1V1€31_J RETURNING MATERIALS: Place in book drop to LJBRARJES remove this checkout from 4—! your record. FINES wiH be charged if book is returned after the date stamped below. A VARIABLE LIFE-TABLE FOR Illinoia pepperi (MacGillivray) BY Robert Delain Kriegel II A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Entomology 1987 ABSTRACT A VARIABLE LIFE-TABLE FOR Illinoia pepperi (MacGillivray) By Robert Delain Kriegel II A variable life-table approach is used to design and parameterize a population dynamics model for the blueberry aphid” IELlinoia pepperi (MacGillivray). in Michigan's commercial highbush blueberry agro-ecosysman. The model addresses aphid growth and maturation, seasonal fecundity, morph determination, vertical within bush distribution, parasitism. and rain induced mortality. Degree day durations for parthenogenic life stages are presented. Fecundity was found to decline seasonally from 24 to 3 young per female for both viviparous morphs. The proportion of alates in the population decreased sigmoidally from 79% at 1400 on 38 changes 511 the aphid's vertical distribution followed to less than 5% by 2000 0D38. Although seasonal changes in host plant phenology; parasitoids remained, predominantly, in the lower third of the bush. While rates of parasitism remained below 3% at a chemically treamai site. parasitism exceeded 15% at an unsprayed site. Storms proved to be an important mortality factor. At times, individual thunderstorms accounted for over 50% reductions in aphid numbers. To my father, who shared the dream but is not alive to see it fulfilled. I'd rather learn from one bird how to sing than teach ten thousand stars how not to dance. e.e. cummings Deadlines and commitments, what to leave in what to leave out. Bob Seiger ii ACKNOWLEDGMENTS First, I would like to thank my graduate advisor, Dr. Mark Whalon, for giving me the opportunity to work on such a challenging project, for allowing me to voice my opinions freely, and for giving me a push when I slowed down. I would also like to formally thank him for a rather unusual musical instrument that keeps me company during the long nights of blacklighting for Lepidoptera. Next” It wish to thank the members of my graduate committee, Drs. Donald Ramsdell, Jim Hancock, and Stuart Gage for their support, suggestions, and critical review of this manuscript. I would especially like to thank Dr. Gage for encouraging my explorations into computer programming and computer simulation techniques. I would also like to express my graditude to Terry Davis and Nancy Cushing for assisting me with the field studies and in the laboratory. Finally, to my fellow students who provided intellectual stimulation, emotional support, and who always knew how to enjoy themselves, thank you all. iii TABLE OF CONTENTS CHAPTER 1. INTRODUCTION . . . . . . . . 1.1 Why study an Aphid? . . . . . . 1.2 History of Blueberry Shoestring 1.3 Disease Symptomatology and Host 1.4 Disease Spread . . . . . . . . 1.5 Virus-Vector Relationship . . . 1.6 Biology of I, pepperi . . . . . 1.7 Thesis objectives . . . . . . . 1.8 Thesis Structure . . . . . . . CHAPTER 2. NATURAL ENEMIES . . 2.1 Introduction . . . . 2.2 Materials and Methods 2.3 Results . . . . . . . 2.3.1 Parasitoids . 2.3.2 Predators . . 2.4 Discussion . . . . . 2.4.1 EntomOphthora 2. 4. 2 Sampling Method 2. 4. 3 Parasitoids . . 2. 4. 4 Predators . . . CHAPTER 3. RANGE AND SPATIAL DISTRIBUTION 3.1 Introduction . . . . . . . . . 3.2 Geographical Distribution . . 3.3 Within Bush Distribution . . 3.3.1 Materials and Methods 2 Results . . . . . . . 3 Discussion . . . . . . CHAPTER 4. APHID GROWTH AND MATURATION ‘4.1 Materials and Methods . 4.2 Experimental Bias . . . 4.3 Results . . . . . . . . 4.4 Discussion . . . . . . CHAPTER 5. FECUNDITY . . . . . . 5.1 Introduction . . . . 5.2 Materials and Methods . 5.3 Results . . . . . . . . . 5.3.1 Frequency Distribution of 5.3.2 Seasonal Aphid Fecundity . 5.4 Discussion . . . . . . . . . . iv 0' O O 0 Disease Range . B r h o 0 Po. 0.. o o c (To 0 o o o o m._. o o OCDQGQU'INHH CHAPTER 6. MORPH DETERMINATION . . . . . . . . . . . . 6.1 ’Introduction . . . . . . . . . . . . . . . . 6.2 Literature Review . . . . . . . . . . . . . . 6.3 Materials and Methods . . . . . . . . . . . . 6.4 Results . . . . . . . . . . . . . . . . . . . 6.5 Discussion . . . . . . . . . . . . . . . . . CHAPTER 7. ABIOTIC MORTALITY . . . . . . . . . . . . . 7.1 ‘Introduction . . . . . . . . . . . . . . . . 7.2 Materials and Methods . . . . . . . . . . . . 7.3 Results . . . . . . . . . . . . . . . . . . . 7.4 Discussion‘ . . . . . . . . . . . . . . . . . CHAPTER 8. DESIGNING A LIFE-TABLE FOR I, pepperi . . . 8.1 Introduction . . . . . . . . . . . . . . . . " 8.1.1. .A variable lif -table for Masonaphis 8.3.1 8.3.2 8.3.3 8.3.4 8.3.5 8.3.6 8.3.7 8.4 Discu CHAPTER 9. APPENDIX 1. APPENDIX 2. BIBLIOGRAPHY maxima (Mason) . . . . . . . . 8.2 Materials and Methods . . . . . . . 8.3 Model Design and Parameterization . SSion C O C C C C O O C O O O O over-View O O C O O O O O 0 How the Model Keeps Time . Sequence of Calculations . Modeling Aphid Maturation . . . Fecundity and Morph Determination Modeling Spatial Distribution Mortality Component . . . . . CONCLUSION I O O O O O O I O I O O O O Voucher Specimens . . . . . . . . . . Distribution Map Data . . . . . . . . 86 86 86 91 92 95 96 96 97 99 102 103 103 104 106 107 107 110 111 113 118 122 123 129 131 136 142 144 LIST OF TABLES Potential predator and parasitoid species of I. pe pperi in Michigan . . . . . . . . . . . 12 Lower developmental temperature thresholds and generation times for several species of aphids and their parasitoids . . . . . . . . . . . . . . . 14 Lower developmental temperature thresholds and generation times for several species of aphid predators . . . . . . . . . . . . . . . . . . . . 15 Observations of coccinellid predators during the 'standard sample' . . . . . . . . . . . . . . . 36 Observations of cecidomyiid predators during the 'standard sample' . . . . . . . . . . 39 Overall percentages of aphids in each third of the bush for samples with N > 20 aphids . . . . . . . 53 Overall percentages of parasitoid pupae collected in each third of the blueberry bush . . . . . . . 56 ANOVA summary of fixed temperature rearing experiment for four nymphal instars of I. pepper reared at 5, 10, 17, and 23°C.. . . . . . . . . . 62 ANOVA summary of fixed temperature rearing experiment for all life stages of I. pepperi reared at 5, 10, 17, and 23 . . . . . . . . . . 64 ANOVA summary of fixed temperature rearing experiment for all life stages of I. pepperi reared at 10, 17, and 23°C . . . . . . . . . 64 Discrete probability density function of the temporal pattern of births for apterous I_. pe pperi 78 ANOVA summary of seasonal fecundity of I, pepperi where fecundity= young in colony plus all embryos 79 ANOVA summary of seasonal fecundity of I. pepperi where fecundity = young in colony plus developed embryos O O O I O I I O O O O O O O O O O O O O O 79 Regression results of seasonal fecundity study where fecundity = young in colony plus all embryos 81 Regression results of seasonal fecundity study where fecundity = young in colony plus developed en‘bryos O O O I O O O C O O O O O I O O I O O O 81 Regression statistics for rain mortality field experiments . . . . . . . . . . . . . . . . . . . 102 Distributed delay parameters and maximum allowable DTs for each life stage of apterous and alate I. m I O O O O O C C I C C O O I O I O O O O O 117 8.2 Table look—up values for the percentage of the aphid population occurring in each third of the bush versus degree days . . . . . . . A2.1 Localities in Michigan where I, pepperi collected . . . . . . . . . . . . . . O O O O I I 124 has been 0 O O O O O 142 vii LIST OF FIGURES Typical sample of I. pe pperi and parasitoids from one blueberry bush— (3 terminals) in mid July (first data entry screen) . . . . . . . . . . Typical sample of predators of I, pepperi from one blueberry bush (3 terminals) during a population explosion in mid July (second data entry screen) . Parasitoid summary for site 1, 1981: (a) daily rainfall. (b) total aphids per sample, (c) aphid mummies containing parasitoids, and (d) empty parasitoid cocoons . . . . . . . . . . . . . . . . Parasitoid summary for site 1, 1982: (a) daily rainfall, (b) total aphids per sample, (c) aphid mummies containing parasitoids, and (d) empty parasitoid cocoons . . . . . . . . . . . . . . . . Parasitoid summary for site 2, 1982: (a) daily rainfall, (b) total aphids per sample, (c) aphid mummies containing parasitoids, and (d) empty parasitoid cocoons . . . . . . . . . . . Percent parasitism versus degree days for (a) site 1, 1981: (b) site 1, 1982; and (c) site 2,1982 . Sample counts of spiders and spider egg cases for (a) site 1,1981: (b) site 1.1982: and (c) site 2' 1982 O I O O O O O O O O O O O O O O O O O 0 Sample counts of chrysopid predators for (a) site 1, 1981; (b) site 1,1982: and (c) site 2, 1982 . Sample counts of syrphid predators for (a) site 1, 1981: (b) site 1, 1982; and (c) site 2, 1982 . . . Distribution of I. pe pperi on wild and cultivated Vaccinium species in Michigan . . . . . . . Vertical distribution of I, pe pperi within highbush blueberry bushes at (a) site 1,1981: (b) site 1, 1982: and (c) site 2, 1982 . . . . . . . . Vertical distribution of the aphidiid parasitoid that pupates within the aphid mummy at (a,d) site 1. 1981: (b,e) site 1, 1982; and (c,f) site 2, 1982 O I C O O O O I O O O O O O I O O O C O O O 0 Vertical distribution of the aphidiid parasitoid that pupates beneath the aphid mummy at (a,b) site 1, 1981: and (c,d) site 2, 1982 . . . . . . Survivorship of reproductive I. pepperi versus (a) physiological time, and (b) calendar time . . . . viii 22 22 27 28 29 33 35 38 40 49 52 54 55 61 Temporal pattern of reproduction for I. pepperi reared at (a) 5°C, (b) 10°C, (c) 17°cT and (a) 23°C. Dotted line marks 90% mortality . . . . . . 72 Temporal pattern of reproduction: (a) cumulative births/female versus OD after molt to adult, (b) transforming the cumulative function to a cumulative density function (CDF), (c) aphid fecundity GDP, and (d) discrete probability density fecundity function (PDF) . . . . . . . . . 75 Morph determination in I. pe pperi: the proportion of alates (a) as a function of crowding and OD, and (b) only as a function of 0D. . . . . . . . . . . . . 92 Data and linear regression of rain induced mortality for the (a) upper, (b) middle, and (c) lower third of the bush . . . . . . . . . . . . . 100 Overview of the I, pepperi population dynamics model . . . . . . . . . . . . . . . . . . . . . . 108 Sequence of calculation in the blueberry aphid variable life-table. Modified from Manetsch & Park 1980 . . . . . . . . . . . . . . . . . . . . 112 Flowchart showing input and output variables for the distributed delay used in the model . . . . . 115 Simulated maturation of one cohort of apterous and alate I. pepperi . . . . . . . . . . . . . 119 Flowchart for the fecundity and morph determination sections of the model . . . . . . . 121 Table look-up function used to simulate the vertical within bush distribution of aphids through the growing season . . . . . . . . . . . 124 Flowchart for the mortality section of the aphid population dynamics model . . . . . . . . . . . . 125 Table look-up functions used to simulate high, medium, and low levels of parasitism . . . . . . . 127 ix CHAPTER 1 . INTRODUCTION 1.1 Why Study an Aphid? The aphid investigated herein, Illinoia pepperi (MacGillivray). is the only known vector of blueberry shoestring virus (BBSSV), an important disease of cultivated highbush blueberry and the most serious virus disease of blueberry in Michigan (Varney 1977. J. Nelson pers. comm.). Berry production of bushes infected with BBSSV declines drastically before the bush.is eventually removed as unproductive. Once a bush is infected, nothing can be done to stop or slow the progress of the disease. For this reason efforts tn) stop the spread of BBSSV must focus on preventing new infections and identifying and removing existing sources of innoculum. To be effective research directed towards such control must investigate all three components of the plant-virus-vector relationship. Of the three components, the virus vector is the least understood. The virus itself has been the subject of study for almost thirty years and some cultivars partially resistant to BBSSV have been known for over a decade (Varney 1970). On the other hand, I, pepperi was implicated as the vector of BBSSV only eight years ago (Ramsdell 1979a). Since that time several aspects of the aphid's biology have been investigated. Laboratory studies have been conducted to determine the lower develOpmental temperature threshold, developmental rate, generation time, and 1 2 fecundity of apterous viviparous I, pepperi (Elsner 1982). Field investigations to determine the aphid's life cycle and seasonal history. alternate host plants, and the identity of natural enemies have been initiated (Elsner 1982). The seasonal timing of alate flights has also been studied (Elsner 1982, Morimoto 1984). The virus-vector component of the disease cycle has also been investigated. Field studies have explored the seasonality of new virus infections and the viruliferousness of the aphid population (Morimoto 1984). Laboratory research has identified the time course of BBSSV acquisition by (Morimoto 1984) and localization within the aphid vector (Peterson 1984). 1.2 History of Blueberry Shoestring Disease The probable viral nature of blueberry shoestring disease was first demonstrated by grafting experiments in the mid 1950's (Varney 1957). The disease had first been reported in highbush blueberry, Vaccinium corymbosum L.. seven years earlier by Hutchinson (Varney 1970). BBSSV was only one of several blueberry diseases first recognized during the 1950's. Although native blueberry species had been partially cultivated in North America since the early 19th century. few diseases of Vaccinium species were known before the introduction of the cultivated blueberry cultivars y, australe Small and V, corymbosum in the 1920's (Varney 1970. 1977). Even more interesting is the fact that 3 (all known virus diseases of cultivated highbush blueberry are believed to be indigenous to New Jersey (Varney 1970). This historical pattern suggests that blueberry shoestring disease spread from native plants to introduced cultivars. This hypothesis is supported by grafting experiments conducted on native lowbush blueberry,‘Z. aggustifolium Ait. (Lockhart & Hall 1962). In that study buds from lowbush blueberry plants were grafted to highbush blueberry cv. 'Jersey' test plants. Although very few of the lowbush plants exhibited symptoms of shoestring disease. all of the 'Jersey' test plants eventually developed typical shoestring systems. BBSSV has also been discovered in wild populations of both V, aggustifolium and y, corymbosum at several sites in Michigan (Ramsdell et a1. 1984). These sites are sufficiently isolated from cultivated fields that .it is doubtful the disease spread from cultivated to wild populations. In his 1957 article Varney stated that shoestring was of minor importance, but he warned that it was a potential threat to the emerging cultivated kdghbush blueberry industry in North America. Since that time BBSSV has indeed become a very real threat to the industry. The disease has been reported from Michigan, North Carolina. and Washington. USA (Ramsdell 1979b) and Nova Scotia, Canada (Lockhart 8: Hall 1962). Financial losses to the blueberry industry from reduced yield and bush removal in Michigan alone amounted to 4 more than three million dollars in 1980 (Ramsdell et a1. 1980). Blueberry shoestring was established as a 'new' virus when isometric particles about 27 nm in diameter were implicated as the disease causing agent (Hartmann et a1. 1973, Lesney & Ramsdell 1976, Ramsdell 1979a, 1979b). Transmission electron microscopy (Hartmann et a1. 1973) revealed these particles in leaf and root tissue but not in phloem. Root xylem contained the largest crystalline arrays of particles indicating that the disease is a systemic root infection. Apparently, the disease only attacks vaccinium species.. .All attempts to innoculate herbaceous plants by sap, dodder, or graft transmission have been futile (Varney 1977. Ramsdell 1979b). Since 1980 researchers from the departments of Botany and Plant Pathology: Horticulture. and Entomology at Michigan State University have been coordinating BBSSV research. In addition to the biological studies of the aphid mentioned above, investigations have focused on developing serological assays to the virus (Morimoto 1984, Peterson 1984). localizing the virus within its aphid vector (Peterson 1984), and screening blueberry cultivars for resistance to both the virus and its aphid vector (Hancock et a1. 1982. Schulte et a1. 1984, Ramsdell et a1. 1984). To date, three serological assays have been developed for detecting blueberry shoestring virus. These techniques are enzyme-linked immunosorbent assay (ELISA), radioimmunoassay 5 (RIA). and immusorbent scanning electron microscopy (ISEM). Of these techniques ISEM is most sensitive for detecting purified virus while RIA is superior for detecting the virus in individual aphids or aphid extract (Gillett et a1. 1982). In 1984 a program was instituted to screen blueberry varieties for resistance to BBSSV using the ELISA technique (Schulte et al. 1984, Ramsdell et al. 1984). 1.3 Disease Symptomatology and Host Range Typical BBSSV symptoms consist of elongate reddish streaks on current and one-year-old stems. Streaking is most noticeable on the side of the stem exposed to the sun (Ramsdell 1979b). Affected leaves are often narrow and strap-like (hence the name 'shoestring'), curled, or crescent shaped. Occasionally leaves may exhibit red vein- banding or oak-leaf patterns. Immature berries on affected bushes develop a purplish cast. This abnormal color disappears as the berries mature with no apparent loss in quality. Often only a portion of an infected bush exhibits disease symptoms and symptomless infections are common. Environmental conditions may alter symptom expression. As a result bushes that exhibit obvious symptoms one year may be symptomless the following year (Elsner 1982). After a few years the yield from infected bushes is dramatically reduced and the bush is eventually removed. BBSSV has been observed in the following cultivars: Burlington, Coville, Barliblue, Jersey, June, Rancocas, 6 Rubel, and Weymouth (Ramsdell 1979b). Under field conditions the disease has never been observed to infect the cultivars Atlantic, Bluecrop, Bluejay. and Nbrthland. However. low percentages of the cultivars Atlantic and BluecrOp do become infected if they are manually inoculated with purified virus (Ramsdell 1979b). 1.4 Disease Spread The disease tends to spread bush by bush along the row (Lesney et a1. 1978). This pattern of spread suggests that apterous aphids are responsible for most of the within field disease transmission. A compound interest rate for disease spread has been calculated at 0.269 bush per year (Lesney et a1. 1978). The infection rate is also influenced by bush size. 'Jersey' seedlings infected with purified virus develop symptoms in five to six months (Lesney et a1. 1978). but healthy bushes planted in a diseased field may not show symptoms for up to four years (Ramsdell 1979b). The quantity of virus (and therefore the number of aphids) needed to infect a mature blueberry bush is not known. 1.5 'Virusévector Relationship BBSSV is transmitted to soft—wood cuttings by I, pepperi after a 10 minute acquisition feeding period and an inoculation period of 100 hours (Morimoto 1984). Longer acquisition feeding periods increase the efficiency of 7 transmission with a maximum reached after about 24 hours (Morimoto 1984). Scanning electron microscopy autoradiography (SEM AR), light microscopy autoradiOgraphy, and ferritin labeling techniques have revealed that BBSSV is transmitted by the aphid 131:3 semi-persistent, circulative manner (Peterson 1984). In the SEM AR experiments 125I-labeled BBSSV ingested by aphids was detected (1) in the stomach six hours after feeding. (2) in stomach and intestines 12 hours after feeding. and (3) throughout the alimentary canal to the anus 72 hours after feeding. Light microscopy autoradiography supported these findings and also indicated that an interaction occurred between embryos and the 125I-labeled BBSSV. The exact meaning of this interaction is not presently known. Finally, indirect ferritin antibody labeling visualized BBSSV particles in the salivary glands of aphids fed on sachets containing virus preparation in sucrose 0 1.6 Biology of I. my; I. pepperi is a holocyclic species of aphid, spending its entire life cycle on Vaccinium species. During population outbreaks the aphid has been observed to feed and reproduce on a few woody plants present in blueberry fields (Elsner 1982). However, none of these colonies was ever Observed to survive for more than two weeks. Reproduction has never been observed in the field on herbaceous hosts. 8 Fiwe morphs of I, pepperi are recognized. These include (1) apterous, viviparous, fundatrix females; (2) apterous. viviparous females: (3) alate. viviparous females; (4) apterous, oviparous females: and (5) alate males (MacGillivray 1958). All morphs are normally green in color: however, a biotype of red morphs has been collected in parts of southwest Michigan (Elsner 1982). In greenhouse colonies green. viviparous females have been cbserved to produce red progeny. but red females were never observed giving birth to green young (Elsner pers. comm.). The seasonal life cycle of this aphid begins in late April or early May with the hatching of overwintering eggs laid in or near the base of a blueberry bush. This first, or fundatrix, generation consists entirely of apterous. viviparous females. The fundatrix, in turn, give birth to Iboth apterous and alate viviparous females. From June to August several such viviparous generations are produced. Most of the alate individuals are produced early in the growing season when the blueberry bush is growing actively. In late August and September the viviparous females begin giving birth to oviparous females and alate males that mate to produce overwintering eggs. 1.7 Thesis Objectives For several years the recommendations for controlling the spread of BBSSV have been (1) rouge out all bushes exhibiting BBSSV symptoms, (2) plant resistant cultivars, 9 (3) plant only certified stock, and (4) limit numbers of the aphid vector with chemical suppression. More recently, an aphid labeling study has indicated that mechanical harvesters can be very inmmutant in the dispersal of apterous aphids (Ramsdell et al. 1984). This study reiterates the important of harvester sanitation as a means of stopping the spread of BBSSV to uninfected fields. (hi the other hand, it is not possible to kill all of the aphids in a blueberry field with chemical suppression strategies. At certain times during the growing season it may be very difficult to even halt their increase. If ‘vector populations in BBSSV infected fields are to be controlled below some threshold level (eg. a minimum level for disease spread, or perhaps, aphid movement) more must be known about the mechanisms that control this aphid's population dynamics. To best utilize all mortality factors, chemical applications must be timed to augment rather than disrupt the actions of predators and parasites. Since many blueberry growers apply pesticides from the air we must also be aware of the within bush distribution of the aphid, and how such applications affect this distribution. For instance. pesticide applications directed against aphids may not be very effective if most of the aphids are at the base of the bush where residues from aerial applications are lowest. 131 light of these facts, several objectives providing more information for better aphid control were identified: 10 1. determine the phenologies of aphid predators and parasites; 2. determine the seasonal within bush distribution of the aphid and its natural enemies; 3. describe seasonal changes in aphid fecundity: and 4. determine the impact of rainfall as an aphid mortality agent. Research was directed towards designing and parameterizing a life-table that would describe the aphid's population dynamics and could be used to simulate a variety of control strategies. 1.8 Thesis Structure Due tx> the diverse nature of the topics contained in this thesis, the work is presented in chapters. Each chapter addresses a different facet of the research. These chapters include discussions of aphid fecundity and mortality, morph determination, spatial distribution, and phenologies of natural enemies. The computer modeling section, 131 turn, integrates information from the various biological studies to structure a variable life-table for I, pepperi. The thesis ends with general conclusions and suggestions for future research. CHAPTER 2. NATURAL ENEMIES 2.1 Introduction The first step in utilizing natural enemies in pest control, whether the insect is a direct or indirect pest, is identifying these biotic mortality agents. The second step is to understand, and to be able to pwedict, when given predators and parasites will be active in the field. Finally, the impact that individual species or species complexes have on pest density must be assessed. Both the vmnflc of Tuttle (1947) and Elsner (1982) provide information.cn1 the identity of potential natural enemies of I, pepperi. Table 2.1 lists potential predators and parasitoids identified by these two researchers. In his research, Elsner observed individuals from seven insect families attacking the blueberry aphid. Of these, coccinellids were important early in the season during the months of May and June. Larvae of flies in the families Syrphidae and Cecidomyiidae were both found to be effective aphid predators. Elsner found syrphid larvae to be common from June to September. Cecidomyiid larvae, while much less common, were found to be very effective predators during July and August. Chrysopids, hemerobiids, and anthocorids were found throughout much of the growing season, but they appeared to have little impact on aphid populations. Whalon and Elsner (1982) have suggested that these three families 11 12 Table 2.1 Potential predator and parasitoid species of I. pgppg51_in.xichigan. Coccinellidae Adalia Coleoptera Diptera Hemiptera Cecidomyiidae Syrphidae Anthocoridae HymenOptera Eulophidae Anatis Chilocorus Coccinella Coleomegilla Hippodomia Hyperaspis Aphidoletes Cartosyrphus Eristalis Melanostoma Mesogramma Metasyrphus Pipizella Platycheirus Sphaerophoria Syritta Syrphus Toxomerus Tropidea Orius Aphelinus Symphiesis Chrysopa Micromus bipunctata lS—punctata bivulnerus novemnotata sanguinea trifasciata maculata lfingi' convergens parentesis 13-punctata signatabinatgta Aphidomyza trista dimidiatus tenax obscurum marginata polita latifasciatus wiedemanni puchella erraticus quadratus cylindrica robusta pipens knabi ribesi geminatus quadrata insidiosus sp. bimacuIatipennis carnea oculata subticus 2,3 _————-——————-——--———-——_——————-—-———.—————_-——————~-———————- —-—-~-—--—-o.--——-“-—’.—-————————-—.——-—.———---——~————_-————‘_—— Neuroptera Chrysopidae Hemerobiidae 1 Unless Tuttle (1947). 2 Elsner (1982). otherwise noted, all 3 Whalon & Elsner (1982). species are cited from 13 may be limited by chemical control practices. Finally, Elsner observed hymenopterous parasitoids in the family Eulophidae frequently attacking aphids during July and August. Although Tuttle collected over 400 species of insects from 300 acres of highbush blueberry, he did not specifically note any of these as attacking aphids. Fortunately, most of the families identified by Elsner are composed of species that feed primarily, or solely, on aphids. For this reason all species collected by Tuttle from these seven families are included in the table as potential natural enemies. Undoubtedly, some of these were feeding (n1 other species of aphids in the neighboring habitat anui do not have any significant impact on populations of I, pepperi. The table also indicates that two large natural enemy complexes exist in the commercial highbush blueberry agro—ecosystem. These complexes are composed of predators in the families Coccinellidae and Syrphidae. Tables 2.2 anui 2.3 contain more detailed information from the literature concerning aphid predators and parasitoids. These tables list lower developmental temperature thresholds and generation times for natural enemies studied in other aphid ecosystems. Most of the parasitoids listed in Table 2.2 are in the ichneumonoid family Aphidiidae. Members of the families Ceraphronidae and Pteromalidae are hyperparasitoids. Notice Table 2.2 Lower developmental temperature thresholds and generation times for several species of aphids and their parasitoids. —————————-—¢o—————-———_—————.———-—.—-——————————————-——————————_ ———————-——————-—-—-—--—n———.n—_-—--—-——————————-———-———--——-—. Aphididae Acrythosiphon pisum Masonaphis maxima Illinoia pepperi Aphidiidae Aphidius ervi ervi Aphidius ervi pulcher Aphidius rubifolii Aphidius smithi Praon pequodorum Ceraphronidae Dendrocerus niger Pteromalidae Asaphes lucens Campbell Campbell Campbell Elsner, Campbell Campbell Campbell Campbell Campbell Campbell Campbell & Mackauer, 1975 & Mackauer, 1975 & Gutierrez, 1973 1982 & Mackauer, 19758 & Mackauer, I975 et al., 1974 & Mackauer, 1975 & Mackauer, 19758 & Mackauer, 197510 & Mackauer, 1975 _~-_—-~-_——-“——————_-—-‘—_-fl————-——~-*——--~-_—-—-“—~-n-—'--—“— ---——“-—-—--—————————-—-——————_—~-———--—.—-—-_-—-—-——-‘~———— 4 t = 5 K = 6 apterae 7 alatae 8 9 maxima. 10 Masonaphis maxima. lower deveIOpmental temperature threshold in OC. time from oviposition to F1 emergence in ODt. Parasitoid of the pea aphid, Acyrthosiphon pisum. Parasitoid of the thimbleberry aphid, Nbsonaphis Hyperparasitoid of both Acyrthosiphon pisum and 15 that time parasitoids listed have higher develOpmental temperature thresholds than their hosts. The table also indicates that all of the parasitoids studied have generation times of approximately 200 degree days. What the table does not show is that the adults are relatively long lived. For example, Gilbert and Gutierrez (1973) were able to keep Aphidius rubifolii Mackauer alive in the laboratory for cum; to two weeks when honey was provided as a carbohydrate food source. A, rubifolii is the primary parasitoid of Masonaphis (=Illinoia) maxima (Mason), an Table 2.3 Lower developmental temperature thresholds and generation times for several species of aphid predators. _———————————---—————-—-——-—-———_-————.—--——-————_--—--——d——- --—-----——--—-————---——-———---———-———_—-—-————-——-———---———- Taxa t11 K12 Citation Coccinellidae Adalia bipunctata 9.0 263 Obrycki & Tauber, 1981 6.8d Obrycki et al., 1983 Coccinella septempunctata 12.1 197 Obrycki & Tauber, 1981 Coccinella transversoguttata 12.2 218 Obrycki & Tauber, 1981 Coleomegilla maculata lengi 11.3 236 Obrycki & Tauber, 1978 13.8 199 Wright & Laing, 1978 Hippodamia convergens 9.0 Baumgaertner et al.. 1981 Chrysopidae Chrysopa carnea 8.3 Baumgaertner et al., 1981 4.4 170 Tauber & Tauber, 1973 Chrysopa harisii 10—14 314 Tauber & Tauber, 1974 Hemerdbiidae Hemerobius pacificus 4.4c, 0.4e, 4.11, 0.6p Neuenschwander, 1975 11 t = lower developmental temperature threshohd in 0C; e=eggs: l=larvae: p=pupae; c=complete development; d=post-diapause, pre-reproductive adults. 12 Time of oviposition to F1 emergence in OCt. l6 aphid closely related to I, pepperi. Frazer & Forbes (1968) reported that levels of parasitism from this species can reach 15%. The species also has the potential for very rapid rates of increase. Females lay 400 eggs in the laboratory and up to 90 eggs per female have been observed in the field (Gilbert & Gutierrez 1973, Gilbert et al. 1976). Potential fecundity for other species of aphidiid wasps ranges from a low of 30 to a high of over 1500 eggs per female (Hagen & Van den Bosch 1968). Two species listed in the table, A, smithi Sharma & Subba Rao and A, ervi ervi Haliday, are not native to North America. Both species were imported, mass reared, and released in 1958 in an attempt to control the pea aphid, Acyrothosiphon pisum (Mackauer & Finlayson 1967). Ibo general, primary parasitoids of monoecious aphids tend to be Species specific. Dioecious aphid species, on the other Tammi, often have different complexes of parasitoids on each of their host plants. Hyperparasitoids are usually not species specific. Aphid predators also tend to be generalists; feeding on mites, Lepidoptera larvae, and insect eggs in addition to a variety of aphid species (Hagen & Van den Bosch 1968). Several of the aphid predators listed in Table 2.3 have been observed in commercial blueberry fields. These species include Adalia bipunctata L., Coleomegilla maculata lengI Timberlake, Hippodamia convergens Guerin-Memeville, and Chrysopa carnea Stephens. Most of the predators listed in 17 the table lmnma lower developmental temperature thresholds much higher tfluni I. pepperi. Exceptions to this generalization include 9, carnea and Hemerobius pacificus. g, carnea overwinters as a pupa. Data from Ithica, NY, indicates that, at least in an apple ecosystem, chrysopid adults are one of the first predators to appear in the spring (Tauber & Tauber 1973). Laboratory studies revealed that post diapause females required only 100 degree days above 4°C to develop into reproductive adults. A, pacificus is a confer dwelling hemerobiid in the Pacific Northwest. Although the same families of predators appear time and time again in studies of different aphid ecosystems, the relative importance of each family is quite variable. For instance, while studying Myzus persicae (Sulz.) in potatoes Mack and Smilowitz (1980) found coccinellids to be the primary predator. On the other hand, syrphids were found to be the most important predator of M, maxima on thimbleberry (Gilbert & Gutierrez 1976). Both syrphids and cecidomyiids were important predators of the cabbage aphid, Brevicoryne brassicae (L.), on Brussels sprouts in a study conducted in southern England (George 1957). George also noted, with some surprise, that coccinellids were completely absent from this system. Zhi a later study of this ecosystem, Harris (1973) identified the cecidomyiid Aphidoletes aphidimyza (Rondani) as the predominant predator. During insecticide trials in highbush blueberries here in Michigan Whalon and 18 Elsner (1982) also found cecidomyiids to be the most common predator. The current study of I, pepperi's natural enemies has two objectives. The first is to determine the phenologies of the aphid's natural enemies with greater precision. The second objective is to continue the process of identifying I, pepperi's parasitoids. 2.2 Materials and Methods Rather than devise separate experiments to address questions concerning phenologies of natural enemies, within bush aphid distribution, and other life-table parameters; a systematic sampling scheme that could be analyzed in a variety of ways was developed. In later references this scheme will be referred to as the 'standard sample'. These studies were conducted at two commercial highbush blueberry, y, corymbosum cv. 'Jersey', sites in Michigan during tflua summers of 1981 and 1982. Data was collected during both years at site 1 near Charlotte, MI. At a second site near Grand Junction, MI, data was only collected in 1982. Plots were sampled every seven to ten days from early June until September. Site 1 was located on the Lowel Cook Farm, 3.5 miles southwestL4 0‘ d a Q E 1.1.! o = 7r 1" v 1 -' ; r - I I " 1 l 1000 1600 2000 2600 3000 3600. 4000 4600 Accumulated Degree Days Figure 2.5 Parasitoid summary for site 2, 1982: (a) daily rainfall”. (b) total aphids per sample, (C) aphid armies containing parasitoids, and (d) empty parasitoid cocoons. 30 synchronized development of a single generation of parasitoids. The rapid disappearance of parasitoid pupae in Figure 2.5c is most likely linked to the collapse of the aphid population. Adult aphidiid wasps require honeydew as a source of carbohydrate fuel (Hagen & Van den Bosch 1968, Gilbert & Gutierrez 1973). Therefore, as the aphid population declined the parasitoids were deprived of both suitable hosts and a food source. On the other hand, the peak in Figure 2.5d cannot be linked to this collapse or to the chemical applications. Since it charts cumulative adult emergence this curve should be sigmoid shaped not bell shaped. The only way to decrease the number of empty ;parasitoid cocoons lll‘bhe sample is to physically remove them from the leaves. 'In this case removal was the result of mechanical harvesting. Several broad generalizations may be drawn from these diagrams. First, the aphidiid species that pupates within the mummified aphid was observed in all plots throughout much of the growing season. In all cases, they first appeared around 1500 0D38 (June 15 to 21). Conversely, the aphidiid species that pupates beneath the aphid mummy was not observed until later in the growing season and then only for a limited time. At site 1 this second parasitoid species was only found on August 31 and September 8. At site 2 it was observed, in larger numbers, from July 27 to September 9. 31 The figures also indicate that empty parasitoid cocoons are a poor indicator of cumulative adult emergence. Counts of empty parasitoid cocoons declined two or three times during the growing season in all plots. With two exceptions (one of these was discussed earlier), all of these declines were associated with heavy rainfall between sampLes. In most cases these accumulations exceeded one inch of rain. Figure 2.6 charts percent parasitism through the growing season for each site. This rate of parasitism combines counts of both kinds of aphidiid parasitoids. Percent parasitism for each sample was calculated as follows: E2eber-e§_ee£e§i§e_eeeee Total number of aphids X 100 % Parasitism = This method of calculating the rate of parasitism is overly simplistic. Since not all aphid life stages are parasitized, the method will tend to underestimate the level of parasitism. One the other hand, since the aphid populations have a relatively stable age distribution (Kriegel & Whalon 1983) more complicated methods of estimation do not alter the general conclusions drawn here. These conclusions are two fold. First, the rate of parasitism increases through the growing season. A similar result was obtained by Gilbert & Gutierrez (1973) in the M, maxima system. Second, the rate of parasitism at the unsprayed plots was an order of nagnitude greater than at the sprayed site. Parasitism reached a high of 18% at 32 Figure 2.6 Percent parasitism versus degree days for (a) Bite 1, 1981: (b) site 1, 1982: and (c) site 2, 1982. Parasltlsm Percent 33 a '1 J 31 o 1 I I T r r I l 1000 I ‘00 RM 2500 3000 3500 4000 4500 g 1 d d 1a a 9. J o D i . °1ooo 2000 soon 4000 a ' T " I ‘ I f fl T T j 1000 1600 2000 2500 8000 3500 4000 0500 “7 . (C) 4 _.J J J .4 J a A I r I I I fi'r : r 1 1m 1‘00 2‘00 3000 3600 4000 4800 2000 Accumulated Degree Days 34 site 1 during 1981. This plot also contained the largest early season aphid population. At site 2 the rate of parasitism peaked at 2% during the collapse of the aphid population. The figures of 25% and 66% parasitism encountered at the end of the season at site 1 in 1982 (Figure 2.6b) are not reasonable figures. I believe these numbers represent diapausing parasitoid pupae remaining on the leaves after the aphid population had all but disappeared. 2.3.2 Predators Very little is known about the role that spiders play as direct predators in the highbush blueberry agro- ecosystenu. Direct evidence of their consuming aphids was only encountered once during this study -- a shriveled alate was discovered in a web. Most often, a single spider was the lone inhabitant of a blueberry terminal. On the other hand, crab spiders (Araneae: Thomisidae) were observed consuming lepidopterous larvae on several occasions. The present study is primarily concerned with when spiders and spider egg cases are present in blueberry fields. 'Fhese results are depicted graphically in Figure 2.7. No attempt was made to differentiate between different families or life stages of arachnids. In general, spiders ‘were present, though not abundant, throughout the growing season. More spiders were observed at site 1 than at 35 D "l (A) 0 Arochnld sp. + Egg cases o-a .. ’f‘ 1 I 1 II 1 I l I l I, ‘ I l -4 I ‘ ID I ‘ fun“. ‘ ‘ e I, ‘ I l l I / l ,’ o e c, e e , e c’, 4 ‘- I- , : 1 - I - q 1000 1 500 2000 2800 3000 3500 4000 4500 21 (B) O I) - I O. I, E 3 I/ a .. ,\ , (D " , / \ / I \ I o ,' \ / 0. ’1 \ l’ e I \ 1' I G u‘ ,\ [I \\ I 'U A ,’ \ -/ \ l - \ I \ I, \ I U) I \ l , \ I \ l ’ \ ’ \ I \e I \ I \ I A I \ I y i l 5 :1 3 i r t ' I 'I ‘ I l' °1000 1500 2000 2000 3000 3000 4000 4600 El (0) d ID -e ’’,e.\\ I;- l’ \ I, A A ’ I L 1 1 o : —‘ [v v 1 V I' I j 1 Y—f V v, 1 ‘ 1000 1500 2000 2600 3000 3500 4000 4600 Accumulated Degree Days Figure 2.7 Sample counts of spiders and spider egg cases for (a) site "1, 1981: 0)) site 1, 1982: and (c) site 2, 1982. 35 a '7 (A) o Arcchmd sp. + Egg cases 2* I l I, ‘ I \ I \ I \ I | I, ‘ I, “ u: , \‘ lr--_. ‘ I “ o I I I \ I / ‘1 ’ O I A A A A A 4”4‘\\‘ I A L A A ‘ 4 a V V1 V V 1 V V 1 V 1V 1 V 1 V V V '1 1000 1500 2000 2500 3000 3 500 4000 4500 ’3‘ (B) O I' '5. 1’ s fi I a a \ / .01 I \ I (D I \ [I I \ l b I \ ,’ O I, \ ’l \ O. i \l, a J [I \ l, O p I \ ’l \\ 1 2 I» I \ ,’ \ ,' a I \ I \ I \ I I \ K I \ I \ U) I \ I \ I \ I \ 1 ’I \ I \ , Y \ I \ ’ \ I I ’ \ ’ \ I X I \ I \ I A, - \‘I A A‘ A o V 1 V V1 ' V 1 ' V 1 '1 ' 1 1' 1W0 1500 2000 2500 3000 3500 4000 4500 37 (C) 1 .J 00 4 ,l.\ r—c I” \ ’1 o V —f rV V 1 V a 1 V V 1 'L T 1 1000 1500 2000 2500 3000 4000 4500 Accumulated Degree Days Figure 2.7 Sample counts of spiders and spider egg cases for ‘(a) site "1. 1981: (1:) site 1. 1982: and (e) site 2. 1982. 36 site 2. Although one egg mass was found in early July, most egg laying appeared to take place in August and September. Surprisingly few coccinellids were observed during the course of this study. Table 2.4 indicates that all such observations were of isolated adults at site 1. Although larvae were occasionally seen in the fields visited, none were observed during the course of the 'standard sample'. One of the adults collected belongs to a species not previously recorded from highbush blueberries. This new member of the already large complex of predatory coccinellids is Brachyacantha ursina (F.). Table 2.4 Observations of coccinellid predators during the 'standard sample'. —————-——————————————————————a-——————————————.————— ---------_-_-—-—--—--—----—---———--—-——---—--——-- Site Date OD Life stage No. Observed 1 JUN-15-81 1487 adults 1 1 JUN-27-82 1686 adults 2 l JUL-06-82 1943 ' adults 2 -————--fl-’~—_—--——-————~-———~-_——-.-——_~————~——~~- .—_—————-—--——-—-—---———------——--————_————--——-— Predatory neuropterans belonging to the families Hemerobiidae and Chrysopidae were collected during the course of this study. Hemerobiids were quite rare and were not collected during the 'standard sample'. However, they and chrysopid larvae were the most abundant predators present in a mechanical harvester at site 2 on August 19, 1982. Sample counts of chrysopids collected during the 'standard sample' are presented in Figure 2.8. Eggs were the most commonly collected life stage. This observation is 37 in accordance with Hagen & Van den Bosch's (1968) review of aphid natural enemies. In their review the authors state that eggs and adults are the most commonly encountered life stages. Indeed, adult chrysopids were also commonly observed in blueberry fields. However, they were normally observed in flight and therefore appear only rarely in 'standard sample' counts. It is very interesting that so few chrysopids were observed at site 2 during 1982 (Figure 2.8c). The reasons why this is so cannot be stated with any certainty. Perhaps they were excluded by chemical suppression strategies practiced throughout the entire 400+ acre clearing. Adult chrysopids are known to be much less resistant to pesticides than their larvae (Bartlett 1964). But why then was such a flush of egg laying observed around 3000 0D? Hagen & van den Bosch (1968) note that fecundity of g, carnea is dependent on the amount of honeydew available to adults. They state that very high aphid densities are required to attract the species and induce egg laying. Indeed, by 3000 0D the number of 1;. pepperi exceeded 3000 aphids per 225 terminals and honeydew covered much of the foliage. Since the fields were approaching harvest, pesticide applications had also»ceased. Therefore, I believe this flush of eggs were produced by immigrant chrysopid adults attracted to the field by large quantities of honeydew on the foliage. 38 T (A) E99 Larva + 0 X A. ,4 °1000 9'“ (B) 0 - ‘s‘ G (D 24 L- O O. O E O. a“ O 0) >4 5- .fl 4‘ o A ‘ - -_ Al”" \\- - o 4%: 1 V V1 ‘V V 1 V 1' 1 V’ 1V1 V ”V1 1 ‘F 1000 1500 2000 2500 3000 3500 4000 4500 c T T l 1000 1500 2000 2500 4000 4500 Accumulated Degree Days Figure 2.8 Sample counts of chrysopid predators for (a) site 1. 1981: (b) site 1.1982: and (c) site 2. 1982. ‘ 39 All dipterans collected during this study are members of the families Cecidomyiidae and Syrphidae. Cecidomyiid larvae were observed only rarely. In fact, during the two years of sampling only three blueberry terminals with cecidomyiid larvae were found. The details of these observations are presented in Table 2.5 Table 2.5 Observations of cecidomyiid predators during the 'standard sample'. Site Date OD Life stage No. Observed 1 JUL-27-81 2842 eggs 1 l SEP-08-82 3811 larvae 1 2 AUG-10-82 2047 larvae 4 Conversely, syrphids were by far the most common aphid predator observed. Sample counts for all syrphid life stages are presented in Figure 2.9. As with some earlier predator species. eggs were the predominant life stage collected in the sample. The only sizeable larval samples were obtained in 1982 at Site 2 (Figure 2.9c). This site also contained aphid counts five to ten times larger than any other plot. In general, both egg and larval counts in Figure 2.9c follow the same pattern. Also, the sharp decline in syrphid numbers between 2500 and 3000 degree days coincides with the applications of Cythion discussed earlier. Syrphlds Per Sample 40 o “l (A) 2d 2.1 d to °1000 3? Bu gut ”d °1000 *‘ (c) o-l w . 9 3 I,\\ x""""\ , \ ’l’ \\ [I \ I, \\ [I \\ II \\ I \ I O \ o‘ ‘- .1: 4: 14: : fir "' “—fi , 1 1000 1500 2000 2500 3000 3500 4000 4500 Accumulated Degree Days Figure 2.9 Sample counts of syrphid predators for (a) site ‘1. 19812113) site 1. 1982: and (c) site 2. 1982. 41 2.4 Discussion 2.4.1 Entomophthora One surprising result of this study was the complete absence of fungal pathogens. Earlier studies using caged bushes reported significant epizootic outbreaks (Elsner 1982, Morimoto pers. comm.). There are two simple reasons why caging bushes may lead to such outbreaks. First. Entomophthora sp. require high relative humidity for spore germination. Second. according to Hagen & Van den Bosch (1968). epizootic outbreaks appear to be a density dependent mortality factor. Caging blueberry bushes encourages both of these conditions. The results of this study suggest that Entomophthora sp. are not normally a significant mortality factor in the commercial highbush blueberry agro-ecosystem. 2.4.2 Sampling method The sampling method used in this study was more successful at capturing some natural enemies than others. Such problems are difficult to avoid when attempting to survey a large number of insect species and life stages with a single sample method. Since it was essentially a static, visual sample, the method was most successful at sampling relatively slow or immobile creatures such as aphids and insect eggs or pupae. The method did not prove useful for sampling highly mobile organisms or life stages. The latter included adults of most natural enemies and larvae of 42 Coccinellidae and Neuroptera. The difficulties in sampling coccinellids, in particular, has been studied by several authors (Frazer & Gilbert 1976, Gilbert et a1. 1976, Mack & Smilowitz 1980). At present it is unclear exactly how useful the method was in capturing larval Diptera. Substantial numbers of syrphid larvae were only obtained when aphid numbers exceeded 2000 per sample. However, at such high aphid densities the temporal patterns of egg and larval counts ‘were very similar. Either (I) mortality is very high for first instar syrphid larvae, or (2) the sample cannot detect small populations of these larvae. Harris (1973) found that a visual sample was a very poor estimator of the numbers of cecidomyiid larvae in a field of Brussels sprouts. Compared to fixing a leaf sample in 70% EtOH, sieving, and examining under a stereo microscope; a visual examination of the leaves only identified 11% of the larvae present. 2.4.3 Parasitoids Several interesting conclusions regarding ;, pepperi's parasitoids can be drawn from this study: (1) Parasitoids first appear in the field at approximately 1500 OD38‘ (2) This parasitoid complex consists of at least two primary parasitoids and one hyperparasitoid. An 43 additional species was observed to attack 1, pepperi in the greenhouse. (3) Samples of empty aphid mummies are a poor indicator of cumulative adult parasitoid emergence. (4) Although parasitoids were a significant mortality factor (lo-18%) in a highbush blueberry field not subjected to regular chemical suppression strategies, the rate of parasitism remained below 2% in a chemically treated field. If these estimates of percent parasitism are at all reliable, parasitoids are not a significant mortality factor in most commercial highbush blueberry operations. Before proceeding udJflu the deve10pment of a parasitoid sub-model for the aphid simulation program it would be appropriate to obtain a better assessment of the importance of aphid parasitoids in Michigan's commercial highbush blueberry agro-ecosystem. Development of such a sub-model would require that (1) cultures of the major parasitoids be established, and (2) fixed temperature rearing experiments be conducted to determine their lower developmental temperature thresholds and other life-table parameters. On the other hand. a field experiment to asses the importance of aphid parasitoids would be both quicker and less expensive. Such an experiment would also yield additional information.cn1 the identity of these parasitoids. It is 44 'very possible that some parasitoid species present in the system were not collected during the above experiment because (1) only two sites were surveyed, (2) they are relatively rare, or (3) they do not affix mummified aphids to blueberry leaves. This assessment could be carried out most economically using an approach modified from Messenger & Force (1963): (I) collect large nymphs of ;, pepperi from several locations, (2) rear aphids in clip cages in the greenhouse, (3) place parasite pupae, ie. aphid mummies, in gelatin capsules until adult emergence, and (4) calculate percent parasitism and identify adult parasitoids. Such a sample should be conducted four to five time during the growing season beginning around 1500 OD38‘ 2.4.4 Predators In general, the phenologies of aphid predators were not consistent among either sites or years. At site 1 chrysopids and syrphids were the most commonly collected predators. Isolated coccinellid adults were also collected during the first half of the growing season. At site 2 syrphids appeared to be the primary aphid predator. Although a previous study found cecidomyiid larvae to be very important, they were very rare in this study. Just how important each of these predators is in reducing aphid numbers is difficult to say. This study suggests that chrysopids and syrphids are the most common predators in the system. Both families also have a high per 45 capita consumption rate of aphids. Chrys0pid larvae require 200 tn) 500 aphids to complete their deve10pment. Syrphid larvae require from 400 tol800 aphids to complete theirs (Hagen & Van den Bosch 1968). This means that the 25 syrphid larvae observed at site 2 on August 10, 1982, ate ten to twenty thousand aphids during the course of their development! Also, remember that only two species of chrysopids have been reported from highbush blueberries. On the other hand, nineteen species of syrphids have been collected there. To accurately model the population dynamics of £3 pepperi it will be necessary to develop methods to account for losses inflicted by these two families of predators. CHAPTER 3. RANGE AID SPATIAL DISTRIBUTION 3.1 Introduction This chapter addresses two important questions regarding the distribution of the blueberry aphid, L. pepperi. First, where in Michigan has the aphid been found? Second, how does the within bush distribution of this aphid and its natural enemies change through the growing season? 3.2 Geographical Distribution Since L. pepperi only infests Vaccinium species, the aphid's range must be a subset of its hosts' ranges. Four blueberry species are found in Michigan: 2, corymbosum L., ‘1. aggustifolium Ait., Z, myrtilloides Michx., and Y, vacillans Torr. (Marvin Pritts pers. comm.). Both cultivated and wild stands of the common highbush species, 2, corymbosum, may be found throughout the lower half of Michigan's lower peninsula. About 12,000 acres of y, corymbosum are currently under cultivation in the state, with over 90% of this acreage located in southwest Michigan near Lake NUchigan (J. Nelson pers. comm.). Late lowbush blueberry, Z, angustifolium, is a native species found throughout the upper peninsula and the upper half of the lower peninsula. Ihn some regions of the upper peninsula this blueberry species is the dominant ground cover in coniferous forests and logged areas. 1, pepperi feeds on both of these blueberry species and could therefore 46 47 potentially be found anywhere in the state. £3 pepperi has only been collected from \_I_. myrtilloides on one occasion (E. Elsner pers. comm.). The blueberry aphid has not yet been observed on Y, vacillans in Michigan. Although ;, pepperi was not confirmed as the vector of blueberry shoestring until 1979, the aphid was a suspected vector of the disease for many years. As early as 1963 the Michigan Blueberry Grower's Association was encouraging research to identify the BBSSV disease vector (Burger 1966). One result of this research was the first distributional study of aphids in blueberry fields. This study was conducted.cn1 ten highbush blueberry farms in southwestern Michigan in Berrien, Van Buren, Allegan, Ottawa, and Muskegan counties. Burger found only two aphid species widely distributed in blueberry fields, Masonaphis (=Illinoia) pepperi and Myzus scammelli (Mason). I. e eri was collected at all ten sites. While Burger was surveying blueberry fields for virus vectors, Francis Giles (1966) was investigating the geographic distribution of aphid species infesting small fruits in Michigan's lower peninsula. Giles also found ;, pepperi to be the prevalent species of aphid in commercial blueberry plantings. Except for one observation in Montcalm County, Giles only found 1, e eri in southwest Michigan near Lake Michigan. Also, he did not report L, pepperi from wild Vaccinium species. Giles' dissertation included a map showing the distribution of L. 48 pepperi and M, scammelli in Michigan's lower peninsuLa. Unfortunatelyn In: documentation on the map's construction was included with the work. However, 16 specimens from these two studies were deposited in MSU's Department of EntomOIOgy museum. Since these studies, 1, pepperi has been investigated in greater detail. The aphid has been collected from wild Y_. mustifolium in both the lower and upper peninsulas. During this time blueberry acreage in the state has continued to expand. With these facts in mind it seems appropriate to compile a new list of localities where ;, pepperi has been collected (Appendix 2) and construct a new distribution map from this data (Figure 3.1). Data for this map comes from several sources; (1) specimens deposited in the Department of Entomology museum at Michigan State University, (2) ;, pepperi study sites used by faculty or graduate students at MSU, and (3) wild Vaccinium sites surveyed by Dr.'s Jim Hancock and Mark Whalon. The distribution map shown in Figure 3.1 includes all documented sites where l, pepperi has been collected. Vaccinium host species are also indicated. Adthough £- pepperi has only been collected from 15 of Michigan's 82 counties, tflua locations of these sitings suggest that the species is found on Vaccinium species throughout the state. 49 "“1 5 I rJ—-—-T- o cult. 1: corymbosum 0 wild y_._ corymbosum I wild ll; angustifolium A wild L myrtilloides F1gure 3.1 Distribution of 1. pepperi on wild and Cultivated Vaccinium species in Michigan. 50 3.3 Within Bush Distribution The remainder of this chapter is devoted to an investigation of the vertical distribution of ;, pepperi and its parasitoids within cultivated highbush blueberry bushes. There are several reasons why it is important to understand how this distribution changes through the growing season. One reason, the impact of vertical distribution on aphid survival following aerial pesticide applications, was already discussed in the introductory chapter. But there is another, less obvious reason. Changes in vertical distribution must be accounted for in aphid sampling schemes. During some parts of the year it may be necessary to sample some portions of the blueberry bush more intensively than others. Uneven distribution may even make it necessary to devise different sampling plans for different times of the year. It would also be useful to know if the vertical distribution of the aphid's natural enemies tracks changes in the aphid's vertical distribution. 3.3.1 Materials and.nethods Data for this study was collected as part of the 'standard sample' described in Section 2.2. Blueberry bushes were partitioned roughly into thirds. One terminal was sampled from the upper, middle, and lower portions of each of 25 bushes in each of three rows. 51 3.3.2 Results Results for the aphid portion of the vertical distribution study are shown in Figure 3.2. Samples are plotted as percent of the population in a given third of the bush versus degree days. Such graphs can be misleading when the number of aphids actually collected in a sample becomes small. In most samples hundreds of aphids were collected; however, in a few cases aphid populations were almost non- existent. Less than 10 aphids were collected in a sample of 225 terminals at the following times: Site 1, 1982 at 4519 0D; and Site 2, 1982 at 1253, 3461, and 3777 0D. The last two dates cited occurred during the collapse of the aphid population following pesticide applications. The figure indicates that early in the season most aphids were found in the bottom third of the bush. In general, vertical distribution followed a trend of increasing percentages of aphids in the upper third of the bush as the growing season progressed. The middle third of the bush consistently contained the smallest share of the aphid population. A temporary increase in the proportion of aphids in the bottom third of the bush was observed around 3500°D in two of the three plots. Table 3.1 lists summary percentages of vertical aphid distribution in the three plots. 52 ‘3' (A) Upper Middle Percent of the Aphid Population in a Given Third of the Bush 2000 2500 3000 3500 ‘500 Accumulated Degree Days Figure 3.2 Vertical distribution of ]_Z. pepperi within Mahbush blueberry bushes at (a) site 1, 1981: 0)) site 1, 19823 and (c) site 2, 1982. 53 Table 3.1. Overall percentages of aphids in each third of the bush for samples with.N > 20 aphids. --——-——---—--——--——--—-———--—-——-—_*-—-—‘—————-—‘ -----—--——------——-—---—-----———--——-——-————--—-- Third of Total aphids Percentage Site Year bush collected of total 1 1981 upper 853 32.5 middle 726 27.6 lower 1048 39.3 1 1982 upper 683 24.6 middle 802 28.9 lower 1292 46.5 2 1982 upper 3917 33.0 middle 4381 36.9 lower 3601 30.1 Results for the parasitoid portion of this study are shown in Figures 3.3 and 3.4. Frequently, only a few parasitoids were collected in a sample. Due to small sample sizes, graphs of percent parasitoids in a given third of the bush would be very misleading. For this reason the parasitoids' vertical distribution have been plotted in the untransformed unit -- parasitoids per sample. Sample counts for the parasitoid that pupates within the aphid mummy are shown in Figure 3.3. The left side of the figure (3.3a,b,c) shows the vertical distribution of parasitoid pupae. The ride side (3.3d,e,f) charts the vertical distribution of empty parasitoid cocoons. Similar diagrams for the parasitoid the pupates beneath the aphid mummy are shown in Figure 3.4. Since the presence of two kinds of primary parasitoids was not detected until 1982, Figures 3.3a and 3.3d contain counts of both parasitoid species. 54 can. “$3 .u my: E6. can 2824 when 3.8 .33 .H 3? 3:3 um. EB Show 93 n3»? gamma and» 33333 333mm 93 no 33333:. 1.63am; n.n 0.33m 930 69500 no.2:an been 0< 88 sown OOON - 8v usng eq; 30 pm; HOME) 8 ug suoocoo dea 38n— ootooa p232§oo< coon aoow sate... . .53: D .23: n— A 0 r A9 coo. coon _ lb g e 000v coon ooou occ~ % ’ l— ’ O! o qsna em 10 mm; uengg a u! esdnd pgoussmd coon coo— P 330.. . S .62: D .025 D qsna em go pm; using 8 ul suooooo K1dw3 Amy 9: 56 The most striking aspect of these graphs is the abundance of parasitoids in the bottom third of the bush. Table 3.2 shows that 50% or more of the primary parasitoid pupae were found in the bottom third of the blueberry bush, while 20% or less of the pupae were collected from the upper third of the bush. Table 3.2. Overall percentages of parasitoid pupae collected in each third of the blueberry bush. Third of Total aphids Percentage Site Year bush collected of total 1 1981 upper 24 17.4 middle 46 33.3 lower 68 49.3 1 1982 upper 10 11.9 middle 20 23.8 lower 54 64.3 2 1982 upper 8 20.0 middle 8 20.0 lower 24 60.0 3.3.3 Discussion Seasonal changes in the vertical distrflnujon of l, pepperi documented in this study agree well with previous knowledge of the aphid's life history. Since fundatrhc individuals hatch from eggs in leaf litter or in the crown of the bush, one would expect to find most of the aphids in the bottom third of the bush early in the growing season. As the season progresses aphids become distributed throughout the bush on actively growing terminals. Since most of these terminals are located in the upper portion of 57 the bush, distribution slowly shifts in favor of aphids being found in the top of the bush. After berries have been harvested the bush experiences a flush of new growth. This late season growth is most noticeable in the crown of the bush. A corresponding increase in the proportion of aphids in the lower third of the bush was observed at this time. The aphid's parasitoids, on the other hand, display no reCOgnizable shift in vertical distribution through the growing season. Perasitoid pupae are found in increasing numbers as one moves lower in the bush. Usually, more than half of the pupae collected were found in the bottom third of the bush. These small wasps are not known to be particularly strong fliers. In the M, maxima system, Gilbert & Gutierrez (1973) found different rates of parasitimm in patches of thimbleberry separated by as little as 50 meters. From this observation tflua authors concluded that individual parasitoids do not fly very far. The present study suggests that parasitoids are also less effective in the upper portions of blueberry bushes. One of the following hypotheses seems most likely. Either (1) adult parasitoids spend a greater amount of time searching the lower reaches of the bush, or (2) parasitoids that venture into and above the bush canopy run the risk of dispersal by strong breezes. CHAPTER 4. APHID GROWTH AND.HATURATION 4.1 Materials and fisthods To parameterize the aphid maturation portion of the model the number of degree days required to complete each life stage must be determined. The mean and variance figures needed to parameterize the model's distributed delays were obtained by re-analyzing Elsner's (1982) fixed temperature rearing experiment. Briefly, Elsner's study was a fixed temperature cohort experiment conducted at six different temperatures (5, 1o, 17, 23, 26, and 29°C). Aphids were reared individually in small vials on leaf discs floating atop a nutrient solution. Molts and young produced were recorded daily. As they were observed, young were removed to eliminate any complicating density dependent reductions in fecundity. Exuviae were also removed. The present analysis treats the rearing experiment as a randomized block design with life stage as the treatment variable and temperature as the block variable (after all, temperature is the gradient here). Statistics were done using the SPSS software package. In the course of this analysis, a serious source of experimental bias was discovered. Since the bias will affect the final form of the analysis, it is considered in detail at this time. 58 59 4.2 Experimental Bias Over 90% of the aphids in the experiment died prematurely by drowning. Premature death is readily identifiable during the nymphal stages. Obviously, an aphid that molts has successfully reached the next instar. An aphid that drowns has not been so successful. On the other hand, premature deaths during the adult life stage are more difficult to interpret -- herein lies the rub. The severity of this drowning bias is most apparent if survivorship curves for reproductiwe adults are examined. If the experiment was completely unbiased one would expect to obtain very similar curves when survivorship is plotted against accumulated degree days for each temperature block. This expectation is based on the assumption that degree days are an accurate measure of physiological time. Since that assumption breaks down at temperatures near developmental thresholds, a significantly different curve would not be surprising for the 5°C run. Additional knowledge about aphid movement yields yet a third possibility. Both field and greenhouse observations indicate that l, pepperi is more likely to move the higher the temperature. If drowning is the result of temperature induced movement, aphids reared at higher temperatures should drown more frequently than aphids reared at lower temperatures. Unfortunately, when survivorship of reproductive adults is plotted against accumulated degree days since first birth (Figure 4.1a) the resulting curves do not agree with any of 60 these hypotheses. Survivorship curves are different for each temperature block. Also, aphids lived pmogressively longer at each higher temperature. What could have caused this unexpected mortality trend? The answer to this question begins with another graph of survivorship. When percent survivorship is plotted against days since first birth (Figure 4.1b) an interesting pattern emerges. Aphid mortality in the experiment was a function of calendar time not degree days. Remember, the experiment was sampled daily and any young were removed to eliminate cxmmdicating density dependent reductions in fecundity. Similarly, during the nymphal instars daily sampling included removal of exuviae. In this light premature death can be viewed as a function of sample interval. Every time a vial was sampled there was a probability that the sampling process itself would perturb the aphid enough to cause it to move. Movement, in turn, was a risk involving a fixed probability of drowning. In effect, a sampling technique designed to be non—destructive was actually quite destructive. In the analysis detailed below, only those aphids that lived long enough to reproduce were included. By placing this restriction on the data, any bias due to premature deaths could at least be eliminated from results for the nymphal instars. Unfortunately, this criteria also results Percent Survivorship Figure 4.1 Survivorship of reproductive 2_[_. (a) physiological tine, and (b) calendar tine. 61 (A0 °o i 700 250 350 T430 ' silo _ Two Degree Days Spent in Reproductive Stage §< \b -‘\\b——b-A (B) J ‘\ ‘\ 34 \ \\ \\ " ‘ ‘b—A ‘P-‘\ o-l ‘ ‘-'\ \ \ -_ ‘\ Y ' 1 25 30 .5 "'5 23 Days Spent in Reproductive Stage verse—ti versus 62 in the exclusion of the two highest temperature blocks, 26 and 29°C. 4.3 Results First, an analysis of variance was conducted to determine the number of degree days required to complete each instar. The ANOVA summary shown in Table 4.1 indicates that the length of the aphid's first three instars is approximately equal. The length of the fourth instar, in turn, is significantly longer (P:0.05) than any of the earlier stages. This result agrees with similar data for l. maxima (Campbell et al. 1974, Gilbert et al. 1976). Table 4.1 ANOVA summary of fixed temperature rearing experiment for four nymphal instars of I, 232225; reared at 5, 10, 17, and 23°C. Instar mean S.E. 95% conf. int. for mean first 33.34 1.7448 29.86 < x < 36.83 second 37.46 1.4082 34.65 < x < 40.28 third 41.27 1.4979 38.28 < x < 44.27 fourth 53.70 2.3824 48.95 < x < 58.46 The next step was to decide how to convert laboratory data containing four nymphal instars to the three nymphal age classes reported in field samples. I believe the discrepancy between this experiment and field observations lies in the inability to distinguish between the first three nymphal instars. As a result, second instar nymphs were sometimes classified as small nymphs and sometimes as medium nymphs iJl the field. Therefore, the analysis of variance was conducted again after the data for each aphid 63 (designated [i]) was transformed from four to three nymphal stages as follows: Small[i] First[i] + Second[i] / 2.0 Medium[i] Third[i] + Second[i] / 2.0 Fourth[i] Large[i] Analysis of variance for all aphid life stages following this transformation is summarized in Table 4.2. The highly significant F values in this table indicate that different degree day accumulations were required to complete a life stage at different temperatures. Examination of results from Scheffe's multiple range test (P10.05) reveals that the problem lies in the 5°C temperature block. This temperature consistently produced the lowest mean values for degree days spent in a stage. A previous analysis of this experiment (Elsner 1982) indicated that the lower developmental threshold temperature for I, pepperi is approximately 3.4°C. Consequently, the consistently low estimates for mean deve10pment time obtained at 5°C are probably the result of being in the non—linear portion of the developmental rate curve. Therefore this block was removed and the ANOVA conducted yet another time for 10, 17, and 23°C. Table 4 . 2 AHOVA experiment for all 10, 17, and 23°C. 64 summary of fixed temperature rearing life stages of L. m reared at 5, 2.501 15.076 F 95% confidence interval 4.179*:* 48.06 x 56.09 25.286** 56.17 x 63.84 5.518 48.95 x 58.46 *** 11.359*** 50.69 x 61.44 19.389*** 129.12 x 192.95 16.805 186.60 x 246.84 Stage mean nymphs small 52.08 medium 60.01 large3 53.70 adults pre-repro 55.68 repro 161.04 total 216.72 Table 4 . 3 ANOVA experiment for all 17, and 23°C. summary of fixed temperature rearing life stages of L, 23223;; reared at 10, F 95% confidence interval 2.164*** 49.77 x 58.07 15.019 59.46 x 66.55 0.127 52.02 x 61.45 7 *** 15.933*** 50.35 x 61.44 16.748*** 147.15 x 212.25 12.569 205.38 x 265.80 Stage mean S.E. nymphs small 53.92 2.073 medium 63.00 1.771 large 56.74 2.355 adults pre-repro 55.89 2.769 repro 179.70 16.255 total 235.59 15.086 13 pre-repro = repro = total = it P < 0.01 *** r P < 0.001 Abbreviations for adult life stages: pre-reproductive adults reproductive and post reproductive adults complete adult life stage 65 Elimination of the 5°C temperature block improves the results markedly (Table 4.3). There are now no apparent block differences for small and large aphid nymphs. The significant F statistic for medium nymphs may be partly due to the regrouping of data. Notice that block differences are still very significant (P<0.001) for the adult life stages. In fact, all three adult categories also fail Bartlett's Box F test for homogeneity of error (PLOTS). The statistical difficulties associated with the adult stages are nmet likely due to the drowning bias discussed earlier. 4.4 Discussion The drowning bias encountered in this experiment is very serious. The only data available for parameterizing the distributed delays simulating aphid growth and reproduction comes from this fixed temperature rearing experiment. Yet, parameters for the adult life stage (representing over 50% of the aphid's lifetime) obtained from the experiment are dubious at best. The length of the adult stage, presently estimated at 250 °D, could be 350 or possibly even 400 (EL. Errors in estimating this parameter will result in error to all model parameters subject to 'tinkering'. Two aspects of the model affected by such error will be the aphid's potential rate of increase and the impact of biotic mortality factors. 66 The fixed temperature cohort study must be conducted again. Individual aphids could be caged on live plants to eliminate drowning. Since the nutritional quality of individual plants can vary greatly, and relatively few plants will be used in the experiment: nmesures must be taken to ensure that the nutritional status of the bushes is as uniform as possible. To accomplish this, host phenology should be synchronized by forcing bushes through a dormant period. Bushes should also be replaced at regular intervals. Once the experiment has been completed, it will be necessary to re—parameterize the aphid growth portion of the model, fit it to field data, and test it against an independent data set. Until these measures have been completed only limited trust should be placed in the simulation. CHAPTER 5. FECUNDITY 5.1 Introduction Determining a specie's fecundity is a central component of any effort to understand or predict that organism's population dynamics. Often, a Specific number is associated with a birth rate for a given organism. But an insect's reproductive rate is often not a constant but a function of environmental factors. Aphid fecundity in particular is known to be affected by many factors such as crowding (Johnson 1965, Shaw 1970, Dixon 1974, 1975), nutrition (Johnson 1966, Mittler & Sutherland 1969, Dixon 1974, Dixon & Dharma 1980, Wellings et al. 1980), time of year (Way & Banks 1968, Dixon 1975), morph (Dixon 1976a, Wratten 1977, Dixon & Dharma 1980), and size (Dixon & Wratten 1971, Wratten 1977). To date, several methods have been used to estimate aphid fecundity in the field. The most common methods employed are (1) cohort studies on caged aphids (Way & Banks 1968, Dixon 1975), (2) counting ovarioles (Dixon & Dharma 1980, Wellings et al. 1980), and (3) counting aphid embryos (Gilbert & Gutierrez 1976, Wratten 1977). For several aphid species teneral adult weight has been shown to be a very good predictor of fecundity (Dixon 1970, 1976a, Dixon & Wratten 1971, Taylor 1975, Wratten 1977, Dixon & Dharma 1980, Wellings et al. 1980). For example, Wratten (1977) used relationships involving adult weight within 24 hours of 67 68 ecdysis auni embryo compliment to predict reproductive potential of apterous and alate morphs of the aphids Sitobion avenae F. and Metopolophium dirhodom Wlk. Conversely, ovariole number alone has been found to be a poor predictor of potential fecundity for several aphid species (Dixon & Dharma 1980, Wellings et al. 1980). Although seasonal changes in ovariole number are very predictable, the number of embryos per ovariole is highly variable. Both of these studies indicate that ovariole number appears to be a programmed feature of aphid life cycles and is not a function of either food quality or adult weight. 5.2 Materials and Methods Data for these fecundity studies come from two sources. First, data collected by Elsner (1982) was re-analyzed to determine how aphid births are distributed over a stem mother's adult life stage. Methods used in that experiment were summarized in Section 4.1. Since no young were produced at the two highest temperatures used in the experiment (26 and 29°C), the current analysis only incorporates data from the lower four temperature treatments (5, 1o, 17, and 23°C). Data for the seasonal aphid fecundity study was collected during the summer of 1982 at the same two sites where the 'standard sample' was conducted (Section 2.2). In this study individual stem mothers found during the course 69 of the 'standard sample' alone or with a group of young were collected and the number of young in the colony was recorded. Stem mothers were then carefully dissected to determine the numbers of deve10ped and undeveloped embryos they contained. Developed embryos were defined as those with pigmented eyes (cf. Gilbert & Gutierrez, Wratten 1977). Apterous and alate aphids were recorded separately. Statistically, this study was designed as a two by two factorial experiment with two sites and two aphid morphs. In a similar study, Gilbert 8 Gutierrez (1976) estimated aphid fecundity as the sum of the young in a colony plus the number of developed embryos. Wratten (1977) however, indicates that developed embryos only represent births that will occur in the next 24 to 48 hours. To determine which of these hypotheses is correct fecundity was calculated using both methods. For convenience the methods will be referred to as 'upper' and '1ower' bounds for seasonal fecundity. The lower bound was estimated as the regression line for young in a colony plus developed embryos, while the upper bound was defined by the regression line for young in the colony plus all embryos. Analysis of variance was done using tflua SPSS software package. Regression analysis was conducted using SPOCS. 70 5.3 Results 5.3.1 Frequency Distribution of Births One consequence of using distributed delays to simulate aphid growth (details of this process will be discussed in Section 8.3.4) is that the computer model requires a discrete frequency distribution describing how fecundity varies during an aphid's reproductive period. tutimately, this distribution must take the form of a.tdstogram conforming to the following three criteria. First, the time (or X) axis must be in physiological time units, ie. degree days, where xmax - xmin = mean time spent in reproductive stage (5.1) In practice, it is often easiest to set xmin equal to 0. Second, the histogram is divided into K equal cells where Cell Width = mean time spent in reproductive stage (5.2) K in this case refers to the order of the distributed delay. Finally, the y value of each cell, Yi' corresponds to the proportion of births that occur during that segment of the reproductive period. It follows by definition that 71 K 2 Y1 = 1.0 (5.3) To date, aphid fecundity data from the fixed temperature rearing experiment conducted by Elsner (1982) has only been presented in the form of mean young per day versus degree days after molt to adult instar. In this format different temperature treatments cannot be compared. Since insect development rate is different at different temperatures, days after molt is not a uniform measure of time. Therefore, the first step in this analysis was to convert calendar time to degree days using a lower deve10pmental threshold of 38°F (Elsner 1982). Figure 5.1 contains graphs of the number of births per aphid per day versus accumulated degree days since adult molt for the four temperature treatments in which young were produced. It is important to note that since mortality occurs throughout the adult life stage, births per day must be reported on a per aphid basis. One consequence of this strategy is that when the sample size becomes very small a single birth can dramatically alter the histogram. This effect is particularly noticeable in the 10°C treatment (Figure 5.1b) when, after 250 degree days, the sample size has been reduced to one. In subsequent analyses this bias has been eliminated by truncating that data sets when the sample size 72 'U'l"'-|' '''''''' (A) (B) o.— >oa cod 2...? con. aim coo—2 22§ZZVW ammo .......... ammmmwmm am 427 (427 QZZQKZQV 6%2Q6226 AQZZQézo 25% 1m 1m mama [222%3 .422Q&M @8222» lazaax r azaaZZEV 2329327, zzazzav 1 2m AZQZZZV, [ZZZQ%ZM 33%??? - 9.0 I.N a O.“ 4 a.— « a o.— m.o «l 0.0 u.u o.u Degree Days After Molt to Adult ri Dotted and (d) 23°C. roduction for I. (c) 17°C, Temporal pattern of rep (b) 10°C. line marks 90% mortality. reared at (a) 5°C, Figure 5.1 73 is reduced by 90 percent. Data from the lowest temperature treatment, 5°C (Figure 5.1a), has also been eliminated from subsequent analyses. This data set simply contained little useful information. Undoubtedly, this is primarily due to the fact that aphids in the treatment only spent an average of 6.4 00 in the reproductive stage. Many of these aphids were still alive when the experiment was terminated. At that point some of them were over 100 days old! Further examination of Figure 5.1 reveals that these histograms really don't look very similar at all. This is a result of transforming the X axis to physi010gical thma units. Since data was originally collected daily, the transformation results in different cell widths for each temperature. This difficulty is corrected in Figure 5.2a by plotting the truncated data sets as cumulative births per aphid per day versus accumulated degree days since adult molt. The data is now in a form to begin satisfying the three criteria detailed at the beginning of this section (see Figure 5.2b for a graphical description of this process). First, it is necessary to window data to cover only the mean length of the reproductive phase of the adult life stage. This is accomplished by eliminating data for degree day totals greater than the mean length of the adult life stage and less than the mean length of the pre-reproductiwe period. ‘1! . "5“ €_ _x.! 1"} "L" .._- 74 Figure 5.2 Temporal pattern of reproduction: (a) cumulative births/female versus 00 after molt to adult, (b) transforming the cumulative function to a cumulative density function (CDF), (c) aphid fecundity GDP, and (d) discrete probability "density fecundity function (PDF). 75 “.3109 31 (916)“ 0‘ M (916“) 053‘ {:5 J W (new 09“ .00“ 5‘ (A) 23°C 0 ll. \ a. 2 E (n O ..>. 9: I! E 0 . °0 5130 Y 000 2 2 g (B) ,,,,,,, E l-Ym, 1.0 """" f 2 3 iYm—mo ’’’’’’ 9 "’0 m 5 """""""""" {ms-m 1‘ W “’0 Degree Days 3- (C) ,,,,,,,,,,, u. x, a IIII “a ,,,,, s vvvvvvv g ””””””” 17°C :5 --------------- 23°C 3 ’’’’’ 0 °. ”’ “’0 60 150 . 160 200 260 Degree Days In Reproductive Period 8":- ‘g (D) E / 23 g 2 a / / . / / 8 ¢ ¢ (I 8;. , , 4 e , , , 2 ° 1 2 a 4 5 6 7 8 9 10 11 12 Substages in Reproductive Period '1 a. VI 0 N 76 Let xpre-repro = mean length of pre-reproductive period xadult = mean length of adult life stage xrepro = mean length of reproductive period t = time axis, in degree days then, Ypre-repro = cumulative births/aphid/day at Xpre-repro Yadult = cumulative births/aphid/day at xadult = cumulative births/aphid/day at x Yrepro repro t_new = transformed time axis, in degree days Since Y values are cumulative it is necessary to transform them as follows: Yt_new = Yt _ Ypre-repro This results in the following identity: Yrepro = Yadult - Ypre-repro If X values are transformed in a similar fashion xt___new = xt - xpre-repro The cumulative fecundity function now exists between (0.0) and (xrepro'Yrepro)' To transfonm this cumulative function into a true cumulative density function calculate 77 Yt__new = Yt new Yrepro Figure 5.2c shows the experimental data following these transformations. All that remains is to convert this cumulative density function to a probability function displayed as a histogram with K cells where xrepro 250 Cell Width = = = 20.83 on K 12 Values for K and Xrepro are from Table 8.4. The probability density functions for the 10°C and 17°C temperature treatments are shown in Figure 5.2d. Results from a Kolomagorov-Smirnov test (KS=.03, n=12. ns) indicate that both treatments are from the same distribution. Therefore an average of the two treatments will be used to parameterize the probability density function (Table 5.1) required by the distributed delay. 5.3.2 Seasonal Aphid Fecundity Data was first analyzed using covariate analysis of variance to determine if (1) seasonal changes. (2) site differences, or (3) morph differences were significant. Upper and lower bounds were analyzed separately. Results of these two ANOVAs are summarized in Tables 5.2 and 5.3. 78 Tdble 5.1 Discrete prdbability density function of the temporal pattern of births for apterous ;. Qg222£_. .-----————-—---——-—-—-—-------_——--—-—-——-——————_—“——-. --------—-—----------——----—-----------—_----—--—------ Degree day interval % births in interval 0.00 to 20.83 10.80 20.84 to 41.67 11.35 41.68 to 62.50 10.45 62.51 to 83.33 9.40 83.34 to 104.17 8.05 104.18 to 125.00 8.05 125.01 to 145.83 8.10 145.84 to 166.67 8.05 166.68 to 187.50 5.75 187.51 to 208.33 8.35 208.34 to 229.17 6.15 229.18 to 250.00 5.55 Results of the two analyses are very similar. In both cases the covariate, degree days, was the most significant variable (P<0.001). Both analyses also indicated no significant difference in fecundity between apterous and alate morphs. The two analyses do differ regarding the question of site differences. When fecundity is defined as the number of young in a colony plus all embryos (upper bound), site differences were highly significant (P<0.01). However. when fecundity is defined as the number of young in a colony plus only developed embryos (lower bound), site differences were not significant. The second step in this analysis was to use linear regression to determine the functions best describing upper and lower bounds for seasonal fecundity. Results from theanalysis of variance indicate that site and degree days are the appropriate variables for the regression. 79 'Table 5.2 ANOVA summary of seasonal fecundity of I, 25222;; *uhere fecundity = young in colony plus all embryos. Source df Mean Square F Covariates 1 4772.941 125.629::: degree days 1 4772.941 125.629** Main Effects 2 248.139 6.479*** site 1 455.280 11.983 morph 1 30.549 0.804 2-Way Interactions 1 0.433 0.011 site X morph 1 0.433 0.011*** Explained 4 1316.413 34.649 Residual 219 37.992 ' Total 223 60.924 Table 5.3 ANOVA summary of seasonal fecundity of I, 222235; where fecundity a young in colony plus developed embryos. Source df Mean Square F Covariates 1 2085.118 79.04o::: degree days 1 2085.118 79.040 Main Effects 2 41.309 1.566 site 1 72.929 2.765 morph 1 8.353 0.317 2-Way Interactions 1 5.849 0.214 site X morph 1 5.849 0.214*** Explained 4 543.346 20.597 Residual 219 26.382 Total 223 35.853 ** P < 0.01 *t* P < 0.001 A"IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIllliilllllllllllllllllii.1 80 Therefore, separate multiple regression equations for upper and lower bounds were determined using the following statistical model 2 Y = Z alaars1 + a2*82 + a3*Sl*°D + a4*Sz*OD + E(0.cr) s=1 where. Y = predicted fecundity per aphid oD accumulated degree days base 38°F S1 1 if site = 1, 0 if site = 2 S2 0 if site = l, 1 if site = 2 a ..a constants determined by regression E10,v random error term with a mean of 0 and standard error of Results for these two regressions are summarized in Tables 5.4 and 5.5. The coefficient of variability for the upper and lower bound regressions were 47.2% (F=46.963. P<0.001) and 59.5% (F=28.429. P<0.001). respectively. In both cases the tables reveal that the 95% confidence intervals for slopes and intercepts at the two sites overlap. 5.4 Discussion I only have one brief comment about the laboratory fecundity study. In general. the form of the analysis used should work for transforming data from other sets of fixed temperature fecundity experiments if one important assumption is met. Births per degree day must be independent of temperature. The data set used here did 81 Table 5.4 Regression results of seasonal fecundity study where fecundity = young in colony plus all embryos. constants S.E. F 95% conf. int. a1 32.015 3.777 0.2901fi* 24.50 < a1 < 39.53 a2 29.980 2.923 105.194*** 24.16 < a2 < 35.80 a3 -8.040E-3 8.886E-4 82.290*** -9.96E-3 < a3 < -6.1zs-3 a4 -6.015E-3 1.202E-3 25.059 -8.63E-3 < a4 < -3.40E-3 Table 5.5 Regression results of seasonal fecundity study where fecundity a young in colony plus developed embryos. constants S.E. F 95% conf. int. a1 22.352 3.141 1.262}fi* 16.10 < a1 < 28.60 a2 18.823 2.431 59.944*** 13.99 < a2 < 23.66 a3 -5.653E-3 7.371E-4 58.803*** -7.24E-3 < a3 < -4.06E-3 a4 -3.740E-3 9.994E-4 14.006 -5.92E-3 < a4 < -1.56E-3 _—---—-_-——-—---—--_-_——-—-_—--------‘-——_-—----—-—-—-——-—- --—--—--—————--———-—-————--—----—-—-P—-———-—-—--————————--- 14 Due to the way SPSS calculates multiple regressions. this F value really represents the significance of the difference between a1 and a2. *** P < 0.001 82 indeed meet this assumption; however, only results from 10, 17. and 23°C treatments were used. It will be interesting to see if this assumption is met once the aphid has been successfully reared over a broader range of temperatures. Several comments about the seasonal fecundity study are in order. The first two comments concern differences in results for the upper and lower boundary conditions. When estimated fecundity is defined as the number of young in a colony plus all embryos (cf. Wratten 1977) site differences are highly significant. However. when fecundity is defined as the number of young in the colony plus only developed embryos (cf. Gilbert & Gutierrez 1976) site differences were not significant. If, as Wratten suggests. developed embryos only represent young born during the next 24 to 48 hours then, indeed, we should expect little difference between sites when potential fecundity is estimated using Gilbert & Gutierrez's criteria. On the other hand, if Gilbert & Gutierrez's criteria is an accurate estimation of potential fecundity then undeveloped embryos must be surplus reproductive material that is a programmed part of the aphid life cycle. If Gilbert & Gutierrez are correct and if site differences do indeed exist, then their method should reveal them and potential fecundity calculated using Wratten's approach should result in no significant difference in fecundity for the two sites. The results from this experiment agree with Wratten's hypothesis and do not support the argument of Gilbert & Gutierrez. 83 Next. I should like to consider the curious relationship between the F statistics and the multiple regression results for the two boundary conditions. The upper boundary definition for seasonal fecundity resulted in higher F statistics throughout the ANOVA tables. Yet, the multiple regression equation for the upper boundary explained 12% less of the variation in the data than did the multiple regression for the lower boundary condition. Also. site differences in the slopes and intercepts for the multiple regression equations were greater for the lower boundary than for the upper boundary. Nowhere however, were they significant at P=0.05. This pattern suggests that the data has been fitted to an inappropriate regression line. Several transformation were tried; however, none of them resulted in higher coefficients of variability. Nor did they changes the pattern of this statistic for the two boundaries. The most reasonable explanation is that seasonal fecundity is not a linear function at all but a polynomial one. Way 5. Banks (1968) demonstrated that such is the case for the black bean aphid, éphis fabae Scop. If changes in the nutritional quality of the host is indeed the primary factor influencing fecundity, then there maybe a very simple biological reason for a polynomial form to the equation. Once the blueberry fruit has matured the bush undergoes a short-lived late season burst of new terminal growth. This abundance of new growth could, in turn. result in a temporary upswing in aphid fecundity. 84 Throughout this experiment variability in estimated fecundity between aphids collected on the same date was quite high. This high variability is probably the result of aphid movement. During periods of hot weather. ie. above 90°F, aphids were observed actively walking on blueberry leaves. This tendency would result in an underestimation of potential fecundity during generations subjected to hot weather in July and August. Natural enemies could also be responsible for changing patterns of aphid dispersal through the growing season. For instance, Wratten (1976) has shown that lime aphids, Eucallipterus tillae L., are sometimes able to escape predation from coccinellid larvae by jumping. kicking, or running away. The percentage of aphids escaping was dependent on the life stage of both the aphid and the predator and on the mode of escape attempted by the aphid. Frazer-ii Gilbert (1976) obtained similar results when studying coccinellid predation on pea aphids, Acyrthosiphum pisum. Predator induced movement of adult aphids would also result in underestimates of potential fecundity. Due to (1) the high variability in the data of this experiment. (2) potential problems due to aphid dispersal, and (3) the great expense (in time and labor) of dissecting large numbers of aphids, I would suggest that future field experiments concerning aphid fecundity be conducted using individually caged aphids. Experiments could be initiated with fourth instar nymphs. This technique should result in 85 less within date variation, and larger sample sizes can be used with less increase in cost. CHAPTER 6. MORPH DETERMINATION 6.1 Introduction The population model detailed in chapter eight includes both apterous and alate. viviparous morphs of I, pepperi. The sole objective of this chapter is to determine how to correctly allocate newborn aphids between these two morphs. No experiments were conducted to investigate the mechanisms influencing morph determination in this aphid. That project would easily be a thesis topic in its own right. Instead, a review of the existing literature was used to reveal which environmental and biological factors consistently play a role in this process throughout the family Aphididae. Once identified, aphid field data was plotted in terms of these factors and analyzed to arrive at an acceptable mathematical function for the model. 6.2 Literature Review Scientific literature exploring the causes of aphid polymorphism is quite voluminous. The subject has been debated in more than 100 scientific papers and several review articles (Hille Ris Lambers 1966, Lees 1966, Mittler & Sutherland 1969. Schaefers 1972). Five factors have been demonstrated to be important in the process of morph determination: crowding, nutrition, temperature. photoperiod, and parentage. In general. the literature can be divided into two groups. The first group of papers 86 87 explores factors involved in determining whether a female aphid will produce sexual or parthenogenic forms. Photoperiod, temperature. parentage. and to a lesser extent nutrition all play a role in governing the appearance of sexuales (Bonnemaison 1951: Lees 1959. 1960, 1963. 1966; Dixon & Glen 1971; and Dixon 1977). A second, and larger body of research addresses factors influencing whether parthenogenic aphids will be apterous or alate. Only this latter group is pertinent to the present discussion. The prevailing hypothesis for many decades was that the quality of an aphid's host plant was the puimary factor influencing wing development (Schaefers 1972). According to this hypothesis alates were a population's means of escape from deteriorating host plants and apterae were viewed as the true adult form. Crowding was also thought to be a factor but only in so far as it resulted in nutritional deficiencies. Then, in 1951 Bonnemaison presented evidence indicating that tactile stimulation by other aphids could also promote wing development (Bonnemaison 1951). Since earlier researchers had not controlled for aphid density in their experiments this development placed much of that earlier work in doubt. The theoretical tide turned again in 1960 when a new theory of the developmental process underlying wing formation was unveiled (Johnson & Birks 1960). This new 88 hypothesis, later dubbed the diversionist theory, asserted that alates are the ancestral and true adult form of aphids. Apterae. in turn. are a neotenic form occurring when alatiform embryos are irreversibly diverted from their true course by nutritional or density effects. Although this theory is by no means universally accepted today, over the last 25 years it has gained wide support and continues to color many of the developmental questions asked by aphid researchers. For the last two decades work has focused on questions of how and when the factors of nutrition and crowding affect wing development. Three types of questions have been most prevalent: (1) which chemicals afect aphid growth and wing development. (2) what density levels are required to alter innate developmental tendencies, and (3) when during development can tactile stimmulation affect wing development. Six amino acids are known to affect wing development, the most important of these being isoleucine and histidine (Mittler & Dadd 1966, Dadd 1968, Sutherland 1969b). Diets lacking these chemicals retard growth and have an apterizing influence on developing aphids. Starvation has also been shown to reduce wing formation in several species (Schaefers 1972). In addition. there is a high correlation between large size (or weight) and the percentage of alates for some species (Tamaki & Allen 1968, Mittler & Sutherland 1969). 89 Shaw (1970b) has suggested that the level of juvenile hormone present during larval development may be responsible for the continual polymorphism observed between apterous adults and fully develOped. functional alates. The effects of crowding on wing development are more complicated and often contradictory. Some species, such as Aphis fabae, only produce alates when reared in isolation (Shaw 1970b). Others, like g, brassicae, Aphis cracivora. and g, persicae respond to crowded conditions by producing many alate emmigrants (Johnson 1965, Sutherland 8: Mittler 1971). And a few species such as the strawberry aphid. Chaetosiphum fragaefolii, respond little to changes in density (Dixon & Glen 1971, Judge & Schaefers 1971). The timing of tactile stimulation also plays an important role in morph determination. Again, different species respond differently. In some species density influences wing formation prenatally, through the crowding of stem mothers (Lees 1961b). In others this development is influenced postnatally, either through the density of stem mothers (Johnson 1965) or nymphs (Shaw 1970b). Still other species are influenced both pre- and postnatally (Dixon &' Glen 1971, Sutherland & Mittler 1971). This wide range of responses reflects the fact that alates play diffement ecological roles in different aphid species. Schaefers (1972) has identified three ecological function for alates: dispersal, emmigration, and escape 90 from natural enemies. Two examples will serve to illustrate these functions. The strawberry aphid, g, fragaefolii, is the first example (Judge & Schaefers 1971, Schaefers 1972). This aphid has a very narrow host range. It exhibits a positive correlation between size and alatism. But only a slight decrease in alate production occurs when aphids are crowded, even at very high densities. Alate production is almost totally a function of nutrition. The sole purpose of alates in this species appears to be dispersal, ie. alates are produced when the host is in the best condition for exploitation by the aphid. Example number two is the bird cherry-oat aphid, R, pagi. a host alternating species (Dixon 1976. Schaefers 1972). 'Ehis species exhibits a negative correlation between adult size and alatism. Crowding results in an increase in the production of alates. In this. and several other species of host alternating aphids poor nutrition and crowding can result in almost an entire generation of aphids being winged emmigrants (Dixon & Glen 1971).. However, it would be a udstake to generalize from these two examples that dispersal is the primary role of alates in holocyclic aphid species: while in heteocyclic species their sole purpose is escape. either from a deteriorating host or natural enemies. Contrary examples also exist. 91 6.3 Materials and Methods The literature indicates that nutrition and crowding are the most important factors influencing wing development in parthenogenic lines of many aphid species. Blueberry bushes undergo an annual cycle of growth. fruiting, and senescence. Therefore our time axis, degree days, was chosen as the variable most readily associated with these seasonal changes in the nutritional quality of the host. Similarly. the total number of aphids collected in a sample was used as an indicator of crowding. Partitioning of newborn nymphs will be accomplished by a function relating the porportion of alates produced in a population to aphid density and degree days. Data for the construction of this function was derived from summary statistics of time 'standard sample' described in Section 2.2. Analysis was conducted using (1) SURFACEZ, a three dimensional computer plotting package (Sampson 1978). and (2) linear regression techniques (Ruppel & Dimoff 1978). Make no mistake. this analysis is correlation ecology. plain and simple. As nentioned previously. no effort was made to study the mechanisms governing morph determination in I. e eri. 6.4 Results Figure 6.1a contains a contour plot of the three variables in question. The X and Y axes are degree days and 900» (A) . J O O ale 0‘! O O Aphids Per Sam (N O O 100 ~ 1000 3000 40100 « . (B) 0’ 1 3 . $8 (I) Site 1, 1981 < «5‘ - ._ A Site 1, 1982 o J 8 3K Site 2, 1982 E 5.: — Non—linear Regression Eq. 0 O a O I- O. o .0 I ‘ " l a 9 ° - - - - 4 - 4 - L 4 -4 °1000 2.000 .. 3030 v w - w V400; v VT Accumulated Degree Days Figure 6.1 Morph determination in 2_II_. ri: the proportion of alates (a) as a function of crowding and °D, and (1)) only as a function of ob. 93 total aphids per sample. respectively. The proportion of alates in a sample is plotted on the Z axis between 0 and 1. Surface contours are drawn at intervals of 0.05 (5%). and labeled in increments of 0.1 (10%). Interpolation of Z values between data points was accomplished using Bessel interpolation (Sampson 1978). The response surface depicted here shows a sharp decline from 75% winged adults early in the growing season to less than 5% alates over an interval of 1500 degree days. Aphid density appears to have little or no affect on the proportion of alate aphids. The area of moderate decline in alate production in the upper left hand corned of the figure is an artifact of the interpolation process. Since very high aphid populations were never observed early in the growing season, this area of the plot is devoid of data. Also, four observation of I, e eri populations with densities greater than 1000 aphids per sample are not included in this plot. These points were too widely separated to be used in reconstructing the topology of this response surface. Figure 6.1a indicates that density has little affect on wing production in parthenogenic lines of I, e eri. This result simplified the analysis since the proportion of alates could then be analyzed as a function of degree days using linear regression. Transformation appropriate for sigmoidally shaped, binomially distributed, prOportional data are sin-1(P). probit, and logit (Anscombe 1948, Rao 94 1965, and Ruppel & Dimoff 1978). Of these transformations, the logit gave the best results. It was also necessary to specify an upper asymptote for Y values and to transform the X axis by taking the Loge(x). The final form of this equation reads as follows: P = (6.1) where proportion of alate adults accumulated degree days upper Y asymptote = constants determined by regression ”WNW! ‘5' II II II .A value of 0.8 was used as the upper Y asymptote. Regression results (R2=0.83) indicated that the appropriate values for a and b were 61.31 and -8.291, respectively. Therefore the equation used in the model to determine the proportion of alates as a function of degree days is 3.42 * 1026 P = (6.2) x8.291 26 + 4.27 * 10 Figure 6.1b shows both this predictive equation and the data used to generate it. The figure is a plot of the proportion of alates versus degree days using three site-years of data. 95 6.5 Discussion The minor role that density appears to play in regulating the number of alates in populations of I, pepperi is quite unexpected. work with a closely related species, M, maxiam, indicates that crowding is a very important factor in morph determination in that species (Gilbert et a1. 1976). In fact, four data points missing from Figure 6.1a suggest that at extreme densities, ie. greater than 2000 aphids per sample, a crowding effect may be present. At such high densities I, pepperi exhibits rates of 1% to 4% alates, at times when smaller samples contained no winged individuals. CHAPTER 7. ABIOEIC HDRTALITY 7.1 Introduction Several kinds of non-biological, or abiotic, mortality agents can play a role in redcuing the numbers of aphids in highbush blueberry fields. The most important of these are climatic factors and cultural practises. Insecticide applications and mechanical harvesting are undoubtedly the most important cultural practises in this respect. Similarly, climatic mortality effects could be due to excessive heat, precipitation, or high winds. This chapter focuses on the role of rainfall as an aphid mortality agent. The work was undertaken because previous research (Elsner 1982) suggested that moderate to heavy rainfall may result in significant aphid mortality. Surprisingly little information on rain induced mortality is present in the aphid literature. Although the impact of predators, parasitoids, and insecticides on aphid pOpulations have been studied intensively, not a single experiment on the effect of rainfall was discovered during the literature search. In fact, the possibility of rain induced mortality was only mentioned in two articles. In his work with B, brassicae Hughes (1963) found that very heavy rainfalls following prolonged dry periods resulted in the disappearance of almost two-thirds of the aphid population. During dry periods B, brassicae populations develop all over host plants. Subsequent heavy 96 97 rains eliminate aphid populations from exposed surfaces. Hughes demonstrated this mechanical effect by holding heavily infested kale plants at an abnormal orientatflmi during a torrential rainstorm. Under these circumstances aphids were washed from leaf surfaces exposed to the rain. Following such storms Hughes also noticed a temporary shift in the age distribution of the population, apparently caused by the re-ascent of late instar aphids that had been washed off the plants. The second reference to rainfall is no more than a comment. In an article on the population dynamics of M, maxima, Gilbert (1980) notes an apparent correlation between rainfall and year to year variation in aphid numbers. He postulated that such an effect, if genuine, might be due to either the direct effects of wind and rain or to indirect effects on the host plants. With this background in mind an experiment was conducted to determine the impact of rain induced mortality on aphid populations and to generate predictive equations relating the percent reduction in aphid numbers of daily rainfall between samples. 7.2 Materials and Methods This experiment was conducted at site 2 (Charlotte, MI) during the summer of 1982. To assess the impact of rainstorms on aphid populations, individual colonies of I, 98 pepper were tagged and sampled twice weekly until they disappeared. As colonies disappeared they were replaced to maintain a relatively constant sample size. Due to occassional difficulties in finding enough replacement colonies, variations in sample size did occur. The sample was stratified vertically into three regions to determine if the mortality effect was uniform throughout the bush. Sampling consisted of recording the tag identification number and the number of aphids present in a colony. Sample size was generally between 15 to 20 colonies in each third of the bush. Each CHE the vertical stratifications was analyzed separately using linear regression. To prepare data for this analysis it was necessary to calculate the rate of change in the population between samples. This was done in several steps. Step number one was to identify which colonies were present at sample time t and t+1. Next, the total number of aphids present in these colonies at the two sample times was calculated. Then a rate of change index (RC) was calculated by dividing the number of aphids present at time t+1 by the number of aphids at time t: RCt = Numbert+1 / Numbert (7.1) Finally, the inches of rainfall that fell during a sample interval was paired with this index value. Only sample 99 intervals that experienced rainfall were included in the final data sets. 7.3 Results Inspection of these data sets revealed that not all rainstorms are detrimental to the aphid population. In fact, the heaviest rainfall of the season, 1.75 inches, resulted in no significant mortality. The RC index value remained slightly above 1.0 for all three regions of the bush, just as it did repeatedly when no rain fell. Although only daily weather information was available for Charlotte, hourly data from Lansing airport indicated that light rain had fallen frequently over a two day period. Therefore, although the total accumulation was quite high, each storm was actually very light. Regressions were conducted after this data point had been removed from each data set. Figure 7.1 contains plots of the rate of change index versus inches of rainfall for the upper, middle, and lower third of the bush, respectively. Both data points and the regression line are illustrated. Solid data points highlight a second, unusual storm. This storm imparted heavy losses to colonies in the upper third of the bush, yet did not affect the rate of increase for populations lower in the bush. If anything the RC index for the bottom third of the bush was elevated slightly. Perhaps the storm is evidence that rainfall may also result in re-distribution of Jpn J c 8 1-60 I 1.00 sir I 0.78 0.50 0.25 L Rate of Change Index (RC) cp.00 '8 1.80 I .00 100 (A) r 0.78 0.50 1.25 (B) 0.26 1.00 .00 Figure 7.1 I T 0075 .25 Inches of Rain 0.25 1.00 Data and linear regression of rain induced nortality*for the (a) upper. (b) middle, and (c) lower third of the bush. 101 aphids within the bush. In any event, removal of this data point from the regression for the bottom third of the bush improved the coefficient of variation by almost 50%, from 23.6 to 70.5%. Statistics for all three regressions are contained in Table 7.1. 'Fhe smaller y'intercept value for the middle portion CHE the bush is probably reflection of its suitability to the aphhd. A smaller rate of increase for aphids iJ1 the middle of the bush, even in the absence of rain agrees well with results from both seasonal fecundity and vertical distribution studies. The seasonal fecundity study highlighted the importance of good nutrition in maintaining high levels of fecundity. The vertical distribution study revealed that the middle of the bush contains tflua smallest proportion of the aphid population. Since the middle of the bush contains few actively growing shoots, it is not surprising that this region also has the lowest rate of increase. The middle of the bush also has the smallest exterior bush surface are to bush volume ratio. Presumably, aphids away from the bush exterior are afforded some protection from the rain. Therefore it is not surprising that aphids in the middle of the bush are least affected by rain induced mortality. What is surprising is that aphids in the bottom third of the bush are subject to greater rain induced mortality than are aphids in the top of the bush. 102 Table 7.1 Regression statistics for rain mortality field experiment. Third of Bush Y intercept lepe R2 Upper 0.990 -0.467 0.677 Middle 0.889 -0.391 0.745 7.4 Discussion The experiment has demonstrated that rainfall can indeed result in significant mortality to I, e eri populations. It has also shown that increasing levels of rainfall generally mean increasing aphid mortality. However, the present method of prediction was not completely successful. The method failed to identify storms that did not affect the aphid population, even though they drOpped significant amounts of rain over a two to three day period. Daily rainfall alone is not the best predictor. Better results should be attainable by incorporating hourly rainfall or wind speed into the prediction. CHAPTER 8. DESIGNING A LIFE-TABLE FOR I. m; 8.1 Introduction The purpose of this chapter is to distill and synthesize information presented earlier in the thesis to design a predictive population model for I, pepperi. Once completed and tested, the model could be used to simulate strategies for limiting the number of aphids in commercial highbush blueberry fields. One strategy of immediate interest is the proper timing of pesticide applications to limit aphid populations and minimize interference with natural mortality agents. The idea of constructing a predictive aphid population dynamics model is certainly not new. Several such models have been developed over the last two decades (Hughes 1963, Hughes & Gilbert 1968, Gilbert & Hughes 1971, Lowe 1973, Dixon & Barlow 1979). A few models have also addressed the role of aphids as disease vectors (Gutierrez et al. 1971, 1974a, 1974b, Frazer & Gilbert 1976). Research on one aphid in particular, Masonaphis maxima, has been instrumental to the deve10pment of the present simulation. Although this model is by no means a clone of the M, maxima research, the earlier work pointed out important ecological relationships in a similar aphid ecosystem and provided a source for initial estimates of some model parameters. Since it is important to understand both the sindlarities and differences between the two 103 104 models, a few moments will be spent reviewing the M, maxima system. 8.1.1 A.variable life-table for Masonaphis maxima (Mason) Details of M, maxima's biology are discussed in Frazer & Forbes (1968) and Campbell et al. (1974). Efforts to model this aphid's population dynamics using a variable life-table approach are presented in Gilbert 8: Gutierrez (1973), Gilbert et al. (1976), and Gilbert (1980). Like I. pepperi, M, maxima is a holocyclic aphid, spending its entire life cycle on thimbleberry, Mppgg parviflorus Nutt. The species range extends along the Pacific coast from British Columbia to California. Although the aphid is not economically important, it is a vector of thimbleberry ring spot virus (Stace-Smith 1958). The species was cudginally chosen as the subject for the construction of a variable life-table as a matter of convenience: (1) because of color differences morphs could be distinguished easily: (2) few, distinct generations made the determination of a population's age structure relatively easy: (3) the aphid's life cycle was very closely tied to its host plant: and (4) the aphid had a compliment of natural enemies. The life-table was developed as a deterministic, discrete time model. It is referred to as a variable life-table because many model components (such as fecundity and stage specific mortality factors) are not constants but functions. Aphid fecundity was simulated in 105 time life-table using three functions. The first function described how M, maxima's birth rate varies during the growing season. This linear decline reflects the influence of food quality on aphid fecundity. A second, negative exponential function described the density—dependent reductirnl in fecundity caused by competition between stem mothers. .A third function reflects the frequency distribution of births suring a stem mother's adult stage. In the earliest forms of the model parasite and predator mortalities were modeled with an arbitrary function. Later, complete submodels were developed to account for syrphhi predation and primary parasitism. The major cricism leveled against this model has been that it was never tested against an independent data set. Although good fit to field data was achieved, the same data had been used to construct the budget in the first place. The modelers' answer to this cricism is that model parameters could only be fitted to a specific population of aphids during one growing season. Because of extreme differences in size, climate, and suitability of different patches of its host, thimbleberry aphid populations have quite different densities, numbers of generations, and proportions of sexual morphs. 106 8.2 Materials and Methods Data for the construction of a variable life—table for I. pepperi come from several sources. First, Elsner's (1982) fixed temperature cohort study provided information about the duration of aphid life stages (Section 4.3), and the distribution of births during a stem mother's reproductive period (Section 5.3.1). Second, much of the life-table was parameterized using information collected via the 'standard sample' studies described in Section 2.2. This information included (1) seasonal changes in the aphid's reproductive rate (Section 5.3.2), (2) seasonal within bush distribution (Section 3.2), and (3) both biotic and abiotic mortality factors (Sections 2.3 and 7.3). As mentioned earlier, where data for a model parameter was not available the parameter was estimated using existing information for M. maxima. The earliest version of this model contained many estimates from the thimbleberry aphid life-table. As the project progressed experiments were devised to replace these estimates with experimental results for I, pepperi. In the current version of the model only the mean development time for large, alate nymphs has been estimated from M, maxima. The computer program detailing the life-table was written in USCD Pascal (version IV.0) on the Columbia Data Products microcomputer described in Section 2.2. 107 8.3 Mbdel Design and Parameterization 8.3.1 Overview Five aspects of I_. pepperi's biology are required to simulate its population dynamics: aphid maturation, fecundity, mortality, morph determination, and spatial distributixn1. The overall design for this variable life- table is flowcharted in Figure 8.1. Following an explanation of this figure each aspect of the model will be discussed in greater detail. The maturation of each aphid is simulated using five distributed delays, represented in the flowchart by circles. Since the dynamics of apterous and alate aphids are dealt with separately, a total of ten delays are needed to model growth. Although fixed temperature experiments indicate that I, pepperi has four nymphal instars, only three sizes of nymphs can be distinguished readily in the field. These are classified as small, medium, and large nymphs and are modeled using three separate delays. Similarly, the adult life stage is separated into pre-reproductiwe and reproducitve adults and mmdeled via two delays. For our purposes the pre-reproductive period is defined as the interval between molt to adulthood and the birth of the first young. Idkewise, the reproductive period is defined as the time interval between the first birth and death. wanna mow—=95”. defined—mom % .m on» No Buffing Haw nun—mam 108 Ir \ warm to 558.65 teams—cot m . I. A - 38> 529E528 / //J //J / / WWII 568a /\ /\ giia%%%\\ >530 p. 5:83.... .9688 V 109 Two facets of the aphid's reproductive strategy have been incorporated into the model. First, seasonal changes in aphid fecundity are estimated as a function of accumulated degree days. Then the numbers of adult aphids in both morphs are pooled and a transfer function describing how an individiual stem mother's reproductive potential is distributed over time is applied to calculate the number of small nymphs born. Finally, newborn aphids are separated into two groups and added to the current pOpulations of small and alate nymphs. The proportion of newborns destined to become alate is a function of accumulated degree days. Aphid mortality is divided into three causal categories: abiotic, biotic, and cultural mortality. Abiotic mortality is modeled as a function of rainfall. Biotic mortality is the result of predation and parasitism. In the current version of the model, parasitism is estimated as a function of accumulated degree days. Predation, on the other hand, is treated as an unknown quantity and will be estimated as part of the 'tinkering' process. CMltural mortality factors include such things as chemical applications, effects of machine harvesting, etc. The implementation of this last category is beyond the scope of the present study and has been left to future modeling efforts. Spatial distribution refers to the vertical distribution of the aphid population within a tflueberry bush. This stratification changes during the growing season 110 anui is modeled using a table look-up function to linearly interpolate between known values. The effects of both cultural mortality and parasitism are affected by the within bush distribution of aphids. 8.3.2 How the Model Keeps Time Individuals using the computer program detailing the life-table observe a model that progresses by calendar date. However, the model is also keeping track of time in degree days (variable WEATHER.DD), an estimate of physiological time for the aphid. Although most model variables are updated daily, those relating to aphid growth are updated in discrete steps composed of a fixed number of degree days (variable DD_PER_DT). Each such step is referred to as one DT (delta time step). The portion of the program that controls time in this way reads as follows: LEFT_OVER_DT := 0.0: FOR J_DATE := FIRST_DAY TO LAST_DAY DO BEGIN GET_WEATHER( J_DATE )3 LAST_DT_TODAY := TRUNC(WEATHER.DD/DD_PER_DT + LEFT_OVER_DT): LEFT_OVER_DT := WEATHER.DD/DD_PER_DT - LAST_DT_TODAY: FOR DAY_SEGMENT := 1 TO LAST_DT_TODAY DO BEGIN CALC_STATES( DAY_SEGMBNT ); CALC_RATES( DAY_SEGMENT )2 END: END: 111 The FIRST_DAY and LAST_DAY of the current run are entered duriru; the initialization process. J_DATE stands for julian date. Once each day the proqram reads weather data from a random access file (procedure GET_WEATHER). The program then calculates how many DTs are needed to simulate the current day (variable LAST_DT_TODAY), taking into account any fraction of a DT not run yesterday (variable LEFT_OVER__DT). The actual model is contained within the procedures CALC_STATES and CALC_RATES. 8.3.3 Sequence of Calculations If the model is to function properly it is important that all variables be calculated in the proper order. Only two correct sequences exist (Manetsch & Park 1980). In either case, during any given DT all state variables must be calculated before corresponding rate variables. To avoid confusion state variables are calculated in one routine and rates in another. Figure 8.2 details the overall sequence of calculations for the entire model. This sequence was chosen because it allows for the greatest flexability in specifying model output. Note that model output is only available at the end of each day. The actual frequency of such output may be changed during the course of a run via the command line in step (1) of the execution phase. This command line is an important feature for future implementation of control 1) 2) 3) 4) 1) 2) 3) 4) 5) 6) 7) 8) 9) 112 INITIALIZATION PHASE Default initialization of all user specifiable variables Interactive initialization of user specifiable state and rate variables and specification of run characteristics Initialize internal state variables at T0 Compute internal rate variables at T0 EXECUTION PHASE Prompt user with command line for possible changes in print frequency and execution interupt control Calculate the number of DT's to run this day Update TIME: T = T + DT Compute STATE variables at T + DT Compute RATE variables at T + DT If not end of current day return to (3) Print output as desired If last day of run terminate simulation and print summary output If an interrupt condition is satisfied return to (1), otherwise return to (2) Figure 8.2 Sequence of calculations in the blueberry aphid variable life-table. .Hodified fromiuanetsch,& Park 1980. 113 strategy simulations. The prompt is the visible portion of an execution interrupt control mechanism that allows users to (1) alter output frequency, (2) view additional model statistics not contained in routine output, (3) specify control measures (not currently implemented), (4) specify how many days the model should run before interrupting execution again, (5) complete the simulation without further interruption, or (6) abort the run. 8.3.4 Medeling Aphid Maturation As computer usage in biology has become more common place, mathematical models of bioloqical systems have become increasingly complex. This trend is particularly true in the science of entomology where systems science methodology and simulation techniques have played a fundamental role in the design, development, and implementation of many pest management models (Ruesink 1976, Tummala et al. 1976, Welch 1984). One simulation technique in particular, the distributed delay, has been widely employed in the simulation of insect growth and maturation (Gilbert et al. 1976, Welch et al. 1978). In this life-table separate distributed delays are used to model each aphid life stage. Several variations of these delays have been implemented on digital computers (Abkin & Wolf 1976, Manetsch 1976, Manetsch & Park 1980). The delay routine used here is formally classified as a time-invariant distributed delay with storage losses. 114 .A block diagram of the distributed delay used in the present model is shown in Figure 8.3. The parameter MEAN equals the average number of degree days spent in a given life stage. The parameter K is an integer referring to the order of the delay. K describes what form the distribution will take and can be determined directly from experimental estimates of the population MEAN and VARIANCE where K = -------- (8.1) VARIANCE Output from the delay is the number of aphids that exited a life stage this DT. The rate of recruitment into a life stage is the input variable. Obviously, output from Stagei is input to STAGEi+1. Exceptions to this rule are (1) input to STAGEl is the number of aphids born during a DT, and (2) output from STAGES represents deaths due to old age. The number of aphids in a particular life stage is contained in the variable TOT_STORAGE. The number of aphids in each of K sub-stages is contained in the matrix STORAGE where K TOT_STOR = E STORAGE i (8 . 2 ) i=1 115 Storage Variables KJ:DISTRIBUTED Input Out ut % p , DELAY Proportion Loss Rate! I Figure 8.3 Flowchart showing input and output variables for the distributed delay used in the model. 116 The final two parameters, CONST_PLR and VARIABLE_PLR are the constant and dynamic components of the proportional loss rate. 'Fhe constant component refers to unexplained stage specific mortality and is entered as an Option during the initialization phase of the program. The variable loss rate is the sum of known mortality facotrs such as violent storms, chemical applications, and parasitism acting during one DT. One final theoretical issue regarding the use of distributed delays remains to be addressed —— model stability. Because of the way distributed delays operate, some choices of DT will result in unstable and unpredictable results. The distributed delay procedure used here employs Euler integration: therefore, allowable DTs must be in the range 0 < DT < 2 * (MEAN / K) (8.3) Such instability is an unavoidable consequence of numerical approximation techniques required to implement differential equations on digital computers. In fact, due to additional feedback from other differential equations scattered throughout the model the maximum allowable DT may be much less than equation 8.3. One facet of model parameterization is the selection of an appropriate DT to ensure that such error remains within allowable limits (for details see Manetsch & Park 1980, vol. 2, p11-14). Whenever changes 117 involving differential equations are made such stability tests on DT should be conducted again. Table 8.1 lists MEAN, K, and maximum allowable DT values for each life stage of both apterous and alate morphs of I, pepperi. These values were calculated from Table 4.3 using equations 8.1 and 8.3. Means have been rounded to integer values for simpicity. All aphids in the fixed temperature rearing experiment summarized in Table 4.3 were apterous individuals. However, rearing experiements conducted with M, maxima (Campbell et a1. 1974, Gilbert et al. 1976) indicate that the only difference in thermal requirements for alate and apterous individuals is the time required to complete the fourth instar (large nymph Table 8.1 Distributed delay parameters and maximum allowable DTs for each life stage of apterous and alate viviparous I, pgppg£_. small nymphs 54 12 9 medium nymphs 63 22 6 large nymphs 57 10 12 pre-reproductive adults 56 7 l6 reproductive adults 250 12 42 small nymphs 54 12 9 medium nymphs 63 22 6 large nymphs 66 10 12 pre—reproductive adults 56 7 16 reproductive adults 250 12 42 ---——---------—-———--——————---——--——————————————————-—_-—-—- .“-—------—---—--———--—-——-——-———--—-——-——-——-———~.—-—--————— In"- 1\I1l.1 118 category). Alate M, maxima nymphs require about 17% longer to complete the fourth instar than apterous nymphs. Using this relationship it is estimated that large I, e eri nymphs require 66.2 degree days to complete development. Both the MEAN and K values for reproductive adults are subjective estimates based on rearing aphids at 23°C. These figures were increased from those contained in Table 4.3 in an attempt to offset the drowning bias present in the rearing experiment. values of DTmax in Table 8.1 indicate that the selection of a DT for the model is limited by stability criteria for medium nymphs. This means that the model variable DD_PER_DT must be less than 6 degree days. The simulated maturation of one generation of apterous and alate I, pepperi generated using these parameters is shown in Figure 8.4. Output was obtained using an initial population of 100 newborn small nymphs of each morph and a ITr«of two degree days. Notice that reproduction of alate aphids is delayed slightly because large alate nymphs take longer to mature. Also notice that the aphid spends over half of its lifetime as an adult. 8.3.5 Fecundity and Mbrph.Deternination Figure 8.5 contains a more detailed flowchart of this portion CHE the model. .As discussed previously, both seasonal changes in aphid fecundity and the temporal 119 E1 (A) ’1 1 1 1 ' 1 ' 1 1 1 0 100 200 300 400 600 800 700 800 i: (B) v j W ' V 1 1 1 1 1 1 0 100 200 300 400 600 600 700 800 i: (C) 1 1 1 1 1 1 1* ' 1 O 100 200 300 400 500 800 700 800 g? (0) Percent of Population in a Stage 80 S- l o ”‘1 1 *1 1 ' 1 ' 1 ' 1 0 400 500 800 700 800 0 Sn . (E) . -——— Apterous morph 1 ----- Alate morph 0‘ l0 4 4 J i O V v I v v I ' Y ' j ‘f 1 0 100 200 500 800 700 800 36 ' .60 Deg:ee Days Since Birth Figure 8.4 Simulated maturation of one cohort of apterous and alate I. 32221- .. 120 distribution of births within a stem mother's reproductive period have been incorporated into the model. The flowchart also shows the relationship of morph determination to other components 1J1 this portion of the model. Note that both morph determination and seasonal fecundity are functions of accumulated degree days (TDD). The seasonal decline in aphid fecundity is simulated using the linear relationships determined in Table 4.4. Since the model will be fitted to data from site 1 (Charlotte, MI), equation 8.4 is used to estimate seasonal fecundity, FECUNDITY = 32.015 - 0.00804 * TDD (8.4) This decline reflects seasonal changes in the nutritional status of the host. Second, the temporal pattern of births during an individual stem mother's reproductive period (REPRO_FREQ) is modeled using the discrete probability function calculated in Section 4.3.1. The values for this matrix were shown in Table 4.1. The birth rate (per DT) is then determined using equation 8.5. The middle term in this equation reflects that portion of an aphid's reproductive period occurring this DT. 121 .Hoooa one no uncapped dawudswamuoo #936 can huwpguow on» HON Haggai mam mun—own A v anti .0 _ .Irllveoza :m 333.6 a. _ $233 5. Econ—coca A. c. omao~a< o>=oaooaoom mean; 5.35:. + 1.200 Azoeasom@A i 1 h N coco—2 + \7 dogma—am 5. e. no}? 355.com o>=oaoocaom & _mcoooom goo cocoon 122 _ K .1 BIRTHS DD_PER_DT ------ = FECUNDITY * ------—-- *2 REPRO__FREQi * STORAGEi DT MEAN ---- i=1 K (8.5) where, K = order of the distributed delay, reproductive stage MEAN = average degree days spent in reproductive stage MEAN/K = degree days spent in the ith substage of the reproductive stage REPRO_FREQi = proportion of births occurring during - , the ith substage STORAGEi = number of reproductives in the ith substage To incorporate alates into the model the proportion of newborn, presumptive alates (P) is determined using equation 8.6 developed in Section 6.4. 3.42 * 1026 T9081291 + 4.27 * 1026 As time flowchart for this section of the model indicates, the number of newborn apterae are then calculated as the difference between total newborns and newborn alates. 8.3.6 Medeling Spatial Distribution Both rain induced mortality and pesticide losses are affected by the vertical within bush distribution of aphids. This distribution is modeled via a table look—up function. The function uses linear interpolation to estimate between table values. Table estimates for each third of the bush are contained in Table 8.2. Table values were estimated from the 'standard sample' vertical distribution results 123 discussed in Section 3.2. Figure 8.6 illustrates this table in the same fashion in which the data was presented in that earlier discussion. 8.3.7 Mortality Component Three classes of martality agents are included in the population model: bdotic, abiotic, and pesticide induced mortalities. Biotic mortality factors include both predators and parasitoids. The relationships between these mortality sub-components and necessary system inputs are shown in Figure 8.7. This flowchart actually protrays two independent groups of mortality calculations. The simplied of these two lines determines the number of parasitized aphids that will be removed from the distributed delays for large nymphs and adults. System inputs into this function are accumulated degree days (TDD), and the level of parasitism (high, medium, low). The more complicated sequence of calculations includes mortality caused by predation, rainfall, and pesticide applications. The last two factors are affected by the vertical within bush distribution of aphids. The proportion of the aphid population killed by each of these two processes is first summed. This result is then multiplied by the total number of aphids present and by the proportion of those aphids in a given third of the bush. 124 I 00 8.1 Upper := fl Ee- O n. 1.. fi 4 O u o 55? Mkkfle O b O Q. Lower °1ooo ' 2600 71 3500 . 4000 Accumulated Degree Days Figure 8.6 Table look-up function used to simulate the vertical ‘within bush distribution of aphids through the growing season. Table 8.2 Table look—up values for the percentage of the aphid population in each third of the bush versus degree days. Degree days Upper Middle Lower 1000 0 26 74 1100 0 30 70 1500 16 30 54 2000 36 30 34 2150 40 30 30 2350 44 30 26 2500 46 30 24 3000 47 29 24 3500 48 26 26 125 Degree H. M. Days Parasitism I Divide Losses ? Between Large Nymphs 81 Adults - Divide Losses Predation + 2 Among + All Stages _a Vertical Total Number Distribution 9 1 w of Aphids of Aphids 1 + a + Pesticide Rain Daily Mortality Rainfall T Figure 8.7 Flowchart for the mortality section of the aphid population dynamics model. pplicafi Mortality l Third of Bush 126 The number of aphids lost is then integrated over the three regions of the bush and added to predation losses. Aphid losses from this second group of calculations are removed from the distributed delays for all life stages. Parasitoid mortality is modeled using one of three levels of parasitism. The level to be used (high, medium, or low) is selected by the user during the initialization portion of the program. The function describing each level of parasitism is generated using a table look-up function. The input table for this function contains an estimate of percent parasitism every 100 degree days from 1000 to 4500 degree days. These tables are reproduced graphically in Figure 8.8. The lowest level of parasitism represents a site subject to regular pesticide applications (site 2). On the other hand, the high level of parasitism was estimated from 1982 data at site 1 where chemicals were not applied regularly and where a substantial early season aphid population was observed. A.third level of parasitism was provided as an intermediate between these two extremes. Modeling aphid losses due to predation would be a very complicated process if any attempt was made to realistically simulate predator prey interactions for even the two most important aphid predators, chrysopids and syrphids. That process would necessitate an understanding of (l) predator phenologies, (2) predator abundance as a function of aphid density, and (3) predator avoidance of or mortality due to 127 .Enauwnmua _ mo nHu>oH 30H can ..a—apps Hog manage on own: ocean—p.55 main—ooh“ manna. m6 aha—own goo cocoon. oo§§oo< oooe. ooom Doom - m r b 1 301- (\1 528.2 :9: coop 10 19 “159393de 128 pesticides. Such detail is obviously beyond the scope of the present project. Instead, predation will be allowed to remain the one unknown process in the model. During model testing, predation will be estimated as the difference between field samples and model output. Once all other (model variables have been fitted to field data, predation estimates will be calculated and stored in a table look-up. The applicability of these estimates can then be tested using an independent data set. Rain mortality is estimated using the regression statistics from Table 7.1. These equations estimate daily survivorship, Yi' as a function of rainfall for three regions in the blueberry bush. Yu = 0.9905 - 0.4674 * X (8.7) Ym = 0.8894 - 0.3907 * X (8.8) Y1 = 0.9901 - 0.4061 * X (8.9) where, X = rainfall, in inches u = upper, m = middle, 1 = lower third of bush Simulating the effects of pesticide applications on the aphid population is also beyond the scope of the present work. After the model has been fitted to field data and predation has been estimated, pesticide mortality can be incorporated into the model. Pesticide mortality will be calculated using existing results from dosage mortality experiments with I, pepperi (Ramsdell et al. 1983, 1984), and published decay curves of residuals for these chemicals. 129 8.4 Discussion The design and parameterization phases of building the model are now complete. However, the model has yet to be fully tested and validated against an independent data set. Until these steps have been completed, only limited trust should be placed in the model. The process of model testing is itself a: time consuming endeavor involving running the model under a variety of conditions, sensitivity testing, and trial and error adjustments. But testing will also have a very tangible result. This result will be an estimate of predation losses across time. A few words must be said regarding the differences between 'natural' sites and chemically intensive site management. Many of the model's parameters were determined from field results for a site where pesticides were not used. This site provided an opportunity to view the system without the perturbations that would result from regular pesticide use. The relative stability of this system made it much easier to identify the ecological relathmufldps affecting the aphid population. However, the addition of pesticides to a system does not merely kill a few insects; it changes the structure of that system. As pesticide use increases, the levels of predation and parasitism are drastically curtailed. The biological feedback mechanisms that keep the aphid population in check are short-circuited. That is why three levels of parasitism were incorporated into the model. The different levels reflect changes in the 130 structure of the ecosystem. Likewise, it may be necessary to provide multiple predation estimates to reflect changes in predator adundance caused by regular chemical suppression strategies. One shortcoming of this simulation is that model output represents expectations for samples conducted using the 'standard sample' technique. Since this method has not been calibrated against an absolute sample, model results cannot be transformed into absolute population estimates. This means that, at present, the model cannot provide estimates of aphids on a per bush or per acre basis. CHAPTER 9. CONCLUSION This thesis has explored many aspects of aphid ecosystems. It has also attempted to structure our knowledge of the I, pepperi agro-ecosystem into a predictive population model. The thesis stresses model design and parameterization. Model testing and validation are equally important: but, being beyond the scope of the present work, have been relegated to future efforts. At this point it seems appropriate to review what has been learned about the I, pepperi system during the course of this research. Discussion revolves around four general topics: (1) natural enemies, (2) vertical within bush distribution, (3) aphid fecundity, and (4) aphid mortality. Conclusions involving this aphid's natural enemies are several. First, as one would expect, natural enemies were much more common at a site where insecticides were not applied, than at a site subject to regular chemical suppression strategies. Second, although fungal pathogens in the genus Entomophthora are an important mortality agent in several other aphid ecosystems, they do not appear to play a significant role in the present system unless bushes are caged. Third, syrphids and chrysopids appear to be the most important predators in this ecosystem. Of the two, syrphids are the best candidate for the title 'most important predator' because they are more tolerant to insecticides, they consume about twice as many aphids during 131 132 their lifetime, and they occur as a large complex of species. ‘Fourth, at least two species of primary parasitoids and one hyperparasitoid species are present in blueberry fields. Both primary parasitoids are members of the braconid family Aphidiidae. The first of the two species is present in the field from about 1500 degree days (base 38F) until the end of the growing season. The second species does not appear before 2500 degree days. Primary parasitoid species are easily distinguishable in the field by the appearance of their pupae. Although parasitoids were a significant mortality factor (lo-18%) at an unsprayed site, the rate of parasitism remained below 2% at a chemically treated site. Results regarding the vertical distribution of I, pepperi within highbush blueberry bushes agree well with what was previously known about this aphid's habits. Early ill the growing season the bulk of the aphid population is located in the bottom third of the bush. As the season progresses the proportion of the population in the bottom third of the bush declines, with most of the aphids residing on new growth in the upper third of the bush. Late in the growing season this trend is reversed as aphids recolonize the base of the bush. By late August the majority of aphids are found on short, late season terminal shoots growing around the crown of the bush. The middle third of the bush consistently contains the smallest proportion of the population. I, pepperi's parasitoids do not exhibit similar 133 shifts in their vertical distribution. Instead, more than half of the parasitoid population is consistently found in the bottom third of the bush. Two aspects of aphid fecundity were addressed in the thesis. First, results from a field experiment indicate that the aphid's fecundity declines seasonally from a high of 24 young per female early in the growing season to less than three young per female by late August. Although this function was determined using a linear model, future work may well reveal that it is in fact a polynomial relationship. Second, a re-analysis of Elsner's (1982) fixed temperature rearing experiment revealed the temporal distribution of births during a stem mother's reproductive period. Both of these relationships were taken into account when modeling I. pepperi's fecundity. Four potential sources of aphid mortality were also designed into the pOpulation model: predation, parasitism, rainfall, and pesticide applications. Of these, parasitism and rain induced mortality were estimated experimentally during the course of this research. Results from field sampling were used to generate three equations describing percent parasitism as a function of degree days. The functions depicted high, medium, and low levels of parasitism. Rain mortality was described using a different linear function for each of three vertical regions in the bush. Rain induced mortality was greatest in the bottom third of the bush, and least in the middle of the bush. 134 Heavy rains were often responsible for more than 50% mortality to the aphid population. The impact of predators remains to be determined during model testing. Pesticide mortality will also be estimated at a later date using results from dosage mortality experiments with I, e eri and published curves of chemical residues across time. Before the aphid population model can be used with any confidence it will be necessary to test and validate it. Testing consists of fine tuning the model to reproduce results from systematic field samples used to generate its various state and rate functions. Validation involves comparing model results against an indeopendent data set, ie. one not used in the construction of the model. One of the weakest links in the present model is the estimate of the length of the aphid's adult life stage. Due to unforeseeable problems in the fixed temperature experiment, the estimate for this variable was arrived at very subjectively. Another shortcoming of the model is that its results are only applicable when the system is sampled using the 'standard sample' method. Since this systematic sample has not been compared against any absolute sampling method, model results cannot be transformed into absolute units, ie. aphids/bush or aphids/acre. To conclude, I should like to reflect for a moment on the uniqueness of aphids as the object of scientific scrutiny. Like many other insects their small size and short generation time makes them ideal experimental 135 subjects. But aphid ecosystems are also a complex web of interdependent relationships. Most aphid species are highly polymorphic, exhibiting a gradation of forms from completely wingless to functionally alate and combining both parthenogenic and sexual reproduction. They are also one of the most multivoltine insect famdlies, with some species undergoing more than twenty generations per year. In addition, they serve as food for a wide range of predators and parasitoids. These complexities increase the difficulties associated with experimentation, particularly regarding proper experimental controls. The study of aphid morph determination provides a case history of some of these potential difficulties, Such difficulties are major challenges to those who desire to gain a better understanding of aphid ecosystems through modeling. Modeling is not merely a process of mathematizing an ecosystem. Of necessity, the process also results in an abstraction and simplification of the system in question. One of the most useful results of the modeling process may well 1M3 that errant model results help identify important relationships of the system that are either absent from the model or poorly understood. Hopefully, the research presented here has succeeded in spotlighting some of the more important relationships in the I, pepperi agro- ecosystem. APPENDIX 1 Voucher Specimens 136 APPENDIX 1 Record of Deposition of Voucher Specimens* The specimens listed on the following sheet(s) have been deposited in the named museum(s) as samples of those species or other taxa which were used in this research. Voucher recognition labels bearing the Voucher No. have been attached or included in fluid-preserved specimens. Voucher No.: 1987-3 Title of thesis or dissertation (or other research projects): A VARIABLE LIFE-TABLE FOR Illinoia pepperi (MaCGillivray) Museum(s) where deposited and abbreviations for table on following sheets: Entomology Museum, Michigan State University (MSU) Other Museums: Abbreviations: OV oviparous female AL viviparous, alate female AP viviparous, apterous female Investigator's Name (3) (typed) Robert Delain Kriegel II Date July 28, 1987 *Reference: Yoshimoto, C. M. 1978. Voucher Specimens for Entomology in Nerth America. Bull. Entomol. Soc. Amer. 24:141-42. Deposit as follows: Original: Include as Appendix 1 in ribbon copy of thesis or dissertation. Copies: Included as Appendix 1 in copies of thesis or dissertation. Museum(s) files. Research project files. This form is available from and the Voucher No. is assigned by the Curator, Michigan State University Entomology Museum. 137 APPENDIX 1.1 Voucher Specimen Data 1 of 5 Pages Page mum: HOBmpso hmmHlQDble ouflH euwmuo>wcb eunum cwwfizofiz one ca uamoomo .Esompz hwoaoEOucm pom mowefiooom noumfia o>obm ecu ow>aooom hlhmma .oz umcoso> Como»: A3352 plobmemoa/sH Homoflnx unonom Amummmmom: we mucosa apnoeuaoom mmsv emmOQmNHoo .N no venom .m .Hoo .Hmmarmmmiam Amcmsaomom Ne mvm onHm .Hz .m>HHO Dmm3 om: um .00 msmnuo "Hz A.oomze anemone macaaaaH .momuob. >0 Edmonemnoo .M so umcmam .m .Hoo .Hmman>021m mem pamflm .Hz .m>aao pmoz sz x .00 esoppo “HE A.womzv flummmom meocHHHH Edmonemnoo .M so panes .m .Hoo .Hmmnummmnam mam mamas .Hz .w>aao name on: >0 .ou nzmuuo “Hz A.womzv Hummmmm maocHHHH Monmam .m .Hou ommfiromoum .Hz sz x .o>HHo ommz ..oo m3muuo "H2 A.oomzv “seamen meocHHHH HmcmHm .m .Hou omma1>oziam .Hz sz x .o>flao Dmoz ..oo nznppo "Hz A.Uonzv Hummmom MHOQAHHH muo+ oouwmoomo pom pom: Mo omuuoaaoo coxmu porno no mmfiomom m e r r m m e .m % mooanoom now some Hmong erOdEIIHPWS s e D.e .n u u D. m .5 Uhettdduyag M waai no A Auar N 1t E “mo pmnfiaz 138 APPENDIX 1.1 Vbucher Specimen Data Pages _5__ of _2_ Page >uumum>wc= wumum :mwfinuwz ucu :« uwmoamc mama paumuau Smflnqobumm 33 .e:om=: >mo~oEOucm new macawumnm vmumwm m>oam msu vm>amumm 733 .3 $625 \ammmauMIHHanQMIll Avmahuv Amvmamz m.u0umwwumm>cn Ahummmmuwc ma mumonm Hm:0auwvvm omav Edmonawuoo .Mvummc any» can ca Hmcmam .m .HOU .ommHIADbIoH Hz .m>aao ummz A.womz. Hummmmm MaocflHHH sz v Afi .00 mzmuuo "Hz .%mmumh. .>o Edmonawuoo .M :0 HMCmHm .m .Hou .HmmHIADbION mam namflm ..Hz .m>HHo puma sauce umu sz md ..00 m3muuo «Hz A.oomzv Hummmmm mwoaHHHH EdmOQEMHoo rm :0 uflumm .m .Hou .Hmmanmmmuam Amcmawommm m. mwm vamflm ..Hz ~9,530 ummz sz A4 ..00 mzmuuo "HS A.womz. Huummmm MHOGflHHH Juo+ confluence cam tum: no amuumafioo coxwu umnuo no mauumam m e r r m m e .m u mausuumnm new mumv Hanna erOdealla W8 “mummMMWMan uuw.a¢L O .n A n: N Tu E "mo umneaz 139 APPENDIX 1.1 voucher Specimen Data Pages .31. of _§_. Page mum: heumpzo bmmfiuqanumm ou2_ .eaomsz >mc~o30ucm %Ufi@k0>dc= Madam Gfiwfigua 0.58 a.“ UdQOn—OU HOW mcwfl—wumn—w UUUOfiH U>ODQ 0:0 00>fl00g Nut. N- m m H o 02 H0£UGO> HammauMIuHMAQMIIII AvmahuV Amvuamz m.u0uaw«umu>cu Aaummmmuoa ma muumsm docouuwvvu ouav Dm:_ Dmy_ Dméa Dmi— .wmmuwh. .>o Edmonamuoo .M :0 nmmmflum .m .HOU..~mmHtZDh+wN m 00m Smam maa .um nuvm ~Hz .aofluocsn manna m4 .00 :muum cm> ”H2 .wmmumh. .>o Edmonemuoo aw :0 nmmmflnx .m .Hou .mmmHIZDblqm m 0mm 3mHm was Hz .coauocsb wcmuw Ad ..00 Gmnsm 20> "H2 . mmuwn. .>o Eamonswuoo,.w no mwmflum .m .Hoo..mmmarzannMH mm 0mm 3mm zme .m3m oamme A4 2 .muuoHumnu ..oo coumm "Hz ESWOQEMHOU .N :0 wunuaso macaw Hmmmflum .m .Hou .mmmfi-mm«-¢ ..mcflm 0mm .smz a: .mcfimcmq ummm ..oo amamcH “Hz :3 Amcmfiflommm m. A.womz. Hummmmm mflocHHHH Amcmaflommm my A.womzv Huwmmmm MHOQHHHH A.womzv Hummmmm mHozHHHH A.oomzv Hummmmm aflocflHHH Museum where depos- ited Other vuufimoauc can now: no vuuumaaou acmeauoam new mama Huang Adults 0' Adults 9 Lannm Eggs Pupae lhmmhs coxmu umsuo no moauuam O“. O Hugsaz 140 APPENDIX 1.1 Vbucher Specimen Data Pages of 5 .£_ Page mama xugmum>gca wumum cquzofiz uzu a“ uumommc acumuao ammfluqonumm wag. .e=mm=z >wo~080ucm yew newEuumam vmumqa m>onm mnu mm>umumm blbmma .oz non—35> Gun»: A3352 mlewUmmZfi Axummmmuw: ma muwmsm unscauanem mmav Fammwux #Hwflom Om:— DmSH Dmgn Dmrm md. ad. 1 1 ammmflux .m .Hoo .Nmmfiuqonumfi m 0mm zmfim may .pm spam .Hz .cofiuocsn wcmuw .OU cmusm cm> “Hz wmmumb. .>o Edmonemuou 4M no Hmmmaux .m .HOU .mmmHIqulom mm 0mm 3mm ZNB .wam oemHmm flz .muuoauwno ..OU coumm ”H2 sea esflHOMflpmnmcm .M wees no .mmmflux .m .Hou .Nmmfiuooaumfl 0mm 35m qus ..CH H0>Hm #50 ..OU omcflxomz ”Hz ammumn. .>o Edmonfimnoo .M co awmmflum .m .HOU .Nmmalzoblwm m omm zmfim may .um spam .Hz .:0flpocsn ozone ..oo cmusm cm> “Hz Amcmfiflommm m. A.UUME~ Huwmmmm MHOCHHHH heada wanna cmufluflmmumm A.womz. flummmmm wwocHHHH QQHOE wmu A.womzv Hummmmm aflocfiHHH A.Uomzv fluwmmmm MMOGHHHH Museum where depos- ited Other Adults 0' Adults 9 Pupae NVmphs Larvae Eggs vmuamomwm mam man: no vmuooaaou mauawuonm new «use Hanna :oxmu umnuo no muuuoam “mo amass: mama neumpao ammauqsnsmm mug. .Eammaz hwoaoEOucm >uamum>wca mumum :mw«:Ufiz Una :« uumommc new mamaaomam vmumwa m>onm mnu vm>wmumm plumma .oz uuzu=o> Ammahuv Hmmmflnm unmnom Amvmemz m.u0umw«umm>cm Axummmmumc «a mummsm anneauwvvm mmsv Pages 5 of APPENDIX 1.1 Vbucher Specimen Data 5 141 Page .%wmumb. .>o EdmonEMHoo .M :0 Ammmflum .m .HOU .mmmHI>OZIm~ mm com 3mm ZNB .w3m oEmme x .52 .ouuoaumsu ..ou :oumm “Hz “mmumb. .>o Edmonfihuoo .M :0 cOHm33 .2 .HOU..mmeIM¢SIMH H 0mm Smam mHB .eumm mxcom .m Afi ~.00 cwnsm cm> ”Hz - \. EdmOQEMHoo .M :0 Hmcmam .m .Hoo .Hmmalqohlvm mwdvm GHme .HS .m>fiHO #wmg ..OU m3muu0 «HZ Q; at A.wom2v wummmmm MHOGHHHH Amcwsflommm N. A.womzv Hammmmm aflocHHHH Amcmfiflommm vv sauce own A.Uomzv Hummmwm MflOGHHHH JVO? vmuumoamc can mum: Ho vmuumaaou :ma omen we a m w a mmwduMMnmms a. .22.. s e.Pn..n nu u .9 8 t pe.d Au u .4 8 WW.@.10AAPM.LE :oxmu umnuo uo moauoam "mo umnaaz APPENDIX 2 Distribution Map Data APPENDIX 2. Table A2.1 Localities in Michigan where I, 23222;; has been collected. - County Location Year Host15 Source16 Alger Adams Trail (Co. 637) 1984 angustifolium J.H. & Ausable Point rd.. T48N, R16W, Sec 29 NW Alger Lk. Superior shoreline 1984 corymbosum J.H. West of Beaver Lk., T48N, R17W, Sec 13 Allegan T4N, R15W, Sec 30 1962 corymbosum ENT Allegan Fennville, MI 1963 corymbosum ENT Berrian Shawnee & Cleveland rds. 1962 corymbosum ENT Berrian Hawthorne & Cleveland 1964 corymbosum ENT rds. Berrian St. Joseph, MI 1964 corymbosum ENT Berrian New Buffalo, MI 1965 corymbosum ENT Berrian North Watervillet rd. & 1965 corymbosum ENT Indiana State line Berrian Sawyer, MI 1965 corymbosum ENT Crawford Hwy I—75 South rest stop 1981 myrrtilloides E.E. Eaton Charlotte, MI, Kalamo 1981 corymbosum R.K. Hwy, T2N, RSW, Sec 22 SW ‘ Ingham Dannsville St. Game ‘ 1981 corymbosum17 E.E. Area. Potter rd.. T2N, RlE, Sec 32 W Mackinaw Cut River rd., 1982 aggustifolium M.W. T43N, R6W, Sec 5 Montcalm Sidney, MI 1961 corymbosum ENT Muskegan Quarterline & Pontaluma 1962 corymbosum ENT rds. Muskegan Giles & Buys rds. 1962 corymbosum ENT Ogemaw Mills Township, 1981 aggustifolium E.E. R3E, T21N, Sec 7 SW 15 All host plants are Vaccinium sp. Unless otherwise noted all specimens collected on Y, corymbosum were captured on cultivated plants. All non corymbosum hosts represent captures from non-cultivated sites. 16 Abbreviations for data sources are as follows: J.H. = Dr. Jim Hancock, ENT = Department of Entomology museum specimens. E.E. = Erwin Elsner, M.W. = Dr. Mark Whalon, K.M. = Kathy Morimoto, R.K. = Robert Kriegel. 17 Collected on wild Y, corymbosum. 142 Table A2.1, continued. County Location Year Host Source Oscoda USFS rd. #4147 1981 angustifolium E.E. R3E, TZSN. Sec 7 SE Ottawa Coopersville, MI 1981 corymbosum E.E. Ottawa West Olive, MI 1981 corymbosum E.E. Ottawa Barry rd. at 144th St., 1983 corymbosum K.M. T6N, R16W, Sec 36 Ottawa 84th St. & Haynes rd.. 1983 corymbosum K.M. T8N, R14W, Sec 32 Schoolcraft M77 at E. Branch Fox R. 1982 angustifolium M.W. T46N, R13W, Sec 4 SW Schoolcraft Sable St. Forest, 1983 aggustifolium R.K. Stanley Lk. Campgrnd, on Little Fox River, T47N, R15W, Sec 11 13 Van Buren 0.3 mi. E. of 72nd St. 1962 corymbosum ENT and 36th Av. Van Buren Hartford, MI 1964 corymbosum ENT Van Buren Grand Junction, Mi., 1982 corymbosum R.K. 54th St, TlS, R15W, Sec 9 SW Van Buren MI. Blueberry Grower's 1982 corymbosum R.K. Association research plot Grand Junction, MI, 54th St.. T18, R15W, Sec 8 Van Buren 54th St.. T18, R15W, 1983 corymbosum M.W. Sec 4 SW Van Buren 56th St., T18, RlSW, 1983 corymbosum M.W. Sec 8 NW Van Buren 57th St.. T18. R16W, 1983 corymbosum M.W. Sec 1 E Van Buren 95th St. & Phoenix rd., 1983 corymbosum M.W. TIS, R16W, Sec 1 18 Wild demonstration site managed by the Michigan Department of Natural Resources. BIBLIOGRAPHY BIBLIOGRAPHY Abkin, Michael H. and Chris Wolf. 1976. Distributed delay routines: DEL, DELS, DELF, DELLF, DELVF, DELLVF. Computer Library for Agricultural Systems Simulathmn (CLASS) document No. 8. Dept. of Agricultural Economics, Michigan State University, East Lansing, MI. Unpaged. Addison III, Edwin S. 1969. Forecast of blueberry harvest dates as determined by climatology. MS Thesis. Rutgers University, New Brunswick, NJ. 95p. Anscombe, FHJZ 1948. The transformation of poisson, binomial, and negative-binomial data. Biometrika. 35:246-254. Bartlett, B.R. 1964. Toxicity of some pesticides to eggs, larvae, and adults of the green lacewing, Chrysopa carnea. J. Econ. Entomol. 57(3):366-369. Baskerville, G.L. and P. Emin. 1969. Rapid estimation of heat accumulation from maximum and minimum temperatures. Ecology. 50(3):514-517. Baumgaertner, J.U., A.P. Gutierrez and C.G. Summers. 1981. The influence of aphid prey consumption and searching behavior, weight increase, developmental time, and mortality of Chrysopa carnea (Neuroptera:Chrysopidae) and Hippodamia convergfns (Coleoptera:Coccinellidae) larvae. Can. Ent. 113:1007-1014. ~Beckwith, Charles EL. 1943. Insects attacking blueberry fruit. New Jersey Agricultural Experiment Station Circular No. 472. 4p. Blackman, R.L. 1971. Variation in the photoperiodic response within natural populations of Myzus persicae (Sulz.). Bull. Ent. Res. 60:533-546. . 1972. 'The inheritance of life-cycle differences in Myzus persicae (Sulz.) (Hem., Aphididae). Bull. Ent. Res. 62:281-294. 144 145 Borror, Donald J., Dwight M. DeLong and Charles A Triplehorn- 1981. IU1 Introduction to the Study 0 Insects. New York: CBS College Publishing. 827pp. rho Broadbent, L. 1957. Insecticidal control of the spread of plant viruses. Ann. Rev. Entomol. 2:339-354. . 1965. The importance of alate aphids in virus spread within crops. Proc. 12th Intern. Congr. Entomol. 1964:523-524. Burger, Thomas In 1966. A survey for possible virus disease vectors occurring on the cultivated blueberry Vaccinium corymbosum Linnaeus. MS Thesis. Michigan State University, East Lansing, MI. 64p. Campbell, A., B.D. Frazer, N. Gilbert, A.P. Gutierrez and M. Mackauer. 1974. Temperature requirements of some aphids and their parasites. J. Appl. Ecol. 11:431- 438. Campbell, A. and M. Mackauer. 1975. Thermal constants for development of the pea aphid (Homoptera:Aphididae) and some of its parasites. Can. Ent. 107:419-423. Cartery Ifl., O.P. Aikman and A.P.G. Dixon. 1978. An appraisal of Hughes' time-specific life-table analysis for determining aphid reproductive and mortality rates. J. Anim. Ecol. 47:677—687. Dadd, R. H. 1968. Dietary amino acids and wing determination in the aphid, Myzus persicae. Ann. Ent. Soc. Am. 6(5): 1201- 1210. Dixon, A.P.G. 1972. Crowding and nutrition in the induction of macropterous alate in Drepanosiphmn dixoni. J. Insect Physiol. 18:459-464. . 1975. Seasonal changes in fat content, form, state of gonads, and length of adult life in the sycamore aphid, Drepanosiphum platanoides (Schr.). Trans. R. Ent. Soc. Lond. 127(2):87-99. . 1976a. Reproductive strategies of the alate morphs of the bird cherry-oat aphid Rhopalosiphum padi L. Journ. Anim. Ecol. 45(3):817-830. . 1976b. Timing of egg hatch and viability of the sycamore aphid, Drepanosiphum platanoids (Schr.). at knui burst of sycamore, Acer pseudoplatanus L. J} Anim. Ecol. 45(2):593-603. 146 . 1977. Aphid ecologyw life cycles, polymorphism, and population regulation. Ann. Rev. Ecol. Syst. 8:329-353. Dixon, A.P.G. and N.D. Barlow. 1979. Population regulation in the lime aphid. Zool. Journ. Linn. Soc. 67:225— 237. Dixon, A.F.G. and T.R. Dharma. 1980. Number of ovarioles and fecundity 1J1 the black bean.aphid, Aphis fabae. Ent. Exp. & Appl. 28(1):1-14. Dixon, A.P.G. and D.M. Glen. 1971. Morph determination in the bird cherry-oat aphid, Rhopalosiphum padi L. Ann. Appl. Biol. 68:11-21. Dixon, A.P.G. and S.D. Wratten. 1971. Laboratory studies on aggregation, size, and fecundity in the black bean aphid, Aphis fabae Scop. Bull. Ent. Res. 61:97-111. Elsner, Erwin A” 1982. Biology of Illinoia pepperi (MacG.), the blueberry aphid, and its relation to blueberry shoestring virus disease in western Michigan. MS Thesis. Michigan State University, East Lansing, MI. 123p. Elsner, E.A. and M.E. Whalon. 1980. Blueberry aphids and blueberry shoestring virus disease in western Michigan. pp. 175-196 in J.N. Moore (ed.). Proc. 4th N. Am. Blueberry Res. Workers Conf., Oct. 16—18, 1979. Fayetteville, Ark. 396p. . 1985. Developmental rate and longevity of Illinoia pepperi (Homoptera: Aphididae) on excised blueberry leaf disks. Great Lakes Enmmmfld 18(4):155-158. Frazer, B.D. and A.R. Forbes. 1968. Masonaphis maxima (Mason) (Homoptera:Aphididae), an aphid on thimbleberry with an unusual life history. J. Entomol. Soc. Brit. Col. 65:36-39. George, K.S. ‘1957. Preliminary investigations on the biology and ecology of the parasites and predators of Brevicoryne brassicae (L.). Bull. Entomol. Res. 48:619-629. Gilbert, 11. 1980. Comparative dynamics of a single-host aphid. II. The evidence. Journ. Anim. Ecol. 49:351- 369. Gilbert, N. and A.P. Gutierrez. 1973. A plant-aphid- parasite relationship. Journ. Anim. Ecol. 42:323-340. 14? Gilbert, N., A.P. Gutierrez, B.D. Frazer and R.E. Jones. 1976. Ecological Relationships. W.H. Freeman and Co., Ltd., Great Britian. 157p. Gilbertq IN. and R.D. Hughes. 1971. A model of an aphid population -- three adventures. JOurn. Anim. Ecol. 40:525-534. Giles, Francis E. 1966. The seasonal history and geographic distribution of aphids infesting raspberry, blueberry and strawberry in the lower peninsula of Michigan. Ph.D. Dissertation. Michigan State University, East Lansing, MI. 149p. Gillett, J.M., K.M. Mbrimoto, D.C. Ramsdell, K.K. Baker, W.G. Chaney and W.J. Esselman. 1982. A comparhun1 between the relative abilities of ELISA, RIA, and ISEM to detect blueberry shoestring virus in its aphid vector. Acta Horticulturae. 129:25-29. Gutierrez, A.P., ILJI, Morgan and D.E. Havenstein. 1971. The ecology of Aphis craccivora Koch and subterranean colver stunt virus I. The phenology of aphid pOpulations and the epidemiology of virus in pastures in south-east Australia. J. Appl. Ecol. 8:699-721. Gutierrez, A.P., D.E. Havenstein, H.A. Nix and P.A. Moore. 1974a. lThe ecology of Aphis craccivora Koch and subterranean clover stunt virus II. A model of cowpea aphid populations in temperate pastures. J. Appl. Ecol. 11:1—20. Gutierrez, A.P., H.A. Nix, D.E. Havenstein and P.A. Moore. 1974b. 'Phe ecology of Aphis craccivora Koch and subterranean clover stunt virus. III. A regional perspective of the phenology and migration of the cowpea aphid. J. Appl. Ecol. 11:21-35. Hagen, K.S. and R. Van den Bosch. 1968. Impact of pathogens, parasites, and predators on aphids. Ann. Rev. Entomol. 13:325~367. Hancock, James F., Nancy L. Schulte, James H. Siefker, Marvin P. Pritts and Janet M. Roueche. 1982. Screening highbush blueberry cultivars for resistance to the aphid Illinoia pepperi. Hortscience. 17(3):362-363. Harris, K.M. 1973. Aphidophagous Cecidomyiidae (Diptera): taxonomy, biology, and assessment of field populations. 148 Harris, Kerry FR anui Karl Maramorosch. 1977. Aphids as Virus Vectors. New York: Academic Press. Hartmann, James, James E. Bath and Gary R. Hooper. 1973. Electron microscopy of viruslike particles from shoestring-diseased highbush blueberry, Vaccinium corymbosum L. Phytopathology. 63:432-436. Hille Ris Lambers, D. 1966. Polymorphism in Aphididae. Ann. Rev. Ent. 11:47-78. Hughes, R.D. 1962. A method for estimating the effects of mortality on aphid populations. JOurn. Anim. Ecol. 31:389-396. . 1963. Population dynamics of the Cabbage aphid, Brevicoryne brassicae (L.). Journ. Anim. Ecol. 32:393—424. Hughes, R.D. and N. Gilbert. 1968. A model of an aphid population -- a general statement. Journ. Anim. Ecol. 37:553-563. Johnson, Bruce. 1965. Wing polymorphism in aphids. II. Interactions between aphids. Ent. Exp. & Appl. 8:49- 64. . 1966. Wing polymorphism in aphids. III. The influence of the host plant. Ent. Exp. & Appl. 9:213—222. Johnson, Bruce and P.R. Birks. 1960. Studies on wing polymorphism in aphids. The developmental process involved 511 the production of different forms. Ent. Exp. & Appl. 3:327-339. Kriegel, Robert D. and Mark E. Whalon. 1982. A simulation model of the population dynamics and dispersal of the blueberry aphid, Illinoia pepperi (MacG.). Ann. Mtg. Entomol. Soc. Am., Toronto, Canada. . 1983. A life—table for the blueberry aphid, Illinoia pepperi (MacG.). Ann. Mtg. Entomol. Soc. Am., Detroit, MI. Krombein, Karl V., Paul D. Hurd Jr., Davis S. Smith and B.D. Burks. 1979. Source Catalog of Hymenoptera 12 America North g_f_ Mexico. Smithsonian Institution Press, Washington D.C. 149 Lees, A.D. 1966. The control of polymorphism in aphids. In Beament, J.W.L., J.E. Treherne and V.B. Wigglesworth (Eds). Advances in Insect Physiology.‘ Third Edition. pp 207-277. London: Academic Press. . 1967. Direct and indirect effects of day length on the aphid Megoura viciae Buckton. J. Insect Physiol. 13:1781-1785. LeFevre, V.F. and R.G. Adams. 1983. Bibliography of the aphid predator, Aphidoletes aphidimzya (Diptera:Cecidomyiidae). Bull. Ent. Soc. Am. 28(2):129—133. Lesney, M.S. and D.C. Ramsdell. 1976. Ihuification and some properties of blueberry shoestring virus. Acta Horticulturae. 66:105-109. Lesney, M.S., D.C. Ramsdell and Mike Sun. 1978. Etiology of blueberry shoestring disease and some properties of the causal virus. Phytopathology. 68:295-300. Lockhart, G.L. and I.V. Hall. 1962. Note on an indication of shoestring virus in the loWbush blueberry Vaccinium angustifolium Ait. Can. J. Bot. 40:1561-1562. Lowe, A.D. (ed). 1973. Perspectives _i_g1_ Aphid Biology. Bulletin No. 2 of the Entomological Society of New Zealand, Aukland, New Zealand. MacGillivray, M.E. 1958. A study of the genus Masonaphis Hille Ris Lambers, 1939 (Homoptera:aphididae). Temminckia. 10:1-131. Mack, T.P. and Z. Smilowitz. 1980. The development of a green peach aphid natural enemy sampling procedure. Environ. Entomol. 9:440-445. Mackauer, Manfred and Thelma Finlayson. 1967. The hymenopterous parasites (Hymenoptera:Aphidiidae et Aphelinidae) of the pea aphid in North America. Can. Ent. 99:1051-1080. Manetsch, Thomas L. 1976. Time—varying distributed delays and their use in aggregative models of large systems. IEEE Transactions of Systems, Man, and Cybernetics. 6(8):547-553. 150 Manetsch, Thomas J. and Gerald L. Park. 1980. System Analysis and Simulation with Applications to Economic and Social Systems. Third Edition, 2 volumes. Department of Electrical Engineering and Systems Science, Michigan State University, East Lansing, MI. Messenger, P.S. and D.C. Force. 1963. An experhmnnal host-parasite system: Therioaphis maculata (Buckton) -- Praon palitans Muesebeck (Homoptera: Aphididae-- Hymenoptera: Braconidae). Ecology. 44(3):532-540. Miller, C.A. 1955. A Technique for assessing spruce budworm larval mortality caused by parasites. Can. J. Entomol. 33:5-17. . 1959. The interaction of the spruce budworm, Choristoneura fumiferana (Clem.), and the parasite Apanteles fumiferanae Vier. Can. Entomol. 41(8):457- 477. Miller, C.A. and T.R. Renault. 1973. Notes on parasitism among endemic budworm populations. Canadian Forestry Service Report, MAritimes Forest Research Centre, Fredericton, New Brunswick. Mittler, T.E. and R.H. Dadd. 1966. Food and wing determination in Myzus persicae (Homoptera:Aphidae). Annals Ent. Soc. Am. 59(6):1161—1166. Mittler, T.E. and H. Kunkel. 1971. Wing production by grouped and isolated apterae of the aphid Myzus persicae on artifical diet. Ent. Exp. & Appl. 14:83- 92. Mittler, T.E. and O.R.W. Sutherland. 1969. Dietary influences on aphid polymorphimn. Ent. Exp. & Appl. 12:703-713. Morimoto, Kathryn M. 1984. Aphid transmission of blueberry shoestring virus and seasonal populations of its vector Illinoia pepperi (MacGillivray). MS Thesis. Michigan State University, East Lansing, MI. 137p. Neuenschwander, P. 1975. Influence of temperature and humidity (”1 the immature stages of Hemerobius pacificus. Environ. Entomol. 4(2):215-220. . 1976. Biology of adult Hemerobius pacificus. Can. Ent. 5:96-100. 151 Obrycki, John J. and Maurice J. Tauber. 1978. Thermal requirements of Coleomegilla maculata (Coleoptera: Coccinellidae) and its paraiste Perilitus coccinellae (Hymenoptera: Braconidae). Can. Ent. 110:407-412. . 1981. Phenology of three coccinellid species, thermal requirements for development. Ann. Entomol. Soc. Am. 74:31-36. Obrycki, John J., Maurice J. Tauber, Catherine A. Tauber and Brian Gollands. 1983. Environmental control of the seasonal life cycle of Adalia bipunctata (Coleoptera: Coccinellidae). Environ. Entomol. 12:416-421. Peterson, Maureen A” 1984. Localization of blueberry shoestring virus (BBSSV) in its aphid vector, Illinoia pepperi (MacGillivray). MS Thesis. Michigan State University, East Lansing, MI. 95p. Propp. Gary D., Maurice J. Tauber and Catherine A. Tauber. 1969. Diapause in the neuropteran Chrysopa oculata. J. Insect Physiol. 15:1749-1757. Ramsdell, D.C. 1979a. Physical and chemical properties of blueberry shoestring virus. Phytopathology. 69:1089- 1091. . 1979b. Blueberry shoestring virus. Descriptions of plant viruses No. 204. C.M.I./A.A.B. Kew, Surrey, England. unpaged. Ramsdell, D.C., M.E. Whalon and J. Hancock. 1980. A proposal for research dealing with an integrated approach to blueberry shoestring virus disease control. Report to U.S.D.A., Washington, D.C. 28pp. Ramsdell, I)J3., J.F. Hancock and M.E. Whalon. 1982. An integrated approach to blueberry shoestring virus disease control. A proposal for renewed funding. Unpaged. . 1983. An integrated approach to blueberry shoestring virus disease control. A proposal for renewed funding. Unpaged. . 1984. An integrated approach to blueberry shoestring virus disease control. A proposal for renewed funding. Unpaged. 152 Rao, C. Radhakrishna. 1965. linear statistical inference and its applications. New York: John Wiley and Sons. 522p. Ribbands, C.R. 1965. The significance of apterous aphids in the spread of viruses within crops. Proc. 12th Intern. Congr. Entomol. London, 1964. p. 525-526. Ruesink, W.G. 1976. Status of the systems approach to pest management. Ann. Rev. Entomol. 21:27-44. Schaefers, G.A. 1972. The role of nutrition in alary polymorphism among the Aphididae -- an overview. Search Cornell Univ. 2:1-8. Schulte, Nancy L., James F. Hancock and D.C. Ramsdell. 1984. IDevelopment of a screen for resistance to blueberry shoestring virus. Michigan State University Ag. Exp. Station Journal. Shaw, M.J.P. 1970a. Effects of population density on aliencolae of Aphis fabae Scop. I. The effects of crowding on the production of alate in the laboratory. Ann. Appl. Biol. 65:191-196. . 1970b. Effects of population density on aliencolae of Aphis fabae Sc0p. II. The effects of crowding on the expression of migratory urge among alatae 1J1 the laboratory. Ann. Appl. Biol. 65:197- 203. . 1970c. Effects of population density on aliencolae of Aphis fabae Cop. III. The effect of isolation (n1 the development of form and behavior of alatae in a laboratory clone. Ann. Appl. Biol. 65:205-212. Southwood, T.R.E. Ecological Methods. Chapman and Hall: New York. Sutherland, O.R.W. and T.E. Mittler. 1971. Influence of diet composition and crowding on wing polymorphism by the aphid Myzus persicae. J. Insect Physiol. 17:321- 328. Stace-Smith, R4 1958. Studies on Rubus virus diseases in British Columbia, thimbleberry ring spot virus. Can. Tauber, Maurice J. and Catherine A. Tauber. 1973. Seasonal regulation of dormancy in Chrysopa carnea (Neuroptera). J. Insect Physiol. 19:1455-1463. 153 . 1974. Thermal accumulations, dispause, and oviposition in a conifer— inhabiting predator, Chrysopa harrisii (Neuroptera). Can. Ent. 106:969-978. Tamaki, George, Lee Fox and B.A. Butt. 1979. Ecology of the green peach aphid as a vector of beet western yellows virus of sugarbeets. USDA Technical Bulletin No. 1599. 16p. Tummala, Ramamohan L., Dean L. Haynes and Brian A. Croft (ed.). 1976. Modeling for Pest Management U.S.A./USSR, Concepts, Techniques, and Applications. Michigan State University Press, East Lansing, MI. Tuttle, Donald M. 1947. A study of the insect fauna of the cultivated blueberry, vaccinium corymbosum Linnaeus. MS Thesis. Michigan State College, East Lansing, MI. Varney, E.H. 1957. Mosaic and shoestring, virus diseases of cultivated blueberry in New Jersey. Phytopathology. 47:307—309. . (ed). 1970. Virus diseases of blueberry and cranberry. pp. 173-188 in Virus Diseases 9: Small Fruits and Grapevines. N.W. Frazer (ed). University (of California division of agricultural sciences, Berkeley, CA. 290pp. . 1977. Viruses and mycoplasma-like diseases of blueberry. Hortscience. 12:476-478. Way, M.J. and C.J. Banks. 1968. Population studies on the active stages of the black bean aphid, Aphis fabae Sc0p., on its winter host Euonymus europaeus L. Ann. Appl. Biol. 62:177-197. Welch, S.M. 1984. Developments in computer-based IPM extension delivery systems. Ann. Rev. Entomol. 29:359-382. Welch, S.M., B.A. Croft, J.F. Brunner and MQF. Michels. 1978. PETE: .An extension phenology modeling system for management of multispecies pest complexes. Environ. Entomol. 7(4):487-494. Wellings, P.W., S.R. Leather and A.P.G. Dixon. 1980. Seasonal variation in reproductive potential: a programmed feature of aphid life cycles. J. Amin. Ecol. 49:975-985. 154 Whalon, Mark E. and Erwin A. Elsner. 1982. Impact of pesticides on Illinoia pepperi and its predators. Environ. Entomol. 75(2):356-358. Wratten, ESJ). 1976. Searching by Adalia bipunctata (L.) (Coleoptera: Coccinellidae) and escape behavior of its aphid and cicadellid prey on lime (Tila X vulgaria Hayne). Ecol. Entomol. 1:139-142. . 1977. Reproductive strategy of winged and wingless morphs of the aphids Sitobion avenae and Metopolophium dirhodum. Ann. Appl. Biol. 85:319-331. Wright, E. Jane and J.E. Laing. 1978. The effects of temperature on development, adult longevity, and fecundity of Coleomegilla maculata lengi and its parasite, Perilitus coccinellae. Proc. Ent. Soc. Ont. 109:33-47.