.....

 

swam, lgéﬁggggﬁﬁsﬁ? E: fummnw

 

gigkwéuﬁﬁn u... . V. a. - amigﬁﬁmﬁ

 

“”4:ng

This is to certify that the
dissertation entitled

DO VARYING NATURAL ENEMY ASSEMBLAGES IMPACT
APHIS GL YCINES POPULATION DYNAMICS?

presented by

ALEJANDRO CARLOS COSTAMAGNA

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in EntomologL

 

r ofessor's Signature
§ 9>/o 5

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

LIBRARY
t Michigar $tate
Eggniversny

 

 

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

 

it?" $83 3%08

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/ClRC/DateDue.indd-p.1

DO VARYING NATURAL ENEMY ASSEMBLAGES IMPACT APHIS
GLYCINES POPULATION DYNAMICS?
By

Alejandro Carlos Costamagna

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Entomology
and
Program in Ecology, Evolutionary Biology and Behavior

2006

ABSTRACT

DO VARYING NATURAL ENEMY ASSEMBLAGES IMPACT APHIS
GLYCINES POPULATION DYNAMICS?
By
Alejandro Carlos Costamagna
The soybean aphid Aphis glycines Matsumura (Hemiptera: Aphididae), was

detected for the ﬁrst time in North America in Wisconsin during 2000, and it is currently
present in 22 states. Since its detection, it has developed outbreak populations that
resulted in up to 50% yield reduction during alternating years. I conducted ﬁeld studies
that demonstrated the crucial role of existing generalist natural enemies in suppressing
soybean aphid population growth below economic threshold levels. By using predator
exclusion cages, I demonstrated that ambient levels of predators exert strong suppression
on soybean aphid (3 to 20-fold reduction) in comparison with open controls and sham
cages. Through work at the NSF-Long Term Ecological Research site at the Kellogg
Biological Station of Michigan State University, I found that this control occurs over a
broad range of agricultural production systems, ranging from conventional practices with
or without tillage, to organic soybeans. This suppression occurs regardless of initial
aphid infestation, with aphid populations being reduced below threshold levels at high,
medium and low initial aphid densities. The strength of this suppression results in a
trophic cascade, with similar soybean biomass and yield in treatments where aphids were
exposed to predation as in treatments completely protected from aphids by aphid
exclusion cages. Direct ﬁeld observations revealed that the coccinellids Harmonia

axyridis and Coccinella septempunctata combined provided most of the mortality on

soybean aphid, were very effective on a per capita basis, and responded to aphid density
by increasing their abundance, time within the aphid patch, and consumption rate.
Exposure to predation signiﬁcantly shifted the aphid’s within-plant distribution towards

lower nodes on the plants, suggesting indirect effects of predation.

I also conducted studies using cages that differentially excluded large versus
small predators and parasitoids, which showed that ambient levels of naturally occurring
parasitoids provide only minor or no reduction of soybean aphid populations, even in
treatments in which parasitoids were provided with a reﬁige from intraguild predation by
coccinellids. Similar results were obtained in a separate study using artiﬁcial infestations
with the parasitoid Lysiphlebus testaceipes. I observed direct evidence of intraguild
predation on parasitoid mummies by generalist predators, in which up to 80% of
mummies were killed by predators, but percentage parasitism did not differ from
controls, suggesting that intraguild predation did not disrupt parasitism. Finally, I
developed a mathematical model for soybean aphid population grth by ﬁtting the data
of ﬁeld studies under predator exclusion conditions, using a novel approach that assumes
a linear decrease in the intrinsic rate of increase of the aphids. Overall, these ﬁndings
provide one of the ﬁrst empirical tests of recent theoretical models in which the role of
generalist predators is predicted to be an important source of insect population regulation.
In addition, these results support theoretical predictions that intraguild predators with
strong impacts on herbivores may not disrupt, or may even enhance biological control.
From the pest management perspective, it suggests the incorporation of natural enemies

as an important factor to take into account when determining soybean aphid thresholds.

To Marisol, Ailén and Chiara, for their continues love and support

iv

ACKNOWLEDGEMENTS

I would like to thank Dr. Doug Landis for being an outstanding advisor, providing
support, enthusiasm, dedication and patience. His vast knowledge on insect ecology and
biological control, high research standards, and encouragement to pursue interesting
questions were fundamental parts of my Ph.D. experience. I would also like to thank the
members of my Ph.D. Committee, Jim Miller, Mike Brewer, Chris DiFonzo and Phil
Robertson, their expertise and support contributed signiﬁcantly to improve the ﬁndings of
this research.

I am very grateful to all the Landis lab people, Chris Sebolt, Anna Fiedler, J eff
Evans and Mary Gardiner, for sharing ideas, resources, and providing support throughout
these studies, including hard days of work in the ﬁeld. Special thanks to Chris, who was
responsible for maintaining the campus study ﬁeld site, and coordinated the ﬁeld workers
very effectively each ﬁeld season. This study would have not been possible without the
hard work of many undergraduate helpers: M. Wood, R. MacDonald, C. Stahlman, J.
Steffen, T. Gibbons, A. Berro, N. Vuljaj, E. Knoblock, C. Richards, M. Wayo, R.
Alderson, T. Lehman, D. Richards, K. Newhouse and S. Gottschalk.

I would also like to thank P. Robertson, J. Simmons, A. Corbin and all the KBS-
LTER staff for their help in this study, including facilitating biomass and yield
measurements from the study site. I am grateful to D. Prokrym and the staff of the
USDA APHIS PPQ Laboratory at Niles, MI, for providing advice and support for rearing
A. glycines. Very helpful statistical advice was given by S. Kravchenko and X. Huang

(SCC-CANR). B. Graff generously helped process soybean samples (Agronomy Farm-

MSU). I am also indebted for the support provided by all the Faculty and staff in the
Department of Entomology at MSU, especially Fred Stehr, Ed Graﬁus, Rich Merritt, and
Jill Kolp. K. Pike, M. Brewer, T. Noma and G. Parsons provided their valuable expertise
for insect identiﬁcations. Finally, I am very grateful to Wopke van der Werf and Felix
Bianchi for all their support during the modeling part of this dissertation and for being
excellent hosts during my visit to Wageningen University.

Funding for this research was provided through a teaching assistantship from the
Department of Entomology, and by a USDA-CSREES grant 2004-35302-14811, the NSF
Long-Term Ecological Research Program at the Kellogg Biological Station, the Michigan
Agricultural Experiment Station, a MSU-Sustainable Agriculture grant, a summer
fellowship from the Ecology, Evolutionary Biology and Behavior program at MSU, and a
C. S. Mott Fellowship in Sustainable Agriculture.

Finally, many thanks to my dear wife Marisol and my daughters Ailén and
Chiara, whose continuous support, encouragement, understanding and love made the

completion of this degree possible.

vi

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES XI
LIST OF FIGURES XIV
KEY TO SYMBOLS AND ABBREVIATIONS ....................................... XVIII
CHAPTER 1
APHIS GLYCINES IN NORTH AMERICA 1
Introduction 1
Top-down and bottom-up forces in herbivore control 3
Trophic cascades - 4
Generalist versus specialist natural enemies 5
Top-down control by multiple enemy assemblages 7
Intraguild predation 7
Advantages and consequences of [GP 8
Factors affecting [GP 10
Impacts of 1GP on top-down control in agroecosystems 10
Intraguild predation in Aphid — Predator— Parasitoid Systems - 12
Types of predator-parasitoid [GP 14
Forage behavior of natural enemies and [GP 14
The Aphis glycines-natural enemy system 16
Aphis glycines life history .................................................................................. 16
Aphis glycines damage to soybeans ................................................................... I 7

vii

Factors affecting A. glycines abundance ............................................................ 18

Bottom-up effects ........................................................................................ 19

Top-down effects ......................................................................................... 19

The assemblage of natural enemies of A. glycines in Michigan ......................... 21
CHAPTER 2

PREDATORS EXERT TOP-DOWN CONTROL OF SOYBEAN APHID ACROSS

A GRADIENT OF AGRICULTURAL MANAGEMENT SYSTEMS

Abstract.

 

 

Introduction

 

Materials and methods

Study site .............................................................................................................
Soybean aboveground net primary production and yield ...................................
Aphid establishment ............................................................................................
Aphid population increase ..................................................................................
Aphid intrinsic rate of increase ..........................................................................
Natural population sampling ..............................................................................

Statistical analysis ..............................................................................................

Results

 

 

Soybean aboveground net primary production and yield ...................................
Aphid establishment ............................................................................................
Aphid population increase ..................................................................................
Aphid intrinsic rate of increase ..........................................................................

Natural population sampling ..............................................................................

Discussion

 

viii

24

24

25

27

27

28

29

30

31

32

33

34

34

35

37

41

41

45

CHAPTER 3
SUPPRESSION OF SOYBEAN APHID BY GENERALIST PREDATORS

 

 

 

 

 

 

 

 

 

 

RESULTS IN A TROPHIC CASCADE IN SOYBEAN S 50
Abstract 50
Introduction 51
Materials and methods 55

Statistical Analysis .............................................................................................. 60
Results 61
Effectiveness of predator manipulations ............................................................. 61
Parasitoid eﬂectiveness and intraguild predation .............................................. 64
Aphis glycines control ........................................................................................ 65
T rophic cascade .................................................................................................. 6 7
Discussion 68

CHAPTER 4

ARE EXISTING SOYBEAN APHID PARASITOIDS LIMITED BY INTRAGUILD

PREDATION? - 75
Abstract 75
Introduction _ - - 76
Material and methods 79

Field site ............................................................................................................. 79
Natural enemy manipulations ............................................................................. 80
Natural population sampling .............................................................................. 83
Statistical analysis .............................................................................................. 83
Results - - 85

 

ix

Aphis glycinesﬁeld versus caged populations ................................................... 85

 

 

 

 

 

 

 

 

 

 

 

Aphid suppression ............................................................................................... 86
Soybean biomass and yield ................................................................................. 90
Predator abundance ........................................................................................... 90
Parasitoid abundance and intraguild predation ................................................ 95
Discussion 97
CHAPTER 5
DOES APHIS GLYCINES HAVE REFUGES FROM TOP-DOWN CONTROL BY
GENERALIST NATURAL ENEMIES? 102
Abstract 102
Introduction 103
Materials and methods 104
Statistical analysis ............................................................................................ I 06
Results 106
Initial A. glycines density ................................................................................. 106
Within-plant distribution ................................................................................... 1 10
Discussion 116
CHAPTER 6
QUANTIFYING PREDATION ON SOYBEAN APHID THROUGH DIRECT
FIELD OBSERVATIONS _ 121
Abstract 121
Introduction 122
Material and methods 124
Statistical analysis ............................................................................................ 126
Results - 127

 

Discussion 139

 

 

 

 

 

 

 

 

 

 

 

 

CHAPTER 7
PREDICTIVE MODEL FOR THE GROWTH OF FIELD POPULATIONS OF
APHIS GL YCHVES (HEMIPTERA: APHIDIDAE) IN SOYBEAN 145
Abstract 145
Introduction 146
Materials and methods 148
Study sites and ﬁeld experiments ...................................................................... 148
Models of A. glycines growth ........................................................................... 151
Statistical analysis ............................................................................................ 152
Results 153
Discussion - - 160
CHAPTER 8
CONCLUSIONS AND FUTURE DIRECTIONS 163
Aphis glycines invasion in North America 163
Top-down versus bottom-up forces in A. glycines population control .............. 164
Aphis glycines natural enemy assemblages and intraguﬂd predation ............... 165
Aphis glycines — predator interactions 167
APPENDIX - 170
Appendix 1 171
REFERENCES CITED 174

 

xi

LIST OF TABLES

Table 2.1. Poisson regression results of the effect of agricultural management system
and predator manipulation treatments on the establishment of A. glycines at the KBS-
LTER site, Michigan, 2003. ..................................................................... 35

Table 2.2. ANOVA results for a) ﬁxed, and b) random effects, and c) slicing tests results
of the effect of agricultural management system and predator manipulation treatments on
A. glycines within large ﬁeld cages (1n [number of aphids / plant + 1]) at the KBS-LTER
site, Michigan, 2003. .............................................................................. 38

Table 2.3. Mean (i SE) of large- and small-size predatorsl (individuals / m2) and
parasitoids (L. testaceipes mummies / m2) for different combinations of agricultural
management system and predator manipulation treatments, within the large ﬁeld cages in
the aphid population increase experiment the KBS-LTER site during 2003. ............. 42

Table 2.4. Mean longevity (days, d: SE), fecundity (total females / female) and intrinsic
rate of increase (r, females/female/day*) of A. glycines reared on soybean produced under
three different agricultural management systems in the KBS-LTER during 2003. ...... 43

Table 2.5. MANOVA table for the effect of agricultural practices on the ﬁve more
abundant A. glycines foliar predators at the KBS-LTER site during 2003. ............... 45

Table 3.1. ANOVA results of predator and parasitoid manipulations on the abundance of
predators, parasitoids and aphids using ﬁeld cages in Michigan during 2004. ........... 62

Table 4.1. Results of the ANOVA for effects of year, agronomic treatment, and natural
enemy treatment on A. glycines abundance. .................................................. 87

Table 5.1. ANOVA results for the effect of initial aphid density and natural enemy
control on the population growth of A. glycines natural populations in Michigan. 107

Table 5.2. MANOVA results for the effects of different initial aphid densities and
natural enemy manipulations on the within-plant distribution of A. glycines. .......... 112

Table 6.1. Predator abundance, total aphids consumed (TC), consumption time and
residence time, observed per 1112 during 2 h periods in a soybean ﬁeld in Michigan. 128

Table 6.2. Percent predation of A. glycines by different predators on different plant parts

and aphid sizes, recorded by direct observation of 211 predation events in a soybean ﬁeld
in Michigan during 2005. ........................................................................ I37

xii

Table 7.1. Characteristics of four experiments studying A. glycines population dynamics
under ﬁeld conditions in predator exclusion cages. For a detailed description of each
experiment see the methods section and Costarnagna 2006). ............................. 150

Table 7.2. Parameters (mean i SE) obtained by ﬁtting three population grth models
to mean population abundances of A. glycines in four ﬁeld experiments in Michigan,
USA (for details on the parameters see the methods section). ............................ 154

Table 7.3. Parameters (mean t SE) obtained by ﬁtting the decreasing r model to

observations scaled to days after infestation, A. glycines degree-days and soybean degree-
days as time scales. At the bottom, results of the statistical analysis testing the assumption
of linear decrease of r in time are presented for each model. .............................. 158

xiii

LIST OF FIGURES

Figure 2.1. Effect of natural enemies and agricultural management system on the
number of A. glycines (mean t SE) alive after 24 h at the KBS-LTER site, Michigan,
2003. Different letters above bars indicate signiﬁcant differences among predator
manipulation treatments (P < 0.05, LSMD , SAS Institute 2001) .......................... 36

Figure 2.2. Effect of natural enemies and agricultural management systems on the
number of A. glycines (mean :I: SE) after a) 6 days, and b) 14 days of manipulation using
large ﬁeld cages at the KBS-LTER site, Michigan, 2003. Different letters above bars
indicate signiﬁcant differences among predator manipulation treatments (P < 0.0001 ,
LSMD , SAS Institute 2001). ................................................................... 40

Figure 2.3. Un-manipulated A. glycines (a) and natural enemy (b-f) populations (mean :l:
SE) in soybeans under conventional, no-till and zero-chemical input treatments at the
KBS-LTER site during 2003. Asterisks above lines indicate signiﬁcant differences
among agricultural management system treatments within sampling dates (P < 0.05)
(LSMD , SAS Institute 2001). ................................................................... 44

Figure 3.1. Design of predator exclusion cages (A) used in aphid and parasitoid
treatments, and sham cages (B) used in predator and predator + parasitoid treatments.
The ﬁne mesh (< 1 mm) prevents aphid, parasitoid and predator movement, whereas the
coarse mesh (2 mm openings) prevents only large predator access (i. e. mainly
Coccinellidae). All cages were covered on top by ﬁne mesh. Ground-dwelling predator
movement was reduced by means of a transparent plastic barrier buried in the soil (see
details in the methods) ............................................................................ 57

Figure 3.2 Mean (i SE) abundance of a) coccinellids, b) small predators (midge larvae
and 0. insidiosus combined), and c) parasitoid mummies in ﬁeld cages with A. glycines
that combine predator and parasitoid manipulations. The arrow indicates the opening of a
canopy-level window (see methods). Different letters within each sampling date
represent signiﬁcant differences (p < 0.05) using slicing tests for the 2 and 3-way
signiﬁcant interactions (Table 3.1); ns = not signiﬁcant. .................................... 63

Figure 3.3 Mean (t SE) number of A. glycines in ﬁeld cages with different combinations
of predators and parasitoids. The arrow indicates the opening of a canopy-level window
(see methods). Asterisks indicate signiﬁcant differences between treatments with
ambient levels of predators versus predator exclusion treatments (predator x date slicing
tests, P < 0.0005). ................................................................................. 66

Figure 3.4. Mean (i SE) soybean yield and above ground dry biomass in ﬁeld cages
without A. glycines (no-aphid control), and with A. glycines and different combinations of
predators and parasitoids. Different letters among columns represent signiﬁcant
differences (P < 0.05, LSMD-TK). .............................................................. 68

xiv

Figure 4.1. Open ﬁeld populations of A. glycines at the biodiversity plots of the Kellogg
Biological Station — Long Term Ecological Research site in Michigan during 2004 and
2005. Asterisks above the line denote signiﬁcant differences between years in equivalent
sampling weeks (pro-planned contrasts, P < 0.001 , performed on initial six weeks of
sampling). ......................................................................................... 86

Figure 4.2. Effect of natural enemy treatments on the number of A. glycines (mean 3:
SE) during a) 2004, and b) 2005. Different letters on each sampling date indicate
signiﬁcant differences (P < 0.05, Least Mean Square Differences adjusted by the
sequential Bonferroni method for multiple comparisons). Statistical comparisons were
restricted to the six initial sampling dates, when aphids reach peak populations. ...... 88

Figure 4.3. Effects of natural enemy manipulations on soybean a) above ground biomass
and b) yield (g of seed / plant), in the biodiversity plots of the KBS—LTER site, during
2004 and 2005. Different letters indicate signiﬁcant differences (P < 0.05, LSMD,
adjusted by the sequential Bonferroni method) among treatments within the same year
(capital letters for 2004, lower case letters for 2005). ....................................... 91

Figure 4.4. Open ﬁeld predator populations sampled in the biodiversity plots of the
KBS-LTER site during 2005. Large-size predators includes nabids, spiders, small
carabids, mirids and lacewing larvae; small-size predators includes midge and syrphid
larvae .............................................................................................. 92

Figure 4.5. Effect of natural enemy manipulations on the abundance (mean + SE) of the
four major groups of predators sampled in the biodiversity plots of the KBS-LTER
during 2004 and 2005. Asterisks above treatment lines indicate signiﬁcant differences
among natural enemy treatments, asterisks below sampling dates indicate signiﬁcantly
higher abundance for all treatments combined in that year (Kruskal-Wallis tests, * = P <
0.10, ** = P < 0.05, *** = P < 0.01). Statistical comparisons were performed for the six
initial weeks. ..................................................................................... 94

Figure 4.6. Effect of predator manipulations on a) number of mummies / plant, b)
percentage parasitism (proportion of mummies to aphids x 100), and c), number of
mummies damaged by predators; averaged across agronomic treatments. Only one
mummy was obtained in the open treatment and none in the aphid treatment, and
therefore these treatments were excluded from statistical comparisons. Treatments were
compared independently for each date using Wilcoxon tests (* = P < 0.10). . . . 96

Figure 5.1. A. glycines population growth from three initial densities under three
predator manipulation treatments. Mean (i SE) aphid abundance is shown at a) before
natural enemy manipulation, and b) one week, and c) two weeks after manipulation.
Different letters represent signiﬁcant differences among natural enemy treatments within
patch density treatments (P < 0.05, Least Square Mean Differences). .................. 108

XV

Figure 5.2. Mean (:l: SE) per capita rate of increase of A. glycines at three initial
densities under three natural enemy manipulation treatments after 14 days. Signiﬁcant
main effects are described in the text. ........................................................ 109
Figure 5.3. A. glycines population growth under three predator manipulation treatments.
Mean (:l: SE) aphid abundance is shown at a) three, c) four, and c) ﬁve weeks after
predator manipulation. Plots were paired according to aphid density in the exclusion
treatment at week 2 and randomly chosen to switch exclusion and sham cages (switched)
or remained un—switched (control). Control treatments were interrupted at week four due
to the extraordinary high aphid densities achieved, and therefore week ﬁve was analyzed
separately (see methods). Different letters represent signiﬁcant differences among
natural enemy treatments (P < 0.05, LSMD adjusted by the sequential Bonferroni method
for multiple comparisons). ...................................................................... 1 1 1

Figure 5.4. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and c) low densities before assigning the predator manipulation treatments.
Cages were placed immediately after sampling aphid abundance. The statistics for this
ﬁgure are presented in Table 5.2. .............................................................. 1 13

Figure 5.5. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and c) low densities one week after predator manipulation treatments. The
statistics for this ﬁgure are presented in Table 5.2. .......................................... 114

Figure 5.6. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and c) low densities two weeks aﬁer predator manipulation treatments. The
statistics for this ﬁgure are presented in Table 5.2. ......................................... 115

Figure 5.7. Within-plant distribution of naturally occurring A. glycines at a) three, and b)
four weeks after predator manipulation treatments. Exclusion and sham cages were not
switched, serving as controls for the switching treatment (see Figure 5.8). ............. 117

Figure 5.8. Within-plant distribution of naturally occurring A. glycines at a) three, and b)
four weeks aﬁer predator manipulation treatments. Exclusion and sham cages were
switched after the second week. ............................................................... 118

Figure 6.1. A. glycines abundance in the soybean ﬁeld observed for predation,
Entomology Farm, Michigan State University, during 2005. .............................. 130

Figure 6.2. Mean t SE, a) predator abundance, and b) total, and c) per capita rate of
consumption of A. glycines by the most abundant predator groups observed during 2 h
periods in a soybean ﬁeld in Michigan, USA. [-1. axyridis and C. septempunctata
combine adults and 3rd and 4th instar larvae, Coccinellid larvae includes all the 1St and 2“d
instar larvae, since they were not identify to species. Different letters indicate signiﬁcant
difference between predator groups (LSMD-TK, P < 0.05). Per capita rates of
consumption of midge and lacewing larvae were not included in the statistical
comparisons due to their low ﬁequency (15% of cases). .................................. 132

xvi

Figure 6.3. Responses of resident (closed symbols) and transient predators (open
symbols) to aphid density: a) predator abundance, b) total consumption rate, and c) per
capita consumption rate. Resident predators included 0. insidiosus, midge, lacewing, and
coccinellid larvae. Transient predators included coccinellids adults, Nabis spp.,
Plagiognathus spp., and C. associatus. Lines indicate linear trends, r is Pearson’s
correlation coefﬁcient for the closest line, * indicates P < 0.10, ** indicates P < 0.05, ***
indicates P < 0.01, ns = not signiﬁcant. ...................................................... 134

Figure 6.4. Correlations between aphid abundance and the abundance of a) H. axyridis
adults, b) C. septempunctata larvae, and c) 0. insidiosus nymphs. Other predator groups
did not show signiﬁcant associations. Lines indicate linear trends, r is Pearson’s
correlation coefﬁcient, * indicates P < 0.10, ** indicates P < 0.05, *** indicates P < 0.01 ,
ns = not signiﬁcant. .............................................................................. 135

Figure 6.5. Correlations between aphid abundance and total aphids eaten by a) H.
axyridis adults, b) C. septempunctata adults, and c) C. septempunctata larvae. Other
predator groups did not show signiﬁcant associations. Lines indicate linear trends, r is
Pearson’s correlation coefﬁcient, * indicates P < 0.10, ** indicates P < 0.05, ***
indicates P < 0.01. .............................................................................. 138

Figure 6.6. Correlation between aphid abundance and per capita consumption of C.
septempunctata adults. Other predator groups did not show signiﬁcant associations.

Line indicates the linear trend, r is Pearson’s correlation coefﬁcient, * indicates P < 0.10.
...................................................................................................... 140

Figure 6.7. Signiﬁcant correlations between aphid abundance and residence time of C.
septempunctata adults. Other predator groups did not show signiﬁcant associations. Line

indicates the linear trend, r is Pearson’s correlation coefﬁcient, ** indicates P < 0.05.
...................................................................................................... 140

Figure 7.1. Observed population dynamics of A. glycines in four ﬁeld experiments
using predator exclusion cages and the ﬁt of three competing models, for experiments 1
(a), 2 (b), 3 (c) and 4 ((1). Observations (mean at: SE) are indicated by points and the
exponential, logistic, and decreasing r models by dashed, dotter and solid lines,

respectively. Parameter values and statistics for these models are given in Table 7 .2.
...................................................................................................... 155

Figure 7.2. Fit of the decreasing r model (solid line) when ﬁtted to observations of the
four ﬁeld experiments combined, scaled to days after infestation (a), A. glycines degree-
days (b), and soybean degree-days (c). Different symbols represent the means of the
four experiments. Parameter values for these models are given in Table 7 .3. . 159

xvii

ANOVA
oC

cm

df

RH
SD

SE

spp.

KEY TO SYMBOLS AND ABBREVIATIONS

analysis of variance
degrees Celsius
centimeters

days

degrees of freedom
Fisher distribution
grams

hours

meters

milimiters

number of observations

not signiﬁcant

probability under the assumption that the null hypothesis is true

Pearson’s coefﬁcient of correlation
coefﬁcient of determination
relative humidity

standard Deviation of the Mean
standard Error of the Mean

species

t-statistic

xviii

CHAPTER 1

APHIS GL Y CINES IN NORTH AMERICA

Introduction

The soybean aphid, Aphis glycines, Matsumura (Hemiptera: Aphididae) is an
important pest of soybean (Glycine max L.) in China and Southeast Asia, causing yield
losses up to 70% (Wang et al. 1962, Hirano et al. 1996, Van den Berg et al. 1997, Wu et
al. 2004). In the year 2000, A. glycines was discovered for the ﬁrst time in North
America and Australia (Fletcher and Desborough 2000, Wedburg 2000, Venette and
Ragsdale 2004). Since 2000, it has expanded its distribution to 22 states in the USA and
three provinces of Canada (Losey et al. 2002, Hunt et al. 2003, Landis et al. 2003,
Venette and Ragsdale 2004, Mignault et a1. 2006). In 2003, over 7 million acres of
soybean were sprayed with insecticides against A. glycines, which infested over 50
million acres of soybean in the USA (Landis et al. 2003). Before the arrival of A.
glycines, soybean in the North Central USA had few serious insect pests and was not
usually treated with insecticides. Thus, soybean has played an important role as reservoir
of natural enemies for other crops (Heimpel and Shelly 2004). Previous studies have
shown signiﬁcant impacts of natural enemies on A. glycines in Asia (Van den Berg et al.
1997, Liu et al. 2004, Wu et al. 2004). In the US, recent studies have also demonstrated a
key role of natural enemies in suppressing A. glycines populations (Fox 2002, Fox and
Landis 2003, Fox et a1. 2004, Rutledge et a1. 2004, Fox et al. 2005, Nielsen and Hajek

2005, Rutledge and O'Neil 2005). Generalist predators currently dominate the

assemblage of A. glycines natural enemies, with parasitoids only rarely observed ﬁom
2000-2002. Interestingly, in 2003, surveys in Michigan, Minnesota and New York have
revealed the presence of both native and introduced parasitoid species beginning to
appear in natural enemy assemblages (Brewer and Noma 2003, Heimpel and Wu 2003,
Nielsen and Hajek 2005). Observations indicate the presence of parasitoids, particularly
Lysiphlebus testaceipes (Cresson) (Hymenoptera: Braconidae) in 8 of 12 North Central
States (Landis et al. 2003).

The invasion of A. glycines into North America provides a novel system to test
ecological theory related to herbivore population dynamics. These include possible tests
of the relative strength of bottom-up and top-down forces in the control of herbivorous
species, and the interactions between generalist and specialist natural enemies,
particularly in regard to the impact of intraguild predation on top-down control. The very
limited levels of ﬁeld parasitism to date provide a unique opportunity to explore the role
of intraguild predation in inﬂuencing the integration of native parasitoids into natural
enemy assemblages. This information could be used to anticipate the potential beneﬁts
and desired traits that introduced parasitoid species (classical biological control agents)
should have to be successful in this system. In this dissertation, I ﬁrst studied the relative
role of top-down versus bottom-up control of A. glycines (Chapter 2). Then, after a
prominent role of top-down regulation was established, I investigated if this resulted in a
trophic cascade and whether it was affected by trophic interactions within the natural
enemy guild, particularly intraguild predation (Chapters 3 and 4). I then investigated the
dynamics of predation at the within ﬁeld level assessing predation at different initial

aphid densities and different plant parts (Chapter 5). Finally, I assessed the relative

importance of each predator group attacking A. glycines through direct ﬁeld observations
(Chapter 6), and combining data from all my experiments I developed a model of A.
glycines population growth under ﬁeld conditions as a tool to investigate potential

controls for this pest (Chapter 7).

Top-down and bottom-up forces in herbivore control

Hairston et al. (1960) proposed that top-down control of herbivore abundance by
consumers was the dominant force shaping natural communities. This view proved to be
controversial and it was subsequently proposed that in several communities bottom-up
(plant resources) are the dominant forces (Murdoch 1966, Ehrlich and Birch 1967), a
view supported by several empirical studies (Hunter et a1. 1997, Denno et al. 2002,
Gratton and Denno 2003b). Latter reviews suggest emerging consensus that both top-
down and bottom-up forces are present in natural communities and that a more
appropriate approach is to estimate their relative strength within ecosystems (Hunter and
Price 1992, Power 1992, Walker and Jones 2001). Most empirical research in the role of
these two forces has been conducted in aquatic ecosystems, but several recent studies
have focused on terrestrial ecosystems (Dyer and Letoumeau 1999, Walker and Jones
2001, Moran and Scheidler 2002, Gruner 2004). Several complex interactions and
factors have been shown to signiﬁcantly affect the relative strength of top-down versus
bottom-up forces in terrestrial ecosystems. Among these are: environmental productivity
(Power 1992, Fraser and Grime 1998, Dyer and Letoumeau 1999, Fraser and Grime
1999, Moran and Scheidler 2002), consumer efﬁciency (Power 1992), temporal versus

spatial variation of herbivore populations (Hunter et al. 1997), herbivore life stages

(Walker and Jones 2001), year-to-year (Walker and Jones 2001 , Gratton and Denno
2003b) and seasonal changes in abiotic and biotic conditions (Boyer et a1. 2003, Gratton
and Denno 2003a), differential predation risk among herbivore species (Denno et al.
2003), and herbivore body size (Sinclair et al. 2003). No studies explicitly testing the
relative role of top-down versus bottom-up forces has been conducted in agroecosystems,
although comparisons of the effects of fertility regimes and host plant resistance (bottom-
up) versus the effects of natural enemies (top-down) was utilized by Walker and Jones

(2001) to infer the strength of this interaction from applied studies.

Trophic cascades

Higher order trophic levels can affect the abundance, biomass or productivity of
populations across more than one link in a food web in a phenomenon termed a “trophic
cascade” (Pace et al. 1999). There is still an open debate regarding the prevalence and
importance of trophic cascades in different ecosystems, and particularly on what
ecosystem characteristics lead to trophic cascades (Pace et al. 1999, Polis et al. 2000).
Empirical evidence shows that trophic cascades frequently occur in more simple systems
(i.e. less reticulated food webs) with many cases documented in aquatic environments,
leading to the postulate that “all trophic cascades are wet” (Strong 1992, Pace et al.
1999). However, several studies have demonstrated trophic cascades in terrestrial
ecosystems (Fraser and Grime 1998, Moran and Hurd 1998, Dyer and Letoumeau 1999,
Schnritz et al. 2000, Halaj and Wise 2001, Terborgh et al. 2001, Moran and Scheidler
2002, F inke and Denno 2004, 2005), and particularly in agroecosystems (Carter and

Rypstra 1995, Snyder and Wise 1999, Colfer and Rosenheim 2001, Snyder and Wise

2001, Cardinale et al. 2003, Lang 2003). Several factors postulated as important
characteristics of systems exhibiting trophic cascades are present in agroecosystems.
They include: homogeneous habitat, prey turnover faster than predator turnover, prey
uniformly edible, and simple trophic stratiﬁcation of the system (Polis et al. 2000, Miiller
and Brodeur 2002). These factors make agroecosystems suitable subjects for study
trophic cascades and is in fact form the theoretical foundation of biological control
practices, i.e. a reduction of plant damage (with expected increased in crop yield) through
deleterious effects of natural enemies (native or imported) on herbivore pests (Polis et al.

2000)

Generalist versus specialist natural enemies

There is increasing consensus supported by empirical evidence that top-down
control is typically exerted by communities of multiple enemies rather than single species
(Losey and Denno 1998a, Sih et al. 1998, Chang and Kareiva 1999, Symondson et al.
2002). This view contrasts with traditional biological control practices that have focused
on the use of one or a few natural enemy species and that have generally favored
specialist (or at least stenophagous species) rather than assemblages of generalist
(polyphagous) species (Chang and Kareiva 1999, Symondson et al. 2002). The role of
generalists as effective top-down regulators of herbivores in agroecosystems has received
more support in recent theoretical and empirical reviews (Chang and Kareiva 1999,
Symondson et al. 2002, Stiling and Comelissen 2005). The traditional view suggest that
specialists have increased capacity to control pests due to 1) a stronger numerical

response (i.e. more offspring produced per prey than generalists) and 2) a greater

functional response that is not limited by satiation in the case of parasitoids or diminished
by the abundance of alternative prey (Symondson et a1. 2002). In addition, Kindlrnann
and Dixon (2001) postulate that large relative generation time ratios between predators
and prey prevent predators from exerting top-down control. They suggest herbivore prey
are either regulated by bottom-up processes or self-regulated.

Utilization of generalists species in classical biological control programs (i.e.
importation of exotic generalists) is not a viable practice due to potential non-target
effects (Howarth 1983, 1991, Simberloff and Stiling 1996, Louda et al. 2003, Stiling and
Comelissen 2005). However, the use of native or previously established exotic
generalists may have several advantages (Symondson et al. 2002). Generalist natural
enemies have the potential to be present early in the season, before the pest arrives or
builds up its population in the ﬁeld (Chang and Kareiva 1999, Symondson et al. 2002).
Using a simple theoretical model, Chang and Kareiva (1999) demonstrated that early
predation by generalist predators has a similar effect as latter immigration of more
specialized natural enemies. Landis and Van der Werf (1997) showed that the
assemblage of generalist predators present in sugar beet ﬁelds early in the season
signiﬁcantly diminished the population of aphids and the impact of viruses transmitted by
the aphids. Generalist predators are typically adapted to use several different habitats and
can therefore re-invade agricultural ﬁelds faster than specialists after disturbances
(Symondson et al. 2002). Finally, assemblages of generalists have the potential to
contain natural enemies that attack all the stages of the pests throughout the season,
avoiding stage speciﬁc or temporal refuges that may limit more specialized species

(Symondson et al. 2002).

Top-down control by multiple enemy assemblages

Multiple enemy assemblages are typically composed of both generalist and
specialist natural enemies (Symondson et al. 2002). Within consumer assemblages,
interactions such as predator interference, cannibalism, predator avoidance behavior, and
intraguild predation may modify the strength of top-down control of herbivores (Sih et a1.
1998, Snyder and Wise 1999, Symondson et a1. 2002), and multiple enemies within an
assemblage may interact to have synergistic, additive, or non-additive effects (Polis et al.
1989, Sih et al. 1998, Dixon 2000). Synergist interactions among natural enemies result
in greater impact on the prey population than that expected by adding individual predator
impacts (Ferguson and Stiling 1996, Dixon 2000). Additive effects occur when the effect
of natural enemies together is equivalent to adding the effects of each one separately
(Ferguson and Stiling 1996, Dixon 2000). For this to occur, there has to be either an
absence of direct (i.e. predation) or indirect (i.e. competition, interference) interactions
among natural enemies or an overall compensation of interactions. Finally, non-additive
effects can result when the control exerted by all the enemies together is less than the
addition of the effect of each one separately, indicating negative interactions among

consumers (Ferguson and Stiling 1996, Dixon 2000).

Intraguild predation
The presence of generalist natural enemies often leads to the existence of intraguild
predation, i.e. trophic interactions among members of the same guild (Polis et al. 1989).

Intraguild predation (IGP hereafter) is a particular case of omnivory in which consumer

species interact eating each other (direct effect, +-), and/or competing for a shared
resource (indirect effect, --) (Polis et a1. 1989). IGP is common in many trophic webs
including ﬁ'eshwater and marine aquatic ecosystems as well as terrestrial ecosystems
(Polis et a1. 1989, Rosenheim et al. 1995, Rosenheim 1998, Brodeur and Rosenheim
2000, Miiller and Brodeur 2002). Two characteristics have been commonly associated
with 1G predators: larger size than other guild members, and a broad range diet (Polis et

al. 1989).

Advantages and consequences of IGP

The widespread existence of IGP interactions in trophic webs could be attributed
to several advantages received by the IG predator. IGP provides an alternative source of
energy to the IG predator, reduces the number of competitors, and in the cases of mutual
IGP also reduces predation risk (Brodeur and Rosenheim 2000). Mutual IGP refers to
symmetrical interactions between natural enemies, i.e. both predators have similar
capacities to prey on each other (Lucas et al. 1998), and thus reducing the number of the
other predator species not only diminishes competition, but also reduces the chances of
been eaten (Brodeur and Rosenheim 2000). For IG omnivores that primarily feed on
plants, IGP could also represent an important source of nitrogen in the diet (Polis et al.
1989). It has been shown that IGP can increase survivorship during periods of low
abundance of the primary prey, by both direct increase in food intake and indirect beneﬁt
in reducing competition (Polis et al. 1989). This may be especially important in
ephemeral habitats, where IG predators that use both producer and consumer trophic

levels (6. g. Orius spp.) can persist under low prey populations by using plant resources,

and therefore be present in the ﬁeld with very low prey densities (C011 and Guershon
2002)

In order to achieve strong IGP, natural enemies or at least those acting as IG
predators should aggregate in response to prey density (Rosenheim 1998, Miiller and
Brodeur 2002). As an example, Tscharntke (1992) showed that parasitoids attacking
hosts at high host densities were more affected by bird predation, whereas parasitoid
species that prefer low host densities were relatively unaffected by avian IGP. Parasitoid
species that showed a density-independent response to hosts were affected by IGP
proportionally to their abundance (Tscharntke 1992). IGP has a direct impact on the
predatory guild by removing more susceptible genotypes or by promoting niche shifting
(Polis et al. 1989). Niche shifting may be achieved through behavioral modiﬁcations,
morphological adaptations or development of toxins (Polis et al. 1989). Theoretical
models predict that stability between guild members that engage in IGP can be achieved
when the IG prey is a better exploiter of the shared resource than the IG predator, a
ﬁnding also documented in the ﬁeld (Polis et a1. 1989, Holt and Polis 1997, Miiller and
Brodeur 2002). When both members of the guild have similar capacities to consume
each other, priority effects (relative initial densities) are predicted to inﬂuence the
outcome of the interaction, with the ﬁrst species arriving in the habitat usually displacing
the other (Polis et al. 1989, Miiller and Brodeur 2002). Finally, equilibrium models
showed that although IGP could result in stable equilibria, these may occur at a higher
density of the shared resource (Briggs and Borer 2005). Therefore, IGP may result in a

lower level of biological control, if the shared resource is an agricultural pest.

Factors affecting IGP

The intensity of IGP is also predicted to vary in habitats with different
productivities (Miiller and Brodeur 2002). In highly productive habitats theory predicts
an increase in the presence of the IG predator, whereas the contrary is true for low
productivity habitats (Polis et al. 1989, Miiller and Brodeur 2002). This prediction was
empirically demonstrated by the differential abundance of two parasitoids of the
California red scale that engage in IGP on three citrus cultivars representing a gradient of
productivity (Borer et al. 2003). Structural complexity is another factor that mediates the
intensity of IGP. Finke and Denno (2002) showed signiﬁcantly less IGP between the
mirid T yttus vagus and the wolf spider Pardosa Iittoralis in patches of vegetation with
dense thatch (structurally complex) than in patches in which thatch was removed. The
abundance of extraguild prey is also predicted as an important factor mediating IGP, and
was shown to modify the outcome of the interactions between Cecidomyiidae,
Coccinellidae and Chrysopidae species that prey on the potato aphid, Macrosiphum

euphorbiae (Lucas et al. 1998).

Impacts of IGP on top-down control in agroecosystems
IGP can cause disruption, enhancement or have no effect on top-down control of
herbivores (Polis et al. 1989, Rosenheim et al. 1995, Rosenheim 1998, Brodeur and
Rosenheim 2000, Miiller and Brodeur 2002, Symondson et al. 2002). Disruptions are
caused by non-additive interactions that can cascade down producing an increase in the
population of the herbivore (Rosenheim et al. 1995). In natural ecosystems top-down

control is exerted more commonly by predators than parasitoids (Hawkins et al. 1999).

10

F inke and Demo (2003) demonstrated that together the the mirid T yttus vagus and the
wolf spider Pardosa Iittoralis exerted less control on Prokelisia planthoppers than the
two predator species alone, indicating strong disruption of top-down control due to IGP .
However, other studies shown no impact of IGP on top-down control of herbivores in
natural ecosystems (Polis et al. 1989).

Although IGP is likely to occur in most natural enemy communities in
agroecosystems, very few studies had been conducted to quantify its impact on herbivore
regulation (Rosenheim et al. 1995). In agroecosystems, top-down control of herbivores
has been shown to be predominantly the result of parasitism rather than predation
(Hawkins et al. 1999). Empirical models predict both enhancement or disruption of
biological control by IGP (Rosenheim et al. 1995). However, most models assume that
attacks on each prey is based on its relative abundance, thus no preference between [G or
extraguild prey, which at least in some systems seems not to be the case (Rosenheim et
al. 1995). Where biological control has been disrupted by IGP, predator-predator IGP
interactions dominate (Rosenheim et al. 1995). This system differs fundamentally from
parasitoid-predator and pathogen-parasitoid IGP in that in these last two systems IGP
occurs within a shared host, and therefore high levels of IGP also imply high host
mortality (Rosenheim et a1. 1995). By contrast, in predator-predator IGP systems, high
levels of IGP may cause a relaxation of the top-down control on the herbivore, leading to
pest increase (Rosenheim et al. 1995, Prasad and Snyder 2006).

Despite this general prediction, several ﬁeld experiments have shown either no
effect, or enhancement of top-down control by assemblages of predators that engage in

IGP. Lang (2003) showed that despite strong IGP, winter wheat aphid p0pulations were

11

signiﬁcantly more impacted by ground beetles (Carabidae) and spiders (Lycosidae,
Lynphiidae) together than in plots were only one group of predators was present. In other
systems, predator facilitation has been demonstrated to result in synergistic effects among
predators. The impact of the coccinellid Harmonia axyridis, the damsel bug Nabis sp.,
and the parasitoid Aphidius ervi together on the aphid Aphis craccivora was signiﬁcantly
higher than the summed impact of the three enemies alone (Cardinale et al. 2003). In
greenhouse conditions, the addition of the predator Delphastus pusillus to the enemy
guild composed by the parasitoids Encarsiaformosa and E. pergandiella also increased
control of the whiteﬂy Bemisia argentifolii, but in this case the predator was more
effective than the parasitoids alone and also showed a signiﬁcant non-preference for older
parasitized hosts or parasitoid mummies (Heinz and Nelson 1996). A similar synergistic
effect was found between introduced parasitoids of the winter moth Operophtera
brumata (L.) and staphilinid predators that attack proportionally less parasitized than
unparasitized winter moth pupae (Roland and Embree 1995). Finally, in parasitoid-
predator systems where the predator showed strong IGP and prey have a refuge ﬁom
predation but not from parasitism, IGP is predicted to signiﬁcantly reduce biological

control (Brodeur and Rosenheim 2000).

Intraguild predation in Aphid — Predator— Parasitoid Systems
Predator-parasitoid IGP is usually asymmetric, with all the stages of the parasitoid
vulnerable to IGP, but with no attack of the predator by the parasitoid (Brodeur and
Rosenheim 2000, Meyhiifer and Klug 2002). In this way, IGP has been suggested as a

mechanism that may explain the failure of aphid parasitoids to regulate their hosts

12

(Ferguson and Stiling 1996, Rosenheim 1998, Brodeur and Rosenheim 2000, Colfer and
Rosenheim 2001). Generalist predators commonly attack immature parasitoid stages,
while the adult parasitoid is not typically consumed because of its mobility (Rosenheim
et al. 1995, Brodeur and Rosenheim 2000). However, predation on adults has been
reported and can achieve levels that limit the impact of parasitoids on herbivores (Rees
and Onsager 1982, 1985, Vblkl and Kraus 1996, Heimpel et al. 1997, Rosenheim 1998).
Usually, predators eat immature parasitoids inside their hosts without discerning
their presence (Brodeur and Rosenheim 2000, Colfer and Rosenheim 2001), termed
coincidental IGP by Polis et al. (1989). However, at least some predators have the ability
to discriminate parasitized hosts, and this ability usually increases as the immature
parasitoid develops inside its host, exhibiting less preference for parasitized than
unparasitized hosts (Rosenheim et al. 1995, Rosenheim 1998). The ability to avoid
parasitized hosts may maximize predator ﬁtness, since there is evidence that parasitized
hosts and mummies are of nutritional less quality than unparasitized hosts (Takizawa et
al. 2000). Conversely, Meyhbfer and Klug (2002) showed that the coccinellid Coccinella
septempunctata consume proportionally more parasitized aphids and mummies than
unparasitized aphids, including consumption of empty mummies, indicating inability of
this species to evaluate the suitability of food items. Predation on mummies is a case of
direct predation and can also reach high levels in agroecosystems (Brodeur and
Rosenheim 2000, Meyhbfer and Hindayana 2000, Colfer and Rosenheim 2001, Meyhbfer

2001, Meyh'dfer and Klug 2002).

13

Types of predator-parasitoid IGP

If there is no discrimination between parasitized and unparasitized hosts the
relative impact of a 1G predator would be a ﬁmction of encounter rates and attack rates
with each type of host (Rosenheim 1998). IGP may occur simply due to a higher ratio
between IGP prey versus non-guild prey (passive IGP), or can be due to a behavioral shift
(active IGP) that results in the inclusion of more IG prey in the diet or an increase in the
rate of encounter with IG prey due to increased predator mobility (Polis et al. 1989).
Passive increase in IGP could additionally occur due to differential behaviours in
parasitized hosts that makes them more susceptible to IGP (Brodeur and Rosenheim
2000). Among them are immobility that resulted in easy prey capture (Brodeur and
Rosenheim 2000, Snyder and Ives 2001), increased dropping behavior that elevated the
risk of predation by other predator species (McAllister and Roitberg 1987, Losey and
Denno 1998a), and increase in honeydew production that could lead to aggregative
responses of predators in patches with high parasitism (Brodeur and Rosenheim 2000).
Conversely, certain immature parasitoids have the ability to manipulate host behavior to
reduce the potential impact of predators and hyperparasitoids (Brodeur and McNeil 1992,
Rosenheim 1998).

Forage behavior of natural enemies and IGP

The foraging patterns of different natural enemies could increase or decrease rates
of IGP over what it may be expected by their relative abundance alone (Rosenheim
1998). Several parasitoids of gall insects have differential preference for gall sizes that in
turn interacts with the gall size preference of predators to determine the intensity of IGP

(Rosenheim 1998). Functional responses of parasitoids and predators also interact to

14

determine different intensities of IGP, and most species of predators and parasitoids seem
to respond positively to aphid density, increasing the chance for IGP interactions to occur
(Rosenheim 1998, Brodeur and Rosenheim 2000). In a multi-year study of old ﬁeld
aphids in Great Britain, aphid parasitoids did not show a positive response to host density
and therefore their effect as IG predators was reduced in comparison with predators that
exhibited positive responses (Miiller and Brodeur 2002). There is also evidence that
adult parasitoids avoid patches where secondary parasitoids and predators are foraging
(Héller et al. 1993, Taylor et al. 1998, Miiller and Brodeur 2002). Other factors affecting
the intensity of IGP against immature parasitoids include differential location of
parasitized hosts in the plant canopy, immature parasitoid developmental time, parasitoid
pupal morphology, aphid density, predator species and relative abundance, aphid-ant
mutualisms and predator foraging behavior (Brodeur and McNeil 1992, Brodeur and
Rosenheim 2000, Snyder and Ives 2001). IGP has been found to have non-additive
effects on predator-parasitoid systems in both laboratory and ﬁeld experiments, due to the
direct effect of predation on immature parasitoids (Press et al. 1974, Ferguson and Stiling
1996). Field studies with the braconid Lysiphlebus testaceipes in combination with the
coccinellid Hipodamia convergens showed that despite high predation on parasitoid
mummies, suppression of Aphis gossypii populations was increased when both enemies
were present (Colfer and Rosenheim 2001). Aphid-ant mutualism could also interfere
with the intensity of IGP since ants can differentially interfere with the enemies in the
guild (V 611d 1992, Rosenheim et al. 1995). Moreover, Meyhbfer and Klug (Meyhéifer

and Klug 2002) conclude that Lysiphlebusfabarum (Marshall) (Hymenoptera:

15

Braconidae) lacks strategies to avoid IGP, probably due to its preference to utilize the

“enemy free space” provided by ant-tended aphid colonies.

The Aphis glycines-natural enemy system

Aphis glycines life history

Aphis glycines is a heteroecious holocyclic aphid species that uses Rhamnus spp.
as primary host and Glycine spp. as a secondary host (Wang et al. 1962, Van den Berg et
al. 1997, Ragsdale et al. 2004, Wu et al. 2004). In China, A. glycines uses R. davurica
Pall as it primary host. Glycine max and wild species of Glycine are used as secondary
hosts (Wu et al. 2004). In North America, the most important primary host is the exotic
invasive species R. cathartica L., however A. glycines can also successfully overwinter
on native R. alniﬂora L’ Héritier (Voegtlin et al. 2004, Yoo et al. 2005). Tests on another
eight Rhamnaceae present in North America failed to produce succesfull overwintering
(V oegtlin et al. 2004, Yoo et al. 2005). The A. glycines life cycle begins with egg hatch in
spring on Rhamnus spp., where up to three generations develop in China (Wang et al.
1962, Wu et al. 2004), and an undetermined number in North America (Ragsdale et al.
2004) . Alate viviparous females from Rhamnus spp. migrate to soybeans on summer,
where several (10 to 22 in Asia) parthenogenetic generations of apterous females are
produced (Wu et al. 2004), On soybean, alates are also produced if densities are crowded
or host quality declines (Ragsdale et al. 2004, Wu et al. 2004). In the fall, a generation of
alate females called gynoparae is produced and migrates back to Rhamnus spp. They

produce both apterous females called oviparae and alate males (Ragsdale et al. 2004, Wu

l6

et al. 2004). These last two morphs will mate in Rhamnus spp. and oviparae will lay the

overwintering eggs on this host (Ragsdale et al. 2004, Wu et al. 2004).

Aphis glycines damage to soybeans

Aphis glycines cause direct damage to soybeans by sap removal as a consequence
of very high densities of aphids feeding per plant. Yield reductions of between 50-70%
have been reported in Asia (Wu et al. 2004) and up to 40% in USA (DiFonzo and Hines
2002). Plant symptoms from heavy infestation include “wrinkled and distorted foliage,
early defoliation, underdeveloped roots, shortened stems and leaves, stunting, reduced
branch number, lower pod and seed counts, reduced seed weight, and under severe
infestations, plant death” (Wu et al. 2004, Mensah et al. 2005). However, a recent study
has also shown that even relatively low aphid densities, feeding can signiﬁcantly affect
photosynthesis and potentially reduce plant yield (Macedo et al. 2003). High aphid
densities also result in increased amounts of honeydew deposited on leaves which allows
development of sooty molds which themselves reduce yield via reduced photosynthesis
(Wu et a1. 2004). Finally, A. glycines is vector of several important plant viruses that not
only affect soybeans, but also other crops such as snap beans and seed potato (V enette
and Ragsdale 2004). Aphis glycines has been shown to vector Soybean mosaic virus
(SMV).in China and North America (Hill et al. 2001, Wang and Ghabrial 2002, Wu et al.
2004, Burrows et al. 2005), as well as Alfalfa mosaic virus (AMV), and Tobacco ringspot
virus (TRSV) (Hill et al. 2001, Clark and Perry 2002). Other viruses that are known as
transmissible by A. glycines in other parts of world and thus represent an important

potential threat in North America, include Bean pod mottle virus (BPMV), Bean yellow

17

mosaic virus (BYMV), Peanut mottle virus, Soybean stunt virus, Abaca mosaic, Beet
mosaic, Tobacco vein-banding mosaic virus, Mungean mosaic virus, Peanut stripe poty
virus, and Peanut mosaic virus (Clark and Perry 2002, Wang and Ghabrial 2002, Wu et
al. 2004). Recently it was demonstrated that Soybean dwarf virus is not transmitted by A.

glycines in Illinois, USA (Harrison et al. 2005).

Factors aﬂecting A. glycines abundance

Environmental factors have been correlated with A. glycines outbreaks in Asia
(Wu et al. 2004). Hirano et al.(1996) showed in laboratory conditions that temperatures
from 22 to 27 °C are optimal for immature development, with survivorship declining
sharply at 32 °C. The threshold temperature for development was estimated as 9.5 °C.
Adults reproduce faster at 27 than at 22 °C, but their longevity is lower. However, the
intrinsic rate of increase (rm) is higher at 27 than at 22 °C. Other studies found the range
of optimal temperatures for A. glycines between 22 and 25 °C (Wang et al. 1962,
Ragsdale et al. 2004). Similar results were reported for A. glycines in North America,
with a basal developmental threshold of 8.6 °C, an upper threshold of 34.9 °C, and an
optimum of 27.8 °C (McComack et al. 2004). The supercooling point for eggs was
estimated as -34 °C, restricting the overwintering likelihood for A. glycines in northen
United States and Canada (McComack et a1. 2005). Models utilizing average
temperatures and precipitation produced signiﬁcant correlations of these factors with
aphid densities in some regions of China, but the conditions varied from region to region

(Wu et al. 2004). The climate in the area of distribution of A. glycines in Asia and

18

Australia allowed Venette and Ragsdale (2004) to conclude that this pest has the potential

to be distributed through most of the soybean production area in North America.

Bottom-up eﬂects

Several sources of bottom-up effects have been reported that can potentially affect
A. glycines. First, Van den Berg et al. (1997) showed in Indonesia that plant phenology
signiﬁcantly affected population growth of soybean aphid, diminishing its population
growth 50% latter in the season. While, the results of Van den Berg et al. (1997) were
not conﬁrmed in trials performed in Indiana (Rutledge et a1. 2004, Rutledge and O'Neil
2006), anecdotal evidence from Michigan suggests less A. glycines damage on early-
planted soybeans (C. DiFonzo, pers. com.) A second potential source of bottom-up
effects is host plant resistance, and resistant varieties have been found in both Asia and
North America, as well in lines of the ancestor of soybean G. soja Sieb. & Zucc. (Hill et
al. 2004a, Li et al. 2004, Wu et a1. 2004). Third, Myers et al. (2005b) showed that A.
glycines fed on potassium deﬁcient plants resulted in lower population growth in
laboratory assays, although they could not replicate this ﬁnding in ﬁeld studies. Finally,
other factors that may affect A. glycines population include cultural practices, such as

cropping, rotation (Wu et al. 2004).

T op-down eﬂects

Top-down effects of natural enemies on A. glycines have been demonstrated in its
native range and more recently in North America. In China, the most important natural
enemies are the coccinellids Propylaeajaponica (Thunberg) and Harmonia axyridis
Pallas, and several species of syrphids and lacewings (Wu et al. 2004). Parasitism by

Lysiphlebiajaponica (Ashmead) (Hymenoptera: Braconidae) reached levels between

19

10.3 — 52.6%, and maintained A. glycines populations below threshold level at rates of
parasitism of 34% (see Wu et. al., 2004, for a complete list of A. glycines natural enemies
reported in China and South Korea). In China, using a combination of exclusion cages
and sampling, Liu et al. (2004) showed a 12 fold reduction in A. glycines population
growth when aphids were not protected from natural enemies, and 4 fold reduction when
populations were exposed only to small size (< 2 mm) natural enemies. The most
abundant natural enemies were the parasitoid Lysiphlebus sp., the coccinellid predators P.
japonica, Scymnus (Neopullus) babai Sasajai, and the syrphid Paragus tibialis (Fallén).
Of them, only S. babai showed a trend towards a numerical response, with its populations
tracking aphid populations during the season. However, when they exposed (i.e.
previously caged) high A. glycines populations, they obtained signiﬁcant aggregation of
P. japonica. Van den Berg et al. (1997) showed that the coccinellid Harmonia arcuata
(Fabricius) was responsible for reducing A. glycines population latter in the season in
Indonesia. In Japan, the absence of A. glycines outbreaks is mainly attributed to the
action of natural enemies (Heimpel and Shelly 2004). Finally, the success in reducing A.
glycines density by intercropping soybean with corn has been attributed to the
concomitant increase in natural enemy abundance (Wu et al. 2004).

In North America, several studies have been conducted demonstrating that the
assemblage of generalist predators naturally occurring in the ﬁeld can provide effective
top-down control of A. glycines. Using a combination of clip—cages that exclude
predators but allowed A. glycines emigration, Fox et al. (2005) showed that generalist
predators could signiﬁcantly impact the initial number of adults that establish in the ﬁeld

at the beginning of the season in Michigan. Excluding predators during 5 weeks in early-

20

mid season resulted in a 20—fold increase in A. glycines populations (Fox et a1. 2004).
Furthermore, exposure of formerly protected populations or conversely, protecting
formerly exposed populations resulted in a reversal of population trends over 3 weeks
(Fox et al. 2004). During these trials in Michigan, the most abundant foliar predators
were the coccinellid H. axyridis, the anthocorid Orius insidiosus (Say), and the
chamaemyiid Leucopis spp. Parasitoids were not observed in these studies in 2001-02.
Similar patterns of species abundance were found in Indiana, although a much more
relevant role is attributed to 0. insidiosus (Rutledge et al. 2004, Rutledge and O'Neil
2005). Interestingly, predator populations that were effectively acting against A. glycines
never exceeded 5 predators / m2 (Fox et al. 2004), supporting the idea that predation early
in the season can successfully maintain low pest levels (Chang and Kareiva 1999,

Symondson et al. 2002).

The assemblage of natural enemies of A. glycines in Michigan

The most abundant and important natural enemy of A. glycines in Michigan is H.
axyridis (Fox 2002, Fox et al. 2004, Rutledge et a1. 2004, Fox et al. 2005). Harmonia
axyridis is a very effective aphid predator in Asia (Osawa 1993, 2000), and it was
introduced in the USA in 1988 (Chapin and Brou 1991). Since its introduction H.
axyridis has become the most abundant coccinellid in several regions, displacing the
previous dominant invasive species, C. septempunctata and other native coccinellid
species (Brown and Miller 1998, Colunga-Garcia and Gage 1998, Michaud 2002, Snyder
et al. 2004). The success of H. axyridis in invading new habitats is at least in part due to
its behavior as IGP of other coccinellid species (Y asuda et al. 2001, Sato et al. 2003,

Snyder et al. 2004). Orius insidiosus is an omnivorous predator commonly present in

21

many crops, including soybeans where it was found in association with A. glycines
(Rutledge et al. 2004, Rutledge and O'Neil 2005). As true omnivore, 0. insidiousus can
survive on plant resources in the absence of insect prey, allowing it to persist at very low
levels of pest density (C011 and Guershon 2002). Lysiphlebus testaceipes is a native
parasitoid widely distributed in North America in many different habitats (Pike et al.
2000). It has a wide host range and has been shown to effectively suppress aphid
populations in several habitats (Pike et al. 2000). In 2003, L. testaceipes was observed
attacking A. glycines in Michigan at very low levels (Brewer and Noma 2003). This
species has been shown susceptible to IGP in cotton, where 98-100% of the mummies of
this parasitoid can be attacked and killed by the coccinellid Hippodamia convergens
Guerin (Coleoptera: Coccinellidae) (Colfer and Rosenheim 2001). However, the same
study showed that suppression of the aphid A. gossyppi was enhanced in the presence of
both natural enemies together. The parasitoid Aphelinus albipodus (Hymenoptera:
Aphelinidae) was introduced in the USA two decades ago to control the Russian wheat
aphid Diuraphis noxia (Brewer and Noma 2003). In Asia it was found parasitizing A.
glycines at very low host densities and a ‘strain’ obtain from Wyoming populations has
been released in Minnesota and Wisconsin in 2002 (Heimpel and Shelly 2004).
Aphelinus albipodus has also been detected parasitizing A. glycines in Michigan in very
low levels (Brewer and Noma 2003). Other natural enemies present in Michigan have
been shown capable to attack A. glycines in pretri dish trials, including Nabis spp.
(Heteroptera: Nabidae), Chrysopa spp. (N europtera: Chrysopidae), Coccinella
septempunctata (L.), Coleomegilla maculata De Geer and H. convergens, F orﬁcula

auricolaria Linnaeus (Coleoptera: Forﬁculidae), Philonthus thoracicus (Grav.)

22

(Coleoptera: Staphylinidae), and eight species of Carabidae (Coleoptera) (Fox 2002).
Finally, syrphid larvae and the larvae of the midge Leucopis sp., have also been shown

attacking A. glycines in the ﬁeld in Michigan (Fox 2002, Fox et al. 2004).

23

CHAPTER 2

PREDATORS EXERT TOP-DOWN CONTROL OF SOYBEAN APHID ACROSS

A GRADIENT OF AGRICULTURAL MANAGEMENT SYSTEMS

Abstract.

The discovery of soybean aphid, Aphis glycines Matusumura, in North America in
2000 provides the opportunity to investigate the relative strength of top-down and
bottom-up forces in regulating pepulations of this new invasive herbivore. At the
Kellogg Biological Station Long Term Ecological Research site in agroecology, I
contrasted A. glycines establishment and population growth under three agricultural
production systems which differed markedly in disturbance and fertility regimes.
Agricultural treatments consisted of a conventional tillage high input system, a no tillage
high input system, and a zero chemical input system under conventional tillage. By
selectively restricting or allowing predator access I simultaneously determined aphid
response to top-down and bottom-up inﬂuences. Irrespective of predator exclusion,
agricultural manipulations did not result in bottom-up control of A. glycines intrinsic rate
of increase or realized population growth. In contrast, I observed strong evidence for top-
down control of A. glycines establishment and overall population growth in all production
systems. Abundant predators including Harmonia axyridis, Coccinella septempunctata,
Orius insidiosus and various predaceous ﬂy larvae signiﬁcantly reduced A. glycines
establishment and population increase in all trials. In contrast to other systems in which

bottom-up forces control herbivore populations, I conclude that A. glycines is primarily

24

controlled via top-down inﬂuences of generalist predators under a wide range of
agricultural management systems. Understanding the role of top-down and bottom-up
forces in this context allows agricultural managers to focus on effective strategies for

control of this invasive pest.

Introduction

Ecologists have long been interested in understanding the relative strength of top-
down and bottom-up forces in shaping natural communities (Hairston et al. 1960, Hunter
and Price 1992, Power 1992, Walker and Jones 2001). In terrestrial ecosystems both
forces are widespread (Gruner 2004), and several factors have been shown to
signiﬁcantly affect the relative strength of their effects, including plant morphology
(Kareiva and Sahakian 1990), productivity (Power 1992, Fraser and Grime 1998, 1999,
Dyer and Letoumeau 1999, Moran and Scheidler 2002), consumer efﬁciency (Power
1992), temporal versus spatial variation of herbivore populations (Hunter et al. 1997),
herbivore life stages (Walker and Jones 2001), seasonal and year-to-year changes in
abiotic and biotic conditions (Walker and Jones 2001, Boyer et al. 2003, Gratton and
Denno 2003b, b), differential predation risk among herbivore species (Denno et al. 2003),
and herbivore body size (Sinclair et al. 2003). Although not explicitly framed in a
bottom-up versus top-down context, several studies in agroecosystems have tested the
effects of fertility regimes and host plant resistance versus the effects of natural enemies
on pest control (Walker and Jones 2001). Agroecosystems often consist of highly
simpliﬁed food webs with strong reciprocal inﬂuences between adjacent trophic levels

(Polis et al. 2000), resulting in ideal systems to manipulate bottom-up versus top-down

25

forces. Despite this potential, no simultaneous manipulations of both forces have been
conducted in agroecosystems (but see Dyer and Stireman 2003). Furthermore,
understanding the relative strength of top-down versus bottom-up forces in agricultural
systems has the potential to provide insights on fundamental ecological questions, as well
as help guide agroecosystem management.

I tested the relative strength of top-down versus bottom-up inﬂuences on the
population growth of soybean aphid, Aphis glycines, Matsumura (Homoptera:
Aphididae). Aphis glycines is an important pest of soybean (Glycine max L.) in China
and Southeast Asia, causing yield losses of up to 70% (Wang et al. 1962, Hirano et al.
1996, Van den Berg et al. 1997, Wu et a1. 2004). In the year 2000, A. glycines was
discovered for the ﬁrst time in North America and as of 2004 was reported present in 21
states in the USA and three provinces of Canada (Hunt et al. 2003, Landis et al. 2003,
Venette and Ragsdale 2004). In 2003, over 7 million acres of soybean were sprayed with
insecticides against A. glycines, which infested over 50 million acres of soybean in the
USA (Landiset a1. 2003). Before the arrival of A. glycines, soybean in the North Central
USA had few serious insect pests and was not usually treated with insecticides. Thus,
soybean has played an important role as reservoir of natural enemies for other crops
(Heimpel and Shelly 2004). Previous studies have detailed signiﬁcant impacts of natural
enemies on A. glycines in Asia (Van den Berg et al. 1997, Liu et al. 2004, Wu et al.
2004), suggesting potential for top-down control. Generalist predators currently
dominate the assemblage of A. glycines natural enemies in the US, with parasitoids only
rarely observed from 2000-2002 (F ox 2002, Fox and Landis 2003, Fox et al. 2004,

Rutledge et al. 2004).

26

In the soybean aphid system, I manipulated different agricultural management
systems as a potential source of bottom-up forces under realistic agronomic practices. At
the same time, I manipulated the existing natural enemy assemblage as a source of top-
down control by using exclusion cages. Speciﬁcally, I tested the relative strength of
bottom-up versus top-down inﬂuences in determining A. glycines 1) early season
establishment and 2) population increase. In addition, I tested the effects of agricultural
management systems on 3) A. glycines survival and intrinsic rate of increase, and on 4)

un-manipulated populations of aphids and their natural enemies during the entire season.

Materials and methods

Study site

Experiments were conducted at the Kellogg Biological Station Long Term
Ecological Research in Row Crops Agriculture site (KBS-LTER) in 2003. The site
contains seven different cropping systems, representing levels of disturbance from
secondary succession to high input agriculture plots that have been managed under these
practices since 1988 (see http://lter.kbs.msu.edu/ for detailed information). These
differing production systems are known to signiﬁcantly affect weed (Menalled et al.
2001) and natural enemy populations, including ground beetles (Clark et al. 1997) and
ladybeetles (Maredia et al. 1992, Colunga-Garcia and Gage 1998), but their inﬂuence on
herbivore populations has not previously been studied.

Three of the annual crop treatments, which were planted to soybean and represent
a broad range of agricultural practices, were utilized. The high input treatment utilizes

conventional tillage and standard chemical inputs to simulate typical grower inputs in the

27

region (conventional). A second high input treatment is identical to the ﬁrst except for
the use of a no tillage regime (no—till). These were contrasted to a conventionally tilled
treatment that received no chemical input of any kind (zero-chemical input) but which
utilizes a winter leguminous cover crop to help maintain fertility. These treatments
effectively bracket the range of practices used to produce soybean in the region and are
likely to affect soybean plant quality to herbivores. Crop yields obtained in this site are
representative of yields reported by the U. S. Department of Agriculture for the North
Central region (Robertson et al. 2000). All treatments were planted between 26 and 29 of
May 2003, at a rate of 72,845 seeds / ha, using the cultivar NK S20-F8 (Syngenta Seeds).
Conventional and no-till treatments were planted on a 38 cm row spacing, and were
fertilized at a rate of N: 1.0 kg/ha, P: 4.8 kg/ha, K= 11.0 kg/ha, on 04/26/2003. No
insecticides were applied in any of the treatments during 2003. The zero-chemical input
treatment was planted on 76 cm rows, with no previous fertilization and weed control by
cultivation only. Treatments were replicated six times in 1 ha plots arranged in a
randomized complete block design (see http://lter.kbs.msu.edu/ for plot layout and details
of all agricultural operations). Using these plots I conducted ﬁve studies to assess the

impact of agricultural management system and predation on A. glycines.

Soybean aboveground net primary production and yield

The effects of bottom-up manipulations on the soybean plants were assessed by
comparing Aboveground Net Primary Production (ANPP) and yield among the three
agricultural management systems. This information is collected systematically as part of
the LTER-KBS sampling protocol and detailed procedures as well as raw data are

available at http://lter.kbs.msu.edu. Sampling of ANPP was performed on 9/02/2003,

28

during soybeans peak biomass, on ﬁve quadrates (1 m2) per plot. On each quadrate
soybeans were clipped at ground level and then dried at 60 °C for 48 h to obtain dry
weight. Yield was determined by harvesting with conventional techniques on 10/06/03

and standardized to 13% moisture.

Aphid establishment

I ﬁrst tested the effects of agricultural management system and predation on aphid
survival during the establishment phase in the ﬁeld. On June 26, A. glycines were placed
within clip-cages in the three agronomic treatments described previously, at a rate of ﬁve
apterous adults per clip-cage. Clip cages were cylinders of a 1.8 (diameter) x 1 (length)
cm Cresline PVC pipe, covered on the top by a ﬁne-mesh brass screen (33 threads/cm)
that allow air exchange but prevent aphid movement (Fox et a1. 2005). The clip-cage was
secured to the abaxial surface of a leaﬂet of the ﬁrst or second trifoliate by a metallic clip,
which allowed cage removal with minimal disturbance of the aphids. Naturally occurring
aphids are common on these plant parts at this crop stage (A. C. Costamagrra, personal
observation). After a 3 h acclimatization period, the groups of ﬁve aphids were
indiscriminately assigned to one of three predator treatments: 1) predator exclusion, in
which aphids were protected ﬁom predators by clip-cages, 2) emigration control, in
which aphids were enclosed by clip-cages with a 3 mm opening (covered by a cork
during the acclimatization period) that allowed aphid emigration but restricted predator
access, and 3) open, in which clip-cages were removed, allowing both predation and
aphid emigration. The 3 mm opening prevents entry by large predators (i.e.
Coccinellidae), but small predators such as Orius insidiosus (Heteroptera: Anthocoridae)

may theoretically enter cages, although this was never observed by us or in previous

29

studies (Fox et al. 2005). Therefore, predation measured as the difference between the
open and the emigration control treatments represents a conservative estimation (Fox et
al. 2005). After 24 h the numbers of live A. glycines adults and nymphs were quantiﬁed
separately. The experiment was conducted on the six replicates of agricultural
treatments, with ﬁve clip-cages per agricultural and predator treatment combination, for a

total of 270 clip-cages.

Aphid population increase

In a second experiment, I tested the effects of agricultural management system
and predation on the population growth of A. glycines enclosed in l m2 areas during the
p0pulation increase phase. In each agricultural treatment plot 1 m2 areas were established
with l) a ﬁeld cage that excluded predators, 2) a sham ﬁeld cage that allowed predator
access, and 3) a 1 m2 open area as a no cage control. The predator exclusion ﬁeld cage
consisted of a PVC ﬂame (1 x 1 x I m) covered by ﬁne-mesh white no-see-um netting
(Kaplan Simon Co., Braintree, MA) on the upper portion to exclude foliar predators, and
a basal plastic barrier (10 cm buried in the soil, 20 cm above soil surface), connected to
the netting by Velcro, to exclude ground-dwelling predators (after Fox et al. 2004). The
sham treatment consisted of a similar ﬁeld cage, but open from 0 to 15 cm above the soil,
and with a second 20 cm wide opening at the canopy level, allowing access to both
ground and foliar predators. The open treatment consisted of a 1 m2 area delimited by
wire ﬂags. In each plot of the three agricultural treatments, three 1 m2 areas were
arranged as the vertices of a triangle, separated by 3 m, and were randomly assigned
using a table of random numbers, to a predator manipulation treatment. In each 1 m2 area

resident predators and aphids were removed and then plots were re-infested at the

30

average aphid density in the ﬁeld at large (110 aphids/m2). Aphids were released at a rate
of 11 aphids per plant on 10 plants interspersed throughout the 1 m2 area. Five blocks
were infested using naturally occurring aphids from the same or nearby block, and one
block was infested using artiﬁcially reared aphids from the USDA Niles PPQ lab at
Niles, Michigan. No subsequent difference in the performance of natural vs. artiﬁcially
reared aphids was detected and therefore all blocks were considered for data analysis.
The experiment was initiated on July 14th with aphid and natural enemy populations
assessed in all cages 6 and 14 days after infestation. Natural enemy sampling consisted
of 3 min non-intrusive visual counting of larger predators (Fox et al. 2004), followed by a
more detailed search on 6 to 10 plants within the cage (reducing sampling effort with
aphid density increase over time) to detect small or cryptic natural enemies (i.e.
predaceous ﬂy larvae, syrphid larvae, lacewing larvae, 0. insidiosus, and ﬁrst and second
instar coccinellid larvae; see results). Sample plants were selected without bias as to size,
presence of natural enemies and aphids, or previous sampling history. The numbers of
small-size predators and parasitoid mummies per plant were adjusted for the total number
of plants per cage to compare with the number of large-size predators. Aphid sampling

consisted of total counts of aphids on the same 6 to 10 plants.

Aphid intrinsic rate of increase

A third experiment was conducted to assess the effect of differing agricultural
practices on survival, longevity, fecundity and intrinsic rate of increase (r) of A. glycines.
A. glycines cohorts were followed using clip-cages in the three agricultural treatments
utilized in the previous studies, during the same time period as the large cage study (July

16 to August 21). Clip-cages were the same as previously described but ventilated with

31

the addition of a 3 mm cylindrical opening covered with the same ﬁne mesh screen as
above to reduce potential heat stress. Three to ﬁve apterous adult A. glycines were
enclosed in a clip-cage to produce offspring. After 24 h all but three nymphs were
removed. These nymphs were followed until the ﬁrst adults were observed (indicated by
the presence of newborn nymphs). At this point all but one adult aphid were removed in
order to obtain estimates of per capita fecundity (van den Berg 1997). From the
survivorship, longevity, and fecundity data I estimated the intrinsic rate of increase (r) for
each block using the method developed for aphids by Wyatt and White (1977). The
experiment was replicated in the six blocks, with 10 clip-cages per agricultural treatment

plot.

Natural population sampling

I further assessed the effect of agronomic management system on A. glycines and
natural enemy populations by conducting weekly sampling of un-manipulated
populations in the six plots of each agricultural treatment from June 24 to August 18,
2003. There were other herbivores present, mostly thrips (Insecta: Thysanoptera) and
leafhoppers (Homoptera: Cicadellidae), but due to their low abundance no attempt was
made to quantify them. Sampling consisted of 3 samples of 1 linear m of soybean plants
within a row (14.6 i 2.7 plants / linear m, mean i standard deviation), in each plot. New
sampling points were interspersed in each ﬁeld plot, for each sampling date. In each
sample, foliar natural enemies were sampled by three-minute non-intrusive visual counts,
followed by more detailed inspection of plant foliage, as described in the aphid
population increase section. Whole plant counts of 4 to 10 plants were used to assess A.

glycines populations (reducing sampling effort as aphid density increased over time).

32

Statistical analysis

Soybean ANPP and yield were analyzed using ANCOVA on a completely
randomized block design with number of plants / m2 and number of plants / ha as
covariates, respectively. The number of surviving aphids in the aphid establishment
experiment was analyzed using a split-plot design with agricultural treatment as the
whole plot factor and predator manipulation as the sub-plot factor. Numbers of adults
and nymphs were analyzed independently to separate adult mortality/emigration from
reproduction. Number of aphids, predators and parasitoid mummies in the aphid
population increase experiment were analyzed using a split-split—plot design, with
agricultural treatment as whole-plot factor and predator manipulation and time as sub-
plot factors. A Poisson regression using the GLIMMIX Macro link of SAS in PROC
MD{ED (SAS Institute 2001) was used to analyze these data due to the absence of a
suitable transformation to normalize them, except in the aphid counts within large cages
which were transformed (In [x+1]) and analyzed by ANOVA. The effect of agricultural
treatments on A. glycines longevity, fecundity and r was assessed by one-way ANOVA.
The effect of agricultural treatments on un-manipulated populations was assessed by one-
way ANOVA (aphids) and MANOVA (natural enemies) utilizing sampling date as a
repeated measures factor. Due to the large number of samples with zeros, natural enemy
data were condensed in three ways: 1) subsamples within each sample were combined
(addition of all individuals per group), 2) the ﬁrst two sampling dates were excluded from
the analysis, and 3) groups representing less than 5% of the natural enemies collected
were not used in the analyses. Standardized canonical coefﬁcients were used to interpret

the discriminant function of the MANOVA (Scheiner 2001).

33

Signiﬁcant interactions in the ANOVA models were further explored via slicing
by main effects (Quinn and Keough 2002), and means were separated using Least
Squares Means Difference (LSMD , SAS Institute 2001). A signiﬁcant agricultural
treatment by date interaction obtained in the MANOVA was explored using pre-planned
contrasts among agricultural practice treatments within each sampling date. When
necessary to meet the assumptions of the ANOVA and MANOVA, variables were
transformed (1n [x+1]) before analysis. Blocks and the interaction terms involving them
were modeled as random effects in all models. ANOVAs were performed using PROC

MIXED and MANOVA was performed using PROC GLM (SAS Institute 2001).

Results

Soybean aboveground net primary production and yield

Agricultural management systems signiﬁcantly affected soybean ANPP and yield.
ANPP was signiﬁcantly lower in the zero chemical input treatment (322.2 i 22.3 g / m2)
than in the conventional treatment (515.9 d: 9.5 g / m2), with an intermediate response in
the no-till treatment (462.3 i 14.1 g / m2) (F 2,9 = 5.10, P = 0.0331; LSMD test: P <
0.02). The number of plants per m2 was not a signiﬁcant covariate (F 1,9 = 0.00, P =
0.9971), and blocking did not affect the results (F 5,9 = 1.07, P = 0.4374). Yield also
differed among treatments (F 2,9 = 5.54, P = 0.0271), with signiﬁcantly higher levels in
the no-till (1854. 7 :t 35.7 kg / ha) and in the conventional (1620.8 i 78.5 kg / ha), than in
the zero-chemical input treatment (1009.5 i 78.7 kg / ha; LSMD tests, P < 0.04). The
number of plants per hectare and block effects were not signiﬁcant (F 1,9 = 0.04, P =

0.8406; and F 5,9 = 1.39, P = 0.3139, respectively).

34

Table 2.1. Poisson regression results of the effect of agricultural management system
and predator manipulation treatments on the establishment of A. glycines at the KBS-
LTER site, Michigan, 2003.

 

 

 

 

 

 

 

a) Fixed eﬂects
Adults Nymphs
df F P F P
Agricultural treatment 2, 10 0.48 0.6314 0.35 0.7103
Predator treatment 2, 30 8.77 0.001 4.41 0.021
Agricultural treat. x pred. treat. 4, 30 0.39 0.8156 1.02 0.4125
b) Random eﬂects
Adults Nymphs
Estimate Z P Estimate Z P
Block 0.06 i 0.05 1.15 0.1241 0.06 :t 0.09 0.73 0.233

Block x Agricultural treat. 0.02 :t 0.03 0.45 0.3265 0.12 :t 0.1 l 1.13 0.1284

Residual 0.35 :t 0.10 1.75 :l: 0.49

 

Aphid establishment

There were signiﬁcant effects of predator manipulation on adult survivorship (F2,
30 = 8.77, P = 0.0010) and nymph production (F2, 30 = 4.41, P = 0.0210) after 24 h of
establishment (Figure 2.1, Table 2.1). Exposure to predation signiﬁcantly reduced adults
and nymphs (ca. 32%) in the open versus the emigration control treatment (Figure 2.1 ).
Adults and nymphs were 12 and 6% less abundant in the emigration control versus the
predator exclusion treatment; however, these differences were not signiﬁcant (Figure

2.1). There was no signiﬁcant effect of agricultural management system on either adults

35

(F110 = 0.48, P = 0.6314), or nymphs (F110 = 0.35, P = 0.7103), or the interaction
agricultural management system x predator manipulation (adults: F4, 30 = 0.39, P =
0.8156; nymphs: F4, 30 = 1.02, P = 0.4125), indicating an absence of bottom-up effects in

this test.

5 _ I Conventional
E3 No—tiil

A El Zero chemical input

 

 

_ i---—i
H-l

 

 

   

 

 

 

 

 

 

 

 

 

 

b) Nymphs
18 7
15 -
12 ‘

Aphids / cage

A A

 

 

 

 

 

 

 

 

 

 

Exclusion Emigration
control

Predator manipulation

Figure 2.1. Effect of natural enemies and agricultural management system on the
number of A. glycines (mean i: SE) alive after 24 h at the KBS-LTER site, Michigan,
2003. Different letters above bars indicate signiﬁcant differences among predator
manipulation treatments (P < 0.05, LSMD , SAS Institute 2001).

36

Aphid population increase

Topdown control signiﬁcantly reduced aphid numbers in the large ﬁeld cages
during the two weeks of the experiment (predator manipulation effect: F2, 27 = 94.63, P <
0.0001, Figure 2.2, Table 2.2). Aphids increased signiﬁcantly from the ﬁrst to the second
week (F1, 42 = 644.46, P < 0.0001), but differences among predator treatments ranging
from 4 to 7 fold more aphids in the exclusion cages were persistent between weeks
(Figure 2.2). The signiﬁcant impact of predation on aphid population growth was
consistent despite a marginally signiﬁcant interaction between agricultural management
system and predator manipulation (F4, 27 = 2.50, P = 0.0661), with higher number of
aphids in predator exclusion cages under all agricultural management systems (all slicing
tests with P < 0.001, Table 2.2). Agricultural treatments did not affect signiﬁcantly the
number of aphids in exclusion (F 2, 27 = 1.43, P = 0.2572), and open (F2, 27 = 2.03, P =
0.1508; Figure 2.2) treatments, but signiﬁcantly higher aphid suppression in sham cages
in the no-till treatment in comparison with conventional and zero chemical input (F2, 27 =
8.28, P = 0.0016) was observed. Sham and open treatments differed in the no-till
treatment (LSMD, P = 0.0144), suggesting a cage effect rather than an effect of the
different agricultural management systems. No cage effects were observed in
conventional or zero-chemical input treatments. Therefore, I found strong top-down
control on A. glycines population growth, but no evidence of bottom-up effects due to
differing crop management systems.

The natural enemy assemblage of A. glycines was dominated by generalist
predators and a generalist parasitoid species. Large-size predators sampled within the

ﬁeld cages consisted of the lady beetles (Coleoptera: Coccinellidae) Coccinella

37

Table 2.2. ANOVA results for a) ﬁxed, and b) random effects, and c) slicing tests results
of the effect of agricultural management system and predator manipulation treatments on
A. glycines within large ﬁeld cages (ln [number of aphids / plant + 1]) at the KBS-LTER
site, Michigan, 2003.

a) Full model with fixed eﬂects

 

 

 

 

 

df F P
Agricultural treatment 2, 10 5.56 0.0237
Predator treatment 2, 27 94.63 < 0.0001
Agricultural treat. x predator treat. 4, 27 2.50 0.0661
Date 1, 42 644.46 < 0.0001
Agricultural treat x date 2, 42 0.55 0.5827
Predator treat x date 2, 42 1.47 0.2424
Agricultural treat. x predator treat. x date 4, 42 0.65 0.6278
b) Random eﬂects
Estimate Z P

Block 0.090 d: 0.071 1.25 0.1053
Agricultural treat x block 0.006 i 0.033 0.19 0.4248
Agricultural treat x predator treat x block 0.094 i 0.048 1.97 0.0427*
Residual 0.143 t 0.031

 

* Graphical exploration of the signiﬁcant 3-way interaction revealed consistently higher
number of aphids on the exclusion cages in comparison with open and sham cages within
each agronomic treatment and block. Therefore, predator manipulation effects were not
affected by the 3-way interaction. However, the number of aphids obtained in the
exclusion cages for each agronomic treatment varied in different blocks, showing an
inconsistent effect of agronomic treatment in this experiment.

38

Table 2.2 (cont’d)

c) Slicing tests for significant interaction terms of fixed effects.

 

 

Agricultural treat. x predator treat. df F P
Conventional 2, 27 49.07 < 0.0001
No-till 2, 27 36.20 < 0.0001
Zero-chemical input 2, 27 17.03 < 0.0001

Predator exclusion 2, 27 1.43 0.2572
Sham 2, 27 8.28 0.0016
Open 2, 27 2.03 0.1508

 

septempunctata (9.4 :l: 4.4% of total predators sampled), and Harmonia axyridis (5.5 i

2.8%), whereas small-size predators consisted of 0. insidiosus (23.3 d: 6.3%), ﬁrst and

second instar coccinellid larvae not identiﬁed to species (18.4 :1: 4.9%), syrphid larvae

(Diptera: Syrphidae) (20.5 :t 6.1%), and predaceous ﬂy larvae (Diptera: Cecidomyiidae

and Chamaemyiidae) (7.9 i 3.7%) (Table 2.3). Mummies of the native parasitoid

Lysiphlebus testaceipes (Hymenoptera: Braconidae) were also detected at low levels

(Table 2.3). Exclusion cages signiﬁcantly reduced the abundance of large-size predators

(F 2 27 = 4.81 , P = 0.0163), but they had no signiﬁcant effect on small-size predators (F227

= 1.39, P = 0.2662; Table 2.3). By contrast, parasitoid mummies showed the opposite

trend, with signiﬁcantly more mummies found in exclusion than in open cages (F227 =

3.29, P = 0.0525; Table 2.3), suggesting that exclusion cages may have protected

parasitoids from intraguild predation. Although parasitoids were not able to enter the

mesh of the exclusion cages, some parasitism may have occurred within exclusion cages

if ﬁeld collected aphids used to initially infest the cages had already been parasitized.

39

a) 6 days after release

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

350 “j
300 l I Conventional
250 .1 A UNo-till
200 l El Zero-chemical input
150 E , ’l’
100 l W” 4 f B B
50 - I
.. o _ ﬁ-zlj]. c,“
5
'5'.
3 b) 14 days after release
:3 2500 7
9..
<1!
2000 ‘
1500 ~
1000 l
. ._ B B
500 -
If; I .L
Exclusion Sham Open
Predator unnipulation

Figure 2.2. Effect of natural enemies and agricultural management systems on the
number of A. glycines (mean :l: SE) after a) 6 days, and b) 14 days of manipulation using
large ﬁeld cages at the KBS-LTER site, Michigan, 2003. Different letters above bars
indicate signiﬁcant differences among predator manipulation treatments (P < 0.0001,
LSMD , SAS Institute 2001).

Agricultural treatments signiﬁcantly affected the abundance of small-size predators (F2, 10
= 6.21, P = 0.0177), with higher number of predators in the conventional treatment
(Table 2.3), but did not affect the abundance of large-size predators (F2, 10 = 0.62, P =

0.5569), or parasitoids (F2, 70 = 0.61 , P = 0.5649; Table 2.3). These results suggest that

40

the signiﬁcant increase in aphid density obtained in predator exclusion cages was
associated with reduction in the action of large-size predators (mainly Coccinellidae)

rather than small-size predators and parasitoids.

Aphid intrinsic rate of increase

Rearing aphid cohorts with clip-cages conﬁrmed the absence of effects of
agricultural management system on A. glycines life history parameters. Agricultural
management systems did not affect signiﬁcantly A. glycines longevity (18.4 d: 0.6 d, n =
179; F2, 10: 1.93, P = 0.1954), fecundity (28.9 :t 1.6 females x female", n = 174; F2, .0 =
0.68, P = 0.5287), or intrinsic rate of increase (0.32 i 0.01 females x female" x day", n =
18; F 2, 10 = 2.38, P = 0.1423; Table 2.4). My estimates (r= 0.30 — 0.33) are consistent

with other ﬁeld estimates of r= 0.310 reported by van den Berg (1997).

Natural population sampling

Un-manipulated ﬁeld populations of A. glycines differed among agricultural
management systems on three of the seven dates of sampling (agricultural management
system by date interaction: F7 2, 33: 2.21, P = 0.0175), but there was no main effect of
agricultural treatment (F2, 10: 2.36, P = 0.1446). Aphids increased exponentially ﬁom the
end of July to mid August (sampling date: F6, 83: 638.95, P < 0.0001), and slicing by the
interaction between agricultural management system and date revealed signiﬁcantly
fewer aphids in the no-till treatment on July 16 (F2, 33: 4.29, P = 0.0167). This trend later
reversed, with signiﬁcantly less aphids in the conventional treatment on August 5 (F2, 38:

6.85, P = 0.0017) and 13 (F2, 88: 3.87, P = 0.0246) (Figure 2.3a).

41

Table 2.3. Mean (:1: SE) of large- and small-size predatorsl (individuals / m2) and
parasitoids (L. testaceipes mummies / m2) for different combinations of agricultural
management system and predator manipulation treatments, within the large ﬁeld cages in
the aphid population increase experiment the KBS-LTER site during 2003.

 

 

 

 

 

 

 

Date Predator treat. Agricultural treatment Bigpredators Small predators Parasitoids
7/21/03 Exclusion Conventional 0.00 :t 0.00 1.34 :t 0.95 2.12 :l: 1.62
No—till 0.00 i 0.00 0.00 i 0.00 0.00 d: 0.00
Zero-chemical input 0.00 :t: 0.00 1.32 i 1.32 1.95 i 0.72
Sham Conventional 0.67 i 0.49 0.71 :i: 0.71 0.42 :i: 0.42
No—till 0.17 :1: 0.17 0.00 d: 0.00 0.43 d: 0.43
Zero-chemical input 0.33 i 0.33 1.07 i 0.50 0.91 :t 0.41
Open Conventional 0.33 :l: 0.21 2.75 i 1.23 0.00 :i: 0.00
No—till 0.17 a: 0.17 0.46 i 0.46 0.46 i 0.46
Zero-chemical input 0.50 :t 0.34 0.28 i 0.28 0.37 3: 0.37
7/28/03 Exclusion Conventional 0.00 :i: 0.00 0.79 i 0.79 13.22 :h 8.66
No-till 0.50 d: 0.50 2.02 at: 1.17 7.66 :l: 7.66
Zero-chemical input 0.00 i 0.00 0.00 i 0.00 21.73 d: 20.27
Sham Conventional 1.17 i 0.48 14.28 :t 5.20 1.29 :t 1.29
No—till 0.50 :t 0.34 0.00 i 0.00 0.00 i 0.00
Zero-chemical input 0.50 :i: 0.34 1.85 i 0.85 0.00 d: 0.00
Open Conventional 0.17 :l: 0.17 4.64 i 2.45 3.07 i 1.57
No-till 0.00 :t 0.00 4.48 i: 4.48 3.21 i 1.56
Zero-chemical input 0.17 i 0.17 1.68 d: 0.79 2.47 :l: 1.35

 

I Large-size predators were counted through 3-minute non-intrusive visual sampling
within the whole cage and included H. axyiridis, C. septempunctata, nabids (Heteroptera:
Nabidae), and spiders (Aracnidae). Small-size predators and mummies were counted by
a detailed, intrusive search on 10 (7-21) and 6 (7-28) plants per cage and were adjusted to
m2 by multiplying the average per plant by the total number of plants within the cage.
Small-size predators included: 0. insidiosus, ﬁrst and second instar coccinellid larvae,
syrphid larvae, predaceous ﬂy larvae (Diptera: Cecidomyiidae and Chamaemyiidae), and
lacewing larvae (N europtera: Chrysopidae).

42

Table 2.4. Mean longevity (days, :t SE), fecundity (total females / female) and intrinsic
rate of incresase (r, females/female/day*) of A. glycines reared on soybean produced
under three different agricultural management systems in the KBS-LTER during 2003.

 

 

Agronomic practice Longevity Fecundity r
Conventional 16.8 :1: 1.0 26.0 :1: 2.6 0.30 :l: 0.01
No-till 18.7 d: 1.0 29.8 :l: 2.7 0.33 :t 0.01
Zero chemical input 19.6 i 1.0 30.8 i 2.8 0.33 :l: 0.01

 

*estimated per block using the method developed for aphids by Wyatt and White (1977).

Five groups of natural enemies were included in the analysis: C. septempunctata
(29.4%), H. axyridis (23.1%), coccinellid larvae (16.0%), 0. insidiosus (10.1%) and
syrphid larvae (Diptera: Syrphidae) (5.5%), accounting for 84.1% of the total foliar
natural enemies observed during the ﬁve sampling dates included in the analysis (Figure
2.3b-f). The rest of the natural enemies included L. testaceipes (3.2%), predaceous ﬂy
larvae (3.2%), nabids (Heteroptera: Nabidae) (2.8%), other coccinellid species (2.6%),
spiders (Arachnida: Araneae) (1.7%), lacewing larvae (N europtera: Chrysopidae) (1.1%),
opilionids (Arachnida: Opiliones) (0.9%), and carabids (Coleoptera: Carabidae) (0.3%).
The results for the MANOVA showed signiﬁcant agricultural management system (Flo,
143 = 2.25, P = 0.0176), date (F4, 75 = 20.73, P < 0.0001), and management system x date
interaction (F3, 152 = 2.84, P = 0.0057) effects on the assemblage of A. glycines natural
enemies (Figure 2.3b-f, Table 2.5). Pre-planned contrasts showed that natural enemies
assemblages in the no-till treatment differed signiﬁcantly from the zero chemical input
treatment (July 22 and July 30, P = 0.05) and from the conventional treatment (August

13, P = 0.01). The standardized canonical coefﬁcients (SCC) of the ﬁrst canonical

43

a) Aphis glycines b) Coccinella septempucntata

       

 

3500 a: '4 a
l 2
l —t—Conventlonal Q
E 2500 J, g
34 2000 l —a— No-rm ' 3
:3 1500 l - ~ -A- - - Zero-chemical input ' g
.c 2
g 1000 - z
500 l B
6/24 7/4 7/ 14 7/24 8/3 8/ 13 6/24 7/4 7/14 7/24 8/3 8/13
c) Harmonia axyridis d) Coccinellidae larvae
6/24 7/4 7/1 4 7/24 8/3 8/1 3 6/24 7/4 7/14 7/24 8/3 8” 3
i
g e) Orius insidiosus 1) Syrphid larvae
%
g 5 7 5
.2 . "
z 41

 

   

8/1 3 6/24 7/4 7/1 4 7/24 8/3 8/1 3

Sampling date

Figure 2.3. Un-manipulated A. glycines (a) and natural enemy (b-f) populations (mean :t
SE) in soybeans under conventional, no-till and zero-chemical input treatments at the
KBS-LTER site during 2003. Asterisks above lines indicate signiﬁcant differences
among agricultural management system treatments within sampling dates (P < 0.05)

(LSMD , SAS Institute 2001).
variate showed that C. septempunctata (SCC = 0.5110) and syrphid larvae (SCC =

0.4004), which were more abundant in the no-till treatment, explained most of the

44

Table 2.5. MANOVA table for the effect of agricultural practices on the ﬁve more
abundant A. glycines foliar predators at the KBS-LTER site during 2003.

 

 

Effect Pillai’s Trace df F P
Agricultural practice 0.2613 10, 148 2.25 0.0176
Block 0.4157 25, 390 1.41 0.0912
Date 0.5251 4, 75 20.73 <0.0001
Agricultural practice x Date 0.2603 8, 152 2.84 0.0057
Block x Date 0.2733 20, 312 1.14 0.3034

 

variability and were negatively correlated with H. axyridis (SCC = -0.4876), which was
less abundant in the no-till plots. There was an increase of all the abundant foliar
predators late in the season, following the increase in aphid populations, with the

exception of 0. insidiosus, which declined in the last two samples (Figure 2.3 e).

Discussion

These results provide strong evidence of top-down regulation of A. glycines due
to predator assemblages, but only weak evidence of bottom-up effects under the range of
conditions tested. Generalist predators dominated the natural enemy assemblages, with
parasitoids only accounting for less than 4% of the natural enemies sampled in the un-
manipulated study. However, parasitoids showed increased numbers within exclusion
cages (46% of total natural enemies in the big cages study), suggesting that cages may
have reduced intraguild predation (Colfer and Rosenheim 2001). Ground predators (i.e.
spiders and carabids) were probably underestimated by mysampling methods, however

the absence of dropping behavior in A. glycines (A. C. Costamagna, and D. A. Landis,

45

personal observations), suggest a prevalent role of foliar predators in this system. The
assemblage of generalist predators was dominated by the coccinellids C. septempunctata
and H. axyridis, which were the dominant species previously reported attacking A.
glycines in North America (Fox et al. 2004, Rutledge et al. 2004). Both species
responded numerically to the increase in A. glycines natural populations later in the
season; however, C. septempunctata was more abundant in the no-till system, whereas H.
axyridis predominated in the conventional and zero-chemical input systems (Figure 2.3b
and c). H. axyridis has been shown to be a strong intraguild predator of other
coccinellids, and in particular of C. septempunctata larvae (Yasuda et al. 2001). Thus,
the different agricultural system preferences found for these coccinellids may result in an
attenuation of negative interactions between them.

Despite the signiﬁcant differences in the assemblages of natural enemies in
different agricultural production systems, overall suppression of A. glycines did not
differ, suggesting a complementary role of the different predator species. Suppression of
A. glycines occurred both during aphid establishment early in the season and during aphid
population increase at mid season. In the aphid establishment experiment, only a third of
the aphids survived when there was no protection against predation. The large impact on
aphid survivorship obtained in the relatively short time frame of this experiment (24 h)
suggests that this is a conservative measure of predation on aphid establishment. My
results are in agreement with other studies that showed signiﬁcant effects of predation on
establishing pest populations. Using a simple theoretical model, Chang and Kareiva
(1999) demonstrated that early predation by generalist predators has a similar effect as

latter immigration of more specialized natural enemies. Landis and Van der Werf (l 997)

46

showed that the assemblage of generalist predators present in sugar beet ﬁelds early in
the season signiﬁcantly diminished the population of aphids and the impact of viruses
transmitted by the aphids. In A. glycines, Fox et a1. (2005) also found a signiﬁcant
reduction on aphid establishment due to generalist predators, independently of the
presence of predator refuge strips within the ﬁeld. Ostman et al. (2001) found signiﬁcant
impacts of farming practices on bird cherry oat aphid (Rhophalosiphum padi [L.])
(Homoptera: Aphididae) establishment on barley, but there was no subsequent difference
in aphid population growth between farming practices mediated by natural enemies.

Top-down control by generalist predators signiﬁcantly reduced A. glycines
population increase (4 to 7 fold) across all agricultural production systems. Top-down
control of herbivores has been shown important in several terrestrial systems, leading to
trophic cascades that release plants from herbivory pressure (Spiller and Schoener 1990,
Carter and Rypstra 1995, Moran and Hurd 1998, Schmitz et al. 2000, Halaj and Wise
2001, Snyder and Wise 2001). In agroecosystems, both parasitoids and predators have
been shown to be effective in suppressing several aphid species (Hopper et al. 1995,
Obrycki and Kring 1998, Colfer and Rosenheim 2001, Symondson et al. 2002, Cardinale
et al. 2003, Lang 2003, Schmidt et a1. 2003, 2004). Natural enemies suppress A. glycines
in its original area of distribution (Van den Berg et al. 1997, Liu et al. 2004, Wu et al.
2004) and in its invaded range in North America (Fox et al. 2004). My results expand
these ﬁndings by showing that generalist predators suppressed A. glycines establishment
and grth across a wide gradient of agricultural production systems.

Several factors have been shown to independently lead to bottom-up effects on

aphid population growth in agroecosystems. Among them are signiﬁcant impacts of

47

fertilization (Cisneros and Godfrey 2001, Morales et al. 2001, Nevo and C011 2001),
tillage (Andersen 2003, Hesler and Berg 2003, Gencsoylu and Yalcin 2004), and cover
crops (Tillman et al. 2004). On A. glycines, Van der Berg et al. (1997) observed a
negative impact of plant age on population increase, whereas Myers et al. (2005b) found
positive effects of soil potassium deﬁciency. In mystudy, differing input regimes
(treatments) resulted in signiﬁcant differences in ANPP and soybean yield, conﬁrming
successful manipulation of bottom-up resources; however, aphid population increase was
not signiﬁcantly affected. This suggests that manipulation of chemical inputs and tillage
under realistic agricultural production systems does not change the quality of the soybean
plant sufﬁciently to cause signiﬁcant bottom-up effects on A. glycines establishment or
population growth, even in the absence of top-down controls. Although other studies
have shown signiﬁcant bottom-up effects on aphid population growth within shorter time
ﬁames than myexperiments (Van den Berg et a1. 1997, Nevo and C011 2001, Myers et al.
2005b), I can not rule out that bottom-up effects may become more evident over longer
periods of time.

In my study, ﬁeld populations of A. glycines eventually exceeded the economic
threshold (250 aphids per plant, University of Wisconsin 2003). However, my
experiment demonstrates that if predators were not continually suppressing aphids,
damaging aphid populations would have occurred much earlier in the season (Figure 2.2).
Other studies have documented that season-long predator suppression of A. glycines can
occur as well (Fox et al. 2004). Understanding the mechanism of aphid escape in some

but not all years is the focus of ongoing studies.

48

Top-down controls dominated the dynamics of the invasive A. glycines
populations during the period of this study. These ﬁndings differ from results obtained in
natural systems in which bottom-up forces has been shown to exert the primary control
on herbivore populations (Stiling and Rossi 1997, Denno et al. 2002), but supports the
prevalence of top-down control on determining herbivore populations found in other
studies (Dyer and Letoumeau 1999, Walker and Jones 2001). The two high input
production systems utilized in my study represent the most common strategies to grow
soybean in North America while the zero chemical input system represents an extreme of
low inputs. Given this wide range, these results suggest that there is little scope to
manage A. glycines impacts via such bottom-up inﬂuences. However, other sources of
bottom-up control on A. glycines such as manipulation of plant phenology (Van den Berg
et al. 1997), and plant resistance (Hill et al. 2004a, b, Li et a1. 2004) should be

investigated.

49

CHAPTER 3

SUPPRESSION OF SOYBEAN APHID BY GENERALIST PREDATORS

RESULTS IN A TROPHIC CASCADE IN SOYBEANS

Abstract
Top-down regulation of herbivores in terrestrial ecosystems is pervasive and can lead to
trophic cascades that release plants from herbivory. Due to their relatively simpliﬁed
food webs, agroecosystems may be particularly prone to trophic cascades, a rationale that
underlies biological control. However, theoretical and empirical studies show that within
multiple enemy assemblages, intraguild predation (IGP) may lead to a disruption of top-
down control by predators. A factorial ﬁeld study was conducted to test the separate and
combined effects of predators and parasitoids in a system with asymmetric IGP.
Speciﬁcally I combined ambient levels of generalist predators (mainly Coccinellidae) of
the soybean aphid, Aphis glycines Matsumura, with controlled releases of the native
parasitoid Lysiphlebus testaceipes (Cresson) and measured their impact on aphid
population grth and soybean biomass and yield. Generalist predators provided strong,
season-long aphid suppression, which resulted in a trophic cascade that doubled soybean
biomass and yield. However, contrary to expectations, L. testaceipes provided minor
aphid suppression and only when predators were excluded, which resulted in non-
additive effects when both groups were combined. I found direct and indirect evidence
of IGP, but since percentage parasitism did not differ between predator exclusion and

ambient predator treatments, I concluded that IGP did not disrupt parasitism during this

50

study. These results support theoretical predictions that intraguild predators which also

provide strong herbivore suppression do not disrupt top-down control of herbivores.

Introduction

Top-down regulation of herbivores has been strongly debated since it was
postulated as a major force organizing terrestrial ecosystems by Hairston et al. (1960).
This view of three deﬁned trophic levels that interact directly and indirectly with lower
trophic levels has lead to the concept of trophic cascades, in which organisms in higher-
order trophic levels affect the abundance, biomass or productivity of other species across
more than one link in a food web (Pace et al. 1999). Empirical evidence shows that
trophic cascades frequently occur in simple systems (i.e. less reticulated food webs) with
many cases having been documented in aquatic environments (Strong 1992, Pace et a1.
1999). Other studies have demonstrated trophic cascades in terrestrial ecosystems
(Schmitz et al. 2000, Halaj and Wise 2001), and particularly in agroecosystems (Carter
and Rypstra 1995, Snyder and Wise 1999, Colfer and Rosenheim 2001, Snyder and Wise
2001, Cardinale et al. 2003, Lang 2003, Matsumoto et al. 2003, Rypstra and Marshall
2005). Several characteristics of agroecosystems contribute to the development of
trophic cascades, including homogeneous habitat, faster prey than predator turnover,
uniformly edible prey, and simple trophic stratiﬁcation of the system (Polis et al. 2000,
Miiller and Brodeur 2002). Moreover, trophic cascades in agroecosystems form the
theoretical foundation of biological control, i.e. a reduction of plant damage (with
expected increased in crop yield) through deleterious effects of natural enemies (native or

imported) on herbivore pests (Polis et al. 2000, Rosenheim et al. 2004a).

51

However, despite the relative trophic simplicity suggested for agroecosystems, the
consumer assemblages present in them typically have multiple natural enemies, and thus
interactions such as predator interference, cannibalism, predator avoidance behavior, and
intraguild predation may modify the strength of top-down control of herbivores (Sih et al.
1998, Snyder and Wise 1999, Halaj and Wise 2001, Symondson et a1. 2002, Rosenheim
and Corbett 2003, Rosenheim et al. 2004a). Multiple consumers within an assemblage
may interact with synergistic, additive, or non—additive effects on herbivore populations
(Polis et al. 1989, Sih et al. 1998, Dixon 2000). Intraguild predation (IGP hereafter), i.e.
trophic interactions among members of the same guild within multiple natural enemy
assemblages, may be particularly important in determining the overall impact of
predators on herbivores (Polis et a1. 1989). Within agroecosystems, empirical studies
show that the presence of multiple natural enemies that engage in [GP can enhance
(Cardinale et al. 2003), have no effect on (Colfer and Rosenheim 2001, Lang 2003,
Snyder and Ives 2003, Rosenheim et al. 2004b), or disrupt (Rosenheim et al. 1993,
Snyder and Ives 2001, Snyder and Wise 2001, Yasuda and Kimura 2001, Persad and Hoy
2004, Philpott et al. 2004, Rosenheim et al. 2004a) biological control. Disruptions are
caused by negative non-additive interactions that cascade down, producing an increase in
the population of the herbivore (Rosenheim et al. 1995). In this case, the IG predator
exerts stronger pressure on an IG prey that in turn is effectively suppressing an herbivore,
releasing the herbivore ﬁom control, thus acting functionally as a fourth trophic level
(Polis and Holt 1992, Rosenheim and Corbett 2003).

In natural ecosystems top-down control is commonly exerted by predators,

whereas parasitoids are postulated as more important control agents in agroecosystems

52

(Hawkins et al. 1999). Interactions between parasitoids and predators often result in
asymmetric IGP, with all the stages of the parasitoid vulnerable to the predator, but with
no attack of the predator by the parasitoid (Brodeur and Rosenheim 2000). As a result,
IGP has been suggested as a possible mechanism to explain the failure of aphid
parasitoids to regulate their hosts (Ferguson and Stiling 1996, Rosenheim 1998, Brodeur
and Rosenheim 2000, Colfer and Rosenheim 2001). Although IGP is likely to occur
within most of the natural enemy communities present in agroecosystems, relatively few
studies have been conducted to quantif3I IGP’s impact on herbivore regulation
(Rosenheim et al. 1995). Colfer and Rosenheim (2001) and Snyder and Ives (2003)
found no impact of IGP on herbivore suppression, while Snyder and Ives (2001)
documented a disruptive effect on biological control.

Even in relatively simpliﬁed agroecosystems many factors may inﬂuence the
extent of herbivore suppression by multiple natural enemies. However, there is a lack of
studies manipulating natural enemies in experiments that allow distinction among their
separate and combined direct effect on herbivores and indirect effect on plants (Lang
2003). A notable exception is Cardinale et al. (2003), in which synergistic interactions
between predators and parasitoids resulted in a reduction of herbivore abundance and an
increase of plant biomass, although they did not directly test for IGP interactions.

The recent invasion in North America by the soybean aphid, Aphis glycines
Matsumura (Hemiptera: Aphididae) provides an ideal system to test the impact of
assemblages of multiple natural enemies. Several studies demonstrate that assemblages
of naturally occurring generalist predators can provide effective top-down control of A.

glycines in North America (Fox et al. 2004, Fox et al. 2005, Costarnagna and Landis

53

2006). Parasitoids were observed at very low levels in these studies and the most
common species collected was Lysiphlebus testaceipes Cresson (Hymenoptera:
Braconidae) (Brewer and Noma 2003, Costamagna and Landis 2006). L. testaceipes is a
generalist parasitoid shown to be the target of intense IGP by coccinellids in cotton
agroecosystems, although remains as an important biocontrol agent of the cotton aphid,
Aphis gossypii Glover (Hemiptera: Aphididae) (Colfer and Rosenheim 2001). Moreover,
in a previous study I found signiﬁcantly greater numbers of L. testaceipes mummies in
cages that excluded A. glycines predators than in sham cages and open treatments. I
postulated that exclusion cages provided refuge from IGP (Costamagna and Landis
2006). These results suggest that in the A. glycines system, IGP may be limiting the
impact of parasitoids, and may impair future attempts to establish additional parasitoid
species by importation biological control.

I conducted a ﬁeld study to test the impact of ambient levels of generalist
predators and controlled releases of the native parasitoid L. testaceipes on A. glycines
suppression and soybean biomass and yield. Using exclusion cages allow separation of
the effects of large predators versus L. testaceipes, and testing whether their combined
effect resulted in additive control. In addition, intraguild predation on parasitoid
mummies was assessed to determine to what extent IGP disrupts biological control by L.
testaceipes. Speciﬁcally, I asked 1) can ambient levels of generalist predators provide
season-long suppression of A. glycines, 2) can parasitoids alone suppress A. glycines, 3)
does the combined impact of predators and parasitoids result in synergistic, additive or

non-additive control of A. glycines, 4) does IGP disrupt biological control by parasitoids,

54

and ﬁnally, 5) do predators, parasitoids or their combined impacts on A. glycines

populations indirectly increase soybean biomass and yield?

Materials and methods

The experiment was conducted in a 99 x 94 m soybean ﬁeld at the Michigan State
University Entomological Research Farm, Ingham County, Michigan, USA. Soybean
(Pioneer 92813) was planted on May 7, 2004, using a Great Plains Seed Drill planter on
38 cm wide rows, at a population of 494,209 seeds / ha, and weeds were controlled using
Roundup WeatherMax (Monsanto Co.) tank-mixed with ammonium sulfate at standard
rates. The experiment was designed as a 2 x 2 factorial in 6 completely randomized
blocks, with predator and parasitoid manipulations as main factors. The four treatments
were: 1) aphid control, with added aphids only; 2) parasitoid, in which parasitoids were
released and large predators were excluded; 3) predator, in which ambient levels of
predators were allowed access to the aphids; and 4) predator + parasitoid, similar to the
predator, but with parasitoids released as in the parasitoid treatment. The rationale for the
approach of augment ambient parasitoid populations was that parasitoid immigration into
the aphid and predator treatments would be minimal due to the very low background
levels of ﬁeld parasitism found in the study area (Costarnagna et al. unpublished results).
To further ensure low levels of parasitism in the aphid and predator treatments, mummies
observed during the weekly samplings were recorded and then removed. Thus,
treatments contrasted ambient levels of generalist predators versus increased levels of

parasitoids.

55

The predator exclusion cage consisted of a PVC frame (1 x 1 x 1 m) covered by
ﬁne-mesh white no-see-um netting (Kaplan Simon Co., Braintree, MA) on the upper
portion to exclude foliar predators, and a basal plastic barrier (10 cm buried in the soil, 20
cm above soil surface), connected to the netting by Velcro, to exclude ground-dwelling
predators (Figure 3.1a) (after Fox et a1. 2004, Costamagna and Landis 2006). To allow
alate aphid and parasitoid emigration from cages, the upper 10 cm of the lateral walls
consisted of a band of mesh with coarse openings (2 mm). A second band of the same
coarse mesh (30 cm height) was placed at the canopy level after the 3rd sampling to
mimic the openings on the sham cages (see below). This design prevented large
predators from entering cages and was used for the aphid and parasitoid treatments.
Sham cages were utilized to allow access of predators in the predator and predator +
parasitoid treatments. The sham cage had the same design as the exclusion cage, but with
the upper and canopy level windows completely uncovered (Figure 3.1b). To ensure
parasitoid establishment, all cage windows were covered during the ﬁve initial days of
the experiment by an external band of the ﬁne mesh secured by Velcro, and this was
performed to all treatments to avoid bias.

To this design I added three control treatments. First, one open reference area of
l m2 (open treatment) was ﬂagged in each block as a control for potential cage effects.
Second, to estimate effects on plant biomass and yield I included an exclusion cage
treatment completely closed (i.e. without lateral bands of coarse mesh) that was
maintained free of aphids by weekly inspections and removal of any aphids present (no-
aphid treatment). The four treatments and the two controls were located in blocks

separated by at least 10 m. Within blocks, all treatments were separated by at least 3 m

56

a) Exclusion cage

10cm 2

20 cm

 

IIIIFIIIWIUUI

 

 

Topwindow
lll’Llll’llIll

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plastic barrier

 

10 cm

30 cm I
I
t

 

 

 

 

.3 1L2 1.

 

 

10 cm ti",

 

 

Plastic barrier

 

Soil line

 

I. .

 

Figure 3.1. Design of predator exclusion cages (A) used in aphid and parasitoid
treatments, and sham cages (B) used in predator and predator + parasitoid treatments.
The ﬁne mesh (< 1 mm) prevents aphid, parasitoid and predator movement, whereas the
coarse mesh (2 mm openings) prevents only large predator access (i. e. mainly
Coccinellidae). All cages were covered on top by ﬁne mesh. Ground-dwelling predator
movement was reduced by means of a transparent plastic barrier buried in the soil (see

I:

 

 

details in the methods).

57

 

 

Soil line

from each other. In all treatments I included Raid® Ant Baits II (S. C. Johnson & Son,
Inc., Racine, WI) to prevent ants from establishing and tending aphids. Finally, I
established a third control consisting of totally un-manipulated 1 m2 areas located 30 m
distant from the main experiment to avoid contamination by insect manipulations (ﬁeld
reference).

In all but the ﬁeld reference, all resident predators and aphids were initially
removed by hand and then plots were re-infested with aphids at similar densities to those
found at the beginning of the season during outbreak years (100 aphids / m2, Fox et al.
2004, Costamagna and Landis 2006). In the no-aphid treatment the same manipulation
was performed, but no aphids were added. Aphids reared on soybeans in the laboratory
were released on June 16m, 2004, at a rate of 10 aphids per plant on 10 random plants per
cage. The soybean plants were at the V3 stage and cages enclosed (mean 3: SD) 29.1 at
2.0 plants / m2. Parasitoids were established in the parasitoid and predator + parasitoid
treatments using mummies obtained from a colony of L. testaceipes maintained by the
USDA APHIS PPQ Laboratory at Niles, Michigan. Mummies were placed in 2 02 plastic
cups (Sweetheart Cup Company Inc., Owings Mills, MD) covered by a coarse mesh (2
mm openings) at the top that allowed parasitoid emigration from the cup but protected
them ﬁom large predators. Cups were held 20 cm above ground level by a wooden stake
and were protected ﬁom rainfall by an inverted white plastic cup (7.5 x 8 cm). Three
consecutive releases were performed on June 16, 23, and 25, at a rate of 1 cup per cage.
For each release, 20 mummies were placed in each cup, which generated a 1:5 (parasitoid
to aphid) ratio for the ﬁrst release. This ratio was successfully used in preliminary tests

in the same ﬁeld establishing L. testaceipes in small mesh cages. During each release, I

58

deployed control release cups completely closed by a ﬁne mesh (11 = 15 randomly
selected cages), that were checked daily for 4 days to assess parasitoid emergence.

Aphid and natural enemy populations were assessed weekly in all cages for six
weeks following aphid and parasitoid establishment. By week 3 it was evident that,
contrary to expectations, the sham cages were interfering with predator activity, resulting
in higher aphid densities than in the open control (see results). Thus, on July 9 windows
at the canopy level for both exclusion and sham cages (previously described) were
opened, which resulted in a signiﬁcant reduction in aphid abundance. Natural enemies
were sampled by conducting visual counts in cages for 3 minutes (Fox et al. 2004),
followed by aphid and parasitoid sampling which consisted of total counts of aphids and
mummies on 4 to 10 plants within the cage (reducing sampling effort with aphid density
increase, Costamagna and Landis 2006). Mummies were classiﬁed as intact (fully closed
mummy), emerged (with openings made by emerging adult parasitoids), or damaged by
predation (mummies opened in irregular patterns, consistent with damage by coccinellid
feeding observed in preliminary laboratory trials). P parasitism was assessed based on
the number of intact mummies, and therefore the measure of parasitism is conservative,
since some aphids could have been parasitized but not yet mummiﬁed. Soybean plants
were harvested for total above ground biomass in 3 randomly selected blocks on August
16‘“, and the remaining 3 blocks were harvested on October 13 to estimate yield. Plants
were dried at 60 °C until no change in weight was detected and then dry weights for

whole plants and seeds were recorded.

59

Statistical Analysis

The effects of predator and parasitoid manipulations on aphid, predator, and
parasitoid populations were analyzed as a 2 x 2 factorial ANOVA using a complete
randomized block design (CRBD) with time as a repeated factor (Proc Mixed, SAS
Institute 2001). Signiﬁcant interactions were further explored by slicing main effects
(Quinn and Keough 2002), and individual means were compared by the Least Square
Means Difference adjusted by the Tukey-Kramer method for multiple comparisons
(LSMD-TK, SAS Institute 2001). I also assessed 1) effects of caging by comparing the
predator treatment with the open reference plots, and 2) effects of predator and parasitoid
manipulations on soybean yield and biomass, using l-way ANOVAs on a CRBD with
time as a repeated factor. Predator, mummy, and aphid counts, and plant biomass and
yield data, were log-transformed, whereas percentage parasitism and proportion of
damaged mummies (mummies damaged/ [mummies damaged + new mummies]) were
arcsine-transformed, to meet assumptions of normality and homoscedasticity.
Additionally, using log-transformed data in the analysis allowed testing interactions using
a multiplicative risk model, which avoids bias when predation levels are high (Sih et al.
1998). I did not record predators and mummies during the ﬁrst two weeks of sampling in
most samples, and therefore these weeks were not included in statistical analyses for

those variables.

60

Results

Eﬂectiveness of predator manipulations

The assemblage of A . glycines predators was dominated by coccinellids
(Coleoptera: Coccinellidae; 53.7% of total number of predators sampled), with the most
abundant species being Harmonia axyridis (Pallas) (69.9%), followed by Coccinella
septempunctata L. (4.2%), Coleomegilla maculata (De Geer) (2.2%), and Hippodamia
variegata (Goeze) (2.2%). First and second instar coccinellid larvae (21.5% of total
coccinellid counts) were not identiﬁed to species, but were most likely also dominated by
H. axyridis. Predator and predator + parasitoid treatments had signiﬁcantly higher
abundance of coccinellids than did aphid and parasitoid treatments after the opening of
the canopy-level windows (signiﬁcant predator x date interaction in Table 3.1, Figure
3.2a), indicating successful manipulation of this group of predators. During the last week
of sampling, this trend reversed, with coccinellids occurring in higher abundances on the
aphid and parasitoid treatments (Figure 3.2a) where higher numbers of aphids occurred
(Figure 3.3). Small predators including midge larvae (Diptera: Cecidomyiidae and
Chamaemyiidae; 58.1%) and Orius insidiosus (Say) (41.9%) accounted for 34.3% of total
predators. In contrast to coccinellids, small predators were not affected by predator and
parasitoid manipulations (Table 3.1 , Figure 3.2b). Although the visual sampling method
utilized may have underestimated small predators hidden in the canopy of the plants, the
detection of small predators at low densities in all treatments indicates an absence of bias
among treatments.

Both coccinellids and small predators increased signiﬁcantly throughout the

season, following the increase in A. glycines densities (Table 3.1, Figure 3.2). Other

61

 

 

 

62

m L Ed AN; 2: .m ﬂood vow Sad 56 Shae nmd on .m 2868a; x .53on x 2mm
NNth mmd o3 .m memo EN 3.3.0 36 good Nod ow .m onmﬁem x 8mm
Soov EBB co“ .m 38V 36 would mm." Soov NOR on .m 5335 x 8mm
_ooo.v modem 2: .m $86 is Noood 5N. 58V v~.c_ ow .m 8mm
ﬂood 36 3 ._ Saved Nmé Emmd mo; Soﬁe mm.“ 2 ._ 2868mm x .8335
236 TN 2 ._ _ooo.v Qumm mood Ed mwomd mm; 2 ._ 3368mm
Seqv 34$ 3 ._ Sod 5&2 336 0: wood 3.: 2 J 8385
326 c: 3 .m moo—d mm; 336 morm v8.50 med 2 .m vLoo—m
& k .3 Q. E k K m k we
BERN .V “summoned“ d £832; =«Em men—oﬁoooo

 

.38 mass sane: a meme 22» man. maﬁa
EB mEozwﬁam .Eouawoa mo Sentence 2: no 2233958 Eozgm EB 8333 .«o 3:52 <>OZ< .3” 03.;

2.0 1 —-o—— Aphid control a) Coccinellids

"a —o— Parasitoid

5 15 J + Predator

4; l _._ Predator + Parasitoid a a
c .

5

0

g b
8 a
9.

g” b

 

2.0 ,
b) Small predators

1.5

Log (small predators + 1) / m2

 

Log (mummies + l) / plant

 

 

 

6/21 6/28 7/5 7/1 2 7/1 9 7/26

Figure 3.2 Mean (d: SE) abundance of a) coccinellids, b) small predators (midge larvae
and 0. insidiosus combined), and c) parasitoid mummies in ﬁeld cages with A. glycines
that combine predator and parasitoid manipulations. The arrow indicates the opening of a
canopy-level window (see methods). Different letters within each sampling date
represent signiﬁcant differences (p < 0.05) using slicing tests for the 2 and 3-way
signiﬁcant interactions (Table 3.1); ns = not signiﬁcant.

63

predator groups were also sampled, but low abundance and irregular distribution among
samples prevented statistical analysis. However, none of these groups showed trends of
association with any of the treatments, suggesting that they probably played only a minor
role in aphid suppression. Those predators included (in order of abundance): syrphid
larvae (Diptera: Syrphidae, 7.4%), lacewing larvae (N europtera: Chrysopidae, 2.6%), and
nabids (Hemiptera: Nabidae, 0.9%), followed by carabids (Coleoptera: Carabidae),
spiders (Arachnidae: Aranae), and staphylinids (Coleoptera: Staphylinidae) (1.1%
combined). Since samplings were conducted during the day, predators with nocturnal

habits such as carabids and spiders were undoubtedly underestimated.

Parasitoid eﬂectiveness and intraguild predation

Data ﬁ'om the control release cups indicated that 46.7 i 12.0% (mean i SD) of
the L. testaceipes mummies emerged successfully within 4 days. Releases resulted in
parasitoid establishment in the cages and a signiﬁcant increase of L. testaceipes
mummies (Table 3.1, Fig 3.2c). This effect was consistent through time, as indicated by
the non-signiﬁcant parasitoid x date interaction (Table 3.1). The presence of predators
signiﬁcantly reduced mummy abundance, but only during the last two sampling dates,
after opening the canopy-level window (Table 3.1, Fig 3.2c). There was a signiﬁcant
predator x parasitoid interaction, indicating non-additive effects of predator and
parasitoid manipulations. The signiﬁcant interaction term resulted ﬁ'om the signiﬁcant
reduction in the number of mummies when predators and parasitoids were combined
(Figure 3.2c). Parasitoid abundance increased signiﬁcantly over time in the parasitoid

treatment (Table 3.1, Fig 3.20).

64

To test whether predators disrupted aphid control by parasitoids, I compared the
percentage parasitism between the parasitoid and the predator + parasitoid treatments.
Despite the signiﬁcantly lower number of parasitoid mummies found in the treatments
with predators, there were no signiﬁcant differences in percent parasitism between
treatments with parasitoids exposed versus protected from large predators (F 1, 5 = 0.01, P
= 0.9455), indicating that predators did not reduce parasitism by L. testaceipes. Overall,
parasitism was very low (parasitoid: 0.46 i 0.16%; predator + parasitoid: 0.59 :l: 0.21%;
mean :l: SE averaged across weeks 3-6, n = 24) and did not vary among blocks (F5, 5 =
1.34, P = 0.3774). In addition, parasitism decreased signiﬁcantly over time (F 3, 30 = 4.67,
P = 0.0085), indicating a lack of response of L. testaceipes to aphid increase. Very low
levels of damaged mummies were observed throughout the season, and despite a trend for
higher total number of mummies damaged per plant in the predator + parasitoid treatment
(1.8 i 1.4, mean t SE) than in the parasitoid treatment (1.1 i 0.6), differences were not
signiﬁcant for total (F1, 5 = 0.01, P = 0.9104) or proportion of mummies damaged (F 1, 5 =
0.86, P = 0.3950). Thus, these results indicate that parasitism by L. testaceipes was not

disrupted by predators through IGP.

Aphis glycines control

Natural enemy manipulations resulted in large and signiﬁcant differences in A.
glycines populations after 6 weeks (Figure 3.3). Large predators (mainly Coccinellidae)
signiﬁcantly reduced aphid populations (Table 3.1), resulting in an average l76-fold
aphid reduction compared with the aphid control and parasitoid addition treatments.
Predator impacts on aphid abundance were underestimated during the initial part of the

experiment due to the restrictive size of the sham cage openings, which resulted in

65

—°— Aphid control

- D- Parasitoid

+ Predator at:
—o— Predator + Parasitoid

A. glycines / plant

 

6121 6128 7/5 7/1 2 7H 9 7R6

Figure 3.3 Mean (at SE) number of A. glycines in ﬁeld cages with different combinations
of predators and parasitoids. The arrow indicates the opening of a canopy-level window
(see methods). Asterisks indicate signiﬁcant differences between treatments with
ambient levels of predators versus predator exclusion treatments (predator x date slicing
tests, P < 0.0005).

signiﬁcant differences between predator and open treatments (F I, 5 = 52.18, P = 0.0008;
Figure 3.3). However, predator effectiveness increased in both treatments after opening
the canopy-level windows, which resulted in a signiﬁcant predator x date interaction
(Table 3.1, Figure 3.3). Ultimately, the predator treatment did not differ signiﬁcantly
ﬁom the open treatment during the last sampling date (slicing by date, F 7, 50 = 0.61, P =
0.4379; for the last sampling date; Figure 3.3). In contrast, parasitoid addition resulted in
a weak reduction of A. glycines populations and only when predators were excluded
(Figure 3.3). Aphid abundance was not affected by the opening of the canopy-level

windows in the parasitoid treatment (Figure 3.3). Natural A. glycines abundance was

very low in the ﬁeld references, reaching a maximtun of 2.3 i 0.6 aphids / plant at the

66

end of July, and was signiﬁcantly lower in the ﬁeld references than in the open treatment
(Figure 3.3, F], |4= 211.70, P < 0.0001).

There was a signiﬁcant predator x parasitoid interaction, indicating non-additive
effects, and those were consistent among sampling dates (non-signiﬁcant predator x
parasitoid x date interaction, Table 3.1). Slicing the interaction by main effects indicates
that predators reduced the number of aphids per plant, with or without added parasitoids
(F 1, 15 = 53.86, P < 0.0001, and F1, 15 = 137.07, P < 0.0001, respectively), whereas
parasitoids reduced aphids when predators were excluded (F 1, 15 = 10.76, P = 0.0051) but
not when predators were present (F 1, 15 = 1.18, P =0.2937). These results show that
generalist predators (mainly coccinellids) provided strong suppression of A. glycines and
that their presence eliminated the small impact of L. testaceipes, resulting in signiﬁcant

non-additive effects of the two groups of natural enemies.

Trophic cascade

Predator manipulations resulted in a trophic cascade that restored plant biomass
and yield (F4, 3 = 7.70, P = 0.0075; and F4, 3 = 12.08, P = 0.0018, respectively) to the same
levels as the no-aphid control (Figure 3.4). Exposure to ambient levels of predators
resulted in a 2-fold biomass and a 1.6-fold yield increase over the aphid control. Adding
parasitoids alone resulted in an intermediate increase in biomass, but did not improve
plant yield over the aphid control (Figure 3.4). Block was not signiﬁcant in either
analysis. These results indicate that the strong suppression of A. glycines by generalist

predators cascades down to result in levels of plant biomass and yield similar to plants

where aphids were excluded.

67

plant yield D plant biomas

A3 A]:

 

    

 

    

 

 

 

 

 

 

 

 

 

 

'o
I .7

 

No-aphid Predator+ Predator Paradtoid Aphid
control Paradtoid control

Figure 3.4. Mean (3: SE) soybean yield and above ground dry biomass in ﬁeld cages

without A. glycines (no-aphid control), and with A. glycines and different combinations of
predators and parasitoids. Different letters among columns represent signiﬁcant

differences (P < 0.05, LSMD-TK).
Discussion

The established assemblage of generalist predators signiﬁcantly impacted A.
glycines populations throughout the entire season. By contrast, controlled releases of the
native parasitoid L. testaceipes had only minor effects in A. glycines suppression when
large predators were excluded and did not add control when combined with them.
Signiﬁcant impacts of generalist predators on A. glycines was previously shown in Asia
(Van den Berg et a1. 1997, Liu et al. 2004), and North America (Fox et al. 2004, Fox et
al. 2005, Costamagna and Landis 2006), but my results extend these ﬁndings by spanning

an entire ﬁeld season, and by demonstrating for the ﬁrst time that this control cascades

down to increase plant biomass and yield.

68

The use of exclusion cages to test for the impact on natural enemies has been
extensively used for aphids in general (e. g. Chambers et a1. 1983, Hopper et al. 1995,
Ferguson and Stiling 1996, Jervis and Kidd 1996, Brown 2004) and also recently for A.
glycines (Fox et al. 2004, Costamagna and Landis 2006, Costamagna et al., unpublished
results). However, others have criticized cages as potentially preventing aphids ﬁom
emigrating and thus artiﬁcially inﬂating apparent predation impacts (e. g. Kindlmann et
al. 2005). Several lines of evidence support the role of predator exclusion rather than
aphid enclosure in explaining higher aphid abundances in predator exclusion cages.
Brown (2004) demonstrated similar levels of increased spirea aphid populations by
excluding predators with cages or with insecticide treatments in open plots. Chambers et
al. (1983) showed that removal of alates from exclusion cages did not result in lower
aphid populations in comparison with cages in which alates were conﬁned. In my study I
did not observe signiﬁcant accumulation of alates in exclusion cages even at high aphid
populations, suggesting that alates were able to successfully emigrate from the cages.
Finally, a signiﬁcant reduction in aphid abundance has been shown for treatments with
both aphids and predators enclosed together in comparison to controls in which only
aphids were enclosed, demonstrating the ability of predators to control aphid populations
(e. g. Cardinale et al. 2003, Snyder and Ives 2003). Thus, I feel conﬁdent that the
reduction in A. glycines in my study was due to predation rather than to artiﬁcial
conﬁnement of aphids.

The different mesh sizes used in the cages allowed us to separate the effects of
large versus small predators on A. glycines. Small predators, such as midge larvae and 0.

insidiosus, did not differ in abundance between exclusion and sham cages. Since aphids

69

increased exponentially in the exclusion cages in the presence of small predators, these
results suggest a minor role for small predators in A. glycines control at this location.
This result contrasts with other studies that suggest that 0. insidiosus is an important
predator of A. glycines (Rutledge et al. 2004, Rutledge and O'Neil 2005).

Although extensive data supports the importance of coccinellids as aphid
predators (Frazer 1988, Hodek and Honek 1996, Obrycki and Kring 1998), there is still
debate concerning their ability to suppress aphid populations (Dixon 2000). My results
demonstrate strong aphid suppression by un-manipulated ambient levels of coccinellids.
Both predator treatments resulted in signiﬁcant A. glycines control during the entire 2004
ﬁeld season, which mimicked un-manipulated ﬁeld populations. By contrast, in

treatments that excluded predation by large predators, A. glycines increased

exponentially, indicating that abiotic factors were suitable and did not limit aphid growth.
The similarity in abiotic conditions in exclusion versus sham cages and open controls was
previously demonstrated (e. g. Hopper et al. 1995, Schmidt et a1. 2003, Fox et al. 2004).
Thus, the absence of outbreak aphid populations in commercial ﬁelds during 2004 was
not directly attributable to unfavorable abiotic conditions. As in previous reports (Fox et
al. 2004, Fox et a1. 2005, Costamagna and Landis 2006), these results suggest that the
coccinellids H. axyridis and C. septempunctata, are the main predators responsible for A.
glycines suppression in the study area.

Controlled releases of the native parasitoid L. testaceipes failed to suppress A.
glycines population growth, even when large predators were reduced. There was only a
small impact of L. testaceipes on A. glycines, and this effect disappeared when ambient

levels of large predators were combined with parasitoids, resulting in non-additive

70

interactions. Although I obtained evidence of predator interference on parasitism, the
very low impact of L. testaceipes on A. glycines in predator exclusion cages indicates that
this disruption was trivial in comparison with the strong impact of large predators in
suppressing A. glycines. However, one important caveat of my experiment is that the low
efficiency of L. testaceipes prevents us from making a generalized conclusion regarding
the disruptive effect of IGP on more effective parasitoids. Nevertheless, the strong top-
down control exerted by coccinellids suggests that even more efﬁcient parasitoids will
add little control and most likely suffer high IGP (Colfer and Rosenheim 2001) under
similar conditions as in my experiment. Therefore, my results conﬁrm the theoretical
prediction that IG predators that have strong impacts on the primary prey (i.e. the
herbivore) will not disrupt top-down control, regardless of how much suppression they
exert upon the IG prey (Polis et al. 1989, Holt and Polis 1997).

The failure of L. testaceipes to suppress A. glycines was unexpected. Other
studies report show that L. testaceipes signiﬁcantly suppressed aphid populations in
cotton (Colfer and Rosenheim 2001), sorghum (Femandes et al. 1998), and cereals in
general (Brewer and Elliott 2004), and is the most abundant parasitoid attacking A.
glycines at the study area (Kaiser, Noma and Brewer, personal communication;
Costamagna and Landis 2006). Moreover, laboratory colonies of L. testaceipes
developed successﬁrlly on A. glycines for several generations (ACC personal
observation) indicating that A. glycines is a suitable host for this parasitoid. Field
observations suggest that the lack of control of A. glycines by L. testaceipes cannot be
explained by unsuitable abiotic factors. Visual inspections of releases performed in

completely closed sleeve cages in the same ﬁeld two weeks prior to the main experiment

71

revealed continued high levels of parasitism throughout the season, with a cumulative
number of 512.2 i 152.1 mummies per plant (mean :t SE, n = 4). In contrast, the
cumulative number of mummies observed in the parasitoid treatment was signiﬁcantly
lower (73.4 i 32.5 mummies per plant, t = -3.61, df = 8, P = 0.0068). Thus, it is likely
that the lower parasitism levels obtained in ﬁeld cages were due to parasitoid emigration
rather than from potential unfavorable weather conditions. Why L. testaceipes would
leave a patch with abundant suitable hosts is unclear, but may include a preference to
search in other habitats where it is adapted to ﬁnd hosts (Vinson 1976), imperfect host
discrimination (Rosenheim and Mange] 1994), variability in host suitability across scales
(Ives et al. 1999) and different responses to cues from different host species (Grasswitz
and Paine 1992).

There were a higher number of mummies in the parasitoid treatment, but since
percent parasitism was not affected by predation, mummy increase may be a response to
increased host density rather than the effect of a refuge from predation. I did not estimate
predation on parasitized aphids before mummy formation or on adult parasitoids, both of
which has been shown as potential sources of IGP (Heimpel et al. 1997, Meyhiifer and
Klug 2002). In addition, I can not rule out a more general, indirect effect of intraguild
interference (sensu Lang 2003) as another potential factor reducing mummy numbers in
the treatment combining parasitoids and predators. Recent studies showed that female
parasitoids spend less time, attack fewer aphids, and avoid leaves on which coccinellids
or their semiochemicals are present (Taylor et al. 1998, Nakashima et al. 2004).

This study adds to the growing body of evidence that generalist predators have

strong effects on prey, and is novel in demonstrating that these effects cascade down to

72

increase plant biomass and yield. Trophic cascades are pervasive in terrestrial
ecosystems (Schmitz et al. 2000, Halaj and Wise 2001, but see Polis et al. 2000), and
have been demonstrated in agricultural systems, including vegetable (Snyder and Wise
1999, 2000, 2001), perennial (Cardinale et a1. 2003), and annual crops (Carter and
Rypstra 1995, Colfer and Rosenheim 2001, Rypstra and Marshall 2005), although some
studies fail to show signiﬁcant impacts (e. g. Vichitbandha and Wise 2002). My ﬁndings
showed that strong top-down control by ladybeetles can completely release soybean
plants from aphid pressure and restore plant biomass and yield. This study differs from
previous experiments investigating top-down control and trophic cascades in
agroecosystems in two aspects. Previously it has been argued that strong trophic
cascades in agroecosystems may be at least partially an artifact of increasing the levels of
natural enemies above ambient levels, and by measuring plant damage, which may not be
an adequate surrogate of plant ﬁtness (Polis et al. 2000, Halaj and Wise 2001). These
concerns were addressed by conducting factorial natural enemy manipulations that
allowed ambient levels of predation to respond to aphid density and avoided forcing
interactions by allowing free emigration of aphids and parasitoids from all treatments.
This also makes me conﬁdent that aphid pressure on plants was not artiﬁcially inﬂated.
Finally, plant biomass and yield were measure directly, this last being a good surrogate
for plant ﬁtness.

In summary, I observed that strong suppression of soybean aphid by generalist
predators led to a trophic cascade that restored soybean biomass and yield. On the other
hand, the most abundant native parasitoid attacking soybean aphid, L. testaceipes, was

ineffective in reducing aphid populations. Although I obtained direct evidence of IGP in

73

ﬁeld conditions, the low suppression exerted by L. testaceipes is not currently limited by
IGP. However, there is still a very poor knowledge of factors leading to A. glycines
outbreaks, and thus it is still an open question whether despite potential IGP more

effective parasitoids may add control during seasons in which aphids escape predation.

74

CHAPTER 4

ARE EXISTING SOYBEAN APHID PARASITOIDS LIMITED BY INTRAGUILD

PREDATION ?

Abstract

Generalist predators are increasingly recognized as important sources of mortality
for agricultural pests. However, their role as intraguild predators has also been
demonstrated, and disruption by intraguild predation (IGP) has been proposed to explain
the failure of some parasitoids to suppress pest populations. The soybean aphid, Aphis
glycines (Hemiptera: Aphididae), present in North America since 2000, is widely
attacked by predators, but only sustains very low levels of parasitism by endemic species.
During 2004 and 2005, I conducted ﬁeld experiments in Michigan, USA, to test whether
parasitism is increased by providing parasitoids with refuges from IGP. I used ﬁeld
cages that differentially excluded large predators (primarily coccinellids) and parasitoids,
and tested their impact on controlled releases of A. glycines. Three species of resident
parasitoids were observed attacking A. glycines. However, even under levels of IGP as
high as 94%, I found no evidence for disruption in the level of parasitism, which was
uniformly low (< 1%). In contrast, coccinellids provided strong suppression of A.
glycines (86 -— 36 fold reduction), which resulted in a trophic cascade that restored
soybean biomass and yield to similar levels as controls without aphids. Parasitoids and
small predators (primarily Orius insidiosus) combined did not prevent exponential
growth of A. glycines, and soybean biomass and yield suffered similar reduction as in the

aphid control treatment. These results indicate that large predators exert strong

75

suppression of A. glycines, but do not disrupt the currently low levels of parasitism.

However, the potential for IGP in this system should be considered when introducing

exotic parasitoids as biological control agents for A. glycines.

Introduction

Exotic species that invade new habitats face an established assemblage of natural
enemies, including generalists which may attack them as a novel resource. The Enemy
Release Hypothesis proposes that invasive species become pests due to the inability of
the existing generalist natural enemies to control them (Shea and Chesson 2002). The
success of classical biological control programs involving importation of natural enemies
from the original area of distribution of a pest provides support to this theory (Shea and
Chesson 2002). Parasitoids are commonly selected for classical biological control of
arthropods, due to their host speciﬁcity and life histories which are frequently tightly
coupled with the pest (Hawkins et a1. 1999, van Lenteren et a1. 2003, Stiling and
Comelissen 2005). However, many attempts to introduce parasitoids as biological
control agents have failed to achieve adequate control and negative interactions with
existing natural enemies have been suggested as one possible reason for lack of success
(Rosenheim et al. 1995, Brodeur and Rosenheim 2000, Miiller and Brodeur 2002).

There is increasing recognition that established generalist predators can provide

pest suppression, due to their voracity and earlier presence in the ﬁeld (Chang and
Kareiva 1999, Symondson et al. 2002, Stiling and Comelissen 2005). However, trophic
interactions are likely within multiple natural enemy assemblages that involve

generalists, a phenomenon termed intraguild predation (IGP hereafter, Polis et al. 1989,

76

Polis and Holt 1992). In addition to direct nutritional gains, intraguild predators obtain
an indirect beneﬁt from eliminating potential competitors for the resource they share
(Polis et al. 1989). IGP has been shown to be widespread among food webs, where it is
postulated to confer stability in trophic relationships (Polis et a1. 1989, Rosenheim et al.
1995, Rosenheim 1998, Brodeur and Rosenheim 2000, Polis et al. 2000, Finke and
Demo 2002, Miiller and Brodeur 2002, Finke and Denno 2003).

Generalist predators usually engage in IGP and their body size is the main
determinant of whether they act as intraguild predators or prey (Polis and Holt 1992).
Thus, generalists have the potential to provide pest suppression, but also to disrupt
control by other natural enemies, resulting in what has been called the ‘the indeterminacy
of predator function’ (Rosenheim and Corbett 2003). By contrast, parasitoids that are
specialized on herbivores typically do not have the potential to attack other members of
the guild, with the exception of other parasitoids species sharing the same host (Brodeur
and Rosenheim 2000, Borer 2002, Borer et al. 2003). Instead, parasitoids are themselves
frequently susceptible to asymmetric IGP by predators in both immature and adult stages
(Rosenheim et al. 1995, Heimpel et al. 1997, Brodeur and Rosenheim 2000, Meyhbfer
and Klug 2002, Miiller and Brodeur 2002). Moreover, IGP has been suggested as a
potential explanation for the failure of parasitoids to suppress their hosts (Ferguson and
Stiling 1996, Rosenheim 1998, Brodeur and Rosenheim 2000, Colfer and Rosenheim
2001)

Empirical evidence shows that IGP can either disrupt (Snyder and Ives 2001,
F inke and Demo 2004) or have no effect (Colfer and Rosenheim 2001, Snyder and Ives

2003) on herbivore suppression by natural enemies. Moreover, additional factors

77

mediate the strength of the effect of IGP on herbivore control. Cardinale et a1. (2003)
showed that the presence of alternative prey for the predator resulted in enhancement of
aphid suppression by parasitoids in alfalfa. Finke and Demo (2002) showed that
increased structural complexity of vegetation resulted in lower levels of IGP and
increased herbivore suppression in a salt marsh community. Thus, the potential effect of
1GP on herbivore suppression is difﬁcult to predict (Rosenheim et al. 1995, Brodeur and
Rosenheim 2000).
The soybean aphid, Aphis glycines Matsumura (Homoptera: Aphididae) is a pest
of soybeans that is originally from Asia and has been found in North America since 2000
(Ragsdale et al. 2004, Venette and Ragsdale 2004). Several studies have shown
signiﬁcant impacts of natural enemies on A. glycines in Asia (Van den Berg et al. 1997,
Liu et al. 2004, Wu et al. 2004) and North America (Fox and Landis 2003, Fox et a1.
2004, Rutledge et al. 2004, Fox et al. 2005, Costamagna and Landis 2006). In Asia, the
complex of natural enemies attacking A. glycines includes the predators Propylaea
japonica (Thunberg), Harmonia axyridis (Pallas), and H. arcuata (Fabricius)
(Coleoptera: Coccinellidae), and several species of syrphids and lacewings (Van den
Berg et al. 1997, Wu et al. 2004). In addition, 10 — 53% parasitism by the parasitoid
Lysiphlebiajaponica (Ashmead) has been reported in China (Wu et al. 2004). In North
America, the assemblage of A. glycines natural enenries is dominated by generalist
predators, mainly the coccinellids H. axyridis and Coccinella septempunctata L. (Fox et
al. 2004, Fox et al. 2005, Costamagna and Landis 2006, Mignault et al. 2006). There are

only anecdotal references of ﬁeld parasitism (Landis et al. 2003, Costamagna and Landis

78

2006). Due to the lack of effective parasitoids in North America, efforts are currently
underway to research and possibly introduce parasitoids from Asia (Heimpel et al. 2004).
One potential explanation for the lack of parasitism of A. glycines in North

America is the occurrence of asymmetric IGP on parasitoids. In a recent survey, Kaiser
et a1. (personal communication) detected six parasitoids species from sentinel A. glycines
in Michigan, suggesting that North American parasitoid assemblages have the potential to
attack soybean aphid. In addition, results of a 2003 ﬁeld cage study revealed higher
number of the native parasitoid Lysiphlebus testaceipes Cresson (Hymenoptera:
Braconidae) in cages in which predators were excluded, supporting the hypothesis of IGP
limitation of parasitism (Costamagna and Landis 2006). Here I report the results of ﬁeld
experiments manipulating natural enemies to test the potential role of IGP in limiting
extant North American parasitoids of A. glycines. Speciﬁcally, I tested whether refuge
ﬁ'om IGP by large predators increased parasitoid attack of soybean aphid. The level of
IGP was determined directly by quantifying the number of mummies damaged by
predation. Finally, I also measured the impact of the different natural enemy groups on

aphid suppression, plant biomass and yield.

Material and methods

Field site

Field experiments were conducted in the biodiversity study of the Kellogg
Biological Station - Long Term Ecological Research site (Hickory Comers, Michigan),
during 2004 and 2005. The biodiversity study has a series of 21 different agronomic

treatments that vary in plant species diversity in time and space, ranging from 0 to >15

79

species in 3-year rotation cycles. Treatments are replicated in four randomized blocks in
9.1 x 27.4 m plots. Within this site, I utilized the three most similar treatments (all com-
soybean-wheat rotations) that varied only in the presence of cover crops in the non-
soybean portions of the rotations. These included system “b” with a legume and a small
grain cover crop in corn and wheat, system “c” with only a legume cover crop in corn,
and system “(1” with no cover crop. My rationale in selecting these systems was
primarily to increase replication, rather than explore the effects of cover crop systems per
se, since previous work had shown little impact of cover crops legacy in A. glycines
population growth (Costamagna and Landis 2006). A detailed description of the layout

of the treatments and protocols of agricultural management is available at

http://lter.kbs.msu.edu/.

Natural enemy manipulations

In each plot aphid and natural enemies were manipulated using three types of
cages. An exclusion cage consisted of a cylindrical framework of wire of 0.4 x 1.0 m,
covered by a ﬁne-mesh white no-see-um netting (Kaplan Simon Co., Braintree, MA),
buried 25 cm in the soil and tied to the top, enclosing three soybean plants. The ﬁne
mesh used prevented most insects (even small aphids) from moving in or out. The
parasitoid refuge cage was similar, but was covered with a coarse mesh that had 2-mm
openings. This cage allowed ﬁ'ee movement of parasitoids and other small insects
(primarily Orius insidiosus (Say) [Heteropteraz Anthocoridae]), but prevented large
predators (i.e. coccinellids) from entering cages. Finally, I used a sham cage identical to
the previous one, except the mesh was held 5 cm above the ground surface, attached with

metal clips to four wooden stakes, and with eight rectangular openings (2 x 20 cm)

80

arranged in 3 circular patterns alternating in a manner such that there were openings in all
directions. The rings of openings were located from 10 to 30 cm above the soil,
encompassing the height range of the plant canopy. Thus, the sham cage was designed to
allow movement of insects of all sizes, while minimizing the differences with the
parasitoid refuge cage.

Using these cages, I established ﬁve natural enemy treatments in each plot: 1) a
no-aphid treatment, in which soybean plants free of insects were enclosed within
exclusion cages, 2) an aphid treatment, in which a controlled number of aphids at the
beginning of the season were enclosed within exclusion cages, 3) a parasitoid refuge
treatment, in which aphids were enclosed in the same manner as in the previous
treatment, but within parasitoid refuge cages, and ﬁnally two treatments that allowed all
natural enemies: 4) a sham treatment, in which aphids were enclosed within sham cages,
and ﬁnally 5) an open treatment, in which soybean plants were infested with aphids but
without enclosing them in a cage. In all treatments groups of 3 soybean plants were
visually inspected and all resident arthropods were removed at the beginning of the
experiment (7/1/04 and 6/24/05). Then, I randomly assigned to each group of plants one
of the ﬁve natural enemy treatments and infested them (except in the no-aphid control) at
a rate of 15 aphids per plant on the central plant. This infestation rate was equivalent to
approximately 165 aphids / m2, which is slightly higher than the number used in previous
studies (i. e. 100 - 110 aphids / m2, Costamagna and Landis 2006, Costamagna et al.,
unpublished results), to ensure aphid establishment in all treatments. Aphids were from a
colony maintained on soybeans at Michigan State University, in growth chambers at 24

°C, 16:8 h photoperiod, and 60 -— 80% RH.

81

Each cage was sampled once a week after initial infestation until aphid
populations clearly started to decline in the aphid and the parasitoid refuge treatments
(mid August to early September). On each sampling date all the aphids and natural
enemies present in each cage were quantiﬁed by carefully inspecting each plant.
Parasitoid mummies were identiﬁed as Braconidae or Aphelinidae, and I also determined
whether they were intact, emerged or damaged by predation. On each sampling date few
individual mummies (1 — 5) were collected for rearing to determine parasitoid species.
After the third week of sampling, all but one plant in each cage were removed (typically
the central one unless damaged), to allow plants to grow uncrowded by the cage. Before
that date, the number of aphids was quantiﬁed on the three plants and calculated the
average per plant in each cage, in order to compare these data with subsequent dates.
During 2004 beginning two weeks after the start of the experiment, I detected ants
tending the aphids in the parasitoid refuge, sham and open treatments in seven of the ﬁve
experimental plots. To prevent this in other plots, I added plastic rings (60 cm diameter)
around the cages, buried 15 cm in the soil and rising 20 cm above the soil, coated with
Tanglefoot (The Tanglefoot Company, Grand Rapids, M1) on the outside. Ant-exclusion
rings were placed around the ant-free plots with no attempt to prevent ant tending in the
seven plots where ants were already present. This allowed us to compare the effects of
ant tending on aphid abundance. In 2005, I included ant-exclusion rings to all treatments
from the outset. I did not observe ants tending aphids in cages with ant-exclusion rings

during either year of the experiment.

82

Natural population sampling

In the surrounding ﬁeld at large, natural populations of aphids and natural
enemies were sampled at the same time cages were sampled. During 2004, sampling
consisted of total counts of aphids and natural enemies on one speciﬁc plant per plot
throughout the season. During 2005 a more comprehensive sampling program consisted
of direct counts of predators in 3 quadrats of 1 m2 in the same plots with natural enemy
treatments, changing their location on each sampling date. In each quadrat the number of
large predators was recorded during 3 minute (weeks 1 - 4) to 1 minute (weeks 5-8)
periods. In order to sample smaller predators (i.e. 0. insidiosus, midge larvae, and
syrphid larvae), a more detailed search was conducted in one quadrat per plot after the
initial non-intrusive observation. All values of natural enemies are expressed as number
of insects per m2. I also determined plant phenology and counted the number of A.

glycines in the plant located in the Northeast comer of each quadrat.

Statistical analysis

Aphid abundance was analyzed using ANOVA (PROC MIXED, SAS Institute
2001). Treatment structure was a 3 x 4 factorial in a split-plot design, with agronomic
treatment as a whole plot factor with 3 levels and natural enemy treatment as the subplot
factor with 4 levels. The experiment was conducted over two years. The statistical model
included year, agronomic treatment, natural enemy treatment and sampling dates, and all
the interactions between them as ﬁxed effects. Blocks and the interactions between them
and other factors were treated as random effects. The error term used for year was the
year x block interaction. Sampling dates were nested within year. I analyzed the ﬁrst 6

weeks after infestation, because after that aphid populations started to decline. In all

83

analyses, sampling dates within each year were treated as repeated measures obtained
from each subplot with autoregressive covariance structure selected as the most
appropriate covariance structure based on AIC (Akaike Information Criteria) values
(Littell et al. 1996). I performed pro-planned contrasts using the ANOVA model above
to assess whether the artiﬁcial aphid releases were representative of natural populations
(open treatment versus ﬁeld sampling), whether ﬁeld populations differed between years
(2004 versus 2005), and whether the no-aphid treatment resulted in signiﬁcantly reduced
aphids (no-aphid treatment versus ﬁeld sampling). To rule out potential effects of ants
tending the aphids during 2004, 1 compared treatments with and without ants using a
similar ANOVA model, but including ant control treatment at the whole-plot level. The
effect of agronomic treatments on the abundance of coccinellids and 0. insidiosus was
assessed with ANOVA for 2005 data, whereas the reduced sampling performed in 2004
prevented statistical comparisons. The cumulative number of mummies (i.e. added
across sampling dates) was analyzed using a Split-Plot ANOVA, with agronomic
treatment in whole-plot in a CRBD, and the natural enemy treatments as the subplot
factor. Individual means were compared using the Least Square Mean Difference,
adjusted for multiple comparisons by the sequential Bonferroni method (Rice 1989).
Data were log- or arcsine-transformed before analyses to meet the assumptions of
ANOVA, when necessary. Data that did not ﬁt the assumptions of ANOVA after
transformation were analyzed using non-parametric statistical tests. Natural enemy
abundance within cages was analyzed using the Kruskal-Wallis tests to compare the
effect of natural enemy manipulations independently within years and dates, and the

effect of year on the total number of natural enemies sampled in each date (SYSTAT

84

Software Inc. 2004). Similarly, the same test was used to assess the effect of agronomic
treatments on un-manipulated populations of large and small predators averaged across
all sampling dates for 2005. Finally, mummy abundance, percentage parasitism and
proportion of mummies damaged by IGP were compared between the parasitoid refuge
and sham treatments using separate Wilcoxon tests for each sampling date (SYSTAT

Software Inc. 2004).

Results

Aphis glycinesfield versus caged populations

Un-manipulated ﬁeld populations of A. glycines differed signiﬁcantly between
years (F = 262.44; df = l, 36; P < 0.001). Aphid populations were low throughout 2004,
but exceeded the action threshold of 250 aphids / plant during 2005 (Figure 4.1). Un-
manipulated ﬁeld populations were signiﬁcantly lower than those of the open treatment
during both years (Figure 4.1 and 2, F = 145.16; df = 1, 36; P < 0.001), suggesting that
the initial infestation levels resulted in higher aphid establishment than the natural aphid
immigration in the ﬁeld. Caging plants successfully reduced aphids in the no-aphid
treatment in comparison with the open treatment and aphid ﬁeld populations (mean
aphids / plant :t SE for the no-aphid treatment, 2004 = 0.4 :t 0.2; 2005 = 31.6 2!: 12.4;
contrasts with ﬁeld, F = 152.93; df= 1, 36; P < 0.001; with open, F= 596.07; df= 1, 36;
P < 0.001). I did not detect signiﬁcant effects of agronomic treatments on aphids (F =

1.92; df= 2, 6; P = 0.2623).

85

600 I +2004

l +2005 *
400 I r
300 l

200 I a:

l 4:

1004* _;

l

Aphids / plant

 

o . A ~~ ,- A, MW i—m—mw—u- ~7-—— ,_

6/27 7/4 7/11 7/18 7/25 8/1 8/8 8/15
Sampling dates

Figure 4.1. Open ﬁeld populations of A. glycines at the biodiversity plots of the Kellogg
Biological Station — Long Term Ecological Research site in Michigan during 2004 and
2005. Asterisks above the line denote signiﬁcant differences between years in equivalent
sampling weeks (pre-planned contrasts, P < 0.001, performed on initial six weeks of
sampling).

Aphid suppression

The presence of ants in 2004 did not signiﬁcantly reduce aphid abundance either
as a main effect (F = 3.49; df = l, 2, P = 0.2027) or as an interaction with agronomic or
natural enemy treatments (all P > 0.20), therefore this factor was removed from
subsequent analysis. Although I did not detect signiﬁcant effects of agronomic
treatments (Table 4.1, P = 0.1116), there was a signiﬁcant agronomic treatment by
natural enemy by date (year) interaction (Table 4.1, P = 0.0365). Slicing by main effects
revealed that aphid abundance was signiﬁcantly affected by agronomic treatment in only
5 out of the 48 potential contrasts, indicating the absence of a consistent effect of
agronomic treatments on natural enemy treatments and sampling dates. Therefore, I

present the results averaged across agronomic treatments (Figure 4.2). Natural enemy

86

14x-.— . A‘

Table 4.1. Results of the ANOVA for effects of year, agronomic treatment, and natural

enemy treatment on A. glycines abundance.

 

Source df F P
Year 1, 3 40.86 0.0078
Agronomic Treatment 2, 6 3.23 0.1116
Agronomic Treatment x Year 2, 6 0.26 0.7804
Natural enemies 3, 54 65.67 <.0001
Agronomic Treatment x Natural Enemies 6, 54 1.38 0.241
Natural Enemies x Year 3, 54 7.96 0.0002
Agron. Treat x Natural Enemies x Year 6, 54 1.12 0.3624
Date (Year) 10, 360 155.02 <.0001
Date (Year) x Agronomic Treatment 20, 360 1.62 0.0457
Date (Year) x Natural Enemies 30, 360 7.85 <.0001
Date (Year) x Agron. Treat. x Natural Enemies 60, 360 1.39 0.0365
Block 3, 1.2 6.66 0.2429
Block x Year 3, 6 0.57 0.6567
Block x Agronomic Treatment 6, 6 1.2 0.4142
Block x Year x Agronomic Treatment 6, 54 1.32 0.2659
Block x Year x Agron. Treat. x Natural Enemies 54, 360 4.45 <.0001

 

manipulations had signiﬁcant effects on aphid abundance that changed in intensity with
year and sampling date (signiﬁcant natural enemy by date [year] interaction, Table 4.1).
Large predators dramatically reduced aphid populations during 2004 (207-fold reduction
comparing aphid and open treatments, and 86-fold reduction comparing parasitoid refuge

and open treatments at peak aphid populations) and although this effect was lower in

87

2005 (63- and 36-fold reduction, respectively) in both cases it was highly signiﬁcant

(Table 4.1, Figure 4.2).

100000 a) 2004

. l a

10000 ,

 

1000 2

100 j

10

1 i ' ' I I“ l " 'I ' TI ' " LT "II ' ' " ”__' "I
7/1 7/11 7/21 7/31 8/10 8/20 8/30 9/9

Aphids / plant

100000 4 b)2005 a a

10000 3 "
1000‘, "“
100
j --«v--Panmmotraeﬁge
10 l
j — - - - Threshold level
1 ,_.. MAT--- , _. I .L _ -__.T._ ___—L_ I _ ___

 

6/23 6/30 7/7 7/14 7/21 7/28 8/4 8/1 1

Sampling dates

Figure 4.2. Effect of natural enemy treatments on the number of A. glycines (mean 3:
SE) during a) 2004, and b) 2005. Different letters on each sampling date indicate
signiﬁcant differences (P < 0.05, Least Mean Square Differences adjusted by the
sequential Bonferroni method for multiple comparisons). Statistical comparisons were
restricted to the six initial sampling dates, when aphids reach peak populations.

88

Aphid populations increased rapidly in the aphid and parasitoid refuge treatments
and exceeded threshold levels three (2004) and two (2005) weeks after initial infestation
with aphids. By contrast, large predators maintained aphids below threshold for the
entire season in 2004 in the open treatment, whereas in the sham treatment a maximum of
approximately 800 aphids / plant were observed 4 weeks after infestation and then
declined to the same level as in the open treatment (Figure 4.2a). In 2005, an outbreak of
A. glycines occurred throughout the US Midwest, and aphids reached threshold
populations four weeks after infestation, with a maximum level of 2,500 aphids / plant at
week 6 in the sham and open treatments. However, these population levels were reached
3 weeks
earlier in the aphid and parasitoid refuge treatments, in which large predators were
excluded (Figure 4.2b).

During 2004, I observed a signiﬁcant delay (approximately two weeks) in the
timing of peak aphid density in the parasitoid refuge versus the aphid treatment (Figure
4.2a). A shorter delay was observed in 2005, although in general aphid populations were
slightly lower in the parasitoid reﬁrge (Figure 4.2b). Since there was much more
parasitism in 2004 than in 2005 (see 3.5), I tested whether the delay in aphid peak
populations may have been a result of parasitism. To test this, aphid abundance was
contrasted for the ﬁrst 6 weeks of the experiment in the parasitoid refuge treatment
between plots in which I found parasitoids (n= 5) versus plots in which I did not ﬁnd
parasitoids (n = 7), irrespective of block and agronomic treatment. There were no
signiﬁcant differences related to the presence of parasitoids (F = 2.70, df = 1, 10; P =

0.1311) and a trend opposite to the expected, with more aphids in plots with parasitoids.

89

Thus, the lower numbers of aphids in the parasitoid refuge treatment in comparison with

the aphid treatment were not attributable to parasitism.

Soybean biomass and yield

Agronomic treatment did not affect soybean biomass (F = 0.71; df = 2, 15, P =
0.5081) or yield (F = 0.53; df = 2, 15, P = 0.5965), and therefore results were averaged
across the three treatments. I observed signiﬁcant differences in biomass (F = 42.69; df =
4, 80, P < 0.0001) and yield (F = 70.11; df = 4, 80; P < 0.0001) due to natural enemy
manipulation treatments (Figure 4.3). The presence of soybean aphids decreased biomass
and yield between 72 to 90% in comparison with the controls with no aphids, but
exposure of aphids to predators in the sham and open treatments resulted in a trophic
cascade that restored both plant yield and biomass (Figure 4.3). Biomass was
signiﬁcantly lower in 2004 than in 2005 (F = 10.77; df = 1, 15; P = 0.0050). I observed a
non-signiﬁcant trend of lower yield in 2005 in all except the open treatment (year by
treatment interaction: F = 5.65; df = 4, 80; P = 0.0005). These results conﬁrm
independent ﬁndings of the potential for extant predator assemblages to reduce A.
glycines populations to levels that prevent signiﬁcant plant damage (Costamagna et al.,

unpublished results).

Predator abundance

In 2004, the single plant sampling yielded only 17 natural enemies, and thus these
data were excluded ﬁ'om statistical comparisons. By contrast, in 2005 a total of 1,197
natural enemies were observed in the quadrats, which were dominated by Coccinellidae
(76.8% of all predators), including H. axyridis (36.4%), Hippodamia variegata (Goeze)

(4.4%), Coccinella septempunctata L. (3.6%), and Coleomegilla maculata (De Geer),

90

a) Soybean biomass

  
     

 

 

 

[32004 El 2005
‘2"
E
Q
E! b
§ B
E
.2 c
in C
Parasitoid Aphid
Refuge
b) Soybean yield
a
4° , A
35 l a 132004 1352005

   

Yield (9 I plant)
8

 

 

 

15 .
10 I :_ , ,.
. . . c
5 I I
0 tA My}? T, V :,;,:,;,.',-' _.T. m ., ‘
N° Aphid Parasitoid Aphid
Refuge

Figure 4.3. Effects of natural enemy manipulations on soybean a) above ground biomass
and b) yield (g of seed / plant), in the biodiversity plots of the KBS-LTER site, during
2004 and 2005. Different letters indicate signiﬁcant differences (P < 0.05, LSMD,
adjusted by the sequential Bonferroni method) among treatments within the same year
(capital letters for 2004, lower case letters for 2005).

Cycloneda munda (Say), and Hippodamia convergens Guérin-Menéville, accounting for

the rest (2.5%). In addition, I sampled ﬁrst and second instar coccinellid larvae (19.6%)

91

and pupae (10.2%), that were not possible to identify to species in the ﬁeld, but most
likely were H. axyridis. Other large-size predators included nabids (3.3%), spiders
(2.6%), mirids (2.5%), and lacewing larvae (1.7%), and their abundance was compared
among agronomic treatments grouping them together as “other large predators” (Figure
4.4). Among small-size predators, Orius insidiosus (Say) (Heteroptera: Anthocoridae)
was the most abundant (9.8%), followed by midge (2.3%) and syrphid (1.2%) larvae
(these last two groups were combined as “other small predators,” Figure 4.4). Overall,
predator abundance was low (Figure 4.4), and was not affected by agronomic treatment
(coccinellids: F = 1.28; df = 2, 6; P = 0.3445; 0. insidiosus: F = 1.45; df = 2, 6; P =
0.3055; and, averaged across sampling weeks, other large predators: x2: 0.546; df = 2; P

= 0.761; and other small predators: x2 = 3.512; df= 2; P = 0.173).

 

 

 

5 .t -—e—Coccinellidae ---x---LargePredators
i +0. insidiosus ---°--- Srmll Predators
N 4
E l ..
E 3 1 l vA>
ii 2 ..
CL. 1
1i " ::: £3; ..... i
l I-‘-«~-.----x ........
O _ = -.__L I
26-Jun 3-Jul 10-Jul 17-Jul 24-Jul 31-Jul 7-Aug

Sampling dates

Figure 4.4. Open ﬁeld predator populations sampled in the biodiversity plots of the
KBS-LTER site during 2005. Large-size predators includes nabids, spiders, small
carabids, mirids and lacewing larvae; small-size predators includes midge and syrphid
larvae.

92

Cage manipulations resulted in signiﬁcant differences in the number of predators
present in each treatment (Figure 4.5). During both years coccinellids were successfully
excluded from the aphid and parasitoid refuge treatments, although in weeks 5 and 6 of
2005, coccinellid larvae were found in parasitoid refuge treatments (Figure 4.5b). The
abundance of 0. insidiosus was initially reduced only in the aphid treatment, and showed
a trend of higher abundance in the parasitoid refuge treatment (Figure 4.5e and d). Large
predators other than coccinellids were signiﬁcantly more abundant in sham cages only
during the fourth week of 2004, although at very low levels (Figure 4.5e and f). Small
predators (i.e. midge and syrphid larvae) were initially higher in the sham and open
treatments at low levels, and then reached high but variable abundances in the parasitoid
refuge and aphid treatments, probably as a response to the increase densities of aphids on
those treatments (Figure 4.5g and h). Thus, cage manipulations successfully excluded
coccinellids, but did not prevent small predators and 0. insidiosus from occurring in
parasitoid refuge cages. To test for a potential difference in the abundance of natural
enemies between 2004 and 2005, I compared the total number of natural enemies
observed, grouped for all four natural enemy treatments, for each sampling date. Overall,
predators were more abundant in 2005, probably responding to increased aphid densities
in the ﬁeld (Figure 4.1). In 2005, signiﬁcantly more 0. insidiosus and coccinellids were
observed in weeks 2, 5 and 6 after the beginning of the experiment, and more large
predators in weeks 2 and 5 (Figure 4.5a - f). The abundance of small predators did not

differ between years (Figure 4.5 g and h).

93

6.0 . a) 2004 6.0 b) 2005 ..

Coccinellids / plant

 

0. insidiosus / plant

 

 

Larger pred. / plant

 

   

18 , h) 2005

g 15 '1

a.

\

23;

a.

'3

E

m - .

7/7 7/17 7/27 8/6 8/16 8/26 9/5 6’26 7/6 7/16 7/26 8/5
Sampling dates

Figure 4.5. Effect of natural enemy manipulations on the abundance (mean + SE) of the
four major groups of predators sampled in the biodiversity plots of the KBS-LTER
during 2004 and 2005. Asterisks above treatment lines indicate signiﬁcant differences
among natural enemy treatments, asterisks below sampling dates indicate signiﬁcantly

94

Figure 4.5 (cont’d)

higher abundance for all treatments combined in that year (Kruskal-Wallis tests, * = P <
0.10, ** = P < 0.05, *** = P < 0.01). Statistical comparisons were performed for the six
initial weeks.

Parasitoid abundance and intraguild predation

Sampling of natural populations did not yield any parasitoids outside cage
manipulations during 2004. However, 1,934 parasitoids were observed in cage
manipulations, most of which were Braconids (85%) and the rest were Aphelinidae
(15%). Despite the aphid outbreak and the more intensive sampling performed in 2005,
only 5 braconid mummies were detected outside of cages, and 42 mummies in cage
manipulations (95% Braconidae). Therefore, only data from 2004 were considered in the
following analyses of parasitism. I collected three parasitoid species: the braconids
Lysiphlebus testaceipes (Cresson) and Binodoxys kelloggensis Pike et Stary, and the
aphelinid Aphelinus asychis (Walker). There were no parasitoids in the aphid treatment
and only one mummy in the open treatment, and therefore these treatments were
excluded from statistical comparisons.

I observed signiﬁcantly higher cumulative number of mummies in the parasitoid
refuge than in the sham treatment (F = 8.30; df = 1, 9; P = 0.0182).- Mummy counts were
not affected by agronomic treatment (F = 0.24; df = 2, 6; P = 0.7953), and therefore this
factor was excluded for the following analyses with parasitoids. Non-parametric
comparisons per date showed that the trend of higher mummy abundance observed in the
parasitoid refuge treatment was signiﬁcantly different on only two dates (Figure 4.6a).
To test whether this difference was due to intraguild predation or to a response to the
higher aphid densities of the parasitoid refuge treatment, I compared the percentage

parasitism

95

a) Total number of mummies
50 1
1 + Parasitoid Refuge

404 a:
1! -O--Sham *

 
 
    

Mummies / plant

b) Percentage parasitism
10.000 7
I

1.000 1
0.100 9
0.010 *

% Parasitism

 

0.001
0.000 -—»- e. _, . “-- -------

c) IPG on mummies

 

 

 

% Mummies damaged /
plant
0 _. to 00 :> 01 a:

I

7/21 7/28 8/4 8/11 8/18 8/25 9/1 9/8
Sampling dates

Figure 4.6. Effect of predator manipulations on a) number of mummies / plant, b)
percentage parasitism (proportion of mummies to aphids x 100), and c), number of
mummies damaged by predators; averaged across agronomic treatments. Only one
mummy was obtained in the open treatment and none in the aphid treatment, and
therefore these treatments were excluded from statistical comparisons. Treatments were
compared independently for each date using Wilcoxon tests (* = P < O. 10).

96

among treatments, and found no signiﬁcant difference between parasitoid refuge and
sham treatments (Wilcoxon tests, all P > 0.176, Figure 4.6b). Finally, IGP was directly
assessed by comparing the number and percentage of mummies damaged by predation
between the parasitoid refuge and the sham treatment. Although I observed a trend of
more damaged mummies in the sham treatment, the highly variable incidence of IGP in
this treatment resulted in an absence of signiﬁcant differences (Wilcoxon tests, all P >
0.180, Figure 4.6c). Contrasting IGP on all the possible pairs in which parasitism was
obtained in both treatments within the same plot, revealed that proportionally
signiﬁcantly more mummies were damaged in the sham (18.1 d: 7.6%, range 0 — 94.4%)
than in the parasitoid refuge treatment (3.7 :1: 2.4%, range 0 — 36.4%, Wilcoxon test 2 =
2.1, P = 0.036). This result shows that IGP is occurring at the very low levels of

parasitism observed in this study.

Discussion

Despite an expected positive role of ant tending on aphids (e. g. Vélkl 1992,
Stadler et a1. 2002), I did not ﬁnd signiﬁcant effects of ants on aphid abundance.
However, this interaction can not be ruled out from the A. glycines system, since my
study was not explicitly designed to test this factor and I did not carry on observations on
this interaction during 2005. As anticipated, cover crops did not affect signiﬁcantly
aphid abundance in my experiment, conﬁrming previous results of experiments using
similar agronomic treatments (Costamagna and Landis 2006).

During the two years of study, I observed strong suppression of A. glycines due to

generalist predators, but I did not ﬁnd signiﬁcant effects of extant parasitoids. These

97

ﬁndings thus support an increasing number of studies showing an important role of
generalist predators controlling A. glycines in North America (Brown et al. 2003, Fox et
a1. 2004, Rutledge et al. 2004, Fox et al. 2005, Nielsen and Hajek 2005, Rutledge and
O'Neil 2005, Costamagna and Landis 2006). Moreover, my results showed that even
during outbreak years such as 2005, predators were able to delay aphid increase above
threshold levels in comparison with controls in which they were excluded, conﬁrming
earlier studies (Fox et a1. 2004, Costamagna and Landis 2006). In addition, I showed that
soybean aphid has the potential to cause a 3.5 to 10-fold reduction on soybean yield and
biomass, but that both were restored to normal levels by exposure to predation, during
both years of study. Similar results were obtained in a different experiment using 1 m2
ﬁeld cages by Costamagna et al. (unpublished results) during 2004. These ﬁndings
suggest that predators are a key factor in soybean aphid suppression and their input
should be taken into account in determining action thresholds.

The different cages used for the natural enemy treatments allowed separation of
the effect of 0. insidiosus and small predators from Coccinellids and other large
predators. Previous studies have suggested that 0. insidiosus, due to its abundance on
soybeans early in the season, and potential for aphid consumption in laboratory
conditions, may play an important role reducing A. glycines population growth (Rutledge
et al. 2004, Rutledge and O'Neil 2005). I observed similar or even higher numbers of 0.
insidiosus and midge and syrphid larvae in the parasitoid refuge cages than in sham and
open cages, conﬁrming that these predators had similar access to aphids in those
treatments. However, aphids increased exponentially to very high numbers in the

parasitoid refuge treatment during both years, despite being exposed to predation by

98

small predators. These results indicate that 0. insidiosus and other small predators were
unable to prevent A. glycines exponential growth. However, ﬁeld experiments were
started at levels of 5 aphids / plant, and thus I can not rule out a potentially greater impact
of 0. insidiosus at lower aphid densities, earlier in the season. In conclusion, I observed
strong reduction in aphid numbers only in the treatments that had high numbers of
coccinellids and large predators (sham and open treatments), indicating that these were
the predators that had the greater impact on aphid populations in the ﬁeld.

On the other hand, my study suggests that 0. insidiosus and other small predators
may have had a low to moderate impact on aphids. There was a delay in peak aphid
populations in the parasitoid refuge treatment with respect to the aphid treatment in 2004,
and slightly less overall aphid densities during 2005 (Figure 4.2). However, there were
very few parasitoids during 2005, and there was no signiﬁcant association between
parasitism occurrence and aphid abundance in 2004. Therefore, I concluded that the
lower aphid densities observed were not a consequence of parasitism. An alternative
explanation is that alate aphids were conﬁned in the aphid treatment due to the ﬁne mesh
of the cage, and inﬂated aphid density with respect to the parasitoid refuge treatment. To
test that, I compared the proportion of alates (number of alates / total number of aphids)
between these two treatments and found that the proportion of alates was signiﬁcantly
higher in the aphid treatment than in the parasitoid refuge only during the fourth week of
the experiment, during both years (natural enemy by date [year] interaction: F = 3.03; df
= 10, 80; P = 0.0014, LSMD tests on week four, P values < 0.001). In general alates
were at low densities (at week 4: 0.3 versus 2.7% in 2004, 1.8 versus 6.4% in 2005, for

the parasitoid refuge and aphid treatment, respectively), indicating that probably a

99

 

combination of alate emigration and the impact of small predators is likely to explain the
difference on aphid densities observed between the parasitoid refuge and aphid
treatments. In addition, these results suggest that to conservatively assess the impacts of
large predators, aphid densities on the sham and open treatments should be compared to
the parasitoid refuge treatment.

Finally, three species of parasitoids were observed attacking A. glycines, but there
was no signiﬁcant effects of parasitism in reducing aphid populations, even when IGP
was minimized with predator exclusion cages. I did observe direct evidence of IGP, by
quantifying damaged mummies, and this measure is likely to be conservative, since
predators may completely dislodge some of the parasitoid mummies during consumption
(Brodeur and Rosenheim 2000). The maximum levels of IGP detected were around 94%
of mummies damaged by predation, although on average, IGP was less than 20%. Colfer
and Rosenheim (2001) studied IGP of L. testaceipes by the coccinellid Hippodamia
convergens in cotton and found that despite levels of IGP of up to 98 — 100%, the
addition of the coccinellid added suppression of cotton aphid over the presence of the
parasitoid alone. By contrast, in the A. glycines system I obtained very low parasitism
levels, typically around 1%, and no evidence of reduction in aphid growth due to
parasitism.

In conclusion, generalist parasitoids are attacking A. glycines in Michigan
soybean production systems and they experience signiﬁcant IGP by generalist predators.
However, IGP did not affect the percent parasitism, suggesting the absence of any
disruptive effect of predation at the levels of parasitism detected. The relatively few

studies on the effect of IGP on herbivore control by parasitoids involves systems where

100

parasitism levels are high, and therefore the potential for disruption is greater (Colfer and
Rosenheim 2001, Snyder and Ives 2001, 2003). In the A. glycines system, the recent
arrival of this exotic pest to North America results in a lack of adaptation of the extant
parasitoid species to use it as a host, as seen with other aphid species (Brewer et al.
2005). My results showed that resident parasitoids of A. glycines are not currently
disrupted through IGP. Generalist predators act as major sources of A. glycines mortality,
although during some seasons aphids escape their control. However, I demonstrate a
potential for IGP that may reach high levels in some cases and may contribute to delay
the adaptation process of resident parasitoid species to A. glycines. Moreover, the
potential disruptive effect of generalist predators may be important if more effective
parasitoids species are added to the system and this should be considered in any classical

biological control effort involving parasitoids.

101

CHAPTER 5

DOES APHIS GLYCINES HAVE REFUGES FROM TOP-DOWN CONTROL BY

GENERALIST NATURAL ENEMIES?

Abstract

The soybean aphid, Aphis glycines Matsumura (Hemiptera: Aphididae) recently
invaded North America, causing signiﬁcant economic losses. Previous studies
demonstrated strong suppression of A. glycines by natural enemies, however, in some
years A. glycines reach outbreak populations. I investigated the existence of refuges from
predation as one of the potential factors leading to A. glycines escape from predation. I
tested the presence of ‘numeric refuges’, in which colonies with higher aphid densities
were released from predator control, and spatial refuges at the within-plant level. I
followed naturally occurring A. glycines populations at high, medium, and low initial
densities and compared them with populations of the same initial level protected by
predator exclusion cages. In addition, I compared the vertical distribution of the aphids at
the within-plant level for the same colonies. Predators provide strong and signiﬁcant
reductions of A. glycines populations regardless of initial aphid density. Interestingly,
there were lower per capita rates of increase at higher than at lower initial aphid
populations. Exposure to predators signiﬁcantly altered A. glycines distribution within
the plant, with proportionally more aphids at the lower plant nodes in colonies exposed to

predation. These results suggest that predator suppression is effective under the naturally

102

occurring range of aphid densities and that the lower portion of the plant may represent a

partial refuge from predation for A. glycines.

Introduction

The soybean aphid, Aphis glycines (Hemiptera: Aphididae) is a new invasive pest
in North America (Losey et al. 2002, Ragsdale et al. 2004). The potential of natural
enemies to suppress A. glycines populations has been shown both in its original area of
distribution in Asia (Liu et al. 2004, Wu et al. 2004), and recently in North America (Fox
2002, Fox and Landis 2003, Fox et al. 2004, Rutledge et al. 2004, Fox et al. 2005,
Rutledge and O'Neil 2005). However, outbreak populations of A. glycines have led to
signiﬁcant yield reductions during alternating years since its ﬁrst detection in 2000
(DiFonzo and Hines 2002, Venette and Ragsdale 2004, Myers et al. 2005a). The causes
of the escape of A. glycines from natural enemy control during those outbreak years
remain poorly understood.

One potential explanation of the lack of control of A. glycines is the existence of
refuges from predation. Here the term refuge is used broadly for any spatial, temporal or
numerical scenario that release prey populations from suppression by predators. Thus,
aphid colonies of high density may constitute a refuge from predation through a dilution
effect on the mortality risk (Turchin and Kareiva 1989). If that is the case I predict that
ﬁeld patches with higher aphid densities will increase aphid density more rapidly than
ﬁeld patches with lower aphid densities, constituting a ‘numerical refuge’ for the aphids.
In addition, differential foraging by natural enemies has been shown to result in refuges

at the within-plant scale (Hacker and Bertness 1995, Gonzales et al. 2001, Magalhﬁes et

103

a1. 2002). I conducted ﬁeld studies to determine the existence of numerical and spatial
refuges from predation in the A. glycines system. Speciﬁcally I tested for the presence of
refuges in ﬁeld patches with different densities of A. glycines and within the vertical

distribution of the aphids on soybean plants.

Materials and methods

Experiments were conducted in a conventional soybean ﬁeld (50 x 100 m) located
on the Entomological Farm of Michigan State University, Michigan, USA during 2003.
Soybeans (Pionner 9283 8) were planted on 05/22/2003, and maintained with
conventional agricultural techniques. In this ﬁeld, naturally occurring colonies of A.
glycines were caged at different patch densities to test whether initial aphid density
inﬂuenced suppression by predators. On 7/3/2003, the ﬁeld was divided in 50,10 x 10 m
patches that were sampled to determine aphid density. Sampling consisted of counting
the total number of aphids (nymphs, apterous adults, alates) on 10 randomly selected
plants per patch. A. glycines densities obtained in the grid sampling were used to classify
ﬁeld patches in three categories according to their initial aphid density: low (< 5 aphids /
plant), medium (5 — 10 aphids / plant) and high (> 10 aphids / plant). In each category, 6
patches were randomly chosen and received three predator manipulation treatments:
predator exclusion, a sham cage or no-cage. The exclusion cage consisted in a cylindrical
framework of wire of 0.4 x 1.0 m, covered by a ﬁne mesh white no-see-um netting
(Kaplan Simon Co., Braintree, MA), buried 25 cm in the soil and tied to the top. The
sham cage was identical except the mesh was held 5 cm above ground surface attached

by metal clips to four wooden stakes, and had eight rectangular openings (2 x 20 cm)

104

arranged in 3 circular patterns alternating in a manner such that there were openings in all
directions. These rings of openings were located from 10 to 30 cm above the soil,
encompassing the height range of the canopy. Finally, the no cage control (“open”)
consisted of three contiguous uncovered plants marked by ﬂags nearby.

Each cage enclosed the focal plant containing a naturally occurring aphid colony
within the range desired and, for the ﬁrst two weeks, two additional soybean plants
initially free of aphids. Aﬁer two weeks, the plants were reduced to one per cage, leaving
typically the central one (unless it was damaged), to allow plants to grth uncrowded by
the cage. The experiment started on 7/03/2003, when plants were at the V3 - V4 stage
(Ritchie et al. 1994). Once a week the total number of aphids and their location within
the plant was assessed by direct counts. After two weeks I calculated the per capita rate
of increase as PCRI = No / N”, where Nto is initial aphid density, and NM is the density
at day 14 after the beginning of the experiment. To rule out potential local effects of the
soybean plants on aphid growth, exclusion and sham cages were switched in a subset of
the cages after two weeks. On 7/17/2003 I matched cages in pairs according to the
density of aphids in the exclusion cages, and in one of the plots the sham and exclusion
cages were switched, and in the other they were left un-switched as controls, for a total of
6 plots switched and 6 control plots. The experiment was conducted for three additional
weeks, but sampling of the control plots was stopped at week 4 due to the extraordinary
number of aphids produced and the fact that alates could not emigrate, artiﬁcially

inﬂating numbers in exclusion versus sham and open treatments.

105

Statistical analysis

The effects of initial aphid density and natural enemy control on A. glycines
population growth and per capita rate of increase were assessed by ANOVA using a split-
plot design, with aphid density as the whole-plot factor and natural enemy as the sub-plot
factor (Proc MIXED, SAS Institute 2001). For the aphid population increase analysis,
sampling date was analyzed as a repeated measures factor. Signiﬁcant interactions were
explored slicing by main effects (Quinn and Keough 2002). Means were compared using
Least Square Mean Difference tests adjusted for multiple comparisons by the sequential
Bonferroni test (Rice 1989). Within-plant aphid distribution was analyzed by MANOVA
on a split-plot design with initial aphid density as the whole-plot factor and natural enemy
as the subplot factor using Proc GLM (SAS Institute 2001). The proportions of aphids
present in each node were analyzed as the dependent variable of the MANOVA and
separate analysis were conducted for each date due to the variable number of nodes.
Roy’s greatest root was reported as the statistic for the MANOVA due to its power to
detect differences among treatments (Scheiner 2001). Data were log or arcsine-
transformed previous to statistical analysis to meet the normality and homocedasticity
assumptions.

Results

Initial A. glycines density
After two weeks of manipulation, predator exclusion resulted in signiﬁcantly
higher aphid densities, ranging 9 to 28 times above the sham and open controls (Figure

5.1, Table 5.1). Initial aphid density also signiﬁcantly affected aphid abundance, with

higher

106

Table 5.1. ANOVA results for the effect of initial aphid density and natural enemy
control on the population growth of A. glycines natural populations in Michigan.

 

 

Source df F P
Aphid density 2, 15 12.36 0.0007
Natural enemies 2, 118 153.29 <.0001
Aphid density x natural enemies 4, 118 2.67 0.0355
Date 2, 118 303.32 <.0001
Aphid density x date 4, 118 0.91 0.4621
Natural enemies x date 4, 118 46.77 <.0001

Aphid density x natural enemies x date 8, 118 0.47 0.8756

aphid abundances at the high than at the medium and low initial aphid densities (Ps <
0.05, LSMD tests). Slicing the signiﬁcant aphid density by natural enemy interaction
revealed that natural enemy treatments differed signiﬁcantly under all initial aphid
densities (Ps < 0.005), and initial aphid density affected aphid abundance in all natural
enemy treatments (Ps < 0.005). The signiﬁcant interaction is therefore due to the lower
aphid abundance observed in the sham and open treatments of the lower aphid density in
comparison with medium and high (Figure 5.1). The signiﬁcant natural enemy by date
interaction is due to the expected lack of differences between natural enemy treatments at
the start of the experiment (P = 0.9734, Figure 5.13). I did not observe signiﬁcant cage
effects at high and low initial aphid densities, and only small differences between sham
cages and open plots at the medium density, suggesting that potential cage effects were
minimal (Figure 5.1). In summary, although there was a signiﬁcant effect of initial aphid

density, this effect was more important in exclusion cages, where treatments exposed to

107

160

140 ‘
120 '

100

40
20

A. glycines / plant

1000

800

600

400

200

0

80“
60‘

a) Before natural enemy manipulation
ns

I Exclmion
C1 Slam

ClOpen

1 ill-ii:

b) One week after natural enemy manipulation
a

b b

     

0) Two weeks aﬁer natural enemy manipulation

‘ a

1 a a

l

l b b 'E c I b b
High LOW

Medium
Initial A. glycines density

Figure 5.1. A. glycines population growth from three initial densities under three
predator manipulation treatments. Mean (i SE) aphid abundance is shown at a) before
natural enemy manipulation, and b) one week, and c) two weeks after manipulation.
Different letters represent signiﬁcant differences among natural enemy treatments within
patch density treatments (P < 0.05, Least Square Mean Differences).

108

predation maintained were well below the 250 aphids per plant threshold level in all cases
(Figure 5.1).

Comparing the per capita rate of increase (PCRI) showed similar results, with
strong effects of natural enemies (F = 129.75; df = 2, 29; P < 0.0001), marginally
signiﬁcant effects of initial aphid density (F = 3.48; df = 2, 15; P = 0.0573) and no
signiﬁcant aphid density by natural enemy interaction (F = 1.05; df = 4, 29; P = 0.3934;
Figure 5.2). Comparing the means of the natural enemy treatments revealed a 16-fold
decrease in PCRI on sham and open treatments in comparison with the predator exclusion
treatment (LSMD tests, Ps < 0.0001), and no signiﬁcant differences between the sham
and open treatments (LSMD test, P > 0.05, Figure 5.2). Interestingly, I observed up to 3-
fold lower PCRI plots that had high versus low initial aphid density, and that difference
was marginally signiﬁcant (LSMD test, P = 0.0638, Figure 5.2). This result suggests a

potential self-limitation on A. glycines population growth.

5001

400 l I EXCIUSIDII
; El Slum
300 ’J 121 Open
200
100 l
l
0 WW TIAT‘ ’ I .
High Med'mm Low

Initial A. glycines density

Aphids / aphid/ 14 days

Figure 5.2. Mean (i SE) per capita rate of increase of A. glycines at three initial
densities under three natural enemy manipulation treatments after 14 days. Signiﬁcant
main effects are described in the text.

109

 

Switching of the cages conﬁrmed high impacts of predation; aphid populations
differed signiﬁcantly among predator manipulation treatments (F = 13.75; df = 5, 25; P <
0.0001), and sampling dates (F = 35.11; df = 1, 30; P < 0.0001). Reversing of the
trajectories of switched versus control cages is indicated by a signiﬁcant cage by date
interaction (F = 4.21; df = 5, 30; P = 0.0051, Figure 5.3a and b). Due to the
extraordinarily high aphid populations observed the control cages were interrupted on
week 4. Therefore, A. glycines densities for the ﬁfth week were compared separately,
showing a complete reversal in the trajectories of the aphid populations, with
signiﬁcantly higher densities in the exclusion than the open treatment (F = 3.98; df = 2,

10; P = 0.0534; Figure 5.3c).

Within-plant distribution

Aphid distributions within plants were not affected by initial aphid density (Table
5.2). Most of the plants were at the V4 stage (few were V3 or V5), and most of the
aphids were located within the ﬁfth node, which represented the youngest tissue available
in most cases (Figure 5.4). One week aﬁer natural enemy manipulation I observed
signiﬁcant effects of aphid initial density on the within-plant distribution (Table 5.2),
with aphids at the low density treatment concentrated more on the youngest tissue (nodes
5 to 7, Figure 5.5c), whereas at the low and medium initial density treatments
proportionally more aphids were at lower nodes (Figure 5.5a and b). Plants were at the
V5, V6 or R1 developmental stage, and in some cases had secondary trifoliates emerging
from the ﬁrst and second node. Two weeks after manipulation, exposure to natural

enemies resulted in highly signiﬁcant differences on aphid distribution, whereas initial

110

aphid density was only marginally signiﬁcant (Table 5.2). I observed that the within-

plant distribution of A.

a) Three weeks after predator manipulation

4000
I Exclusion

. El Sham
3000 1:] Open
2000 ab
1000 I

. c be be bC

0

A. glycines / plant

 

c) Five weeks aﬁer predator manipulation

12000 1
‘ a

9000 ‘

6000 *1

3000 ‘ .

Ended prevrous ab b
week
0 . -

Control Switched

Figure 5.3. A. glycines population growth under three predator manipulation treatments.
Mean (:1: SE) aphid abundance is shown at a) three, c) four, and c) ﬁve weeks after
predator manipulation. Plots were paired according to aphid density in the exclusion
treatment at week 2 and randomly chosen to switch exclusion and sham cages (switched)
or remained un-switched (control). Control treatments were interrupted at week four due

lll

to the extraordinary high aphid densities achieved, and therefore week ﬁve was analyzed
separately (see methods). Different letters represent signiﬁcant differences among
natural enemy treatments (P < 0.05, LSMD adjusted by the sequential Bonferroni method
for multiple comparisons).

Table 5.2. MANOVA results for the effects of different initial aphid densities and
natural enemy manipulations on the within-plant distribution of A. glycines.

 
 
 

 

 

 

df F’ P
Before manipulation
Aphid density 6, 26 1.37 0.2619
Natural enemies 6, 26 2.12 0.0847
Aphid density x natural enemies 6, 28 2.67 0.0357
One week after manipulation
Aphid density 9, 20 3.44 0.0103
Natural enemies 9, 20 2.24 0.0644
Aphid density x natural enemies 9, 22 2.05 0.0814
Two weeks after manipulation
Aphid density 12, 18 2.23 0.0607
Natural enemies 12, 18 6.04 0.0004

 

1 F values of Roy’s greatest root statistic (Scheiner 2001).

112

a) High initial density

 

 

 

I Exclmion
B Sham
.1 Dove"
6 F b) Medium initial density

 

 

Plant node
b.)

 

 

 

 

0 0.2 0.4 0.6 0.8 1
Proportion of A. glycines population

Figure 5.4. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and 0) low densities before assigning the predator manipulation treatments.
Cages were placed immediately after sampling aphid abundance. The statistics for this

ﬁgure are presented in Table 5.2.

113

 

a) High initial density

 

I Exclusion
El Sham

ClOpen

~wau~0xxloo©
_.i’ll" ‘“ g
l

b) Medium initial density

Ox \1 00 so
‘ .L_l_.-

L

i

L

 

Plant node
H N W «5 Vi
WEI HE?!“

C) Low initial density

 

 

 

0 0.2 0.4 0.6 0.8 1
Proportion of A. glycines population

Figure 5.5. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and c) low densities one week after predator manipulation treatments. The
statistics for this ﬁgure are presented in Table 5.2.

114

 

12 - a) High initial density I EXCllBiOD
11 El Sham

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plant node

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E we ....... _.--
0

0.1 0.2 0.3
Proportion of A. glycines population

Figure 5.6. Within-plant distribution of naturally occurring A. glycines at a) high, b)
medium, and c) low densities two weeks after predator manipulation treatments. The

statistics for this ﬁgure are presented in Table 5.2.

115

 

glycines in the predator exclusion treatment tended to be shifted towards the upper
portion of the plant, whereas in the sham and open treatments a relatively high proportion
of the population was concentrated in lower parts of the plant (Figure 5.6). Plants were
mostly at the R1 stage and many had secondary trifoliates up to the fourth and ﬁfth node.
After switching exclusion and sham cages, a similar pattern was obtained (Figure
5.7). After three and four weeks of manipulation, aphid within-plant distribution
continue to differed signiﬁcantly among natural enemy treatments in unswitched controls
(F = 23.26; df = 12, 5; P = 0.0014, and F = 14.60; df= 12, 5; P = 0.0041, respectively).
Aphids were more evenly distributed in exclusion cages, probably due to the
extraordinary densities achieved (average around 12,000 aphids / plant, Figure 5.3b)
forced them to move to less prefered portions of the plant. In switched cages, aphid
densities were much lower and with a good proportion of the aphids remaining at lower
nodes of the plant, despite protection from predation, which resulted in only marginally
differences in week 3 and signiﬁcant but small differences in week 4 (F = 4.63; df = 12,

5; P = 0.0511, and F = 5.22; df= 12, 5; P = 0.0399, respectively, Figure 5.8a and b).

Discussion
In addition to a remarkable reproductive potential, the existence of refuges from
predation, in combination with other factors may provide an explanation for the
development and persistence of outbreak populations in aphids (Dixon 1987, 1998).
Turchin and Kareiva (1989) demonstrated that Aphis varians (Hemiptera: Aphididae) at
high density colonies beneﬁted from a ‘dilution effect’ of lower predation risk in

comparison with colonies with lower aphid densities. I postulated that if such a

116

14 l a) Three weeks after manipulation
13 .— I Exclusion
12

11 ESham
10 ;— DOpen

- l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

v—tNUJ-hUIQQOOO

Plant node

 

 

 

 

 

 

 

 

 

 

 

 

 

u—ANUJ-bUIQQOOC

 

 

0 0.1 0.2 0.3
Proportion of A. glycines population

Figure 5.7. Within-plant distribution of naturally occurring A. glycines at a) three, and b)
four weeks after predator manipulation treatments. Exclusion and sham cages were not
switched, serving as controls for the switching treatment (see Figure 5.8).

117

14 i a) Three weeks after manipulation
13 j I Exclusion
if - Ea Sham
.9
10 &= ‘3 Open
9 __

l

_
Wm- H.
_

I

 

 

ﬁr

 

 

 

 

 

 

 

 

 

 

 

 

i—INUJ-BUIONQOO

Plant node

 

 

 

 

 

 

 

 

 

 

 

 

 

 

V—NUJAUIQQOOO

0 0.1 0.2 0.3
Proportion of A. glycines population

Figure 5.8. Within-plant distribution of naturally occurring A. glycines at a) three, and b)
four weeks after predator manipulation treatments. Exclusion and sham cages were
switched after the second week.

‘numerical refuge’ exists for A. glycines, I would detect lower impacts of predation on

high density aphid patches that may lead to an eventual escape from predation. My

results showed that predators exerted strong suppression at all the aphid densities within

118

the range detected in naturally occurring A. glycines populations during 2003. Moreover,
this pattern was even more convincingly demonstrated when aphid population trajectories
were reversed as a result of switching exclusion and sham cages. Interestingly, I observed
that A. glycines populations showed lower per capita rate of increase at higher aphid
densities, suggesting a potential self-limitation in population growth. Thus, the potential
beneﬁts of lower predation risks at high density patches may be offset by a reduction in
reproduction. Since I did not determine rates of predation in this study, it is not possible
to separate the effect of these two factors determining A. glycines abundance. This
pattern of reduce reproduction at higher aphid densities provides support to the model
developed in Chapter 7, which assumes a linear decrease in the intrinsic rate of increase.
There was a signiﬁcant shift in the within-plant distribution of aphids as a result
of exposure to predation. Similar changes in spatial distribution in response to natural
enemies have been shown for different mites and aphid species. Magalhaes et al. (2002)
showed that the green cassava mite, Mononychellus tanajoa Bondar, migrated vertically
in speciﬁc patterns in response to odors from different predatory mites. Hacker and
Bertness (1995) showed higher Uroleucon ambrosiae (Thomas) densities on shorter than
taller plants as a response to the forage behaviour of its coccinellid predator. Hopkins
and Dixon (1997) showed lower predation levels on the nymphs of the aphid Monaphis
antennata (Kaltenbach) adapted to live on the upper surfaces of leaves and petioles, an
unusual place for other aphids and even the adults of this species. Gonzales et al. (2001)
showed that Rhopalosiphum maidis (Fitch) located in the whorl of corn were less
susceptible to attacks ﬁom the parasitoid L. testaceipes. An extreme adaptation is the

dropping behavior exhibited by the pea aphid, Acyrthosiphon pisum (Losey and Demo

119

1998b), although this behavior could result in increased mortality by ground based
predators (Losey and Demo 1998a, 1999).

I observed that A. glycines exposed to predation shifted its distribution towards
lower nodes in the plant. This is consistent with avoiding predation by coccinellids that
are observed more frequently foraging at the upper portion of plants (Hacker and
Bertness 1995, Nakashima and Akashi 2005). In the A. glycines system, there were
signiﬁcantly more predation events on the upper portion of the plants (Chapter 6),
supporting the existence of ‘enemy free space’ at lower portions of the plant. The
development of predator avoidance behavior usually involves a tradeoff of reduced
reproduction or other ﬁtness components (Hacker and Bertness 1995, Lima 1998,
Magalhaes et al. 2002). The quality of plants for aphid feeding differs with leaf age, with
younger leaves more suitable due to the concentration of metabolites (Dixon 1998).
Thus, it is likely that shifts in A. glycines within-plant distribution as a consequence of
predation result in indirect negative effects on aphid population growth. Further studies
should be conducted to address the importance of this potential indirect effect of

predation and its consequences in A. glycines suppression.

120

CHAPTER 6

QUANTIFYING PREDATION ON SOYBEAN APHID THROUGH DIRECT

FIELD OBSERVATIONS

Abstract

The soybean aphid, Aphis glycines (Hemiptera: Aphididae), is a new pest
attacking soybeans in North America since 2000. Several studies have documented
strong impacts of generalist natural enemies on A. glycines populations using predator
exclusion cages and correlating predator and aphid abundances. However, to date no
studies were conducted to directly determine the natural enemy groups that attack A.
glycines in North America under ﬁeld conditions. In 2005, 72 h of direct observations of
predation on natural populations of A. glycines were conducted in Michigan.
Observations were conducted during three consecutive weeks during morning, afternoon
and evening, in replicated 1 m2 areas. A total of 643 predators within 11 groups and 211
predation events on A. glycines were observed. Transient predators such as Harmonia
axyridis and Coccinella septempunctata (Coleoptera: Coccinellidae) accounted for most
of the mortality observed and were very effective on a per capita basis, despite relatively
short residence time in the observed patches. Transient predators responded positively to
increased A. glycines ﬁeld densities. By contrast, resident predators, particularly Orius
insidiosus (Hemiptera: Anthocoridae), were abundant and accounted for many predation
events, but they were not as effective on a per capita basis and did not respond to changes

in aphid density. I observed lower predation on A. glycines located at the bottom portion

121

of the plants and on alate aphids. I concluded that coccinellids exerted most of the
mortality observed and have the potential to rapidly respond to changes in aphid density

with high per capita rates of predation.

Introduction

The soybean aphid, Aphis glycines Matsumura (Homoptera: Aphididae) is a new
pest of soybeans in North America and has rapidly extended its distribution since its
detection in 2000 to over 22 States in the USA and 3 Canadian provinces (Ragsdale et al.
2004, Venette and Ragsdale 2004, Mignault et al. 2006). Several studies have
demonstrated strong impacts of natural enemies on soybean aphids in Asia (Van den
Berg et al. 1997, Liu et al. 2004, Wu et al. 2004). In North America, ﬁeld experiments
suggest suppression of A. glycines by generalist predators (Fox et al. 2004, Rutledge et al.
2004, Fox et al. 2005, Nielsen and Hajek 2005, Rutledge and O'Neil 2005, Costamagna
and Landis 2006, Costamagna et al., unpublished results). However, most studies to date
relied on indirect methods to establish predator effects, including comparison of predator
exclusion versus sham cages (Fox et al. 2004, Liu et al. 2004, Fox et al. 2005) and
correlations between aphid and predator abundances (Rutledge et a1. 2004, Nielsen and
Hajek 2005, Rutledge and O'Neil 2005). These techniques have distinguished the major
predator groups of A. glycines according to their size, based on different types of
exclusion cages (Costamagna et al, unpublished data), or suggested the most important
predator species based on their co-occurrence with the aphids in the ﬁeld. However,
direct evidence of which are the most important predator species attacking A. glycines in

the ﬁeld is very limited.

122

The only study assessing the relative importance of different predator species
attacking A. glycines in North America was conducted in laboratory conditions by
Rutledge et a1. (2004). This study suggests that within the North American guild of
predators, the coccinellids Harmonia axyridis Pallas, Coccinella septempunctata L., and
Hippodamia convergens G.-Mén., the earwig F orﬁcula auricularia L., the nabid Nabis
sp. and lacewing larvae of the genera Chrysopa sp. have the highest potential to consume
A. glycines. Predators with nocturnal habits, such as carabids and staphylinids showed
lower impacts. However, these were 24 h trials carried out using Petri dishes and
realized predation rates under ﬁeld conditions may differ.

A more realistic, although time consuming method to assess predation consist of
conducting direct observations of predation events on the ﬁeld. Direct observation is the
most useful technique to establish trophic relationships and determine predation rates
under un-manipulated ﬁeld conditions and has been used for a variety of organisms,
including leaﬂroppers and aphids (Luck et al. 1988, Powel et al. 1996). Van den Berg et
al. (1997) conducted direct observations of predation events on A. glycines by three
predator groups in Indonesia: Harmonia arquata (Fabricius) adults, Harmonia larvae,
and the staphylinid Paederus fuscipes Curtis. In that study, the authors followed
individual predators for 10 minutes, obtaining information on the per capita consumption
for each predator, but they did not assess the impact of the entire predator guild on A.
glycines.

In this study, I present observations assessing the relative importance of different
diurnal predator species attacking A. glycines under un-manipulated ﬁeld conditions. The

speciﬁc objectives of the study were to determine: a) which species attack A. glycines in

123

 

ﬁeld conditions and their relative importance, b) their response to naturally occurring
aphid densities, and c) the potential existence of refuges from predation. I considered
refuges in a broad sense as locations on the soybean plant and sizes of the aphids that

experienced lower predation rates.

Material and methods

The interaction of A. glycines and its natural enemies was observed in a soybean
ﬁeld located at the Entomology Research Farm of Michigan State University, East
Lansing, MI. The ﬁeld was planted with variety Pionner 92B38, in 37.5 cm rows, on
5/5/05, and was maintained using standard agronomic techniques. Observations were
conducted on 7/8, 7/12, 7/14, 7/19 and 7/21/2005, on plants ranging from V4 to R2 stage
(Ritchie et al. 1994). On each observation date, three 1 in2 areas (hereafter patches) were
established. Patches were delimited by placing quadrates made by PVC tubes on the
soil, each of which encompassed two soybean rows. The three patches were observed
simultaneously during 2 h periods with one observer per patch during three time slots per
day: 10 -— 12 AM (total = 30 h), 2 - 4 PM (30 h), and 7 — 9 PM (12 h). In each patch, all
predation events on A. glycines and information on predator species and behavior were
recorded.

Based on preliminary observations, I divided A. glycines predators in two groups.
Transient predators (i.e. coccinellid adults, mirids, nabids), only resided in an observation
patch for a short time and consumed aphids very rapidly, requiring continuous attention.
By contrast, resident predators (i.e. 0. insidiosus, spiders, and lacewing, midge and

coccinellid larvae) usually stayed in the observed patch during the entire observation

124

period and consumed aphids more slowly. By noting where these individuals were
foraging, I was able to periodically check on their activity. In order to balance
observations of predation events by transient and resident predators, I set three rules for
observations. First, priority was assigned to follow the behavior of the predators that
consumed aphids more rapidly and were the most mobile. This included, in order of
priority for observation, coccinellid adults > third and fourth instars coccinellid larvae >
mirids > lacewing larvae > ﬁrst and second instar coccinellid larvae > 0. insidiosus >
syrphid larvae > midge larvae. Second, individual resident predators were relocated and
observed at least every 20 minutes (typically much more fragrnentally), interrupting
observation of transient predators if necessary to ensure observation of predation events
by these predators. Finally, 0) observation of small predators (i.e. 0. insidiosus, ﬁrst and
second instar coccinellid larvae, syrphid larvae, and midge larvae) was focused on the
proximal soybean row, whereas transient predators were followed in both rows. To
facilitate observation of resident predators, at the beginning of each observation period
the location of each resident predator was noted on a chart containing the plants of the
proximal row. Total number of small predators per m2 (N SP / m2) was estimated as: NSP
/ m2 = NSP in focal row / number of plants in focal row x total number of plants / m2, and
the number of predation events due to small predators was similarly adjusted. Due to
logistical reasons, observations during the evening were shorter (84.0 i 17.1 minutes,
mean :1: SD) and thus observations were adjusted for comparison with the morning and
afternoon observations.

The distribution of predation events within plants and among aphid size-classes

was also characterized. Plants were divided in three parts: the upper-third (“top”), the

125

middle-third (“middle”), and the lower-third (“bottom”) and predation events were
classiﬁed according to where they occurred. The size of the aphids eaten was recorded as
large, medium, and small-size apterous, and alate aphids. In addition to predation events,
the residence time (RT), i.e. the time spent by transient predators in the observed patch,
and the consumption time (CT), i.e. the time needed by predator to consume each A.
glycines after they located an aphid, were recorded. Not all predation events and RT
were timed, and thus I restricted statistical comparisons to groups with sufﬁcient data
collected (n 2 9). Predation by coccinellid adults was typically very fast, with individual
aphids consumed in a few seconds. There was a trade off between recording and timing
of predation events by coccinellids, and therefore to facilitate observation of predation
events, I recorded CT of coccinellids only to the half minute. Consequently, CT by
coccinellid adults is undoubtedly overestimated and comparisons with other groups
should be interpreted with caution. Immediately after the two hour periods of
observation, the number of plants per patch and A. glycines present on 4 — 6 random
plants, were recorded. With this information I estimated the total number of aphids per

m2 for each patch.

Statistical analysis

Predator abundance (predators / m2 / 2 h), number of aphids eaten (aphids eaten /
m2/ 2 h) and per capita consumption of aphids (aphids eaten / predator/ 2h, PCCR) were
analyzed with two-way ANOVAs using a factorial design (time of the day x predator
group), and date of observation as a random blocking factor (PROC GLM, SAS Institute
2001). Residence time and consumption time were compared among predator groups

using one-way ANOVAs considering each predation event an independent observation.

126

Similarly, the distribution of predation events among plant portions and on different
aphid size-classes was analyzed with one-way ANOVAs. Data was log-transformed to
meet the assumptions of ANOVA before analysis when necessary. Means were
compared using the Least Square Mean Difference, adjusted by the Tukey method for
multiple comparisons (SAS Institute 2001). The responses of predator abundance,
residence time and predation on aphids to aphid density were investigated by calculating
Pearson’s correlation coefﬁcient using SYSTAT 11 for Windows (SYSTAT Software
Inc. 2004). The distribution of predation events by different predator groups on different
plant parts and on different aphid size-classes were compared calculating the Pearson’s )8

statistics for two-way tables (SYSTAT Software Inc. 2004).

Results

A total of 643 predators in 12 groups that attacked A. glycines were observed
during the 72 h of observations (Table 6.1). In addition, I found two unidentiﬁed 3rd
instar coccinellid larvae eating 1 aphid each, and adults of Hippodamia convergens
Guérin-Méneville (n = 1) and Cycloneda munda (Say) (n = 1), syrphid larvae (n = 33),
small carabid beetles (n = 11) and small staphyllinids (n = 3), but I did not observe
predation by these taxa and therefore they were excluded from statistical analyses. Two
predator species not previously reported feeding on A. glycines were observed:
Plagiognathus spp. Fieber, and Chlamydatus associatus (Uhler) (Hemiptera: Miridae).
Very little about the biology of these two species is known. The ﬁrst genera is broadly

consider phytophagous, with no previous references of predation, whereas the second

127

 

   

  

   

 
 

  
  

  

 

 

    

 

a 9m H 0.: _ dd_ md _ 3 end H dmd _Nd H mmd 2d H dvd 2.532
t- .1 ed 5 N m ddd H odd Ed H nod Ed H end $3.383 U
a cd H o._m _ d.m H 9: ed E N 2 odd H odd 2d H Ed 2d H Rd .QO uzﬁezmﬁmsm
-1 _ 9: ed 5 N we _wd H mm." Sd H cod Nod H ZN madam
a d._ H _.m _ md d; E m mm id H am; :d H no; _md H dmd 333. 33:er U
an Wm H of m dd H md d.m E 0 mm :d H :d wmd H 8.. 3d H cm; 3.33. Sewers» E
.1 o a dd H ed Md 8: E 2 Odd H ddd mmd H end wmd H de 8535383
E V dd H mi 30 E 2 _N wmd H wmd 2d H 2d 3d H 3; onﬁﬂomEZ
E Q 2. 2 H EN dd 3: 3 _m wmd H mud vmd H mmd 2d H dvd 3:8 meats
ca dd H of em a _d H dd dd :2 mm 2 2d H mmd 2d H ovd .Sd H Rd .< Seenzmsmﬁsa U
.1 2 a wd H Nd ~42 3: ow :: mm; H mad cm; H nod dwd H w~.~ ~28 _ oats—3.05800
--- Nv .3 md H e; mi 7.3 3 F 2d H Rd omd H mm; end H pd.— vcsm meets
a o; H d.: mm a _d H wd wd W3 em _2 mod H 3: wmd H cod mwd H mmé $3.2. 33x58 E
E 2 a n; H dd _.m_ a: 3 2m cod H San do; H 36 S; H wed 3952
E mm a v._ H ed ndm :2 no Em m_.~ H 33 mm; H odd 3”.— H hmd 3:64.. «38.8me U
*.05? 8:02QO : Lb .x. No... 30,—. wcﬁo>m 8083.2 wear—22 _uoﬁeoum

 

ANS \ £9323 02895.2 mm H :82

 

.cawEosz 5 20m savage. a 5 £02.89
a N marge NE com @35me .08: 8528“ was 08: cozaﬁaaoo ADC 883:8 35% ~88 decades? “0335 Ad 03:.

128

 

was 5523 .80 v a massage 0:85:00 2865 000:2 E0006 m802:2 .

.0050 038.8 M : A308

VEHDSEA .mdd v mv m00¢000t€ ~585qu 0:00:05 000:0— EBMEE ”08305 \ 2:90 \ 00558 mm H 508 ”083 5:983:00 "BU m
.3520 8% $5508 003 mam—mom 0.8.39 32030 838:: 3800 03 30x85 55:» 009852 .5. \ NE \ :83 023% :38 H OH N
028% 8 005802 8: 003 £023 £02030 0.505 EN 28 0% =m 8 000.02 0023 E=0ﬁoooo .333 n .< _

3.283 ud 03:.

129

species is also reported as phytophagous, but have been documented feeding on pollen
and eggs as well (Schuh 2001, Schuh and Schwartz 2005).

Time of the day did not affect signiﬁcantly the number of predators observed (F =
1.85; df = 2, 66; P = 0.1648), aphids eaten (F = 1.20; df= 2, 66; P = 0.3089) and per
capita consumption rate (PCCR) (F = 1.18; df = 2, 39; P = 0.3191), although in general
there were lower values for these parameters during the evening observations (data not
shown). The number of A. glycines per plant tended to increase with sampling date
(Figure 6.1), but there were no signiﬁcant effects of date on predator abundance (F =
0.70; df = 2, 66; P = 0.5923), number of aphids eaten (F = 0.25; df = 2, 66; P = 0.9111),
and PCCR (F = 1.54; df = 2, 39; P = 0.3098). Therefore, I present results averaged

across observation dates and observation times (Figure 6.2).

 

 

600 1 1 16,000
1 1_ .~ :A
50011 f 14,000 03‘
’S ._ 1 12,000 3
a 1 1 3
\ 400 1. 10,000 13
a l 1 \
i: 300 1 “ 1. 8,000 51»
N 1 1
0, ' _ '-
ws 200 1 1 6’0“)
1 4,000
100 1 1 2,000

7/8 7/10 7/ 12 7/ 14 7/ 16 7/ 18 7/20
Sampling dates

Figure 6.1. A. glycines abundance in the soybean ﬁeld observed for predation,
Entomology Farm, Michigan State University, during 2005.

130

The most abundant predator observed was the minute pirate bug, 0. insidiosus
(49.4% of total predators), which was present during all observation periods and was
approximately three-fold more abundant than the next more abundant species (H.
axyridis, 14.9%) (F = 38.95; df = 6, 66; P < 0.0001; Fig 6.2a). Accordingly, signiﬁcantly
more predation events were observed by 0. insidiosus (35.9% of total predation events)
than other predators with the exception of H. axyridis (23.6%) (F = 7.67; df = 6, 66; P <
0.0001; Figure 6.2b). On a per capita basis, however, the most effective predators were
the coccinellids C. septempunctata and H. axyridis, followed by ﬁrst and second instar
coccinellids (F = 3.96; df = 4, 39; P = 0.0086; Figure 6.2c). Other coccinellids and 0.
insidiosus were the least effective, eating between half to a ﬁfth of the number of aphids
than the most voracious coccinellids (Figure 6.2c). The low frequency of predation
events observed for midge and lacewing larvae (n = 6, for each one) prevented statistical
analysis.

The time required to consume A. glycines varied signiﬁcantly among predators (F
= 39.16; df = 7, 153; P < 0.0001; Table 6.1). Both adult and nymphal 0. insidiosus
required on average almost 10 minutes to consume an individual A. glycines, and a
similar result was obtained for other hemipterans and midge larvae. For coccinellids,
adults of H. axyridis, C. septempunctata, and Hippodamia variegata (Goeze) consumed
aphids in less than 1 min (mean between 0.5 — 0.8 min), while large larvae (third and
fourth instars) required 1.7 to 2.7 min, and small larvae (ﬁrst to second instars) 4.2 min
(Table 6.1). Lacewing larvae required similar consumption times as coccinellid larvae
(Table 6.1). This result suggests there was a higher chance of missing predation events

from coccinellids (particularly adults), than by 0. insidiosus and other resident predators.

131

 

20 1 a) Predator abundance

 

 

 

 

 

 

161 1
S 12 E 1
Na 1 ,
E -1
9 8 i b
'3 r be cd
Q, 4 I. _.. 1.. d
.l. e e e
0 . [If] :i: can

 

 

 

 

 

 

 

 

 

 

4 .1 b) Total consumption of A. glycines

a

31 .T.

’ .1. ab
2 1 I 1 b be
— 11 1
I . .7 1 .1. c c C
0:. w ._ * [if] Eh [1:1-

 

 

l

 

 

Aphids eaten / m7- / 2 11

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-= f
N . a
it
e 1
g 1 ab ab .1L T
-8 4 b T- b I ‘1 I
.‘9 $9 0‘ 0 e
O§ ‘06}. 0°C? . $12? @452 - be? 4.19%
$6 cf? a"; ‘9 “a“?
O Q 0. V

Figure 6.2. Mean :t SE, a) predator abundance, and b) total, and e) per capita rate of
consumption of A. glycines by the most abundant predator groups observed during 2 h
periods in a soybean ﬁeld in Michigan, USA. H. axyridis and C. septempunctata
combine adults and 3rd and 4‘h instar larvae, Coccinellid larvae includes all the 1st and 2nd
instar larvae, since they were not identify to species. Different letters indicate signiﬁcant
difference between predator groups (LSMD-TK, P < 0.05). Per capita rates of
consumption of rrridge and lacewing larvae were not included in the statistical
comparisons due to their low ﬁ'equency (15% of cases).

132

However, since the sampling scheme was accommodated to follow more closely
coccinellids, and high density clusters of aphids were not observed, I feel this potential
bias was minimized. The residence time of transient predators in the patch also differed
signiﬁcantly (F = 6.35; df = 5, 213; P < 0.0001; Table 6.1). While abundant,
Coleomegilla maculata De Geer (Coleoptera: Coccinellidae) spent signiﬁcantly less time
in each plot (mean = 5.1 min) compared to other coccinellids, nabids and Plagiognathus
spp., and had a very low consumption of A. glycines (Table 6.1).

Observations were conducted over a range of naturally occurring aphid densities
from 1,290 to 24,678 aphids / m2 (48 — 784 aphids / plant; Figure 6.1), and thus I
explored the response of predators to aphid density within that range. I did not observe a
signiﬁcant correlation between the abundance of all predators combined and aphid
density (r = 0.001, P = 0.996). However, when predators were classiﬁed according to
their vagility, there was a signiﬁcant correlation between transient predator abundance
and aphid density (r = 0.479, P = 0.002), whereas a marginally signiﬁcant but opposite
trend in the response of resident predators to aphid density was observed (r = -0.281, P =
0.083; Figure 6.3a). When predator groups were considered separately there were
signiﬁcant positive associations between aphid density and the abundance of H. axyiridis
adults, and negative associations of immature C. septempunctata and 0. insidiosus
(Figure 6.4).

By contrast, there was a signiﬁcant positive correlation of aphid density and
predation by all predators combined (r = 0.338, P = 0.036), although this response was
largely due to predation by transient predators (r = 0.514, P = 0.001), since resident

predators did not show a signiﬁcant correlation with aphid abundance (r = - 0.144, P =

133

Predators / m2 / 2 h

Aphids eaten / rn2 / 2 h

PCCR/ m2/2 h

50 1 a) Predator abundance
° —0 Resident

. . - ‘ " ‘0 Transient

 

1 o
0'8 l r = 0.394 1"

  

0.4 0 ...........

0.21 D r=0.087 ns

0.0 4mm a a memoir.» 1 . - , lav-“o i -1 F
1,000 6,000 1 1,000 16,000 21,000 26,000

Aphids / m2

Figure 6.3. Responses of resident (closed symbols) and transient predators (open
symbols) to aphid density: a) predator abundance, b) total consumption rate, and c) per
capita consumption rate. Resident predators included 0. insidiosus, midge, lacewing, and
coccinellid larvae. Transient predators included coccinellids adults, Nabis spp.,
Plagiognathus spp., and C. associatus. Lines indicate linear trends, r is Pearson’s
correlation coefﬁcient for the closest line, * indicates P < 0.10, ** indicates P < 0.05, ***
indicates P < 0.01, ns = not signiﬁcant.

134

1.4 ~ a)H. axyridis adults
12: r=0.568***

1.0 1 . .
0.8 f
0.6 ‘
0.4 .
0.2

0.0

 

0.7 1 b) C. septempunctata larvae

0.641 . °’ r=-0.521***
0.5 . .

0.4 1
0.3 1
0.2
0.11
0

Log (predators / m2 +1)

 

1.6 --1 c) 0. insidiosus nymphs
1.4 , =-0.306*

1.2 «E
1.0 .
0.8 i
0.6 4
0.4 5
0.2

0,0 o 0 ~ 0 ,_.__-,,,1 o. o

3 3.5 4 4.5

 

Log (A. glycines / m2 +1)

Figure 6.4. Correlations between aphid abundance and the abundance of a) H. axyridis
adults, b) C. septempunctata larvae, and c) 0. insidiosus nymphs. Other predator groups
did not show signiﬁcant associations. Lines indicate linear trends, r is Pearson’s
correlation coefﬁcient, * indicates P < 0.10, ** indicates P < 0.05, *** indicates P < 0.01,

ns = not signiﬁcant.

135

 

0.383; Figure 6.3b). Separate analysis for each predator group revealed signiﬁcant
positive correlations with aphid density for adults of H. axyridis and C. septempunctata,
and negative correlations for C. septempunctata larvae, with the rest of the predator
groups not showing signiﬁcant responses (Figure 6.5). There was a signiﬁcant
correlation between aphid abundance and PCCR (r = 0.310; P = 0.055), but again this
positive correlation was due to the response of transient predators (r = 0.3 94 P = 0.017),
since resident predators did not respond signiﬁcantly to aphid increase (r = 0.087; P =
0.602; Figure 6.3c). At the species level, only C. septempunctata adults showed a
marginally signiﬁcantly positive correlation between PCCR and aphid abundance (r =
0.625, P = 0.054; Figure 6.6). Lastly, I assessed the potential relationship between
residence time in the patch and aphid density for the adults of ﬁve transient predator
groups in which there was at least 10 observations (C. septempunctata [n =12], H.
aiyirids [n = 95], Plagiognathus spp. [n =10], C. maculata [n =72], and H. variegata [n
=23]). Only C. septempunctata adults exhibited a positive response to aphid abundance
(Figure 6.7). None of the other predator groups showed a change in their residence time
in the patch associated with aphid density.

Finally, I compared predation impacts on different parts of the plant and on
different aphid stages. Signiﬁcantly more aphids were killed in the upper-third of the
plants than in the bottom-third when predation by all predators was combined (F = 8.38,
df = 2, 114; P = 0.0004; Table 6.2). When considered separately, different predator
groups varied in their impact on aphids at different parts of the plant (Pearson’s
correlation f = 54.9, df= 18, P < 0.001), with H. axyridis larvae and 0. insidiosus

nymphs having proportionally higher predation rates in both the upper and lower thirds of

136

A888 Vince/Hm: .86 v my momma 2:8 waoEu 80:88.86 88:8: 888

88.832 EQBEE 880:? .858 :33 macaw 8:86 8:88: me 8:88 0:8 5 8880*:6 E8536
88:2: 888 _838 88th .85888 8888: =« x: : N \ NE \ =88 88:8 :0 8:88 me 882 m
.20: E 888:8:

8m o\om 8:8 8:8: mow—8880a .moﬁm boo: 2:8 888%: 8.: 838:0 888 8:88: go «880: N

.22: E @8865
8a fan 55 8:3: 888880: .8288: EB: 8:88.86 05 so 888:9 885 80:88; me 8880: _

 

8o .. 3.0.-

 

 

 

ems. N3 .1 $5 (Nmo , , N3. -, . . ‘- - - .- mm ..
8 a: a ”2.. 2 _S a £3 < N3 m< a? m is mass
0 --- 9m We ed 332 Sewer? .&
S f f 3 no _.N S o.o 8&2 8583
e _.N no no 3 no 5. 9o 3:2 822
2 Z Z _.N 2 S. 2 3 43383
_N _.N 3 f 2 3: I 2. 333. 3325883. U
2 we 3 _.N 3 3. mm 2 N 25 8 8&2 2:888
3 ed ed 2. 3 we 2: am 4 Ba m 22:3
E Z 8. _.N 3. a.” _.N E 333. 8:88 E
2 mm mm .2 no «em I 2 £832
2 NM 3 an .3 e: 2 3 £33.. 3338 .e
3o: :25 5:602 owe). an? 8.8. 22:: 69:5
N8 0% 2.3. _g 5:83 :35

 

.38 waist 53:82 E 20: S838 a E 885 coca—85 :N we 8888.830 886 x: 8888
.mommm 2:8 85 8.8a 28E E80b€ no 8888: EobbE x: 82.65% .Vmo 80:88: 8880.: .N6 «Bah.

137

1,0 1 a) H. axyridis adults

0.8 -:

0.6 1

0.4 ;

0.2 '

0.0

1.4 ‘

0.6

Log (aphids eaten / m2 +1)

1.2
1.0 -
0.8 ‘

0.4 i

l

0.2
0.0 : ~

0.7 .

0.5
0.4
0.3
0.2

l
l

l

r = 0.346 **

 

b) C. septempunctata adults
r = 0.521 "*

O /
Mo 00 M a ~

0) C. septempunctata larvae
r = - 0519*"

O O.

 

~ -4

3.5 4 4.5
Log (A. glycines / m2 +1)

Figure 6.5. Correlations between aphid abundance and total aphids eaten by a) H.
axyridis adults, b) C. septempunctata adults, and c) C. septempunctata larvae. Other
predator groups did not show signiﬁcant associations. Lines indicate linear trends, r is
Pearson’s correlation coefﬁcient, * indicates P < 0.10, ** indicates P < 0.05, ***
indicates P < 0.01.

138

the plants, whereas the rest of the predators concentrate their attacks on the upper and
middle portions of the plants (Table 6.2). Predation also differed signiﬁcantly among
different aphid sizes (F = 6.68, df = 3, 152; P = 0.0003; Table 6.2), with alates being
preyed less on average than apterous adults and nymphs. The response of the different
predator groups to the different aphid size-classes also differed signiﬁcantly (Pearson’s
correlation )(2 = 43.1, df = 24, P = 0.01), with H. axyridis adults consuming
proportionally more alates and their larvae proportionally more small aphids than the rest
of the predators, that concentrate their attacks on medium and large size aphids (Table
6.2)
Discussion

Coccinellids were the most important predators of A. glycines in this study.
Altogether, H. axyridis, C. septempunctata, ﬁrst and second unidentiﬁed coccinellid
larvae, and adult coccinellids of other species, accounted for 37.4% of all predators, and
53.7% of predation events on A. glycines. In absolute terms, H. axyridis eat
proportionally more aphids (23.6%), than C. septempunctata (13.0%), but the former was
more abundant, and thus on a per capita basis there was a trend of more aphids eaten by
C. septempunctata (Figure 6.2c). Therefore, these results conﬁrm previous reports of
ﬁeld (Fox et al. 2004, Fox et al. 2005, Costamagna and Landis 2006, Costamagna et al.
unpublished results) and laboratory experiments (Rutledge et a1. 2004), that suggested
these two species as the most important predators of A. glycines in North America. The
second most abundant coccinellid, C. maculata, was very mobile, spending between a

half to a third of the time per patch in comparison with other coccinellids, and its impact

139

Log (aphids eaten / predator / m2 / 2h +1)

1.4 "
L r=0.625*
1.2 * o

1.0 :
0.6 5

0.2 3

0.0 l _-. ,
3 3.5 4 4.5

 

Log (A. glycines / m2 +1)

Figure 6.6. Correlation between aphid abundance and per capita consumption of C.
septempunctata adults. Other predator groups did not show signiﬁcant associations.
Line indicates the linear trend, r is Pearson’s correlation coefﬁcient, * indicates P < 0.10.

Log (min / m2 +1)

2.5 i

r = 0.682 **
2.0 ‘. O
1.5 3

1.0 ‘

 

0.5 T

0.0 -
3 3.5 4 4.5
Log (A. glycines / m2 +1)

Figure 6.7. Signiﬁcant correlations between aphid abundance and residence time of C.
septempunctata adults. Other predator groups did not show signiﬁcant associations. Line
indicates the linear trend, r is Pearson’s correlation coefﬁcient, ** indicates P < 0.05.

on A. glycines was minor in absolute terms (1.4% of predation events) and on a per

capita basis (Figure 6.2c, combined with other coccinellids). This result is in agreement

140

with previous ﬁndings of low predation by C. maculata on A. glycines obtained conﬁning
adults in Petri dishes (Rutledge et a1. 2004).

The role of 0. insidiosus in A. glycines suppression is less clear. Manipulative
laboratory experiments and ﬁeld studies correlating the abundance of 0. insidiosus and A.
glycines suggest a potentially important role of this predator suppressing aphid increase
(Rutledge et a1. 2004, Rutledge and O'Neil 2005). However, ﬁeld manipulations using
cages that differentially excluded coccinellids but not 0. insidiosus resulted in
exponential increase of A. glycines to similar levels as populations in which all predators
were excluded, indicating that aphid suppression observed in open plots and sham cages
were mainly due to predation by coccinellids (Costamagna et al., unpublished results).
The difﬁculty of completely excluding 0. insidiosus in ﬁeld experiments makes difﬁcult
to establish the separate contribution of this species to A. glycines suppression, and
whether it has additive effects with coccinellids (Costamagna et al., unpublished results).
In the present study there was only a marginally signiﬁcant correlation between the
abundance of A. glycines and 0. insidiosus, and no correlation between A. glycines
abundance and total and per capita aphid consumption by 0. insidiosus, suggesting a
limited role for this species in controlling A. glycines. However, my studies were
conducted at a relatively high density of A. glycines per plant, and I can not rule out a
more important role of 0. insidiosus at lower aphid densities and on younger and smaller
plants (Rutledge and O'Neil 2005).

I did not detect differences in the mortality of aphids relative to the size of
apterous adults and nymphs, although lower predation was observed on alate aphids.

This could be due to the proportionally lower number of alates observed (mean 1: SD, 1.9

141

 

i 2.2% of total aphids). Selective mortality on larger-size A. glycines was demonstrated
for the parasitoid Aphidius colemani (Hymenoptera: Braconidae) and resulted in stronger
aphid suppression than mortality factors concentrated on smaller aphids (Lin and Ives
2003). In my study there were no signiﬁcant differences in the distribution of predation
events according to the size of nymphs and apterous A. glycines adults. However, I did
not establish the proportion of each aphid size in the population, so I could not assess the
relative impact of predation on each size category in apterous individuals.

Predation was more intense on the upper third of the plant, and diminished in
intensity towards the bottom. Since the within-plant distribution of aphids was not
surveyed, it is not possible to assess whether this was due to higher densities of aphids at
the upper third of the plant or more intense searching of predators in that portion of the
plant. However, in a separate experiment (Chapter 5), I observed proportionally more
aphids on the lower part of plants when predators were present versus when they were
excluded. Together these results suggest that the lower portion of the soybean plant may
constitute a partial refuge from predation for A. glycines.

Resident and transient predators showed different patterns of interactions with A.
glycines, which resulted in different predator impacts. When combined, resident
predators did not change their abundance and total and per capita consumption rates, in
response to increased aphid abundance. However, individual responses of C.
septempunctata larvae and 0. insidiosus nymphs showed a signiﬁcant negative
correlation with aphid density. Since patch immigration/emigration by these resident
predators was negligible, these results indicate that lower aphid populations could be the

result of consumption by these two predators or predator avoidance by the aphids. By

142

 

contrast, transient predators such as H. axyridis and C. septempunctata adults
signiﬁcantly aggregated, stayed longer within the patch, and increased total and per
capita consumption of aphids in function of aphid abundance. Similar responses of
increased abundance and residence time were obtained for adult C. septempunctata and
H. variegata to varying densities of ﬁreweed aphis at different scales (Ives et al. 1993).
Several studies have demonstrated the response of different coccinellid species to aphid
cues that result in the orientation towards aphid clusters (Dixon 2000 and references
therein). In addition, C. septempunctata is attracted to A. glycines-induced volatiles
produced by soybeans (Zhu and Park 2005), supporting the pattern that I observed in the
ﬁeld. These observations suggest that resident predators provide a basal mortality level
that does not react rapidly to changes in aphid abundance, at least in the scales measured
in this study. To the contrary, transient predators mainly consisting of adult coccinellids,
spent relatively little amount of time per patch (on average, less than 18 minutes of the
120 observed), but provide strong aphid suppression (Chapters 1 - 5) coupled with
positive responses to aphid increase.

In summary, my observations indicate that the coccinellids H. axyridis and C.
septempunctata are the major consumers of A. glycines. Moreover, transient predators in
general and adult coccinellids in particular, have the ability to respond to changes in
aphid density by increasing their abundance, and total and per capita feeding. By
contrast, resident predators in general and 0. insidiosus in particular, contribute to A.
glycines mortality due to their abundance, but they do not respond signiﬁcantly to
changes in aphid abundance. Predation is not equally distributed among plant portions

and aphid size classes. This study is the ﬁrst one to provide a direct assessment of the

143

importance of different predator species attacking A. glycines in North America, using
un-manipulated ﬁeld populations. More studies focusing on the interaction of A. glycines
and its predators under ﬁeld conditions, including different times of the season and
observation of nocturnal predators, are needed to completely assess the role of the

different predator species in the regulation of this pest.

144

CHAPTER 7

PREDICTIVE MODEL FOR THE GROWTH OF FIELD POPULATIONS OF

APHIS GLYCINES (HEMIPTERA: APHIDIDAE) IN SOYBEAN

Abstract

The soybean aphid, Aphis glycines (Hemiptera: Aphididae) is an invasive pest that
was detected in North America in 2000, and rapidly became the most important insect
pest of this crop in the American Midwest. Strong suppression by natural enemies occurs
in some years, but A. glycines populations escape natural suppression in other years. The
potential causes of this escape are poorly understood. Here I present a mathematical
model parameterized using data from four ﬁeld experiments in which A. glycines
populations grew in predator exclusion cages in two locations in Michigan, USA,
between 2003 and 2005. These data were ﬁt an exponential grth model in which the
intrinsic rate of increase (r) decreases linearly with time. This model, in contrast to the
traditional exponential and logistic growth models, accurately predicts A. glycines
population dynamics in soybeans, including the population decline towards the end of the
season. The model is robust against variability within and between experiments and a
generalized model combined the data of the four experiments. This generalized model
has better explanatory power (R2 = 0.95) when is ﬁtted on an A. glycines degree-days
scale, suggesting the important role of temperature in aphid population dynamics.
Potential causes of r declining with time are discussed. This model provides a simple and
accurate tool to model A. glycines populations and explore the effects of natural enemies

and different control measures to develop management strategies for this pest.

145

Introduction

The soybean aphid, Aphis glycines Matsumura (Hemiptera: Aphididae) is an
invasive pest originally from Asia that was observed for the ﬁrst time in North America
in 2000 (Ragsdale et al. 2004, Venette and Ragsdale 2004). Since its detection, A.
glycines has expanded its distribution throughout the soybean producing region in North
America, including both the United States and Canada, and became the most important
insect pest of this crop (Venette and Ragsdale 2004, Mignault et al. 2006). Yield losses
due to A. glycines ranging from 50 - 70% have been reported in Asia (Wu et al. 2004)
and up to 40% in USA (DiFonzo and Hines 2002). Yield reduction is mostly caused by
direct feeding of outbreak populations, but even small populations may cause yield losses
due to reducing the photosynthetic rates of plants (Macedo et al. 2003). Indirect damage
due to A. glycines includes facilitating the development of sooty mold and virus
transmission, although severe viral outbreaks have not yet been reported in North
America (Hill et al. 2001, Clark and Perry 2002, Wang and Ghabrial 2002, Wu et al.
2004). Strong impacts of natural enemies on A. glycines have been demonstrated in Asia
(Van den Berg et al. 1997, Liu et al. 2004, Wu et a1. 2004) and in North America (Fox
and Landis 2003, Fox et al. 2004, Fox et a1. 2005, Nielsen and Hajek 2005, Costamagna
and Landis 2006). However, despite substantial suppression by natural enemies, during
some years A. glycines develops outbreak populations, and the causes of this escape
remain unclear (Costamagna and Landis 2006). Thus, accurate mathematical models to
describe A. glycines population dynamics may provide an important tool to manage this

pest and to help elucidate potential causes for population outbreaks.

146

Besides predation, aphid population dynamics are inﬂuenced by several other
factors under ﬁeld conditions, including temperature, and host plant quality (Dixon 1987,
1998). Temperature effects on aphids are well documented affecting growth,
developmental and respiration rates, and honeydew production (Dixon 1998). At the
population level, these effects result in changes in the intrinsic rate of increase that is
expressed (Dixon 1998). In general, aphids increase their individual growth rates from a
basal temperature until an optimum and then growth rates decline until a maximum
tolerable temperature (Dixon 1998). In the case of A. glycines, temperature effects have
been characterized under the range of summer temperatures, with a basal temperature of
8.6 °C, optimum around 20 - 25 °C, and a maximum threshold of 34.9 °C (McComack et
al. 2004).

Several bottom-up factors inﬂuencing host plant quality have been shown to
affect aphids populations, including fertilization (Cisneros and Godfrey 2001, Morales et
al. 2001, Nevo and Coll 2001), tillage regime (Andersen 2003, Hesler and Berg 2003,
Gencsoylu and Yalcin 2004), and cover crops (Tillman et al. 2004). Manipulating
potential bottom-up factors resulting from different regimes of tillage, fertilization and
weed control, did not affect population growth of A. glycines in ﬁeld experiments
conducted in Michigan (Costamagna and Landis 2006), however, deﬁciencies in soil
potassium have been shown to increase A. glycines populations in Wisconsin (Myers et
al. 2005b). Finally, plant phenology has been shown as an important factor affecting
aphid population growth (Rossing et a1. 1994, Williams et al. 1999). In the A. glycines

system, Van den Berg et a1. (1997) demonstrated a linear decrease of the intrinsic rate of

147

increase on aphids reared at three different soybean phenological stages under ﬁeld
conditions in Indonesia.

The objective of this work was to develop an analytical model to accurately
describe A. glycines population growth under ﬁeld conditions in the absence of predation.
The ultimate goal of this study is to develop a mathematical model that may serve as a
tool to explore the consequences of different levels of predation, control measures, use of
resistant varieties and abiotic factors in A. glycines management. I utilized data from four
ﬁeld experiments in which A. glycines was reared in predator exclusion cages to
determine parameters of population growth. Previous evidence suggest that the intrinsic
rate of increase (r) declines with time due to host plant quality on other aphid species
(Rossing et al. 1994, Williams et al. 1999) as well as on A. glycines (Van den Berg et a1.
1997). Thus, I incorporated a linearly decreasing r into an exponential growth model
(further referred to as the “decreasing r model”) and compared this model with the
traditional exponential and logistic growth models, to test which one best describes the
ﬁeld data. Then data from all experiments were combined to create a more general
predictive model and its accuracy was further improved by ﬁtting two additional models
accounting for temperature effects (aphid degree-days since initial infestation) and plant

phenology effects (soybean degree-days since sowing).

Materials and methods

Study sites and ﬁeld experiments
I utilized data obtained in four ﬁeld experiments in which 48 A. glycines

populations were reared in predator exclusion cages. All ﬁeld experiments were

148

conducted at one of two locations in Michigan (USA): the Entomology Farm or the
biodiversity study site of the Kellogg Biological Station Long Term Ecological Research
site in row crop agriculture (KBS-LTER) of Michigan State University, between 2003
and 2005. The details of the experimental designs are summarized in Table 7.1, and are
ﬁrlly described in the previous Chapters. The experiments were conducted by caging
aphids naturally infesting plants, or by placing controlled numbers of aphids reared in the
laboratory on soybean plants using a camel-hair brush (Table 7.1). Planting date varied
among experiments, resulting in different plant phenology at equivalent dates. All aphid
inoculations were performed between the V2 — V4 stage of plant phenology (Ritchie et
al. 1994).

Two cage designs were utilized, sleeve cages that enclosed aphids on one to three
plants, and larger cages that enclosed a l m2 area (29.1 i 2.0 plants / m2, mean at SD).
Sleeve cages consisted of a cylindrical frame of wire (0.4 m of diameter by 1.0 m of
height), that was completely covered by ﬁne-mesh white no-see-um netting (Kaplan
Simon Co., Braintree, MA) and that was buried 15 cm in the soil. The larger ﬁeld cages
consisted of a PVC frame (1 x 1 x 1 m) covered by the same material on the upper
portion and with a basal plastic barrier (10 cm buried in the soil, 20 cm above soil
surface), connected to the netting by Velcro. Two bands of mesh with coarse openings (2
mm) were included around the sides of the larger cages, one located at the top (10 cm
wide) and the other at the canopy level (30 cm wide). These bands of coarse mesh were
required to allow alate aphids to escape, but also allowed some small predators to enter
(primarily Orius insidiosus (Say) [Heteroptera: Anthocoridae]). My previous

experiments have shown a very limited impact of 0. insidiosus on A. glycines population

149

$83388 29:2: com @052: noggioxsh 2.: 3 363.8“ cocobwtm £802 896m 384 .86 v a:

358530 muons $32056 ESmeB 286E 888— Eocommc douﬂmomcm cows 99% mm “we 2: 8m EBQBQEB mm um :82 v
.283 \ 8:.»ch «do 59832 n

.<mD 42 56.53:: 88m Swwﬁoaz 8 “833032“ no.8

>28 5 BB 588.3% :3waon 8.5% mac; I :ouﬁm Rama—2m wwo=ov~ ”Mmkq-mmv_ c5 .55“ zwo_o80~:m ”gm .BEoEm— N
$23 wE—aﬁwm 22 BE 2: com momma c 9 @0262 ~ .98 E .6085on En momma mo 58:: Z _

 

a mam. as §-9 ,. m w 803- :28: RM? rumirmmw .3 98W. -38 w. ,,..-
n no a 0.8 ~> m 33 €28 $5: $5-89 2 252m 38 m
0 ed a 3: 9 am 2 83 .228 in: as» .3835 0 Na: 38 N
a no a «.2 «>. 9 3 32 22a mm a: 85 .3895 M: 052m 88 _

 

«.9881 502 $2823 ENE maﬁa 2mm 880m Bea magma accuses _: ammo 80> «5853mm

 

cote—sooﬁ 83ch .V

 

.Goom mcwaaﬁmoo 98 5:23. $8508 05 com 258598 :08 mo countomou 3:82. a com .8on 55:88
883E E 32:28 Eva 823 888:? magi—mom mmﬁcbm .V @556 3.8.5.830 .58 no 85388320 AS 03:.

150

growth in these cages, however, my results should be consider a conservative measure of
aphid increase. Sampling was performed weekly by counting the total number of aphids
on all the enclosed plants in the sleeve cages, and on 4 to 10 random plants, diminishing
sampling size as the population increased, on the larger cages (Costamagna and Landis

2006). All counts are expressed as number of A. glycines per plant.

Models of A. glycines growth

The A. glycines population growth curves obtained in the four experiments were
ﬁtted separately to exponential, logistic and a decreasing r models of population growth.
Exponential growth was modeled using the equation:

N, = Noe” (Eq. 1)
where t is time (days), N, is the aphid population size at time t (aphids / plant), No is the

initial aphid population (aphids / plant), and r is the intrinsic rate of increase (aphids /

 

aphid / day).
Logistic growth was modeled using:
K
N t =
K — NO _,, (Eq. 2)
1 + —— e
N 0

where K is the carrying capacity (aphids / plant). Finally, A. glycines growth was

modeled assuming a linear decrease of r using the equation presented in Williams et

al.(1999):
rd
N, = NoeI t (Eq. 3)
with
1
Jirdt=rmaxt[l—§at] (1314)

151

By substitution of Eq. (4) in Eq. (3) we obtain:
rmaxt[l-§at )

NtzNO-e (Eq.5)
where rm is the maximum intrinsic rate of increase (aphids / aphid /day), and a is the
reciprocal of the time to peak population (1 / days) . Thus, this model has an r that
decreases in time, starting at a maximum rmax when t = O and decreasing linearly until r =
0, when t = 1 / a and the population is at its peak, and after that the population declines
(Williams et a1. 1999). Finally, I calculated the intrinsic rate of increase r between
sampling intervals using:

= 1n(Nt+At)_ ln(Nt—At)
2At “39 6)

 

r

Statistical analysis

Fitting of the three models was performed by nonlinear regression on log-
transformed data and parameter values 2%: SE and R2 are reported for each model (Proc
NLIN, SAS Institute 2001). Potential within-experiment variability that may result in
discrepancies with the linear decreasing r model, was assessed by ﬁtting models for the
individual aphid populations followed in each experiment (i.e. considering each cage a
separate aphid population). Differences in mean temperature among experiments were
assessed for the ﬁrst 28 days after infestation, which was the duration of the shortest
experiment, using ANOVA with day as a blocking factor (Proc MIXED, SAS Institute
2001). I ﬁt the decreasing r model to the combined data of all experiments, and tested
whether inclusion of aphid physiological or plant phenological scales resulted in a higher
R2. I expect that when population dynamics of A. glycines are sensitive to changes in

temperature, a better ﬁt will be obtained using a degree-day scale based on aphid growth

152

(basal temperature = 8.6 °C, McCornack et al. 2004) rather than time since inoculation
(calendar time). Similarly, if plant phenology were the most important factor driving the
population dynamics of A. glycines, the model will provide the best ﬁt when scaled on a
scale of degree-days since sowing, based on soybean phenology (basal temperature = 10
°C, Zhang et al. 2001). In addition, linear regressions of the intrinsic rate of increase r
with calendar time, aphid degree-days and soybean degree-days were performed to check

which scale provides the best ﬁt (Proc REG, SAS Institute 2001).

Results

The decreasing r model was the only mode] that accurately represented population
dynamics of A. glycines in all four experiments (R2 > 0.94; Figure 7.1, Table 7.2).
Exponential and logistic models produced very good ﬁts for experiments 1 and 2 (R2 >
0.98, Fig 7.1a and b, Table 7.2), which were conducted before aphids reached peak
populations, but there was a clear lack of ﬁt for experiments 3 and 4 when the number of
aphids decreased after having reached their peak (R2 = 0.42 — 0.90, Figure 7.1a and b,
Table 7 .2). In experiments 3 and 4, exponential and logistic models underestimated r to
account for the latter decrease in aphid population, resulting in low to intermediate r
values in comparison with values of rm, estimated by the decreasing r model (Table 7.2).
Thus, the decreasing r model provides a complete description of A . glycines population
dynamics on soybeans during the whole season.

To test the robustness of the decreasing r model against variability within
experiments, I ﬁtted separate models for each aphid population within each of the four

experiments. All the models (n = 36 aphid populations) were signiﬁcant (P < 0.01) and

153

 

 

 

 

53 v 1-50.5 v. .- 33w 8% .- m ..
$3 $3 $3 8.3 mm
Sod a 82 89¢ a Mano Sod a ammo ~de « 33 a
$3 a $3 $3 a 23 an; an S3 33o a $8 22v 3
Sod « 82 Sod a mono Sod a Name $3 a and sea
xmﬁaumth
:56 v :56 v :56 v Sod m
ES 82 52 83 mm
33 a 33 8.8 a 8% £3 a $3. mm...“ a 3:... 032
8.3 a 23 ﬁne a OS: :3 a :2 2.3 a was 2% we.
wed in $3 $05 a ES 23 a £3 £3 a and a
BEMQN
89o Rod :55 v 33 a
£46 33 £3 :35 Na
33. a 8M: 83 a 3: 83 a zoo $8 a 33 35 ms
owed a 23 Rod um god Sod a 2g 33 a «.86 a
NawEgomﬁw
v m N _
mucoﬁtomxm

 

 

€288 $5508 05 com £08883 2:
:0 £83 cob <mD .5353: E $585on Eva Bow 5 82.83% .33 305.25% :oaﬁaom
508 8 2258 538w noun—38 085 maﬁa E 8:850 Amm “m 525 8225.53 .NK 03.;

154

 

5 f 3) Exp. 1 . Observations
Exponential

4 ~§ ------ Logistic

‘ Decreasingr .

   

C
.
a".
I

   

Log (A. glycines / plant)

 

O 10 20 3O 40
Days after infestation

Figure 7.1. Observed population dynamics of A. glycines in four ﬁeld experiments
using predator exclusion cages and the ﬁt of three competing models, for experiments 1
(a), 2 (b), 3 (c) and 4 ((1). Observations (mean :t SE) are indicated by points and the
exponential, logistic, and decreasing r models by dashed, dotter and solid lines,
respectively. Parameter values and statistics for these models are given in Table 7.2.

155

Log (A. glycines / plant)

Figure 7.1. (cont’d)

20 3O 40

Days after infestation

156

 

' *W
I

all the values for the parameters rmax, ln(No), and a, had overlapping 95% conﬁdence
intervals within each experiment. The R2 values indicate a very good ﬁt of individual
aphid populations to the model (exp. 1: all R2 > 0.99 [n = 6]; exp. 2: all R2 > 0.97 [n = 6];
exp. 3:112 > 0.79 [n = 7] and R2 > 0.90 [n = 5]; and exp. 4: R2 > 0.69 [n = 1], R2 > 0.79 [n
= 5], and R2 > 0.90 [n = 6]). These results show that the decreasing r model accurately
describes the growth of individual A. glycines populations and therefore supports the use
of the means obtained in each experiment to investigate differences among experiments.
A more generalized model of A. glycines population growth was developed by
ﬁtting the model to the data of all four experiments combined. The ﬁt of the model to the
observed data expressed at the scale of days aﬁer infestation resulted in very good ﬁt (R2
= 0.90, Figure 7.2a, Table 7.3). Mean temperatures registered during the ﬁrst 28 days
after infestation differed signiﬁcantly among experiments, with higher mean temperatures
recorded in 2003 and 2005 (F = 17.82; df = 3, 84; P < 0.0001, Table 7.1). Thus, to
incorporate the effect of temperature on the population dynamics of A. glycines, the
model was ﬁt using A. glycines degree-days instead of days aﬁer infestation, obtaining
substantially higher explanatory power on this scale (R2 = 0.95, Figure 7.2b, Table 7.3).
Similarly, using soybean degree-days improved the model explanatory power over using
calendar days, however it was slightly less predictive than the model using A. glycines
degree-days (R2 = 0.943, Figure 7.20, Table 7.3). In all cases, the assumption of a linear
decrease of r in time was met, with all three linear regressions being highly signiﬁcant

(Table 7.3).

157

 

 

 

1530- . w .. 83v? 830 m
vino Rune wand NM
«Seed H .8506- Zoood H caged- ﬂood H 256- _n
ﬂowed H $350 3350 H 0393 N306 H £93 an
x 23.3.3me L885
Sod v Sod v Sod v R
mvod mmod nowd NM—
«886 H meOd 3899 H maood good H $86 a
Emcee 0 8.89m- Evie H 83% 885 H 33 A53
mwmood H Smood 0586 H ommmod wowed H wand as:
385 L MawweoxomQ
mxmcoocwou masoaxom mam—Yoocwou 8523‘ .V sensuous: Eta QED

 

 

 

.688 some 8m @8585 03
0:5 5 x me 888% Soc: mo 839553 05 9:38 mains“ _moumuﬁm 05 me 3:52 .8983 05
E .838 0:5 mm waxwoﬁwot 58.38 98 £8-02va umﬁubm .V .co_§mo.«:_ has 998 8 “co—mom
maoaatomno 8 $on L mammmocoow 2: wave 3 3:850 Amm H 525 E20883 .mﬁ «Bah.

158

 

0 . ~ - ~
0 10 20 30 40 50 60 70
Days after infestation

Log (A. glycines / plant)

 

O t d 7 .‘ " " 1 ‘ l' l """1
0 100 200 300 400 500 600 700 800 900
A. glycines degree-days

 

200 3004005006007008009001000

Soybean degree-days

Figure 7.2. Fit of the decreasing r model (solid line) when ﬁtted to observations of the
four ﬁeld experiments combined, scaled to days after infestation (a), A. glycines degree-
days (b), and soybean degree-days (c). Different symbols represent the means of the
four experiments. Parameter values for these models are given in Table 7.3.

159

Discussion

The population dynamics of A. glycines on soybean in the absence of predation
can be well described using a model for exponential growth with the intrinsic rate of
increase r that decreases linearly in time. The model was robust against variability
among four experiments that were conducted in two separate locations during three
different years. In addition, the model provided high explanatory power when ﬁtted
separately to data of the 36 individual aphid populations followed in the four
experiments. Similar models assuming a decrease of r with time were previously used to
simulate aphid population dynamics of Sitobion avenae (F.) (Hemiptera: Aphididae) in
wheat (Rossing et al. 1994, Bianchi and van der Werf 2004) and three aphid species
colonizing sugar beets in England (Williams et al. 1999). In addition, Van den Berg et a1.
(1997) used a similar approach to model A. glycines population growth under ﬁeld
conditions in Indonesia, decreasing the value of r according to estimations obtained
rearing cohorts of aphids on plants at three different phenological stages. These studies
measured the effects of plant quality either on individual aphids (Van den Berg et al.
1997, Williams et al. 1999) or on aphid ﬁeld populations but without excluding predation
(Rossing et a1. 1994). My study presents a similar decreasing r model predicting A.
glycines population dynamics based on data of ﬁeld populations excluding mortality due
to predation. Therefore, these studies show that plant quality affects population dynamics
of a range of aphid species on a variety of crops and thus, can be considered a general
phenomenon.

The generalized model developed for A. glycines population growth is highly

predictive under the range of environmental conditions of the four experiments

160

conducted. The model provided a very good ﬁt when ﬁtted to data expressed at the scale
of days aﬁer infestation. This is most likely caused by the relatively narrow range of
phenological stages of the plants when infested with aphids (V 2 — V4) and temperatures
that generally fell within the optimum range for A. glycines development in all
experiments (McComack et al. 2004). Notably, A. glycines populations in the
experiments where lower mean temperatures occurred (2 and 3), were below the curve
predicted by the model, whereas in experiments 1 and 4 populations were above the
predicted values (Figure 7.2a). However, when the model was ﬁtted using physiological
scales for aphids and soybeans, these differences disappeared, indicating an effect of
temperature on A. glycines population growth (Figs. 7.2b and c). The very similar results
obtained using aphid and soybean degree-days may be due to two factors. First the basal
temperatures for A. glycines and soybeans are very close, resulting in a relatively small
difference in scale (8.6 versus 10 °C, respectively, Zhang et al. 2001, McCornack et al.
2004). Second, and most importantly, my experiments were conducted under a narrow
range of plant phenology; greater effects could be expected on populations reared on
more separated soybean phenological stages (Van den Berg et a1. 1997).

These results supports the use of population growth models with decreasing r for
systems in which the quality of the resources vary signiﬁcantly with time and this
change is directly translated into reduced reproduction. The observed decrease in the
intrinsic rate of increase of A. glycines in time could be due to several factors. First,
plant phenology may have a deleterious effect on the quality of resources for aphids, as
was previously mentioned (Van den Berg et al. 1997). Second, rapidly increasing aphid

populations may deteriorate host plant quality, as documented for several aphid species

161

(Dixon 1998, Williams et al. 1999). Finally, programmed reductions in aphid fecundity
in response to seasonal changes in habitat quality have also been demonstrated (W ellings
et al. 1980). Since my experiments were not designed to distinguish among these
possible mechanisms, I can not rule out any of them affecting A. glycines. In summary, I
developed a mathematical model for the population dynamics of A. glycines in the
absence of predators, which provides season-long accurate and robust predictions. This
model can be used as a simple tool to explore effects of different control strategies and

predation levels on ﬁeld populations of A. glycines.

162

CHAPTER 8

CONCLUSIONS AND FUTURE DIRECTIONS

Aphis glycines invasion in North America

The soybean aphid, Aphis glycines Matsumura (Hemiptera: Aphididae) was
detected in North America for the ﬁrst time in Wisconsin in 2000 (Ragsdale et al. 2004,
Venette and Ragsdale 2004). Since then, it has rapidly expanded to 22 States and three
Canadian provinces, causing yield losses up to 40% in the USA (DiFonzo and Hines
2002, Losey et al. 2002, Hunt et al. 2003, Landis et al. 2003, Venette and Ragsdale 2004,
Myers et al. 2005a, Mignault et al. 2006). During 2003, A. glycines was detected in over
50 million acres in the USA, of which 7 million acres were sprayed with insecticide to
control this aphid (Landis et al. 2003). Thus, A. glycines arrival dramatically changed the
dynamics of the insect communities associated with soybeans, a crop that was very
seldom treated with insecticides before soybean aphid arrival (Heimpel and Shelly 2004).

Several studies were conducted in response to A. glycines invasion, including
ﬁeld studies for development of sampling protocols and threshold levels (Hodgson et a1.
2004, Myers et al. 2005a, Onstad et al. 2005a, Onstad et al. 2005b), experiments
determining optimal abiotic conditions for A. glycines development (McCornack et al.
2004, McCornack et al. 2005), and trips to its original area of distribution exploring for
natural enemies (Heimpel and Shelly 2004). In addition, Fox and Landis (2003), Fox et
a1. (2004, 2005), Rutledge et al. (2004), and Rutledge and O’Neil (2005), assessed the

impact of resident natural enemies on A. glycines suppression. In this dissertation, I

163

investigated several aspects of the interactions between A. glycines and its resident

natural enemies in Michigan.

Top-down versus bottom-up forces in A. glycines population control

The source of herbivore population control has been intensively debated since the
proposal by Hairston et a1. (1960) of a dominant top-down control on natural
communities (Hunter and Price 1992, Power 1992). At present, both forces have been
shown determining herbivore abundances on several ecosystems and their relative
importance within each system is the focus of current research (Hunter et al. 1997, Fraser
and Grime 1998, Dyer and Letoumeau 1999, Fraser and Grime 1999, Walker and Jones
2001, Demo et a1. 2002, Gratton and Denno 2003a). In the soybean aphid system, I
observed strong suppression of A. glycines due to generalist predators both at aphid
establishment and population growth phases, but only a weak, non-signiﬁcant and
variable effect of bottom-up forces (Chapter 1). I explored bottom-up factors by
conducting experiments over a range of agricultural treatments, from conventional to
organic agriculture, present at the Long Term Ecological Research Site at the Kellogg
Biological Station of Michigan State University. These agricultural treatments altered
plant quality as measured by above ground net productivity and yield, but did not
signiﬁcantly affect aphid populations even in the absence of top down controls,
suggesting little scope for the management of A. glycines by manipulating agricultural
practices. Similar results were obtained in experiments conducted on soybeans under
different rotation systems (Chapter 4). However, other sources of bottom-up factors have

been shown to affect A. glycines, including potassium soil deﬁciencies (Myers et al.

164

2005b), soybean resistant varieties (Hill et al. 2004a, b, Li et a1. 2004, Wu et al. 2004,
Mensah et al. 2005), plant phenology (Van den Berg et al. 1997), and alternative
cropping systems (Wu et al. 2004). These should also be investigated to explore the
possibility of combining them with existing sources of top-down control.

Strong top-down control may result in a trophic cascade. Trophic cascades occur
when higher trophic levels signiﬁcantly affect the abundance of lower trophic levels
across more than one link in a food web (Pace et al. 1999, Polis et al. 2000). In
agroecosystems in which pest suppression is exerted by natural enemies, a trophic
cascade of increased plant yield represents a measure of biological control efficacy (Polis
et al. 2000). In three separate experiments over two years I observed trophic cascades, as
evidenced by increased soybean biomass and yield, due to the reduction of A. glycines
populations by ambient levels of generalist predators (Chapters 3 and 4). These ﬁndings
suggest that natural enemies play a prominent role in suppressing aphids during non-

outbreak years and during much of the season in outbreak years.

Aphis glycines natural enemy assemblages and intraguﬂd predation
In a multiple natural enemy guild, trophic interactions among its members may
result in a cumulative effect on the prey, leading to increased (synergistic interactions) or
reduced prey suppression (non-additive interactions) (Polis et al. 1989, Ferguson and
Stiling 1996, Sih et al. 1998). Thus, in addition to attacking the focal prey, generalist
predators may attack other members of the natural enemy guild, thus acting as intraguild
predators (Polis et al. 1989, Polis and Holt 1992). In particular, aphid parasitoids have

been shown to be susceptible to high levels of intraguild predation, since their immature

165

stages develop within the aphids and later murnmify (pupate) in close association with
them (Rosenheim et al. 1995, Brodeur and Rosenheim 2000, Miiller and Brodeur 2002).
The assemblage of A. glycines natural enemies in North America is currently dominated
by generalist predators, and parasitoids have only occasionally been detected in ﬁeld
infestations (Landis et al. 2003, Rutledge et al. 2004). By contrast, parasitism levels up
to 53% have been reported in China, potentially contributing substantially to A. glycines
suppression (Wu et al. 2004).

Thus, A. glycines provides an ideal system to test the potential limitation of
parasitoid effectiveness by generalist predators through intraguild predation. I obtained
evidence of intraguild predation on ambient and augmented levels of North American
parasitoids of A. glycines in two separate ﬁeld experiments (Chapters 3 and 4). Intraguild
predation did not disrupt parasitism in these experiments, since the percentage parasitism
was very low (< 1% in ambient levels of extant parasitoids). However, intraguild
predation may slow down the process of adaptation of local parasitoids to this new host.
Moreover, my results show that intraguild predation may become more important under
higher levels of ﬁeld parasitism. Thus, efforts directed at introducing more effective
parasitoids of A. glycines in North America should consider detrimental effects of
intraguild predation and try to assess whether new natural enemies will provide additive
or synergistic aphid suppression or, in contrast, their effect will be diluted by non-

additive interactions with the resident assemblage of natural enemies.

166

Aphis glycines -— predator interactions

Despite strong predator suppression, A. glycines still increase to outbreak levels
that result in signiﬁcant yield reductions and economic losses. The causes of these
outbreak populations are not well understood but may include a combination of abiotic
and biotic factors. I developed a mathematical model of A. glycines population growth
under ﬁeld conditions as a ﬁrst step in understanding the population dynamics of this pest
(Chapter 7). The model is an exponential growth model in which the intrinsic rate of
increase of A. glycines decreases linearly with time, and provides very good ﬁt to the data
obtained from four ﬁeld experiments where predators were excluded (Chapters 1 to 5). A
decreasing intrinsic rate of increase could be the result of decreasing resource quality due
to plant phenology (Van den Berg et a1. 1997), aphid feeding (Dixon 1998, Williams et
al. 1999) or programmed responses to seasonal changes (Wellings et al. 1980). This
model could be used to determine the effects of different control strategies and mortality
due to natural enemies to successfully manage A. glycines.

Aphis glycines shows a highly aggregated spatial distribution (Hodgson et al.
2004), suggesting that certain areas in the ﬁeld will develop higher aphid populations and
may constitute a relative reﬁrge ﬁom predation due to a dilution effect (Turchin and
Kareiva 1989). By comparing predator protected versus unprotected aphid populations I
determined that predators suppressed aphids below the threshold level irrespective of the
initial aphid density (Chapter 5). Interestingly, I found a trend of lower per capita
increase rates in high density aphid colonies, suggesting self limitation for A. glycines. In
addition, exposure to predation signiﬁcantly affected aphid distribution within the plant,

with proportionally lower aphid densities on the upper nodes of the plant when they were

167

exposed to predators (Chapter 5). This could have signiﬁcant consequences for A.
glycines population growth if the suitability of the soybean plant varies at different nodes
and should be the focus of future research.

A signiﬁcant role of the entire assemblage of predators suppressing A. glycines in
North America was demonstrated in previous studies (Fox 2002, Fox and Landis 2003,
Fox et al. 2004, Fox et al. 2005) and Chapters 1 to 5. However, these studies involved
excluding partial or entire assemblages of predators, thus not allowing determination of
the individual contribution of each predator group. During 2005, I conducted direct ﬁeld
observations and recorded 211 predation events on un-manipulated A. glycines
populations. These results showed that the coccinellids H. axyridis and C.
septempunctata were the most effective predators of A. glycines, consuming more aphids
per capita and responding to changes in aphid density with increased aggregation in space
and time and per capita consumption rate (Chapter 6). Thus, investigating the population
dynamics of these two species across the landscapemay provide insights on the causes of
A. glycines outbreak populations.

In summary, these studies provide strong support for the key role of natural
enemies, in particular generalist predators, in suppressing A. glycines populations across a
range of conditions in Michigan. However, during outbreak years A. glycines is able to
escape control from natural enemies and the causes of this escape remain unclear. In
addition, several aspects of the interaction between A. glycines and its natural enemies are
still poorly understood. The current major gap in our knowledge is regarding the
population dynamics of the main predators of the system, the coccinellids H. axyridis and

C. septempunctata. Since these are vagile predators that develop early season and

168

populations overwinter in areas outside soybean crops, a study of their populations at a
landscape level should provide insights on their interaction with A. glycines. In addition,
their foraging behavior and factors inﬂuencing it should also be investigated. A more
complete understanding of the interactions between A. glycines and its natural enemies
will allow the implementation of better management practices and may represent the key

to control of this pest.

169

APPENDIX

170

Appendix 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.: m

Title of thesis or dissertation (or other research projects):

DO VARYING NATURAL ENEMY ASSEMBLAGES IMPACT APHIS
GLYCINES POPULATION DYNAMICS?

Museum(s) where deposited and abbreviations for table on following sheets:

Entomology Museum, Michigan State University (MSU)

Other Museums:

lnvestigator's Name(s) (typed)
Ale andro m

 

 

Date 05/01/20i5

*Fleference: 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.

171

Appendix 1 .1

Voucher Specimen Data

Page i of _2_ Pages

\

2mm

 

50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

WRWXaR 0.x 8cm 5. 28
So Saw
3225 saw 5.953 a: 5 ﬁance
.6. mccEBcaw 3%.: 90am 9: cozcocm magi—coo 3.30 0.65-o2
8.08de accozo> 609$ vacEmz mLoﬁmzmgs
$838: m 335 _wcoEouw $2
88 mm .5 Em“. 320525 - :ms.
2 323 Emu .8 52?. :2 <8 3mg mamssms 3:0
Bow 5 __> Emu. 32323 - :ms.
0 9:28.. 6mm .60 529.. :2 <w3 .86 on EmSQmE $395300
88 mm .5 .58“. >8_oeecm - :ms.
2 Ease. new :8 59.9.. :2 <9 A889 Smears. Heavens:
meow mu __> .55”. Boasoam - 3w:
0 0523 6mm :00 Ewan... :2 (m: 1. «.mﬁcagmﬁmm $3.530
meow mu __> .55”. >8652cm - :ms.
2 u «.533 am. .8 529.. :2 <3 amino arcs sense:
88 _> .2330 93 - 5.983
om om 0% win? <8: .mo__z .3 <9 93536.2 «36% «En...
m M 10 0+ 8
ue.s r s s e h e 8.588.058?
m m m. m W W. W m. m «.3 .0 362.8 29:8ch .2 Sun .33 :98. .550 no mgooam
h e t d d U y a g
ded_OAAPNLE
no 8952

 

 

 

172

 

 

 

 

 

 

2ND .99 :0
885m 2.3

.anm22 >mo_oEoEm

2.99.5 9% 896.5. 9.. c. .688
.2 9.950on beam: 93m 9: 8289”. acne—c.3000 3.30 223.02
8-88 .02 .88.; .83. @982 $28.82.

$888: __ 385 68:68 83

 

Appendix 1.1
Page _2_ of __2_ Pages

Voucher Specimen Data

58 am __> .228 3.2985 mmxémﬁ.
m 88.8.3. :8 88.53. .2 <2. 85 .o 9.x. 8882.9. 868.3

8m 9 ___> $85 2.29.85 wmx¢w§
m 858.8. .8 88.53. :2 <9. 5.3.5 8.9.8 8.....ch...

meow : __> .285 2.29.8... mmvEmS
m 85.693. .60 oonEmﬁv. =2 <w: Eowmoav monBoSmﬁ «333$qu

88 F __> Em“. 620595 - :ms.

 

 

 

 

Museum
deposite

where

 

 

 

 

 

P 9.653 6mm .60 629.. 5 <9. dam masmcmoﬁﬁl
88 F __> .53”. 3.8595 - :92

F @523 “mam roo Emanc- =2 <w: 322 $3683 mambkszU
88 mm __> Emu. .8685 - :ms.

2 8.83 5mm .8 528. __2 <2. 8% £82
0+ m e 3538 new no»:

6 D. .
m Way m. m w 5 8838 mcmgooam .5. Sun _33 :98. 550 6 «283
A P N m E

 

Other

 

 

 

173

 

REFERENCES CITED

Andersen, A. 2003. Long-term experiments with reduced tillage in spring cereals. 11.
Effects on pests and beneﬁcial insects. Crop Protection 22: 147-152.

Bianchi, F. J. J. A., and W. van der Werf. 2004. Model evaluation of the function of
prey in non-crop habitats for biological control by ladybeetles in agricultural
landscapes. Ecological Modelling 171 : 177-193.

Borer, E. T. 2002. Intraguild predation in larval parasitoids: implications for coexistence.
Journal of Animal Ecology 71: 957-965.

Borer, E. T., C. J. Briggs, W. W. Murdoch, and S. L. Swarbrick. 2003. Testing
intraguild predation theory in a ﬁeld system: does numerical dominance shift
along a gradient of productivity? Ecology Letters 6: 929-935.

Boyer, A. G., R. E. Swearingen, M. A. Blaha, C. T. Fortson, S. K. Gremillion, K. A.
Osborn, and M. D. Moran. 2003. Seasonal variation in top-down and bottom-up
processes in a grassland arthropod community. Oecologia 136: 309-316.

Brewer, M. J., and T. Noma. 2003. Detecting new natural enemies of soybean aphid.
NCR-125 Arthropod Biological Control. Michigan State Report,
http://www.cips.msu.edu/ncrl 25/StateRpt82003MI.htm.

Brewer, M. J., T. N oma, and N. C. Elliott. 2005. Hymenopteran parasitoids and
dipteran predators of the invasive aphid Diuraphis noxia after enemy
introductions: Temporal variation and implication for future aphid invasions.
Biological Control 33: 315-323.

Briggs, C. J., and E. T. Borer. 2005. Why short-term experiments may not allow long-
terrn predictions about intraguild predation. Ecological Applications 15: 1111-
1 1 17.

Brodeur, J., and J. N. McNeil. 1992. Host behavior modiﬁcation by the endoparasitoid
Aphidius nigripes - A strategy to reduce hyperparasitism. Ecological Entomology
1 7: 97-1 04.

Brodeur, J., and J. A. Rosenheim. 2000. Intraguild interactions in aphid parasitoids.
Entomologia Experimentalis et Applicata 97: 93-108.

Brown, G. C., M. J. Sharkey, and D. W. Johnson. 2003. Bionomies of Scymnus
(Pullus) louisianae J. Chapin (Coleoptera : Coccinellidae) as a predator of the
soybean aphid, Aphis glycines Matsumura (Homoptera : Aphididae). Journal of
Economic Entomology 96: 21-24.

174

Brown, M. W., and S. S. Miller. 1998. Coccinellidae (Coleoptera) in apple orchards of
eastern West Virginia and the impact of invasion by Harmonia axyridis.
Entomological News 109: 143-151.

Burrows, M. E. L., C. M. Boerboom, J. M. Gaska, and C. R. Grau. 2005. The
relationship between Aphis glycines and Soybean mosaic virus incidence in
different pest management systems. Plant Disease 89: 926-934.

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.

Carter, P. E., and A. L. Rypstra. 1995. Top-down effects in soybean agroecosystems -
Spider density affects herbivore damage. Oikos 72: 433-439.

 

Chang, G. C., and P. Kareiva. 1999. The case of indigenous generalists in biological
control. In H. V. Cornell [ed.], Theoretical Approaches to Biological Control.
Cambridge University Press, Cambridge.

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.

Cisneros, J. J., and L. D. Godfrey. 2001. Midseason pest status of the cotton aphid
(Homoptera: Aphididae) in California cotton: Is nitrogen a key factor?
Environmental Entomology 30: 501 -5 1 0.

Clark, A. J., and K. L. Perry. 2002. Transmissibility of ﬁeld isolates of soybean viruses
by Aphis glycines. Plant Disease 86: 1219-1222.

Clark, M. S., S. H. Gage, and J. R. Spence. 1997. Habitats and management associated
with common ground beetles (Coleoptera: Carabidae) in a Michigan agricultural
landscape. Environmental Entomology 26: 519-527.

Colfer, R. G., and J. A. Rosenheim. 2001. Predation on immature parasitoids and its
impact on aphid suppression. Oecologia 126: 292-304.

Coll, M., and M. Guershon. 2002. Omnivory in terrestrial arthropods: mixing plant and
prey diets. Annual Review of Entomology 47: 267-297.

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.

Costamagna, A. C., and D. A. Landis. 2006. Predators exert top-down control of
soybean aphid across a gradient of agricultural management systems. Ecological
Applications In press. ‘

175

 

Denno, R. F., C. Gratton, H. Dobel, and D. L. Finke. 2003. Predation risk affects
relative strength of top-down and bottom-up impacts on insect herbivores.
Ecology 84: 1032-1044.

Denno, R. F., C. Gratton, M. A. Peterson, G. A. Langellotto, D. L. Finke, and A. F.
Huberty. 2002. Bottom-up forces mediate natural-enemy impact in a
phytophagous insect community. Ecology 83: 1443-1458.

DiFonzo, C. D., and R. Hines. 2002. Soybean aphid in Michigan: update from the 2001
season. MSU Extension Bulletin E-2748. Michigan State University, E. Lansing
MI.

Dixon, A. F. G. 1987. Parthenogenetic reproduction and the rate of increase in aphids. In
A. K. Minks and P. Harrewijn [eds], Aphids. Their biology, natural enemies and
control. Elsevier, Amsterdam, The Netherlands.

Dixon, A. F. G. 1998. Aphid ecology. An optimization approach. Chapman and Hall,
London, UK.

Dixon, A. F. G. 2000. Insect predator-prey dynamics. Ladybird beetles and biological
control. Cambridge University Press, Cambridge.

Dyer, L. A., and D. K. Letoumeau. 1999. Relative strengths of top-down and bottom-
up forces in a tropical forest community. Oecologia 119: 265-274.

Dyer, L. A., and J. O. Stireman. 2003. Community-wide trophic cascades and other
indirect interactions in an agricultural community. Basic and Applied Ecology 4:
423-432.

Ehrlich, P. R., and L. C. Birch. 1967. The "balance of nature" and "population control".
American Naturalist 101: 97-107.

Ferguson, K. 1., and P. Stiling. 1996. Non-additive effects of multiple natural enemies
on aphid populations. Oecologia 108: 375-379.

Finke, D. L., and R. F. Denno. 2002. Intraguild predation diminished in complex-
structured vegetation: Implications for prey suppression. Ecology 83: 643-652.

F inke, D. L., and R. F. Denno. 2003. Intra-guild predation relaxes natural enemy
impacts on herbivore populations. Ecological Entomology 28: 67-73.

F inke, D. L., and R. F. Denno. 2004. Predator diversity dampens trophic cascades.
Nature 429: 407-410.

Finke, D. L., and R. F. Denna. 2005. Predator diversity and the functioning of
ecosystems: the role of intraguild predation in dampening trophic cascades.
Ecology Letters 8: 1299-1306.

176

Fletcher, M. J., and P. Desborough. 2000. The soybean aphid, Aphis glycines, present
in Australia. NSW Agriculture, Document 8489,.

Fox, T. B. 2002. Biological control of soybean (Aphis glycines Matsumura) aphid in
Michigan, Department of Entomology. Michigan State University, East Lansing.

Fox, T. B., and D. A. Landis. 2003. lrnpact of habitat management on generalist
predators of the soybean aphid, Aphis glycines Matsumura, pp. 250-255,
International Symposium on Biological Control of Arthropods. Unites States
Department of Agriculture, Forest Service, Forest Health Enterprise Team,
Morgantown, WV., Honolulu, HI.

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.

Fraser, L. H., and J. P. Grime. 1998. Top-down control and its effect on the biomass
and composition of three grasses at high and low soil fertility in outdoor
microcosms. Oecologia 113: 239-246.

Fraser, L. H., and J. P. Grime. 1999. Experimental tests of trophic dynamics: towards a
more penetrating approach. Oecologia 1 19: 281-284.

Gencsoylu, 1., and I. Yalcin. 2004. Advantages of different tillage systems and their
effects on the economically important pests, T hrips tabaci Lind. and Aphis
gossypii Glov. in cotton ﬁelds. Journal of Agronomy and Crop Science 190: 381-
388.

Gonzales, W. L., E. Gianoli, and H. M. Niemeyer. 2001. Plant quality vs. risk of
parasitism: within-plant distribution and performance of the corn leaf aphid,
Rhopalosiphum maidis. Agricultural and Forest Entomology 3: 29-33.

Gratton, C., and R. F. Denno. 2003a. Seasonal shift ﬁ'om bottom-up to top-down
impact in phytophagous insect populations. Oecologia 134: 487-495.

Gratton, C., and R. F. Denno. 2003b. Inter-year carryover effects of a nutrient pulse on
Spartina plants, herbivores, and natural enemies. Ecology 84: 2692-2707.

Gruner, D. S. 2004. Attenuation of top-down and bottom-up forces in a complex
terrestrial community. Ecology 85: 3010-3022.

Hacker, S. D., and M. D. Bertness. 1995. A herbivore paradox: why salt-marsh aphids
live on poor-quality plants? American Naturalist 145: 192-210.

177

Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure,
population control, and competition. The American Naturalist 94: 421-425.

Halaj, J., and D. H. Wise. 2001. Terrestrial trophic cascades: How much do they
trickle? American Naturalist 157: 262-281.

Harrison, B., T. A. Steinlage, L. L. Domier, and C. J. D'Arcy. 2005. Incidence of
Soybean dwarf virus and identiﬁcation of potential vectors in Illinois. Plant
Disease 89: 28-32.

Hawkins, B. A., N. J. Mills, M. A. Jervis, and P. W. Price. 1999. Is the biological
control of insects a natural phenomenon? Oikos 86: 493-506.

Heimpel, G. E., and Z. Wu. 2003. Biological control of Soybean aphid, Aphis glycines.
NCR-125 Arthropod Biological Control. Michigan State Report,

http://www.cips.msu.edu/ncr125/StateRpt32003 MN.htm#heimpel.

Heimpel, G. E., and T. E. Shelly. 2004. The soybean aphid: a review of its biology and
management. Annals of the Entomological Society of America 97.

Heimpel, G. E., J. A. Rosenheim, and M. Mangel. 1997. Predation on adult Aphytis
parasitoids in the ﬁeld. Oecologia 110: 346-352.

Heimpel, G. E., D. W. Ragsdale, R. Venette, K. R. Hopper, R. J. O'Neil, C. E.
Rutledge, and Z. S. Wu. 2004. Prospects for importation biological control of
the soybean aphid: Anticipating potential costs and beneﬁts. Annals of the
Entomological Society of America 97: 249-258.

Heinz, K. M., and J. M. Nelson. 1996. Interspeciﬁc interactions among natural enemies
of Bemisia in an inundative biological control program. Biological Control 6:
384-393.

Hesler, L. S., and R. K. Berg. 2003. Tillage impacts cereal aphid (Homoptera:
Aphididae) infestations in spring small grains. Journal of Economic Entomology
96: 1792-1797.

Hill, C. B., Y. Li, and G. L. Hartman. 2004a. Resistance to the soybean aphid in
soybean germplasm. Crop Science 44: 98-106.

Hill, C. B., Y. Li, and G. L. Hartman. 2004b. Resistance of Glycine species and various
cultivated legumes to the soybean aphid (Homoptera: Aphididae). Journal of
Economic Entomology 97: 1071-1077.

Hill, J. H., R. Alleman, and D. B. Bag. 2001. First report of transmission of Soybean
mosaic virus and Alfalfa mosaic virus by Aphis glycines in the new world. Plant
Disease 85: 561.

178

Hirano, K., K. Honda, and S. Miyai. 1996. Effects of temperature on development,
longevity and reproduction of the soybean aphid, Aphis glycines (Homoptera:
Aphididae). Applied Entomology and Zoology 31: 178-180.

Hodgson, E. W., E. C. Burkness, W. D. Hutchison, and D. W. Ragsdale. 2004.
Enumerative and binomial sequential sampling, plans for soybean aphid
(Homoptera : Aphididae) in soybean. Journal of Economic Entomology 97: 2127-
21 36.

Hiiller, C., C. Borgemeister, H. Haardt, and W. Powell. 1993. The relationship
between primary parasitoids and hyperparasitoids of cereal aphids. An analysis of
ﬁeld data. Journal of Animal Ecology 62: 12-21.

Holt, R. D., and G. A. Polis. 1997. A theoretical framework for intraguild predation.
American Naturalist 149: 745-764.

Hopkins, G., and A. F. G. Dixon. 1997. Enemy-free space and the feeding niche of an
aphid. Ecological Entomology 22: 271-274.

Hopper, K. R., S. Aidara, S. Agret, J. Cabal, D. Coutinot, R. Dabire, C. Lesieux, G.
Kirk, S. Reichert, F. Tronchetti, and J. Vidal. 1995. Natural enemy impact on
the abundance of Diuraphis noxia (Homoptera, Aphididae) in wheat in southern
France. Environmental Entomology 24: 402-408.

Howarth, F. G. 1983. Classical biological control: panacea or Pandora's box?
Proceedings of the Hawaii Entomological Society 24: 239-244.

Howarth, F. G. 1991. Environmental impacts of classical biological control. Annual
Review of Entomology 36: 485-509.

Hunt, D., R. Foottit, D. Gagnier, and T. Baute. 2003. First Canadian records of Aphis
glycines (Hemiptera : Aphididae). Canadian Entomologist 135: 879-881.

Hunter, M. D., and P. W. Price. 1992. Playing chutes and ladders: heterogeneity and
the relative roles of bottom-up and top-down forces in natural communities.
Ecology 73: 724-732.

Hunter, M. D., G. C. Varley, and G. R. Gradwell. 1997. Estimating the relative roles
of top-down and bottom-up forces on insect herbivore populations: A classic
study revisited. Proceedings of the National Academy of Sciences of the United
States of America 94: 9176-9181.

Ives, A. R., P. Kareiva, and R. Perry. 1993. Response of a predator to variation in prey
density at three hierarchical scales - Lady beetles feeding on aphids. Ecology 74:
1929-1938.

Kareiva, P., and R. Sahakian. 1990. Tritrophic Effects of a Simple Architectural
Mutation in Pea-Plants. Nature 345: 433-434.

179

Kindlmann, P., and A. F. G. Dixon. 2001. When and why top-down regulation fails in
arthropod predator-prey systems. Basic and Applied Ecology 2: 333-340.

Landis, D. A., and W. Van der Werf. 1997. Early-season predation impacts the
establishment of aphids and spread of beet yellows virus in sugar beet.
Entomophaga 42: 499-516.

Landis, D. A., M. J. Brewer, and G. E. Heimpel. 2003. Soybean aphid parasitoid
questionnaire 2003. NCR-125 Arthropod Biological Control. Michigan State
Report, http://www.cips.msu.edu/ncr125/StateRpt32003MI.htm.

Lang, A. 2003. Intraguild interference and biocontrol effects of generalist predators in a
winter wheat ﬁeld. Oecologia 134: 144-153.

Li, Y., C. B. Hill, and G. L. Hartman. 2004. Effect of three resistant soybean genotypes
on the fecundity, mortality, and maturation of soybean aphid (Homoptera :
Aphididae). Journal of Economic Entomology 97: 1106-1111.

Lima, S. L. 1998. Nonlethal effects in the ecology of predator-prey interactions. What
are the ecological effects of anti-predator decision-making? Bioscience 48: 25-34.

Lin, L. A., and A. R. Ives. 2003. The effect of parasitoid host-size preference on host
population growth rates: an example of Aphidius colemam' and Aphis glycines.
Ecological Entomology 28: 542-550.

Littell, R. C., G. A. Milliken, W. W. Stroup, and R. D. Wolﬁnger. 1996. SAS system
for mixed models. SAS Institute Inc., Cary, NC.

Liu, J., K. M. Wu, K. R. Hopper, and K. J. Zhao. 2004. Population dynamics of Aphis
glycines (Homoptera: Aphididae) and its natural enemies in soybean in northern
China. Annals of the Entomological Society of America 97 : 235-239.

Losey, J. E., and R. F. Denno. 1998a. Positive predator-predator interactions: Enhanced
predation rates and synergistic suppression of aphid populations. Ecology 79:
2143-2152.

Losey, J. E., and R. F. Denno. 1998b. The escape response of pea aphids to foliar-
foraging predators: factors affecting dropping behaviour. Ecological Entomology
23: 53-61.

Losey, J. E., and R. F. Denno. 1999. Factors facilitating synergistic predation: The
central role of synchrony. Ecological Applications 9: 378-3 86.

Losey, J. E., J. K. Waldron, E. R. Hoebeke, L. E. Macomber, and B. N. Scott. 2002.
First record of the soybean aphid, Aphis glycines Matsumura (Hemiptera :
Sternorrhyncha : Aphididae), in New York. Great Lakes Entomologist 35: 101-
1 05.

180

Louda, S. M., R. W. Pemberton, M. T. Johnson, and P. A. Follett. 2003. Nontarget
effects - the Achilles' Heel of biological control? Retrospective analyses to reduce

risk associated with biocontrol introductions. Annual Review of Entomology 48:
365-396.

Lucas, E., D. Coderre, and J. Brodeur. 1998. Intraguild predation among aphid
predators: Characterization and inﬂuence of extraguild prey density. Ecology 79:
1084-1092.

Luck, R. F., B. M. Shepard, and P. E. Kenmore. 1988. Experimental methods for
evaluating arthropod natural enemies. Annual Review of Entomology 33: 367-
391.

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.

Magalhaes, S., A. Janssen, R. Hanna, and M. Sabelis. 2002. Flexible antipredator
behaviour in herbivorous mites through vertical migration in a plant. Oecologia
132: 143-149.

Maredia, K. M., S. H. Gage, D. A. Landis, and J. M. Scriber. 1992. Habitat use
patterns by the seven-spotted lady beetle (Coleoptera: Coccinellidae) in a diverse
agricultural landscape. Biological Control 2: 159-165.

McAllister, M. K., and B. D. Roitberg. 1987. Adaptive suicidal. Behavior in pea
aphids. Nature 328: 797-799.

McCornack, B. P., D. W. Ragsdale, and R. C. Venette. 2004. Demography of soybean
aphid (Homoptera: Aphididae) at summer temperatures. Journal of Economic
Entomology 97: 854-861.

McCornack, B. P., M. A. Carrillo, R. C. Venette, and D. W. Ragsdale. 2005.
Physiological constraints on the overwintering potential of the soybean aphid
(Homoptera: Aphididae). Environmental Entomology 34: 235-240.

Menalled, F. D., K. L. Gross, and M. Hammond. 2001. Weed aboveground and
seedbank community responses to agricultural management systems. Ecological
Applications 1 1 : 1586-1601 .

Mensah, C., C. DiFonzo, R. L. Nelson, and D. C. Wang. 2005. Resistance to soybean
aphid in early maturing soybean germplasm. Crop Science 45: 2228-2233.

Meyhiifer, R. 2001. Intraguild predation by aphidophagous predators on parasitised
aphids: the use of multiple video cameras. Entomologia Experimentalis Et
Applicata 100: 77-87.

181

 

Meyhiifer, R., and D. Hindayana. 2000. Effects of intraguild predation on aphid
parasitoid survival. Entomologia Experimentalis Et Applicata 97: 115-122.

Meyhr’ifer, R., and T. Klug. 2002. Intraguild predation on the aphid parasitoid
Lysiphlebusfabarum (Marshall) (Hymenoptera: Aphidiidae): mortality risks and
behavioral decisions made under the threats of predation. Biological Control 25:
239-248.

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.

Mignault, M.-P., M. Roy, and J. Brodeur. 2006. Soybean aphid predators in Québec
and the suitability of Aphis glycines as prey for three Coccinellidae. BioControl
51: 89-106.

Morales, H., I. Perfecto, and B. Ferguson. 2001. Traditional fertilization and its effect
on corn insect populations in the Guatemalan highlands. Agriculture Ecosystems
& Environment 84: 145-155.

Moran, M. D., and L. E. Hurd. 1998. A trophic cascade in a diverse arthropod
community caused by a generalist arthropod predator. Oecologia 113: 126-132.

Moran, M. D., and A. R. Scheidler. 2002. Effects of nutrients and predators on an old-
ﬁeld food chain: interactions of top-down and bottom-up processes. Oikos 98:
116-124.

Miiller, C. B., and J. Brodeur. 2002. Intraguild predation in biological control and
conservation biology. Biological Control 25: 216-223.

Murdoch, W. W. 1966. Community structure, p0pu1ation control, and competition: a
critique. American Naturalist 100: 219-226.

Myers, S. W., D. B. Hogg, and J. L. Wedberg. 2005a. Determining the optimal timing
of foliar insecticide applications for control of soybean aphid (Hemiptera :
Aphididae) on soybean. Journal of Economic Entomology 98: 2006-2012.

Myers, S. W., C. Gratton, R. P. Wolkowski, D. B. Hogg, and J. L. Wedberg. 2005b.
Effect of soil potassium availability on soybean aphid (Hemiptera : Aphididae)
population dynamics and soybean yield. Journal of Economic Entomology 98:
113-120.

Nakashima, Y., and M. Akashi. 2005. Temporal and within-plant distribution of the
parasitoid and predator complexes associated with Acyrthosiphon pisum and A.
kondoi (Homoptera : Aphididae) on alfalfa in Japan. Applied Entomology and
Zoology 40: 137-144.

182

Nevo, E., and M. C011. 2001. Effect of nitrogen fertilization on Aphis gossypii
(Homoptera : Aphididae): Variation in size, color and reproduction. Journal of
Economic Entomology 94: 27-32.

Nielsen, C., and A. E. Hajek. 2005. Control of invasive soybean aphid, Aphis glycines
(Hemiptera : Aphididae), populations by existing natural enemies in New York
State, with emphasis on entomopathogenic fungi. Environmental Entomology 34:
1036-1047.

Obrycki, J. J., and T. J. Kring. 1998. Predaceous Coccinellidae in biological control.
Annual Review of Entomology 43: 295-321.

Onstad, D. W., S. Fang, and D. J. Voegtlin. 2005a. Forecasting seasonal population
growth of Aphis glycines (Hemiptera : Aphididae) in soybean in Illinois. Journal
of Economic Entomology 98: 1157-1162.

Onstad, D. W., S. Fang, D. J. Voegtlin, and M. G. Just. 2005b. Sampling Aphis
glycines (Homoptera : Aphididae) in soybean ﬁelds in Illinois. Environmental
Entomology 34: 170-177.

Osawa, N. 1993. Population - ﬁeld studies of the aphidophagous ladybird beetle
Harmonia axyridis (Coleoptera, Coccinellidae). Life-tables and key factor-
analysis. Researches on Population Ecology 35: 335-348.

Osawa, N. 2000. Population ﬁeld studies on the aphidophagous ladybird beetle
Harmonia axyridis (Coleoptera : Coccinellidae): resource tracking and population
characteristics. Population Ecology 42: 115-127.

Ostman, 6., B. Ekbom, and J. Bengtsson. 2001. Landscape heterogeneity and farming
practice inﬂuence biological control. Basic and Applied Ecology 2: 365-371.

Pace, M. L., J. J. Cole, S. R. Carpenter, and J. F. Kitchell. 1999. Trophic cascades
revealed in diverse ecosystems. Trends in Ecology & Evolution 14: 483-488.

Pike, K. S., P. Stary, T. Miller, G. Graf, D. Allison, L. Boydston, and R. Miller. 2000.
Aphid parasitoids (Hymenoptera : Braconidae : Aphidiinae) of Northwest USA.
Proceedings of the Entomological Society of Washington 102: 688-740.

Polis, G. A., and R. D. Holt. 1992. Intraguild predation - the dynamics of complex
trophic interactions. Trends in Ecology & Evolution 7: 151-154.

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., A. L. W. Sears, G. R. Huxel, D. R. Strong, and J. Maron. 2000. When is
a trophic cascade a trophic cascade? Trends in Ecology & Evolution 15: 473-475.

183

Powel, W., M. P. Walton, and M. A. Jervis. 1996. Populations and communities, pp.
223-292. In M. A. Jervis and N. Kidd [eds], Insect natural enerrries. Practical
approaches to their study and evaluation, lst ed. Chapman & Hall, London, UK.

Power, M. E. 1992. Top-Down and Bottom-up Forces in Food Webs - Do Plants Have
Primacy. Ecology 73: 733-746.

Prasad, R. P., and W. E. Snyder 2006. Polyphagy complicates conservation biological
control that targets generalist predators. Journal of Applied Ecology 43: 343-352.

Press, J. W., B. R. Flaherty, and R. T. Arbogast. 1974. Interactions among Plodia
interpunctella (Lepidoptera: Pyralidae), Bracon hebertor (Hymenoptera:
Braconidae), and X ylocoris flavipes (Hemiptera: Anthocoridae). Environmental
Entomology 3: 183-184. .

Quinn, G. P., and M. J. Keough. 2002. Experimental design and data analysis for
biologists. Cambridge University Press, Cambridge, UK.

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.

Rees, N. E., and J. A. Onsager. 1982. Inﬂuence of predators on the efﬁciency of the
Blaesoxipha spp.(Diptera, Sarcophagidae) Parasites of the Migratory Grasshopper
Melanoplus sanguinipes (F) (Orthoptera, Acrididae). Environmental Entomology
11: 426-428.

Rees, N. E., and J. A. Onsager. 1985. Parasitism and survival among rangeland
grasshoppers in response to suppression of robber ﬂy (Diptera, Asilidae)
predators. Environmental Entomology 14: 20-23.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-225.

Ritchie, S. W., J. J. Hanway, H. E. Thompson, and G. O. Benson. 1994. How a
soybean plant develops? Iowa State Cooperative Extension, Iowa State
University, Ames.

Robertson, G. P., E. A. Paul, and R. R. Harwood. 2000. Greenhouse gases in intensive
agriculture: Contributions of individual gases to the radiative forcing of the
atmosphere. Science 289: 1922-1925.

Roland, J., and D. G. Embree. 1995. Biological control of the winter moth. Annual
Review of Entomology 40: 475-492.

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
function: Can a knowledge of foraging behavior help? Ecology 84: 253 8-2548.

184

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.

Rossing, W. A. H., R. A. Daamen, and M. J. W. Jansen. 1994. Uncertainty analysis
applied to supervised control of aphids and brown rust in winter-wheat .1.

Quantiﬁcation of uncertainty in cost-beneﬁt calculations. Agricultural Systems
44: 419-448.

Rutledge, C. E., and R. J. O'Neil. 2005. Orius insidiosus (Say) as a predator of the
soybean aphid, Aphis glycines Matsumura. Biological Control 33: 56-64.

Rutledge, C. E., and R. J. O'Neil. 2006. Soybean plant stage and population growth of
soybean aphid. Journal of Economic Entomology 99: 60-66.

Rutledge, C. E., R. J. 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. 2001. SAS/STAT User's manual. Version 8.2. SAS Institute Inc., Cary,
NC.

Sato, S., A. F. G. Dixon, and H. Yasuda. 2003. Effect of emigration on cannibalism and
intraguild predation in aphidophagous ladybirds. Ecological Entomology 28: 628-
633.

Scheiner, S. M. 2001. Manova. Multiple response variables and multiespecies
interactions, pp. 99-115. In S. M. Scheiner and J. Gurevitch [eds], Design and
analysis of ecological experiments, 2 ed. Oxford University Press, New York.

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., A. Lauer, T. Purtauf, C. Thies, M. Schaefer, and T. Tscharntke.
2003. Relative importance of predators and parasitoids for cereal aphid control.
Proceedings of the Royal Society of London Series B-Biological Sciences 270:
1 905-1 909.

Schmitz, O. J., P. A. Hamback, and A. P. Beckerman. 2000. Trophic cascades in
terrestrial systems: A review of the effects of carnivore removals on plants.
American Naturalist 155: 141-153.

Schuh, R. T. 2001. Revision of new world Plagiognathus Fieber, with comments on the
palearctic fauna and the description of a new genus (Heteroptera : Miridae :
Phylinae). Bulletin of the American Museum of Natural History: 1-264.

185

Schuh, R. T., and M. D. Schwartz. 2005. Review of North American Chlamydatus
Curtis species, with new Synonymy and the description of two new species
(Heteroptera : Miridae : phylinae). American Museum Novitates: 1-55.

Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for
biological invasions. Trends in Ecology & Evolution 17: 170-176.

Sih, A., G. Englund, and D. Wooster. 1998. Emergent impacts of multiple predators on
prey. Trends in Ecology & Evolution 13: 350-355.

Simberloff, D., and P. Stiling. 1996. How risky is biological control? Ecology 77: 1965-
1974.

Sinclair, A. R. E., S. Mduma, and J. S. Brashares. 2003. Patterns of predation in a
diverse predator-prey system. Nature 425: 288-290.

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.

Snyder, W. E., and A. R. Ives. 2001. Generalist predators disrupt biological control by a
specialist parasitoid. Ecology 82: 705-716.

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., G. M. Clevenger, and S. D. Eigenbrode. 2004. Intraguild predation and
successful invasion by introduced ladybird beetles. Oecologia 140: 559-565.

Spiller, D. A., and T. W. Schoener. 1990. A terrestrial ﬁeld experiment showing the
impact of eliminating top predators on foliage damage. Nature 347: 469-472.

Stadler, B., A. F. G. Dixon, and P. Kindlmann. 2002. Relative ﬁtness of aphids: effects
of plant quality and ants. Ecology Letters 5: 216-222.

Stiling, P., and A. M. Rossi. 1997. Experimental manipulations of top-down and
bottom-up factors in a tri-trophic system. Ecology 78: 1602-1606.

Stiling, P., and T. Comelissen. 2005. What makes a successful biocontrol agent? A
meta-analysis of biological control agent performance. Biological Control 34:
236-246.

Strong, D. R. 1992. Are trophic cascades all wet? Differentiation and donor-control in
speciose ecosystems. Ecology 73: 747-754.

186

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.

SYSTAT Software Inc. 2004. SYSTAT User's Manual. Version 11. SYSTAT Soﬁware
Inc., Chicago, IL.

Takizawa, T., H. Yasuda, and B. K. Agarwala. 2000. Effects of parasitized aphids
(Homoptera: Aphididae) as food on larval performance of three predatory
ladybirds (Coleoptera: Coccinellidae). Applied Entomology and Zoology 35: 467-
472.

Taylor, A. J ., C. B. Muller, and H. C. J. Godfray. 1998. Effect of aphid predators on
oviposition behavior of aphid parasitoids. Journal of Insect Behavior 1 1: 297-302.

Terborgh, J., L. Lopez, P. Nunez, M. Rao, G. Shahabuddin, G. Orihuela, M.
Riveros, R. Ascanio, G. H. Adler, T. D. Lambert, and L. Balbas. 2001.
Ecological meltdown in predator-free forest ﬁagments. Science 294: 1923-1926.

Tillman, G., H. Schomberg, S. Phatak, B. Mullinix, S. Lachnicht, P. Timper, and D.
Olson. 2004. Inﬂuence of cover crops on insect pests and predators in
conservation tillage cotton. Journal of Economic Entomology 97: 1217-1232.

Tscharntke, T. 1992. Cascade effects among four trophic levels: bird predation on galls
affects density-dependent parasitism. Ecology 73: 1689-1698.

Turchin, P., and P. Kareiva. 1989. Aggregation in Aphis varians, an effective strategy
for reducing predation risk. Ecology 70: 1008-1016.

University of Wisconsin. 2003. Soybean plant health.
http://www.plantpath.wisc.edu/sovhealth/aglycine.htrn.

Van den Berg, H., D. Ankasah, A. Muhammad, R. Rusli, H. A. Widayanto, H. B.
Wirasto, and I. Yully. 1997. Evaluating the role of predation in population
ﬂuctuations of the soybean aphid Aphis glycines in farmers' ﬁelds in Indonesia.
Journal of Applied Ecology 34: 971-984.

 

van Lenteren, J. C., D. Babendreier, F. Bigler, G. Burgio, H. M. T. Hokkanen, S.
Kuske, A. J. M. Loomans, I. Menzler-Hokkanen, P. C. J. van Rijn, M. B.
Thomas, M. G. Tommasini, and Q.-Q. Zeng. 2003. Environmental risk
assessment of exotic natural enemies used in inundative biological control.
BioControl 48: 3-38.

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.

187

i It. I :lu'n. a."

Voegtlin, D. J., R. J. O'Neill, and W. R. Graves. 2004. Test of suitability of
overwintering hosts of Aphis glycines: Identiﬁcation of a new host association
with Rhamnus alniflora L'Héritier. Annals of the Entomological Society of
America 97: 233-234.

Viilkl, W. 1992. Aphids or their parasitoids: who actually beneﬁts from ant-attendance.
Journal of Animal Ecology 61: 273-281.

Viilkl, W., and W. Kraus. 1996. Foraging behaviour and resource utilization of the
aphid parasitoid Pauesia unilachni: Adaptation to host distribution and mortality
risks. Entomologia Experimentalis Et Applicata 79: 101-109.

Walker, M., and T. H. Jones. 2001. Relative roles of top-down and bottom-up forces in
terrestrial tritrophic plant-insect herbivore-natural enemy systems. Oikos 93: 177-
187.

Wang, C. L., N. I. Siang, C. S. Chang, and H. F. Chu. 1962. Studies on the soybean
aphid, Aphis glycines Matsumura. Acta Entomologica Sinica 11: 31-44.

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.

Wedburg, J. 2000. Important update on aphids in soybeans. Wisconsin Crop Manager,
http://ipcmwisc.edu/wcm/pdfs/2000/00-22insect1.htrnl.

Wellings, P. W., S. R. Leather, and A. F. G. Dixon. 1980. Seasonal variation in
reproductive potential: a programmed feature of aphid life cycles. Journal of
Animal Ecology 49: 975-985.

Williams, I. S., W. van der Werf, A. M. Dewar, and A. F. G. Dixon. 1999. Factors
affecting the relative abundance of two coexisting aphid species on sugar beet.
Agricultural and Forest Entomology 1: 119-125.

Wu, Z., D. Schenk-Hamlin, W. Zhan, D. W. Ragsdale, and G. E. Heimpel. 2004. The
soybean aphid in China: a historical review. Annals of the Entomological Society
of America 97: 209-218.

Wyatt , I. J ., and P. F. White. 1977. Simple estimation for intrinsic increase rates for
aphids and tetranychid mites. Journal of Applied Ecology 14: 757-766.

Yasuda, H., T. Kikuchi, P. Kindlmann, and S. Sato. 2001. Relationships between
attack and escape rates, cannibalism, and intraguild predation in larvae of two
predatory ladybirds. Journal of Insect Behavior 14: 373-384.

Yoo, H. J. S., R. J. O'Neil, D. J. Voegtlin, and W. R. Graves. 2005. Plant suitability of
Rhamnaceae for soybean aphid (Homoptera : Aphididae). Annals of the
Entomological Society of America 98: 926-930.

188

Zhang, L., R. Wang, and J. D. Hesketh. 2001. Effects of photoperiod on growth and
development of soybean ﬂoral bud in different maturity. Agron J 93: 944-948.

Zhu, J., and K.-C. Park. 2005. Methyl salicylate, a soybean aphid-induced plant volatile
attractive to the predator Coccinella septempunctata. Journal of Chemical
Ecology 31: 1733-1746.

189

   

l11111111111111111