fiEflE PIES:
25¢ per W per it.»
“TWINS LIBRARY MATERIALS:

Place in book return to move
charge from circulation record:

     
   
 

    
 
 

r;
is”;
* ¢ :2 ‘p

.E If .‘;.‘",," . 3:2", '

l
‘,l_. .

 

 

 

DEVELOPMENT OF MANAGEMENT

CONCEPTS IN PARASITE SYSTEMS

by

Forrest William Ravlin

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Entomology

1980

 

ABSTRACT

DEVELOPMENT OF MANAGEMENT
CONCEPTS IN PARASITE SYSTEMS

By

Forrest William Ravlin

Biological control is recognized as an essential component of
integrated pest management programs. However, for the most part, it
is not operational in most agricultural production systems. In this
study generalized system designs are formulated for the management of
parasites. Concepts are generated concerning the ecology of different
classes of parasite-host systems and management of those systems.

Data were collected from a prototype system composed of the eastern
tent caterpillar, fall webworm, and a parasite common to both pests,

Campoplex validus. Analysis of this natural system proceeds with

 

descriptive statements concerning the phenology and survival of the
discrete components of the system. Interactions between components are
examined using a simulation model capturing development and survival
characteristics, as well as, density relationships.

The model is validated using actual field data including: temporal
occurrence, larval population maturity, survival patterns, yearly popula-

tion trends, and trends in parasitism rates. Sensitivity analyses are

Forrest William Ravlin

performed primarily considering the components of parasite-host synchrony
and how management of parasites can be effected through alterations in
synchrony.

Analysis of the tent caterpillar-webworm system showed that an
expanded system conceptualization is necessary for understanding the
determinants of parasitism. Thus, the basic one-to-one parasite-host
interaction is important, but host-host interactions may ultimately explain

the majority of the variability in this multiple host system.

Once the analysis of the tent caterpillar-webworm system was
completed, a framework was produced to conceptualize various types of
parasite systems and considerations for management of those systems.

The study as a whole emphasizes the need to view each management
option (i.e., parasites/predators) as though it were the objective of
control. By doing so, there is a need to increase the complexity of
system design concepts and consider all factors impinging on parasite

production as potentially controllable.

To my parents, Forrest and Wilma

and my wife, Susan

You made it all possible.

11

ACKNOWLEDGEMENTS

I would like to extend my sincere appreciation to Dean L. Haynes
and Richard W. Merritt for serving as Co-Major Professors. Their
friendship, insight, and diversity of attitudes made my Ph.D program
a truly unique and rewarding experience. .

I would also like to thank Drs. Thomas C. Edens, Gary A. Simmons,
and R. Lal Tummala for serving on my Guidance Committee and immeasurably
adding to the quality of my education.

Special thanks go to Dr. James E. Bath who, as Department Chairman,
has provided an academic and professional atmosphere for myself and all
those who have the opportunity to work with him.

Throughout the last 7 years at Michigan State I've had the privilege
of working with a tremendous group of fellow students. Sincere thanks
go to Emmett Lampert, Dan Lawson, Alan Sawyer, Gary Whitfield, and
Michael Mispagel. In particular, I value my association with Tom and
Alice Ellis, Ray Carruthers, Joe Noling, and Roger Varadarajan. They

have truly made the hard times bearable.

iii

 

TABLE OF CONTENTS
Page
LIST OF TABLES O O O I O O O O O O O 0 O O O O O O o O O O O 0 Vi

LIST OF FIGURES. . . . . . . . . . . . . . . . . . . . . . . . vii

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . 1
THE EASTERN TENT CATERPILLAR-FALL WEBWORM SYSTEM . . . . . . . 3
Problem Statement . . . . . . . . . . . . . . . . . . . 3
System Definition . . . . . . . . . . . . . . . . 4
Previous Studies in the ETC-FWW System. . . . . . . . . . 6
Methads O O O O O C O O O O O O O l O O O O I O O O O 9
Analytical approach. . . . . . . . . . . . . . . . 9

Plot layout and sampling program . . . . . 10

Development and survival . . . . . . . . . . . . . . 12
Measurement of leaf area . . . . . . . . . . . . . . 13
Basic modeling techniques. . . . . . . . . . . . . . 14

PHENOLOGICAL MODELS IN THE ETC-FWW SYSTEM. . . . . . . . . . . 18

Leaf Growth in Wild Black Cherry. . . . . . . . . . . . . 18
A generalized model. . . . . . . . . . . . . . . . . 23
Phenology and Survival in the Eastern Tent Caterpillar. . 23
Egg and first instar phenology . . . . . . . . . . . 23
Leaf growth and larval maturity. . . . . . . . . . . 29

Survival in the eastern tent caterpillar . . . . . . 33
The distribution of mortality. . . . . . . . . . . . 35
Stage-specific surivivals. . . . . . . . . . . . 43
Phenology and Survival in the Fall Webworm. . . . . . . . 45
The Parasite Component. . . . . . . . . . . . . . . . . 50

Immature parasite survival . . . . . . . . . . . . . 50

A MODEL OF PARASITISM AND POPULATION INTERACTION IN THE
EASTERN TENT CATERPILLAR-FALL WEBWORM SYSTEM . . . . . . . . . 58

The Aggregate System. . . . . . . . . . . . . . . . . . . 60
The Eastern Tent Caterpillar. . . . . . . . . . . . . . . 61
Campgplgx_Attack on the ETC . . . . . . . . . . . . . . . 64

The attack model . . . . . . . . . . . . . . . . . . 66
Models for Development in g, validus. . . . . . . . . . . 69
The Fall Webworm. . . . . . . . . . . . . . . . . . . . . 69
Simulation of Mating, Development and Survival in

the FWW . . . . . . . . . . . . . . . . . . . . 70
mpoplex Attack on the FWW . . . . . . . . . . . . . . . 76

Mbdel Validation. . . . . . . . . . . . . . . . . . . . . 77

iv

Eastern tent caterpillar validation. .

Fall webworm validation. . . . . . . . . . . .
Population growth and parasitism . . . . . . .
Parasitism and encapsulation in the fall webworm .

Model Sensitivity . . . . . . . . . . . . . . . . .

Temporal synchrony . . . . . . . . . . . . . . .
Temperature regimes. . . . . . . .

Effects of phenotypic composition in the fall
webworm. . . . . . . . . . . . . . . . . . . .
Harvesting strategies. . . . . . . . . . . .
Numerical sensitivity to ETC densities . . . . . .
Host plant effects on development and parasitism .
Parasite longevity and parasitism. . . . . . . .

CONCLUSIONS FOR THE ETC-FWW SYSTEM . . . . . . . . . .

DISCUSSION OF CONCEPTUAL NEEDS FOR MANAGEMENT OF PARASITE

SYSTEMS.

Framework for System Conceptualization. . .

Single Host-Parasite Systems. . . . . . . . . . . .
Multivoltine Single Host-Parasite Systems . . . . .
Multiple Host-Parasite Systems. . . . . . . . . . .

SUMMARY AND CONCLUSIONS. . . . . . . . . . . . . .

LIST OF REFERENCES .

APPENDICES

A. Simulation Model of the ETC-FWW System. . . . .

B. Degree-Day Accumulations for Gull Lake, 1977- 1979 .

C. Determination of Developmental Temperature
Threshold for ETC Eggs. . . . . . . . . .

D. Study Areas at the Kellogg Biological Station . . .

E. Population Estimators for Parasites and Colonial
Hosts . . . . . . . . . . . . . . . . . .

F. Species Concepts in the Fall Webworm. . .

Page
79

87
95
98
98
99

103
108
110
114
115

115

118
118
122
124
128
137

140

146
166

172
174

175
181

Table

10.

11.

12.

13.

F.l.

LIST OF TABLES

Summary of regression statistics for WBC leaf area .
Life—tables for the eastern tent caterpillar . . . .

Comparison of ETC and FWW with reference to suscepti-
bility to parasite attack. . . . . . . . . . . . .

Eastern tent caterpillar delay parameters. . . . . .
Q, validus delay parameters. . . . . . . . . . . . .
Fall webworm pupal delay parameters. . . . . . .

Fall webworm delay parameters (excl. pupae). . . . .
Fall webworm encapsulation coefficients. . . . . . .

Comparison of model and observed survivorship curves
for FWD 0 O O O O O O O O O O O O O O O O O O O O 0

Comparison of model and field results with the model,
reinitialized with the previous year's output. . . .

Comparison of model and field results with model
reinitialized with actual field counts . . . . . .

Regression statistics for Hyposoter egg encapsulation.

Host plant effects on Campoplex parasitism in the fall

webworm O O O O O O O O O O O I O O O O O O O O O O 0

Relation of color in the fifth-instar larvae of the
fall webworm to temperature and collection area. . .

vi

Page
22

44

56
63
69
72
74

78

87

9O

94

96

. 114

184

LIST OF FIGURES

Figure Page

1. Potential parasite-host interactions, gross
temporal occurrences and environmental stimuli . . . . . 5

2. System model depicting the flow of parasites and
hosts through time . . . . . . . . . . . . . . . . . . . 7

3. Relationship between length times width and actual
lea f at ea in WBC O O O O O O O O O O 0 O O O O 0 O O O O l 5

4. Relationship between mean leaf area and degree-
day accumulations in WBC . . . . . . . . . . . . . . . . 20

5. Relationship between leaves per branch and degree-
day accumulations in WBC . . . . . . . . . . . . . . . . 21

6. Qualitative phenology model for leaf area in WBC . . . . 24

7. Distribution of degree-days required for ETC egg
hatCh. O O O O O O O O O O O O O O I O O O O O O O 25

8. Cumulative distribution of ETC egg hatch indicating
the 50% mark 0 O O O O O O O O O O O O O O O O O C O O O 2 7

9. Field relationships of ETC egg hatch, larval
silking and tent formation . . . . . . . . . . . . . . . 28

10. Relationship between ETC egg hatch and leaf area
Of WBC O O O O O O I O O O O O O O O O O O O O O O O 0 O 30

11. Relationship between ETC weighted mean instar and
leaf area. . . . . . . . . . . . . . . . . . . . . . . . 34

12. Percent ETC egg survival for Gull Lake, Michigan,
1976-1979. . . . . . . . . . . . . . . . . . . . . . . 36

13. Relationship between eggs per mass and ETC egg
survival . . . . . . . . . . . . . . . . . . . . . . . . 37

14. ETC larvae per tent (i .95 C.L.) in relation to
emulative D-D9 O I O O O O O O I O O O O O I I O O O O O 39

15. Probability of ETC larvae dispersing from the colony . . 41

16. Probability of ETC larvae remaining in the colony
due to dispersal and survival. . . . . . . . . . . . . . 42

17. Frequency distributions of Kp for Gull Lake 1977
and 1978 . . . . . . . . . . . . . . . . . . . . . . . . 48

vii

Figure Page

18. Survivorship curves for Gull Lake, 1977-1978 and
Hartford, Michigan, 1978 . . . . . . . . . . . . . . . . 49

19. Probability of Hyposoter surviving to adult as a
function of the time of attack on ETC. . . . . . . . . . 53

20. Functional block diagram of eastern tent cater-

pillar component. (Inset: Generalized time

varying distributed delay (Manetsch 1976)) . . . . . . . 55
21. Graphical representation of Campoplex attack model . . . 63

22. Functional block diagram for the FWW component.
One of 10 flows is represented . . . . . . . . . . . . . 71

23. Discrete mating delays . . . . . . . . . . . . . . . . . 75
24. Mating model . . . . . . . . . . . . . . . . . . . . . . 75

25. Comparison of observed and predicted ETC egg hatch
rates 0 O O O O O O O 0 O O O O O O O O O O O I O O O 0 O .80

26. Comparison of observed and predicted ETC weighted
mean instar for 1977 . . . . . . . . . . . . . . . . . . 32

27. Comparison of observed and predicted ETC weighted
mean instar for 1978 . . . . . . . . . . . . . . . . . . 32

28. Comparison of observed and predicted ETC survivor-
ship curves. . . . . . . . . . . . . . . . . . . . . . . 33

29. Comparison of observed and predicted FWW weighted
mean instar for 1977 . . . . . . . . . . . . . . . . . . 85

30. Comparison of observed and predicted FWW weighted
mean instar for 1978 . . . . . . . . . . . . . . . . . . 86

31. Comparison of observed and predicted FWW survivor-
ship curves. . . . . . . . . . . . . . . . . . . . . . . 33

32. ETC larval and Campoplex adult incidence curves. . . . . 92

33. Probability of oviposition and encapsulation for
Hzposoter in different portions of a FWW larva . . . . . 97

34. Locations in Michigan for simulation runs. . . . . . . . 100

35. Incidence curves for ETC, FWW, and Campoplex . . . . . . 101

viii

Figure

36.

37.

38.

39.

40.

41.
42.
43.
44.
45.

46.

C01.

F.1.

Input/output relationship for effects of pheno-
typic composition in the FWW . . . . . . . . .

Pupal production for the FWW and Campoplex as

a function of phenotypic composition in the FWW.

Effects of time of harvest on pupal production,
parasite/host ratio, and generation index. . .

Campoplex production in response to varying ETC
den81ties. O O I O O I O O O O I O O O O O O 0

Fall webworm population response to varrying ETC

input 8 O O O O O O O O O O O O O O O I O O O O

Generalized succession of hosts and parasites.

Classes of single and multiple host-parasite systems .

The within generation components of SHP systems. . .

Single host-parasite systems . . . . . . . .

Multiple host-parasite systems . . . . . . .

Proportion that H1 forms 0f P1 versus the proportion

it forms of total hosts (prey), F1 (taken from
MurdOCh and oaten 197 5) O O I O O O O O O O O O

Time-temperature functions and determinations of

To for ETC eggs 0 O O O O O I I O O I O O O . O 0

Distribution of FWW head capsule color . . .

ix

Page

. 104

. 106

. 109

, 112

, 113

119

, 121

, 123

125

130

132

, 173

. 183

INTRODUCTION

Since its conception, biological control has been looked on
theoretically as the most desirable method of controlling pest popula—
tions. Classic examples, such as the introduction of the vedalia beetle
for control of cottony cushion scale, demonstrated that such methods
were viable control options. However, reviews on biological control
suggest that these successes are infrequent and for the most part, not
operational in agricultural production systems (Beirne 1975, Monroe
1971, Turnbull and Chant 1961). Because of this, pest control methods
employing pesticide applications represent the only reliable management
option. Presently, this type of energy input into agroecosysteme is
relatively inexpensive. However, geometric increases in development
costs for new compounds, a shrinking energy budget, increased pesticide
resistance and concern for environmental quality are generating further
interest in non-chemical management techniques and the promotion of an
integrated pest management (1PM) philosophy.

The acceptance of IPM dictates an optimal combination of management
techniques taking into account constraints occuring internal and external
to the system. Keeping this in mind, it is necessary to research each
management option as though it were the "object of control" and consider
all variables impinging on its efficacy. By doing so, resources are
focused on those factors determining the trajectory of parasite popula-
tions as opposed to past efforts focusing on pest numbers. Using this
approach, parasites and predators become controllable variables and
manipulation of numbers and attack rates is possible. This outlook differs

from previous concepts in the following way. Using classical methods,

parasite introductions are followed only by the assessment of populations
in terms of percent parasitism. Therefore, parasites are controlled at
the instant of release. Further, in those instances where manipulations
are made to control (increase) parasitism (e.g., with food sprays), the
state of the system is not sufficiently known and cannot be "fine-tuned"
toward desirable population levels.

The inadequacy of our knowledge of parasite systems and the inability
or unwillingness to deal with a significant amount of system complexity
has given managers little capacity to make predictive statements concerning
parasite performance. Assuming that models can be formulated to make
predictions for each component of a system, we can look at profit as the
objective function and manage the system as a whole in an anticipatory
fashion. Without this predictive ability it is clear that we are locked
into making short term decisions on long term ecological problems. At
best, we will only be able to determine what "has been" rather than what
"will be."

In this study parasites will be viewed as the object of control
resulting in an explicit appraisal of factors determining parasite numbers
and consideration of those factors in the management of parasites.
Specifically, the goal of this research is to determine the major manage-
ment concepts and techniques necessary for successful implementation of
biocontrol agents in agroecosystems. This will be accomplished by
determining the role that parasites play in population regulation in a
real world system. The system selected for study includes the eastern
tent caterpillar, the fall webworm and a parasite common to both pests.
Analysis will proceed with descriptive statements concerning individual

components followed by a simulation model examining population interactions.

Information gained from the analysis will be used in re—evaluating

concepts of parasite systems in ecological and management contexts.

THE EASTERN TENT CATERPILLAR-FALL WEBWORM SYSTEM

Problem Statement

 

Regulation of host populations depends on the host's ability to
accept or reject successful parasite attack. Some hosts pass through
susceptible life-stages very rapidly and in this way avoid parasitism.
Conversely, other organisms move very slowly through time and avoid
parasitism through a large variability in temporal occurrence. As a
population progresses through different stages of development, mortality
factors act differentially on those stages. Thus, age-specific behaviors
interact with mortality factors to produce characteristic survivorship
curves. Parasites occuring internal to their host experience these
same patterns of mortality. If we plan on developing management programs
for parasites, it is clear that techniques involving timing and mortality
patterns will be of utmost importance. For a monophagous parasite our
only concern is the temporal placement of the one-to-one interaction and
related mortality patterns. In a multiple host system one-to-one
relationships are important, however, indirect effects between host
organisms are also of significance (i.e., host-host interactions). This
is particularly true in systems where host organisms are separated in
time, and multivoltine parasites experience different survival strategies
in different generations. In these examples, the number of parasites
produced to attack a given host are largely determined by the previous

host. Therefore, understanding the dynamics of the system and its

ultimate management is dependent on the one-to-one interaction as well

as host-host relationships. Keeping the above concepts in mind, objectives
for this study fall into categories of system definition, description of
development and survival characteristics, and analysis of population
interaction. These studies lead to determining a parasite's role in

population regulation.

§ystem.Definition
The system chosen for study includes the eastern tent caterpillar
(ETC) (Malacosoma americanum (L.)), ugly nest caterpillar (UNC) (Archips

cerasivoranus (Fitch)), the fall webworm (FWW) (Hyphantria cunea (Drury)),

 

and a complex of multivoltine polyphagous parasites native to the state
of Michigan. The parasite pool contains only those species which attack
at least 2 hosts throughout the course of a season. Therefore, other
parasites, predators, pathogens, etc., become part of the biotic environ-
ment which is not explicitly included within this system. Other environ-
mental factors include solar radiation, precipitation, temperature and
any other abiotic variables driving the system.

Figure 1 portrays the universe of concern showing the potential
parasite-host interactions, gross temporal occurrences, and environmental
stimuli.

Within the parasite pool each generation of the individual species
is represented by discrete rectangles (Fig. 1). Larger vertical enclosures
pertain to the potential parasite guild attacking any one particular host.
The term potential is used because the composition of each guild may
change as a function of time and geographic location. This listing merely

represents documented one-to-one interactions from which we hypothesize

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MONITORED ENVIRONMENT é
—"""" —"'—_ HOST PLANT :
use: cum ' ::
AOOOTIC FACTORS fig, 3'
N0! Raoul-on I '
213°“ F / f \ . _
Ififl'fi. on I :
|V I
s... a... ' / HOST COMPLEX \
a..." : —— r—I-I— :
: A L . u A 1
mm mm mats mm. m .......
m an. .-
Gmmu Von-Into and ————? I— I
hunch" mm F. ‘ 1 T :
”than T
Waves-cu
Ono: Putnam No! Cm PARASITE POOL
TO LO LO." T- m
Other "MM mu
7”. 00.0", ‘
r V s I I
mm W m
m ' m ' an... '
m Mnm W ‘
9.0m... W W :
. m I W .
“I! 4 m m
Inn-u ' Dunn. '
.- I“
;° -- H """""""""""""""""""""" .v
I“ I“:
“I am
A I M"... .
(Lora-om ‘F—-? ———1
n on... m an
. M ' ‘ Q
1 . new .III ' um. '
""""'"" . ‘fiL .....
L - Loam ’
1 —._..., ___M
I I m m “I
' '12:: ' """ “-
u a I'm.“ , . 1 .m. l. m
m ‘ m 1

 

 

 

Figure 1. Potential parasite-host interactions, gross temporal occurrences
and environmental stimuli.

 

population flow in the manner described. Information for Figure 1 was
taken principally from Muesebeck et a1. (1951) and Witter and Kulman
(1972).

If each member of the defoliator complex was included along with all
parasite species, the basic understanding of the system would be obscured.
A sufficient amount of complexity must be retained while eliminating
components lacking data and detracting from the objective of determining
the parasite's role within the system.

The tent caterpillar and webworm are clearly dominant in the
succession of hosts. Baseline information is available from the litera-
ture as well as from preliminary data gathered in this study. A parasite

common to the ETC and FWW is Campoplex validus (Cresson). g, validus

 

was chosen for study on the basis of its multiple host characteristic
along with availability of basic ecological information. Figure 2
presents the gross aspects of the defined system. The plant component,
wild black cherry (WBC) (Prunus serotina Ehrh.), is included in this
conceptualization and further examined in the subsequent discrete
component analysis. However, explicit inclusion into the interaction
model is not done due to lack of information on plant-herbivore relation-
ships. Effects of the plant are taken into account in development and

survival coefficients and considered as static estimates.

Previous Studies in the ETC-FWW System

Information concerning the dynamics of the ETC is fragmentary.
Research has been directed primarily at natural controls including
parasites, predators (Witter and Kulman 1972) and pathogens (Clark 1958,

Nordin 1974, 1975, 1976). There are no studies which examine population

.oawu nwsounu mumon one moufimmuma «0 30am one moauofiaov Hoooa Eoumxm .N ouswam

 

Emozmmz
44<u

 

 

 

autism mh<qln_z

 

 

 

. >m¢mru
mDG~4<> ¥U<Jm man—12>

5:856 3 E xmioazfi

 

 

 

 

 

 

 

 

 

><4mo .2 .o

m<44~mxwh<u
hzmh
zmmhm<m

 

 

 

 

 

processes in this species of Malacosoma. Because of this limited treat-
ment, one may only theorize as to the dynamics of the ETC based on

research done on a related species, M. califoricum_p1uviale (Dyar),

 

the western tent caterpillar (WTC).

Wellington's (1960) analysis of the WTC indicates that populations
are made up of a number of different types of individuals. There is a
gradient ranging from larvae which are very active, consume large amounts
of foliage, and have a high survival value to sluggish individuals having
an overall decrease in fitness. Regional populations may be represented
by a frequency distribution of activity which, in part, characterizes
the fitness of the population. Changes in the distribution and its
magnitude occur as a function of mortality factors such as parasites,
predators, and weather variables. Simulation studies point to the
retention of desirable character states in portions of the population
which are able to avoid these mortality factors. This ability appears
to be highly correlated with varying degrees of habitat heterogeneity.
Portions of WTC populations which are vigorous enough to seek out spatial
refugia are selected for activity and other correlated fitness responses
(Wellington et a1. 1975, Thompson et a1. 1976, 1977, 1979).

The ETC exhibits many of the same traits as the WTC, producing
colonies of varying levels of activity, occuring in approximately the
same temporal position, and feeding on the same kinds of roseaceous host
plants. Furthermore, many of the parasites attacking the WTC in the west
are conspecific or congeneric with the eastern species. For example,
Hyposoter fugitivus pacificus Cushman attacks both the WTC and FWW in
the west. Similarly, the ETC and FWW are attacked in the east by a

subspecies of H, fugitivus, g, f, fugitivus (Say) (Langston 1957,

Muesebeck et a1 1951, Witter and Kuhlman 1972). Even though there are
some minor variations in species composition between the two systems,
many of the parasite-host interactions that apply to the WTC-FWW system
may well apply to the ETC-FWW system.

Much research has been done concerning population processes in the
FWW. As early as 1917 Baird reviewed its distribution and natural history.
Tothill (1922) examined the parasite complex, and Morris from 1965 through
1976 produced a series of models examining development and survival.

Other significant works on the FWW have been conducted in Japan
entitled "Biology of Hyphantria cunea Drury in Japan" (Ito and Miyashita
1968). For the most part, these latter works verify Morris' findings and
put them in the context of lifetables for the Japanese environment. Other
studies conducted in various parts of the world are listed in warren and
Tadié (1970). It should be noted that much data are available for incor-
poration into a comprehensive model. With this in mind, emphasis is
placed on examining the FWW in Michigan as it relates to other parts of

North America and particularly its place in the ETC-FWW system.

Methods

Analytical approach. The analytical approach used is that of the
"systems approach" (Churchman 1968). This approach is an explicit
problemesolving methodology utilizing a succession of model development,
validation, and reformulation of those models. There are two prominent
attributes of the systems approach: 1) it overtly seeks to include all
factors which are important in arriving at a "good" solution to the
given problem; and 2) it makes use of quantitative models whenever deemed

appropriate. Therefore, in addition to qualitative thinking, it includes

10

computer simulation techniques, mathematical systems theory, stochastic
models, and optimization techniques (Varadarajan 1979).

With reference to parasite systems or ecological systems in general,
the approach described finds its maximum value in hypothesis generation
and data summarization. In addition, if at any point in time we are
able to provide probabilistic statements concerning any given performance
criterion, we are that much closer to explaining 100% of the variability
in nature. This latter aspect should only be viewed as a goal with our
objective to pose interesting questions relevant to real world problems.

Plot layout and samplingiprogramm Field studies were performed in

 

and around the Kellogg Biological Station (KBS) located in Kalamazoo and
Barry Counties, Michigan. Three major study areas were selected at the
KBS. These sites are typical of what might be termed an "old field
ecosystem" (Odum 1959). These are areas of land which were once in
agricultural production and have since been allowed to revert back to

a natural state. The 3 plots at the KBS have been out of production for
approximately 30 years and are characterized by varying densities of wild
black cherry and to a much lesser extent choke cherry (P. Virginians L.),
American crab apple (Malus coronaria L.),and black walnut (Juglans

gigga L.). Because these stands have a relatively open canopy, a signifi-
cant amount of light is able to reach the "forest" floor and hence promote
growth of many herbs such as Solidaga spp. Queen Anne's lace (Daucus

carota L.), and common ragweed (Ambrosia artemisiifolia L.).

 

Each of the 3 plots was subdivided into 5 subplots based on character-
istics such as tree density, openness and geographic separation. Sub-
dividing the major plots made it possible to examine characteristics of

subplots in terms of densities of the eastern tent caterpillar, fall

11

webworm and the parasite complex. In addition, sampling was facilitated
allowing random sampling of much smaller areas, as well as providing the
ability to partition and reduce variance terms in population parameters

(see Appendix E). Positions and areas of each subplot are presented in

Appendix D.

The sampling program was developed around a 2 phase scheme. The
first phase involved host population density estimates for both the
entire larval population and individual colonies. To accomplish this,
absolute counts of the number of colonies in each plot were made after
the population had established itself. Along with this, weekly samples
were taken of 15 whole colonies and the number of larvae recorded. This
produced an absolute number of colonies within the study area, estimates
of the larval population, and estimation of individuals per colony.

Phase 2 dealt with the determination of parasitism rates. Within
each subplot 5 colonies were selected at random. From these colonies,

14 larvae were removed and returned to the laboratory for dissection,
totaling 350 larvae per sample period (week). Dissections were done

under a stereoscopic microscope, recording the head capsule width of

the host, whether or not the larva had been parasitized, and if so, the
species of parasite. Data concerning the parasites themselves included
visual estimation of instar, the position in the host larva's body cavity
and whether or not the parasite had been encapsulated. Along with routine
dissections, selected numbers of larvae were reared under laboratory
conditions. This provided adult parasites for species determinations
along with developmental information on both host and parasite.

Appendix E presents the necessary methods used in deriving estimators

for parasite and host population means and variances.

12

Development and survival. Studies to determine developmental and

 

survival functions were conducted under both field and laboratory
conditions. Developmental rate functions were determined in the
laboratory and later compared with field derived incidence curves.
Because the egg stage in the ETC is of utmost importance in initial-
izing sample programs, much more extensive rearings were done on this
life-stage. Ten egg masses were placed in each of 6 temperatures ranging
from 45° F (7.22° C) to 95° F (35.29° C). Each egg mass was monitored
daily with the number of individuals emerging recorded. Listing of this
rearing data and determination of developmental temperature threshold
are presented in Appendix C.
Survival rate functions were determined under field conditions.
This was done by obtaining weekly samples of both whole and partial
colonies and deriving stage specific incidence curves. The area under
each stage specific curve (N1) was calculated and divided by the develop-
mental time of that particular instar (Eq. 1) (Southwood 1978).

t N1(t)+Ni(t-dt)

 

The results of these calculations were determined for each instar and
compared with the results of the previous instar. Survival rates for each
life-stage were calculated by equation 2.

s - N1/N1_ [21

i 1
Survival was also calculated in a continuous manner (i.e., as a

function of physiological time). This was done by determining the number

of individuals per tent for each sample period producing a survivorship

 

13

curve. Problems and applications of this technique will be reviewed in
a later section.

Due to the fact that the FWW was occurring at endemic levels, larval
sampling was optimized so as to create only minor changes in colony
density without sacrificing information content in the sampling program.
In order to achieve this goal, census' began after the population had
reached a weighted mean instar (WMI) of 3.5 (see Eq. 13). This allowed
reasonable estimates to be made on the early state of population para-
meters through back—calculations. For example, emergence time of the
adult stage can be determined from the knowledge of the instar of an
individual and the developmental requirements to that particular life-
stage. Since we do not know how long a webworm has been in an instar,
we must assume median values for degree-day accumulations along with a
normal distribution of counts.

Changes occurring after a WMI of 3.5 were analyzed in a direct
manner through weekly samples. These samples were treated as specified
by the sampling program previously described.

Measurement of leaf area. Coincident with weekly host and parasite
samples, measurements were taken of leaf growth in wild black cherry
(WBC). At the beginning of each season 9 WBC trees were selected for
monitoring of leaf area. From each of these trees, 1 branch was chosen.
All leaves were counted on each branch and length and width measurements
of 20 randomly selected leaves were taken. This method of measurement
was chosen because initial study indicated it had a high correlation
with actual leaf area as measured electronically (Li-Cor(33). It allowed

rapid accurate estimates to be made without destructively sampling or

14

taking equipment to the field. The relationship between the length times

width (L x W) and metered leaf area (A) is shown in Figure 3 with

A - (O.8819)(LxW)°-9125
[31
r2 - 0.9646*.

Basic modeling techniques. The development of lifestages of the

 

insects are highly correlated with the prevailing weather conditions,
i.e., temperature, humidity, etc., the predominant factor being tempera-
ture. Development takes place only when the temperature (T(t)) exceeds a
threshold To, which is characteristic for that particular insect species.
Thus,each life-stage requires a specific amount of heat accumulation
(DD) above the threshold To, in order to pass from one stage to the next.
These heat accumulations are expressed in "degree-days" (D-D). When the
temperature profile is given, the degree days are calculated as follows:
F(t) - MAX [0,(T(t)-To)]v1‘(t) [4]
Specifically, the mean delay in days, DEL to move from one distinct
life-stagetx1another, is dependent on the heat accumulation in degree-

days, TDD. That is,

x
TDD - J F(t)dt [5]
0

where "x" represents the mean number of calendar days required to complete

development in a lifestage.
If it is assumed that the growth rate is linearly related to the

temperature above the threshold, then the instantaneous value of the

 

*For clarity many equations have been returned to the original units of
measure. Coefficients of determination (r2) throughout this discussion
apply only to transformed equations.

LN HETERED RREH (ChuuZ)

(54.6)

4

3 (mil)

2 (L4)

(1.0)

0

1 (2.7)

 

15

LN Y = -.1257 + .9125 LNIX)

 

Ruiz = .9646

 

L7 l I 1

r
0 (1.0) l (2.7) 2 (7.4) 3120.1) 4 (54.6) 5 (148.4)

LN [LENGTH x WIDTH) (CflxeJ

Figure 3. Relationship between length times width and actual leaf area

in WBC.

l6

maturation delay (the inverse of the rate of growth), DEL(t), is given
by:

DEL(t) . TDD/MAX[0,(T(t)-To)] = TDD/F(t). [6]

However, it is not necessary to make the above assumption, or to

restrict ourselves to considering temperature alone. Developmental
velocity curves that provide information on the maturation rates for
insects with reference to weather parameters, can be used in conjunction
with "table look-up functions," to obtain instantaneous values of DEL(t),
for given weather parameters (Fulton 1975). Otherwise equation 6 can be
used. The threshold temperature (To), the heat requirement in degree-
- days (TDD), and developmental velocity curves are easily estimated from
laboratory and field studies. Once the developmental rates are estimated,
the development of the insects in general can be modeled as discrete
delays using difference equations or delay-differential equations. The
underlying assumption in using a discrete delay to model insect develop-
ment is that all_individual entities of the insect population at any one
stage develop in an identical manner. In other words, it is assumed
that the insects move as homogeneous groups from one life-stage to another.
In reality, this is seldom the case. Insects develop as heterogeneous
groups, and hence there exists differences in development between
individuals. Therefore, it is more realistic to model the developmental
delays of insects by modeling their aggregative behavior which is char-
acterized by a mean delay and associated variance. For insects this mean
delay is dependent on temperature and is time-varying. Consequently,
the basic building-blocks of the subsequent simulation model are time-
varying distributed delays (Forrester 1961, Abkin and Manetsch 1972,

Manetsch 1976). In general,distributed delays are useful in modeling

17

processes that are: l) irreversible, and 2) described by flow-rates or
aggregates of entities that move at different rates through a given
process.

The two important parameters that completely specify a distributed
delay are DEL and k. DEL is the expected value of the transit time of
an individual entity through a process. In other words, DEL is the
mean of the probability density function describing the transit times
of the population of entities. The parameter k specifies a member of
'the "Erlang" family of density functions that is used to describe the

transit times. The Erlang density function is given by:

k .
f(1') - PEEL (1:) (k-1)exp[- D—kE‘EI-J/(k-l) ! . [7]

The density function f(r) desribes the transit times of individuals
II, 12, ..., Tn through the distributed delay process. The mean and
variance of the random variable I are respectively,

u.r - DEL . [8]

a: - DELZ/k. [91

In. the limit as k+~, the probability density function f(T)
approaches a normal distribution with mean DEL and zero variance. In
modeling real-world problems, DEL and k can be chosen or estimated from
time-series to approximate the properties of the process being modeled.

Manetsch (1976) has extended the concept of distributed delays to
the case where the mean delay varies with time; these are referred to as
time-varying distributed delays (TVDD). Using weather data which normally
contains daily maximum and minimum temperatures, a continuous temperature

profile T(t), can be constructed using the "sine-curve method"

(Baskerville and Emin 1969). Thus, equation 6 or developmental velocity

 

18

curves can be used to compute the instantaneous maturation delay, DEL(t).
With this instantaneous delay, DEL(t), and a chosen value for k, the time-
varying distributed delay can be used for modeling insect development.
A generalized delay is included as an inset in Figure 20.

Population models can now be constructed linking TVDD's in series and
providing survival coefficients either during the delay process or at the
end of each instar. This technique constitutes the basis for discrete

component phenology models and, ultimately, the aggregate system model.

PHENOLOGICAL MODELS IN THE ETC-FWW SYSTEM

Phenology models are an important aspect of any population study
for many reasons. Their main contribution lies in the insight gained in
host-parasite and herbivore-plant synchronies. In addition, they are
essential in the timing of population samples and other field oriented
activities. In the following section, each component of the defined
system will be described with reference to its placement in time and
some factors governing this placement. The individual components will
also be related in terms of direct relationships with other components.
Thus, developmental rates and patterns of mortality are discussed for
the ETC and FWW. In turn, each is related to significant events in WBC

phenology.

Leaf Growth in Wild Black Cherry

 

Leaf expansion in WBC begins at approximately 65 D-D9*, considerably

earlier than the majority of other deciduous trees occupying the same

 

*Because there are no empirical data available for the determination of a
developmental threshold, 9° C will be used because of the close association
with the ETC (see Appendix C).

19

habitat. Expansion continues in a linear fashion for the next 300 D-Dg
peaking at 400-500 D-Dg (Fig. 4). This peak marks the end of the major

leaf growth period. The second phase of leaf phenology is marked by a
significant decrease and leveling off coinciding with the flowering period
and an initial drop of some of the larger leaves. Additional leaves are
added at this time (Fig. 4, pt. A) which have the effect of lowering

the mean leaf area but increasing total surface area (i.e., A x No. leaves).
The end of this period is marked by fall leaf drop and a concurrent decrease
in the mean and total leaf area (Fig. 4, pt. B, ca.1200 D-Dg).

There are a number of factors which contribute to the variability in
the rate and timing of the above phases of leaf phenology. I have already
mentioned that the number of leaves dropping from larger size classes has
an effect. In addition, temperature and precipitation would be hypothesized
to exert a significant effect on leaf area in WBC. Independent variables
which may influence leaf area include: cumulative D-Dg's, incremental
D-Dg's between sample points, incremental precipitation, precipitation
per incremental D-Dg, and number of leaves per branch. A forward step-
wise multiple regression was used on loge transformed data for 1977 and
the following variables proved significant: precipitation per D-Dg,
cumulative D-Dg's and number of leaves per branch. These results verify
the original hypothesis. However, problems with multicolinearity restrict
the use of cumulative D-Dg and number of leaves, with a Pearson correlation
of -.94l9. In the final analysis only cumulative D-Dg's are used along
with precipitation per incremental D-D9 due to a greater proportion of
the variability being explained by degree-days alone (76.832 vs. 58.40%).

In addition, leaves/branch applies only to those branches sampled whereas,

cumulative D-Dg have a more universal application. A summary of the

20

30

77 311 B?!

26
L l L kl L l L LJ

LENGTH u NIDTH (CMIxZJ
1

 

 

 

0 T ‘ 300 ‘ I 360 I r79:30 ' I 1200'
DEGREE—DRYS (>9 C]

l
1500

Figure 4. Relationship between mean leaf area and degree-day accumula—
tions in WBC.

21

1200

J14"

1978

1000
1 L

L

800
L LL L

LERVES/BRHNCH
400 800

NERN N0.

L

1977

L

200
1

L

 

 

 

T I f I U r

0 300 600 900
DEGREE-DRYS (>9 0)

T T V T—

l l
1200 1500

Figure 5. Relationship between leaves per branch and degree-day accumula-
tions in WBC.

22

regression statistics is presented in Table l with the equation:

wa - (2.3857) (CUMDD)-21°6(CMDD)'-1651 [101
where:
CUMDD - cumulative D-Dg's, and
CMDD - centimeters of precipitation divided by D-Dg during the
sample period.

Analysis of 1978 leaf data provides similar results; however,
increased precipitation produced no significant effects on leaf area
(Table l). The regression equation is:

wa - (.1195)(CUMDD)'78“7 [111

It appears that precipitation becomes a significant factor in leaf
area when occuring in limited quantities (i.e., 1977). Throughout the
1978 season, rain occurred often enough and in sufficient quantities so

as to produce no apparent growth effects on a within-season basis.

Table 1. Summary of regression statistics for WBC leaf area.

Variable F to Enter r2 Overall F

1977 Leaf Area:

Cum. Degree-Days 12.12*** .5026 12.12***

CM/D-Dg 12.62*** .7683 18.24***
1978 Leaf Area:

Cum. Degree-Days 20.84*** . .7764 20.84***

 

***(p < .01)

23

A generalized model. I will conclude this brief analysis of leaf

 

growth by presenting a model of leaf area which summarizes and explains
8the observed patterns. There are 3 major stages to the phenology of leaf
area in the WBC (Fig. 6A). Stage I is characterized by bud break and
leaf expansion. During this period, leaf area due to leaf expansion
(LAE) is greater than the leaf area lost from the drop of large size
class leaves (LAD). Stage I ends with LAE - LAD at approximately 500 D-Dg
(Fig. 6B). Stage II represents a quiescent period with LAD > LAE due to
no significant losses of leaves. Finally, there is an increase in LAD
(ca. 1200 D-Dg) or the fall leaf drop which continues until all leaves
have been lost.

The model assumes that factors such as precipitation are not wanting.
Were we to superimpose the findings of the regression analysis on this
conceptual model, the rates of change in leaf area would be altered
proportional to amounts of rainfall. Thus, the difference between LAD

and LAE during State 11 and III would further be increased.

Phenology and Survival in the Eastern Tent Caterpillar

Egg and first instar phenology. During early summer (June-July),

 

ETC adults emerge and deposit egg masses on the branches of WBC and other
roseaceous host plants. Within 2-3 weeks pharate first larval instars
are fully formed and undergo obligate diapause. Diapause or quiescence
for pharate larvae continues until the following spring. Egg hatch
begins very early in the season at approximately 20 D-Dg and extends over
a 100-140 D-Dg period peaking at 65 D-Dg (Fig. 7). The distribution of
egg hatch was derived under laboratory conditions and thus serves only as

an initial model for field verification. This was done by observing egg

24

 

 

 

 

 

 

8'1
+

[0-1

N E
a: o
u.) : :
¢:”1 I a ll 2 In
a: i E E
C m: : :
uJ - : :
-J E E
z 1 E E
a: : :
w 9.. LEAF E 5 FALL LEAF
1: - EXPANSION 5 5 DROP

o V V l V 5— T V I r V T I: f V I

0 300 600 800 900 1200 1600

DEGREE-DRYS (>9 C)
A
T.
c:
.J
a:
m
(z
c:
u.
c:
m
.J
a'
x
w
r f r Y I V r I Y T I i V
°0 300 500 600 900 1200 1508
DEGREE-DHYS (>9 C)
Figure 6. Qualitative phenology model for leaf area in WBC.

Lear RRER LOST (L90)

PERCENT HRTCH

25

 

 

 

 

 

 

 

 

 

 

 

 

Figure 7.

 

r r I Ti

30 80 90 ‘ 120 150 180
DEGREE-DHYS (>9 0)

Distribution of degree-days required for ETC egg hatch.

26

masses under field conditions and recording whether or not at least one
larva had emerged. In this context, 100% emergence indicates that all
egg masses had begun to emerge. The cumulative emergence curve for the
laboratory population shows that all egg masses initiated emergence by
65 D-Dg or approximately 502 of the total larval population (Fig. 8).
Under field conditions the point of 100% initial emergence occurred also
at 65 D-Dg (Fig. 9, curve H).

Within the first 2 days after emergence (ca. 5-10 D-Dg) larvae form
a silken mat on the surface of the egg mass and remain there for an
additional 5-10 D-Dg. After this silking and resting period, the colony
moves to an adjoining fork of the tree to begin tent formation. The
relationship between these 3 activities is shown in Figure 9. The lag
in tent formation and initial decrease in larval activity is further
demonstrated in the differences of the rates of hatch and tent building.
If larvae were able to silk and form tents at the same rates as hatching,
then we would expect the T and S curves in Figure 9 to mirror the hatch
(curve H). Larvae are initially able to hatch at a rate of 1.12%lD-Dg,
whereas tent formation is lagged by at least 5 D-Dg and increases at
.27Z/D—D9. Once the population has begun to produce significant numbers
of tents, increased variability tends to decrease the rate of tent
formation in relation to hatch rates. Analagous portions of H and T
curves may be compared with rates of 6.36Z/day and 4.78Z/day, respectively.

Possibly one factor adding to increased variability is sample bias.
After 50 D-Dg bias enters the calculations. Samples taken during this
time favor those colonies forming tents due to visual bias of the sampler,
causing proportionately more tents to be observed than unhatched egg masses

and further decreasing the T slope.

27

 

PERCENT HRTCH

CUM.

 

 

 

D1 I r r I I T I r r I V r r T r j
0 30 80 90 120 150 180

DEGREE-DRYS (>9 C)

Figure 8. Cumulative distribution of ETC egg hatch indicating the 50%
mark .

28

100
l

CUMULFITIVE PERCENT

 

 

 

20 30 40 50 60 70
DEGREE-DRYS (>9 0)

Figure 9. Field relationships of ETC egg hatch, larval silking and tent
formation.

 

29

Following tent formation it is of utmost importance that first
larval instars are well synchronized with the succulent high quality
buds of WBC. This is due to the fact that larvae which are forced to
feed on fully developed leaves have an extremely difficult time piercing
leaf tissues. Further, those that are successful in consuming foliage
are unable to do so at a rate comparable to bud feeders. At least 1
individual from every egg mass (50% of the total hatch) emerged during
the bud stage with only 33% of peak leaf expansion completed at 100%
hatch (Fig. 10). Of 150 egg masses placed on trees in early June only
53% of the colonies were able to establish themselves compared with over
an 80% success rate for normal masses (April). Admittedly, there are
other factors acting on ETC masses at that time; however, laboratory
larvae placed on fully developed leaves in the spring produced only
slightly better results. These findings support the idea that ETC is
highly dependent on the quality of the foliage.

Leafgrowth and larval maturity. In most cases temperature is used
in driving phenological models, and thus population phenomena are viewed
as a function of heat unit accumulations (D-D). This is desirable for
making predictions concerning timing of samples due to availability of
temperature data. On the other hand, on-site measurements are not always
possible. Thus, it might be expedient to utilize the plant as an integrator
of aggregate weather conditions, assuming that the "pest" and plant are
highly correlated in their development.

Throughout the ETC egg hatch period it was shown that there is a
close association between bud and leaf development and the rate that first
instars enter the field. In fact, overall larval development is highly

correlated with Stage I of the WBC phenology (Fig. 6). The ETC requires

 

30

100
1

H

 

PERCENT HHTCH
40

20

 

‘ I ‘7 r U i I I U j
°0 2 4 6 a 10

LENGTH a NIDTH (CflnuZJ

Figure 10. Relationship between ETC egg hatch and leaf area of WBC.

 

31

approximately 400 D—Dg to complete larval development. Similarly 400-500
D-Dg are required for maximum leaf expansion in Stage I.

In order to utilize WBC leaf growth in a predictive mode we must
also have a suitable measure of population maturity in the tent cater-
pillar. Population maturity may be measured in a number of different
ways. One of the most common measures is the mean larval instar (MLI).
The MLI is defined as:

k k

MLI - 151N11/151N1’ [12]

where:
Ni . the number of larvae in the 1th instar with a total of

k instars.
The major problem with the MLI is that it assumes that equal portions of
time are spent in each instar. An alternative measure is the weighted
mean instar (WMI) which takes into account differences in instar length.
Specifically it includes the proportion of the total larval period spent

in each instar (pi) or:

k k

WMI . 1§1N11P1I1§1N1p1 [13]

(Fulton 1978).
Regressing the WMI (loge transformed) for 1977 as a function of leaf
area (loge transformed) produces a significant relationship (p < .01)
with the resulting equation:
ln(WMI) - -1.452 + 1.094[1n((LxW)+1)] [14]
r2 - .9218.
Data from 1978 can now be used to verify this model by direct comparison

of regression lines. The equation for 1978 data is:

32

ln(WMI) - -.6151 + .8020[1n((LxW)+1)] [15]

r2 - .8868.
Regression lines were compared using a t-test for a common 8 (Steel and
Torrie 1960). Significant differences in the slopes of the lines were
obtained (t - 8.069, p <.01) (Fig. 9). This was not entirely unexpected
due to precipitation effects on leaf area, as described earlier (Table 1).
An additional factor creating differences in slopes is possible differences
in developmental thresholds and/or rates of development. In 1977 D-D
accumulations began February 23 and continued to March 15 with a total
accumulation of 25 D-Dg. After that, a cold period continued for 10 days
in which there was no heat accumulation. Under normal circumstances ETC
hatch would have begun at that time but at very low levels (< 1%).
Significant hatch in the field did not occur until the 55 D-Dg mark or
20 D-Dg after the cold period. It would appear then that ETC hatch had,
more or less, been "reset" and that an additional 20 D—Dg were required
for first instar emergence. This is a feasible hypothesis in that first
instar development is completed during the first month after egg deposi-
tion. Therefore, the developmental threshold concept is not applicable
in this case. Rather a hatching or activity threshold may have more
validity for spring hatch of ETC eggs. The tree, on the other hand,
was able to begin bud break and leaf expansion causing some asynchrony
with 50% larval emergence and bud break (Fig. 10).

Adjusting the 1977 data by 25 D-Dg (i.e., subtracting 25 D-Dg of

development from the LxW) produces a new equation:

ln(WMI) - -.9492 + .9385[1n((LxW)+l)] [16]

r2 - .8826.

33

Even with this adjustment the 2 regression coefficients proved to be
significantly different, though much less so (t = 4.05, p < .01).

The above analysis has shown that significant differences in
herbivore-host synchrony may occur between years due to temperature and
precipitation. Because of these differences and lack of substantial data
concerning the WMI-LxW relationship, a precise model cannot be formulated
at this time. On the other hand, estimates of WMI can be obtained which
have considerably wider confidence limits but still provide at least

preliminary prediction. Combining both years data produces the model:

 

WMI - (.15198)(wa+1)-8°26 [17]

r2 - .8717.
Figure 11 shows this relationships and includes the 95% confidence limits
on WMI.

Along with the regression models which have a plant orientation, a
simulation model was developed. This particular model is temperature
driven and may be used in situations where on-site data are available
and predictions are required from a remote laboratory. Further, it
appears that this model has considerably more repeatability on a year-to-
year basis. Because this technique was extensively used in the development
of the system model, I will defer discussion of those results to a later
section.

Survival in the eastern tent caterpillar. There are 2 ways in which

 

survivorship in organisms is viewed. First, the number of individuals
occurring per unit area over the length of a generation can be transformed
into a distribution of probabilities over time resulting in a survivorship

curve. This technique has the advantage of viewing survival as a continuous

process and provides a graphic representation of the pattern of survival.

 

 

“2d Y = --6544 + .8027 X

RIIZ = .8717

LN (WMI)

 

 

 

 

T r u ' u . . . 4—fi
0 1 2 3 4

LN ((LxN]+1)

Figure 11. Relationship between ETC weighted mean instar and leaf area.

35

The second approach, life tables, involves survivorship as it pertains

to any particular life-stage or group of life-stages. This method provides
survival coefficients which are utilized in subsequent analyses (i.e.,
models). In either case, one may determine what life stage or points in
time are particularly susceptible to mortality.

The distribution of mortality. Survivorship in the ETC may be

 

described as a combination of 2 theoretical generalized curves. A Type I
curve restricts the majority of mortality to the latter stages of life,
and Type III to the earlier stages (Price 1975).

we may begin by first examining egg mortality. Eggs are laid early
in the summer and remain in that stage until the following spring. In
actuality, complete development of first instar larvae occurs within the
first month after oviposition (Mansingh 1974). However, for the purposes
of this discussion they will be referred to as eggs. With approximately
80% of their life spent in the egg stage one would suspect it to be
highly resistant to the rigors of the environment. This appears to be
the case with a mean survival rate of .8425 i .0440 over the 4 years in
which data were available. No appreciable mortality could be attributed
to weather effects, parasitism or predation, and no significant differences
were detected from 1976 through 1979 (Fig. 3.12) (P < .05). An additional
hypothesis is that as the number of eggs laid increases, the quantity and
quality of yolk for the developing embryos decreases. Pooling the 4 years
of egg mass data it was found that this is not the case. Concurrent with
the above hypothesis, a significant negative slope would be expected with
respect to percent survivals as a function of eggs/mass. No significant
density dependence was detected with a constant positive slope of 0.9430

(Fig. 13).

 

 

 

 

 

 

 

 

 

 

 

 

 

SSSSSS
333333333333333333

NUMBER SURVIVING

 

37

 

 

m
E- m/
693/

4 El/ ,.96 c1
8- Y = -21.06 + .9430 x / //
0‘)

R'.2 = 08939 /

‘ .1 m
84 m
N
O
Q-a
N

.4

Bang”

§§.1 Eylflfls’: El

. / m m

/ K E m

o ,6 II!
CD-

. /

/

8‘15

. m
D ‘ ' '— I V I 1— r r j

90 140 190 240 290 340 390

EGGS/H983

Figure 13. Relationship between eggs per mass and ETC egg survival.

38

Immediately after eclosion and just prior to tent formation, ETC
first instars are susceptible to a number of mortality factors. Yet,
throughout the course of this study only 5-10% additional mortality
was detected after egg hatch. Much of this mortality may be attributed
to attacks by ants. Studies indicate that foraging by ants such as
Formica obscuripes Forel is a common phenomenon. Further, this foraging
is coincidental with bud break and extrafloral nectaires which are active
during this period (Tilman 1978). This would also be correlated with
peak ETC hatch at approximately 65 D-Dg as derived from historical
records and dates recorded by Tilman. Tilman's data suggest that ant
predation is patchily distributed and dependent on the proximity of
ant colonies to ETC colonies. Thus, it appears that the effects of preda-
tion will be felt on a localized basis and would not affect regional
population significantly.

Once the colony has been able to form a tent the rate of mortality
levels off,and by 150 D-D9,7OZ of the colony is still intact. By 200
D-Dg feeding and general activity has increased considerably. Older life
stages (instars 5-6) appear in the colonies and greater amounts of
mortality are realized. At this same time the WBC has ended Stage I,
bloom has taken place, and the quality of the foliage has begun deteriorating
(Fig. 6). Thus, between 150 and 200 D-Dg, survival rates drop from 70%
to 552. Sixth instars are prevalent within colonies and begin to move out
on the foliage, drop to the ground,and disperse toward pupation sites.

Figure 14 shows the number of individuals remaining in the colony
over a 400 D-Dg period. For one-half of the 400 D-Dg the proportion
remaining is equated only with survival. However, once prepupational

dispersal begins at 200 D-Dg the sampling technique used (tent samples)

39

250
J

200
L,

l

 

LHRVRE/TENT
150
1 1
Z

 

 

 

g4 4L
4

O—

to

o W I r I r t r r ‘1
0 100 200 300 400 500

UEGREE-DRYS (>9 C]

Figure 14. ETC larvae per tent (I .95 C.L.) in relation to cumulative
D-Dg.

40

becomes confounded with 2 processes: survival and dispersal. Thus, the
later part of the curve represents the probability of surviving (PS) and the
probability of dispersing (PD) at that point in time. The probability of
still remaining in the tent (PR) is therefore:

PR - (PS)(1-PD). [18]

Knowing P allows a calculation of PS as it relates to survivorship

D
within the colony or:

PS I PR/l-PD. [l9]

PD was estimated in the field by placing large sheets of plastic
beneath 9 trees with single ETC colonies. The perimeter of each sheet
was raised so as to form a funnel-like collection device for dispersing
larvae to fall into. Collections were made, for the most part, on a
daily basis. Figure 15 shows the cumulative percentages or probability

of dispersing, P Survivorship curves can now be corrected for dispersal

D'
through equation 19 and plotted (Fig. 16). The estimates obtained may
slightly overestimate survival. This is due to individuals which have
died and fallen into the collection "funnel" being counted as dispersing
larvae rather than expiring.

Looking at the overall pattern of survival it appears that the tent
affords an extreme amount of protection for ETC larvae. The first 100
D-Dg is spent in the hatching phase with a 20% reduction in the colony
(Fig. 16A, pt.A). Once tent formation has occurred little mortality takes
place; approximately 10%. From the point where sixth instars first start
appearing in the colony (pt. B) until the end of prepupational dispersal,

an additional 452 of the colony is lost. This would indicate that with

larger instars spending greater amounts of time away from the tent or

41

100
.J

PROBRBILITY 0F DISPERSING

 

 

 

 

ca-a

on

c, T r fF* r l r ‘1
O 100 200 300 400 500

DEGREE-DHYS (>9 C)

Figure 15. Probability of ETC larvae dispersing from the colony.

42

.Hm>«>e:m com ammuwamfiv cu 03v acoHoo mnu a“ wcwcfimEm» mm>um~ 09m mo xufiaqnmnoum .cH muswwm

 

    

 

.u mA_ m»¢o-mm¢omo .a
com co. can cow . on. o
F. r e t p t .p r p .u
a---a -u
r
$225... Tm
. e
$22.8 / 433.68
a a
9
To
1
a
1.0
T

 

 

Al I 'IIQUQOEH

cow
_

 

     

 

Au mA. m»¢o-mm¢omo a

Gov can can con 0
a . . . (P . .p t 0
Bl lfl rm
.2223 ..m

0
4¢>_>¢:m acacumw_o

9
lnu

r
a
lo
um
0

Al I “86808:!

43

on its surface, factors such as predation, parasitism, and weather
variables have a much greater chance of affecting ETC larvae. Thus,

65% of the mortality occurs while larvae are either building the tent or
foraging away from its confines.

Stage-specific survivals. Once again,the pattern of survival

 

described above emerges with the majority of mortality in the very early
stages and the very late stages (Table 2). Instars 3 and 4 were combined
in 1978 due to poor timing of samples missing a significant portion of
the third instar. For purposes of comparison these instars were also
combined for 1977.

Determination of fifth and sixth instar survival presents a problem
due to the tent-oriented sampling technique and the dispersal behavior of
sixth instars. An approach to this problem was to take quadrat samples
at the soil's surface to collect any dispersing larvae or pupae. In
addition, emergence traps were placed in the field to determine adult
densities. Unfortunately, neither technique proved adequate. One-
hundred quadrat samples (lmz) were taken in each subplot for the 3 study
areas (1500 samples) in 1977 producing a survival rate for instars 5 and
6 of less than 12. This rate seemed unreasonably low with 50 emergence
traps producing a similar result for instars S-P. The only recourse was
to utilize the fact that ETC females deposit only one egg mass. Late
stage survival was then calculated from the number of colonies in the
following season which was equated with the number of surviving females
in the present season. This produced an aggregate survival including
instars 5 through adult, as well as any immigration or emmigration. Rates

of .0544 and .0413 were calculated for 1977 and 1978, respectively.

44

Table 2. Life-tables for the eastern tent caterpillar.

 

 

No. Alive at No. Dying M* as Survival
Age Interval at the Beginning During X Percentage Rate
(X) of X (Nx) (M3) of Nx Within X
1211:
Eggs 219.41 38.06 17.35 .8265
Instar I 181.35 0.036 0.02 .9998
Instar II 181.31 7.56 4.17 .9583
Instar III 173.75 39.82 22.92 .7708
Instar IV 133.92 67.59 50.47 .4953
Instar III-IV 151.49 96.07 63.42 .3658
Instar V-A 66.33 62.72 94.56 .0544*
1218:
Eggs 218.38 33.00 15.11 .8489
Instar I 185.38 22.65 12.22 .8778
Instar II 162.73 26.00 15.98 .8396
Instar III-IV 136.63 77.23 56.52 .4348
Instar V-A 59.40 56.95 95.87 .0413*

 

*Calculated from the number of tents in the following year equated to the
number of emerging females in the present year (see text).

45

Phenology and Survival in the Fall Webworm

Studies involving population processes in the FWW have been done
principally in Canada and Japan. Canadian researchers have centered on
the idea of genetic control of fitness responses such as fecundity and
larval and pupal survival. These fitness responses are highly correlated
with the length of the pupal stage, symbolized as Kp (Morris and Fulton
l970a,b) . Kp has been shown to vary, both between geographic locations
and between years. Because of this, the so-called "quality" of any given
webworm population will be exemplified by the frequency distribution of
Kp. This frequency distribution is,therefore,a changing entity responding
primarily to factors which act on large portions of a population such as
temperature. Shifts in the distribution (fitness) may occur as a function
of heat unit production (degree-days) for a given location in a given
year. In years which exhibit extremely short and/or cool summers
individuals with high Kp are suppressed due to the inability to accumulate
sufficient heat units for pupation. Since diapause is restricted to the
pupal stage, groups of individuals caught in these circumstances suffer
high rates of mortality. Conversely, groups with high Kp are favored in
extremely warm and/or long summers. Reduction in fat stores after pupation
suppresses low Kp cohorts in this instance, as they will spend more time
in the ground at high temperatures. This, in turn, reduces fat stores
and has significant effects on pupal weight, fecundity, and survivorship
(Morris and Fulton l970a). The average fitness or quality of the popula-
tion is then a function of the total heat available in the field and the
history of the population determining heat requirements. The population

of FWW occurring at Gull Lake, Michigan exhibits many of the characteristics

46

described above. The mechanisms, on the other hand, are not necessarily
restricted to temperature.

Synchrony of the FWW larvae with Stage II of WBC phenology is
extremely important, as food quality degrades rapidly throughout the
larval stages. In terms of the WBC, the webworm's oviposition period
occurs in Stage II and larvae feed in Stages II and III (Fig. 6). The
time of adult emergence (KP) and oviposition will therefore, determine
the type of foliage available to a webworm colony. Larvae which are
forced to spend significant amounts of time in Stage III will be exposed
to very low quality food and experience reductions in overall fitness.
Mbrris' (1967) study of the effects of foliage age on survival and
fecundity quantitatively demonstrates these effects. Larvae reared on
early and mid season foliage produced no significant differences in larval
and pupal survival. However, larvae reared on late season foliage had a
15 to 20% reduction in survival of larvae and pupae, respectively.
Further, fecundity was affected by each treatment with early foliage
producing a mean of 604 eggs/female, mid season foliage 372, and late 128.

Foliage age is not necessarily related only to degree-day accumulations.
As pointed out earlier, foliage may "age" as a function of precipitation.
Changes in the make-up of FWW populations in relation to WBC foliage
quality would be expected.

The frequency distribution of Kp was calculated for 1977 and 1978.
This was done through back-calculation from the weighted mean instar of
each colony sampled, to the time of emergence (Fig. 17). Emergence for
1977 began at 618 D-D11 and continued until 1267 D-D11. This is compared
with 1978 where adult emergence began at 678 D-D11 and ended at 989.

Referring once again to the WBC phenology, in 1977 it was found that

47

significant changes in leaf area occurred at approximately 1100 D-Dg.
Translating the degree-day accumulation (base 9) to the webworm threshold
(base 11) it was found that these changes occurred at 949 D-Dll. If we
accept Morris' data concerning genetic control of degree-day requirements
and the fact that food quality affects larval fitness, then a truncation
of the frequency distribution at 900-1000 D-D11 would be expected. Both
distributions began during the same time period (600-700 D-D11) and have
median values of 820 D-Dll for 1977 and 809 D-D11 in 1978. Therefore,
the major differences lie in later portions of the 1978 curve. The

1978 data suggest that this is the case with the last adults emerging at
989 D-Dll.

Survivorship curves for 1977 and 1978 also demonstrate a shift in
the webworm populations (Fig. 18). As much as 125 D-D11 separate the
2 curves. However, it is not clear what other factors are contributing
to changes in temporal occurence.

Because of the low levels of FWW at Gull Lake, a second population
was sampled in Hartford, Michigan. The overall form of the survivorship
curve bears a close resemblance to both of the Gull Lake curves.
Differences between years and between locations then lie primarily in
the placement of the population in physiological time.

No additional data are available to support hypotheses concerning
frequency dependent changes in population make-up for Michigan. However,
the sum total of each piece of information presented lends a more than

feasible.explanationtxaobserved differences between years.

48

‘ *= 1977
c3_a
"' ‘ m: 1978

l

l

FREQUENCY

 

 

 

 

 

 

 

 

500 T 700 ' 900 1100 ' 1500 ' 1:500
n-n REQUIRED FOR EHERGENCE

Figure 17. Frequency distributions of Kp for Gull Lake 1977 and 1978.

 

49

250
1

GULL GULL
1978 1977

L

200
41

k

 

..HHRTFORD
1978

150

LHRVRE/NEB

100
L

 

 

f r r U I f V I
800 900

I I T r I I
1000 1100 1200 1300 1400 1500

DEGREE-DRYS (>11 C)

Figure 18. Survivorship curves for Gull Lake, 1977-1978 and Hartford,
Michigan, 1978.

50

The Parasite Component

 

It was hypothesized earlier that Q, validus was a likely candidate
for inclusion into a more simplified system (Fig. 2). It attacked the
ETC early in the year and the FWW later in the year. In addition, a
sufficient amount of information was available with regard to attack
rates and phenology. Throughout the 3 years of study at the Gull Lake
research area 9, validus was not detected. While this was unfortunate
from the standpoint of overall model validation, the general character-
istics of C. validus allowed flexibility with regard to this component.
Hyposoter fugitivus (Say), a closely related species, was recovered from
both ETC and FWW and hence studied in the field in the absence of g,

validus. While much of_§yposoter's ecology will be dealt with in simula-

 

tion, the following paragraphs will prove useful in terms of examining
some factors contributing to parasite production. These factors will be
discussed in terms of how Hzposoter interacts with each host independently.
The interaction between all 3 of these populations will later be analyzed
through use of the simulation model.

Immature parasite survival. Parasites which occur internal to their

 

hosts are subjected to the same mortality factors as the host. This means
that the parasite's distribution of mortality will follow very much the
same patterns as the host. The probability of surviving to adult will
therefore be dependent on the survivorship curve and time of attack, in
addition to other phenomena such as egg and larval encapsulation, which
add to mortality of the parasite alone.

Hyposoter adults prefer to attack early instars, and as such, are
synchronized with early portions of ETC and FWW populations. For the

tent caterpillar, Hyposoter adults emerge just prior to egg hatch and

51

attack larvae issuing from egg masses before tent formation. In order
for these adults to emerge earlier than ETC larvae they must overwinter
in a stage which requires very little heat unit accumulation. The mode
of overwintering is not known. Throughout this study adults occurred
in the field as late as October when little or no host material was
available. This suggests that Hyposoter may overwinter in the adult
stage and have the ability to begin activity as soon as spring tempera-
tures permit. Dunstan (1921) held adults of H, pilosulus in a cage
under field conditions along with potential host material (Ctenucha
virginica Charp.). Attacks on Ctenucha occurred late in the season

and overwintering took place within the host larva. Along with this
treatment, adults were held without host material, and a few individuals
successfully overwintered as adults. It appears that g, pilosulus and
presumably E. f. fugitivus may survive the winter in the adult stage
providing that suitable hosts are unavailable. This is particularly
important from the standpoint of attacking ETC larvae.

As mentioned,Hzposoter attacks the early instars, primarily the
first through the third. In the ETC the first 3 instars occur in the
field through 250 D-Dg. This means that successful attacks may occur
throughout this period. However, if we take into account ETC's pattern
of survival,then a slightly different picture emerges.

Because the survivorship curves in the ETC are nonlinear they
provide Hzposoter certain periods of time which are optimal. Optimal,
in the sense that there is a greater probability of producing adult
parasites (PA)' This probability can be calculated by the following
equation:

PA - l-(Ps(t)-Ps(t+dt)). [20]

52

Here, Ps(t) is derived from equation 19 and is the probability of
suriving to time t. Ps(t+dt) is the probability of a larva surviving

an additional dt degree-days. This dt represents the developmental time
lag from deposition of a parasite egg until the host is killed.

The probabilities shown in Figure 16 apply to both the ETC and
its parasites. However, they represent a mean condition for the tent
caterpillar population and do not directly equate with parasite survival.
Parasite eggs which are deposited in a host at any time t produce a
probability of 1.0 that both host and parasite are there at that time.
What this means is that a "sliding" scale must be used where PA is 1.0
minus the difference between P8(t) (the time of deposition) and Px(t+dt)
(the developmental time lag).

Hzposoter deposits eggs beneath the integument of ETC larvae and
requires, under laboratory conditions, approximately 150 D-Dg to kill
its host. Further, susceptible life-stages of ETC occur in the field
until about 250 D-Dg. Therefore, the time period of interest ranges

from 0 to 250 D-Dg. Figure 19 presents P as a function of time of

A
attack (t).

If one considers only the curves generated there are 2 periods of
time when there will be a comparatively high probability for PA (662 to
87%). The first period occurs within 100 D-Dg of eclosion of tent cater-
pillar eggs. As might be expected, most Hyposoter oviposition occurred
during this period. The second high probability region occurs between
225 and 250 D-Dg. During this time PA was as high as 872; however, this
only considers ETC survival within the tent. Larvae which are attacked

later than 200 D-Dg will be in the dispersal phase. This means that

estimates of tent related survivals will not apply. In fact, late stage

Figure 19.

PROB. 0F SURVIVING T0_RDULT

40

0

 

53

 

T

T’ I r Y I" T

50 100 150 200 ' 250 300
TIME OF RTTRCK (o-n>9 c1

Probability of Hyposoter surviving to adult as a function
of the time of attack on ETC.

54

survival away from the colony appears to be less than 10%, based on life
table estimates (Table 2). Some individuals do not disperse from the
colony and remain in the tent to pupate. Of those individuals, over 95%
in 1977 through 1979 were parasitized by another parasite, Itoplectis

conquisitor (Say). Itoplectis is able to out-compete or hyperparasitize

 

 

Hyposoter, causing great amounts of mortality in both ETC and Hyposoter.

Assuming the host larva was not killed by other mortality factors,
encapsulation of eggs or larvae is highly probable in instars 3-6.
Morris' data (1976) shows encapsulation rates of 0, 36, and 742 for
instars 1-3 in laboratory studies on the fall webworm. In a related
species, 3, exiguae (Viereck), 1002 of eggs or larvae were encapsulated
by host larvae which were third instar or older of Peridroma saucia
(Hubner) (Puttler 1961). If these data are accepted as reasonable for
ETC, then few Hzposoter eggs will produce adult parasites.

It should be noted that throughout this analysis we are under the
assumption that survival of parasitized and nonparasitized larvae is
the same. It may well be that behavioral changes in parasitized individuals
avails them to greater or lesser amounts of mortality. Differential
mortality of this sort has been demonstrated for the western tent cater-
pillar. Predators, parasites, and pathogens respond to the activity level
of any particular WTC larva (Iwao and Wellington l970a,b). Activity levels
are a function of the individual and any factor such as parasitism which
affects the individual. Hence, differences occur in an individual's
survival value.

For g, fugitivus to attack the ETC, there is extreme selection

pressure for attacks to occur within the first 125 D-Dg. This ensures

55

that there will be at least a 50% probability of obtaining an adult
parasite.
In terms of the FWW, the nature of the data collection (after a

WMI of 3.5) does not allow calculation of P We may, therefore, only

A'
speculate on Hyposoter's survival within the webworm colony and make
comparisons with the ETC.

During the 3 years observed, Hyposoter parasitism in the FWW has
been 50-952. This is contrasted with the ETC where Hyposoter has attacked,
at most,252 of the larvae. There are many factors which may be responsible
for the disparity in parasitism rates such as host density, parasite
density, weather factors, and so on. However, possibly more significant
than these are a series of parameters, most of which are intrinsic to
either the ETC or FWW. In the analysis of Hyposoter survival in ETC it
was observed that the length of time host larvae spent in larval stages
was important in successful parasite attack. The time spent in susceptible
life-stages was also an important factor. Other parameters which may be
significant in producing parasites are colony size, number of generations
(Hzposoter) attacking a particular host species, encapsulation rates, and
the number of competing parasite species. Each of these variables is
listed in Table 3 and their comparison provides the following results.

For tent forming insects the amount of time spent with the colony
(- larval stages) affords the individual a certain amount of increased
survival value. FWW larvae spent almost 3 times the amount of time in
the larval stage as the ETC. Hence, the rate of mortality per D-D will
be less in the webworm and allow more Hyposoter larvae to survive. In

fact, when rates calculated from Morris and Fulton (l970a) are compared

56

Table 3. Comparison of ETC and FWW with reference to susceptibility to
parasite attack.

 

 

ETC FWW
Time required for larval period
(days at 21 C) 27°25 80
Time spent in susceptible stages
(days at 21 C) 8.75 61.7
Mean colony size at 502 larval 130 200
development
Number of potential Hyposoter l 2
generations
Encapsulation O-lOZ O-lOOZ
Number of competing parasite 3 5

species during study

 

S7

with values for ETC found here, rates of larval decline are .60 and .47
larvae/colony/D-D for ETC and FWW, respectively.

The time spent in susceptible stages affords the parasite more or
less opportunity to discover and parasitize hosts. Here, the FWW
spends approximately 7 times longer in instars 1-3 than the ETC. Once
again, higher probabilities of being parasitized are expected in the FWW.

Even though the size of a colony has no effect on the number of
attacks in the colony, sheer numbers may provide a refugium for individuals.
In other words, with increasing tent size the probability of being attacked
decreases. By comparing the ETC and FWW, one finds that webworm colonies
are 1.5 times as large as ETC colonies at 50% of their larval development.
The 50% mark is used here due to sample program characteristics in the
FWW. This allows us to utilize actual sample data from Gull Lake plots
as opposed to extrapolations from the literature. All of the above
comparisons directly apply to mortality rates in the hosts and indirectly
to Hyposoter. In most cases the FWW has shown higher susceptibility to
parasitism over the ETC and supports the parasitism rates found. Further
support is given by the number of generations of Hyposoter which may
potentially attack hosts with 2 in the webworm and l in the tent cater-
pillar. In opposition to these variables is encapsulation which commonly
ranges from 50-100% in the FWW and less than 5% in the ETC. The number
of observed encapsulations may, however, be related to the intensity of
attack and thus comparison is not entirely justified. Also in opposition
is the number of parasites competing for host resources and contesting
Hyposoter. Three species were recovered from the ETC whereas 5 were
frequently reared from the FWW. Multiple parasitism, however, was not a

usual occurrence (< 3%).

58

Even though comparisons between tent caterpillars and webworms were
done in a qualitative fashion, each piece of information lends support
to the idea that the ETC moves through the system very rapidly and in
this way avoids mortality factors. In contrast, the FWW is characterized
by variability in emergence times, larval periods, and colony size, and
in this way disperses mortality throughout the population. Hyposoter
has also adapted to these strategies. In the spring, adults emerge
before or simultaneously with ETC larvae and insure a maximum of success-
ful attacks. Later in the summer Hzposoter produces 2 generations to
encompass the entire length of time spent in susceptible life-stages by
the FWW.

The interaction of ETC and FWW through their common parasites is
the overriding theme of this research. A specific system was defined
and analyzed in terms of its individual components and their direct
interaction with related components. The following section will combine
the available information in a comprehensive simulation model including

all of the components discussed previously.

A MODEL OF PARASITISM AND POPULATION INTERACTION
IN THE EASTERN TENT CATERPILLAR--
FALL WEBWORM SYSTEM

The utility of phenology and survival studies lies in the researcher's
ability to use the results in further analyses and applications. In an
analytical mode, the temporal interaction of populations and their impact
on each other is basic to ecological theory. In an applied mode, questions
concerning when and how many can be answered for population sampling and

management considerations. In this portion of the study on the ETC-FWW

system I will address both the analytical and applied areas.

59

From an analytical perspective this study seeks to examine how popula-
tions interact with system and environmental parameters. For example,
what role does temperature play in seasonal parasite production? How do
density relationships function within the context of a multiple host-
parasite system? What effect does host population heterogeneity have on
parasite success and long-term relationships? These and other questions
are central to an ecological understanding of the ETC-FWW system.

Currently, there is no economic justification for management of tent
caterpillars, webworms, or related parasite populations. However, from
a heuristic standpoint, the system design naturally leads to applications
in agroecosystems of similar character. One of the reasons for choosing
this system was because it is thought that the principles generated are
amenable to extrapolation.

In order to deal with the above questions, two approaches can be
taken. First, a series of field-oriented studies could conceivably
examine short-term population processes. In turn, long-term questions
can and have also been studied in the field. However, system noise, a
posteriori realization of data gaps, time constraints and resource
limitations reduce their effectiveness.

The second approach involves the use of computer simulation models.
In this case, the only constraints imposed on analyses are lack of funds
designated for such uses, computer size, creativity of the researcher and,
of course, the data base. Depending upon the objectives of the modeling
effort, possibly the key factor here is creativity. This implies that
the researcher be specifically concerned with a set of sufficiently
interesting queries regarding input/output relationships. Parameters

which are not available empirically can be substituted with reasonable

60

assumptions and designated as researchable for later study. The end
result provides a set of "haves" and "have nots" as well as parameters
and functional relations to be further explored. In the past, research
has often ended with a modeling effort. Although these models have
performed in each of the above capacities, they need to be used as an
aid in mental processes at each stage of a project. The initial stages
of research in particular require rigor and systems analysis utilizing
simulation models to provide well defined structure to research in
applied ecology.

Prediction is still another use of a simulation model. Numerous
models have been put together with the objective of making predictive
statements both quantitative and qualitative. Even though the end result
may provide good probabilistic statements as to system response, their
principal role in research still remains with the interesting questions

posed and needed parameter studies.

The Aggregate System

The ETC-FWW system represents a system which is at best imperfectly
known. A simulation model was developed at the outset of research and
used to define and summarize the state of knowledge. The model was also
and most importantly used in the development of concepts. Because of this,
it is not necessary to mimic the real world system verbatim. For example,
a priori model development dictates that all pest and parasite components
be explicitly defined. For reasons stated earlier, the system includes

the tent caterpillar, webworm, and Campoplex validus. It was shown that

 

Q, validus was not recovered during field studies. However, a related

species, Hyposoter fugitivus fugitivus, was recovered. This would seem

 

61

to immediately invalidate the model. On the other hand, 9, validus was
chosen because of its generalizeable characteristics. As such, statements
concerning component and system response can still be made. In fact, a
qualitative validation of the model under these circumstances demonstrates
its ability to extrapolate to similar systems and the robustness of this
technique.

I have stated in very general terms the intellectual pursuits of the
model. In addition, a number of specific ecological questions have been
asked. They are as follows:

1) How do system and environmental parameters affect
time synchronies and rates of parasitism?

2) What is the system's response to various density
relations within a growing season?

3) How do 1) and 2) affect long term system responses
(multiseason)?

4) How does the system respond to management-oriented

harvesting strategies?

The Eastern Tent Caterpillar

The ETC component is made up of a single series of time-varying
distributed delays (TVDD) simulating a normal unparasitized flow of
individuals. As mentioned earlier, each delay is characterized by DEL
and k related to the mean and variance of the delay process (Eqs. 8, 9)
(k - 12). DELi (days to develop) is determined at every point in time
by instantaneous temperatures (T) generated from the sine wave technique
and time-temperature functions derived experimentally. These functions
take the form:

DEL1 ' l/A + BT [21]

62

where A and B are regression coefficients. This equation is used through-
out the larval stages of the ETC model with Table 4 presenting the stage-
specific coefficients. The functional form of the equation for tent cater-
pillar eggs is slightly different and is specifically:
DELE - T/-32.29 + 0.64T. [22]

Thus, an impulse type input is given to the series of egg and larval delays
and individuals flow through stages at rates specified by equations 7, 21,
22, and temperature derivations. This model is portrayed in the functional
block diagram in Figure 20 labeled NORMAL FLOW. For this component I have
only included egg and larval stages, as pupal and adult dynamics are
assumed to be insigificant insofar as parasitism is concerned.

Survival coefficients (Si) are treated in terms of mean values and
are allowed to randomly vary 1 10%. Coefficients are applied at the end
of each life-stage. These coefficients are listed in Table 4 and derived
in an earlier section treating stage-specific survivals. As mentioned
in that section, stages 5-A were combined for an aggregate survival rate
(ca. 0.05). Because of this aggregation, 85 and 36 had to be estimated.
This was done by dividing the 0.05 survival among the four life-stages
equally. An alternative to this would be to weight survival coefficients
as a function of the amount of time spent in the particular instar. This
proved to be unsatisfactory in that a fifth instar survival of 0.67 appears
to grossly overestimate the trend indicated by previous instars and
survivorship curves presented earlier. Equal weighting produces a more
satisfactory late stage survival.

The output produced from the normal flow includes the number of
individuals in each life-stage and the maturity of the population at any

point in time (WMI). Output from the sixth instar gives the exact number

63

Table 4. Eastern tent caterpillar delay parameters.

 

 

 

 

Degree-Day Time-Temp.
Requirements Parameters (°F) Survival
Life-stage D'D48 D-Dg A B Coefficient
Egg 118.75 65.97 +-32.28929 0.63973 0.84
L I 63.00 35.00 -0.76377 0.01585 0.94
II 63.00 35.00 -0.76377 0.01585 0.90
III 63.05 35.03 -0.76377 0.01585 0.77
IV 82.33 45.74 -0.57937 0.01203 0.49
V 102.30 56.78 -0.47243 0.00980 0.47*
VI 214.88 119.38 -0.22435 0.00465 0.47*
Pupa 370.03 205.57 -0.l3028 0.00270 0.47*
Adult 72.00 40.00 -- -- 0.47*

 

*Individual coefficients are not available. Survival is distributed
evently from V-A. Survival from V-A is 0.05

+See text for form of equation.

64

of larvae entering the pupal stage. This, along with pupal and adult
survival and fecundity, can then be used to make predictions as to the

overwintering egg population.

Campoplex Attack on the ETC

9, validus has been shown to attack only instars 1-3 in the webworm
and it is assumed that the same preference occurs in the tent caterpillar.
Campoplex deposits a single egg beneath the integument of these early
stage larvae. Once parasitized these larvae are not reattacked. Thus,
parasites are able to discriminate between parasitized and non-parasitized
individuals. This aspect is handled explicitly in the model by removal
of individuals from the normal ETC flow and using them as inputs to a
parasitized flow (Fig. 20).

A unique aspect of the removal of parasitized individuals is that
the within-instar age of an individual is retained after parasitism. In
other words, a larva which has developed 80% through a given instar will
be placed in exactly that same position in a parallel parasitized flow.
Further, the fact that parasitized larvae are taken directly from within
the delay insures that parasites respond only to non-parasitized densities.
This technique is not without its faults. Laboratory studies have demon-
strated that parasites will respond in some manner to larvae which have
been parasitized. Even though the parasite may ultimately reject the
host, a significant amount of searching time can be used in this activity.
Thus, attack rates are reduced as a result of this type of interference
(Griffiths and Holling 1969, Hassell and May 1974).

By the end of the third instar, no further parasitism takes place.

Larvae move through parallel flows in identical fashion with the parasitized

Figure 20. Functional block diagram of ETC component. (Inset:
Generalized time varying distributed delay (Manetsch
1976)).

65

 

82m 586 s
.88... .o 346::
3 822m

 

82m 263 32m 8.5
282. as. -553 :82. :3.

835-2333. .. u . . . do a.

 

 

 

301E omN_._..m<z<a

 

om om cm mm mm .m muodpxuhm
whagmuts Ch
:52. Emma

 

 

 

 

 

. E... . ...,... . ..é... . ,_ e E E a 8.3.33.5...

a , a a 4 45.2.

mm hoodxwogmo om h9<\.uo._uo cm B+<\.»¢43 mm hu+¢s7n4mo Nm h0+<\.»~._uo .m hm.(\_n.43 um 5+4:qu 00m

7 5 Ohm

x03. #4
b. . m4C< .

66

larvae producing the second generation of Campoplex to attack webworm
larvae. Because of insufficient data it is assumed that parasitized
and non-parasitized larvae behave similarly and have the same fitness.
Therefore, survival coefficients apply equally to both flows.

The attack model. The attack function is a result of modifications

 

made to the Holling (1959) "disc equation" and the Griffiths and Holling
(1969) competition submodel. Attacks per parasite (A) are generated
using:

aTH .
A 1+aTH [23]

where:
a - the instantaneous rate of discovery,
T - the time available for search,
TH - the handling time, and
H - the host density.
Total attacks for the population are a linear function of the parasite
density P or:
NA - AP [24]
Use of equations 23 and 24 assumes a random search by parasites
(i.e., each individual in the host population has an equal opportunity
to be attacked). We may expand this idea to the case where attacks are
distributed other than randomly. The equation:

N -k
NG-H 1- 1+; [25]
Hk

takes into account the fact that attacks may be distributed in a contagious

fashion (Griffiths and Holling 1969). Here:

67

NG = the gross number of attacks,
H - the host density, and
k = the negative binomial parameter, or in its simplified

form:

2

k . T—o‘u-u [26]

for mean u attacks and variance 02. The parameter k varies between zero
(strong clumping) and infinity where the negative binomial becomes poisson.
For this model (Eq. 25) the proportion of hosts attacked is one minus the
probability of zero attacks, q, where q is given by:

NA

a.1+_H-l-C— [2.7]

For Campoplex, k is set of 100 indicating a random distribution of attacks
(Salt 1934). Strictly speaking, equation 25 can be reduced to a random
distribution but is retained for flexibility.

For this type of model, where system dynamics are simulated on a
within day basis, it is not reasonable to assume that parasites are always
in an attack mode. This aspect is considered in two ways. First, the
time available for attack is assumed to include only the daylight hours.
Second, temperature affects are taken into account by reducing attacks
during temperatures less than 65° F (18.33° C) and greater than 90° F
(32.22° C). Attacks are reduced by a linear function with a slope of 0.1
at temperatures between 60° F (15.56° C) and 65° F (18.33° C). Between
90° F (32.22° C) and 110° F (43.33° C) the slope is 0.05. Figure 21
demonstrates this function along with effects of host density. This is

essentially a graphical representation of the attack model.

69

Mbdels for Development in C. validus

Simulation of life-stages in Q, validus is accomplished using the
delay techniques described earlier. A simulated growing season begins
with an impulse type input into the pupal delay. Adults are produced
from the output and used as inputs to the attack model. Immature stages
of Campoplex flow through host larval stages in rates determined by the
host. Upon reaching the end of the sixth instar (in ETC and FWW) output
from this delay produces the second generation when attacking ETC, or the
overwintering generation when attacking FWW. Table 5 gives the delay
parameters used for adult and pupal stages of Campoplex. The time-

temperature function is that of equation 21.

Table 5. g, validus delay parameters.

 

Time-Temp.

 

 

Survival
Life-Stage D-Duz D-D5.5 Pirameters (0:) Coefficient
Adult 625 347 -0.06719 0.00159 1.00
Pupa 175 97 -0.23993 0.00571 1.00

 

The Fall webworm

Characteristic of FWW populations is a tremendous amount of vari-
ability in adult moth emergence. More specifically, a larger variance is
associated with the heat requirements for diapause termination in over-
wintering pupae (Morris and Fulton 1970a). Correlated with this variability
are certain fitness responses such as fecundity and larval survival. It

appears that heat requirements are under genetic control and therefore,

70

in part, responsible for the survival value of that particular phenotype.
Morris (1971, 1976) and Morris and Fulton (l970a,b) have examined these
relationships and produced a series of regression models to describe

the relationship between requirements for adult moth emergence and
survival. These models, along with the developmental delays, form the
basis for the FWW component.

The genetically controlled emergence of the webworm dictates that
different portions of the population will be subjected to different types
of mortalities (e.g., changing food quality). In addition, differential
overwintering success and larval ability to encapsulate parasite eggs are
correlated with the genetics of the individual. Because of the extreme
differences in individuals, survival becomes a dynamic entity and must

be explicitly modeled in the FWW component.

Simulation of MatingiiDevelgpment and Survival in the FWW

Within the model the degree-day requirements for adult moth emergence
(i.e., the pupal delay) are divided into ten discrete blocks or phenotypic
classes symbolized as "Kp classes" ranging from 280-720 D-D11 in 40 D-D11
increments. Each of the Kp classes is simulated separately, producing a
total of ten flows for the unparasitized webworm component (Fig. 22).
Independent simulation of the ten classes enabled the assignment of
coefficients to govern encapsulation, survival, and developmental rates
(Table 6). Each Kp class is then represented by a series of TVDD's
simulating all life-stages of the FWW.

As with the ETC and Campoplex, instantaneous values for the delays
are calculated from laboratory data generated in this study and information

from Mbrris and Fulton (l970a).

DELA (T)

   

DELPlT)

FEClKPI SA”) DELEtT) SE”)

  
     
 

PRE
0V 1 P05-
ADULT S

     
  

    

PRE
OVIPOS.
ADULTS

  

    

 

 

 

 

 

 

 

nus": DELP")
Input
pl
I!
Hfllfi'i ATTACK
nut: arr;
nootL
Tuost ousnv
PARASITE
r w 1
on. m sum out!" sum scum sum mum sum
Lm. ' 0 LA'V‘ 2 0 u a 0 u a
u.» m sum tnunuuuo LIMA!
I no vu: nu- now
I
~ tucnn , tncnn 0 men.
‘ Lulu I 0 LANA 2 Leave 3
at: «1 sum SL3 m an m an m

 

In. "'8 "HIV 0.
"HI I! 'AIASIVI ".0.

Figure 22. Functional block diagram for the FWW component. One of 10
flows is represented.

72

Table 6. Fall webworm pupal delay parameters.

 

 

 

Degree-Day Time-Temp.
Requirements Parameters Survival Pupal
KP D-D51 D-011 A B Coefficient Weight
1 320 178 -0.15944 0.00313 0.40 150
2 360 200 -0.l4178 0.00278 0.40 153
3 400 222 -0.12755 0.00250 0.40 155
4 440 244 -0.11593 0.00227 0.40 155
5 480 267 -0.10624 0.00208 0.40 154
6 520 289 -0.09810 0.00192 0.40 153
7 560 311 -0.09107 0.00179 0.40 150
8 600 333 -0.08499 0.00167 0.40 144
9 640 356 -0.07969 0.00156 0.40 139
10 680 378 -0.07499 0.00147 0.40 128

 

73

Survivorship or the transfer coefficients for each life-stage (i.e.,

8 SI, 82 ...) are applied at the end of each stage. These coefficients,

E’
like those of the ETC, were determined through life-table studies and
allowed to randomly vary 1 10%. Because of the endemic level of webworms
throughout this study, coefficients were taken from Morris (1967, 1972)
and Ito and Miyashita (1968). Coefficients for development and survival
are listed in Table 7.

Simulation of the webworm component begins by providing impulse
inputs to the pupal stage of each of the Kp classes (Fig. 22). Webworms
are held within the pupal delays as specified in Table 6. Upon adult
emergence, mating begins and is completed by the third day after eclosion
(Morris and Fulton l970a). As the mating period for an individual is
short (1-3 days) and the total emergence time is quite long, mating
is positively assortive with respect to Kp. That is, early individuals
(low Kp) mate with early individuals and late (high Kp) with late.

Because the output from pupal delays is time distributed, there
is some overlap in emergence. This, together with the fact that some
individuals may wait as long as 3 days to mate, ensures that there will
be gene flow throughout the population. Within the mating model, 65% of
the moths mate the day of eclosion (A). The remaining 35% enter one of
two discreet delays, the first of which holds 23% of the moths for one
day (B). The second holds 12% for a two-day delay (C) before mating
(Fig. 23). Kp values for mating moths are calculated from the cumulative
degree-days (CUMDD(t)) and the length of stay in the premating delay.
Moths in mating classA have a Kp value equal to CUMDD(t), while B and C
have Kp values of CUMDD(t) minus the degree-day accumulation while in

their respective mating delays. Since three types of mating individuals

74

Table 7. Fall webworm delay parameters (excl. pupae)

 

 

 

 

Degree-Day Time-Temp.
Requirements Parameters Survival
Life-stage D—D51 D—D11 A B Coefficient
Egg 240 133 -0.21237 0.00416 0.85
L I 114 63 -0.44778 0.00878 0.55
II 91 51 -0.55997 0.01098 0.75
III 91 51 -o.55997 0.01098 0.70
IV 114 63 -0.44893 0.00880 0.60
V 166 92 -0.30672 0.00601 0.45
VI 223 124 -0.22863 0.00448 0.45
Adult 150 83 -0.34018 0.00667 0.75

 

75

.Hmmos wcfiumz .qN muawfim

«\2~-3a?2-52.~+§uo~78 --58+2..58§Y3
«\A.~-pvaawaq-».aa+a=:=uo-uu< .«upvcaoaaxauln-

N\.~«-~vaa+==:=uouvln< a:x=uu¢¢

Aadmv occuumazoamo QM Houcwummlowz

 

 

 

  

 

 

 

 

 

 

 

 

 

3-58.2-51— E 3.585.:
8.83.... 8.83....

a 1 c
2-58.83... 758.83...
. .
85!. 85?...
¢ ¢

 

 

.mxmauo wcfiume oumuomfin .mw ouswum

Siamwwup—

 

*3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

*3

:izasiz

76

are found at any time, the possible cross-matings are: AA, AB, AC, BB,
BC, and CC. Figure 24 shows these cross-matings along with equations
for calculating individual Kp and mid-parental Kp for each mating
combination. Through this type of recombination evolves the population
heterogeneity in terms of development and survival throughout the range
of Kp classes.

The mid-parental Kp (pr) is used directly in determining the
distribution of Kp for the resulting offspring (colony). The mean Kp
for the colony (ipc) is calculated by the regression equation:

" 0.602
(2.456 + Kpp ) [28]

ipc - e
(Morris and Fulton l970b). The resulting Epc is then used in determining
the within colony variance and standard deviation (SD) where:

so - -1328.82 + Epc229-555. (29]
Within a colony it is assumed that Kp's are normally distributed. Means
' obtained from equation 28 are normalized, distributed about the normal
curve, and once again divided into the ten discreet Kp classes. The number
of eggs/colony are determined from Kpp where:

Eggs . -l34.0 + 3.8 (pupal wt). [30]
Kpp is used here in its relationship with pupal weight. Pupal weight is

determined for each Kp class derived under laboratory conditions (Morris

and Fulton 1970b). Table 6 gives these values.

Campoplex Attack on the FWW

 

Parasite attack on the FWW is simulated as described in the ETC.
Significant differences, however, lie in two areas. First, the 10 Kp
classes each have parallel flows for parasitized larvae. Second, once in

a parasitized flow, attrition may occur due to reductions proportional

77

to larval survival coefficients and encapsulation of Campoplex eggs.
Encapsulation is modeled similar to parasite attack in the unparasitized
flows. Individuals are removed as a function of instar and Kp class and
serve as inputs to encapsulated flows (Fig. 22, Table 8). Because of the
perfect discrimination of parasitized larvae (encapsulated and not
encapsulated) by Campoplex, 30 flows must be individually simulated.

Host larvae are only parasitized once. At the end of the third instar,
the encapsulated flows join the unparasitized flows due to the fact

that parasite attack is not involved after the third instar. Thus, there
is no reason to continue to "track" larvae with encapsulated parasite
eggs. The parasitized flows continue until the sixth instar, during
which death of the host larva occurs and parasite pupae are generated.

Listing of the entire simulation model is given in Appendix A.

Model Validation
Validation of a simulation model demonstrates the degree of realism

embodied in the model. Comparisons can be made utilizing field observa-
tions of both quantitative and qualitative nature concerning various
aspects of model structure and response to stimuli. For the ETC-FWW
simulation, outputs are verified with field observations made at Gull
Lake, Michigan, in 1977 through 1979. Comparisons will be made with
reference to each modeled population for the following aspects:

1) temporal occurrence,

2) larval population maturity,

3) patterns of survival,

4) trends in yearly population build-up or decline, and

5) trends in rates of parasitism.

78

 

 

oo.a o.o o.o o~.o m~.o am.o as.o He.o oh.o cm.o oo.H HHH
m.~ o.o o.o ~H.¢ m~.o mm.o as.o om.o mo.o mk.o om.o HH
o.H o.o o.o mo.o mH.o -.o a~.o km.o Na.o mm.o oo.o Ho
”hmu mGH
.. . . . . . . . . . . camouwmomoo
o o o c we o mm 0 am o mu o em 0 me o Nn o as o uflcfiomam as
camouscmmoc cm a o a c n s m N H sumac am

owmaoommlowmum

 

.wucuauaumoou soaumHsmmmooo auoaauz damn .m canoe

79

Eastern tent caterpillar validation. In an earlier section,

 

laboratory hatch rates were compared with field observations and were
highly correlated. Field observations were taken, however, on a per
mass basis, whereas laboratory measurements could be taken for a popula-
tion of individuals. Because the field and laboratory populations
compared favorably and because laboratory estimates were done on a per
individual basis, I will examine the simulation results of egg hatch
with reference to lab results. This method is not entirely valid
because delay values for the egg stage were derived from a summary of
this laboratory data. This means that our validation contains some
elements used in determining DEL and K values for the time-varying dis-
tributed delays. We do know, however, that the l, 50, and 100% cumulative
hatch points on the curve do have validity under field conditions (see
"Phenology Models in the Eastern Tent Caterpillar").

Figure 25 presents the simulated and observed egg hatch data. It
is apparent that the two curves are quite similar but differ in temporal
placement.

During the ETC hatch period degree-day accumulations average approxi-
mately 3 D-D9/day (1976-1979). In terms of percent hatch, this means
that the actual hatch may be overestimated by as much as 25%, were the
original values to be used. In order to determine the degree to which
the model and field observations differ, observed values were regressed
on model estimates of percent hatch. If the values are identical, then
we expect to obtain a y-intercept of 0.0 and a slope of 1.0. Results of
the analysis produced the equation:

Obs - -12.86 + 0.9729 Mod [31]

r2 - .8439

100
J

PERCENT HRTCH

 

 

Figure 25.

80

 

I I U r U V V r

53 150 190 200
DEGREE-DHYS (>9 0)

Comparison of observed and predicted ETC egg hatch rates.

81

Without resorting to tests of significance, a slope of .97 is reasonably
close to 1.0 and indicates that the rate of egg development is not
different from observed values. The intercept of -12.86 indicates the
number of degree-days that the curves are shifted. Thus, 12.86 D-Dg, or
its equivalent in chronological time, is added to the time-temperature
function for egg development.

Another model output which can be compared with field results is the
maturity of the larval population through the growing season. In order
to examine this, a weighted mean instar was calculated (Eq. 13) from
population samples taken at Gull Lake and compared with simulation
results (Figs. 26, 27). These estimates are not significantly different
(p < .05) with model results providing a good prediction of larval popula-
tion maturity for 1977 and 1978.

Based on the results of ETC egg hatch and larval population maturity,
the simulation model behaves well phenologically. Rates of development
do not differ significantly from observed values and temporal occurrence
of the immature life-stages agree with the natural population.

The pattern of mortality in simulated and observed populations
provides an additional piece of validation data. Results of this compari-
son are shown in Figure 28. What the model actually presents is the
distribution of mortality within the colony for nondispersive larval
stages and survival associated with dispersal from 200 D-Dg on. Because
of this, it would be expected that the model may provide good survival
estimates of the whole population, but differ from the larval population
examined in tent sampling. For the most part, the model agrees with
observed values for 1978 but deviates early in the year (< 175 D-Dg) by

about 10%. This discrepancy may be explained by the allowance of the

82

.mumm you umumsq some counmuoa Dem
mouoqooue com om>uomso mo comfiumeaoo .mm.o»=wfim

.0 ma. m>¢ouuu¢ouo
can ecu sea on. so. on o

   

4992.
ou>¢uaoa

 
 

mbmfi

 

UUISNI N83" 031N013"

.muma now pounce cums mounwwus Dem

omuoaooua mam oo>umunc mo somfiumnsoo

Gnu

.u an. m>¢ouww¢¢mo

2133 1‘23

w
d

ow—

oon on

Juan:
ou>¢umoo

bhmfi

 

.om moswoe

HUlSNI N83“ 031H013H

83

   
   

-——-= HUDEL
--—= 033ERVED 1977
‘---= OBSERVED 1978

 

PROBHBIL I TY 0F SURVIVING

 

U 7 Ti I U T I 1 V

330
DEGREE-OHYS r>9 C)

I I
100 200

Figure 28. Comparison of observed and predicted ETC survivorship curves.

84

model to randomly fluctuate i 10% about the mean values. Also, by
applying survivals at the end of life-stages rather than continuously,
there is a tendency to overestimate populations. Thus, the period from

0 to 175 D-Dg overestimates due to random variability and the method

used in application of mortality. The 1977 curve has a somewhat different
form and plateaus from 150 to 250 D-Dg. It is not known why the survivor-
ship curve assumes this form, although factors such as precipitation may
play a role in larval dispersal. Colonies often remain intact longer
during periods of precipitation and hence provide a different form to
survival and dispersal.

It should be noted that Figure 28 indicates that tent caterpillars
remain in the larval stage from approximately 20 to 400 D-Dg and that
this is the same range observed under field conditions for 1977 and 1978.

The technique of using laboratory-derived developmental information
and time-varying distributed delays provides reliable estimates of
phenological events. Adequate predictions of time of batch, rate of
hatch, time and rate of larval development, and seasonal population decay
were produced from simulation results. It is felt that a realistic picture
of ETC phenology has been produced from the model and may be used in the
field or in future simulation experiments. Additional parameters such as
parasitism and rates of population increase or decline will be examined
with reference to the system as a whole in a later section.

Fall webworm validation. Validation of the webworm component of the
simulation model presents problems which were not seen in the ETC. First,
because Michigan populations were at low levels throughout the study period,
data were not accumulated to the extent that they were in the ETC. Second,

the webworm component was designed around populations studied by R. F. Morris

85

in New Brunswick, Ontario. Morris' populations emerged as adults between
280 and 720 D-D11 whereas Gull Lake webworms emerged between 600 and 1100
D-D11.

Based on Morris' studies and the analysis of species concepts provided
in Appendix F, it is expected that this type of shift could occur in emer-
gence patterns. Developmental rates for larval stages appear to remain
unchanged when moving south in the FWW's distribution. However, degree-
day requirements for adult moth emergence (Kp) do vary. If this idea is
accepted, then shifting the Canadian population by the appropriate number
of degree-days should enable predictions to be made concerning any other
webworm population. Shifting the median of the New Brunswick population
by 420 D-D11 produces estimates of larval incidence and weighted mean
instar which are not significantly different from the Gull Lake population
(9 < .05).

Temporal placement of the FWW population and maturation rates are
examined in Figures 29 and 30. If observed values are regressed on model
estimates after taking into account the 420 D-Dll difference, we obtain:

Obs a -0.1854 + 1.047 Mod [32]
r2 - .9691
The regression indicates that the intercept and slope are not different
from 0 and 1, respectively. Thus, the hypothesis of changes in Kp
relative to geography and climate appears justified and is validated for
the two years examined.

In terms of larval survival, the model may be partially validated
because of low level populations and characteristics of larval sampling.
Figure 31 presents model estimates along with curves for 1977 and 1978

Gull Lake populations. Clearly, values obtained in simulation underestimate

86

.mmmH sou pmumsa some mousmuuz 33m
monogoocm mom om>ummno mo comaumaeoo .om unawam

no -A. m>¢oummxomo

now. com can con 31
r b bI . a P P b ‘4» D b

\
\ .

oom—
—

b b

 

\ T
\ 1:.
i 1
u
\ cut
\ chco m_¢¢o: null:
\ an“ M. I
\ - a.
m 3 1 .

 

UUlSNI NUBN 031H013N

.mmmfi HON poems“ some mouswfioa 33m
couu«omhe cam oo>ummno mo somwumasoo .mm muzmfie

Au __A. m>couumeomo
own can can o

I
b . h — LP D h b P

coma ecu.
P b —

b

E I

\ c w
\ 1 m
\ . . MW
1 a...
\ - m
\ coca u_¢¢o: -.... . WW
Juno: "III:
\ 8:88 «a :9
\
E: .

 

 

87

survivorship from 1000 to 1500 D-Dll. The three curves are compared by

the regression technique used earlier (Table 9).

Table 9. Comparison of model and observed survivorship curves for FWW.

 

 

Model 1977 1978
Intercept 361.29 210.8 382.3
Slope -.2558 -.1150 -.2558
r2 .8631 .7601 .9219

 

For 1978, the model compares favorably with the rate of decline
having idential slopes of -.2558, the differences occurring in the
intercepts of the regression. 0n the other hand, 1977 values differ
significantly from the model. It is not readily apparent why the slope
is less than half that of 1978 values or the model estimates. Changes in
foliage quality and timing of fall leaf drop may, however, affect the
rate of decay. These factors are not taken into account in the model.

Because such differences in the magnitude and rates of mortality may
occur from year-to-year and between locations, the coefficients used in
simulation (Table 7) are retained as reasonable estimates with the decay
rate corresponding well with 1978 data.

Population gpowth and parasitism. Criteria for this portion of the

 

validation include rates of growth for each population and characteristics
of parasitism. Two methods were used in validation: 1) the model was
initialized with field data for 1977 and allowed to run through 1979; and

2) the model was initialized at the beginning of each year with field

88

NOBEL
OBSERVED 1977
OBSERVED 1978

PERCENT SURVIVING

 

 

 

F

r
°1000 1100

Y I f

f f F
1200 1300 1400 1500
DEGREE-DQYS (>11 C)

I

 

I
1600

Figure 31. Comparison of observed and predicted FWW survivorship curves.

89

derived estimates. The results of these simulations were compared to
field data for a 90 Ha. Gull Lake plot from 1977 through 1979.

For the first series of runs, the model was initialized with 1977
data and comparisons made of population statistics as listed in Table 10.
In general, the output from 1977 overestimates each statistic for the
tent caterpillar. Overwintering eggs are 24.07% greater, the number
of colonies produced for the succeeding year are 4.28% greater, and the
generation index, N(t)/Nt-1), is 0.08 larger than the observed value.
This is not unacceptable in that the prediction of colonies for the 1978
season overestimated the true number by only 34 tents. This could con-
ceivably be accounted for by the conservative estimate of egg mass size
used (220 eggs/mass) or by any errors in population counts.

In terms of Campoplex, it is not possible to assess spring or
initializing densities. As mentioned earlier, Campoplex was not detected
within the field plots at Gull Lake. Hence, qualitative statements
concerning the performance of the related species, H, g, fugitivus can
only be made. It should be emphasized that this model represents the
first iteration of a continuous process of reformulating concepts of
system composition and parameterization. Further study would seek to
build on the insight gained here. For 1977, the input to simulation was
10 parasites. This value was chosen based on preliminary runs and on the
observed parasitism rate for that year. Even with this seemingly low
density of parasites (0.11/Ha.) 7.94% of the simulated ETC population was
parasitized as compared with 0.71% in the field situation. Clearly, a
number of factors account for the discrepancy, the most obvious of which
are species differences between Campoplex and Hyposoter. Under laboratory

conditions, Campoplex may deposit as many as 22 eggs/day while Hyposoter

90

 

 

 

 

 

 

00.0 I- 00.0 00. 00. 00. 00. 00. 00000 0000000000
000 I- 000 00 00 00 00 00 00+00 00000000
00.0000 00.0000 00.0000 0000 00.000 00.0000 00.000 00.0000 00000000 000000000
00.00 00.00 00.00 00.00 00.00 00.00 00.00 00.00 0000000000 0
000 00 000 000 000 000 000 000 00000 00000 230
00.0 I- 00.0 00.0 00.0 00.0 00.0 00.0 00000 0000000000
0000 .. 0000 000 000 000 000 000 00+00 00000000
00.0 00000 000 000 000 00.000 000 00.000 00000000 000000000
00.0 00.0 00.0 00. 00.0 00. 00.0 00. 0—000000000 0
0000000 -- 000000 000000 000000 000000 000000 000000 -0000 00 0mmw0wm0
000000 000000 000000 000000 00000\ 00000 00000 00000 00000 0000 000
Homo: vm>0umno ammo: oo>uumno Home: vu>ummno Home: ow>uumno
0000 0000 000000 0000>0000 0000

shad

 

.uaeuao m.umoh

mooa>oue may 5003 @00000000a000 .Hovoa 0:0 5003 0005000 @0000 cam Hoooa mo 6000009300 .oH canoe

91

may lay 3 (Morris 1976). If the attack function is altered proportional
to the difference in eggs laid, this aspect can be examined. Even with
the reduction in attacks, parasitism was only decreased by 1.79% and this
does not account for differences with field data (Table 10).

Parasite-host synchrony could also account for differences in attack
rates. Figure 32 shows the temporal coincidence of susceptible host
stages and parasite populations as simulated. The ETC and Campoplex
are well synchronized in this example for each year examined. Because of
Campoplex's longevity, the entire susceptible host population experiences
some level of parasite pressure. Under field conditions, Hzposoter
attacks ended at 192 D-Dg, 134 D-Dg, and 139 D-Dg for 1977 through 1979,
respectively. Field data suggest that attacks end somewhat earlier for
Hzposoter and that the later portions of the host incidence curve
(> 140 D-Dg) experience considerably less parasitism than simulated,
particularly so in 1978. Attacks are also reduced in simulation with
1978 showing the largest decline in adult Campoplex incidence, and 1977
showing the least decline, beginning approximately 125 D-Dg (Fig. 32).
Statistical significance cannot be detected between the 3 years, but the
trend in simulation and field estimates is similar, possibly due to
differences in yearly temperature regimes.

Reinitializing the model with the previously simulated year's outputs
produces instability in the model. By the end of 1977, 916 parasites were
produced to overwinter. If this number is used along with ETC and FWW
outputs, the tent caterpillar quickly goes to extinction by the end of
1978. Reinitializing in this manner implicitly assumes that 100% of the
parasites will survive and remain within the area for the following spring.

Under field conditions over 8000 parasites (Hyposoter) were produced by

92

Al I SNEIU 31196860

 

 

   

 

.mm>u=o oucmvwocu uaavw godmomsmo can Hm>uma 09m .mm muawfim
no m3 w>¢ouumm¢uo
omw cow OB 900 cm 9
0 0 L . . . _ 0 0 01 0
II II II II \
Illlh.’ ' I \
. 0. i. .0 .
Z 7..
nuJ .Inu
O
J '[II'I' I
are I H
M 2.3 ulll TM W
4 00.00 0|! 0 1
0000000000 0
0 . 3
w
00 03 71
0.0 1m I.
IA
0..
J I I
O
a.
3 9
nuJ [no
000 o
m1 «m
0 0
0

93

the end of 1977 and only 0.41% parasitism resulting in 215 parasites

from the ETC at Gull Lake in the spring of 1978. In contrast, 5759
parasites overwintered at Gull Lake in 1978 and 14,937 were produced

from the ETC in 1979 (Tablell». Thus, the relationship of overwintering
parasites such as Hzposoter and Campoplex and those emerging in the spring
is not clear. Further field studies are required to determine if the
variation in spring densities is due to immigration or emmigration, or

if hosts other than the FWW are being attacked later in the season. A
number of other hosts are parasitized by both Campoplex and Hyposoter and
may add to the overwintering generation. Low webworm densities can also
serve to cause parasite dispersal away from areas and further reduce spring
densities on a local basis. In either case, these factors are not con-
sidered in the model.

If the parasite density in the spring is held constant and the
webworm and tent caterpillar are reinitialized with previous outputs, the
generation index for ETC remains at 2.36 for 1978 and 1979. Under field
conditions, the generation index for 1978 was 1.23. Thus, the rate of
increase in the ETC is 1.9 times greater in simulation (Table 10)-

An alternative to allowing the model to run continuously from year
to year is initializing it each spring with known densities. This takes
into account problems of estimating overwintering survivals and dispersal.
Results of this provide a better estimate of ETC increase for 1978. Once
again the observed generation index for 1978 was 1.23 with a revised
model estimate of 1.88 (Table 11).

For the FWW, continuous model runs also lead to unacceptable results.
As seen in Table UL each successive year produces increasing numbers of

webworms while actual field estimates indicate a decline in webworm

94

 

 

 

 

 

 

00. - 00. 00. 00. 00. 00000 0000000000
00.00 - 00 00 00 00 00+20 00000000
. . . . 00000000
00 000 0000 0 000 0000 00 000 00 0000 000008080
00.00 00.00 00.00 00.00 00.00 00.00 0000008000 0
00 00 000 000 000 000 00000 00000 330
00.0 - 00.0 00.0 00.0 00.0 00000 0000800000
00.0000 - 00.0000 000 000 000 00+00 00000000

. . vmuawoum
000 00000 000 00 000 000 00 000 000000000
00.0 00.0 00.0 00. 00.0 00. 0000008000 0
umuawa
000000 - 000000 000000 000000 000000 -00>0 00 0000 000
000000 000000 000000 000000 00000 00000 00000 000 000
Homo: wm>ummno Hove: um>ummno Honor cm>ummno
0000 0000 0000

 

.mucaoo vfimfiu Hmnuum £003 vmuwamfiuaaumu HmvoE £003 muaammu waoww can Hence no somwumnaoo .HH manna

95

density. Initialization of each year's run with known densities provides
more acceptable results with overestimation of the number of colonies for
90 Ha.'s of 34 and 30 for 1978 and 1979. Further, the rate of decrease
in population numbers prediced is similar to observed values (-43 and -47,
respectively) (Table 10).

Parasitism and encapsulation in the fall webworm. Rates of para-
sitism in the webworm remained remarkably constant during simulation even
though parasite production from the ETC ranged from 172 to 796. This is
due, in part, to the impact of encapsulation on parasitism. As described
earlier, encapsulation varies with instar and with the time of adult
emergence (Table 8). This factor was examined using data collected from
the Gull Lake study. Colonies were selected on the basis of having at
least one attack in the larvae dissected from a colony. In addition to
ascertaining whether an individual had been parasitized and the egg of
Hyposoter encapsulated, the number of eggs laid, the position of the egg
in the host, and the age of the host were also determined. A forward
step-wise multiple regression was performed with percent encapsulation
(PCTE) regressed on the above factors. To take into account the aspect of
time of adult emergence (Kp) and encapsulation, it was assumed that, at
the colony level, if a group of larvae had emerged earlier than another,
then the average age of those individuals would be greater. This age
factor was measured in terms of a weighted mean instar (Eq. 13).

Results of the regression showed that position in the host and number
of eggs laid were significant in determining encapsulation levels (p < .05).
Age of the individual, on the other hand, was not significant (Table 12).
The fact that the time of emergence (Kp) and encapsulation were not signi-

ficantly correlated does not necessarily invalidate assumptions made in

96

the model. That is, calculating Kp in the manner described will add signi-
ficant amounts of variability to the relationship. For example, each
colony examined was subjected to different temperatures, food qualities,
and other factors considered a part of the microclimate. Each of these
factors contributes to the variance associated with colony maturity along

with the time of emergence.

Table 12. Regression statistics for Hzposoter egg encapsulation.

 

 

F to Enter Signifi- Overall Signifi-
Variable or Remove cance r2 F cance
Position (POS) 7.25** 0.01 0.13 7.25 0.010
No./Host (N/H) 4.41* 0.04 0.20 6.08 0.004
Wtd. Mean Ins. 0.42 n.s. 0.52 0.21 4.14 0.011

 

PCTE - 28.99 + 55.06 ln P08 + 48.49 In N/H

 

* (p .05)
** (p .01)

In terms of position, two factors are significant. First, 89.59% of
all Hzposoter eggs laid were deposited posterior to the fourth set of pro-
legs. Of the remaining ovipositions, 8.782 were placed in the area of the
prolegs and 1.63% were anterior to that. Second, given that an egg had
been laid, there was a 0.1868 probability of being encapsulated in the
posterior section, 0.3953 in the mid-section, and 0.50 for the anterior
portion. Combining these numbers produces the probability of oviposition
agd_encapsulation for each section; 0.1679, 0.0347, and 0.0082 for posterior,

mid, and anterior sections, respectively. Figure 33 shows the relationship

97

monMHwom

.m>uoH 33m a mo mooauuon
ucouommwv :« umuomomxm MOM coaumasmaoocm can acauumoa«>o mo muaaunmnoum

mqoomz

 

Laozm + L~>o

zouhcaammcozu

 

 

h\\\\\\\\\\

zouhuwom~>o

 

b

 

W

mommMHzm

 

W

 

 

 

7'0 2’0 0'0

AllWIQUBDHd

9'0

8’0

I

.mn ouswfim

98

between the section of the host's body, ovipositional preference, and
probability of encapsulation.

The validation of the ETC-FWW system model has produced some promising
results. Phenological aspects of the model compared well with field data
as did rates of host population increase. The inability to quantify
parasites not contained within hosts, however, makes it extremely
difficult to track the flow of individuals between host species. This
is particularly true for the overwintering period. At this point in
time, it is not possible to fully examine the model's performance relative
to parasitism and encapsulation and we may only make qualitative infer-

ences based on the general timing and magnitude of events.

Model Sensitivity

Temporal synchrony. The ETC-FWW model, being phenological in nature,
lends itself to a series of investigations involving time synchrony rela-
tionships. Temporal synchronies have been examined in the past principally
at the macro level, that is, considering how parasite and host incidence
curves interface. For the most part, these data indicate a positive out-
come from varying degrees of asynchrony. The creation of temporal refugia
for the host organisms acts as a stabilizing force in long-term relation-
ships (Miinster-Swendsen and Nachman 1978). Reasons for temporal coincidence
or incoincidence have not been examined to any great extent. From a
management standpoint, the restriction of study to the incidence curves
rather than causal mechanisms greatly reduces the number of control options.
In terms of ecology of parasite-host systems, analysis of incidence curves
can only describe the outcome of temporal relations (usually percent

parasitism) and the effect on system stability.

99

Throughout earlier portions of this work, a reductionist approach
has been taken. I will continue to utilize these methods here by con-
sidering time synchronies as a function of their subcomponents. With
regard to the host, its placement in time depends upon the rate that
susceptible life-stages enter and leave the system. Therefore, develop—
mental rates of those stages are of importance. In light of this, system
responses were examined from the standpoint of altering the host popula-
tion's developmental characteristics. These alterations were performed
through changes in the temperature regime, changes in the population's
phenotypic composition (Kp), and changes in species of the host plant.
From the parasite's viewpoint, the temperature regime will produce varying
emergence patterns. In addition, longevity of adult females increases
or decreases the time available for attack activities, as well as the
degree of overlap in incidence curves.

Temperature regimes. The effects of different temperature regimes

 

were simulated by obtaining temperature data from 3 different areas in
the state of Michigan for 1977: Gull Lake, Houghton Lake, and Marquette
(Fig. 34). These data were input into the model as specified earlier.
For this series of runs, population densities were initialized with field
data from Gull Lake (Table 11) with the frequency distribution of Kp
for the FWW derived from field samples. The proportion of individuals
in Kp 1 through 10 are 0.0351, 0.1170, 0.1871, 0.2047, 0.1579, 0.1170,
0.0760, 0.0585, 0.0292, and 0.0175, respectively.

As the ETC-FWW system is moved on the south to north gradient,
significant changes occur in the temporal synchrony of all 3 populations
(Fig. 35, AFC). The Gull Lake simulation demonstrates a very close

association between Campoplex and both host species (Fig. 35A). Moving

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

kilo/'22

 

 

 

 

(
A ”W
L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EE

 

 

 

 

 

 

 

 

 

 

 

 

.— I”. 0.. I.“ I w 0-
.I.‘

Figure 34. Locations in Michigan for simulation runs.

101

uOIlALISNBO JHUD ‘ HHJ

.xmmmoaemu cam .33m .09m new mo>u=u mocmvucca

 

    

 

    

 

 

 

 

 

 

 

.n
>¢o 2:.435
ova can can 0.. 0.— Su ON—O
P p u h b r b b u L b
a
to
no
3
a
3
nuaxogu ul l 9 3
two N
2.... "III 1
v 3
U
3
Lam
N
, n
‘
10
v
[m
.m
0024. > om- . 82L 80
at .u
l f
3
m. .m m.
a
A v 3
N
s
. . I
0; To 1
u
a . ...u.
.
“L T“
0.;
ml loo

uOIIAiISNBO JHUS ‘ HHJ

IOIlAlISNJO JHUJ 0 HHJ

 

    

 

 

 

 

 

    

 

 

.mm muswwm
:3 2:32. .o
oe.~.om~.£~.fl_.3..en..om_.2.:.moo
l 1
A v 3
9 9 I.
0... 103
a
. l m
S
«1 wan“
A
I
. nude-eecu . mm
"L I; u: I“
. . mecmzamcz sem_ .
m. .m
>¢a 2¢_4:fi f<
am.“ on“ can 9.. . on. L a". ., ems A. small. ago
J x
A a V
.. 1 ..
01 § '0
A ~ v
_ 3
s _ o m
~ .3
u
A . m
“4 flag I“
I: a:
A memo saga semi fl

 

 

N!

102

to Houghton Lake, a shift in the FWW-Campoplex interaction occurs (Fig.
358). This shift continues further as can be seen in the Marquette

run (Fig. 35C). Figure D allows a direct comparison of each of the

runs by plotting cumulative percent emergence curves for Campoplex

adults and the webworm larvae over time. The movement toward asynchrony
is a direct result of differences in developmental temperature thresholds.

Campoplex's base of 42° F (5.56° C) allows it to accumulate more degree-

 

days than the webworm whose base is 51° F (10.56° C). This will most
affect populations in cooler temperature regimes where Campoplex pupae
will develop during periods of quiescence in the FWW.

As expected, parasite production decreases on the south to north
gradient. At Gull Lake, 772 Campoplex pupae were generated from attacks
on ETC, and 916 from the FWW. Houghton Lake produced 751 and 923,
while Marquette produced 699 and 897 from the ETC and FWW, respectively.
Though the trend in declining parasitism is apparent, it does not reflect
the extent of asynchrony shown in Figure 35D. Reasons for the smaller
than expected decrease in parasitism with high levels of asynchrony are
not clear. The most feasible explanation lies in a high ratio of hosts
to parasites, presenting low densities of Campoplex with a seemingly
endless supply of susceptible larvae. Each adult parasite may attack the
maximum number of webworms that it possibly can.

It is not known whether this shifting phenomenon actually occurs
under field conditions. The assumption is that characteristics of all
3 populations remain constant as one moves to different geographic areas.
As shown earlier in this study and in Morris' work (Morris 1971, Morris
and Fulton l970a,b), emergence patterns may change annually and with

geographic location primarily in response to temperature, precipitation

103

and related food quality. Thus, the simulated results may be applicable
only under situations where these other selective forces are minimized.
Some credence is, however, lent to these simulations in that Campoplex

is synchronized with the early portions of FWW populations in New Brunswick
and areas inland (Morris 1976). This also occurs in the model for
Marquette which is approximately at the same latitude.

From a management standpoint the resultant asynchrony on the south
to north transect may have both positive and negative results. A posi-
tive outcome is that the opportunity exists to implement additional
control options for later emerging portions of the FWR population.
However, on the negative side, if asynchrony is severe enough that
little impact can be made on the host (pest) population, then one must
resort to a series of manipulations favoring later emergence of Campoplex.
Further, if the system is relatively insensitive to these changes, examin-
ation of alternative management techniques may be warranted.

Effects of phenotypic composition in the fall webworm. In an earlier
section, I alluded to the idea that the phenotypic (genotypic) composition
of a webworm population may play an important role in parasite production.
In turn, the amount of parasite pressure placed on the ETC in the following
season may be largely dependent on the distribution of webworm phenotype.
This idea can be tested in simulation by varying the frequency distri-
bution of Kp in the FWW. Four distributions were examined emphasizing
early, middle (a normal distribution),and late portions of the population,
as well as an equal number of individuals in each Kp class (Fig. 36A).

The total number of FWW input was 10,000 with an equal number of ETC

and 10 Campoplex pupae for spring initialization.

 

104

 

 

 

 

 

 

O
O
ID-n
n
4 LEFT NORflflL RIGHT
O
O n; n
o—w
n
o 4
D
IO-
N
.- 4
D 8
s s:
H 4
.4 o
2 a
E . count. ‘ ~ ..
o»
.8:
84
ID
d
a
o l I
0 1 2 3 4 5 8 7 8 9 10
A. KP CLRSS
o
8
‘1
J ‘8 LE?!
Uslflflflm.
(a O .& Rt“!
5 §‘ +. count.
:3 t
3
m
m d
8 I
c: O "
2 8.4
0-0 Q
I:
u:
..—
2: ‘ t
:3
a O
In“ 8d
>ȢI t
o .
O T I T T I I I I T I
0 1 2 3 4 5 8 7 0 9 10
KP CLFISS
B.
Figure 36. Input/output relationship for effects of phenotypic

composition in the FWW.

105

Figure 36B shows how the composition of the population can change
after a single generation. Because of differential fecundity and
encapsulation, selection tends to favor those individuals in the first
5-6 Kp classes. Therefore, the Kp distribution changes as a result of
varying fitness characteristics. In the distribution skewed LEFT, little
change occurs. However, the mode for NORMAL and RIGHT distributions
shifts by 1.0 and 1.5 Kp classes, respectively. The EQUAL distribution
assumes the pattern of fecundity and peaks in classes of 3-4. Were this
trend to continue for another 3-generations, each of these populations
would assume the frequency distribution similar to EQUAL and LEFT treat-
ments .

These results are somewhat counter-intuitive in that a mortality
agent (Campoplex) appears to be contributing to forcing its host into a
higher degree of synchrony with parasitism. One would expect the webworm
to opt for a strategy of avoidance as in the RIGHT treatment. However,
the ability to encapsulate parasite eggs and reproduce more rapidly,
outweighs parasitism. Therefore, the webworm coexists with Campoplex
rather than avoiding it and compensates for parasitism through its capacity
to increase. In fact, if we compare the numbers of parasites and FWW
colonies produced, little change occurs in colony production for each
distribution treatment (Fig. 37). In terms of Campoplex significant
differences occur with over an 85% decrease in pupal production. The
mechanisms involved are compensatory in nature and involve encapsulation,
fecundity and temporal avoidance, depending on synchrony with the adult
Campoplex incidence curve.

With reference to the system as a whole, the distribution of pheno-

type (genotype) plays a large role in ETC parasitism and resultant

106

.33m mnu :« noduamoaaoo

ouahuococa mo :ofiuoasw m an xmamomaoo cam 33h mSu you couuosvoua Hanan .mm muswam

.J

anew-

p.

:a

0—0
0:

 

 

K\\\\\\\\

E

 

 

L\\\\\\\\\

L\\\\\\\\\

 

 

\

 

 

 

 

mucous Hum; 0
fig
no
“N

M.d
0.d
0.”
.d
”U
T .3
mm
Im.u
0
0.3
M”
M”
no
mm
1mnu
am
a.
.L a no
33m
mzco

 

r
0008

107

parasite production 2 generations removed. As the distribution becomes
skewed to the right, fewer parasites are produced to attack the ETC.

In terms of pupal production, the FWW is little affected by its own
phenotypic distribution. The tent caterpillar, on the other hand, experi-
ences increased parasite pressure.

A factor which has not been discussed explicitly but is an underlying
mechanism in selection for early emerging webworms is food quality. It
has been shown that Kp is under genetic control with early individuals
begetting early individuals. The aspect of timing with wild black cherry
was also shown to favor those individuals with low Kp. Year after year,
this situation will hold true even though in certain seasons conditions
allow latter portions of the webworm population to expand in time. When
compared with a sporadic mortality factor such as parasitism, it is not
surprising to see selection for temporal avoidance overridden by the
need for high quality foliage. Throughout Morris' (1976) study of
parasitism, the encapsulation is implied to be under genetic control.

This may be true, but only to the extent that time of emergence (Kp) will
directly determine food quality and the amount of stress placed on larvae.
Larvae which are under greater nutritional stresses (high Kp) would have

a more difficult time in allocating energy to encapsulation than those
which are not (low Kp). This does not necessarily mean that encapsulation
is independent of individual host differences (genetic), but these
differences are masked by the interaction of emergence and quality of
foliage. The encapsulation coefficients used in simulation (Table 8)

may actually be an assay of nutritional stress as opposed to directional

selection combatting parasitism.

108

As with the series of runs examining geographic variability, pheno-
typic variability may also play a role in determining the makeup of
management systems. Selection forces other than parasites determine
a population's susceptibility to parasitism. Thus, the mere determination
of pest density will not only have little to do with resulting parasitism
rates, but also restrict management options to those which affect the
pest population as a whole. Strategies which utilize population structure
represent a class of options not used to date. In the following section
I will analyze the effects harvesting strategies aimed at predetermined
portions of the webworm population.

Harvesting strategies. Phenotypic composition in the FWW was shown
to significantly affect parasite production. This composition, however,
was varied only in the initial stages of the growing season. The question
here is "What effects do explicitly timed harvests have on webworm and
Campoplex production?"

The model was initialized with 10,000 ETC, 10 Campoplex,and 10,000
webworms distributed in proportions determined from Gull Lake 1977.
Survivorship larval stages of the FWW were decreased to 99% throughout a
specified time interval (in degree-days). By doing this, an impulse—like,
timed, mortality factor was introduced into the system, much like a short
residual pesticide. Performance of the system was evaluated with
reference to the numbers of FWW colonies and Campoplex pupae generated
in the succeeding generation (T + 1).

If the system is evaluated with reference to the number of parasites,
then it is possible to actually increase parasite production by harvesting
very early in the webworm larval incidence curve. Figure 38A shows the

relationship between time of harvest and parasites in the next generation

109

 

 

 

\\\\\\\\\§ \\\\\\\§
\\\\\\\ 3 \ \§\\ 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\\\\\\\\\¥ \\\\\\\\\\

rrrrrrr r fi
L o ’ o o

\\\\\\\\a \\\\\\
X R\\\\\\\\a \\\\\\\\\\s

 

 

 

 

 

 

00’ SL’O
N ON 740118837430

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\\\\\\\\\\\s§ \\\\\\.§
\\\\\\\\\\§h \\\\\\§: ‘“

 

 

 

 

 

 

N\\\\\\\\\\z \\\\\

 

 

\\\\\\\\é§= \\\\\\\:§
\\\\\\\\\=; \\\\\\\s;
E\\\\\\\\\\s— \\\\\\\\\\.— ~

\\\\\\\\ \ \\\\\\\\\\

110

(T + l). The increase at the 450 D-D11 harvest, though not large, suggests
that early portions of the FWW population (low Kp) are selected against.
Thus, individuals with high encapsulation ability are removed from the
population. By doing so, the webworm's susceptibility to parasitism is
increased. Because the 450 D-D11 harvest occurs so early in the season,
only a small portion of the FWW larvae are affected. Thus, there is
virtually no affect on colonies produced (Fig. 38B).

In terms of parasite increases relative to host increase, harvesting
later in the season maximizes Campoplex, while minimizing FWW population
growth. This aspect is shown in Figure 38C with the largest ratio of
parasites to hosts occurring at a 750 D-D11 harvest. Here, much of the
webworm's reproductive capability is removed with Campoplex able to
increase at a greater rate. In fact, in all previous harvesting
strategies the generation index (i.e., colonies in the following
generation/colonies in the previous generation) for FWW was greater
than 1.00 indicating population increase. Here, the index drops below
1.00 indicating population decrease (Fig. 38D).

Again, the model suggests that variability throughout the webworm
population produces significant differences in response to parasite
attack and indirect effects on ETC populations. Further, it was shown
that the phenotypic composition of FWW populations produces a larger
affect on overwintering Campoplex populations than on the webworm itself
(a host-host interaction). Similarly, this host-host interaction will
be seen in strategies which manipulate only ETC numbers.

Numerical sensitivity to ETC densities. Numerical sensitivity
refers to how the system responds to changes in host-host and parasite-

host density relationships. One explicit assumption of the model is that

lll

parasite pressure applied to the webworm is dependent on parasite produc-
tion in the ETC. Thus, it is desirable to examine this interaction from
the standpoint of varying ETC populations and determining the impact of
the FWW.

In order to evaluate the interaction, the model was initialized
with 10,000 webworms (1,000/Kp class), 10 parasites and ETC densities
ranging from 5,000 to 50,000. Because temperature is considered
explicitly in the parasite attack function, simulations were run using
a maximum temperature of 70° F (21.11° C) and a minimum of 50° F (10° C)
for each day simulated. This was done to remove any variability caused
by spring and summer temperature regimes.

Continual additions to the ETC population results in a saturation
of the parasite production curve (Fig. 39A). Inputs of tent caterpillars
after 40,000 (ca. 182 colonies) produce no significant increases in second
generation parasites. Similarly, there were no significant gains in
overwintering Campoplex after this point (Fig. 39B).

The decreasing rate of Campoplex production is particularly dynamic
when viewed with reference to the 10 Kp classes in the webworm. Figure
40 shows these responses with classes 2-5 most sensitive to changes in
ETC density. Reasons for this lie in the interaction of differential
fecundity, encapsulation and Campoplex-webworm synchrony. As second
generation Campoplex are augmented through the addition of ETC, the
magnitude of the parasite incidence curve increases to better synchronize
with FWW larval incidence and increase parasite pressure.

Thus, a viable management option may revolve around the augmentation
of ETC populations and any factors which can be manipulated in the one—to-

one interaction of Campoplex and ETC or FWW. Techniques which allow

PRRRSITES PRODUCED

OVERHINTERING PHRRSITES

B.

Figure 39.

112

 

 

 

fi’ fl 1 V

r I 1* I
20000 30000 40000 80000

ETC INPUT

I
10000

 

 

 

V T

o I r I I 'T r I
0 10000 20000 30000 40000 80000
ETC INPUT
Campoplex production in response to varying ETC densities.

113

  
  
   
  

 

 

 

 

 

D
D
01
4')
O
8.4 arcmm
“03 N o: soooouaooo
._ += 40000::0000
'5; g ' 0: 30000310000
3; 8‘ an 10000::0000
0: m. 5000:10000
0.
O
s g4
H
1
LL]
'- o
E 84
3 d
m L-
U
>
o 8-
ID
0 I I I Ti r I I "
0 1 2 3 4 6 B 7 8 9 10
A- KP CLRSS
O
O
O-
F
ETCIFHH
3 ma 5000310000
8q a: 10000::0000
a, 0: 20000::0000
g g 40000310000
g m 50000::0000
8
O
3 8d
a V
2
Izzgd
:3 n
z
z s
5 .r
>
O O
O
g-w
c, I I I I I I l’ I I I
0 1 2 3 4 6 a 7 a 9 10
B. KP CLRSS

Figure 40. Fall webworm population response to varying ETC inputs.

114

parasites to spend more time in contact with susceptible host life-
stages (i.e., developmental rates of host larvae and adult parasite
longevity) are two such options and will be examined in the following
sections.

Host plant effects on development and parasitism. Plants directly
affect developmental rates of herbivores. The FWW is no different with
significant changes in the amount of time spent in each larval instar
(Kovacevié 1954). In turn, synchrony of parasite susceptible life-
stages with adult parasites should be affected. Two extremes in slow and
fast larval development were examined for the FWW reared on apple (Malus

pumila Mill.) and maple (Acer saccharum Marsh.). Data for these runs are

 

presented in Table 13.

Results suggest that as more time is spent in susceptible instars,
higher rates of parasitism occur. Of those webworms reared on apple,
19.352 of the larvae were parasitized. For maple, there was 24.69% para-
stitism. Increases in developmental time affords parasites more time to
locate hosts and increases rates of parasitism. Thus, parasite-host
synchrony can be positively affected through the herbivore-plant inter-

action.

Table 13. Host plant effects on Campoplex parasitism
in the fall webworm.

 

 

 

Malus Acer
Colonies (T + l) 6522 6535
Campoplex pupae 7822 8212

Percent parasitism 19.35 24.69

 

115

Parasite longevity and parasitism. In addition to the host's develop-

 

mental times, adult parasite longevity is a component of synchrony and
parasitism rates. Many studies have examined the factors affecting longe-
vity and allude to how it might affect parasitism (Leius 1967; Barney et a1.
1969, Fusco et a1. 1978, Miller 1977). Typically, data are derived under
laboratory conditions and extrapolations are made to field situations.

The objective of this series of runs is to assess the system's sensitivity
to alterations of Campoplex's longevity.

Again, 10,000 ETC, 10,000 FWW (distributed as in Gull Lake, 1977) and
10 Campoplex were used to initialize the model. In order to alter the
adult's life span, the delay function used in all previous simulations
was taken as a liberal estimate (ca. 30 days), then halved and quartered,
resulting in longevity ranging from 7.5 to 30 days.

For both the ETC and FWW, an exponential decrease occurred in parasite
production. When Campoplex was allowed to live for 30 days, 456 parasites
were generated from the tent caterpillar, and 7,822 from the webworm. A
lS-day life span produced 326 and 2,098, and 7.5 days 8 and 2 parasite
pupae for the ETC and FWW, respectively. Therefore, significant changes
occur as a result of factors such as food availability, temperature,

pesticide intervention, or any variable altering parasite longevity.

CONCLUSIONS FOR THE ETC-FWW SYSTEM

The study of population dynamics includes two basic areas of study,
development and survival. Development relates to temporal placement of
populations, while survival studies (within generation) attempt to identify

patterns and causes of mortality. In the previous analysis, development

116

and survival were examined in two ways. First, the major components of
the system (i.e., ETC, FWW, Campgplex, Hyposoter, and WBC) were examined
in isolation. This was done only to the extent that developmental rates
could be determined, stage-specific survivals calculated, and patterns
of mortality described. In particular, two distinct survival strategies
were found in the ETC-FWW system. The ETC synchronized its emergence
with high quantity and quality food. Further, because of rapid develop-
mental rates, particularly in instars 1-3, the ETC was able to minimize
the length of time that individuals remain in life-stages susceptible to
parasitism and other mortality factors. Eighty percent of the year was
spent in the egg stage (actually pharate larvae), and it was found that
between 85 and 95% of the individuals survived in that stage.

The FWW makes use of variability to disperse mortality throughout
the population. Early emerging individuals (adults) initiate colonies
on high quality foliage and consequently are more fecund and have a higher
fitness overall. This includes the ability to encapsulate parasite eggs
and avoid early frosts prior to pupation. Late emerging webworms feed
on a lower quality food but temporally avoid parasites such as Campoplex
so that there is some compensation for decreased fecundity and suscepti-
bility to cooler temperatures. Webworm populations, therefore, respond to
environmental pressures through phenotypic variability and spread risk
to various types of individuals (Den Boer 1968, Reddingius and Den Boer
1970).

It is virtually impossible to view population processes without
relating components. Hence, interactions between the tent caterpillar,
fall webworm, and any common parasites (e.g., Hyposoter or Campoplex)

complete the flow of energy within and between growing seasons. Parasites

117

produced by the ETC go on to attack the FWW and other hosts before
returning to tent caterpillars the following season. The model that was
developed points to a number of factors linking the three components.

One of the more intriguing ideas generated is the concept that the pheno-
typic composition of a fall webworm population can have major impact on
parasitism in the tent caterpillar. Determination of host and parasite
densities as single numbers is not sufficient to explain the resultant
parasitism rates.

Concepts generated in simulation are indeed useful,though. as seen
in validation comparisons do not always approximate real world phenomena.
In this example, the model's greatest use is derived from its inability
to make predictions. The fact that increases in parasite production in
the FWW can result in decreased attack in the ETC points to further
complexity. In this case, system conceptualization makes no allowances
for immigration or emmigration processes in the parasite component, and
certainly not for other potential hosts. The utility lies in directing
future studies to examine these factors along with an overwintering
component examining spatial patterns and movement of parasites into host
inhabiting areas.

The plant component of the system model is considered implicitly in
the stage-specific survivals and developmental rates. However, it is
static and by no means impacts the simulated system as it does in the
real world. This was seen, to a limited degree, when altering webworm
development as a function of host plant species in simulation. One of
the largest impacts on the system may be within a species of host plant

and effects on the plant by successive feeding of ETC and FWW. Recent

118

studies by Feeney (1975), Haukioja and Niemela (1979), and others show
the dynamic aspects of the defoliation-refoliation process. Feeding by
ETC and subsequent refoliation will cause changes in the quality of
foliage for the FWW and other late season defoliators. This in turn,
affects their mortality patterns, fecundity, susceptibility to parasitism,
and most importantly, the underlying distribution of phenotype and/or
genotype for the population.

Research in the ETC-FWW system has pointed to the need to accept
complexity within and between components. This need further points to a
new set of system concept requirements not presently adhered to in manage-

ment systems using natural controls.

DISCUSSION OF CONCEPTUAL NEEDS FOR
MANAGEMENT OF PARASITE SYSTEMS
Given both the holistic philosophy promoted throughout this work and
information gained in the analysis of the ETC-FWW system, "How is our
concept of parasite-host interactions altered?" The following discussion
considers different categories of parasite systems and outlines those

factors functioning as determinants of parasitism.

Framework for System Conceptualization

Parasite systems in general may be viewed as subsets or simplifica-
tions of the conceptual model shown in Figure 41. Here, a generalized
succession of hosts is attacked by a parasite pool. Hosts, as well as
the parasite pool, may either be represented by single species or a
group of species. Further, the magnitude of the developmental time lag

(At) then determines whether the model applies to a single growing season

 

PARASITE
POOL (t)

 

 

 

 

 

HOST N
(t + n(At))

 

 

 

119

 

HOST I (t)
A

 

 

 

 

 

HOST
PLANTS

 

 

 

 

 

PARASITE
POOL (t + At)

 

 

 

 

HOST'Z
(t + At)

 

 

 

 

PARASITE
POOL (t + n(At))

 

 

Figure 41. Generalized succession of hosts and parasites.

 

120

or a series of N growing seasons. For example, Host 1 at time t appears
in the spring and is attacked by parasites at time t. After developing
in Host 1 the multivoltine species re-emerge to attack Host 2 at t + At.
Host 2 is either a second generation of Host 1 or another species. The
time period, At, represents the developmental time lag for the parasite
pool (i.e., length of immature life-stage) and affects their synchrony
with Host 2 at t + At. This can be generalized for N numbers of hosts
and n time lags (i.e., the parasite pool at t + n(At) and Host N(t + n(At)).
Variations of this generalized model-are as follows:
1) singlehost-parasite systems (SHP) (Fig. 42A),
2) multiple host-parasite systems: synchronous host
availability (MHPS) (Fig. 42B),
3) multiple host-parasite systems: asynchronous host
availability (MHPA) (Fig. 42C), and
4) multiple parasite versions of 1—3. (This final
variation will not be considered explicitly in this
study. Considering all of the possible combinations
it is felt that the inclusion of number 4 will only
serve to cloud the more general concepts behind the
determinants of parasitism.)
In most cases, parasite competition can be looked upon as an additional
mortality factor for immature parasites becoming part of the biotic
environment. In addition, little information is available concerning

parasite-parasite interactions other than the mortality aspects.

 

121

 

 

 

 

 

 

 

 

 

 

 

[dParasite ————> Host T—b Plant

 

A. Single host

 

 

 

 

 

“"""‘1
—-—>Parasite ————> H2(t) —9

\E/

v Plants

 

 

 

 

 

 

 

 

 

B. Synchronous multiple hosts

H (t)
Spring /' 1
p ' I \P(t + At)

 

 

 

 

 

 

 

 

t

 

 

 

Overwinter Host
[Delay

[flN(t + At)

C. Asynchronous multiple hosts

 

 

 

 

 

 

”ut+ufi

 

 

 

/\3
v?

 

 

(t + At)

 

Figure 42. Classes of single and multiple host-parasite systems.

122 .

Single Host-Parasite Systems

Single host-parasite systems (SHP) are defined as in Figure 42A and
composed of a single species of host and parasites. The basis of
parasite—host interaction theory has been developed around this model and
serves as a point of departure for more complex systems. One of the more
important concepts to be drawn from the SHP model is the idea that the
host organism is, to a large extent, responsible for its own fate in
terms of parasite pressure in succeeding generations. For an SHP system
parasite pressure will only be a function of previous parasitism rates
and factors inherent in the host population making it more or less
susceptible to parasitism, such as phenotypic/genotypic composition (i.e.,
self regulation), all other factors being constant. Other factors do not
remain constant, however, and must be explored by a within generation
basis and formated as in Figure 43.

The within generation components of SHP systems may be viewed as
a series of interconnected processes leading to a complex, yet organized,
model of the determinants of parasite numbers. This approach is not
altogether different from the experimental component analysis of Holling
(1963) (i.e., each of the components of a particular process is
examined in terms of a set of variables making up the process). "It is
based on the belief that the characteristics of any specific example of
a complex process can be determined by the action and interaction of a
number of discreet components" (Holling 1966). Each of the discrete
components (effective and absolute density) are arranged with respect to
related components at lower levels (vertical plane), as well as analogous
or related processes in horizontal positions of the model (e.g., effective

parasite and host density).

Figure 43. The within generation components of SHP systems.

123

 

 

hégxu

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N 0...... an .55..
1 y - J - 1| . d -
fig”. a. a... e. _ ..s. a ...ss
an...“ 3%. tea.— .:. ,
a; t . h - Swat
H dz...
LF
0“" Paags
.a ufimmfiw a Jm>u4
a swan
5029.28 3.95.. .2... . 502828. 59. l
a
H sm>ms

 

 

 

 

124

Each component is dynamic and may change its characteristic value
or frequency distribution of values with time. For example, within any
given period of an insect generation, the distribution of fecundity
throughout the population may change in response to food availability,
temperature or the genetics of the population. Ultimately, changes in
any component can be traced back to fluctuations in the environment.
Environment is defined as that portion of the universe of concern which
we do not explicitly include in the model (Tummala et a1. 1975). Environ-
mental factors, though usually considered to be of the abiotic type,
(temperature, relative humidity, etc.) may also include biotic factors
such as,other parasites, predators and pathogens. The performance of
the system (adult parasite production) is then a function of the behavior
of each of the discrete components, their interaction, and environmental
variables driving the system. With these things in mind, the total number
of adult parasites produced is an integrator of all variables below it and
serves as the performance criterion for the system as a whole. Under—
standing of within generation dynamics is facilitated when viewing
relevant processes in this manner. However, between generation dynamics
are not considered. The remainder of this discussion will focus on those

intergeneration factors affecting parasite production.

Multivoltine Single Host-Parasite Systems

Let us now examine some multivoltine variants of SHP interactions.
Figure 44 graphically illustrates some common systems. Within each of
these models, daily parasite-host interactions are determined by the
same factors as discussed and presented in Figure 43. However, the

dynamics of the system as a whole presents some additional considerations.

125

 

cacao. =92—

 

 

9352:08—

”NE

T!!!

 

 

 

 

cacao->3:—

TlflE

 

 

 

 

cacao—=9:

Single host—parasite systems.

Figure 44.

126

Performance relative to a complete cycle must be viewed with reference
to an expanded time frame. Referring to Figure 41, a complete cycle is
formed when the numbers of parasites produced to attack Host 1(t) have
gone through Host 2(t + At) and Host N. This alteration in a time frame
is not significantly different than our view of single generation SHP
systems except that control of the first generation of the host is now
not self-regulating. It is dependent on previous generations and their
behavior, genetics, and the environment in which they exist.

Closely related to the univoltine SHP system (Fig. 44A) is the
multivoltine situation in which the host and parasite remain synchronized
on a per generation basis (Fig. 44B). In other words, the length of
either the parasite or host generation is not sufficiently large enough
to encompass more than a single generation of the other (see Figs. 44C,D).

If it is assumed that the characteristics of each population and the
environment were the same for all generations, then there is merely a
repetition of the single generation case. However, properties of the
system do not remain constant. Changes in environmental factors, such
as temperature, can significantly alter parasite production. An example

of this is demonstrated with Aphytis maculicornis (Masi), a parasite of

 

the olive scale in California. During the spring Aphytis is highly
successful in attacking the scale; however, the summer generation is
greatly inhibited by high temperatures and low precipitation (Huffaker

and Kennett 1966). An extreme example of this occurs in many agricultural
situations. Parasite populations often experience varying levels and types
of pesticidal pressure depending on the time of year and geographic

location (Huffaker et a1. 1962, Rabb 1969, DeBach and Rose 1977).

127

Changes may also occur in the developmental characteristics and
survival patterns of the host. This may be due to either a change in
species of host plant or in quality of the host plant and result in
asynchrony of both populations and increased mortality (Feeney 1976,
Morris 1967).

Another variant of the SHP system is shown in Figure 44C where
the parasite's generation time is at least twice as long as the hosts'.
The same sort of phenomena that occurred in B may also occur here. Of
greater significance, however, is the period of time in which hosts.
are either at low levels or lacking (shaded area). Parasites attacking
during this time must either be well equipped to locate hosts at low
levels or perish without being able to deposit progeny. The length of
this period is, therefore, a function of three factors: the parasite's
ability to locate prey at low levels, the developmental rates of the
host and parasite, and environmental factors such as temperature governing
the developmental velocity of each of the system's components. In some
situations, such as with aphids, the rate of reproduction is so rapid
that there may be 3 or 4 host generations for every 1 parasite generation
acting essentially as a single generation host (Passlow and Roubicek
1967, Kralifa and Sharif El-Din 1965). Thus, the between generation
host lag becomes nonexistent.'

Another feature of the multivoltine host system relates to the
selective pressure applied to different portions of the host population.
In Figure 440 (arrows) hosts which emerge early in the first generation
receive comparatively less parasite pressure than late emerging individuals.
On the other hand, late emerging individuals in the second generation

see comparatively less parasitism. The evolutionary and phenological

128

aspects of this concept are far reaching, Possibly moving the system to
2 parasite generations through decreased host availability during the
developmental lag period (Fig. 44C, shaded area).

Similar to the multivoltine host example is the multivoltine parasite
(Fig. 44D). Here again a developmental lag occurs, in this case, with
the parasite component. During this period parasite pressure is at its
lowest. Potential hosts occurring in the lag will have a high probability
of escaping parasitism. Selectively, the extremes of the host incidence
curve will be forced to adapt to the majority of the parasite attack by
shifting in phenotypic (genotypic) composition or to accept this differen-
tial mortality (Morris 1976). In an applied situation, the developmental
lag can be utilized in the application of management strategies which are
antagonistic to adult parasites. The cereal leaf beetle system is an
example where pesticides should not be applied outside of the lag period
or "biological window" (Haynes et al. 1974).

There are other types of SHP systems, however, the ones described
above represent a readily identifiable family of interactions. Variations

from this basic framework may produce more simple or complex systems.

Multiple Host-Parasite Systems

Multiple host-parasite systems (MHP) are defined as interactions
involving one or more generations Of a single parasite species and more
than one host species (i.e., polyphagous parasites). They are further
defined by dichotomizing them into two classes: those with synchronous
host availability (MHPS) and those with asynchronous host availability

(MHPA).

129

A synchronous system is pictured in Figure 45A and is composed of
a parasite (P) with l to N host species (Hl-HN). It is synchronous
because each of the hosts are occurring, to some extent, at the same
point in time with the parasite attacking one or more during that time
period.

If the objective of study is to determine total parasite production
for such a system,then the problem reduces to that of a SHP system with
an extended host generation. However, if objectives involve cause and
effect and the differnetial impact on the various host populations, then an
SHP systsn concept will not be sufficient. The first consideration is
to determine the attack rates on any one of the HN species while in the
presence of the other HN-l species. If it is assumed that the parasite
has no preference for one host over another, then the proportion of H1,

taken over all other hosts will be:

N H
1 l

N 3 NF""" [331*
X N -N1 2 H -H1

i=1 1 1-1 1

where:
Ni 8 the number of individuals attacked in the ith species.
If there is a preference for H1 over the other species then H1 must be

more heavily weighted, hence:

N1 CH1
N N - [34]
2 N -N1 2 u -H1
1-1 1 i=1 i

 

*Equation 33 and the following discussion are modified from the work done
by Murdoch and Oaten (1975).

130

 

 

 

TINE

TTHE

 

 

a4¢=a~>_oz~ .oz

 

 

 

o4¢:o_>~oz— .08

TIN!

Multiple host-parasite systems.

Figure 45.

131

For H1 preference, the coefficient c will be a constant greater than 1. If
H1 is preferred less than the other species, c will be less than 1.
Assume now that c is not a constant and that it is an increasing

-H1. In other words, the number of H1 taken becomes

N
f notion of H IE H
u 11-1 1

proportionately greater as the numbers of H1 increase relative to the
other hosts. As the frequency of H1 changes relative to the other host
species, the parasite begins to spend a disproportionate amount of time
attacking H1. This concept is termed "switching" (Murdoch 1969). There-

fore, P1, the proportion of the total hosts attacked, is:

CH1

P1. N o
CH141TZ HN-Hl]

i-l

 

[35]

Letting the proportion of H1 in the total host complex be F1, then:

CFI

a l-F1+CF1 [36]

P1

N
If c is assumed to be a linearly increasing function of Hl/iZIHi-Hl

then:

2
cF1

P1 ' (l-F1)2+cF12

 

[37]

The resulting family of curves appears as in Figure 46 with curves A and
B demonstrating no preference at unity (i.e., 50:50) and C and D demon-
strating preference for hosts other than H1. In curve C, for example,
F1 - .5 and P1 - .2, thus, even though H1 makes up 50% of the host
complex, the attack rate on H1 is less than unity.

Throughout the above discussion the models have only included what
might be called frequency dgpendent responses. In other words, any
change in attack rate N as a function of H or the total potential host

1 i
N
population (ilei) is ignored. Murdoch and Oaten (1975) have approached

132

0.50 0.75 1.00

P1

0.25

 

 

 

I T I
0-00 0.25 0.50 0-75 1-00

Figure 46. Proportion that H1 forms of P1 versus the proportion it
forms of total hosts (prey), F1 (taken from Murdoch and
Oaten 1975).

133

this problem using two methods, the first of which will be discussed.
Using this method, an existing attack model was modified to include N
host species and preference coefficients. The Holling "disc equation"
(1959) is used as an example where the gross attack rate per parasite

for a single host is:

N ..e$§_. [38]

c ' l+aThH

where:
H - the host density,
T - the total time available for attack,
a - the attack rate as defined by Holling (1965, 1966), and
T . the handling time, which for parasites is a function
of the time spent pursuing and attacking or ovi-
positing (larvipositing) in or on the host.
Modifying the above model for a MHPS system for i hosts we obtain:

a TH
i i
NG1 N [39]

1+1£1a1ThiHi

(Murdoch 1973). This model holds if host species are attacked independent
of one another. Incorporating frequency dependence into the model

produces:

N - ciF'H'T [401
G1 N

1+1§1“1F1H1

 

Here, switching is incorporated into equation 39 by assuming ai to
linearly increase with the proportion of the total host population occupied

by the ith species or:

134

 

 

31 ' “31 [41'
where:
F1 = Hi [42]
N
1:1Hi
and “i is a constant for the ith species. The model becomes:
N .. «iFiHiT . [43]
G1 N
1+151¢1F1ThiH1

The models discussed above have been developed primarily for use in
predator-prey systems. However, the concepts are largely applicable to
parasite-host systems. Empirically little is known about the dynamics of
switching in MPHS interactions. Studies have centered on laboratory
examination of host preference and density relations, while frequency
dependency has not been dealt with. On a one parasite-one host basis,
factors which are of importance in the host selection process are size,
color, movement and a number of other physical factors (Vinson 1976).

A conceptual model of how these factors interact with parasite behavior
and environment is presented in Vinson (1975). Aside from the confounding
effects of frequency dependence and possible host switching, one-to-one
models of host selection such as Vinson's are, however, probably valid.

Another factor which separates MHPS from SHP systems in this context
is preimaginal conditioning. This relates to preference for one species
over another based on the host that a parasite was reared on. Preference
is defined in terms of the actual selection process, changes in fecundity,
production of female progeny, and searching behavior (Vinson 1975, Legner

and Thompson 1977, Taylor and Stern 1971, Marston and Ertle 1973).

135

In MHPS systems host preference is not only due to physical character-
istics of each host and its microenvironment along with frequency depend-
ence, but is also due to conditioning of the previous generation. Attack
of a particular host species may, therefore, be mediated by preimaginal
conditioning and either strengthen or weaken preferences with each
succeeding generation. These factors along with frequency and density
dependence play a unique role in MHPS systems not found in single host-
parasite systems.

The second class of ME? interactions is the asynchronous system
(MHPA). An asynchronous system is pictured in Figure 45B and is made up
of a multivoltine parasite with different host species separated in time.

The within generation dynamics of an MHPA system does not differ
from its single host counterpart (Fig. 44B) in that there is still only
a one-to-one interaction of populations. However, the between generation
dynamics are different due to spatial, temporal, and behavioral character-
istics of the different hosts which affect parasite production. The
second generation host and parasite are not only affected by conditions
at that point in time but are "preconditioned" In! system and environ-
mental parameters in the first generation.

An interesting variant of the MHPA system is a combination of
synchronous and asynchronous host availability (Fig. 45C). Here, aspects
of preimaginal conditioning may play an important role in the amount of
parasite pressure applied to a particular host. If preimaginal condition-
ing does occur, and there are at least 2 generations of one host,
preference in the second generation will be a function of frequency,

density, and parental host. However, parasite production in the first

136

generation will only be a function of density, and of course, the full
range of factors covered in SHP systems.

At any point in time in a MHP system, parasitism is dependent on the
one-to-one interactions of parasite and host. However, when viewed in
a larger time frame, changes in the components of the system significantly
alter the parasite's performance and the characteristics of the inter-
actions in succeeding generations. Enlargement of the time frame allows
us to view these interactions relative to one another and arrive at
causal mechanisms for parasite production in general. Factors pertaining
to host preference such as frequency dependence and preimaginal condi-
tioning may actually produce changes in system composition. These changes
can occur in an evolutionary context altering the range of hosts available
for attack through permanent shifts in preference. In the short-run, non—
permanent shifts in attack may be effected through frequency and condi-
tioning.

In terms of managing MHP systems the above factors will determine
the success or failure of the project. The relationship between one host
(pest) and its parasite will be strengthened if frequency dependence
and conditioning prove favorable. However, periods of low pest numbers
can force parasites to preferentially attack nontarget hosts. Therefore,
long-term relationships for MHP systems must be conceptualized and managed

within this context.

137

SUMMARY AND CONCLUSIONS

Successful use of parasites in IPM programs requires that parasites
must be researched as the object of control. In the past they have been
viewed as uncontrollable management options with little appreciation for
the complexity inherent in their ecology. This outlook has lead to a
pesticidal approach to application and evaluation of parasites (i.e.,
controlling the amount being applied at application time and evaluating
their performance in terms of percent kill). Therefore, resources in this
classical context have been channeled toward those factors controlling
pest numbers rather than parasite numbers.

Analysis of the ETC-FWW system showed that an expanded system con-
ceptualization is a necessity for understanding the determinants of
parasitism._ Thus, the basic one-to-one parasite-host interaction is
important, but host-host interactions may ultimately explain the majority
of the variability in multiple host systems. Host-host interactions
include those processes mediated through common parasites and through
common host plants.

In addition to increasing the required number of system components,
a reductionist approach is needed to understand their interactions. In
the model developed in this study, results of sensitivity analyses of the
varying phenotypic composition of FWW populations exemplified this need.
In this case, parasite pressure applied to ETC in the subsequent genera-
tion was significantly affected by composition. In addition, this
"micro" view suggests that attempting to characterize pest populations
as a single number is not sufficient to evaluate a parasite's effective-

ness on one or more hosts.

138

Fromznlanalytical perspective the increase in system complexity
(within and between components) complicates the analysis of parasite
systems. However, from a management perspective, many new control options
are generated. For example, within a component explicitly harvesting
certain portions of a pest population provides managers with the ability
to alter a population's susceptibility to parasitism. Thus, the prob-
ability of being parasitized becomes a controlled management option.
Techniques involving population interaction are also possible with mani-
pulations made on first generation hosts and parasites aimed at impacting
second generation hosts.

Once the analysis of the ETC-FWW system was completed, a framework
was produced to conceptualize various types of parasite systems.
Basically, there were two major classes: single host and multiple host
systems (SHP and MHP, respectively). These two types were further
dichotomized into multiple generation single host systems, and synchronous
(MHPS) and asynchronous multiple host systems (MHPA). For SHP systems,
results of the analysis provided a model in which to view the various
processes determining parasite production. Many of the factors
identified in the model have been explored in field situations and
deemed viable management techniques for one-to-one interactions. In
terms of MHP systems, factors such as frequency dependence and pre- ’
imaginal and imaginal conditioning will determine the success or failure
of a program. Changes can occur in a MHP system altering the range of
hosts available for attack in the long-run. In the short-run, non-
permanent shifts in attack may be effected through frequency and

conditioning.

139

Development of generalized system concepts for the management
of parasites has provided a framework to examine specific techniques
for pest management systems. Use of this framework insures that studies
will be goal-oriented and resources directed through appropriate channels.
By using a holistic philosophy along with the altered concept of para-
sites as the object of control, it is anticipated that parasites will
become controllable variables and viable options in low energy IPM

programs.

LIST OF REFERENCES

Abkin, M. H. and T. J. Manetsch. 1972. A development planning oriented
model of the agricultural economy of Southern Nigeria. IEEE Trans.
Syst., Man and Cybern. 2:422-486.

Baird, A. B. 1917. An historical account of the forest tent cater-
pillar and of the fall webworm in North America. 47th Ann. Rep.
Ent. Soc. Ont. 1916:73-87.

Barney, R. J., D. P. Bartell, and W. G. Ruesink. 1977. Influences of
host density, temperature, and parasite age on the reproductive
potential of Bathyplectes curculionis (Hymenoptera: Ichneumonidae),
an endoparasite of the alfalfa weevil (Coleoptera: Curculionidae).
Great Lakes Entomol. 10:191-198.

 

Baskerville, G. L. and P. Emin. 1969. Rapid estimation of head accumu-
lation from maximum and minimum temperatures. Ecology 50:514-517.

Beirne, B. P. 1975. Biological control attempts by introductions
against pest insects in the field in Canada. Can. Ent. 107:225-236.

Casagrande, R. A. 1971. An approach to alfalfa weevil management in
Michigan. M.S. Thesis, Michigan State University. 60 pp.

Churchman, C. W. 1968. The Systems Approach. Dell Pub., NY. 243 pp.

Clark, E. C. 1958. Ecology of the polyhedroses of tent caterpillars
(Malacosoma spp.). Ecology 39:132-139.

Danks, H. V. 1979. Canada and its insect fauna. Mem. Ent. Soc. Can.
108:1-573.

DeBach, P. and M. Rose. 1977. Environmental upsets caused by chemical
eradication. Calif. Agric. 31:8-10.

Dobzhansky, T. 1970. Genetics of the Evolutionary Process. Columbia
Univ. Press, NY. 505 pp.

Dunstan, Au G. 1921. Some notes on the habits of Campoplex pilosulus,
a primary parasite of the fall webworm. Proc. N. S. Ent. Soc.
1920:81-88.

 

Feeney, P. P. 1976. Plant apparency and chemical defense. Rec. Adv.
Phytochem. 10:1-40.

Forrester, J. W. 1961. Industrial Dynamics. MIT Press.

140

141

Fulton, W. C. 1975. Aster yellow disease: Systems approach to its
management. Presented at the 13th Ann. Mtg. of the North Central
Branch of the Ent. Soc. Amer., E. Lansing, MI. March, 1975.

Fulton, W.C. 1978. Development of a model for on-line control of the
cereal leaf beetle (Oulema melanopus (L.)). Ph.D Dissertation,
Michigan State University. 130 pp.

 

Fusco, R. A., L. D. Rhoads, and M. Blumenthal. 1978. Compsilura
concinnata: Effect of temperature on laboratory propagation.
Env. Ent. 7:15-18.

 

Griffiths, K. J. and C. S. Holling. 1969. A competition submodel for
parasites and predators. Can. Ent. 101:785-818.

Hassell, M. P. and R. M. May. 1974. Aggregation in predators and
insect parasites and its effect on stability. J. Anim. Ecol. 43:
567-594.

Hattori, I. and Y. Ito. 1973. Status of black-headed and red-headed
Hyphantria ,cunea (Drury). 11. External characteristics of the
two types and their hybrids. App. Ent. 2001. 8:172-182.

Haukioja, E. and P. Niemela. 1979. Birch leaves as a resource for
herbivores: Seasonal occurrence of increased resistance in foliage
after mechanical damage of adjacent leaves. Oecologia 39:151-159.

Haynes, D. L., S. H. Cage, and W. Fulton. 1974. Management of the
cereal leaf beetle pest ecosystem. Quaes. Ent. 10:165-176.

Holling, C. S. 1959. The components of predation as revealed by a
study of small mammal predation of the European pine sawfly. Can.
Ent. 91:293-320.

Holling, C. S. 1963. An experimental component analysis of population
processes. Mem. Ent. Soc. Can. 32:22-32.

Holling, C. S. 1966. The functional response of invertebrate predators
to prey density. Mem. Ent. Soc. Can. 48:1-86.

Huffaker, C. B. and C. E. Kennett. 1966.‘ Studies of two parasites of
olive scale Parlatoria oleae (Colvee). IV: Biological control of
Parlatoria oleae (Colvee) through the compensatory action of two
introduced parasites. Hilgardia 37:283-335.

 

Huffaker, C. B., C. E. Kennett, and G. L. Finney. 1962. Biological
control of olive scale, Parlatoria oleae (Colvee), in California
by imported Aphytis maculicornis (Masi) (Hymenoptera: Aphelinidae).
Hilgardia 32:541-636.

Ito, Y. and K. Miyashita. 1968. Biology of Hyphantria cunea Drury in
Japan. V. Preliminary life tables and mortality data in urban
areas. Res. Popul. Ecol. Kyoto Univ. 10:177-209.

142

Ito, Y. and L. 0. Warren. 1973. Status of black-headed and red-headed
Hyphantria cunea (Drury). I. Biology of two types and results of
crossing experiment. App. Ent. Zool. 8:157-171.

 

Iwao, S. and W. G. Wellington. l970a. The western tent caterpillar:
Qualitative differences and the action of natural enemies. Res.
Popul. Ecol. 12:81-99.

Khalifa, A. and N. Sharaf El-Din. 1965. Biological and ecological
study of Aphis gossypii. Soc. Ent. Egypt Bull. 48:131-153.

Kovacevié, Z. 1954. Utjeacaj hrane na bioticki poteneijal dudovca
Hyphanteia cunea Drury. Biljne proizvodnje 2:65-78.

Langston, R. L. 1957. A synopsis of hymenopterous parasites of Malacosoma
in California (Lepidoptera: Lasiocampidae). Univ. Calif. Publ.
Ent. 14:1-50.

Legner, E. F. and S. N. Thompson. 1977. Effects of the parental host
on host selection, reproductive potential, survival and frcundity
of the egg-larval parasitoid, Chelonus sp. near curvimaculatus,
reared on Pectinophora ggssypiella and Phthorimaea gpercullela.
Entomophaga 22:75-84.

Leius, K. 1967. Influence of wild flowers on parasitism of tent cater-
pillar and codling moth. Can. Ent. 99:444-446.

Lyman, H. H. 1902. The North American fall webworm. 32nd Ann. Rept.
Ent. Soc. Ontario 1901:57-62.

Manetsch, T. J. 1976. Time-varying distributed delays and their uses
in aggregative models of large systems. IEEE Trans. Syst. Man and
Cybern. 6:547-553.

Mansingh, A. 1974. Studies on insect dormancy. II. Relationship of
cold-hardiness to diapause and quiescence in the eastern tent
caterpillar, Malacosoma americanum (Fab.). Can. J. Zool. 52:
629:636.

Marston, N. and L. R. Ertle. 1973. Host influence on the bionomics of
Trichogramma minutum. Ann. Ent. Soc. Amer. 66:1155-1162.

 

Merritt, R. W. 1970. Melanistic variation in Compsobata mima (Diptera:
Micropezidae). J. Kan. Ent. Soc. 43:451-455.

Miller, D. J. 1977. The bionomics of Diaparsis N.S.P. (Hymenoptera:
Ichneumonidae), a larval parasitoid of the cereal leaf beetle,
Oulema melanopus (L.) (Coleoptera: Chrysomelidae). Mich. State
Univ. P.M. Tech. Rep. 12:1-165.

 

143

Munroe, E. G. 1971. Status and potential of biological control in
Canada. Biol. Cont. Prog. Tech. Comm. 4:213-255.

Morris, R. F. 1963. The effect of predator age and prey defense of
the functional response of Podisus maculiventris Say to the density
of Hyphantria cunea Drury. Can. Ent. 95:1009-1020.

 

Morris, R. F. 1963. Synonymy and color variation in the fall webworm,
Hyphantria cunea Drury (Lepidoptera: Arctiidae). Can. Ent. 95:
1217-1223.

Morris, R. F. 1967. Influence of parental food quality on the survival
of Hyphantria cunea. Can. Ent. 99:24-33.

Morris, R. F. 1971. Observed and simulated changes in genetic quality
in natural populations of Hyphantria cunea. Can. Ent. 103:893-906.

Morris, R. F. 1972. Fecundity and colony size in natural populations
of_Hyphantria cunea. Can. Ent. 104:399-409.

Morris, R. F. 1976. Influence of genetic changes and other variables
on the encapsulation of parasites by Hyphantria cunea. Can. Ent.
108:673-684.

 

Morris, R. F. and W. C. Fulton. l970a. Models for the development and
survival of Hyphantria cunea in relation to temperature and humidity.
Mem. Ent. Soc. Can. 70:1-60.

Morris, R. F. and W. C. Fulton. l970b. Heritability of diapause inten-
sity in Hyphantria cunea and correlated fitness responses. Can.
Ent. 102:927-938.

Muesebeck, C. F. W., K. V. Krombein, and H. K. Townes. 1951. Hymenop-
tera of America north of Mexico synoptic catalog. USDA Agr. Mono.
2:1-1420.

MHnster-Swendsen, M. and G. Nachman. 1978. Asynchrony in insect host-
parasite interaction and its effect on stability, studied by a
simulation model. J. Anim. Ecol. 47:159-171.

Murdock, W. W. 1969. Switching in general predators: Experiments on
predator specificity and stability of prey populations. Ecol.
Monog. 39:335-354.

Murdoch, W. W. 1973. The functional response of predators. J. Appl.
Ecol. 14:335-341.

Murdoch, W. W. and A. Oaten. 1975. Predation and population stability.
Adv. Ecol. Res. 9:1-131.

Nordin, G. L. 1974. Pathogens of Malacosoma americanum in Kentucky.
Kan. Ent. Soc. 47:249-253.

 

144

Nordin, G. L. 1975. Transovarial transmission of a Nosema sp. infect-
ing Malacosoma americanum. J. Invert. Pathol. 25:221-228.

 

Nordin, G. L. 1976. Influence of natural Nosema sp. infections on
field populations of Malacosoma americanum (Lepidoptera: Lusio-
campidae). J. Kan. Ent. Soc. 49:32-40.

Odum, E. P. 1959. Fundamentals of Ecology. W. B. Saunders Co. 546 pp.

Oliver, A. D. 1963. A behavioral study of two races of the fall webworm,
Hyphantria cunea (Lepidoptera: Arctiidae) in Louisiana. Ann. Ent.
Soc. Am. 57:192-194.

Passlow, T. and M. S. Roubicek. 1967. Life-history of the cucurbit
aphid (A, gossypii). Queensland J. Agr. Anim. Sci. 24:101-102.

Price, P. W. 1975. Insect Ecology. Wiley-Interscience, NY. 514 pp.

Puttler, B. 1961. Biology of Hyposoter exiguae (Hymenoptera: Ichneu-
monidae), a parasite of Lepidopterous larvae. Ann. Ent. Soc. Amer.
54:25-30.

Rabb, R. L. 1969. Environmental manipulation influencing populations
of tobacco hornworms. Proc. Tall Timbers Conf. Ecol. Anim. Control
by Habitat Mgt., No. 1., Tallahassee, FL (1969):175-l9l.

Salt, G. 1934. Experimental studies in insect parasitism. II. Super-
parasitism. Proc. R. Soc. Lond. B. 114:455-476.

Southwood, T. R. E. 1978. Ecological Methods. Halstead Press, NY.
524 pp.

Steel, R. G. D. and J. H. Torrie. 1960. Principles and Procedures of
Statistics. McGraw-Hill, NY. 481 pp.

Taylor, A. T. and V. M. Stern. 1971. Host-preference studies with the

egg parasite Trichogramma semifumatum (Hymenoptera: Trichogramma-
tidae). Ann. Ent. Soc. Amer. 64:1381-1390.

Thompson, W. A., P. J. Cameron, W. G. Wellington, and I. B. Vertinsky.
1976. Degrees of heterogeneity and the survival of an insect
population. Res. Popul. Ecol. 18:1-13.

Thompson, W. A., W. G. Wellington, I. B. Vertinsky, and E. M. Matsumura.
1977. Harvesting strategies, control styles and information levels:
A study of planned disturbances to a population. Res. Popul. Ecol.
18:160-176.

Thompson, W. A., I. B. Vertinsky, and W. G. Wellington. 1979. The
dynamics of outbreaks: Further simulation experiments with the
western tent caterpillar. Res. Popul. Ecol. 20:188-200.

14S

Tilman, D. 1978. Cherries, ants and tent caterpillars: Timing of
nectar production in relation to susceptibility of caterpillars
to ant predation. Ecology 59:686-692.

Tothill, J. D. 1922. The natural control of the fall webworm (Hyphan-
tria cunea Drury) in Canada, together with an account of its
several parasites. Bull. Can. Dept. Agric. (N.S.) 3. 107 pp.

 

Tummala, R. L., W. G. Ruesink, and D. L. Haynes. 1975. A discrete
component approach to the management of the cereal leaf beetle
ecosystem. Env. Ent. 4:175-186.

Turnbull, A. L. and D. A. Chant. 1961. The practice and theory of
biological control of insects in Canada. Can. J. Zool. 39:697-753.

Varadarajan, R. V. 1979. Applications of modern control theory to
.the management of pest ecosystems. Ph.D Dissertation, Michigan
State University. 167 pp.

Vinson, S. B. 1975. Biochemical coevolution between parasitoids and
their hosts. .39: Evolutionary Strategies of Parasitic Insects and
Mites, Price, P. W. (Ed.). Plenum Press, NY. 225 pp.

Vinson,,S. B. 1976. Host selection by insect parasitoids. Ann. Rev.
Ent. 21:109-133.

Warren, L. O. and M. Tadié. 1970. The fall webworm, Hyphantria cunea
(Drury). Agric. Exp. Sta., Univ. of Arkansas Bull. 759.

Wellington, W. G. 1960. Qualitative changes in natural populations
during changes in abundance. Can. J. 2001. 38:289-314.

Wellington, W. G. 1977. Returning the insect to insect ecology: Some
consequences for pest management. Env. Ent. 6:1-8.

Wellington, W. G., P. J. Cameron, W. A. Thompson, 1. B. Vertinsky, and
A. S. Landsberg. 1975. A stochastic model for assessing the
effects of external and internal heterogeneity on an insect popu-
lation. Res. Popul. Ecol. 17:1-28.

Witter, J. A. and H. M. Kulman. 1972. A review of the parasites and
predators of tent caterpillars (Malacosoma spp.) in North America.
Agr. Exp. Sta. Univ. of Minn. Tech. Bull. 289:1-48.

Zuska, J. and C. O. Berg. 1974. A revision of the South American genus
Tetanocheroides (Diptera: Scimyzidae), with notes on color varia-
tions correlated with mean temperatures. Trans. R. Ent. Soc. Lond.
125:329-362.

 

APPENDIX A

SIMULATION MODEL OF THE ETC-FWW SYSTEM

146

PROGRAM WORM (INPUT=65,0UTPUT=65,TAPE1=65,TAPE2=65,
+TAPE3=65,TAPE"=65,TAPES:65,TAPE6=65,TAPE7=65,TAPE8=65,TAPE9=65,
+TAPE10=65,TAPE11=65,TAPE12=65)

COMMON /DEBUG/ IFPRINT(10)

COMMON /FOLK/ ECUMDD,CUMDD,PCUMDD,ETHOST,THOST,PARA,RNORM(8,10)

COMMON /FOLK/ RKPTOT(10).RZPAR(8,10),R2KPTOT(10),R3ENC(8,10)

COMMON /FOLK/ R3KPTOT(10),MINSTAR,TOWWEB,PARAP,PPCTKP(10)

COMMON /FOLK/ EPCTKP(10),UNPARIN(8),PARIN(8),ENCIN(8)

COMMON /FOLK/ UNPARKP( 10) ,PARKP(10) ,ENCKP( 10)

COMMON/FOLK/STRG(10,10,3),PARHOST,ENCHOST,PTHOST

COMMON/BILL/WTDMINE ,WTDMINF

REAL PRIN(10),PROUT(10),PR(10,10),ZSTRG(10),AM(3),Al(2)

REAL PAMT(6),XKP(6),ROUT(9,10,3),P(10),DEL(10),DELP(10)

REAL
REAL
REAL
REAL
REAL
REAL
REAL
REAL
REAL
REAL

DATA

HOST(8),R(10,8,10),SURV(8),ESURV(7),PARACO(3),ENCAP(10)
PHOST(8),EHOST(8),DE(10),DEP(10),R3(10,8,10),0WWEB(10)
RA(100),RB(200),EKP(10),PLRN(8),R2(10,8,10)
A(10),B(10),STORE(10,8,10),STORE1(10,1,l),TOTKP(10)
PROUTI(10),ROUTI(8),TOM(10),PESTRG(7),ETSTRG(7),EDEL(7)
EDELP(7),E(7),F(7),PUPALWT(10),EGGS(10),PAR(10),PPAR(10)
ETOUTI(7),PEOUTI(7),TEOUTI(7),PHOSTI(8),EHOSTI(8),THOSTI(8)
OWPARA(10),IPAWS(10),PENCP(10),EENCP(10)
NEWX,NEWY,INSTAR(6),MINSTAR

XINS(6),PROPF(6)

C INITIALIZATION AND PARAMETERIZATION OF SOME VARIABLES

PARACO(1),PARACO(2),PARACO(3)/.3,.7,l.O/,

+ENCAP(1),ENCAP(2),ENCAP(3).ENCAP(4),ENCAP(5),ENCAP(6),
+ENCAP(7),ENCAP(8),ENCAP(9).ENCAP(10)/O.,O.,O.,1.66,1.5,1.o,
+0.,0.,O.,0./,EKP(1),EKP(2),EKP(3),EKP(4),EKP(5),EKP(6), ,
+EKP(7),EKP(8),EKP(9),EKP(10)/.6,.52,.42,.37,.29,.22,.15,.08,
+0.,O./,K/10/,APLRN/O./
+.A(1),A(2).A(3).A(u).A(5).A(6).A(7).A(8).A(9),A(10)/-O.228631,
+-O.306721,-O.nua932,-O.559971,-O.559971,-O.uu7777,-O.21237u,
+-O.390179,
+.0.067198,-0.0239933/.B(1).B(2),B(3),B(u),B(5),B(6),B(7),B(8),B(9)
+,B(lO)/0.004482,0.006013,0.008802,0.010977,0.010977,0.008778
+,0.00n164,0.006669,0.001600,0.005713/

DATA
DATA
DATA
DATA
DATA
DATA
DATA

IFLAGl,IFLAGZ/0,0/

STRG/300'0./

AM,A1/5'0./

ROUTI/8’0./

TOM/10'O./

RA,RB/300'0./
E/‘022u3519‘0u72u311-05793709-07637691-07637699'0763769y

+—27.946828/

DATA
DATA

F/.004654,.009802,.012025,.015851,.015851,.015851,.550767/
PUPALWT/150.0,l53.0,155.0,155.0,154.0,153.0,150.0,144.0,

+139.0,128.0/

DATA OHPARA/10'0./

DATA IPAWS/lO'O./

DATA PROPF/.279,.208,.142,.114,.114,.143/

C

147

DATA BASEl,BASE2,BASE3/51.,42.,48.2/

C 5:00otnnnnnuessoosiosuggGIN SIMULATIONeuausaaiuaiisisnsosiaa
C VARAIBLES INITIALIZED

C

1500

60

70

80

666
667

C
C---=

REWIND 1

DT=0.1

Q1=DT'ZA.

HOUR: (22./7.)/24.

TIME=0.

N=l./DT+.5

DO 1500 1:1,10

READ',PRIN(I)
PRINT',"PRIN",I,"=”,PRIN(I)
PRIN(I)=PRIN(I)/DT

ZSTRG(I)=PRIN(I)

CONTINUE

PRINT*,"ENTER NDAYS TO RUN..."
READ',NDAYS

PRINT',"NDAYS=",NDAYS

WRITE 60

FORMAT(1X,'ENTER PARASITE DENSITY.)
READ 70,SUMPP

FORMAT(F10.2)

PRINT',SUMPP

SUMPP=SUMPPIDT

WRITE 80

FORMAT(1X,'ENTER EASTERN TENT CAT. DENSITY“)
READ 70,ETCIN

PRINT',ETCIN

ETCIN=ETCINlDT

FORMAT (1X,'DEBUG TABLE, PRINT STEP (212)’)
READ 667,I,J

FORMAT (212)

IF (I .LT. 1 .OR. I .GT. 9) GO TO 2
IFPRINT(I) = J

GO T0 1

CONTINUE

PRINT‘,"ENTER HARVEST INTERVAL AND EFFICACY (SB,SE,EFF)..."
READ.,SPRAYB,SPRAYE,EFFIC
PRINT.,SPRAYB,SPRAYE,EFFIC

 

C

C CALCULATION OF THE NUMBER OF EGGS LAID/FEMALE FOR EACH KP CLASS

111
C

D0 111 I=l,10
EGGS(I)=3.8'PUPALWT(I)-134.0
CONTINUE

SET VALUES FOR YEARS

PAIN=O.

 

1900
1800

1720
1710
1700

123

148

TWPARA:O.

DOUTI=O.

PAOUTI=O.

CUMDD=O.

PCUMDD=O.

ECUMDD=0.0

PLRPzO.

DO 1700 I=l,lO
PROUTI(I)=0.0
OWWEB(I)=O.
PROUTI(I)=O.
PAR(I)=O.
PENCP(I)=O.
EENCPCI)=O.
PPAR(I)=O.

DO 1800 J=1,1O
PR(I,J)=O.

DO 1900 JJ=1,8
R(I,JJ,J)=0.
R2(I,JJ,J)=0.
R3(I,JJ,J)=0.
STORE(I,JJ,J):O.
IF(I.GT.9)GO TO 1900
IF(JJ.GT.3)GO TO 1900
ROUT(I,J,JJ)=0.0
CONTINUE

CONTINUE
IF(I.GT.3)GO TO 1700
PHOSTI(I)=O.
THOSTI(I)=O.
EHOSTI(I)=O.
RB(I)=0.

PLRN(I)=O.
IF(I.GT.5)GO TO 1700
XKP(I):O.
IF(I.GT.N)GO TO 1700
RA(I)=O.

DO 1710 11:1,8

DO 1720 J=l,lO
RNORM(II,J)=O.
RZPAR(II,J):O.
R3ENC(II,J)=O.
CONTINUE

CONTINUE

CONTINUE

DO 123 I=l,10
RKPTOT(I)=0.0
R2KPTOT(I)=0.0
R3KPTOT(I)=0.0
CONTINUE

149

C
C OINEOCIQDDOQGNIONNIIOTHE 1uOO LOOP IS THE DAILY LOOPODOIGIINOON
C
DO 1uOO KJ=1,NDAYS
HTIME=O.
C READ FROM TAPEl--THE TEMPERATURE FILE
READ(1,50)IDAY,TMIN,TMAX
50 FORMAT(13X,I3,32X,2(F8.2,2X))
C
c CALCULATE DEGREE-DAYS
CALL DEGDAY (TMAX,TMIN,BASEl,FRFDD)
CALL DEGDAY (TMAX,TMIN,BASE2,FRPDD)
CALL DEGDAY(TMAX,TMIN,BASE3,FREDD)
HRANO=(TMAx-TMIN)/2.
TMEAN=(TMAX+TMIN)/2.
C
C 0!!!{CIIOQOQOIOIONIDQOTHE 99 LOOP IS THE DT Loopfiiillfliiiiiil
C
DO 99 MM=1,N
C
C ACCUMULATE DEGREE-DAYS AND CALCULATE INSTANTANEOUS TEMPS.
CUMDD:CUMDD+FRFDD*DT
PCUMDD:PCUMDD+FRPDD*DT
ECUMDD=ECUMDD+FREDD*DT
RTIME=HTIME+Q1
THETA=(HTIME-9.)'HOUR
TEMP=TMEAN+HRANG'SIN(THETA)
TIME=TIME+DT

c---_

 

CALL DELAY (TEMP,DEL,EDEL,A,B,E,F)
IF(KJ.GT.1.0R.MM.GT.1)GO TO 6
DO 2100 I=l,10
DELP(I)=DEL(I)
IF(I.GT.7) GO TO 2100
EDELP(I):EDEL(I)
2100 CONTINUE
FLAGzl.

 

C..-
C

C CALCULATE SURVIVALS FOR ETC (ESURV) AND FUN (SURV)

6 CALL SURVIV(CUMDD,SPRAYB,SPRAYE,EFFIC,SURV,ESURV)

C

C——— A —===-
C

C PARASITE PUPAL AND ADULT DELAYS

C

 

CALL DELLVF(PAIN,PAOUT,PAR,PARA,DEL(9).DELP(9),DT,K,STORE1(1,
+l,1))
CALL DELLVF(SUMPP,PAIN,PPAR,PARAP,DEL(10),DELP(10),DT,K,STORE

 

150

+1(1,1,1))

 

 

 

SUMPP:0.
PAOUTI=PAOUTI+(PAOUT'DT)
C
c nu... _ - _ =
C
C CALLING ETC SUBROUTINE TO CALCULATE DELAYS AND NEW SUMPP
C
CALL EASTC(ETCIN,EDEL,EDELP,DT,ESURV,PARA,PARACO,SUMPP,
+PESTRG,ETSTRG,FLAG,ETOUTI,PEOUTI,TEOUTI,TEMP,ETHOST,PTHOST)
C
c _ __ _
IF(ETOUTI(u).LT.2.) GO TO 968
C
C THE FOLLOWING IS USED IN THE COMPUTATION OF THE ADULT DELAYS
C

IF ( ETSTRG(N) .GE. 1.0 ) GO TO 968
IFLAG1 = IFLAG1 + 1
IF ( IFLAG1 .EQ. 1 ) IETC = IDAY
968 DOUT=0.0
SPUPAE=0.0
DO 100 J=1,10
BIGX:.01'PRIN(J)
IF(ZSTRG(J).LE.BIGX)GO TO 100
ABEL=(CUMDD-(280.+J'u0.))/12.
IF(ABEL.LT.-3.)GO TO 1713
IF(ABEL.GE.3.)GO TO 171”
XDENOM=0. .
IF(ABEL.GE.-.1.AND.ABEL.LT.0.0)ABEL=-.1
IF(ABEL.GE.0.0.AND.ABEL.LT.0.1)ABEL=.1
CALL ZSCORE(ABEL,BAK,EBAR)
GO TO 1715
1713 BAK=0.
GO TO 1715
171“ BAK=1.0
1715 DICK=BAK'PRIN(J)
PROUT(J)=DICK-TOM(J)
ZSTRG(J)=PRIN(J)-DICK
TOM(J)=DICK
SPUPAE=SPUPAE+ZSTRG(J)
DOUT=DOUT+PROUT(J)
PROUTI(J)=PROUTI(J)+(PROUT(J)'DT)
100 CONTINUE
DOUTI=DOUTI+(DOUT'DT)

 

 

 

C
C
C ==== — MATING MODEL---
C

AM(1)=.65'DOUT
A1(1)=.23*DOUT

151

A1(2)=.12'DOUT
CALL DCTDEL(A1(1),AM(2),RA,N)
CALL DCTDEL(A1(2),AM(3),RB,N*2)
TMATE=AM(1)+AM(2)+AM(3)
Do 300 1:1,3
IF(TMATE.EQ.0.)GO To 300
AM(I):AM(I)/TMATE
300 CONTINUE
C--- _
c
C PAMT(I) EQUALs THE DENSITY IN EACH MATING CLAss
PAMT(1)=(AM(1)'*2)*TMATE
PAMT(2)=(AM(2)**2)*TMATE
PAMT(3):(AM(3)'*2)*TMATE _
PAMT(H)=AM(1)*AM(2)*2*TMATE
PAMT(5)=AM(1)'AM(3)'2.'TMATE
PAMT(6)=AM(2)*AM(3)'2.*TMATE

 

C---
C
C XKP(I) EQUALS THE MEAN KP OF THE OFFSPRING BETWEEN EACH MATING
C CLASS
XKP(1)=EXP(2.H56+0.602“AL00(CUMDD))
Z=CUMDD-20.
IF(Z.LE.0.)GO T0 5
XKP(2)=EXP(2.HS6+O.602'ALOG(CUMDD-20.))
Z=CUMDD-HO.
IF(Z.LE.0.)GO T0 5
XKP(3)=EXP(2.H56+O.602'ALOG(CUMDD-HO.))
Z=(2.'CUMDD-20.)/2.
IF(Z.LE.O.)GO T0 5
XKP(H)=EXP(2.H56+O.602‘ALOG((CUMDD‘2.-20.)/2.))
Z=(2.'CUMDD-AO.)/2.
IF(Z.LE.0.)GO TO 5
XKP(5)=EXP(2.HS6+O.602'ALOG((2.’CUMDD-40.)/2.))
Z=(2.'CUMDD-60.)/2.
IF(Z.LE.0.)GO T0 5
XKP(6)=EXP(2.AS6+O.602*ALOG((2.“CUMDD-60.)/2.))

 

CLEARING INPUT ARRAY FOR PREOVIPOSITIONAL ADULT DELAY
DO ”00 I:1,10
ROUT(9,I,1)=O.
“00 CONTINUE
C _ _
C
C CALCULATION OF STANDARD DEV. AND PLACEMENT OE EGGS INTO NEW CLASS
DO “10 J=1,6
IF(XKP(J).LE.330) GO TO ”10
SD:-1328.82+229.665'ALOG(XKP(J))
NEWY=(300.-XKP(J))/SD

C
C
C
5

 

 

MA

A15
“20

1299

“30
“10
C
c---

152

CALL ZSCORE(NEWY,PY,EBAR)

DO “20 I=1,10

X=300.+(I'u0.)

NENX=((X-XKP(J))/SD)

IF(NENX.LT.-3.0)GO TO “15
IF(NEWX.GT.3.0)GO TO “1“
IF(NENX.GE.-0.1.AND.NEWX.LT.0.0) NEWX=-0.1
IF(NEWX.GE.0.0.AND.NEWX.LT.+0.1) NEWX=+0.1
CALL ZSCORE (NEWX,P(I),EBAR)

GO TO ”20

P(I)=1.0

GO TO ”20

P(I)=0.000001

CONTINUE
ROUT(9,1,1)=P(1)'PAMT(1)+ROUT(9,1,1)
IF(P(1).GE..98)GO T0 “10

DO ”30 I:2,10

IF(P(I).GE.0.99)GO TO 1299
ROUT(9,I,1)=(P(I)-P(I-1))'PAMT(J)+ROUT(9,I,1)
GO TO ”30
ROUT(9,I,1):(1.-P(I-1))'PAMT(J)+ROUT(9,I,1)
GO TO ”10

CONTINUE

CONTINUE

 

C

C DELAYS FOR PREOVIPOSITIONAL ADULTS TO PUPAE

C

699

695

C

D0 699 I = 1,8
HOST(I)=O.
PHOST(I)=O.
EHOST (I) =
UNPARIN (I)
PARIN (I)
ENCIN (I)
CONTINUE
D0 695 J = 1,10
UNPARKP (J) : 0,
PARK? (J)
ENCKP (J)
CONTINUE
DO 700 I=1,8

II=I+1

DO 800 J=1,10
IF(I.LT.”.OR.I.GT.6) GO TO Run

00

COMO

O.

C ATTRITION APPLIED TO NORMAL, PAR. AND ENCAP. FLOWS.

C

D0 900 JJ=1,10

153

ALOSS=PLRN(I)*R(JJ,I,J)
R(JJ,I,J)=R(JJ,I,J)-ALOSS
ELOSS=ALOSS*ENCAP(I)'EKP(J)
R3(JJ,I,J)=R3(JJ,I,J)+ELOSS
R2(JJ,I,J)=RZ(JJ,I,J)+ALOSS-ELOSS
RNORM(I,J)=RNORM(I,J)+R(JJ,I,J)
RZPAR(I,J)=R2PAR(I,J)+R2(JJ,I,J)
R3ENC(I,J)=R3ENC(I,J)+R3(JJ,I,J)
900 CONTINUE
AHA DE(I)=DEL(I)
DEP(I)=DELP(I)
C
C CALL DELAYS FOR ALL LIFE STAGES OF THE FUN
C
CALL DELLVF(ROUT(II,J,1),ROUT(I,J,1),R(1,I,J),STRG(I,J,1),
+DE(I),DEP(I),DT,K,STORE(1,I,J))
ROUT (I,J,1)= ROUT (I,J,1) ' SURV(I)
ROUTI(I)=ROUTI(I)+(ROUT(I,J,1)'DT)
IF(I.EQ.8) ROUT(I,J,1)=ROUT(I,J,1)'EGGS(J)/2.0
IF(I.GT.6)GO TO 800
DE(I)=DEL(I)
DEP(I):DELP(I)
CALL DELLVF(ROUT(II,J,2),ROUT(I,J,2),R2(1,I,J),STRG(I,J,2),
+DE(I),DEP(I),DT,K,STORE1(1,1,1))
ROUT(I,J,2)=ROUT(I,J,2)'SURV(I)
IF(I.LT.A)GO TO 800
DE(I)=DEL(I)
DEP(I)=DELP(I)
CALL DELLVF(ROUT(II,J,3),ROUT(I,J,3),R3(1,I,J),STRG(I,J,3),
+DE(I),DEP(I),DT,K,STORE1(1,1,1))
ROUT(I,J,3)=ROUT(I,J,3)*SURV(I)
IF(I.NE.A)GO T0 800
ROUT(A,J,1)=ROUT(H,J,1)+ROUT(A,J,3)
EENCP(J)=EENCP(J)+ROUT(H,J,3)'DT
800 CONTINUE
C
C INTEGRATES RATES FOR PAR., ENCAP. AND TOTAL NEBWORMS.
C
PHOSTI(I)=PHOSTI(I)+PHOST(I)*DT
EROSTI(I)=EHOSTI(I)+EHOST(I)'DT
THOSTI(I)=EHOSTI(I)+PHOSTI(I)
700 CONTINUE
DO 2000 1:1,8
DELP(I)=DEP(I)
2000 CONTINUE

C

C SUMMING ACROSS KP CLASSES FOR EACH LARVAL INSTAR.
DO 33 I = 1,6
DO 3“ J:1,10

UNPARIN (I) = UNPARIN (I) + STRG (I,J,1)

154

PARIN (I) = PARIN (I) + STRG (I,J,2)

ENCIN (I) = ENCIN (I) + STRG (I,J,3)
3n CONTINUE

HOST (I) = HOST (I) + UNPARIN (I)

PHOST (I) = PHOST (I) + PARIN (I)

EHOST (I) = EHOST (I) + ENCIN (I)

33 CONTINUE
C SUMMING ACROSS LARVAL INSTARS FOR EACH KP CLASS.
D0 35 J = 1,10
DO 35 I = 1,6
UNPARKP (J) = UNPARKP (J) + STRG (I,J,1)
PARKP (J) PARKP (J) + STRG (I,J,2)
ENCKP (J) ENCKP (J) + STRG (I,J,3)
35 CONTINUE
C SUMMING AVAILABLE HOSTS FOR PARASITISM
THOST = HOST(H) + HOST(5) + HOST(6)
C SUMMING PARASITIZED HOSTS
PARHOST = PHOST(u) + PHOST(5) + PHOST(6)
C SUMMING ENCAPSULATED HOSTS
ENCHOST = EHOST(H) + EHOST(5) + EHOST(6)
C SET UP FLAG T0 COMPUTE IFWW TO BE USED IN CONNECTION WITH IDELAY
IF ( HOST(6). LE . 1.0 ) GO TO 38
IFLAGZ = IFLAG2 + 1
IF ( IFLAG2.EQ.1 ) IFWW = IDAY

 

C __
C CALLING ATTACK MODEL FOR THE FWW

38 CALL ATTACK(DT,THOST,PARA,PATT)

c----- ___
C REDUCES ATTACK RATE IN RESPONSE TO TEMPERATURES

C LESS THAN 65 AND GREATER THAN 90. ATTACK IS REDUCED

C BY A LINEAR FUNCTION WITH A SLOPE OF .1 AT TEMPERATURES

C BETWEEN 60 AND 65. ATTACK IS ALSO REDUCED BY A LINEAR

C FUNCTION AT TEMPERATURES BETWEEN 90 AND 110

C WITH A SLOPE OF .05.

 

 

C
IF(TEMP.LE.60.)PATT=0.0
IF(TEMP.GT.60..AND.TEMP.LE.65.)PATT=(TEMP-60.)'.2'PATT
IF(TEMP.GT.90..AND.TEMP.LE.110.)PATT=(1.-.05'(TEMP-90.))*PATT
IF(TEMP.GT.110.)PATT=0.

c--- _

C

DO 1000 I=1,3
PLRN(I+3)=(PARACO(I)'PATT)
1000 CONTINUE
C
C_____ _.
C
C SUMMING OVERWINTERING NUMBERS
TOWWEB=0.0
D0 1100 J=1,10

 

155

C

C OUTPUT NOT INCLUDED IN SUBROUTINE DEBUG
WRITE(11,137)IDAY,ECUMDD,(ETSTRG(JJ),JJ:1,7)

137 FORMAT(1X,I3,1X,8(E9.3,1X))

1&00 CONTINUE

 

 

C--- END DAILY LOOP (1uoo)--—
C
C OUTPUTS FOR INITIALIZATION OP MULTIPLE YEAR RUNS
C
PRINT‘,"ON FWW PUPAE PER KP"
PRINT 138,(0WWEB(JJ),JJ=1,10)
138 FORMAT(5X,10(E9.3,1X))
DO 10 JKP=1,1O
WRITE(12,139)JKP,OWWEB(JKP)
139 FORMAT(I3,10(E9.3))
1o CONTINUE
ENDFILE 12
DO 11 JKP:1,10
WRITE(12,139)JKP,OWPARA(JKP)
11 CONTINUE
ENDF'ILE 12

PRINT',"ETC PUPAE"
PRINT 138,ETOUTI(1)
C ETCEGG COMPUTED FROM 220 EGGS PER MASS AND SURVIVAL
C OF .A7 AND ."7 FOR PUPAL AND ADULT STAGES.
ETCEGG=ETOUTI(1)’H8.H'.5
PRINT‘,"ETC EGGS TO OVERWINTER"
PRINT 138,ETCEGG

PRINT’,"CAMPOPLEX FROM ETC"
PRINT 138,PEOUTI(1)
PRINT',"OW PARASITES PER KP”
PRINT 138,(0WPARA(JJ),JJ=1,10)
PRINTS,"TOTAL OW PARASITES"
PRINT 138,TWPARA

C

C COMPUTE DELAY = IFWW - IETC
IDELAY = IFWW — IETC
PRINT 39.IDELAY

39 FORMAT (5X, ” DELAY IN DAYS = ",1” )
END

 

 

156

OWWEB(J)=0WWEB(J)+ROUT(1,J,1)*DT
TOWWEB=TOWWEB+OWWEB(J)
OWPARA(J)=0WPARA(J)+ROUT(1,J,2)‘DT
TNPARA=TNPARA+ROUT(1,J,2)*DT
1100 CONTINUE
C ---- ----
c
C THIS PORTION OF THE PROGRAM DETERMINES MEAN LARVAL INSTAR
DO 79 1:1,6
INSTAR(I)=HOST(I)+PHOST(I)
IF(I.LE.3)GO TO 79
INSTAR(I)=INSTAR(I)+EHOST(I)
79 CONTINUE
XNUM=0.
WINSTAR=0.
TINSTAR=0.
XDENOM=0 .
NTDMINE=O.
MINSTAR=O.
DO 81 I=1,6
TINSTAR=TINSTAR+INSTAR(I)
HINSTAR=NINSTAR+INSTAR(I)*(7-1)
XINS(I)=INSTAR(I)
XI=I
XNUM=XNUM+(PROPF(I)'(7.-XI)'XINS(I))
XDENOM=XDENOM+(PROPF(I)'XINS(I))
81 CONTINUE
IF ( TINSTAR.EO.O.O) GO TO 99
IF(XDENOM.EQ.O.)GO TO 99
C COMPUTE MEAN LARVAL INSTAR
MINSTAR=NINSTAR/TINSTAR
WTDMINF=XNUMIXDENOM
99 CONTINUE
C

 

 

 

END OF DT LOOP (99)==- ———
C
C DETERMINE _ PARASITIZED AND ENCAPSULATED FOR EACH KP CLASS.
DO 36 J = 1,10
TOTKP (J) = UNPARKP (J) + PARKP (J) + ENCKP (J)
IF ( TOTKP (J).EQ.0.0 ) GO TO 37
PPCTKP(J) : PARKP (J) / TOTKP (J)
EPCTKP(J) : ENCKP (J) / TOTKP (J)
GO TO 36
37 PPCTKP(J):0.
EPCTKP(J)=O.
36 CONTINUE
C
C SUBROUTINE DEBUG GENERATES OUTPUT TABLES
DO 668 ITBL = 1,9
CALL DEBUG (ITBL, IDAY)
668 CONTINUE

 

157

SUBROUTINE EASTC (x,EDEL,EDELP,DT,ESURV,PARA,PARACO,
+SUMPP,PESTRG,ETSTRG,FLAG,ETOUTI,PEOUTI,TEOUTI,TEMP,ETHOST,PTHOST)
COMMON/BILL/HTDMINE,NTDMINE

DIMENSION ETOUT(8),ETCR(12,7),ETSTRG(7),EDEL(7),EDELP(7),
+ESTORE(12,7),ESURV(7),PEOUT(8),PETCR(12,7),PESTRG(7),
+EDEL1(7),EDELP1(7),PSTORE(12,7),PARACO(3),ETOUTI(7)
+,PEOUTI(7),TEOUTI(7)

DIMENSION EINSTAR(6),PROPE(6)

DATA PROPE/.3058,.1u60,.1172,.0897,.0897,.0897/

IF(FLAG.EQ.O.)GO TO 6

INITIALIZING DELAY VARIABLES FOR FIRST OF YEAR

000

DO U I:1,7
ETOUTI(I)=0.
PEOUTI(I)=0.
ETOUT(I)=0.
ETSTRG(I)=O.
PEOUT(I)=0.
PESTRG(I):0.

DO 5 J=1,12
ETCR(J,I)=0. m
ESTORE(J,I):0
PETCR(J,I)=0.
PSTORE(J,I)=0.
CONTINUE
CONTINUE
ELAG:0.
PEOUT(8)=0.

.zwn

BEGINNING OF ETC DELAYS

O‘OOOO

K=12

ETOUT(8)=X

DO 1 I=1,7

II=I+1

EDEL1(I)=EDEL(I)
EDELP1(I)=EDELP(I)

CALL DELLVF(ETOUT(II),ETOUT(I),ETCR(1,I),ETSTRG(I),EDEL1(I),
+EDELP1(I),DT,K,ESTORE(1,I))
ETOUT(I)=ETOUT(I)'ESURV(I)
ETOUTI(I)=ETOUTI(I)+ETOUT(I)'DT
IF(I.GT.6)GO TO 1

7
C
C DELAYS FOR PARASITIZED ETC
C
CALL DELLVF(PEOUT(II),PEOUT(I),PETCR(1,I),PESTRG(I),EDEL(I),

+EDELP(I),DT,K,PSTORE(1,I))
PEOUT(I)=PEOUT(I)*ESURV(I)

 

 

158

PEOUTI(I)=PEOUTI(I)+PEOUT(I)*DT
TEOUTI(I)=PEOUTI(I)+ETOUTI(I)

1 CONTINUE
EDELP(7)=EDELP1(7)

C SUMMING PARASITIZED HOSTS
PTHOST:PESTRG(A)+PESTRG(5)+PESTRG(6)
SUMMING HOSTS AVAILABLE FOR ATTACK

ETHOST=ETSTRG(A)+ETSTRG(5)+ETSTRG(6)
CALL ATTACK(DT,ETHOST,PARA,PATT)

 

C ___ _
C REDUCES ATTACK RATE IN RESPONSE TO TEMPERATURES
C LESS THAN 65 AND GREATER THAN 90. ATTACK IS REDUCED
C BY A LINEAR FUNCTION WITH A SLOPE OF .1 AT TEMPERATURES
C BETWEEN 60 AND 65. ATTACK IS ALSO REDUCED BY A LINEAR
C FUNCTION AT TEMPERATURES BETWEEN 90 AND 110
C WITH A SLOPE OF .05.
CO!!!QIIIIOOIO!!!COIOQQQGCICCQOQIGCQOOCQOCINICIfilfiifilllilflliliiiil
IF(TEMP.LE.60.)PATT=0.0 ‘
IF(TEMP.GT.60..AND.TEMP.LE.65.)PATT=(TEMP-60.)*.2'PATT
IF(TEMP.GT.90..AND.TEMP.LE.110.)PATT=(1.-.05'(TEMP-90.))'PATT
IF(TEMP.GT.110.)PATT:O.
cl!!!allallaiunsaa!iiiniilafliiiianalcloElsilnnacalsiiaaiaiiou
C
C APPLICATION OF PARASITE ATTACK
C
00 2 1:4,6
EPLRN=PARACO(I-3)'PATT
D0 3 J=1,12
ALOSS=EPLRN'ETCR(J,I)
ETCR(J,I)=ETCR(J,I)-ALOSS
PETCR(J,I)=PETCR(J,I)+ALOSS
3 CONTINUE
2 CONTINUE
SUMPP=PEOUT(1)
X=0.
DO 99 I=1,6
EINSTAR(I)=0.
99 CONTINUE
XNUME=0.
XDENOME=O.
WTDMINE30.
C CALCULATE WEIGHTED MEAN INSTAR FOR ETC.
DO 100 1:1,6
YI=I
EINSTAR(I):EINSTAR(I)+PESTRG(I)+ETSTRG(I)
XNUME:XNUME+(PROPE(I)'(7.-YI)'EINSTAR(I))
XDENOME:XDENOME+(PROPE(I)*EINSTAR(I))
100 CONTINUE

 

 

159

IF(XDENOME.EQ.O.)RETURN
WTDMINE=XNUME/XDENOME

 

 

RETURN
END
C
C _ _ _ _ ___-
C
SUBROUTINE ZSCORE(X,P,D)
AX=ABS(X)
T:1.0/(1.0+.2316H19'AX)
D=O.3989u23'EXP(-X'X/2.0)
P=1.0-D'T'((((1.33027NFT-1.821256)'T+1.781478)'T-0.3565638)*T+O.3
+193815) -
IF(X)1,2,2
1 P:1.0-P
2 RETURN
END
C—-—— ___ _
C

SUBROUTINE ATTACK(TAG,ANO,P,PATT)

C HOLLING DISC EQUATION-—N0. OF ATTACKS/PARASITE
A=(.0H6'ANO)/(1.+.000u6'ANO)

C CALCULATE NO. OF ATTACKS FOR PARASITE POPULATION
ANA=A'TAG'P
IF(ANA.LE..1)GO To 1
IF(ANO.LE.O.1)GO TO 1

c GRIFFITHS AND HOLLING COMPETITION SUEMODEL
ANHA=ANO'(1.0-(1.0+ANA/(ANO'50.))"(-50))
PATT=ANHA/ANO
RETURN

1 PATT=O.
RETURN
END

 

SUBROUTINE DELLVF(VIN,VOUT,R,STRG,DEL,DELP,DT,K,STORE)
DIMENSION R(K),STORE(K)
FK=FLOAT(K)
A=DTFFKIDEL
V=VIN
DELD:(DEL-DELP)/(DT'FK)
DELP=DEL
DO 1 1:1,K
DR:R(I)
R(I):DR+A'(V-DR.(1.+DELD))
V=DR

1 CONTINUE
VOUT=R(K)
STRG=0.
DO 2 I:1,K

 

160

STORE(I)=R(I)‘DEL/FK
STRG=STRG+STORE(I)

 

2 CONTINUE
RETURN
END

C

C---_

C

SUBROUTINE SURVIV(CDD,SB,SE,EFF,SURV,ESURV)

DIMENSION SURV(8),ESURV(7)
THE DATA FOR THIS SUBROUTINE WAS TAKEN FROM MORRIS(1967,5TH
PAPER) AND ITO AND MIYASHITA,1968.

SURVIVAL COEFICIENTS FOR FWW

SURV(1) IS APPLIED TO THE 6TH INSTAR OUTPUT BUT CONTAINS'
TWO COMBINED MORTALITY FACTORS, THAT OF THE 6TH INSTAR AND THE
OVERWINTERING PUPAE.

SURV(1):(.A5+RANF(X)'.1)'(.H+RANF(X)'.1)

C)C)C)C)CIC)C)C)

SURV(2)-SURV(6) ARE THE SURVIVAL COEFICIENTS FOR LARVAL INSTARS
5 THRU 1 RESPECTIVELY. SURV(7) IS THE EGG SURVIVAL COEF.
AND SURV(8) IS THE ADULT SURVIVAL COEFICIENT.
M......G'HGGGCMGMGMMMMMMMMC...MGEM.....GCEHMGMMMMGMMMGMMCMMM
SURV(2)=.NS+RANF(X)'.1
SURV(3)=.60+RANF(X)'.1
SURV(N)=.7+RANF(X)'.1
SURV(5)=.75+RANF(X)..1
SURV(6)=.55+RANF(X)..1
SURV(7)=.85+RANF(X)' .1
SURV(8)=.75+RANF(X)'.1
CM...GM......MMMMMMMMGGMMMMHGMMM'.....'......UMMGMMHMGGMMGHGM
C
C SURVIVAL COEFFICIENT FOR ETC
C DATA IS TAKEN FROM RAVLIN UNPUBLISHED DATA 1980.
C ESURV(1)-(7) ARE COEFFICIENTS FOR EGGS THROUGH INSTAR 6.
C
C5......‘MGMMMHMM...G'.‘..........'.....'.....GGMMMMCMMGMMGGM
ESURV(1)=.N7+RANF(X)..1
ESURV(2)=.N7+RANF(X)'.1
ESURV(3)=.‘19+RANF(X)' .1
ESURV(u)=.77+RANF(X)'.1
ESURV(5)=.90+RANF(X)' .1
ESURV(6)=.9N+RANF(X)..1
ESURV(7)=.8“+RANF(X)'.1
CM...MMG......‘MGGGGHM...‘GMMMMGGMMGM...‘MMMMHMMGMGGMHMGMHGHM
IF(CDD.LE.SB)GO TO 2
IF(CDD.GT.SE)GO TO 2
DO 1 IS=1,7
SURV(IS)=SURV(IS)'EFF

C)C)C)C)C)

 

161

 

1 CONTINUE
2 RETURN
END
C
C---== _ ___
C

SUBROUTINE DEGDAY(XMAX,XMIN,BASE,XHEAT)
DATA TPIE/6.283181/, HPIE/1.570795/
IF(XMAX.GT.BASE) GO TO 1
XHEAT:0.00001
RETURN
C
C IF MAXIMUM TEMP GREATER THAN BASE ENTER HERE
C
1 Z=XMAX-XMIN
XM=XMAX+XMIN
IF(XMIN.LT.BASE) GO TO 2
XHEAT=XM/2.-BASE
C
C ROUNDOFF- ODD UP--EVEN DOWN
C
IF(XHEAT.GT.0.)GO TO 3
XHEAT=.OOO1
RETURN

3

C

C IF MINIMUM TEMP LESS THAN BASE ENTER HERE
C

2 TBASEzBASEFZ.

A=ASIN((TBASE-XM)/Z)
XHEAT=(Z'COS(A)-(TBASE-XM)‘(HPIE-A))/TPIE

 

 

C
IF(XHEAT.GT.O)GO TO A
XHEAT=.00001
u RETURN
END
C
C--- _ _-
C
SUBROUTINE DCTDEL(VIN,VOUT,VINT,N)
DIMENSION VINT(N)
VOUT=VINT(1)
DO 1 I=Z,N
1 VINT(I-1)=VINT(I)
VINT(N)=VIN
RETURN

END

162

 

 

C _ = _ _ _=____ _ _____ _ __
C
SUBROUTINE DELAY (TEMP,DEL,EDEL,A,B,E,F)
DIMENSION DEL(10),A(10),B(10),
+EDEL(7),E(7),F(7)
C CALCULATE DEVELOPMENTAL DELAYS FOR WEBWORM.
DO 10 I=1,8
IF ( TEMP.LE.55.) DEL(I)=100.
IF (TEMP.GT.55 ) DEL(I)=1./(A(I)+B(I)'TEMP)
10 CONTINUE
C
C CALCULATE DEVELOPMENTAL DELAYS FOR THE PARASITE.
DO 20 K=9,10
IF (TEMP.LE.N5.) DEL(K)=700.
IF (TEMP.GT.A5.) DEL(K)=1./(A(K)+B(K)'TEMP )
20 CONTINUE

C DELAY FOR ETC EGGS

IF(TEMP.LE.52.)EDEL(7)=100.
IF(TEMP.GT.52.)EDEL(7)=TEMP/(E(7)+F(7)'TEMP)

C DELAYS FOR ETC L1-L6.

25

c---

DO 25 L=1,6

IF(TEMP.LE.52.)EDEL(L)=100.
IF(TEMP.GT.52.)EDEL(L)=1./(E(L)+F(L)'TEMP)
CONTINUE

RETURN

END

 

SUBROUTINE DEBUG (ITBL, IDAY)
PRINT STEP -IDAY- OF TABLE -ITBL-.

COMMON /DEBUG/ IFPRINT(10)

COMMON /FOLK/ ECUMDD,CUMDD,PCUMDD,ETHOST,THOST,PARA,RNORM(8,10)
COMMON /FOLK/ RKPTOT(10),R2PAR(8,10),R2KPTOT(10),R3ENC(8,10)
COMMON /FOLK/ R3KPTOT(10),MINSTAR,TOWWEB,PARAP,PPCTKP(10)
COMMON /FOLK/ EPCTKP(10),UNPARIN(8),PARIN(8),ENCIN(8)

COMMON /FOLK/ UNPARKP(10),PARKP(10),ENCKP(10)
COMMON/FOLK/STRG(10,10,3),PARHOST,ENCHOST,PTHOST
COMMON/BILL/WTDMINE,WTDMINF

REAL MINSTAR

DETERMINE WHETHER TO PRINT STEP -IDAY- OF TABLE -ITBL-.

WE DON-T PRINT THE TABLE AT ALL UNLESS SPECIFICLY REQUESTED,
IF IFPRINT .GT. 0

IF (IFPRINT(ITBL) .LE. 0) RETURN

IF EVERY STEP OF THE TABLE IS REQUESTED, WE DO PRINT.

 

 

A00
91

115

88

500
996

105

600
997

106

700
998

800
999

I63

WRITE(N,88)
RETURN
T A B L E u

WRITE(",91)IDAY,CUMDD,PCUMDD
FORMAT('0*,9X,'IDAY=',I3,2X,'CUMDD=',F9.3,NX,
+FPCUMDD=F F9.3,/,9X,'TABLE NF,11X,'FWW-ENCAPSULATED'
+F INDIVIDUALS',/,9X,57('-‘),/,9X,'KP',“X,'1F,9X,*2F,9X,
+'3',9X,'u',9x,.5',9X,'6.,5X,FTOTALF,/,9X,67(’-*))

DO 115 IK=1,10
WRITE(N,90)IK,STRG(6,IK,3),STRG(5,IK,3),STRG(N,IK,3)
+,STRG(3,IK,3),STRG(2,IK,3),STRG(1,IK,3),ENCKP(IK)
CONTINUE

WRITE(H,88)
WRITE(N,87)ENCIN(6),ENCIN(5),ENCIN(N),ENCIN(3)yENCIN(2)
+,ENCIN(1)

NRITE(u,88)

FORMAT(9X,67('-'))

RETURN

IF(IDAY.EQ.91)WRITE(5,996)

FORMAT(*1*,3X,*TABLE 5',/,1X,*IDAY',5X,*ECUMDD*,SX,
+*WTDMINE',HX,*CUMDD*,6X,'WTDMINF*,2x,'MINSTAR',/,1X,65('-*))
WRITE(5,105)IDAY,ECUMDD,WTDMINE,CUMDD,WTDMINF,MINSTAR
FORMAT(1x,I3,3x,5(E9.3,2x))

RETURN

IF(IDAY.EQ.91)WRITE(6,997) ,

FORMAT('1',1OX,'T A B L E 6*,/,1X,'IDAY CUMDD',
+6X,'PCUMDD',5X,*TOWWEB',5X,*PARAP*,/,1X,50('-'))
WRITE(6,106)IDAY,CUMDD,PCUMDD,TOWWEB,PARAP
FORMAT(1x,I3,ux,u(E9.3,2X))-

RETURN

IF(IDAY.EQ.91)WRITE(7,998)

FORMAT(*1!,u1x,!T A B L E 7',/,28X,10('PCTPARA/',2X),
+/,1X,'IDAY CUMDD',6X,'PCUMDD',7X,'KP 1',5X,*KP 2*,5x,
+£KP 3',5X,'KP u!,5x,'KP 5',5X,'KP 6',5X,*KP 7',5X,'KP 8*,
+5X,'KP 9',5x,'KP 10*,/,1X,128('-'))

WRITE(7,109)IDAY,CUMDD,PCUMDD,(PPCTKP(J),J=1,10)

RETURN

IF(IDAY.EQ.91)WRITE(8,999)

FORMAT('1*,A1X,'T A B L E 8',/,28X,10('PCTENCP/*,2X),
+/,1X,'IDAY CUMDD',6X,'PCUMDD',7X,'KP 1',5X,'KP 2',5X,
+‘KP 3',5X,'KP A',5X,'KP 5',5X,'KP 6',5X,*KP 7‘,5X,*KP 8',
+5X,'KP 9',5X,'KP 10',/,1X,128('-*))

10

100
83

85

200
86

90

113

87

300

89

11A

164

IF (IFPRINT(ITBL) .EQ. 1) GO TO 10

PRINT IF MUDULUS SHOWS CORRECT STEP NUMBER.

IF (MOD(IDAY,IFPRINT(ITBL)) .NE. 1) RETURN
BRANCH ON TABLE NUMBER TO CORRECT SET OF PRINTS.

GO TO (100,200,300,u00,500,600,700,800)ITBL
T A B L E 1
IF (IDAY .EQ. 91) WRITE (3,83)
FORMAT('1',7X,FTABLE 1',2NX,FTOTAL NUMBER OF HOSTS PER',
+i DAY',/,7X,105('-'),/,7X,*IDAY ECUMDD CUMDD!
+9 PCUMDD NORMETC PARETC NORMFWW PARFWW.
+' ENCFWW CAMPOPLEX',/,7X,105('-.))
WRITE(3,85)IDAY,ECUMDD,CUMDD,PCUMDD,ETHOST,PTHOST,THOST
+,PARHOST,ENCHOST,PARA
FORMAT(7X,I3,3X,9(E9.3,1x))
RETURN
T A B L E 2
WRITE (u,86) IDAY,CUMDD,PCUMDD
FORMAT('1',9X,*IDAY:*,13,2X,'CUMDD=*,F9.3,AX,
+FPCUMDD=F,F9.3,/,9X,'TABLE 2.,11X,'FWW-UNPARASITIZEDF
+9 INDIVIDUALS.,/,9x,67('-.),/,9X,'KP',ux,.1.,9x,'2.,9x,
+‘3',9X,'N',9X,*5',9X,'6',5X,.TOTAL',/,9X,67(.-'))
DO 113 IK=1,1O
WRITE(N,90)IK,STRG(6,IK,1),STRG(5,IK,1),STRG(N,IK,1),STRG(3,IK,1)
+,STRG(2,IK,1),STRG(1,IK,1),UNPARKP(IK)
FORMAT(9X,I2,1X,7(E9.3,1X))
CONTINUE
WRITE(H,88)
WRITE(N,87)UNPARIN(6),UNPARIN(5),UNPARIN(N),UNPARIN(3)
+,UNPARIN(2),UNPARIN(1)
FORMAT(12X,6(E9.3,1X))
NRITE(u,88)
RETURN
T A B L E 3
WRITE(N,89)IDAY,CUMDD,PCUMDD
FORMAT('0*,9X,'IDAY=',I3,2X,*CUMDD=' F9.3,ux,
+*PCUMDD=',F9.3,/,9X,*TABLE 3',11X,*FWW-PARASITIZED*
+F INDIVIDUALS',/,9X,67('-'),/,9X,.KP',NX,'1.,9X,.2',9X,
+‘3',9X,'u',9x,'5',9X,.6',5X,'TOTAL',/,9X,67('-.))
DO 11A IK=1,10
NRITE(A,9O)IK,STRG(6,IK,2),STRG(5,IK,2),STRG(A,IR,2)
+,STRG(3,IK,2),STRG(2,IK,2),STRG(1,IK,2),PARKP(IK)
CONTINUE
WRITE(u,88)
WRITE(N,87)PARIN(6),PARIN(5),PARIN(u),PARIN(3),PARIN(2)
+,PARIN(1)

 

165

WRITE(8,109)IDAY,CUMDD,PCUMDD,(EPCTKP(J),J=1,10)
109 FORMAT(1x,I3,2x,2(E9.3,2X),1O(F8.u,2X))

RETURN

END

APPENDIX B

DEGREE-DAY ACCUMULATIONS FOR GULL LAKE

1977-1979

xooofioxmsz-n

1977 DEGREE-DAYS (> 9 C)

166

 

AUG

SEPT

 

DAY MARCH APRIL MAY JUNE JULY
1 95 190 520 818 1289 1659
1 99 195 525 828 1295 1667
1 52 201 530 890 1309 1677
1 53 206 591 858 1325 1689
1 53 217 555 876 1390 1709
1 53 227 563 895 1359 1715
1 59 239 566 913 1368 1726
2 59 238 570 929 1381 1738
‘1 55 2110 5711 9‘13 1393 1750
7 61 293 580 956 1907 1761
12 66 2118 586 969 11120 1768
18 75 255 593 985 1929 1776
20 83 266 . 599 999 1990 1782
20 88 276 609 1013 1952 1788
29 95 287 618 1033 1961 1795
26 103 299 631 1052 1979 1803
26 113 313 697 1069 1983 1813
26 125 326 662 1089 1990 1826
26 136 391 679 1109 1997 1838
26 197 356 689 1129 1509 1899
26 157 372 692 1193 1511 1898
26 162 387 702 1156 1521 1856
26 169 902 713 1169 1531 1869
26 166 918 728 1182 1598 1879
26 166 ”33 7'42 1196 1555 18811
27 168 997 759 1206 1566 1893
30 173 958 768 1219 1582 1899
35 178 972 781 1226 1599 1907
39 181 985 799 1290 1612 1911
93 185 996 805 1259 1623 1917
95 509 1270 1638

167

1977 DEGREE-DAYS (>11 C)

 

 

DAY MARCH APRIL MAY JUNE JULY AUG SEPT
1 0 32 199 936 691 1109 1932
2 0 35 153 939 700 1119 1999
3 0 37 157 999 710 1131 1952
9 0 37 162 953 727 1196 1963
5 0 37 171 966 793 1159 1976
6 0 37 180 972 760 1171 1985
7 0 37 185 979 777 1189 1995
8 1 37 188 977 791 1196 1505
9 2 38 189 980 809 1206 1516

10 9 93 192 985 815 1219 1525
11 8 98 196 990 826 1229 1530
12 13 55 201 996 891 1237 1537
13 19 62 211 500 859 1296 1592
19 19 65 220 508 866 1257 1596
15 17 71 229 516 885 1265 1552
16 18 77 290 527 902 1276 1558
17 18 86 252 592 917 1289 1567
18 18 96 269 555 931 1289 1578
19 18 106 277 565 950 1295 1589
20 18 115 290 579 968 1300 1593
21 18 129 305 581 985 1306 1596
22 18 127 319 589 996 1319 1603
23 18 128 332 599 1008 1323 1608
29 18 130 396 612 1019 1338 1617
25 18 130 360 625 1032 1399 1626
26 19 131 373 635 1090 1359 1633
27 21 135 382 697 1098 1368 1638
28 29 190 399 659 1058 1389 1693
29 27 191 905 670 1070 1395 1697
30 30 199 915 680« 1082 1909 1652
31 31 927 1097 1918

168

1978 DEGREE-DAYS (> 9 C)

 

 

1 O 6 65 299 621 1008 1910
2 O 8 66 311 630 1021 1923
3 O 11 69 318 690 1039 1938
9 0 19 72 325 651 1093 1997
5 0 15 72 332 662 1051 1960
6 0 16 75 393 677 1061 1979
7 O 19 78 357 692 1079 1989
8 0 20 83 365 707 1088 1506
9 O 20 87 372 718 1099 1523
10 O 23 92 382 727 1111 1539
11 0 26 97 399 739 1123 1556
12 O 28 105 907 792 1136 1568
13 O 29 113 913 755 1151 1579
19 0 30 118 917 767 1167 1585
15 0 30 120 926 781 1185 1599
16 O 30 126 935 791 1201 1609
17 0 32 132 950 802 1215 1616
18 O 33 190 963 816 1230 1628
19 0 39 151 973 833 1296 1692
20 0 35 163 985 851 1255 1658
21 0 36 166 997 868 1265 1670
22 0 38 172 505 886 1277 1679
23 O 39 180 519 900 1293 1679
29 O 93 189 526 911 1309 1686
25 0 95 200 590 925 1329 1691
26 0 99 213 553 992 1337 1696
27 0 52 227 568 956 1351 1703
28 0 56 292 582 965 1366 1706
29 0 60 257 595 978 1378 1711
30 O 69 273 608 986 1390 1716
31 4 285 995 1900

169

1978 DEGREE-DAYS (>11 C)

DAY MARCH APRIL MAY JUNE JULY AUG SEPT

5 95 239 518 857 1212

 

1 0
2 0 6 96 250 525 868 1223
3 0 8 98 255 539 880 1237
9 0 10 50 261 593 887 1299
5 0 11 50 267 553 899 1255
6 0 12 52 276 566 903 1268
7 o 13 59 289 580 919 1281
8 0 19 58 296 592 926 1296
9 0 19 61 301 603 936 1312

10 0 16 69 309 610 997 1327

11 o 18 68 320 615 957 1392

12 0 20 79 332 622 969 1353

13 0 21 81 337 639 982 1357

19 0 21 89 390 699 996 1366

15 o 21 85 398 656 1012 1379

16 0 21 89 355 665 1027 1383

17 o 22 99 368 675 1039 1393

18 o 22 101 380 687 1053 1903

19 o 23 111 388 702 1068 1915

20 0 23 120 398 719 1075 1930

21 0 23 123 909 735 1083 1990

22 0 25 127 916 750 1099 1999

23 o 26 139 923 763 1108 1998

29 0 28 192 933 773 1122 1953

25 0 30 151 996 785 1136 1957

26 o 33 162 957 800 1198 1961

27 0 35 175 971 813 1160 1967

28 0 38 188 983 820 1173 1969

29 0 92 202 995 831 1189 1973

30 0 99 217 506 838 1199 1976

31 3 227 896 1203

 

 

170

1979 DEGREE-DAYS (> 9 C)

DAY MARCH APRIL MAY JUNE JULY AUG SEPT

27 89 285 605 1008 1363

 

1 0

2 0 27 95 292 616 1022 1378
3 0 27 101 301 626 1035 1393
4 0 27 102 312 636 1050 1406
5 0 27 103 323 643 1063 1419
6 0 27 109 336 651 1076 1432
7 O 27 119 350 660 1092 1446
8 0 27 132 365 672 1109 1456
9 0 27 145 380 687 1121 1460
10 0 27 159 392 699 1137 1465
11 0 27 170 399 714 1145 1477
12 0 33 173 405 730 1152 1488
13 0 40 176 413 746 1161 1500
14 0 41 180 425 762 1168 1514
15 0 41 184 439 779 1173 1523
16 0 42 188 453 792 1179 1528
17 1 43 193 467 805 1188 1535
18 6 45 205 476 814 1197 1544
19 8 48 216 486 824 1208 1554
20 9 53 222 499 835 1219 1560

21 10 59 228 512 847 1231 1566
22 14 64 231 525 861 1242 1575
23 20 69 237 531 876 1256 1580
24 22 77 240 537 891 1270 1584
25 22 82 244 543 907 1279 1590
26 22 86 247 551 919 1288 1598
27 22 87 251 564 933 1299 1607
28 22 88 254 575 949 1311 1616
29 23 88 261 589 964 1322 1625
30 26 88 269 598 979 1336 1636
31 27 276 995 1350

1979 DEGREE-DAYS (>11 C)

171

xooo-qoxmzwm—I

WWNNNNNNNNNN—I—I—Id—hd—a—I—a—n
doom-4mmsz—aoooo—qosmsz-ao

A
OO‘O‘U’IZOOOOOOOOOOOOOOOOO

d—fi—l—I-A—l—ld—J
m‘lO‘U'IU'IU'IU'IU'IU'I

389
398
909
921
432
437
992
997
959

465 .

975
987
999

701
712
723
737
751
765
775
788
803
816
829
844

1000
1005
1012
1020
1029
1039
1049
1059
1071
1083
1091
1099
1107
1118
1128
1140
1152

APPENDIX C

DETERMINATION OF DEVELOPMENTAL TEMPERATURE THRESHOLD

FOR ETC EGGS

 

172

DETERMINATION OF DEVELOPMENTAL TEMPERATURE THRESHOLD
FOR ETC EGGS

Because the egg stage in the ETC is of importance in initializing
sampling programs, extensive rearings were done on this life-stage. Ten
egg masses were placed in each of six temperatures ranging from 45° F
(7.22° C) to 95° F (3S.29° C). Each egg mass was monitored daily with
the number of individuals emerging recorded.

Development was recorded in terms of the number of days required to
hatch as a function of temperature (Fig. C.lA). After tabulating these
values, each is inverted so as to transform them to percent development
per day. Threshold determination is.done by regressing percent/day on
temperature and calculating the point at which there is 0% development/
day. This analysis was performed on the linear portion of the data with
10.26° C calculated as the base temperature (Fig. C.lB). This number
represents only an initial estimate for the second portion of the
analysis.

The standard error technique (Casagrande 1971) was used in lieu of
possible nonlinearity. Here, To's bracketing the suspected true base
(as estimated through regression) are used in calculating degree-day
accumulations. Standard errors are calculated for each assumed base
over the range of rearing temperatures. The point at which the
standard error is minimized provides the best fit for the given data

set To ' 9° C (Fig. C.lC).

DAYS 10 "RICH

IO
1

 

 

Y - Rattan-4.0)

173

 

 

 

Figure 6.1.

V

v

V

T

1.

fit r f V V Y

a;
trnrennruat °c

8‘

13 20 8‘ 30
1 L 1 L

PCT. DEVELOPMENT/DRY

10
L

 

3‘

8'4
.39
5 .J
:53
U
D 4
C
C
3?
553‘
m
3‘
4
' ' ' l o ' r ' ' r v 1 Tfi
‘0 O 4 O 11 o I. 20
B. ESTIHHTED ans: c

 

 

U
24
TEMPERATURE °c

Time-temperature functions and determination of To

for ETC eggs.

APPENDIX D

STUDY AREAS AT THE KELLOGG BIOLOGICAL STATION

174

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SECTION 5 SECT'ON 4
LN
\ I. '|
5 c 7 . 1|
9 I .3 I
‘r
O I II D I. Lhufi----q‘
I i 11
:LJ L
i -'—q 1_. m
’ I
a“ '8 '
u 4 '1 1'1
N V
. . . L--Ju--JV
SECT1ON 8 SECTION 9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 L— " H
7- hm " fl 3'
Coward: .7 : Io to
0.1.1ij --
.. .. 1: 0
g
C
\ 1
t 1

 

 

 

 

 

Study areas within the KBS.

APPENDIX E

POPULATION ESTIMATORS FOR PARASITES AND COLONIAL HOSTS

175

POPULATION ESTIMATORS FOR PARASITES AND COLONIAL HOSTS

Before undertaking any sampling program there is a need to identify
the population parameters required as dictated by the project objectives
and characteristics of the sample units. As alluded to above, the
objectives are to examine parasitism rates and production of adult
parasites throughout the course of a growing season. These rates and
numbers will then be related back to the size of the host population,
its age, and any other parameters of interest.

The need to provide probablistic statements concerning population
size is characteristic of many host-parasite studies. This need, in most
instances, is adhered to only with reference to the host population.
Parasitism is typically viewed as a mean value (percent) with no refer-
ence to the precision of that point estimate. The following treatment
serves to present estimators for both host and parasite populations and

methods for reducing variance terms.

HOST POPULATION ESTIMATORS

Because of the highly visible characteristic of both the ETC and
FWW we have the ability to make very precise counts of the number of
colonies per unit area. With reference to the total population the problem
concerns estimating the number of larvae per colony. In order to accomplish
this, whole colony samples were taken for each sample period and the total

number of individuals determined. The mean number of larvae per colony

G) is:

176

_ 1_ nc
Y'ri 3'1 [1]
c i=1
where:
y1 = the number of larvae in the ith colony, and

nc ' the number of colonies sampled.

The association variance term is:

n

2.._1__ c :2
8c nc-l 1E1 (yi y) ' [2]

Because of the knowledge of the actual number of colonies, (Nc)v the
variance of the mean (§) can be reduced with a "finite population
correction" (FPC) or:

N -n S 2

var(y)- :1 c c [3]
c c

 

The FPC weights the variance term by the proportion of the population
sampled (NC-nc/Nc).

Estimates arrived at, at the colony level, can now be projected to
the entire host population (i.e., the total number of host larvae (§))

with:

c i-l
Nc no
8 -- 2 y [4]
DC 181 1

The variance of y is:

var(y) - Nc (vary(y))

S 2(N -n )
a N c c c [5]
c n
c

 

177

It should be noted once again that there is only a single component in

var (y) because of the knowledge of absolute colony density.

PARASITISM ESTIMATORS

Parasite studies in general dictate that host individuals fall
principally into two classes, parasitized and unparasitized. In some
instances a third class must be recognized, that being parasite eggs
or larvae which have been encapsulated. This problem reduces to estimating
the preportion of the host population parasitized and the proportion of
parasite attacks which have been nullified by encapsulation. I will
first derive the equations necessary for estimates at the colony level
and further calculations necessary in the plot/subplot design. Within
this context, a given number of colonies are sampled in_each of the
subplots pictured in Appendix D. This produces a 2-phase scheme. Phase 1
provides estimates of colony size, colony density, and total larval
numbers (described above). Phase 2 includes samples taken from intact
colonies and estimates of parasitism rates calculated.

The total number of attacks over a given time period is, of course,
of great interest in our study of host-parasite relations. This includes
both those individuals that have viable parasites and those containing
encapsulated eggs and larvae. Let:

a1 - the number of unparasitized hosts in the sample/colony,

a2 - parasitized hosts, and

a3 a the number of hosts containing encapsulated parasites.
Therefore, the proportion of the hosts attacked (Pa) is:

a2+a3

Pa = a1+a2+a3 [6]

178

The variance of P8 is:

32 - nlpaqa

pa n-l [7]

As was done earlier (Eq. 3) the FPC is used to reduce SPa or:

N1(N1-n1)

T Page ‘81

var(pa) =
with:

98 - l-pa.
In this instance, an estimate of the number of larvae per colony (N1) or

§ (Eq. 1) must be used in place of the absolute numbers in equation 3._.

Hence:

P q [9]

where:
n1 - a1+a2+a3
- the number of larvae sampled in the colony.
In addition to the attack rate (pa) the number of attacks producing
viable parasites (pp) (i.e., those escaping encapsulation) is calculated.
The proportion of the colony being effectively parasitized is:

82

pp - a1+a2+a3 [10]

Like equation 9 the variance of pp is:

§(§-n1)
---- 11
n1_1 ppqp I 1

var -
(pp)
The methodology concerning encapsulation is similar to attack and

parasitism rates, however, here the population of concern is now the

parasite. With this in mind the encapsulation rate for the colony is:

179

a
3 [12]

Fe 32+83

 

Further, using a slightly different FPC a variance estimate for pe
is:

-_ I -
pay a2 .pay)peqe
pa? az+83

 

var(Pe) ' [13]

The value of pay estimates the parasite population on a per colony basis
and a2 representing the sample size analagous to n1 in equation 11.

This concludes the derivation of colony level parasitism parameters
and with the inclusion FPC's applies well to samples taken in the tent
caterpillar-webworm system because of their colonial nature. In cases
where the sample frequency is, or can be assumed to be, less than 52 the
FPC can be dropped from variance estimates. In this example all variance
calculations include FPC's. This is due to the fact that early in the
host generation correction factors may only reduce negligible amounts of
variance (i.e., with a sample frequence of < 52). However, as the genera-
tion progresses the FPC becomes more significant and is retained throughout

calculations for continuity and generality.

MULTISTAGE PARASITISM ESTIMATES

Methods derived at the colony level can further be applied to subplot,
plot, and regional levels of the study area. This type of sampling program
is termed a multistage or nested design. Calculation of mean values for
each level is straightforward 2h: that for any level, estimates are derived
as a mean from the next lower level. For example, a subplot parasitism
mean (SP) is the summation of the colony estimates divided by the number

of colonies sampled (nc) or:

180

P [14]

Calculation of variance terms at any level is equally straight—
forward as mean estimates. Each level contains components of variance
from lower levels. In order to clarify this relationship further an
analysis of variance table is presented (Table E.l). The reader should
notice that variance components are additive. Subplot parasitism

variances are the sum of colony and subplot variance such that:

s 2 - -
_ nc-nc] Pa ] y(y-ni) 1
- +*
var(pp) Nc J Nc J p q I 5]

 

n-l p p

Equation 15 then includes the between colony variance (subplot) with an
FPC to take into account the proportion of the colonies sampled in that
subplot. The FPC in this case is Né-nc/Nc where N6 is the total number
of colonies in the subplot and me, the number sampled. The within colony
variance (Eq. 11) is added directly to determine the subplot variance
estimate. These methods can extend to the plot level where needed, with

an additional variance component included as defined in Table E.l.

Table E.l. ANOVA table for components of variation in parasitism rates.

 

Components of

 

Source of Variation df Variance
Total N n n -l
s c 1
Plots n-l s 2+ns 2+nn s 2
s c 8 c1 p
_ 2 2
Subplots ns(n1 1) SC +nlsS
_ 2
Colonies nsnc(nl 1) SC

 

APPENDIX F

SPECIES CONCEPTS IN THE FALL WEBWORM

181

SPECIES CONCEPTS IN THE FALL WEBWORM

Since Lyman's (1902) original treatment of the genus Hyphantria

 

there has been continual reference made as to taxonomic status of black
and red headed "races" (BB and RH). In fact, in Lyman's work 2 distinct
species were recognized g, cunea Drury (BH) and fl, textor Harris (RH).

Since that time the dichotomdzation of gyphantria has been retained by

 

the recognition of species, subspecies, or races. Morris (1963) comments;
"The taxonomic status of the webworm with light-headed larvae is not yet
clear. In population studies, at least, it should be treated as a separate
species." He states further, "This webworm (RH) and ggng§.(BH) may prove
to be sibling species, occurring sympatrically from New Brunswick to
Georgia." These ideas are given additional substance when other characters
such as feeding rhythms, methods of web construction, diapause inducing
photoperiods, lengths of larval development, and range of host plants are
considered (Oliver 1963, Ito and Warren 1973). On the other hand, each

of these studies has, by design or by convenience, utilized populations
which occur toward the extremes in the range of the webworm. Thus,
populations from New Brunswick and Nova Scotia are compared with those

from Georgia and Arkansas. This, in effect, produces some rather striking
differences in those characteristics mentioned even though BH and RH
individuals, when mates, produce viable intermediate offspring (Morris
1963, Ito and Warren 1973). Because of these rather obvious differences
questions arise as to the taxonomic placement and more importantly,
treatment in population studies. The approach taken here will synthe-

size the available information with reference to possible changes in

population make-up in response to gross climatic differences for different

182

parts of North America. The objective of this discussion is not to solve
the problem but to view the webworm in light of basic postulates concerning
the species and its variability. Also, it allows us to place the Michigan
population in relation to others in North America.

As mentioned above, the view taken to date has been one of dis-
cretizing the BH and RH components of the FWW. This is quite under-
standable if we compare entities from the extremes in its range. However,
taking the data supplied by Warren and Tadic (l970),Ito and warren (1973),
Hattori and Ito (1973), and Morris (1963) provides quite a different
picture. Figure F.l presents 2 types of information. First the bar
graphs located in Nova Scotia, Michigan, and Georgia picture the contin-
uous nature of head color. The x-axis of each is a color continuum
ranging from black to red and y the frequency of phenotype. It is clear
from the Nova Scotia and Georgia plots (Morris 1963) that populations tend
toward the BH penotype in the north while RH is favored in the south.
Michigan apparently represents an intermediate with a distribution not
significantly different from normal (x2 - .1838, 3 df) (data from this
study). Looking at Morris' temperature treatments it is intuitive that
this might occur. A colony from the Nova Scotia population was reared
under different temperature regimes in the laboratory and resulted in
significant phenotypic shifts (Table F.l). This indicates a classic
example of melanistic variation which occurs in many other insects in
response to changes in latitude and/or altitude, hence temperature
(Merritt 1970, Zuska and Berg 1974). This trend continues throughout
Canada and the United States when looking at data supplied in Warren and
Tadic (1970). The ratios (BH:RH) in Figure F.l present these data.

Data from Washington and Kansas indicate colonies which became mixed

183

 

-- Cold climate with moist winter, cool sullner

.- Cold climate with moist winter, hot sulmaer
.- Humid temperate climate, hot suIner

l- Tropical wet and dry

33' Semiarid climate or steppe

g- Desert climate

.- Warm climate, cool dry stunner

.- Humid temperate climate, cool summer

0 - Cold climate with moist winter, cool summer

Figure F.1. Distribution of FWW head capsule color.

184

amass matte: aotmv
.H canoe a“ Aev moauom mo mammoumm

.H manna Ca ANV magnum mo xsmwoum:

.ah0w cumnusom one
mo maasmm uwwuma m mcw>oua cu umuuo cw .mcwaouoo :uuoz can mkuoam :« cmuomHHoo om>uoH meow mmvsdoaHm

.Anv magnum ta can» amxaaou

.mmv sumo mason c you o=Hm> edema:
one cu ousumumaaou Ham panama scans mQEoH use: cu H>I>H modems“ weapon vomoaxo one am On an voummmi

 

 

e we me o mace "outage m.m.z x oawaomu H

o as on o mock "coaamm aaamaooo :

o as ms nu sauna ca umaumsaoo naawaoao w

om No ma o paoam cw wouooaaoo mauoom o>oz m
«sea o o o mowm : : : m

OCH 0 o o maeo : : = v

m cm a o boos : : : o

o w: an o HmommION : : : n

o mm mm o Anomaloh uvmumom ofiuoom m>oz m
xuma aafivwz uzwfia uswwq huo> unwauooua mouaom mowumm

 

om>pm4 mo mwmucwouom

 

.mmum couuomaaou
can munumuoeamu ou Escapes Hana ocu uo m>uma umumawlnumam one :H uoaoo mo :owunaom .H.m manna

185

in transit (12 and 5, respectively). The data were derived from sample
colonies taken from these various areas and classified only as to BE or
RH. Even with this restricted classification the same pattern emerges
with BB favored in colder climes. In addition to the frequency of BH
and RH phenotypes Figure F.l also shows a gross climatic classification
of North America (Danks 1978) which supports the hypothesis of tempera-
ture induced melanization.

Until this point I have only discussed head color as an indicator
of population type and this has been the approach taken by other authors.
The implicit assumption has been that head color, behavioral characters,
and other variables were somehow genetically associated, presumably
through pleiotropism. Therefore, the BH "race" is equated with loosely
formed webs, day and night feeding, a shorter developmental period, and
a diapause inducing photoperiod of 10-14 hrs. While the RH group is
associated with compact webs, nocturnal feeding, a longer developmental
period and a critical photoperiod'of 10-14 hrs. While the RH group is
associated with compact webs, nocturnal feeding, a longer developmental
period and a critical photoperiod of 18 hrs. These equations counter
the ideas of clinal attributes in populations and numerous examples of
character gradation (Dobzhansky 1970). Morris' temperature experiment
demonstrates environmental intervention yet color and all other characters
are considered as one. Additional studies by Morris further support the
idea of clinal changes (Morris and Fulton l970a,b, Morris 1971).
Selection pressure due to temperature has been shown to truncate the
distribution of adult emergence with early emerging individuals favored
in cool years and late individuals in warm years (Morris and Fulton 1970a,

b). This tends to explain not only the later emergence of RH individuals

186

but also changes in melanism from one population to another. It may well
be that developmental rates are also selected for by the temperature
regime and hence correlated with head color.

These data suggest that the FWW represents a highly variable species
showing clinal changes on climatic gradients. The fact that intermediate
forms occur (behaviorally and otherwise) in Michigan, New Brunswick and
Vancouver tends to invalidate the 2 "race" concept. In addition, the 2
race idea provides only a static view of a gene system which may be
evolving in a number of directions. This does not negate the possibility
of a pleiotropic gene but at this point there is no data to back the
assumption. It is more reasonable to assume that a number of genes are
operating forming the observed continuum of phenotypes and that they are
highly correlated through spatial and temporal selection pressures such
as temperature. This points to the need to view webworm populations, or
animal and plant populations in general, as heterogeneous assemblages.
Whether or not our goals are of a taxonomic nature, aimed at population
processes or in a management mode we cannot consider things as only
"black or red." As Wellington (1977) points out, the need is to return
the insect to insect ecology: "Insect populations no longer appear to
be inert masses passively responding to changing environmental pressures."
The webworm problem is a classic example of the need to define the frequency
distribution of phenotypes (genotypes) in natural populations. Populations

exhibit a mean and variance and one without the other is meaningless.