"Land and Food Systems, Faculty of"@en . "DSpace"@en . "UBCV"@en . "Smith, Risa Barbara"@en . "2011-01-24T21:41:55Z"@en . "1990"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "This thesis was designed to address three seldom studied aspects of the relationship between herbivores and vegetational diversity. 1. Interactions between vegetational diversity and herbivore mortality due to predation were assessed by experimentally manipulating both the species diversity of plants and the densities of a common generalist predator, the spider Enoplagnatha ovata. 2. The importance of plant size to herbivore densities was examined by quantifying plant size (measured as plant height, width, leaf area and growth rates) and adjusting for it through covariate analysis. 3. Differences in population responses of several species of herbivores to both vegetational diversity and a predator, were compared by concurrently studying four lepidopterans.\r\nThe main experiment used a two factor design, with two planting treatments and two predator treatments. The planting treatments consisted of plots planted with monocultures of brussels sprouts (Brassica oleraceae) and dicultures of brussels sprouts intercropped with peppermint (Mentha piperita). The natural enemy treatments involved augmentations of E ovata in some plots and untreated controls. Two of the herbivores studied, Plutella xylostella and Pieris rapae are monophagous lepidopterans, specializing on crucifers, while the others, Autographa californica and Mamestra configurata are polyphagous.\r\nFor two species, P. xylostella and M. configurata responses to augmentations of the spider, E. ovata. were different in monocultures and dicultures. Reduced densities of these two species were found in monoculture plots with added spiders; in dicultures increased densities were found in plots with added spiders. This interaction effect points out that generalist predators can be effective in monocultures. I suggest that the importance of natural enemies in monocultures is often overlooked because only the initial colonization phase is being studied. By augmenting predator populations I was able to simulate densities equivalent to those in more established cropping systems. The increased herbivore densities in dicultures with added spiders might be explained by possible predation by E. ovata on other natural enemies of P. xylostella and M. configurata in dicultures but not in monocultures. Supporting evidence for this interpretation lies in the fact that percent parasitism of P. xylostella by the ichneumonid, Diadegma insulare was lower in plots with added spiders than in control plots. Furthermore, parasitism of P. xylostella by D. insulare increased with host density in diculture plots, but not in monoculture plots. Mamestra configurata was not subject to parasitism in this study, precluding assessment of a similar relationship. No A. californica larvae were found in plots with additional spiders. In contrast, P. rapae larvae were not affected by the experimental treatments.\r\nPlant size was a crucial determinant of both herbivore populations and percent parasitism of those herbivores. Most importantly, had plant size not been accounted for, the importance of vegetational diversity to both herbivore densities and percent parasitism would have been overestimated. For example, the incorrect conclusion, that vegetational diversity alone was important in determining the abundance of both of the generalist feeders would have been reached. The greater densities of A. californica in monocultures and M. configurata in dicultures were accounted for by plant size. Without plant size adjustments, percent parasitism of P. xylostella by D. insulare would have been misinterpeted as being greater in monocultures than dicultures. With plant size adjustments, the importance of E. ovata augmentations on lowering percent parasitismwas unmasked. All important interaction effects were discovered only after adjustments for plant size had been made. \r\nDespite the low densities of all herbivore species, significant responses to experimental treatments were found in three of the four species studied. Only P.l rapae was unaffected by any of the treatments. However, conclusions based on the feeding ng habits of the herbivores could not be made. The polyphagous feeders were affected by generalist predation as much as the crucifer specialist. Parasitism was found in only two of the species, P. xylostella and A. californica. Of these two species parasitism of the specialist, P. xylostella was affected by both vegetational diversity and generalist predation, whereas parasitism of A. californica was not.\r\nMy study emphasizes multifaceted interactions between the size and diversity of a primary resource and several trophic levels of consumers. Multifactor models, involving several aspects of a cropping system, are required to uncover the important mechanisms behind variable herbivore responses to vegetational diversity."@en . "https://circle.library.ubc.ca/rest/handle/2429/30791?expand=metadata"@en . "PLANT SIZE, RESOURCE CONCENTRATION AND NATURAL ENEMIES: A COMPARISON OF FOUR HERBIVORES IN MONOCULTURES OF BRUSSELS SPROUTS AND DICULTURES OF BRUSSELS SPROUTS/PEPPERMINT by RISA BARBARA SMITH B.Sc. University of Toronto, 1973 M.Sc. University of British Columbia, 1984 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF PLANT SCIENCE We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA March 1990 \u00C2\u00A9Risa B. Smith, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Plant Science The University of British Columbia Vancouver, Canada Date April 30, 1990 DE-6 (2/88) i i ABSTRACT This thesis was designed to address three seldom studied aspects of the relationship between herbivores and vegetational diversity. 1. Interactions between vegetational diversity and herbivore mortality due to predation were assessed by experimentally manipulating both the species diversity of plants and the densities of a common generalist predator, the spider Enoplagnatha ovata. 2. The importance of plant size to herbivore densities was examined by quantifying plant size (measured as plant height, width, leaf area and growth rates) and adjusting for it through covariate analysis. 3. Differences in population responses of several species of herbivores to both vegetational diversity and a predator, were compared by concurrently studying four lepidopterans. The main experiment used a two factor design, with two planting treatments and two predator treatments. The planting treatments consisted of plots planted with monocultures of brussels sprouts (Brassica oleraceae) and dicultures of brussels sprouts intercropped with peppermint (Mentha piperita). The natural enemy treatments involved augmentations of E ovata in some plots and untreated controls. Two of the herbivores studied, Plutella xylostella and Pieris rapae are monophagous lepidopterans, specializing on crucifers, while the others, Autographa californica and Mamestra configurata are polyphagous. For two species, F\ xylostella and M, configurata responses to augmentations of the spider, E. ovata. were different in monocultures and dicultures. Reduced densities of these two species were found in monoculture plots with added spiders; in dicultures increased densities were found in plots with added spiders. This interaction effect points out that generalist predators can be effective in monocultures. I suggest that the importance of natural enemies in monocultures is often overlooked because only the initial colonization phase is being studied. By augmenting predator populations I was able to simulate densities equivalent to those in more established cropping systems. The increased herbivore densities in dicultures with added spiders might be explained by possible predation by E ovata on other natural enemies of FL xylostella and configurata in dicultures but not in monocultures. Supporting evidence for this interpretation lies in the fact that percent parasitism of P. xylostella by the ichneumonid, Diadegma insulare was lower in plots with added spiders than in control plots. Furthermore, parasitism of FL xylostella by TX insulare increased with host density in diculture plots, but not in monoculture plots. Mamestra configurata was not subject to parasitism in this study, precluding assessment of a similar relationship. No californica larvae were found in plots with additional spiders. In contrast.JL rapae larvae were not affected by the experimental treatments. Plant size was a crucial determinant of both herbivore populations and percent parasitism of those herbivores. Most importantly, had plant size not been accounted for, the importance of vegetational diversity to both herbivore densities and percent parasitism would have been overestimated. For example, the incorrect conclusion, that vegetational diversity alone was important in determining the abundance of both of the generalist feeders would have been reached. The greater densities of A. californica in monocultures and M , configurata in dicultures were accounted for by plant s ize. Without plant size adjustments, percent parasitism of P. xylostella by D. insulare would have been misinterpeted as being greater in monocultures than dicultures. With plant size adjustments, the importance of E. ovata augmentations on lowering percent parasitism was unmasked. All important interaction effects were discovered only after adjustments for plant size had been made. i v Despite the low densities of all herbivore species, significant responses to experimental treatments were found in three of the four species studied. Only f l rapae was unaffected by any of the treatments. However, conclusions based on the feeding ng habits of the herbivores could not be made. The polyphagous feeders were affected by generalist predation as much as the crucifer special ist. Parasitism was found in only two of the species, P. xylostella and A. californica. Of these two species parasitism of the specialist, P. xylostella was affected by both vegetational diversity and generalist predation, whereas parasitism of A. californica was not. My study emphasizes multifaceted interactions between the size and diversity of a primary resource and several trophic levels of consumers. Multifactor models, involving several aspects of a cropping system, are required to uncover the important mechanisms behind variable herbivore responses to vegetational diversity. V TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS v LIST OF TABLES vii LIST OF FIGURES viii ACKNOWLEDGEMENTS ix CHAPTER 1. INTRODUCTION A. Resource Concentration and Natural Enemies 1 B. Plant Quality 5 GENERAL METHODS AND EXPERIMENTAL APPROACH The Cropping System i) The Principle Crop 8 ii) The Intercrop 9 Study Site and Field Layout 10 Experimental Design 10 Predator Manipulations i) Choice of a Predator 12 ii) Life History of Enoplaanatha ovata 14 iii) Augmentation Methods 14 Background Experiments on E ovato i) Experiment 1: Diurnal Activity Pattern 17 ii) Experiment 2: Longevity of E ovata in Augmentation Plots 20 iii) Experiment 3: Comparison of E ovata Densities in Augmentation and Control Plots 25 The Herbivores i) Choice of Herbivores 28 ii) Sampling Method 28 iii) Life Histories of Herbivores : 29 Summary 33 CHAPTER 2. PLANT QUALITY, A CONFOUNDING FACTOR IN DETERMINING THE DISTRIBUTION OF FOUR LEPIDOPTERANS IN MONOCULTURES AN DICULTURES INTRODUCTION Plant Quality 34 Predation 35 MATERIALS AND METHODS Measurements of Brussels Sprouts Plant Quality 37 i) Plant height, width and leaf area 38 ii) Growth rates 38 Herbivore Densities 39 i)Data Analysis 40 RESULTS Measurements of Brussels Sprouts Plant Quality i) Plant height 42 ii) Plant width 42 iii) Leaf Area and Leaf Loss 42 iv) Growth rates 47 Herbivore Populations 1. Plutella xylostella 47 i) July 51 ii) August 51 v i 2. Autographa californica 56 3. Mamestra configurata 60 4. Pieris rapae 66 DISCUSSION Host Plant Quality and Herbivore Density 68 Confounding Effects of Predation 70 Herbivore Density Differences Attributable To The Diversity of A Planting 71 Causes of Larval Distribution 72 Herbivore Feeding Rates 73 Summary 74 CHAPTER 3. PARASITISM IN RELATION TO PLANTING TYPE, SPIDER DENSITY AND PLANT QUALITY INTRODUCTION Plant Quality 75 Predation 76 MATERIALS AND METHODS The Parasitoids 78 Determination of Percent Parasitism 80 Plant Quality, Herbivore Density and Experimental Design 80 Data Analysis 81 RESULTS Parasitism of Plutella xylostella by Diadeama insulare i) Parasitism and host life stage 82 ii) Plant Quality 82 iii) Percent parasitism and treatments 85 iv) Period of Sampling 95 Parasitism of Autographa californica with Voria ruralis 100 i) Parasitism and host life stage 105 ii) Plant quality 105 iii) Percent parasitism and treatments 105 iv) Percent parasitism and host density 107 DISCUSSION 110 Plant Quality and Experimental Treatments 110 Predation by Enoplagnatha ovata 112 Host Density and Parasitism 114 Assessing Percent Parasitism 116 Comparing the Two Herbivores 116 Summary 117 CHAPTER 4. YIELD INTRODUCTION 119 METHODS 120 RESULTS 121 DISCUSSION 128 CHAPTER 5. GENERAL DISCUSSION 129 REFERENCES 135 v i i LIST OF TABLES Number Title page Table 1.1 Species of Foliar Dwelling Spiders Collected in 1985 13 Table 1.2 Schedule of E. ovata Augmentations 16 Table 2.1 Mean Plant Height (cm) Per Plot 43 Table 2.2 Sample Three-Way ANOVA Summary 44 Table 2.3 Mean Plant Width (cm) Pec Plot 45 Table 2.4 Mean Plant Leaf Area (cm^) Per Plot 46 Table 2.5a Total Leaf Area Eaten 48 Table 2.5b Percent of Available Leaf Area Eaten 48 Table 2.6 Growth Rates of Brussels Sprouts Plants 49 Table 2.7 Analysis of Covariance Summary 52 Table 2.8 Density of xylostella larvae and pupae 53 Table 2.9 Sample Two-Way ANOVA Summary for Herbivore Densities 54 Table 2.10 Comparison of Results With and Without Plant Quality Parameters 55 Table 2.11 Densities of A,, californica Instars 59 Table 2.12 Densities of JyL configurata 63 Table 2.13 Densities of P. rapae 67 Table 3.1 a Comparison of the Proportion of Parasitised P. xylostella Collected as Larvae and Pupae 83 Table 3.1 b Percent of Each Life Stage Collected and Parasitised 83 Table 3.2 Analysis of Covariance Summary 92 Table 3.3 Analysis of Covariance Summary 93 Table 3.4 Percent Parasitism of FL xylostella by D.insulare 94 Table 3.5 Summary of Percent Parasitism Differences Between Spider Augmentation Plots and Control Plots, Without Spider Augmentations 97 Table 3.6 Analysis of Variance Summary 98 Table 3.7a Percent Parasitism of A^ californica by V. ruralis 106 Table 3.7b Percent Parasitism of A^ californica Per Plot, Separated by Experimental Treatments 106 Table 4.1 Results of One-Way ANOVAS Comparing Yield on the October and November Harvest Dates 122 Table 4.2 Two-Way ANOVA Summary on Yield Data for October Harvest 123 v i i i LIST OF FIGURES Number Title page Fig. 1.1 Layout of Field Plots 11 Fig. 1.2 Diurnal Activity Pattern for E ovata 18 Fig. 1.3 E ovata Densities and Temperature 19 Fig. 1.4 Longevity of E ovata in augmentation plots 23 Fig. 1.5 Predator/Herbivore Ratio, 1984 24 Fig. 1.6 Difference Between of E ovata Densities in Augmentation and Control Plots 1986 27 Fig. 2.1 Frequency of Occurrence of Each Life Stage of FL xylostella 50 Fig. 2.2 E . xylostella Density and Plant Width 57 Fig. 2.3 Freuqency Distribution of A. californica Life Stages 58 Fig. 2.4 Fifth Instar A^ californica and Plant Width 61 Fig. 2.5 Frequency of Occurrence of Each M. configurata Larval Instar 62 Fig. 2.6 configurata density and plant quality 65 Fig. 3.1 July - Plant Quality and Parasitism of F l xylostella by D.insulare 84 Fig. 3.2 July - Leaf Area and Parasitism of JL xylostella with Parasitism >5% 86 Fig. 3.3 August - Plant Quality and Parasitism of P. xylostella by D.insulare 87 Fig. 3.4 August - Plant Quality and Parasitism of P. xylostella by D.insulare 88 Fig. 3.5 July - Plant Quality and Density of FL xylostella 89 Fig. 3.6 August - Plant Quality and Density of P. xylostella 90 Fig. 3.7 August - Plant Quality and Density of P. xylostella 91 Fig. 3.8 Parasitism of FL xylostella in July and August 96 Fig. 3.9 Parasitism of FL xylostella in August as a Function of Parasitism in July 99 Fig. 3.10 Density of P. xylostella and Parasitism by D.insulare 101 Fig. 3.11 Parasitism, Host Density and Planting Type 102 Fig. 3.12 Parasitism, Host Density and Spider Density 103 Fig. 3.13 /V californica Density and Parasitism by V. ruralis 108 Fig. 3.14 Parasitism, Host Density and Planting Type-A. californica 109 Fig. 4.1 Yield in Fresh Weight of Sprouts 124 Fig. 4.2 Yield in Number of Sprouts/Plant 126 i x ACKNOWLEDGEMENTS I would like to thank the people who provided support throughout this study. Dr. William Wellington provided me with the philosophical perspective to keep me on track and financial support when scholarships lapsed. Dr. Bryan Frazer provided invaluable advice throughout the thesis. I am indebted to him, as co-supervisor, for the patience with which he dealt with my questions and challenges, the time he generously gave encouraging me when I had doubts, and the skill with which he is able to teach. Dr. Judith Myers kindly agreed to take me on as her student when Dr. Wellington retired. Her thoughtful insights helped me gain a new perspective on my data when I was becoming weary. I am also indebted to Ms. Laura Richardson and Ms. Louise Waterhouse for their excellent assistance during my field seasons. Dr. Tom Mommsen's continued moral support throughout my tenure as a graduate student and infectious joy of science gave me the enthusiasm to continue. The University of British Columbia provided me with the Leonard Klinck Fellowship and a University Graduate Fellowship, both of which allowed me to complete this work. The Department of Plant Science provided the field station where the research was conducted and Agriculture Canada offered their greenhouses for starting my plants. All insect and spider identifications were done by the Biosystematics Research Institute, Agriculture Canada. Mr. J. Redner identified the spiders, Dr. B. E. Cooper the Diptera, Dr. H. E. Bisdee the Hymenoptera and Dr. E. W. Rockburne the Lepidoptera. 1 CHAPTER 1 INTRODUCTION A . RESOURCE CONCENTRATION AND NATURAL ENEMIES Applied ecologists generally believe that pest management problems are aggravated in monocultures (Cromartie 1981). Diseases and insect pests of annual and perennial crops (Cromartie 1981) and of forests (Crook et al. 1979, Kemp and Simmons 1979) seem to thrive where biological diversity is low. Agricultural monocultures and associated problems of increased herbivore populations as a result of low vegetational diversity, have probably been studied more than other monocultures. A vast theoretical and experimental literature specifically addresses the questions of how and why herbivore populations differ in monoculture and polyculture cropping systems. The subject has been thoroughly reviewed in recent years (Altieri and Letourneau 1982, Risch et al. 1983, Kareiva 1983, Altieri and Liebman 1986, Sheehan 1986, Russell 1989, Vandermeer 1989). It is generally accepted that herbivore densities tend to be reduced in polycultures. However, experimental results are highly variable, with many examples supporting and contradicting the folklore (Risch et al. 1983). Results vary with the cropping system, the herbivore species, the geographic location and between researchers (Altieri and Liebman 1986, Vandermeer 1989). For example, two different groups of workers found lower rates of attack of the fall armyworm, Spodoptera frugiperda (J.E. Smith), in a maize-bean intercrop as compared to a maize monoculture (Francis et al. 1978, Van Huis 1981). However Vandermeer (1989) was unable to detect 2 such a pattern in the same geographic location, on the same crop system and with the same pest. Several workers have hypothesized mechanisms to explain the perceived pattern of reduced herbivore loads in polycultures (c.f. Vandermeer 1989). A seminal paper by Root (1973) has stimulated not only a vast amount of experimental work meant to clarify the role of vegetational diversity in pest management, but also a lively, and controversial discussion. Root (1973) proposed two alternative mechanisms to explain the differences in insect abundance in monocultures and polycultures. The resource concentration hypothesis argues that the densities of specialist herbivores on host plants grown in monoculture are higher because the plants are easier to find, or more apparent in the terminology of Feeny (1976). Also, specialized herbivores, who can meet all their life requirements on a single plant species, tend to have longer tenure times in monocultures, resulting in higher fecundity. The resource concentration hypothesis stresses that a concentration of one food source can result in the accumulation of a large biomass of specialized herbivores, but not of species requiring a more varied diet, to complete their life cycle. The enemy hypothesis proposes that herbivores are less abundant in vegetationally diverse ecosystems because generalist predators and parasitoids are most effective where plant diversity is higher. Several mechanisms were proposed to explain natural enemy efficiency in diverse settings: 1. the availability of a wide variety of prey species, present at different times throughout a season, coupled with alternate food sources such as pollen and nectar, result in an environment where food is always abundant for generalist predators and parasites; 2. the large number of prey refuges found in diverse settings prevent natural enemies from driving their prey to extinction, resulting in stable predator-prey dynamics; 3. natural enemies switching to consume the most abundant prey prevents any particular herbivore 3 species from reaching high densities. The flurry of work which followed Root's paper has generally, although not always, provided evidence for the importance of resource concentration over natural enemies. In particular, herbivore movement has been identified as the most important determinant of high herbivore densities in monocultures (particularly the work of Bach 1980a,b; Risch 1980,1981; Kareiva 1982a,b, 1983; Risch et al. 1983; Elmstrom et al. 1988). However, the original controversy, encompassing the role of natural enemies, has not been resolved. One problem has been that almost all experiments have manipulated one single parameter, the resource concentration, and made assumptions about the other parameter, natural enemies, from correlations with unmanipulated enemy abundance and herbivore densities (Bach 1988b). Results of this work have shown that generalist predators usually occur at higher densities in polycultures (Pimentel 1961, Root 1973, Tukahirwa and Coaker 1982, but see Schultz 1988). Of only two experiments in which natural enemies were manipulated, Letourneau and Altieri (1983) used cages to include and exclude the predacious minute pirate bug, Orius tristicolor (White), and Tukarhiwa and Coaker (1982) erected barriers to exclude ground beetles. In one case the predator was important in determining reduced herbivore densities in polycultures (Letourneau and Altieri 1982) and in the other case the predatory beetles did not seem to be important (Tukarhiwa and Coaker 1982). The use of cages and barriers is confounded by microclimate effects (in cages) and the inhibition of natural immigration and emigration of predators (both cages and barriers), two problems which can bias results. Most of the research supporting the resource concentration hypothesis has been done on highly mobile Chrysomelid beetles, which have few natural enemies (Root 1973; Bach 1980a,b, 1981, 1984, 1986; Kareiva 1982b; Tukahirwa and Coaker 1982; Risch et al 1983; Elmstrom et al. 1988). 4 Root (1973) points out that adult flea beetles, Phvllotreta cruciferae Goeze, the chrysomelid beetles which constituted the bulk of herbivore biomass in his study, are rarely attacked by generalist predators and have only one specialized parasitoid. A test of the relative importance of resource concentration and natural enemies should include an examination of herbivore species more subject to predation than Chrysomelids. Finally, sampling methods often capture few of the more highly elusive generalist predators, such as wasps, ants and wandering spiders. Mortality caused by these generalist predators is the cornerstone of the enemy hypothesis. The diet breadth of both the herbivores and the natural enemies is central to Root's ideas (1973) about the importance of resource concentration. Recent laboratory findings suggest that specialized herbivores, who Root (1973) hypothesized would flourish in monocultures because of the concentration of their food plant, are less susceptible to predation by generalist predators, such as wasps and ants, than are herbivores with more polyphagous diets (Bernays 1988, Bernays and Cornelius 1989). This finding suggests that it may be difficult to separate the importance of a concentrated resource, per se, as a determinant of specialist herbivore density, from the lack of predator induced mortality on specialist herbivores simply because predators do not eat them. A fair comparison of the resource concentration and natural enemy hypotheses would have to not only consider herbivores who are susceptible to predation, but also several herbivores in the same system, some with wide diet-breadths and some with narrow diet-breadths. This thesis, in part, focusses on the effects of a generalist natural enemy on Lepidopteran populations in monocultures and dicultures. The monocultures were brussels sprouts, Brassica oleraceae L. (Cruciferae), or peppermint, Mentha piperita L . (Labiatae), and the dicultures consisted of brussels sprouts strip-cropped with peppermint. Populations of the most abundant invertebrate predator in the 5 system, Enoplagnatha ovata Clerck. (Araneida: Theridiidae), frequently found feeding on lepidopterous larvae, were manipulated in a factorial design experiment. Four lepidopterans were considered, two with polyphagous feeding habits, Mamestra configurata (Walk.) (Noctuidae) and Autoarapha californica (Speyer) (Noctuidae), and two crucifer specialists, Plutella xylostella (L.) (Plutellidae) and Pieris rapae L . (Pieridae). By the use of a two-factor experimental design, the relative importance of the concentrated resource (manipulated by planting types) and natural enemies (densities manipulated by augmentations) could be directly compared. The consideration of polyphagous and specialized herbivores, in the same experimental system, allowed me to compare interactions between the diet-breadth of herbivores and the importance of the resource versus natural enemies. B. PLANT QUALITY From the point of view of the consumer, vegetational diversity encompasses many aspects of a plant resource. Attempts have been made to isolate components like plant density (Bach 1980a,b, 1981,1984), patch size (Cromartie 1975) and microclimate differences (Risch et al. 1983, Altieri and Liebman 1986) to clarify important mechanisms (reviewed by Kareiva 1983). However, Kareiva (1983) emphasizes that plant quality, measured as lushness, size, moisture and nitrogen content or growth rates, can also vary with both plant density and patch size. Therefore, results of those experiments designed to separate components of vegetational diversity are often confounded by plant quality. Even subtle differences in plant quality can be detected by herbivores. For example, Bach (1981) demonstrated, in laboratory experiments, that Acalymma vittata (Fab.), the striped cucumber beetle, chose cucumber leaves taken from monocultures more 6 frequently than those taken from cucumber plants intercropped with tomatoes. A related beetle, A, innubum was able to distinguish between leaves from its host plant Cayaponia americana (Cucurbitaceae) grown in preferred 'open' patches and those grown in 'forest patches' (Bach 1984). An extensive literature exists on the relationship between herbivorous insects and 'plant quality'. All aspects of plant quality, including chemistry (Schultz 1988), nitrogen and water content (Scriber 1984) and growth form (Bach 1981) can alter herbivore feeding and oviposition behaviour. Inspite of the obvious possibility of differences in many aspects of plant quality, as a result of the planting diversity, few researchers interested in the behaviour of herbivores in monocultures and polycultures have measured plant quality. Neither have they accounted for plant quality in their experimental designs. The work of Bach (1980a,b, 1981) on the striped cucumber beetle has provided a notable exception. She clearly demonstrated that many aspects of plant quality, including leaf area, growth rate, vine length or growth forms are altered by diversity. Bach presented evidence that beetle abundance is correlated with some of these plant characteristics in monocultures, but not in polycultures. In another experiment, broccoli plants, Brassica oleracea L , were larger in monocultures than in dicultures with white clover, Trifolium repens L. (Leguminosae) (Elmstrom et al. 1988). In Root's (1973) experiments, in one of three years collards grown in pure stands were smaller than those grown adjacent to a meadow (diverse stands). In two other years the collards were larger in the pure stands. He attempted to control for differences in plant growth rate by using an index of \"herbivore load\", equal to the mean biomass of all herbivores per 100 grams of consummable plant material. Others have noted visible differences between plants grown in monocultures and polycultures, but have not accounted for these differences in their results (Altieri and Liebman 1986, 7 Letourneau 1987). Although the relationship between predators and plant quality is not well understood, some evidence exists that predators too can be influenced by plant quality (Perfecto et al. 1986). Consequently, a second aim of this thesis was to directly address the problem of plant quality in the system studied. By measuring several aspects of plant quality, and factoring them out in my analysis, I was able to determine the relative importance of plant quality as compared to vegetational diversity and natural enemies. In the remainder of this chapter I introduce the details of the cropping system, my general experimental approach, life history and background experiments on the manipulated predator, E ovata. and the life history of the four herbivores studied. In Chapter 2,1 discuss the interaction between the densities of these four lepidopterans, plant quality, resource concentration and natural enemies. In Chapter 3,1 present data on parasitoids, and how they are affected by plant quality, resource concentration and the generalist predator. Chapter 4 includes a brief presentation of the yield differences between monocultures and dicultures. Finally, general conclusions are presented in Chapter 5. 8 GENERAL METHODS AND EXPERIMENTAL APPROACH THE CROPPING SYSTEM i) The Principal Crop The principal crop in this study was brussels sprouts, Brassica oleracea v. aemmifera cv Long Island Improved 47, Catskill Strain (family: Cruciferae). A cruciferous crop was chosen because several previous studies comparing herbivore populations in monocultures and polycultures have already established a comparative framework for studies on cruciferous crops (Pimentel 1961, Dempster 1969, Smith 1969, Root and Tahvanainen 1969, Tahvanainen and Root 1972, Root 1973, Cromartie 1975, Kareiva 1982). Crucifers harbour a large number of well characterized lepidopterans, all of which have a natural enemy complex in the study area. The brussels sprouts were seeded, on April 30,1985 and on April 22,1986 in a greenhouse and transplanted into field plots, May 27 to June 1, 1985 and June 2 to 4, 1986. Spacing was 45 cm between rows and 60 cm within rows, resulting in 23 to 27 plants in a strip. Fertilizer treatments, made on the basis of recommendations from a soil analysis, consisted of two applications of urea, at the concentration of 45 kg N/ha, at the time of transplanting and again two weeks later. Boron, at the level of 3.5 kg/ha was applied two weeks after transplanting. 9 ii) The Intercrop Peppermint, Mentha piperita L , was used as the intercrop. The choice of an appropriate intercrop can be difficult. In the tropics, where intercropping is a common agricultural practise, crop combinations are usually based on historical precedent. In temperate climates, however, experimentation with new crop combinations is more common (Vandermeer 1989). Intercrops can be selected on the basis of direct yield enhancement potential, as in intercropping of grains with nitrogen fixing legumes, or on the basis of yield enhancement through reduced herbivory on the principal crop. Because my interest is in pest management, I chose an intercrop whose facilitative characteristics would be in deterring herbivores. Of several candidate crops for intercropping with crucifers, peppermint was particularly interesting because volatile chemicals emitted from it are purported to deter crucifer pests by masking host-finding olfactory cues (Riotte 1975). Peppermint also has been reported to improve the flavour of nearby cruciferous plants (Riotte 1975). A final reason for choosing peppermint was for its potential to attract parasitoids to adjacent brussels sprouts plants. It is well established that nectar producing plants attract parasitoids because the food source they provide enhances longevity and fecundity for parasitoids (Price et al. 1980). Peppermint flowers mid-July to August, early enough to attract parasitoids at a time when caterpillars are susceptible to parasitism. The peppermint was started from seedlings grown in a greenhouse in 1985 and planted in the field plots April 28 to May 2. Plants were spaced 30cm apart in 4 rows per strip. This resulted in 72 plants per strip, with 8 strips per plot in monocultures, and 4 peppermint strips per plot in dicultures. A crop was harvested by mowing in 1985. Peppermint is a fast-growing perennial crop and new growth appeared within a 10 week of mowing in 1985. Plants overwintered and peppermint strips were so dense by 1986 that individual plants were no longer distinguishable. In both years, the peppermint was top-dressed with urea at a level of 75 kg N/ha the second week in May. A second crop was harvested by mowing during the last week of July, 1986. All plants were irrigated once a week, from June to September, using a high pressure irrigation system. STUDY SITE AND FIELD LAYOUT The experiment was conducted in a 50m X 100m field at the Plant Science Field Station, University of British Columbia, Vancouver, Canada. Eighteen plots, 13.5m by 5.5m, were laid out with a 5.5m cultivated margin around each plot (Fig. 1.1). Each plot was further subdivided into 8 strips, 1m X 5.5m, with 0.7m wide, weed-free paths between strips. Margins between plots, and paths between strips were cultivated weekly to maintain their vegetation-free status. Strip-intercropping, rather than row-intercropping, was used to minimize below ground competition for water and nutrients and above ground competition for light. In the system used here, due to regular cultivation between strips, it was highly unlikely that roots of the two species came within 0.7m of each other. EXPERIMENTAL DESIGN A two factor experiment was conducted, with 2 replicates in 1985 and 3 replicates in 1986. Plots were randomly assigned a planting type: monoculture of brussels sprouts, monoculture of peppermint, or diculture of brussels sprouts/peppermint. Half of the plots assigned to each planting type were further randomly assigned one of 11 FIGURE 1.1: FIELD LAYOUT OF EXPERIMENTAL PLOTS 1985 1986 70 cm w Layout of the field plots and surrounding vegetation in 1985 and 1986. 5.5 m wide borders around the individual plots (5.5 m x 13.5 m) were cultivated in weekly intervals. Plots consisted of 8 strips (1 m x 5.5 m) planted either in diculture or one of two monocultures. The top inset illustrates a diculture plot with alternating strips of mint and brussels sprouts. The inset on the right delineates an individual strip of brussels sprouts. Bordering vegetation consisted of: N: 67% cultivated field, 33% potatoes; E: lawn, bordering on a two-lane road; S: bare field, cultivated regularly (1985); rye grass (1986); W: a thin strip of shrubs (mainly alder), bordering on a two-lane road. The strawberry patch used to study the diurnal pattern and temperature dependence of E. ovata was 50 m to the South of the 18 experimental plots. 12 three spider treatments in 1985, and one of two spider treatments in 1986. In 1985, populations of the theridiid spider Enoplaanatha ovata (Clerck) were either augmented or not manipulated (i.e. control plots). In a third treatment all foliar dwelling spiders were removed (i.e. removal plots). For reasons outlined below, removals were not done in 1986 and only two spider treatments, augmentations and controls, were used. PREDATOR MANIPULATIONS i) Choice of a Predator In 1985 all invertebrate predators found on brussels sprouts foliage were collected weekly from the removal plots and identified to species. Although several coccinellids were present, the only predator group collected which was known to, or observed to feed on lepidopterans was the order Araneae (spiders). As spiders are often abundant predators in agroecosystems (Nyffeler and Benz 1987), including rice (Kiritani et al. 1972), apples (Mansour et al. 1980, Dondale et al. 1979), peanuts (Agnew and Smith 1989), and cotton (Nyffeler et al. 1989), it was not surprising that they were abundant at the site of this study as well. The species list of all spiders found is presented in Table 1.1. Enoplagnatha ovata. the most numerous spider, was chosen as the species to be manipulated. Enoplaanatha ovata was observed feeding on lepidopterans on many occasions during the course of this study. It is not known how frequently the other spider species fed on lepidopterans. 13 TABLE 1.1: SPECIES OF FOLIAR DWELLING SPIDERS COLLECTED IN 1985 Species Family Enoplagnatha ovata (Clerck) Theridiidae Lepthyphantes tenuis (Blackwall) Linyphiidae Theridion bimaculatum (Linnaeus) Theridiidae Philodromus rufus pacificus (Banks) Philodromidae Xvsticus sp. Thomisidae Bathyphantes brevipes (Emerton) Linyphiidae Eriaone spp. Erigonidae Linyphantes aeronauticus (Petrunkevitch) Linyphiidae Theridion varians (Hahn) Theridiidae Eperigone sp. Erigonidae Zygiella x-notata (Clerck) Araneidae Pardosa sp. Lycosidae Tibellus sp. Philodromidae Clubiona lutescens Westring Clubionidae Tetragnatha sp. Tetragnathidae Species of spiders removed from the 'predator removal' plots are listed in descending order of abundance. Only those species for which two or more specimens were collected are listed. 14 ii) Life History of Enoplaanatha ovata Enoplaanatha ovata is a cosmopolitan theridiid spider. Its life history has been determined by Seligy (1971). It is a univoltine spider that overwinters in leaf litter as second instar spiderlings. Third instars disperse in early spring and move on to shrubs and herbs. The fifth, and last instar stage is the first web-building stage. Adult males are entirely cursorial while females are sedentary web-builders. Both sexes feed. At my field sites opposition occurred between mid-August and mid-September. Although juvenile E. ovata were first seen in the field plots as early as June 11,1985 none were observed after August 22, in either 1985 or 1986. The instars are easily distinguishable by total length (cephalothorax to opisthosoma) (Seligy 1971). Length was used to estimate the age distribution of spiders used in the spider augmentation plots. iii) Augmentation Methods All augmentations were made using individuals in the penultimate instar or adults, except on August 25 and September 2,1986. On those days second instar spiderlings, collected from a laboratory colony, were used. For introductions of penultimate and adult spiders, E. ovata were handpicked from the forest edge of the University of British Columbia Endowment Lands. Before being released, spiders were kept in vials with mesh tops for 12 hours, sprayed with water, and fed five lab-reared Drosophila melanogaster Meigen, as a method of preventing immediate dispersal. In 1985, due to problems in finding sufficient sources of E. ovata for augmentations, spiders were not all introduced on the same day. In 1986 E. ovata augmentations were made in all augmentation plots on the same days. In order to be 15 able to collect enough E. ovata to make introductions on the same day, the total number of spiders introduced per plot was reduced from 240 in 1985 to 70 in 1986. The augmentation schedules for both years are presented in Table 1.2. 16 TABLE 1.2: SCHEDULE OF ENOPLAGNATHA OVATA AUGMENTATIONS Date of Introduction Days from Planting Plot No. Introduced 1985: June 12 June 19 June 24 June 26 July 1 July 2 July 9 July 10 July 12 July 18 July 29 July 31 12 19 24 26 31 32 39 40 42 49 59 61 3 7 9 18 11 12 3 7 9 18 12 11 120 132 124 120 120 120 120 120 120 120 120 120 1986: June 6 2 July 4 30 July 15 41 August 1 59 August 25 83 September 1 90 all all all all all all 10 16 19 25 125 (2nd instars) 400 (2nd instars) In 1985, a total of 240 E ovata were introduced into individual augmentation plots on different days. The schedule of introductions is presented above. In 1986, equal numbers of E ovata spiders, to a total of 70 per plot, were introduced into all augmentation plots on the same day. 17 BACKGROUND EXPERIMENTS ON ENOPLAGNATHA OVATA i) EXPERIMENT 1: DIURNAL ACTIVITY PATTERN OF ENOPLAGNATHA OVATA Introduction: Background information on the diurnal activity patterns of E ovata had to be determined in order to select appropriate times for sampling. Methods: Enoplaanatha ovata was abundant in a 5-year old strawberry patch approximately 50 metres from my field. A 14.5 metre length was marked in the middle of two 0.7 metre wide strawberry beds. June 18 to 19, and again on June 25 to 26,1985, a census of E. ovata was done every two to three hours, between 4:00 a.m. and midnight, by counting all individual E. ovata found in a 20 minute period in the two marked beds. For analysis, sampling times were categorized as day, 6 a.m. to 7 p.m., and night, 7 p.m. to 4 a.m. Temperatures were recorded at the beginning of each sampling period. Student's t-tests, ANOVA and linear regression were used for the analyses, using the statistical package SPSS/PC+. Significance testing was done at the 0.05 probability level, except where otherwise noted. Results: Significant differences in activity patterns of E ovata were not a function of the time of day = -1-59, p>0.05; Figure 1.2). However, 28 percent of the variability in E. ovata numbers was accounted for by temperature (r 2 = 0.28; Figure 1.3). Discussion: Enoplagnatha ovata did not exhibit a distinct diurnal activity pattern. Therefore, it was reasonable to commence a census of E ovata during daylight hours and complete it after dark~a practise which was followed because of the long time involved in completing a census of E ovata in all plots. 18 FIGURE 1.2: DIURNAL ACTIVITY PATTERN FOR ENOPLAGNATHA OVATA Enoplaanatha ovata did not display a diurnal acvtivity pattern (tr-m = -1.59, p>0.05; with temperature as a covariate F-j j 3 = 1.07, p > 0.05) 19 FIGURE 1.3: E. OVATA DENSITIES AND TEMPERATURE E ovata densities decreased as temperature increased (JT = 0.28, p<0.05 that slope of the regression=0). Data were collected in a five-year old strawberry plot, 50 metres from my experimental plots. 20 Temperature was an important factor in determining the number of E ovata observed during a census. Therefore, when a census was taken in the experimental plots, as described below, care had to be taken to control for some of the variability due to fluctuating temperatures throughout the day. One strip per plot was counted in all plots before a second strip in each plot was counted. With this method, densities in an entire plot were not taken in the morning, when temperatures were lower, and in another plot in the afternoon, when temperatures were higher. This experiment also revealed that sampling E. ovata densities would be difficult. Repeated census in the same strawberry bed on the same day resulted in 72 percent unaccounted-for variability in E. ovata counts (Figure 1.3). From personal observations I suspected that the variability in E. ovata densities was due to both the adeptness with which this spider can hide under litter and in the top centimetres of the soil surface, and E. ovata's mobile nature. During future census in my fields I was careful to search for E. ovata under plants, on both sides of leaves and immediately under the soil surface. ii) EXPERIMENT 2: LONGEVITY OF ENOPLAGNATHA OVATA IN AUGMENTATION PLOTS Introduction: The success of the predator manipulations was dependent on E ovata densities being maintained at higher levels in augmentation plots than control plots, and lower levels in removal plots than control plots. Because I suspected that it would be difficult to keep E ovata densities at high levels in augmentation plots from a single introduction, I made repeated introductions throughout the season. In 1985 I conducted an experiment to determine how long E ovata densities could be maintained 21 at higher levels in augmentation plots than control plots. As well, I conducted a census to ensure that the number of E ovata per herbivore was greater in augmentation plots and lower in removal plots than in control plots. Methods: After augmentations, a census of the number of E ovata was taken every two to seven days for 21 days. In both the brussels sprouts and peppermint all plants in four strips per plot (i.e. 100 brussels sprouts plants per plot and 288 peppermint plants per plot) were carefully examined for the presence of E. ovata. As well, the soil under plants was searched. After 21 days, the census of E. ovata was done on a less regular basis. Considerable patchiness in plant quality was observed throughout the field. To reduce some of the variability in E ovata densities, which might be caused by plant quality, augmentation plots were paired with the nearest control plots for analysis. Paired Student's t-tests were calculated on E ovata densities using arcsin transformed data. Statistical significance was tested at the 0.05 probability level. Comparisons were made between E ovata densities in the strawberry patch, used in experiment 1, and E ovata densities in my augmentation plots in order to assess whether or not the level of augmentation I used resulted in biologically realistic densities. On a regular schedule of seven to fourteen days from planting the brussels sprouts, commencing at day 33, 20 sample plants per plot were non-destructively sampled for the presence of lepidopteran larvae and E ovata. All larval species were pooled to calculate a predator/herbivore ratio. The species of larvae found are listed below under 'Choice of Herbivores'. The E oyata/herbivore ratio (E.o/h) was calculated as: E oyata/herbivore = Total number of E. ovata per plant Total number of lepidopterans per plant 22 A oneway ANOVA was calculated for each day, to determine differences in E.o/h among augmentation, control and removal plots. Statistical significance was tested at the 0.05 or 0.01 probability levels. Results: Enoplaanatha ovata densities were significantly higher in augmentation plots than control plots up to 29 days from introductions (monoculture, brussels sprouts: t[9] =2.907, p=0.02; monoculture, peppermint: trg] =3.214, p=0.01; diculture, brussels sprouts strips: t[irj] =2.33, p=0.04; peppermint strips: t[-jo] =3.147, p=0.01; Figure 1.4). No significant differences in E. ovata density were found between removal and control plots. In the strawberry plots sampled in experiment 1, densities ranged from 0.19 to 1.39 E ovata per metre of bed. Approximately five brussels sprouts plants fit into the same area as one metre of strawberry bed. Therefore, E. ovata densities in the strawberries were divided by five in order to make densities in the strawberry plot equivalent to the per plant densities in my experimental plots. By this method densities in the strawberry plots were equivalent to 0.038 to 0.278 spiders per brussels sprouts plant. In experiment 2, E ovata densities in augmentation plots ranged from 0 to 0.17 spiders per brussels sprouts plant. In the brussels sprouts strips the number of E.o/h was significantly greater in the augmentation plots than either the control plots or the removal plots on Days 45, 52 and 60, but not on Days 33 and 74 (Figure 1.5a). In the peppermint strips, the number of E.o/h was significantly greater in the augmentation plots on Days 33, 60 and 74, but not on Days 45 and 52 (Figure 1.5b). Discussion: Because higher densities of E ovata were maintained in augmentation plots for 29 days, and the E.o/h ratio was higher in augmentation plots on most days, FIGURE 1.4: LONGEVITY OF E. OVATA IN AUGMENTATION PLOTS MONOCULTURE. BRUSSELS SPROUTS MONOCULTURE. PEPPERMINT -0.01 0 ADDITION \u00E2\u0080\u00A2 C O N T R O L 1L 14. U II. 11 \u00C2\u00AB \u00C2\u00BB1 <1 DAYS F R O M A U G M E N T A T I O N i OM 0.13 -0.06 -0.01 I. 5. 6. 7. 11 13. 2a 2Z D A Y S F R O M A U G M E N T A T I O N DICULTURE. BRUSSELS SPROUTS DICULTURE. PEPPERMINT 0 ADDITION \u00E2\u0080\u00A2 CONTROL DAYS F R O M A U G M E N T A T I O N w -0.01 11. II. KX 74. D A Y S F R O M A U G M E N T A T I O N Predator augmentation and control plots were paired according to planting type and proximity in the field. Significantly higher E. ovata densities were maintained in auamentation plots up to 29 days from introductions (monoculture, brussels sprouts: trqi = 2 907 p=0.02; monoculture, peppermint: trgi =3.214, p=0.01; diculture, brussels {jprouts strips: tr 1 0] =2.33, p=0.04; peppermint strips: t[1 0] =3.147, p=0.01). F I G U R E 1.5: P R E D A T O R / H E R B I V O R E RATIO, 1985 A. BRUSSELS SPROUTS 33. 45. 52 60. 74. DAYS FROM PLANTING 0 ADDITION \u00E2\u0080\u00A2 CONTROL \u00E2\u0080\u00A2 REMOVAL B. PEPPERMINT 33. 45. 52. 60. 74. DAYS FROM PLANTING 0 ADDITION \u00E2\u0080\u00A2 CONTROL \u00E2\u0080\u00A2 REMOVAL The predator/herbivore ratio was calculated as: E. ovata per plot divided by pooled counts of lepidopteran larvae per plot. Significant differences between treatments were determined by oneway ANOVA on each day. A probability level of 0.05 is indicated by * and a probability level of 0.01 is indicated by **. 25 the predator augmentation treatment was considered successful. However, neither densities of E ovata nor the E.o/h ratio were less in removal plots than control plots. Therefore, a decision was made not to pursue removals in 1986. Densities of E ovata in augmentation plots were in the same order of magnitude as densities in the nearby strawberry plot used for experiment 1, indicating that densities in the augmentation plots were biologically realistic. iii) EXPERIMENT 3: COMPARISON OF ENOPLAGNATHA OVATA DENSITIES IN AUGMENTATION AND CONTROL PLOTS, 1986 Introduction: A comparison between E ovata densities in augmentation and control plots was made in 1986, to confirm that the population of E ovata was maintained at higher levels in augmentation plots. Methods: In 1986, a census of E ovata was taken every ten to fourteen days in two Brussels sprouts strips per plot. Counts were made on brussels sprouts plants only, using the same searching methods as in 1985. Also, as in 1985, augmentation and control plots were paired according to proximity in the field and a paired Student's t-test was used to analyse the data. All analyses used arcsin transformed data. A comparison was made between E ovata densities, in augmentation plots, in 1985 and 1986 in order to determine whether or not the reduced numbers introduced in 1986 affected overall plot densities. Results: No significant differences in E. ovata densities were found between monoculture and diculture plots (for augmentations t[34] =-1.267, p>0.05; for control plots t[34] =0.897, p>0.05). Therefore, data.for the two planting types were pooled. Enoplagnatha ovata densities were not significantly higher in augmentation plots at the 26 first sampling, which was done six days after the brussels sprouts had been transplanted from the greenhouse at the six leaf stage (trsj=-2, p > 0.05). However, on all other sampling dates E. ovata densities were significantly higher in augmentation plots than control plots (\[29] =2.472, p=0.02, Figure 1.6). Mean E. ovata densities were not significantly greater in augmentation plots in 1985 (0.049 \u00C2\u00B1 0.01 per plant) than in 1986 (0.041 \u00C2\u00B1 0.01 per plant). Discussion: As E. ovata densities were maintained at higher levels in augmentation plots, the predator manipulation treatment was considered successful in 1986. The lack of difference between augmentation plots and control plots on the first sampling date was likely because the brussels sprouts plants were small and did not yet harbour many herbivores as prey. As explained under the section 'Augmentation Methods', above, more than three times as many E. ovata were introduced in augmentation plots in 1985 than in 1986. From the density data presented in experiments 2 and 3, it appears that the reduction in total numbers introduced did not affect overall E. ovata densities in augmentation plots. A considerable amount of cannibalism was observed in 1985 immediately after augmentations were made. Cannibalism was not a common occurrence in 1986, indicating that augmentation densities were unnaturally high in 1985 and the spiders self-regulated their densities to biologically appropriate levels. FIGURE 1.6: DIFFERENCE BETWEEN ENOPLAGNATHA OVATA DENSITIES IN AUGMENTATION AND CONTROL PLOTS, 1986. 27 3 O Pi H O O <\u00C2\u00BB Z O I\u00E2\u0080\u0094i & z. o < H < PL, H o rii 40 50 60 70 80 DAYS FROM PLANTING 90 In 1986 E. ovata augmentation plots and control plots were paired according to proximity in the field. The difference between the number of E. ovata per plant in augmentation plots and control plots is presented on the y-axis. A number greater than 0 indicates that the augmentation plot in a pair had higher densities of E. ovata than the control plot. Over the season, higher densities in augmentation plots were maintained ( t [ 2 9 ] =2.475, p=0.02). 28 THE HERBIVORES i) Choice of Herbivores The number of herbivore species present in my field plots was very low. Five lepidopterans and two aphids were found in 1985. The lepidopterans were the diamondback moth. Plutella xylostella (L) (Plutellidae), the alfalfa looper, Autographa californica (Speyer) (Noctuidae), the bertha armyworm. Mamestra configurata (Walk.) (Noctuidae), the imported cabbage worm Pieris rapae L. (Pieridae) and the variegated cutworm, Peridroma saucia (Hbn.) (Noctuidae). The aphids were the cabbage aphid, Brevicoryne brassicae (L.) and the green peach aphid, Myzus persicae (Sulz.). The thesis focussed on the population dynamics of the four most abundant lepidopterans; FL xylostella. A, californica. M. configurata. and JL rapae. ii) Sampling Method Sampling for herbivores involved more plants, sampled less frequently, in 1986 than in 1985. In 1985, five randomly selected plants, in alternate strips, (i.e. 20 plants per plot) were sampled for the presence of lepidopterans every 10 to 14 days. In 1986 all brussels sprouts plants in two strips were sampled on July 22 to July 30, and again August 16 to 29. This method resulted in sampling a total 48 to 54 plants per plot. Both sides of all leaves on every plant, as well as the ground directly under plants, were examined for lepidopteran larvae. In 1985 larvae were identified to species and instar in the field and left on plants; in 1986 all larvae were collected, placed in covered petri dishes and maintained in a screenhouse. They were fed with fresh brussels sprouts leaves, taken from potted plants grown in a greenhouse and leaves were replaced on alternate days. Larvae were identified to species and instar and reared until they 29 pupated, died or parasites emerged from them. Thus, percent parasitism could be calculated in 1986, but not in 1985. Sampling was non-destructive in 1985 but destructive in 1986. Therefore, in the second sampling of 1986, a quarter of the population of herbivores per plot had already been removed a month earlier. As comparisons were subsequently made between control and treatment plots, all of which experienced the same destructive sampling, the sampling method in 1986 did not affect my results. iii) Life Histories of the Herbivores 1. Plutella xylostella The diamondback moth. Plutella xylostella (L.) (Plutellidae), was introduced from Europe, into North America mid 19th Century (Metcalf and Flint 1962, Beirne 1971). Because of its habit of migrating long distances on the prevailing winds, its distribution reaches as far north as Yellowknife, North West Territories (Beirne 1971). P. xylostella feeds on plants in the family Cruciferae, and can occasionally cause serious damage to cruciferous crops. Because the larvae are small-i.e. each larva of Pieris rapae L , one of the most common cruciferous pests, eats 7.5-times as much as a larva of R xylostella (Beirne 1971 )-the pest status of R xylostella is often underestimated. The moths breed continuously in southern North America, but overwinter as adults north of the 36th parallel of latitutde. Plutella xylostella does not overwinter in Canada. Each year immigrants from the south migrate to Canada to re-infest annual cruciferous crops. In Ontario, where most of the work on F\ xylostella has been conducted, first generation eggs are laid on cruciferous weeds because there are not yet any cruciferous crops in the field. Continuous migrations from the south, coupled with immigration into nearby 30 fields from offspring of previous immigrants, can result in large increases in the population throughout a summer (Harcourt 1957,1986). There are four to five generations a year in Ontario (Harcourt 1986), and two to three in British Columbia (Beirne 1971). Fecundity averages between 31 and 180 eggs per female, depending on the photoperiod (Harcourt 1986). First instar larvae hatch in 4 to 8 days, bore through the epidermis of the lower surface of leaves, and feed in mines. Second to fourth instar larvae are surface feeders on leaves. However, they also burrow into the sprouts of brussels sprouts. When disturbed, larvae make a characteristic wriggling movement, and often drop from leaves on a silken thread (Harcourt 1957, Beirne 1971). Pupation takes place on the host plant. Adults are active after dusk, resting on plants during the day, and feeding on nectar at dusk. Mating and oviposition occur after dark (Harcourt 1957). Harcourt (1957, 1960, 1986) has done extensive work on the population dynamics of P\ xylostella. Larvae are very susceptible to direct mortality from rainfall. Reduced fecundity throughout the season is also a major cause of reductions in populations in later generations. The most important biotic agent causing significant mortality is the ichneumonid parasitoid Diadegma insulare (Cress) (see Chapter 3). 2. Autoarapha californica Autographa californica (Speyerl (Noctuidae), the alfalfa looper, is a polyphagous lepidopteran, occasionally causing defoliation in alfalfa and vegetable crops in southern British Columbia (Beirne 1971). Its range is the western half of the United States, southern Saskatchewan and British Columbia (Milne and Milne 1980). The female lays 31 individual eggs on the underside of leaves, which hatch within approximately 7 days. There are 5 larval instars and pupation takes place on leaves or nearby structures. In California, the entire life cycle from egg to adult, takes 26 to 48 days (Puttarudriah 1953). Because the adults are seen in May and again from July to September, A californica most likely has two generations a year in British Columbia. It is not known whether this insect overwinters in British Columbia or is a regular immigrant (Beirne 1971). Although the preferred foodplants are alfalfa, clover and lettuce, A californica has been known to cause damage in a wide variety of fruit and vegetable crops, including both crucifer crops and mint. When alfalfa fields are cut down the larvae sometimes migrate to other crops (Beirne 1971). 3. Mamestra configurata Mamestra configurata (Walk.) (Noctuidae), the bertha armyworm, is native to North America. In Canada, it occurs from Manitoba west to British Columbia (Wylie and Bucher 1977). It is highly polyphagous, feeding on a wide variety of cultivated plants and weeds, but is economically most important as a pest of cruciferous crops, particularly canola in the prairies (Turnock 1984). In Canada, M, configurata completes one generation per year. In some years a second generation emerges in late August, but the offspring of these individuals do not survive the winter (Turnock and Philip 1977). Mamestra configurata overwinters in the pupal stage and adults emerge over a long period, from early June to early August. Eggs, laid in clusters, on the underside of leaves, take approximately one week to hatch. After 6 larval instars, M , configurata pupates in the soil around late August 32 (Turnock 1984, 1988). JN/L configurata hosts a large parasitoid complex, which is thought to keep its populations at low densities. Very little was known about this species until 1971, when an unprecedented outbreak occurred in canola on the Canadian prairies (Turnock 1984). Its populations seem to fluctuate, with outbreak levels reaching as high as 13.8 larvae per (Turnock 1988). 4. Pieris rapae Pieris rapae L (Pieridae), the cabbage white butterfly, was introduced into North America from Europe. It is a common pest of cruciferous crops. Because it is easily reared and readily sampled, P. rapae has been the subject of a wide range of studies, including studies on population dynamics (Pimentel 1961, Harcourt 1961,1966, Parker 1970, Parker et al. 1971), insect behaviour (Jones 1977a,b, Jones and Ives 1979), fecundity and genetics (Gilbert 1984), dispersal (Jones et al. 1980), and parasitism (Nealis 1983). Eggs are laid singly on the underside of leaves of crucifers. A complete generation requires five to six weeks at an average temperature of 17 C (Richards 1940). P. rapae overwinter as diapausing pupae. There are three to four generations a year in England (Richards 1940), three in Ontario (Harcourt 1963) and only two in Vancouver, British Columbia (Jones and Ives 1979). 33 SUMMARY The experiment described in this thesis was set-up in 1985 and repeated in 1986. Due to the patchiness of residues from a variety of fertilizers, soil sterilants and herbicides used in previous experiments, differences in plant quality from plot to plot rendered most of the quantitative results collected in 1985 unreliable. Hence, results presented in subsequent chapters, are from 1986 only. However, the 1985 season provided crucial background data used to improve the experiment in 1986. For example, although plant quality was not quantified in 1985, its importance in determining herbivore population densities was recognized. Therefore, in 1986 plant quality was quantified. The choice of predator species to be manipulated was made from 1985 surveys which identified the most abundant predator. Likewise, the species of herbivores followed in the study were chosen after 1985 surveys determining the most abundant herbivores at my study site. 34 CHAPTER 2 PLANT QUALITY, A CONFOUNDING FACTOR IN DETERMINING THE DISTRIBUTION OF FOUR LEPIDOPTERANS IN MONOCULTURES AND DICULTURES INTRODUCTION PLANT QUALITY Adult moths and butterflies can distinguish between individual plants of the same species by several characteristics, including height and width (Myers et al. 1981, Courtney 1981, Forsberg 1987), leaf area (Rausher 1983) nitrogen content (Myers 1985), crude protein (Myers et. al. 1981), water content (Wolfson 1980), growth rate (Price 1975) age (Jones 1977b) and plant chemistry (Renwick 1989). These \"plant quality\" parameters can be indicators of the suitability of plants for insect development (Southwood 1973, Onuf 1978, Vince et al. 1981, White 1978, Andow 1984). Not surprisingly, female lepidopterans often select oviposition sites on plants best suited for larval development. Some lepidopteran species choose plants on which larval development is rapid (Chew 1975, Jones and Ives 1979, Myers 1985, Forsberg 1987), while other species select plants on which developmental rate is slower (Courtney 1981). Through the mechanism of female choice for oviposition sites, plant quality can have a strong effect on lepidopteran distribution and densities. Many aspects of plant growth, and hence plant quality, are influenced by the species (Bach 1980b, 1988a) and density (Bach 1988a,b) of neighbouring plants. As a result, plants grown in monocultures and those grown in polycultures can be qualitatively different, as described in Chapter 1. The confounding effect of plant 35 quality makes it difficult to design experiments suitable for distinguishing the various effects of planting patterns on herbivore densities (Ralph 1977, Thompson and Price 1977, Solomon 1981, MacGarvin 1982, Kareiva 1983, Letourneau 1987). The first part of the study reported in this chapter was designed to test the hypothesis that lepidopterans do not respond to the diversity of plants in a plot alone, but rather to the differences in plant quality which might accompany differences in vegetational diversity. The effects of several plant quality parameters on the densities of the four Lepidopteran pests of crucifers, introduced in Chapter 1, were compared in the brussels sprouts monocultures and brussels sprouts/peppermint dicultures. f PREDATION Besides plant quality and planting type, a third parameter was investigated in this experiment-the importance of a natural enemy. Sometimes natural enemies can be more important determinants of herbivore distribution than plant quality. For example, avoidance of mortality from natural enemies determines the feeding preference of the larvae of a pyralid moth, Omphalocera munroei Martin. C\ munroei feed on older leaves of their host plant Asimina spp, even though their development rate is slower on old leaves than on young leaves. The reason is that the more rigid older leaves provide better protection from natural enemies when they are rolled (Damman 1987). In this experiment, as described in Chapter 1, the densities of a spider, Enoplaanatha ovata. were enhanced in some plots and not in others. I hypothesized that once the variability in lepidopteran densities due to plant quality was accounted for, and hence the importance of the concentrated resource on its own diminished, the impact of natural enemies would become more 36 apparent. Because predators can respond to vegetational diversity, just as herbivores (Price et al. 1980) I also hypothesized that the effect of augmenting E. ovata populations would differ in monocultures and dicultures. 37 MATERIALS AND METHODS MEASUREMENTS OF BRUSSELS SPROUTS PLANT QUALITY Several indicators of plant quality were measured on the brussels sprouts crop: plant height, width, leaf area and growth rate. A short explanation on why these particular parameters were chosen is in order. I had two main criteria in choosing plant quality measurements: 1.1 wanted to do as little destructive sampling as possible so that I could take plant measurements throughout the growing season without altering important components of the cropping environment, such as plant density and spacing; 2.1 wanted to choose bioassays which would represent a large range of plant characteristics of potential importance to both herbivores and their natural enemies. The first criterion ruled out measuring nitrogen levels. Although nitrogen content is, undoubtedly, an important attribute of plant quality from the perspective of herbivores (for review see Mattson 1980, Scriber 1984a,b), it is highly correlated with plant growth (Mattson 1980) and I was able to measure growth using non-destructive plant height measurements. Plant width was measured because it represents a plant characteristic affecting important elements of the crop ecosystem, such as openness of the canopy cover, light capturing ability of the plants and microclimate under the canopy (Harper 1977). It also correlates with leaf dry weight (Hunt 1982). Leaf area is used in several plant growth indices, and is known as a good indicator of photosynthetic potential (Causton and Venus 1981). In addition, leaf area describes the physical area small parasitoids must search when looking for hosts (see Chapter 3). 38 i) Plant height, width and leaf area Measurements of plant height and width were taken in two ways: i) non-destructive sampling of the same five, randomly chosen plants per strip, from a total of 4 alternate strips per plot (i.e. 20 plants per plot) and; ii) destructive sampling of 4 plants per plot on two occasions, corresponding to mid-season and late season. The non-destructive sampling was done for all 20 plants per plot on June 25 and July 12 to 15. Measurements of plant height were taken from the base of the plant, at soil level, to the axil of the newest leaves. Plant width was defined as the widest expanse of leaves present on the plant. Destructive sampling was done on July 18 and Sept. 1 to 3 . The most easterly plant, in the middle row, of alternate strips was collected (i.e. 4 plants per plot). Height and width measurements were taken as well as leaf areas. All leaves were removed and leaf area (cm2) was measured with a Licor area meter (Model 3100). Holes, presumed to be caused by herbivore feeding, were covered with paper and a second reading of leaf area was taken to estimate the amount of leaf area consumed. ii) Growth rates Early season growth rates (egr) were calculated by subtracting the differences in heights of the non-destructively sampled plants as follows: egr = (ht2 - ht1)/18 days where ht1 was the plant height on June 25 and h2 was the plant height July 12 to 15. Late season growth rates (Igr) were calculated using the destructive sampling data as follows: Igr = (ht4-ht3)/47 days 39 where ht3 was the plant height on July 18 and ht4 was the plant height on September 3 to 4. Because measurements were not taken on the same plant, data were ranked according to relative plant size, such that the height of the tallest plant in a plot on the first date was subtracted from the height of the tallest plant in the plot on the second date, and so on. iii) Data Analysis All data presented in this thesis were analyzed using the statistical package SPSS/PC + , and all significance testing was done at the 0.05 probability level, unless otherwise noted. nwo separate analyses were calculated for the plant quality data. Oneway ANOVAs, for each plant quality parameter, were calculated to determine whether or not plot means differed. Two-way ANOVAs, with planting type (i.e. monocultures and dicultures) and spider density (i.e. spider augmentations or no spider augmentations) were used to identify whether or not plant quality varied with the experimental treatments. HERBIVORE DENSITIES Herbivore densities from 1986 only are presented in this chapter. In 1985 plant quality was not quantified, hence the obvious variability in plant quality throughout the experimental field could not be accounted for. Details of the methodology for sampling herbivores is described in Chapter 1. Sampling was done on two occasions, first between July 22 and 29, and second between August 26 and 29, corresponding to mid-summer and late summer. Densities of each of the four lepidopteran species found in the Brussels sprouts crops, P, xylostella. A californica. M. configurata and JL rapae. were determined as follows: 40 number per plant = number collected in a plot number of plants sampled Densities were calculated for each of the twelve plots that contained Brussels sprouts plants and analyzed separately for each sampling time. i) Data Analysis The G-testfor goodness of fit (Sokal and Rohlf, 1981, pp. 692-703) was used to establish whether or not each instar, or each species, was found at the same frequency as all other instars. If the null hypothesis, stating that each instar was found with equal frequency, was supported, then the data for all instars were pooled; if the null hypothesis was rejected then the data for each instar were analyzed separately. The experimental design was a two-factor completely randomized design, with three replicates (cf. Chapter 1 for details of field layout). Planting type and spider density were the factors and mean herbivore density per plot was the dependent variable. Unpredictable and complicating effects of plant quality were accounted for by using analysis of covariance (ANCOVAR). In it, adjustments for each covariate were made before analyzing the main treatments. Appropriate transformations of the data were done when this was necessary to meet the assumption of parametric tests (Sokal and Rohlf 1981). In most studies of this sort, direct adjustments for plant quality are not made. I wanted to uncover potential errors in interpretation of the importance of planting types, which can result when plant quality is not included as a variant. Thus, analyses of variance, without plant quality adjustments, were also calculated and results of the two tests compared. 41 Direct relationships between plant quality and herbivore density were determined using linear regression and testing the null hypothesis that herbivore density did not increase or decrease as any of the plant quality parameters increased or decreased. Significance testing was done at the 0.05 probability level. 42 RESULTS MEASUREMENTS OF BRUSSELS SPROUTS PLANT QUALITY PARAMETERS i) Plant Height The Brussels sprouts plants were significantly taller in some plots than in other plots (June25: F i i > 2 2 8 = 2 - 6 . p<0.01; July 12-15: F-| 1 | 228= 2 - 0 4 . P<0.05; September 3-4: F-| 1 35 = = 1.18, p > 0.05, Table 2.1). However, these differences were not associated with planting type on any date except June 25. On June 25 the plants in dicultures were significantly taller than the plants in monocultures (p>0.05 for all main effects except on June 25;on June 25, planting type, F1 fQ=7.244, p<0.05; sample ANOVA summary, Table 2.2). ii) Plant width The Brussels sprouts plants were wider in some plots than in others (June 25, F11,228=9-39> P<0.001; July 12-15, F11 I 228= 4 - 2 3 . p<0.001; September 3-4, F11,36 = 1 - 9 6 ' P > 0-05, Table 2.3. As with plant height, these differences were independent of the planting type. iii) Leaf Area and Leaf Loss Leaf area was significantly larger in some plots than in others (July 18, F 11 ,36 = 3 - 3 2 > p< 0.01; September 3-4, F11,35=5.10, p< 0.0005, Table 2.4 for means). As with height and width, differences in leaf area were independent of planting type. 43 TABLE 2.1: MEAN PLANT HEIGHT (cm) PER PLOT Plot June 25 Type Plot Type July 12-15 Plot Type Sept.3-4 M 7.2 + 0.4 M 14.2 + 0.9 M 28.3 + 3.3 M 7.6 + 0.5 D 15.8 + 0.8 M 29.6 + 1.0 M 7.6 + 0.4 D 16.0 + 0.8 D 33.3 + 4.9 M 7.7 + 0.5 D 16.4 + 0.9 M 34.5 + 3.9 D 7.8 + 0.4 M 17.0 + 1.0 D 35.4 + 4.2 M 8.0 + 0.5 D 17.1 + 1.0 M 36.5 + 2.6 D 8.1 + 0.6 D 17.3 + 1.2 D 37.8 + 1.7 M 8.7 + 0.4 M 17.5 + 1.2 M 38.8 + 7.2 D 8.9 + 0.4 M 18.5 + 1.0 D 38.9 + 2.0 D 9.0 + 0.4 D 18.8 + 1.1 D 39.9 + 2.5 D 9.2 + 0.6 M 19.1 + 1.2 D 41.2 + 6.3 D 9.8 + 0.5 M 20.1 + 1.9 M 42.0 + 3.1 M = Monoculture, D = Diculture Means on June 25 (ht1) and July 12-15 (ht2) represent mean heights of the same 20 plants per plot, taken on two separate days. Means on September 3-4 represent mean heights of 4 destructively sampled plants per plot. The results of oneway ANOVAs testing for differences in mean plant height among plots are: June 25: F11 228= 2 - 6 . P< 0.01; July 12-15: F i i 228 = 2 - 0 4 > P<0.05; September 3-4: Fi 1 = = 1.18, p > 0.05. Solid vertical 'bars join means which are not significantly different at the 0.05 probability level, as determined by Student-Neuman-Keuls multiple range test. Where no bars are presented, there is no significant difference between any of the means. 44 TABLE 2.2: SAMPLE THREE-WAY ANOVA SUMMARY ANOVA SUMMARY FOR MEAN PLANT HEIGHT PER PLOT ON JUNE 25 SOURCE OF SUM OF DF MEAN VARIATION SQUARES SQUARE Main Effects Planting type (Pt) 2.784 1 2.784 7.244 <0.05 Spicier density (Sd) 0.011 1 0.011 0.874 >0.05 Interactions P t X S d 0.853 1 0.853 0.175 >0.05 Residual 3.075 8 0.384 Separate ANOVA tests, using the same model as summarized here, were calculated for all plant quality parameters, namely width, leaf area and growth rates, on all sampling dates. 45 TABLE 2.3: MEAN PLANT WIDTH (cm) PER PLOT PLOT JUNE 25 TYPE PLOT JULY 12-15 TYPE PLOT SEPT. 1-3 TYPE D' M M M D D M M M D D 31.9 +. 32.8 \u00C2\u00B1 35.9 \u00C2\u00B1 38.3 \u00C2\u00B1 39.0 \u00C2\u00B1 39.0 +. 41.1 \u00C2\u00B1 41.3 + 1.06 1.41 1.43 1.08 1.03 1.35 1.32 0.89 41.4+. 1.13 42.4 _+ 0.85 42.4 +. 1.66 43.5 + 1.33 D D M D M M D D M M M D 51.6 54.6 55.4 57.7 57.8 57.9 57.9 59.4 62.1 62.3 63.0 + 1.51 + 1.87 + 1.97 + 3.05 + 2.11 32 56 76 21 21 70 63.6 + 1.36 M D D M D M M D M M D D 45.9 47.3 48.0 50.8 51.7 54.0 54.2 55.2 56.3 56.4 59.6 60.6 + 3.12 + 3.63 + 2.21 + 4.49 \u00C2\u00B1 1 . \u00C2\u00B1 3 . + 3. \u00C2\u00B1 3 . + 3. \u00C2\u00B1 2 . \u00C2\u00B1 3 , + 2. 59 97 36 87 76 93 67 78 M = Monoculture, D = Diculture. Means on June 25 and July 12-15 represent mean widths of 20 non-destructively sampled plants per plot. Means on September 3-4 represent mean heights of 4 destructively sampled plants per plot. The results of oneway ANOVAs testing for differences in mean plant width among plots are: June 25 - F-| 1 228 = 9 - 3 9 -p<0.001; July 12 to 15- F n n 2 2 8 = 4 - 2 3 \u00C2\u00BB P< 0.001; Sept.3 to 4 - F i -j 36 = 1.96, p>0.05. Solid vertical bars join means which are not significantly different at the 0.05 probability level, as determined by Student-Neuman-Keuls multiple range test. Where no bars are presented, there is no significant difference between any of the means. 46 TABLE 2.4: MEAN PLANT LEAF AREA (cm2) PER PLOT PLOT TYPE JULY 18 PLOT TYPE SEPT.3-4 M D M D D D M M D D M M 3828 4443 4507 5039 6340 6547 6640 6669 7261 7273 7731 8690 + 365 \u00C2\u00B1 7 8 5 \u00C2\u00B1 7 1 1 \u00C2\u00B1 7 0 5 \u00C2\u00B11369 \u00C2\u00B11022 \u00C2\u00B11155 \u00C2\u00B1 3 5 1 \u00C2\u00B1 2 8 5 \u00C2\u00B11127 \u00C2\u00B1 1 4 1 + 654 D M D M D M M D M M D D 4017 4510 4852 4933 5366 5392 5468 5479 6098 6820 7529 9023 \u00C2\u00B1 5 9 3 \u00C2\u00B1 7 1 4 \u00C2\u00B1 6 3 9 \u00C2\u00B1 6 3 5 \u00C2\u00B1 4 5 2 \u00C2\u00B1 6 1 2 \u00C2\u00B1 3 6 9 \u00C2\u00B1 3 4 5 \u00C2\u00B1 5 9 2 \u00C2\u00B1 9 5 2 \u00C2\u00B1 5 2 9 + 753 M = Monoculture, D = Diculture. Number of plants sampled per plot and the results of oneway ANOVAs for each date are: July 18-n=4, F-|-^36=3.32, p<0.01; September 3 to 4 - n=4, F i 1,36=5.10, p< 0.0005). Measurements of leaf areas were obtained through destructive sampling. Solid vertical bars join means which are not significantly different at the 0.05 probability level, as determined by Student-Neuman-Keuls multiple range test. Solid vertical bars join means which are not significantly different at the 0.05 probability level, as determined by Student-Neuman-Keuls multiple range test. 47 Although leaf area did not differ between monocultures and dicultures, the observed damage to the leaves did. Leaves were significantly more damaged in diculture plots than in monoculture plots on both sampling dates (for July: F i ) 6 = 11.29; p < 0.025; for September: F ^ g = 6.36; p<0.05; Table 2.5a for means). Generally low levels of herbivory were found in this study, with percent of leaf area eaten ranging from means of 1.5 percent to 4.3 percent (Table 2.6b). iv) Growth Rates Early season growth rates also differed according to plot (egr, F-| -| ) 2 2 8 = 3 - 8 6 . p< 0.0001), while late season growth rates did not (Igr, F-| -| 35 = 1.02, p>0.05; Table 2.6). Both early season and late season growth rates of the brussels sprouts plants were independent of planting type. HERBIVORE POPULATIONS 1. Plutella xylostella At each sampling time, end of July and end of August, the frequency of occurrence of each life stage, of P\ xylostella. was not the same (log likelihood goodness of fit, testing the hypothesis that the frequency of all instars is equal: July, G[3]=35.4, p<0.001, Figure 2.1a; August, G[3j=27.18, p<0.001, Figure 2.1b). Consequently, data for each life stage were analyzed separately. 48 TABLE 2.5a: TOTAL LEAF AREA EATEN (cm 2 + s. e. m) July September M no additional predators 101 \u00C2\u00B1 33 114 \u00C2\u00B1 27 M predators augmented 126 \u00C2\u00B1 26 137 \u00C2\u00B1 22 D no additional predators 223 \u00C2\u00B1 5 7 216 \u00C2\u00B1 22 D predators augmented 186 \u00C2\u00B1 37 186 \u00C2\u00B1 25 TABLE 2.5b: LEAF AREA EATEN BY LEAF-CHEWING HERBIVORES (IN PERCENT OF AVAILABLE LEAF AREA) July September M no additional predators 1.53 \u00C2\u00B1 0.53 2.09 \u00C2\u00B1 0.48 M additional predators 2.57 \u00C2\u00B1 0 . 6 1 2.76 \u00C2\u00B1 0.56 D no additional predators 3.66 \u00C2\u00B1 0.85 4.28 \u00C2\u00B1 0.48 D additional predators 3.10 \u00C2\u00B1 0.34 2.75 \u00C2\u00B1 0.24 M = Monoculture; D = Diculture a. Total leaf area eaten was determined by the following calculation: leaf area with all feeding holes covered - leaf area with feeding holes exposed. Leaves were significantly more damaged in diculture plots than monoculture plots, July: F i >e = 11.29, p<0.025; September: Fi i6=6.36, p<0.05. b. The amount of leaf area eaten was low, ranging from a mean of 1.53 to 4.28 percent of the total leaf area. 49 TABLE 2.6: GROWTH RATES OF BRUSSELS SPROUTS PLANTS BY PLOT (cm of growth per day) PLOT TYPE EGR PLOT TYPE LGR M D D D D D M M M D M M 0.347 \u00C2\u00B1 0.399 0.404 0.412 0.444 0.513 0.520 0.541 0.543 _ 0.546 + 0.657 + 0.696 + 0.04 0.05 0.04 0.04 0.05 0.04 0.05 0.06 0.05 0.05 0.05 0.10 M D M M M D D D M D M D 0.307 +. 0.334 \u00C2\u00B1 0.359 +. 0.362 \u00C2\u00B1 0.362 \u00C2\u00B1 0.436 \u00C2\u00B1 0.445 +. 0.453 \u00C2\u00B1 0.480 \u00C2\u00B1 0.520 +. 0.552 +. 0.570 + 0.03 0.10 0.05 0.09 0.07 0.07 0.02 0.06 0.14 0.17 0.06 0.09 M = Monoculture; D = Diculture; EGR=early season growth rate, in cm per day, from June 25 to July 12-15; LGR = late season growth rate, in cm per day, from July 18 to September 3. Growth rates were calculated by subtracting heights at the two sampling dates and dividing by the number of days between samples. Egr was significantly different among plots (n=20, F- |- | ) 228 = 3 - 8 6. p< 0.001), Igr was not significantly different among plots (n=4, F= l l ) 36=1.02, p>0.05). Vertical bars represent means which are not significantly different at the 0.05 probability level, as calculated from Student-Neuman-Keuls multiple range test. 50 FIGURE 2.1: FREQUENCY OF OCCURRENCE OF EACH LIFE STAGE OF PLUTELLA XYLOSTELLA * JULY 200 AUGUST 200 | 1 1 r 150 -INSTAR FL xylostella pupae predominated in the July sample (G[3] = 35.4, p<0.001). Fewer FL xylostella first instar larvae than any other life stages were collected in August ( G i ^ i = 27.18, p<0.001). 51 i) July In July, the densities of all P. xylostella larval instars and pupae were independent of both planting type and spider density (Table 2.7 sample ANCOVAR Summary; Table 2.8 means). F\ xylostella density did not vary with any of the covariates, plant height July 12-15, plant width July 12-15, leaf area July 18, nor early season growth rate (Table 2.7 for sample ANCOVAR). However, some significant differences did occur between monoculture and diculture plots for second and third instar larvae (F-| 4=13.28 for interaction effect, Table 2.7). Within the monocultures more second to third instar larvae were counted in plots without added spiders. In contrast, within the diculture plots higher numbers of second and third instar larvae were present in plots where spider densities had been augmented (Table 2.8 for means). The interpretation of the importance of planting types on P.xylostella densities for July was not different when adjustments for plant quality were not made (Table 2.9 for sample ANOVA summary), because none of the plant quality covariates were statistically significant. As well, no linear relationships were found between any of the plant quality parameters and larval or pupal densities. ii) August In August significantly higher densities of fourth instar larvae were found in monoculture plots than diculture plots (F-| ,3=21.03). This difference really applied to plots with spider additions only (F-| ,3=26.06, for interaction of planting treatment by spider density). The higher densities of fourth instar larvae in monoculture plots were not detected when adjustments for plant quality were omitted (Table 2.10). 52 TABLE 2.7: ANALYSIS OF COVARIANCE SUMMARY Second and third instar P. xylostella larvae per Brussels sprout plant, July 22 to 29 sample by PT Planting Type (monoculture or diculture) SD Spider Density (augmentations or no augmentations) with HT Mean plant height/plot, July 12-15 (20 samples per plot) WD Mean plant width/plot, July 12-15 (20 samples per plot) LA Mean plant leaf area/plot, July 18 (4 samples per plot) GR Mean growth rate from June 25 to July 12 (egr) Source of Variation Sum of Squares DF Mean Square Covariates HT .005 1 .005 2.134 >0.05 WD .005 1 .005 2.197 >0.05 LA .003 1 .003 1.055 >0.05 GR .004 1 .004 1.775 >0.05 MAIN EFFECTS PT .002 1 .002 0.999 >0.05 SD .001 1 .001 0.213 >0.05 2-Way Interaction .033 1 .033 13.277 <0.05 Residual .010 .003 This is a sample ANCOVAR summary. Separate analyses for all instars of all four herbivores were calculated using the same model as summarized here. Significance testing was done at the 0.05 probability level. 53 TABLE 2.8: DENSITY OF F\ XYLOSTELLA LARVAE AND PUPAE (MEAN NUMBER PER PLANT\u00C2\u00B1 S.E.M.) A. JULY SAMPLING TRT H l 2-3 l 4 PUPAE MNP 0.07 \u00C2\u00B1 0 . 0 1 0.12 \u00C2\u00B1 0 . 0 5 0.10 \u00C2\u00B1 0 . 0 5 0.22 \u00C2\u00B1 0 . 0 8 MAP 0.02 \u00C2\u00B1 0 . 0 0 0.03 \u00C2\u00B1 0 . 0 1 0.06 \u00C2\u00B1 0 . 0 2 0.15 \u00C2\u00B1 0 . 0 6 DNP 0.08 \u00C2\u00B1 0 . 0 5 0.03 \u00C2\u00B1 0 . 0 1 0.16 \u00C2\u00B1 0 . 0 6 0.12 \u00C2\u00B1 0 . 0 1 DAP 0.07 + 0.01 0.15 + 0.03 0.07 + 0.06 0.17 + 0.05 B. AUGUST SAMPLING TRT H l 2-3 I4 PUPAE MNP 0.11 \u00C2\u00B1 0 . 0 1 0.14 \u00C2\u00B1 0 . 0 2 0.16 \u00C2\u00B1 0 . 0 1 0.20 \u00C2\u00B1 0 . 0 2 MAP 0.10 \u00C2\u00B1 0 . 0 4 0.27 \u00C2\u00B1 0 . 0 5 0.24 \u00C2\u00B1 0 . 0 2 0.17 \u00C2\u00B1 0 . 0 6 DNP 0.15 \u00C2\u00B1 0 . 0 6 0.23 \u00C2\u00B1 0 . 0 4 0.17 \u00C2\u00B1 0 . 0 5 0.23 \u00C2\u00B1 0 . 0 4 DAP 0.11 + 0.02 0.30 + 0.10 0.17 + 0.04 0.21 + 0.08 *TRT = Treatment MNP = Monoculture, no additional predators MAP = Monoculture, additional predators DNP = Diculture, no additional predators DAP = Diculture, additional predators The only significant differences, at the 0.05 probability level, between means presented here were: July sample, significantly more P, xylostella I2-3 in monoculture without added spiders and dicultures with added spiders (F-|3=21.03); August sample, significantly more P. xylostella I4 in monocultures wit'h added spiders (Fi ,3=26.06). 54 TABLE 2.9: SAMPLE TWO-WAY ANOVA SUMMARY FOR HERBIVORE DENSITIES ANOVA SUMMARY FOR SECOND AND THIRD INSTAR F\ XYLOSTELLA DENSITY BY PLANTING TYPE AND SPIDER DENSITY SOURCE OF VARIATION SUM OF SQUARES DF MEAN SQUARE F P Main Effects Planting Type Spicier Density 0.001 0.001 1 1 0.001 0.001 0.276 0.276 p>0.05 p>0.05 2-Way Interactions Planting Type by Spicier Density 0.035 1 0.035 14.420 p<0.05 Explained 0.037 3 0.012 4.991 p<0.05 Residual 0.020 8 0.002 Total 0.056 11 0.005 Separate ANOVA tests, using the same model as summarized here, were calculated for all larval instars and pupae of all four herbivore species. These two-way ANOVAs were calculated strickly for comparisons with the results of the ANCOVAR tests to determine whether or not misinterpretations of the importance of planting type can result when adjustments for plant quality are not made. Significance testing was done at the 0.05 probability level. 55 TABLE 2.10: COMPARISON OF RESULTS WITH AND WITHOUT PLANT QUALITY PARAMETERS WITH PLANT QUALITY WITHOUT PLANT QUALITY ADJUSTMENTS ADJUSTMENTS P. xylostella: -higher densities of fourth instars in monocultures with spider augmentations A. californica: -no differences due to planting types 3. M. configurata: -no difference in first instar densities related to planting type -no difference in first instar densities related spider treatments -more second instars in monoculture plots without additional spiders -no difference in third instar densities due to spider treatments -no differences in density related to planting types or spider treatments -fourth and fifth instar densities greater in monoculture plots -first instar densities greater in diculture plots -first instar densities greater in spider augmentation plots -no difference in second instar density related to planting types or spider treatments -more third instar larvae in plots without spider augmentations Differences in interpretation of the data with and without covariate adjustments for plant quality are summarized here. With adjustments for plant quality parameters, all significant relationships (p < 0.05) between herbivore densities and experimental treatments were a result of interaction effects; without adjustments for plant quality it appeared that planting treatment alone, as well as spider density alone, could account for the densities of some species and some instars. 56 The density of P . xylostella was independent of planting type and spider density for all other life stages sampled. For the August sampling period, the effects of two plant quality covariates were statistically significant. More fourth instar larvae were found in plots with shorter plants at the time of sampling (F-| 3 = 12.12, p<0.05) and smaller leaf widths at the July sampling (F-| 3=23.81, p < 0.05). Fourth instar density was a linear function of plant width (F-| -|o=5.02, p<0.05, r2=0.33,14 density = 0.74 - 0.0096 X plant width at the July sampling date, Figure 2.2). The relationship between density and plant height was not a simple linear function ( F i i 1 0 = 1.45,p>0.05). 2. Autographa californica As with P. xylostella. densities for each instar, of A californica. could not be pooled (G[4] =70.876, p<0.001). Second and third instar larvae predominated during the July sampling period (Figure 2.3). Data for August were not included for this species because only 13 individuals were collected on a total of 616 plants examined. The densities of first, second and third instar larvae did not vary with planting type or spider densities, regardless of whether or not the four plant quality parameters were included as covariates in the analysis (see Table 2.12 for means). However, the plant quality parameters did affect the results with fourth and fifth instar larvae (Table 2.10 for comparisons with other herbivores). The monoculture plots had higher numbers of fourth and fifth instar larvae than the diculture plots (fourth instars: F-i.,8 = 5.59, p<0.05; fifth instars: F i 8 = 8.00; p<0.05), before the covariates were included (Table 2.11). These differences were accounted for by the covariate early season growth rate (for fourth instars, growth rate, Fi>4=7.39, F I G U R E 2.2: P. X Y L O S T E L L A D E N S I T Y A N D P L A N T W I D T H \u00C2\u00A9 < I H 55 < PH w H S CO z E 0.30 0.25 0.20 h 0.15 h 0.10 h 0.05 50 55 60 65 MEAN PLANT WIDTH PER PLOT (CMSKJULY Fourth instar P. xylostella density in August decreased as plant width, at the July sample, increased (F-| -JQ = 5.02). FIGURE 2.3: FREQUENCY DISTRIBUTION OF AUTOGRAPHA CALIFORNICA LIFE STAGES 58 Second and third instar Autographa californica larvae predominated during the July sample (G[4j =70.88, p<0.001). 59 TABLE 2.11: DENSITIES OF AUTOGRAPHA CALIFORNICA INSTARS (MEAN NUMBER PER PLANT \u00C2\u00B1 S.E.M.) TRT H l 2 I 3 MNP 0.013 \u00C2\u00B1 0.007 0.09 \u00C2\u00B1 0 . 0 4 0.10 \u00C2\u00B1 0 . 0 4 MAP 0.007 \u00C2\u00B1 0 . 0 0 7 0.06 \u00C2\u00B1 0 . 0 0 0.09 \u00C2\u00B1 0 . 0 5 DNP 0.027 \u00C2\u00B1 0 . 0 1 3 0.07 \u00C2\u00B1 0 . 0 1 0.10 \u00C2\u00B1 0 . 0 4 DAP 0.013 \u00C2\u00B1 0 . 0 1 3 0.02 + 0.02 0.09 + 0.03 TRT l 4 l 5 MNP 0.04 \u00C2\u00B1 0 . 0 2 0.03 \u00C2\u00B1 0 . 0 1 MAP 0.08 \u00C2\u00B1 0.03 0 DNP 0.01 \u00C2\u00B1 0 . 0 1 0.01 \u00C2\u00B1 0 . 0 1 DAP 0.01 + 0.01 0 T R T = Treatment MNP = Monoculture, no additional predators MAP = Monoculture, additional predators DNP = Dicultures, no additional predators DAP = Dicultures, additional predators No significant differences in A californica density were found due to either planting type or spider density once plant quality was accounted for. 60 p=0.05; for fifth instars F-| 4=14.30, p <0.05). For fifth instars, leaf area was also significant as a covariate (F1 4=28.45, p<0.01). No fifth instar A, californica larvae were found in the spider addition plots, irrespective of planting type (for spider density treatment, F1,4=41.86, p<0.005). The density of fifth instar A. californica increased with an increase in plant width (F1 jo=4.92, p=0.05, r2=0.33, fifth instar density = -0.14 + 0.0025 x (x=plant width), Figure 2.4). However, as there were only 4 non-zero densities this relationship cannot be considered a strong one. No other significant linear relationships were found between any of the plant quality parameters and A,. californica density. 3 . Mamestra configurata Only one individual configurata larva was found in the August sample. Therefore, all analysis was restricted to the July sample, when a total of 577 larvae of M. configurata were collected. As the frequency of first and second larval instars was greater than that of any other instars (G[3] = 198.28, p<0.001, Figure 2.5) densities of each instar were analyzed separately. Fifth and sixth instar larvae were not found in the field. Higher densities of second instar larvae were found in monoculture plots without spider augmentations, and in diculture plots with spider augmentations (planting treatment, F-|,4=27.59, p<0.01; spider density, F-| ,4=40.26, p<0.01; interaction F-|4 = 15.76, p<0.05, Table 2.12 for mean densities). FIGURE 2.4: FIFTH INSTAR AUTOGRAPHA CALIFORNICA AND PLANT WIDTH Fifth instar >V californica density increased as plant width increased. The slope of the regression line shown is significantly greater than 0 at the 0.05 probability level. 62 FIGURE 2.5: FREQUENCY OF OCCURRENCE OF EACH MAMESTRA CONFIGURATA LARVAL INSTAR Significantly more first and second instar Mamestra configurata larvae were collected than third and fourth instars (G[3] = 198.28, p<0.001). 63 TABLE 2.12: DENSITIES OF MAMESTRA CONFIGURATA MEAN NUMBER PER PLANT + S.E.M. TRT \"2 \"3 MNP 0.22 \u00C2\u00B1 0 . 1 2 MAP 0.21 \u00C2\u00B1 0 . 1 3 DNP 0.40 \u00C2\u00B1 0 . 2 4 DAP 0.63 + 0.43 0.41 \u00C2\u00B1 0 . 1 7 0.07 \u00C2\u00B1 0.05 0.05 \u00C2\u00B1 0.03 0.97 + 0.32 0.21 \u00C2\u00B1 0 . 1 1 0.05 \u00C2\u00B1 0.03 0.24 \u00C2\u00B1 0 . 1 0 0.12 + 0.04 0.03 \u00C2\u00B1 0 . 0 1 0.03 \u00C2\u00B1 0.03 0.07 \u00C2\u00B1 0 . 0 1 0.07 + 0.03 T R T = Treatment MNP = Monoculture, no additional predators MAP = Monoculture, additional predators DNP = Dicultures, no additional predators DAP = Dicultures, additional predators Significantly more second instar M^ configurata were found in monoculture plots without spider augmentations and diculture plots with spider augmentations (planting treatment, F-|4=27.59, p<0.01; spider density, F-| 4=40.26, p<0.01; interaction F-|i4 = 15.76, p<0.05). Differences in densities of all other instars were not associated with my experimental treatments. 6 4 The densities of all other larval instars of hL configurata did not vary with either planting treatment or spider density (p > 0.05). Several errors in interpretation of these results would have occurred, if adjustments for plant quality parameters had not been made (Table 2.10 for comparison with other herbivores): i) first instar larval densities would have been considered greater in diculture plots (F 1 j 8=5.251,p=0.05); ii) first instars would have been interpreted as greater within dicultures in spider augmentation plots (F-| 8=7.091, p<0.05), when, in fact, the difference is actually not statistically significant with covariate adjustments; iii) the differences in second instar densities would have gone undetected (planting type Fi 8 = 1-328, p>0.05, spider density, F-| ,8=0.171, p>0.05, interaction F 1 ) 8=0.216, p>0.05); iv) in monocultures, third instar larval densities would have been considered greater in plots without spider augmentations (interaction of planting type by spider density Fi ,8 = 11.819, p<0.01), regardless of planting type (Table 2.12). Significant variability due to plant quality parameters (i.e. significant covariates) accounted for these differences between the two analyses (width July 12-15,12, F-|4=9.07, p<0.05,14, F i j 4 = 7.99, p<0.05; height July 12-15, I3, F i j 4 = 14.32, p<0.05). Mamestra configurata larval density did vary as a direct function of several plant quality parameters (Figure 2.6): i) more third instar larvae were found where growth rate was slower (13 density = 1.13-1.04 Xlog10 growth rate, F-|io=8.67, p<0.05, r2=0.46); ii) more third instar larvae were found where shorter plants prevailed 65 FIGURE 2.6: MAMESTRA CONFIGURATA DENSITY AND PLANT QUALITY LOCK10) EARLY SEASON GROWTH RATB LOG<10> PLANT HEIGHT (CM) H _ 2 I , , 1.70 1.75 1.80 1.85 LOG(IO) PLANT WIDTH (CM) Third instar M L configurata density decreased as early season growth rate increased and as plant height increased. Fourth instar jVL configurata density decreased as plant width increased. The slopes of the regression lines presented are all significantly different from 0 at the 0.05 probability level. 66 (13 density = 3.40 - 2.63 X Iog10 plant height, Fi j - |n = 14.99, p<0.01, = 0.60); iii) more fourth instar larvae were found where plants were narrower (14 density = 1.67-0.92 XloglO plant width, F 1 ) IQ=6.46 , p<0.05, r 2 = 0.39). 4 . Pieris rapae In the July sample a total of 20 P\ rapae larvae were collected on 616 plants sampled, while in August, this lepidopteran pest was more abundant. As a consequence, only August data were used in the analysis. The August data for all instars were pooled because too few individuals could be collected to allow for a separate analysis of age classes (for means, cf. Table 2.13). Larval densities were not significantly different in monoculture or diculture plots (Fi4=0.12, p>0.05) nor in plots with different spider treatments (F-| 4=0.15, p>0.05). As well, none of the plant quality parameters assessed affected P\ rapae density. The interpretation of results was the same, whether or not plant quality adjustments were made. Finally, linear relationships between P\ rapae density and the plant quality parameters could not be detected. 6 7 TABLE 2.13: DENSITIES OF PIERIS RAPAE (MEAN NUMBER PER PLANT \u00C2\u00B1 S.E.M.) TRT = Treatment MNP = Monoculture, no additional predators MAP = Monoculture, additional predators DNP = Dicultures, no additional predators DAP = Dicultures, additional predators TRT MNP MAP DNP DAP All Instars Pooled 0.01 \u00C2\u00B1 0 . 0 1 0 0.03 \u00C2\u00B1 0 . 0 1 8 0.03 \u00C2\u00B1 0 . 0 1 0 0.03 + 0.030 The densities of R rapae were not significantly different in monoculture versus diculture plots, nor in spider augmentation versus control plots. 6 8 DISCUSSION HOST PLANT QUALITY AND HERBIVORE DENSITY As hypothesized, host plant quality had a significant impact on the density of P\ xylostella. A, californica and M, configurata. In this study, unlike results in other studies (Root 1973, Altieri and Liebman 1986, Elmstrom et al. 1988, Bach 1980b, 1988a), plant quality did not vary with the planting type (i.e. monocultures and dicultures). However, all plant quality parameters did vary among plots, regardless of the planting type. This was possibly a function of the checkered history of the field where this experiment was conducted. Variable fertilizer, herbicide and soil sterilant treatments in previous experiments may have increased the heterogeneity in plant quality observed among plots. As this experiment was conducted at a university research station, typical of many plot experiments of this sort, unquantifiable variability in soil quality, and the resultant heterogeneity in plant quality, is probably not an unusual occurrence and might explain why experimental results vary widely between species, experiments and researchers. When plant quality was accounted for through covariate analysis, interpretations of the way in which herbivore densities vary with planting type changed fundamentally (changes summarized in Table 2.10). Plant quality did not always have the same directional effect. However, when adjustments for plant quality were made all single factor effects disappeared and only interaction effects between planting type and spider treatments were statistically significant. For instance, higher densities of fourth instar R xylostella were found in monoculture plots with spider additions than in diculture plots or monoculture plots without spider additions. This difference went undetected unless the required adjustments for 69 plant quality were included in the analysis. In contrast, fourth and fifth instar A californica densities did not differ between monoculture and diculture plots with adjustments for plant quality. Without plant quality adjustments, densities of A. californica would have incorrectly been interpreted as higher in monoculture plots than in diculture plots. By the same token, first instar M L configurata densities would have been incorrectly considered greater in dicultures than monocultures, if adjustments for plant quality had not been made. A fourth example lies with P. xylostella second to third instar larvae and M L configurata second instar larvae. For these two species and instars, higher densities were found in monoculture plots without spider augmentations, and in diculture plots where the spider abundance had been enhanced. This interaction effect, was uncovered only after adjustments for plant quality were made. The important concept, that variability in herbivore densities due to unqualified plant quality can distort the results of monoculture/polyculture experiments, is clearly illustrated in this study. This result confirms, for a lepidopteran/crucifer system, a previous suggestion from an experiment with the cucumber beetle/cucurbit system (Bach 1988a), that variation in plant quality may account for decreased herbivore densities, found in diverse habitats, by many researchers. The results of this study, as in Bach's work, also emphasize that herbivore response to different planting types is sometimes important, even if variable plant quality is taken into account. In some cases herbivore density varied as a linear function of a plant quality parameter. Fourth instar P. xylostella density and third instar M. configurata densities, for instance, decreased as plant width increased; fifth instar A californica density increased as plant width increased; and third instar jVL configurata density decreased with increasing plant growth rates as well as plant height. Densities of 70 first instar larvae did not vary directly with any plant quality parameter. As first instar densities are a good approximation of egg densities, this result suggests that female choices for oviposition sites, based on plant quality parameters, did not determine larval distribution. Preferences of a mortality agent within plots, such as the spider E. ovata. or parasitoids (see Chapter 3), for specific plants could better account for the linear relationships found between later larval instar densities and plant quality parameters. CONFOUNDING EFFECTS OF PREDATION With two herbivore species, R xylostella and ML configurata. responses to the addition of the generalist predator E ovata were different in monocultures than in dicultures. In monocultures reduced densities of these two species were found in plots with added spiders; in dicultures increased densities were found in plots with added spiders. Different responses to planting types, such as this, can result from complicated relationships between insects operating at different trophic levels and plants (Price et al. 1980). For example, generalist predators can indirectly cause a decrease in populations of parasitoids by disturbing oviposition behaviour or preying preferentially on parasitised larvae. This response would result in an increase in herbivore densities as a result of the presence of additional generalist predators, if parasitoids are an important mortality factor of the target herbivore. Because parasitoid responses can differ in monocultures and dicultures, indirect effects of generalist predators, such as the one suggested, would only be of biological significance in situations where parasitoids are an important mortality factor. These results also emphasize that predators can be very effective at reducing herbivore populations in monocultures-perhaps more effective than in dicultures. 71 The enemy hypothesis, and the literature which supports it (Russell 1989) suggests that predators are only effective in polycultures. However, I hypothesize that predators may take longer to colonize a monoculture. Hence, their importance over a growing season cannot be assessed accurately in experiments which cover one season, or one generation. Low predator numbers often found in experiments in which natural enemies are deemed unimportant confirm this hypothesis. For example Bach (1980b) found a maximum of 4.7 +_ 2.9 hemipteran predators in a 49m 2 plot, which is a density of half that of introduced spiders in my augmentations plots. Fifth instar A californica larvae were not found in any plots with additional spiders. Whether or not this was directly due to E. ovata predation is not clear. However, the presence of additional E. ovata obviously has an effect on A californica populations. HERBIVORE DENSITY DIFFERENCES ATTRIBUTABLE TO THE DIVERSITY OF A PLANTING In plots without spider augmentations, which are comparable to other experiments in which predator populations were not manipulated, densities of both second and third instar P\ xylostella, and second instar configurata were greater in monocultures than dicultures. This result is in direct agreement with that found by many other researchers in a number of other cropping systems (reviews in Risch et al. 1983, Russell 1989, Vandermeer 1989). However, this difference disappeared in second generation P\ xylostella. confirming suspicions that a short experimental period, lasting only one season, or spanning only one generation of a multivoltine species, can skew results. As my results illustrate, it is possible that results of 72 short-term experiments are over-interpreted since they consider only the rate of initial colonization (Kareiva 1983). CAUSES OF LARVAL DISTRIBUTION As was the case with plant quality parameters, no changes in herbivore densities due to the experimental treatments were found in first instars of any of the four species studied. Differences did not appear until later instars. Thus, it is unlikely that female preferences for oviposition sites in monocultures, were a determinant of the higher populations of P. xylostella and M L configurata in monocultures without additional spiders . Clearly, other factors, including within plot mortality factors, such as predation, or movement of older instars, resulting in emigration from plots, could account for these differences. Although the importance of herbivore movement is well documented for some insects (Kareiva 1983, Bach 1980a,b, 1981,1984), movement is an unlikely explanation in my particular experimental setting. Both P. xylostella and M , configurata larvae can move from plant to plant within plots. However, the likelihood of movement between plots separated by 5.5m, cultivated margins seems small. Larvae of neither species is highly mobile and the frequent cultivation of margins between plots eliminated the possibility of finding refugia between plots. In the literature the greatest support for the hypothesis that movement into and out of concentrated resources is the driving mechanism behind increased herbivore densities in monocultures comes from studies on chrysomelid beetles (Acalymma vittata. the striped cucumber beetle, Bach 1980b, 1981,1984: Phyllotreta cruciferae (Goeze). Root 1973, Tahvaneinen and Root 1972; 6 chrysomelids, Risch et al. 1983). For lepidopteran pests, where mobility is important to adults, a life stage which is not phytophagous, but not to the more sedentary but phytophagous larvae, it is not surprising that mortality within 73 plots, like that due to natural enemies, is more important than movement. In agreement with this suggestion is a study by Dempster (1969) on P. rapae. in which mortality rates within plots, not oviposition by females nor movement out of plots, accounted for differences in density. HERBIVORE FEEDING RATES Leaf areas did not differ between plants in monoculture and diculture plots, but total leaf area consumed by herbivores, was greater in dicultures. Higher herbivore densities in some diculture plots, i.e. second instar configurata larvae in dicultures with additional spiders, were cancelled out by interaction effects, i.e. second instar ML configurata larvae were more abundant in monocultures with no additional spiders. Therefore, herbivore density could not account for the increased consumption of leaves in dicultures. However, reduced feeding rates by the complex of phytophagous insects present could account for the differences in herbivory found. Two possible explanations for this novel finding can be hypothesized: 1. Percent parasitism, discussed in Chapter 3, is higher in monocultures. Because parasitised larvae often feed less than unparasitized ones (eg. Hill 1988), although not always (Price et al. 1980), increased parasitism in monocultures could result in decreased feeding rates; 2. Differences in microclimate between planting types might result in lower feeding rates in monoculture. There is evidence with several insect herbivores that microclimate can indeed affect feeding rate (Osisanya 1970, White 1978, Risch 1981, Harrison 1987). Herbivore densities, and resultant herbivory, were low in this study. The percent of leaf area eaten ranges from a mean of 1.5% to 4.3%-hardly the result of a population explosion of phytophagous insects. 74 SUMMARY The results presented here do not support the dogma that herbivore populations are greater where a plant resource is concentrated. Without exception, differences due to the planting treatment were either accounted for by differences in plant quality or by interactions with predator augmentations. The contention that mortality within plots, and particularly predation, is of secondary importance to herbivore movement patterns is refuted. Although responses to crop diversity are variable, determinants of herbivore density in this study were always a result of mortality within plots, rather than colonization by females. Larval movement was an unlikely explanation. The significant effects caused by the manipulation of the most common generalist invertebrate predator indicate that predation could be an important within-plot mortality factor. Most importantly, the effects of the generalist predator were deary distinct in monoculture and diculture plots. 75 CHAPTER 3 PARASITISM IN RELATION TO PLANTING TYPE, SPIDER DENSITY AND PLANT QUALITY INTRODUCTION PLANT QUALITY An important determinant of the total percent parasitism of insects is the efficiency with which parasitoids are able to find their hosts. Volatiles, emanating from host food plants, can provide important cues to insect parasitoids searching for host habitats, at a long range, and hosts, at a shorter range (Vinson 1976, Price et al. 1980, Vinson 1984). The role of chemical cues from adjacent non-host food plants is not clearly understood. In some cases olfactory cues from host food plants, or hosts themselves, can be masked so that parasitoids are unable to locate their hosts (Atsatt and O'Dowd 1976, Price et al. 1980). An example of this phenomenon was found in a natural system, in which odours from non-host plants (Pjcea) interfered with location of larch sawflies, Pristiphora erichsonii (Htg) by two tachinid parasitoids, Drino bohemica Mesn and Bessa harveyi Tns. (Monteith 1960). Altieri and colleagues (1981) found increased parasitism in fields sprayed with water extracts of weeds, implying that non-host food plants can actually attract parasitoids. Results have been ambiguous in agroecosystems, when the total percent parasitism of herbivore species in monocultures have been compared to polycultures. Although percent parasitism of herbivores is most often higher in polycultures (Altieri et al. 1981, Letourneau and Altieri 1983, Letourneau 1987), many examples can be found where parasitism is higher in monocultures (Pimentel 1961, Smith 1976, Andow and Risch 1987). 7 6 The results presented in Chapter 2 showed that the densities of four lepidopteran species in brussels sprouts were as strongly influenced by plant quality as by vegetational diversity. In this chapter I hypothesize that percent parasitism of those lepidopterans, like the densities of the hosts themselves, is independently influenced by plant quality. If my hypothesis is supported, variation in plant quality might explain some of the ambiguity of results in monoculture/polyculture experiments. Only two of the lepidopterans, Plutella xylostella and Autographa californica. found in this study harboured insect parasitoids. To test my hypothesis, I measured the percent parasitism of these species in monocultures of brussels sprouts and compared it with percent parasitism in dicultures of brussels sprouts/peppermint. Plant growth rate, leaf area, height and width were measured and percent parasitism was adjusted, using covariate analysis, to account for differences due to plant quality parameters. PREDATION The effects of generalist predators on percent parasitism can be directly affected by vegetational diversity in the same way that herbivore densities can be affected. In particular, predators may forage more efficiently in particular habitats, thereby disturbing parasitoid oviposition in some situations but not others; microclimates may be more favourable for predators in polycultures, as compared to monocultures; differences in the susceptibility of parasitised and unparasitised hosts to predation can be altered by vegetational diversity. Complicated interactions between different trophic levels, like those suggested above, have received little attention in discussions on insect populations in 77 monocultures and polycultures (Sheehan 1986). In this chapter, I also hypothesize that generalist predators exert an important influence on the percent parasitism of the lepidopterous herbivores studied, and that their impact varies between monocultures and polycultures. To test this hypothesis I conducted a two-factor experiment in which the densities of a common invertebrate predator, the theridiid spider Enoplaanatha ovata Clerck, were manipulated in the monocultures and dicultures (as described in Chapter 1). I then determined the relative importance of vegetational diversity and generalist predation on percent parasitism of F\ xylostella and A californica. 78 MATERIALS AND METHODS THE PARASITOIDS Plutella xylostella was parasitised by Diadeama insulare (Cress.) (Hymenoptera: Ichneumonidae) and A californica was parasitised by Voria ruralis (Fallen) (Diptera: Tachinidae). Both of these parasitoids frequently occur in surveys of the natural enemies of pests of crucifer crops (Puttarudriah 1953, Oatman 1966, Harcourt 1986, Lim 1986). Diadegma insulare forms an integral part of a large, cosmopolitan, complex of natural enemies attacking F\ xylostella (Harcourt 1986, Lim 1986). Its life history is well synchronised with the development of its host. TX insulare has several generations a year, as does its host, P. xylostella. It overwinters as a pupa in the host cocoon. \X insulare has been documented to be a key mortality agent in a study of R xylostella in Ontario, where parasitism by CX insulare ranged from 0 to 89% with a mean of 38% (Harcourt 1986). Voria ruralis has been studied in more detail than D. insulare. It, too, is a cosmopolitan parasitoid. Although it is most often recorded as a parasite of the cabbage looper, Trichoplusia ni (Hubner), it infrequently parasitises other hosts (McKinney 1944, Butler 1958, Oatman 1966, Brubaker 1968, Martin et al. 1981). It was the only parasitoid of the alfalfa looper, A. californica. recovered in this study. In one Californian study, Voria ruralis was the most important natural mortality agent on A californica. accounting for 24% of the host mortality (Puttarudriah 1953). Voria ruralis has been repeatedly recorded as a mortality agent of J_. Di (Oatman, 1966; Brubaker, 1968; Oatman and Platner, 1972; Oatman et al. 1983). 79 The life history of NA ruralis is well documented. Females lay their eggs directly on the larvae, the eggs hatch within 1 minute of oviposition and the first instar larva penetrates the integument and enters host muscle fibres. Voria ruralis displays no clear ovipositional preference for host instar, but it is most successful in completing its development when second or third instar hosts are parasitised. This permits pupation by the end of the host's fifth instar. There seems to be little host defense against NA ruralis (Elsey and Rabb 1970). The life cycle of this parasitoid takes an average of 26 days at 24 C. (Brubaker 1968). A californica parasitised by V. ruralis cease feeding. Black scars, indicating the location of developing V. ruralis are highly visible through the integument of fifth instar A. californica (Puttarudriah 1953, and pers. observation). Multiple oviposition events by the same female result in as many as 10 NA ruralis emerging from one host larva (Elsey and Rabb 1970; Grant and Shepard 1983). Superparasitism by several other parasitoid species also occurs, and is exaggerated at low host densities (Brubaker 1968). In this study up to seven NA ruralis emerged from one larva, but no superparasitism with other species was observed. Individual T. nj larvae parasitised by NA ruralis are often infected with a nuclear polyhedral virus (NPV). The parasitoid can develop from infected hosts, although survivorship is lowered when NPV is prevalent (Martin et al. 1981). The virus is voided from adult parasitoids soon after emergence and before oviposition, thus it is assumed that V . ruralis cannot act as a biological vector of the virus (Vail 1981). 80 DETERMINATION OF PERCENT PARASITISM As described in Chapter 1, on July 22 to July 29 and one month later, on August 26 to 29, approximately 50 plants per plot were sampled for all herbivorous insects. Numbers per plot varied from 46 to 53 and a total of 616 plants were sampled. Larvae collected this way were reared, as previously described, until each larva died, pupated or parasitoids emerged from it. Parasitoids emerged from P . xylostella and A. californica only. For P. xylostella parasitised by D. insularis. percent parasitism was calculated as: total number of D. insularis emerging X 100 number of F\ xylostella larvae & pupae For A. californica parasitised by the gregarious fly \A ruralis. percent parasitism was calculated as: No. of A. californica with V. ruralis emerging X 100 number of A. californica PLANT QUALITY, HERBIVORE DENSITY, AND EXPERIMENTAL DESIGN A detailed description of the methods for measuring plant quality and herbivore density are outlined in the Methods section, Chapter 2 . Plot to plot variability of the four plant quality parameters, height width, growth rates and leaf area, are given in the Results section, Chapter 2. As well, the experimental design is described in Chapter 1. 81 DATA ANALYSIS All analyses were done in a similar fashion to those for herbivore densities (Chapter 2), using the statistical package SPSS/PC+, unless otherwise indicated. Adjustments for significant plant quality parameters were made by using them as covariates in ANCOVAR. As in Chapter 2, comparisons between results with and without adjustments for plant quality were made by calculating an ANOVA test and comparing the results of ANCOVAR and ANOVA. Finally, linear regression was used to test for relationships between percent parasitism and plant quality, herbivore density and plant quality, percent parasitism in July and August, and percent parasitism and herbivore density. In all the linear regressions the null hypothesis, that the slope of the relationship being examined was equal to zero, was tested. Probability testing was done at the 0.05 probability level. All analyses using percent parasitism were calculated on both arcsin transformed data and untransformed data. In cases where there was no difference in interpretation between the two methods, untransformed test results are presented. 82 RESULTS A. PARASITISM OF PLUTELLA XYLOSTELLA BY DIADEGMAINSULARE. i) Parasitism and host life stage Diadegma insulare emerged more frequently from F\ xylostella collected as pupae than from f l xylostella collected as larvae (July, t[-| -\ ] = 1.2, p > 0.05; August, t[-| 1 j = 5.4, p<0.001, Table 3.1a). However, pupae were rare (Table 3.1b), probably because they represent only a short period of the life history, and host density was low (July, JL xylostella per plant = 0.401 \u00C2\u00B1 0.05; August JL xylostella per plant = 0.731 \u00C2\u00B1 0.06). Therefore, data for all life stages had to be pooled to achieve sufficient numbers for a meaningful analysis. Since the effects of planting types and spider densities on parasitism were the main focus of this study, rather than the impact of the parasite on its host, and the frequency of pupae to larvae did not change with either planting type or the spider density (July: planting type, F-|)8 = 1.28, p>0.05, spider density, F-jj8 = 1.39, p>0.05; August: planting type, F-j 3=0.0001, p>0.05, spider density, F-|8= 2- 6 8> p>0.05), pooling the life stage data did not create a serious problem (Van Driesche 1983). ii) Plant Quality For the linear regressions presented in this chapter, transforming the percentages with the arcsin transformation did not account for more than an extra 2% of the variability, above that accounted for using untransformed data. Therefore, results are presented on untransformed data. Percent parasitism of f l xylostella with a insulare decreased as the mean leaf area per plot increased, in July (r2=0.33, p=0.05, Figure 3.1 a). The linear relationship explained more of the 83 TABLE 3.1a: A COMPARISON OF THE PROPORTION OF PARASITISED E. XYLOSTELLA COLLECTED AS LARVAE AND PUPAE Mean Proportion Parasitised Per Plot (+_ s.e.): collected collected d.f. t p as larvae as pupae July 0.060\u00C2\u00B10.02 0.117+.0.05 11 1.2 >0.05 Aug. 0.104 + 0.02 0.380\u00C2\u00B10.05 11 -5.4 <0.001 TABLE 3.1b: PERCENT OF EACH LIFE STAGE COLLECTED AND PARASITISED Percent of Number Percent of F\ xylostella Parasitised Total collected Parasitism July larvae 145 58.9 8 47.1 pupae 101 41.1 9 52.9 August larvae 329 72.8 34 43.6 pupae 123 27.2 44 56.4 a. D^ insulare emerged more frequently from F\ xylostella collected as pupae than those collected as larvae, although the difference was only statistically significant at the 0.05 probability level for August. b . Pupae represented 41 % of all R xylostella collected in July and 27% of F\ xylostella collected in August. However, of the total percent parasitised in July and August, 53% and 56%, respectively, were pupae. The higher percent parasitism in the rarer life stage resulted in too few parasitised individuals for a meaningful analysis separated by life stage. Therefore, percent parasitism for all life stages was pooled. 84 FIGURE 3.1: J U L Y - PLANT QUALITY A N D PARASITISM OF PLUTELLA XYLOSTELLA BY DIADEGMA INSULARE 3000 4000 5000 6000 7000 8000 9000 0.3 0.4 0.5 0.6 0.7 L E A F AREA (cm ) EARLY SEASON GROWTH RATB (cm/day) I I I I\" I ' p>0.05 < rt Pi w 30 20 10 p>0.05 -10 14 15 16 17 18 19 PLANT HEIGHT (cm) 20 21 -10 50 55 60 PLANT WIDTH (cm) 65 Percent parasitism of P. xylostella by [X insulare. in July, decreased as leaf area of the brussels sprouts plants increased. However, no linear relationships were found between percent parasitism in July and any other plant quality parameters. 85 variability in the data if percent parasitism was greater than or equal to 5% (n=0.63, p < 0.05, Figure 3.2). It did not vary directly with any other plant quality parameter in July (Figures 3.1 b to d). Percent parasitism increased as early season growth rate increased in August (1-2=0.33, p<0.05, Figure 3.3b). In August, for all other plant quality parameters, percent parasitism of R xylostella could not be explained by an increase or decrease in the parameter (Figures 3.3a,c,d, 3.4a-c). R xylostella density did not vary with the same plant quality parameters as percent parasitism (Figure 3.5a, Figure 3.6b). Plutella xylostella density, in August, decreased with an increase in plant width at the July sample (Figure 3.6d). This was the same relationship between density and fourth instar larvae found when life stage data were separated (Chapter 2, Figure 2.3b). Scattergrams of R xylostella density and all plant quality parameters are presented in Figures 3.5, 3.6, 3.7. iii) Percent parasitism and treatments After adjusting for the effects of leaf area and early season growth rate (as covariates) in July, and early season growth rate alone in August, the mean percent parasitism did not significantly differ in monoculture and diculture plots (July: F-|6=5.51; p>0.05, August: F-\j = 3.21, p>0.05; Tables 3.2 and 3.3, ANCOVAR summaries; Table 3.4 for means). In contrast, the manipulations of spider densities exerted a measureable effect above that associated with plant quality. In July, plots with spider augmentions had significantly greater parasitism than plots without spider augmentations (F-|>6=7.17; p<0.05; percent parasitism = 3.2+_ 2% without augmentations and 11.5 \u00C2\u00B1 3% with augmentations). In August, the opposite result occurred. Percent parasitism was greater in plots without spider augmentations (Fi j = 6.05; p<0.05; percent parasitism = 20.3 +_ 3% without augmentations and 14.7 \u00C2\u00B1 2% FIGURE 3.2: J U L Y - L E A F AREA AND PARASITISM OF P L U T E L L A XYLOSTELLA WITH PARASITISM > 5% 86 With percent parasitism _> 5%, leaf area accounted for 63% of the variability in percent parasitism, as compared to 33% when percent parasitism < 5% was included in the regression analysis. This indicates that there is a lower threshold of parasitism below which leaf area was not important. 87 FIGURE 3.3: AUGUST - PLANT QUALITY AND PARASITISM OF PLUTELLA XYLOSTELLA BY DIADEGMA INSULARE 3000 4000 5000 6000 7000 8000 9000 0.3 0.4 0.5 0.6 0.7 LEAF AREA (cm > - JULY SAMPLE EARLY SEASON GROWTH RATE (cm/day) 14 IS 16 17 18 19 20 21 PLANT HEIGHT (cm) - JULY SAMPLE SO 55 60 65 PLANT WIDTH (cm) - JULY SAMPLE In August, percent parasitism of P. xylostella by D. insulare increased as early season growth rate increased (b). Percent parasitism did not increase or decrease with any other plant quality parameter (i.e. the slopes of the regression lines were not significantly different than zero). 88 FIGURE 3.4: AUGUST - PLANT QUALITY A N D PARASITISM OF PLUTELLA XYLOSTELLA BY DIADEGMA INSULARE 4000 5000 6000 7000 8000 9000 10000 LEAF AREA (cm ) - AUGUST SAMPLB 20 30 40 SO PLANT HEIGHT (cm) - AUGUST SAMPLE 40 50 60 70 PLANT WIDTH (cm) - AUGUST SAMPLE Percent parasitism of FL xylostella by [X insulare in August was not a linear function of any plant quality parameter measured at the time of sampling (i.e. the slopes of the regression lines were not significantly different than zero). FIGURE 3.5: J U L Y - PLANT QUALITY A N D DENSITY OF P L U T E L L A XYLOSTELLA 89 3000 4000 S0O0 6000 7000 8000 9000 LEAF AREA (cm ) 1.0 i 0 8 \u00E2\u0080\u00A2J CM fit 3 0.6 h 3 W 0.4 51 0.2 0.0 \u00E2\u0080\u00A2 p>0.05 03 0.4 0.5 0.6 0.7 EARLY SEASON GROWTH RATE (cm/day) 0.4 H 0.2 h -i 1 r \u00E2\u0080\u00A2 p>0.05 J L 14 IS 16 17 18 19 20 21 PLANT HEIGHT (cm) 1.0 0.8 a-1 CM Pi 3 0.6 0.4 02 0.0 SO p>0.05 SS 60 PLANT WIDTH (cm) 65 In July, f l xylostella density did not increase or decrease with any of the plant quality parameters (i.e. the slopes of the regression lines were not significantly different than zero). 90 FIGURE 3.6: AUGUST - PLANT QUALITY A N D DENSITY OF PLUTELLA XYLOSTELLA 15 < fc 1.0 Pi \u00C2\u00AB \u00E2\u0080\u00A2< W 0.5 0.0 n 1 r p>0.05 3000 4000 3000 6000 7000 8 000 9000 LEAP AREA tea?) - J U L Y SAMPLE 0.3 0.4 05 0.6 0.7 E A R L Y SEASON GROWTH RATB (cm/d\u00C2\u00BBy> 0.0 14 IS 16 18 19 20 21 0.0 SO 55 60 65 PLANT HEIGHT (cm) - JULY SAMPLE PLANT WIDTH (cm) - J U L Y SAMPLB In August, P. xylostella density decreased as the width of brussels sprouts plants increased in July (d) . That is. P. xylostella density was a function of plant width five weeks before the August sample. P\ xylostella density was not as a linear function of any other plant quality parameter (i.e. the slopes of the regression lines were not significantly different than zero). 91 FIGURE 3.7: AUGUST - PLANT QUALITY A N D DENSITY OF PLUTELLA XYLOSTELLA is i i i i r ft < a 1.0 -Qi -p>0.05 0.0 4000 5000 6000 7000 8000 9000 10000 LEAF AREA (cm1) - AUGUST SAMPLB 20 30 40 SO PLANT HEIGHT (cm) - AUGUST SAMPLE 40 50 60 70 PLANT WIDTH (cm) - AUGUST SAMPLE In August, P\ xylostella density did not increase or decrease with any of the plant quality parameters, measured at the time of sampling (i.e. the slopes of the regression lines were not significantly different than zero). 92 TABLE 3.2: ANALYSIS OF COVARIANCE SUMMARY Proportion of P\ xylostella parasitised by [X insulare. July sample. by PT Planting Type (Monoculture and Diculture) SD Spider Density (augmentations or no augmentations) with EGR Early Season Plant Growth Rate JLA Leaf Area Source of Variation Sum of Squares DF Mean Square F P Covariates EGR JLA 0.016 0.017 1 1 0.016 0.017 14.37 15.09 <0.01 <0.01 Main Effects PT SD 0.006 0.008 1 1 0.006 0.008 5.51 7.17 >0.05 <0.05 2-Way Interactions PT ST 0.000 1 0.000 0.125 >0.05 Explained Residual 0.050 0.007 5 6 0.010 0.001 8.770 <0.01 Total 0.057 11 0.005 Plant quality covariates that were not statistically significant were eliminated from this ANCOVAR. Statistical significance was tested at both the 0.01 and 0.05 probability levels. Once adjustments for the plant quality covariates were made, percent parasitism was not significantly different in monocultures as compared to dicultures, but it was significantly different in the spider augmentation plots as compared to the no augmentation plots. 93 TABLE 3.3: ANALYSIS OF COARIANCE SUMMARY Proportion of F\ xylostella parasitized by [X insulare. August sample by PT SD with EGR Planting Type (Monoculture and Diculture) Spider Density (Augmentations and No Augmentations) Early Season Plant Growth Rate Source of Variation Sum of Squares DF Mean Square F P Covariates EGR 0.016 1 0.016 8.881 <0.05 Main Effects PT SD 0.006 0.011 1 1 0.006 0.011 3.212 6.048 >0.05 <0.05 2-Way Interactions PT SD 0.002 1 0.002 1.225 >0.05 Explained Residual 0.036 0.013 4 7 0.009 0.002 4.922 <0.05 Total 0.048 11 0.004 Plant quality covariates that were not.statistically significant were eliminated from this ANCOVAR. Statistical significance was tested at both the 0.01 and 0.05 probability levels. Once adjustments for the plant quality covariates were made, percent parasitism was not significantly different in monocultures, as compared to dicultures, but it was significantly different in the spider augmentation plots as compared to the no augmentation plots. 94 TABLE 3.4: PERCENT PARASITISM OF P\ XYLOSTELLA BY LI INSULARE (Means \u00C2\u00B1 s . e . ) A . JULY Monoculture Diculture TOTAL No Spider 4 . 8 \u00C2\u00B1 3 1 . 5 \u00C2\u00B1 2 3.2\u00C2\u00B12 Augmentations Spider 16.1 \u00C2\u00B1 2 6.5 \u00C2\u00B1 4 11.5 \u00C2\u00B13 Augmentations TOTAL 10.5 + 3 4.0 + 2 B. AUGUST Monoculture Diculture TOTAL No Spider 25.0 \u00C2\u00B1 3 15.6 \u00C2\u00B1 1 20.3 \u00C2\u00B1 3 Augmentations Spider 17.5 \u00C2\u00B1 1 11.7 \u00C2\u00B1 4 14.6 \u00C2\u00B12 Augmentations TOTAL 21.2+2 13.6+2 In July (a) percent parasitism of FL xylostella by D. insulare was significantly greater in the spider augmentation plots than in the plots without spider augmentations (FH 6] =7.17). In August (b) percent parasitism of FL xylostella by \X insulare was significantly greater in the plots without spider augmentations than the plots with spider augmentations (^ [1 7] =6.05). Significance testing was done at the 0.05 probability level. 95 with augmentations). As well, in plots with spicier augmentations, percent parasitism did not increase significantly from July to August (t[5j=-1.69, p>0.05). In plots without spider augmentations, percent parsitism increased 6-fold from July to August (t[5]=-8.931,p< 0.001, Figure 3.8). A summary of the differences in percent parasitism in spider augmentations plots, as compared to control plots without spider augmentations, is presented in Table 3.5. As in Chapter 2, two-way ANOVA's were calculated to uncover errors in interpretation that would have resulted had adjustments for plant quality been omitted from the analysis. Without adjustments for the plant quality covariates, percent parasitism was significantly greater in monoculture plots than diculture plots (for July, planting type, Fis = 4.89, p=0.058; for August, planting type, F - ia = 6.83, p<0.05; ANOVA summaries, Tables 3.6a and b). Conversely if plant quality had not been taken into account, in August, the importance of predation by the spider, B ovata. on percent parasitism, would have been masked (July, spider density, F - | 8 = 7 - 7 7 > p<0.05; August, spider density, F-| 3 = 3.88; p>0.05; Tables 3.6a and b). iv) Period of Sampling Percent parasitism in the experimental plots in August was not correlated with percent parasitism in July (Figure 3.9), although it was greater in August than July $[11] = -3.03, p<0.01; Tables 3.4a and b). 96 FIGURE 3.8: PARASITISM OF PLUTELLA XYLOSTELLA IN J U L Y AND AUGUST SPIDER AUGMENTATIONS 20 -2 JULY AUGUST SAMPLE DATB NO SPIDER AUGMENTATIONS JLY AUGUST SAMPLE DATE Percent parasitism of PV xylostella by EX insulare was not significantly greater in August than July in spider augmentation plots (t[5]=-1.69, p > 0.05). In plots without spider augmentations percent parasitism was significantly greater in August than July (trsj =-8.931, p<0.001). TABLE 3.5: SUMMARY OF PERCENT PARASITISM DIFFERENCES BETWEEN SPIDER AUGMENTATION PLOTS AND CONTROL PLOTS WITHOUT SPIDER AUGMENTATIONS WITH SPIDER AUGMENTATIONS 1. July sample: % parasitism decreases with increasing host density 2. August sample: % parasitism increases with increasing host density 3. July to August samples: No increase in % parasitism from July to August 4. July sample: % parasitism greater than no spider augmentations 5. August sample: % parasitism less than no spider augmentations NO SPIDER AUGMENTATIONS % parasitism increases with increasing host density no linear relationship between parasitism and host density 6-fold increase in percent parasitism from July to August % parasitism less than spider augmentations % parasitism greater than spider augmentations 98 TABLE 3.6 ANALYSIS OF VARIANCE SUMMARY Percent Parasitism of R Xylostella by D. Insulare by PT Planting Type (Monoculture and Diculture) SD Spider Density (Augmentations and No Augmentations) A. JULY Source of Variation Sum of Squares DF Mean Square F P Main Effects PT SD 0.013 0.020 1 1 0.013 0.020 4.89 7.77 >0.05 <0.05 2-Way Interactions PT SD 0.003 1 0.003 1.18 >0.05 Explained Residual 0.036 0.021 3 8 0.012 0.003 4.61 <0.05 Total 0.057 11 0.005 B. AUGUST Source of Variation Sum of Squares DF Mean Square F P Main Effects PT SD 0.017 0.010 1 1 0.017 0.010 6.825 3.884 <0.05 <0.05 2-Way Interactions PT SD 0.001 1 0.001 0.387 >0.05 Explained Residual 0.028 0.020 3 8 0.009 0.003 3.698 >0.05 Total 0.048 11 0-004 ANOVAs were calculated to compare percent parasitism without plant quality adjustments to percent parasitism with plant quality adjustments (i.e. ANCOVAR). Significance testing was at the 0.05 probability level. 99 FIGURE 3.9: PARASITISM OF PLUTELLA XYLOSTELLA IN AUGUST AS A FUNCTION OF PARASITISM IN J U L Y -10 0 10 20 PERCENT PARASITISM. J U L Y 30 Percent parasitism of R xylostella by a insulare in August was not a linear function of percent parasitism in July (i.e. the slope of the regression line was not significantly different than zero). 3 100 v) Percent parasitism and host density When the data for all plots, regardless of treatment, were pooled, percent parasitism by D. insulare did not change in a density-dependent manner, with the density of the host FL xylostella (Figures 3.10a and b). However, the relationship between percent parasitism and host density differed in plots with different experimental treatments. Percent parasitism and host density were not correlated in either monocultures or dicultures in July (Figures 3.11a and b). In August percent parasitism increased, as P. xylostella density increased (r2=0.79, p<0.05, in diculture plots (Figure 3.11 d) but not in monoculture plots (Figure 3.11c). As well, the augmentation of spiders affected the relationship between mortality due to CX insulare and FL. xylostella density. In July, in plots with spider augmentations, percent parasitism decreased as P. xylostella density increased (r2=0.89, p<0.05, Figure 3.12a), but only when FL xylostella density was below 0.5 individuals per plant (Figure 3.11c). In contrast, in July, without spider augmentations, percent parasitism increased as FL xylostella density increased ((^=0.79, p<0.05, Figure 3.12b). In August, with spider augmentations, percent parasitism increased with host density (r2=0.54, p<0.1, Figure 3.12d). Without spider augmentations mortality in August did not vary as a result of host density (Figure 3.12e). These results are summarized in Table 3.5. B. PARASITISM OF AUTOGRAPHA CALIFORNICA WITH VORIA RURALIS As explained in Chapter 2, the analysis for A, californica was done only on the data from the July sample because so few A californica larvae (only 13 on 616 plants sampled) were collected during the August sampling period. 101 FIGURE 3.10: DENSITY OF PLUTELLA XYLOSTELLA A N D PARASITISM BY DIADEGMA INSULARE 0.0 0.3 0.6 0.9 12 P. XYLOSTELLA PER PLANT - JULY 0.0 03 0.6 0.9 12 P. XYLOSTELLA PER PLANT - AUGUST These linear regressions show that percent parasitism of FL xylostella with H insulare did not increase or decrease as a linear function of host density when data for each sampling date were pooled (i.e. the slopes of the regression lines were not significantly different than zero). 102 FIGURE 3.11: PARASITISM. HOST DENSITY A N D PLANTING T Y P E M O N O C U L T U R E 0.0 0.3 0.6 0.9 1J2 P. XYLOSTELLA PER PLANT - JULY D I C U L T U R E 0.0 0.3 0.6 0.9 1_2 P. XYLOSTELLA PER PLANT - JULY In July percent parasitism of P. xylostella by D, insulare did not vary as a linear function of host density in either monocultures or dicultures (i.e. the slopes of the regression lines were not significantly different than zero). However, in August, percent parasitism increased as host density increased in diculture plots, but not in monoculture plots. 103 FIGURE 3.12: PARASITISM, HOST DENSITY AND SPIDER DENSITY. The augmentation of spiders affected the relationship between percent parasitism and host density. In July percent parasitism decreased as host density increased in spider augmentation plots (a), but only when host density was below a threshold of 0.5 larvae per plant (c). Percent parasitism increased as host density increased in plots without spider augmentations (b). In August, percent parasitism increased as host density increased in spider augmentation plots (d) but not in plots without spider augmentations (e). However, density was less variable in plots without spider augmentations than with spider augmentations. All linear regressions tested the null hypothesis that the slope of the regression line was equal to 0 at the 0.05 probability level. ]04 FIGURE 3.12: PARASITISM. HOST DENSITY A N D SPIDER DENSITY SPIDER AUGMENTATIONS a OLO O J O.S 0.9 1.2 P . X Y L O S T E L L A P E S . P L A N T - J U L Y NO SPIDER AUGMENTATIONS b 0.0 O J 0.6 0.9 1.2 P . X Y L O S T E L L A P E R P L A N T - J U L Y 0.0 0 J 0.\u00C2\u00AB 0.9 1.2 P . X Y L O S T E L L A P E R P L A N T - J U L Y 0.0 0.3 0.6 0.9 12 T. X Y L O S T E L L A P E R P L A N T - A U G U S T 0.0 0.3 0.6 OS 12 P. X Y L O S T E L L A P E R P L A N T - A U G U S T 105 i) Parasitism and host life stage Voria ruralis emerged more frequently from A. californica collected as older instars than from those collected as younger instars (F^ 113=2.73, p < 0.05, Table 3.7a). As with E L xylostella. A californica density was low, with a mean number of 0.22 larvae per plant (+_ 0.03). The frequency of older instars (i.e. third, fourth and fifth instars) to younger instars (i.e. first and second instars) did not change with either planting type (Fi>8=4.53, p>0.05) or spider density (F-|)8=0.035, p>0.05). In addition, the focus of the study was the relationship between experimental treatments and parasitism, not the impact of the parasite on its host. For these reasons, data were pooled for the analyses. ii) Plant quality In contrast to the results for P. xylostella parasitised by insulare. percent parasitism of A californica by V. ruralis did not change, in a linear fashion, with any of the plant quality parameters measured. Moreover, none of the plant quality parameters were statistically significant covariates. iii) Percent parasitism and treatments Percent parasitism of A. californica by \A ruralis did not differ due to either the planting type or the spider density (planting type: F-| g= 1.34, p>0.05; spider density: F-| s = 0.34, p>0.05; Table 3.7b for means). 106 TABLE 3.7a: PERCENT PARASITISM OF A, CALIFORNICA BY V. RURALIS (MEAN \u00C2\u00B1 S.E.) INSTAR N MEAN 1 11 18.2^12 2 45 19.1 \u00C2\u00B1 6 3 35 31.4 \u00C2\u00B1 8 4 21 47.6 \u00C2\u00B1 1 1 5 6 66.7 \u00C2\u00B1 2 1 The means represent the percent of A californica per plot that died of parasitism. The 'n' represents the total number of plants, of the 616 sampled, that were infested with A californica. For example, first instar A. californica were found on 11 plants, of the total 616 sampled. Of these larvae, 18.2\u00C2\u00B112%, per plot, had \A ruralis parasitoids emerge from them. TABLE 3.7b PERCENT PARSITISM OF A, CALIFORNICA PER PLOT, SEPARATED BY EXPERIMENTAL TREATMENTS (Means \u00C2\u00B1 s.e.) Monoculture Diculture Spider 32.1 \u00C2\u00B1 6 19.4 \u00C2\u00B1 1 2 Augmentations No Spider 40.8 \u00C2\u00B1 7 25.6 \u00C2\u00B1 21 Augmentations Percent parasitism of all A, californica instars, combined, was calculated as the percent of A californica larvae from which V, ruralis emerged. Using the simple calculation of number of parasitoids per larvae was not suitable because \A ruralis is a gregarious parasitoid. 107 iv) Percent parasitism and host density Percent parasitism did not change in relationship to host density (Figure 3.13), regardless of whether or not the analyses was done separately for treatments. Scattergrams of percent parasitism and A. californica. separated by experimental treatments are presented in Figure 3.14. 108 FIGURE 3.13: AUTOGRAPHA CALIFORNICA DENSITY A N D PARASITISM BY VORIA RURALIS 0.1 0.2 0.3 0.4 A. CALIFORNICA PER PLANT 0.5 Percent parasitism of A californica by V. ruralis did not increase or decrease as host density increased or decreased when data for all experimental treatments were pooled (i.e. the slope of the regression line was not significantly different than zero) 109 FIGURE 3.14: PARASITISM. HOST DENSITY A N D PLANTING T Y P E -AUTOGRAPHA CALIFORNICA MONOCULTURE DICULTURE 0.1 0.2 0.3 0.4 A. CALIFORNICA PER PLANT 0.5 S 8 Pi W fe 0.1 02 OS 0.4 A. CALIFORNICA PBR PLANT 0.5 SPIDER AUGMENTATIONS 0.0 0.1 0.2 0J 0.4 A. CALIFORNICA PBR PLANT 0.5 NO SPIDER AUGMENTATIONS d 0.1 0.2 0.3 0.4 A. CALIFORNICA PBR PLANT 0.5 Percent parasitism of A californica by NA ruralis did not increase or decrease with host density for any of the experimental treatments (i.e. the slopes of the regression lines were not significantly different than zero). DISCUSSION Parasitism of Plutella xylostella by the ichneumonid Diadegma insulare. varied directly with plant quality, spider density and host density. However, percent parasitism did not vary with vegetational diversity once plant quality was accounted for. In contrast to P\ xylostella. parasitism of Autoarapha californica. by the tachinid Voria ruralis. collected from the same host plants as Pv xylostella. did not change with any of the plant quality parameters, experimental manipulations or host density. PLANT QUALITY AND EXPERIMENTAL TREATMENTS Percent parasitism of p\ xylostella by EX insulare. varied with two of the indicators of plant quality, leaf area and early season growth rate (Figs. 3.1-3.4). Specifically, parasitism was greater on plants with smaller leaf areas and faster growth rates. At close range EX insulare walks over the surface of a leaf searching for hosts. It is not surprising, then, that the probability of finding a host would be greater on leaves with a smaller surface area. In an experiment designed to examine parasitoid performance on different sized leaves, percent discovery of European corn borer, Ostrinia nubilalis (Hbnj) egg masses, by the chalcid, Trichogramma nubiale Ertle and Davis, increased with decreasing leaf area (Need and Burbutis 1979). When either measurements or casual observations have been made, plants are always smaller in polycultures than monocultures (Bach 1980b, Altieri and Liebman 1986, Letourneau 1987). Increased parasitism in polycultures, found by others, might consequently, be I l l explained by the smaller leaf areas in polycultures. In the large monoculture plantings, which are the earmark of modern agriculture, much more than reduced vegetational diversity has been manipulated. Crop varieties are usually selected for increased yield and often an associated increase in plant size. Decreased parasitism, as a result of the larger leaf area of plants, rather than the concentration of a plant resource, could account for higher herbivore densities in these monocultures. Host density did not vary with the same plant quality parameters as percent parasitism (i.e. host density decreased with an increase in plant width, Fig.3.6), implying that CI insulare responds to host plants independently of its host P. xylostella. These data support the speculation of others that parasitoids use plant clues independent of their hosts (Vinson 1984). Although a few vegetational diversity studies have accounted for plant quality in the analysis of herbivore densities (Bach 1980b, Elmstrom et al. 1988), the relationship between parasitism and plant quality has been largely ignored. In this study, when adjustments were made for leaf area and early season growth rate, through covariate analysis, interpretations of how percent parasitism varies with the experimental treatments (planting type and spider densities) had to be altered. In the first case, had adjustments not been made for plant quality, percent parasitism would have been misinterpreted as being greater in monocultures than dicultures, when in fact there was no significant difference between planting types. In the second case, once plant quality adjustments were included in the analysis, percent parasitism in August was unmasked as being lower in plots with spider augmentations than in those without spider augmentations. The reverse was true in July. However, the importance of additional spiders in July was already apparent before adjustments for plant 112 quality were made. PREDATION BY THE SPIDER, ENOPLAGNATHA OVATA Interestingly, the generalist predator, E ovata may be instrumental in organizing the relationship between P. xylostella and its ichneumonid parasitoid D. insulare. The addition of E. ovata to plots had a dampening effect on percent parasitism in August, the later generation of host and parasite sampled, as compared to July, the early generation (Table 3.5). In August, when Pv xylostella densities were the highest of the two sampling periods, decreased percent parasitism was associated with the addition of E ovata to plots. As well, mortality due to EX insulare increased 6 fold from July to August in plots without additional spiders. In contrast, in plots with added spiders, the slight increase in mean percent parasitism from July to August was not statistically significant (Figure 3.8). These results, coupled with the fact that P\ xylostella densities increased with the addition of spiders (Chapter 2), implies a direct relationship between E. ovata density and the populations dynamics of [X insulare. Several mechanisms could be at work here. Studies examining only one generation of a parasite and its host would not capture cumulative effects of either planting diversity or natural enemies. As with herbivore numbers, first generation studies of parsitoids only shed light on the dynamics of initial colonization. Important elements of the interaction between different trophic levels can be detected only when the temporal scale is extended to include more than one generation. In the example of this study, the addition of a generalist predator was associated with increased parasitism of the herbivore in the initial 113 colonization stage (July) and a decrease in parasitism in later stages (August). As population densities of P\ xylostella were lower in the earlier generation, it is probable that E. ovata did not 'switch' to F l xylostella as a prey item until the later generation. At the low f l xylostella densities in July, the spider was more likely to be consuming prey items other than Fl xylostella. including other natural enemies of f l xylostella. In fact, several incidents of E. ovata predation on coccinellids were noted in the experimental plots. Hence, in July, high densities of spiders might actually allow less interference with the parasite and its host, through the elimination of other natural enemies. In spider augmentation plots, preferential predation on parasitized hosts (Tostowaryk 1972, Frazer and van den Bosch 1973,Jones 1987), although not studied, could account for the decrease in percent parasitism in the August generations of f l xylostella and insulare. f l xylostella larvae normally drop from the plant on a silken thread when disturbed by parasitoids or predators (Waage 1983 and personal observation). Parasitised hosts are more likely to be captured by a predator because they are often sluggish and unable to exhibit their usual defense mechanisms against predation. Whether the effects of the spider are indirect or direct cannot be determined from these results. For example, it is possible that the presence of additional spiders attracted other generalist predators, such as birds, which then preyed more heavily on parasitized than unparasitized H xylostella larvae and pupae. Casual observations of birds at the study site hinted that bird predation may have played a significant role. 114 HOST DENSITY AND PARASITISM Although there is some evidence that parasitoids concentrate their attacks in regions of high prey density (Hassell and Southwood 1978, Murdoch et al. 1985) , results can vary widely. In the field, percent parasitism is just as likely to vary inversely with host density as it is to vary positively or not to vary at all with host density (Morrison and Strong 1980, Lessells 1985, Cheson and Murdoch 1986) . Discussion of this topic has focused on the relative importance of factors such as host plant distribution (Hassell and Southwood 1978, Murdoch et al.1985), parasitoid aggregation (Beddington et al. 1978, Waage 1983), patch quality (Kareiva 1986) and parasitoid egg or time limitations (Lessells 1985). Generally, the theoretical discussion has not considered vegetational diversity as an important determinant of the relationship between percent parasitism and host density. This study demonstrates that vegetational complexity can exert unexpected influences over host/parasitoid relationships. For example, parasitism by D^ insulare increased with host density, but only in diculture plots (Figure 3.11). The foraging movements of a parasitoid, such as EX insulare. would likely be altered in dicultures in much the same way as the foraging movements of predators are altered depending on the patchiness of a plant resource (Kareiva 1985). The probability of alighting on a non-host plant, and thus having to move more frequently, could result in a larger sampling of the available hosts, and hence a greater likelihood of responding positively to host density, in dicultures. Because P. xylostella is a crucifer specialist it is likely that its parasitoid, \X insulare uses host plant cues, as well as host density cues, to dictate its searching efforts. In monoculture plots the likelihood of encountering host plants is high. Regardless of host density, the parasitoid receives a continuous cue to stimulate searching. In contrast, in diculture plots 115 cues from host plants would not be as strong, simply because the probability of encountering a non-host plant would be increased. Therefore the only cue the parasitoid has to assess the profitability of a patch in dicultures is host density, and accordingly, its efforts are concentrated where host density is highest. Not only did parasitism by D\u00C2\u00B1 insulare increase as Fl xylostella density increased in diculture plots, but the relationship between parasitism and host density was also found to depend on the spider density treatments. In spider augmentation plots, percent parasitism was inversely density-dependent (Figure 3.12a) in July, the early generation of host and parasitoid, and density-dependent in August, the later generation of host and parasitoid (Figure 3.12d). Without spider augmentations, percent parasitism was density dependent in July (Figure 3.12b) and not dependent on host density in August (Figure 3.12e). The inverse density-dependence found in spider augmentation plots, in July, was probably related to the lower host densities (i.e. fewer than 0.55 f l xylostella per plant). Positive density-dependence occurred at higher host densities (i.e. greater than 0.4 H xylostella per plant). Generally, studies examining the relationship between parasitism and host density do not look for a response at densities as low as those found in this study. For example, percent parasitism of Diadegma eucerophaga Horstmann, a parasitoid of H xylostella closely related to H insulare. did not increase with host density (Waage 1983). However, the lowest density used in Waage's study was 2 larvae per plant. In my experiment, at mean host densities between 0.25 and 1.2 larvae per plant, percent parasitism was a linear function of host density. It is possible that Waage (1983) was unable to detect density dependence because his densities were too high for this group of parasitoids. 116 ASSESSING PERCENT PARASITISM The percent parasitism in this study was likely an underestimate for both host species. Calculating percent parasitism over all life-stages, when there are differences in susceptibility of different stages, usually results in an underestimate of the impact of a parasitoid on its host (van Driesche 1983). Mortality caused by parasitoid feeding or host death before parasitoid emergence, was not measured although these types of mortality can often be substantial (Murdoch et al. 1985). As my main interest was in the effects of planting types and spider densities on parasitism, rather than on the impact of the parasitoid on its host, the underestimate of parasitism did not create a serious problem. However, caution would have to be taken in using these results for predictions about the importance of either parasitoid in the population dynamics of its host. COMPARING THE TWO HERBIVORES Parasitism of F\ xylostella with L>. insulare changed with several of the parameters examined; parasitism of A californica with \A ruralis did not change with any of the parameters examined. For both species the host density was very low, as described in Chapter 2. (P\ xylostella: July, 0.401 _+ 0.05 per plant; August, 0.931 _+ 0.08 per plant. A californica: July, 0.217 \u00C2\u00B1 0.03 per plant). Interestingly, inspite of the low host densities percent parasitism was substantial for both examples. The mean percent parasitism for P. xylostella was 7% in July and 38% in August. For A californica it was 30 % in July. Some generalizations emerge when the two host/parasitoid associations 117 are compared: 1. One host.JP. xylostella. is a specialist feeder on crucifers, while the other host, A californica is a generalist feeder on many plants, including both crop species in this experiment. Several authors have predicted differences in the importance of natural enemies to the population dynamics of specialist and generalist herbivores (Root 1973, Sheehan 1986), and my results support this notion. However, in contrast to the what would be predicted from the laboratory work of Bernays (1988) and Bernays and Cornelius (1989), in this study the generalist predator E ovata had an impact on the specialist herbivore, while the polyphagous herbivore was not affected. 2. Although both host densities were low, that of A californica was lower and may account for the lack of response of V. ruralis to any of the plant quality parameters studied. SUMMARY Plant quality had an importance influence on percent parasitism. These results clearly demonstrate that plant quality can complicate mortality due to parasitism in monoculture/polyculture experiments to the same extent as plant density and spatial and temporal scales. As well, the potential for generalist predators to organize the relationship between herbivores and their parsitoids should not be underestimated. Interestingly, the form this organization takes differs with vegetational diversity. Thirdly, at low host densities parasitism can be high. It is not surprising that the importance of parasitoids such as D. insulare is overlooked when mortality due to parasitism is only studied at high host densities. Perhaps the importance of 118 pest/parasitoid relationships would be better appreciated if it were more frequently studied at low herbivore densities, before pest status is achieved. 119 CHAPTER 4 YIELD INTRODUCTION The ultimate purpose of intercropping is to increase yield. Most research on intercropping in temperate climates is conducted by crop physiologists interested in facilitative or competitive effects of two crops planted together (Francis 1986, Vandermeer 1989). In this study, strip-cropping was used specifically to prevent interspecific competitive effects between the two crops. Facilitative effects of peppermint on brussels sprouts yield were expected to be a result of reduced herbivory, alone. As already described in Chapter 2, all the herbivore densities in this study were very low, and the resulting damage to leaves was also low. It was, therefore, not expected that herbivores would have any significant effect on the crop yield. The facilitative effects of the peppermint on the brussels sprouts would not be apparent from a yield analysis. In the following Chapter I present the yield data, which confirm these expectations. 120 METHODS All sprouts from the twenty plants per plot, which had previously been sampled for plant quality parameters (Chapter 2) were later harvested to obtain yield data. Sprouts from ten plants per plot were harvested between October 1 to 27, for data on an early harvest; sprouts from the remaining ten plants per plot were harvested November 18 -27, for data on a late harvest. Harvesting was done by cutting all sprouts from plants and putting them into a paper bag, in the field. Counts and weights were taken within four hours of harvesting. Both the total fresh weight of sprouts per plant and the number of sprouts per plant were determined. Sprouts were then divided into four sub-groups, according to their marketability, reweighed and counted. Sub-group classifications were: marketable, damaged, oversized, undersized. In order to be considered marketable the sprout had to be compact, greater than one centimetre in diameter, and with no visible signs of damage from either insects of disease. Damaged sprouts were all those sprouts with visible signs of insect damage or disease symptoms. Sprouts which appeared marketable were cut in half on the first two days of harvest to insure that they did not harbour f l xylostella larvae. It became obvious, after two days, that all damage was detectable from the outside. Therefore, sprouts were not cut in half on subsequent harvest days. Oversized sprouts were no longer compact. Undersized sprouts were less than one centimetre in diameter. One-way ANOVA was used to compare the yield data for the October and November harvest dates. Two-way ANOVA, with planting type and spider density, was used to determine differences in yield according to experimental treatments. 121 RESULTS Yield data for the October and November harvest dates could not be pooled because of significant differences between October and November in almost all yield parameters (Table 4.1). Neither fresh weight nor number of sprouts per plant differed significantly due to either planting type or spider density for any of the parameters (p > 0.05; sample two-way ANOVA Table 4.2). The percentage of the total number of sprouts which were marketable sprouts was 43% by weight for October and 28% by weight for November (Figure 4.1; by number of sprouts, Figure 4.2). Although some of the damage was due to P\ xylostella feeding, most of it was caused by an unidentified disease. 122 TABLE 4.1: RESULTS OF ONE-WAY ANOVAS COMPARING YIELD ON THE OCTOBER AND NOVEMBER HARVEST DATES Parameter F-| 238 P total weight of sprouts 9.023 p<0.01 weight of marketable sprouts 0.304 p>0.05 weight of damaged sprouts 112.5 p<0.01 weight of oversized sprouts 7.99 p<0.01 weight of undersized sprouts 140.56 p<0.01 number of sprouts 4.41 p<0.05 number of marketable sprouts 13.59 p<0.01 number of damaged sprouts 136.11 p<0.01 number of oversized sprouts 10.60 p<0.01 number of undersized sprouts 131.18 p<0.01 For all yield parameters, except weight of marketable sprouts, yields in November were high than October. Therefore, date for October and November could not be pooled. 123 TABLE 4.2: TWO-WAY ANOVA SUMMARY ON YIELD DATA FOR OCTOBER HARVEST Weight of Marketable Brussels Sprouts by Pt (Planting Treatment) and Sd (Spider Density) Source of df Sum of Mean Variation Squares Square Within cells 108 389682 Replicate 2 6542 pt 1 63 sd 1 25687 p tXsd 1 2631 pt X rep 2 13357 sd X rep 2 30819 pt X sd X pt 2 2695 3608 0.46 >0.05 3271 0.42 >0.05 63 0.008 >0.05 25687 3.288 >0.05 2631 0.34 >0.05 7812 The denominator used to calculate F for all the main effects was the total sum of squares for all interaction effects with replicates and 6 degrees of freedom. A similar analysis as this was used for all yield parameters, on both sampling dates. 124 FIGURE 4.1: Fresh weight of brussels sprouts for the early harvest, October, and the late harvest, November. No significant differences in yield were found between monocultures and dicultures or spider augmentation plots and controls. All data were pooled for the graphs shown here. The proportion of marketable sprouts was small (i.e. 43% October, 28% November). A comparison, after harvesting, with brussels sprouts in Vancouver supermarkets indicated that my criteria for marketability were more stringent than the commercial criteria. FIGURE 4.1: YIELD IN FRESH WEIGHT OF SPROUTS OCTOBER HARVEST YIELD CATEGORY NOVEMBER HARVEST YIELD CATEGORY 126 FIGURE 4.2: Yield, in terms of number of sprouts per plant, was not significantly different in any of the experimental treatments. The total yields for all treatments are presented here. FIGURE 4.2: YIELD IN NUMBER OF SPROUTS/PLANT OCTOBER H A R V E S T YIELD CATEGORY N O V E M B E R H A R V E S T YIELD CATEGORY 128 DISCUSSION As predicted, no differences in yield were found between either monocultures and dicultures, or spider augmentations plots and plots without spider augmentations. The differences in herbivore densities and percent parasitism discussed in Chapters 2 and 3 did not have an impact on crop yield. The lack of difference in yield between the monocultures and dicultures provides support to my contention that strip-cropping, in this case, would prevent competitive interactions between the two crops. The high percentage of damaged sprouts, found at both sampling times, is mostly indicative of the high standards I used for 'marketability'. On subsequent examination of brussels sprouts in a supermarket I realized that my standards of marketability were substantially higher than those of the industry. 129 CHAPTER 5 GENERAL DISCUSSION This thesis does not support the widely held belief that resource concentration is of overriding importance in determining high herbivore densities in monocultures, as compared to polycultures. The abundance of the four lepidopterans in this study varied with plant quality and predation by the spider E. ovata. Resource concentration was only of minor importance. Plant quality, measured as plant height, width, leaf area and growth rates, was the most important determinant of lepidopteran densities. Had I not accounted for plant quality in my analysis I would have overemphasized the importance of the concentration of plants (Table 2.10) and erroneously concluded that A californica was more abundant in monocultures and M. configurata was more abundant in dicultures. In many studies, whether it is quantified or not, some aspect of plant quality varies with vegetational diversity. In studies where this is not accounted for (eg. Risch 1980, 1981, Risch et al. 1983, Letourneau 1987), it is impossible to determine whether herbivores are responding to a more preferred quality of plant in monocultures or to the concentration of host plants. The key to separating effects of plant quality from resource concentration involves controlling for plant quality in plant treatments. I did this by strip-cropping, thereby preventing interspecific competition between the primary crop and the intercrop. Other practises, such as fertilizing an intercrop more heavily than a monoculture might have a similar effect. Since one purpose of this line of research is to determine whether the modern practise of planting large monocultures will always result in high herbivore loads, 130 clarifying the effects of plant quality is crucial. If, as my research suggests, the quality of the resource is more important than its concentration, it would be reasonable to consider alterations in plants within monocultures to make them less attractive to herbivores. Breeding for yield enhancement alone, and hence larger plants, may be more responsible for the high herbivore loads found in monocultures than the concentration of one species of plant. If temperate-climate agriculture is to continue to promote monoculture plantings then plant breeding should consider more than yield enhancement and disease management. Admittedly, including herbivore deterrence in plant breeding programs is not a simple task, as different herbivores may respond differently to the same aspect of plant quality. For example, in this study the correlation between plant width and abundance was positive for A californica but negative for F\ xylostella and configurata. Manipulating plant quality is not a matter of determining herbivore preferences alone. Parasitoids can respond to host plant quality, independent of their insect hosts, as occurred in this study. The current fascination with the potential for the use of biotechnology in pest management overlooks the reality that pest management is not a simple matter of altering one gene in a plant. Interactions between the quality of a primary resource and consumers at several trophic levels means that pest management must always consider more than the simple relationship between a plant and a herbivore. Natural enemies can exert a significant impact on herbivore abundance, both directly, through predation on the herbivore, and indirectly, probably through predation on parasitised herbivores. Enoplagnatha ovata appeared to have a regulatory effect on parasitism of F\ xylostella by D. insulare. However, both the higher populations of the host in the second generation, and the higher densities of the spider in augmentation plots were necessary for this important relationship to become apparent. The 131 probable 'switching' of E. ovata to P. xylostella as a prey item at higher densities, as well as preferential predation on parasitised hosts could account for this result. An interesting aspect of my results was the finding that responses to the addition of the generalist predator, E ovata. could depend on the planting type. For example, E ovata augmentations were associated with decreases in P\ xylostella and M. configurata densities in monocultures and increases in densities in dicultures. The importance of predation may have been overlooked in previous experiments for several reasons. 1. Most studies have been done on highly mobile herbivores (i.e. chrysomelid beetles) who readily leave host patches (Root 1973, Bach 1980a,b, 1981, 1984,1986, Risch et al. 1983) and have few predators. Predation is likely more important to lepidopterans, whose adults are mobile but whose offspring are relatively sedentary. The lepidopterans studied in this thesis were all readily preyed upon both by predators and parasitoids. 2. Predators are often highly mobile, and elusive. As a result, obtaining reliable estimates of predator abundance is difficult. The approach taken here, in which predator abundance was experimentally manipulated, eliminated a dependence on measuring natural predator abundance. 3. The impact of predators can differ in the initial colonization stage, when new plant associations are first set-up, from that in a well established ecosystem. In my study, E ovata densities in the augmentation plots were similar to densities in a nearby five-year old strawberry patch. Therefore, although this experiment was conducted for only two years, I was able to simulate predator abundance at a temporal scale past the initial colonization stage. The effects of plant quality on generalist predators could not be directly determined from this study. In a recent experiment addressing this question, no differences in spider or ground beetle abundance were found as a result of fertilization 132 treatments in roadside habitats (Snodgrass and Stadilbacker 1989). More work of this kind is necessary to clarify the role of plant quality in determining the effectiveness of predators. In particular, the way factors such as plant size affect the success of searching predators should be elucidated. The importance of vegetational diversity, aside from effects due to the predators, was confirmed, but also qualified. In plots without spider augmentations, which would be equivalent to experiments conducted by others (egs. Root 1973, Letourneau and Altieri 1983, Bach 1980a,b, 1981, Andow and Risch 1987, Risch 1980, Letourneau 1987), I indeed found higher densities of second to third instar P\ xylostella (in July) and second instar ML configurata. in monoculture plots. However, the difference disappeared in both cases with the addition of spiders, supporting my contention that when predators are studied at densities reached after the initial colonization phase, the importance of generalists predators over resource concentration becomes apparent. As Vandermeer (1989) points out, Root's original hypotheses were never meant to be mutually exclusive. In the effort to find simple, single factor explanations for a phenomenon the importance of more than one factor, along with possible interactions among factors, may have been overlooked. By studying four lepidopterans concurrently in the same system I had expected to uncover generalizations concerning life style potentially accounting for different responses to both vegetational diversity and predation. In this aspect I was not successful. F\ xylostella. a crucifer specialist, and JVL configurata. a polyphagous herbivore, revealed close resemblance to each other in their specific responses; closer, in fact than compared to species with similar feeding habits. P\ rapae was not affected by any of my treatments. Although P. xylostella. A, californica and M^ configurata. responded to experimental treatments even at low densities, it is likely that for P. rapae higher densities are required to elicit a response. A final problem was that only two of 133 the species, P\ xylostella and A californica were parasitised. Although predators may prefer polyphagous herbivores as prey over specialized herbivores (Bernays 1988, Bernays and Cornelius 1989), interactions between plants, predators and parasitoids add a dimension to herbivore population dynamics that might override effects due to the range of host plants included in the diet. Some points of caution in interpreting these data are in order. The appropriateness of spatial and temporal scale in this study are not known. As far as spatial scale is concerned, the appropriate scale on which to do an experiment of this sort depends on the species of insect being studied. Some insects, especially those with limited mobility, are easily studied in relatively small plots, such as the ones chosen for this study. However, from the point of view of a highly mobile insect, the sum of all small plots may merely represent a single heterogenous field. It could be argued that the scale was too small for the colonizing stage of the lepidopterans to recognize the plots as distinct choices (and hence no differences in colonization were found due to planting type) but large enough for differences within plots due to predation to be noticeable. Positive results always indicate that the scale was sufficiently large for the species under consideration. Unfortunately, negative results always leave open the possibility of scale problems. Criticisms due to temporal scale do not seem as crucial. Although this was a one season study, the cropping system was set up the previous year. Initial colonization was, therefore, not really the focus of the study. Also, the one species that had several generations in a season FL xylostella. was studied during two of those generations, and, interestingly, the most significant effects did not surface until the second generation. The method of intercropping can also be important. For example, more 134 cabbage root flies (Delia brassicae (Wied)) eggs are found on brussels sprouts intercropped with clover, than on brussels sprouts in monoculture, if the clover is planted 20 cm from the brussels sprouts. This effect declines as the distance between the clover and brussels sprouts exceeds 50 cm (Tukahirwa and Coaker 1982). In this study the distance between strips was 70 cm, which may have dampened the potential effects of the intercropping on herbivore densities. In conclusion, several points raised in this thesis challenge the widely held belief that predation is of secondary importance to resource concentration in determining differences in herbivore abundance between monocultures and polycultures. Host plant quality was of overriding importance in determining the abundance of lepidopteran larvae. Generalist predators too can be important determinants of herbivore abundance, particularly through predation on parasitised individuals. LITERATURE CITED Agnew C.W. and J.W. Smith. 1989. Ecology of spiders (Araneae) in a peanut agroecosystem. Environ. Entomol. 18: 30-42. Altieri M.A. and D.K. Letourneau. 1982. Vegetation management and biological control in agroecosystems. Crop Prot. 1:405-430. Altieri M.A. and M. Liebman. 1986. Insect, weed and plant disease management in multiple cropping systems. In: Multiple Cropping Systems C A . Francis (ed.), Macmillan Publishing Co., New York.pp. 183-218. Altieri M.A, W.J. Lewis, DA. Nordlund, R.C. Gueldner and J.W. Todd. 1981. Chemical interactions between plants and Trichogramma wasps in Georgia soybean fields. Prot. Ecol. 3: 259-263. Andow DA. 1984. Microsite of the green rice leafhopper, Nephotettix cincticeps (Homoptera: Qicadellidae), on rice: plant nitrogen and leafhopper density. Res. Popul. Ecol. 26:313-329. Andow DA. and S.J. Risch. 1987. Parasitism in diversified agroecosystems: phenology of Trichogramma minutum (Hymenoptera: Trichogrammatidae) Entomophaga 32: 255-260. Atsatt P.R. and D.J. O'Dowd. 1976. Plant defense guilds. Science 193:24-29. Bach C.E. 1980a. Effects of plant diversity and time of colonization on an herbivore-plant interaction. Oecoloaia 44: 319-326. Bach C.E. 1980b. Effects of plant diversity and density on the population dynamics of a specialist herbivore, the striped cucumber beetle, Acalymma vittata (Fab.). Ecology 61: 1515-1530. Bach C.E. 1981. Host plant growth form and diversity: effects on abundance and feeding preference of a specialist herbivore, Acalymma vittata (Coleoptera: Chrysomelidae). Oecologia 50: 370-375. Bach C.E. 1984. Plant spatial pattern and herbivore popoulation dynamics: plant factors affecting the movement patterns of a tropical cucurbit specialist Acalymma innubum. Ecology 65:175-190. Bach C.E. 1986. A comparison of the responses of two tropical specialist herbivores to host plant patch size. Oecologia 68: 580-584. 136 Bach C.E. 1988a. Effects of host plant patch size on herbivore density: patterns. Ecology 69:1090-1102. Bach C.E. 1988b. Effects of host plant patch size on herbivore density: underlying mechanisms. Ecology 69:1103-1117. Beddington J.R., C.A. Free and J.H. Lawton. 1978. Modelling biological control: on the characteristics of successful natural enemies. Nature (London) 273:513-519. Beirne B.P. 1971. Pest insects of annual crop plants in Canada. I. Lepidoptera, II. Diptera, III. Coleoptera. Mem. Ent. Soc. Canada 78:1-124. Bernays E.A. 1988. Host specificity in phytophagous insects: selection pressure from generalist predators. Entomol. exp. appl. 49:131-140. Bernays E.A. and M.L Cornelius. 1989. Generalist caterpillar prey are more palatable than specialists for the generalist predator Iridomyrmex humilis. Oecolooia 79:427-430. Brubaker R.W. 1968. Seasonal occurrence of Voria ruralis. a parasite of the cabbage looper, in Arizona, and its behavior and development in laboratory culture. J. Econ. Entomol. 61: 306-309. Butler G.D. 1958. Tachinid flies reared from lepidopterous larvae in Arizona. J. Econ. Entomol. 51:561-562. Causton D.R. and J.C. Venus. 1981. The Biometry of Plant Growth. Edward Arnold, London. pp16-36. Chesson P.L. and W.W. Murdoch. 1986. Aggregation of risk: relationships between host-parasitoid models. Am. Nat. 127:696-715. Chew F.S. 1975. Coevolution of Pierid butterflies and their cruciferous foodplants. Oecolooia 20:117-127. Courtney S.P. 1981. Coevolution of Pierid butterflies and their cruciferous foodplants. III. Anthocharis cardamines survival, development and oviposition. Oecologia 51:91 -96. Courtney S.P. 1985. Apparency in coevolving relationships. Oikos 44: 91-98. Cromartie W.J. 1975. The effect of stand size and vegetational background on the colonization of cruciferous plants by herbivorous insects. J. Appl. Ecol. 12: 517-533. Cromartie W.J. 1981. The environmental control of insects using crop diversity. In: Handbook of Pest Management. D. Pimentel (ed.) C.R.C.Press, Boca Raton, F L , USA, pp. 223-251. Crook G., P. Vezina and Y. Hardy. 1979. Susceptibility of balsam fir to spruce budworm defoliation as affected by thinning. Can. J. For. Res. 9: 428-435. Damman H. 1987. Leaf quality and enemy avoidanced by the larvae of a pyralid moth. Ecology 68: 88-97. Dempster J.P. 1969. Some effects of weed control on the numbers of the small cabbage white Pieris rapae on Brussels sprouts. J. Appl. Ecol. 6:339-345. Dondale C D . , B. Parent and D. Pitre. 1979. A six-year study of spiders (Araneae) in a Quebec apple orchard. Can. Ent. 111:377-380. Elmstrom K.M., DA. Andow and W.W. Barclay. 1988. Flea beetle movement in a broccoli monoculture and diculture. Environ. Entomol. 17:299-305. Elsey K.D. and J.F. Chaplin. 1978. Resistance of tobacco introduction 1112 to the tobacco budworm and green peach aphid. J^ Econ. Entomol. 71: 732-725. Elsey K.D. and R.L.Rabb. 1970. Biology of Voria ruralis (Diptera: Tachinidae). Ann. Entoml. Soc. Amer. 63:216-222. Feeny P.P. 1976. Plant apparency and chemical defense. Rec. Adv. Phytochem. 10:1-40. Forsberg J. 1987. Size discrimination among conspecific hostplants in two pierid butterflies; Pieris napi L. and Pontia daplidice L. Oecologia 72: 52-57. Francis C.A. 1986. (ed.) Multiple Cropping Systems. Macmillan Publishing Co., New York. 218pp. Francis C.A., C A . Flor and M. Prager. 1978. Multiple cropping potentials of beans and maize. HortScience 13:12-17. Francis C.A., C A . Flor and M. Prager. 1978. Effects of bean association on yields and yield components of maize. Crop Sci. 18:760-764. Frazer B.D. and R. van den Bosch. 1973. Biological conttrol of the walnut aphid in California: the interrelationship of the aphid and its parasite. Environ. Entomol. 2: 561-568. Gilbert N. 1984. Control of fecundity in Pieris rapae L I. The problem. II. Differential effects of temperature. III. Synthesis. J. Anim. Ecol. 53:581-607. Grant J.F. and M. Shepard. 1983. Biolgocial characteristics of a South American population of Voria ruralis (Diptera: Tachinidae), a larval parasitoid of the soybean looper (Lepidoptera: Noctuidae). Environ. Entomol. 12: 1673-1677. Harcourt D.G. 1957. Biology of the diamondback moth. Plutella maculipennis (Curt.) (Lepidoptera: Plutellidae), in Eastern Ontario. II. Life-history, behaviour, and host relationships. Can. Ent. 79: 554-564. Harcourt D.G. 1960. Distribution of the immature stages of the diamondback moth. Plutella maculipennis (Curt.) (Lepidoptera: Plutellidae), on cabbage. Can. Ent. 92: 517-521. Harcourt D.G. 1961. Spatial pattern of of the imported cabbage worm R rapae (Lep. Pieridae) on cultivated Cruciferae. Can. Ent. 93:945-952. Harcourt D.G. 1963. Biology of cabbage caterpillars in eastern Ontario. Proc. Ent. Soc. Ont. 93:61-75. Harcourt D.G. 1966. Major factors in survival of the immature stages of R rapae. Can. Ent. 98:653-662. Harcourt D.G. 1986. Population dynamics of the Diamondback Moth in Southern Ontario. In: Diamondback Moth Management. N.S. Talekar and T.D. Griggs (eds.), Proceedings of the first international workshop, Tainan, Taiwan, 11-15 March 1985. Asian vegetable research and development center. AVRDC Publication No.86-248. pp 3-15. Harper J.L. 1977. Population Ecology of PJants. Academic Press, London, New York, 892p. Harrison S. 1987. Treefall gaps versus forest understory as environments for a defoliating moth on a tropical forest shrub. Oecologia 72: 65-68. Hassell M.P. and T.R.E. Southwood. 1978. Foraging strategies of insects. Ann. Rev. Ecol. Syst. 9:75-98. Hill M.G. 1988. Analysis of the biological control of Mythimna separata (Lepidoptera: Noctuidae) by Apanteles ruficrus (Braconidae: Hymenoptera) in New Zealand. J. Appl. Ecol. 25:197-208. Hunt R. 1982. Plant Growth Curves. The functional approach to plant growth analysis. University Park Press, Baltimore. 242p. 139 Jones R.E. 1977a. Search behaviour: a study of three caterpillar species. Behaviour 60: 237-259. Jones R.E. 1977b. Movement patterns and egg distribution in cabbage butterflies. J. Anim. Ecol. 46:195-212. Jones R.E. 1987. Ants, parasitoids, and the cabbage butterfly Pieris rapae. J. Anim. Ecol. 56: 739-749. Jones R.E. and P.M. Ives. 1979. The adaptiveness of searching and host selection behaviour in Pieris rapae L. Aust. J. Ecol. 4: 75-86. Jones R.E., N. Gilbert, M. Guppy and V. Nealis. 1980. Long-distance movement of Pieris rape. J. Anim. Ecol. 49:629-642. Jones R.E., V.G. Nealis, P.M. Ives and E. Scheermeyer. 1987. Seasonal and spatial variation in juvenile survival of the cabbage butterly Pieris rapae: evidence for patchy density-dependence. J. Anim. Ecol. 56. 723-737. Kareiva P. 1982a. Experimental and mathematical analyses of herbivore movement: quantifying the influence of plant spacing and quality on foraging discrimination. Ecol. Monoar. 52, 261-282. Kareiva P. 1982b. Exclusion experiments and the competitive release of insects feeding on collards. Ecology 63: 696-704. Kareiva P. 1983. Influence of vegetation texture on herbivore populations: resource concentration and herbivore movement. In: Variable Plants and Herbivores in Natural and Managed Systems. R.F. Denno and M.S. McClure (eds.). Academic Press, New York. pp. 259-289. Kareiva P. 1985. Finding and losing host plants by Phyllotreta: patch size and surrounding habitat. Ecology 66:1809-1816. Kareiva P. 1986. Trivial movement and foraging by crop colonizers. In: Ecological Theory and Integrated Pest Management Practice. M. Kogan (Ed.). John Wiley & Sons, New York. pp. 59-82. Kemp W.P. and G.A. Simmons. 1979. Influence of stand factors on survival of early instar spruce budworm. Environ. Entomol. 8: 993-996. Kiritani K., S. Kawahara, T. Sasaba and F. Nakasuji. 1972. Quantitative evaluation of predation by spiders on the green rice leaf hopper, Nephotettix cincticeps Uhler, by a sight-count method. Res. Popul. Ecol. 13:187-200. 140 Lessells C M . 1985. Parasitoid foraging: Should parasitism be density dependent? J. Anim. Ecol. 54: 27-41. Letourneau D.K. 1987. The enemies hypothesis: Tritrophic interactions and vegetational diversity in a tropical agroecosystem. Ecology 68:1616-1622. Letourneau D.K. and M.A. Altieri. 1983. Abundance patterns of a predator, Orius tristicolor (Hemiptera: Anthocoridae), and its prey, Frankliniella occidentalis (Thysanura: Thripidae): habitat attraction in polycultures versus monocultures. Environ. Entomol. 12:1464-1469. Lim G.-S. 1986. Biological control of diamondback moth. In: Diamondback Moth Management. Proceedings of the First International Workshop, Tainan, Taiwan, March 1985. Talekar N.S. and T.D. Griggs (Eds.). Asian Vegetable Research and Development Center. AVRDC Publication No.86-248. pp. 159-171. MacGarvin M. 1982. Species-area relationships of insects on host plants: herbivores on rosebay willowherb. J. Anim. Ecol. 51: 207-223. Mansour F., D. Rosen, A. Shulov and H.N. Plaut. 1980. Evaluation of spiders as biological control agents of Spodoptera littoralis laravae on apple in Israel. Ecol. Applic. 1:225-232. Martin P.B., P.D. Lingren, G.L Greene and E.E. Grissell. 1981. The parasitoid complex of three Noctuids (Lep.) in a northern Florida cropping system: seasonal occurrence, parasitization, alternate hosts, and influence of host-habitat. Entomophaga 26: 401-419. Mattson W.J. 1980. Herbivory in relation to plant nitrogen content. Ann. Rev. Ecol. Syst. 11:119-161. McKinney K.B. 1944. Cabbage looper as a pest of lettuce in the Southwest. USDA Tech. Bull. 84, 30p. Metcalf and Flint 1962. Destructive and Useful Insects. Tata McGraw-Hill Publ. Co., New Dehli. pp. 666-667 Milne L. and M. Milne. 1980. The Audubon Society Field Guide to North American Insects and Spiders. Alfred A. Knopf, New York. pp. 797. Montieth L G . 1960. Influence of plants other than the food plants of their host in host-finding by tachinid parasites. Can. Ent. 92:641-652. 141 Morrison G. and D.R. Strong. 1980. Spatial variations in host density and the intensity of parasitism: some empirical examples. Environ. Entomol. 9:149-152. Morrison G. and D.R. Strong. 1981. Spatial variations in egg density and the intensity of parasitism in a neotropical chrysomelid (Cephaloleia consanguinea). Ecol. Entomol. 6:55-61. Murdoch W.W. and J.D. Reeve. 1987. Aggregation by parasitoids and the detection of density dependence in field populations. Oikos 50:137-141. Murdoch W.W., J. Chesson and P.L Chesson. 1985. Biological control in theory and practice. Am. Nat. 125:344-366. Myers J.H. 1985. Effect of physiological condition of the host plant on the ovipositional choice of the cabbage white butterfly Pieris rapae. J. Anim. Ecol. 54: 193-204. Myers J.H., J. Monro and N. Murray. 1981. Egg clumping, host plant selection and population regulation in Cactoblastis cactorum (Lepidoptera). Oecologia 51:7-13. Nealis V.G. 1983. Development, diapause and seasonal ecology of the insect parasite, Apanteles rubecula (Hymenoptera: Braconidae). Ph.D. Dissertation. University of British Columbia, Canada, 126pp. Need J.T. and P.P. Burbutis. 1979. Searching efficiency of Trichogramma nubilale. Environ. Entomol. 8:224-227. Nyffeler M. and G. Benz. 1987. Spiders in natural pest control: a review. J. Appl. Ecol. 103:321-335. Nyffeler M., D.A. Dean, and W.L. Sterling. 1989. Prey selection and predatory importance of orb-weaving spiders (Araneae: Araneidae, Uloboridae) in Texas cotton. Environ. Entomol. 18:373-380. Oatman E.R. 1966. An ecological study of cabbage looper and imported cabbageworm populations on cruciferous crops in southern California. J. Econ. Entomol. 59: 1134-1139. Oatman E.R. and G.R. Platner. 1972. An ecological study of lepidopterous pests affecting lettuce in coastal Southern California. Environ. Entomol. 1: 202-204. Oatman E.R., G.R. Platner, J.A. Wyman, R.A. van Steenwyk, M.W. Johnson and H.W. Browning. 1983. Parasitization of lepidopterous pests on fresh market tomatoes in Southern California. J. Econ. Entomol. 76: 452-455. 142 Onuf C P . 1978. Nutritive value as a factor in plant-insect interactions with an emphasis on field studies. In: The Ecology of Arboreal Folivores. Montgomery G.G. (ed.). Smithsonian Press, Washington, D.C. pp. 85-96. Osisanya E.0.1970. Effect of shade on the rate of infestation of Triplochiton scleroxvlon by Dichlidophlebia species. Entomol. exp. appl. 13:125-132. Parker F.D. 1970. Seasonal mortality and survival of Pieris rapae (Lepidoptera: Pieridae) in Missouri and the effects of introducing an egg parasite, Trichoaramma evanscens. Ann. Entomol. Soc. Amer. 63:985-994. Parker F.D., F.R. Lawson and R.E. Pinnell. 1971. Suppression of P. rapae using a new control system: mass releases of both pest and its parasites. J. Econ. Entomol. 64:721-735. Perfecto I. B. Horwith, J. Vandermeer, B. Schultz, H. McGuinness and A. dos Santos. 1986. Effects of plant diversity and density on the emigration rate of two ground beetles, Harpalus pennsylvanicus and Evarthrus sodalis (Coleoptera: Carabidae) in a sytem of tomatoes and beans. Environ. Entomol. 15:1028-1031. Pimentel D. 1961. The influence of plant spatial patterns on insect populations. Ann. Entomol. Soc. Amer. 54:61-69. Price P.W. 1975. Insect Ecology. John Wiley and Sons, New York, N.Y. 514p. Price P.W., C.E. Bouton, P. Gross, B.A. McPheron, J.H. Thompson and A.E. Weis. 1980. Interactions among three trophic levels: influence of plants on interactions between insect herbivores and natural enemies. Ann. Rev. Ecol. Syst. 11:41-65. Puttarudriah M. 1953. The natural control of the alfalfa looper in Central California. J. Econ. Entomol. 56: 723. Ralph C P . 1977. Effect of host plant density on populations of a specialized, seed-sucking bug, Oncopeltus fasciatus. Ecology 58: 799-809. Rausher M. 1983. Ecology of host-selection behavior in phytophagous insects. In: Variable Plants and Herbivores jn Natural and Managed Systems. Denno R.F. and M.S. McClure (eds.). Academic Press, New York. pp. 223-257. Renwick J.A.A. 1989. Chemical ecology of oviposition in phytophagous insects. Experientia 45:223-228. Richards O.W. 1940. The biology of the small white butterfly (Pieris rapae) with special reference to the factors controlling its abundance. J. Anim. Ecol. 9:243-282. 143 Riotte L. 1975. Secrets of companion planting for successful gardening. Garden Way Publishing. Charlotte, Vermont. 226p. Risch S.J. 1980. The population dynamics of several herbivorous beetles in a tropical agroecosystem: the effect of intercropping corn, beans and squash in Costa Rica. J. Appl. Ecol. 17:593-612. Risch S.J. 1981. Insect herbivore abundance in tropical monocultures and polycultures: an experimental test of two hypotheses. Ecology 62:1325-1340. Risch S.J., D. Andow and M.A. Altieri. 1983. Agroecosystem diversity and pest control: data, tentative conclusions and new directions. Environ. Entomol. 12:625-629. Root R.B. 1973. Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards, Brassica oleracea. Ecol. Monogr. 43: 95-124. Root R.B. and J.O. Tahvanainen. 1969. Role of winter cress Barbarea vulgaris as a temporal host in the seasonal development of the crucifer fauna. Ann. Entomol. Soc. Amer. 62:852-855. Russell E.P. 1989. Enemies hypothesis: A review of the effect of vegetational diversity on predatory insects and parasitoids. Environ. Entomol. 18: 590-599. Schultz J.C. 1988. Many factors influence the evolution of herbivore diets, but plant chemistry is central. Ecology 69: 896-897. Scriber M. 1984a. Host-plant suitability. In: Chemical Ecology of Insects. Bell W.J. and R.T.Carde (eds.), Chapman and Hall, London, pp. 159-202. Scriber M. 1984b. Nitrogen nutrition of plants and insect invasion. In: Nitrogen in Crop Production. Hauck R.D. (ed.), American Society of Agronomy, Crop Science Society of American, Soil Science Society of America, Madison, Wisconsin USA. pp441-460. Seligy V . L 1971. Postembryonic development of the spider Enoplognatha ovata (Clerck) (Araneae: Theridiidae). Zool. J. Linn. Soc. 50:21-31. Sheehan W. 1986. Response by specialist and generalist natural enemies to agroecosystem diversification: selective review. Environ. Entomol. 15:456-561. Smith J.G. 1969. Some effects of crop background on populations of aphids and their natural enemies on Brussels sprouts. Ann. Appl. Biol. 63:326-30. Smith J.G. 1976. Influence of crop background on natural enemies of aphids on Brussels sprouts. Ann. Appl. Biol. 83:15-29. Snodgrass G.L and E.A. Stadilbacker. 1989. Effect of different grass and legume combinations on spider (Araneae) and ground beetle (Coleoptera: Carabidae) populations in roadside habitats in the Mississippi delta. Environ. Entomol. 24:370-378. Sokal F.J. and R.R. Rohlf. 1983. Biometry. 2nd. ed. W.H. Freeman and Co., New York, N.Y. USA, 859p. Solomon B.P. 1981. Response of a host-specific herbivore to resource density, relative abundance, and phenology. Ecology 62:1205-1214. Southwood T.R.E. 1973. The insect/plant relationship - an evolutionary perspective. In: Insect/Plant Relationships, van Emden H.F. (ed.), The Royal Entomological Society of London Symposium No. 6, Blackwell Scientific, London, England, pp. 3-30. Tahvanainen J.O. and R.B. Root. 1972. The influence of vegetational diversity on the population ecology of a specialized herbivore, Phylotreta cruciferae (Coleoptera: Chrysomelidae). Oecologia 10:321-346. Thompson J.N. and P.W. Price. 1977. Plant plasticity, phenology, and herbivore dispersion: wild parsnip and the parsnip webworm. Ecology 58:1112-1119. Tostowaryk W. 1972. Relationship between predation and parasitism of diprionid sawflies. Ann. Entomol. Soc. Amer. 64: 1424-1427. Tukarhiwa E.M. and T.H. Coaker. 1982. Effect of mixed cropping on some insect pepsts of brassicas; reduced Brevicoryne brassicae infestations and influences in epigeal predators and the disturbance of oviposition behavious in Delia brassicae. Entomol. exp. appl. 32:129-140. Turnock W.J. 1984. Mamestra configurata Walker, bertha armyworm (Lepidoptera: Noctuidae). In: Biological Control Programmes against Insects and Weeds in Canada 1969-1980. Kelleher J.S. and M.A. Hulme (eds.), Commonwealth Agr. Bur. London, U.K. pp. 49-55. Turnock W.J. 1988. Density, parasitism, and disease incidence of larvae of the bertha armyworm, Mamestra configurata (Lepidoptera: Noctuidae), in Manitoba, 1973-1986. Can. Ent. 120: 401-413. Turnock W.J. and H.G. Philip. 1977. The outbreak of bertha armyworm Mamestra configurata (Noctuidae: Lepidoptera) in Alberta 1971 to 1975. Manitoba Entomologist 11:10-21. 145 Vail P.V. 1981. Cabbage looper nuclear polyhedrosis virus-parasitoid interactions. Environ. Entomol. 10:517-520. Vandermeer J. 1989. The Ecology of Intercropping. Cambridge Univ. Press. Cambridge. p237. van Driesche R.G. 1983. Meaning of \"percent parasitism\" in studies of insect parasitoids. Environ. Entomol. 12:1611-1622. van Huis A. 1981. Integrated pest management in the small farmer's maize crop in Nicaragua. Meded. Landbouww. 81: [1-6]. Vince S.W., I. Valiela and J.M. Teal. 1981. An experimental study of the structure of herbivorous insect communities in a salt marsh. Ecology 62: 1662-1678. Vinson S.B. 1976. Host selection by insect parasitoids. Ann. Rev. Entomol. 21:109-133. Vinson S.B. 1984. The parasitoid-host relationship. In: Chemical Ecology of Insects. Bell W.J. and R.T. Carde (eds.), Chapman and Hall, London, pp. 205-233. Waage J.K. 1983. Aggregation in field parasitoid populations: foraging time allocation by a population of Diadegma (Hymenoptera, Ichneumonidae). Ecol. Entomol. 8:447-453. White T.C.R. 1978. The importance of relative shortage of food in animal ecology. Oecologia 33:71-86. White T.C.R. 1984. The abundance of invertebrate herbivores in relation to the availability of nitrogen in stressed food plants. Oecologia 63: 90-105. Wolfson J.L. 1980. Oviposition response of Pieris rapae to environmentally induced variation in Brassica nigra. Entomol. exp. appl. 27: 223-232. Wylie H.G. and G.E. Bucher G.E. 1977. The bertha armyworm. Mamestra configurata (Lepidoptera: Noctuidae). Mortality of immature stages on the rape crop, 1972-1975. Can. Ent. 109: 823-837. "@en . "Thesis/Dissertation"@en . "10.14288/1.0100450"@en . "eng"@en . "Plant Science"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Plant size, resource concentration and natural enemies : a comparison of four herbivores in monocultures of brussels sprouts and dicultures of brussels sprouts/peppermint"@en . "Text"@en . "http://hdl.handle.net/2429/30791"@en .