0.64 0.05
Bembidion nitidum 0 0.33 0.11 0
Bembidion obtusum 0 0.01 0.00 0
Bembidion

quadrimaculatum 0 0.19 1.84 0.48
Bembidion rapidum 0.16 0.15 0 0
Bembidion rupicola 0 0.06 0 0
Brachinus ovipennis 0 0.35 0 0.93
Calathus gregarius 0.03 0.01 0 0
Calosoma calidum 0 0 0.04 0
Chlaenius sericeus 0 0 0 0.05
Chlaenius platyderus 0.01 0 0 0
Chlaenius pusillus 0.51 0.10 0 0.03
Chlaenius tomentosus 0.01 0.00 0 0
Chlaenius tricolor 0.77 0.36 0.32 0
Clivina bibustulata 0.29 0.21 0 0
Clivina impressefrons 0.03 0.36 0 0
Clivina spp. 0.01 0 0.04 0
Cyclotrachelus

seximpressus 0.03 0 0 0
Cyclotrachelus sodalis 51.10 11.88 4.25 58.80
Dicaelus elongatus 0.28 0 0 0
Dyschirius globulosus 0 0 0 0.03

104

Table 4.4 (continued)

 

Species (continued) Iowa Michigan Wisconsin Minnesota
Elaphropus anceps 0.20 0.28 0.39 0
Galeritajanus 0.03 0 0 0
Geopinus incrassatus 0 0 4.96 0
Harpalus afﬁnis O 0.43 0.07 0.08
Harpalus caliginosus 0.28 0 0.04 0.03
Harpalus erraticus 0.01 0.01 46.37 0
Harpalus herbaivagus 0.05 0.17 0.35 3.64
Harpalus pensylvanicus 2.94 8.90 12.12 20.89
Lebiafuscata 0 0.01 0 0
Lebia spp. 0.01 0 0 0
Leptotrachelus dorsalis 0.01 0 0 0
Poecilus chalcites 18.91 61.39 4.47 8.88
Poecilus lucublandus 5.90 0.72 1 .3 l 6.62
Pterostichus

commutabilis 0.01 0.08 0.25 0
Pterostichusfemoralis 0.04 0 0 0
Pterostichus melanarius 0.07 2.96 16.27 0.10
Pterostichus permundus 13.13 4.67 0.25 0.80
Scarites quadriceps 4.88 0.98 0.82 0.23
Stenolophus comma 0.04 0.97 2.45 0
Stenolophus lecontei 0.01 0 0 0
Stenolophus ochropezus 0.03 0 0 0
Stroluphus chropezus 0.01 0 0 O
Synuchus impunctatus 0 0.01 0.07 0
T rechus quadristriatus 0.00 0.37 0.00 0
unknown 1 0 0.01 0.07 0
unknown 2 0 0 0.04 0
unknown 3 0 0 0.04 0
unknown 4 0.01 0 0 0
unknown 5 0.01 0 0 0

105

Table 4.5. Spider community composition in soybean ﬁelds in Iowa, Michigan,

Minnesota and Wisconsin in 2005-2006. All spiders were identiﬁed to family, Lycosidae

were identiﬁed to genus and where possible, species.

 

Percentage of Total Araneae

 

 

Family IA MI MN WI
Agelinidae 0.25 0.23 2.42 2.81
Araneidae 0 0.15 0 0
Gnaphosidae 0.98 0.30 1.61 0.56
Linyphiidae 2.66 28.33 53.23 34.27
Lycosidae 88.77 66.31 40.32 61.42
Oxyopidae 0.25 0 0 0
Salticidae 0.45 1.66 0 0.37
Tetragnathidae 0 0.23 0 0
Thomisidae 6.66 2.79 2.42 0.56

 

Percentage of Total Lycosidae

 

 

Lycosidae Species IA MI MN WI
Allocosa modesta 0.18 0.34 0 0
Hogna helluo 4.37 1.36 0 1.22
Pardosa saxatilis 26.55 34.09 0 21.65
Pardosa sp. 1 0 2.61 0 0
Pardosa sp. 2 0 0 0 3.05
Schizocosa avaida 6.53 3.64 80.00 3.96
Schzocosa crassipes 27.06 0.23 0 4.57
T rochosa terricola 0.97 25.00 8.00 35.98
Varacosa avara 0.64 0.11 2.00 0
Immature 31.75 30.45 10.00 29.27
Damaged 1.94 2.16 0 0.30

106

bicep 33:8 0338.5 53, 9280322 gunmen a 83%.: memoaaocma E $35.55 N

no.0 V m an 835:»?- 83 053:3 a 82865

 

 

 

107

.8 _

NNd de N58 N 2.8- em 5 2

So So cod 8.8 m 8.8- ANoecm + em 8. on

eNd N3 N58 N 2.8- § 5 0.8

So 85 85 2.8 m 8.8- A825 + an 8. c8

”Nd 2... N28 N 2.8- em an «N

3 8.o . 85 No.8 8 R8- swear-gm E «N

2 .o :2 N58 N 2.8- em as. oN

S 86 8.0 8.3 m 8.8- 302$ + em E oN

:5 82 N58 N 2.8- em mu. 3

:.o 86 85 2.8 .N 8.8- NEE + an E m.—

2..o- “N8125- "Gm :.o 25 EN 8.8 4 SN. +382va +2023 + em 5. on
26 8.2 N58 N 2.8- em as. 3

2... 81.. 8... Ed. 8 8.8- +£9.53 em as. 3

macaw—ebony 35m Nu _3 24 002 _M vowﬁx: N 2282 awn—wwm

e333
.2258 mauoaaeo

mm @3203 can 82 20 N mo 6 a NE? 2268 >15 .6on 3230 «won 05 823:2: 28— .358 00?. 8:838 05 mm 228 38%

some “a 38: _oueE “ma 8:- .annﬁmU MO 588% 33:8 oﬁ weaning com 8323?. aouooﬁom 638 we .CaQESm 6.? 28d.

 

108

8.8.0 .80- "NE 80.0 8.0 $2 8.8 e 8.8- $8028 + 08 + 08 =2 8N

8.0 8.0 8.0 8.8 m 8.8- tL.8028 + 02 :2 8

8.80 .80- "N02 .80- £00 8.0 2.0 8.0 No.8 8 8.8- +0 + £8028 +0028 + an 8 0N
8.80 .80- "NB 8.0 8.0 8.0 8.8 v 8.8- £8028 + 08 + an ax QN

8.810 .80- "NE .80- uSm 8.0 8.0 NS 8.8 8 8.8- +0 + £8023 £025 + em ex 8
8.0 8.0 8.0 8.8 m 8.8- £8028 + S 8 oN

Nb. - "N08 .880 8.0 2 .o 82 8.8 e 8.8- 58023 + 08 + 8 8 2

Ntona .80- uN0m .8.? 108 0.0 8.0 8.0 N2 .8 m 8.8- +0 + £8028 +2028 + 08 an. 3
8.0 N00 8.0 8.8 m 2.8- £8028 + an 8 8

8.0- "82 .80- £08 8.0 8.0 8.2 8.8 v 8.8- £8028 3028 + an :2 3

8.0 no .80- nN0m .80- £00 8.0 8.0 82 8.8 m. 8.8- 0 + 2.8028 +2028 + 90 E 8
8... S... 8... 8.8 m 8.8- EAN02£ + 0: .5. 8

80220800 202-am N. 20000.8(. _3 _2 00?. MM BMW—emu“: m N 20202 mmﬁwm

 

 

20208 8808800 m0 202208
08 82 .8 N .20 _< 0 525 20208 >80 A0208 =80>0 802 05 8000228 202 ._0208 002 88888 05 mm 200m 280% 2000 H0 208:

_0208 «8m 05- 0023800 w8808300>82820 .20 38802 326.00 08 880305 How 038208 80.5028 _0208 Mo 80885 8.2 0302

38:02 32500 525 030820—00 0>20w0= 0 000825 0000500009 E 0030ﬂ0>

 

 

0
0000 v m 03803 E .80 v m 8.823 t- .000 v m 0 008000000 003 03020, 0 8002030 N
b03000 0950-200— 22 05 5 00030: mo coupon—Bani 2:0 «GEN: H Q 000:? ~£0003Q 0.0009H2m M Q _

80$ .00. - “800 80.0 8.0 8.0 00.8 v 00.8- E80280 + 000 + 0m E. 0.0

:0 00.0 00.0 00.00 m 8.8- 580200 + 00 an. 0.0

8.0uo .80- n~00 3.0 8.0 80.0 8.00 v 80.8- E80280 + +000 + 00 80 0.0

30$ .80- u~00 .80- L00 00.0 8.0 00.0 8.00 0 8.8- 0-0 + £80200 +3020 + 00 :2 0.0

80.0 8.0 00.0 2.00 0 8.8- 58028 + 0m E 0.0

00020—00000 328m 0 «3 6 002 2M 2083002— 3202 €06
N . . M N — W:
-ws 3
0200.60.

 

8000080 5- 030-0

109

 

$.on8m .om. - "Gm do a; v: 8.: v 3%.. RES 592m + gm 5. 3

ad. un atonmg do 8d 8: NW: v 2.2. $23 + ABS + am no. 3

No as 86 3.? m 3%. RES + an E. 3

mg. "a .QduSm 25 Ed ”2 8.? v ”3m- KER 18$ + em 5. QN

Souwom «3. "Ba Go 3o 92 on: v 3.2. +8.5 +2on + an 5. 3

2... 3... 8... =3» m 5.8. +8.1 a + gm .5. 3

all“: .2.? "Sm mo M: .o e: 3.: v 3%. momwm +2053 + an :2 3

mg- "a among: 8o 25 o; a: v 2 .2. N83 + ABS + an 5. 3

”3 mac 8d 3.? m 8.2- +8.1 m + an as. 2

3 NS oo.~ 8.? m 3.x- ABS + an :2 3

«3.. no .0212 ”3 Ed 3: on? v 3%. RES 19$ + an no. 3

:6 v3 86 8.? m 3.2- moan + gm 5 3

macaw—oboo Baum N S» _< 002 _M wowﬁx: N 20602 mMW—vum
333

 

110

.2358 wquomaoo mm @332: 0.8. mm»:
.8 N no 6 a 5m? 20on >30 .388 :80; “won 05 830me Ron :0on oUH< 82558 09 mm 038 Emu—am :03 um Bum: 338

“we 2:. .32.?va pruaEBbgAmE— mo bﬁaoc 3.538 2: waumﬁgo 8% 859% 83028 $35 no ng .wé 033.

EVA—ow 5:58 HEB aEmaowEE gamma 5 8888 3858.39 8 moﬁmtaxw m

no.0 v m an 88$:me mg oﬁatg a m88€5+ N
bemoaao $80.98— A: 05 8 “553 .8 aouuoaoanﬁ EB NAEVN: u D 96:3 33.830 paoagm H D _

 

 

o; 8.0 cod 3:: m 6.2- +83% + cm :2 nm
3on8; .Mz. - “8m Ed 2.0 92 3.? v 3%. mommm +32va + em 5 3
Ed 33 cod 8.? . m «v.2- Roam + gm 3 3
$8522.80 Etna mu _3 _< 002 _M voonuovﬁ N $282 980 958%
33.64 .88

 

Gosquaoov ad 033.

111

88% 83.88 5; 8380838 o>umwoq a 8888 8858089 8 mainta> m

mod v K 8 8on8me mm? 05883 a 88085

 

 

bowoﬁo 880-882 g: 05 8 8833 8 88288H8 88w NOEN: H D 20:? ~88ng Macmgm u n“- N
8o 85 8.? N www.m- gm no. 3.
86 m2 cod 8.2. m $8. 888 + an :2 nm
Edna .26. “88 86 :.o 22 8.: v 8.9.. 3083 + 88 + an .8. 9m
86 £3 8.? m mnem- cm 5 3
mod one cod #8 m 8.8. 588 + an E 3
So- 26 o: 8.3. m 3.2. Em + am no. on
Edna .98- "88 go :.o of 8.: v 3.2- 20.88 + OS + 8 E. 3
3d one 8.2 N 92.n- gm :3 3
cod Rd cod 8.? m 3.8. aomrm + on no. 3
85 v2 8.3- N 8.x- am E 3
S 2.0 cod :8 m 8.9”. aumrm + 8 8 ca
25 9d cod 2.: m 8.2- +888 + an Ex 3
2... me... 8... 8.: «- 8.7-.- ..ANUBE + 8m .5. 3
8882880 38$ N.— _3 a 002 g vowwomx: N 2082 mMMWWM
38?.

 

112

2288 mac-2880 mm
8.888 0.8 82 8 N8 6 a at? 2088 .880 .888 3898 «won oﬁ 8:388 Eon 4088 00?V 888 05 8 23m 38% :08

«a 83m: 888 8a 2:- .0888-«0 b08808 .8 .8me 338m 05 88830 8m 8888 882% 888 8 E8888 .86 2an

 

 

25 N2 go 8.3 m 82:- .285 + an 8x mm

mmu~8 .3518 36 m2 cod 2.:- V 2.:- ~8~m {8S + am n2 3

So Rd 33 on? m :2- t-Bﬁm + em 5 o.~

3E8 .33 8 go 9.0 28 8.? v 8.2- 883 +8128N m + an 82 3

2 8o 62 3? m 3:- 288 + an E E

3.38 anon“?— ﬁ... «3. 8... and. v 3.3- ~83 928$ + gm .5. m.—

Qdu~8 isno : .o So 0: mos- v .8:- m8mm + harm + om 5. _

So. no .338 :6 So 2; was. v «.2:- 283 + ABS + an E H

Ed 23 mod 2.3 m :3:- 28£ + an E _

mmon~8 .5318 go 35 cod 8.3. v 3:- 883 +285. + an .8. H
macaw—ebony 38am 8 _3 _< 002 _M @8885 N $082 Aqua

3%.? .84 3:5

.20on

wcuomaoo mm @8205 2a 32 8 m .8 5 a £3» £288 >80 .3on =8o>o «won on... 83083 Eon ,8on 0014- 85888 0% mm 088

€3QO some “a 3%: 888 BE 2E- .omoauz .8 baa-«6 >868 05 88396 8m mom-Bung 8828 888 .8 855m .3 .v 03.5%

113

32:6 3338 at» 933358 gamma: a 3363 momoﬁgq E 83323 m

mod V m «a Enema—swam an? Egg.» a m8mo€£+ N

Cowoaao agreeing 5m 05 5 «853 m0

aoanoHn—HE can «AMEN: M Q 95:3 \EEUZQ Macadam u G _

 

 

w; 26 «to was. m .32. toga + cm .5. on
goumunm @6qu :6 one cod 2.3 V IL. momma +285 + an E on
S .o 2.0 $5 3.3 m 3.2- 288 + an as am
moonmom $.0qu 33 and 8d 8.3. v no.2- 8.3m +29% + an E 3
gown—280 Evan .— 65 _< 002 _M 602$qu E52 Anus
m -woq N _ 33
38.5%

 

ABEESV o _ .v $5

114

 

 

115

2d 3d 2.3 N $2. em :2 3
3o «.3 85 mm? m 2.2- 85. + gm .8. 3
86 NS 2: 2.3 m 32:. 8.5 + an 5. od
:6 o} 2.2. N 88. em E 3
Rona“: gonna 2 .o o? 2: 2.3. v 3:- ~83 tuna + am no. 3
2. a... 3.... «3.. m 2.2- «35+ gm .5. .2
omoua isnSm 3 a; ”5 $9. v o3? momma + QE + an an 2
26 02 2.3 N 3.8- em 83 2
mnoumom .de£8 26 26 who 3.? w 3.2- momma 18$ + em .5 3
8 3o 85 8.3. m $2- 8.5 + em .8. 3
Ed NS 3.2. a $8- . em 5 3
@3qu .338 NS :6 w: 8.3. v 8.2- NBS +6.5 + an .6. 3
3ch .QduSm NS :5 c: 8.8 v 3.8- +85% + Qm + cm 5. 3
3 one 85 53 m 3.3- 8.5 + an no. 3
30:29:00 393m N.— Sw 5 002 _M vowﬁﬁ~ N 20.52 mmﬁvnm
33.8....
20on wﬁuomﬁoo

mm vows—05 03 $2 Ho m .«o 6 a 5:5 Eugeéco .3on 2896 “men 05 830%“: Eon 4258 002 Snead: 05 mm 030m 333m

:08 “a 3%: 3on “Eu 23. .32893 «c bung brEom 05 maugﬁﬁ How 859% 8:028 «#58 mo $883 A 3. £an

mmdﬂQ .ovduNUm

mod v m an “5..chme 33 03¢? a 330va
bowoﬁo uo>ooéq£ 5m 06 E ”853 m0

+N

comp—acacia can NAEVN: u D 20:? bﬁoﬁﬂ @5385 u G _

 

3 2.0 3._ 3.3 3 NNQ- *NUNNN + GS + an 5. Om

amount «Noun: NS 26 0N; 2.3 v 2.2- NONE imam + on so— 2

Ed 3N.“ 2.3 N NMSN- cm 5 on

3 So 23 No.3 m N37 Noam + gm :2 Wm

META .3.onNom 86 N3 3: 8.3 3 3.3- .NomNm + OS + 3 Ex 3.

Ed 32 2.3 N NNdN- an Ex 3

mnonNoN .33qu NS :6 o? 2.3 3 2i- NomNm ioﬁm + on E. 0.».

3 N2 85 N33 N 23:- Noam + gm 5. gm

8an .NQHNB mod N; 32 N33 3 82- N83 + 95 + on E. N

Nod N3 N? 2.3 m SON- 8.5 + an ac. 0N

3.onNom .30an :.o 2.0 o: 3.3 3 02:- NomNm :83 + gm go. ON

mucus—9:00 Etna— NH 35 5 o0? _M vowwwmé N 2262 by
333.

 

32583 :3 03¢

116

 

 

83 v m @863 E. .85 v m a :aomama 33 £3.53, 3 mesaﬂ _
~3.o 85 23 OS: m 33. Eaﬁm + 3m E. on
o3.o Gd cod 93: m :3. + 3. +6.: m + 3m a. 9m
o3.o 3o cod 3.2 m an. :29: + 3m E 3
3o 33 8.0 2.2 m 8.3- $223 + 3m as. 3
woo 3o 85 232 m 2.3- 1.233 + 3m an. 3
32 3o cod 3.: m ”E- :Sﬁm + an .5 3

mu #3 5 00?. WM gown—ox: $032 9mg:
32333 - 3 63m
20on wquomaoo

ma 86205 08 m8— .5 N no a a ﬂ?» 2358. 3:0 .388 :30; “won 05 8:863 Eon .3on 00733 Eng 06 mm 038 33%

:03 E 8%: 3on 5E 2F .omEEBAEA mo Emmﬁé 335m 06 m§m3m>o How 8533 deuce—om 6on mo Egan—m .NH .3. 2an

117

Figure 4.1.

 

Figure 4.1. Map of soybean study sites sampled for Carabidae, Opiliones and Araneae in
2005 (1 -13) and 2006 (14-26) using pitfall traps. Sites were selected so that the 3.5 km
landscape radius surrounding each ﬁeld varied in diversity and the abundance of semi-

natural and agricultural habitat.

118

Figure 4.2.

Total Carabidae

 

lllllllll

C yclotrachelus sodalis

 

1111111111

25
20 ~
15 .
10 <

Poecilus chalcites

 

llllllllll

/
I

 

 

 

 

 

 

llllllllll
f

6/5 6/19 7/3 7/17 7/318/14 6/4 6/l8 7/2 7/16 7/30 8/13

2005

2006

Figure 4.2. Activity density by date of (A) total Carabidae (B) C. sodalis, (C) P.

chalcites, (D) P. permundus, (E) H. pensylvanicus, (F) H. erraticus, (see page 120) G) P.

lucublandus, (H) S. quadriceps, and (I) P. melanarius (see page 121) collected per pitfall

trap for the 26 soybean ﬁelds sampled in 2005 and 2006. Species are shown in order of

overall abundance.

119

Figure 4.2 (continued)

Haipalus pensylvanicus Pterostichus permundus

Harpalus erraticus

 

 

 

 

 

 

 

 

 

 

12 / ,4
D
10 -
8 3
6 1
4 .
2‘ w
O .
1 J 1 1_1 1 1 1 71 fl 1111111111
12 7/ I,
1‘3 '
10
8 3
6 .
4 .
.L
2 .
O ..
111111111 lﬁ/ ll 1 1 1 1 1 1 1 1 1 1
20 / /
F
15 1
10 -
5 3
0 W—O—H
1111111 1 L 1 7/ r] 1 g 1 1 1 1 1 1 1 1
6/5 6/19 7/3 7/17 7/318/14 6/4 6/18 7/2 7/16 7/30 8/13
2005 2006

120

 

 

 

Figure 4.2 (continued)

Scarites quadriceps Poecilus lucublandus

Pterostichus melanarius

/

I

 

4

3.1

I

M.

I

m

 

 

LLIllllllll

 

I

I

 

 

I

/

 

 

I

M

I I I I l I I l I I

I

Ll_l_l l I l l I l l

 

I
I
6/5 6/19 7/3 7/17 7/318/14
2005

I
6/4 6/18 7/2 7/16 7/30 8/l3
2006

121

 

Figure 4.3.

 

 

 

 

 

 

 

 

 

10 2’ /
A

8 .1

6 ..
d)
8
E 41
<

2 ..

o 3

30

25 -

20 ~
8’3
8
a 15 ‘

10 ~

5 ..

O I I I J l l I I I I4 ,1 I I I I I I l l I I

6/5 6/19 7/3 7/17 7313/14 6/4 6/18 7/2 7/16 7/30 8/13
2005 2006

Figure 4.3. Activity density by date of (A) total Araneae (B) Opiliones, and the two most
abundant families of Araneae: (C) Linyphiidae (see pg. 123), and (D) Lycosidae (see pg.

123) collected per pitfall trap for the 26 soybean ﬁelds sampled in 2005 and 2006.

122

Figure 4.3 (continued)

 

 

 

 

 

 

 

 

 

 

 

 

2.5 / /
C
2.0 -
o 1.5 "
ﬁ
4;. 1.0 -
.s
'4 0.5 -
0.0 ~
10 7/ I]
D
8 .1
6 d
3%
2
a 4 -
E
2 J
0 -
6/5 6/19 7/3 7/17 7/318/14 6/4 6/18 7/2 7/16 7/30 8/l3
2005 2006

123

Figure 4.4.

 

 

 

 

 

 

 

 

1.2 7’ /
1.0 -
§ 0.8 -
,o
g 0.6-
a
o 0.4 —
é
a 0.2 -
0.0 -
1.8 B
1.6 -
L4 a
.23 1.2 3
3 1.0 3
§ .
5’. 0.8
Q -1
E; (16
0.. 0.4 -
0.2‘ -
0.0 -

 

 

 

 

6/5 6/l9 7/3 7/17 7/318/14 6/4 6/18 7/2 7/16 7/30 8/13
2005 2006

Figure 4.4. Activity density by date of the two most abundant species of Lycosidae (A)

T. terricola and (B) P. saxatilis collected per pitfall trap for the 26 soybean ﬁelds

sampled in 2005 and 2006.

124

Figure 4.5.

 

 

 

 

0.6
0.4 < . Grassland
O
T. 0.2 - . 0 Water
§ Corn Urban
0 O
c. 0.01
a O
O O
2 Soybean Forest
"a -0.2
O
c:
'5
9" -O.44
O
-O. 6 . Wheat
Other Crops
-O.8 . 3 . T 1
-O.6 -O.4 -0.2 0.0 0.2 0.4 0.6

Principal Component 1

Figure 4.5. PCA ordination for principal components 1 and 2 of landscape elements
surrounding soybean ﬁelds sampled. Data for a radius of 2 km is shown, plots are similar
for all spatial scales. Sites with positive loadings on PCI were correlated with the
variable Forest while negatives loadings on PCI were correlated with the variables Corn
and Soybean. Sites with positive loadings on PC2 were correlated with the variable

Grassland while negative loadings on PC2 were correlated with the variables Other Crops

and Wheat.

125

Figure 4.6.

 

 

 

 

 

 

Log (Mean Carabidae +1)
DJ
0
o.
. 0
Log (Overwmtenng Larvae + 1)

 

 

-4 -3 -2 -1 O 1 2
Principal Component 2 (1 km)

 

 

   

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

I c

= +

23’ E

a, ~§

c e

.‘s‘ E

i 2

<5, :3

ED -4 v r u T t t I 3 -1 1 I 1 ﬁ I 1 T
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 -5 -4 -3 -2 -l 0 1 2 3

Principal Component 2 (1 km)

E I

s .43

:3

~§ U

°~ E

5 a:

E e

3) no

_I 1 1 1 1 1 1 1 3 1 I r r I I
-5 -4 -3 -2 -l O l 2 3 _4 -3 -2 -1 o 1 2

Principal Component 2 (1-5 km) Principal Component 2 (1 km)

Figure 4.6. Relationship between (A) activity density of Carabidae and PC2 at 1 km, (B)
adult-overwintering Carabidae and PC2 at 1 km, (C) P. chalcz'tes and PC2 at 1.5 km, (D)
larval-overwintering Carbidae and PC2 at 2.5 km, (E) C. sodalis and PC2 at 2.5 km, and
(F) predatory Carabidae and PC2 at 1 km. Carabidae data were collected using pitfall

traps. Means are log (x+1) transformed.

126

Figure 4.7.

 

 

 

 

 

5?

2 o'
g +
E ‘6‘
a 3.
t.“ ...J
a V
a. 1:
O U

m

l.2 ﬁ I I I I I I
-5 -4 -3 -2 -l 0 l 2 3

 

Partial Residuals

 

 

Log (Linyphiidae + l)

 

 

Principal Component 1 (1.5 km)

 

 

 

 

3.0
2.51 C ' '
O
1 o o
2.0 .. .
1.53 °
W
1.03 0 ' ° 0
. C. O .
0.5 I ﬁr T T I I
-3 -2 -1 o 1 2 3 4

Principal Component 1 (2 km)

 

 

 

 

 

Principal Component 1 (3.5 km)

Figure 4.7. (A and B) Araneae activity density in 26 soybean ﬁelds was best explained

by a model containing both PC] and PC2 at a radius of 1.5 km. Partial residual plots

illustrate a positive correlation with both PCI and PC2. Plots (C) and (D) illustrate the

relationship between Lycosidae activity density and PC2 at 2.5 km, and the relationship

between Linyphiidae activity density and PC] at 3.5 km respectively.

127

CHAPTER 5

LANDSCAPE COMPOSITION INFLUENCES PATTERNS OF NATIVE AND

EXOTIC LADYBEETLE ABUNDAN CE.

Abstract

Landscape structure provides the foundation for species coexistence and
resulting ecosystem services. Coccinellid beetles are important predators contributing to
pest suppression in agricultural landscapes. Since the introduction of the exotic
coccinellids Coccinella septempunctata L. and Harmonia axyridis Pallas into the U.S.,
several studies have reported a decline of native Coccinellidae in agroecosystems. As
part of a two-year study on the biological control of the soybean aphid, Aphis glycines
Matsumura, I examined coccinellid communities in soybean ﬁelds across Michigan,
Wisconsin, Minnesota and Iowa. In 2005-06, exotic species made up 90% of the total
coccinellid community in Michigan soybean ﬁelds followed by Wisconsin (84%),
Minnesota (66%) and Iowa (57%). Harmonia axyridis was the dominant exotic
coccinellid in all states comprising 45-62% of the total coccinellid community, followed
by C. septempunctata (13-30%). Two additional exotic species, Hippodamia variegata
(Goeze) and Propylea quatuordecimpunctata (L.) were also discovered to be established
in the region. Overall, the most abundant native coccinellid was Hippodamia convergens
Guerin-Meneville, however, its abundance varied across the region, comprising 0%
(Michigan) to 28% (Iowa) of the total coccinellid community. This is the ﬁrst study to

illustrate that patterns of native and exotic coccinellid abundance are inﬂuenced by

128

landscape structure. I found that native coccinellids were most abundant in low-diversity
landscapes with an abundance of grassland habitat while exotic coccinellids were
associated with the abundance of forested habitats. I propose that grassland dominated
landscapes with low structural diversity and low amounts of forested habitat may be
resistant to exotic coccinellid build-up, particularly H. axyridis, and therefore represent
landscape-scale refuges for native coccinellid biodiversity.
Introduction

Across the majority of the north central U.S. agricultural ecosystems form a
human-mediated landscape matrix containing small patches of formerly dominant natural
habitats. These natural habitats may serve as source populations of organisms which
colonize the highly disturbed matrix, inﬂuencing the abundance of pest and beneﬁcial
insects and the ecosystem services such as biocontrol and pollination they provide in
croplands (Marino and Landis 1996, Menalled et al. 1999, Ostman et al. 2001, den Belder
et al. 2002., Steffan-Dewenter et a1. 2002, Thies et a1. 2003, Thies et al. 2005, Tscharntke
et al. 2005b, Marino et al. 2006, Westphal et al. 2006, Gardiner et al. in press).
Landscapes supply both native and exotic species to agricultural ecosystems resulting in
agricultural food webs containing intentionally and unintentionally introduced species at
each trophic level. Although it is well established that the two largest threats to native
biodiversity are a loss of habitat and the introduction of exotic species, (Wilcove et al.
1998) very little is known regarding how the amount of natural habitat remaining in a
landscape inﬂuences the abundance of native species and their exotic competitors within
the agricultural landscape matrix. Didham et al. (2007) found that of 11,588 studies that

examined the inﬂuence of land-use change and 3,528 studies that examined the effects of

129

invasion published between 2002 and 2007 only 1.2 % considered both potential impacts
on native biodiversity. From both a conservation and ecosystem services perspective, it
is critical to determine how both of these factors inﬂuence populations of natives species.
Herein, I investigate the inﬂuence of landscape composition on populations of declining
native Coccinellidae and their exotic competitors.

The use of exotic lady beetles in classical biological control projects was ﬁieled
by the success of the vedalia beetle, Rodalia cardinalis (Mulsant), introduced ﬁom
Australia to suppress populations of the cottony cushion scale, Icerya purchasi Maskell,
in 1889. Since this early biological control effort, over 100 species of exotic lady beetles
have been either intentionally or accidentally released into North America (Harmon et al.
2007). Although the majority of introduced exotics have not signiﬁcantly affected native
communities, a small number successfully established throughout the US. have been
implicated as contributing to native coccinellid decline. The two most prominent of these
are Coccinella septempunctata L. and Harmonia axyridis Pallas. Intentional
introductions of C. septempunctata began in 1958, and over the next few decades beetles
were released in several states (Schaefer et al. 1987). This species was ﬁrst detected in
1973 in New Jersey and Quebec, however, it remains unknown whether these established
populations were the result of intentional or accidental introduction (Schaefer et al.
1987). Harmonia axyridis was ﬁrst detected in the US. in 1988 (Chapin and Brou 1991).
Like C. septempunctata, this species was intentionally released; but its eventual
establishment has been attributed to an accidental introduction via shipping (Day et al.
1994). In the north-central U.S. two additional exotic species have been detected

recently; Propylea quatuordecimpunctata (L.) (Gardiner et al. in press), and Hippodamia

130

variegata (Goeze) (Gardiner and Parsons 2005). Both species were released as part of an
effort at biological control of the Russian wheat aphid (Ellis et al. 1999).

Although exotic coccinellid species contribute to biological control of agricultural
pests (Cardinale et al. 2003,.Snyder and Evans 2006, Costamagna and Landis 2007a,
Gardiner and Landis 2007) increasing evidence suggests that some species of exotic lady
beetles may also displace native coccinellid species (Putnam 1955, Wheeler and Hoebeke
1995, Elliott et al. 1996, Colunga-Garcia and Gage 1998, Michaud 2002, Alyokhin and
Sewell 2004, Snyder et al. 2004a, Harmon et al. 2007). Putrnan (1955) was one of the
ﬁrst to document this phenomenon, indicating that introduction of Stethorus punctillus
Weise resulted in the displacement of the native Stethorus punctum (LeConte) in orchards
on the Niagara Peninsula. Wheeler and Hoebeke (1995) attributed the decline of
Coccinella novemnotata Herbst to an increase of C. septempunctata in the northeastern
U.S. Once common, only 10 specimens of C. novemnotata have been collected in North
America in the last 10 years (Losey et al. 2007). A similar pattern was documented in
South Dakota by Elliott et al. (1996) who attributed a 20-fold reduction in populations of
Coccinella transversoguttata richardsoni Brown and Adalia bipunctata (L.) to the
establishment of C. septempunctata. Colunga-Garcia and Gage (1998) reported a
decrease in populations of Brachiacantha ursina (F.), Cycloneda munda (Say), and
Chilocorus stigma (Say) following the establishment of H. axyridis in Michigan.
Similarly, Alyokhin and Sewell (2004) found a signiﬁcant decline in C. transversoguttata
and Hippodamia tredecimpunctata (Say) in Maine potato ﬁelds following the invasion by

exotic lady beetles. The introduction of H. axyridis has also been reported to have led to

131

a decline in other exotic species. Brown and Miller (1998) found that C. septempunctata
declined in apple orchards in West Virginia following the establishment of H. axyridis.

Like many natural enemies, adult coccinellids are transient predators, foraging
within several habitats during the growing season (Evans 2003). Therefore, their
diversity and abundance is likely to depend on both the abundance of prey within crop
habitats and the structure and composition of the surrounding landscape (Marino and
Landis 1996, Colunga-Garcia et al. 1997, Bommarco 1998, Elliott et al. 1999, Thies et al.
2003, Schmidt and Tscharntke 2005, Tscharntke et al. 2005b, Gardiner et al. in press).
Landscape variables such as habitat composition, quality, and patchiness as well as
dispersal capability, all impact the abundance of coccinellids in agricultural crops (Elliott
et al. 1999, Thies et al. 2003, Schmidt and Tscharntke 2005). Agroecosystems are
ephemeral habitats where disturbances such as tillage, pesticide application and
harvesting require both pests and their natural enemies to frequently re-colonize ﬁelds
(Wissinger 1997). The surrounding landscape supplies a species pool of these organisms,
and may impact the invasion success of exotic coccinellids as well as interactions
between native and exotic coccinellid species.

Several recent studies show that coccinellids supply a valuable ecosystem service
to soybean farmers, suppressing populations of the soybean aphid, Aphis glycines
Matsumura, an invasive pest of soybean in the US. (Venette and Ragsdale 2004, Fox et
al. 2005, Costamagna and Landis 2007a, Gardiner and Landis 2007, Gardiner et al. in
press, Chacon et al. in review). Given the patterns of native species decline, and the
importance of landscape diversity and composition in supplying beneﬁcial insects to

croplands, I was interested in determining if soybean ﬁelds in landscapes that varied in

132

diversity and composition were colonized by a similar community of coccinellids. My
objectives were to: 1) measure variation in the composition of the coccinellid community
across the north-central states of Iowa, Michigan, Minnesota and Wisconsin 2) determine
if the abundance of native and exotic coccinellids was related to the diversity and
composition of the surrounding landscape. My hypothesis was that landscapes with a
high proportion of natural habitat such as forests and grasslands would supply the largest
population of coccinellids to soybean ﬁelds.
Methods

Field sites

Between June and August of 2005 and 2006, I examined the abundance of exotic
and native coccinellids in 33 soybean ﬁelds in Iowa, Michigan, Minnesota, and
Wisconsin (Figure 5.1). Twenty-two sites were part of a USDA Risk Avoidance and
Mitigation (RAMP) funded multi-state study of soybean aphid management. In each
RAMP site a randomized complete block design with 4-6 blocks and either 4 (2005) or 5
(2006) treatments was established in 0.4 ha (2005) or 0.2 ha (2006) plots. Multiple
experiments occurred within these ﬁelds; with the study reported here was conducted in
control plots not treated with insecticide. An additional 11 sites included in this study
were commercial production ﬁelds (2 in 2005 and 9 in 2006), each containing four 0.4 ha
(2005) or 0.2 ha (2006) untreated plots. Each year, a minimum distance of 20 km
separated each site. Across years, I maintained the relative regional distribution of
soybean ﬁelds, but sampled different sites. The average distance between soybean sites
sampled in 2005 or 2006 within a given region was 2.4 km (range = 0.2-14.6 km, Figure

1). Field size averaged 16.0 ha (range = 13.6-48.1 ha).

133

Aphid and coccinellid sampling

Beginning in June of 2005 and 2006 and continuing until mid to late August,
coccinellid diversity and abundance was estimated by placing an unbaited yellow sticky
card (PHEROCON AM, Great Lakes IPM, Vestaburg, MI) in each plot at all study sites
(4 cards per site in both 2005 and 2006). Yellow sticky card traps have been shown to be
an effective sampling technique for coccinellids in croplands, with greater capture
efﬁciency than other sampling techniques such as visual plant counts or vacuum
sampling (Mensah 1996, Parajulee and Slosser 2003, Schmidt et al. 2008a). In the
center of each plot, a metal “T” fence post was erected with holes every 10 cm vertically.
A 0.61 cm diameter dowel was placed through a hole such that it suspended a 22.9 x 27.9
cm sticky card just above the plant canopy. As plants grew, the dowel was moved up the
post to keep the trap just above canopy level. Sticky traps were replaced every 7 days.
All adult coccinellids were counted and identiﬁed to species. The number of exotic and
native coccinellids found per sticky trap was averaged across all sampling dates for each
site. As soybean aphid populations varied across the study region, I also measured prey
availability as a potential predictor of coccinellid abundance by conducting weekly
destructive whole plant counts during the timeframe of the coccinellid sampling. In each
plot, ﬁve randomly selected plants were removed from the ground and the number of
apterous and alate aphids were counted on each plant. For analysis, prey abundance was
averaged across all sampling dates for each site.
Landscape analysis

Field geospatial data was collected using a handheld GPS receiver using Wide

Area Augmentation System (WAAS) correction. The spatial coordinate for the center of

134

each site was used to obtain ortho-rectiﬁed digital aerial imagery. I digitized the habitats
surrounding each study site to a radius of 3.5 km using ARC GIS 9.1. Aerial images
were captured between 1998 and 2006. Land use changes between the image date and
study period were recorded by ground-veriﬁcation in June - August annually, with
corrections made during the digitization process. At the same time, I also determined the
speciﬁc land cover (n = 25 categories) present in all areas of each landscape (Table 5.1).
Some locations included polygons that were not visible from a roadway and permission
to access private lands could not always be obtained. These polygons were given a value
of zero and were excluded ﬁom further analysis. The area of each site that could not be
identiﬁed varied ﬁ'om 0-4.5%. The smallest polygons identiﬁed included ﬁeld plots on
university research farms and small patches of fallow ﬁeld (< 5 m2); the largest were
contiguous urban areas, lakes, and forests (< 11.9 kmz).

Landscape diversity was measured using Simpson's Index (D) (Simpson 1949).
Simpson’s Index is typically used to examine the variance of species abundance
distributions. Here I applied it to examine variance in the proportion of area covered by
each of 25 land use categories. This index was calculated for each of the landscapes at
each of the spatial scales examined. The equation for Simpson’s Index (D) is: D =
1/Z(pi)2 where pi=proportion of habitat in the it“ land-cover category (D increases as
diversity increases). Using methods modeled after Thies et al. (2003), I measured
landscape diversity at 6 spatial scales ranging from 1 km to 3.5 km radii (at 0.5 km

intervals) ﬁ'om the ﬁeld center.

135

Statistical analysis

To examine variation in coccinellid communities found in Iowa, Michigan,
Minnesota and Wisconsin I ﬁrst performed a likelihood ratio chi square analysis
assuming a multinomial distribution (Shao 2003). This test compared the proportion of
the coccinellid community composed of individual species at the state level across both
2005 and 2006 to determine if the coccinellid community within each state is
signiﬁcantly different from the null hypothesis that the proportional distribution is
equivalent across all four states.

To evaluate the relationship between coccinellid abundance and landscape
variables, I performed a principal components analysis (PCA) on the landscape variables
in order to reduce the dimensions of the data. The 25 observed land cover categories
were combined into seven variables included in the PCA: four crop variables and three
non-crop (Table 5.1). The four crop variables included the three most abundant crops as
separate categories: Corn, Soybean, and Wheat. The fourth variable was “Other Crops”
which included all other crops planted within a 3.5 km radius of our soybean sites. This
included many regionally-important and small acreage specialty crops. The three non-
crop variables were Forest, Grassland, and Urban. The Forest variable included all
forested land, the majority of which was deciduous forest. Some landscapes also
included small acreage of planted coniferous stands as well as mixed stands of conifers
and deciduous trees. The Grassland variable included all perennial habitat lacking
dominant woody vegetation. This included old ﬁeld and restored prairie, grazed
pasturelands and forage crops. The ﬁnal non-crop variable was Urban which included

residential and urban areas. The water and wetlands variable was dropped from the

136

principal components analysis as this habitat constituted a low (< 2%) percentage of the
majority of landscapes. Principal component axes were extracted using correlations
among variables and the resulting factors were not rotated (McCune and Mefford 1999).

I restricted the analysis to the ﬁrst two eigenvectors which explained between 51.3-57.3%
of the variability in landscape data. This was done for each spatial scale (1.0-3.5 km
radii).

Akaike’s Information Criterion was used as a model selection procedure to
examine the inﬂuence of six variables on the abundance of native and exotic
Coccinellidae. These variables were: Year (year site was sampled, 2005 or 2006), Prey
(average abundance of soybean aphid present in each site), Exotic (average abundance of
exotic Coccinellidae present in each site, used only in native Coccinellidae models), D
(Simpson’s Diversity (D)), PCl (principal components 1 interpreted from PCA), and PC2
(principal component 2 interpreted ﬁom PCA) (Table 5.1). For exotic coccinellids 17
models were compared: an intercept only model, sampling year model and models
containing all combinations of the two interpreted principal components, Simpson’s D,
and prey abundance (Table 5.2). For native coccinellids, 33 models were compared,
including an intercept only model, sampling year model, and models containing all
combinations of the two interpreted principal components, Simpson’s D, prey abundance,
and exotic coccinellid abundance (Table 5.1). At each spatial scale I present the model
with the minimum AICc value, i.e. with the best support for the data, and any competing
models with a AICc difference of less than two (Ribic and Sample 2001, Bumham and
Anderson 2002). For each model, I present the maximum log-likelihood estimate, the

Akaike weights, which estimate the relative likelihood of a given model against all other

137

models, and AICc differences (A,). I calculated adjusted r2 for the minimum AICc model
and competing models to evaluate how well the models explained the variation in the
data. I deﬁne the best overall model as the model with the lowest AICc score across all
spatial scales. After determining the best overall model for each predictor, I examined
the relationship between this model and the two most abundant species of native or exotic
coccinellids. I calculated partial correlations for all variables in models with more than
one predictor. Partial correlations were used to assess the importance of individual
independent variables after adjusting for additional variables in the model. The AICc
analysis and adjusted 1'2 were determined using R version 2.1.1 (R Development Core
Team 2005). Partial correlations were obtained using PROC CANCORR in SAS v. 9.1
(SAS Institute 1999). The mean number of native and exotic coccinellids, mean numbers
of individual coccinellid species as well as mean prey abundance were log (x+1)
transformed prior to analysis to meet the assumptions of normality and homogeneity of
variances (SAS Institute, 1999).

While no site had an overlapping 3.5 km landscape buffer within a given year, 24
sites did have overlapping buffers across years. Therefore, prior to interpreting the
results of the AICc analysis, I examined potential spatial autocorrelation of the residuals
of the best ﬁt and competing models for both response variables (exotic and native
Coccinellidae). This was done between each site and the nearest neighboring site using
Moran’s I statistic. I did not ﬁnd evidence of spatial autocorrelation between neighboring

sites for any of the best ﬁt or candidate models examined for either response variable.

138

Results
Coccinellid diversity and abundance across the region

Four exotic and eight native species of coccinellids were observed on sticky cards

in 2005 and 2006 (Table 2) across the study area in the north-central US. There were
signiﬁcant differences in the composition of this community across the four states (X224=

756.34, P<0.0001). Michigan had the highest percentage of exotic coccinellids (90.0%)
followed by Wisconsin (84.3%) (Table 5.3). Exotic species were less dominant in
Minnesota and Iowa, comprising (65.6%) and (57.2%) of the coccinellid communities
respectively. In all four states, the most abundant exotic species was H. axyridis
followed by C. septempunctata. Two additional exotic species were detected: H.
variegata in Michigan and Wisconsin and P. quatuordecimpunctata in Michigan
(Gardiner and Parsons 2005, Gardiner et al. in press).

The most abundant native species was Hippodamia convergens Guerin-
Meneville, which comprised a greater proportion of the community in Iowa (28.3%) and
Minnesota (19.5%) compared to Michigan (0%) and Wisconsin (4.9%). The second most
abundant native species was C. maculata which comprised 2.2-6.2% of the coccinellids
detected in soybean ﬁelds (Table 5.3).

Landscape diversity

The landscape surrounding each of the ﬁeld sites varied from agriculturally;
dominated to forest and grassland-dominated. Within the 3.5 km landscape radius
surrounding each of the 33 sites, landscape diversity values (D) ranged ﬁ'om 2.4-6.4. The
percentage of the landscape composed of annual cropland ranged from 11-9l%.

Landscapes with high and low percentages of annual cropland were sampled in each state

139

(Michigan 11- 91%, Wisconsin 20—7 1%, Minnesota 44-84%, and Iowa 27-89%). At a 3.5
km radius grassland habitat comprised l-69% (Michigan 1-21%, Wisconsin 4-21%,
Minnesota 6-18%, and Iowa 5-69%) while forested habitat comprised 0-58% of the
landscape surrounding ﬁeld sites (Michigan 3-32%, Wisconsin 3-58%, Minnesota <1-
16%, and Iowa 0-6%)
Principal component analysis of landscape variables

Both of the principal components interpreted in this study were measures of
landscape composition (Figure 5.2). PC] was correlated with the variable Forest while
negatives loadings on PC] were correlated with the variables Corn and Soybean.
Therefore sites with positive values of PCI suggest a landscape with an abundance of
wooded habitat while sites with negative values of PCI indicate a landscape dominated
by com and soybean agriculture (Figure 5.2). For PC2, sites with positive loadings were
correlated with the variable Grassland and negative loadings were correlated with the
variables Other Crops and Wheat. Sites with high values of PC2 had a high proportion of
pastures, old ﬁelds, and restored prairies. Sites with negative values had a high
proportion of locally important fruit, vegetable, ornamental and small grain crops. Both
principal components indicate the intensity of landscape disturbance, with high values
indicating less disturbed grassland and forested habitats and low values indicating greater
agricultural disturbance.
Model ﬁtting of relationships between coccinellid abundance, prey and landscape
variables

For exotic coccinellids, the PCI model had the lowest AICc value and the

PC1+Prey model was a competing model at all spatial scales examined (radii of 1-3.5

140

km) (Table 5.4). The PCl model at 2 km was the best ﬁt model overall, with the lowest
AlCc value across spatial scales. PCl was a signiﬁcant predictor of exotic coccinellid
abundance (P = 0.003, 2 km). Exotic coccinellid abundance was highest in soybean
ﬁelds in landscapes with an abundance of forested habitat (Figure 5.3a). The two most
abundant species of exotic coccinellid both increased with PCI. This relationship was
statistically signiﬁcant for H. axyridis (P = 0.004, 2 km) but not for C. septempunctata (P
= 0.139, 2 km) (Figure 5.3b-c). In the competing PC1+Prey model, prey abundance was
not a signiﬁcant predictor of exotic coccinellid abundance (P = 0.209, 2 km) but was for
H. axyridis (P = 0.022, 2 km), which increased with soybean aphid populations. There
was not a signiﬁcant correlation between C. septempunctata and prey abundance in the
PC1+Prey model (P = 0.475, 2 km) (Figure 5.3d-f).

For native coccinellids, the best ﬁt and candidate models varied by spatial scale.
The Year model had the lowest AICc value at spatial scales of 1-1.5 km, while at radii of
2-3.5 km the Diversity+PC2 model had the lowest AICc value. Several competing
models were found across spatial scales including: Year, PC2, Diversity, Diversity+PC2,
PC1+PC2, PC2+Exotic, and Diversity+PC2+Exotic. The Diversity+PC2 model at 2 km
had the lowest AICc value across spatial scales. Native species were signiﬁcantly
correlated with both predictors in this model; being most abundant in low diversity
landscapes (Diversity, P=0.043 at 2 km) with an abundance of grassland (PC2, P=0.038
at 2 km) (Figure 5.4 a and d). The most abundant native coccinellid species, H.
convergens was also signiﬁcantly correlated with both variables included in the best ﬁt
model. This species was also most abundant in low diversity landscapes (Diversity,

P=0.024 at 2 km) with an abundance of grassland (PC2, P=0.024 at 2 km) (Figure 5 .4 b

141

and e). There was not a signiﬁcant correlation between the second most abundant native
coccinellid, C. maculata and any of the variables in the best ﬁt model (Figure 5.4 c and
1).
Discussion

Over the past 100 years the proportion of native coccinellids found in lady beetle
communities has declined dramatically (Harmon et al. 2007), primarily since the mid
1980’s. Harmon et al. (2007) stated that in studies between 1914-1985, native species
averaged 95% of all coccinellid individuals found, while between 1987-2001 natives
declined to 67.5%. Several authors concluded that this decline in native species
abundance is tied to the establishment of exotic coccinellids (Putnam 1955, Wheeler and
Hoebeke 1995, Elliott et a1. 1996, Michaud 2002, Evans 2004, Snyder et al. 2004a,
Snyder and Evans 2006, Harmon et al. 2007). Here, I investigated the inﬂuence of
landscape structure on the abundance of both native and exotic coccinellid populations
across Iowa, Michigan, Minnesota, and Wisconsin. I found that coccinellid populations
in soybean ﬁelds across a 4-state region varied signiﬁcantly in diversity and abundance.
The proportion of the lady beetle community comprised of native species in soybean
increased from Michigan (10.0%) west to Iowa (44.8%). In all states, H. axyridis was the
most abundant exotic species, followed by C. septempunctata. The most abundant native
species in Wisconsin, Minnesota, and Iowa was H. convergens. This species was not
detected in any of the 14 Michigan soybean ﬁelds sampled, although it was documented
by previous investigators in soybean in recent years (Fox et al. 2005, Costamagna and

Landis 2007a).

142

Native and exotic coccinellid populations in soybean fields

This study illustrates that temporal, within-ﬁeld, and landscape variables all
inﬂuence the abundance of coccinellids in soybean ﬁelds. Year was a signiﬁcant
predictor of native coccinellids with populations higher in 2005 than 2006. I did not see
a signiﬁcant correlation between the within-ﬁeld average soybean aphid abundance and
trap catches of native Coccinellidae, however, yearly differences may be tied to variation
in the abundance of A. glycines across the study region. On average, soybean aphid was
more abundant in 2005 compared with 2006 across the four states studied. Fields with
particularly high aphid populations may have acted as sources of native species, resulting
in greater foraging across the landscape. An alternative explanation for the signiﬁcant
year effect is the positive correlation between the abundance of H. axyridis and soybean
aphid. This coccinellid is important in the biological control agent of A. glycines
(Gardiner and Landis 2007) and may act as more of an A. glycines specialist than the
other coccinellids sampled. Increases in populations of this species with soybean aphid
may have led to greater intraguild predation pressure on native species in late 2005,
reducing overwintering and 2006 populations.

Landscape variables were also signiﬁcant predictors of native and exotic
coccinellid populations. Landscapes with an abundance of forested habitat supported
greater numbers of exotic coccinellids, particularly H. axyridis, in soybean ﬁelds. Scares
et al. (2008) describe the “champion hypothesis” to describe the successful invasion of H.
axyridis. This hypothesis states that morphological, physiological, and behavioral traits,
including a large body size, high fecundity, high foraging efﬁciency and a high

conversion efﬁciency of prey to biomass, lead to the success of H. axyridis throughout a

143

large invaded range (Soares et al. 2008). While H. axyridis was present throughout the
study region, I found it in signiﬁcantly greater abundance in landscapes with an
abundance of forested habitats. The forested habitats present in these landscapes are
highly fragmented and intermixed within an agricultural matrix. This type of landscape
supplies exotic ladybeetles with an abundance of habitat edge and variation in vertical
structure which may favor species such as H. axyridis which is an arboreal species in its
native range (Chapin and Brou 1991). This coccinellid is highly mobile and disperses in
and out of forest patches throughout the growing season (Gardiner 2008). It also exhibits
a hypsotactic behavior when searching for overwintering sites, ﬂying to prominent
objects in the landscape (Koch 2003) such as forest edges. I propose that a landscape
lacking abundant forested habitat may limit the success of H. axyridis and potentially its
impacts on native coccinellids.

This study illustrates that the abundance of native species in soybean ﬁelds is
greatest in low diversity landscapes with an abundance of perennial grassland habitat.
These include pasturelands, perennial forage crops, restored prairies, and old ﬁelds. I
hypothesize that these perennial landscape elements are likely to provide early season
prey and overwintering habitat for native species, while not supporting high populations
of exotic competitors. These perennial habitats may serve as source populations of native
species that disperse into annual cropping systems such as soybean embedded within the
landscape. In addition to their association with grassland habitats, native coccinellid
abundance was negatively correlated with landscape diversity (D). While seemingly

counter-intuitive, landscapes with the highest values of Simpson’s D were highly

144

fragmented and contained high proportions of forested habitat, while grassland patches
were frequently embedded in low diversity agricultural landscapes.
Coccinellidae and landscape scale

I measured the response of coccinellid species to landscape structure at 6 spatial
scales ranging from a l-3.5 km radius from the focal soybean ﬁeld. Season-long
coccinellid populations were best predicted by the abundance of forests for exotic species
and grasslands for native species at a radius of 2.0 km. These results are consistent with
other studies examining the abundance of herbivores, natural enemies, and biocontrol
services (Thies et al. 2003, Schmidt and Tscharntke 2005, Gardiner et al. in press).
Gardiner et al. (in press) found that the amount of biocontrol services supplied by a
landscape to soybean ﬁelds was correlated with increased landscape diversity at a radius
of 1.5 km surrounding soybean ﬁelds. Thies et al. (2003) found that herbivory and
parasitism in wheat ﬁelds were positively correlated with percent non-crop area at
landscape diameter of 1.5 km.
Implications

This study demonstrates that native and exotic coccinellids are favored by
different types of landscapes, and that competitive and predatory interactions between
these species may contribute to this partitioning. It is possible that native species like H.
convergens and C. maculata may be adapted to open savanna and prairie landscapes that
formerly dominated large areas of the north-central US. and may still favor landscapes
with an abundance of open grassland and ﬁeld crop habitat. However, this does not
explain the decline in native coccinellids in the last few decades, long after landscapes

were transformed by human settlement. Alternatively, native species may only be able to

145

thrive in landscapes which are less favorable to invasion by exotic species many of which
are known to be strong intraguild predators. If exotics are favored by forested habitat, a
lack of these landscape elements in portions of Minnesota and Iowa may restrict the
abundance and dominance of exotics in these areas. Harmonia axyridis, the most
abundant exotic species found, was detected in 1994 in both Michigan and Iowa
(Colunga-Garcia and Gage 1998, Rice 2006) therefore it does not appear that these
differences are the result of an invasion front that has not yet fully colonized the western
end of the study region. Instead, I propose that landscape structure may be limiting the
success of H. axyridis and other exotics in intensely agricultural landscapes and allowing
persistence of native coccinellid communities in such regions. If so, understanding the
role of perennial grasslands in supporting source populations of native coccinellids is a
conservation priority. Evans (2000) discussed the habitat compression hypothesis, where
after the introduction of an exotic, native species decline in agricultural habitats as they
must compete with exotics for prey. Natives are forced back into ancestral habitats
where prey are sufﬁcient to maintain populations (Evans 2000). We must determine if the
different types of perennial grassland, such as pasturelands, restored prairies and
abandoned old ﬁelds vary in their ability to support native species. Within these habitats,
an understanding of the amount competitive displacement or intraguild predation pressure
from exotics incurred by native populations will demonstrate which serve as important

refuges in the conservation of these species.

146

Table 5.1. Land cover categories used in the calculation of Simpson’s Diversity Index

(D) and as variables included in the principal components analysis.

 

 

Simpson's Diversity (D) PCA

 

 

Cropland Categories

 

Corn Corn
Soybean Soybean
Wheat Wheat

Sugar Beet Other CropsI
Dry Bean

Tree Fruit

Sunﬂower

Cabbage

Canola

Potato

Carrot

Squash

Strawberry

Tomato

Bell Peppers

Melons

Mint

Ornamental and Rare Crops
Tilled Field

Non-Crop Categories

 

Forage Grassland2
Old Field and Restored Prairie

Pastureland

Forested Forest

Wetlands and Open Water3
Urban Urban

 

l . . .
Other Crops contarns all Simpson's D cropland categories
except corn, soybean, and wheat.

Grassland contains the Simpson's D non-crop categories
Forage, Old Field and Restored Prairie, and Pastureland

3 Proportion Wetlands and Open Water was dropped from the
PCA analysis due to rarity.

147

3% none £63» 26.505080 oﬂoxo .«o .8255 own—93 23. n ouoxm

 

 

 

8mm some 35m? :83 5m 33% 5238 .«o .5982. own—96 25. u .8.—m M

N 83888 agent; 1 8m 3

_ .8838 8qu 1 6m m

>18ng 53,869: 5m ca 3 “853 we aoquEQnE can NAEVN: u D 20:? >58an Mnemgm u Q N

88 8 88 .3295... 2......» 8305080 30> 1 new _

seam + 8m mm + 53$ + Qm + 3m

ouoxmnm + ca 3 + 8n mm + 63. + Qm + 3m 3:3 + 8.5 + mam + 3m
ouoxmlﬁ + .35 mm + 88$ + mam + 3m 35 mm + 6.53 + Qm + 3m
ouoxm£ + .65 mm .3 6.? + mam + 3m 35 mm + 8.33 + 6am + 3m
08me + ca mm + 8.5 + 6am + an exam + 8am + 3m
ouoxmmm + base + 8am + 3m . same + 6am + 3

82am + 3&3 + 6am + 3m 35$ + mam + 3m

ouoxmmm + 35$ + arm + 3m “seam + 3m

ouoxmmm + 35$ + 3m 8.23m + 5.5 + arm + 3m

83am + 8.3 .3 Guam + OS + 3m mean + arm + 3m
oaoxmmm + 8me + 05 + em Seam + mam + 3m

ouoxmmm + 6.5 + arm .3 3m 8.? + 6am + 3m

ouoxmmm + 8.5 + 885 + 3m 38m£ + 3m

ouoxmmm + 8.5 + 3m moam + 3m

ouoxmmm + 6am + 3m Noam + em

ouoxmmm + Q5 + 3m _aecm + 3m

308me + 3m 3m

«SE—«53:0 95.2 .8.. £352 128533. 265050er 95.2 “Ea 958nm— .Su 223—2

.5 one; .8 828m 33% a mine. so? 3 3888 £082 :3 0533

148

 

 

 

 

 

 

 

 

 

 

 

 

88-88 a 952 as 380qu .3332 .32 a ﬂow 533” a 82838 bangs E0588 .3 2.3

 

2... u a..." I... a :2 8... a N3 2... a a: 83.0585 .5;
8... a an... 8... a E... 8... u «.2 S... a 3... ha «.3 3: «.3 862; 26.2 .2
o o 83 a 83 So a 8.9 o o 3 to 2?: cs§§£8£
o o o :3 u :3 o o o no 33835 ﬁgw
o o :3 u :3 o o o 9o o Ewan 3:8ch
So a :3 So a :5 8.0 « So So a god to 3 3 we 53:2 €30th
86 a mod :3 H 3o 83 a :3 Sod a :3 E 3 no to 38§m§8n3 SEES:
So a 86 So a 3.0 So a So 8d a 8.0 3. 3 2 3. aasnsa saunoﬁﬁ
8.0 « go 8.0 a 30 So a o; So a o; 3 3 9m 3 33:85 Sawmsssu
86 a 2d 8d « ado o 86 H who 3 2: o 3a gmsée Eggnog:
830mm 33.2
E... u m5 2... a a... 8 a a: S... a 2: 3a 98 33 gm £84m scam .2
o Q So a 8.0 o o o «.2 o s§§m§8§5§u SEEN
god a 33 o 85 a 2 .o o 86 o Sw C Sawmtg SEQEEE
86 a 23 36 H and 8.0 a moo 86 a v3 0.2 mi 98 ME S§§§§a 33398
NE a a: o; a $3 So u wad So H m3 vs Nov gm 3% ant3 séﬁum
86on area

53085, 8805:: gwﬁog «32 53035, «Smog—:2 QWEEE «32

2mm « 9E beam 5&550588 $32 0830588 ace mo 383

149

 

 

 

150

:3 v .2 @323 t. .23 v m a 382.2»; 33 29.5.» a 82862.. N
towns”. 5,865. 52 as a .332 .6 Sﬁoaeana 2a “saw: u o 923 2835 €8.35 u a _
So u .65 .23 n 6.2 a; a; 2.2 No: v 3.2. 3&3 + 252$ + gm :2 nm
26 «no 86 2.2 m 2:1 28.23 + em .5 3
So u 32 .23 u 6.2 3o 86 2.2 2.2 2. NE- .623 + 2.6.32 + em .62 3
86 2d 8.0 :2 m 3.? 28$ + cm 82 3
«No u 22.2 53 u 5.2 2.0 .86 go 8.2 v 13. 22.23 + £6.52 + gm 5. 3
:6 m8 cod 2.2 m 2.? $6.25 + gm 52 3
mac n 52 .mg n 6.2 m3 Rd 23 2.2 v 22- 28.2mm + £6.28 + gm .8. ed
an... mm... 8... an: m m5- .2 25.1: + 2m .5. a."
mg n 28.2 .25 u Em and 2.3 mg 2.2 v 2%. 22.23 + :6: + em 52 2
26 one cod a2 m 2.2.. 20.1% + am so. 3
8.0 u 32 .23 u 6.2 Ed 85 Ed 2.2 2. mm? 3&3 + 26.5 + em .6. 3
26 3o cod :2: m 26. 26.32 + .5 so. 3
2532280 35$ 2 _3 6 00?. WM vooHEouﬁ Ego—2 And—V

82%;”... .2: N _ .83

.2258 $5388

mm 83:08 0.8 32 no N mo 6 a :23 20on 3:0 2258 3226 EB 05 82863 Eon 4288 002 Eng 05 2 038 33%

:03 “a Bum: 622: “PE QC. .omE__o:BooU 0:88 we oogvﬁﬁm 2t wags—go How motmuﬁm 83023 338 MO Egm .v.m 2an

2.0- n 038nm .wmd N NOAH .de- H D

 

 

151

v3 86 a: 2: m 83 6:383 #883 :83 + on :2 3
:6 8.0 N: N? m 02 +8.5 + an .8. 3
m; 25 3; x: m 88 +3883 + an :2 3
2.0 H mom .2.? u a 3c 8.0 86 93- v «3 $83 + harm + an .5. 3
:6 mod 88 «8 m :3 RES + am so. o.~
w; 85 mm: x: m 22 +8883 + on E 3
:6- u 82m £3 u Sm .30- n a mud mod c: mg m Sn 336mm 38.5 + SE + an as. 3
52. "8.8 an... n: on... a... 8... at... v 8m $23 + 28$ + .m .5. 3
3.6 u mom .9.? u a ”3 25 3: and v SA 8.: + KEG + an E 3
2d 26 Ed S; m ”2 +88 + an no. 3
25 ad 86 x: m can , +A§t£ + em E 3
8.0 V2 23 x: m 88 +3868 + an 8x 3
adorn—0800 188$ mu _3 a 00?. _M @0958:— m N move: 85¢
33.3 .98 3.6%
.2258 @8388

mm wows—08 as $2 Ho N we 6 a :85 2038 >80 .6608 =Eo>o “won 05 8:868 Eon 4358 oOH< 8:888 05 mm 23m 33%

some an 3%: 3808 88m 2? .oawEoEoooO 033: no 838:3“ 05 wﬁﬁﬁgo 8m motmaﬁm 8382me $608 mo $885 .m.m 2an

858% owvﬁoaoooo ”$88 55 83823—8 gnawed a 88:8 momoﬂgm 8 m03a5>

8.0 v m 383:8

.4;

.23 V m an 8858me 33 0588.» a 82835

m
in

30330 “gooﬁng 5m 05 8 «338 m0 :OEOQBQHE ES NAMENE .I. Q 82—3 33.830 Macmngm M Q _

 

 

2d- u 33m .Qd u 8.8 .93. u a mud . 8d a: v: m m? cuoxmdmm 3.8.83 + SE + dm 5. nm

”dd 8d d: 2: m 93 +8888 + dm Ex 2

2d- n 225 .dvd u 88 ad wdd d2 3d v v3. 33me +£8.1m + dm E. 2

odd n mom .8.? u Ga 8d odd ”2 Ed v 2;. tuommm inﬁrm + dm as. 3

mmd :d vod Ed- m Rd. £888 + dm 5 3.

3d n mom dmd- u a 8d 2d 8d Ed- v odd momma 18$ + dm :2 on

:d n mom .2. - H Um dmd 8d 33 v: v 3d +883 +£0.88 + dm 5. dd

2 d- u 32m dvd n mom dad- n 8 Ed 8d a: N: m Ed 3283 £883 + 85 + dm Ex dd

8.? n 288 dvd u mom dad 8d d: of V 8m 338mm #888 + dm as. dd.

2d 2d v2 2: m odd 8A§>Xm+dm :2 ed

2 d 2d and 9d m 8m +883 + dm :2 dd.

8d u 88 dd. u 8 Ed 2d 8d 2d- v Ed +883 + ABS + dm nu. dd
882050 338 m _3 a do? i doowdoé m N node: and

33?. - 3 33

 

ABaeqsd 2 29¢

152

Figure 5.1.

 

Figure 5.1. Location of 33 soybean ﬁelds sampled for coccinellids in 2005 (ﬁelds 1-13)
and 2006 (ﬁelds 14-33) in Iowa, Michigan, Minnesota, and Iowa. Sites were selected so
that the 3.5 km landscape radius surrounding each ﬁeld varied in diversity and the

abundance of natural and agricultural habitat.

153

Figure 5.2.

 

 

 

 

I

L l
I Grassland

0.5 - I .
l
l
l
I
Corn : . Urban
l
3 o o '
. ~ ——————————————————— 1- ———————————————————
m g I

. I Forest

L Soybean |
I
l
l
l

-0.5 4 Wheat . : O h

I . t er Crops

. I

. l .
-0.5 0.0 0.5

PC1

Figure 5.2. PCA ordination for principal components 1 and 2 of landscape elements
surrounding soybean ﬁelds sampled at a radius of 1.5 km. Points indicate the principal
component loadings of each variables included in the PCA analysis. Sites with positive
loadings on PC1 were correlated with the variable Forest while negatives loadings on
PC1 were correlated with the variables Corn and Soybean. Sites with positive loadings
on PC2 were correlated with the variable Grassland while negative loadings on PC2 were

correlated with the variables Other Crops and Wheat.

154

Figure 5.3.

 

 

Mean Exotic Coccinellidae
Partial Residuals
O
O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.22
g
l)
«’6 S
.15 3%
c
E’ d
<0 .‘2
I E
‘L '
2 '0'“ . -
£9. 0 2 4 6 8 10
£2
0
§ 0.8 . F
CE1 2 0.6-
co
3 3 0
Q :9 0.4 -
m a, o
0’ ° 02 - o ' '
g E ' . o ' '
E g 0'0 ‘W
8 § .02 - .f a" ‘ °
8 ' . '
§ l f 1 '0.4 I U r l l
2 -3 -2 -1 O O 2 4 6 8 10
Principal Component 1 (2 km) log (Prey+1) (2 km)

Figure 5.3. Relationship between the mean weekly catch of all exotic Coccinellidae (A),
H. axyridis (B) and C. septempunctata (C) from yellow sticky card traps and the overall
best ﬁt model, PC1 at 2 km. Partial residual plots illustrate the relationship between the
mean number of all exotic Coccinellidae (D), H. axyridis (E) and C. septempunctata (F)
and soybean aphid abundance in the competing model, PC1+Prey at 2 km. Models were
compared using an AICc model selection procedure. Raw data means are shown in A-C,

data were log (x+1) transformed for analysis.

155

Figure 5.4.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1’ _‘Q
.‘é E
g :6
a E
2 0.6 1 ii 0.6 a o
m a
13 0.4 - . g
m o
g 0.2 ‘ . .. . g
E (T:
0.0 -
‘3 - 75. . 5
CL -0.2 o ' . o 0-
-O.4 .
2 3 4 5 6 7 8
0.3 C
% 0'2 l .. . (LII3
: :s
'o 0 U
(7) 0.1 ' 0 17:
a) o ' m
a: O a:
1: . o 0.0 . O E
§ -0.1 - o 0 Q
'0 2 f v T r . '0.2 l 1 r I l r
2 3 4 5 6 7 8 -4 -3 -2 -1 O 1 2 3
Simpson's Diversity (D) (2 km) Principal Component 2 (2 km)

Figure 5.4. Partial residual plots illustrate the relationship between the mean weekly
Catch of all native Coccinellidae (A and D), H. convergens (B and E), and C. maculata
(C and F) from yellow sticky card traps and the best ﬁt model Diverstiy+PC2 at 2 km

determined using an AICc model selection procedure.

156

CHAPTER 6

NATIVE COCCINELLIDAE ARE SUBJECT TO SIGNIFICANT EGG

PREDATION [N SOYBEAN FIELDS.

Abstract

Intraguild predation (IGP) is a major factor shaping predator communities.
Assemblages of predatory coccinellids are frequently important in suppressing herbivores
in agroecosystems. However, in the past several decades, coccinellid communities across
North America have been invaded by exotic coccinellids resulting in dramatic declines in
some native species. I used the coccinellid community attacking soybean aphid in
soybean in the north-central US. to test the hypotheses that IGP by exotic competitors is
a major factor shaping native predator communities across differing agricultural
landscapes. Previous studies have shown that coccinellid communities vary across the
region and are related to the composition of the landscape surrounding soybean ﬁelds.
My objectives were 1) to determine if observations of reduced native coccinellid
abundance in Michigan were explained by greater IGP by exotic species and 2) to
ascertain whether the area of forested patches across the study region was related to the
amount of IGP incurred by native coccinellids. I measured the intensity of native
coccinellid IGP in soybean ﬁelds in Michigan and Iowa using Coleomegilla maculata as
a sentinel species and found that IGP occurred in 43% of eggs exposed to predators
within 48 h. The majority of this predation is attributable to diurnally active predators.
The intensity of C. maculata egg predation was consistent across both states; however,

the composition of the potential egg predators varied. In Michigan, the exotic

157

coccinellids C. septempunctata and H. axyridis were the most abundant predators
whereas in Iowa, the majority of those found were native species including C. maculata,
Hippodamia parenthesis and the soft-winged ﬂower beetle Collops nigriceps. Thus, in
Michigan landscapes, consumption of native coccinellid eggs favors exotic coccinellids
whereas in Iowa, this resource is more likely to sustain other native species. This study
illustrates that it is not egg IGP per se but rather the overall composition of the IGP
community in soybean that is contributing to the reduced native coccinellid community in
Michigan versus Iowa.
Introduction

Following successful 1889 introduction of the vedalia beetle, Rodalia cardinalis
(Mulsant) into the U.S., over 100 species of exotic lady beetles have been either
intentionally or accidentally released into North America (Harmon et al. 2007). Although
many have either not established or are present at low densities, a few have become
extremely abundant in agricultural habitats. In the north-central U.S., the two most
prevalent species are Coccinella septempunctata L. and Harmonia axyridis Pallas.
Although these exotic coccinellid species can be beneﬁcial in the control of several
agricultural pests (Brewer and Elliott 2004, Snyder et al. 2004b, Costamagna and Landis
2007a, Gardiner et al. in press), increasing evidence suggests that their presence has
resulted in the displacement of native coccinellid species (Putnam 1955, Wheeler and
Hoebeke 1995, Elliott et al. 1996, Colunga-Garcia and Gage 1998, Michaud 2002,
Snyder et al. 2004a, Harmon et al. 2007, Gardiner et al. in press). Understanding the
factors that structure coccinellid communities is an important topic of both biological

control and conservation science.

158

Both C. septempunctata and H. axyridis were originally introduced into the US.
for biological control; however, it remains unclear whether their eventual establishment is
due to intentional or accidental introductions (Schaefer et al. 1987, Day et al. 1994).
Coccinella septempunctata was ﬁrst detected in the US. in 1973; H. axyridis was
detected 15 years later in 1988 (Schaefer et al. 1987, Chapin and Brou 1991). Recently,
two additional exotic coccinellids, Propylea quatuordecimpunctata (L.) (Gardiner et al.
in review) and Hippodamia variegata (Goeze) (Gardiner and Parsons 2005) have been
detected in agricultural ﬁelds in the north-central U.S.

Over the last three decades, the food webs of native coccinellid species have
undergone signiﬁcant restructuring. Not only have additional competitors been added,
but new prey species have also been introduced, including the Russian wheat aphid,
Diuraphis noxia (Mordvilko) and the soybean aphid, Aphis glycines Matsumura. Exotic
coccinellids have been shown to be important predators of these exotic pests, and in turn
the presence of these prey may have allowed exotic species to increase in abundance and
expand their ranges.

The success of exotic coccinellids has been correlated with a decline in several
native coccinellids in the US. One of the most dramatic declines is that of Coccinella
novemnotata Herbst (Wheeler and Hoebeke 1995). Once common in the northeast, only
10 specimens have been collected in North America in the last 10 years (Losey et al.
2007). Wheeler and Hoebeke (1995) attribute this loss to the introduction of C.
septempunctata. A similar pattern was documented in South Dakota by Elliott et a1.
(1996) who observed a 20—fold reduction in populations of Coccinella transversoguttata

richardsonz' Brown and Adalia bipunctata (L.) following the establishment of C.

159

septempunctata. With the establishment of H. axyridis came additional reports of native
coccinellid decline. Colunga-Garcia and Gage (1998) reported a decrease in populations
of Brachiacantha ursina (F.), Cycloneda munda (Say), and Chilocorus stigma (Say) in
Michigan. More recently, Gardiner et al. (in review) failed to detect Hippodamia
convergens Guérin-Méneville across two years of season-long sampling in Michigan,
despite it being collected by previous investigators in recent years (Fox et al. 2005,
Costamagna and Landis 2007a).

Gardiner et al. (in review) measured coccinellid diversity and abundance in
soybean ﬁelds across Iowa, Michigan, Minnesota, and Wisconsin and found that the
abundance of native species varied signiﬁcantly across this region, from a low of 10%
natives in Michigan to 44.8% natives in Iowa. They also found that the structure of the
coccinellid community was tied to the composition of the landscape surrounding soybean
ﬁelds. Landscapes with an abundance of forested habitat had the greatest proportion of
exotic coccinellids while landscapes with an abundance of grasslands supported higher
populations of native species.

Several studies have attributed native coccinellid decline to IGP by exotic species.
While the propensity of exotic coccinellids to act as intraguild predators has been well
documented (Cottrell 2004, Snyder et al. 2004a, Cottrell 2005, 2007, Gardiner and
Landis 2007) the intensity of egg predation experienced by native coccinellids had not
been measured at a regional scale. My goal was to determine if the pattern of native
coccinellid abundance in Michigan and Iowa illustrated by Gardiner et al (in review) was
explained by differences in the extent of egg predation incurred by native coccinellid

populations in these states. My hypotheses were that 1) egg predation would be greater

160

in Michigan soybean ﬁelds due to a greater abundance of exotic coccinellids compared
with Iowa and 2) landscapes with extensive forested habitat within both states would
support greater egg predation than grassland or cropland dominated landscapes.
Methods

Field sites

During July of 2007, I measured the intensity of predation on sentinel egg masses
of C. maculata in n=14 soybean ﬁelds in Michigan (n=8) and Iowa (n=6) (Figure 6.1). A
minimum distance of 10 km separated each ﬁeld site. Field size averaged 12.5 ha (range:
3.0-60.4 ha). Within each state, soybean ﬁelds were located within landscapes which
ranged from very simple landscapes comprised primarily of cropland to complex
landscapes with an abundance of natural habitats. Within each soybean ﬁeld, four 0.2 ha
plots were established where all sampling took place. Soybean ﬁelds were not treated
with insecticide during the study period.
Egg mass collection

Beginning in May 2007, C. maculata adults were ﬁeld collected from grasslands
and alfalfa ﬁelds in Iowa using sweep nets. Females were placed individually into Petri
dishes with strips of paper to serve as an egg-laying substrate. Beetles were provided
daily with water, honey and eggs of the corn earworm, Helicoverpa zea. Dishes were
checked daily for C. maculata eggs and any found were frozen (- 80 °C). I determined
that 15 eggs per mass was the average size deposited by C. maculata under these
conditions and therefore used egg masses of this size in the ﬁeld experiment. To prepare
sentinel egg masses, individual eggs were either added or removed ﬁ'om existing masses

using a paint brush, to reach n=15 eggs per strip. Added eggs were attached to the paper

161

strip using water soluable glue (Elmer’s Products, Columbus, OH). The standardized egg
masses were cut out and glued onto ﬁlter paper disks (12 mm diameter). Egg masses
remained frozen until they were transported to the ﬁeld.
Measuring predation of frozen egg masses

Prior to ﬁeld experiments, I compared predation of previously frozen and live C.
maculata eggs by four commonly found intraguild predators: C. septempunctata, H.
axyridis, H. parenthesis and Nabis sp. Individual predators were released into a Petri
dish arena containing three thawed and three ﬁesh eggs, randomly arranged in a grid
pattern. All arenas were checked hourly for 8 h then again at 18 and 24 h and the number
and type of eggs consumed by each predator were recorded. Ten replicates were
completed for each predator species. None of the species tested showed signiﬁcant
preference for either thawed or live eggs (24 h: C. septempunctata P=O.4595, H. axyridis
P=1.000, H. parenthesis P=0.6132, Nabis sp. P=O.1480).
Egg predation experiment

To begin the experiment, two ﬁlter paper disks containing egg masses were glued
onto the top side of a soybean trifoliate leaf approximately 15 cm ﬁom ground level on
adjacent plants in the center of each plot. The exclusion egg treatment consisted of one
egg mass enclosed in a 22 cm clip cage to prevent predators from accessing the eggs.
The open egg treatment consisted of an un-caged egg mass open to predation. Both
treatments remained in the ﬁeld for 48 h after which they were collected and examined to
determine the number of undamaged eggs remaining. After 48 h of exposure in the ﬁeld
40% of eggs in the caged treatment were partially collapsed, mimicking damage that

could be caused by piercing sucking predators. To eliminate potential confusion I only

162

considered eggs to be predated-upon if there was clear evidence of damage ﬁom a
chewing predator. Thus, my measure of egg predation does not account for losses due to
piercing sucking predators and is thus likely to be conservative compared with the ﬁeld
reality.
Nocturnal egg predation experiment

To account for what proportion of egg predation was due to noctumally active
predators, I conducted a nocturnal predation experiment in eight sites (four per state).
Open and caged egg mass treatments were placed within the center of each plot on
adjacent plants at dusk 2000-2200 h and removed at dawn 0500-0700 h and the number
of undamaged eggs recorded.
Predator and soybean aphid survey

During the 48 h egg predation experiment, I measured the activity of potential egg
predators by placing one unbaited yellow sticky card (PHEROCON AM, Great Lakes
IPM, Vestaburg, MI) in the center of each plot. A metal “T” fence post was erected and a
22.9 x 27 .9 cm sticky card was suspended just above the plant canopy. Sticky traps
remained in the ﬁeld for 48 h. All potential egg predators were counted and identiﬁed to
species. I also measured the abundance of predators by collecting a 20—sweep sample
from two rows of soybean plants in each plot. As the abundance of prey may effect the
intensity of egg predation, I also measured soybean aphid populations within each ﬁeld
site using destructive plant counts. In each plot, ﬁve randomly selected plants were
removed from the ground and the number of apterous and alate A. glycines were counted

on each plant.

163

Landscape analysis

Field geospatial data were collected using a handheld GPS receiver using Wide
Area Augmentation System (WAAS) correction. The spatial coordinate (WSG 1984) for
the center of each site was used to obtain ortho-rectiﬁed digital aerial imagery for the site.
I digitized the habitats surrounding each study site to a radius of two km using ARC GIS
9.1 and conducted ground veriﬁcation of each landscape from July-August 2007. Each
landscape polygon within two km of the center of the study sites was given a value
corresponding to one of seven landscape categories: corn, soybean, other crops, forest,
grassland, wetland and open water, and urban. Some locations included polygons that
were not visible from a roadway and permission to access private lands could not be
obtained. These polygons were given a value of zero and were excluded from further
analysis. The area of each site that could not be identiﬁed was very low (<1%). The
smallest polygons identiﬁed included ﬁeld plots on university research farms and small
patches of abandoned crop ﬁeld (< 5 ml) the largest were contiguous forests, grasslands
and crop ﬁelds (5 1.3 kmz).

Landscape diversity was measured using Simpson's Index (D) (Simpson 1949).
Simpson’s Index is typically used to examine the variance of species abundance
distributions, here I applied it to examine variance in the proportion of area covered by

each of seven land cover categories. The equation for Simpson’s Index (D) is: D =
1/Z(pi)2 where pi=proportion of habitat in the ith land-cover category (D increases as

diversity increases).

164

Statistical analysis

For both egg predation experiments, a split-plot mixed effects analysis of variance
model (ANOVA) was used to determine if a signiﬁcant number of C. maculata eggs
were removed by predators. This model included Treatment and State as ﬁxed effects
and a State by Treatment interaction. Random effects were Site nested within State and
Plot nested within Site and State.

A log likelihood chi-square test assuming a multinomial distribution was used to
examine the variation in the community of potential egg predators in Michigan and Iowa
soybean ﬁelds (Shao 2003). This test determines if the species composition of the
coccinellid community within each state is signiﬁcantly different. The null hypothesis
for this test was that the proportional distribution of predator species within Michigan and
Iowa did not differ. This test was completed using compiled species data from sticky
cards and sweep net samples. Both the ANOVA models and log likelihood chi-square
test were completed using SAS v. 9.1 (SAS Institute 1999).

To asses the inﬂuence of landscape on the intensity of egg predation, four models
were compared using Akaike’s Information Criterion, adjusted for a small sample size
(AICc) (Bumham and Anderson 2002). These included an intercept only model and
models containing the variables; proportion forested habitat, proportion grassland habitat,
overall landscape diversity (Simpson’s D) and extraguild prey abundance (Table 6.1).

The maximum log-likelihood estimate, the Akaike weights (estimate the relative
likelihood of a given model against all other models) and AICc differences (Ai) are

presented. The model with the minimum AICc value is considered to have the best

support for the data. Any model with a A, of less than two is termed a “competing

165

model” (Ribic and Sample 2001) and is also considered to also have substantial support
for the data (Bumham and Anderson 2002). Once the minimum AICc and competing
models were determined, I also examined the relationship between these models and the
average number of potential egg predators collected per 20-sweep sample in soybean

ﬁelds. For each model I calculated the r2 to evaluate how well the models explained the

variation in the data. The AICc analysis was conducted using R version 2.1.1 (R
Development Core Team 2005). The average number of predators per 20-sweep sample
was log (x+1) transformed prior to analysis.
Results

Egg predation experiment

Egg masses of C. maculata were subject to signiﬁcant intraguild predation in
soybean ﬁelds across Michigan and Iowa after 48 h of ﬁeld exposure (F1,54=45.73,
P<0.001) (Figure 6.2). In the open treatment, 43% of the C. maculata eggs were
removed. Predation events were recorded on 60.7% of egg masses, with an average of
4.4 eggs remaining aﬁer 48 h. There was no signiﬁcant difference in the amount of

predation incurred between the states (State*Treatment interaction: F1,54=3.18,

P=0.0804).
Nocturnal egg predation experiment

Predators removed 6.2% of the available eggs during the nocturnal predation
experiment (experiment run for 9 h, 2000-2200 h to 0500-0700 h). There was no

signiﬁcant difference between the number of eggs remaining in caged and open

treatments (Fl,30=1.83, P=O.l725) (Figure 6.2). Aﬁer exposure, the average number of

166

eggs remaining in open egg masses was 13.9 :1: 0.6 compared with 14.8 i 0.1 in the
caged treatment.
Aphid and predator populations

Aphid populations in soybean ﬁelds during the egg experiment were low across
the 14 sites, varying from 0 per plant (Monroe and Britton M1) to 21.4 i 4.5 per plant
(Arnes, IA). In both states, low numbers of chewing predators were detected averaging
0.3 :t 0.1 and 0.7 :t 0.1 per 20 sweeps in Iowa and Michigan respectively (Table 6.2).
Chewing predators on yellow sticky cards averaged 0 per card in Iowa and 1.6 :i: 0.2 in
Michigan. Six species of potential egg predators were collected: H. axyridis, C.
septempunctata, P. quatuordecimpunctata, C. maculata, H. parenthesis and the soft-
winged ﬂower beetle Collops nigriceps (Say) (Melyridae) (Table 6.2). The species

composition of these communities in Michigan and Iowa soybean ﬁelds were
signiﬁcantly different (X25=28.28, P<0.0001) (Table 6.2). In Michigan, the community

was dominated by the coccinellids H. axyridis and C. septempunctata. In Iowa, C.
nigriceps comprised 50% of the predator community followed by C. maculata, H.
parenthesis and H. axyridis which each accounted for 16.7%. As C. nigriceps was not
included in preliminary testing, I subsequently measured its potential as a predator of C.
maculata. Four specimens were placed into individual Petri dishes each with a thawed C.
maculata egg mass. Within 3 h, 3 of 4 C. nigriceps consumed all 15 eggs, in the fourth
replicate the beetle did not consume any eggs after 24 h.
Landscape diversity

Within the 2 km landscape radius surrounding each of the 14 sites, landscape

diversity values (Simpson’s D) ranged from 2.12-5.31. The percentage of the landscape

167

composed of cropland ranged from 15-86%. Landscapes with high and low percentages
of cropland occurred in both Michigan 18-95% and Iowa 23-88%. Forested habitat
comprised from 0.04-22% of the landscape in Iowa sites and 2-21% of Michigan
landscapes.
AICc analysis of egg predation and landscape variables

The diversity model predicting the abundance of eggs after 48 h had the lowest
AICc value (Table 6.3). There was a marginally signiﬁcant (P = 0.0774) negative
relationship between D and the number of eggs removed in the open treatment after 48 h
(Figure 3). This indicates that the intensity of egg predation in soybean ﬁelds is slightly
higher in simple, agriculturally-dominated landscapes compared to diverse landscapes
with an abundance of non-crop habitat. The intercept model was a competing model with
a A, of 1.33, indicating the intercept alone provides nearly equal support to the data
compared with the predictor Simpson’s D (Table 6.3). Predator abundance in ﬁelds was
positively correlated with Simpson’s D, the average number of predators found in 20-
sweep samples was greater in diverse landscapes with an abundance of forests and
grasslands.

Discussion

Over the last 20 years several studies have documented declines in native
Coccinellidae populations in the US. Many have attributed this to predation of native
eggs and larvae by exotic coccinellid species, primarily H. axyridis and C.
septempunctata. In this study, I measured the intensity of intraguild predation of native
coccinellid eggs within soybean ﬁelds in two states that vary in the structure of the

coccinellid community. Michigan communities were composed of 10% native species

168

versus 44.8% native species in Iowa (Gardiner et al. in review). This difference has been
linked to the composition of the landscape surrounding soybean ﬁelds, with landscapes
dominated by grasslands supporting higher native coccinellid populations while forested
landscapes support a greater abundance of exotics. My hypothesis was that the reduced
amounts of forest habitats in Iowa would limit the success of exotic coccinellids and thus
the intensity of IGP, leading to a greater abundance of native species. In contrast, I found
signiﬁcant native coccinellid egg predation in soybean ﬁelds throughout both Michigan
and Iowa. Although the extent of egg predation detected was consistent across the states
studied, the guild of egg predators exploiting this resource varied. The community of egg
predators in Michigan was dominated by the exotic coccinellid species whereas the Iowa
community was comprised primarily of native predators.
Egg predation of C. maculata

This study illustrates that the intensity of C. maculata egg predation in soybean
ﬁelds is high, with predators consuming 43% of available eggs within 48 h. This is the
ﬁrst account of egg predation of a native coccinellid, illustrating that egg predation likely
plays a signiﬁcant role in its population dynamics. However, depending on the landscape
context of any given ﬁeld, the community of predators consuming native coccinellid eggs
will vary. Since exotic species were the most abundant coccinellids collected from
Michigan soybean ﬁelds both in this study and previous studies (Gardiner et al. in
review) they are likely to have contributed signiﬁcantly to the predation of C. maculata
eggs. Both laboratory studies and ﬁeld observations illustrate that H. axyridis and C.
septempunctata larvae and adults are signiﬁcant predators of native coccinellid eggs and

larvae (Cottrell 2004, Snyder et al. 2004a, Cottrell 2005). These exotic species are more

169

likely to act as intraguild predators of native species, rather than intraguild prey (Cottrell
2004, Snyder et al. 2004a, Cottrell 2005, Yasuda 2006). In addition, H. axyridis is able
to complete its development on a diet of native coccinellid eggs, while native coccinellid
larvae cannot reach adulthood on a diet of H. axyridis eggs (Cottrell 2004). The presence
of H. axyridis has also been found to result in interference competition with
heterospeciﬁcs, resulting in reduced fecundity in native species (Soares and Serpa 2007).

Whereas exotic coccinellids were the most abundant predators in Michigan
soybean ﬁelds, the potential egg predator community in Iowa was composed primarily of
native species. Although it is likely that exotic species play a signiﬁcant role in the
intraguild predation of native coccinellid eggs throughout their invaded range, it is likely
that native species also consume coccinellid eggs. Several native coccinellids are known
to consume the eggs of other coccinellids as well as conspeciﬁcs (Cottrell and Yeargan
1998a, b). I also determined that the melyrid C. nigriceps found in Iowa soybean ﬁelds
fed on C. maculata eggs.
Contribution of nocturnal egg predators

To accurately identify predation pressure on a speciﬁc prey, it is critical to
account for activity of both diurnal and nocturnal natural enemy species (Pfannenstiel and
Yeargan 2002). I found that 6.2% of eggs exposed to nocturnal predators were preyed
upon in a single night. Assuming equal predation for both nights of the study, I estimate
that 12.5% of available eggs in the 48 h egg predation study were consumed at night, or
approximately 30% of the total eggs consumed. Other studies in soybean have also
found signiﬁcant nocturnal predation. Pfannenstiel and Yeargan (2002) observed that C.

maculata larvae, Phalangiidae, Clubiona abbotii (Clubionidae), Elateridae, Carabidae,

170

and Lygus lineolaris (Lygeidae) were all noctumally active predators of lepidoptern eggs
in corn and soybean ﬁelds. As I only recorded damage from chewing predators, I did not
account for potential predation by spiders, Opiliones, or true bugs. Of the remaining
chewing predators found by Pfannenstiel and Yeargan (2002), only C. maculata larvae
are known to consume C. maculata eggs. Pfannenstiel and Yeargan (2002) report that C.
maculata adults were diurnal feeders while larvae remained active at night.
Local landscape eﬂects on C. maculata egg predation

A marginally signiﬁcant negative correlation was detected between landscape
diversity (D) and the number of C. maculata eggs consumed within 48 h. At a spatial
scale of 2 km, soybean ﬁelds in low diversity agricultural landscapes experienced greater
C. maculata egg predation compared with high diversity landscapes with an abundance
of non-crop habitat. My hypothesis was that egg predation would increase in soybean
ﬁelds in diverse landscapes as these landscapes would supply a greater number of
potential egg predators. Although potential egg predator abundance did increase with
landscape diversity, egg predation declined. It is possible that the predator species found
in simple, agricultural landscapes were more likely to consume native coccinellid eggs
than those supplied by diverse landscapes. Alternatively, although the abundance of
soybean aphid did not impact egg predation in soybean ﬁelds, diverse landscapes may
have a greater abundance of total available prey, reducing the likelihood of egg predation
in the open egg treatment in ﬁelds in these landscapes.
A new hypothesis for native coccinellid decline

This study investigated one potential mechanism to explain the differences in

coccinellid community composition found by Gardiner et al. (in review). I investigated

171

the hypothesis that lower populations of exotic coccinellids in Iowa soybean ﬁelds would
result in reduced intraguild egg predation of native coccinellids, sustaining native
communities. However, the data support a different hypothesis for greater exotic
dominance in Michigan compared with Iowa. 1 found that it was not the intensity of egg
predation in soybean that varied, but the community of likely intraguild predators. I
propose a landscape-mediated IGP hypothesis, where the abundance of a coccinellid
species is driven by landscape structure and perpetuated by within-ﬁeld interactions
between guild members (Figure 6.4). For example, the abundance of exotic coccinellids
in soybean ﬁelds is positively related to the amount of forested habitat within the
landscape. Therefore, exotic species are likely to dominate coccinellid communities in
soybean ﬁelds in forested landscapes. In these exotic-dominated food webs, native
coccinellid eggs are likely to contribute to the maintenance of exotic populations.
Alternatively, landscapes with an abundance of grassland habitat support higher
populations of native coccinellids in soybean ﬁelds. In food webs with a greater
proportion of native species, intraguild predation of coccinellid eggs is likely to beneﬁt
both native species and exotic competitors. Therefore, a greater proportion of the
resources from native coccinellid eggs will build and sustain native food webs in
grassland-dominant landscapes (Figure 6.4).
Summary

Gardiner et al. (in review) were the ﬁrst to illustrate that patterns in the abundance
of native and exotic lady beetles are affected by landscape composition. Although
Gardiner et al. (in review) demonstrated that native and exotic coccinellids populations in

soybean ﬁelds are favored by different types of landscapes, it was unknown if

172

competitive or predatory interactions between these species caused this partitioning. Two
scenarios were discussed: 1) native species may be adapted to more open savanna and
prairie landscapes that formerly dominated large areas of the north-central US. and thus
still favor landscapes with an abundance of grassland habitat. However, this did not
explain the relatively recent decline in native coccinellids within the last few decades,
long after landscapes were transformed by human settlement. 2) Alternatively, it may be
that these native species are still thriving in landscapes that are less favorable to invasion
by exotic coccinellids, which act as strong intraguild predators. Since exotics are favored
by forested habitat, a lack of these landscape elements may have restricted H. axyridis
dominance of coccinellid communities in these areas favoring persistence of native
coccinellid communities.

This study illustrates that it is not the intensity of IGP that is contributing to the
decline in native coccinellids but more likely the structure of the predator guild
consuming this resource. In landscapes that favor exotic coccinellids, native coccinellid
eggs are more likely to be consumed by exotic predator guilds. This is likely contributing
to the decline in several native coccinellid species in forested states such as Michigan
were populations of B. ursina, C. munda, C. stigma, and H. convergens have noticeably
declined (Colunga-Garcia and Gage 1998, Gardiner et al. in review). Alternatively,
despite overall equal egg predation in Iowa, the presence of fewer exotic competitors is
likely to favor the persistence of native coccinellid populations and could protect native
biodiversity.

Although it may be impossible to remove established exotic coccinellids from the

environment, we can protect native biodiversity by protecting and promoting landscape

173

attributes which favor native coccinellid populations. Grasslands appear to be very

important in supplying native coccinellids to soybean ﬁelds (Gardiner et al. in review)
and thus should be a high conservation priority to preserve these species. Further, the
majority of work on exotic and native coccinellids has focused on their populations in
agricultural ﬁelds (Snyder and Evans 2006), which likely function as sink populations
supplied by surrounding natural habitats. Future work must focus on understanding

interactions between source populations of native and exotic coccinellids, in grassland

and forested habitats.

174

>880

8080008 30 8 wv 08 886 0880.8 0880 20m 8000 883 808 808 08880 800080 .80 800.808 0w080>0 08 8 >088
>80w0800 89/00680— 88 05 8 000308 .«0 888080.8qu v80 «GEN: u G 08083 38080005 0.800985 N 31.80005 8

N

 

8080008 ww0 05080000 08008 808008
808880 :83 00800850 >08 3880.880 8032 0 :8? 020m 800£>0m

.0208

8000500 8 8080008 wwo 05080000 08008 808008 8 8508 EB 083.
0808080 08 085080000 08080 080 0,808 >385 808 0808308 8080
.808 .00 00800850 80000.8 0 00808 8. >005 00 008000088 0808a

88800.8 wwo 82080000 08008
800008 808880 :8? 020m 800£>0m 8 8880008 m0 808.8808 8mg
005080000 03808 .80 080808808 888 808880 8083 0080008804

80800088 ww0 05080000
0808 80000.8 808880 :8? 020m 800£>00 8 90080.8 mo 80880808 8mg

«>055 + cm H >
082 88

388208 + .8 u .2
082 0828

088808 + .8 u 0
.082 0588.8

80080m~m + 0% H >

 

005080000 08080 .80 080808808 888 808880 8083 0080000808 .0802 00.0.3.8
.mEom em .I. >

8000>00 .8 8080008 ww0 05080000 08808 .«0 88080 08 8088 a
008 00 008008.50 >08 80 003083 08000880: 8008233 :8 Z _0802 0 0080.5
88508.8 .082

 

175

00030.8 080 8 #0008 8000 88008 80.8 80088088 8.8.088 0 08:. A0083 080 08800 =080 0 80.8. 0058.500 803080 0?. w808
80808800 0803 20008 05. >08 2880880 .80 >EB0=0>0 08 080 800508 0880008 880 «0080.8 .80 00808850 >880>€ w8>80> m0

0080000808 8 008002 0208 800880 032 080 88.8082 8 8080008 30 .80 808008 08 80880 00 20008 0880808 03.n— 80 0308.

 

 

 

 

 

 

 

176

 

00.0 0 03 0 000 0 .500.
00.0 um 00.0 0 Z 0 08.0288. .0
00.0 a 00.0 0 . 0.00 0 33000050000 .0
00.0 a 00.0 0 0.8 0 00008 .0

0080082 0308 003003 0308

200 a 082 0000. 00 0030800

800,—. .3085 30:0?

000 a 00.0 00.0 a 00.0 000 000 .300.
0 00.0 a 2.0 0.0 0.00 088.200. .0
00.0 3.0 a 3.0 0.0 000 0.803808 .0
0 0.0 0 0.0 0.0 0.00 3383. 0
00.0 a 00.0 0 0.0 0.0 03008500085000 .8
:0 a 00.0 0 :0 0.0 08000850000 .0
00.0 a 00.0 00.0 0 3.0 0.00 000 000.08 .0
0080002 0308 0080002 0308 008008m 03000.5

200 a 082 0000 00 0030800

002 80005

 

600800800 00000008 30 0 wv 08
w883 0 ww .008 0.00 08 8 008088 000 >80000 000—8 08 0>0n0 00.0 00008 0.003 00000 >800 30:0? 0808 000p>00 80 0300 00B
80 088800 80000.8 0 80 00000000 0088800 802% .0808 000 5800.0 30:0> 000 0088800 80030 ”0080008 858800 003 0800 00800

0000.>00 0080082 000 0308 8 00008 00000008 80 82mm H 00000850 0008 000 38008800 00000008 00000 08 80 0.000.008 N0 0308.

.3950 “nu—Q 333.53 mam: @05ng 3:25.593

donavoa wwo a 3 05 .«o 08: 05 8 mEom ”Boga S wmﬁobm .3me Bags 5038 we 8592K? omﬂguu >95 v
$.22.“ So can 339E 689$.— mug—83mg wag—:03

4333 van—mama .«o 389:8 Bow 989mg €50.55 3:3.— 5 N a 5.5» &§§_ .«0 «890905 .I. @5635
4332 3380.“ m0 @3038 20m 533m maﬁa—8.5m 3.38 Eu— N a 55 ogaoqua— mo “Bahama u awoken m

bowoﬁo 550653 a: 08 5 .352 .«o 8:585qu 98 «QENE n G 80:3 .0335 Egan—mm M Q N

noumvoa 30 .33 awn—macaw?“ gamma: a 33:93 3858qu E maﬁa—S _

 

 

86. w; So 33% m Ed? .955 + om u .A
8.? mi 85 38 m and: megaeova + cm H x
86. won 86 Emma m 3%. mcmeomrm + on n z
«.2 ma... 3.: N «E... can .m
S... 8... as 8% m 3.3.. "again—VS + S n »
NH a E 8?. M wooﬁoé 2252
33.32 .34

 

.28 5 825m 30on mnuomﬁoo ES 3on 00?. 82538 BC.

.SwwEBE can «32 5 35¢ 509mg E 898on wwu mo 336:: 05 wags—go 5m moumwﬁm 8:023 3on mo Egm do 2an

177

Figure 6.1.

 

Figure 6.1. Location of 14 soybean ﬁelds in Michigan and Iowa where C. maculata egg
predation was measured during July of 2007. The distance between each location was 10
km or greater. Sites were selected so that the 2 km landscapes surrounding each ﬁeld

varied in diversity and the abundance of natural and agricultural habitat within each state.

178

Figure 6.2.

 

 

 

 

 

 

 

16 a
m 14 ‘
3A
3.“: 12
‘93 10
E g? 8 DExclusion

E

g a 6 .Open
2 2

0

48h Noctural

Figure 6.2. Mean number of eggs remaining in the exclusion cage treatment (white bars)
and open treatment (black bars) in predation experiments. In the 48 h experiment,
treatments remained in soybean ﬁelds for 48 h, while nocturnal experiment treatments

remained in ﬁelds for 9 h.

179

Figure 6.3.

16 ' 1.5

.—n

N
T

0

Mean C. maculata eggs
consumed (48 h)
b. 00

Mean predators per 20-sweeps

 

 

 

   

Simpson’s Diversity D Simpson’s Diversity D

Figure 6.3. Relationships between landscape diversity (D), the number of C. maculata
eggs consumed after 48 h of exposure to predators in soybean ﬁelds, and the average
number of egg predators collected per 20-sweep sample of soybean foliage. Landscape

diversity was estimated from 7 land-use categories using the Simpson’s Diversity Index
(D): D = l/Z(pi)2 where pi=proportion of habitat in the ith land-cover category (D

increases as diversity increases). The mean number of predators per 20-sweep sample

was log (x+1) transformed prior to analysis. Untransforrned means are shown here for

interpretation.

180

 

0:080:33 3300:: m0 ﬁwgbm 0000000:

05 0000203 00:: 000.00 .30 505» mg 00:30: 0000005 00:: 00000 0:03 0:050:38 3090030005 30000:: 00:: 020m .903

000.3 05 :02? 0000000 .30 00:00::30 03020: 06 00000003 008 2000 .003 000.0 0>0n0 05 :H 0:002:00: .005 wﬁﬁﬁmsm n3030000:
0330: >0. 00020: 00' 0: 30M: 0:08 0:0 www0 03:0: 8:080:00 0030:0080 0:07.me E 030 05050000 0300: .«0 50.8005 003000w 0
0:50:00 020 0:0 3030005 23035 ”00:30: m0 28w 05 mo 50.8000: 50003 0 >038 00 £005 000:: 00 000000 .0003 .0000000:0_
00:03:80 300000 3 0200 :00980 5 000.008 :80 00:6 038% 00:: 00 0:000:50: 05050000 0.50: 005 00000:: 00:00::50
05030000 0330: :00000m 0210305 0:0Em00w .«0 00:00:20 :0 £05 000000005 5 020m 509.30 2.33 Agwﬁouzv 005050000
00008 %n 0000:0080 0:03 000000053 000.0000.“ 3 020m :0098m 3:080? 0:080:03 3:000.m:0 605 003080000 0003 0200 :005080

E 0:003:00: 05030000 00000 0:0 03.0: «05 0:000 $0.50: 05 .00 30 03000 0605093 :9 00300080000032 0E. .00 03E

181

a..£:.£...(§ .11]. .2....
0.50.4834... L? .4 t .9 «.4....
.611... (Aft?)

ﬁlﬂftﬁb... 00303000 3 0005030000 m
: 0000.53 003050005 ,0

     

       

           

 

0 u. 0
u. 0 of. n. 00 380 00
mm. 0.0 . . 00000 00000003 00030: 500m
. o . ,. . .u. x .0
0 an o o M
u 0 .
.. .m
u r 0 o.
. 00 00
o o»
n. a
0 0
0hr 0
I 0.0
0 00
o 00
n a
o. n.

182

CHAPTER 7

CONCLUSIONS AND FUTURE DIRECTIONS

The soybean aphid, Aphis glycines Matumura, is an invasive species ﬁrst detected
in the US. in 2000. This pest has now spread throughout the north-central soybean
growing states and Canadian provinces. Previous work has demonstrated that natural
enemies are key drivers of soybean aphid population dynamics within soybean ﬁelds
(Fox et a1. 2004, Rutledge et al. 2004, Fox et a1. 2005, Costamagna and Landis 2007a,
Costamagna et al. 2007a, Costamagna et al. 2007b, Costamagna et al. in press). The
objectives of my dissertation research were to investigate how landscape diversity and
composition inﬂuenced the species pool of potential biocontrol agents supplied to
soybean ﬁelds and determine how within-ﬁeld interactions between these species impact
biocontrol services. I investigated these objectives in replicate sites across Iowa,
Michigan, Minnesota, and Wisconsin from 2005-2007.

Overall, I found that both landscape composition and within-ﬁeld interactions
affected the diversity, abundance, and biocontrol services provided by the predator
community in soybean ﬁelds. The distribution and abundance of semi-natural and
agricultural habitats within the landscape greatly inﬂuenced the abundance of key
predator species including Coccinellidae, Carabidae, Araneae and Opiliones. While I
expected all taxa to increase with landscape diversity and the amount of semi-natural
habitat present, I found variation in the response of the study organisms based on their
trophic level, phenology, and whether they were native to the north-central US. The

activity density of Araneae increased in soybean ﬁelds in landscapes with an abundance

183

of forests and grasslands. Relationships with Carabidae activity density varied, with
adult overwintering and predatory species increasing and larval overwintering species
declining with the abundance of grassland in the landscape. The coccinellid community
also exhibited varied responses to landscape diversity and composition. I found that
native coccinellids were most abundant in low-diversity landscapes with an abundance of
grassland habitat whereas exotic coccinellid increased with the abundance of forested
habitats. The biocontrol services supplied to soybean ﬁelds by these species is a function
of landscape diversity, with diverse landscapes supplying greater biocontrol service
compared with simple landscapes dominated by agricultural habitats.

My studies illustrated that landscape diversity and composition were key in
supplying a species pool of predators to provide biocontrol services, and that within-ﬁeld
interactions that occurred between these species inﬂuenced their community dynamics.
Exotic coccinellids acted as signiﬁcant intraguild predators of native coccinellids,
lacewings and predatory ﬂies. Despite this IGP, the presence of H. axyridis and C.
septempunctata had a positive impact on soybean aphid biological control.

Biocontrol services at the landscape scale

My work illustrates that to understand biological control we must consider the
inﬂuence of both landscape and within-ﬁeld community dynamics. Landscapes contain
source and sink populations of natural enemies. My data show that semi-natural habitats
such as forests and grasslands are positively correlated with species abundance and
biocontrol services in soybean, thus these habitats are likely to serve as source
populations which supply organisms to agroecosystems. As landscape diversity has a

signiﬁcant positive correlation with biocontrol services, maintaining overall landscape

184

diversity, and especially the presence of perennial grassland and forested habitats in the
landscape, will favor suppression of soybean aphid.

However, to advance biological control of A. glycines, we must also begin
examining the landscape distribution of the primary host plant of this pest, buckthom
species in the genus Rhamnus. Rhamnus cartharica and R. frangula are widely
distributed exotic shrubs which invades forest edges, ditches and fencerows (Voegtlin et
al. 2004, Voegtlin et al. 2005). These plants are more abundant in northern sections of
the invasive range of A. glycines. Since its detection in 2000, soybean aphid has often
been detected and reached threshold earlier in the season in areas with an abundance of
buckthom. Soybean ﬁelds in southern Ohio, Illinois, Indiana, and Iowa where buckthom
is less abundant or absent often see alate ﬂights later in the season. It is thought that
these aphids are migrating from northern ﬁelds into these areas. A critical next step in
soybean aphid biological control research is to determine if the abundance and
distribution of buckthom in the landscape inﬂuences the establishment and population
dynamics of soybean aphid in soybean and at what spatial scale.

Although my work has concentrated on the biological control of soybean aphid,
the positive correlation between landscape diversity and biocontrol services is likely to
apply to other pests in agricultural habitats. The generalist natural enemies found in
soybean ﬁelds are also important antagonists of many other herbivores. To fully
understand the inﬂuence of landscape on biocontrol services, it should be measured in
multiple habitats. I have begun to develop a sentinel measure of biocontrol services

using corn earworm eggs which can be deployed in multiple habitats to measure the

185

activity of natural enemies. This will yield information on the landscape-wide biocontrol
service provided by natural enemies.

This is a critical time to understand the inﬂuence of semi-natural habitats on
natural enemy abundance and biocontrol services. Currently, corn acreage is on the rise
in response to increased demand for corn ethanol, with harvested acres rising by 15
million acres between 2006 and 2007 (NASS 2008). Increasing pressure to produce
biofuel crops on agricultural lands could have a negative effect on biocontrol services in
these landscapes particularly if corn production increases, reducing landscape diversity
and displacing habitats associated with greater predator abundance. Our focus must be
on how these landscape changes will affect arthropod biodiversity, pollination and pest
management in food and biofuel crops. There is likely to be major differences in the
sustainability of annual monoculture biofuel crops such as corn grain compared with
perennial polyculture crops such as prairie mimics.

Landscape scale approach to native species conservation

In addition to measuring biocontrol services, my work also illustrated a potential
negative outcome of soybean aphid biological control: the loss of native coccinellid
species due to competition from exotic species. While I illustrated that the exotic species
H. axyridis and C. septempunctata are key in suppressing A. glycines, my work also
illustrates that these coccinellids are intraguild predators of native species. Management
plans for native species conservation must consider the role of landscape elements in
inﬂuencing native species abundance and native and exotic species interactions.

Grasslands appear to positively inﬂuence the abundance of native species.

186

Understanding the mechanism behind this pattern is the critical next step in
preserving native coccinellid biodiversity. It is possible that native coccinellids are
adapted to open savanna and prairie landscapes that formerly dominated large areas of
the north-central US. and may still favor landscapes with open grasslands. However,
this does not explain the decline in native coccinellids in the last few decades, long aﬁer
landscapes were transformed by human settlement. Alternatively, native species may be
able to thrive in landscapes that are less favorable to invasion by exotic Coccinellidae
many of which are known to be strong intraguild predators. Since exotics are favored by
forested habitat, a lack of these landscape elements in portions of Minnesota and Iowa
may restrict the abundance and dominance of coccinellid communities by exotics in these
areas. I propose that landscape structure may be limiting the success of H. axyridis and
other exotics in intensely agricultural landscapes and allowing persistence of native
coccinellid communities in such regions. If so, understanding the role of perennial
grasslands in supporting source populations of native coccinellids is a conservation
priority. We must determine if the different types of perennial grassland, such as
pasturelands, restored prairies and abandoned old ﬁelds vary in their ability to support
native species.

Within these habitats, understanding the role of competitive displacement and
intraguild predation by exotics will be crucial in the conservation of native species. I
began this work by investigating the extent of IGP that occurs within soybean ﬁelds.
This work illustrates that, at least in soybean ﬁelds, it is not the intensity of IGP that is
contributing to the decline in native coccinellids but more likely the structure of the

predator guild consuming this resource. In landscapes that favor exotic coccinellids,

187

native coccinellid eggs in soybean are more likely to be consumed by exotic predator
guilds. This is likely contributing to the decline in several native coccinellid species in
forested states such as Michigan were populations of B. ursina, C. munda, C. stigma, and
H. convergens have noticeably declined (Colunga-Garcia and Gage 1998). Alternatively,
despite overall equal egg predation in intensely agricultural landscapes such as Iowa, the
presence of fewer exotic competitors is likely to favor the persistence of native
coccinellid populations.

Although it is impossible to remove established exotic coccinellids from the
environment, we can protect native biodiversity by protecting and promoting landscape
attributes which favor native coccinellid populations. The majority of research on exotic
and native coccinellids has focused on their populations in agricultural ﬁelds (Snyder and
Evans 2006), which likely function as sink populations supplied by surrounding natural
habitats. Future work must focus on understanding interactions between source

populations of native and exotic coccinellids, in grassland and forested habitats.

188

APPENDICES

189

Appendix A

Table A1 .1. Record of deposition of voucher specimens*

The specimens listed on the following sheet(s) have been deposited in the named
museum(s) as samples of those species or other taxa, which were used in this research.
Voucher recognition labels bearing the Voucher No. have been attached or included in
ﬂuid-preserved specimens.

Voucher No.: I 2008-07

Title of thesis or dissertation (or other research projects):

LANDSCAPE SCALE AND WITHIN-FIELD INFLUENCES ON PREDATOR
ABUNDANCE AND BIOCONTROL SERVICES IN SOYBEAN FIELDS.

Museum(s) where deposited and abbreviations for’table on following sheets:

Entomology Museum, Michigan State University (MSU)
Other Museums

Investigator’s Name(s) (typed)

 

Mary M. Gardiner

 

Date 8-22-08

*Reference: Yoshimoto, C. M. 1978. Voucher Specimens for Entomology in North
America. Bull. Entomol. Soc. Amer. 24: 141-42.

Deposit as follows:
Original: Include as Appendix 1 in ribbon copy of thesis or dissertation.
Copies: Include as Appendix 1 in copies of thesis or dissertation.
Museum(s) ﬁles.

Research project ﬁles.

This form is available from and the Voucher No. is assigned by the Curator, Michigan
State University Entomology Museum.

190

 

 

.l x f
090.30%
0.9025 00022.2 9.. c. 20800

.00 05.0.0000 00.0: 0>000 05 0020001

 

 

0305.02 x00 0:200 020 0:08.030 .2..

 

wooNNNE 0000

 

 

.050me .2502

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00.008 .02 L0co2o> .0098 @0802 012000095
3.000000: L. 0.00:0 3000.000 003
202 . 0802.3 .2 .0880 08.0.:
202 . 88.3 .2 00:80 08.0...
20.2 . 0802.3 .2 005.8200
.0 s 20.2 . 080.3 .2 006.850 09.00 835.
m m... 20.2 N 08002. .3 0205.2
A m new 202 . 88.03 .3 0200.3.
m m 7 202 . 8000.... 5 000.0520
.m .w f 20.2 . 0802.... .3 0085..
m. an» o 20.2 . 080.02. .3 00800..
n. r 1 20.2 . 0802.3 .2 3.5.8200
A m e 202 . 0832. .2 005.850 220.00.: 252000
w m. 20.2 . 880.2. .2 000080.800
v P 20.2 . 08.0.23 .2 005.8200
202 N 800...... .2 0828.00 00:00.58 22:00...
:ms. 5 0032 Q0 :2 .«0.8.0055 55802030000 52203..
20.2 . 88: 0.0 .2 000080.800 0320.8 25:00...
20.2 . 0800.0 .3 .2880: 3.20.20: 30.023...
cumin—«.30
mews. .r w w... e m e 02.080020 8..
m a W m m w W m M $ 0002 .0 00.02.00 00068000 .2 0.00 .0002 000200.050 .0 00. m
M m 0 m M M m m. m _m.
#0 23:52

 

 

 

191

Appendix A

Voucher Specimen Data

0.0m. 8.0.20

0.20025. 60.0525 .

00.02:: 20.0 00022.2 a... c. .088
.0. 000020000 00.0.. 0>000 0... 0020001

 

03000.02 x00 0.500 0.0 000020000 .7..

 

080.00... 26.

 

00.0.06 .5. 2.0.2

 

 

Page 2 of 7 Pages

 

 

 

 

 

 

 

 

 

 

 

 

8.88 02 .0622 .89... 0.00.02 0.200.005.
3.0000000 .. 0.0000 .000.._000 00:.
20.2 . 080.... 22 0.508003. .
20.2 . 0000.0... .3 0.08000.
202 . 08...... .3 0085.. 550.802.0003.. =32§00
20.2 . 080.00.. .3 008.0...
20.2 . 88.0.. .3 0.085..
20.2 . 88.0.. .2 000080.85
202 . 080.20 .2 002.8080 320...... 20.20500
20.2 . 38.. E .3 0.088...
20.2 . 88.0.0.0 .3 0.085..
20.2 N 08...... .3 0.089... 53.03532220500
202 . 0000...... .3 0.088..
20.2 . 08....” E .2 00008.85
20.2 . 88.0.. .2 000080.080
392 . 08.0.3.0 E £000.83 0.030.000.0200 “Sbuuhoﬂﬁx
20.2 . 08000.0 .2 .2208 .080...
202 . 080.0 .0 .2 0.880 .080...
20.2 N 88.2.0 .2 0.088 .080... .2800: 3.08.8020
00050.00
m w W W M0 M w m m 8 00000000 000 00x0. .050 .0 00.00am
s e m .w o. m 02 .0 00 00 000020000 .0 0 .
m m m. w M M 0U. W. m m 00 00. .. . . .00.000._
“.0 00.022

 

 

192

C>>chcz xmm .353 m.m WCQPEOOQm =<¢

 

 

 

 

 

 

 

 

Appendix A
Voucher Specimen Data

Page 3 of 7 Pages

 

 

 

 

 

 

 

 

 

 

 

0.00 8.0.20
08000.0 0.00
0.20025. 30.00.95.
0.0.0202 0.0.0 000.0002 00. 0. .0800 60.0.00 ..2 00.2
.0. 0008.800 00.0.. 0>000 00. 002003. i
3.008 .02 .00020> 6002.. 0.00.02 0..0.00_.00>0.
$000800 .. 0.0000 .0000000 002.
. 080.0 :0 .2 £00.00 30.0.0303 003....
. 080.... .2 00.0.8000
. 080.0... .3 0.08000.
. 080.00.. S 02.007. .
. 080.0... S 000007. 00.00.... 00000.00
. 080...... S 000002
. 080.20 <. 02.007.
. 080.0 .0 2.. .85.
. 88.0.0 <. .802 3....30. 3000.00
. 0000.0... .2 00.0.8000 0:00.200 00000.00
. 08.0.2.0 .2 0.00.00 08.0...
. 88.8.0 0. £53 3.000000 0008.00
. 080.80 .2 .300000 00000.06 30.00000
. 080.0... .2 000.000 053.00 00.0.0500
. 88.0.0 .2 00.0.8000 50.0.00. 00.0.0200
. 080.. 0.. 7.2 00000003. 0.000.805.0000. 00.0.0500
000.0200
D. u u D. Q .
mmmwmmmMmﬁB 00.8.. . .:....
”.0 .3052

 

 

193

 

0.00 5.0.00

0.00002 30.00.25
0.20%: 0.0.0 80.5.5. 0... c. .0800
.0. 00950000 00.0.. 0>000 0.... 0020001

 

Cgocv—c: xmw ﬁzz—um mum wcmchoam =<¢

 

08000.0 0.0%

 

50.0.00 .2 b0:

 

Appendix A

 

Voucher Specimen Data
Page 4 of 7 Pages

 

 

 

 

 

 

 

 

 

 

 

 

 

3.008 .02 .0..000> .0098 0.00.02 0.209.002:
$000000: .. 0.09.0 0:00.000 000.
3.0.2 . 0030020 .3 £000.00... $033.0... 00.0%0...
000 0 080.3 .3 0.088.. 0002.0 38.8..
:92 . boomkmk. <. .09....05:m 0:005M...00 00.050...
00.2 . 080.0 :0 .2 .3800 0.0.00 300.0...
:22 v 2.on . \0 .3 £0009... 05000000.... 00......00w
00.2 . 0083.... S 080.. 3:0??300
00.2 . 080.00.. .3 0.089...
000 . 080.0... .3 0.08s...
3.01.). . moonNR .3 £000.00...
00.... . 88.. S. .3 0.085.. 8.8% 0:00.0030
000 . 080.0 :0 22 0508003. 38.330 0.30.0005
3m: . coomktn S 000002
3.0.2 . 0coNR<0 3 002.002 050M230 03000.8.
:92 N mooN\NN\0 .2 .30...w0m 0.00000 050000.00N0AU
00.2 . 08$... .3 .0089... .0... 3......0
00.2 . 88.8.0 .2 £05.00 .080... 3000.80.03. 3.3.50
3.0.2 . 383.80 :2 .BmaEam 0.0.0.3500 00.3.5
00030.30
m e M .r M w... “w m. m 00.00000 0:0 00x0. .050 .0 0060.5
8 M e W. w o. m m... 00 .0 00 00 00086000 .0 0 0 0 .
mewmmnwwmcwg 00... . ..0.00...
”.0 .0952

194

 

Appendix A

Voucher Specimen Data

0.00 5.0.00

 

0.00022 60.00.25
0.00200 0.0.0 c0000.... 0... c. .0800
.0. 000020000 00.0.. 0>000 0... 002000”.

 

c>>ocxc: x00 £53.00 000 9.053000 :4...

 

wooNNNE 0.00

 

00.0.00 .2 2.0.2

 

 

Page Lof 7 Pages

 

 

 

 

 

 

 

 

 

 

 

3.008 .02 .00000> 0.00.02 0.0.00..00>0_
000000000 0.. 0.0000 .000...000 00:.
30.2 . 00.82030 .3 .0.00050... 00.0.0039... 00000.3
30.2 . 00083.. 4. .0000. 00.000.00.00 03.30.0003
30.2 . 0082.0 .2 6005.00.85 00.500 00.30.0003
00.2 . 08.0.00 .2 000000.00...
:05. . mOOQ . 0R :2 .0w0tcox000m. 00.00.50.030 000.2000
00.2 . 080.30 .2 00.000.00.020
30.2 . mooﬁon .2 .305w0m 000008.00. 05.0.0000:
00.2 . 88.30 .2 0000002080
:05. . 95.0% C0 .2 .0w0t00x000m 00.202030. 0000.00.80.
30.2 . 00.833 5 .0290. 0.030.350.00 000.0...000000. ,
30.2 . 00.8.0.0 S .000... 00.020.00.00. 00.8000.
00.2 N 0800 :0 .2 000000.020 0000.000 300$.
20.2 . 008280 S 00$
00.2 . 02.0.0.0 S .0003 0.0.0 0.00..
00.2 _. 0800.0 .2 000.8200 0.830.000
:05. N 38% <0 :2 $00.03 000.000.09.00. 00.030m
00.2 . 080.00 .2. 00800.. 0.0000000 3000....
00030.00
m m m .r Wm... M w W m 00.00000 000 00x0 0 0 .0 00.000
W W. m. W0 W W 0% m. m 0% 0000 .0 00.0260 050.0000 .0. 0.00 .000. .. 0. . 0
“.0 .00 E02

 

 

 

195

Appendix A

Voucher Specimen Data

0.00 0.0.00
0.00005. 20200.95

200.0203 0.0.0 000.00.... 0.... 0. 200000
.0. 00050000 00.0: 0>000 0... 002.000”.

 

C>>chc3 x00 .0333 ®.G 00.05380 :6...

 

080.000 26

 

.00E00 ._2 2.0.2

 

 

Page 6 of 7 Pages

 

 

 

 

 

 

 

 

 

 

 

 

3.88 .oz 0.8.5 .08.... @0502 £200.09...
300080.. .. 0.0020 .0803... 000.
00.2 . 0800.0 .2 23.0.0 800000
00.2 . 080.000 .2 £828.00 8000020...
30.2 . 000N\0N\.. E). .303000 0.0.03.0... 0000.00.
305. . 000m\m.\0 3 000.6500 0.00.2.0. 00000005
30.2 F 000N\0m\0 S .008... 03.0.. 0200...
00.2 . 080000 .2 00.2.8280 0.0.5.. 383.200
305. . 000m... . .0 5 0.5002 0000000....» 000000300
302 . 000.0824
00.2 N 300...... .2 02.00.... .30 2.0.0....
002 . 300...... .2 00.05.. .30 30.08030
00.2 a 080.0... .2 00.08.. .30 80.0.50
00.2 m 88...... S 5.0 0.0.8.. .0800... 2.0.20
00.2 m 080...... .2 00.05.. .000 3000.22 5500000....
002 . 88.0.... .2 00.2.8.8... 0.8.200... 550.800..
00.2 . 88...... .2 28.00 30.82.. 0.20.5830
30.2 . 000N\m :0 :2 0500.900 0.0.0=3&50.&00 00.00.0000
00.2 . 008.0 E .2 60008.85 0.0.23 0.0800...
0005050000
00.2 . 0800 .0 .2 50.000 300.0003... 38...
mew .r M w... e m e 00.08002...
N W W m W w W m m m... 0000 .0 00.09.00 000.5030 .0. 0.00 .000. 00x0. .050 .0 00.0000
M w w mu. M M m m. m .m.
”.0 500.02

 

 

196

Esocxc: xow .353 2m 9586ch __<.

 

 

 

 

 

 

 

 

 

7 Pages

Appendix A
7 of

Voucher Specimen Data
Page

 

 

 

 

 

 

 

 

 

 

 

 

End .2930
moowﬁua EB.
.5332: 30.9.5.5
£292: 05m amazes. 9: s «38% 85.28 .s. be:
.8 wcmEBmam 0.2m: m>onm 2: 8200mm
No.88 6: Lmzozo> 635 @252 wzoﬁmume.
Abmmmmooc%ﬂomcw _mcoauum $3 i
am: P 8830 S .82 Sagas—h
m m 6 0+ m e 350an ucm
m s *r b .6 e P .
m .m W m .lm. w W m. m w vow: B 862.8 mcmgooam .8 Emu .83 :98. 550 .0 860mm
M w w 0 A A P N ha E
co .8832

 

 

197

 

Appendix B

Table B1.1. Appendix B includes land cover maps for all ﬁeld sites sampled in 2005 and
2006 in Iowa, Michigan, Minnesota and Wisconsin. Land cover is grouped into
categories used in the principal component analyses presented in chapters 3-5. Below,
the ﬁgure number, location, study year, and diversity of each site is listed.

 

 

198

Study Site and Season Coordinates of Field Center
_Eiggre # State Location Year N W Simpson's D
81.1 IA Ames 2005 41° 58' 50.6 " 93° 38' 13.86" 4.35
81.2 IA Ames 2006 41° 58' 48.91" 93° 38' 49.42" 4.19
81.3 IA Lucas 2005 40° 58' 25.75" 93° 24' 54.64" 3.96
81.4 IA Lucas 2006 40° 57' 43.47" 93° 24' 04.54" 4.82
81.5 IA Nashua 2005 42° 55' 46.98" 92° 33' 57.72" 3.81
81.6 IA Nashua 2006 42° 56' 07.99" 92° 34' 22.75" 3.44
81.7 IA Sutherland 2005 42° 55' 37.99" 95° 31' 53.18" 2.68
81.8 IA Sutherland 2006 42° 55' 50.06" 95° 32' 17.52" 2.43
81.9 MI Breckenridge 2005 43° 22' 51.35" 84° 31' 33.94" 4.50
81.10 MI Breckenridge 2006 43° 23' 01.09" 84° 31' 33.58" 5.11
81.11 MI Britton 2006 41° 56' 18.62" 83° 48' 27.53" 3.28
81.12 MI East Lansing 2006 42° 41' 26.34" 84° 29' 37.68" 4.07
81.13 MI Elsie 2006 43° 4' 3.86" 84° 29' 0.57" 6.07
81.14 MI Hickory Corners 2005 42° 24' 15.25" 85° 22' 42.83" 5.61
81.15 MI Hickory Corners 2006 42° 24' 23.86" 85° 23' 16.64" 6.40
81.16 MI Monroe 2006 41° 56' 56.22" 83° 28' 13.43 5.15
81.17 MI Saginaw 2005 43° 23' 10.89" 84° 06' 49.96" 5.28
81.18 MI Saginaw 2006 43° 22' 41 .39" 84° 06' 49.94" 5.36
81.19 MI Schoolcraft 2005 42° 07' 45.66" 85° 36' 42.02" 5.70
81.20 MI Schoolcraft 2006 42° 07' 41.0 " 85° 36' 50.01" 5.55
81.21 MI Sheppard 2006 43° 31' 33.11" 84° 42' 18.88" 5.51
81.22 MI Westphalia 2006 42° 54' 51.58" 84° 48' 50.60" 4.95
81.23 MN Lamburton 2005 44° 08' 29.31" 95° 10' 30.10" 2.73
81.24 MN Lamburton 2006 44° 14' 32.72" 95° 17' 32.02" 3.51
81.25 MN Rosemount 2005 44° 43' 57.94" 93° 06' 20.25" 4.60
81.26 MN Rosemount 2006 44° 42' 13.32" 93° 06' 23.76" 4.53
81.27 WI Arlington 2005 43° 18' 59.94" 89° 20' 05.85" 2.78
81.28 WI Arlington 2006 43° 17' 44.72" 89° 23' 00.94" 3.35
81.29 WI Hancock 2005 44° 06' 01.87" 89° 32' 4.91" 4.04
81.30 WI Hancock 2006 44° 07' 06.72" 89° 32' 35.72" 5.56
81.31 WI Menominee 2006 44° 58' 31.80" 88° 27' 43.22" 2.64
81.32 WI West Madison 2005 43° 03' 51.77" 89° 32' 18.52" 2.55
81.33 WI West Madison 2006 43° 04' 04.49" 89° 32' 38.17" 2.95

2005
Ames, IA

 

El Corn
D Soybean

D Wheat

D Other Crops

I] Grassland

Forest

D Wetland and Open Water
Urban

Figure B1.1. Land cover map of Ames IA, 2005. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

199

  
  
  
 

L
-

 
 
   
    

I"
i“-

_u is

      
 
  

     
 

     
 

u.
l ‘l

“3 Com

Soybean

El Wheat

D Other Crops

D Grassland

ﬂ Forest

D Wetland and Open Water
E Urban

Figure B1.2. Land cover map of Ames IA, 2006. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

200

2005
Lucas, IA

 

El Com

[1 Soybean

El Wheat

D Other Crops

D Grassland

E Forest

I] Wetland and Open Water
ﬂ Urban

Figure B1.3. Land cover map of Lucas IA, 2005. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

201

2006
Lucas, Iowa

 

D Corn

E Soybean

I] Wheat

[:1 Other Crops

El Grassland

ﬂ Forest

El Wetland and Open Water
D Urban

Figure B1.4. Land cover map of Lucas IA, 2006. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

202

2005
Nashua, Iowa

 

 

El Corn

E] Soybean

El Wheat

D Other Crops

E] Grassland

Forest

:1 Wetland and Open Water
El Urban

Figure B1.5. Land cover map of Nashua IA, 2005. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

203

2006
Nashua, Iowa

 

E] Corn

D Soybean

I] Wheat

I] Other Crops

El Grassland

El Forest

I] Wetland and Open Water
a Urban

Figure B1.6. Land cover map of Nashua IA, 2006. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

204

2005
Sutherland, Iowa

 

El Forest
D Wetland and Open Water
ﬂ Urban

Figure B1.7. Land cover map of Sutherland IA, 2005. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

205

2006
Sutherland, Iowa

 

D Soybean

El Wheat

[3 Other Crops

Cl Grassland

a Forest

E Wetland and Open Water
a Urban

Figure B1.8. Land cover map of Sutherland, IA, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

206

2005
Breckenridge, Michigan

 

El Com
5 Soybean

L—J Wheat

D Other Crops

C] Grassland

Forest

D Wetland and Open Water
El Urban

Figure 81.9. Land cover map of Breckenridge, MI, 2005. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

207

2006

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breckenridge, Michigan
J1
.~ . a
"F; _‘Ig‘j'md 52“,
a t ~.f
. ,,
4* . . ' .
w: ..,.
. -3
I _' "If .
.. , , ,7;
. I I ,§ '
II ‘ .
-. . ..
.1 .‘ ’7
‘ 3
U . 1, .. ".£1.
'1'. ‘ \tn 0 I!
.1 «5.
a. . . "
r- ,.
:‘I
' 1' 2M [7
Lb". 2~ .
i .

 

El Corn

D Soybean

El Wheat

El Other Crops

D Grassland

D Forest

[1 Wetland and Open Water
3 Urban

Figure B1 . 10. Land cover map of Breckenridge, MI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

208

2006
Britton, Michigan

 

B Soybean

D Wheat

[:1 Other Crops

El Grassland

ﬂ Forest

I] Wetland and Open Water
Urban

Figure B1.11. Land cover map of Britton, MI, 2006. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

209

2006
East Lansing, Michigan

D's: <
:1 ~91

“Iii

 

El Corn
D Soybean

D Wheat

[1 Other Crops

D Grassland

D Forest

D Wetland and Open Water
a Urban

Figure B1.12. Land cover map of East Lansing, MI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

210

2006
Elsie, Michigan

 

D Soybean

El Wheat

El Other Crops

E] Grassland

[3 Forest

D Wetland and Open Water
E Urban

Figure B1.13. Land cover map of Elsie, MI, 2006. Focal soybean ﬁeld indicated with dot
in the center of the image. The landscape circle has a radius of 3.5 km.

211

2005
Hickory Comers, Michigan

 

E] Corn
Soybean

El Wheat

C] Other Crops

D Grassland

I'J Forest

:3 Wetland and Open Water
ﬂ Urban

Figure B1.14. Land cover map of Hickory Comers, MI, 2005. Focal soybean ﬁeld

indicated with dot in the center of the image. The landscape circle has a radius of 3.5 km.

212

2006
Hickory Comers, Michigan

III

 

E Soybean

El Wheat

El Other Crops

E] Grassland

Forest

a Wetland and Open Water
5 Urban

Figure Bl . 15. Land cover map of Hickory Comets, MI, 2006. Focal soybean ﬁeld
indicated with dot in the center of the image. The landscape circle has a radius of 3.5 km.

213

2006
Monroe, Michigan

 

D Corn
Soybean

'3 Wheat

D Other Crops

D Grassland

D Forest

D Wetland and Open Water
ﬂ Urban

Figure B1.16. Land cover map of Monroe, MI, 2006. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

214

2005
Saginaw, MI

 

 

Soybean

D Wheat

E] Other Crops

El Grassland

U Forest

3 Wetland and Open Water
5 Urban

Figure B1.17. Land cover map of Saginaw, MI, 2005. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

215

2006
Saginaw, Michigan

 

El Corn
D Soybean

El Wheat

D Other Crops

E] Grassland

Forest

E] Wetland and Open Water
Urban

Figure B1.18. Land cover map of Saginaw, MI, 2006. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

216

2005
Schoolcraft, MI

 

 

[3 Corn
[3 Soybean

D Wheat

D Other Crops

El Grassland

Forest

U Wetland and Open Water
Urban

Figure B1.19. Land cover map of Schoolcraft, MI, 2005. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

217

2006
Schoolcraft, Michigan

‘ l
{and

 

 

El Com

D Soybean

D Wheat

D Other Crops

El Grassland

Forest

D Wetland and Open Water
a Urban

Figure 81.20. Land cover map of Schoolcraft, MI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

218

2006
Shepherd, Michigan

 

D Soybean

D Wheat

El Other Crops

El Grassland

I] Forest

a Wetland and Open Water
Urban

Figure B1.21. Land cover map of Shepherd, MI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

219

2006
Westphalia, Michigan

 

D Soybean

El Wheat

D Other Crops

D Grassland

El Forest

D Wetland and Open Water
5 Urban

Figure Bl.22. Land cover map of Westphalia, MI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

220

2005
Lamberton, Minnesota

 

E3 Soybean

El Wheat

D Other Crops

E] Grassland

El Forest

D Wetland and Open Water
E Urban

Figure Bl.23. Land cover map of Lamberton, MN, 2005. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

221

2006
Lamberton, Minnesota

'-
“5;? "Ir-

.‘
l.’ .
‘ i ll ‘
u, i- ___ k
' 3"» ___—=
-1! . — —

l
Illllllll '

" - ' mums
:- l M .-

 

D Soybean

L—J Wheat

D Other Crops

El Grassland

Forest

D Wetland and Open Water
I! Urban

Figure Bl.24. Land cover map of Lamberton, MN, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

222

2005
Rosemount, Minnesota

 

El Corn
[I Soybean

El Wheat

D Other Crops

El Grassland

D Forest

El Wetland and Open Water
I! Urban

Figure Bl.25. Land cover map of Rosemount, MN, 2005. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

223

2006
Rosemount, Minnesota

 

El Corn
Soybean

E] Wheat

D Other Crops

El Grassland

Forest

D Wetland and open Water
9 Urban

Figure 81.26. Land cover map of Rosemount, MN, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

224

2005
Arlington, Wisconsin

 

‘_.‘_':"l l _,
. [1'4 E:
.:.:"3 ‘ 1y .
lz. ' . - n .L ,_

 

 

D Other Crops

D Grassland

Forest

D Wetland and Open Water
E Urban

Figure B1.27. Land cover map of Arlington, WI, 2005. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

225

2006
Arlington, Wisconsin

 

 

 

El Corn

El Soybean

El Wheat

El Other Crops

El Grassland

E Forest

E Wetland and Open Water
Urban

Figure B128. Land cover map of Arlington, WI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

226

2005
Hancock, Wisconsin

 

El Corn
Soybean

D Wheat

El Other Crops

El Grassland

El Forest

El Wetland and Open Water
ﬂ Urban

Figure B1.29. Land cover map of Hancock, WI, 2005. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

227

2006
Hancock, Wisconsin

 

 

 

 

 

 

 

D Soybean

El Wheat

D Other Crops

D Grassland

Forest

D Wetland and Open Water
El Urban

Figure B130. Land cover map of Hancock, WI, 2006. Focal soybean ﬁeld indicated with
dot in the center of the image. The landscape circle has a radius of 3.5 km.

228

2006
Menominee, Wisconsin

>w...a

 

El Com
m Soybean

U Wheat

El Other Crops

D Grassland

ﬂ Forest

ﬂ Wetland and Open Water
Urban

Figure B131. Land cover map of Menominee, WI, 2006. Focal soybean ﬁeld indicated
with dot in the center of the image. The landscape circle has a radius of 3.5 km.

229

2005
West Madison, Wisconsin

 

El Corn
D Soybean

El Wheat

D Other Crops

E] Grassland

Forest

[3 Wetland and Open Water
E Urban

Figure B1.32. Land cover map of West Madison, WI, 2005. Focal soybean ﬁeld
indicated with dot in the center of the image. The landscape circle has a radius of 3.5 km.

230

2006
West Madison, Wisconsin

 

D Soybean

D Wheat

[:1 Other Crops

El Grassland

I] Forest

D Wetland and Open Water
Urban

Figure B1.33. Land cover map of West Madison, WI, 2006. Focal soybean ﬁeld
indicated with dot in the center of the image. The landscape circle has a radius of 3.5 km.

231

REFERENCES CITED

232

REFERENCES CITED

Addicott, J. F., J. M. Aho, M. F. Antolin, D. K. Padilla, J. S. Richardson, and D. A.
Soluk. 1987. Ecological Neighborhoods - Scaling Environmental Patterns. Oikos
49: 340-346.

Alomar, 0., M. Goula, and R. Albajes. 2002. Colonisation of tomato ﬁelds by
predatory mirid bugs (Hemiptera : Heteroptera) in northern Spain. Agriculture
Ecosystems & Environment 89: 105-115.

Alyokhin, A., and G. Sewell. 2004. Changes in a lady beetle community following the
establishment of three alien species. Biological Invasions 6: 463-471.

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.

Armer, C. A., R. N. Wiedenmann, and D. R. Bush. 1998. Plant feeding site selection
on soybean by the facultatively phytophagous predator Orius insidiosus.
Entomologia Experimentalis Et Applicata 86: 109-118.

Baseompte, J., and C. J. Melian. 2005. Simple trophic modules for complex food webs.
Ecology 86: 2868-2873.

Bianchi, F. J. J. A., C. J. H. Booij, and T. Tscharntke. 2006. Sustainable pest
regulation in agricultural landscapes: a review on landscape composition,
biodiversity and natural pest control. Proceedings of the Royal Society B-
Biological Sciences 273: 1715-1727.

Bommarco, R. 1998. Reproduction and energy reserves of a predatory carabid beetle
relative to agroecosystem complexity. Ecological Applications 8: 846-853.

Brewer, M. J., and N. C. Elliott. 2004. Biological control of cereal aphids in North
America and mediating effects of host plant and habitat manipulations. Annual
Review of Entomology 49: 219-242.

Brown, M. W. 2003. Intraguild responses of aphid predators on apple to the invasion of
an exotic species, Harmonia axyridis. Biocontrol 48: 141-153.

233

Brown, M. W., and S. S. Miller. 1998. Coccinellidae (Coleoptera) in apple orchards of
eastern West Virginia and the impact of invasion by Harmonia mridis.
Entomological News 109: 143-151.

Bumham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference
: a practical information-theoretic approach. Springer, New York.

Cardinale, B. J., C. T. Harvey, K. Gross, and A. R. Ives. 2003. Biodiversity and
biocontrol: emergent impacts of a multi-enemy assemblage on pest suppression
and crop yield in an agroecosystem. Ecology Letters 6: 857-865.

Chacon, J., D. A. Landis, and G. E. Heimpel. in review. Potential for biotic
interference of a classical biological control agent of the soybean aphid.
Biological Control.

Chapin, J. B., and V. A. Brou. 1991. Harmonia axyridis (Pallas), the 3rd species of the
genus to be found in the United-States (Coleoptera, Coccinellidae). Proceedings
of the Entomological Society of Washington 93: 630-635.

Clark, A. J., and K. L. Perry. 2002. Transmissibility of ﬁeld isolates of soybean viruses
by Aphis glycines. Plant Disease 86: 1219-1222.

Colfer, R. G., and J. A. Rosenheim. 2001. Predation on immature parasitoids and its
impact on aphid suppression. Oecologia 126: 292-304.

Colfer, R. G., J. A. Rosenheim, L. D. Godfrey, and C. L. Hsu. 2003. Interactions
between the augmentatively released predaceous mite Galendromus occidentalis

(Acari : Phytoseiidae) and naturally occurring generalist predators. Environmental
Entomology 32: 840-852.

Collinge, S. K., and T. M. Palmer. 2002. The inﬂuences of patch shape and boundary
contrast on insect response to fragmentation in California grasslands. Landscape
Ecology 17: 647-656.

Colunga—Garcia, M., and S. H. Gage. 1998. Arrival, establishment, and habitat use of
the multicolored Asian lady beetle (Coleoptera : Coccinellidae) in a Michigan
landscape. Environmental Entomology 27: 1574-1580.

234

Colunga-Garcia, M., S. H. Gage, and D. A. Landis. 1997. Response of an assemblage
of Coccinellidae (Coleoptera) to a diverse agricultural landscape. Environmental
Entomology 26: 797-804.

Costamagna, A. C., and D. A. Landis. 2007a. 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. 2007b. 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 A.
glycines by generalist predators results in a trophic cascade in soybean. Ecological
Applications 17: 441- 451.

Costamagna, A. C., D. A. Landis, and M. J. Brewer. in press. The role of natural
enemy guilds in Aphis glycines suppression. Biological Control.

Costamagna, A. C., W. van der 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.

Cottrell, T. E. 2004. Suitability of exotic and native lady beetle eggs (Coleoptera :
Coccinellidae) for development of lady beetle larvae. Biological Control 31: 362-
371.

Cottrell, T. E. 2005. Predation and cannibalism of lady beetle eggs by adult lady beetles.
Biological Control 34: 159-164.

Cottrell, T. E. 2007. Predation by adult and larval lady beetles (Coleoptera :
Coccinellidae) on initial contact with lady beetle eggs. Environmental
Entomology 36: 390-401.

Cottrell, T. E., and K. V. Yeargan. 1998a. Inﬂuence of a native weed, Acalypha
ostryaefolia (Euphorbiaceae), on Coleomegilla maculata (Coleoptera :
Coccinellidae) population density, predation, and cannibalism in sweet corn.
Environmental Entomology 27: 1375-1385.

235

Cottrell, T. E., and K. V. Yeargan. 1998b. Effect of pollen on Coleomegilla maculata
(Coleoptera : Coccinellidae) population density, predation, and cannibalism in
sweet corn. Environmental Entomology 27: 1402-1410.

Day, W. H., D. R. Prokrym, D. R. Ellis, and R. J. Chianese. 1994. The known
distribution of the predator Propylea quatuordecimpunctata (Coleoptera,
Coccinellidae) in the United-States, and thoughts on the origin of this species and
5 other exotic lady beetles in eastern North-America. Entomological News 105:
244-256.

den Belder, E., J. Elderson, W. J. van den Brink, and G. Sehelling. 2002. Effect of
woodlots on thlips density in leek ﬁelds: a landscape analysis. Agriculture
Ecosystems & Environment 91: 139-145.

Dennis, P., and S. D. Wratten. 1991. Field manipulation of populations of individual
staphylinid species in cereals and their impact on aphid populations. Ecological
Entomology 16: 17-24.

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.

Didham, R. K., J. M. Tylianakis, N. J. Gemmell, T. A. Rand, and R. M. Ewers.
2007. Interactive effects of habitat modiﬁcation and species invasion on native
species decline. Trends in Ecology & Evolution 22: 489-496.

DiFonzo, C., and R. Hines. 2002. Soybean aphid in Michigan: Update from 2001 season
Michigan State University Extension Bulletin E-2746.

Donaldson, J. R., Meyers, S. W. and Gratton, C. 2007. Density-dependent responses
of soybean aphid (Aphis glycines Matsumura) populations to generalist predators
in mid and late season soybean ﬁelds. Biological Control.

Elliott, N., R. Kieckhefer, and W. Kauffman. 1996. Effects of an invading coccinellid
on native coccinellids in an agricultural landscape. Oecologia 105: 537-544.

Elliott, N. C., R. W. Kieckhefer, J. H. Lee, and B. W. French. 1999. Inﬂuence of
within-ﬁeld and landscape factors on aphid predator populations in wheat.
Landscape Ecology 14: 239-252.

236

Ellis, D. R., D. R. Prokrym, and R. G. Adams. 1999. Exotic lady beetle survey in
northeastern United States: Hippodamia variegata and Propylea

quatuordecimpunctata (Coleoptera: Coccinellidae). Entomological News 110: 73-
84.

Evans, E. W. 2000. Morphology of invasion: body size patterns associated with
establishment of Coccinella septempunctata (Coleoptera: Coccinellidae) in
western North America. European Journal of Entomology: 469-474.

Evans, E. W. 2003. Searching and reproductive behaviour of female aphidophagous
ladybirds (Coleoptera: Coccinellidae): a review. European Journal of Entomology
100: 1-10.

Evans, E. W. 2004. Habitat displacement of North American ladybirds by an introduced
species. Ecology 85: 637-647.

Fahrig, L. 1992. Relative Importance of Spatial and Temporal Scales in a Patchy
Environment. Theoretical Population Biology 41: 300-314.

Fahrig, L. 2001. How much habitat is enough? Biological Conservation 100: 65-74.

F ahrig, L. 2003. Effects of habitat ﬁagmentation on biodiversity. Annual Review of
Ecology Evolution and Systematics 34: 487-515.

Ferguson, K. I., and P. Stiling. 1996. Non-additive effects of multiple natural enemies
on aphid populations. Oecologia 108: 37 5-379.

Fortin, M., and Y. Mauffette. 2001. Forest edge effects on the biological performance
of the forest tent caterpillar (Lepidoptera : Lasiocampidae) in sugar maple stands.
Ecoscience 8: 164-172.

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.

237

Galecka, B. 1966. The role of predators in the reduction of two species of potato aphids,
Aphis nasturii Kalt. and A. ﬁangulae Kalt. Ekol. Polska 16: 245-274.

Gardiner, M. M. 2008. Landscape scale and within-ﬁeld inﬂuences on predator
abundance and biocontrol services in soybean ﬁelds, pp. 282, (Department of
Entomology. Michigan State University, East Lansing, MI.

Gardiner, M. M., and G. L. Parsons. 2005. Hippodamia variegata (Goeze)
(Coleoptera: Coccinellidae) detected in Michigan soybean ﬁelds. Great Lakes
Entomologist 38: 164-169.

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. P. Schmidt, M. O’Neal, E. D.
Mueller, J. Chacon, G. E. Heirnpel, and C. D. Difonzo. in review. Landscape
composition inﬂuences patterns of native and exotic ladybeetle abundance.
Diversity and Distributions.

Gardiner, M. M., D. A. Landis, C. Gratton, C. D. DiFonzo, M. O’Neal, J. Chacon,
M. Wayo, N. Schmidt, E. Mueller, and G. E. Heimpel. in press. Landscape
diversity enhances biological control of an introduced crop pest in the north-
central U.S. Ecological Applications.

Grasswitz, T. R., and E. C. Burts. 1995. Effect of native natural enemies and
augmentative releases of Chrysoperla ruﬁlabrz’s Burmeister and Aphidoletes
aphidimyza (Rondani) on the population dynamics of the green apple aphid, Aphis
pomi Degeer. International Journal of Pest Management 41: 176-183.

Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community Structure,
Population Control, and Competition. American Naturalist 94: 421-425.

Harmon, J. P., E. Stephens, and J. Losey. 2007. The decline of native coccinellids
(Coleoptera: Coccinellidae) in the United States and Canada. Journal of Insect
Conservation 11: 85-94.

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.

238

Holland, J., and L. Fahrig. 2000. Effect of woody borders on insect density and
diversity in crop ﬁelds: a landscape-scale analysis. Agriculture Ecosystems &
Environment 78: 115-122.

Holt, R. D., J. H. Lawton, G. A. Polis, and N. D. Martinez. 1999. Trophic rank and the
species-area relationship. Ecology 80: 1495-1504.

Isenhour, D. J., and N. L. Marston. 1981. Seasonal Cycles of Orius-Insidiosus
(Hemiptera, Anthocoridae) in Missouri Soybeans. Journal of the Kansas
Entomological Society 54: 129-142.

Kaiser, M. E., T. N oma, 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 midwestem United States invasion. Annals of the
Entomological Society of America 100: 196-205.

Kindlmann, P., and K. Houdkova. 2006. Intraguild predation: ﬁction or reality?
P0pulation Ecology 48: 317-322.

Klug, T ., Gathmann, A., Poehling, H. M. and Meyhofer, R. 2003. Area dependent
effects of landscape structure on the colonisation of spinach cultures by the silver
Y moth (Autographa gamma L., Lepidoptera: Noctuidae) in western Germany.
IOBC wprs Bulletin 26: 77-82.

Lee, J. C., F. B. Menalled, and D. A. Landis. 2001. Refuge habitats modify impact of
insecticide disturbance on carabid beetle communities. Journal of Applied
Ecology 38: 472-483.

Legendre, P. 1993. Spatial Autocorrelation - Trouble or New Paradigm. Ecology 74:
1659-1673.

Losey, J. E., and R. F. Denno. 1998. Positive predator-predator interactions: Enhanced
predation rates and synergistic suppression of aphid populations. Ecology 79:
2 143 -2 1 52.

Losey, J. E., and M. Vaughan. 2006. The economic value of ecological services
provided by insects. Bioscience 56: 311-323.

239

Losey, J. E., J. E. Perlman, and E. R. Hoebeke. 2007. Citizen scientist rediscovers rare
nine-spotted lady beetle, Coccinella novemnotata, in eastern North America.
Journal of Insect Conservation 11: 415-417.

Lucas, E. 2005. Intraguild predation among aphidophagous predators. European Journal
of Entomology 102: 351-363.

Lucas, E., D. Coderre, and J. Brodeur. 1998. Intraguild predation among aphid
predators: Characterization and inﬂuence of extraguild prey density. Ecology 79:
1084-1092.

Macedo, T. B., C. S. Bastos, L. G. Higley, K. R. Ostlie, and S. Madhavan. 2003.
Photosynthetic responses of soybean to soybean aphid (Homoptera : Aphididae)
injury. Journal of Economic Entomology 96: 188-193.

Marino, P. C., and D. A. Landis. 1996. Effect of landscape structure on parasitoid
diversity and parasitism in agroecosystems. Ecological Applications 6: 276-284.

Marine, P. C., D. A. Landis, and B. A. Hawkins. 2006. Conserving parasitoid
assemblages of North American pest Lepidoptera: Does biological control by

native parasitoids depend on landscape complexity? Biological Control 37: 173-
1 85.

McComack, 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.

McCune, B., and M. J. Mefford 1999. PC-ORD, Multivariate Analysis of Ecological
Data, Version 4 computer program, version By McCune, B., and M. J. Mefford,
Gleneden Beach, Oregon.

Menalled, F. D., P. C. Marino, S. H. Gage, and D. A. Landis. 1999. Does agricultural
landscape structure affect parasitism and parasitoid diversity? Ecological
Applications 9: 634-641.

Menalled, F. D., A. C. Costamagna, P. C. Marina, and D. A. Landis. 2003. Temporal
variation in the response of parasitoids to agricultural landscape structure.
Agriculture Ecosystems & Environment 96: 29-35.

240

Mensah, R. K. 1996. Evaluation of coloured sticky traps for monitoring populations of
Austroasca vividigrisea (Paoli) (Hemiptera: Cicadellidae) on cotton farms.
Australian Journal of Entomology 35: 349-353.

Meyhofer, R. 2001. Intraguild predation by aphidophagous predators on parasitised
aphids: the use of multiple video cameras. Entomologia Experimentalis Et
Applicata 100: 77-87.

Michaud, J. P. 2002. Invasion of the Florida citrus ecosystem by Harmonia axyridis
(Coleoptera: Coccinellidae) and asymmetric competition with a native species,
Cycloneda sanguinea. Environmental Entomology 31: 827-835.

Muller, C. B., M. D. E. Fellowes, and H. C. J. Godfray. 2005. Relative importance of
fertiliser addition to plants and exclusion of predators for aphid growth in the
ﬁeld. Oecologia 143: 419-427.

NASS. 2008. National Statistics, Field Com. USDA National Agricultural Statistics .
Service.

Ostman, O. 2002. Distribution of bird cherry-oat aphids (Rhopalosiphum padi (L)) in
relation to landscape and farming practices. Agriculture Ecosystems &
Environment 93: 67-71.

Ostman, 0., B. Ekbom, and J. Bengtsson. 2001. Landscape heterogeneity and farming
practice inﬂuence biological control. Basic and Applied Ecology 2: 365-371.

Parajulee, M. N., and J. E. Slosser. 2003. Potential of yellow sticky traps for lady
beetle survey in cotton. Journal of Economic Entomology 96: 239-245.

Pfannenstiel, R. S., and K. V. Yeargan. 2002. Identiﬁcation and diel activity patterns of
predators attacking Helicoverpa zea (Lepidoptera : Noctuidae) eggs in soybean
and sweet corn. Environmental Entomology 31: 232-241.

Phoofolo, M. W., and J. J. Obrycki. 1998. Potential for intraguild predation and
competition among predatory Coccinellidae and Chrysopidae. Entomologia
Experimentalis Et Applicata 89: 47-55.

Polis, G. A. 1991. Complex Trophic Interactions in Deserts - an Empirical Critique of
Food-Web Theory. American Naturalist 138: 123-155.

241

Polis, G. A. 1998. Ecology - Stability is woven by complex webs. Nature 395: 744-745.

Polis, G. A., C. A. Myers, and R. D. Holt. 1989. The Ecology and Evolution of
Intraguild Predation - Potential Competitors That Eat Each Other. Annual Review
of Ecology and Systematics 20: 297-330.

Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an integration of landscape
and food web ecology: The dynamics of spatially subsidized food webs. Annual
Review of Ecology and Systematics 28: 289-316.

Prasad, R. P., and W. E. Snyder. 2004. Predator interference limits ﬂy egg biological
control by a guild of ground-active beetles. Biological Control 31: 428-437.

Putnam, W. L. 1955. Bionomics of Stethorus punctillum Weise (Coleoptera:
Coccinellidae) in
Ontario. Canadian Entomologist 87: 9-33.

R Development Core Team 2005. R: A language and environment for statistical
computing computer program, version By R Development Core Team, Vienna,
Austria.

Raffy, M., and A. Tran. 2005. On the dynamics of ﬂying insects populations controlled
by large scale information. Theoretical Population Biology 68: 91 -104.

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.

Ragsdale, D. W., B. P. McComack, R. C. Venette, B. D. Potter, 1. V. Macrae, E. W.
Hodgson, M. E. O'Neal, K. D. Johnson, R. J. O'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.

Ribic, C. A., and D. W. Sample. 2001. Associations of grassland birds with landscape
factors in southern Wisconsin. American Midland Naturalist 146: 105-121.

Rice, M. E. 2006. 04-05-06 and the multicolored Asian lady beetle, pp. 84-85, Integrated
Crop Management News.

242

 

Roschewitz, 1., M. Hucker, T. Tscharntke, and C. Thies. 2005. The inﬂuence of
landscape context and farming practices on parasitism of cereal aphids.
Agriculture Ecosystems & Environment 108: 218-227.

Rosenheim, J. A. 1998. Higher-order predators and the regulation of insect herbivore
populations. Annual Review of Entomology 43: 421-447.

Rosenheim, J. A., and A. Corbett. 2003. Omnivory and the indeterminacy of predator
ﬁinction: Can a knowledge of foraging behavior help? Ecology 84: 253 8-2548.

Rosenheim, J. A., L. R. Wilhoit, and C. A. Armer. 1993. Inﬂuence of Intraguild
Predation among Generalist Insect Predators on the Suppression of an Herbivore
Population. Oecologia 96: 439-449.

Rosenheim, J. A., H. K. Kaya, L. E. Ehler, J. J. Marois, and B. A. Jaffee. 1995.
Intraguild Predation among Biological-Control Agents - Theory and Evidence.
Biological Control 5: 303-335.

Rosenheim, J. A., D. D. Limburg, R. G. Colfer, V. Fournier, C. L. Hsu, T. E.
Leonardo, and E. H. Nelson. 2004. Herbivore population suppression by an
intermediate predator, Phytoseiulus macropilis, is insensitive to the presence of an
intraguild predator: an advantage of small body size? Oecologia 140: 577-585.

Rutledge, C. E., R. J. O'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.

SAS Institute 1999. SAS/STAT Users Guide First Edition, Vol. 1. computer program,
version By SAS Institute, Cary, NC.

SAS Institute 2007. JMP Version 7 computer program, version By SAS Institute, Cary,
NC.

Schaefer, P. W., R. J. Dysart, and H. B. Specht. 1987. North-American distribution of
Coccinella septempunctata (Coleoptera, Coccinellidae) and its mass appearance
in coastal Delaware. Environmental Entomology 16: 368-373.

243

 

Schmidt, M. H., and T. T scharntke. 2005. Landscape context of sheetweb spider
(Araneae : Linyphiidae) abundance in cereal ﬁelds. Journal of Biogeography 32:
467-473.

Schmidt, M. H., U. Thewes, C. Thies, and T. Tscharntke. 2004. Aphid suppression by
natural enemies in mulched cereals. Entomologia Experimentalis Et Applicata
113: 87-93.

Schmidt, M. H., I. Roschewitz, C. Thies, and T. Tscharntke. 2005a. Differential
effects of landscape and management on diversity and density of ground-dwelling
farmland spiders. Journal of Applied Ecology 42: 281-287.

Schmidt, M. H., I. Roschewitz, C. Thies, T. Tscharntke, and W. Nentwig. 2005b.
Spiders in Space: How landscape-wide movement of generalist predators
inﬂuences local density, species richness, and biocontrol, Second International
Symposium on Biological Control of Arthropods, Davos, Switzerland.

Schmidt, N. P., M. E. O'Neal, and P. M. Dixon. 2008a. Aphidophagous predators in
Iowa soybean: A community comparison across multiple years and sampling
methods. Annals of the Entomological Society of America 101: 341-350.

Schmidt, N. P., M. E. O'neal, and P. M. Dixon. 2008b. Aphidophagous predators in
Iowa soybean: A community comparison across multiple years and sampling
methods Annals of the Entomological Society of America 101: 341-350.

Shao, J. 2003. Mathematical statistics. Springer, New York.

Shi, A. [ed.] 1993. Effects of ecological interactions on forager diets: competition,
predation risk, parasitism and prey behavior. Wiley and Blackwell, Boston.

Simpson, E. H. 1949. Measurement of diversity. Nature 163: 688-688.

Snyder, W. E., and D. H. Wise. 1999. Predator interference and the establishment of
generalist predator populations for biocontrol. Biological Control 15: 283-292.

Snyder, W. E., and D. H. Wise. 2001. Contrasting trophic cascades generated by a
community of generalist predators. Ecology 82: 1571-1583.

244

Snyder, W. E., and A. R. Ives. 2003. Interactions between specialist and generalist

natural enemies: Parasitoids, predators, and pea aphid biocontrol. Ecology 84: 91-
107.

Snyder, W. E., and G. M. Clevenger. 2004. Negative dietary effects of Colorado potato
beetle eggs for the larvae of native and introduced ladybird beetles. Biological
Control 31: 353-361.

Snyder, W. E., and E. W. Evans. 2006. Ecological effects of invasive arthropod
generalist predators. Annual Review of Ecology Evolution and Systematics 37:
95-122.

Snyder, W. E., G. M. Clevenger, and S. D. Eigenbrode. 2004a. Intraguild predation
and successful invasion by introduced ladybird beetles. Oecologia 140: 559-565.

Snyder, W. E., S. N. Ballard, S. Yang, G. M. Clevenger, T. D. Miller, J. J. Ahn, T. D.
Hatten, and A. A. Berryman. 2004b. Complementary biocontrol of aphids by
the ladybird beetle Harmonia axyridis and the parasitoid Aphelinus asychis on
greenhouse roses. Biological Control 30: 229-235.

Soares, A. 0., and A. Serpa. 2007. Interference competition between ladybird beetle
adults (Coleoptera : Coccinellidae): effects on growth and reproductive capacity.
Population Ecology 49: 37-43.

Scares, A. 0., I. Borges, P. A. V. Borges, G. Labrie, and E. Lucas. 2008. Harmonia
axyridis: What will stop the invader? Biocontrol 53: 127-145.

Steffan—Dewenter, 1., U. Munzenberg, C. Burger, C. Thies, and T. Tscharntke. 2002.
Scale-dependent effects of landscape context on three pollinator guilds. Ecology
83: 1421-1432.

Swinton, S. M., F. Lupi, G. P. Robertson, and D. A. Landis. 2006. Ecosystem services
from agriculture: looking beyond the usual suspects. American Journal of
Agricultural Economics 88: 1160-1166.

Symondson, W. O. C., K. D. Sunderland, and M. H. Greenstone. 2002. Can generalist
predators be effective biocontrol agents? Annual Review of Entomology 47: 561 -
594.

245

Tewksbury, J. J ., D. J. Levey, N. M. Haddad, S. Sargent, J. L. Orrock, A. Weldon,
B. J. Danielson, J. Brinkerhoff, E. I. Damschen, and P. Townsend. 2002.
Corridors affect plants, animals, and their interactions in fragmented landscapes.
Proceedings of the National Academy of Sciences of the United States of America
99: 12923-12926.

Thies, C., and T. Tscharntke. 1999. Landscape structure and biological control in
agroecosystems. Science 285: 893-895.

Thies, C., I. Steffan-Dewenter, and T. Tscharntke. 2003. Effects of landscape context
on herbivory and parasitism at different spatial scales. Oikos 101: 18-25.

Thies, C., I. Roschewitz, and T. Tscharntke. 2005. The landscape context of cereal
aphid-parasitoid interactions. Proceedings of the Royal Society B-Biological
Sciences 272: 203-210.

Thompson, R. M., and C. R. Townsend. 2003. Impacts on stream food webs of native
and exotic forest: An intercontinental comparison. Ecology 84: 145-161.

Tscharntke, T., and R. Brandl. 2004. Plant-insect interactions in fragmented
landscapes. Annual Review of Entomology 49: 405-430.

Tscharntke, T., T. A. Rand, and F. J. J. A. Bianchi. 2005a. The landscape context of
trophic interactions: insect spillover across the crop-noncrop interface. Annales
Zoologici Fennici 42: 421-432.

Tscharntke, T., I. Steffan-Dewenter, A. Kruess, and C. Thies. 2002. Characteristics of
insect populations on habitat fragments: A mini review. Ecological Research 17:
229-239.

Tscharntke, T., A. M. Klein, A. Kruess, I. Steffan-Dewenter, and C. Thies. 2005b.
Landscape perspectives on agricultural intensiﬁcation and biodiversity -
ecosystem service management. Ecology Letters 8: 857-874.

VandenBerg, 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 farmers' ﬁelds in Indonesia.
Journal of Applied Ecology 34: 971-984.

246

 

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.

Venzon, M., A. Janssen, and M. W. Sabelis. 2002. Prey preference and reproductive
success of the generalist predator Orius laevigatus. Oikos 97: 116-124.

Voegtlin, D. J ., R. J. O'Neil, and W. R. Graves. 2004. Tests of suitability of
overwintering hosts of Aphis glycines: Identiﬁcation of a new host association
with Rharnnus alnifolia L'Heritier. Annals of the Entomological Society of
America 97: 233-234.

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, R. Y., and S. A. Ghabrial. 2002. Effect of aphid behavior on efﬁciency of
transmission of Soybean mosaic virus by the soybean-colonizing aphid, Aphis
glycines. Plant Disease 86: 1260-1264.

Weibull, A. C., O. Ostman, and A. Granqvist. 2003. Species richness in
agroecosystems: the effect of landscape, habitat and farm management.
Biodiversity and Conservation 12: 1335-1355.

Westphal, C., I. Steffan-Dewenter, and T. Tscharntke. 2006. Bumblebees experience
landscapes at different spatial scales: possible implications for coexistence.
Oecologia 149: 289-300.

Wheeler, A. G., and E. R. Hoebeke. 1995. Coccinella Novemnotata in northeastern
North-America - historical occurrence and current status (Coleoptera,
Coccinellidae). Proceedings of the Entomological Society of Washington 97: 701 -
716.

Wheeler, A. G., J. T. Hayes, and J. L. Stephens. 1968. Insect Predators of Murnmiﬁed
Pea Aphids. Canadian Entomologist 100: 221-&.

Wiens, J. A., R. L. Schooley, and R. D. Weeks. 1997. Patchy landscapes and animal
movements: Do beetles percolate? Oikos 78: 257-264.

247

 

Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying
threats to imperiled species in the United States. Bioscience 48: 607-615.

Wissinger, S. A. 1997. Cyclic colonization in predictably ephemeral habitats: a template
for biological control in annual crop systems. Biological Control 10: 4-15.

With, K. A., and A. W. King. 1999. Dispersal success on fractal landscapes: a
consequence of lacunarity thresholds. Landscape Ecology 14: 73-82.

With, K. A., S. J. Cadaret, and C. Davis. 1999. Movement responses to patch structure
in experimental ﬁactal landscapes. Ecology 80: 1340-1353.

Wyckhuys, K. A. G., and G. E. Heirnpel. 2007. Response of the soybean aphid
parasitoid Binodoxys communis to olfactory cues from target and non-target host-
plant complexes. Entomologia Experimentalis Et Applicata 123: 149-158.

Wyckhuys, K. A. G., R. L. Koch, and G. E. Heirnpel. 2007. Physical and ant-mediated
reﬁiges from parasitism: Implications for non-target effects in biological control.
Biological Control 40: 306-313.

Wyckhuys, K. A. G., J. E. Strange-George, C. A. Kulhanek, F. L. Wackers, and G.
E. Heirnpel. 2008. Sugar feeding by the aphid parasitoid Binodoxys communis:
How does honeydew compare with other sugar sources? Journal of Insect
Physiology 54: 481-491.

Yasuda, H. 2006. Special feature: biological control: theory and practice - Preface.
Population Ecology 48: 293-295.

Yoo, H. J. S., R. J. O'Neil, D. J. Voegtlin, and W. R. Graves. 2005. Plant suitability of
Rharnnaceae for soybean aphid (Homoptera : Aphididae). Annals of the
Entomological Society of America 98: 926-930.

248

 

 

 

11111111111311111111