r . . ‘1 J1 I,-.I.-"D
. F . a
33.71,? 61.!!! .> ,HHW,
.):r:. .ufllytflbfl . 1393-...” . A; ..
1“ .va'lka...h.>lfl.xfim q tr.‘-..QU It.

. a . . l

1.5.,1 3].: V7 .»9AY):11.I.'

..>.v.ovo.. :lvflr.

pip: v10

‘ z I...
‘ 75:527.")...-

 

 

 

 

l‘zfi...

:1; vv .‘tAII.
ill. ..

.

r.
ti

U
f

v
.
.1.
1....

- r

y .

,
filrfi

). .ly‘fi .rv p I
3?:

in:“.
I

n . ' xv '- “r
.7» ”f“, ,I of v. ,
- .'.v v a A U
°:a.’;,f: " 'v‘“
l"u~'

t

 

at}. 1.

MI

:3 h.

'J

5333”:
mafia

mi
‘4
CO

‘v’
‘ 5‘1!

5
a

 

Date

23> we m5...

r

WWW

This is to certify that the

dissertation entitled

MANIPULATING OVIPOSITION OF THE ONION FLY,
DELI/1 ANTIQUA (MEIGEN): A STIMULO—DETERRENT
DIVERSIONARY APPROACH

presented by

RICHARD STEVEN COWLES

has been accepted towards fulfillment
of the requirements for

Doctoral in Entomology

degree

 

 

Ma' r professor

QM” ///Z//;/
V :4

May 21, 1990

 

MS U i: an Afl'mnariw Action/Equal Opportunity Institution 042771

 

 

I

 

k

*—

LIBRARY
Inchlgan State
University J

T

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or betore date due.

,___________________.
DATE DUE DATE DUE DATE DUE

 

 

P

 

ll

 

 

 

 

 

L“:JI
‘fl

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[__

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution

ammo-m

 

 

 

MANIPULATING OVIPOSITION OF THE ONION FLY, DELIA ANT I QUA
(MEIGEN): A STIMUID-DETERRENT DIVERSIONARY APPROACH

By

Richard Steven Cowles

A DISSERTATION

Submitted to
Michi an State University
in partial ment of the requirements
for the degree of

DOCTOR OF PHIIDSOPHY

Department of Entomology

1990

5x01,-

 

ABSTRACT

MANIPULATING OVIPOSITION OF THE ONION FLY, DELIA ANT IQUA
(MEIGEN): A STIMULO—DETERRENT DIVERSIONARY APPROACH

By

Richard Steven Cowles

Onions and onion models were used to bioassay onion fly host acceptance
behavior, with the goal of developing strategies for controlling onion fly (Delia
amiqua (Meigen)) oviposition. Stimulo-deterrent diversion (SDD) was developed,
where the valued crop is treated with chemical deterrents, and simultaneously, a
highly stimulatory ovipositional resource (onion culls) is deployed to concentrate
eggs away from the crop.

A wide range of non-onion chemicals deterred onion fly oviposition. In
laboratory choice experiments, pungent spices deterred oviposition by 88 to 100%,
but were ineffective in no-choice conditions. Compounds with appreciable
deterrency are: C8 to C13, intermediate in polarity, and possess either oxygen-
containing or nitrile functional groups. When formulated in polyethylene pellets,
(E )-cinnamaldehyde had a BR90 (concentration eliciting 90% deterrency) of 1.0%
and (E)-4-methoxycinnamaldehyde had a BR“) of 0.38%. The air concentration of
(E)-cinnamaldehyde at its BR” was 1.7 ng/ml. Deterrents alone may not be
sufficient for control; increased oviposition due to deprivation would require high
deterrent concentrations.

The interaction of visual (red) and chemical (cinnamaldehyde) deterrent
stimuli fit a purely multiplicative model, consistent with separate processing of host

stimuli from different modalities during distinct host examining behaviors. Video

recordings of examining behavior revealed that red foliage decreased overall activity
on a resource, while cinnamaldehyde diminished transitions to ovipositor probing.
A greenhouse test of SDD also revealed a multiplicative response when deterrents,
plus culls protected seedling onions. Total eggs laid on seedlings were: seedlings-.
only (3185), seedlings + culls (1531), seedlings + deterrent (127), and seedlings +
deterrent + culls (69). The probability of an onion fly accepting seedlings was -
reduced independently by the presence of deterrent and culls. :

SDD can reduce pest density in a valued crop. It is suggested that increased
pest densities in ~a diversionary crop will enhance biological control. A population-
genetics model using two—allele loci for avoidance and physiological resistance traits
suggested that SDD combined with conventional insecticide could prevent or‘
reverse pesticide resistance development. Requirements are: 1) higher suitability of
the diversionary'crop, 2) high finding of the diversionary crop, and 3) deterrents to
which a pest is preadapted to respond.

To Elizabeth, for her love and her faith in my abilities,
to my parents, for nurturing my interest in plants and insects,

and to Jim, for being the quintessential mentor.

iv

ACKNOWLEDGEMENTS

Iwould like to thank my committee: Drs. George Ayers, Guy L. Bush,
William Mattson, Edward Grafius, and especially James R. Miller for having
patiently helped revise manuscripts generated through this work. I would
especially like to thank Dr. George Ayers for having stimulated thought on
the importance of the diversionary crop concept.

I would like to thank co-workers, who have shared in the toil and less
in the spoils. They have been co-authors on some manuscripts: James R.
Miller, James E. Keller, and Robert M. Hollingworth, from Michigan State
University, and George Matolcsy, F erenc Szurdoki, Taha Abdel-aal, and K.
Bauer from the Plant Protection Institute, Budapest, Hungary. I would also
like to thank David F. Grant for bringing my attention to Fred Gould's
population genetics models so early in my work with Jim Miller. Special
thanks go for the technical assistance from James E. Keller, Brian K.
Mitchell, David Arnosky, and Michael G. Rashid. They helped make my
work easier. JEK and MGR have gone the extra mile and have also made

substantive suggestions for improving my manuscripts.

TABLE OF CONTENTS
LIST OF TABLES

 

LIST OF FIGURES

 

General Introduction

 

Introduction and Overview

 

Perspective on the Problem

 

Host Finding and Acceptance

 

The Nature of Deterrents

 

The Onion / Onion Fly System

 

Delia antiqua Colonization of Onions

 

Stimulo-deterrent diversion

 

Chapter 1. The effects of depth of planting, physical damage, and larval

mfestation on acceptability of cull onions to Delia antiqua ...........

Introduction

 

Methods and Materials

 

 

Experiment 1 - Depth of Planting
Experiment 2 - Damage and Infestation

 

Results and Discussion

 

Experiment 1 - Depth of Planting

 

Experiment 2 - Damage and Infestation

 

General Discussion

 

Chapter 2. Evaluating choice-test methods for onion fly ovipositional
deterrents

 

Introduction

 

Methods and Materials

#Nchth-u—A

HH

 

Onion flies

 

Deterrent

 

Bioassays

 

Experiment 1 - Treatment:Contol Ratio

 

Experiment 2 - Two - vs. Multiple-Choice

 

Data analysis

 

Results and Discussion

 

 

Experiment 1 - Treatment:Contol Ratio

 

Experiment 2 - Two - vs. Multiple-Choice
General Discussion

 

vi

N

E:

HHH H 5—.

N
O\

28
29
30
30
30
31
31
32
32
32

36

Chapter 3. Deterrencyoof pungent spices and capsaicin-containing products

 

to Deha antiqua “

Introduction

 

Methods and Materials

 

Delia antiqua culture

 

Laboratory ovipositional bioassays -

 

Botanicals

 

Dose-response series

 

N o-choice test

 

Field trials

 

Results and Discussion

 

Botanicals

 

Dose-response series

 

No-choice test

 

Field trials

 

General Discussion

 

Chapter 4. Cinnamyl derivatives and monoterpenoids as non-specific

 

owpositional deterrents of the onion fly
Introduction
Methods and Materials

 

 

Delia antiqua culture
Chemicals

-- 53

54

-55

55

'.55

 

Bioassays

 

Vapor concentration

 

Results and Discussion

 

Dose-response series

 

Discriminate-dosage series

 

Vapor concentration

 

Discussion

 

Advantages of discriminate bioassay

 

Structure-activity relationships

 

Biological activity and concentration

 

Behavioral effects and volatility

 

Suggested mode of action

 

Consideration of stimulus interaction

 

 

Ecological interpretations
Practical consrderatrons

 

Chapter 5. Interaction of visual and chemical onion fly ovipositional

 

deterrents

Introduction

 

Methods and Materials

 

General methods

 

vii

.55

59
59

-59
-62

62

-69

70

-70
.71
.72

74

75
76

78
78

 

Experiment 1 - Chemical syner ' m 79

 

 

 

 

 

 

 

 

 

 

 

Experiment 2 - Visual-chemi synergism 79
Experiment 3 - Video analysis 80
Results and Discussion - - 81
Experiment 1 - Chemical syner,- 'sm 81
Experiment 2 - Visual-chemica synergism 83
Experiment 1 - Video analysis _ - 84
General Discussion ...... 88
Chapter 6. Stimulo-deterrent diversion for manipulating onion fly
oviposition: a greenhouse test 90
Introduction 91
Methods and Materials 92
Results and Discussion _ 94

 

Chapter 7. Integrating stimulo-deterrent diversion into pesticide resistance
management: Directing pest evolution towards sustainable

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

control 102
Introduction - 103
Pesticide management approaches - 104
Mixtures and rotation 104
Pesticide plus synergist 105

High dose and selective timing 105
Pesticide strategy assumptions - 106

High vs. low selection strategies - 108
Stimulo-Deterrent Diversion 109
Modeling the impact of SDDS on Resistance 110
Simulation results and discussion 117
Interpretation of figures 117
Genetic factors 120
Operational factors 121
Practical considerations for SDDS - 124
Summary 125
Thesis Summary ...... 127
Diverting oviposition with cull onions 128
Significance of ovipositional deterrents 128
Interactions and host-plant acceptance 129
SDD and natural systems 130
General conclusions 131
Important unanswered questions 132

 

viii

References . - - -- -- 133

 

Appendix 1 BASIC Population genetics model . 146

 

 

Appendix 2 Voucher specimen depository forms _ _ - 149

ix

LIST OF TABLES
Chapter 1.

Table 1. Means for the ln(x+ 1) transformed and back-transformed
(B. T.) means for the numbers of elg‘gs laid upon cull
onions planted at varying depths. e standard error
for all ln(x+ 1) transformed values is 0.22

 

Table 2. Average stand counts and egg count totals for 2.5 meter
gong 1:lsouble row plots of cull onions planted at varying
ept -

 

Table 3. Mean 1- S.E. Delia antiqua eggs laid per 24 hours on
variously treated cull onions in the laboratory -

 

Chapter 2.

Table 1. The effect of treatedzcontrol ratio on the measured
ovipositional deterrency of 1% cinnamaldehyde in
PEG. Deterrency is the difference between ln(x+ 1)
transformed treatment and control eg COuntS

Chapter 4

Table 1. Estimated mean and 95% confidence limits of the BR30
ovipositional deterrency of cinnamaldehy e
derivatives.

 

Table 2. Estimated mean and 95% confidence limits of the BRgo
ovipositional deterrency of monoterpenoids

 

Table 3. Estimated percent ovipositional deterrency (stimulation)
for compounds tested at 0.1% in PEG

 

Chapter 5.

Table 1. Treatments, means for egg counts (ln(x+ 1) transformed),
and back-transformed deterrency values for testing

binary mixtures of three compounds for synergism .............

Table 2. Totals (and expected totals from X2 analysis) for
behavioral events in two-choice assays analyzing the
effects of visual and chemical deterrents on behavioral
transitions.

 

21

-22

_-25

33

60

63

65

82

86

Chapter 6.

Table 1. Analysis of variance for ln(x+ 1) transformed numbers of
onion fly eggs deposited on seedling onions (valued
crop) protected by deterrent (cinnamaldehyde) and
diversionary (cull onion) treatments. Rows represent
sequential presentation of all four treatment
combinations while columns coded the order of
treatment presentation. - --

95

 

Chapter 7.

Table 1. Parameters used in a two-locus, two-allele model
investigating the effects of physiolo 'cal resistance and
behavioral avoidance on the evo ution of pesticide
resistance.

xi

- 113

LIST OF FIGURES
Chapter 2.

Figure 1. Dose response relationships for methyl salicylate generated
when all deterrent concentrations are resented
simultaneously in a bioassay chamber (multiple c oice) or
when single concentrations are compared to control cups (two-
choice). Mean deterrency and standard errors are from back-
transformed logarithmic values

 

Chapter 3.

Figure 1. Number of eggs laid with 5-7 mg of the following $8568
placed within 1.5 cm of surrogate onion in choice tests: N,
control; GIN, gin er; DIL, dill; PPK, paprika; RPP, red

epper; CHI, c ' powder; BPP, black pepper. Means

Tollovgegfliy the same letter are not significant y different (SNK

est, .

 

 

Figure 2. Dose response relationship for oviposition when ground
cayenne pepper was placed m choice tests within 1 cm of
surrogate omon folia‘ge. Straight line fit by eye. Confidence
interval based on 7 d

Figure 3. Dose re use curve for oviposition when substrate was
s rayed in c oice tests with 0.175 ml of 6lfifir-igard Insect

 

epellent solutions. Standard error based on

Figure 4. Dose response relationship for oviposition with synthetic
capsaicin placed within the top cm of sand.

 

Figure 5. Cumulative oviposition for M factorial experiment
investigating pre-reproductive and reproductive exposure to
GCP. - - o - -, no exposure; - -El - -, exposure only pre-
reproductively; - - A - -, exposure only
reproductively; —- o —, continual exposure. - -

 

Figure 6. Percent damage to seedling onions by onion fly. W, water
control; VG, Vaporguard control; AGR, Agrigard; SBR,
Sevana Bird Re llent Powder; CHE, conventional chemical
control (0.05 g c orpyri hos m'l). Amounts given are for 3 m
of row. Means followe by the same letter are not statistically

different (SNK Test, 5 df).

 

xii

35

45

45

-46

46

-48

48

Chapter 4

Figure 1. Method for determining the B from the dose-
relationship, as demonstrated with cinnarlhhonitrile (see text for
further explanation). -

58

 

Chapter 5.

Figure 1. Sample data collection sheet for transcribing behavioral
transitions from videotape recordings of onion fly

preovipositional examining and egg deposition behaviors. ..............

Chapter 6.

Figure 1. Theoretical forms of interactions between deterrent and
diversionary treatments, with non-transformed and log-
transforrned hypothetical data. (A) - no interactions when
non-transformed; slightly negative interactions when
transformed. (B) - Positive interaction. The log-transformed
data show no interaction, hence, the factors are independent.
(C) - Negative interaction.

85

96

 

Figure 2. Experimental data for interactions between
cinnamalde yde deterrent and cull onions, as both non-
transformed (A) and (Hg log-scale graphs. These data are
consistent with Figure 1 , a synergistic effect with an additive
model but with parallel lines indicating treatment

99

 

independence with log-transformation .
Chapter 7.

Figure 1. Schematic for the life cycle of a st and its biology
relevant to the pesticide resistance simu ation model. - - - -
Immature stages (eg through pupae) are exposed to selection;
— Adult stage: mates randomly yet chooses habitat to lay
eggs based on A allele

 

Figure 2. Relationshi between lagooncentration) of a pesticide and
rcent mo ty for rr, , and RR genotypes. Arrows
abeled A, B, and C correspond to pesticide concentrations
simulated in Figures 3 AC .

 

Figure 3. Genetic factors influencing development of avoidance (A
allele) or b iolo 'cal resistance (R allele). (A) R allele is
recessive; §B R a1 ele is dominant; (C) A allele IS dominant;
(D) A alle e is recessive; (E) no fitness costs; (F) fitness cost
exists for R allele; (G) fitness cost exists for a allele; (H) fitness
cost exists for R and a alleles.

 

xiii

112

114

118

(Aal alele) or ph iological resistance allele). (A—C)
pesticide 1sapl1e at low, moderate and gh concentrations,
respectively see arrows in );§D-F) diversionary crop
suitability ts eater than, c to1 an less than the suitability
of the value crpp; (1G ) hi uavailability of diversionary crop;
(H) Low availabi ty of diversion
dicrimination by Aa and AA enotyp will)“ an S0 DSan situation,
including low availability but igh d1scr1mination of Aa and AA
cnotypes for the diversionary crops and low survivorship in
oth d1versionary and valued crops. 119

Figure 4. erational factors influencing develo§ment of avoidance

 

xiv

”... I shall be telling this with a sigh
Somewhere ages and ages hence:
Two roads diverged in a wood, and I-
I took the one less traveled b ,

And that has made all the ' erence.”

- Robert Frost, The Road Not Taken

Introduction and Overview

Today's farmers increasingly face a predicament. The pesticides on which
they depended for the last fifty years are becoming ineffective, unavailable, and
: unacceptable in the market-place (Dover and Croft, 1984; Schneider, 1989). A
major reason for loss of insecticides is physiological resistance. Under strong
selection, many pests have proven capable of adapting to new pesticides faster than
products with new modes of action can be registered (Dover and Croft, 1984). The
process of product registration has become increasingly arduous and expensive,
requiring eight to ten years of research and development at a cost of 25 to 40 million
dollars (Patton, et al., 1982; Storck, 1984). Partly to limit overall registration costs,
manufacturers limit new pesticide registration to those materials that are
marketable on the most profitable major crops. Growers of minor use crops are
thus often forced to use older and less efficacious compounds (Patton, et al., 1982).

Though AlarR is a plant regulator rather than a pesticide, its recent
withdrawal from sale has foreshadowed an era in which scare tactics and consumer

demand for residue-free produce set more stringent pesticide limits than those set

1

by the Environmental Protection Agency. Following the AlarR scare, several
national grocery chains declared that, by the year 1995, they will not sell produce
sprayed with 64 ”hazardous" pesticides. Such action would close large markets to
those growers following conventional and entirely legal practices (Ames and Gold,
1989; Schneider, 1989).

Having this century started down the road of using biocides, is there any way
growers can switch to a different and "less traveled road" that promotes pesticide
alternatives? This thesis investigates how behavioral manipulation of the onion fly
could assist the transition to more rational pest control. I report in this thesis
studies of ovipositional deterrents and cull onions, which could be used for
manipulating onion fly distribution. This, in turn, has important implications for: i)
reducing maggot damage to seedling onions, ii) maintaining biological control
agents, and iii) managing pesticide resistance. Together, these developments

suggest that manipulating pest behavior is crucial to sustainable pest management.

Perspective on the Problem

Wages - From an ecological point of view, pesticides are simply
human-applied antibiotics (Gould, 1984; Brattsten, 1988). By design, the most
successful insecticides, like DDT or chlorpyrifos, have broad-spectrum activity. The
manufacturer benefits from broad-spectrum toxicity, both because of the larger sales
volumes and the concomitant decrease in unit costs of production. The relatively
low cost of pesticides extended to growers partly explains their prevalence today. In
apple production, for example, multiple pests can be controlled with a single
reduced-rate application of synthetic pyrethroid insecticide at a chemical cost of ea.
$2.00 per acre (Cowles, unpublished data).

Applying pesticides as defensive compounds differs significantly from natural

systems where insects may coevolve with plant defences and slowly be selected for

detoxification adaptations. When insecticides with varying modes of action are
available, a grower can alternate or combine these chemicals. This leads to a
situation where pests may not be able to adapt genetically to changes in synthetic
protective chemistry (Curtis, 1985).

Effective pesticides give agriculturists a distinct marketing advantage. Farms
freed from normal ecological constraints can change so that maximal yields of the
most valued cultivar can be grown in monoculture with the same sort of efficiency of
scale that benefits pesticide producers. Ultimately, consumers realize short-term
economic benefit from pesticides, because market forces decrease the cost of
produce.

WW - Iarge-scale, high-yield agriculture is conducive to
pest outbreaks. Monoculture, high soil fertility, and susceptible crops stray from
natural conditions, in which populations often are suppressed by spatial
heterogeneity, low host-plant availability and nutritional quality, toxins and genetic
diversity in host-plants, as well as by predators, parasitoids, and pathogens (Edens et
al., 1985; Tahvanainen and Root, 1972; Atsatt and O'Dowd, 1976; Kogan, 1988;
Bernays and Graham, 1988).

Although pests adapted to monoculture crops live on an ecological "easy
street,” their biological control agents can be severely disadvantaged. Monoculture
and use of broad-spectrum pesticides together can accelerate the spiralling loss of
biological control agents. A hypothetical and simplistic relationship might be that
ecological suppression of pest populations is proportional to species diversity (Atsatt
and O'Dowd, 1976; Price et al., 1980) and inversely proportional to the extent of
pesticide suppression (Levins, 1986).

Insecticides are not the only biocides destroying biological control.
Herbicides may reduce competition between weeds and crop plants, but they also

decrease within-field diversity, and in some cases may be directly toxic to beneficial

insects (Carruthers et al., 1985). Eliminating weeds may improve host-plant finding
by some insects (Tahvanainen and Root, 1972; Thiery and Visser, 1987). In
addition, parasitoids lose sources of pollen and nectar (Atsatt and O'Dowd, 1976),
and ground-cover, favoring predatory insects, also disappears (Ryan et al., 1980;
House and Alzugaray, 1989). Fungicides controlling plant diseases have broad-
spectrum activity, so they often disrupt insect pathogens. For example, onion flies
often are decimated by epizootics of Entomophthora muscae, until fungicides are
applied to control Boaytis leaf blight (Carruthers and Haynes, 1986).

Entomophagous insects are directly and indirectly affected more adversely by
most insecticides than their phytophagous hosts (Croft and Brown, 1975). Gordon
(1961) suggested that since predators and parasitoids are one trophic level above
the plant origin of defensive compounds, they are less likely to have detoxification
enzyme preadaptations that could allow survival in pesticide-laden environments.
Furthermore, since these beneficials depend on pest populations as food, they may
' starve when pests are suppressed to low levels (Croft and Brown, 1975).

Production ”improvements” made possible by applying pesticides, such as
monoculture, high fertility, and large fields, are unlikely to be abandoned by growers
who have made large capital investments in equipment specific to these large scale
practices. When pest control fails, the usual response is to find a new and more
effective pesticide. The ”pesticide treadmill" phenomenon (Van den Bosch, 1978)
results, where the negative impact of this strategy on biological control agents

exacerbates dependence on pesticides.

Ahematixes - On first appearances, growers have "burned their bridges.”
Current practices guarantee that if pesticides fail, biological control alternatives will

not respond quickly enough to prevent severe crop losses. Today's agriculture

critically needs approaches that will allow conventional agronomic practices to use
biological control and other biotic means for pest suppression.

One possible approach toward enhancing biocontrol would be to subdivide
the pest population into two habitats, consisting of a valued crop (initially protected
by insecticides), and a diversionary crop (where pesticides are not used).
Concentrating the pest in a pesticide-flee diversionary crop should have three major
impacts: 1) direct reduction of pest pressure on the valued crop, 2) facilitation of
host finding and use by biological control agents, and 3) providing refugia for
insecticide-susceptible pest genotypes. Reduced insecticide applications on the
valued crop favors survival of biocontrol agents immigrating from the diversionary
crop; this would further reduce the need for insecticides. This positive feedback
could reduce or eliminate the need to apply insecticides to the valued crop.

The key to implement this concept is efficient concentration of the pest in a
diversionary crop. This requires designing the maximum preference differential
between the two habitats, perhaps by simultaneously deploying deterrents on the
valued crop while planting a maximally stimulatory diversionary crop.

Host Finding and Acceptance

Principles of host finding and acceptance have been reviewed by Dethier,
1982; Miller and Strickler, 1984; Courtney et al., 1989; and Courtney and Kibota,
1989. Host colonization is now divided into: finding, examining, and consuming
(Miller and Strickler, 1984). Host finding is the process of arriving at or near host
resources. Behaviors that enable host stimuli to be sensed and processed constitute
examining. Consuming consists of end result behaviors, such as feeding or
oviposition, which indicate that the host has been accepted.

Host finding, the process of arriving at or near host resources, can be

governed by two major strategies. Sensory stimuli that operate at a distance, such as

vision and olfaction could direct movement to a host (Prokopy and Owens, 1983).
Otherwise, insects may follow movement patterns that are independent of host-
associated stimuli, and simply be arrested when host stimuli are encountered.
Tracldng insects in natural or semi-natural environments is difficult. Nevertheless,
the actual path taken by an insect must be analyzed to distinguish between host-
finding mechanisms.

Examining involves active sampling of host stimuli, usually at close quarters,
and leads to host acceptance or rejection. Host acceptance is a complex process
‘ involving integration of physiological state with external stimuli (Dethier, 1982).
The sensory capabilities of each insect shapes how a potential host is perceived.
Thus we should expect each species to respond to a complex of plant stimuli in some
unique manner, and possibly with some degree of genetic variation (Singer, 1982).
Physiological state or experience adds another dimension, individuals may not
always perceive stimuli the same way (Visser, 1983).

Although the physical process involved in the integration of sensory inputs
leading to host acceptance or rejection is not well understood, Miller and Strickler's
(1984) rolling fulcrum model provides a physical representation of the behavioral
process without assuming a particular underlying mechanism. In this model, all
sensory information (visual, olfactory, tactile, etc.) is processed together as a
Gestalt. The relative "weight" of external excitatory inputs are balanced against the
"weight'I of the external inhibitory inputs (deterrents). The likelihood that the insect
will respond to these stimuli, such as by laying eggs, is also affected by the internal
physiological milieu. Ovipositional activity is most likely when a female is mated,
reproductively mature and presented with all pertinent host stimuli (Spencer,
unpublished data). Lack of food required to mature eggs can cause females to be
unresponsive to host stimuli (Klowden, 1989). Conversely, females deprived of a

host can be induced to display ovipositional behaviors even when some host stimuli
are lacking (Harris and Miller, 1991).

Other models of host acceptance behaviors have recently been developed.
The rolling-fulcrum model has been reworked from the perspective of end-result
predictions of individual host selection, and renamed the heirarchy-threshold model
(Courtney et al., 1989). In this model, the rank order of potential hosts is genetically
determined. A highly ranked host will always be accepted by a gravid female, while
lower-ranked hosts will only be accepted following deprivation. If a low-ranked host
is accepted, then any higher-ranked host will be acceptable, while the opposite is not
true.

Experience (learning) can have subtle effects if included with the heirarcby-
threshold concept. J aenike (1983) has observed cross-induction in acceptance of
low-ranked hosts, e.g., an insect is more likely to accept a low-ranked host following
oviposition on a different low-ranked host. The heirarchy-threshold model predicts
cross-induction if acceptance of a low-ranked host is explained as an accompanying
change in specificity (Courtney and Kibota, 1989). Host ranking, the core of the
heirarchy-threshold model, can be dramatically altered through early adult
experience. The change in ranking of apple over hawthorne (Papaj and Prokopy,
1988) by Rhagoletis pomonella is only compatible with the heirarchy-threshold
model if experience re-programs how an individual insect perceives potential hosts.

A radically different approach for modeling host choice has been pioneered
by Mange] (1989a; 1989b; Mange] and Roitberg 1989). He modifies standard
optimal foraging models (Stephens and Krebs, 1986) by replacing expected energy
values with the expected increase in lifetime fitness. A dynamic model is generated
by using a state-variable approach to maximize lifetime fitness (Mangel, 1989b).
This model is important because it predicts changes in ovipositional behavior in

individuals based on the assumption that evolutionary processes have selected for

behaviors that maximize reproductive success. Mangel's dynamic state-variable
approach quantitatively predicts how fast the ”fulcrum rolls” (Miller and Strickler,
1984) relative to the rate of egg maturation and the life expectancy of an individual
female.

The Nature of Deterrents

The complexity of host examining behaviors and use of a Gestalt for host
recognition suggest there may be multiple opportunities for manipulating stimuli to
effect rejection of a potential hostplant. Manipulation could involve host color, if
cultivars with poorly accepted reflectance spectra are developed (Prokopy et al.,
1983). When less-accepted host cultivars are not available, the entomologist may
achieve desired preference differentials by applying deterrents. Theoretically,
almost any sensory stimulus could become deterrent under selective pressure
(Landis and Gould, in press).

Most research in the area of ovipositional deterrents has been shaped by the
' assessment of Dethier (1980) and earlier workers that secondary (”defensive”) plant
compounds are well suited, through insect-plant evolution, for use as deterrents.
Examples of work in this area are investigations on how the composition of
secondary compounds in crucifers determines host specificity of Pierid butterflies
(Renwick and Radke, 1987); and various works (T‘abashnik, 1987; Tingle and
Mitchell, 1984; 1986; Lundgren, 1975), on the possibility of using non-host
secondary compounds to deter oviposition in Pieris species and Heliothis virescens
(F .).

The mode(s) of action of chemical deterrents and repellents is not well
understood. Early discussion of ”deterrent receptors,” (J ermy and Szentesi, 1978)
suggested a similarity to ”token stimuli” and the "labeled line hypothesis” then
invoked for explaining host acceptance (F raenkel, 1959). Under this hypothesis,

single stimuli will produce a stereotyped (reflexive) response independent of other

factors. This idea had to be modified when neurophysiological work (Dethier, 1980)
proved that receptors once thought to be specific are indeed more broadly sensitive.
Broad tuning implies that single compounds elicit complex responses from several
receptor populations whose output must then be interpreted by the central nervous
system.

Davis (1985) proposed several modes of action for repellents. Interference
with perception of host-attractant signals can be brought about by: 1) exciting a
receptor for competing behavior, 2) switching a sensory message from attraction to
repulsion, 3) activating several different receptor systems so that the repellent, in
effect, ”jams” significant sensory information, and/or 4) exciting a repellent (i.e., a
noxious substance) receptor.

A concept common in deterrency literature is that certain compounds may
"mask,” host odors (J ermy and Szentesi, 1978; Stadler, 1983; Visser, 1983)
”Masking," is not yet a defined physiological process; so, like “repellents," this
operational description cannot serve indefinitely in the place of understanding the

underlying sensory-behavioral mechanisms (Davis, 1985).

The Onion / Onion Fly System

Onion, Allium cepa (I..), is a member of the Family Liliaceae.
Approximately 2900 ha are planted to onions in Michigan each year (Clement,
1987), almost exclusively on muck soil. Onions here are mostly grown from seed
planted from mid-April to the first week in May. Gerrninating seedlings first pass
through the “loop” and ”flag” or ”seven” stages, descriptive of the appearance of the
cotyledon. For the remaining interval of active growth, the onion develops an
additional new leaf approximately every week. These tubular leaves emerge from
the center of the plant, displacing the next oldest leaf in an alternating pattern. In

late August through September, these leaves die back as metabolic resources are

10

stored as swollen leaf bases, or bulb scales, which form concentric layers in the
familiar onion bulb (Raven and Curtis, 1970). This process is accelerated by
uprooting the onion and allowing the plant to dry in the field in preparation for
harvest (a common practice in New York, but not in Michigan).

The major onion insect pest, Delia antiqua (Meigen) (Diptera:
Anthomyiidae), usually completes three generations each year in Michigan. Onion
fly adults start emerging from overwintering pupae in mid-April, and feed for one to
two weeks before ovipositing (Loosjes, 1976). Eggs are laid in the soil at the base of
onion plants. When larvae hatch, they first may feed on or around smaller roots,
then usually enter the base of the onion at the meristematic disk. This tissue is so
critical for plant survival that the plant rapidly shrivels and dies. During the first
maggot generation, the number of plants required to complete development
depends on the quantity of tissue available in each plant. When onions are in the
loop stage, as many 28 seedlings may be consumed by a single larva (Workman,
1958). Typically though, several eggs are laid in a batch, and 10 to 20 adjacent
seedlings are destroyed by each group of larvae (Loosjes, 1976). The first
generation larvae requires about one month to develop, so they are present during
the cotyledon to six- or seven-leaf stage of onions.

Second generation flies begin emergence in late June through July. Onions
may have 6-10 leaves at this time, so second-generation larvae often do not kill the
plant (Kendall, 1932; Workman, 1958). Damage from second generation larvae is
largely restricted to plants earlier damaged but not killed by first generation larvae
(Kendall, 1932), or by farm equipment (Finch, et al, 1986b). Third generation flies
are active through harvest and up until a killing freeze (late October or November).
These flies predominantly lay eggs on bulbs damaged by earlier generation larval
feeding or farm machinery, and on sprouting onions left in the field after harvest
(Finch and Eckenrode, 1985; Finch, et al., 1986b). Pupae fi'om this generation, and

11

a small number of diapausing pupae left from the first and second generations,

overwinter to the next growing season (Perron and LaFrance, 1961; Finch and
Eckenrode, 1985).

W - Control of D. antiqua is achieved by using in-
furrow insecticides targeting first generation larvae. These soil pesticides contact
larvae traveling from egg-to-plant and plant-to-plant (Loosjes, 1976). Currently
available in-furrow insecticides, chlorpyrifos and fonofos, are losing (or have lost)
efficacy at labeled rates (Grafius, et al., 1987).

Adulticide sprays used to be directed at first, second, and third generation
adults, often on intensive 3 to 5 day schedules. Spraying contact insecticides has
largely been discontinued; they were ineffective because of insecticide resistance
and because the diurnal movement of flies out of fields coincided with the usual
timing of spray applications (Finch et al., 1986a).

Howitt (1958) and Guyer and Wells (1959) gave the first reports of onion fly
insecticide resistance. These cases reported cross resistance to a large number of
chlorinated hydrocarbons insecticides in the Pacific Northwest and Michigan,
respectively. Harris et a1. (1982) reviewed onion fly insecticide resistance in
Michigan and Ontario, and concluded from comparisons of laboratory selected vs.
field collected flies that field populations could rapidly be selected for high levels of
resistance to organophosphate and carbamate insecticides.

Other areas that have been investigated for control of onion flies include
control with natural enemies. Much of this work has been done at MSU under the
purview of the IPM approach of the 1970's and early 1980's. Several beneficial
insects are of importance, including Aleochara bilineata (Gryll.) (Coleoptera:
Staphylinidae), Aphaereta pallipes (Say) (Hymenoptera: Braconidae), and
Bembidion quadtimaculatum (Coleoptera: Carabidae) (Groden, 1982; Grafius and

Warner, 1990). Conditions affecting Entomophthora muscae, an entomopathogen

12

attacking calypterate muscoid flies, have been studied by Carruthers (1981;
Carruthers and Haynes, 1986). Intensive work in the Netherlands focused control
efforts on sterile male onion fly release (Loosjes, 1976), a method that is
theoretically feasible but expensive and technologically demanding. A more
complete discussion of current control methods for Delia flies is given in Finch
(1989).

Delia antiqua colonization of onions

Onion fly habitat finding has been studied by Martinson et al. (1988). Flies
emerging from overwintering puparia tend to emigrate from fields more than
second generation adults. This difference could be caused by diapause vs. non-
diapause conditioned flies, or alternatively to differences in the quality of host
habitat. First generation onion flies were captured up to 1.5 km from likely
overwintering sources, supporting Loosjes's (1976) calculation of a fivefold increase
in movement rates in non-host habitat.

Onion fly habitat finding probably fits the model, discussed earlier, of non-
directed movement followed by arrestment by host odors. This assessment is
supported by: 1) good fit of an exponential function, relating damage to distance
from overwintering source of flies (Martinson et al., 1988), 2) fly movement out of
fields when onions are producing small quantities of volatiles (Martinson et al.,
1988), and rapid movement between onion fields (Loosjes, 1976; Martinson et al.,
1988).

In contrast to Martinson et al. (1988), Judd and Borden (1989) suggest that
long—distance host finding is mediated by anemotactic response to onion volatiles.
Using 14~day old virgin females, these workers demonstrated slightly cross-wind
biased flight for flies exposed to low n-propyl disulfide concentrations and upwind
flight at higher concentrations (Judd and Borden, 1989). These observations may be

subject to another interpretation, however. Considering that some mating

13

encounters occur on the host-plant (Cowles, personal observation) and that the age
of onion fly mating in lab conditions normally occurs at 5 or 6 days (Joseph Spencer,
Michigan State University), we can speculate that the onion fly females used by
Judd and Borden were highly mating-deprived. The observed response to onion
volatiles may have been more closely related to mate-finding rather than host-
finding behavior.

Odor and vision may both play a role in short-distance host finding. Onion
flies made short, upwind, non-zigzag flights to a source of volatile compounds, most
likely a combination of onion odors combined with microbial products (Dindonis
and Miller, 1980; Hausmann and Miller, 1989b). Harris and Miller (1988) conclude,
based on the use of surrogate onion foliage, that color, shape and size all influence
alighting behavior in the onion fly.

Onion fly host examining is characterized by stereotyped preovipositional
behaviors, such as foliar runs, substrate runs and substrate probing. During these
behaviors, there is concomitant repetitious sampling by mouthparts and ovipositor,
and very likely visual and tarsal stimulation as well (Harris and Miller, 1988).
Harris and Miller (1988, et ante) have established that optimal release of
ovipositional behavior requires simultaneous presentation of a combination of
chemical, visual and tactile stimuli. For example, host models with subooptimal
chemical and visual stimuli will not receive as many eggs in choice tests as the
optimal 4 mm diameter green or yellow cylinders, emitting dipropyl disulfide from
both foliage and substrate.

Harris and Miller (1982) noted that there was a significant interaction
(”synergism") between visual and chemical cues. Stimulus interaction suggests that
onion flies either integrate these cues via central nervous system cross-fiber

patterning or by temporal processing of information.

14

The consumatory stage, oviposition, for Delia antiqua has been examined in
great detail through analysis of video records (Mowry et al., 1989b). Unlike
preovipositional behaviors, which have complex feedback to previous activities, eg
depositional behaviors appear to be deterministic. Egg depositional behavior begins
with subsurface probing with the ovipositor. Once an egg begins moving to the
bursa copulatrix, all ancillary behaviors (such as grooming) cease, and the fly
becomes stationary. The increasingly deterministic nature of behaviors, as a fly
approaches egg deposition, suggests that behaviors occuring late in preovipositional
sequences may be more difficult to avert (with deterrents) than earlier

preovipositional examining activity.

Stimulcdeterrent diversion

The objective of the work embodied in thesis was to assay onion fly host
acceptance behavior, with the goal of developing strategies for preventing
oviposition on young seedling onions. Stimulo-deterrent diversion (SDD) was
developed, where the valued crop is treated with chemical deterrents, and
simultaneously, a highly stimulatory ovipositional resource (sprouting cull onions) is
deployed to concentrate eggs away from the crop. Basic and applied questions are
addressed in this work. Use of deterrents to protect crops is not a well deveIOped
field, and with the exception of work by Rice (1986; Pyke et al., 1987), I am not
aware of other workers considering the importance of bipolar manipulation of pests'
consumatory behaviors.

The thesis is divided into chapters corresponding to units involved with SDD:
Chapter 1 involves stimulatory ovipositional resources, Chapters 2 through 5
investigate deterrent stimuli and their interactions, Chapter 6 describes a
greenhouse test of SDD, and Chapter 7 discusses how SDD may prevent or reverse

pesticide resistance.

Chapter 1
Acceptability of cull onions to Delia antiqua (Meigen) oviposition: the effects of

planting depth,
physical damage, and previous larval infestation

15

Introduction

Onion maggot (Delia antiqua (Meigen)) is the principle insect pest of onions
in temperate regions (Loosjes, 1976; Eckenrode, 1988). In Michigan, onion maggots
are typically trivoltine, with spring, mid-summer, and autumn larval development
(Whitfield, 1981). Young seedlings are especially susceptible to damage caused by
the spring generation of maggots; a single larva typically consumes four to ten
onions (Kendall, 1932; Workman, 1958). Manipulating onion fly ovipositional
activity of D. antiqua on waste onion bulbs (cull onions), may protect seedlings by:
1) reducing the overwintering population and 2) diverting oviposition away from
seedlings.

Third generation densities ranging from 10,000 to 600,000 overwintering
pupae per hectare may result from colonization of onions left in the field after
harvest (Drummond, 1982; Finch and Eckenrode, 1985). Preventing oviposition on
culls left after harvest, and making these bulbs unsuited for larval development
should be effective cultural control practices (Finch and Eckenrode, 1985;
Eckenrode and Nyrop, 1986).

Sprouting cull onions are 60 to 200 times more acceptable to onion flies than
small seedlings (Lovett, 1923; Hammond, 1924; Mowry, unpublished; Chapter 6).
This ovipositional preference may be exploited by planting cull onions in the same
field as small seedlings, either as a trap crop (Lovett, 1923; Hammond, 1924), or a
diversionary crop (Miller and Cowles, 1990). Maximizing preference of onion flies
for the sprouted onion bulb should enhance these strategies for protecting seedlings.

This chapter investigates physical and biological effects on ovipositional ‘
preference of onion flies for sprouted onion bulbs. Both physical and maggot

16

17

feeding damage to onion bulbs were predicted to enhance oviposition through
increased production of chemicals stimulatory to onion flies (Dindonis and Miller,
1981; Hausmann and Miller, 1989). Investigating the effect depth of planting has
for stimulating D. antique oviposition on cull onions simultaneously addresses: 1)
how we should manage bulbous onion residues after harvest to minimize the
overwintering population, and 2) how we may best use these onions to manipulate
ovipositional behavior of spring adults. .-
Methods and Materials

Warmth: - large sprouting red
onion bulbs (ca. 6 cm diam., unknown cultivar) were obtained from Riley Farms,
Plainwell MI. To ascertain which planting depth stimulated the most onion fly
oviposition, these bulbs were planted so the neck heights were at 5 cm above (+ 5
cm), even with (0 cm), 5 cm below (-5 cm), and 12 cm (~12 cm) below the soil
surface

Plots were laid out in a commercial onion field (Bath, MI, Clinton Co.),
within the previously planted rows. Three rows, 50 meters apart, were used to lay
out experimental plots. Four replicates were placed in each row, for a total of 12
replicates in a randomized complete block design (RCBD). Blocks were spaced 20
meters apart and were located at least 20 meters from the edge of the onion field.
To enhance the element of fly ”choice,” it was desireable to maintain close proximity
and nearly equal distances between the four depth treatments. This was
accomplished by arranging treatments in a rectangle. Two treatments were located
end-to—end within one row, and the other two treatments were laterally located two
rows (1 meter) away.

Seeds and fonofos granular insecticide previously planted by the grower were
removed along with soil by digging trenches 25 cm wide, 2.5 meters long and 5, 10,

or 17 cm deep. Bulbs were planted touching each other in two rows 10 cm apart for

18

each depth. For the + 5 cm treatment, pesticide-free soil was transported from the
edge of the field to fill the 5 cm deep trench before the bulbs were pushed into the
loose soil. For the other treatments, bulbs were placed at the bottoms of the
trenches, then covered with pesticide-free soil. Blocks were planted between April
21 through 23.

Eggs deposited near cull onions were sampled on May 20-21, June 3-4, and
June 18. A plastic spoon was used to remove soil and eggs from the proximity of
onion plants. Two subsamples of eggs laid near three onions were taken from
within the middle third of the plots (to avoid edge effects). Eggs and muck soil were
stored in one liter wax paper cups for up to five days at 4° C until the eggs were
counted. Stand counts of bulbs with green foliage were taken May 20 and June 3.

Eggs were separated from muck soil by flotation in water. Residual floating
muck and eggs were passed through a 1 mm mesh sieve to remove the larger muck
fragments; the resulting eggs and fine muck were deposited on a nylon filter. This
' residue was washed into a black enamel pan. The contrast of the white eggs against
the black background enabled the eggs to be easily counted as they were removed
with a vacuum aspirator.

Egg counts required ln(x + 1) transformation to establish homogeneity of
variance (Bartlett's test, P > 0.1). Egg counts from each sample date were subjected
to analysis of variance and the means separated by the Student-Newman-Keuls' test
(Steel and Torrie, 1980). The average number of eggs per plant was multiplied by
the stand count to estimate the total eggs per treatment plot. These total eggs per

plot were then averaged to give average eggs per plot for each planting depth.

 

effects and interaction of bulb damage and larval infestation on onion fly oviposition
were studied via a two x two factorial design. Larval and adult onion flies were

obtained from a laboratory culture initiated in September, 1986, from pupae

19

collected from onions left in harvested fields in Grant, MI (Newaygo Co.). Flies
were housed in 80 x 65 x 60 cm screened cages provisioned with food (Schneider et
al., 1983), water, and ovipositional resources (Harris et al., 1987a). The rearing
room was held at 16:8 L:D photoperiod, 21 t 1° C, and 70 t 5% RH. Larvae were
reared on bisected onions. Flies used in this experiment were reared in the lab for
ca. eight generations.

Neonate larvae were obtained by collecting eggs from the culture cage by
flotation from sand, and holding the eggs on damp filter paper at 4° C. Under these
conditions, eg hatch is prolonged, so larvae were available for four to five days.

Commercially obtained yellow onion bulbs (unknown cultivar) were
refrigerated at 4° C from July 27 until September 14, 1987 to break bulb dormancy.
Bulbs were then planted in a 50 x 35 x 10 cm flat filled with steam sterilized muck
soil. Bulbs were positioned with their necks at the soil surface, and allowed to grow
within a 1.5 x 1.5 x 3.5 m screened cage in a greenhouse. Plants were kept within a
cage to prevent infestation with onion maggots prior to the start of the experiment.
On September 30, sprouted onions were then blocked by bulb size and foliar
development into groups of four. Two bulbs from each of the 1 1 blocks were
subjected to damage by sagittally slicing a 1 cm thick piece from the side of the bulb.
Single bulbs were planted so the necks were 5 cm beneath the surface of steam
sterilized muck soil held in 20 cm x 20 cm diameter pots.

Half of the bulbs from the damaged and undamaged treatments were then
subjected to larval infestation. Twenty neonate larvae were transferred with a fine
camel hair brush to the foliar-soil interface of one damaged and one undamaged
bulb treatment from a replicate, two weeks before an ovipositional bioassay.

Bioassays were conducted in a one meter x one meter diameter cylindrical
cage held in a growth chamber under the same environmental conditions as for

maintaining cultures. The cage had a plywood base and a suspended floor, into

20

which pots could be inserted. The soil level in pots placed in the cage was even with
the suspended cage floor. Sides of the cage were made of MylarTM plastic and was
ventilated at the top with aluminum window screen.

Fifty females and 50 male onion flies of mixed ages were transferred from
the onion fly culture to the bioassay cage provisioned with food and water. The four
treatment-combinations in one replicate of the factorial design were placed in
randomized order within the cage. After allowing the flies to oviposit for ca. 20 hr,
the pots were removed, dead flies were replaced, and the next block of pets was
introduced to the cage.

Eggs were sampled from the soil in the same manner as for Experiment 1,
except that all the muck to a depth of 4 cm was removed, and onion plants were
dissected to wash eggs from leaf axils. Egg counts did not require transformation

before analysis of variance.

Results and Discussion

WWW - Depth of cull
planting greatly affected onion fly oviposition (Table 1). F-tests for treatment
effects were highly significant both for the late May and early June eg sample dates
(F033) = 14.55 and 20.6, respectively; P < 0.0001). For the May 21 - 22 eg counts,
the rank order of preference (descending order) was -5, 0, -12, and 5 cm treatments.
The June 3 egg counts had a rank order of -12, -5, 0, and 5 cm treatments. On the
June 18 sample date, very few eggs were found and there were no significant
differences between treatments (F933) = 2.75, P > 0.05).

The preference of onion flies for these treatments probably was determined
by the number and quality of plant parts projecting through the soil surface. Cull
onions pressed into the soil surface (+5 cm) were the least accepted treatment.

These sprouting bulbs made a broad circumference of soil contact and were not

21

.36 v m J... a .32 22m. 320...... 2.53.2»: .2. can .23. use» a... .3 332.8 2.89.8 55.3 2.32.

 

 

 

 

 

 

a Q. 03.. a 9% NS... 9 .d ..~.~ 2.
a 0.. 036 a .dN 3: a E... «36 n-
a 9.. .3... 0 ad 03.. a néu mvma c
a v.n «3.. c ~.~ 2.... 0 new ~.~.. n
.30.). ammo :82 name :82 fine 383m :8
H .m 58:. .... .m :82 .... .m e. 2.3.3.
sun: .82
M: 2:: in 2:: --_~ .32
mamm— =82

 

dud a. 8...? 38.889: 2+5... .3 8.. echo Emcee: 2.... .333 9.3.9
E .323... 83.8 =3 .89. Ba. «.38.... go 3.. 2:3... A... .m. 608.888.1303 US. 38.88.... 2+5... 0... 3.. 2.33. .. 0.9a...

22

Table 2. Average stand counts and egg count totals for 2.5 meter
long double row plots of cull onions planted at varying depths.

 

Plants per plot Eggs per plot

 

 

Neck height
relative to 5/21 6/3
soil surface Mean (t 8.13) Sample Sample Total

 

5 26.8 (3.1) 111 173 284
o 24.6 (2.7) 681 851 1532
-5 18.1 (1.9) 859 1470 2329

-12 7.8 (12) 159 825 984

 

23

green, two characteristics that make poorly accepted host models (Miller and
Harris, 1985). Variation in acceptability among the remaining treatments probably
was due to the varying numbers of leaves projecting through the soil at different
sampling dates. The deepest planted bulbs were the most stimulatory, once their
leaves reached the soil surface. For the June 3 - 4 sample dates, this treatment had
an unsurpassed mean of 55 eggs per plant (Table 1). However, fewer bulbs in this
treatment had leaves reaching the soil surface; thus, fewer eggs were laid per plot
than for -5 cm (Table 2). Egg counts per plant were lower on -12 cm for the May
sample date (Table 1), probably because so few leaves per plant reached the soil
surface.

The decreased relative acceptance of bulbs planted at -5 cm compared to -12
may also have been due to the changing qualities of leaves at the soil surface. When
cull onion foliage first breaks through the soil, a large number of foliar resources
may be present upon which the onion fly may lay eggs. As the onion grows,
however, the floral stalk emerges and the leaves, which earlier were at the soil
surface, are well above ground. An onion thus changes from having a large number
of leaves that are a nearly optimal diameter for eliciting onion fly oviposition, to
having one or two large diameter (suboptimal) floral stalks. Based on these
observations, -12 cm on June 3 probably closely resembled the -S cm treatment on
May 21.

In addition to foliar cues, chemical stimuli influencing acceptability of culls
probably were changing. Colonization of bulbs by onion maggots and microbial
decomposition each initially increase the production of onion fly ovipositional
stimulants (Hausmann and Miller, 1989a; 1989b). Maximal production of these
volatiles probably was reached within two weeks of foliage having reached the soil
surface. After this time, bulb resources may have become depleted and

microsuccessional changes in bacteria would have diminished ovipositional

24

acceptability. By June 3, the acceptability of the -5 cm treatment probably were
diminished compared to the May 21 sample date.

WW - The mean

numbers of eggs laid upon each of the four treatment combinations were not
significantly different from each other (Table 3, F930) = 0.08, P > 0.1).
Partitioning variability into one degree of freedom main effects and interaction
suggested that neither damage to the bulbs, larval infestation, nor their interaction
caused significant differences in oviposition on bulbous onions (F0310) = 0.08, 0.09,
and 0.34, respectively; P > 0.1).

This experiment suggests that sprouting bulbous onions planted at an optimal
depth will remain highly stimulatory under highly variable conditions. This
greenhouse experiment also confirmed some personal field observations on the
range of ovipositional resources accepted by onion flies. Most strikingly, treatments
with foliage that was becoming flaccid fi'om maggot feeding remained highly
attractive to gravid onion flies as long as the leaves were green and moist. In two
replicates, foliage from the damaged + infested bulbs had become detached from
the bulb and had dried out; these plants did not receive eggs. However, when five
other pots (both damaged and undamaged bulbs) had been fed upon sufficiently to
cause leaves to detach and begin to dry, ovipositional activity was not diminished. It
is possible that increased chemical cues generated by feeding larvae offset lower
quality foliar cues, so that differences between treatments were not detected. This
experiment only ofiered a single time point for assaying ovipositional acceptance.
By allowing larvae to develop for two weeks, both damaged and maggot infested
bulbs may have passed their peak chemical acceptability to ovipositing onion flies
(Hausmann and Miller, 1989a; 1989b). Earlier bioassays could then have generated

treatment differences caused by damage and/ or infestation.

25

Table 3. Mean :t SE. Delia antiqua e laid per 24 hours on
variously treated cull onions1 in the la rato1y.

 

 

 

 

DAMAGE EFFECTS
INFESTATION
EFFECTS No Damage Damage
0 larvae 62 a: 16 a2 54 2 19 a
201arvae 59: 17a 66: 16a

 

1 2 x 2 factorial design with 11 replicates.

2 Means within the table followed by the same letter are not
significantly different, SNK test, P < 0.05.

26

General Discussion

Important features of bulb onions that elicit oviposition fiom D. antiqua are:
1) their sprouted leaves projecting through the soil surface, 2) large quantities of
characteristic volatiles from decomposing onion bulbs, which are known to attract
adults (Dindonis and Miller, 1981; Hausmann and Miller, 1989) and 3) crevices in
the soil that facilitate ovipositor probing (Mowry, et al., 1989a), generated by the
expansion of below-ground onion foliage. The upright, ca. 4 mm diameter yellow to
green colored leaves projecting through the soil surface closely resemble optimally
stimulatory surrogate onion foliage (Harris and Miller, 1991; et ante), and probably
had the greatest influence in eliciting oviposition.

Eckenrode and Nyrop (1986) found damage to bulbs to be of great
importance in colonization by the overwintering population of onion maggot in New
York State. Experiment 2 would suggest that damage to bulbs may increase
overwintering populations more through enhanced bulb suitability to onion flies
rather than through increased acceptability as an ovipositional resource. However,
the onions in the New York study were lifted from the soil, exposing damaged onion
tissues more than in Experiment 2.

To minimize the overwintering population of onion maggots, I recommend
from this study and other work (F inch and Eckenrode, 1985; Eckenrode and Nyrop,
1986), that bulbous onion crop residue should be damaged as little as possible, to
restrict colonization of the onion tissues, and should either be: 1) plowed deeply
under the soil, so that sprouting foliage will not reach the soil surface or 2) be
removed from the field and placed in deep cull piles (Finch and Eckenrode, 1985).
Leaving bulbs on the surface of the field is not recommended, because some bulbs
will sprout and will be colonized. However, this last option is probably better than
shallowly discing the residues. Damage to the bulbs during discing causes maximal
sprouting of bulbs and allows entry by maggots.

27

The preference of onion flies for deeply planted sprouted bulbs over small
seedling onions has been documented since the 1920's (Lovett, 1923; Hammond,
1924). Deeply planted sprouting onion bulbs elicit 60 to 200 times as much onion fly
oviposition as small onion seedlings (Hammond, 1924; Chapter 6; Mowry,
unpublished data). To maximize the number of eggs laid around cull onions used as
either a trap or a diversionary crop, bulbs should be planted so that the largest
number of leaves is emerging through the soil surface when seedlings need to be
protected. For the field work in this experiment, planting cull onions with their
necks 5 cm below the soil surface gave optimal conditions for deploying them as a

trap crop.

Chapter 2

Evaluating choice test methods for onion

fly ovipositional deterrents

28

Introduction

Investigators wishing to evaluate deterrents of insect oviposition by counting
numbers of eggs deposited on treated resources face further methodological
choices: the single treatment ("no-choice”) test offers the greatest assurance of
freedom from treatment interactions, and over time can address important shifts in
acceptance resulting from ovipositional deprival. Weaknesses are high demands on
time and insect stocks, and difficulties in comparing activities of various compounds.
By contrast, multiple-treatment (”choice”) tests offer the converse strengths and
weaknesses.

Experience in searching for deterrents of onion fly (Delia antiqua (Meigen))
oviposition (Cowles, et al., 1990) has led us to favor the ”choice” test in primary
" screening for the most deterrent compounds. Vast time savings can be realized
when testing is limited to discriminating dosages. Follow-up ”no-choice" tests with
the most promising deterrents can address the influence of ovipositional deprival.

Prokopy, et al. (1988), reviewed efficiency of choice bioassays for tephritid
flies, and concluded that multiple-choice assays with continual behavioral recording
were efficient and more sensitive (could detect ovipositional discrimination better)
than two-choice assays. Throughout that study, however, non-observed flies biased
deterrency measurements due to deposition of oviposition-deterring pheromone
during assays. Girolami, et al. (1981) ran ovipositional deterrent trials with Dacus
oleae using only two-choice arenas. They explained that in preliminary multiple-
choice trials, there were too few eggs deposited on deterrent treatments to conduct
analyses, because of competition from non-deterrent treatments. These two studies

illustrate complexities inherent with designing deterrent choice-tests;
2 9

30

communicational cues deposited during ovipositional activities, and the physical
number of treatment and control resources can limit and bias measures of
deterrency. The present study tested the quality of onion fly ovipositional choice-
test data. Specific questions of interest were: 1) Are measures of percent
deterrency in a ”choice test" influenced by the number of negative controls
included?, and 2) Do two-choice and multiple-choice dose-response assays give

similar deterrency estimates and precision?

Methods and Materials

Qnignflies - An onion fly culture was established in September 1987 from
field collected pupae and maintained as described earlier (p. 18 - 19). Flies used in
experiments were reared in the lab for 5-10 generations.

Deterrent - Methyl salicylate (98%, Sigma Chemical) was diluted in
polyethylene glycol (PEG)(Carbowax PEG 8000, Fisher Scientific) in five decade
steps to a 0.001% concentration (w/w). (E)-Cinnamaldehyde (99+ % purity) was
purchased from Aldrich Chemical and formulated in PEG at a 1% concentration.
PEG made an appropriate release matrix, as upon moderate heating it readily
dissolved these compounds and after cooling retarded their evaporation. Pre-
weighed PEG was melted at ca. 56°C, deterrent compound was added, and the
mixture was vortexed. The mixture was dispensed from a Pasteur pipette onto
aluminum foil to form ca. 10 mg pellets easily weighed for subsequent dilution steps.
These pellets were coarsely crushed, then stored at -18°C until bioassayed.
Experiments used 20 mg formulated mixture per ovipositional cup.

W - Ovipositional bioassays were conducted in a walk-in
environmental growth chamber having conditions identical to the rearing chamber
except the bioassay room was free of onion volatiles. Experiments were conducted '

in 30 cm diameter x 30 cm cylindrical top-ventilated plastic cages provisioned with

31

diet and water and stocked with ca. 10 male and 20 female flies. These cages
rotated on their axis once every 13 min; thus, environmental gradients were
distributed evenly over all treatments (Weston and Miller, 1985).

W - Each ovipositional dish consisted of 50 g white silica
sand (U nirnin Granusil, Grade 40, Oregon, IL) moistened with 3 ml distilled water,
then tamped into a 4 cm diameter x 4 cm plastic cup. Standing upright in the center
of each dish was a surrogate onion leaf consisting of a 0.4 cm diameter x 12 cm glass
tube painted onion-green with oil pigments (Winsor and Newton Cadmium Yellow,
Flake Everwhite No. 2, Winsor Green, and Lamp Black in the weight ratio
132:7:5z2) and coated with paraffin wax containing 0.05% n-propyl disulfide (Harris
et al., 1987). The surrogate onion was swiveled in the sand to provide a 1 mm space
around its base for ovipositor probing (Mowry et al., 1989a). This surrogate onion
foliage is competitive with similarly sized onion foliage (Harris and Miller, 1991),
and offers a highly controlled ovipositional resource. Deterrents were evenly
dispensed on the sand within 1 cm of the juncture of foliage and soil, where the most
critical host examining behaviors and, oviposition occur (Harris et al., 1987). PEG
granules applied on moist sand partly dissolved and became incorporated in the top
1 mm layer.

Wm - Four cups were placed in each of
the five rotating cages. Combinations were: 1) four control (blank PEG) cups, 2)
three control and one cinnamaldehyde-treated cup, 3) two control and two treated
cups, 4) one control and three treated cups, and 5) four treated cups.
”Combinations" 1 and 5 were included to contrast no-choice conditions with the
choice assays. Combinations 1 through 5 were randomly assigned each day to five
cages. There were ten replicates in this randomized complete block design

(RCBD).

32

WW - The dose-
response relationship for methyl salicylate was compared fiom two-choice (two

types of treatment per mge) versus multiple choice protocols. For the multiple-
choice test, the five decade-step concentrations of methyl salicylate and a control
(blank PEG) were randomly ordered in a single cage, constituting one replicate in a
five-replicate RCBD. To maintain the same fly densities in the cages, yet avoid fly
crowding on oviposition cups, the number of resources in the two-choice protocol
was kept the same as in the multiple-choice experiment; e.g., three treated cups
were compared with three control cups. Treated cups (one concentration per cage)
and control cups were alternated in a hexagonal pattern in the cages. Egg counts
within treatment and cage were pooled, constituting one replicate. There were
three replicates of each concentration tested, for the same five concentrations as in
the multiple.choice assay. The five replicates of the multiple choice series and the
three replicates of the five concentrations of the twoochoice tests were randomly
assigned to the five rotating cages over four days.

W - Log-transformed (ln(x+1)) egg counts fulfilled homogeneity
of variance assumptions for analysis of variance (Bartlett’s test, P > 0.1). Therefore,
deterrency was measured as the difference between ln(x+ l) transformed treatment
and control eg counts. This difference is readily converted to percent deterrency
(percent deterrency z {l-e“m+ ”hm“ 1)1} x 100%).

Results and Discussion

Will? - The ratio of treatmentzcontrol cups
significantly affected the logarithmic deterrency measurement (F (3 .27) = 8.58, P <

0.0001). Converting from logarithms yielded 87.6, 92.5, and 94.6% deterrency for
one, two, and three treated cups per cage, respectively (T able 1). The linear

33

Table 1. The effect of treatedzcontrol ratio on the measured ovipositional
deterrency of 1% cinnamaldehyde in PEG. Deterrency is the difference between
ln(x+ 1) transformed treatment and control eg counts.

 

 

 

Total Eggs
Treated:Control Deterrency Mean
Ratio Treated Control Combined (S.E) % Deterrency‘

0:4 - 1300 1300 - -

1:3 91 1692 1783 «2.085 (0.314)b 88 %
2:2 181 1856 2037 -2.591 (0.299)ab 93

3:1 369 1972 2341 .2925 (0.249)a 95

4:0 995 - 995 -0.246 (0.314)c 22

 

£Me3n65§ollowed by the same letter are not significantly difierent (SNK test, 9 d.f.,
< . .

34

orthogOnal contrast (Steel and Torrie, 1980) demonstrated systematic bias in the
deterrency measurements (F037) =6.19, P < 0.025).

This increase in measured deterrency with increasing treatmentzcontrol ratio
would not be expected to hold if the total number of cups increased so the one
control cup became less findable. This condition would probably mimic the no-
choice situation. Cinnamaldehyde at 1% concentration in PEG was only 22%
deterrent in a no-choice comparison (compared to the 90+ % deterrency from the
choice test).

A possible explanation for this increase in deterrency with increasing
treatmentzcontrol ratio is that ovipositing onion flies became more concentrated on
a single resource and that there is a ”group effect" where ovipositional behaviors of
flies or perhaps pheromonal substances associated with freshly laid eggs stimulate
further oviposition. This hypothesis is consistent with observations of intense
oviposition by aggregations of onion flies in culture cages. Two way analysis of
variance and linear, quadratic, and cubic orthogonal contrasts were conducted on
control eg counts to investigate this possible source of deterrency bias. If
pheromonal cues biased deterrency, a significant inverse linear relationship should
exist between the number of control cups and the number of eggs laid on the
controls. This linear orthogonal contrast was suggestive but not statistically
significant (F027) = 1.30, P > 0.25), so the source of the bias in deterrency remains
unexplained.

WWW - The standard
errors in the two-choice test were slightly smaller (Figure 1) than for the multiple-
choice test, even though there were two more replicates for the latter. The higher
precision in the two—choice test largely can be attributed to the pooled triplicate
subsamples.

35

 

E
8
o

e - MULTIPLE-CHOICE
- - TWO-CHOICE } i

on
O
r

a:

o

I l
r-—-.—-t

.b
O
1

l

N
O
I

 

 

 

MEAN PERCENT DETERHENCY (_-l_-_ S. )

O

0.601 0:01 0:1 7 1'0
PERCENT METHYL SAUCYLATE IN PEG

Figure 1. Dose-response relationships for methyl salicylate generated when all
deterrent concentrations are presented simultaneously in a bioassay chamber
(multi le-choice) or when single concentrations are compared to control cups (two.

choice . Mean deterrency and standard errors are from back-transformed
logarithmic values.

36

To compare dose response curves, the deterrency for each chemical
concentration was calculated separately for the two protocols, using the difference
between ln(x+ 1) transformed treated and control eg counts, then the two sets of
deterrency values were subjected to a paired t-test. There was not a significant
over-all bias in deterrency (t(4 df) = 0.901, P > 0.4). However, there was a trend
toward higher deterrency for intermediate concentrations in the two-choice assay
(Figure 1). This effect is biologically consistent with: 1) adaptation or habituation to
low concentrations of deterrent when presented with high concentrations, and 2)
greater ability of the flies to discriminate between treatments when they are
physically close, so that intervals are short between examining bouts on the different
"choices” (Mowry et al., 1989a).

Having multiple deterrent treatments in dose-response choice tests of
deterrents should cause increased acceptance of low concentration treatments.
Flies examining low-deterrency treatments after having visited high-deterrency
' treatments should be more likely to lay eggs than had she previously just visited an
untreated control. This influence could readily be tested by designing an
experiment in which each deterrent concentration in a multiple-choice assay is
alternated with an untreated control. If the cause of deviation from a two-choice
assay is deprivationally-caused, then there will no longer be differences in
deterrency compared to the two-choice assay. If differences in deterrency are
caused by habituation due to high ambient concentrations of deterrents, then the

differences between multiple-choice and two-choice tests will persist.

General Discussion
The results of choice are context-dependent, because measures of deterrency
are a function of the number and quality of all the host resources present. The

quantitative differences in deterrency shown in Experiment 1 were small, however,

37

and probably detectable only because of the high experimental precision.
Experiment 2 suggested that two-choice assays may eliminate underlying
interactions caused by either deprivation or habituation. Furthermore, two-choice
tests can give a standard response unaffected by the ratio of treatmentzcontol
ovipositional resources.

Multiple-choice assays have the advantages of allowing direct comparisons
between all treatments (rather than comparisons through a control), as is necessary
for conducting factorial designs. Also, multiple choice assays allow removal of
experimental variation due to differences in overall oviposition activity, and are less
labor intensive. The dose-response assays generated by multiple-choice and two-
choice tests converge at low and high concentrations of deterrents; this suggests
estimates of threshold responses and agronomically "practical" levels of deterrency
may be estimated at nearly the same deterrent concentrations with either bioassay
method. The advantages of multiple-choice assays are probably most readily
realized in discriminating-dosage bioassays (Cowles, et al., 1990). In a
descriminating-dosage assay, several treatments are compared via a multiple-choice
design. However, unlike the dose-response assay in this paper, concentrations are
kept low enough to prevent deprivation or habituation.

For the chapters to follow, multiple choice is justified as deterrency measures
were similar enough between two-choice and multiple choice assays to satisfy the

requirements of large chemical screening tests.

CHAPTERS

Deterrency of Pungent Spices and Capsaicin-containing
Products to Delia antiqua (Meigen)

38

Introduction

In accepting hosts, insects integrate via multiple sensory modalities a
diversity of external excitatory and inhibitory inputs with internal excitatory and
inhibitory inputs, that establish internal "physiological status” (Dethier, 1982; Miller
and Strickler, 1984). As for many insect herbivores, external excitatory inputs
affecting onion fly oviposition have been investigated much more extensively (Harris
et al., 1987; et ante), than external inhibitory inputs. Distinctive allelochemicals
(Levin, 1971) from non-hostplants could potentially interfere with normal
processing of host cues if brought into the region where intensive host examining
behaviors occur.

After ascertaining that various .spices deter D. antiqua oviposition in choice
tests, ground cayenne pepper was chosen for detailed study involving both choice
and no-choice conditions because: (1) a synthetic analogue of the principal flavor
ingredient was readily available, (2) commercial products, including an insect
repellent mixture available for field use, contain capsicum oleoresin as a principal
ingredient, and (3) at least in vertebrates, capsaicinoids cause stimulation of
chemoreceptors, heat and pain receptors (Virus and Gebhart, 1979), making animal
behavioral responses likely.

39

40

Materials and Methods
W - An onion fly culture was established in September 1986
from field collected pupae and maintained as described earlier (p. 18 - 19). Flies

used in experiments were reared in the lab for 3-5 generations.

WWW - Ovipositional assays were conducted in a

second walk-in environmental growth chamber having conditions identical to the
rearing chamber except this room was free of onion volatiles. ' All choice
experiments were conducted in 57 cm diameter x 57 cm tall cylindrical cages stocked
with ca. 50 male and 50 female flies provisioned with diet and water. These cages
rotated once every 13 min, thus experimental error was minimized by distributing
environmental gradients evenly over all treatments (Weston and Miller, 1985).

Ovipositional dishes and onion foliar surrogates were as previously described
(p. 31).

With the exception of the no-choice experiment, all tests were randomized
complete block designs, with treatments placed 5 cm fi'om the perimeter of the cage
in randomized order and blocked over time intervals, usually by day. Replicates
with combined treatment counts of less than 50 eggs were pooled to minimize
sampling error. Egg counts were not distributed normally, however log-transformed
(ln(x+ 1)) counts fulfilled assumptions for analysis of variance. A two-way analysis
of variance general linear models procedure (SAS Institute, Cary, NC) was used,
and means were separated with the Student-Newman-Keuls' Test for multiple

comparisons.

Bojanieals - Pungent spices were presented as choices in one cage, along
with a foliar surrogate control. Treatments consisted of 5-7 mg of each of the
following spices scattered within 1.5 cm of the surrogate onion foliage: crushed red

pepper, consisting of 3 x 5 mm flakes with seeds; chili powder, containing chili

41

pepper, onion, cumin, garlic, oregano, cayenne pepper, black pepper, caraway and
silicon dioxide; dill weed, flakes 1 cm x 1 mm; ground ginger; and coarsely ground
black pepper (R. T. French Co., Rochester, NY 14692). This experiment was
replicated three times, flies were allowed to oviposit S, 6, and 48 h, respectively for

each replicate.

Wes - Dose response choice tests were conducted in three
cages, each cage with various concentrations of one test material. Ground cayenne
pepper (GCP) (McCormick & Co., Baltimore, MD) or Sevana Bird Repellent
Powder (SBR) (Sevana Co., Fresno, CA) were applied in quantities of 0 (control), 1,
2, 5, 10, 22 and 46 mg, placed within 1 cm of the surrogate onion. Agrigard Insect
Repellent (AGR) (Sevana Co., Fresno, CA) diluted to 1, 3.2, 10, 32, 100 and 320 ppt
in deionized distilled water was sprayed (ca. 0.175 ml) on the sand surface with a
TLC atomizer.

Ovipositional deterrency of synthetic capsaicin was assayed in a dose
response choice test with 50 males and 50 females per cage. Synthetic capsaicin
(97% n-vanillyl-n-nonamide, Pfaltz & Bauer, Waterbury, CT) was dissolved in 95%
ethanol and diluted to 6.32, 20.0, 63.2, 200, 632, 2000 and 6320 ppm. Twenty ml of
each solution was added to 200 g white silica sand, mixed thoroughly and allowed to
air dry. Final concentrations were 0.316, 1.00, 3.16, 10.0, 31.6, 100 and 316 ug/g
sand. Each sand treatment was moistened with 10 ml deionized distilled water and

added in a 1 cm layer on top of clean sand in oviposition cups.

W - GCP was assayed in a no-choice context that employed a two
x two factorial completely randomized design that quantified effects of exposures
during both pre-reproductive and reproductive periods. Pre-reproductive exposure
was effected by placing 120 flies (not sexed, less than 24 hr post-eclosion) for 5 days

in a cage with food, water, and a foliar surrogate treated with 10 mg GCP. A second

42

group received similar treatment but no GCP. Females from these two groups were
transferred individually to 9 cm diameter screen sided cages and provided with 2
males, food, water and an ovipositional dish. Half of the females experiencing GCP
pre-reproductively were provided standard foliar surrogate; half were provided
foliar surrogate plus 10 mg GCP. Similar reproductive exposure treatments were
provided to females that had previously not experienced GCP.

Egg counts were taken daily from day 6 to day 15 post eclosion, to measure
days until first oviposition as well as daily oviposition. Onion flies in this experiment
tended to lay most eggs on alternating days, which hindered data analysis because of
zero counts. This problem was eliminated by pooling counts every two days. Egg
counts for flies that died during the experiment (18% of total flies) were not used in
the analysis, however their ovipositional record was included when comparing days
until first oviposition. Days to first oviposition data could not be analyzed by
parametric methods due to non-homogeneity of variance. These data were
therefore analyzed with the Kruskal-Wallis Test (Steel and Torrie, 1980); treatment
comparisons were then conducted with the Wilcoxon-Mann-Whitney Test (Steel and
Torrie, 1980).

We hypothesized that if habituation or adaptation to GCP had occurred in
the no-choice experiment, that flies with the greatest exposure would show
diminished GCP deterrence in a follow-up choice test, e.g., as when compared to
flies not exposed or exposed only pre-reproductively. Therefore, groups of 3
females were pooled fiom each treatment and placed in 9 cm diameter screen cages
with food, water and two oviposition cups, one with 10 mg GCP, one without. The

percent ovipositional deterrency was then compared for the 4 treatment groups.

WWW - Field trials of AGR and SBR

ovipositional deterrency to D. antiqua were carried out in 1986 in a commercial

43

onion field (Eaton Rapids, MI) with historically moderate onion fly population
pressure. Plots consisted of 3 m of treated row, separated by 3 m of buffer row, in a
linearly arranged randomized complete block design replicated 6 times. Blocks
were separated by ca. 35 m. SBR treatments were 0.05, 0.5 or 5 g per plot, applied
to the soil at the base of onion plants with salt shakers. AGR treatments were 1.5
mg, 15 mg, 150mg, 1.5 g or 15 g plus 15 mg Vaporguard (Miller Chemical Co.,
Hanover, PA), in 70 ml distilled water per plot, using Chapin compressed air
sprayers (Model No. 1 10, R. E. Chapin Manufacturing Works, Inc., Batavia, NY
14020), with Teejet 730077 flat fan nozzle at 20 psi (Spray Systems Co., North Ave,
Wheaton, IL 60188). Negative controls for these treatments were 70 ml water and
15 mg Vaporguard in 70 ml water applied in 3 m plots in each block. There were 9
applications made of SBR and AGR, on a 7 to 10 day schedule starting May 27.
Damage estimates in conventionally pesticide treated rows (liquid chlorpyriphos, ca.
1.1 kg A.I./ha) were made in 3 m of row 1 m laterally from untreated plots. Stand
' counts were made May 27 and at 2 week intervals. Analyses were conducted on
percent damage [(stand count on August 6 / stand count May 27) x 100%]; these
data did not require transformation.

Results and Discussion

Meals - In choice experiments, pungent spices all significantly deterred
onion fly oviposition (Figure 1). Mean percent ovipositional deterrencies of these
commercial spices were: paprika (88.6%), red pepper (95.9%), ginger (99.0%), dill
(99.3%), chili powder (99.8%), and black pepper (100%).

Black pepper, red pepper and ginger were pungent spices that showed great
promise in choice tests. Unfortunately, the pungent flavor components of black
pepper, predominantly piperine, are suspected carcinogens (Buchanen, 1978), while
gingerol, the pungent principle ingredient of ginger (Merck Index, 1976), is

44

relatively expensive. Red pepper was chosen for further studies based on its
relatively well known physiological effects on taste perception in mammals and the
availability of synthetic capsaicin. Capsaicinoids, the pungent principal flavor
ingredients in red pepper, cause bursts of activity from contacted chemoreceptors
and nociceptors (Virus and Gebhart, 1979). Capsaicinoids have some similarities to
warburganal, a potent Spodoptera exempta anti-feedant from the bark of an African
tree (N altanishi, 1980). Both capsaicin and warburganal have a pungent flavor and
are used as spices (Todd, et al, 1977; Nakanishi, 1980); in both cases these chemicals
cause rapid firing from chemoreceptors that then become unresponsive to
stimulation (Virus and Gebhart, 1979; Ma, 1977). It may be that such non-specific
activity at the level of gustatory chemoreceptors enables these spicy substances to

interfere with normal host acceptance behavior.

We; - In dose-response choice tests, oviposition of onion
flies was reduced 78 to 99% by the presence of 1 to 46 mg GCP (Figure 2) (F =

26.1; df=6,2A; P<0.001). Increasing quantities of GCP clearly caused greater
reductions in the number of eggs laid next to treated surrogate foliage. Agrigard
deterred oviposition at the highest rates tested (F = 10.0; df = 6,36; P<0.001), with
98% deterrence at 100 ppt and 100% deterrence at 320 ppt (Figure 3);
recommended field rates correspond approximately to 32 ppt, which was not
significantly different from the control. Sevana Bird Repellent showed no
deterrence, even at the highest rate tested (F = 0.58; df = 6,36). Synthetic capsaicin
incorporated into the top 1 cm of ovipositional substrate significantly deterred
oviposition when present at concentrations greater than 600 ppm (F = 35.4; df =
7,77) (Figure 4).

As summarized by Dethier (Dethier, 1947), many chemicals are deterrent or

repellent at high concentrations, including host-plant chemicals that at lower

45

 

 

 

 

500‘ .1
a < 4
U a
c 400-I .1
I"

E
‘8
O 300‘ "
I“ 4
fl
‘1’, 2004 .
z 4
<
:4; 100- b .
4 b
. . g . .,
c - — i
CON GIN DIL PPK RPP CHL app
TREATMENT

FIG. I. Number of eggs laid with 5-7 mg of the following spices placed within 1.5 cm
of surrogate onion in choice tests: CON. control: GIN. ginger: DIL. dill: PPK. paprika:
RPP. red pepper: CHL. chili powder: BPP. black pepper. Means followed by the same
letter are not significantly different (SNK test. 2 (if).

\\

“I I ‘ I 'I'lri j ' F

 

 

 

 

100.5 3 E

50-: :

4 l

m ‘ O l
D

0 101 1

5 so: 0 5

a. ‘ '1 ‘1

U) ‘ 4

0 1 .
8

2 1.0.: . I I

g 0.51 95%cr .3

a 4 1

r i

0.1. I

I Ԥ\ 1 I ' I V I 'U'I I ' I '
0 I 2 4 6 8 10 20 4O 60

WEIGHT (mg) OF GCP WITHIN 1 CM OF SURROGATE ONION

FIG. 2. Dose-response relationship for oviposition when ground cayenne pepper was
placed in choice tests within 1 cm of surrogate onion foliage. Straight line lit by eye.
Confidence interval based on 7 d].

46

 

 

 

 

so I I ' ' ' 'f"'I ' f """l i '4
1
1 a 1
‘ ab abc abc 4
I \
“- bc
:3 10 '\ 1
0 1 I 3
E 51 1
Q ‘ ‘
m 1
O i l
0
Ill
2 "‘ 1:86 cl
(
I.”
2 c
z’ \, 45
\C
0 . - . n" . . . n" v I‘
5 I i 5 1'0 5'0 160

SPRAY CONCENTRATION AGR (PPT)

FIG. 3. Dose-response curve for oviposition when substrate was sprayed in choice tests
with 0.175 ml of Agrigard Insect Repellent solutions. Standard error based on 6 df.

 

 

 

 

r ' I * fi 1 I v
a
100‘ I a . a -
g 8011 \a a d
0 so. \- .
C
III
a 40‘ \b d
O
G I
III
a 20- .
:1 1155
2
g 10- q
a 8T -
5.4 -c c.
."A V v I v I fiL
0 10 100 1000

CONCENTRATION (ppm) SYNTHETIC capsarcm IN SAND

Fro. 4. Dose-response relationship for oviposition with synthetic capsaicin placed within
the top centimeter of sand.

47

concentrations are involved with host acceptance. These dose response choice tests
were conducted to be certain that low rates of a promising material did not
stimulate oviposition. Any material showing stimulatory activity at low
concentrations would cause logistical problems when applied in the field, because
chemical decomposition would eventually decrease the concentration to levels that
could stimulate insect damage. Dose response experiments with capsaicin-based
products elicited no increased oviposition through the range of rates tested, unlike
dipropyldisulfide, which shows deterrent activity at high concentrations (Matsumoto
and Thorsteinson, 1968) and stimulatory activity with an optimum concentration in
surface wax of about 0.05% (Harris, et al, 1987). Direct visual observations of fly
behavior suggested that even at the highest rates tested, onion flies did not orient
away from GCP; so, the end result appears to be mediated by deterrency upon

contact rather than repellency.

W - In the no-choice experiment, there was no evidence that
GCP deterred oviposition (Figure 5). Pre-reproductive exposure showed no
significant effect on two-day egg counts (F = 0.19; df = 1,39); exposure during peak
reproductive activity also was not significant (F = 0.39; df = 1,39). There were
significant differences between treatments in days to first oviposition (Kruskal-
Wallis Test, “/13 = 1125, P< 0.025) (Steel and Torrie, 1980). Multiple comparison
of treatments showed only one significant difference, flies not exposed to GCP laid
eggs sooner (mean of 6.9 days) than flies of the reproductive-exposure-only
treatment (mean of 7.8 days) (Wilcoxon-Mann-Whitney Test, '1‘ = 89.5, n1 = 10,
112 = 14, P<0.05) (Steel and Torrie, 1980).

No—choice tests of deterrents are a rigorous test of deterrency, because
insects under these conditions face increasing internal excitatory inputs that can

override the presence of external inhibitory inputs. No—choice situations may

48

200 -,-,-

 

ISO-r

I00-

CUMULATIVE EGGS PER FEMALE

 

 

 

 

  

u 2 4 3 5 r10 - l'z . 1'4 firs
DAYS POST ECLOSION

no.5. Cumulative oviposition for 2 x 2 factorial experiment investigatlng prerepro-
ductrve and reproductive exposure to GCP. --O--. no exposure: -- C --. exposure
only prereproductively: - - A --. exposure only reproductively; —C -. continual expu-
sure.

100
901

.0: I...

70« ' a

ecu a

 

50- ab
40..
sol
zol
lO-r

MEAN PERCENT DAMAGE
a

LlAlgLJlilAlAlA‘L‘L

 

 

 

WVGls rs Isorsoorsoooso soosooo
CONT AGR (mg) SBR (my)

TREATMENT

FIG. 6. Percent damage to seedling onions by onion fly. W. water control: VG. Vapor-
guard control: AGR. Agrigard: SBR. Sevana Bird Repellent powder: CHE. conventional
chemical control (0.05 g chlorpyriphosxm). Amounts given are for 3 m of row. Means
followed by the same letter are not statistically different (SNK test. 5 df ).

I
m

49

simulate some field situations (where alternative acceptable ovipositional substrates
may not be available), and also assist in determining whether choice test deterrency
is caused predominantly by sensory vs. sublethal physiological effects (Landis and
Gould, in press). Our no-choice tests effectively ruled out the possibility that effects
of GCP seen in choice tests were caused by sublethal toxicity, since all no-choice
groups laid similar numbers of eggs. This is in contrast to preliminary screening
with pyrethroids, in which ”deterrent” concentrations were accompanied by
convulsions or tremors in onion flies (unpublished data).

In the choice test bioassay of flies previously used in the no-choice
experiment, all groups laid similar percentages of eggs on the control (84 i 2.7%,
mean .+. SE) (F = 0.49, df = 3,6), implying that these groups retained the same
ability to sense and respond to the presence of GCP. Whether habituation or
adaptation were responsible for the lack of differences in the no-choice experiment
is still unclear; either dishabituation occurred rapidly or else another mechanism
‘ was responsible for acceptance of GCP treated ovipositional substrates. Perhaps

flies simply tolerated deterrents because they were becoming deprived.

Eieldjjals - Field tests of AGR and SBR in 1986 for the most part agreed
well with laboratory studies, however the trends in field data were not statistically
significant (Figure 6). There was a trend toward higher mean damage for all three
rates of SBR in the field than in the control plots. These field results suggested that
SBR may have stimulated oviposition, in contrast to the lack of response
(stimulatory or deterrent) to SBR in the laboratory. Under the wet field conditions
experienced in 1986, there exists the possibility that SBR, which consists of 10%
ground red peppers and 4% ground garlic, could have decomposed to form
microbial products stimulatory to onion fly oviposition (Coley-Smith and King, 1969;
Dindonis and Miller, 1980).

50

The most highly concentrated sprays of AGR suppressed damage to a level
intermediate to the untreated controls and the conventionally pesticide treated
areas. These sprays were irritating to the applicators, and at elevated rates caused
onions to have stunted, yellow foliage. These field results with AGR agree with the
laboratory studies that indicated that ovipositional deterrency only occurred at

concentrations exceeding the field recommended rates.

General Discussion

Ovipositional and feeding deterrents should be tested with hosts or host
models that accurately reflect normal sensory input (Sta dler, 1983). Using natural
stimuli allow normal sequences of behaviors to proceed. Allowing the full
repertoire of behaviors involved with host acceptance to be expressed implies that
each behavior has an Opportunity to be influenced by the presence of deterrents.
When using highly artificial substrates, internal excitatory inputs may override
. inputs associated with normally non-stimulatory substrates to provide abnormal
behavioral responses. An example demonstrating how ovipositional substrates can
influence results while studying ovipositional deterrents is the recent work of Tingle
and Mitchell (1986). Their data show that in choice tests, elderberry extracts were
more deterrent to Heliothir virescens when applied to tobacco (a preferred host)
compared to a standard paper towel substrate.

Ovipositional deterrents have not previously been tested for the onion fly
with host models. Wiens, et al, (1978) were able to reduce oviposition by ca 78%
over controls in choice tests using hydrated bean (Phaseolus vulgaris) extracts.
Alfaro, et a1 (1981), using cedar (Thuja plicata) leaf oil found an 84% ovipositional
deterrency. Both of these experiments used inverted beaker ovipositional bioassays
(Vernon et a1, 1977). Deterrency assays with pine oil (J aver et al, 1987) similarly

used halved onion bulbs rather than onion foliage or foliar models.

51

External excitatory inputs for the onion fly, Delia mrtr'qua (Meigen), involve
synergism between visual and chemical stimuli (Harris and Miller, 1982). Yellow or
green vertical cylinders ca 4 mm diameter that emit n-dipropyl disulfide from
surface waxes elicit pre—ovipositional behaviors, including foliar runs, ovipositor
examining of foliage and soil, and ovipositor probing (Harris, et al., 1987). These
surrogate onions provided a highly standardized ovipositional resource that
accurately simulated host stimuli. We suggest that they are highly suitable and
convenient for studying ovipositional deterrency.

Loss of GCP deterrency in a nO-choice situation is an excellent
demonstration of the dynamic influence physiological state has on the acceptance of
otherwise deterrent hosts (Dethier, 1982; Miller and Strickler, 1984). Other
examples of how deprivation affects oviposition are available from fruit fly
literature. Fitt (1986) deprived three species of Dacine fruit flies (Family
Tephritidae) for 4 to 7 days and observed the host range that flies oviposited into.
In Dacus tryoni Frogg, a rapid increase in egg load was accompanied by a propensity
to accept Solanum maruirianum fruits which normally would not be acceptable. The
other two species tested, D. jam'si and D. cacamiriatus apparently have a
physiological feedback mechanism that allows resorption of eggs to take place; these
species did not show altered host acceptance patterns when deprived. Effects of
deprivation of suitable hosts mu affect ovipositional preferences over a much
shorter time scale. Roitberg and Prokopy (1983) found that Rhagolerir pomonella
accepted hosts marked with ovipositional deterring pheromone following only a 5-10
min deprivation of fruits.

Loss of GCP deterrency in no-choice situations also indicates that there may
be great difficulties in obtaining control of onion flies in the field. It may be
necessary to make highly accepted alternative hosts available, in the form of trap or

diversionary crops, so that flies do not enter an ovipositionally deprived state. The

52

low effimcy of capsaicin-containing products in 1986 field tests may have been due
partially to extremely wet weather or to the lack of a such an alternative
ovipositional resource.

While pungent spices appear to be highly deterrent materials for onion fly
oviposition in laboratory choice tests, and could perhaps be of some use in the small
home garden, I do not feel that they hold much promise for commercial field use.

Chapter 4

Cinnamyl derivatives and monoterpenoids as
non-specific ovipositional deterrents of the onion fly

53

Introduction

Previous work with ovipositional deterrents for Delia flies has largely focused
on mixtures of compounds from essential oils or other plant products. Havukkala
(1982) had little success in field trials using turpentine-soaked stakes to control D.
radicum, while Den Ouden and Theunissen (1980) achieved control of D. bram'cae
approaching standard chemical treatment in trials using slow-release formulations
containing 2 - 8 g naphthalene per plant. Wiens et al., (1978) reported that hydrated
bean seed extracts reduced onion fly oviposition by ca. 78% over controls in choice
tests. The active constituent was a non-volatile gustatory deterrent and was not
further characterized. Alfaro et al., (1981), found that high levels (300 pg per
ovipositional site) of red cedar leaf oil deterred D. antr'qua oviposition by 84%. This
activity was attributed to thujones, the major monooxygenated monoterpenoid
components of this oil. Pine oil, a mixture of terpenes and oxygenated terpenoids,
significantly reduced onion fly oviposition in a no-choice situation when 5 mg was
applied per onion bulb (Javer et al., 1987).

Cowles et al., (1989) showed that black pepper, ginger, dill, and various
materials containing red pepper deterred onion fly oviposition. Synthetic capsaicin,
an analog of the principal flavor ingredient of red peppers, deterred oviposition by
95% when present at 320 ppm in the top cm of sand. Red pepper's efficacy
disappeared in no—choice conditions, however, and was not effective in field tests.

In this study, I quantified the ovipositional deterrency of a range of cinnamyl,
cinnamoyl, monoterpenoid, and phenethyl alcohol derivatives presented around

surrogate onions exposed to D. antique: in the laboratory. The goal was to explore

54

55

structure-activity relationships and to target those compounds most promising for

field assays.
Materials and Methods

Wm - Onion flies were maintained as described earlier (pp.18-
19) in a culture started in 1987. Flies used in all experiments were reared in the lab

for 3-10 generations.

Chemicals - Most cinnamyl, cinnamoyl, and phenethyl derivatives were
purchased from Aldrich Chemical Co., (Milwaukee, WI, 53233) or synthesized
(Szurdoki et al., in prep.) in the Plant Protection Institute (Budapest, Hungary),
according to published methods.

Compounds were diluted in polyethylene glycol as described earlier (p. 30),
in five decade steps to a 0.001% concentration (w/w). Experiments used 20 mg
formulated mixture per treatment.

Bioassays - Ovipositional assays were conducted in a walk-in environmental
growth chamber having conditions identical to the rearing chamber except the
bioassay room was free of onion volatiles. Experiments were conducted in 30 cm
diameter x 30 cm cylindrical cages provisioned with diet and water and stocked with
ca. 10 male and 20 female files. These cages rotated on their axis once every 13
min; thus, experimental error was minimized because environmental gradients were
distributed evenly over all treatments (Weston and Miller, 1985).

Ovipositional dishes and surrogate onion foliage used in bioassays are
described on p. 31.

For initial screening, the deterrency (percent reduction in the number of eggs
laid on a treatment vs. the control) of ca. 1 mg of neat compound was compared '

with an untreated control. Four treatments were tested simultaneously per cage. If

56

there was at least 80% deterrency in the first replicate, the compound progressed to
dose-response analysis. If there was less than 80% deterrency recorded after three
replicates, the BR90 (concentration required to elicit 90% behavioral response) was
considered unattainable and the compound was listed as inactive.

Dose-response choice tests were conducted in a randomized complete block
design. A replicate consisted of egg counts from six ovipositional resources (5
decade-step dilutions of a compound in PEG, plus a PEG control) placed 5 cm from
the perimeter of the cage in randomized order and blocked by day. Five compounds
were tested each day (one chemical in each of five cages) and flies were used once
per compound; this ensured reasonable replicate independence and minimal fly
learning. Replicates with combined treatment counts of less than 50 eggs were
discarded to minimize sampling error. Every concentration series was replicated ten
times. Phenethyl alcohol was tested separately from the other compounds listed in
Table 1. It was tested two days apart; all five cages were simultaneously used to
obtain the ten replicates.

Egg counts were not distributed normally; however, log-transformed
(ln(x+ 1)) counts fulfilled homogeneity of variance assumptions for analysis of
variance (Bartlett's test, P > 0.1; excluding treatments with all zero egg counts and
therefore zero variances). A two-way analysis of variance (general linear models
procedure, SAS Institute, 1985) generated the treatment variance (mean square
error). Ninety-five percent confidence limits were calculated for percent eggs
relative to the control for each concentration in the dosage response relationship for
a compound. A procedure analogous to calculating the least significant ratio (LSR)
(Steel and Torrie, 1980) was used, except that the confidence limit values
(mean x /+ eon /2 JW’») were calculated. By connecting the upper limits, and
similarly connecting the lower limits, a 95% confidence belt was constructed for
each dose-response relationship. The intersection of the confidence belt with the

57

90% deterrency response graphically estimated the concentration of chemical
eliciting 90% ovipositional deterrency (BR90) for a compound (Figure 1). This
graphical method was used instead of regression analysis because: 1) dose response
functions were non-linear, 2) there were few deterrent concentration levels;
therefore the regression degrees of freedom were low and confidence limits inflated,
and 3) the graphical method makes better use of the knowledge that at high
concentrations behavioral response converged towards absolute deterrency,
generating zero variance. Differences between apparently stimulatory responses of
compounds at low concentrations and the control treatment were compared via the
Student-Newman-Keuls' test.

Upon establishing that only the most active compounds at 0.1% were
detectably different from control PEG, this concentration was chosen for conducting
discriminate-dosage bioassays. In one large experiment, the activity of 35 additional
compounds was ranked. Four randomly chosen materials and two blank PEG
' controls were placed in each of five rotating cages; cumulative presentation of all
compounds constituted one replicate. 'Ihirty female and 15 male flies were used
for each of the four replicates of this modified randomized complete block design.
Data were analyzed under the assumption that the number of eggs laid on each
treatment relative to the number laid on the control is independent of the presence
of other treatments (see pp. 36-37). This allowed direct comparison of each
compound to the control via a paired t-test. When these data were log(x + 1)
transformed, the value used to test the null hypothesis, [TRTo(CON1 + CON2)/2]
(SAS Institute, 1985), was a convenient transformation of percent acceptance,
(TRT/CON)x100%, since subtraction with logarithms is equivalent to forming this
ratio. One slight difference is that averaging the two blank PEG controls in the
former equation is equivalent to taking the geometric mean of the non-transformed

counts. Using the difference between log-transformed egg counts satisfactorily

58

 

0|
0

b
O

PERCENT EGGS
re
a

(Treatment/Control)“ 00
cc
0

.a
O

 

 

O

 

 

 

CONCENTRATION IN PEG (%)

Figure 1. Method for determining the BR90 from the dose- response relationship, as
demonstrated with cinnamonitrile (see text for further explanation).

59

transforms, and is symmetrical, for both deterrent (negative values) and stimulatory
treatments (positive values). These data were then transformed to percent

deterrency (or stimulation), for presentation in Table 3.

WWW - Ovipositional CUPS were
prepared with 20 mg of 1% (w/w) (E )-cinnamaldehyde or (E)-cinnamonitrile in
PEG placed on moistened sand, but without an onion foliar surrogate. One cup was
placed in each of four cylindrical cages (the same as used for bioassays), and
allowed to equilibrate in the 21° C conditions for one hour. A one ml air sample
was withdrawn over a 45 sec interval with a Hamilton gas-tight syringe needle
positioned within 5 mm of the PEG granules. The syringe needle was immediately
inserted into a gas chromatograph injector port. The syringe was heated for 30 see
with a hand-held hair dryer, to assure that the vapors did not condense on the
syringe wall, then the plunger was immediately depressed to inject the sample. The
Varian 3700 gas chromatograph was equipped with a 30 m x 0.32 mm (ID) DB-5
capillary column. Running conditions were 120° C isothermal, 1 ml/min He carrier
gas flow rate, with a flame ionization detector and HPBB90A integrator. Standard
curves were generated by injecting 1, 2, or 3 ng of each of the two test compounds in
1 to 2 ul of C82 (3 replicates).

Results

Wes - Among cinnamyl related compounds (Table 1), allyl
benzene (1 ), (E)-fi -methyl styrene (2), (E)-cinnamic acids (3, 4, 5 and 6), and (E)-
cinnamamide (7 ) were less than 80% deterrent in the initial screening, so the BR90
was not attained. The mean BRgo's for active compounds ranged from 0.38% in
PEG for (E)-4-methoxycinnamaldehyde (20) to 3% for (E)-cinnamaldehyde
dimethyl acetal (8 ).

60

Table 1. Estimated mean and 95% confidence limits of the BRQO ovipositional deterrency of
cinnamaldehyde derivatives.

 

 

bop-IPM. W
[OCH/{mm} Source.
Compound (inn) (°c1 Purity Structure Mean 31190 95% c. r.
1 allyl benzene 156 A, 98% 0f Not active
2 (5)18 -methyl styrene 175 A, 99 W Not active
3 (E)-cinnamic acid 300 A,99+ sz“ Not active
4 (E)-4-nitrocinnamic acid (285-286) H, 93 cot" Not active
dec. 0;: < >
s (E)-3,4-methylenedioxy- (242-243) H, 97 r° co,"
cinnamic acid dec. (LG-r Not active
6 sinapinic acid (203-205) A, 98 C1430 co rt Not active
3,5-dimethoxy-4— dec. “DJ— 1
hydroxycinnamic acid 0’
[predominantly (5)] ”‘1
7 cinnamamide (147) A, 97 W com-r, Not active
[predominantly (5)]
8 (ED-cinnamaldehyde OCH:
dimethyl acetal 257' 1,90 °°‘a 3.0% 1.3 - 5.0%
9 phenylpropargyl ‘ 0013C":
aldehyde diethyl acetal 110-115/15 H, 95 < > - ( 2.5 0.92 - 5.6
0042013
10 a -pentyl '
cinnamaldehyde 287 A, 97 2.6 0.99 - 4.8
CHO
3" 00110-13

1 1 (2)-a -bromocinnam-
aldehyde diethyl acetal 145-150/5 H, 95 oer-rial, 2.1 0.91 - 42

C110

12 (E)-cinnarnaldehyde 248 A, 99 W 1.0 0.60 - 2.8
crraotr

13 (E)-cinnamyl alcohol 250 A, 98 W 0.88 0.60 - 1.8

14 ethyl cinnamate 271 A, 99 W 0.80 0.50 - 1.4
[predominantly (5)] m’

CHO
15 hydrocinnarnaldehyde 224 A, 95 Gr 0 0.84 0.50 - 1.2

“5 (IO-methyl cinnamate 260 A, 99 {DJ-«OCH: 0.70 035 - 0.96

 

Table 1 (cont'd.).
17 2-phenylvinyl formate N, 41 Wopo 0.50 0.13 - 0.70
H
' Br

18 (2)4 obromo CHO

cinnamaldehyde (67-68) H, 98 0.44 0.22 - 0.64
10 cinnamonitrile 255 A, 97 CN 0.43 0.15 - 0.74

[predominantly (5)] W
20 (E)-4-methoxy CH G10

cinnamaldehyde (59-61) H, 95 ’° 038 0.16 - 0.60

crrrorr

21 2-phenethanol 220 A, 99 ©——’ 032 0.12 - 0.54
' Boiling point corrected to 760 mm.

Chemical sources: A, Aldrich Chemical, Milwaukee, WI, 53233; H, Plant Protection Institute, Budapest.
Hungary; 1, Internat. Flavors and Fragrances, Inc., NY, NY; N, Dr. M. G. Nair, Horticulture Dept,
Michigan State Univ.

62

Ovipositional deterrent activity among terpenoids paralleled that of cinnamyl
derivatives with similar functional groups (Table 2). The non-oxygenated
monoterpene p-cymene (22), having the same carbon skeleton as terpenoids 24, 25,
and 27 - 30 in Table 2, was inactive in preliminary screening. The remaining
oxygenated compounds had overlapping 95% confidence limits for their BRgo's, so
their D. antiqua ovipositional deterrency did not differ statistically from each other.

WW]; - Compounds tested at 0.1% concentration are
ranked by increasing deterrency in Table 3. Because compounds were relatively
dilute and variability was high, only the most deterrent compounds were
significantly different from the control. Mean percent deterrency given in
parentheses (compounds 31 - 37) reveals instances when more eggs were laid on
the chemical treatment than the control. Though none of these compounds was
statistically different from the blank PEG, the 67% mean percent stimulation with
benzalpinacolone (31) suggests that this compound is a likely ovipositional
stimulant. Similarly, the close agreement of ca. 34% increase in numbers of eggs
with 4-nitrophenethyl alcohol and (E)-4-nitrocinnamyl alcohol suggests that they are
possible stimulants at 0.1%. Compounds 34 through 50, with standard
errors approaching or overlapping 0% deterrency, are probably neutral at the tested
concentration. The remaining chemicals in the table were variably deterrent.
Among those that were significantly different from the control, (E)-cinnamaldehyde
dimethyl acetal (54) was 58% deterrent (lowest) while hydrocinnamonitrile (64) was
88% deterrent (highest).

WW - The concentration of active
compounds in air was quantified by gas chromatography. (E )-Cinnamaldehyde at its
BR90 (1% formulation in PEG) was (present at 1.7: 0.3 ng (mean :1: S. E.) per ml of

air. (E)-Cinnamonitrile is more active than cinnamaldehyde, so the 1% formulation

63

Table 2. Estimated mean and approximate 95% confidence limits of the BR90 ovipositional

 

 

deterrency of monoterpenoids.
W
b.p./pres. Source,
Compound Per/[m] Purity structure Mean 81190 95% c. I.

22 p—cymene 176-178 A, 96% <:> < Not active

4-isopropyltoluene .

O

24 menthone 210 A, 75 _d_< 1.0 080 - 2.8

[3% isomenthone]
25 carvacrol 36-37 A, 98 "0 0.90 0.62 - 1.7

S—isopropyl- a : > <

2-methylphenol
26 citronellol 222 G, 92 -<-\=< 088 0.72 - 1.2
011011
OH

27 menthol 216 A,99 < 5 < 086 0.64- 1.0
28 a -terpincol 217 A, 98 ‘O'ILW 085 052 - 24
20 cumic alcohol 135-136/26 A, 97 "00,,@_< 0.66 0.40 - 085

4-isopropylbenzyl

alcohol

a We... .97 m o... .12....

 

Chemical sources: A. Aldrich Chemical, Milwaukee, WI, 5333; G, Givaudan Corp, Clifton, NJ, 07014;
S, SCM OrganicChemicak,JacksonviIle.FI..32201

64

would give greater than 90% deterrency; its vapor concentration was measured at
2.3 :1: 0.3 ng/ ml.

Discussion

AdxamaacLQLdiicnnfinatedcsagehinassay - Besides being able to select

compounds with high activity, the discriminate-dosage bioassay could be completed
approximately eight time faster than a full dose-response series. Furthermore, being
able to test a large number of compounds in one experiment permitted a larger
number of compounds to be compared directly. The 90% deterrency estimate for
(E)-4omethoxycinnamaldehyde (20), which was included as a positive control in half
of the discriminate~dosage replicates, agreed well with the estimated BR90
concentration of 0.4% from the dose-response assays.

A large amount of structure-activity information was Obtained by combining
data from dose-response and discriminate-dosage bioassays. Instead of using the
B1190 values from Tables 1 and 2, deterrency observed at the 0.1% PEG
concentration in dose response assays was directly compared to the discriminate-
dosage tests. 7

Wins - Key features investigated in this study were:
1) unsaturation and length of alkyl side chains, 2) functional groups at the end of an
alkyl side chain, 3) ring substitution, and 4) presence of an aromatic ring.
Unsaturation and the accompanying conjugation between the aromatic ring and
carbonyl in cinnamaldehyde were not necessary for activity. (E)-Cinnamaldehyde
(12) and hydrocinnamaldehyde (15) had similar BRgo's of about 0.9% (Table 1),
and had similar dose-response profiles. Hydrocinnamonitrile (64) and
cinnamonitrile (19) had similar deterrency (ca. 85%) at the 0.1% concentrations.
However, the alkyne bond in phenylpropargylaldehyde (59) and its dimethyl

65

Table 3. Estimated percent ovipositional deterrency (stimulation) for compounds tested at 0.1%
concentration in PEG.

 

 

b-p-/prc& W
l°Cl/lmml Some.
Compound . (m.p.)[°C] Purity Structure Mean SE.
31 hemlpinncoroue (41-43) H, 95% ° (67)% (87)-(16)%
2,2-dimethyr-5-phenyr- M
4-penten-3-one

C1130"
32 4-nitrophenethyl alcohol 177/16 A, 99 o.u—©_/ (37) (Sm-(21)
33 (E)-4-nitrocinnamyl alcohol (122-125) H, 90 ”@17— °"‘°" (31) (47)-(10)

34 (2)-a -bromocinnamalde- 3' 0

hyde ethylene acetal H, 95 W0] (26) (42)-( 5)
35 (£/Z)-3,4-dirnethoxy ““0 c"

cinnamonitrile A, 97 “”‘W (26) (34)-(17)

CHO
36 (E)-4-nitrocinnnmaldehyde (139-141) H, 95 WWW (14) (47)- 28
37 (E/Z)-4-isopropoxy-3-
methoxy-cinnamonitrile H, 95 .6‘GJ— c" (5) (28)- 20

(E/Z - 1.1)

38 (5)-cinnamyl bromide 100-105/20 H, 90 4 (32). 38

W
39 naphthalene P, 99 6 (35)' 44

40 (S)-(+)-methyr mandelate 171) A, 99+ Qflm'c" 8 (53)- 60

OH
41 (E)-6-phenyl-2-hexenal H, 96 mam 11 (14)- 32
42 (5)-4-pheuy'r-3-huteu-1-ol H, 95 Warm 21 (14)- 46
43 (R)-(-)-methyl moderate 171) A, 99+ <:>_ 4:” 21 (15)- 48
mic-Hi
44 (E/Z)-6-phenyl-5-hexenal H, 90 (3 fl P‘O 22 (20) 51
(E/Z - 8:2)

ation
4s 4-methoxyphenethyl alcohol 334 A, 99 Caro—Q—J 23 (10)- 47

Table 3 (cont'd.).

46

47

49

51

52

57

61

62

ferulonitrile

(E/Z)-4-hydroxy-3-methoxy-
cinnamonitrile
(E /Z - 1:1)

(5)-methyl 4-nitrocinnamate

(E)-6-phenyl-2—hexen-1-ol
homovanillyl alcohol

(E/Z)-4-acetoxy-3-methoxy-
cinnamonitrile
(E/Z - 1.4:1)

(5)-methyl 3,4-
methylenedioxycinnamate

fl -phenethyl formate

benzalacetone
(E)-4-phenyl-3-buten-2-one

(5)-cinnamaldehyde
ethylene acetal

4-chlorOphenethyl alcohol
phenethyl acetate
2-cyclohexylethanol

3,4-methylenedioxy
cinnamyl alcohol

phenylpronarylaldehyde
(E)-5-phenyl-4-pentenal

cinnamyl formate

4-methylphenethyl alcohol

66

160/1 H, 95
(158-161) H, 90
H, 91
A, 99
H, 95
(131-133) H, 93
P, 95
(38-41) H, 97
H, 95
310) A, 99
232-235 H,97
207 A, 99
(71-74) H, 97
118/13 A, 96
11,92
P, 97
224 A, 99

59
62’
63
64‘

69..

71‘
76‘

78

(22)- 56

(12)- 44

(10)- 56

1-52

4-51

38-65

43-69

21-79

45-73

64-71

68-70

60-79

74-77

50-90

67

Table 3 (cont'd.).
63 1-nonanol 215 A, 99 NWCHzOI-I 81‘ 68 - 88
64 hydrocinnamonitrile 2501 A, 99 W CN 88‘ 78 - 93
65 (E)-3,4-methylenedioxy- (O crro
cinnamaldehyde (84-86) H, 95 OW 88 70 - 95
20 (E)-4-methoxy- (59-61) H, 95 C30 90” 86 - 92
cinnamaldehyde Gho'Q—r

 

‘P < 0.05, ”P < 0.001; for t-tcsts of ln(x+ 1) treatment eggs versus controls, n-4 for all compounds
except20, n-20 for20.

1 Boiling point corrected to 760 mm.

Cinnamaldehyde: 3-phenyl-2-propenal; cinnamic acid: 3-phenyl-2-propenoic acid; cinnamyl: 3-phenyI-2-
propen-l-yl

Chemical sources: A, Aldrich Chemical, Milwaukee, WI, 5333; H, Plant Protection Institute, Budapest,
Hungary; P, Pfaltz and Bauer, Inc., Waterbury, CT, 06708.

68

acetal (9) appeared to enhance activity compared to the saturated or alkene
compounds (15, 12, 8).

Increasing the length of the alkyl side chain of aromatic compounds
appeared to decrease activity. (E)-5-phenyl-4-pentenal (60) was deterrent while
(E)-6-phenyl-5-hexenal (44) was not. An alkyl side-chain also decreased activity of
10 vs. 12. Phenethyl alcohol (21 ), included as a comparison to cinnamyl alcohol
(13), had the highest activity (BR90 or 032%), suggesting that a two-carbon side
chain is no less active than a three-carbon side chain. Part of the loss in activity with
alkyl group size is probably due to the attendant decrease in volatility. The vapor
pressure of these compounds is inversely related to the boiling points given in the
tables.

Moderately polar functional groups appear to be required for ovipositional
deterrency. Hydrocarbons (1 , 2, 22) were inactive, at the extreme for non-polar
compounds. Carboxylic acids (3 - 6) and cinnamamide (7) are relatively polar
‘ compounds, and also were inactive. Functional groups associated with deterrency
were ketones, aldehydes, alcohols, nitriles, and esters. The contrast between
aldehydes and alcohols (12 vs. 13, 23 vs. 26), seems to indicate that alcohols had
greater activity.

Modifications of aldehydes included formation of their acetals and a -
bromination. Acetals (8, 11 , 34, and 54) were investigated because they might
allow slow chemical release (Pickett et al., 1984) of the parent aldehydes (12, 18).
Acetalation reduced activity; this could be attributed to reduced fit to a receptor
and/ or decreased volatility. a-Bromo derivatives (18, 11) were more active than
the non-halogenated compounds (12, 8).

Ring substitution at the para position of phenethyl alcohol (21 , 32, 45, 55,
62), decreased deterrency in the following order: unsubstituted, methyl, chloro,

methoxy, nitro. Nitro-substitution converted deterrent phenethyl alcohol (21),

69

cinnamaldehyde (12), and cinnamyl alcohol (13 ), into possibly oviposition-
stimulating chemicals (32, 36, and 33, respectively). Except for the
3,4-methylenedioxy-substituted cinnamaldehyde (65 ), multiple ring substitution
seemed to reduce deterrency compared to unsubstituted or monosubstituted rings.
For example, none of these compounds (37, 46, 49, 50, and 51) were significantly
deterrent in the discriminate assay. One half as much terpinene-4-ol (30) was
required as carvacrol (25) to achieve 90% deterrency, suggesting that the OH ring
position may affect deterrency.

The closely similar activity of carvacrol (25) and citronellol (26) suggests that
the aromatic ring is not a key deterrent feature. This is confirmed by the deterrency

of l-nonanol (63 ).

3.1.1.. 'fl “1.! i-Ass 'd

by Dethier (1947), many chemicals are deterrent or repellent at high concentrations,
including host-plant chemicals that at lower concentrations mediate host
acceptance. For example, n-propyl disulfide is deterrent at high concentrations
(Matsumoto and Thorsteinson, 1968; Harris et al., 1987) and stimulatory at an
optimum concentration in onion surface wax of about 0.05 %. Demonstrating that
these compounds were not simply deterrent because of elevated dosages was
particularly important because phenethyl alcohol, the most active deterrent in these
assays, had been described by other workers as an attractant for both onion and
seed-com flies (Ishikawa et al., 1983). The dose-response choice tests were well
suited to ascertain deterrent potency at a range of concentrations; the discriminate-
dosage bioassays also address stimulation at low concentrations. Any material that
is stimulatory at low concentrations would be problematic when applied in the field,
because the concentration eventually would decrease to levels stimulating

undesirable insect behaviors. None of the cinnamaldehyde or terpenoid derivatives

70

tested in dose-response assays elicited increased oviposition through the range of
rates tested, though, as mentioned above, some nitro-containing compounds assayed
via discriminate-dosage may be somewhat stimulatory.

The concentration of compounds from both groups of compounds required
to deter ovipositional activity, and the threshold for detecting EAG response to
these compounds (Cowles, unpublished data), indicate that onion flies have
approximately a 10-100 fold greater sensitivity to the stimulatory compound, n-
propyl disulfide, than to these deterrents. These generalizations agree well with
Dethier's (1980) proposition that receptors sensitive to deterrents should be
relatively less specialized, and therefore have higher sensitivity thresholds, than
receptors that are adapted for detecting host-specific stimuli.

WWW - At the highest dosages

of deterrent compounds, flies were observed to move away from the source after
approaching to within ca. 1 cm. Thus, they act as repellents. Part of the reason why
some compounds in this study are more active than capsaicin-containing products
(Cowles et al., 1989) is that deterrency could be mediated not only by short range
gustation (as with capsaicin) but also through olfaction.

Deterrency is not a simple function of volatility, however. This is
demonstrated with hydrocarbons, which are highly volatile but inactive, and by the
high deterrency of less volatile compounds, such as 4-meth0xycinnamaldehyde (20)
and a-bromocinnamaldehyde (18).

. - Deterrents

 

can be hypothesized to have various modes of action (Davis, 1985). They could: 1)
block perception of host stimuli by binding to and inhibiting their receptors (Davis,
1985), 2) cause more permanent damage to chemoreception by cross-liking or

otherwise disturbing receptor proteins (Gothilf and Bar-Zeev, 1972), 3) act upon

71

specialized deterrent receptors responsive to non-host odors or eidectic pheromones
(Prokopy, 1981) or, 4) be detected by generalized receptors and interpreted as
deterrents by the CNS. The broad array of deterrent compounds, including
monoterpenoid ketones and alcohols, substituted phenethyl alcohols, l-nonanol, and
cinnamyl-related aldehydes, their acetals, alcohols and nitriles, is unsupportive of
the idea that D. antl'qua deterrency requires functional group specificity, and thus is
unsupportive of hypotheses 1, 2, and 3 (above). Indeed, the similar deterrency
ranking of positional (41 and 44) or optical (40 and 43) isomers argues against
highly specific receptor geometry. The large variation in deterrent structure, both in
size and functional moieties, requires that D. antiqua possess either a very broadly
tuned population of one chemoreceptor type, or more likely, a variety of receptor
types that detect these compounds. For onion fly ovipositional deterrency, I favor
hypothesis 4 (above), and recognize that dealing with multiple deterrent receptors
complicates optimization of compound ”fit" to receptors via structure-activity

relationship analyses, particularly if stimulus inputs interact in the CNS.

considerannuhstimulusnnteracion - Host odors eon act synergistic-11y
during host finding or acceptance (Finch, 1978; Visser, 1983; Davis and Sokolove,
1976), probably via central integration of signals from various receptors (Visser,
1983). At the ecological level, integrating signals from various receptors provides
far greater information than relying on signals from one receptor type. Combining
inputs, both within and across sensory modalities (Miller and Strickler, 1984) thus
reduces uncertainty in the host acceptance decision-making process. Such a scheme
also holds for interpreting compounds as deterrents, hence presentation of single
deterrent compounds may not be the best way to maximize deterrence. Instead,
synergistic combinations of compounds that act upon several receptor populations

might be most deterrent.

72

Complex involvement of multiple receptor populations could help explain
inconsistencies in the rank order for deterrency based on specific chemical
modifications. For example, para-methoxy-substitution (20 , 45) increased activity
for (5)-cinnamaldehyde (12), but not for phenethyl alcohol (21). Another example
is the reduced activity of 3,4-methylenedioxycinnamyl alcohol compared to the
aldehyde (58 vs. 65), which reverses the trend seen with other alcohol/ aldehyde

comparisons.

Emlnsisalimemtetatinns - Why are onion flies deterred by such a broad
array of chemicals? The phenylpropenoids, phenolics, and terpenoids tested in this

study are well known allelochemicals. In plants, they prevent competition by
inhibiting root growth or seed germination (Rice, 1984; Nair et al., 1988), inhibit
microbes (Levin, 1971; Mitscher, 1975; Mandava et al., 1980), and are toxic to or
block detoxification enzymes in insects (Brattsten, 1983; Singh et al., 1989). These
same compounds can be sequestered by or synthesized by insects for defense
(Eisner, 1970) and can act as cues for the presence of interspecific competitors
(Jones et al., 1988). More complex multitrophic interactions (Howard et al., 1988)
can be governed by plant defensive compounds. Indeed, the ability of herbivores 0r
predators to sense these chemicals allows them to function as interspecific warning
signals (Lovett et al., 1989). Such signals serve the evolutionary interests of both the
sender, who benefits by deterring enemies, and the receiver, who benefits by
avoiding a source of potentially toxic substances. For the insect not adapted to
detoxify phenolics or terpenoids in their host plant or environment, these chemicals
could, over evolutionary time, become grouped to indicate poor suitability.
Sensitivity to phenolics and terpenoids appears to be a general phenomenon
in insects, and suggests that chemoreceptors for these compounds might be ancient.

Mosquito feeding repellents (Bunker and Hirschfelder, 1925) and ovipositional

73

deterrents (Klocke et al., 1989), have structure-activity relationships very similar to
those of the present study; activity is largely governed by oxygenation of C7 to C13
compounds. Thus, the 'oxygenated' derivatives of inactive hydrocarbons (aldehydes,
esters, ketones and alcohols) exhibited strong repellent activity. If terpenoids and
phenylpropenoids are detected by the same receptors, then I would expect the
magnitude of insect response (attraction or avoidance) to be correlated with fit of
compounds to those receptors. In some cases, species are attracted to the same
compounds that are deterrent to onion flies. For example, in trapping experiments
using bait pails for codling moths, citronellal and methyl cinnamate have been
shown to be attractants (Dethier, 1947). Specific structure-activity relationships
have been elucidated for attraction of Diabrotica species to cinnamyl derivatives.
Among these derivatives, Diabrotica species responded most strongly to 4-
methoxycinnamaldehyde, 4-methoxycinnamonitrile, cinnamyl alcohol, and
cinnamaldehyde (Metcalf and Lampman, 1989), the same compounds with greatest
onion fly deterrency.

Though general sensitivity of insects to these compounds may be exploitable,
specific ecological relationships are certain to fine tune the behavioral response and
re-order the ranking of activity. For example, even though codling moths are
sensitive to the same classes of compounds as are onion flies, their rank order for
sensitivity to acids, esters, and alcohols (Dethier, 1947) is the reverse of what I
observed for onion flies, probably because they are adapted to fruit feeding.
Recently, Jones et al., (1988) discovered that 3,5-dimethoxy-4-hydroxycinnamic acid,
produced in the frass of a lepidopteran pest of cabbage, deters oviposition by D.
radicum, suggesting that this Delia species may have greater sensitivity to carboxylic
acids than does the onion fly.

74

W - There are practical advantages and disadvantages
to the finding that deterrents or repellents act upon fairly generalized
chemoreceptors. On one hand, generality could allow a single product to be broadly
useful in agriculture because of the wide range of pests that could be deterred. On
the other hand, higher concentrations are required than for detecting host cues.
Field studies with thymol, eugenol, cinnamaldehyde and its dimethyl acetal,
phenethyl alcohol, and cinnamyl formate (Cowles, unpublished data) suggest that
practical use of ovipositional deterrents will rely on placing these compounds in
close enough proximity to a host plant, and at sufficiently high concentrations. For
the volatile compounds (E )-cinnamaldehyde and (E)-cinnamonitrile, the
concentration of active material required for at least 90% deterrency appears to be
1 to 2 ng/ml of air, which is 9 orders of magnitude more concentrated than for
obtaining the BR50 for some lepidopteran pheromones (Miller, unpublished data).
Because of the high concentrations of these compounds needed for control, care will
be required that formulations be designed to avoid phytotoxicity.

I believe that these ovipositional deterrents should be explored further as
possible agents of pest control. It is not yet clear whether deployment would involve
controlled release formulations of synthetics, companion plants chosen for release
of these deterrents, or using plant breeding or genetic engineering to incorporate
these agents into the crop plants. Whatever the deployment method, it will be
important that onion flies are given highly attractive alternative sites in which to
oviposit (i.e., the stimulo-deterrent diversionary strategy, Miller and Cowles, 1990),

so that pest deprival does not lead to loss of deterrency.

Chapter 5

Interaction of visual and chemical

onion fly ovipositional deterrents

75

Introduction

Acceptance of host-plants by many phytophagous insects is triggered by
particular combinations of chemical, visual, and physical stimuli sensed during
examining behaviors (Miller and Strickler, 1984; Papaj, 1986; Prokopy, 1986). The
onion fly is an excellent example: Oviposition is stimulated by a combination of
dipropyl disulfide released at optimal concentrations from foliage and substrate, and
by yellow or green cylindrical foliage 4 mm in diameter standing upright in soil
(Harris and Miller, 1991).

The nature of stimulus interaction during host examining is not well
understood. Miller and Harris (1985) likened the neurophysiological decision
making process to an electronic lock, in which pushing buttons (receiving modality-
specific stimuli) Opens the lock (signifying host-plant acceptance). There are at least
four distinct models that can be deScribed with this analogy; all have the same
possible end-result of integrating sensory information across-modalities. 1) In
"classical behavioral chaining", sequential behaviors leading to host-plant
acceptance are released by ”Sign stimuli" (Miller and Harris, 1985; Kennedy, et al.,
1961). Modality-specific sensory information is serially processed: particular
behaviors are responsible for collecting specific types of information. The electronic
lock "opens” when separate buttons are pushed in an unvarying order. 2) Across-
modality stimulus summation requires convergence of information from different
sensory modalities at an interneuron (Miller and Harris, 1985; Carnhi, 1984;
Meredith and Stein, 1983). StimMation of this cell then triggers host-plant

acceptance. An insect then may be able to perceive the host Gestalt instantaneously

76

77

because of parallel processing of sensory information. The electronic lock, in this
case, opens when a particular combination of buttons are simultaneously pushed. 3)
The "classical” behavioral chain concept does not apply when specific examining
behaviors may be followed by a number of other behaviors. A Markov-chain
process may be more appropriate, in which a behavioral web replaces the linear
"classical behavioral chain.” Markov-chains have probabilities associated with
transitions between any two behaviors (Hoppensteadt, 1982). When modality-
specific sensory input is associated with each behavioral state, the result may be
temporal summation. The electronic lock opens when required buttons are pushed
in any order. 4) Last, a mixed model may apply, in which Markov-chain transitions
between examining behaviors are involved with processing sensory information
across sensory-modalities. The electronic lock opens when specific combinations of
buttons are pushed at various times.

These models can be evaluated with respect to the expected behavioral
- patterns they would generate. Models 1 and 2 would be characterized by highly
deterministic behaviors, perhaps appearing to the viewer like a fixed action pattern
(Dawkins, 1983). Models 3 and 4 involve highly probabilistic examining behaviors.

Manipulative experiments involving combinations of stimuli affecting
different sensory modalities should be able to distinguish between models involving
parallel sensory processing (Models 2 and 4) vs. serial processing (Models 1 and 2).
Complex interactions may take place at a recognition interneuron in Models 2 and
4, so any number of outcomes can be programmed for a combination of sensory
inputs (Moore and Christensen, 1985). Under these conditions, marginal
acceptance probabilities for factors affecting different modalities should be
dependent. With temporal summation (Models 1 and 3), the probability of

acceptance is the product of probabilities that each behavior has occurred. This

78

model predicts independent marginal acceptance probabilities for factors affecting
different sensory modalities.

A detailed investigation of onion fly examining behaviors has suggested that
temporal summation (Miller and Harris, 1985) of sensory information occurs
through behavioral webs, but did not distinguish between Models 3 or 4. This paper
extends work on onion fly host acceptance by investigating interactions between
combinations of deterrent visual and chemical stimuli. Chemical deterrents with
different polar functional groups were tested either singly or in mixtures constituting
a test for synergistic interactions within gustatory and olfactory modalities. A
factorial design investigated the between-modality interaction of visual and
chemical deterrents. Behavioral observations from videotape recordings were then
used to study whether specific D. antiqua examining behaviors were primarily

responsible for assessing chemical vs. visual sensory stimuli.

Methods and Materials

Whom - Flies were from the same population as reported in pp.
18-19 and were reared in the lab for 10-15 generations.

Phenethyl alcohol, hydrocinnamonitrile, and (E )-cinnamaldehyde were
purchased at 99% purity from Aldrich Chemical. (E)-Cinnamyl formate was
purchased from Pfaltz and Bauer (Waterbury, CT, 06708) at 97% purity. These
compounds were diluted on a weight /weight basis in polyethylene glycol (PEG),
according to the same methods as in Chapter 2 (p. 30). Experiments used 20 mg
formulated mixture per treatment.

Ovipositional dishes were the same as described on p. 31. Deterrents were
evenly dispensed on the sand within 1 cm of the juncture of foliage and soil, where
most host examining behaviors and oviposition occur (Harris et al., 1987). PEG

79

granules applied on moist sand partly dissolved and became incorporated in the top
1 mm layer.

We - Ovipositional bioassays
were conducted in a growth chamber with the same rotating cylindrical cages and
standard foliar surrogates painted with oil pigments as described on p. 30.

Chemical treatments were the single compounds, phenethyl alcohol (0.4%),
hydrocinnamonitrile (0.4%), or cinnamyl formate (1%) in PEG. Binary mixtures
consisted of these compounds at halved concentrations, for all pairwise
combinations of chemicals. Mixtures were formulated at half the single compound
concentrations to test for synergism, following the logic that if a single compound
were used as a control for testing synergism, the full dosage would have been
reconstituted for the "combination” treatment. Synergism would be supported if a
combination were more deterrent than a full dose of a single compound, while
additivity would generate deterrency intermediate to the single compound
deterrencies.

There were three assays (Table 1, p. 82). Each binary mixture was presented
in a cage with the corresponding single compounds and one control (blank PEG).
The four cups were placed on the cage floor in a square pattern 15 cm apart. Flies
were allowed to lay eggs for 24 hours, then the eggs were removed from the sand by
flotation and counted. Analysis of variance was conducted on ln(x+ 1) transformed
egg counts, which established homogeneous variances. Each assay was replicated

ten times in a randomized complete block design (RCBD).

 

location, cages, and foliar surrogates were the same as Experiment 1, except foliar

surrogates were painted with chromium oxide green or napthol crimson acrylic

80

paints (Binney and Smith, Inc., Easton, PA, 18042). Cinnamaldehyde was
formulated at 1% concentration in PEG.

The four treatment-combinations deployed in a two x two factorial design
were: green onion + blank PEG, green onion + 1% cinnamaldehyde in PEG, red
onion + PEG, and red onion + cinnamaldehyde. The four cups were arranged in
random order in a 15 cm. square pattern, for ten replicates in a RCBD. Flies were
allowed to lay eggs for 24 hours, then the eggs were removed from the sand by
flotation and counted. Data were subjected to analysis of variance on ln(x + 1)
transformed egg counts, which established homogeneous variance. X 2 contingency
analysis was also conducted, on treatment totals, to confirm independence of main
effects.

° ., - Green onion

+ 1% cinnamaldehyde in PEG, red onion + blank PEG, and red onion +

 

. cinnamaldehyde were treatments paired with the green onion + PEG control for
observations of predepositional examining behaviors. A pair of treatments was
centered 8 cm apart in a 15 x 15 x 25 cm glass-fronted plexiglass cage, having
screened top and sides. The cage was provisioned with food and water, and had a
10 x 4 cm mirror obliquely positioned behind the oviposition dishes to observe flies
eclipsed by the foliage. Four female onion flies were removed from the onion
surrogate foliage in the main culture cage at 1:00 pm and added to the observational
cage. The observations were conducted in a different room than Experiment 2, but
with similar temperature and light conditions. A JVC Model BY-110 VHS video
camera was positioned to view the surface of the sand in the two cups, the entire
length of the foliar surrogates, and the image of the foliage in the mirror. Video
recording was conducted for 6 hrs, flies were removed the next morning and the

cage prepared for another pairwise comparison of ovipositional treatments.

81

The following behavioral states and transitions between them were recorded:
arrive on substrate, rest or walk on sand, land on foliage, rest or walk on foliage,
foliar run, ovipositor probing of substrate, oviposition, and leaving the ovipositional
resource. Focal animal sampling, e.g., observing one animal for a specified duration
(Martin and Bateson, 1986; Harris and Miller, 1991) was not necessary, because
video tapes could be reviewed for all fly visits to ovipositional resources.
Transitions between these behaviors were transcribed from video tape and

subjected to X2 analysis for differences in transitional probabilities.

Results and Discussion

Waterman - In all three t:xrlelriments.
the deterrency was slightly, though not significantly, greater for the mixtures than
the single compounds (P > 0.05, SNK test on log(x+ 1) egg counts, Table 1). The
consistently greater deterrency for combinations of deterrents provides weak
evidence for synergistic activity. Differences between treatments at high deterrency
are not expressed well with percent deterrency. Expressing these data as number of
deposited eggs or percent acceptance relative to the control highlights the possible
importance for synergism at high percent deterrency. For Combination 2, the
mixture received half as many eggs as either cinnamyl formate or phenethyl alcohol
alone.

Synergistic interactions within the chemical senses of gustation and olfaction
have been noted for plant allelochemicals involved with hostplant finding and
acceptance by insects (Visser, 1983). Such interactions have basic significance
because they indicate likely across-fiber pattern involvement in processing sensory
information (Dethier, 1982). The presence of similar interactions between
deterrent compounds would be significant, because mixtures of low concentrations

of deterrents could be applied to protect plants from phytophagous insects rather

82

Table 1. Treatments, means for egg counts (ln(x+ 1) transformed), and back-
transformed deterrency values for testing binary mixtures of three compounds for
synergism.

 

 

Concentration Mean ln(e +1)
Treatment in PEG (%) (is. % Deterrency

 

Combination 1.

Control - 5.028 (0.20) a -
phenethyl alcohol (P) 0.4 1.845 (0.30) b 95.8
hydrocinnamonitrile (H) 0.4 1.514 (0.39) b 97.0
P + H 0.2 + 0.2 1.128 (0.50) b 97.9
Combination 2.
Control - 4.755 (0.20) a -
cinnamyl formate (F) 1.0 2.195 (0.35) b 92.2
phenethyl alcohol (P) 0.4 2.081 (0.30) b 93.1
F + P 0.5 + 0.2 1.262 (0.28) b 96.9
Combination 3.
Control - 4.850 (0.43) a -
cinnamyl formate (F) 1.0 2.565 (0.38) b 89.8
hydrocinnamonitrile (H) 0.4 1.435 (0.39) c 96.7
F + H 0.5 + 0.2 1.381 (0.41) c 96.9

 

‘Means not followed by the same letter are not significantly different, SNK test,
P > 0.05.

83

than high concentrations of single compounds. The implications have a parallel
significance for natural plant-herbivore interactions; i.e., plants would be sufficiently
protected by generating an array of defensive compounds at low concentrations
rather than high concentrations of single compounds.

The lack of significant synergism in this experiment could be due to: no
differences or extremely broad overlap in receptor response to these deterrent
compounds, or if these chemicals were perceived as being difi'erent, non-interacting
neural processing of chemical sensory cues could lead to an additive behavioral
response. A critical follow-up experiment would be to test the physiological basis
for lack of synergism. This could be accomplished by analyzing differential receptor
saturation via electroantennography. If antennal receptors do not distinguish these
compounds as being different, then receptors saturated with one odor will not
generate further stimulation when exposed to a second odor (Payne and Dickens,
1976; Miller et al., 1977).

We - Mean e88$

deposited on the four treatment combinations (back-transformed from ln(x+ 1)
counts) were: green foliage -r- PEG, 148; green foliage + 1% cinnamaldehyde in
PEG, 17.3; red foliage + PEG, 79.1; and red foliage + cinnamaldehyde, 4.52.

The synergistic interaction was explored with analysis of variance on log-
transformed egg counts ln(x+ 1) (LTAN OVA). A significant interaction term in
LTAN OVA signifies lack of independence of main effects, similar to X2 analysis,
however it allows block-to-block variation to be partitioned. Analysis of variance
suggested independence of main effects (F027) = 1.01, P > 0.3). Independence
implies that the deterrent foliar color, red, decreased oviposition by approximately

45% compared to green foliage, irrespective of the chemical treatment. Similarly,

84

cinnamaldehyde decreased egg counts by 86% compared to blank PEG, irrespective
of foliar color.

Contingency X2 analysis did not confirm main effect independence. The
calculated X2 was 4.14, significant at P < 0.05 (Tabular X2“ 40 0.05 = 3.84),
indicating significant non-independence. Contingency analysis is designed for
testing probabilities of independent events. One concern when applying
contingency analysis to these egg-count totals is that a single acceptance event is the
act of initiating an ovipositional bout, rather than deposition of a single egg. There
are approximately 2.2 eggs deposited per bout (Mowry, et al., 1989b); dividing egg
count totals by 2.2 eggs per bout transforms these data to the approximate number
of egg deposition bouts. The revised X2 agrees with the conclusion from
LTAN OVA that the vision and chemical deterrents act independently (X20 do =
1.838, P > 0.1). This is good support for separate examining behaviors being
responsible for detecting host cues of different sensory modalities (Model 3 from
the introduction).

-1-1-111'1 01-125. o n -, r'n-r- "'I‘! alt-'11-Row
totals from behavioral transition data tables (Figure 1) gave the number of times a
behavioral event occurred. Because individual cells in the transition matrix had
small counts, behaviors were grouped by physical location or by function. Visits
(sand arrive + stem land), sand (sand rest/walk + sand run), foliar rest/walk, foliar
run, and probe + oviposit were the resulting categories. When totals from each
choice comparison (Table 2) were subjected to X2 analysis, only the red foliage +
1% cinnamaldehyde in PEG yielded significant differences in transition
probabilities (X2(4 u.) = 32.68, P < 0.001) from the green foliage control. Most
behavioral frequencies were similar on red foliage + cinnamaldehyde and on green

foliage, when these treatments were compared. For example, the numbers of visits

85

 

Dole:_-----_---_-
Irenlrnent:_.. ._ SUBSEQUENT BEHAVIOR
Dole Sheet 1.. of _ SAND STEM OVIPOSITOR
.‘
[8331" (19:53:14)} $013 \yxko qS-c’lifif 93)“ QQQQQ’ 0480‘; 659$

        
      

       
 
 

SAND

PRECEDING BEHAVIOR
STEM

Fi re 1. Sample data collection sheet for transcribing behavioral transitions from
vi 1:rotape recordings of onion fly preovipositional examining and egg deposition
be aviors.

86

Table 2. Totals (and expected totals from X2 anal is) for behavioral events in two-
choiceoassays analyzing the effects of visual and c emical deterrents on behavioral
transrtrons.

 

 

 

Numbers of Behavioral Events
Stem rest/ Probe + Marginal

Treatment Visit Sand walk Stem run Oviposit Totals
Comparison 1.

Green 70 (71.5) 98 (96.6) 38 (43.3) 46 (42) 4 (2.5) 256

Red 44 (42.5) 56 (57.4) 31 (25.7) 21 (25) 0 (1.5) 148
Comparison 2.

Green 45 (43.5) 49 (51.4) 24 (28.9) 17 (16.3) 14 (8.9) 149

Green

+ cinn. 38 (39.5) 49 (46.6) 31 (26.1) 14 (14.7) 3 (8.1) 135

Comparison 3.

Green 47 (58.7) 76 (82.2) 18 (21.7) 48 (40) 36 (22.3) 225

Red

+ cinn. 53 (412) 64(57.8) 19 (152) 20 (28) 2(15.7) 158

 

Green, green foliar onion surrogate + blank (polyethylene fiycol (PEG ; Red, red
foliar surrogate + PEG; + cinn., 20 mg. of 1 o cinnamalde yde in PE .

 

87

to ovipositional cups were 53 and 47 for the two treatments, respectively. The
difference between probing + oviposition between these two treatments suggested
that these behaviors may have contributed to the significant differences between
treatments. When this behavioral category was removed from analysis, there were
still significant differences in behavioral frequencies (X2(3 if.) = 9.86, P < 0.05).
Further removing the effect of foliar runs generated an insignificant X2 value of
1.31. Therefore, combined red foliage and cinnamaldehyde significantly affected
the foliar run and ovipositor probing behaviors.

Separating the effects of visual versus chemical deterrents can be
accomplished by examining the remaining choice comparisons. Though the
combined X2(4 a) value of 8.23 was not significant in the green vs. green +
cinnamaldehyde comparison, the probe + oviposition category was by itself
significant (X2(1 d.f.) = 6.474, P < 0.02). All other activities on these two resources
occurred at approximately the same frequency, therefore we can conclude that
cinnamaldehyde mostly influenced behavioral transitions to ovipositor probing and
any subsequent transitions to egg deposition. The lack of a significant X2 value for
behaviors in the red vs. green foliage may at first glance be misleading. If there
were a general reduction in activity on cups that had red foliage, proportionally
distributed across all behaviors, then the X2 analysis would not detect differences
between red and green foliage treatments. The total level of activity on cups is
measured by the marginal totals used for X2 analysis, and indeed suggests overall
reductions in activity on treatment cups with red foliage. A paired t-test on general
activity level was performed by pairing the number of events for each behavior in
the red vs. green foliage treatments. Both the red + cinnamaldehyde vs. green
foliage + cinnamaldehyde and the red vs. green foliage comparisons were used,
yielding ten pairs of behavioral event totals. There was indeed a reduced level of

activity on cups with red foliage (paired t-test = 3.079, P < 0.02). The stem rest +

88

stem walk behavioral frequencies were similar between treatments; the most
consistent efiect of the presence of red foliage was to decrease the transitions to

foliar runs.

W11 - Possible synergism or multiplicative effects from using
combinations of deterrents are important, both from basic and applied interests.
Multiplicative effects have been demonstrated with across-modality stimuli that
release onion fly oviposition (Harris and Miller, 1988). It appears that the across-
modality interactions are as important with deterrent stimuli as for excitatory cues.

Why do multiplicative models for between-modality interactions make
sense? Neurophysiological models for integrating information from different
sensory modalities and observations of behavioral transitions suggest that onion flies
probably process visual and chemical stimuli during different behaviors. The
probability of accepting a host model consequently is determined by the joint
probability (a multiplicative function) that color and odor are acceptable. A
temporal summation interpretation of across-modality stimulus synergism by onion
flies is as follows: The presence of dipropyl disulfide increases the probability an
onion fly will alight on vertical objects (Harris and Miller, in manuscript). Once
having alighted on a leaf, sitting and grooming behaviors could be involved with
assessment of foliar color (evidence from this work), which increases the probability
of a transition to a foliar run. Changing foliar color from green to red would
decrease the probability of a fly proceeding from landing to foliar runs. Decreasing
dipropyl disulfide presence would "synergize” the deterrent effect of a red onion leaf
because, by reducing the probability that an onion fly lands, it is less likely that fly
will assess foliar color. The presence of deterrent would synergize decreased

dipropyl disulfide concentration and red foliage, because the probability eggs would

89

be deposited in this combination would jointly be reduced by foliar landing,
transition to foliar runs, and transition to ovipositior probing.

While the evidence from both excitatory and inhibitory cues is consistent
with temporal summation of sensory stimuli in onion flies, there are additional
complexities that need to be addressed. One concern with this simple model is that
behavioral categories used in Experiment 3 were low resolution, allowing behavioral
components like proboscis extension and touching the ovipositor tip to the foliar
surface to be lumped under "foliar run.” Would high resolution behavioral analysis
correlate with the interpretations of across-fiber or with temporal summation?
Manipulative experiments with very fine resolution behavioral observations, or
neuroethological methods may have to be applied for us to know how the onion fly

really perceives a host plant

Chapter 6

Stimulo-deterrent diversion for manipulating

onion fly oviposition: a greenhouse test

90

Introduction

Onion fly, Delia antiqua (Meigen), females deposit about 60 times more eggs
on onion plants grown from deeply planted large bulbs than on small onion
seedlings (Hammond, 1924; Finch and Eckenrode, 1985; Cowles and Miller,
unpublished). However, onion seedlings are usually the only hosts available for
onion flies to colonize. Damage to small seedlings from the first larval generation
may be extensive, as small plants may be completely consumed and are highly prone
to dehydration and microbial infection. Moreover, each larva may destroy ca. 10
adjacent seedlings while completing development (Workman, 1958; Loosjes, 1976).
The result is economically injurious non-random thinning of onion stands.

Cull onions, consisting of unmarketably small or sprouting bulbs, are a waste
' product from the onion industry. It has been suggested that highly susceptible onion
seedlings could be protected by planting cull onions as a trap crop, to which onion
fly oviposition would be diverted (Hammond, 1924; Lovett, 1923).

Another idea for non-insecticidal control of onion maggot is treating
seedlings or adjacent soil with deterrents. In laboratory studies, onion fly
oviposition is suppressed by a wide variety of inexpensive and readily available
compounds including phenolics and monoterpenoids; however, the dosages required
for control are several orders of magnitude greater than that for insecticides
(Alfaro, et al., 1981; Javer, et al., 1987; Cowles et al., 1990). As is true for
consumatory behaviors for various insect/ plant interactions, (Raffa and Frazier,
1988; Chapman, 1988; Jermy, 1971), onion fly ovipositional deterrents can lose
their efficacy when females are given no ovipositional outlet (Cowles et al., 1989).

Thus, it is doubtful that onion fly can be controlled by deterrents alone.
9 1

92

Miller and Cowles (1990) proposed stimulo-deterrent diversion (SDD) and
reported preliminary data suggesting that behavioral forces of "push" and "pull”
might be multiplicative rather than simply additive. We predicted for onion flies
that presence of culls would improve the protection of deterrent-treated seedlings,
by both diverting oviposition and preventing the gravid onion flies from becoming
ovipositionally catholic due to deprivation. In the present greenhouse study, a
factorial design was used to test the stimulo-deterrent diversion concept. D. antiqua
oviposition on onion seedlings was assessed for the following treatments: 1)
‘ seedlings-only, 2) seedlings + culls, 3) seedlings + deterrent, and 4) seedlings +

deterrent + culls.

Methods and Materials

The laboratory culture of onion flies was initiated in 1987 and maintained as
described on pp. 18-19. Flies used in experiments were reared in the lab for 10-15
generations.

About 100 onion seeds, 'Sassy Brassy‘ cultivar (Ferry-Morse Seed Co.,
Modesto, CA 95354), were planted in muck soil per 13 cm diameter pot. To prevent
undesired encounters with D. antr'qua, onions were grown and kept in a large
screened cage held in a greenhouse. Onion plants were watered on alternate days.
Seedlings were transplanted when at the two-leaf stage, four per 13 cm diameter
pot, in a 5 cm square pattern. Seedling onions used in experiments were at the 4-5
leaf stage.

Onion bulbs ('Spartan Banner') were collected from a grower's field in May
and September 1989, and stored at 3°C until use. Culls collected in May had been
held in the grower's storage through the winter and spread back into the field. Bulbs
collected in September were gleaned after harvest. Two days before use, two

sprouted cull onions were planted in non-sterile muck soil, in a 19 cm pot. The

93

onion neck was positioned 5 cm below the soil surface, the optimal depth for
maximizing D. antiqua oviposition (unpublished).

The deterrent used was cinnamaldehyde (Aldrich Chemical, Milwaukee, WI
5333) 50% (w/w) in activated charcoal (aquarium charcoal). Granules (5 mm
diameter) were formulated in May 1988, and stored in a NalgeneR bottle at 20°C
until needed. These large granules were coarsely crushed to form < 2 mm diameter
granules. About 50 mg of formulated material was placed within 2 cm of the base of
each onion seedling.

To assay for a possible interaction between deterrents and culls when
protecting seedlings, 100 mature females and 20 males were placed in a 1.5 x 1.5 x
3.5 m screened cage, held within a greenhouse. This large cage was provisioned
with food, water, and one of the four onion treatments. To minimize possible
treatment-treatment interactions, the four treatment combinations resulting from
the 2 x 2 factorial design were sequentially presented at two day intervals according
to a Latin square protocol. Four pots of seedlings were linearly positioned at 50 cm
spacing. When culls were present their pot was positioned at the center of the row,
with 50 cm to the seedling pots on each side. After allowing flies to oviposit for two
days, the potted seedlings and culls were removed and the next onion treatment was
placed in the cage.

The eggs from the top 4 cm layer of soil in each pot were collected and
separated from muck soil by flotation in water. Residual floating muck and eggs
were passed through a 1 mm mesh sieve to remove the larger muck fragments; the
resulting eggs and fine muck were deposited on a nylon filter. This residue was
washed into a black enamel pan. The contrast of the white eggs against the black
background enabled the eggs to be easily counted as they were removed with a

vacuum aspirator.

94

Egg counts were analyzed for significant main efiects and their interaction
with analysis of variance. The numbers of eggs laid on seedlings were not
distributed normally, however, logarithmic (ln(x + 1)) transformation established
homogeneous variances (Bartlett's Test, P > 0.1, Steel and Torrie, 1980).
Contingency X 2 analysis of treatment-combination totals was used to verify
independence of treatment effects.

Results and Discussion

The total numbers of eggs laid on seedlings for the four treatments were
3185 (seedlings-only), 1531 (seedlings + culls), 127 (seedlings + deterrent), and 69
(seedlings + deterrent + culls). Both main effects were significant (P < 0.01,
analysis of variance on log-transformed egg counts, LTAN OVA, Table 1). Because
calculating the differences between logarithms is equivalent to forming the ratio for
non-transformed data, main effects generated from LTAN OVA are a
transformation of percent eggs laid on treated vs. control seedlings. The numbers of
' eggs laid on seedlings were reduced 96 2 1 and 58 2 10 percent (mean :I: S.E.), by
cinnamaldehyde and culls, respectively.

The total numbers of eggs laid on culls were 3867 and 4211 when deterrent
was absent or present on seedlings, respectively. The slightly (but not significantly)
greater number of eggs laid on culls when cinnamaldehyde was present argues
against possible whole-cage ovipositional depression when deterrent was present.

Analysis of the interaction between the deterrent and culls was a major
objective of this study. Possible relationships are non-interaction, positive
interaction (synergism), and negative interaction (see Figure 1). The graphs in the
right column of Figure l are logarithmically transformed data from the graphs in the
left column. Non-interaction (Figure 1A) would be expected if each factor, culls or

deterrent, prevented some absolute number of eggs from being laid on seedlings,

95

Table 1. Analysis of variance for ln(x+ 1) transformed numbers of onion
fly eggs de osited on seedling onions (valued crop) protected by deterrent
(cinnamal ehyde) and diversionary (cull onions) treatments. Rows
represent sequential presentation of all four treatment combinations while
co umns coded the order of treatment presentation.

 

Sums of Mean

 

Source df Squares Square F Value Pr > F
Row 3 12472 4.157 37.92 0.00031
Column 3 1.471 0.490 4.47 0.057
Treatments 3 42.805
Cinnamaldehyde 1 39.755 39.755 362.56 0.0001
01118 1 2.982 2.982 27.19 0.002
Cinn. x Culls 1 0.068 0.068 0.62 0.460
Error 6 0.657 0.109
Total 15 57.406

 

1 Si 'ficant row effect may indicate differences in relative stimulatory
11 3’4)“ culls collected in springtime (Rows 1 and 2) and autumn (Rows
an .

>

ID

96

 

 

 

 

 

 

 

 

NON-TRANSFORMED LOG-TRANSFORMED
00 - 40
1 \ \ CE E IE”
g .0 . ceremem f 154 ABSENT
9 2° ‘ \ DETEFRENT f 3.0 m
PRESENT PFESENT
0 25
UVEFSION WEFBION m m
ABSENT PKSENT ANENT “SENT
N a 4.0 .
a...) \ e a, \ 8:“
tu ceremem to
g * ABSENT g
19 20 .. \ m 5 10., \ m
PFESENT “SENT
0 2.5
m m m m
mean PFESENT m m
n - 6.0 .

\m

151

TOTAL £008
3 8
MEOOS + 1)

10.

 

 

 

 

PESENT
J
0 2.5
orvensrou cum 001mm m
m7 man man m

Figure 1. Theoretical forms of interactions between deterrent and diversionary
treatments, with non-transformed and log-transformed hypothetical data. (A) - no
interactions when non-transformed; slightly negative interaction, indicatin
dependence, when transformed. (B) - Positive interaction. The log-transforme
data Show no interaction, hence, the factors are independent. ( ) - Negative
interaction.

97

irrespective of the presence or absence of the other factor. This pattern generates
parallel lines in the interaction diagram showing non-transformed egg counts.

Synergism can be concluded if there is a positive value for the culls x
deterrent interaction in non-transformed egg counts (Fig 1B). The linear contrast
(Snedecor and Cochran, 1967) calculates the mean for this interaction. For the
present experiment this value is: (3185-127-1531+69)/8 = 199.5. The conclusion
from this additive model is that approximately 200 fewer eggs were laid per 16
seedlings in the deterrent+culls treatment than expected from the main effects.
Hence, there was a synergistic interaction between deterrent and culls. Synergism
may result when independent factors (in the probabilistic sense) govern insect
behavior. In this experiment, culls could reduce the probability of an onion fly
finding a seedling, while deterrent applied to a seedling could, independently of host
finding, reduce the probability of acceptance. The seedling + deterrent treatment
was expected to cause some degree of ovipositional deprivation among gravid onion
flies. This deprivation would increase the probability of an onion fly accepting
seedlings in this treatment relative to the other three treatment combinations, and
thus cause non-independence. I

Contingency X2 and LTAN OVA detect non-independence between
diversionary treatments and deterrents, because both analyses reveal non-parallel
lines in log-transformed interaction diagrams (Figure 1A and C). LTAN OVA has
additional advantages: the calculated direction of the interaction should be negative
when calculated from log-transformed values if there were significant deprivation;
contingency analysis doesn't indicate the type of interaction. Also, LTAN OVA
partitions experimental variation from the Latin square design.

This predicted negative interaction (non-independence) for deterrent-treated
seedlings presented in the no-choice situation was not detected by either

LTANOVA (Table 1) or by X2 analysis (P > 025). in fact, the slight interaction

98

illustrated in the log-transformed diagram of the experimental data (Figure 2), is
positive, though not statistically significant. All analyses consistently support a
multiplicative model for ovipositional response of onion flies to the combined

presence of culls and deterrent.

MW - Unlike pesticidal control strategies,

SDD involves several main components: valued crop, diversionary crop, deterrent,
and pest. Relative to a pesticidal approach, the time frame for control is extended
with SDD, because the adult pest is allowed to survive and reproduce. Multiple
components and a long time-frame for control suggest that understanding
component interactions is crucial for implementation of SDD.

The good fit of data in this greenhouse study to a multiplicative model for
the interaction of deterrents and culls is encouraging for understanding how to apply
the stimulo-deterrent diversionary strategy in the field. For example, if the present
growth state of culls and seedling onions gave a 100:1 preference ratio, we would
need at least 83% (5:1 ratio) deterrency to reach the overall goal of a 500:1
preference ratio.

The crop area that should be planted to diversionary culls, and the optimal
distance and distribution of culls relative to the valued crop are unknown. Marsolan
and Garrett's (1977) simulation model determined that the critical variables for
optimizing a trap crop are maximum distance between trap crop resources, insect
diffusivity, and ovipositional preference. The proximity of seedlings and culls in our
experimental arena (maximum distance was 1 meter), may have maximized the
"comparison” of treated seedlings and culls, and may have contributed to the
effectiveness of the deterrent + cull combination in this small-scale test. However,
the probability that a gravid onion fly would accept deterrent-treated seedlings did
not change over two days, meaning that there may be a long time interval during

TOTAL EGGS

99

 

 

 

NON-TRANSFORMED LOG-TRANSFORMED
4000 A 5000 B
"’100 \ DETERRENT
0 ABSENT
3000
8 500
:‘r’
2000 1-100 \ OETERRENT
OETERRENT .9 5 PRESENT
ABSENT
1000 10
OETERRENT 5
PRESENT 1
0 ' CULLS CULLS CULLS CULLS
ABSENT PRESENT ABSENT PRESENT

Figure 2. Experimental data for interactions between cinnamaldehyde deterrent
and cull onions, as both non-transformed (A) and log-scale (B) graphs. These data
are consistent with Figure 1 B, a synergistic effect with an additive model but with
parallel lines indicating treatment independence with log-transformation.

100

which ”choice” may be executed. Over a two-day interval, onion flies within crop
habitat would be expected to move on the order of 100 to 300 meters, via multiple,
short flights (Loosjes, 1976; Martinson et al., 1988). During movement, several
encounters with highly acceptable culls could elicit oviposition, reset the
physiological state, and consequently maintain sensitivity to deterrents. Less
effective deterrents than in the present study would be expected to become more
rapidly acceptable for oviposition; hence, the diversionary crop would have to be
more closely spaced to increase the frequency of encounters.

So that the results could legitimately be subjected to X2 analysis (Cochran,
1954), relatively large seedlings and moderate deterrent dosage were chosen to
allow appreciable oviposition on the cull + deterrent protected seedlings. Four to
five-leaf stage onions, as used in this experiment, stimulate more oviposition than
the small seedlings to be protected in the field (Groden, 1982; Harris, et al., 1987).
The 480:1 preference ratio of eggs laid on cullszseedlings (on a per-plant basis)
measured in this experiment is close to the 500:1 theoretical value calculated for
commercially acceptable crop protection (Miller and Cowles, 1990).

While the total picture for field deployment of SDD for controlling D.
antiqua is complex, a major piece of the puzzle is in place. Specifically, a simple
multiplicative model adequately described the interaction between deterrents and
culls. Such a simple model suggests that the complexities of the multifactorial SDD
approach may be decomposed into deterrency and diversion. These main
components can be studied separately, and their combined effect predicted to be the
product of the two separately determined probabilities of acceptance (where
acceptance is defined as 1 minus the percent reduction in number of eggs laid on
seedlings). Although this greenhouse test was only a microcosm of the field, the
high mobility of onion flies in the field and lack of deprivation in this experiment

over a two-day interval suggest that the simultaneous use of ovipositional deterrents

101

and culls has promise for field manipulation of onion fly oviposition, provided that
more potent deterrents can be found and formulated to last for several weeks. An
alternative to applying deterrents would be to generate onion cultivars antixenotic
to onion flies. SDD may have greater potential for controlling other mobile and
discriminating pests, for which formulated deterrents or antixenotic crop varieties
are known, and where the plant/pest interaction is shorter than the ca. six weeks
over which onion seedlings must be protected from first generation onion maggot

larvae.

Chapter 7

Integrating stimulo-deterrent diversion into pesticide
resistance management: directing pest evolution
toward sustainable control

102

Introduction

Synthetic organic insecticides have demonstrated tremendous utility for over
fifty years (Metcalf, 1980). The very nature of their success is the ability to kill large
portions of treated pest populations. This ability, unfortunately, is a built-in flaw
with respect to evolutionary processes. Successively removing susceptible
individuals from a population may rapidly select for physiological resistance to an
insecticide. As early as 1946, houseflies could no longer be controlled with DDT,
and by 1980, more than 400 species of arthropods were resistant to insecticides
(Georghiou and Mellon, 1983). Reports of pest resistance to the most recently
developed classes of insecticides, such as pyrethroids, formamidines, insect growth
regulators, and Bacillus thufingiensis endotoxin, make questionable the premise that
. resistance can be mitigated by substituting new modes of action (Dover and Croft,
1984; Georghiou and Mellon, 1983; Miller et al., 1983; Sparks and Hammock, 1983;
Van Rie, et al., 1990).

In response to the possible threat of losing insecticides as a pest management
tool, various strategies have been suggested for slowing or reversing pesticide
resistance development. These approaches may be subdivided into pesticide and
ecological management schemes (Dover and Croft, 1984). Pesticide management
refers to tactics such as applying insecticide mixtures, rotating insecticides, applying
insecticides with synergists, using high doses, applying low doses, or using short
residual materials. Ecological management approaches endeavor to replace
insecticide suppression of pests with cultural, physical, or biological control

practices. These ecological approaches distribute selection pressure among several

103

104

mortality factors. The result is minimized use of insecticides, which in turn implies
reduced selection pressure for insecticide resistance.

Tabashnik's recent (1989) review of resistance management concluded that
pesticide combinations may not consistently prevent resistance development. In
fact, pesticide combinations may be more disruptive than approaches that minimize
pesticide use. This paper supports his critique of pesticide management strategies,
and contrasts these strategies with an ecological approach we all the stimulo-
deterrent diversion (SDD) (Miller and Cowles, 1990). SDD manipulates pest
feeding or oviposition by simultaneously deploying a highly acceptable diversionary
crop while protecting the valued crop with deterrents. This chapter addresses the
theoretical consequences to pesticide resistance management of being able to
manage pest distribution.

Pesticide management approaches

Mixtures and rotation - Successfully avoiding resistance by using insecticide
mixtures assumes that different mechanisms are responsible for causing resistance
to each insecticide (Ozaki, 1983; Curtis, 1985; MacDonald, et al., 1983; Van Ric, et
al., 1990). If this assumption were true, and the genes coding for these resistance
traits were present at low frequencies, the probability of finding individuals resistant
to both pesticides should then be extremely small (Orrtis, 1985; Tabashnik, 1989).
Applying a mixture containing component insecticides at concentrations sufficient to
kill homozygous susceptibles would be expected to leave rare double heterozygote
and the extremely rare double homozygote resistant survivors. After selecting a
population with an insecticide mixture, immigration of susceptible immigrants can
decrease the gene frequency for resistance. The impact of gene flow on delaying
resistance is contingent on immigrant arrival between the time that spraying

occurred and mating of the resident population. If resistant individuals mate with

105

susceptible immigrants, then resistance alleles will mostly be maintained among
heterozygotes (Curtis, 1985).

Using chemicals in rotation arose from the idea that chemicals may have
negatively correlated cross resistance, meaning that selection with one compound
would increase susceptibility to the other compound(s). While one chemical is
being used to control a pest, susceptibility to another chemical would increase, so
using the compounds in alternation could conceivably indefinitely prevent resistance
(Georghiou, 1983). Straightforward genetic mechanisms could cause genetic trade-
offs. One possibility is that different and independent resistance loci could be
responsible for resistance to each pesticide, and each resistance allele has an
associated fitness cost. Another possibility is that the resistance mechanism
involves different, recessive alleles at the same locus. Selection for resistance to one
compound would then preclude resistance to the second compound.

Pesticide plus synergist - A synergist is a compound that, when combined
' with an insecticide, increases the toxicity in more than an additive manner. By
blocking detoxification mechanisms, synergists may restore insecticidal activity after
resistance has occurred (Ranasinghe and Georghiou, 1979; Ozaki, 1983). Use of
synergists has also been suggested for preempting resistance development by
blocking in advance a likely detoxification pathway (Georghiou, 1983).

The use of synergists may readily extend the life of some insecticides.
However, in an evolutionary view, depending on synergists presumes that 1) variants
in detoxification enzyme insensitive to synergists do not exist (or are mutually
incompatible with detoxifying the insecticide), and 2) that the blocked enzyme is the
only important route available for adaptive change.

High dose and selective timing - Comins (1977) and Taylor and Georghiou
(1979) developed a strategy for pesticide resistance management based on high

pesticide doses. If resistance is controlled by one, two-allele locus with co-dominant

106

inheritance, then a concentration of insecticide could be applied that kills both
homozygous susceptible and heterozygous genotypes. If the gene frequency for
resistance is low, then there should be few if any homozygous resistant individuals
present. These few individuals would be expected to survive a high dose insecticide
application; their mating with susceptible immigrants would maintain most
resistance genes among heterozygotes.

In closed pest populations, the probability of resistance development would
be dependent on the initial resistance gene frequency and on mate finding, rather
than on immigration. If the resistance allele is common, mating between resistant
individuals is likely and the gene frequency for resistance would rapidly increase.
Thus, high insecticide doses can be a prescription for immediate and high level
resistance development. On the other hand, if the population is susceptible and
potential resistance traits have low gene frequencies, then the high dose strategy
could effectively suppress the pest population.

Pesticide strategy assumptions - Rotation, mixtures, and synergist strategies
assume that pests have negatively correlated cross resistance or lack alternate
adaptive routes. These assumptions are questionable, based on experience with: 1)
the presence of cross resistance, even when selecting populations with insecticides of
differing modes of action, and 2) the multiple forms of resistance possible when
selecting with a single compound.

The first experiences with cross-resistance suggested that selection with one
compound preselects the population for resistance to related compounds (Hoskins
and Gordon, 1956). Cross resistance may explain why, in the case of
organophosphate and carbamate insecticides applied to control rice leaflloppers, the
effective field life for carbamates was only four years (Ozaki, 1983). In this

situation, the assumption of independent modes of action was clearly violated by

107

synergistic interactions at the physiological level between these classes of
compounds (Ozaki, 1983).

Cross-resistance can also be expected between unrelated compounds;
ignoring this possibility can lead to poor predictions of evolutionary response. For
example, in 1967 Williams predicted that insects would not develop resistance to
juvenile hormone (JH) growth regulators used as insecticides (Sparks and
Hammock, 1983). Williams' claim may have been based on an expectation that
coordinated changes in JH target site and hormone structure would be an
improbable mode of resistance development. However, within five years, strains of
insects resistant to juvenile hormone applications were selected (Dyte, 1972; Cerf
and Georghiou, 1972). The JH resistant strains had mechanisms of decreased
penetration and increased metabolism; mechanisms pre-selected with conventional
insecticides (Sparks and Hammock, 1983).

Houseflies have multiple resistance adaptations to toxins. DDT resistance in
houseflies may be due to any combination of increased dehydrochlorinase or a-
hydroxylation, decreased penetration, or target site insensitivity (Fukuto and
Mallipudi, 1983; Matsumura, 1983). When the dehydrochlorination detoxification
pathway was blocked with a synergist, selection with DDT lead to high resistance
due to a-hydroxylation (Georghiou, 1983).

Correlated resistance traits is a weakness of all the high-kill strategies.
Chemically unrelated compounds should be expected to cause cross resistance to
each other because relatively non-specific resistance mechanisms exist, like
decreased absorption or increased activity of generalized detoxification enzymes
(Dyte, 1972; Brattsten et al., 1986; Tabashnik et al., 1987). Detoxification enzymes
can be genetically correlated without being genetically linked, either through
pleiotropy or coordinated gene expression (Plapp and Wang, 1983; Grant, 1986;
Via, 1986).

108

Genetic correlation is expected for a deeper reason. Genes do not exist by
themselves, rather they interact with all other genes that constitute the genetic
endowment of an individual, the unit of selection. Changes in one enzyme, for
example, should impact the milieu in which related enzymes function. Complex
feedback through physiological machinery implies that positive and negative
interactions exist. When a population is exposed to selection with pesticides,
evolution should optimize population fitness by maximizing survivorship and
minimizing reproductive disadvantages for the environmental conditions. If one
enzyme system is unavailable to selection, perhaps because it is blocked by a
synergist, then selection should act upon any remaining resistance traits. After a
particular resistance allele is nearly fixed, continued selection should allow gene
frequencies for alleles interacting through physiological processes to shift, so that
the population will approach a new adaptive peak (Crow, 1957).

High vs. low selection strategies - Clearly, the pesticide management
' techniques of rotation, mixtures, synergists, high doses, and selective timing all are
based on high mortality, and consequently on high selection potential. Tabashnik
and Croft (1982) have pointed out several practical flaws with these high mortality
strategies. Immigration of pests with susceptible genotypes are largely responsible
for delaying resistance in these strategies, however, these individuals may not
survive long enough in the treated habitat to mate and reduce the number of
homozygous resistant individuals. In addition, I would add a concern that local
mate choice from cohorts surviving pesticide sprays could change the expected
genotype frequency. Inbreeding would greatly increase the number of homozygous
or double heterozygote resistant progeny.

The most troublesome aspect of high-kill strategies is that they add to both
environmental contamination and insecticide costs (T abashnik and Croft, 1982;

Tabashnik, 1989). Replacing insecticide sprays with photosynthetically derived crop

109

resistance traits is compatible with sustainable agriculture principles (e.g.,
preservation of natural enemies and environmental quality). However, whether a
high level of antibiosis is caused by pesticides or plant traits (Gould, 1986),
ecologically unsophisticated applications of high mortality strategies are likely to be
short lived.

Pesticide management approaches using low doses, short residual materials,
and fewer sprays are all consistent with reduced selection intensities, but may less
effectively suppress pest populations (Georghiou, 1983). These approaches are
more consistent with ecological aspects of integrated pest management, in which
mutually compatible and varied techniques, such as cultural and biological control,
replace some insecticide use.

Stimulo-Deterrent Diversion

The remainder of this paper examines the potential of an ecological
management approach, stimulo-deterrent diversion (SDD) (Miller and Cowles,
1990), as a tool for pesticide resistance management. The generality of the pesticide
resistance model used to evaluate SDD may contribute to increased interest in crop
ecosystem management and behavioral manipulation for managing pesticide
resistance.

Crop protectants have largely been limited to toxins. Insecticides are lethal
agents applied to suppress insect pest populations, so they are ecologically
analogous to naturally produced antibiotics found in plants (Gould, 1984; Brattsten,
1988). Other possible modes of action for protectants are antixenosis or tolerance
(Painter, 1968).

Antixenosis, or deterrency, commonly manifests itself as differential
acceptance of a crop or variety by a pest insect (Hokkanen et al., 1986; Riley, 1871,
cited in Casagrande, 1987). Though a few examples are known, deterrents are
rarely effective in no—choice situations (Painter, 1968). Miller and Cowles (1990)

110

have suggested deploying highly stimulatory alternative host resources to prevent
deprivation-caused failure of deterrents protecting a valued crop. This bipolar
behavioral approach has been named stimulo-deterrent diversion. SDD affects the
distribution of individuals colonizing the two sub-habitats. If the diversionary crop
is not treated with pesticides, it can serve as a refuge for susceptible genotypes in a
pest population, divert pests from damaging the valued crop, and act as a nursery for
biological control agents. '

Modeling the Impact of SDD on Pesticide Resistance - The impact of
refugia on the development of pesticide resistance has been modeled by Gould
(1984) and Tabashnik and Croft (1982). Gould (1984) modeled the interactions
between behavioral and physiological resistance to pesticides, a convenient starting
point for modeling the effect of SDD on pesticide resistance development. In my
model, the refuge is a diversionary crop designed as part of the pest management
system; the pesticide-treated areas are the valued crop of the SDD system. The
' purpose of this model was to gauge the relative importance of genetic, management,
and ecological parameters when designing SDD crop systems to prevent or reverse
physiological resistance development.

The assumptions for my population genetics model were: the pest
population is diploid and has two unlinked autosomal genes, each with two alleles
that control: i) physiological resistance (R) to the pesticide residues, or ii) avoidance
(A) to either pesticide residues, an applied deterrent, or an antixenotic valued crop.
Adults emerging from pesticide-treated valued—crop and untreated diversionary crop
habitats randomly mate and belong to an infinitely large population (i.e., there is no
drift; thus, this a deterministic model). The behavioral effect of the A allele was
implemented by assigning probabilities that avoiding and non-avoiding homozygote
females lay eggs in the valued crop (variables X and Y in the model). Mortality

occurs in the immature stage, considered to be incapable of moving between crops.

111

Differences in fitness were assumed to be solely determined by survivorship of the
immature stage (Figure 1).

Instead of assigning survivorships for each genotype in the pesticide treated
and untreated habitats (Gould, 1984), survivorship in this model was composed of:
habitat suitability (THS and UHS for treated and untreated habitat suitability,
respectively), fitness costs for carrying physiological and / or avoidance alleles (F CR
and FCA), and an explicit dose-response model for pesticide mortality (PMORT,
Figure 2). Pesticide concentration, mode of inheritance of R and A alleles, valued
and diversionary crop suitability and availability, and fitness costs for R and a alleles
were each varied to determine the impact of SDD systems on evolution of
physiological resistance genotypes. Parameters used for all the simulations are
summarized in Table 1.

The pesticide component of this model calculated mortality for rr, Rr, and
RR genotypes, based on a linear log(concentration) vs. probit mortality relationship
(Taylor and Georghiou, 1979). The slope and ID50 values for each genotype and
the concentration of pesticide were program variables. Functional dominance or
recessiveness was modeled with co-dominant R inheritance and low or high
pesticide concentrations, respectively (Taylor and Georghiou, 1979). Dominant or
recessive inheritance for the R allele was modeled by setting the slope and LD50 for
the heterozygote genotype equal to the R or rr genotypes, respectively. The
program calculated the standard normal deviation from the LD50 concentration for
the pesticide concentration, which then determined PMORT, the probability of

112

 

 

 

5 Select on Random mating Habitat choice Oviposition
Fl genotype §-> -~ with respect to -

: E A genotype

'. .............. .0

 

 

 

 

 

 

 

 

 

. Figure 1. Schematic for the life cycle of a pest and its biology relevant to the

pesticide resistance simulation model. - - - - Immature stages (e through upae)
are exposed to selection; ——- Adult stage: mates ran omly yet ooses
habitat to lay eggs based on A allele.

113

Table 1. Parameters used in a two-locus, two-allele model investigating the effects
of physiological resistance and behavioral avoidance on the evolution of pesticide
resrstance.

 

 

Proportion in Habitat Fitness
PMORT valued crop‘ Quality Costs

 

 

 

Figure rr Rr R a An AA UHS THS FCR FCA

 

Genetic Factors
2A 0.92 0.92 0.10 0.9 0.65 0.4 0.5 0.5 0.0 0.0
B 0.92 0.10 0.10 0.9 0.65 0.4 0.5 0.5 0.0 0.0
0.92 0.54 0.10 0.9 0.4 0.4 0.5 0.5 0.0 0.0
0.92 0.54 0.10 0.9 0.9 0.4 0.5 0.5 0.0 0.0
0.92 0.54 0.10 0.9 0.65 0.4 0.5 0.5 0.0 0.0
0.92 0.54 0.10 0.9 0.65 0.4 0.5 0.5 0.2 0.0
0.92 0.54 0.10 0.9 0.65 0.4 0.5 0.5 0.0 0.2
0.92 0.54 0.10 0.9 0.65 0.4 0.5 0.5 0.2 0.2
Operational Factors

310-110-100

3A 0.23 0.02 0.00 0.9 0.65 0.4 05 05 0.0 0.0
B 0.92 054 0.10 0.9 0.65 0.4 05 0.5 0.0 0.0
c 0.99 0.86 039 0.9 0.65 0.4 05 0.5 0.0 0.0
D 0.92 054 0.10 0.9 0.65 0.4 0.8 02 0.0 0.0
E 0.92 054 0.10 0.9 0.65 0.4 05 05 0.0 0.0
F 0.92 054 0.10 0.9 0.65 0.4 02 0.8 0.0 0.0
G 0.92 054 0.10 0.6 035 0.1 05 0.5 0.0 0.0
H 0.92 054 0.10 0.9 05 0.1 05 05 0.0 0.0
r 0.92 054 0.10 0.9 05 0.1 03 0.2 0.0 0.0

 

' Proportions for AA and a are variables X and Y, respectively.

114

 

 

 

 

 

 

 

 

 

E 8°“ )
_, .
(‘5 -
L. l
E ‘ 5° 4.- ¢ 8'
ur 40‘ i
o ..
tr:
'ti.’ 20-
00.1 1-0 1°
PESTICIDE CONCENTRATION

Figure 1. Relationshi between log(concentration) of a gesticide and percent

mortality for rr, Rr, an R enotypes. Arrows labeled A, and C correspond to
pesticide concentrations simu ated in Figures 3A-C.

115

mortality for the RR, Rr, and the rr genotypes in the pesticide treated habitat
(Figure 2 and Table 1).

Fitness costs for carrying R or a alleles reflect direct or indirect (pleiotropic)
effects of these alleles on survivorship and reproduction. In this model, if the fitness
of the RR genotype is 0.8 relative to the "wild-type” rr, the ”fitness cost" for the R
allele is defined to be 0.2. For this example, the relative survivorship of rr, Rr, and
RR genotypes would be 1, 10.1, and 102, respectively.

Mean survivorship was calculated for the AARR genotype as follows:

SW“ = [X‘(1-PMORTRR)‘THS + (l-X)‘UHS]‘FCR

The probability of finding an AARR genotype larva in the valued crop (treated
habitat) was X; the remaining portion of this genotype (1-X) was attributed to the
diversionary crop (untreated habitat). Survivorship in the treated environment was
governed by both the presence of a pesticide (1-PMORTRR), and also by the quality
of the host resources (TI-IS), while the survivorship in the untreated habitat was only
governed by the quality of that resource (UHS). The fitness cost for carrying the R
allele, F CR, was calculated to cause the same proportional reduction in survivorship
within both habitats.

Equations calculating survivorship of the other eight genotypes were:

S MRRc[(X+Y)/2‘(l—PMORTRR)‘THS+(1-(X+Y)/2)'UHS]‘FCR‘(1+FCA)/2
Sma-[Y‘(l-PMORTRR)‘THS+(1-Y)‘UHS]"FCR‘FCA

s m-[X°(r-PMORTR,)'THS+(1-X)'UHS]'(1+PCR)/2
Sm:[(X+Y)/2‘(1-PMORTR,)‘THS+(1-(X+Y)/2)‘UHS]‘(1+FCR)‘(1+FCA)/4
Sam-[Y‘(l-PMORTRI)‘TT-IS+(1-Y)‘UHS]‘(1+FCR)/2‘FCA
smsx'(1-PMOan)mrS+(1-X)-Urrs
Sm-[m+W/2‘(l-PMORTn)¢HS+(1-(X+ Y)/2)‘UHS]‘(1+FCA)/2

Sana-[Y‘u-PMORTHYTHS-r(1-Y)‘UHS]‘FCA

116

The genotype frequency for the first generation in each simulation was
assumed to be at Hardy-Weinberg and linkage equilibrium. Subsequent generations
had non-equilibrium genotype frequencies (due to linkage disequilibrium). To
calculate the genotype frequencies following each generation of selection, the
gamete frequencies were first calculated. For example:
frquR = [SWR+.5‘(SMR,+SMRR)+.25‘SMH,]/TOTAI, where TOTAL is the
sum of all genotype survivorships. After the frequencies of Ar, aR, and ar gametes
were calculated, the genotype frequencies for the subsequent generation were
calculated, based on random mating [i.e., frean, = (freqm. “ freqm) 4» (frqu, ‘
freqan)].

The multiplicative model for calculating survivorship, and the genotype times
environment interaction present when dividing the population into valued and
diversionary crops caused an over-all epistatic effect between the physiological
resistance and behavioral avoidance loci. Selection under epistatic conditions
generated linkage disequilibrium, hence there were multiple genotype frequencies
possible for particular gene frequency combinations. Adaptive landscapes (Gould,
1984) therefore could not predict evolutionary outcomes (Lewin, 1988). Easily
interpreted graphs (Figures 3 and 4) of evolutionary results were generated instead.
Four regions may be present in these graphs, delineating the absorbing states aarr,
aaRR, Mn, and AARR (the four corners of the graphs), to which populations
evolve.

To generate such graphs, trajectories of gene frequencies from successive
generations of a population were calculated, and initial gene frequencies for
populations were investigated at 0.01 gene frequency intervals. Typically, within 10
to 20 generations, each trajectory asymptotically approached one of the four
absorbing states. When gene frequencies approached a corner, the genotype to

which the population evolved was determined. Two successive trajectories with

117

different final genotypes indicated that a transition between end results had
occured, and the initial gene frequency from the last trajectory was stored as a
transition point. When trajectories from all 10,000 initial gene frequency
combinations had been calculated, the gene frequencies identified as transition

points were written to a file, and later used for generating Figures 3 and 4.

Simulation results and discussion
Interpretation of figures - For all the graphs generated from the simulation
. model, there are either three or four comers to which all the initial gene frequencies
will eventually evolve. For example, in Figure 4A four regions are demarcated. The
lines separating the regions are boundaries (unstable equilibria); populations
initialized at gene frequencies on either side of these lines will evolve to different
genotypes (different corners).

The evolutionary state most desirable for pest and pesticide resistance
. management is selection to the AArr genotype (upper left corner, Figures 3 and 4).
The AArr pest genotype would be most easily managed. Most of this genotype
would develop in the diversionary crop, because of their avoidance trait; the few
individuals developing in the valued crop would be physiologically susceptible to the
applied insecticide. The pest management strategy would be evolutionarily stable
when a pest population is selected to this genotype.

Selection to the AARR genotype (upper right corner) would diminish the
value of applying an insecticide in the valued crop, because of physiological
resistance. However, expression of avoidance would reduce pest density in the
valued crop.

Selection to the aaRR genotype (lower right corner) yields a non-avoiding,
physiologically resistant population, the worst evolutionary consequence for pest

management. Finally, selection to aarr may occur in rare, low selection intensity

118

 

 

 

 

 

 

 

 

 

 

1.0 1.0
30.33 303-1
3 r a .
£0.61: £0.61:
204-: go»; 0:} .................
" o ‘ ....... - , ......
° Oh meougn‘cv 8:953 Alfie 1.0 o Fgéousficv 83a ALgLE 1.0

1.0

ALLELE
g ’3

A‘A

o
h

FREQUENOCV OF A
3

av
o

 

 

ALA-x ALIA

Al-A

 

 

A
..

 

 

 

0.2 0.4
HIEOUENCYOFRALLELE

vvvvvvvvvvv

o
0.6 0.3 1.0 0

.-
O
0"

.

C .0

C O
0'
C
.C
'

 

 

 

   

vvvvv

0.2

vvvvvv

0.6 0.! 1.0

FREQUENCYOFRMLELE

Figure 3. Genetic factors influencing development of avoidance (A allele) or
ph iological resistance (R allele).

(A R allele is recessive,
allele is recessive, (E) no

,Bg ) R allele is dominant, (C) A allele is dominant, (D) A
tness costs; (F) fitness cost exists for R allele; (G) fitness

cost exists for a allele; (H) fitness costs exist for R and a alleles.

r‘
o

 

 

 

 

 

 

 

 

 

 

‘ A
‘ El
303-: a
J
< i <
0.6-l
as ; s
E 0.41 g
5’ . 3
3024 g
E j u.
o ................. "
o 0.2 0.4 0,5 0,3 1,0 0 0.2 0.4 0.6 0.8 1.0
FREQUENCY OF R ALLELE FREQUENCY OF R ALLELE '

 

 

 

FREQUENCY OF A ALLELE
o
?

 

 

 

 

 

 

 

 

vw'vvv'fivv

0,2 0,4 0,5 0.3 1.0 0 0.2 0.4 0.6 0.8 1.0
o msousucyopam FREQUENCYOFRNlELE

 

 

.0 .
O
LALLAAL.

resouegcvonmas
meouerfvonm
‘

9
o

o
b

 

 

 

 

 

O

 

0 LA. L
O
N
O
.
a
a
o
O
1"
o
O
o
N
O
‘
O
O
o
0
d
O

Figure 4. Operational factors influencing development of avoidance (A allele) or
physiological resistance (R allele). (A-C) pesticide is applied at low, moderate, and

igh concentrations, respectively (see arrows in Fig. 1); (DP) diversion crop
suitability is greater than, equal to and less than the suitability of the value crop;
(G) High availability of diversionary crop; (H) Low availability of diversionary crop
and high discrimination by Aa and AA genotypes; (I) an SDDS situation, including
low availability but high discrimination of Aa and A‘Aagenotypes for the diversionary
crops and low survivorship in both diversionary and ued crops.

120

situations. This genotype would be easily controlled with insecticides, because it is
physiologically susceptible, even though it does not avoid the valued crop.

The simulation results have been divided between genetic (mode of
inheritance and fitness costs for R and A alleles, Figure 3) and operational factors
(pesticide concentration, habitat quality, habitat availability, and an SDD example,
Figure 4). The distinction is made because operational factors are management-
determined variables (T abashnik and Croft, 1982).

Genetic factors - Dominance of the R allele shifted the line separating the
' AArr- and AARR-destined regions further to the left (Figure 3 B vs. 3 A), meaning
fewer gene frequency combinations were selected to the Mn genotype.
Dominance of the A allele (Figures 3 C and 3 D) had less effect on selection than
dominance of the R allele; the boundary shifted significantly only when A and R
alleles were both at low frequencies. The co—dominant expression of R and A
alleles (Figure 3 E) yielded results intermediate to recessive and dominant
inheritance. {

1f fitness of individuals carrying resistance alleles were greater than that for
carriers of susceptible alleles, gene frequencies for resistance would increase
without selection with insecticide, making these resistance alleles common in
unselected populations. Insecticides are not labeled for preadapted pests, so
resistance alleles for pesticides are usually at very low initial frequencies. Hence,
fitness of resistance alleles less than or equal to susceptible alleles should be a
robust assumption. Fitness costs have been measured in several instances; they
reduced intrinsic rates of growth in resistant strains by as much as 20 percent (Crow,
1957; Argentine, et al., 1989). Costs for carrying A or a alleles should also be
possible. Pests may develop greater sensitivity to insecticides, and evolve avoidance
responses (Gould, 1984; Lockwood, et al., 1984; Sparks, et al., 1989). A more

efficient use of avoidance would be to take advantage of a pre-existing common

121

behavioral response. By the same reasoning used for insecticides above, applying a
deterrent would not be practical unless the target pest is preadapted to respond;
hence evolution of non-avoidance and fitness costs for the a allele should be
considered.

Fimess costs subtly biased selection in these simulations. The greatest effect
of a fitness cost for an a allele (Figures 3 G vs. 3 E and Figures 3 H vs. 3 F) was to
shift the proportion of initial gene frequencies evolving to AARR rather than to
aaRR. The effect of including a fitness cost for the R allele (Figure 3 F vs. 3 B and
Figure 3 H vs. 3 G), was an increase in the proportion of gene frequencies evolving
to the AArr genotype. Fitness costs for both the R and a alleles (Figure 3 H vs. 3 E)
simply combined the effects mentioned above.

Operational factors - Pesticide concentration (Figure 2, arrows A, B, and C)
greatly affected selection of R and A alleles (Figure 4 A-C). At low pesticide
concentrations (Figure 2, arrow A) and low to moderate gene frequencies for
physiological resistance, selection operates to diminish physiological resistance
(Figure 4 A). Interestingly, if the A allele is also at low frequency, selection may
drive the population to the aarr genotype. This simulation corresponds to a
functionally dominance R allele, because at low pesticide concentrations, the Rr and
RR genotypes suffer similar mortality (Taylor and Georghiou, 1979).

The greatest selection intensity for physiological resistance occurs when
pesticide concentrations maximally discriminate between rr, Rr, and RR genotypes
(Taylor and Georghiou, 1979). At a moderate pesticide concentration (Figure 2,
arrow B; Figure 4 B) selection was greatest for the R allele and the AARR
genotype. At a still higher pesticide concentration (Figure 2, arrow C and Figure 4
C), a greater number of initial gene frequency combinations evolved to the AArr
genotype (note that the upper-left region in Figure 4 C is larger that the same region
in Figure 4 B).

122

Habitat suitability is a measure of the quality of the environment for pest
development. Suitability is itself a composite of abiotic and biotic factors
contributing to survivorship within a habitat. To be realistic, suitability should
include the effects of predators, parasitoids, pathogens, host-plant quality,
competitors, sub-optimal abiotic conditions, and even density-dependent
development restraints.

Maintaining a moderate pesticide dosage and changing the relative habitat
quality of valued and diversionary crop demonstrated the profound influence of the
habitat suitability on selection of the avoidance trait (Table 1; Figures 4 DP).
When the valued crop was more suitable for pest development than the diversionary
crop (not considering the effect of pesticides), avoidance was selected against unless
R alleles were absent (Figure 4 F). Lower survivorship in the diversionary crop may
be common when the only ”diversionary crop” in conventional agriculture is wild
hosts.

As diversionary crop suitability improved relative to the valued crop, the
region representing gene frequencies evolving to the AArr state greatly increased,
and the number of gene frequency combinations evolving to aaRR was greatly
reduced (Figure 4 D-F). If physiological resistance was below a critical value, the
AArr genotype can evolve rather than physiological resistance genotypes.

Variables X and Y in the BASIC model were the probabilities AA and aa
genotypes will lay eggs in the valued crop. Low values imply a complementarily high
probability the eggs would be deposited in the diversionary crop. Manipulation of
these values represents changes in the availability of or relative discrimination by
the avoidance genotypes for the diversionary crop. Figure 4 B represents low
availability of diversionary crop habitat (X =0.4, Y = 0.9). Figure 4 G represents
higher availability of diversionary crop (X=O.1, Y=0.6) relative to Figure 4 B.

123

Figure 4 H simulates low availability of diversionary crop, but very highly
discriminating avoidance genotypes (X=O.1, Y=0.9).

There is an interaction between habitat availability and the avoidance trait.
Having a readily available diversionary crop reduced selection for physiologically
resistant genotypes, as expected from the reduced selection intensity on that trait
(Figure 4 G vs. 4 B). Nearly the same selection diagram was generated in Figure 4
H as in 4 G, implying that greater discrimination by Aa and AA genotypes can be
exchanged for greater availability of diversionary habitat. The greatest difference
between these two graphs occurred when A and R alleles were at low frequencies.
In Figure 4 G, this region evolved to the aarr genotype, while in Figure 4 H, the
same initial gene frequencies evolve mostly to the aaRR or AArr genotypes.

SDD optimized both pest and pesticide resistance management when: 1) the
diversionary crop habitat was of higher quality for pest development than the valued
crop, 2) an insecticide was used to protect the valued crop, and 3) a deterrency trait
was exploited. Under these conditions, a great number of initial gene frequencies
will evolve to the most desired genetic state, an AArr population (Figures 4 D and 4
I vs. Figure 4 B).

An operational factor hidden in the graphs is choice of a deterrent to which
the pest is or is not preadapted to respond. If a pest is preadapted to avoid a
deterrent, then the population would have a large initial A allele gene frequency;
non-preadaptation implies low initial avoidance. In all the simulation conditions,
the line separating the areas evolving to the AArr vs. AARR genotypes slopes to the
right. This is important - if a deterrent is chosen to which the pest is not adapted to
respond, the A allele is at low gene frequency and there is a high probability that the
population will be evolve to AARR. If a deterrent is chosen to which the pest is
preadapted to respond, the population is likely to evolve to the AArr genotype.

124

Practical considerations for SDD

Deploying suitable diversionary crops for pests permits expression of
normally precluded selective difierences. Diversionary crops have been criticized
because they may act as breeding grounds for the pest (Cromartie, 1981). This need
not be a problem, if the number of pests emerging from the diversionary crop can be
managed by destroying part of the infested crop. Manipulating pest survivorship
differences in valued vs. diversionary crops should be viewed as necessary for both
pest and pesticide resistance management. This is one feature distinguishing SDD
strategy from previous pesticide resistance management models exploiting pest
populations developing on wild hosts or abandoned orchards (T abashnik and Croft,
1982).

Suppressing the pest population in a diversionary crop should be facilitated
by first using behavioral modifying cues (chemical or physical) to confine the pest
population. Under high pest densities, intraspecific competition may reduce
. survivorship. At somewhat lower populations, pest densities may exceed Reed-Frost
epizootic thresholds (Hoppenstaedt, 1982), or facilitate host finding by predators or
parasitoids.

One major advantage of SDD is that pesticides remain useful tools.
Pesticides might be compatible with biological control and SDD because the
majority of beneficial agents would be maintained in the non-treated habitat.
Insecticides in an SDD system could actually help select for continued control, by
driving selection for avoidance of the valued crop, and hence, improving partitioning
of the population to the diversionary crop. If SDD partitioned the pest away from
the valued-crop and reduced the pest density below economic thresholds, pesticides
rates could be reduced or intermittently deleted.

Idealized use of resources (deterrents, pesticides, space used for diversionary

crops) for accomplishing pest and pesticide resistance management may be very

125

challenging. Obviously it requires a thorough knowledge of a particular pest's
behavioral and physiological traits, as well as the economics for growing the crop.
Because selection requires several generations, temporal optimization is also
possible; e.g., allocation of space for diversionary crops and pest emergence allowed
from this habitat could initially be great enough to maximize evolutionary rate
toward the AArr genotype, while simultaneously reestablishing viable populations of
biological control organisms. Later, as the AArr genotype is being selected, space
allocation to the diversionary crop, pesticide concentrations in the valued crop,
and/ or pest emergence from the diversionary crop could be reduced.

In a broader, sustainable view of pest management, plant breeders and
genetic engineers should eventually replace insecticide applications with antibiotic
and antixenotic crop traits. These traits have the same selective consequences for
pest populations as conventional pesticides. Thus ecological management tools,
such as SDD, should be evaluated to prevent squandering the limited genetic
resource of pest susceptibility (Gould, 1984; 1986; Dover and Croft, 1984).

Summary

High-kill strategies for managing insecticide resistance may be short lived.
Multiple resistance mechanisms make eventual failure of these approaches nearly
certain, unless isolated populations lack genetic variance sufficient to respond to the
high selection, or if a continual swamping of resistance genes occurs due to
immigration of susceptible individuals. These pesticide management approaches
are also incompatible with ecological closure aspects of sustainable agriculture,
unless antibiosis crops are substituted for insecticide applications (Edens and
Koenig, 1980). Even then, resistance development is likely unless ecological aspects
of pesticide resistance are taken into consideration.

Ecological management of pesticide resistance through SDD is
revolutionary. By taking advantage of behavior, pests could be manipulated to be

126

less damaging to the valued crop and concentrated in a diversionary crop, where
biological control can be effective. If an SDD approach is implemented before gene
frequencies for physiological resistance pass a critical value beyond which the
population would be selected to the AARR genotype, and a deterrent is used on the
valued crop, then avoidance behavior will be selected for and physiological
resistance selected against. The consequence is an evolutionarily stable pest
management strategy that removes conflict between the agriculturist and pest. Such
an ecological approach redefines the time scale for successful pesticide resistance
management, and re-castes the IPM practitioner as a selective breeder for beneficial
pest traits. The critical role of host acceptance behavior for implementing SDD
emphasizes the role insect behavior can play in generating new ecologically based

IPM and sustainable agriculture methods.

Thesis Summary

127

Thesis Summary

Onions and onion models were used to bioassay onion fly host acceptance
behavior, with the goal of developing strategies for controlling onion fly (Delia
antiqua (Meigen)) oviposition. Stimulo—deterrent diversion (SDD) was developed,
where the valued crop (seedling onions) are treated with chemical deterrents, and
simultaneously, highly stimulatory ovipositional resources (sprouting cull onions)
are deployed to concentrate eggs away from the crop.

Wm - Sprouting onion bulbs planted with
the neck at ca. 5 cm below the soil surface stimulated more oviposition in a field
experiment than bulbs planted either shallowly or at greater depth (Chapter 1). The
stimulatory nature is probably attributable to the large number of leaves that project
through the soil surface, each of which is an ovipositional resource. The visual and
physical characteristics of these leaves closely match optimally stimulatory surrogate
onion foliage (Harris, et al., 1987). In addition to the visual/physical stimuli,
chemical cues generated through larval feeding and microbial decomposition of the
bulb also enhance onion fly oviposition (Hausmann and Miller, 1989a; b). A
factorial experiment investigating bulb damage and larval feeding on onion fly
oviposition in the greenhouse suggested that bulbs remain highly stimulatory under
great latitude of plant conditions. As long as onion bulbs are planted at ca. 5 cm
depth, they can be expected to divert eggs from being deposited near seedlings.

’ .u ... w Hutrtsur . r e urn”! um ”no. .- 14. -
Onion flies were sensitive to a wide array of deterrent compounds (Chapter 3 and
4), such as pungent spices, monoterpenoids and cinnamyl derivatives. Appreciably

deterrent compounds were C8 to C13, intermediate in polarity, and possessed either
128

129

oxygen-containing or nitrile functional groups. In general these compounds were
active at 10-100 fold higher concentrations than is dipropyl disulfide, a strong
ovipositional stimulant (Harris and Miller, 1988). Broad but low sensitivity to
phytochemicals does not mean such chemicals are unimportant in shaping insect-
plant relationships. In an environment where there is a veritable pharmacopoeia of
secondary compounds present in non-hosts, sensory apparatus for stenophagous
herbivores like onion flies may detect ”foreign" compounds when they are at their
usual high, and consequently potentially toxic concentrations.

At a practical level, broad but low sensitivity has trade-offs. Various insect
pests may respond (Dethier, 1947; J ermy, 1983) to compounds that deter onion flies,
making these deterrents interesting because they could have "broad spectrum"
activity. To be effective though, these deterrents may need to be applied at
concentrations that currently make commercial development a challenge. A more
practical approach might be to engineer deterrency into the crop plants. Traits such
as antixenotic color (Prokopy, et al., 1983), or physical structure could enhance host-
plant chemical deterrents. An alternative to deterrent host-plants-could be the use
of intercrops that are themselves repellent (Atsatt and O'Dowd, 1976).

WWW - PhytOPhagous insects integrate
information from different sensory modalities to assess host-plant quality (Miller
and Strickler, 1984; Miller and Harris, 1985). The neurophysiological basis for
sensory integration is not well understood; paradigms for sensory integration and
their expected behavioral characteristics are described in Chapter 5. Deterrents are
expected to be most effective under the paradigm of "classical behavioral chaining"
(Miller and Harris, 1985). Deterministic behavioral sequences could be disrupted
at multiple points, each of which , is necessary for the end result of host-plant
acceptance. Deterrents are expected to be less effective when sensory information '

is temporally integrated via probabilistic behavioral transitions. Behavioral webs

130

allow some behavioral sequences to proceed to oviposition, in spite of the presence
of suboptimal cues.

Interactions were studied within sensory modalities with chemical deterrents
(Chapter 5), between modalities for chemical and visual deterrents (Chapter 5), and
between positive and negative stimuli (Chapter 6). Behavioral observations
supported temporal summation of across-modality sensory cues, which would
generate independence of marginal acceptance probabilities and multiplicative
effects. Analysis of variance on log-transformed egg counts was an innovative
approach for studying end results of across-modality interactions and testing their
multiplicative nature.

Chapters 2, 3, and 6 discussed interactions between external sensory and
internal factors in governing host-plant acceptance. The loss of deterrency under
conditions of ovipositional deprival suggests that deterrents may only be practical in
field conditions if pests are offered highly acceptable ovipositional alternatives
(SDD). A greenhouse test of SDD confirmed that the combination of deterrents
applied to seedlings and adjacent sprouted cull onions more effectively protects
seedlings from oviposition than either deterrents or culls alone (Chapter 6). The
multiplicative effect for reduced numbers of eggs laid on seedlings can be explained
through independently reduced acceptance probabilities when deterrent or culls are
present.

W - A population genetics model using two-allele loci
for avoidance and physiological resistance traits (Chapter 7) suggested that SDD
combined with conventional insecticides could prevent or reverse pesticide
resistance development. Requirements were: 1) higher suitability of the
diversionary crop, 2) high finding of the diversionary crop, and 3) deterrents to
which a pest is preadapted to respond.

131

This model has important parallels in natural systems, if suitability of plants
and adaptation to toxins are substituted for the pesticide / physiological resistance
component. The model generated three or four ”adaptive peaks" in which the
homozygous populations of AArr, AARR, or aaRR genotypes had higher fitness
than heterozygotes. These results suggest that evolution of specialization may be
driven by the quality and quantity of alternative host-plant resources and gene
frequencies for physiological and behavioral traits. The fitness disadvantage for
heterozygotes implies that assortative mating may evolve; the avoidance allele could
' facilitate mate finding, thus causing assortative mating to take place. Hence, there
is the possibility for host races to form on host plants with different suitabilities, and
the pesticide resistance model converges with models for sympatric speciation (Bush
and Diehl, 1982).

The population genetics model covered situations in which the quantities and
qualities of alternative host-plant resource types were constant over time. In natural
' conditions, resource types may vary over time. Under these conditions, populations
may respond genetically, through changes in gene frequencies for traits allowing
adaptation to different host plant types. An alternative to genetic change is offered
by behavioral flexibility. The observed effects of deprivation, and learning in other
insects (Papaj, 1986), suggest that individual ovipositing insects can respond to
paucity of suitable host resources by ovipositing on sub-optimal plants.

Won: - Cull onions, or deterrents applied to seedling onions
by themselves probably would not provide economic control of onion flies unless the
onion fly population is already very low, or the onion seedlings had strong
antixenotic/ antibiotic properties. The combined use of deterrents / antixenosis and
stimulatory resources (SDD) could prevent economic damage from onion maggots.
This SDD approach may have general appeal for other crops and pests, especially A

when: 1) deterrents or antixenotic cultivars are already available, 2) trap crop

132

species have already been identified, 3) reduced pesticide application is highly
desirable, or 4) biological control agents require pest densities that cause economic
damage under conventional agronomic practices.

The expected importance of SDD concepts for implementing biological
control and pesticide resistance management suggests that behavioral manipulation
can play a central role in developing a transition from conventional chemocentric
management practices to sustainable agricultural systems.

WM - Does stimulus summation occur within or
across sensory modalities, at the neurophysiological level?

Do onion flies vary the number of eggs laid per depositional bout in response
to host quality? If they do, this argues for either 1) the fly having a Gestalt (across
modality summation) perception of host plant quality or 2) concurrent examining
and depositional behaviors.

Is there genetic variation for deterrency? How can this genetic variation best
be measured, considering that acceptance is dynamically influenced by deprivation?
Is there deterrency to pesticides in onion flies? If there is, is it an evolved response?

What are the implications of quantitative genetics for SDD? Could
intermediate levels of avoidance or physiological resistance evolve, rather than
fixation at physiological or behavioral extremes?

What would the population genetics model predict under conditions of
overdominance in the behavioral trait? Would host races form if mating is
associated with choice of ovipositional habitat? If host races are formed, would the
the race specialized on the valued crop be more difficult to control than if the SDD
approach had not been used?

LIST OF REFERENCES

LIST OF REFERENCES

Alfaro, R. 1., Pierce, H. D., Jr., Borden, J. H., and Oehlschlager, A. C. 1981. Insect
Esq-in 33911-28 oviposition deterrents from western red cedar foliage. J. Chem
0 : .

Ames, B. N. and Gold, L. S. 1989. Pesticides, risk, and applesauce. Science
244:755-757.

Argentinez J. A., Clark, J. M., and Ferro, D. N. 1989. Relative fitness of insecticide-
resrstant Colorado tato beetle strains (Coleoptera: Chrysomelidae).
Environ Entomol. 18: 05-710.

Atsatt, P. R. and O'Dowd, D. J. 1976. Plant defense guilds. Science 193:24-29.

Bemays, E. and Graham, M. 1988. On the evolution of host specificity in
phytophagous arthropods. Ecology 69:886-892. ‘

Brattsten, L. B. 1983. Cytochrome P-450 involvement in the interactions between

Klimt terpenes and insect herbivores. pp. 173-195 In: Plant Resistance to

cts. P. A. Hedin (ed.).Amer. Chem. Soc. Symp. Series 208. ACS.
Washington, DC.

Brattsten, L. B. 1988. Enzymic adaptations in leaf-feeding insects to host-plant
allelochemicals. J. Chem Ecol. 14:1919-1939.

Brattsten, L. B., Holyoke, C. W., Jr., Leeper, J. R., and Raffa, K. F. 1986.
Insecticide resrstance: challenge to pest management and basic research.
Science 231:1255-1260.

Buchanen, R. I.. 1978. Toxicity of spices containing methylene dioxybenzene
derivatives: a review. J. Food Safety 1:275-293.

Bunker, C. W. O. and Hirschfelder, A. D. 1925. Mosquito repellents. Am. J. Trap.
Med. 5:359-383.

Bush, G. L. and Diehl, S. R. 1985. Host shifts, genetic models of sgnpatrjfi
speciation and the origin of parasitic insect species. p. 297-305 In: oc.
Internat. S . Insect-Plant Relationships. J. H. rsser and A. K. Minks,
(eds.). P . Wageningen, Netherlands.

Camhi, J. M. 1984. Neuroetholo : Nerve Cells and the Natural Behavior of
Animals. Sinauer. Sunderlan MA.

133

134

Carruthers, R. I. 1981. The biology and ecology of Entomo hthora muscae (Cohn)
i1n7 1the onion ecosystem. h.D. Dissertation. Mic 'gan State University.
PP-

Carruthers, R. I. and Haynes, D. L. 1986. Temperature, moisture, and habitat
effects on Entomopthora muscae (Entomopthorales: Entomopthoraceae).

Conidial germination and survival in the onion ecosystem. Environ. Entomol.
15:1154-1160.

Carruthers, R. 1., Whitfield, G. H., and Haynes, D. L. 1985. Pesticide-induced
mortality of natural enemies of the onion maggot, Delia antiqua
(Dip.:Anthomyiidae). Entomophaga 30: 15 1-161. .

Casagrande, R. A. 1987. The Colorado tato beetle: 125 years of
mismanagement. Bull. Entomol. Soc. Am. 3 :142- 150.

Cerf, D. C. and Georghiou, G. P. 1972. Evidence of cross-resistance to a juvenile
hormone analogue in some insecticide-resistant houseflies. Nature 239:401.

Chapman, R. F. 1988. Sensory aspects of host-plant recognitin by Acridoidea:
questions associated with the multiplicity of receptors and variability of
response. J. Insect Physiol. 34:167-174.

Clement, D. B. 1987. N CA summer regional meeting, optimistic attitudes
prevailed. Onion World 3(5):6-10.

Cochran, W. G. 1954. Some methods for strengthening the common Chi squared
tests. Biom. 10:417-451.

Coley-Smith, J. R. and King, J. E. 1969. The production by species of Allium of
alkyl sulphides and their effect on ermination of sclerotia of Sclerorium
cepivorum Berk. Ann. Appl. Biol. 64: 9-301.

Comins, H. N. 1977. The development of resistance in the presence of migration.
J. Theor. Biol. 64:177.

Courtney, S. P., Chen, G. K. and Gardner, A. 1989. A general model for host
selection. Oikos 55:55-65. .

Courtney, S. P. and Kibota, T. T. 1989. Mother doesn't always know best: selection
of hosts by ovipositing insects. In: Insect-Plant Interactions, Vol. 2. E. A.
Bernays (ed.). CRC Press. Boca Raton, Florida

Cowles, R. 8., Keller, J. E. and Miller, J. R. 1989. Pungent spices, ground red
grippers sing 3(thetic capsaicin as onion fly ovipositional deterrents. J. Chem.
1 : 1 - .

Cowles, R. S., Miller, J. R., Hollingworth, R. M., Abdel-aal, M. T. Szurdoki, F.,
Bauer, K., and Matolcsy, G. 1990. Cinnamyl derivatives and
monoterpenoids as non-specific ovipositional deterrents of the onion fly. J.
Chem. Ecol. In press.

135

Croft, B. A. and Brown, A. W. A. 1975. Re use of arthropod natural enemies to
insecticides. Ann. Rev. Entomol. 20: -355.

Cromartie, W. J ., Jr. 1981. The environmental control of insects using crop
diversity. pp. 223-251 In: CRC Handbook of Pest M ement in
Agriculture, 01. 1. D. Pimentel (ed.). CRC Press. Boca Raton,

Crow, .12.2F2.7 3557. Genetics of insect resistance to chemicals. Ann. Rev. Entomol.

Curtis, C. F. 1985. Theoretical models of the use of insecticide mixtures for the
management of resistance. Bull. Entomol. Res. 75:259-265.

Davis, E. E. .1985. Insect repellents: concepts of their mode of action relative to
gotentral sonsory mechanisms in mosquitoes (Diptera: Culicidae). J. Med.
ntomol. 22:273-243.

Davis, E. E. and Sokolove, P. G. 1976. lactic acid-sensitive rece tors on the
antennae of the mosquito, Aedes aegypti. J. Coma Physiol. 105:4 -54.

Dawkins, M. S. 1983. The organization of motor patterns. pp. 75-99 In: Animal
Ezeshaviour, Vol. 1. T. R. Halliday and P. J. B. Slater. (eds.). W. H. Freeman.
pp.

Den Ouden, H. and Theunissen, J. 1980. Controlled release of naphthalene: a
fipf’llirt a‘17n3t5 oviposition of the cabbage root fly, Delia brassrcae. Neth. J.
at : - .

Dethier, V. G. 1947. Chemical Insect Attractants and Repellents. Blakiston.
- Philadelphia, PA. 289 pp.

Dethier, V. G. 1980. Evolution of receptor sensitivity to secondary plant substances
with special reference to deterrents. Am. Nat. 115:45-66.

Dethiegag.9 ?6 1982. Mechanism of host-plant recognition. Entomol. exp. appl.

Dindonis, L. L. and Miller, J. R. 1980. Host finding responses of onion and
seedcom flies to healthy and dewmposifbg onions and several synthetic
constituents of onion. Environ. Entomo 9: 7-472.

Dindonis, I... L. and Miller, J. R. 1981. Onion fly and little house fly host finding
flelctgvilggggiated by decomposing onion and microbial volatiles. J. Chem.
co : .

Dover, M. and Croft, B. 1984. Gettin Tough: Public Policy and the Management
of Pesticide Resistance. World esources Institute. Washington, D. C. 80

pp.

Drummond, F. 1982. Post-harvest biology of the onion maggot, Hylemya antiqua.
MS. Thesis, Michigan State University.

Dyte, C. E. 1972. Resistance to synthetic 'uvenile hormone in a strain of flour
beetle, Tribolium castaneum. Nature 8:48.

136

Eckenrode, C. J. 1988. Delia antiqua (Meigen): an increasing threat in North
America. In: International Symposium on Diptera sts in vegetable crops.
J. Narkiewicz-Jodko and B. Nawrocka, (eds.). pp. 2 -30.

Eckenrode, C. J. and Nyrop, J. P. 1986. Impact of physical injury and commercial
' ' on d e to onion bulbs by larvae of onion maggot (Diptera:
Anthomyiidae). . Econ. Entomol. 79: 1606- 1608.

Edens, T. C., Fridgen, C. and Battenfield, S. L. (eds.) 1985. Sustainable Agriculture
and Integrated Farming Systems. Michigan State University Press.

Edens, T. C. and Koenig, H. E. 1980. oec tern management in a resource-
limited world. Bioscience 30:697-‘fil811: osys

Eisner, T. 1970. Chemical defense ainst predation in arthropods. p . 157-217 In:
l(grhemrcal Ecology. E. Son eimer and J. B. Simeone (eds.)). Academic
ess.

Finch, S. 1978. Volatile plant chemicals and their effect on host plant finding by
the cabbage root fly (Delia brassicae). Entomol. exp. appl. 24: 150- 159.

Finch, 8.1989. Ecological considerations in the management of Delia pest species
m vegetable crops. Ann. Rev. Entomol. 34:117- 13 .

Finch, S., Cadoux, M. E., Eckenrode, C. J ., and Spittler, T. D. 1986a. Appraisal of
current strategies for controlling onion maggot (DipterazAnthomyiidae) in
New York State. J. Econ. Entomol. 79:736-7 .

Finch, S. and Eckenrode, C. J. 1985. Influence of unharvested, cull-pile, and
volunteer onions on fulations of onion maggot (Diptera: Anthomyiidae).
J. Econ. Entomol. 78: 4 -546.

Finch, S., Eckenrode, C. J ., and Cadoux, M. E. 1986b. Behavior of onion maggot
(DipterazAnthomyiidae) in commercial onion fields treated regularly wrth
parathion sprays. J. Econ. Entomol. 79:107-113.

Fitt, G. P. 1986. The influence of a shortage of hosts on the specificity of
oviposition behaviour in species of Dacus (Diptera, Tephritidae). Physiol.
Entomol. 11:133-143.

Fraenkel, G. 1959. The raison d 'etre of secondary plant substances. Science
129: 1466- 1470.

Fukuto, T. R. and Mallipudi, N. M. 1983. Suppression of metabolic resistance
throu chemical structure modification. pp. 557-578 In Pest Resistance to
Pesticides. G. P. Georghiou and T. Saito (eds.). Plenum. NY.

Georghiou, G. P. and Mellon, R. B. 1983. Pesticide resistance in time and s ace.
. 1-46 In Pest Resistance to Pesticides. G. P. Georghiou and T. aito
(Eds). Plenum. NY.

Geor 'ou, G. P. 1983. Management of resistance in arthropods. . 769-792 In
8hFest Resistance to Pesticides. G. P. Georghiou and T. aito (£55.). Plenum.
NY.

137

Girolami, V., Vianello, A., Strapazzon, A., azzi, E. and Veronese, G. 1981.
Ovipositional deterrents in Dacus oleae. nt. exp. & appl. 29:177-188.

Gordon, H. T. 1961. Nutritional factors in insect resistance to chemicals. Ann. Rev.
Entomol. 6:27-54.

Gothilf, S. and Bar-Zeev, M. 1972. Effect of va r repellents on the electrical
response of insect antennae. Experientia 28: 1-603.

Gould, F. 1984. Role of behavior in the evolution of insect ada tation to
insecticides and resistant host plants. Bull. Entomol. Soc. Am. 30: -51.

Gould, F. 1986. Simulation models for predicting durability of insect-resistant germ
lasm: Heft??? £13 (Diptera: Cecrdomyiidae)-resistant winter wheat. Envzro' n.
ntomo : - .

Grafius, 13., Hayden, J ., and Jenkins, M. 1987. Management of insect pests of
omons. 1987 Research Report to the Michigan Omon Research Committee.

Grafius, E. and Warner, F. W. 1990. Predation by Bembidion quadrimaculatum
(5Coleoptera: Carabidae) on the eggs of the onion maggot (Delia antiqua).
nviron. Entomol. In Press.

Grant, D. 1986. Glutathione S-transferase in Aedes aegypti. PhD. Dissertation.
Michigan State University.

Groden, E. 1982. The interactions of root maggots and two parasitoids Aleochara
bilineata (Gryll.) and Aphaereta pallipes (Say). Master's Thesis, Michigan
State University. 152 pp.

Guyer, G. and Wells, A. 1959. Evaluation of insecticides for control of the
chlorinated h3ydrocarbon resistant onion maggot. Quar. Bull. Mich. Agric.
Expt. Sta. 41( ):614-623.

Hammond, G. H. 1924. The onion maggot in the Ottawa District. A. Rep. Entomol.
Soc. Ontario 1923 55:39-41.

Harris, C. R., Tolman, J. H. and Svec, H. J. 1982. Onion maggot
(Dlilptera:Anthomyiidae) resistance to some insecticides following selection
wi parathion or carbofuran. CanEnt. 114:681-685.

Harris, M. 0., Keller, J. B., and Miller, J. R. 1987. Responses to n-dipropyl
disulfide oviposi ' onion flies: Effects of concentration and site of
release.J. Ecol. 3:1261-1277.

Harris, M. O. and Miller, J. R. 1982. Synergism of visual and chemical stimuli in

the ovi sition behavior of Delia antiqua. pp. 117-122 In: Proc. 5th Int.

S p. ct Plant Relationships. J. H. Visser and A. K. Minks (eds.).
ageningen, The Netherlands. UDOC, Wageningen.

Harris, M. O. and Miller, J. R. 1983. Color stimuli and ovipositional behavior of
the onion fly, Delia “#0 (Meigen) (Diptera: Anthomyiidae). Ann.
Entomol. Soc. Am. 76:766- 1.

138

Harris, M. O. and Miller, J. R. 1984. Foliar form influences ovipositional behavior
of the onion fly. Physiol. Entomol. 9: 145-155.

Harris, M. O. and Miller, J. R. 1988. Host acceptance behavior in an herbivorous
fly, Delia antiqua. J. Insect Physiol. 34:179-180.

Harris, M. O. and Miller, J. R. 1991. Quantitative analysis of onion fly egglaying
behavior: effects of a host-plant chemical. J. Insect Behavior In Press.

Harris, M. 0., Miller, J. R., and de Ponti, O. M. B. 1987. Mechanisms of resistance
to onion fly egg-laying. Entomol. exp. appl. 43:279-286.

Hausmann, S. M. and Miller, J. R. 1989a. Ovipositional preference and larval
survryal of the onion magfit (Dgptera: Anthomyiidae) as influenced by
prevrous maggot feeding. . on. ntomol. 82:426-429.

Hausmann, S. M. and Miller, J. R. 1989b. Production of onion fly attractants and
ggqmiistiggfltémulants by bacterial isolates cultured on onion. J. Chem.

Havukkala, I. 1982. Deterring oviposition of the cabbggse root fly, Delia radicum
(firgtlesrazAnthomyiidae), by non-chemical meth . Acta Entomol. Fenn.

Hokkanen, H., Granlund, H., Husbe G.-B., and Markkula, M. 1986. Trap crops
used successfully to control eligethes aeneus (Col., Nitidulidae), the rape
blossom beetle. Ann. Entomol. Fenn. 52:115-120.

Hoppenstaedt, F. C. 1982. Mathematical methods of population biology.
Cambridge Univ. Press. NY. 149 pp.

Hoskins, W. M. and Gordon, H. T. 1956. Arthropod resistance to chemicals. Ann.
Rev. Entomol. 1:89-122. .

House, G. J. and Alzugaray, M. R. 1989. Influence of cover croppinfiland no-tillage
practices on community composition of soil arthropods m a orth Carolina
agroecosystem. Environ. Entomol. 18:302-307.

Howard, J. J., Cazin, J ., Jr., and Wiemer, D. F. 1988. Toxici of terpenoid
deterrents to the leafcutting ant Ana cephalotes and its mutuailis’ tic fungus. J.
Chem. Ecol. 14:59-69.

Howitt, A. J. 1958. Chemical control of lernia antiqua gMgié.) (Diptera:
Anthomyiidae) in the Pacific Northwest. . Econ. Entomol. 1: -887.

Ishikawa, Y., Ikeshoji, T., Matsumoto, Y., Tsutsumi, A., and Mitsui, Y. 1983. 2-
Phenethanol: an attractant for the onion and seed-com flies, lernya '
and H. platura (Diptera: Anthomyiidae). Appl. Ent. Zool. 18:2 0-277.

J aenike, J. 1983. Induction of host preference in Drosophila melanogaster.
Oecologia 58:320-325.

139

Javer, A, Wynne];1 AdD., Bordefn, Jth, and Judd, G. J 0R1.” 1987. PinFMoil: ar)1
ovi rtro eterrent or e onion ot, e ' antiqua eigen
(Dripcfzra: Anthomyiidae). Can. Ent. 119:60-5afil9.

J ermy, T. 1971. Biolo 'cal background and outlook of the antifeedant approach to
insect control. cta Phytopathol. Acad. Sci. Hung. 6:253-260.

J ermy, T. and Szentesi, A. 1978. The role of inhibitory stimuli in the choice of
ovrposition site by phytophagous insects. Entomol. exp. appl. 24:458-471.

J ermy, T. 1983. Multi licity of insect antifeedants in plants. p . 223-226 In:
NaturalProducts or Innovative Pest Management. Vol. 2, ent Themes
in Troplrclal Science. D. L. Whitehead and W. S. Bowers (eds.). Pergamon
Press. .586 9?-

J ones, T. H., Cole, R. A. and Finch, S. 1988. A cabbage root fly oviposition

deterrent in the frass of arden bble moth cate illars. Entomol. ex . a l.
49:277-282. g pe rp p pp

Judd, G. J. R. and Borden, J. H. 1989. Distant olfactory response of the onion fly,
Delia antiqua, to host-plant odour in the field. Physiol. Entomol. 14:429-441.

Kendall, E. W., Jr. 1932. Notes on the onion ot (Hylemya antiqua, Meigen).
A. Rep. Entomol. Soc. Ontario 1931 62:82-83.

Kenne , J. S., Booth, C. O. and Kershaw, W. J. S. 1961. Host finding by aphids in
e field. III. Visual attraction. Ann. Appl. Biol. 49:1-21.

Klocke, J. A., Balandrin, M. F., Barnby, M. A. and Yamasaki, R. B. 1989.
Limonoids, phenolics, and furanocoumarins as insect antifeedants, repellents,
and growth inhibitory com unds In: Insecticides of Plant Origin, ACS
Symposium Series 387. J. . Arnason, B. J. R. Philogne, and P. Morand
(eds.). Amer. Chem. Soc., Washington, D. C.

Klowden, M. J. 1989. Influence of the ovaries and fat body on the initiation and

termination of re-ovi sition behavior in the mosquito, Aedes aegypti. J.
Insect Physiol. 3 :567-5g3.

Kogan, M. 1988. Integrated pest management theory and practice. Entomol. exp.
appl. 49:59-70.

Landis, D. A. and Gould, F. Some considerations in the screening and use of
feeding deterrents for crop protection. Environ. Entomol. In press.

Levin, 1138.1A. 1971. Plant phenolics: an ecological perspective. Amer. Nat. 105:157-

Levins, R. 1986. Perspectives in integrated pest management: from an industrial to
an ecological model of pest management. K? -8 In: Ecological theory and
integrated pest management practice. M. gan (ed.). Wiley. NY. 362 pp.

Lewin, R. 1988. The uncertain perils of an invisible landscape. Science 240: 1405-
1406.

140

Lockwood, J. A., Sparks, T. C., and Story, R. N. 1984. Evolution of insecticide
resistance to insecticides: a reevaluation of the roles of physiology and
behavior. Bull. Entomol. Soc. Amer. 30(4):41-51.

Loosjes, M. 1976. Ecolo and Genetic Control of the Onion Fly, Delia antiqua
(Meigen). P Wageningen.

Lovett, A. L 1923. The onion maggot. 0mg. Agric. Expt. Stn. Circ. 37.

Lovett, J. V., Ryuntu, M. Y., and Liu, D. L. 1989. Allelo athy, chemical
communication, and plant defense. J. Chem. Ecol. 15:1193-1 02.

Lundgren, L. 1975. Natural plant chemicals acting as ovi sition deterrents on
mbb e butterflies (Pieris brassicae (L), P. ropae ) and P. napi (1.)).
Zool. cripta 4:253-258.

Ma, W.-C. 1977. Alterations in chemoreceptor function in armyworm (Spodoptera
exempta) by a plant-derived sesquiterpenoid and by sulfhydryl reagents.
Physrol. Entomol. 2: 199-207.

MacDonald, R. S., Surgeoner, G. A., Solomon, K. R., and Harris, C. R. 1983.
Effect of four spray regimes on the development of permethrin and
dichlorvos resistance in the laboratory by the house fly (Diptera: Muscidae).
J. Econ. Entomol. 76:417-422.

Mandava, N. B., Orellana, R. G., Warthen, J. D., Jr., Worley, J. F., Dutkey, S. R.,
F inegold, H. and Weathington, B. C. 1980. Phytotoxins in Rhyzoctonia
solani: Isolation and biological activity of m-hydroxy and m-
methoxyphenylacetic acids. J. Agric. Food Chem. 28:71-75.

Mangel, M. 1989a. An evolutio inte retation of the "motivation to ovi sit.”
J. Evol. Biol. 2: 157-172. nary rp p0

Mangel, M. 1989b. Evolution of host selection in parasitoids: Does the state of the
parasitoid matter? Amer. Naturalist 133:688-705.

Mangel, M. and Roitberg, B. D. 1989. Dynamic information and host acceptance
by a tephritid fruit fly. Ecol. Entomol. 14:181-189.

Marsolan, N. F. and Garrett, D. L. 1977. Optimal sizing of trap crops. Proc.
Pittsburgh Conf. on Modeling and Simulation 8:751-754.

Mng0 and Bateson, P. 1986. Measuring Behaviour. Cambridge Univ. Press.
pp.

Martinson, T. E., Nyrop, J. P., and Eckenrode, C. J. 1988. Dispersal of the onion fly
gDipterazAntho "dae) and larval damage in rotated onion fields. J. Econ.
ntomol. 81:508- 14.

Matsumoto, Y. and Thorsteinson, A. J. 1968. Effects of organic sulfur compounds
on ovi ition in onion maggot Hylemia antiqua (Diptera: Anthomyiidae).
Appl. nt. 2001. 325-12.

141

Matsumura, F. 1983. Penetration, binding and target insensitivi as causes of
resistance to chlorinated hydrocarbon insecticides. pp. 36 -386. In: Pest
Resistance to Pesticides. G. P. Georghiou and T. Saito (eds.). Plenum. NY.

Merck [Ipsdzx 1976. M. Wildholz (ed.). 9th Edition. p. 570. Merck & Co., NJ,

Meredith, M. A. and Stein, B. E. 1983. Interaction among converging sensory
inputs in the superior colliculus. Science 221:389-391.

Metcalf, R. L. 1980. Changing role of insecticides in crop production. Ann. Rev.
Entomol. 25:219-256.

Metcalf, R. L. and Lamgnan, R. L. 1989. Estragole analogues as attractants for
corn rootworms( leoptera: Chrysomelidae) J. Econ. ntomol. 82: 123-129.

Miller, J. R. and Cowles, R. S. 1990. Stimulo-deterrent diversion: A concept and its
possrble application to onion maggot control. J. Chem. Ecol. In Press.

Miller, J. R. and Harris, M. O. 1985. Viewing behavior-modifying chemicals in the
context of behavior: lessons from the onion {3.138361 In: Semiochemistry:
Flavors and Pheromones. T. E. Acree and . Soderlund (eds.). Walter
de Gruyter & Co., Berlin.

Miller, J. R. and Strickler, K. L. 1984. Finding and accepting hostlplants. pp. 127-
155 In: Chemical Ecology of Insects. W. J. Bell and R. . Cardé, (eds.).

Chapman and Hall, NY.

Miller, J. R., Mori, K., and Roelofs, W. L. 1977. Gypsy moth field trapping and
%emfifgnogm studies with pheromone enantiomers. J. Insect Physiol.

Miller, T. A., Salgado, V. I... and Irving, S. N. 1983. The kdr factor in gyrethroid
resistance. p . 353-366 In: Pest Resistance to Pesticides. G. P. eorghiou
and T. Saito eds.). Plenum. NY.

Mitscher, L. A. 1975. Antimicrobial agents from higher plants. Recent Adv.
Phytochem. 9:243-283. .

 

Moore, L. E. and Christensen, B. N. 1985. Synaptic integration in a single neuron.
pp. 323-334 In: Contemporm Sensory Neurobiology. Correia, M. J. and
erachio, A. A. (eds.). Liss. .

Mowry, T. M., Keller, J. E., and Miller, J. R. 1989a. Oviposition of Delia '
(Diptera: Anthomyiidae) as influenced by substrate holes and particle srze.
Ann. Entomol. Soc. Am. 82:126-131.

Mowry, T. M., Spencer, J. L, Keller, J. E., and Miller, J. R. 1989b. Onion fly
(Delia ) egg depositional behaviour: grinpointing host acceptance by
and insect erbivore. Insect Physiol. 35 :33 1-3 9.

Nair, M. G., Epp, M. D. and Burke, B. A. 1988. Ferulate esters of higher fatty
alcohols and allelopathy in Kalanchoe diagremontiana. J. Chem. Ecol. 14:589-
603.

142

Nakanishi, K. 1980. Insect antifeedants from lants. p. 603-611 In: Insect Biology
ihrlrythe Future "VBW 80". M. Locke an D. S. mith (eds.). Academic Press.

Ozaki, K. 1983. Sugpression of resistance through synergistic combinations with

em basis on anthoppers and leaflroppers mfesting rice in J a an. pp. 595-

iii In: P135; esistance to Pesticides. G. P. Georghiou and . Saito (eds.).
enum. .

Painter, R. H. 1968. Insect Resistance in Crop Plants. University Press of Kansas.
2nd edition. 520 pp.

Papaj, D. R. 1986. Conditioning of leaf-sha discrimination by chemical cues in
the butterfly Battus philenor. Anim. Begin 34:1281-1288.

Papaj, D. R. and Prokopy, R. J. 1988. The effect of prior adult experience on
components of habitat preference in the apple maggot fly (Rhagoletis
pomonella). Decologia 76:538-543.

Patton, 8., Craig, I. and Conway, G. R. 1982. The Pesticide industry. p. 58-73 In:
Pesticide Resistance and World Food Production. G. R. nway (ed.)
Imperial College Centre for Environmental Technology. London.

Payne, T. L. and Dickens, J. C. 1976. Adaptation to determine rece tor specificity
m insect olfactory communication. J. Insect Physiol. 22: 1569-1 2.

Perron, J. P. and LaFrance, J. 1961. Notes on the life history of the onion maggot,
Hylemya antiqua gleig.) (Diptera: Anthomyiidae), reared in field cages.
Can. Ent. 93:101-1 .

Pickett, J. A., Dawson, G. W., Griffiths, D. C., Xun, L, Macaulay, E. D. M., and
Woodcock, C. M. 1984. Propheromones: an approach to the slow release of
pheromones. Pet-tic. Sci. 15:261-264.

Plapp, F. W. and Wang, T. C. 1983. Genetic origins of insecticide resistance. p .
1137-70 In: NCYSt Resistance to Pesticides. G. P. Georghiou and T. Saito (edsg.
enum. .

Price, P. W., Bouton, C. B., Gross, P., McPherson, B. A., Thom son, J. M., and
Weis, A. E. 1980. Interactions among three trophic leve : Influence of
lants on interactions between insect herbivores and natural enemies. Ann.

ev. Ecol. Syst. 11:41-46.

Proko , R. J. 1981. Oviposition-dete ' pheromone system of apple maggot
'es. pg. 477-494 In: Management of Insect Pests with Semiochemicals. E.
R. Mitc ell (ed.). Plenum Press. NY.

Prokopy, R. J. 1986. Visual and olfactory stimulus interaction in resource findifirf lg
msects. pp. 81-89 In: Mechanisms in Insect Olfaction. T. L. Payne, . .
Birch, and C. E. J. Kennedy (eds.). Clarendon Press. Oxford.

Prokopa,l R. J., Collier, R. H. and Finch, S. 1983. Leaf color used by cabbage root
'es to distinguish among host plants. Science 221:190-192.

143

Prokopy, R. J. and Owens, E. D. 1983. Visual detection of plants by herbivorous
rnsects. A. Rev. Ent. 28:337-364.

Prokopy, R. J., Powers, P. J., Heath, R. R., Dueben, B. D., and Tumlinson, J. H.
1988. Comparative laboratory methods for assaying behavioral responses of

Rhagoletis monella flies to host marking pheromone. J. Appl. Entomol.
106:437-44?

Pyke, B., Rice, M., Sabine, G., and Zalucki, M. 1987. The push-pull strategy -
behavioral control of Heliothis. The Australian Cotton Grower, May-July.

Raffa, K. F. and Frazier, J. I... 1988. A generalized model for uantifying
behavioral de-sensitization to antifeedants. Entomol. exp. appl. 46: 3-100.

Ranasinghe, L. E. and Georghiou, G. P. 1979. Comparative modification of
insecticide resistance spectrum of Culex pipiens fatigans Weid. by selection
with temephos and temephos/synergist combinations. Pestic. Sci. 10:502.

Raven, P. H. and Curtis, H. 1970. Biology of Plants. Worth. N.Y.

Renwick, J. A. A. and Radke, C. D. 1987. Chemical stimulants and deterrents
regulating acceptance or rejection of crucifers by cabbage butterflies. J.
Chem. Ecol. 13:1771-1776.

Rice, E. 1.. 1984. Allelopathy. Academic Press. NY.

Rice, M. 1986. Semiochemicals and sensory manipulation strategies for behavioral
management of Heliothis spgi Ochsenheimer (Lepidoptera: Noctuidae), pp.
27-45 In: Heliothis Ecology orkshop 1985 Proceedings. M. P. Zalucki and
P. H. Twine (eds.). Queensland Department of Primary Industries, Brisbane,

Australia.
Rodriguez, E. and Levin, D. A. 1976. Biochemical parallelism of re llents and
ggaactants in higher plants and arthropods. Recent Adv. Phyt 10:214-

Roitberg, B. D. and Prokopy, R. J. 1983. Host deprivation influence on response
of Rhagoletis pomonella to its oviposition deterring pheromone. Physio ogical
Entomology 8:69-72.

Ryan, J., Ryan, M. F. and McNaedhe, F. 1980. The effect of interrow plant cover
on pop-grlgti-qgs of the cabbage root fly Delia brassicae (Weidemann). J. Appl.
Ecol. 1 : 1 .

SAS Institute Inc. 1985. SAS User's Guide: Statistics, Version 5 Edition. Cary, NC:
Sas Institute Inc. 956 pp.

Schneider, K. 1989. Five supermarket chains open effort against pesticide use. The
New York Times Vol. 138. Sept. 12, 1989. p. B9.

Schneider, W. D., Miller, J. R., Breznak, J. A. and Fobes, J. F. 1983. Onion maggot,
Delia mfigua, survival and development on onions in the presence and
absence 0 microorganisms. Entomol. exp. appl. 33:50-56.

144

Singer, M. C. 1982. Quantification of host preferences by manipulation of
tzrzvi9position behavior in the butterfly, Euphydryas editha. Oecologia 52:224-

Singh, D., Siddiqui, M. S. and Sharma, S. 1989. Reproduction retardant and
fumigant properties in essential oils ainst rice weevil
(Coleoptera:Curculionidae) in stored wheat. J. Econ. ntomol. 82:727-733.

Snedecor, G. w. and Cochran, w. G. 1967. Statistical Methods. 6“ Edition. Iowa
State Univ. Press. 593 pp.

Sparks, T. C. and Hammock, B. D. 1983. Insect growth regulators: resistance and
the future. pp. 615-668 In: Pest Resistance to Pestrcrdes. G. P. Georghiou
and T. Saito (eds.). Plenum. NY.

Sparks, T. C., Lockwood, A. J., Byford, R. I... Graves, J. B., and Le
1989. The role of behavior in insecticide resistance. Pestic. Sci. 2(2511..‘ailr8(?£’-3139.R

Sta dler, E. 1983. Attractants, arrestants, feeding and OViposition stimulants in
insect-plant relationships: application for pest control. p. 243-253 In.-
Natural Products for Innovative Pest Management. D. L. “Eitehead and W
S. Bowers (eds.). Vol. 2 of Current Themes of Tropical Science. Pergamon.

Press, NY.

Steel, R. G. D. and Torrie, J. H. 1980. Principles and Procedures of Statistics, 211d
Edition. McGraw-Hill. NY.

Stephens, D. W. and Krebs, J. R. 1986. Foraging theory. Princeton University
Press. Princeton, NJ.

Storck, W. J. 1984. Pesticides head for recovery. Chem. and Engin. News 62:35-59.
April 9, 1984.

Tabashnik. B. E. 1987. Plant secon compounds as ovi sition deterrents for
cgbgm 3 gutterfly, Plens' ' rapae Leprdoptera: Pieri ). J. Chem. Ecol.
1 : - 1 .

Tabashnik, B. E. 1989. Managing resistance with multiple pesticide tactics: theory,
evidence, and recommendations. J. Econ. EntomoL 8 :1263- 1269.

Tabashnik, B. E. and Croft, B., A. 1982. Managing sticide resistance in cr0p-
arthropod complexes: mteractrons between iological and operational
factors. Environ. Entomol. 11:1137-1144.

Tabashnik, B. E., Cushing, N. E., and J ohnson, M. W. 1987. Diamondback moth
(Le idoptera: Plutellidae) resistance to insecticides in Hawaii: intra-island
variation and cross-resistance. J. Econ Entomol. 80: 1091-1099.

Tahvanainen. J. O. and Root, R. B. 1972. The influence of vegetational diversi on
the population ecolo of a specialized herbivore, Phyllotreta cruci erae
(Coleoptera: Chrysome 'dae), Oecologia (Berlin) 10:321.

Taylor, C. E. and Georghiou, G. P. 1979. Suppression of insecticide resistance by
alteration of gene dominance and migration. J. Econ. Entomol. 72:105-109.

145

Thiery, D. and Visser, J. H. 1987. Misleadigg the Colorado potato beetle with an
odor blend. J. Chem. Ecol. 13:1139-11 .

Tingle, F. C. and Mitchell, E. R. 1986. Behavior of Heliothis virescens (Fg in

pggsence of oviposition deterrents from elderberry. J. Chem. Ecol. 12:1 23-
1.

Tingle, G. C. and Mitchell, E. R. 1984. Aqueous extracts from indigenous lants as
oviposition deterrents for Heliothis Vtrescens (F.). J. Chem. Ecol. 10:1 1-113.

Todd, P. H., Jr., Bensin er, M. G. and Biftu, T. 1977. Determination of ungency
due to capsicum gas-liquid chromatography. J. Food Sci. 42:660-&5,

Van degosch, R. 1978. The Pesticide Conspiracy. Doubleday. Garden City NY
pp- ’ °

Van Ric, J., McGaughey, W. H., Johnson, D. E., Barnett, B. D., and V
H. 1990. Mechanism of insect resistance to the microbial mmfiygangffil
thuringiensis. Science 247:72-74.

Vernon, R. S., Borden, J. H., Pierce, H. D., Jr. and Oehlschlager, A. C. 1977. Host

selection by Hylemra antiqua laborato bioassa and methods f ° -
host volatiles. J. Chem. Ecol. 3:359-3681:y y 0 Obmmng

Via, S. 1986. Quantitative genetic models and the evolution of pesticide resistance.
In Pesticide Resistance: strategies and tactics for management. National
Academy of Scrence, Washington, D. C.

Virus, R. M. and Gebhart, G. F. 1979. Pharmacological actions of capsaicin:
apparent involvement in substance P and serotonin. Life Sci. 25: 1273- 1284.

Visser, J. H. 1983. Differential sensory perceptions of plant compounds by insects.
In: Plant Resrstance to Insects. pp. 21 -229. Amer. Chem. Soc. Symp. Ser.
208. P. A. Hedin (ed.). ACS. Washington, DC.

Weston, P. A., and Miller, J. R. 1985. Influence of ca e design on precision of tube
tlrlapalgig‘tasgay for attractants of the onion fly, Delia antiqua. J. Chem. Ecol.

Whitfield, G. H. 1981. Spatial and temporal po ulation analysis of the onion
maggot, Hylemya W. In Michigan. IPhD. thesis. Michigan State
Umversity, East Lansmg.

Wiens, M. N., Rahe, J. E., Vernon, R. S., and McLean, J. A. 1978. Ovipositional

deterrents for lemya antiqua in hydrated seeds of Phaseolus vulgaris.
Environ. Entom. :165-167.

Workman, R. B., Jr. 1958. The biolo of the onion ma lemya antiqua

0t.
§Meigen). under field and gree ouse conditions. h. Dfihesis, Oregon
tate College. ‘

APPENDICES

APPENDIX 1
BASIC P0pulation genetics model

APPENDIX 1

BASIC Population genetics model

10 CLS
zo DIM Dcso(3) : DIM 3(3) : DIM PMORT(3) : DIM STDCON(3)

30 AS="1" : DIM DIST(2) : DIM (3(4) : DIM CORNER(2)

4o DIM P(3,3): DIM NS(3,3) : DIM 3(3,3)

so TOTAL=0 : CORNER(1) =0

60 INPUT 'Path:filename.ext for datafile on disk is:';Z$

7o OPEN -o-,#1,zs

so INPUT "Untreated habitat quality';UHS

9o INPUT 'I‘reated habitat quality';THS

100 INPUT “Fitness cost for R allele';FCR

no FCR: l-FCR

120 INPUT 'Fitness cost for a allele';FCA

130 FCA: l-FCA

14o INPUT 'Probability of AA being in treated area- -;x

150 INPUT ”Probability of a being in treated area-1'91

160 INPUT 'Do you want to use default toxicity values';J$

170 IF 1: .. 'n' THEN GOTO 200

180 LC50(1)=1:LC50(2)=2 : LC50(3)=4

19o B(1)=2 : 3(2)-2 : B(3)=2 : ooro 280

200 PRINT “Enter the LCSO for the following genotypesz'

210 INPUT "rr';LC50(l)

220 INPUT 'Rr';LCSO(2)

230 INPUT 'RR';LC50(3) .

240 PRINT “Enter the slope to: the ln(conc)-probit mortality W”

750 INPUT “for the following genotypes: tr“; 3(1)

260 INPUT 'Rr';B(2)

270 INPUT 'RR';B(3)

280 INPUT “Concentration of pesticide';CONC

290 FOR I s 1 TO 3

300 IF CONC< zooms THEN PMORT(I)=0 : com 410

310 STDCONa)-=B(I)‘(LOG(CONC)-LOG(LC50(I))) : FLG = o

320 IF STDCON(I) > =0 THEN 340

330 FLG .. 1 : STDCON(I) e ABS(STDCON(I))

340 IF STDCON(I) < 3.7 THEN 360

350 PMORT(I) . 1 : ooro «too

360 IFSTDCON(I) < 2.641'HEN380 .

37o PMORTO) .. .959oss+ .0216767‘STDCON(I)-.00291‘STDC0N(I) 2 = GOTO 40° ,
380 PMORT(I)= .49528+ .47095‘STDCON(I)-.14163‘STDCON(I)'2+.01347'STDC0N0) 3
390 IF PMonTa) > 1 THEN PMoara) - 1

400 IF FLG . 1 THEN PMonTa) - 1 . PMORT(I)

410 NEXT I
420 S(1,1) .. (X‘(1-PMORT(3))'THS+ (l-X)'UHS)‘FCR

430 5(1.2)'((X+Y)/2‘(1-PMORT(3))‘THS+ (1-(X+Y)/2)‘UHS)‘FCR‘(1+FCA)/2
440 S(1,3)= (Y‘(1-PMORT(3))“I‘HS+ (1-Y)‘UHS)‘FCR’FCA

450 8(2.1) = (X‘(1-PMORT‘(2))'TI-IS+ (l-X)"UHS)‘(1+FCR)/2

460 S(2,2)-((X+Y)/2‘(l-PMORT(2))"I'HS+ (1-(X+Y)/2)‘UHS)‘(1+FCR)‘(1+FCA)/4
470 80.3) -= (Y’(l-PMORT(2))‘THS+ (l-Y)‘UHS)‘(1+FCR)/2‘FCA

146

480 S(3,1)=X‘(1-PMORT(1))‘THS+(1-X)‘UHS .

490 S(3,2) 8 ((X + Y)/2‘(1-PMORT(1))‘THS + (1-(X+ Y)/2)‘UHS) (1 + FCA)/2
500 S(3,3) = (Y‘ (l-PMORT(1))’THS + (1-Y)‘UHS) ‘FCA

510 FOR A- 0 TO 100

520 CLS

530 PRINT “Simulation is ";A;" percent done.“

540 FOR R a 0 TO 100

550 GFR a R/ 100

560 GFA = A/ 100

IF I=1 THEN PRsGFR‘Z
IF I=2 THEN PR=2‘GFR'(1-GFR)

§§§

610 IF I=3 THEN PR= (l-GFR)‘2

620 IFJ-lT'I-IENPAsGFA‘Z

630 IF r=2 THEN PA=2‘GFA‘(1-GFA)
640 IF J=3 THEN PA: (1-GFA)‘2

650 P(I,J)-PR'PA

660 NEXT 1

670 NEXT I

680 FOR I: 1 TO 3

690 FOR I- 1 To 3
700 NSGJ)=P(U)‘S(U)
710 T0TAL=ToTAL+ NS(I,I)
no NEXT J
730 NEXT I
740 DIST(1) cor-12:2.- GFA‘2
750 GFR = s 1,1 + NS(1,2) + NS 1.3) + .5‘(NS(2,1)+NS(2.2) + NS(2.3)))/1'0TAL
760 GPA- (ONIsEm; + NS(2,1) + NSE3.1)+ .5'(NS(1,2) +NS(2,2) + NS(3,2)))/T0TAL
' 770 DIST(2)=GFR“2+GFA‘2
730 11“ ABS(DIST(1)-DIST(2)) < .00005 THEN ooro 940
790 0(1) a (NS(3,3) + .5‘(NS(3,2) + NS(2,3)) + .25‘NS(2.2))/1‘0TAL
800 6(2) = (NS(1.3)+ £01802) +NS(2.3))+ 2914533))” 0m-
810 (3(3) = (NS(3,1) + .5“(NS(2,1)+NS(3.2)) + 25‘NS(2.2))/'1‘0TAL
820 6(4) = (NS(1,1) + .5“(NS(2,1)+NS(1,2))+ zerS(2.2))/1‘OTAL
830 P(L1)=G(4)‘2

840 P02) =G(2)‘G(4)
850 P(1,3) =o(2)*2

860 P(Zl) =G(3)‘G(4)

870 P(2.2) =o(1)ro(4) +o(2)ro(3)

380 P(2.3) = 6(1)‘G(2)

890 P(3,1) = o(3)‘2

900 P(3,2) = 0(1)'o(3)

910 P(3,3) = 5(1)?

920 TOTAL - 0
930 GOTO 680
940 CORNER(2) - INT(GFR + GFA‘2+ .5)
950 IF CORNER(1)< >CORNER(2) THEN GOTO 970
960 GOTO 1030

970 IF R = 0 THEN GOTO 990

980 PRINT #1, R/1oo,-,-;A/100

99o CORNER(1) - CORNER(2)

1000 IF coma) < > 1 THEN ooro 1030

1010 LET a -o

147

1020 GOTO 1040

1130 NEXT R

1040 NEXT A

1050 CLOSE _ . ..
1060 INPUT “Would you like to run another simulation ;IS
1070 IF IS - 'n' THEN END

1080 GOTO 50

148

APPENDIX 2
Voucher specimen depository forms

APPENDIX 2

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.: 1990-05

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

Manipulating Oviposition of the Onion Fly, Delia antiqua (Meigen):
A Stimulo-deterrent Diversionary Approach

Museum(s) where deposited and abbreviations for table on following sheets:
Entomology Museum, Michigan State University (MSU)

Other Museums:

Investigator's Name (5) (typed)
Richard Steven Cowles

 

 

 

Date _§-20-90

*Reference: Yoshimoto, C. IL 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: Included 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 mseum.
149

mama so mu

 

oauouln wuma

 

 

 

..cw. \V
nmch\1adm‘m:. . s4

. was: >moH m

 

zuwmuo>fica oumum cowasofiz msu cw afimoaov
you mcwswowam vmumfia m>onm osu cm>wmuom

monommfi .oz uozo=o>

 

Asmas.o

moasoo so>oum oumnuwm
Amvmemz m.uoumwfiumw>cu

Amummmmuo: ma muwwcm Hocowuwvvm mmsv

 

Pages

1

of

1

APPENDIX 2. l
Voudher Specimen Data

Page

amazoo .m .m
cows an: as

“was .uamuo

.oo owmmaoz “30H:
.muauaso no;

 

 

Acmwwozv maufiucm «Hana

 

 

 

 

 

 

 

 

 

 

 

Ca Ca
muov wouwmommc can wow: so vmuumHHoo coxmu umzuo no mmfiuwam
m . s s s e mcoswooam you some Hope;
umwduhhwmas
mmwmmamwmwg
deiOAAPNMk.
”mo ponfizz

 

 

150

NICHIGRN STAT

All! *fiufifilwfijflfiflf

L