Leguminous Herbs and their Herbivores: Interaction during Early Secondary Succession

by

Ian Michael Evans

A thesis for the degree of Doctor of Philosophy of the University of London and for the Diploma of Membership of Imperial College.

Department of Pure and Applied Biology Imperial College at Silwood Park Silwood Park Ascot December 1988 Berkshire SL5 7PY Abstract

The thesis describes leguminous herbs and their associated insect herbivores in natural plant communities, characteristic of different stages in early plant succession. The studies were undertaken on a series of experimental sites of known successional age at Silwood Park, Berkshire, U.K. Ecological trends in plant taxonomic composition, cover, structure, distribution and diversity of legumes, as well as the contribution of each legume species to the overall plant community are described and compared along the early successional gradient. The insect herbivores associated with individual species of legume were assessed by Univac sampling and the herbivores and their host plants were investigated using an ordination method, Detrended Correspondence Analysis. By using individually marked plants of different species of legume in control and insecticide-treated plots, the effects of insect herbivory are examined. In addition, the cyanogenic and acyanogenic phenotypes of L. are investigated in the same way. These manipulative field experiments have been carried out at different stages of early succession according to the abundance of the legumes. Differences in vegetative and reproductive characteristics, leaf persistence and leaf damage by chewing are described and discussed in terms of plant/insect herbivore interaction theory. Laboratory-based experiments using adult weevils are included to demonstrate host plant preferences and investigate frequency-dependent grazing in one cyanogenic and three acyanogenic phenotypes of T. repens by adult Sitona lineatus (L.). The effects of cyanogenic/acyanogenic phenotypes of T. repens on the reproduction and survival of Acyrthosiphon pisum (Harris) are also investigated experimentally. Acknowledgements

I would like to thank my supervisor Dr. V.K. Brown for her help and encouragement during the course of this study. Statistical advice was kindly given by Dr. M.J. Crawley, Dr. C.W.D. Gibson, Dr. A.R. Ludlow and Dr. A.J. Morton, however, much of the help with all statistical problems must be credited to Mark "Remedial" Rees. Dr. A.C. Gange, M. Jepsen and A.L. Storr are gratefully acknowledged for their help with field work and also the technical staff at Silwood Park for their help during the course of this study. Discussion of ideas were held with many of the academic staff and students at Silwood Park, particularly Dr. A.C. Gange and Dr. S. McNeill. Additionally, many stimulating conversations were held with Dr. G. Edwards-Jones and A. "Wacky" Wilcox. I am also very grateful to those people who have identified specimens or provided insects and other materials necessary for my experiments; they are all acknowledged individually in the relevant sections. Finally, I would like to express my gratitude to Professor M.P. Hassell for allowing me the use of the facilities at Silwood Park. Table of Contents

Page Abstract 2 Acknowledgements 3 Table of Contents 4 List of Tables 8 List of Figures 12 Chapter 1: Insect-Plant Relationships: A Community Approach 1.1. Introduction 16 1.2. Theory of Plant Succession 18 1.3. Plant-Plant Relationships 20 1.4. Plant-Insect Relationships 21 1.4.1. Plant Architecture 21 1.4.2. Species-Area Relationships 21 1.4.3. Resource Concentration Hypothesis 22 1.4.4. Plant and Insect Life-History Strategies 22 1.5. Insect-Plant Interactions 24 1.5.1. Herbivory 24 1.5.2. Plant Defences 25 Chapter 2: Composition, Pattern and Structure of the Legume Community 2.1. Introduction 28 2.2. Materials and Methods 30 2.2.1. Site Description 30 2.2.2. Plant Recording Techniques 31 2.3. Analysis of Point Quadrat Data 31 2.3.1. Percentage Cover 31 2.3.2. Plant Structure 32 2.3.3. Grass/Forb Index 32 2.3.4. Legume/Forb Index 32 2.3.5. Pattern of Distribution 33 2.3.6. Taxonomic and Spatial Diversity 33 2.3.7. Importance Values 34 2.4. Results: General Trends 35 2.4.1. Cover 35 2.4.2. Species Richness and Composition 35 2.4.3. Grass/Forb and Legume/Forb Indices 39 2.4.4. Plant Structure 39 2.4.5. Pattern of Distribution 43 2.4.6. Taxonomic and Spatial Diversity 48 2.5. Trends in Individual Species 52 2.5.1. 52 2.5.2. Lotus uliginosus 52 2.5.3. 52 2.5.4. Medicago sativa 58 2.5.5. Trifolium dubium 58 2.5.6. Trifolium hybridum 58 2.5.7. 59 2.5.8. Trifolium repens 59 2.5.9. Vida hirsuta 62 2.5.10. Vida sativa 62 2.5.11. Vida tetrasperma 65 2.6. Discussion 65 Chapter 3: Insect Herbivore Survey 3.1. Introduction 72 3.2. Materials and Methods 73 3.2.1. Insect Sampling 73 3.2.2. Insect Sorting 74 3.2.3. Insect Identification 74 3.3. Ordination 74 3.4. Results 76 3.4.1. General Trends 77 3.4.2. Changes in Insect Abundance and Species Composition 80 within different Insect Groups 3.4.3. Multivariate Analysis 90 3.5. Discussion 93 Chapter 4: Insect Herbivory: its Effects on Plant Reproductive and Vegetative Characteristics 4.1. Introduction 96 4.2. Materials and Methods 98 4.3. Results 102 4.3.1. Effects of Insect Herbivory on Plant Height and Leaf Number 102 4.3.2. Leaf Life Expectancy and Accumulated Damage by Insects 106 4.3.3. Effects of Insect Herbivory on Reproductive Characteristics 109 4.3.4. Successional Trends 118 4.4. Discussion 118 Chapter 5: Cyanogenesis and its Effect on Insect Herbivory in T. repens 5.1. Introduction 124 5.2. Materials and Methods 128 5.2.1. Detecting Cyanide using the Guignard Picrate Paper Test 128 5.2.2. Indentifying each Acyanogenic Phenotype 129 5.2.3. Survey of Cyanogenesis 130 5.2.4. Monitoring the Performance of Cyanogenic and 131 Acyanogenic T. repens 5.3. Results 133 5.3.1. Frequency of Cyanogenesis in Succession 133 5.3.2. Effects of Cyanogenesis on Vegetative Characteristics 135 5.3.2. (z). Stolon Survival 135 5.3.2. (z7). Stolon Height and Leaf Number 138 5.3.2. (z77). Leaf Life Expectancy and Insect Damage 138 5.3.3. Effects of Cyanogenesis on Reproductive Characteristics 148 5.3.4. Successional Trends 154 5.4. Discussion 158 Chapter 6: The Effect of the Plant on Preference, Fecundity and Survival of the Insect Herbivore 6.1. Introduction 164 6.2. Materials and Methods 166 6.2.1. Stock Plants 166 6.2.2. Choice Experiments 166 6.2.3. A Test of Frequency-Dependent Grazing by S. lineatus on 170 Cyanogenic and Acyanogenic T. repens 6.2.4. A Study of Fecundity and Longevity in Acyrthosiphon pisum 172 Reared on Cyanogenic and Acyanogenic T. repens 6.3. Results 174 6.3.1. Choice Experiments 174 6.3.2. Frequency-Dependent Grazing 177 6.3.3. Fecundity and Longevity in A. pisum Reared on Cyanogenic 181 and Acyanogenic T. repens 6.4. Discussion 187 Chapter 7: General Discussion 192 References 197 Appendix 1: Importance Values of Each Plant Species Recorded in Sites of 219 Different Successional Age Appendix 2: British Species of Phytophagous Insects Recorded as Occurring on 224 Leguminosae (particularly L. corniculatus, M. lupulina, T. pratense, T. repens, V. hirsuta and V. sativa). Appendix 3: Number of Phytophagous Insects Recorded by Univac Suction 233 Sampling on L. corniculatus, M. lupulina, T. pratense, T. repens, V.and hirsuta V. sativa in Sites of Different Successional Age 8

List of Tables

Page Table 2.1. Seasonal variation in the percentage cover for each species 37 of legume in sites of different successional age. Table 2.2. Grass/Forb Index and Legume/Forb Index in parenthesis 40 for 7 sampling occasions in 1983/4 in sites of different successional age. Table 2.3. Pattern of distribution of the legume herbs and, in 44 parenthesis, the proportion of sub-plots in which they occurred for each sampling occasion in 1983/4 in sites of different successional age. Table 2.4. Seasonal species diversity for the legumes and total 49 vegetation from early to mid-successional sites. Table 2.5. Spatial diversity of the legumes and total vegetation for 51 different sampling occasions during the season in sites of different successional age. Table 2.6. Importance values for the five top ranking species plus all 54 legume species for 6 sampling occasions during the season for the 1984 site. Table 2.7. Importance values for the five top ranking species plus all 55 legume species for 2 sampling occasions in sites of different successional age. Table 2.8. Percentage of total touches in each height category for 57 each species of legume on six sampling occasions for the 1984 site and two sampling occasions for all other sites. Table 3. The density of insect herbivores (insects/ m2) recorded from 79 five insect groups on Vicia hirsuta and Vicia sativa on two sampling occasions in sites of different successional age. Table 4.1. The number of plants marked on each site for performance 100 studies with L. corniculatus, M. lupulina, T. pratense, T. repens, V. hirsuta and V. sativa. 9

Table 4.2. Maximum height attained by each plant in Lotus 103 corniculatus and Medicago lupulina and Trifolium repens in sites of different successional age. Table 4.3. Maximum number of leaves per plant in Lotus 104 corniculatus and Medicago lupulina and maximum number of leaves per stolon in Trifolium repens in sites of different successional age. Table 4.4. F-values of split plot analysis of variance for vegetative 105 characteristics of Lotus corniculatus, Medicago lupulina and Trifolium repens. Table 4.5. Life expectancy of individual leaves of Lotus corniculatus, 107 Medicago lupulina, Trifolium pratense and Trifolium repens in sites of different successional age. Table 4.6. Relationship between life expectancy and insect damage in 108 Lotus corniculatus, Medicago lupulina, Trifolium pratense and Trifolium repens in sites of different successional age. Table 4.7. The effect of reducing levels of insect herbivory on 113 reproductive characteristics in Lotus corniculatus and Medicago lupulina. Table 4.8. The effect of reducing levels of insect herbivory on 114 number of floweiheads and seedheads per stolon in Trifolium repens. Table 4.9. The effect of reducing levels of insect herbivory on 115 reproductive characteristics in Trifolium repens in two early successional sites. Table 4.10. The effect of site age on seedhead variables in Trifolium 116 repens (control plots only). Table 4.11. Seed germination and plant establishment of Lotus 117 corniculatus, Medicago lupulina, Trifolium pratenseand Trifolium repens under natural and reduced levels of insect herbivory. Table 5.1. The number of stolons of each phenotype involved in tests 132 for cyanogenesis in T. repens marked in sites of different successional age. 10

Table 5.2.a Mean stolon height of each cyanogenic/acyanogenic 139 morph (phenotype) of T. repens in sites of different successional age. Table 5.2.b Results of analysis of variance of mean height of each 140 cyanogenic/acyanogenic phenotype of T. repens in sites of different successional age. Table 5.3.a Mean leaf number per stolon of each 141 cyanogenic/acyanogenic morph (phenotype) of T. repens in sites of different successional age. Table 5.3.b Results of analysis of variance of mean leaf number per 142 stolon of each cyanogenic/acyanogenic phenotype of T. repens in sites of different successional age. Table 5.4. Leaf life expectancy of individual leaves of each 147 cyanogenic/acyanogenic morph (phenotype) of T. repens in sites of different successional age. Table 5.5. Contingency x2 for the partition of eating scores between 151 cyanogenic and acyanogenic stolons in sites of different successional age (control treatment only). Table 5.6. The effect of reducing levels of insect herbivory and site 152 age on mean inflorescence number in 4 phenotypes of T. repens. Table 5.7. The effect of reducing levels of insect herbivory and site 153 age on mean seedhead number in 4 phenotypes of T. repens. Table 5.8. The effect of reducing levels of insect herbivory on 155 reproductive characteristics of cyanogenic and acyanogenic morphs of T. repens in a ruderal and two early successional sites. Table 5.9. The effect of site age on maximum inflorescence and 156 seedhead number in 4 phenotypes of T. repens (control plots only). Table 5.10. The effect of site age on reproductive characteristics of 157 cyanogenic and acyanogenic morphs of T. repens (control plots only). 11

Table 6.1. Number of replicates in each pair-wise preference test 168 comparing four species of legume. Table 6.2. Number of replicates for each pair-wise preference test 169 comparing four phenotypes of cyanogenic/acyanogenic Trifolium repens. Table 6.3. Mean leaf area removed from each leaf in pair-wise tests 175 comparing four species of legume and nine species of phytophagous Curculionoidea. Table 6.4. Mean leaf area removed from each leaf in pair-wise tests 176 comparing four phenotypes of cyanogenic/acyanogenic Trifolium repens and nine species of phytophagous Curculionoidea. Table 6.5. Regression parameters a and 6 for Sitona lineatus feeding 180 on six pair-wise combinations of four phenotypes of cyanogenic/acyanogenic T. repens. Table 6.6.Demographic and biological statistics of A. pisum reared on 186 four phenotypes of cyanogenic/acyanogenic T. repens. 12

List of Figures

Page Figure 2.1. Percentage cover of the legumes throughout the season 36 for sites of different successional age. Figure 2.2. Height profiles of the total vegetation and legumes for 6 41 sampling occasions during the season in the 1984 site Figure 2.3. Height profiles of the total vegetation for May and July 42 1984 in sites of different successional age. Figure 2.4. Changes in the taxonomic diversity of the total vegetation 50 and legumes between October 1983 and October 1984 for sites of different successional age. Figure 2.5. Changes in the percentage cover of Lotus corniculatus 53 between October 1983 and October 1984 in sites of different successional age. Figure 2.6. Changes in the percentage cover of Medicago lupulina 56 between October 1983 and October 1984 in sites of different successional age. Figure 2.7. Changes in the percentage cover of Trifolium pratense 60 between October 1983 and October 1984 in sites of different successional age. Figure 2.8. Changes in the percentage cover of Trifolium repens 61 between October 1983 and October 1984 in sites of different successional age. Figure 2.9. Changes in the percentage cover of Vicia hirsuta between 63 October 1983 and October 1984 in sites of different successional age. Figure 2.10. Changes in the percentage cover of Vicia sativa between 64 October 1983 and October 1984 in sites of different successional age. 13

Figure 3.1. Total insect load recorded between May and November 78 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.2. The density of Coleoptera recorded between May and 81 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.3. The density of Heteroptera recorded between May and 82 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.4. The density of Homoptera recorded between May and 83 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.5. The density of recorded between May and 84 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.6. The density of Symphyta recorded between May and 85 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.7. The density of Thysanoptera recorded between May and 86 November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense and (d) T. repens. Figure 3.8. The number of insect species in each insect herbivore 88 group in sites of different successional age for: (a) Lotus corniculatus; (b) Medicago lupulina; (c) Trifolium pratense; (d) Trifolium repens. Figure 3.9. The number of insect species in each insect herbivore 89 group in sites of different successional age for: (a) Vicia hirsuta; (b) Vicia sativa. 14

Figure 3.10. Sample ordination for all six species of legume 91 irrespective of site or sampling date Figure 3.11. Sample ordination for each site and sampling occasion 92 for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifolium pratense, (d) T. repens, (e) Vida hirsuta and (f) Vida sativa. Figure 4.1. Seasonal variation in the percentage of leaves damaged 110 under natural and reduced levels of insect herbivory in (a) Lotus corniculatus (1983 site), (b) Medicago lupulina (1985 site) and (c) M. lupulina (1984 site). Figure 4.2. Seasonal variation in the percentage of Trifolium pratense 111 leaves damaged under natural and reduced levels of insect herbivory in sites of different successional age: (a) 1985 site, (b) 1984 site and (c) 1983 site. Figure 4.3. Seasonal variation in the percentage of Trifolium repens 112 leaves damaged under natural and reduced levels of insect herbivory in sites of different successional age: (a) 1985 site, (b) 1984 site, (c) 1983 site and (d) 1971 site. Figure 5.1. Variation in the frequency of cyanogenic and acyanogenic 134 morphs at different times of the season, in sites of different successional age, for Trifolium repens: (a) July 1984, (b) September 1984, (c) February 1985, (d) May 1985; and Lotus corniculatus: (e) September 1984. Figure 5.2. Changes in the frequency of each acyanogenic morph of 136 T. repens under natural and reduced levels of insect herbivory with successional age. Figure 5.3. Stolon survival of the (i) cyanogenic, (ii) glucoside only, 137 (c) enzyme only and (iv) homozygous acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in: (a) 1985 site; (b) 1984 site; (c) 1983 site; (d) 1971 site. Figure 5.4. Seasonal variation in the percentage of damaged leaves 143 recorded on cyanogenic/acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in the 1985 site. 15

Figure 5.5. Seasonal variation in the percentage of damaged leaves 144 recorded on cyanogenic/acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in the 1984 site. Figure 5.6. Seasonal variation in the percentage of damaged leaves 145 recorded on cyanogenic/acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in the 1983 site. Figure 5.7. Seasonal variation in the percentage of damaged leaves 146 recorded on cyanogenic/acyanogenic morphs of T. repens under natural levels of insect herbivory in the 1971 site. Figure 5.8. The percentage of leaves recorded in each damage 149 category between June 1985 and May 1986 from cyanogenic and all acyanogenic morphs of T. repens. Figure 5.9. Percentage of leaves recorded in each damage category 150 between June 1985 and May 1986 in each acyanogenic morph of T. repens. Figure 6.1. A test of frequency dependent selection between 178 cyanogenic and acyanogenic morphs of Trifolium repens by Sitona lineatus. Figure 6.2. A test of frequency-dependent selection between 179 acyanogenic morphs of T. repens by S. lineatus. Figure 6.3. Variation in the age specific survival of Acyrthosiphon 182 pisum reared on each cyanogenic/acyanogenic morph of Trifolium repens. Figure 6.4. Changes in the number of progeny/aphid/2 day age 183 interval for A. pisum on each cyanogenic/acyanogenic morph of T. repens. Figure 6.5. The cumulative progeny production/female/day for A. 185 pisum on each cyanogenic/acyanogenic morph of T. repens. Chapter 1 Insect-Plant Relationships: A Community Approach

1.1. Introduction

In any study of insect-plant relationships, three factors must be considered. The first is the effect of the insect on the plant, the second is the effect of the plant on the insect, while the third is associated with the influence of neighbouring plants. The plant-soil interaction normally has only an indirect effect on insect-plant relationships and is not considered here. Interactions between the former three factors occur and cannot be ignored. For instance, Whittaker (1979) suggested that plant competition might only be important when invertebrate grazing occurred at the same time. Plant-herbivore interactions, such as the facultative defences of many plants, provide further evidence. Indeed, much of the theory which relates the ecology and evolution of plants and their insect herbivores in specific habitats or along successional gradients take plant-insect, insect-plant and plant-plant interactions into account. This study is part of a long-term investigation into the relationships between insects and plants during secondary succession, which was initiated in 1977 by Professor Sir Richard Southwood and Dr.V.K. Brown, but since 1979 the work has been led by Dr.V.K. Brown. The first results demonstrated the association between insect herbivore numbers and diversity in relation to plant taxonomic and structural diversity (Southwood et al. 1979). Further work has not only encompassed general aspects of insect-plant relationships (e.g. Brown 1982a,1984,1985,1986, Southwood, Brown & Reader 1983), but also specific insect groups (Brown 1982b, Brown & Hyman 1986, Brown & Llewellyn 1985, Brown & Southwood 1983, Godfray 1982,1985, Stinson 1983, Stinson & Brown 1983). In addition to the main interest in insect-plant relationships,other groups such as birds (Southwood et al. 1986) and small mammals (Churchfield & Brown 1987) have been considered. Currently, the work is endeavouring to take on a broader perspective with comparable studies on other soil types (Brown, Jepsen & Gibson 1988, Gibson, Brown & Jepsen 1987, Gibson et al. 1987) and different geographic regions (Brown, Hendrix & Dingle 1987, Hendrix, Brown & Dingle 1988, Hendrix, Brown & Gange in press,). Manipulative field experiments are also being undertaken to investigate the effect that insect herbivoiy has on the plant community in different stages of 17 succession (Brown et al. 1987, Brown, Jepsen & Gibson 1988, Gange et al. in press, Gibson, Brown & Jepsen 1987). It is within the bounds of this latter investigation that this study falls, although only a restricted part of the successional gradient is being examined; namely, early-mid succession. The plants examined form part of a distinct group, the legumes, which are specialized in the sense that they are free from the almost universal dependence on fixed nitrogen in the soil by virtue of a symbiotic association with nitrogen-fixing organisms (Harper 1977). As a consequence, the legumes are freed from competition for this resource. Examination of this group of plants and their associated insect herbivores aims to extend current knowledge of this part of the plant community. The study investigates the insect-plant, plant-insect and plant-plant relationships separately as well as looking for interactions between them. Such relationships may be obscured if large sections or even complete communities are studied. In the remainder of this chapter, general community processes and theories are outlined in relation to succession, while more comprehensive accounts of relevant topics within plant-insect herbivore interaction theory are given at the beginning of the relevant chapters. Chapter 2 describes the legume community found in the early part of succession at Silwood Park, in terms of abundance (cover), pattern, relative importance, structure as well as taxonomic and structural diversity. These community characteristics are used to interpret the composition of the legume community through the effects of plant life-history strategies, neighbouring species and site attributes. The extensive data set on which this study is based is not presented in the thesis, although the primary data are available at Imperial College. In Chapter 3, the changes in abundance and species composition of the insect herbivores associated with each species of legume are described and interpreted in relation to current theories and predictions. A multivariate ordination was used to identify the insect communities characteristic of each species of legume. The effects of insect herbivoiy on the plant reproductive and vegetative characteristics are examined for each species of legume in Chapter 4. This study evaluates the impact of insect herbivores by monitoring the legume community in their presence and absence (by the application of insecticide). While in Chapter 5 the same approach is used to evaluate the impact of insect herbivory on morphs of a single species (Trifolium repens L.) with polymorphic defences i.e. in respect of cyanogenesis. In Chapter 6, differences in the levels of herbivory between plants of T. repens are investigated in terms of variations in plant chemistry. The latter not only affects the preference and level of damage received, but also the fecundity and longevity of the herbivore. This chapter also incorporates laboratory bioassays which attempt to determine host plant preferences. Finally, Chapter 7 comprises the general discussion and conclusions. 18

1.2. Theory of Plant Succession Succession can be viewed as "the process whereby one plant community changes into another" (Crawley 1986). However, it cannot be ignored that these underlying changes influence and are influenced by the associated community. As a consequence, the course and rate of plant succession can be altered depending on the strength of these interactions. In all cases, the changes are directional and long-term, although short-term seasonal (or cyclical) changes may be superimposed upon them. Succession is initiated by the creation of a new habitat, such as after a radical disturbance or after the opening of a new patch in the physical environment. Traditionally, succession has been categorised as primary, occurring on a completely new site that has never supported living organisms (e.g. a larva flow) or secondary, where the previous occupants have been removed or damaged as a result of a disturbance (e.g. fire, ploughing, flood), leaving their propagules to re-establish. As a result, secondary successions are usually more rapid than primary ones (Brown 1984), although both follow similar lines of development. The initial colonizers are commonly referred to as pioneers which are gradually replaced by a sequence of species until a relatively stable community develops. Many of the theories of succession provide a mechanistic explanation by describing the sequence of species by habitat factors and community characteristics. Other models predict community changes in terms of diversity, biomass and production. In addition to the community approach, there are attempts to understand the characteristics of individual plant species of different successional stages in terms of selection operating on their life-history strategies (Peet & Christensen 1980, Picket 1982). The original concept of plant succession, proposed by Clements (1916), dominated ecological thought on the subject for 50 years. His idea emphasised the sequential nature of the process, known as "reaction", whereby each plant community modifies the site in such a way that it becomes unsuitable for the present plant assemblage but more suitable for the next. This sequence of events continues until the process of reaction stops and a stable, self-perpetuating "climax” community develops. Another term for this process is "relay floristics" (Egler 1954) but the concept has been criticized by certain authors (e.g. Connell 1972, Drury & Nisbet 1973) since it is mainly applicable to primary successions. A variation of this model was first developed by Egler (1954), using the term "initial floristic composition", since most of the species are present from the start and the course of succession solely depends on relative growth rates, generation times, persistence and other life-history traits of the constituent species. This model is most likely to apply to 19 secondary successions initiated by small scale disturbances in large areas of climax vegetation or under larger-scale disturbances where most species are adapted to the disturbance (see Gibson & Brown 1985). Connell and Slatyer (1977) recognised three models of succession, in all of which the early colonisers are important. The first was referred to as the "facilitation" model and is essentially as that described by Clements 61 years previously. The other two models lie between the extremes of "initial floristic composition" and "relay floristics". The "tolerance" model assumes that alteration of the environment by the early successional colonists neither inhibits nor assists the development of later successional colonists. Instead, the sequence of species is solely determined by their life-history characteristics. Generally, the late successional colonists, can grow in the presence of the early successional colonists but their growth is slower and their dispersal and reproductive powers smaller. Consequently, dominance of the community takes much longer, but once this is achieved the late successional colonists remain due to their superior competitive abilities. However, there is little convincing evidence to support this model. By contrast, the "inhibition" model envisages that the initial colonists pre-empt resources in such a way that they inhibit the invasion or reduce the growth and survival of the later colonists. These later colonists can only overcome this domination by the damage or death of the early colonists, resulting in resources being released. As it is assumed that the longevity of the early colonists is shorter than those plants which colonize later, the later colonists will therefore accumulate and eventually dominate. However, when the longevity of the early colonists is high, the rate of succession is very much reduced (see Whitmore 1985). It is this latter model which Connell and Slatyer (1977) conclude as the most important with regard to secondary successions and cite considerable evidence to support it. However, the applicability of many of the cited studies have been criticized by McIntosh (1980), while Hils and Vankat (1982) argue that both "tolerance" and "inhibition" models may be acting concurrently on different species in the same plant succession. Another qualitative model was proposed by Noble and Slatyer (1980), in which they considered a small number of life-history characteristics termed "vital attributes". These were considered to determine the replacement sequence which depicts the major shifts in species composition and thereby defines the place each species holds in the succession. Other researchers have developed models describing the changes in the structure and function of the community as succession proceeds (Margalef 1968, Odum 1969). The use of matrix models has provided the mathematical framework necessary to develop quantitative descriptions of secondary succession. Horn (1976) utilized matrices of "replacement probability". These matrices define Markov chains which predict the convergence of the community to an equilibrium state (Usher 1987). However in all 20

successions, where suitable data have been gathered, complex Markov chain models will be needed for any degree of realism. These models have led to the suggestion that the process of succession is non-random, non-stationary (in time and space) and dependent on the history of the system as well as on its present configuration (Usher 1987).

1.3. Plant-Plant Relationships

All plants have three basic requirements which are light, water and nutrients; however, the relative proportions of these needs varies between each plant species. As a result, different species of plant occupy different positions along the resource continuum by seeking the same resource at different times and in different places (Harper 1977). Some plants have evolved specialized mechanisms that allow them to escape the limitations of particular resources by forming symbiotic associations with bacteria and/or fungi. Other plants have solved the same problem by becoming insectivorous or even parasitic. However, most plants do not have this ability but do coexist, avoiding interference from other plants by occupying specialized microenvironments within the community mosaic. In heterogeneous habitats, many microenvironments will be available through variations in micro-topography, edaphic factors, microclimate, plant growth form, microsite requirements and so on. Such variability in the environment is termed lateral heterogeneity, an example of which is the distribution of different tree species along an altitudinal gradient (Whittaker 1969). There is also a vertical component in the distribution of resources, whether it be for light, water or nutrients. Light gradients are commonly found in many environments such as broad-leaf woodlands where there may be an upper, middle and lower canopy consisting of various trees and shrubs as well as a field layer composed of herbs. Leaves in the lower canopies absorb light that has not been intercepted by the upper canopy, thus resulting in greater shade tolerance in the lower levels of the canopy. In addition to the spatial distribution of resources, there is also an uneven distribution through time. However, both spatial and temporal differences in environmental heterogeneity frequently coincide. In many cases, plant species coexist simply because of seasonal variation in shoot expansion and flowering. Such complementary phenological patterns are exhibited in a number of pasture herbs and grasses and tail-herb communities (Grime 1979). Other changes in temporal heterogeneity may result from cyclical fluctuations in nutrients (see Turkington & Harper 1979c). 21

When the requirements of different plant species are similar, direct conflict results in the form of interspecific competition, while direct conflict between two (or more) plants of the same species results in intraspecific, intergenetic competition. Plants with the ability for vegetative growth, additionally experience intraspecific, intragenetic interference (Turkington & Harper 1979b) since clones of exacdy the same genetic composition may compete for the same resources. The outcome of competition may either cause one plant species (or phenotype) to replace another or cause the niches of the different competing species (or phenotypes) to be more closely defined. In the former case, species diversity may decrease since a few competitively dominant plants may emerge, while in the latter case species diversity may well increase since it favours greater "species packing".

1.4. Plant-Insect Relationships 1.4.1. Plant Architecture This term was introduced by Lawton (1978) in a bid to explain why insect diversity changes with plant size and complexity of form. The idea assumes that large, complex plants support more insect species on account of their size presenting a bigger target for colonizing insects. Large size and greater persistence will also enable larger insect herbivore populations to exist, and these will have a reduced chance of extinction through stochastic events . The concept of plant architecture was expanded by Southwood et al. (1979) who also provided the first quantification of it on a community scale. This incorporated the variety of plant structures which provide a greater diversity of niches or microhabitats for insect herbivores to exploit. However, not all plant structures are edible and these may provide non-trophic resources such as sites for overwintering, oviposition, hiding places (enemy free space) and so on.

1.4.2. Species-Area Relationships This phenomenon was first documented by Southwood (1960), who observed a greater number of insect herbivores associated with more geographically- widespread plants than from local or rare species. This can be partly attributed to the fact that the commoner, widespread plants occur in a wider range of habitats and as a result come into contact with a greater number of local insect faunas. In addition, it may also depend on the probability of a mobile insect herbivore discovering or encountering a potential host plant and this will be related to its abundance and geographic range (Strong, Lawton & 22

Southwood 1984). However, it cannot be concluded that widespread plants support more insect herbivores per individual than local, but similar-sized, plants. However, species-area relationships can help explain variations in the number of insect herbivore species as a result of local changes in host-plant patch size, since individual plants or clump of plants do not host all the insect species able to feed on that plant over its whole geographic range. Instead, the insect herbivore species are drawn from a regional pool of potential colonists and their numbers will be proportional to the patch size of the host plant (Lawton 1978, Ward & Lakhani 1977). The mechanisms governing these local species-area relationships are presumed to be broadly similar to those yielding the geographic species-area relationships (Strong, Lawton & Southwood 1984).

1,4.3. Resource Concentration Hypothesis This hypothesis relies on the interaction between two parameters: host-plant density relative to other plant species and absolute host-plant density. Both these factors influence insect herbivore abundance on the host plant. Root (1973) stated that "herbivores are more likely to find and remain on hosts that are growing in dense or nearly pure stands". As a result, the influence of other species of plant (e.g. polyculture) may affect the way in which an insect herbivore makes use of a host. Thus in a plant community, neighbouring plants may make the host less easy to find (especially true for monophagous herbivores) by masking the chemical attractants of the host (Tahvanainen & Root 1972), reducing the contrast between the host and its background (Smith 1976) or by making the host less visible to the herbivore (Rausher 1981). Re-take-off rates may be increased or residence time reduced in polyculture (Bach 1980), especially if the herbivore alights on a non-host plant or if the non-host plants alter the microclimate (Risch 1980). The effect of host-plant density on insect herbivore abundance is not consistent within all insect phytophagous groups. Some insect herbivores are attracted in greater abundance to high-density patches of host (Lemen 1981, Ralph 1977), while others are attracted to either intermediate or low-density patches (Thompson & Price 1977) and some are apparently indifferent (Rausher & Feeny 1980). Thus different insects may attack precisely the same distribution of a single plant species in different ways (see Cromartie 1975).

1.4.4. Plant and Insect Life-History Strategies The life-history strategies of organisms along a successional gradient can be described in terms of the r-K continuum (MacArthur & Wilson 1967). These two parameters refer 23 to the logistic model: r-selected organisms suffer mainly density-independent mortality and a high rratx is continually being selected for; K-selected organisms experience mainly density-dependent mortality and selection favours increased competitive ability at K, their carrying capacity. Thus, as early and late colonists experience different selection pressures at opposite ends of this continuum, contrasting life-history characteristics will evolve. Focusing on the plant community, early successional habitats are characterized by r-selected species due to their ephemeral and unpredictable nature. Consequently, plant species in these habitats will tend to channel their resources into rapid development and high reproductive output, since site suitability is rather brief, hence, the necessity for small, easily dispersed seeds. By contrast, K-selected plants will be most frequent in late successional habitats, since they are able to survive under conditions of competitive stress. As a result, their growth rate is comparatively low and this is paralleled in their reproductive potential, since these attributes are sacrificed for increased competitive ability, longevity, resource-stress tolerance and large size. Fewer seeds are produced by the late successional colonisers but these are larger as more resources are needed for establishment in more crowded conditions. It might be expected that the life-cycle strategies of the insect community at either end of the r-K continuum should be closely tuned to the vegetational characteristics of the habitat and in herbivorous insects often directly to their host-plants’ life cycle. Brown (1986) has considered a range of insect herbivore life-history characteristics along a successional gradient and it is from this study that many of the following assumptions are based. Thus, insects associated with early successional habitats have comparable opportunistic life-cycle strategies to their host plants, such as high dispersal ability for rapid invasion of newly-created habitats, small size, high reproductive potential and short generation time. There are also more specific characteristics, such as a predominantly adult overwintering stage and wide host-plant range or niche breadth. On the other hand, insects appearing later in succession show a general progression towards a lower reproductive potential, longer generation time, the egg or adult as the dominant overwintering stages, increased size and narrower host plant range or niche breadth, although in the latter case, there are exceptions (Brown & Hyman 1986). 24

1.5. Insect-Plant Interactions 1.5.1. Herbivory The impact of the insect herbivore on the plant depends not only on the amount of plant material removed but also on the kind of tissue removed and on the timing of attack relative to the plant’s development (Crawley 1983). The effect of removing photosynthetic area (by leaf removal and leaf mining) can directly reduce the net production of the plant. However, the timing of defoliation is important since as the leaf ages, its value to the plant changes. The removal of young leaves is probably most damaging, since the resources used to produce it will be lost as well as its future photosynthetic production. At the other extreme, removal of older, senescent leaves may not have any effect since their contribution to production will be minimal or even non-existent. Reduced production will also occur as a result of changes in the carbohydrate balance (due to phloem sap-feeders) as well as in the water and nutrient status (due to root and xylem sap-feeders). In all these cases, reduced production is manifested in reduced growth, survival, fecundity and ultimately fitness. Decreased growth rate may cause a delay in maturation and change plant size distribution. This latter effect is important particularly when the impact of the herbivore changes the size distribution of a susceptible but dominant plant, thereby giving competitive advantage to previously suppressed plants (see Tansley & Adamson 1925). However, preferential attack on the susceptible small plants may increase their death rate substantially (Gange et al. in press). The effect of defoliation (or decreased production) on fecundity may decrease the number of flower buds set, increase flower abortion, reduce the number of seeds produced (Bentley, Whittaker & Malloch 1980) or reduce seed size (Brown et al. 1987, Hendrix 1979). Timing of defoliation may be important, since Tinker (1930) found that early cutting of two pasture grass species delayed flower production, while late defoliation stopped flower production altogether. Delay in flower production may also result in reduced seed output due to lack of pollinators or increased susceptibility to frost damage. The effect of sap-feeding insects on fecundity is equally dramatic (Waloff & Richards 1977). However, these effects may be further exacerbated when herbivores such as aphids are ant-tended, as in the case of Aphis fabae Scopoli on Viciafaba L. (Banks & Macaulay 1967). The effects of root herbivory are not well known, although it is expected that root-feeding insects will reduce water and nutrient uptake as well as reduce carbohydrate storage.The latter will in turn, reduce the amount of assimilates translocated above ground 25 for seed production. It is therefore expected that root herbivory will reduce fecundity, although in some trees root pruning has been shown to increase seed production and such anomalies may well occur in other plants (see Crawley 1983). Flower and fruit predation by insects will obviously have a deleterious effect on fecundity, although fruit predation may be beneficial in the case of vertebrates since it may facilitate seed dispersal. The occurrence of seed predation in the pre- and post-dispersal stage can have a profound effect on the dynamics of a plant by directly affecting its fitness. Seed predation in the pre-dispersal stage will have a similar effect to flower removal, since it will reduce seed yield and thereby decrease the number of seeds entering the seedbank. Whether this will decrease the number of recruits in the next generation will depend on the subsequent pattern of seed dispersal (Harper 1977) and on the size of the seedbank. After the seed has left the parent (post-dispersal stage) it may germinate immediately or enter the seedbank and germinate much later. However, the impact of the seed predator on the plant’s population dynamics will depend on the insect’s ability to deplete the seedbank of the plant. Seedling herbivory is also an important component in the dynamics of a plant population, since it is easier for an insect herbivore to kill a seedling than a mature plant. Hence, the density and distribution of seedlings is often very different from that of mature plants (see Dritschilo et al. 1979). The impact of insect herbivory can be offset to a certain degree by compensation by the plant. This process encompasses a number of mechanisms, such as increased longevity of leaves, increased photosynthetic rate, mobilisation of stored carbohydrates and proteins, excess flower production as well as altered patterns of the distribution of photosynthate. However, how well a plant can compensate will depend on the severity of attack and the amount of stored resources available. In addition, to the direct effects that insects have on plants, they also influence the plant indirectly by injecting chemicals into the plant, transmitting viral diseases, causing resources to be diverted to wound repair and production of secondary chemicals, increasing susceptibility to stress and so on (Crawley 1983). Thus it is not surprising that plants are continually evolving new methods of reducing all forms of herbivory and it is this subject which will be addressed in the next section.

1.5.2. Plant Defences Plants, particularly angiosperms, provide the food material for over half of all insect species (Fraenkel 1959) and as a result, plants have evolved and are still evolving novel 26 methods of counteracting this selection pressure. Reducing food quality is one method commonly used and this can be done in two ways. Firstly, by evolving surface or physical defences and secondly, by chemical means. Both forms of defence serve to reduce the edibility of the plant. Surface defences can be described as those which provide a physical barrier between the insect and its food source. Often many of these defences are associated with the epidermis (hence the name "surface defence") since this outer layer of cells often contains lignin and silica or is covered with layers of wax (cuticle), cork (bark) or trichomes (hairs). However, many of these structures also provide other functions for the plant such as structural support and the reduction of desiccation, and can thus be alternatively regarded as proximal defences. This does not mean that they are unimportant; for instance, Singh et a/.(1971) demonstrated that hairy genotypes of supported fewer leaf hoppers than glabrous genotypes. Feeny (1970) found that leaf toughness in oak leaves prevented late larval feeding by the winter . A more specialized form of structural defence is shown by Passiflora L. species (Gilbert 1975), which possess leaves that mimic the shape of non-host plants as well as butterfly "egg mimics" which deter oviposition by their Heliconius L. herbivores. Another defence, which lies between physical and chemical deterrents, is the reduction in nutrient content of the plant. Such measures can be deleterious to the plant, since the insect herbivores may simply consume more of the plant to compensate for its low quality. However, a number of researchers have attributed a low foliar nitrogen content as an adaptive response to insect herbivore pressure (Auerbach & Strong 1981, Rausher & Feeny 1980). Chemical defences represent the ultimate factors in the effort to decrease the levels of insect herbivory, which would ultimately reduce plant fitness. Fraenkel (1959) proposed the "raison d’etre" of these so-called secondary plant substances or metabolites was specifically for this function. Plants contain a vast array of these secondary metabolites, encompassing many chemical groups (e.g. glucosides, saponins, tannins, alkaloids, essential oils, organic acids, analogues of insect hormones and many more), perhaps indicating the scale of selection imposed by insects. Recently, there have been attempts to classify plants in terms of the chemical defences they possess. For instance, Feeny (1976) and Rhoades and Cates (1976) suggested that r- and K-selected plants (and plant organs) should have different defences due to differential selection pressure by their insect herbivores. However, these chemical defences fall into two broad categories, of which the first is concerned with interference with digestive processes by binding to enzymes and complexing starch and protein within the gut. Whereas, the second act as toxins which 27 inhibit specific metabolic processes. Some of these chemical deterrents have been circumvented by some insects, especially the monophages or specialists and, as a result, many of these secondary metabolites represent relative and not absolute defences. Although many of the defences described above may be present throughout the life of the plant, a number are inducible. For instance, the prickles on the stems of Rubus L. are longer and sharper when the plants are grazed by cattle (Abrahamson 1975) and thus represent a form of facultative physical defence. Increases in secondary compounds following herbivory have been reported by Haukioja (1980) and Rhoades (1979) and represent facultative chemical defence. Finally, the possession of extra-floral nectaries in some plants has acquired protection by ants (and parasitoids) from some insect herbivores and even competing plant species. In some neotropical trees, the ants are obligate ’’body guards” (see Janzen 1979), although in other plants the relationship is less close (Koptur & Lawton 1988). Chapter 2

Composition, Pattern and Structure of the Legume Community

2.1. Introduction Plant succession can be viewed as a process of community development that is considered to be directional and therefore predictable. The whole process is controlled by the plant community, although the physical environment often determines the pattern, rate and end point (Odum 1969). Harper (1977) considered that a plant population growing in one place at a point in time was the consequence of a catena of past events and he likened community development to a play where: "The climate and substrate (physical environment) provide the scenery and stage for a cast of plant and animal players that come and go. The cast is large and many members play no part, remaining dormant. The remainder act out a tragedy dominated by hazard, struggle and death in which there are few survivors. The appearance of the stage at any moment can only be understood in relation to previous scenes and acts, though it can be described and, like a photograph of a point in the performance of a play, can be compared with other points in other plays." Such comparisons between "other points in other plays" have been made in many studies, often between sites adjacent to each other (Bazzaz 1975, Nicholson & Monk 1974, Southwood et al. 1979) or between sites in different geographic regions (Brown, Hendrix & Dingle 1987). Few investigations have concentrated on single sites studied over a period of several years (but see Brown & Southwood 1987, Tramer 1975), although such investigations are limited to early successional communities due to the time span involved. However, similar patterns have emerged from many of these studies, mainly in terms of growth form and diversity. These patterns have been identified in four recognized types of communities: ruderal (first year of succession); early-successional (second to fifth year); mid-successional (fifth to fifteenth year); and late successional from when shrubs and trees dominate (60 years in Brown & Southwood 1987). The ruderal community is dominated by annuals but during the second stage of succession (early succession) annual and biennial herbs decrease in abundance and perennial herbs and grasses increase. It is during this period that taxonomic diversity (William’s a) is 29 highest due to a high species turnover and a range of plant life forms (Southwood et al. 1979). The third stage in succession (mid succession) is dominated by grasses and perennial herbs, although shrubs and trees begin to invade. In the last stage, the latter attain dominance and contribute to a high architectural diversity (Southwood et al. 1979). These stages in the community are, however, not clearly defined and the timing may vary between secondary successions in different areas (e.g. Tormala 1982). In this study only the first three stages will be considered. The studies mentioned above mainly consider community characteristics, rather than emphasize the importance of the life-history characteristics of individual species, since it is the latter that help determine the patterns that drive succession. Whether a plant can exist in a particular place will be determined by its response to local physical and chemical variations in the soil (Pemadasa & Lovell 1974, Snaydon 1962), competition (Jackson 1981), growth form and phenology (Turkington & Harper 1979b), climate and other environmental variables (Beinhart 1963, Rackham 1980), allelopathy (Rice 1974), "ecological combining ability" (Turkington et al. 1977) and ability to withstand disturbance such as grazing (Ellison 1960) or fire (Biswell 1974). As a consequence of interactions between these factors, one population of plants will certainly influence other populations since plants are relatively immobile and are forced to live in the same lateral relationship with their neighbours throughout their lives. Hence, pressures from neighbours are continuous and, where environmental conditions are relatively homogeneous, are the principal factors directing community change (Turkington & Cavers 1979). These changes may result in species replacement or selection for more specialized phenotypes. Wemer and Platt (1976) found that in the Solidago L. six species were strung along a moisture gradient. However, in the old field communities the niche overlap was greater than in mature prairie, which they attribute to lower levels of competition and greater site pre-emption. Similarly, studies carried out by Tuikington (1979) and Turkington & Harper (1979a) demonstrated that specific strains of legumes were selected for by different grass species. Thus during community development, each species in the community will be continually changing as a result of selection pressures from other species. Some species fail to keep up with these changes and will therefore be lost from the community. Consequently, in the struggle for existence, there appears to be no overall dominants, since each species is continually trying to maximize its share of the limited resources. If dominance is achieved (during early succession at least) it tends to be brief, due to rapid changes in species composition as a result of increasing levels of competition. 30

In this chapter, the species composition of the plant community is attributed to a wide range of characteristics. Here, some of the key features are discussed in relation to the abundance (in terms of cover), relative importance in the community, pattern and diversity of the legumes in a sequence of sites along the early part of a successional gradient. 2.2. Materials and Methods

22.1. Site Description The study area was situated at Imperial College at Silwood Park, Ascot, Berkshire, U.K. (51° 21’N and 0° 39’W, elevation 91 metres). The experimental sites represent a secondary succession from ruderal to mid-seral stages i.e. from 1 to 14 years. Soils in the area consist mainly of Bagshot Sands and gravel of Eocene Bracklesham Beds. The climate is temperate, maritime with rainfall varying little throughout the year. Sites, created since 1977, have been prepared on a long-standing arable area. The site created in 1981 had a different history, since the area was under permanent grassland for at least two decades prior to the start of the experiment. All sites were established in the same way: in autumn of the year prior to establishment each site was treated with herbicide ("Round-up" obtained from Murphy Chemical Ltd., Wheathampstead, Herts.) to eliminate perennial weeds and was shallow ploughed, harrowed and lightly rolled in mid-March of the following year. Thereafter, the vegetation was left to develop naturally. All sites created after 1977 were near to each other with a boundary zone of at least 5 metres between them. The site created in 1982 (designated "1982 site") has an area of 252m2 and was subdivided into 28 subplots, each 3 metres by 3 metres, arranged in a 7x4 pattern. All other sites have an area of 405m2 and were sub-divided into 45 3mx3m subplots. A fence surrounded all sites to exclude rabbits, although this has only been partly successful and light grazing still occurs from time to time. The site created in 1971 (referred to as "1971 site") is within 250m of the younger sites and it is considered that there is "no great spatial barrier to colonization ....by the same organism" (Southwood et al. 1979). The site was established where soil, removed from the construction of the "Nuclear Reactor Centre", was tipped. The area was subsequently covered with top soil and left to colonize naturally. Rabbits were free to graze the area prior to its enclosure in 1977, but since then some light grazing has occurred, mainly near the edges where patches of Rubus fruticosus L. occur. Throughout the thesis the sites are referred to by year of establishment to avoid confusion with subsequent sampling seasons. 31 22.2. Plant Recording Techniques The vegetation was sampled using point quadrat pins, each pin being 500mm long and 3mm in diameter. In the 1977 site, saplings were establishing and longer pins had to be used. In order to record the vertical structure of the vegetation, each pin was marked off at intervals of 2, 4, 6, 8, 10 and successive 5cm intervals from ground level. During sampling, each height-profile pin was placed vertically on the soil surface so that each interval represented the height of the vegetation above the ground level. The vertical position of the pin was critical since even a slight inclination introduces inaccuracies in the height measurement and also in the number of touches (Greig-Smith 1983). In the ruderal site (designated "1984 site"), 30 individual pins were thrown randomly into each subplot. The number of touches of each species at each height interval were recorded for 10 of the pins (referred to as multiple touch data), while for the remaining 20 pins only the plant species touching the pin were recorded (referred to as presence/absence data). Single points were used in preference to frames of pins, since the degree of precision is greater as each point is independent of each other (Greig-Smith 1983). Sampling of the 1984 site was undertaken at monthly intervals between May and October 1984. In all other sites, 10 individual pins were thrown randomly into each subplot on six occasions between May and October 1984 and species touching each pin recorded. The number of touches of each species at each height interval was recorded for 5 randomly thrown pins in each subplot on two occasions (May and late July/early August). One of the major disadvantages of using point quadrats of finite diameter is that cover values are overestimated because some of the plants that are touched do not make contact with the axis of the pin. However, as the same sampling procedure is used throughout, the error will be consistent between sites and is therefore acceptable (Greig-Smith 1983). All plant species touching the pins were identified with the aid of the following keys: Clapham et al (1981), Fitter et al (1978), Hubbard (1984).

2.3. Analysis of Point Quadrat Data 2.3.1. Percentage cover This was calculated from the presence/absence data as follows: number of pins touched bv each legume species in a site x 100 total number of pins for the site 32

In addition to cover estimates for each legume species (Figures 2.5.-2.10. and Table 2.1.), estimates of total legume cover were also calculated (Figure 2.1.). However, in the latter case where more than one species touched the same pin it had to be regarded as a single legume species.

2.3.2. Plant Structure Height profiles illustrate the vertical structure (stratification) of the vegetation (Figure 2.2. & 2.3.). Each profile utilises the multiple-touch data and was based on the proportion of the touches in each height category. Profiles for the legumes are also given in Figures 2.2. and 2.3. For individual legume species, the height categories were reduced to four: 0-10, 11-30, 31-60 and >60cm. The proportion of total touches in each category were expressed as a percentage and are tabulated in Table 2.8.

2.3.3. Grass/Forb Index The proportion of grasses to forbes (i.e. any broad-leaves herb) was calculated as follows: Total number of touches of grasses Total number of touches of forbes Thus, a ratio of zero indicates no grass cover, whereas a ratio of one indicates an equal frequency of grasses and forbes. In older sites, where Luzula L. species were found, the number of touches of this genus were included in the total number of grass touches. Thus, in these situations the index is more accurately a monocotyledon/dicotyledon index. Indices were calculated for each sampling occasion and site.

2.3.4. Legume/Forb Index This describes the proportion of legumes which make up the herb community and was calculated from: Total number of touches of legumes Total number of touches of forbes A ratio of zero is interpreted as no legumes present and a ratio of one means that all the forbes are legume species. Legume/forb ratios were calculated for each sampling occasion and site. 33

2.3.5. Pattern of Distribution The use of density data in measuring the degree of randomness within it is dependent on the size and sometimes shape of the quadrat area used. The quadrat area in this study was the 3x3m subplot, but since the number of pins per subplot was generally small, the values for the mean density of individual legume species will be low. As a result of large quadrat size and low density of individuals there may be an undue proportion of empty quadrats, which will indicate a non-random distribution. Thus, any statistical test in which the Poisson series applies cannot be used. However, the data can be used comparatively to indicate homogeneous or heterogeneous distributions in each site for all legumes and for each species taken separately. As the expected variance for a random distribution cannot be calculated, the ratio given by Hill (1973b) is modified slightly to the form: Observed variance / (mean)2 This ratio has the property of retaining roughly the same value if the same disproportionate decrease is applied to all values, although for frequency and cover data this may not apply (Greig-Smith 1983). If the ratio is equal to zero, a completely homogeneous distribution is indicated, if greater than zero, a heterogeneous distribution. The greater the departure from zero, the more heterogeneous is the pattern. The pattern of distribution was calculated for the presence/absence and multiple touch data sets, the mean for the former was the proportion of subplots in which a legume occurred, while for the latter, it was based on the mean number of touches per subplot. The variance of each mean was calculated from the formula: Zx2- variance = ------n — 1

2.3.6. Index of Diversity Diversity was measured using William’s a, (see Southwood 1978 and Greig-Smith 1983) and was calculated using the maximum likelihood method of: S = aln (1 +N/a) Fisher et al. (1943) first suggested that a two parameter logarithmic series was applicable to species abundance relations. This index refers to the number of species in a community sample, but fails to recognize changes in species composition between samples (see Crawley 1986). 34

Calculation of a relies on two parameters, S and N being known. In the calculation of taxonomic diversity based on the presence/absence data; S represented the total number of species in the sample and N the total number of touches. Spatial diversity, on the other hand, was assessed from the multiple touch data, where S was the summation of the number of species in each height class and N the total of number of touches in the sample. Both taxonomic and spatial diversity was calculated for the total vegetation and for the legumes in each site for each sampling occasion. In a few samples when N = S, a could not be calculated.

2.3.7. Importance Values These were calculated as described in Brown, Hendrix & Dingle (1987): _ . number of subplots containing a species Frequency4 y of a species y = ------total number of subplots sampled „ , . „ frequency of a species Relative Frequency = ------sum of—— the frequency ------of all—— species------;— number of individuals of a species Density of a Species = area sampled density of a species Relative Density = sum of the densities of all species number of touches of a species from height profile pins’ Relative Cover = total number of touches of all species from height profile pins’ Importance value for a species = Relative Frequency + Relative Density + Relative Cover The calculation of relative density makes the assumption that, where a height profile pin is touched a number of times by a given species, only one individual of that species is sampled by that pin. Observations made over two years by Brown, Hendrix & Dingle suggest that this assumption is realistic. Importance values were calculated from the multiple touch data for each plant species on every site on all sampling dates. Table 2.6. & 2.7. gives values for the 5 most common species plus all legume species. 35

2.4. Results: General Trends 2,4,1. Cover Species of Leguminosae were a major component of the flora. Eleven species were found and displayed a range of reproductive strategies and growth forms. In addition, each species has its own nitrogen supply as a result of a symbiotic association with nitrogen-fixing bacteria. Trifolium dubium Sibth., Vicia hirsuta (L.) S.F. Gray, V. sativa L. and V. tetrasperma (L.) Schreber are annuals reproducing exclusively by seed. Medicago lupulina L. was observed to be an annual, although in other studies (Parvone & Reader 1985, Brown, Jepsen & Gibson 1988) biennial and short-lived perennial life forms were found. M. sativa L., T. hybridum L., and T. pratense L. are perennials which reproduce solely by seed, whereas Lotus corniculatus L., L. uliginosus Schkuhr. and T. repens are perennials with creeping stems able to reproduce vegetatively by clonal growth as well as sexually. The cover of the legumes shown in Figure 2.1. changes during the season and with successional age. Peak cover occurred in late June/early July in the 1981, 1980, 1979, 1977 and 1971 sites. In the 1984 site peak cover was not attained until mid July, since plants were establishing from seedlings in the spring. However, the pattern of temporal changes in these sites were relatively similar. In most sites the cover in October 1983 was similar or slightly higher to that at the beginning of the following season and indicated some vegetative growth during the winter and early spring relative to other species. There were, however, two exceptions (1982 and 1971 sites) where cover decreased and in the former site this was dramatic from 68% in October 1983 to 42% in May 1984. This trend may be the result of increased mortality or the relatively superior growth of other species. This trend continued in the 1982 site while recovery was evident in the 1971 site.

2.4.2. Species Richness and Composition There were 11 species of leguminous herbs encountered during the course of this study, of which eight was the maximum number recorded in the 1984, 1980, 1979, 1977 and 1971 sites. The 1983 site had a maximum of seven species, while only 4 species were found in the 1982 and 1981 sites. Not all species were present throughout the season, with the highest numbers being recorded in June and August (Table 2.1.) Only three species ( T. repens, T. pratense and V. sativa) were common to all sites, although the relative proportions of these species changed with successional age. V. sativa reached up to 45% cover in the older sites, whereas T. repens and T. pratense were % Cover Figure 2.1. Percentage cover of the legumes throughout the season season the throughout legumes the of cover Percentage 2.1. Figure for sites of different successional age. Symbols: (•) 1984 site; (■ ) 1983 1983 ) (■ site; 1984 (•) Symbols: age. successional different site; of sites for ie a 17 site.. 1971 (a) siter 93 1984 1983 ( a ) ) 1982 site; (o) 1981 site; (♦ ) 1980 site; (▼ ) 1979 site; (o) 1977 1977 (o) site; 1979 ) (▼ site; 1980 ) (♦ site; 1981 (o) site; 1982 36 37

1984 site 1983 site Species^^Sample Oct May Jun Jul Aug Sep Oct Oct May Jun Jul Aug Sep Oct of Legume 1983 1984 1983 1984 Lotus corniculatus 0.1 1.3 2.7 1.3 2.0 3.3 3.1 3.8 Lotus uliginosus Medico go lupulina 0.2 1.0 6.0 7.0 8.4 10.7 Medicago sativa Trifolium dublua 1.6 0.2 0.2 Trifolium hybridum 1.3 1.1 1.6 1.5 1.8 1.3 1.8 1.8 0.9 Trifolium pratense 0.1 0.3 0.3 0.5 1.0 24.2 25.3 33.8 39.3 41.8 38.7 34.9 Trifolium repens 0.4 0.9 0.7 0.7 1.3 0.4 10.7 15.6 19.8 12.9 8.2 13.3 Vida birsuta 1.2 12.9 31.0 2.6 0.7 2.2 1.1 1.3 0.2 Vida sativa 0.8 2.9 0.4 0.1 0.1 1.8 1.1 0.4 0.2 1.6 VIda tctraspcrma 0.3

Table 2.1.a Seasonal variation in the percentage cover for each species of legume on sites of different successional age <1984 & 1983).

1982 site 1981 site Spec ie s^ ^ S a m p le Oct May Jun Jul Aug Sep Oct Oct May Jun Jul Aug Sep Oct of Legume 1983 1984 1983 1984 Lotus corniculatus 0.4 Lotus uliginosus Medicago lupulina Medicago sativa Trifolium dubiua Trifolium hybridua Tritoliua pratense 61.4 18.2 16.1 16.1 12.9 7.9 3.2 0.9 0.4 0.2 0.2 0.7 1.3 0.7 Trifolium repens 6.1 25.4 18.2 8.2 3.2 2.5 2.1 0.4 1.1 0.2 0.4 Vida hlrsuta 0.7 2.2 3.1 0.7 0.2 0.7 Vida sativa 1.4 1.4 0.4 1.1 0.7 0.9 7.6 12.7 0.4 6.4 10.7 7/cia tetrasperma

Table 2.1.b Seasonal variation in the percentage cover for each species of legume on sites of different successional age (1982 & 1981). 38

1980 s ite 1979 s ite Spec 1 es^'\Sa»ple Oct May Jun Jul Aug Sep Oct Oct May Jun Jul Aug Sep Oct of Legume 1983 1984 1983 1984 Lotus cornieulatus 0.4 0.9 0.7 0.2 0.4 0.4 1.8 4.7 1.8 0.4 1.6 0.7 1.1 Lotus uliginosus 0.4 0.7 0.4 Medlcago lupulina 6.7 39.3 44.9 40.9 35.6 31.1 3.6 35.8 43.8 38.4 35.8 29.8 Medico go sativa Trifolium dubium 0.2 0.2 0.7 0.2 0.2 Trifolium hybridua 0.2 Trlfolium pretense 13.1 0.4 1.3 2.0 1.8 1.8 2.0 14.7 0.7 0.4 0.4 Trifolium repens 0.4 0.9 0.4 0.4 0.2 0.2 Vida hirsute 0.4 1.3 4.2 2.9 0.4 7.8 5.6 24.9 21.3 0.9 2.2 2.4 Vida sativa 5.8 15.1 24.7 10.4 3.8 O ( 6.0 12.2 20.2 35.6 16.0 2.2 3.3 12.2 Vida tetrasperaa

Table 2.1.C Seasonal variation in the percentage cover fo r each species of legume on sites of different successionai age (1980 & 1979).

1977 s ite 1971 s ite Spec i e s v>^>vSample Oct May Jun Jul Aug Sep Oct Oct May Jun Jul Aug Sep Oct of Legume v 1983 1984 1983 1984 Lotus corniculatus 2.2 1.1 4.0 4.0 4.0 2.7 2.4 7.6 5.3 8.2 11.3 16.0 14.2 12.7 Lotus uliginostis 1.1 0.9 0.9 Medics go lupulina 3.3 2.0 1.3 1.8 1.1 0.4 0.2 0.2 0.7 Medicago sativa 0.7 0.4 0.2 0.2 Tri folium dtiblum 0.7 5.1 0.7 0.2 0.2 Tri folium hybridum Tri folium pretense 2.4 0.2 0.2 0.2 0.2 0.2 0.7 0.2 0.4 1.3 0.2 Tri folium repens 0.4 0.4 28.0 10.9 20.0 14.4 10.9 11.6 10.4 Vide hirsute 18.4 6.9 19.8 14.0 0.7 1.3 3.1 24.9 2 U 45.3 14.2 1.3 0.4 5.3 Vida sativa 13.8 21.8 45.1 14.4 1.1 3.6 9.3 4.2 10.9 11.8 2.2 0.4 0.4 2.0 Vide tetrasperma

Table 2.1.d Seasonal variation in the percentage cover for each species of legume on sites of different 3uccessional age (1977 & 1971). 39 generally more evident in younger sites with cover reaching 20% and 42% respectively. V. hirsuta and L. corniculatus were common to seven sites, while M. lupulina and T. dubium occurred in five sites. The percentage cover for V. hirsuta and M. lupulina reached a maximum of 45% and 44% respectively, although both were minor components of the flora in three of the sites in which they were found. L. corniculatus only reached 16% cover in one site and like T. dubium never exceeded 6% in the other sites. For the remaining four species, cover was always below 2% and the distribution limited to three sites for T. hybridum, two sites for L. uliginosus, while M. sativa and V. tetrasperma were found in a single (but different) site. There was no evidence of the expected trend of annuals being more prevalent in younger sites and perennials in older sites.

2.4,3. Grass/Forb and Legume/Forb Indices These are tabulated in Table 2.2. In the 1984 site, the forbes (or herbs) were a dominant component of the vegetation with very little contribution from the grasses, although more grasses colonized later in the season. The proportion of legumes in the 1984 site was also small with peak contributions in July and October. In later sites grasses increased in cover and this was particularly noticeable at the beginning and end of the season; a trend which corresponds to their phenology of early flowering and late seasonal regrowth. However, in the 1982 site there was a steady increase in the grass/forb index between October 1983 and October 1984 and a corresponding decrease in the proportion of legumes. The frequency of forbes increased during the mid-season and equals the grasses in the 1983 and 1981 sites, but exceeded them in the 1980 and 1979 sites. The proportion of legumes contributing to the forbes in the 1981 site was low and followed a similar trend to the 1984 site, while in the 1983, 1980 and 1979 sites between 40-60% of the forbes were legumes (in all but 2 samples).

2.4.4. Plant Structure Height profiles are used to describe the structural complexity of the vegetation in the vertical plane and display both seasonal and successional changes (Figures 2.2. & 2.3.). In the 1984 site (Figure 2.2), the increase in height and stratification was initially rapid but slowed down by July with the development of the flowering canopy. A reduction of the basal part of the profile then occurred as leaves were lost as a result of shading and senescence. The pattern was similar in August and September, but by October the upper canopy diminished due to senescence of the flowering stems and the basal structure 40

1983 1984 Site October May June July August September October

_ 5.77X10"3 0.03 0.03 0.04 0.96 0.19 1984 - (0.038) (0.089) (0.157) (0.048) (0.076) (0.112) 1.10 1.86 1.21 0.95 0.75 1.04 1.08 1983 <0.285) (0.678) (0597) (0526) (0550) (0523) (0516) 1.0S 1.32 150 1.44 1.63 1.95 2.09 1982 (0.325) (0504) (0.430) (0503) (0557) (0.218) (0.104) 1.58 1.71 1.13 0.94 0.97 1.08 1.15 1981 (0.041) (0.068) (0.181) (0.017) (0.008) (0.096) (0.150) 1.09 0.91 059 054 051 0.83 0.96 1980 (0.186) <0.269) (0.441) (0.414) <0.402 ) (0.420) (0.418) 0.88 1.06 0.64 059 0.70 1.11 152 1979 (0.279) (0.409) (0583) (0572) (0530) (0.497) (0504) 1.19 158 0.90 1.30 1.49 1.65 150 1977 (0577) (0.483) (0.606) (0.444) (0511) (0541) (0508) 0.99 157 0.88 1.40 1.17 1.02 1.11 1971 (0.424) (0.491) (0511) (0547) (0557) (0.248) (0.288)

Table 25. Grass/Forb Index and Legume/Forb Index In parentb.esLa for 7 sampling1 occasions in 1983/4 in sites of different successional age. E u Legumes

-C O) August September October X’«

L Touches in each height category

Figure 2.2. Height profiles of the total vegetation and legumes for 6 sampling occasions during the season in the 1984 site. Height Category icm) iue 3. egt rfls f h ttl eeain n lgms o My n Jl 18 i sts f ifr0 scesoa g. aiu egt is height Maximum age. successional differ®0! of sites in 1984 July and May for legumes and vegetation total fhe of profiles Height . .3 2 Figure niae we i eces h scale. the exceeds it when indicated t-0 iP- 43 increased, as grasses and perennial herbs became established resulting in a profile similar to June. The contribution of the legumes to the vegetation structure in the 1984 site is also illustrated as height profiles in Figure 2.2. The profiles show a height increase similar to the total vegetation and indicate that there was a major contribution by the legumes to the upper height categories between May and July. By August, height declined and structural complexity was reduced due to senescence (mainly of Vicia species). In September and October vegetative growth (mainly in the Trifolium species) and autumn germination of the Vicia species, increased the contribution to the lower part of the profile. In the older sites height profiles are only given for the beginning of the growth period (May) and for the period of most active growth (July/August), (Figure 2.3.). Both height and structural stratification increased during this period, although even at the beginning of the season the vegetative structure was complex compared to the 1984 site. The 1983 site had very little structure above 10 cm in May, since the majority of plants only became established in the autumn of the previous year. By July/ August the basal structure had declined. However, the lowest height class always appeared to be under-represented, especially in the older sites, and was probably related to the development of a litter layer. In the 1977 site, saplings were establishing and, as a result, the height of the vegetation was substantially greater than other sites. The legumes followed the same general trend as the total vegetation with increased height and stratification between May and July/August. In particular, the legumes tended to be a major component of the upper height categories, especially in July/August in most sites. However, it must be remembered that each legume height profile was based on the summation of data for all legume species, and differences that occurred may well be due to individual species.

2.4.5. Pattern of Distribution It can be seen from Table 2.3. that legumes occur in a large proportion of the subplots, although their distribution varies during the season in different sites. Heterogeneity appeared to be negatively correlated to percentage cover, so that when cover was high, heterogeneity was low and vice-versa. The distribution was expectably heterogeneous at the beginning of the season in the 1984 site, but decreased in July and this was coincident with peak legume cover. Subsequently, during legume senescence, heterogeneity increased, although over 70% of the subplots contained legumes. By contrast, the legumes were more homogeneously distributed (especially towards the end of the season) in the 1983 site, even though the number of subplots containing legumes between June and October was similar. Both cover and the frequency of subplots occupied by legumes 44

C sec p<*jye 33 ) Table 2.3.a Pattern of distribution/^of the legume herbs and, In parenthesis, the proportion of sub-plots in which they occurred for each sampling occasion in 1983/1984 in sites of different successional age <1984 & 1983). 45

( s e e . Table 2.3J) Pattern of dlstribution^of the legume herbs and, in parenthesis, the proportion of sub-plots in which they occurred for each sampling occasion in 1983/1984 in sites of different successional age (1982 & 1981). 46

sub-plots In which they occurred for each sanpling occasion in 1983/1984 in sites of different successional age (1980 & 1979). 47

(see po^e "ilO Table 2.3.d Pattern of distributionAof the legume herbs and, in parenthesis, the proportion of sub-plots In which they occurred for each sampling occasion in 1983/1984 in sites of different successional age (1977 & 1971). 48 decreased during the season in the 1982 site, but this decrease was not consistent throughout the site. In the older sites, distribution tended to be least heterogeneous in June, coincident with the peak in legume cover. However, the abundance of the legumes was much lower in the 1981 site and the degree of heterogeneity was much higher, especially in August when the distribution was highly clumped. This appears to be associated with the senescence of the early flowering annuals: a similar pattern occurred in the 1977 site. In the 1980 and 1979 sites, the distribution remained fairly homogeneous with between 78% and 100% of the subplots occupied by legumes during the season.

2.4.6. Taxonomic and Spatial Diversity There was a general increase in taxonomic and spatial diversity along the successional gradient. Both temporal and structural differences within a single season showed that diversity changed between sites and at different times during the year. The pattern of change in spatial and taxonomic diversity was similar but was relatively larger in the former. General successional trends in William’s a may be seen when values for the entire season are considered in Table 2.4. The high species turnover in the 1984 site gave rise to a diversity similar to the older sites. The early colonists in this site were annuals and these were replaced by biennials and even by perennials later in the season, thereby increasing the species richness of the site. The lowest diversity was recorded in the 1983 site, whereas in subsequent sites diversity progressively increased up to a maximum in the 1977 site. The diversity of the legumes in each site was 3 to 8-fold smaller and displayed some differences in rank to the remainder of the vegetation, although the 1977 site was the most diverse in both cases. However, the most significant difference was the maintenance of the high first year diversity (1984 site) into the second season (1983 site). The 1982 and 1981 sites were similarly characterized by low diversity. Changes in the taxonomic diversity (Figure 2.4.) of the vegetation during the season were most pronounced in the 1984 site where there was a sharp rise during May, although there was no further increase until the grasses and perennial herbs established in October. In contrast, the diversity of the legumes remained relatively stable. A similar pattern in legume diversity occurred in the four oldest sites and in each case there was no correlation with the diversity of the total vegetation. However, in the 1983, 1982 and 1981 sites, the legume diversity does follow a similar pattern to the total vegetation. The Legumes Total vegetation Site William1 s S.E. Will iam's S.E. a a 1984 6. 11 0.58 37.30 1.28 1983 6. 11 0.56 22.95 1. 10 1982 3.39 0.44 26. 27 1.36 1981 4.86 0.71 27. 48 1.29 1980 7.00 0.63 36.47 1.52 1979 5.79 0.53 43. 10 1.69 1977 8.27 0.73 50.91 1.92 1971 7.08 0.63 42.39 1.61

Table 2.4. Seasonal species diversity for the legumes and total vegetation from early to mid-successional sites. W illia m ’s a Diversity Index 10r i°r • 2 6 . 6 - 2 8 - 8 0 . 0 6r 8 0 • 0 4 . 4 0 * 0 4 . 4 6 » 6 4 • 4 2- 2- * Symbols: ■ ■ = Symbols: ewe Otbr 93 n Otbr 94 o sts f ifrn successionai age. different of sites for 1984 October and 1983 October between iue24 Cagsi h aooi dvriy f h ttl eeain n legumes and vegetation total the of diversity taxonomic inthe Changes Figure2.4.

93 94 93 1984 1983 1984 1983 c My ue uy u Sp Ot c My ue uy u Sp Oct Sept Auq July June May Oct Oct Sept Aug July June May Oct 11111 < f —0 1— oa vgtto; a=legumes. vegetation; total —■-f *— ----- 0 (g) 1977 1977 (g) (c) a 18 site 1984 (a) " 92 site 1982 v

ie (h) site 0 - b 18 site 1983 (b) d 1981 site (d) o ----- 1971 Site o— 0 50

51

Legumes Total vegetation Site Date of William's S.E. William's S.E. sample a index a index May 1984 2.91 1.04 7.80 0.81 June 5.66 0.77 28.29 1.50 July 10.80 1.00 50.77 1.96 1984 August 13.69 2.00 52.62 2.14 September 12.62 1.95 56.14 2.58 October 11.36 1.61 52.80 2.49 May 1984 4.56 0.82 16.80 1.29 1983 July 11.05 1.13 34.58 1.80 May 1984 5.11 1.04 23.48 1.79 1982 July 9.45 2.04 47.63 3.05 May 1984 3.36 0.97 21.32 1.51 1981 July to 45.35 2.59 May 1984 8.73 1.55 29.71 2.08 1980 July 12.36 1.19 54.14 2.64 May 1984 7.96 1.14 40.96 2.58 1979 July 11.87 1.15 70.51 3.38 May 1984 6.62 1.13 52.63 3.09 1977 July 16.31 2.03 105.08 4.71 May 1984 11.49 1.45 49.94 2.95 1971 July 9.00 1.10 70.22 3.33

Table 2.5. Spatial diversity of the legumes and total vegetation for different sampling occasions during the season in sites of different successional age. 52 sharp increase in August in the 1981 site may be anomalous, since legume abundance was very low at this time of year and consequently N may approach S giving rise to the high but very variable value for the index. The pattern of spatial diversity (Table 2.5.) for the 1984 site was similar for both the legumes and the total vegetation, but unlike the taxonomic diversity (Figure 2.4.a), continued to rise until July. In the other sites, spatial diversity was only compared at the beginning and middle of the season. However, there were similarities to the taxonomic diversity of the vegetation, since the 1983 site had the lowest diversity and the 1977 site the highest. The latter site also had the highest spatial diversity for the legumes.

2.5. Trends in Individual Species 2.5.1. Lotus corniculatus This species occurred in all sites except the 1981 site, although in all but the oldest site the distribution was rather patchy (Table 2.3.) and with cover (Figure 2.5., Table 2.1.) less than 5%. The importance values (Table 2.7.) were predictably low for this species, although there was an increase between May and late July/early August as new foliage developed from the overwintering rootstock. Cover in the younger sites showed no seasonal trend, although, in the 1971 site there was a gradual increase until August. Height profile data (Table 2.8.) showed that L. corniculatus reached a height of up to 60cm, but its contribution to the structural complexity of the vegetation varied between sites.

2.5.2. Lotus uliginosus Confined to the 1980 and 1971 sites, this species was very similar in appearance and growth form to L. corniculatus, but appeared to have a preference for damper situations. Consequently, the distribution within a site was patchy (Table 2.3.) and the contribution to the legume community small.

2.5.3. Medicago lupulina This species is an annual herb first appearing in the 1984 site in May and increasing in both cover (Figure 2.6.a, Table 2.1.) and importance (Table 2.6.) during the season. In the two oldest sites, it was only a minor component of the vegetation, but was much more abundant in the 1980 and 1979 sites. Here the cover (Figure 2.6.b,c) increased rapidly in Figure 2.5. Changes in the percentage cover of of cover percentage the in Changes 2.5. Figure October 1984 in sites of different successional age. successional different of sites in 1984 October % Cover 20 20 20 20 0 0 0 0 93 1984 1983 c My ue uy u Sp Oct Sept Aug July June May Oct fr — rffrr -- A 7" ______* r—— — ______A cul us tu la u ic n r o c s u t o L t 1 between October 1983 and and 1983 October between b 18 site 1980(b) (a). site 1983 c 17 site 1979 (c) d 17 site 1977(d) e 17 site 1971 (e) Q ------° ° ■ 53 Nay June July August September October *Imp. Rank Imp. Rank Imp. Rank Imp. Rank Imp. Rank Imp. Rank Value Value Value Value Value Value Raphauus raphanistrum 21.4 5 25.1 3 Stellaria media 21.1 3 Spergula arvensis 72.5 1 98.7 1 94.1 1 84.4 1 79.9 1 69.1 1 Cbenopodium album 38.8 2 Polygonum aviculare 26.2 5 17.4 5 25.8 4 26.8 2 25.3 2 Polygonum perslcarla 33.8 4 30.1 2 30.8 2 28.5 2 24.4 4 24.0 4 Fallopia convolvulus 36.3 3 Tripleurospermum inodorum 19.7 5 23.1 4 27.1 3 24.7 3 22.5 5

Medic a go lupulina 1.3 18 1.7 19 4.9 15 5.0 14 6.5 13 7.3 11 Trifolium dubium T. bybridum 1.7 19 2.2 23 2.7 20 2.6 18 T. pratense 0.3 32 0.7 27 0.7 30 1.2 25 2.0 22 T. repens 1.0 24 1.8 16 2.1 24 1.4 23 2.3 19 Lotus cornlculatus 0.2 40 Vida hirsuta 8.0 10 20.7 4 30.1 3 18.4 5 21.4 6 17.1 7 V. satlva 3.6 13 6.8 12 0.7 27 0.4 34 0.2 36 0.2 41 V. tetraspermum 0.6 32 Number of species 21 41 40 42 40 50 in each sample

Table 2.6. ^Importance values for the five top ranking plant species plus all legume species for 6 sampling occasions during the season for the 1984 site. 1083 s ite 1082 :*ltr 1081 s ite 1080site 1070site 1077 site 1971 site

Date of sample In 1084 May July May July May July May July May July May July May July

«Imp. lap. Imp. Imp. tlmp. Imp. Imp. Imp. »Imp. Imp. Imp. Imp. *Imp. Imp. Plant Species value rank value ran k value rank value rank value rank value ran k value rank value rank value rank value ran k value ran k value rank value rank value rank

Equlaetum paluatre 33.8 4 Ranunculus repens 30.0 2 20.5 5 Stellarla gramines 20.17 4 31.0 4 24.3 5 20.5 5 23.1 5 Plaotago lanceolata 10.0 5 26.8 3 41.32 3 60.8 2 Achillea alllefollum 18.03 5 30.7 5 Clralum arveaae 38.0 3 37.8 3 26.1 5 Hypochaeria radtcata 15.3 5 21.7 5 L uiula app 20.5 3 42.0 i Poa pratcusts 30.8 2 Poa trlvlalla 16.4 3 Dactylla gloaerata 53.3 2 41.3 3 25.11 « 25.1 3 Bromus a ter ilia 28.5 3 Agropyroo rcpcna 42.2 2 Arrhcnatherum elatlua 28.7 2 Holcua lanatua 103.0 1 30.4 3 112.0 1 41.6 2 127.50 1 46.3 3 53.5 1 34.0 4 55.8 1 45.2 1 65.2 1 57.0 2 40.7 1 23.8 4 Uolcua moil la 23.6 4 Agroatla caplllarla 83.8 2 102.5 1 37.7 3 108.3 1 50.02 2 60.0 1 34.8 4 55.5 1 42.35 2 41.8 3 39.1 3 60.8 1 Med lea go lupullna 20.0 6 55.4 2 18.58 6 42.4 2 0.0 28 2.0 21 0.7 25 Trifolium dublum 0.7 17 1.08 21 0.7 31 T. hybriduB 1.4 13 T. p rateoa e 34.2 3 45.5 2 10.8 6 10.6 5 0.0 16 0.0 10 2.3 15 1.0 24 0.5 20 T. repcna 18.8 4 20.8 6 20.5 4 14.3 6 0.0 20 0.0 10 0.8 21 0.74 25 0.8 23 1.5 23 16.5 9 13.4 10 Lotus corniculatua 2.7 12 3.5 10 1.2 17 1.3 19 1.4 20 1.6 17 6.1 13 5.1 14 12.1 11 L. ullginosua 0.8 21 Vida hirsute 2.1 12 4.1 8 0.0 16 0.0 10 3.4 12 14.93 8 28.6 4 16.1 8 16.6 4 25.1 5 10.7 7 V. a a ttv a 1.4 14 12.8 6 13.8 7 14.5 7 28.44 4 16.6 7 21.0 4 12.6 3 16.1 10 2.7 21 Number of species 16 17 18 17 15 22 22 26 25 32 31 37 20 31 In each sample

Table 2.7. ^Importance values for the five top ranking plant species plus all legume species for 2 sampling occasions in sites of different successional age cn cn October 1984 in sites of different successional age. successional different of sites in 1984 October Figure Figure

2.6. % Cover Changes in the percentage cover of of cover percentage the in Changes 1984 Medicago lupulina Medicago between October 1983 and 1983 October between a 18 site 1984 (a) 56 57

1984 1983 1982 1981 1980 1979 1977 1971 site site site site site site site site

Height Category Hay Inn Jnl Ang Sep Oct Kay Jnl Hay fnl Hay M Hay Jal fay ini [ay Jnl Hay Jnl (ci)

o - t o 0.9 0.6 0.3 0.5 1.2 0.5 1.3 Lotas 11-30 0.2 0.1 0.1 0.3 1.4 0.6 3.0 cortical atos 31-60 0.1 0.1 >60 0-10 Lotas 11-30 0.1 alifitosas 31-60 0.2 >60

0-10 0.3 0.3 0.4 1.2 1.9 3.7 3.5 2.6 3.9 1.4 0.2 teiicago 11-30 0.7 0.6 1.7 1.0 2.6 14.4 1.8 10.7 0.1 1 opal Its 31-60 0.2 0.1 0.1 6.4 8.7 0.1 0.3 >60 0.3

o - t o 0.5 Trifolios 11-30 0.1 dabias 31-60 >60

0-10 0.1 0.1 0.1 0.5 Trifolias l t -30 0.03 0.1 0.1 kybridias 31-60 0.1 >60

o - t o 0.1 0.2 0.1 0.3 8.8 1.2 3.2 0.9 0.1 0.1 0.3 Tr ifolios 11-30 0.1 0.3 0.1 1.0 5.2 1.3 2.3 0.1 0.1 0.2 0.1 pratetse 31-60 0.1 9.4 1.2 0.3 0.1 >60 1.2

0-10 0.04 0. 1 0.2 0.4 3.2 1.5 6.0 2.0 0.1 0.2 2.8 1.8 Tr Ifolios 11-30 0.1 0.1 0.1 2.0 0.2 1.3 0.1 0.1 0.2 0.1 1.2 1.8 repeas 31-60 0.1 >60

o - t o 2.5 1.3 0.7 0 . 1 0. 1 0.7 0.1 0.6 0.1 1.9 0.1 2.2 Vida l t -30 5.8 6.7 0.7 0.5 0.4 0.1 0.9 0.1 2.6 1.9 2.3 1.8 7.9 2.3 kiroots 31-60 1.8 4.2 0.2 0.1 0.4 0.2 0.2 1.0 6.7 3.9 0.1 5.8 >60 0.3 0.7 0.4

0-10 0.5 1.8 0.3 1.0 1.0 0.1 3.2 0.3 4.9 0.1 2.5 Vida 11-30 0.5 0.1 0.1 1.2 2.8 0.6 5.6 1.2 1.7 0.5 2.2 0.3 sativa 31-60 0.4 1.7 0.7 1.5 1.2 0.4 >60 0.3 0.1 0.5 0-10 Vida l t -30 0.1 tetraspersa 31-60 0.1 >60

Table 2.8. Percentage of total touches in each, height category for each species of legume on six sampling occasions for the 1984- site and two sampling occasions for all other successionai sites. 58

May and June, up to a peak of around 45% in late July/early August. At this time the species was most homogeneous (Table 2.3.) with over 70% of the subplots occupied in the 1980 site and between 78% and 96% in the 1979 site. As expected, the importance values (Table 2.7.) also showed a dramatic change between May and the end of July with over a 2-fold increase. The July values were among the highest recorded for the legumes and this was further highlighted by the species being the second most important contributor to the entire plant community. The structural profile data (Table 2.8.) showed that the height and structural complexity were low at the beginning of the season, a feature most obvious in the 1984 site, where the plants established much later from seedlings. By July/August, plants generally reached over 60cm in height, although in the 1984 site they were much smaller and contributed 4.5% of the total plant structure between 0-30cm. By contrast, in the 1980 and 1979 sites, M. lupulina contributed between 6-9% and 10-15% to the total plant stmcture in the 10-30cm and 31-60cm height classes respectively. This represented the largest contribution made by a legume to the structural complexity of a site.

2.5.4. Medicago sativa This species was recorded in one subplot in the 1977 site.

2.5.5. Trifolium dubium The species was the only annual representative of this genus, but was found in a wide range of sites (i.e. 1983, 1980, 1979, 1977 and 1971 sites). However, the species only occurred in a few subplots in each site (Table 2.3.). These small patches of T. dubium reached peak cover (Table 2.1.) in June, but in all cases the relative contribution to the legume community was low.

2.5.6. Trifolium hybridum This perennial species occurred in the 1984, 1983 and 1979 sites, but had a very patchy distribution (Table 2.3.) with low cover (Table 2.1.) and importance values (Table 2.7.). The species had a very similar growth form to T. pratense, and this was reflected by the structural profiles (Table 2.8.). However, the contribution made to the structural complexity of the vegetation was consistently low. 59

2.5.7. Trifolium pratense T. pratense an early colonizing perennial, first appeared in June as seedlings in the 1984 site. The pattern of cover (Figure 2.7.a-c) was similar to T. repens in the 1984-1982 sites and the high importance values (Table 2.7.) in the 1983 site showed that T. pratense became a dominant component of the vegetation, with a fairly homogeneous distribution (Table 2.3.). These attributes decline in the third year (1982 site) and, in the older sites, the abundance remained low (Table 2.1.) and the distribution highly clumped (Table 2.3.). The growth form at the beginning of the season was a compact rosette and consequently the plant structure was concentrated near ground level. However, later in the season, most of the structure occurred between 10-60cm above ground level (Table 2.8.), due to the extension of flowering shoots. In the 1983 site, the species structure was distributed over an even greater height range with contributions of over 5% and 9% to the total plant structure in the 10-30cm and 30-60cm height classes respectively.

2.5.8. Trifolium repens This is a perennial species found in every site, although it was more abundant in the 1983, 1982 and 1971 sites. It appeared very early in succession, being first detected by sampling, in the 1984 site in June. Subsequently, cover (Figure 2.8.a) and patch size (Table 2.3.) increased until mid-July and again in mid-September/ October. Plants in the 1983 site increased their cover from 10% in May to a maximum of 20% by August (Figure 2.8.b) with up to 60% of the subplots occupied (Table 2.3.). T. repens ranked among the top 6 contributors to the vegetation in terms of importance value (Table 2.7.) in this site. In contrast, T. repens was most abundant in May in the 1982 site, but thereafter cover and overall contribution (as assessed by importance values) decreased during the season and, as a result, heterogeneity increased. By October, the frequency of subplots occupied had fallen by 75% (Table 2.3.), indicating substantial mortality amongst the plants. These 3 sites each describe a phase in the population dynamics of T. repens: (i) colonization from the seedbank; (ii) population increase by vegetative growth; and (iii) population decline. The decline in the 1983 site coincided with an increase in other herbs and grasses (Table 2.2.). In later sites, T. repens persisted at very low levels with relatively few individuals surviving. The 1971 site was somewhat of an anomaly, since T. repens occurred in over 50% of the subplots (Table 2.3.) and cover reached a maximum of 20% (Figure 2.8.d) in July. The importance values showed that this species ranked among the top 10 in the site. Figure 2.7. Changes in the percentage cover of of cover percentage the in Changes 2.7. Figure October 1984 in sites of different successional age. successional different of sites 1984in October

% Cover 2.0 40 60 20 20 40 0 0 0 93 1984 1983 c May Oct x ue July June Trifolium pratense Trifolium * » u et Oct Sept Aug between October 1983 and 1983 October between a 18 site 1984 (a) (b) 1983 site 1983 (b) (c) 1982 site 1982 (c) j 60 61

2.Or (a) 1984 site

0L

(b) 1983 site

(d) 1971 site □

a

0 L ------Oct »------May 1------June 1------July 1------1 Aug ------Sept :------Oct 1983 1984

Figure 2.8. Changes in the percentage cover of Trifolium repens between October 1983 and October 1984 in sites of different successions! age. 62

The contribution of T. repens to the structural complexity of the vegetation (Table 2.8.) was highest in the 1983, 1982 and 1971 sites. As a result of the prostrate growth form, most of the species contribution was to the lower height classes (i.e. up to 10cm above the ground surface). However, later in the season there was some contribution to the higher classes due to the extension of leaf petioles and flowering stems.

2.5.9. Vicia hirsuta A widespread, early-flowering annual which was very similar in habit and growth-form to V. sativa. Unlike its congener, it occurred abundantly in the 1984 site, appearing as seedlings in late April/early May and reaching a peak cover of over 30% in mid-July, 2-3 weeks later than in other sites (Figure 2.9.). During this period, the species was the most abundant legume in this site, and as the importance values show (Table 2.6.), the third most abundant plant species. It was also highly homogeneous (Table 2.3.), occurring in up to 98% of the subplots. There was a similar general pattern of cover in the older sites, although cover, overall importance and homogeneity tended to increase with site age. However, the cover and mid-season importance values in the 1971 site decreased earlier and more abruptly than in other sites suggesting an earlier senescence. The height profile data (Table 2.8.) show that in older sites, V. hirsuta made a larger contribution to the 30-60cm height class during late July/early August and may even exceed 60cm in height. In the 1984 site, where plant structure was monitored throughout the season, the height profile data reflected seed germination (late April/early May), rapid growth (June), flowering (July), senescence (August) and the germination of the second cohort (October). In contrast to other sites, the largest contribution (5-7%) to the total structure was made to the 10-30cm height class during June and July, although in July there was just over 4% in the 30-60cm height class and plants reached over 60cm in height.

2.5.10. Vicia sativa A common early-flowering annual which occurred in every site. During the season, cover (Figure 2.10.) and spatial heterogeneity (Table 2.3.) increased up to a maximum in each site in late June/early July, followed by a rapid decline due to plant senescence after reproduction. However, at the end of the season, there was another increase as new individuals established. This general pattern occurred in every site, although the younger sites were characterized by lower importance values (Table 2.7.) and cover (and highest Figurethein Changes2.9. percentage cover of 1984 in sitesof different successional age. 40 r 20

% Cover "I [ ” 40 20 0 0L 0 L 0 0L 0 93 1984 1983 c My ue uy u Sp Oct Sept Aug July June May Oct _ ___ o- o Viciahirsuta between October 1983betweenOctober andOctober ‘fti c 18 site 1981 (c) (a) 1984 site 1984 (a) e 17 site 1979 (e) (d) g 17 site 1971 (g) 1980 a . a i Site 63 % Cover 20 40r 0 20- Figure 2.10. Changes in the percentage cover of thecoverof percentage in Changes 2.10. Figure 20r 1984 in sites of successionalage.different 1984of sites in 0 0 93 1984 1983 c My ue uy Aug July June May Oct ■■ i ■ i # —* # i # i i ■ > i ■ ■ » Vida sativa Vida JL between October 1983and October October between (a) 1984 site 1984 (a) b 18 site 1983 (b) c 18 site 1982 (c) d 18 site 1981 (d) et Oct Sept h 17 site 1971 (h) 64 65 heterogeneity). As a consequence of lower cover in these sites, the percentage contribution to each height class (Table 2.8.) was low and the total height attained was only 30cm, while in the older sites the species was more abundant and reached over 60cm by late July. At this time, the percentage contribution to the height profile was greatest in the 30-60cm height class as a result of the large number of reproductive structures. In the 1971 site, trends tend to be intermediate between the younger and older sites in terms of cover, importance, pattern and plant structure

2.5.22. Vida tetrasperma This annual species was almost indistinguishable from V. hirsuta in the vegetative stage and was only found in the 1984 site, where it was very restricted in distribution and made only a small contribution to the legume community.

2.6.Discussion Variation in community structure of the legumes in sites of different successional age can be attributed to the individual life-history strategies of each species since it is these properties which determine the part which each species plays in the community (Harper 1977). However, the legumes are part of a larger plant community and changes within the whole community will effect the legumes. There are two main factors which determine the composition of each plant community, these are: neighbouring species and site characteristics (Turkington & Cavers 1979). Neighbouring species mediate their influence through the effects of inter- and intraspecific competition. Thus, any two plants with precisely the same requirements will tend to exclude one another, while plants with different requirements can coexist. What determines when a species can enter the community, initially, depends on the germination from the seedbank. This in turn depends on the availability and frequency of "safe sites" i.e. locations that provide the precise conditions for a particular seed to germinate (Harper 1977) as well as the composition of the surrounding species. Only a small proportion of the seedbank germinates at any one time and many seedlings that emerge perish due to biotic and abiotic factors. In the ruderal site, ploughing produced an enormous amount of "safe sites" for many species. Many of the species that emerge are dependent on severe disturbances for colonization. For example, Vida hirsuta is well suited to this type of habitat because of a persistent, well-dispersed seedbank, rapid rate of increase and an annual life cycle. 66

Indeed, much of the overall legume abundance and pattern in this site can be credited to V. hirsuta and, as a result of the lower levels of interspecific competition, this legume quickly became a dominant component of the vegetation. However, V. sativa and V. tetrasperma have similar characteristics, but were not as abundant in the 1984 site which indicates either a difference in the germination requirements and therefore lack of "safe sites" or a smaller seedbank. In the older sites, the annual legumes (M. lupulina, T. dubium, V. hirsuta and V. sativa) made considerable contributions to the legume flora, demonstrating the appearance of "safe sites" each year. Such sites for colonization commonly occur as gaps in the canopy (Grime 1979) and these coincide with the usual late summer/early spring germination of these annuals. Similar patterns of regeneration occur in many pasture grasses (e.g. Arrhenatherum elatius (L.) Beauv. ex. J.& C. Presl. , Bromus mollis L., Bromus sterilis L., Lolium perenne L.) in Europe (Grime 1979). Many are either annuals or short-lived perennials and lack a persistent seedbank. Parvone and Reader (1982,1985) demonstrated that it was the differential survival of seedlings and mature plants of M. lupulina which accounted for the change in density with microtopography and not differential reproduction and seedling recruitment. Exploitation of (large) gaps was also mentioned by Grime (1979), who recognized that relaxation of stress (i.e. detrimental abiotic and biotic factors) increased gap size and resulted in eruptions of this "competitive ruderal" (Grime 1979). The occurrence of "eruptions" of M. lupulina in the 1979 and 1980 sites is therefore suggestive of many large favourable gaps. These were commonly produced by the die back of other annuals (e.g. B. mollis, B. sterilis, Vida species). In the other sites, the abundance of M. lupulina was low and resembled T. dubium, a feature correlated with the lower frequency of gaps suitable for colonization and survival. V. hirsuta and V. sativa can also be termed "competitive ruderals", since they persist in localized areas where competition from perennials is temporarily reduced. Both have a similar growth form to Galium aparine L., another "competitive ruderal" referred to by Grime (1979). This annual has similar germination periods, long straggling stems and leaves that have a mechanism (though not the same as the Vida species) of clinging onto the surrounding vegetation so that individuals can scramble over neighbouring perennials, thereby reducing competition by occupying the upper levels of the canopy. The large seed size of the Vida species (Brown et al. 1987) may also allow rapid germination and establishment by giving each seedling extra energy reserves to germinate quickly and to colonize smaller gaps than M. lupulina and T. dubium. 67

The perennial legumes, L. corniculatus, L. uliginosus, M. sativa, T. hybridum, T. pratense and T. repens do not rely on the annual creation of gaps in the vegetation. Although individual plants in each site may well be the survivors of a single act of colonization several years ago, since recruitment of new individuals (or genets for those with clonal growth) can only be by seed germination and this event is rare. Cahn and Harper (1976a), Turkington (1985) and Turkington et al. (1979) found seedlings of T. repens in their study areas but survival was very low and confined to local patches of disturbance such as molehills and patches of dung. In this study, seedlings of both T. pratense and T. repens were only observed in the 1983 site, but no plants established from them. Poor recruitment of individuals also occurs in other perennial herbs e.g. Plantago L. species (Sagar & Harper 1960), Rumex acetosella L. and Rumex acetosa L. (Putwain & Harper 1968). L. corniculatus, L. uliginosus and M. sativa tend to occur in the older sites, although the latter two species were very local and no discernible trends were apparent. However, L. corniculatus was more widespread and tended to be restricted to well-lit areas (Jones & Turkington 1986). The growth form is a rosette with a well developed tap-root and ability for clonal growth. Grime (1979) described L. corniculatus as a "competitive stress-tolerant raderal" strategist. Such species are characteristic of unproductive grasslands in the British Isles and would relate well with the distribution on the study sites described here. The foliage reaches over 30cm in height (Table 2.8.) allowing some persistence amongst taller species, although it is known to be competitively inferior to A. elatius, D. glomerata and P. pratensis (Jones & Turkington 1986) as well as the legumes, M. sativa and T. pratense (Turkington & Cavers 1979). However, the competitive ability does depend greatly on conditions (Jones & Turkington 1986). The negative association of L. corniculatus with T. pratense may explain the low frequency of the former species in the 1983 site (Turkington & Cavers 1979). However, the taproot enables the plant to exploit reserves of moisture, inaccessible to grasses and other shallow rooted plants (Walter 1973), and thus competitive advantage might result during periods of water stress. r. hybridum, T. pratense and T. repens grow more slowly than the annual legumes and only contribute significantly to community structure and cover in the second and third years of colonization of bare soil. If the vegetation is left undisturbed for several years, these "ruderal perennials" (Grime 1979) are progressively excluded. The oldest site was an exception to this pattern, since the abundance of T. repens resembled that of the 1983 site. This discrepancy can be accounted for by the greater levels of disturbance as a result of rabbit grazing. Pickett (1982) found, in his 20 year study of old field succession, that T. pratense was the only perennial herb to go extinct. This is a possibility in the current 68 study since the older sites have very small populations. Unlike T. repens lateral spread by clonal growth was limited, hence, T. hybridum and T. pratense are forced to live and die in the same lateral relationship with their neighbours as that in which they were first established (Turkington etal. 1977). As a consequence, plants are likely to develop positive and negative associations with their neighbours. Aarssen et al. (1979) defined 3 categories of stability, these are: consistent associations; seasonal associations; and temporary associations. Consistent associations evolve to reduce interspecific competition between species, such as utilizing the soil environment more effectively as suggested by Turkington et al. (1977) and Salisbury (1952). Turkington and Harper (1979a) demonstrated the differential selection pressure by different grass neighbours on genetic individuals of T. repens at a local scale. Further studies on M. sativa and T. repens (Turkington 1979, Turkington & Harper 1979a) concluded that if a plant has a history of growth and selection in the presence of "a", it will survive and grow best when grown in the presence of "a" (Turkington 1979). Such microevolutionary processes must be rapid since Turkington (1979) showed that after 10 years, at least some individuals of M. sativa and T. repens became adapted to the differential selection pressure from their neighbours. In this study, such selection will only be evident in the two oldest sites, however, negative associations may quickly develop on younger sites especially when interspecific competition between species is intense. This may lead to spatial separation, especially when there is a similarity of growth form e.g. M. lupulina, M. sativa, T. hybridum and T. pratense (Turkington 1979, Turkington & Cavers 1979), or superior competitive ability e.g. T. pratense and L. corniculatus (Jones & Turkington 1986). However, in these studies, T. pratense/T. repens (1983 site), T. repensfVicia species (1971 site) and V. hirsutafV. sativa (1980, 1979 and 1971 sites) overlap in their distribution in these sites (Figures 2.7, 2.8, 2.9 & 2.10). Turkington and Cavers (1979) postulated that T. repens can cohabit with other legume species (e.g. T. pratense) by virtue of its stoloniferous guerilla-type growth form (i.e. more contacts with its neighbours than itself). The association with the clump-forming T. pratense in the 1983 site tends to weaken later in the season as T. pratense becomes more evenly distributed and T. repens more aggregated (Table 2.3.). T. pratense is unlike T. repens, since it shows what Harper (1977) describes a "diversity of age states" i.e. its growth form varies with age. At the beginning of the season the height of the foliage of both species was similar (Table 2.8.) so that T. repens can invade clumps of T. pratense. As the season progresses the foliage of T. pratense reaches 85cm and the formation of a large stand in the 1983 site must increase the competition with T. repens (and L. corniculatus), which has a lower canopy but high light requirements (Burdon 1983). 69

Hence, there is a temporal separation in peak cover. T. repens and Vicia species have completely different growth forms so spatial separation results, the former being more prevalent in short vegetation. However, the ability of V. hirsuta and V. sativa to coexist without interference is far from clear, since both have synchronous life cycles and similar vine-like growth-forms. Both tend to make a greater percentage contribution to the upper part of the canopy (see Table 2.8.). In the case of the former, older leaves are shed as new ones are produced resulting in most of the plant material being in the upper region of the plant, whereas in the latter, older leaves are not shed as quickly resulting in a more evenly distributed canopy (Brown et al. 1987). It seems probable that subtle differences like this in the plant’s characteristics, as well as microsite requirements, will determine the spatial distribution between these two species within the community. Seasonal associations were also evident and it was recognized by Turkington and Harper (1979b) that species with seasonal growth can cohabit provided that their growth periods are asynchronous. Temporal separation could be identified between some of the legumes e.g. T. pratense/T. repens (1983 site), M. lupulina/ V. hirsuta (1979 and 1980 sites) and L. corniculatus/T. repens (1971 site). Asynchrony in growth between the legumes and other species have been studied, for instance, the temporal separation between T. repens and L. perenne has been well studied (Turkington & Harper 1979c) as well as T. repens and other grass species (Turkington & Harper 1979b). In these studies, a temporal separation between other legumes and grasses was indicated. The grass/forb index (Table 2.2.) showed that in the younger sites, the grasses tend to be more abundant at the beginning and end of the growing season, whereas several legumes ( Medicago, Trifolium and Vicia species) have their peak cover between June and August, precisely the time of low grass abundance. In the two oldest sites, grasses tend to dominate, although grass abundance remains negatively correlated with legume abundance. Thus, asynchrony in life cycles will allow potentially competing species to coexist. However, this coexistence may only be temporary, since in the 1982 site, the abundant legume community (consisting of mainly T. pratense and T. repens) declined during the season, while the grasses increased. It is known from pasture management that if regular defoliation (by grazing or mowing) does not occur, then T. repens will rapidly disappear and the same may well apply to T. pratense, although the decline in this species occurred later in the season (Figure 2.7.). Thus, "species associations with other species are predominantly loose and changeable, and community evolution is net-like in the sense that species are variously combined and recombined into communities in evolutionary time" (Whittaker 1975). Competition 70 between species therefore varies with changes in the composition of the plant community. Since some species are better competitors than others, domination by a few species will occur. This leads to three major categories based on abundance: dominant, intermediate and rare. The importance values (Tables 2.6. & 2.7.) show that the youngest sites have a few species which are dominant, although in the older sites, the composition of the dominant species changes and the level of dominance is reduced. The legumes tend to follow a similar pattern. To achieve dominance, a particular species must be more adept at capturing a large part of the limited resources available in a site. Intermediate species capture a similar fraction of the resources remaining and the rare species utilize what is left. This is the essence of the niche pre-emption theory of Whittaker (1965). However, the composition of the plant community is determined by its constituent species, so that as new species enter the community, changes in the competitive balance will occur. As a result, dominants in the community are progressively ousted by competitively superior species which in turn alter the interactions between other species leading to shifts in species’ associations and thereby causing the community to change gradually with time. Such changes in community development probably represent some of the forces that drive succession. Two community characteristics used here are taxonomic and spatial diversity, both increased with successional age. General increases in a diversity have also been found by Southwood et al. (1979) and in species diversity by Bazzaz (1975) during the same time scale. In contrast, the legumes do not follow the same general pattern, since taxonomic diversity remains relatively stable except in the 1982 and 1981 sites. Finally, site characteristics also help mould the plant community, mainly by the influence of abiotic factors, although previous land-use history does influence the community that develops through the impact on the seedbank. The seedbank in the 1981 site was very different from that found in other sites, since it had a lower proportion of annual species and higher proportion of grasses (Stinson 1983), thereby explaining the lower abundance of legumes. The 1971 site also had a different history but factors such as rabbit grazing and higher soil moisture content probably have an increasingly greater influence on the developing plant community. The effect of rabbit grazing depresses the structural profile of the vegetation (Figure 2.3.g) in a similar way to the sheep grazed paddocks in Gibson et al. (1987). The result is that low growing, grazing tolerant yet competitively inferior species such as L. corniculatus and T. repens can persist at higher levels of abundance than if the site was 71 ungrazed. Factors responsible for the more advanced phenology of V. hirsuta in the 1971 site were, however, less clear since grazing tends to reduce flower number and delay flowering (Crawley 1983). Successional age will also modify the community through an increase in more stable species associations and a decrease in temporary associations (Aarssen et al. 1979). This follows the assumptions of many ecologists that succession is a process of increasing stability, although the developing communities consist of constantly evolving components (Antonovic 1976). Thus, interactions between neighbouring species and site characteristics will help maintain small scale pattern and heterogeneity of the vegetation in the sites, of which the legumes are a major part. Chapter 3 Insect Herbivore Survey

3.1. Introduction

There are many studies of the insect fauna associated with one plant species (e.g. Davis 1973, Lawton 1976). However, there are far fewer studies whereby the insect fauna of a number of closely related plant species growing in close proximity have been monitored. The aims of this chapter, are to look at the insects actually present on different host plants by sampling individual plants and to relate this to subsequent studies on the performance of these plants under different levels of insect herbivore pressure. The sampling methods employed allow estimates of density to be made and from this the impact of the insect herbivores may be extrapolated. Literature surveys of the insect herbivores known to feed on a particular host plant were made (Appendix 2). However, unlike some studies (e.g. Lawton & Price 1979, Lawton & Schroder 1977, Southwood 1961a) which utilized literature records in place of field data, the records employed in this study were used as a reference to the possible herbivore species which might be encountered. There are dangers of relying solely on literature records because generally they are incomplete or even inaccurate. Deficiencies in the literature records were demonstrated by Compton (1983) in his studies of the insect herbivores on L. corniculatus. Although many of the herbivores encountered were known to feed on this legume, individuals of a number of widely polyphagous species were also found. Additionally, the host plants were sampled in only part of their geographic range and thus certain species may well be absent (Strong, Lawton & Southwood 1984). Unfortunately, the sampling methods employed here could not differentiate between those species feeding on the host plant and those merely resting on it. Thus, some of the polyphagous feeders that are potential herbivores may have been ignored. Southwood (1961b) noted that the herbivores present on a particular host plant represent only a fraction of the initial colonizers. Hence, each sample obtained is only representative of the herbivores present at a specific time. To reduce this error, samples were taken over a period of several months. By sampling plants growing in habitats of different successional age, changes in insect abundance and insect diversity during succession noted by Southwood et al. (1979) will be taken into account. 73

In this chapter, changes in the abundance and species composition of a range of insect herbivores, sampled from six species of leguminous herbs in field sites of different successional age, are reported. Temporal changes are considered in terms of bivariate and multivariate analyses, although for the two Vida species only two samples were taken and thus the data are merely tabulated.

3.2. Materials and Methods 32.1. Insect Sampling A "Univac portable insect suction sampler" (Burkhard Scientific (Sales) Ltd., Rickmans worth, Herts.) was used to sample insects from the above-ground structures of legumes. Although similar to the D-vac suction sampler, used by Southwood et al. (1979), the collection tube of the Univac has a smaller head area (8.3xl0-3 m2) enabling specific plants to be sampled. In this survey, 45 individual plants from each common legume species were sampled on each sampling occasion in each site. Wherever possible one plant was sampled from each of the 45 subplots, but since species were often heterogeneously distributed more than one sample had to be taken in some of the subplots. To simplify sorting, nine samples were taken successively without changing the collecting bag. In this way, there were five pooled samples for each species from each site on each sampling occasion. The collection tube of the Univac apparatus was held in position for 30 seconds, after which the area was quickly searched by eye for any insect herbivores not sucked up: these were mainly large lepidopteran larvae. Sampling was undertaken monthly between May and November, although, because of their short life span, the Vida species were sampled on only two occasions between late May and early July. Sampling on each occasion was carried out between 11.00 and 18.00, when the vegetation was completely dry. The suction sampling method was chosen because all the above ground parts of a plant could be sampled. Even so, the capture efficiency does vary with the height of the vegetation and more importantly with the insect group and sometimes species (Henderson & Whittaker 1977). Insects feeding internally on plant tissues (e.g. leaf miners) were not sampled, although evidence of their work was noted. However, suction sampling is generally effective, since it samples a defined area and is less prone to operator error than other methods (Henderson & Whittaker 1977). 74 32.2. Insect Sorting Samples were stored in a refrigerator while awaiting primary sorting from debris. This was done by hand in a glass-fronted screen illuminated by a bench lamp. Insects emerging from the sample debris were attracted by the light and were caught using a "pooter" (see Southwood 1978). Samples were killed and stored in 70% industrial methylated spirits to which a few drops of glycerol was added. All material was then sorted to the main phytophagous groups: Coleoptera, Hemiptera, Lepidoptera and Thysanoptera. Other minor groups (e.g. Sminthuridae), where herbivores of legumes are known to occur were ignored, since keys to the immature stages of these species are not available.

32.3. Insect Identification Lists of phytophagous insects, associated with individual legume species, were obtained from the literature (Appendix 2). A series of type-specimens were created from the samples with the help of Dr. A.C. Gange (Lepidoptera), Dr. P.S. Hyman (Curculionoidea), Dr. P. Kirby (Heteroptera) and Ms. J. Palmer (Thysanoptera). Generally, all species were identified using a Kyowa Optical stereomicroscope, although for the Thysanoptera slides of individuals were prepared in Berlese Fluid and examined under a Kyowa Optical compound microscope. Taxonomic works employed included Joy (1932), Mound et al. (1976), Southwood & Leston (1959) and Stroyan (1977, 1984). All aphids were confirmed by Dr. V. Eastop (British Museum (Natural History)) and Longitarsus Berthold species and Hypera Germar species by Dr. M. Cox (B.M.(N.H.).) The polyphagous Cicadellidae were not included, since it could not be ascertained whether specimens were associated with the legumes or neighbouring plant species.

3.3. Ordination

Ordination is a multivariate statistical technique which serves "to summarize community data by producing low dimensional ordination space (of typically one to three dimensions) in which similar species and samples are close together and dissimilar entities far apart" (Gauch 1982). Thus, environmental factors inherent within such data can be identified (Fishpool 1982). The ordination technique used in this section, Detrended Correspondence Analysis (DECORANA), was developed by Hill (1979) and is considered to be superior to other techniques. It is based on a simpler method of ordination, referred to as reciprocal averaging or R.A. (Gauch 1982, Hill 1973a), since the species scores are derived from averages of the sample scores and reciprocally, the sample scores 75 are averages of the species scores. These values are derived by simple matrix algebra and involve an iterative process of scaling (arbitrarily set from 0-100) and repeated cross-calibration. The iterative process is continued until the scores stabilize to a final unique solution, which is not dependent on the initial scores, although the number of iterations required to reach them is (Hill 1979). This method gives a one-dimensional ordination (first axis) where the species and stand scores are obtained simultaneously. Further axes are generated by taking out a regression on the final scores of the preceding axes; thus making the first axes uncorrelated to the second, the third to the first two and so on. The variance of each axis is expressed as an eigenvalue, which is the square of the correlation coefficient between the stand and species scores. Most of the variance (and therefore highest eigenvalue) is found in the first axis, and subsequent axes account for a decreasing proportion of the variance within the data matrix. DECORANA or D.C.A. can be thought of as a more refined R.A. ordination, whereby two main faults of the latter, which hinder the interpretation of results, are corrected. The first fault, which is considered by some to be the worst, is the "arch effect" (Gauch et al. 1977) or "horseshoe effect" (Kendall 1971). This arises because the second axis, which is uncorrelated to the first, is not independent. In fact, it is a quadratic function which may result in an interesting secondary gradient being deferred to higher axes which have greater independence from the first. The other main fault is that the scale of the first axis is not clearly defined, since there is a contraction of the ends of the axis relative to the middle. Elimination of the arch effect requires that higher axes have no systematic relation to the first. (Hill & Gauch 1980). The process of detrending is carried out by dividing the first axis into segments and adjusting the sample scores within each segment so that they have zero means. The detrended sample scores are then used to calculate new species scores by the same iterative process as R.A. When the scores stabilize the final sample scores are derived by taking the average of the species scores, but in this case the process of detrending is omitted. The solution of the above ordination gives rise to the second axis and a third is obtained by a similar detrending procedure, applied with respect to the second and to the first axes. Higher axes are generated in exactly the same way, but in each case detrending is applied with respect to each axes taken separately. This method has a slight drawback in that so-called interaction axes can occur where higher axes have a systematic relation with respect to the lower axes when taken together, although remain independent of each one when taken separately. Fortunately, the small eigenvalues of such axes generally render them unimportant with field data (Hill & Gauch 1980). 76

The problem of no clearly defined scale in the first axis is solved by a process called rescaling. It is based on the within-sample dispersion of the species scores and is achieved by expanding or compressing small segments along the species axis so that species turnover occurs at a constant rate along the whole length of the species gradient. This is achieved by trying to equalize the within-sample dispersion of the species scores at every point along the species axis. Only the species ordination is rescaled for technical reasons (see Hill 1979, Hill & Gauch 1980) and sample scores are derived by averaging the scores of the species that occur within them (as in R.A.). As a consequence of standard scaling, the within sample variance is set at a constant of one. This results in a root mean square species abundance profile having the same unit standard deviation. Thus, the unit length of the ordination maybe called average standard deviation of species or S.D. (Gauch 1982, Hill 1979). Characteristically, a full turnover of species composition of samples occurs in about 4 S.D. although a 50% change in sample composition can occur within 1 S.D. The advantage of a more meaningful axis scaling is that it defines gradient length and this is useful when comparing ordinations from different data sets. Since the two main faults inherent to R.A. also affect other ordination techniques such as polar ordination, principal components analysis, nonmetric multidimensional scaling, principal coordinates analysis, factor analysis and canonical correlation analysis, the use of D.C. A. in preference to these tests is therefore justified. However, there are a number of features of D.C.A. which warrant mention (Gauch 1982, Hill & Gauch 1980). First, outliers are a problem and are the result of species of rare occurrence or low abundance together with the samples which contain them. Second, discontinuities in the data if large (i.e. greater than 3 S.D.) will result in an overestimation of the gap width, although with smaller discontinuities the estimation of gap width is more reliable. The only way of dealing with outliers is to remove them, and with disjunct data subsets it is better to isolate them into more coherent subsets prior to ordination. Third, species ordinations are less reliable than sample ordinations because in particular samples there are species which have extreme or truncated distributions. Finally, van der Maarel (1980) recognized the problem of axis interpretation remaining unsolved.

3.4. Results

Species lists for insects sampled from six species of legume are given in Appendix 3. 77 3.4.1. General Trends The total number of insect herbivores sampled from each species of legume between May and November 1984 is given in Figure 3.1. The scale of the Y axis is logarithmic, since there was a large variation in the total insect load between species. Both Vicia species were only sampled on two occasions (between late May and early July) and as a result patterns could not be ascertained graphically; hence, these results are tabulated in Table 3. Generally, insect loads were higher in the younger sites, with the 1971 site consistently having the lowest densities of herbivores. The time of peak insect herbivore load may also be important particularly on the short-lived Vicia species, since it may probably be related to changes in plant phenology and growth stage. On V. hirsuta, there was very little difference in the total insect herbivore load between late May and late June/early July in the 1977 and 1981 sites, whereas, in the 1984 and 1971 sites higher insect loads were recorded in early July. By contrast, in the 1979 site, the highest insect load was recorded in early June. On V. sativa, insect load was highest in late May/early June in the 1981, 1979, 1977 and 1971 sites, although in the 1980 site, the highest load was recorded in early July. On L. corniculatus, insect abundance in the 1971 site remained fairly constant throughout the season, while in the 1983 site insect load increased between June and early August and thereafter remained relatively stable. M. lupulina was the only species which had a comparable insect load in the two sites where it was sampled and showed a general decline throughout the season. On T. pratense, the insect herbivore load was comparable in the two sites where it was sampled, but showed a decline in June in the 1982 site with an increase at the end of the season. By contrast in the 1983 site, insect abundance increased in the first half of June and thereafter remained relatively constant. Insect abundance on T. repens between May and September remained relatively constant in the 1971 site whereas in the 1982 and 1983 sites it increased during the season declining only in October. Differences in the insect load recorded between species were apparent. The overall levels of insect abundance increased in the following species in ascending order: V. sativa, V. hirsuta, M. lupulina, T. pratense, L. corniculatus with the highest levels of abundance being recorded on T. repens. However, there was much variation in insect abundance within a species between sites. Even so, one cannot ignore differences in insect abundance between different species, especially in sites of comparable age. 78

(a)

(d)

Figure 3.1. Total insect load recorded between May and November 1984 in sites of different successional age for (a) Lotus corniculatus, (b) Medicago lupulina, (c) Trifoliumpratense and (d) T. repens. Symbols: (•) 1984 site; (O) 1983 site; ( a ) 1982 site; (□ ) 1980 site; (a ) 1979 site; (■ ) 1971 site. Vicia hirsuta Vicia sativa 1984 site 1981 site 1979 site 1977 site 1971 site 1981 site 1980 site 1979 site 1977 site 1971 site Sampled 1984 4/7 30/5 30/6 4/6 3/7 31/5 1/7 6/6 4/7 31/5 30/6 6/6 3/7 4/6 3/7 31/5 2/7 6/6 7/7

Coleoptera - 1 8 2 17 - 6 - 5 3 17 - 12 4 12 1 21 - 4 -

Heteroptera - 2 19 16 97 38 44 29 16 19 85 21 74 86 125 48 85 12 32 -

Homoptera - 122 1 15 3 17 2 10 8 30 - 14 1 22 - 15 2 1 2 3

Lepidoptera --- - 1 - 1 ------1 -

Thysanoptera - 23 - 1 2 13 - 7 3 11 - 7 2 10 1 9 1 3 --

Total insect load - 148 28 34 120 68 53 46 32 63 102 42 89 122 138 73 109 16 39 3

In insect load - 5.00 3.33 3.54 4.79 4.22 3.97 3.83 3.47 4.14 4.62 3.74 4.49 4.80 4.93 4.29 4.69 2.77 3.66 1.10 Table 3. The density of insect herbivores (insects/ m2) recorded from five insect groups on Vicia hirsuta and Vicia sativa on two sampling occasions in sites of different successional age. 80 3,4.2. Changes in Insect Abundance and Species Composition within different Insect Groups The two Vicia species have been treated separately since they are short-lived and therefore more comparable with each other than with the other species sampled (Table 3.)* In the 1984 site, large numbers of Homoptera (mainly Acyrthosiphon pisum Harris) were recorded on V. hirsuta. Thysanoptera were also more abundant in this site than in other sites, whereas Coleoptera and Heteroptera occurred in much lower numbers. In general, heteropteran abundance decreased during the season on both Vicia species, while the abundance of Coleoptera, Homoptera and Thysanoptera increased. Lepidopteran abundance was very low on both species. Generally, no successional patterns could be seen in any insect group. Changes in the abundance of different insect groups during the season in sites of different successional age on L. corniculatus, M. lupulina, T. pratense and T. repens are shown in Figures 3.2.-3.7. There were a number of general trends within these groups. However, one striking feature is that much of the pattern of the total insect load on L. corniculatus, T. pratense and T. repens was due to the dominance of the Thysanoptera (Figure 3.7.), and as a result, trends that occurred in other insect groups were obscured. On M. lupulina, Thysanoptera only accounted for a small proportion of the total insect load and the Heteroptera were more important in terms of abundance (Figure 3.3). The most important hosts for the Coleoptera (mainly Curculionoidea) were L. corniculatus, T. pratense and T. repens particularly in the younger sites (Figure 3.2.). The 1971 site had the lowest abundance of Coleoptera and, as a result, indicated a trend for decreasing abundance with successional age. The highest coleopteran abundance was recorded on T. pratense with peak numbers reaching over 900 individuals per square metre in the 1982 site. The pattern of abundance of the Coleoptera on T. pratense and T. repens in the 1982 site was the same (Figure 3.2.). On M. lupulina, the abundance of Coleoptera was generally low but relatively constant throughout the sampling period. The pattern of abundance of the Heteroptera was similar for all sites and species of legume (Figure 3.3.). Abundance initially increased during May and June, peaked in early July and thereafter decreased sharply during the remainder of the season. There appears to be no relationship between successional age and abundance. Homopteran herbivore load (Figure 3.4.) was much lower than Coleoptera or Heteroptera. In contrast to the latter, however, there was a general increase in abundance during the season until October and November when numbers declined rapidly. On L. corniculatus, the highest abundance of the Homoptera (mainly aphids) was found in the 1971 site, although on M. lupulina, T. pratense and T. repens the homopteran load was Insect density (insects/m*) Lotus corniculatus, November 1984 in sites of different successional age for (a) for age successional different of sites in and 1984 May November between recorded Coleoptera of density The 3.2. Figure and (d) (d) and site; site. ( a ) 1982 site; (□ ) 1980 site; site; 1980 ) (□ site; 1982 Trifolium repens.

(b) Medicago lupulina,

Symbols: (•) 1984 site; (o) 1983 1983 (o) site; 1984 (•) Symbols: (b) ( (c) (d) a ) 1979 site; (■ ) 1971 ) (■ site; 1979

(c) Trifolium pratense

81 Insect density (insects/m2) Lotus corniculatus, Figure 3.3. The density of Heteroptera recorded between May and and May between recorded Heteroptera of density The 3.3. Figure November 1984 in sites of different successional age for (a) for age successional different of sites in 1984 November and(d) and(d) site; (±) 1982 site; (□ ) 1980 site; site; 1980 ) (□ site; 1982 (±) site; site. Trifolium repens.

(b) (b) Medicago lupulina,

Symbols: (•) 1984 site; 1984 (•) Symbols: Month ( a ) 1979 site; (■ ) 1971 ) (■ site; 1979

(c) Trifolium pratense (o) (o) 1983

82 Insect density (insects/m2) Lotus corniculatus, Figure 3.4. The density of Homoptera recorded between May and and May between recorded Homoptera of density The 3.4. Figure November 1984 in sites of different successional age for (a) for age successional different of sites in 1984 November and(d) and(d) site; site; site. ( a ) Trifolium repens. 1982 site; (□ ) 1980 site; 1980 ) (□ site; 1982

(b) Medicago lupulina,

Symbols: (•) 1984 site; 1984 (•) Symbols: ( a ) 1979 site; (■ ) 1971 ) (■ site; 1979

(c) Trifolium pratense (o) (o) 1983

83 Insect density (insects/m2) Lotus corniculatus, Figure 3.5. The density of Lepidoptera recorded between May and and May between recorded Lepidoptera of density The 3.5. Figure November 1984 in sites of different successional age for (a) for age successional different of sites in 1984 November and(d) and(d) site; site; ( a ) 1982 site; (□ ) 1980 site; site; 1980 ) (□ site; 1982 Trifolium repens.

(b) Medicago lupulina,

Symbols: (•) 1984 site; (o) 1983 (o) site; 1984 (•) Symbols: ( a ) 1979 site; (■ ) 1971 site. 1971 ) (■ site; 1979

(c) Trifolium pratense

84 Insect density (insects/m2) Lotus corniculatus, Lotus November 1984 in sites of different successional age for (a) for age successional different of sites in and 1984 May between November recorded Symphyta of density The 3.6. Figure and(d) and(d) site; site; ( a J A O N O S A J J M ) Trifolium repens. Trifolium 1982 site; (□ ) 1980 site; site; 1980 ) (□ site; 1982 (b) No Symphyta sampled. Symphyta No Medicago lupulina, Medicago Symbols: (•) 1984 site; (o) 1983 (o) site; 1984 (•) Symbols: Month (a) (d) (b) (c) ( a ) 1979 site; (■ ) 1971 site. 1971 ) (■ site; 1979 (c) Trifolium pratense Trifolium

Lotus corniculatus, Lotus Figure 3.7. The density of Thysanoptera recorded between May and and May between recorded Thysanoptera of density The 3.7. Figure November 1984 in sites of different successional age for (a) for age successional different of sites in 1984 November and (d) (d) and site; (±) 1982 site; (□ ) 1980 site; site; 1980 ) (□ site; 1982 (±) site; Insect density [ln(insects/m2)] Trifolium repens. Trifolium (b) Medicago lupulina, Medicago Symbols: (•) 1984 site; (o) 1983 (o) site; 1984 (•) Symbols: ( a (a) ) 1979 site; (■ ) 1971 site. 1971 ) (■ site; 1979 (c) Trifolium pratense Trifolium

36 87

highest in the younger sites. One common feature of the four legume species was that abundance in older sites tended to peak and decrease earlier than in the younger sites. The highest levels of homopteran abundance were reached on V. hirsuta, although T. repens also had a high homopteran load in the 1984 site. Larval Lepidoptera (Figure 3.5.) and Symphyta (Figure 3.6.) were only rarely found in the samples. Undoubtedly this was due, at least in part, to the sampling method used. However, the highest number of Lepidoptera were recorded on L. corniculatus during late August. The Symphyta were even rarer, with only a few individuals being recorded on M. lupulina, T. pratense and T. repens in only four sites. The number of species in each insect group for each species of legume in each site is shown in Figures 3.8. and 3.9. One striking feature is that the number of insect species found on both Vicia species (Figure 3.9.) was much lower than on other legume species (Figure 3.8.). Two groups that were noticeably lacking on the Vida species were the Lepidoptera and Symphyta. However, on L. corniculatus, M. lupulina, T. pratense and T. repens the number of lepidopteran species was high bearing in mind the low abundance of the group. L. corniculatus had the highest number of species recorded and, in common with the other three legumes, the species included specialist feeders (e.g. Zygaena Fabricius species.) as well as more polyphagous feeders (e.g. Artia caja L., Noctua pronuba L. and Phlogophora meticulosa L. ). The number of homopteran species was similar on all the legumes although only one species, Acyrthosiphon pisum, was common to all the legumes sampled. The Coleoptera were particularly well represented on L. corniculatus, T. pratense and T. repens of which the main contributors were from the Apion Herbst and Sitona Germar genera (see Appendix 3). The number of Heteroptera varied between 2-3 species on the Vicia species to 4-7 on the other legumes. One exception was T. repens in the 1984 site where only two species of generalist Heteroptera were found. Finally, Thysanoptera were not represented by many species although one, Sericothrips abnormis (Kamy), reached very high levels of abundance on L. corniculatus, T. pratense and T. repens but was present in much smaller numbers on M. lupulina and V. hirsuta. Another thrip, Frankliniella intonsa (Trybom) was found on all the legumes sampled but was commoner on the two Vida species. Some successional trends in species numbers were apparent, although their clarity varied between different species of legume. For instance, in the Vicia species successional trends were not clearly visible as only a small number of insect species were sampled. However, in L. corniculatus, M. lupulina, T. pratense and T. repens there appears to be an increase in the number of species of Lepidoptera and Heteroptera with successional age, 88

(a)ii. 1971 site

(b)ii. 1979 site

(c)i. 1983 site (c)ii. 1982 site 15 10 5 0 (d)i. 1984 site (d)ii. 1983 site 15 15 10 10 5 5 0 0 (d)iv. 1971 site

Insect Group

Figure 3.8. The number of insect species in each insect herbivore group in sites of different successional age for: (a) Lotus corniculatus; (b) Medicago lupulina; (c) Trifolium pratense; (d) Trifolium repens. Symbols: (■ ) Coleoptera; (■ ) Heteroptera; ( n$ Homoptera; ( ip) Lepidoptera; ( na) Symphyta; ( EU) Thysanoptera 39

(a)i. 1984 site (b)i. 1981 site 15 15 10 - 10 - 5 - 5 ■ 0 __ i iw M ___ m __ 0 — ■ ■ S___ r—i ii. 1981 site ii. 1980 site

iii. 1979 site iii. 1979 site 15 15 10 10 - 5 : 1 1 0 ■ f a n~i iv. 1977 site iv. 1977 site 15 15 10 10 5 5 * 0 --- H n s a _[=1__ 0 __ HHsra____cm__ v. 1971 site v. 1971 site

Insect Group

Figure 3.9. The number of insect species in each insect herbivore group in sites of different successional age for: (a) Vida hirsuta; (b) Vidasativa. Symbols: (■ ) Coleoptera; (■ ) Heteroptera; ( fflf) Homoptera; ( H) Lepidoptera; (EH) Symphyta; ( £U) Thysanoptera. 90 this trend being most apparent in the former group. The number of species of Coleoptera showed an increase in species until the second year of succession after which species number decreased with successional age.

3.4.3. Multivariate Analysis Ordination techniques have been used to define both plant and insect communities at various stages of succession (Brown & Hyman 1986). In the present study, the insect communities on specific species of legume were defined. Only the sample scores are presented here (Figure 3.10. & 3.11.), since they reflect the abundance and composition of the insect herbivore fauna. Congruous sample scores indicate that the composition and abundance of the insect species within those samples are similar, and the more divergent the sample scores, the more dissimilar the insect fauna. Species scores were also calculated, but they did not show clear patterns mainly because the number of samples taken from each species of legume in each site was not always equal. In the analysis, four axes were generated, although, only the first two could be interpreted biologically. Figure 3.10. compares all the sample scores and gives a general picture of the ordination for each legume species. Figure 3.11. shows the same sample scores, but in this case each species of legume is plotted separately. Axis 1 described a weak seasonal gradient from left to right (Figure 3.11.), whereas Axis 2 described insect species composition and abundance. As a result of these differences, there were shifts in the vertical and horizontal distribution of the sample scores from which secondary gradients could be recognized in addition to the primary characterizations of both axes. The secondary gradients identified in Figure 3.10. and 3.11. differentiated the insect fauna of each species of legume. L. corniculatus, T. pratense and T. repens shared very similar sample scores and occupy the lower left hand comer of Figure 3.10. The two Vicia species also share similar sample scores, but in this case they tend to occupy a band to the far right of Axis 1. M. lupulina, on the other hand, occupies another distinct band which lies between T. pratense and the two Vicia species. Variation within the sample scores was visible in a number of legumes sampled, but such outliers were the result of differences in the composition and abundance of the insect herbivores. There were, however, no apparent trends with site age in this ordination, although the sample scores for both T. repens and V. hirsuta in the 1984 site do indicate that insect abundance and composition were different in this site. In V. sativa, 1971 site (second sample), the sample score can be classed as an outlier and this was due to the fact that only Homoptera 350

300 -

c£ o 250 ©CO • o v

+ + X X ++ ++ + 1 A +

■ap * A O ■ v X >*< ■ + *

100 200 300 Axis 1 (Eigenvalue = 0.633) Figure 3.10. Sample ordination for all six species of legume irrespective of site or sampling date. Symbols: (A ) Lotus corniculatus; ( o ) Medicago lupulina; ( + ) Trifolium pratense; ( ■ ) Trifolium repens; ( V) Vida hirsuta; ( x ) Vida sativa. pratense, Figure 3.11. Sample ordination for each site and sampling occasion occasion sampling and site each for ordination Sample 3.11. Figure for (a) (a) for Symbols: (*) 1984 site; (□ ) 1983 site; (*) 1982 site; (♦ ) 1981 site; site; 1981 ) (♦ site; 1982 (*) site; 1983 ) (□ site; 1984 (*) Symbols: site; (•) 1980 site; site; 1980 (•) site; sample; (5) September sample, (6) October sample. October (6) sample, September (5) August (4) sample; sample; July (3) sample; June (2) sample; May (1) Axis 2 (Eigenvalue = 0.335) Lotus corniculatus, Lotus (d) (d) Trifolium repens Trifolium (a) (o) Axis 1 (Eigenvalue = 0.633) = 1(Eigenvalue Axis 1979 site; 1979 (b) (e) , Medicago lupulina, Medicago ( a Vida hirsuta Vida ) 1977 site; (■ ) 1971 site; 1971 ) (■ site; 1977 and (f) and (c) Trifolium (b) Vida sativa Vida

92 93

contributed to the sample score (Table 3.). Similar explanations can be given for the outlying sample scores in L. corniculatus, M. lupulina and T. pratense except in this case, the variation was due to differences in other insect groups.

3.5. Discussion A number of studies have focused on the ideas formulated by Margalef (1968) and Odum (1969) and have concentrated on revealing the different adaptive strategies of a number of groups of phytophagous insects associated with different plant communities along a successional gradient (e.g. Brown 1982b, 1985, 1986, Brown & Hyman 1986, Brown & Llewellyn 1985, Brown & Southwood 1983, 1987). However, there are few studies of insect/plant relationships whereby the phytophagous insect load from individual plants within one plant family is monitored along the early part of a secondary successional gradient. Such studies are essential, since they provide an opportunity to measure the abundance and composition of the insect herbivore community feeding on closely-related species in habitats of increasing durational stability. The suggestion that the insect load (at least within solely phytophagous groups) is governed by plant-species composition (Brown & Hyman 1986) was not tested here, although it is apparent that the insect load not only varied between species but also within species, tending to be higher in the younger sites. However, the within-species variation between sites of similar age was very small. Such differences in insect abundance between young and old sites in the legumes may be attributed to differences in abundance of natural enemies and probably plant chemistry (Lawton & McNeill 1979). It is known that the structural attributes of the plant are important in determining the abundance of natural enemies. Increases in plant architecture have been shown to occur along the successional gradient (Southwood et al. 1979) and together with the findings that insect diversity is higher early in succession (Brown 1982a, Brown et al. 1987), one can infer that insect herbivore abundance should decrease with successional age as predicted by Lawton & McNeill (1979). Sites of similar age should have similar plant architecture and species composition, hence, similarities between insect herbivore load may be expected. In the ruderal site, insect herbivores will be colonizing from adjacent areas and thus, initially, a high frequency of polyphagous generalist feeders will be expected since these are generally better adapted to the conditions of an ephemeral habitat (Brown 1985). Differences in plant chemistry (Feeny 1976 and Rhoades & Cates 1976) and payability (Reader & Southwood 1981) during succession have been demonstrated between plant 94 species, but whether such differences occur within a species are not known. However, such differences can not be ruled out, since seasonal differences are common (Feeny 1970, Lawton 1976). Two features apparent in the insect communities on each legume species are common to many other species (see Lawton & McNeill 1979). Firstly, of the individual species that contribute to the total insect load only a few become very abundant (Appendix 3). Sericothrips abnormis becomes abundant on L. corniculatus, T. pratense and T. repens especially towards the end of the season. On all six species of legume, Plagiognathus chrysanthemi was the most abundant heteropteran and was in fact the dominant herbivore on M. lupulina, V. hirsuta and V. sativa. However, as with many other heteropteran species, the high abundance can be attributed to the incidence of large numbers of nymphs which occur at the beginning of the season (c.f. Holometabolous taxa). This stage in the life cycle experiences a high degree of mortality, as reflected in the smaller number of adults, despite their greater mobility making them more difficult to capture. Within each phytophagous group, only one or two species occur commonly with the time of peak abundance varying. Secondly, of all the phytophagous insect species recorded on their host plants, the majority never become very abundant, although this does not mean they are unimportant to the community. This fact is most clearly illustrated by the Lepidoptera and Symphyta, since although these groups are rather uncommon, their biomass and therefore impact on the vegetation may be considerable. Tumipseed and Kogan (1976) categorized three main components of the native insect fauna colonizing the legume soybean, Glycine max (L.) Merr. These same three components can be used to describe the insect fauna colonizing the legumes in different successional sites. Polyphagous species represent the first category, particularly the Lepidoptera. The larvae of this group were most abundant on cyanogenic L. corniculatus and T. repens (see Chapter 5). Indeed, over 30 species of Lepidoptera have on rare occasions been recorded feeding on L. corniculatus (Compton 1983). Krieger et al. (1971) suggested that polyphagous Lepidoptera may be particularly well equipped to detoxify a wide range of secondary compounds such as cyanide (Compton & Jones 1985). Two more polyphagous feeders were the Miridae, Adelphocoris lineolatus (Goeze) and Lygus rugulipennis Poppius. The second group of colonizing phytophages, include stenophagous species (e.g. Curculionoidea) adapted to wild legumes, and these species probably represent the core fauna of the six legumes studied. However, variation in the abundance of some insect herbivores between different legumes was evident and this is probably a reflection of host preference. What was even more puzzling was the variation in insect abundance within the same species of legume in different successional sites e.g. Lygus rugulipennis on T. repens and Acyrthosiphon pisum on V. hirsuta (and T. repens) 95 in the 1984 site. These differences can be explained by the fact that plants in different environments are subjected to different conditions and, hence, differences in their insect fauna might occur, since some species might be better suited to the particular conditions prevailing in a particular site. Very few purely monophagous species (i.e. those only found on a single host plant as recorded in the host plant records) were found. The third group of insects consist of certain stenophagous species that seem to have shifted their host preferences, sometimes from other plant families. This last category contributes very litde to the legume fauna (except on an evolutionary scale). It is doubtfid whether the particular sampling methods employed would have been sensitive enough to detect any new additions to the host plant fauna, since these are rare events, and take place over large time spans. The ordination of the samples has contributed complementary information and has indicated that growth form and life-form are important in defining the insect community present. There were three life-forms apparent: perennial (L. corniculatus, T. pratense and T. repens); long-lived annual/short-lived perennial (M. lupulina); and short-lived annual (V. hirsuta and V. sativa). Three growth forms could also be defined: low-growing prostrate stems (L. corniculatus and T. repens); upright rigid stems (M. lupulina and T. pratense); and climbing, scrambling stems (V”. hirsuta and V. sativa). Thus, the insect communities present can be defined, since at one extreme, there is an insect community associated with the perennials, while at the other, one is found on the annuals. Between these two extremes lies another insect community associated with a single species representing the long-lived annual/short-lived perennial life-form. Even within the perennial life-form, slight differences occur which could be attributable to differences in growth form. Hence, L. corniculatus and T. repens were most similar in respect of their insect fauna, whereas T. pratense lies between these two legumes species and M. lupulina. Similar insect communities between congeneric legume species are to be expected, since there are probably close biochemical and morphological links. Plant architecture might also influence the composition of the insect community, since most stenophagous legume herbivores only exploit a small part of their host plant, and as a result, legumes having similar plant structures may share the same herbivores. Another factor which could contribute to the insect herbivore distribution is the occurrence of different legume species in the same site. Thus stenophagous herbivores, colonizing one species, may then colonize other legumes in close proximity especially when intraspecific competition is intense. The lack of successional trends in this ordination does suggest that the insect communities within a legume species were remarkably similar (although differences in abundance have been noted). However, the ruderal site due to its ephemeral nature appears to be distinct. Chapter 4 Insect Herbivory: its Effects on Plant Reproductive and Vegetative Characteristics

4.1. Introduction The impact of insect herbivory on natural plant communities is now being realised as a potent force in their organisation and structure. Traditionally, it was thought that herbivores did not play an important part in the regulation of plant populations (e.g. Hairston, Smith & Slobodkin 1960) and plants were assumed to be regulated by competition. Maybe one reason for this view was indicated in Wilson’s review of biological control of weeds by insects (Wilson 1964), in which he concluded that in most successful cases of biological control, a biologist viewing a plant at equilibrium with its herbivores would have little reason to suspect that insects had a significant impact on that plant species. Whittaker (1979) suggested that the outcome of competition between different plants maybe altered by insect herbivory, since removal of plant parts such as leaves may reduce the individual plant to a subordinate position within the hierarchy of a dense population (Harper 1977). Such interactions between herbivory and competition have been shown to occur in natural plant communities (e.g. Parker & Salzman 1985). The effect of herbivores on their host plants tends to reduce overall fitness, although there is some evidence that herbivory may be beneficial to some plants (McNaughton 1983, McNaughton, Wallace & Coughenour 1983). However, this view point has been disputed recently by Belsky (1986). A common feature of insect herbivory is a reduction in growth rate, fecundity (Kinsman & Platt 1984, Parker & Salzman 1985, Rai & Tripathi 1985, Hendrix 1988) and dry matter yield (Bentley & Whittaker 1979, Parker & Salzman 1985, Whittaker 1982). A reduced growth rate may delay flowering (Rai & Tripathi 1985) and increase the exposure of the plant to abiotic and biotic stresses. The effect of herbivory may not always be apparent in the generation of plants in which it occurs, since reductions in seed number and seed weight (Bentley, Whittaker & Malloch 1980, Hendrix 1979, Kinsman & Platt 1984) may affect germination, survival and recruitment (Bentley, Whittaker & Malloch 1980, Hendrix 1979, Louda 1982) of the next generation. This is important since differences in seed germination time (Waller 1985), starting capital or 97 seed size (Waller 1985) as well as neighbourhood effects (competition) influence the final size of the plant and have been implicated in determining which individuals become dominant (Rabinowitz 1979). Such features fuel the generation of size hierarchies (Weiner & Solbrig 1984) which have potential effects on fitness, as size is often related to fecundity and survivorship (Solbrig 1981), but such inequalities have rarely been measured in natural populations (Scheiner 1987). What is even less clear is the role of herbivory in size variability. One way of evaluating the impact of insect herbivores on a plant community is to monitor the community in the presence and absence of insect herbivores. However, one of the major difficulties of this method is excluding, or substantially reducing, the number of insect herbivores. This feat is most easily achieved with vertebrates, and has led to a large number of studies showing vertebrate herbivores influencing the distribution and abundance of their food plants (Harper 1977). Experiments designed to simulate insect herbivoiy involve artificial defoliation by manual clipping (e.g. Becker 1983, Lee & Bazzaz 1980, Maun & Cavers 1971, Rockwood 1973), but discrepancies between insect and simulated defoliation have been demonstrated by Bentley & Whittaker (1979) and Havlickova (1982). Other studies have only looked at single herbivore species on one of their host plants (Hendrix 1979, Rausher & Feeny 1980). However, one of the most successful ways of excluding insect herbivores to date is by the use of repeated applications of insecticide. Such studies do not quantify the effects of different insect herbivore guilds, but show the overall effect of herbivory. Some of these early manipulative studies (Cantlon 1969, Waloff & Richards 1977) demonstrated that such treatments had a marked effect on growth, survival and reproduction. Subsequent studies by Brown (1982a, 1985) and Stinson (1983) have shown that reduced levels of natural insect herbivory on a plant community can dramatically influence the species richness, plant cover, seedling establishment as well as affect growth, survival and reproduction of individual species (Brown 1985). Insect herbivoiy is also known to cause a reduction in the rate and affect the direction of secondary succession (Brown 1982a, 1984, Gibson, Brown & Jepsen 1987, Gibson et al. 1987, Stinson 1983) as well as alter the vegetational structure (Brown, Gange & Gibson 1988). Other workers have found that the rate of secondary succession may well increase following outbreaks of phytophagous beetles (McBrien, Harmsen & Crowden 1983). However, these differences are probably the result of changes in plant demography within different plant communities although such changes have been little studied. 98

In this chapter, the performance of six species of early successional legumes (L. corniculatus, M. lupulina, T. pratense, T. repens, V. hirsutaand V. sativa) under natural and experimentally reduced levels of insect herbivoiy are compared. The effects of insect herbivoiy on vegetative and reproductive characters of each species are reported. Reference is made to the plant communities occurring in the field sites of different successional age, since levels of plant competition and insect herbivore abundance change at different stages in early succession (Brown & Southwood 1987).

4.2. Materials and Methods

The six species: L. corniculatus, M. lupulina, T. pratense, T. repens, V. hirsutaand V. sativa were monitored in 1985 in sites of known successional age. The species, all from one plant family (Leguminosae), represent a range of plant strategies and growth forms (see Chapter 2) each able to exist in a range of plant communities. The sites used in the study were created in 1984,1983 and 1971 and were the same as those used for the point quadrat survey in 1984 (Chapter 2). However, only 20 of the 3m x 3m subplots were utilized in each site so that a 3m gap was maintained between neighbouring subplots. Additionally, a new site was created in 1985 with the same number of subplots as previous sites although in this case the distance separating each subplot from its neighbour was reduced to 2m. All subplots were systematically allocated to a treatment; 10 control subplots in each site were sprayed with 45ml of water, while the remaining 10 were sprayed with 45ml of the insecticide, Malathion-60 solution (obtained as an emulsifiable concentrate from Berks, Bucks & Oxon Fanners Ltd., Twyford, Berks.). Each application of insecticide contained 1.134ml of active ingredient equivalent to the recommended U.K. agricultural rate of 1.26kg a.i.ha-1 (Martin & Worthing 1976). Both insecticide and water were applied using a hand-held U.L.V. sprayer ("Micron ulva"; Micron Sprayers Ltd., Bromyard, Herts.) at intervals of approximately 10 days from late April to October. Applications were undertaken at dawn since the cool, calm conditions at this time of day reduced drift to a minimum. Assumptions that Malathion-60 has no effect on the vegetation of the sites was investigated in a separate study by Brown, Leijn & Stinson (1987), since some insecticides are known to cause phytostimulation (e.g. Allen & Casida 1951, Brown, Cathy & Lincoln 1962), alter the rate of nutrient cycling (Malone 1969) or induce phytotoxicity in some plant species (Shure 1971). The results obtained by Brown et al. 99

(1987) showed that the low persistent, non-systemic insecticide, Malathion-60, had no direct effects on a range of early successional plants in both field and experimentally controlled conditions. Plants in both treatments were protected from mollusc grazing by an application of molluscicide pellets ("Mifa Slug" obtained from Farmers Crop Chemical Ltd., Worcester.) around each plant every two weeks. This precautionary measure was included even though mollusc numbers were generally low in the sites. Grazing damage caused by Microtis agrestis L. was more difficult to eliminate. However, the level of damage was generally low and was also distinctive especially in L. corniculatus, T. repens and T. pratense where stems (stolons), leaves and petioles were severed from the plant and used to line the mammal runs. Damage to Vicia plants was also easily recognised as stems were severed at the base. However, V”. sativa was only seen to be damaged in the 1979 site and was not common. Table 4.1. gives details of the sites in which each species studied was monitored and the number of replicates in each treatment. The basic procedure was to mark, by coloured plastic labels around the base of the stem, up to five plants in each of the 20 subplots (10 control and 10 insecticide-treated). Plant performance was monitored at approximately every 2 weeks in the insecticide-treated plots and 3-4 weeks in the control plots from late May to the end of the growing season (October). In T. pratense and T. repens above-ground parts do not senesce during winter and plants were also monitored in mid-December. The following plant characteristics were measured for each species on each sampling occasion: plant height, total leaf number, number of leaves damaged by insects, number of flowers/flowerheads and number of pods or seedheads. The analysis of the two Vicia species is given in Brown et al. (1987) and T. pratense in Gange et al. (in press). 100

1985 1984 1983 1979 1971 site s ite s ite s ite s ite

I C I C I C IC I C

Lotus corniculatus - -- - 30 35 -- - Hedicago lupulina 5 1 1 20 40 - - --- Trifolium pratense 18 20 8 34 43 44 - - - Trifolium repens 45 44 38 50 35 50 - - 41 Vida hirsuta 40 45 50 50 - - 48 50 - Vida sativa 20 17 20 25 -— 50 50 — —

Table 4. 1. The number of plants marked on each, site for performance studies with. L. corniculatus, M. lupulin a, T. prat ease, T. r opens, V. hirsuta and V. sativa. Symbols: I = Insecticide-treated; C = Control. 101

The fate of individual leaves was assessed by marking the youngest expanding leaf on each sampling occasion with a piece of coloured cotton around the petiole. On subsequent occasions any insect damage was recorded independently on scale: Damage rating Estimated leaf area removed (%) 0 No damage 1 1-5 2 6-25 3 26-50 4 51-75 5 76-99 6 Total removal (only petiole remaining) Leaf turnover was assessed by calculating life expectancy (see Southwood 1978) using the number of marked leaves which were alive on each sampling occasion, irrespective of damage. During mid-August, 100 mature pods from the two Vicia species and L. corniculatus were collected at random from both treatments in each site. The number of seeds per pod were recorded for each species but the pod length was measured only for the Vicia species. Seeds were oven dried and weighed individually, and for L. corniculatus, mean seed weight and seed number per pod was calculated. Regression relationships were obtained between total seed weight and pod length as well as seed number and pod length for both Vicia species. These regression relationships were used to estimate the seed weight and seed number of 20 mature pods in mid-August on each of the marked plants. In T. repens and M. lupulina, up to 70 seedheads per treatment were collected during mid-August and the number of florets per seedhead and seeds per seedhead recorded. In T. pratense, only the total number of seeds were recorded from mature seedheads collected in mid-August. Seeds were separated from their florets, oven dried and weighed individually. Seedhead samples from M. lupulina, T. pratense and T. repens were collected from plants adjacent to those marked to avoid destructive sampling. However, there were insufficient mature plants of M. lupulina, T. pratense and T. repens in the 1985 site, and as a result, no samples were taken. In addition, stolons of T. repens were monitored on the same sampling occasions in a site created in 1971 (see Southwood et al. 1979). Approximately 70 seedheads were collected in mid-August, oven dried and the number of florets per seedhead, seeds per seedhead as well as individual seed weights recorded. The stolons were not sprayed with 102 water or insecticide, as the experimental programme did not permit treatment of this site. However, the stolons were subjected to light grazing by rabbits, and the performance of these stolons is included for comparitive purposes, although the main relevance of this part of the work is in Chapter 5. Plant recruitment was monitored in late-April 1986, by recording the number of seedlings germinating in an area of 500cm2 around the base of each marked plant. The same procedure was repeated in May when the distinction between seedlings (i.e. plants with two cotyledons and no more than two leaves) and established plants was made.

4.3. Results

A complete description of the results for both Vida species is given in Brown et al. (1987) and T. pratense in Gange et al. (in press). The latter study describes the performance of T. pratense for three years, although it is the first year results which are relevant here. The following sections describe the performance of L. corniculatus, M. lupulina and T. repens. Leaf life expectancy data for T. pratense are also included, since leaf persistence was not covered in Gange et al. (in press).

4.3.1. Effects of Insect Herbivory on Plant Height and Leaf Number These two parameters were used to estimate the size of the plant, as size is one of the factors relating to fitness. Maximum height was attained in July in M. lupulina (1985 and 1984 sites), T. repens (1984, 1983 and 1971 sites), and insecticide-treated plants of L. corniculatus. In T. repens (1985 site) and control plants of L. corniculatus maximum height was not reached until early August. Treatment had no effect on maximum height in M. lupulina and T. repens (1984 site), however, in L. corniculatus and T. repens (1985 and 1983 sites) maximum height was greater in control plants (Table 4.2.). To determine whether height varied thoughout the season, a split plot analysis of variance was carried out (Table 4.4.). In general, similar results were obtained to those reported in Table 4.2., however, T. repens plants in control plots were significantly taller throughout the season. Maximum leaf number was attained in July in M. lupulina, August in L. corniculatus and September in T. repens but in each species maximum leaf number did not vary between treatments (Table 4.3.). However, when leaf number was analysed on a seasonal basis it was greater in control plants of L. corniculatus (Table 4.4.) although the other two species showed no significant difference. Plant Species Site Treat- Mean Plant S.E. t d.f. P ment H eight

Lotus corniculatus 1983 I 27.7 ±1.4 A C 32.8 ±2.2 -2.01 58

M edicago lupulina 1985 1 44.4 ±2.5 C 45.9 ±2.9 -0.32 14 N.S.

1984 I 50.6 ±3.5 C 43.78 ±1.7 1.97 58 N.S.

Trifolium repens 1985 I 18.8 ±0.6 C 22.1 ±1.0 -2.75 88 *1 1984 I 34.2 ±1.4 C 37.2 ±1.2 -1.66 86 N.S. 1983 I 26.1 ±1.3 C 30.4 ±1.5 -2.04 83 *

1971 I C 19.8 ±0.8

Table 4.2. Maximum height attained by each plant in L o t u s corniculatus, Medicago lupulina and Trifolium repens in sites of different successional age. Symbols: I = insecticide-treated; C = control (natural levels of insect 3 0 1 herbivory); * = p<0.05; ** = p<0.01; N.S. = not significant. Plant Species S ite T reat- Mean Leaf S.E. t d .f. P ment Number

Lotus corniculatus 1983 I 59.6 ±5.8 C 75.5 ±7.9 -1.62 58 N.S.

M edicare lupulina 1985 I 49.2 ±3.3 C 53.4 ±5.2 -0.51 14 N.S.

1984 I 70.6 ±7.2 C 81.3 ±6.1 -1.07 58 N.S.

Tri folium repens 1985 I 145.7 ±11.8 C 139.6 ±8.3 0.41 88 N.S.

1984 I 29.6 ±2.7 C 30.0 ±2.8 -0.11 86 N.S. 1983 I 24.9 ± C 29.9 ± -0.90 83 N.S.

1971 I _ C 18.4 ±1.52

Table 4.3. Maximum number of leaves per plant (mean ±1 S.E.) in L o t u s corniculatus and Nedicagv lupulina and maxlmimum number of leaves per stolon In Trifolium repens in sites of different successlonal age. Symbols: I = insecticide-treated; C = control (natural levels of Insect 4 0 1 herblvory); N.S. = not significant. Lotus corniculatus Medicago lupulina Trifolium repens (1983 site) (1984 site) (1985,1984 & 1983 sites) Leaf Leaf Leaf Source d.f. Height number d.f. Height number d.f. Height number Layer 1 Sites (S) 2 38.9 *** 136.8 *** Treatment (T) 1 16.0 *** 5.6 * 1 1.7 N.S. 0.0 N.S. 1 15.8 *** 0.4 N.S. SxT 2 0.6 N.S. 0.4 N.S. Total 35 39 167

Layer 2 Dates (D) 5 27.6 *** 11.6 *** 2 94.2 *** 62.8 *** 4 207.7 *** 108.1 *** S x D 8 29.5 *** 71.6 *** TxD 5 9.0 *** 1.2 N.S. 2 9 9 *** 15.4 *** 4 y 7 *** 2.1 N.S. S x T x D 8 0.6 N.S. 0.5 N.S. Total 215 119 839 Table 4.4. F-values of split plot analysis of variance for vegetative characteristics of Lotus corniculatus, Medicago lupulina and Trifolium repens. Analysis performed on square root transformation of data. Symbols: *** = p <0.001; ** = p <0.01; * = p <0.05; N.S. = not significant. 5 0 1 106

43.2. Leaf Life Expectancy and Accumulated Damage by Insects The leaf life expectancy of individually-marked leaves produced at different times in the season by L. corniculatus, M. lupulina, T. pratense and T. repens are given in Table 4.5. The effects of reducing insect herbivory in L. corniculatus and T. pratense appear to be similar in the 1983 site, since application of insecticide tends to increase leaf life expectancy only at the beginning of the season, while leaves produced at the end of the season tend not to live as long as controls. Indeed insecticide-treated plants started to lose their leaves and senesce before the control plants. By contrast, leaf life expectancy of T. pratense in the 1985 site was higher in insecticide-treated plants than control irrespective of the time of year the leaves were produced. However, leaves produced early in the season lived longer in both treatments. A similar pattern occurred in M. lupulina in the 1985 sites, although in the 1984 site, control plants of M. lupulina lost their leaves and senesced before the insecticide-treated plants. The life expectancy of T. repens leaves showed no clear pattern between different treatments and sites. However, the general trend was for values to be highest at the beginning of the season, decline in July, increase in August and decline during mid-September. The relationship between insect damage accumulated during the life of the leaf (i.e. final damage rating) and leaf life expectancy for four species is given in Table 4.6. Generally, insecticide-treated plants received lower levels of damage in each case. In M. lupulina, there was no significant relationship between damage as leaf persistence in insecticide-treated plants in both sites and damage throughout the season was low. By contrast, control leaves tended to accumulate damage with increasing age. Hence, early season leaves which have relatively high life expectancy were damaged more than shorter lived leaves produced later in the season. This same relationship also occurred in L. corniculatus leaves under natural levels of insect herbivory, and to a lesser extent in insecticide-treated leaves. In T. pratense, only control plants in the 1985 site and insecticide-treated plants in the 1983 site showed any significant relationship between damage and leaf life expectancy. In the latter case damage increased with leaf age, but in the former, damage was high initially and did not increase with leaf age. In the 1984 and 1983 sites, high initial damage followed by little or no damage was also indicated in the control treatments. This indicated that leaves were damaged mainly when they were young and confirms my own personal observations. Damage in T. repens was positively correlated with leaf age in insecticide-treated plants in 1985 and 1983 sites and control plants in 1984 site. In the other treatments (excluding the 1971 site) damage appears to be constant and not dependent on age. 107

Lotus Medicago lupulina corniculatus 1983 site 1985 site 1984 site Date IC IC I C 25/5 6.00 4.00 -- 1.85 1.40 26/6 5.50 - - - 2.05 - 10/7 5.03 4.36 -- 1.20 1.23 24/7 3.83 3.36 3.50 3.50 0.95 0.70 13/8 1.83 - 3.10 2.68 0.74 - 30/8 1.02 2.44 2.30 1.90 0.50 - 12/9 0.58 1.61 1.30 0.94 -- 2/10 - 0.87 0.50 0.50 - - Trifolium pratense 1985 site 1984 site 1983 site Date I C I C IC 25/5 - - 2.13 2.42 3.29 2.57 26/6 -- 2.38 - 2.74 - 10/7 -- 1.63 1.62 2.24 1.57 24/7 3.78 3.45 1.00 0.90 1.63 1.17 13/8 3.28 2.75 0.63 - 1.90 1.90 30/8 2.44 2.05 1.90 2.21 1.64 - 11/9 1.67 1.60 1.25 2.18 1.25 1.48 30/9 1.56 1.35 - 1.31 1.29 1.23 25/10 -- - 0.79 - 0.78 Trifolium repens 1985 site 1984 site 1983 site 1971 site Date I C Date I C Date I C Date I C 3/6 2.29 2.35 5/6 1.99 1.58 9/6 1.96 1.90 12/6 - 2.09 11/7 1.90 2.14 15/7 1.20 1.32 18/7 1.36 1.46 23/7 - 1.35 9/8 2.16 2.27 13/8 1.42 1.39 15/8 1.35 1.61 19/8 - 1.66 29/8 1.47 2.27 3/9 1.17 1.74 5/9 1.74 1.50 9/9 - 1.61 16/9 1.43 1.39 20/9 0.98 1.17 23/9 1.33 0.96 25/9 - 1.11 (See Southwood 1978) Table 4.5. Life expectancy/(of individual leaves of Lotus corniculatus, Medicago lupulina, Trifolium pratense and T. repens in sites of different successional age. Symbols: I = insecticide-treated; C = control; - = no sample available. Lotus corniculatus Intercept P Slope P rl d.f. P 1983 I 0.16±0.15 N.S. 0.07+0.03 * 0.031 146 * site C 0.82+0.29 ** 0.33+0.09 *** 0.081 162 *** Medicago lupulina 1985 I 0.48+0.74 N.S. 0.26+0.27 N.S. 0.051 17 N.S. site C 1.35±0.77 N.S. 0.69+0.29 * 0.146 32 ** 1984 I 0.59+0.66 N.S. 0.33+0.40 N.S. 0.014 47 N.S. site C -0.41+0.30 N.S. 0.93+0.25 *** 0.137 89 *** Trifolium pratense 1985 I 0.21+0.16 N.S. 0.31+0.57 N.S. 0.004 84 N.S. site C 1.35+0.35 *** 0.27+0.15 N.S. 0.037 89 * 1984 I 1.03+0.47 * 0.34+0.26 N.S. 0.056 29 N.S. site C 1.21+0.32 *** 0.23+0.17 N.S. 0.015 128 N.S. 1983 I -0.18+0.14 N.S. 0.41+0.06 ♦ ♦ ♦ 0.208 187 5k * * site C 1.65+0.25 *** 0.09+0.14 N.S. 0.002 201 N.S. Trifolium repens 1985 I 0.35+0.37 N.S. 0.43+0.20 * 0.024 186 ★ site C 2.06+0.45 -0.11+0.21 N.S. 0.002 173 N.S. 1984 I 1.19+0.33 *5k5k 0.30+0.23 N.S. 0.015 109 N.S. site C 0.60+0.59 N.S. 0.86+0.41 5k 0.026 170 5k 1983 I 0.21+0.47 N.S. 0.63+0.31 * 0.028 145 * site C 1.40+0.50 *sk 0.31+0.33 N.S. 0.005 170 N.S. 1971 I ------site C 0.53+0.41 N.S. 0.46+0.25 N.S. 0.027 117 5k Table 4.6. Relationship between life expectancy and insect damage in Lotus corniculatus, Medicago lupulina, Trifolium pratense and T. repens in sites of different successional age. Symbols: I = insecticide-treated; C = control; *** = p < 0.0 0 1; ** = p < 0.0 1; * = p < 0.05; N.S. = not significant. 109

The percentage of leaves damaged on a plant in insecticide and control treatments varied between species and sites (Figures 4.1.-4.3.). In control plants, the percentage of damaged leaves in L. corniculatus and M. lupulina increased during the season. Whereas in T. repens and T. pratense, there was an overall decrease in all sites except the 1985 site where damage decreased in mid season but increased again during September. Generally, damage was above 40% in all species although T. pratense and T. repens experienced a higher level of damage throughout the season. Application of insecticide reduced the level of damage in all species but the extent of reduction varied between species and sites. Insecticide treatment appeared to be least effective in reducing the leaf damage in T. repens. In M. lupulina (1985 and 1984 sites) and T. pratense (1984 and 1983 site) the effectiveness of the insecticide treatment was also rather poor at least during part of the season.

4.3.3, Effects of Insect Herbivory on Reproductive Characteristics There was no difference in the number of flowerheads or seedheads produced by M. lupulina (Table 4.7.) and T. repens (Table 4.8.) when insect herbivory was reduced, but increased pod production occurred in L. corniculatus (Table 4.7.). The effect of reducing insect herbivory on L. corniculatus was to increase seed number and seed yield per pod, although individual seed weight per pod was not affected (Table 4.7.). In M. lupulina reducing insect herbivory significantly increased the number of florets and seeds per seedhead. Unfortunately, no information is available on the weight of individual seeds per seedhead in M. lupulina although insecticide application resulted in heavier seeds in both L. corniculatus and M. lupulina. There was no difference in the number of florets per flowerhead in T. repens but seed number per floret was increased when insect herbivory was reduced, and as a result, the total number of seeds per seedhead and seed yield was greater than control plants. Surprisingly, the effect of insect herbivoiy on individual seed weight was to increase seed weight (see seed weight per treatment: Table 4.9.), a trend also seen in the 1984 site when individual seed weight per seedhead was assessed. However, in the 1983 site individual seed weight was higher in insecticide-treated plants than in controls. To determine the effect of insect herbivory on seed germination and survival of seedlings, the number of seedlings that emerged and established around L. corniculatus, M. lupulina, T. pratense and T. repens plants were recorded the following spring (Table 4.11.). Although seedling numbers were small in each case, more seedlings emerged and represented are: (•) insecticide-treated; (o) control. (o) insecticide-treated; (•) are: represented Figure 4.1. Seasonal variation in the percentage of leaves damaged under under damaged leaves of percentage the in variation Seasonal 4.1. Figure Percentage of leaves damaged by insect herbivory site). Values are means (arcsine square root transformed) ±1 S.E. Symbols Symbols S.E. ±1 transformed) root square (arcsine means are Values site). (1983 site), (b site), (1983 natural and reduced levels of insect herbivory in (a) (a) in herbivory insect of levels reduced and natural )Medicago lupulina )Medicago (1985 site) and (c and site) (1985 )M. lupulina )M. Lotas corniculatus Lotas (1984 (1984

sites of different successional age : (a) 1985 site, (b) 1984 site and (c) and site 1984 (b) site, 1985 (a): age successional different of sites Figure 4.2. Seasonal variation in the percentage of of percentage the in variation Seasonal 4.2. Figure Percentage of leaves damaged by insects leaves damaged under natural and reduced levels of insect herbivory in in herbivory insect of levels reduced and natural under damaged leaves 1983 site. Values are means (arcsine square root transformed)^ S.E. transformed)^ root square (arcsine means are Values site. 1983 Symbols represented are : (•) insecticide-treated; insecticide-treated; (•) : are represented Symbols 90 ( ) (a - (o ) ) (o Trifolium pratense Trifolium control.

Percentage of leaves damaged by insects ± 1 S.E. Symbols represented are : (•) insecticide-treated; (O) control. insecticide-treated; (•) : are represented ±1Symbols S.E. Figure 4.3. Seasonal variation in the percentage of of percentage the in variation Seasonal 4.3. Figure leaves damaged under natural and reduced levels of insect herbivory in in herbivory insect of levels reduced and natural under damaged leaves sites of different successional age : (a) 1985 site, (b) 1984 site, (c) 1983 (c) site, 1984 (b) site, 1985 (a) : age successional different of sites site and (d) 1971 site. Values are means (arcsine square root transformed) transformed) root square (arcsine means are Values site. 1971 (d) and site Month Trifolium repens Trifolium

112 113

Lotus Medicago coniculatus lupulina 1963 site 1965 site 1984 site I C I C I C Mean number of flowers 7.010.6 5.010.9 11.212.2 8.110.8 22.813.5 18.712.1 (flowerheads) per plant (n=30> (n=35) (n=5> (n=ll) (n=20) (n*40) t= l .79 p=N.S. t=l.69 p=M.S. t=1.59 p=N.S. d .f. =63 d .f .=14 d. f . =58 Mean number of pods 10.711.1 6.011.5 40.615.5 38.614.2 19.612.8 23.412.9 (seed heads) per plant (n«99) (n»100) t=8.75 p<0.001 d .f. = 197 Mean individual seed 1.2110.02 1.1610.03 weight per pod (mg) (n=702> (mg). t=3.87 p<0.001 t=1.48 p<0.001 d. f . = 1024 d. f . = 1400

Table 4.7. The effect of reducing levels of Insect herbivory on reproductive characteristics (mean l 1 S. E. ) in Lotus corniculatus and Medic ago lupullna. Analysed by two sample t-Test. Symbols: I = insecticide-treated; C = control; number of replicates; N. S. = not significant.

i 1985 site 1984 site 1983 site 1971 site

ICI C I C l C Bffect d.f. P P

lean naiber of 0.73tO.20 0.9810.20 1.5010.22 1.6610.16 1.7410.23 0.9610.19 - _ Site 2 6.93 III flower heads Treatient l 0.46 1.3. per stolon SitelTreat 2 3.63 1 (45) (44) (38) (50) (35) (50) Total 270

lean naiber of - 0.9710.19 - 1.6610.17 - 0.9610.17 0.5910.19 Site 3 6.45 III flower heads per Total 193 stolon (control plots only) (44) (56) (50) (41)

lean nuiber of 0.8910.22 0.9310.22 1.8710.24 1.9310.19 1.6610.25 0.9610.20 - Site 2 11.44 III seedheads Treatient l 0.49 II. 8. per stolon SitelTreat 2 1.69 R.S. (45) (36) (35) (501 Total 270

lean naiber of - 0.9310.19 - 1.9310.17 - 0.9610.18 0.5110.20 Site 3 10.98 Ill seedheads per Total 193 stolon (control plots only) (44) (50) (50) (41)

Table 4.6. The effect of reducing levels of insect herb ivory aid site age on naiber af fl overheads and seedheads (leant l S.B) per stolon in TrltoltuM repets. Figures in parenthesis represent the anther of replicates. Syibols: I = insecticide treated; C = control; lit = p<0.001; l = p(0.05; I.S. = not significant. 4 1 1 115

1984 site 1983 site IC I C Bffect d. f. F P

Slte(S) 1 23.2 ttt Mean nnmher of florets 84.311.5 61.611.5 56.311.5 55.511.5 Treatment t 1.5 N.S. per floverhead (70) (70) (70) (70) StTreat l 0.4 N.S. Total 279 Slte(S) 1 0.6 N.S. Mean nnmher of seeds 1.7410.08 1.0210.08 1.610.1 1.110.1 Treatment l 62.6 itt per floret (70) (70) (70) (70) StTreat l 1.6 N.S. Total 279

Site(S) 1 7.8 tt Mean nnmher of seeds 108.714.9 63.814.9 86.814.9 58.314.9 Treatment l 56.3 ttt per seedhead (70) (70) (70) (70) StTreat l 2.8 N.S. Total 279 Slte(S) l 0.1 N.S. Mean Indlvidnal seed we1gbt 0.3710.02 0.5310.2 0.4710.02 0.4110.02 Treatment l 6.2 N.S. per seedhead (mg) (70) (70) (70) (70) StTreat l 34.7 ttt Total 279 Site(S) l 3.5 N.S. Mean seed yield 45.012.8 32.812.8 41.912.8 25.312.8 Treatment l 26.8 fit per seedhead (my) (70) (70) (70) (70) StTreat l 0.6 N.S. - Total 279 Slte(S) 1 4.9 « Mean individual seedwelyh.t 0.4210.01 0.4810.01 0.4610.01 0.4710.01 Treatment l 26.7 ftt per treatment (mg’) (511) (524) (500) (501) StTreat l 10.5 tt Total 2035

Table 4.9. Tbe effect of reducing insect herbivory on reproductive characteristics (mean ± 1 S.E.) in Trlfoliua repens in two early successional sites. Figures in parenthesis represent the number of replicates in each sample. Symbols: I = insecticide-treated; C = control; = p<0.001; « = p<0.01;* - p<0.05; N.S. - not significant. 1984 site 1983 s ite 1971 site Effect d. f. F p

Mean number of florets 61.6111.26 55.4611.26 42.3911.26 Site 2 60.70 *#♦ per seedhead (70) (70) (70) Total 209

Mean number of seeds 1.01710.057 1.05810.057 0.52810.057 Site 2 27.10 per flo re t (70) (70) (70) Total 209

Mean number of seeds 63.7913.62 58.3013.62 22.8713.62 Site 2 37.66 **t per seedhead (70) (70) (70) Total 209

Mean seed weight 0.52610.019 0.41010.019 0.39410.019 Site 2 14.61 *fi per seedhead (mg) (70) (70) (70) Total 209

Mean seed yield 36.61011.996 25.29311.996 9.21211.996 Site 2 35.98 *** per seedhead (mg) (70) (70) (70) Total 209

Mean seed weight 0.48210.008 0.47110.008 0.39810.008 Site 2 31. 15 ft* per treatm ent (mg) (524) (501) (501) Total 1525

Table 4.10. The effect of site age on seedhead variables (mean 1 1 S.E) in T r i f o l i u m r e p e n s (control plots only). Figures In parenthesis represent the number of replicates in each sample. Symbols: *** = p<0.001

03 * 1985 aite 1984 site 1983 site 1971 site

I CIC I c I C

Biteof Legcie Saiple 3 E 3 B 8 E 3 E 3 E 3 E 3 E 3 B

3.42 0 Lotus 19/4/86 tQ. 54 corticalstus 0.25 1.06 0 0 30/5/86 tO. 12 tO. 21

0.10 0.09 0.30 0.03 fcdicsgo 19/4/86 tO. 20 tO. 09 tO. 16 tO.03 lupalltd 0.20 0.60 0.30 0.10 0.16 0.05 0 0 30/5/86 tO.20 iO.4010.15 10.10 tO.09 tO. 05

0 0 0.45 0.43 5.07 0 Tr I fo il at 19/4/86 tO. 21 tO. 43 tO.72 prstcusc 0.33 0 0.10 0 0 0.08 0 0 1.35 0.09 0 0 30/5/86 tO. 14 10.01 10.08 10.22 10.04

19/4/86 0.05 0.03 0 0 0 0 0 r r I to lla t tO.04 tO. 03 rcpcts 30/5/86 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Table 4.11. Seed germination and plant establishment ofLotus cornlculatus, Medicago lupulina, Trlfolium pratense and Trifotlium repens under natural and reduced levels of insect herb ivory. Values are mean number of plants in 500 cm* around masked plants ±1 S.E. Symbols: S = seedling; E = established plant; I = insecticide-treated; G = control. 117 118 established when insect herbivory was reduced. Generally, recruitment around plants subjected to natural levels of insect herbivory was low and mainly confined to the 1985 site.

43,4, Successional Trends Successional trends can not be reported for L. corniculatus as sampling was only possible in one site. In M. lupulina and T. repens changes in both vegetative and reproductive characteristics were apparent in sites of different successional age. In the 1985 site all plants established from seed in the spring, whereas in older sites, plants may be as old as the site itself. Similar trends in leaf life expectancy are shown by M. lupulina, T. pratense and T. repens with slightly higher longevity in the youngest site (Table 4.5.). The percentage of damaged leaves in T. pratense and T. repens was also the lowest in this site (Figures 4.2. and 4.3.). However, in M. lupulina the proportion of leaves damaged did not vary between 1985 and 1984 sites (Figure 4.1a and b), although the amount of damage each leaf received may be higher in the 1985 site (Table 4.6.). There were no differences in the maximum height attained by M. lupulina in the 1985 and 1984 sites. However, T. repens plants reached their greatest height in the 1984 site, while the shortest plants were found in the 1985 and 1971 sites (Table 4.2.). M. lupulina has the lowest number of leaves in the youngest site while T. repens stolons have most leaves in this site and least in the 1971 (Table 4.3.). Changes in reproductive characteristics also occurred during succession. In M. lupulina, there was an increase in flowerhead number and decrease in seedhead number from the first to the second years after site establishment (Table 4.7.). Control plants of T. repens reached peak flowerhead and seedhead production in the 1984 site after which there is a progressive decrease with successional age (Table 4.8). The same trend was also seen in the seedhead characteristics (Table 4.10), the between site differences being highly significant.

4.4. Discussion

Changes in the structure and dynamics of plant populations and communities are influenced to varying degrees by herbivory. Insect herbivores make a large contribution to these changes by either directly consuming leaves, stems, roots, flowers, seeds and 119 fruits (chewers) or indirectly by using piercing/suctoral mouthparts to extract plant sap or cell contents (sap feeders). The effects of these herbivores are mediated through changes in vegetative and reproductive characteristics of their host plants. Variation in the number of insect herbivores during the season and along the successional gradient are important, since these will reflect changes in the structure and diversity of the vegetation. In the 1985 site all potential herbivores have to immigrate from the surrounding area, consequently, there is a lag in the effects these herbivores have on their host plants. However, as the season progresses the number of invading insect herbivore species increase (Southwood et al. 1979). Futhermore, initial colonizers are comprised of mainly highly mobile and highly fecund generalist feeders, but as succession proceeds, by less mobile specialist feeders (Brown 1986), many of which overwinter on or near their hosts and are thus able to colonize earlier in the season. The leaf longevity was reduced by natural levels of insect herbivory in all species except T. repens. The reduction in leaf turnover tended to be greater in the middle of the season in the youngest site and at the beginning of the season in the older sites which may be a reflection of insect herbivore pressure. However, L. corniculatus and T. pratense (in 1984 and 1983 sites) showed a deviation to this pattern, since leaves produced at the end of the season live longer when the plant experiences natural levels of insect herbivory. In addition, control plants senesce after insecticide-treated plants. Such compensatory responses have been described in other plants especially after defloration (Crawley 1983, Hendrix 1984) and may be an attempt to boost stored carbohydrate levels to maintain survival through to the following season. Leaf longevity in T. repens showed no general pattern between reduced and natural levels of insect herbivory and may be due to a reduction in the efficiency of insecticide application as a result of shielding effects of other plants on the prostrate, low growing stolons. Also T. repens growing on its own forms a dense canopy and this may "protect" the stolons, leaf buds and expanding leaves lying close to the ground from a contact insecticide, and hence, only give a relatively small reduction in insect herbivory. These effects may substantiate the patterns seen in Figure 4.6. since the proportion of damaged leaves was only slightly reduced by insecticide application. Variation in leaf turnover appears to be associated with the plant’s life cycle strategy, since the annual Vicia species have the highest turnover rates followed by the annual/short-lived perennial, M. lupulina, and the perennials, L. corniculatus, T. pratense and T. repens. These changes can be explained in terms of plant apparency theory (Feeny 1976) and may reflect variations in plant secondary chemistry (Rhoades & Cates 1976). L. corniculatus and T. pratense appeared to be most apparent in time and space, T. repens 120

was apparent in time, but not in space as a result of its stoloniferous growth habit. M. lupulina, V. hirsuta and V. sativa tend to be less apparent in time and, by virtue of the growth of other plant species associated with them, often in space as well. Consequently, it can be predicted that leaves of the more apparent legumes should be better defended chemically than their less apparent relatives (Feeny 1976). As a result apparent species will be the least palatable to insect herbivores (Reader & Southwood 1981) and have a greater life expectancy (Southwood, Brown & Reader 1986). Changes in leaf apparency also occurred within a species, since leaves were most apparent at the beginning of the season. In both Vicia species insect damage was more likely to occur in the shorter-lived unapparent leaves produced later in the season and this supports current hypotheses which propose that investment in antiherbivore defences will reflect the degree of predictability of each plant part as a resource for herbivores (Coley 1980). However, in L. corniculatus, M. lupulina and T. repens (1984 site) increased apparency was coupled with increased damage, so the predictability hypothesis does not hold. In this case damage may simply accumulate with age due to an increased probability of being eaten. In T. pratense and T. repens (1985 and 1983 sites) the young expanding leaves appear to represent an ephemeral resource which were either not as heavily defended as older leaves (Feeny 1976; Rhoades & Cates 1976) or were more nutritious (Feeny 1970), thus explaining the high level of damage incurred. It is interesting that reducing insect herbivoiy in V. sativa decreased leaf production (Brown et al. 1987), however, this anomaly may be explained by the fact that this plant species is ant attended (Koptur & Lawton 1988). It has been shown by Bendey (1977) that ants visiting extra floral nectaries often give plants some protection against insect herbivores. In V. sativa protection against insect phytophages was not complete since hard-bodied insects were not deterred by ants (Koptur & Lawton 1988). This must be the reason behind the fact that V. sativa in Silwood Park supported mainly well armoured weevils such as S. lineatus and the number of other leaf and stem feeding insects were generally much lower compared to its congener V. hirsuta (Brown et al. 1987). Thus control plants of V. sativa may have a reduced herbivore load when ant attended and also ants feeding at extra floral nectaries may remove excess assimilates, which can inhibit photosynthesis, and therefore increase plant productivity by acting as sinks similar to aphids (Geiger 1976). No extra floral nectaries occur on the other species of legume studied and as a result their responses to insect herbivory were different. Natural levels of insect herbivory reduced height and leaf production in T. pratense and V. hirsuta and this translates to an overall decrease in plant size. This has important implications, since plant size is correlated with survival and fecundity. A continuation of the studies on T. pratense (Gange et al. in press) has demonstrated that plant size is reduced by insect herbivory and 121

small plants are less fecund and have a higher mortality, even though large plants are more likely to be attacked. Insect herbivory appears to increase plant height in L. corniculatus and T. repens, although leaf production is unaffected. Both these species are able to reproduce vegetatively by cloning and so the balance between vegetative growth and sexual reproduction may be affected by insect herbivory. Thus, reducing herbivory may favour sexual reproduction, while increasing it may stimulate vegetative growth (see Crawley 1983). M. lupulina shows no change in height and leaf number, although plants subjected to natural levels of insect herbivory were damaged more and as a consequence plant size (in terms of photosynthetic area) will be effectively reduced. Changes in reproductive characteristics were not consistent between species and may reflect differences in the plant’s reproductive strategy. Insect herbivory did not change the number of inflorescences or seedheads in M. lupulina and T. repens, while in L. corniculatus flower number remained unchanged but pod number was reduced. Flower production in both Vicia species and inflorescence production in T. pratense was considerably reduced by natural levels of insect herbivory and this lead to a reduction in pod and seedhead number in these species. Reducing insect herbivory increased seed production in all species. However, the effect on individual reproductive structures in each species showed a varied response. In L. corniculatus and V. sativa seed number per pod was increased but individual seedweight remained unchanged, while in V. hirsuta seed number per pod remained consistent (at two seeds per pod) but seedweight was increased. Seed number in M. lupulina was increased by insecticide application and since each floret produces only a single seed, the increased seed production was solely a result of increased floret number, since floret abortion was reduced. T. pratense plants protected from insect herbivory produced a similar number of seeds per seedhead to unprotected plants, however, seeds from protected plants were heavier. By contrast, floret number in T. repens was unaltered by reducing insect herbivory, but seed number per floret, seed yield and seed number per seedhead were all increased. The fact that seed number was never reduced by insecticide application indicates that the efficiency of the pollinators was not reduced. Indeed, Louda (1983) found that insecticide application may well enhance pollinator success by reducing their predators. Reducing insect herbivory has the general effect of increasing the plant’s resources available for reproduction. However, there are trade-offs within individual reproductive structures, since seed number may only be increased at the expense of seed weight and vice versa. Thus, contrary to early beliefs, seed size is not always constant (see Harper 1977) and can vary between species and individual plants (Harper et al. 1970, Janzen 122

1977, Hendrix 1984, 1988). Variation in seed weight can be produced by insect herbivory (Bentley, Whittaker & Malloch 1980, Kinsman & Platt 1984), although plants can compensate by reducing seed number (Brown et al. 1987). Seed size variation can affect subsequent germination rates since small seeds have been shown to germinate faster than larger ones (Maun & Cavers 1971, Bentley, Whittaker & Malloch 1980, Hendrix 1984), although fewer small seeds may germinate overall (Hendrix 1984, Kinsman & Platt 1984) . Large seeds may have greater absolute growth rates of roots, stems and leaves at least initially (Weis 1982) and hence may be able to withstand greater enviromental stress (Kinsman & Platt 1984), competition from neighbouring plants (Crawley & Nachapong 1985) and insect herbivory. In only one of the species studied {T. repens), the seeds entering the seed bank in the insecticide treatments were smaller than controls and such seeds might well be at a competitive disadvantage. However, increased seed production might lead to increased seedling recruitment especially if seedling mortality is not a limiting factor (Louda 1983). This strategy is probably a good adaptation for colonizing recently disturbed areas such as molehills (Cahn & Harper 1976a). To support this, Crawley & Nachapong (1985) found no difference in recruitment between large and small seeds of Senecio jacobaea L. in cultivated soil. Thus, increasing seed number may increase the probability of reaching a disturbed site especially if the latter are distributed heterogeneously (Kinsman & Platt 1984). However, it cannot be ruled out that decreases in flower and seed/ovule predation and subsequent increases in seed number will also be followed by increases in seed weight, since in T. repens application of insecticide did not substantially reduce leaf damage. As a result, resources needed to fill these seeds may be lacking, hence, lighter seeds from insecticide-treated plants. Reducing insect herbivory increased seedling recruitment and plant establishment probably through an increase in seedling success. This may be one of the contributory factors modifying the direction and rate of secondary succession (Brown 1982a, Gibson, Brown & Jepson 1987, Gibson et al. 1987). Seedling establishment does vary between different sites and generally reflects differences in microsite availability (Louda 1983). Plant competition is another factor which helps shape plant communities. Both competition and herbivory may interact and exacerbate the effects of each other, so that whether a certain amount of grazing may affect the plant may depend on the level of competition with other plants (Lee & Bazzaz 1980, Parker & Salzman 1985). Thus, the full impact of herbivory may only be expressed when the plants are also experiencing interference (Whittaker 1979) and this may be accentuated under enviromental stress (Whittaker 1982, Dirzo 1985). Changes during secondary succession alter the balance between herbivory and competition. Generally, competition becomes more severe as succession proceeds due to changes in plant abundance, taxonomic diversity and 123 architecture (Southwood et al. 1979). In the two Vida species the effect of insect herbivory on vegetative and reproductive characteristics appeared more pronounced in the youngest site, although plant growth and reproduction declined with successional age, a feature presumably linked to increased plant competition (Brown et al. 1987). In M. lupulina only the 1985 and 1984 sites could be compared since very little M. lupulina occurred in the older sites in 1985, although the previous season it was also common in the 1980 and 1979 sites (see Chapter 2). Thus, trends in vegetative and reproductive characteristics along the early successional gradient could not be ascertained. The fact that M. lupulina is less common in older sites may be due to the life-history strategy of the plant (short-lived perennial herb), but also demonstrates reduced microsite availability due to higher levels of competition. The Trifolium species are longer lived in successional terms. However, peak vegetative and reproductive performance were attained in the second year of succession after which performance declined, probably as a result of increased specialization of the insect fauna (Brown 1986) and increased plant competition, mainly from grasses (Chapter 2, Brown & Southwood 1987). Thus, plant competition and insect herbivory may act independently but it would appear that it is the interaction between the two which is a major structuring force in determining the distribution and abundance of different plant species. Chapter 5 Cyanogenesis and its Effect on Insect HerbivoryT. repens in

5.1. Introduction

Studies of plant-herbivore interactions aim to understand how influence the distribution and abundance of plant species. These interactions will not affect every individual within a species in the same manner, since it has been noted by Crawley (1988) that herbivore attack varies between individuals of the same species. In addition to herbivory, other factors may also influence the population dynamics of plant species in a differential manner. Thus, natural selection may favour the development of different morphs depending on the factors acting on individuals of the same species. This may result in genetic variation within a species being greater than between species (Jones 1959) and this in turn may lead to the generation of polymorphisms. Such polymorphisms can only result if the different morphs coexist, occurring together in the same habitat (Ford 1940). The polymorphism for cyanogenesis in T. repens fits this definition and it was proposed by Dirzo & Harper (1982b) that "studies into the existence of this polymorphism in local populations will have to be seen in the same context as the study of coexistence in the same area of species occupying different niches". The genetics of cyanogenesis in T. repens have been elucidated by Corkill (1942) and Dawson (1941) while the biochemical properties have been well studied by Coop (1940), Corkill (1940) and Melville and Doak (1940). In summary, there are four phenotypes involved in cyanogenesis. Only one is cyanogenic, possessing cyanoglucosides and the appropriate B-glucosidases. It is only in this phenotype that HCN is evolved when the tissues are damaged. The control of cyanogenesis is through two unlinked loci designated Ac/ac and Li/Li. The former is responsible for the production of cyanoglucosides and the latter for 6-glucosidases. The alleles Ac and Li are completely dominant. Thus the cyanogenic phenotype is designated AcLi, while the three acyanogenic phenotypes are designated Acli (only cyanoglucoside present), acLi (only 6-glucosidase present) and acli (homozygous double recessive containing no glucoside or enzyme). Early studies looked for single factors influencing the frequency of cyanogenesis. Daday (1954a, b) found that both dominant alleles (Ac and Li) in T. repens were at a selective disadvantage at high altitude and latitude and correlated the reduced fitness to 125

lower mean winter temperatures. Later, Daday (1965) showed by direct experimentation that sexual reproduction and vegetative growth of the AcLi, Acli and acLi phenotypes were more prolific in warm conditions. From this, he concluded that low temperatures were responsible for maintaining the polymorphism through the locus responsible for cyanoglucoside production, by being genetically linked to genes concerned with fitness responses to temperature. The AcLi phenotype was also affected directly by its susceptibility to frost damage, since ice crystals forming inside leaf cells disrupt membranes and bring about cyanogenesis resulting in tissue death. Other workers have subsequently shown that other physical factors can influence the frequency of cyanogenesis in T. repens. Foulds and Grime (1972b) demonstrated that under experimental conditions of severe drought, the cyanoglucoside containing morphs suffered a greater mortality than the morphs lacking this glucoside. The former morphs under moist conditions attained higher yields than the latter suggesting another linkage of the Ac/ac allele, this time with genes affecting vegetative growth. The cyanogenic plants were found to have lower sexual vigour and under severe moisture stress flowering was completely inhibited. It is interesting that L. corniculatus does not show the same responses to temperature (Jones 1970, 1972) or moisture stress (Foulds & Grime 1972a), although the genetic polymorphism of cyanogenesis is essentially the same. More recent work (Foulds 1977, Foulds & Young 1977) has shown the conclusions reached by Daday (1954a, b, 1965) and Foulds and Grime (1972a, b) should be treated with caution and some of the early work has been criticized by Jones (1973), who warns of the dangers of judging the effects on alleles rather than on phenotypes. The fact that a dine exists does not explain the existence of a polymorphism (Cahn & Harper 1976a). In any case, all phenotypes are often found growing together in the same habitat, thus cyanogenesis is not always necessary for the survival of the plant under all normal conditions (Jones 1972). There may be components within the habitat that are related to the distribution of phenotypes, such factors are probably important in maintaining the polymorphism. So far only physical factors have been described as selective agents, but biotic forces also contribute and may interact with physical factors (Jones 1970). One of the much studied biotic factors influencing the polymorphism for cyanogenesis is the differential grazing of the cyanogenic and acyanogenic morphs by molluscs. What is apparent from these studies is that cyanide is a relative and not an absolute defence (Dirzo & Harper 1982a), since there are no morphological characteristics to differentiate between cyanogenic and acyanogenic phenotypes and plant tissue has to be damaged before cyanide is released. The protective function of cyanogenesis in T. repens against mollusc herbivores has been demonstrated in the field and laboratory (Angseesing 1974, 126

Angseesing & Angseesing 1973, Crawford-Sidebotham 1972, Dirzo & Harper 1982a,b). Cyanogenesis in L. corniculatus also serves a similar function (Crawford-Sidebotham 1972, Keymer & Jones 1978, Jones 1962, 1966). Indeed good circumstantial evidence obtained by Ellis, Keymer & Jones (1977a) indicated that in areas exposed to wind-borne salt, the frequency of selective mollusc herbivores and cyanogenic plants was low. In less exposed areas, mollusc herbivores were more numerous and consequently the frequency of cyanogenic plants was higher. Grazing intensity and herbivore distribution has also been shown to influence the frequency of cyanogenesis in T. repens (see Dirzo & Harper 1982a). However, what is apparent in most of the studies of T. repens is the high degree of variability in the palatability of cyanogenic and acyanogenic morphs. It might be expected that laboratory experiments should show the least variation, since most of the leaves/plants have been chosen for their high cyanogenic response (Angseesing 1974, Dirzo & Harper 1982a). Even in many field experiments, cuttings were propagated in greenhouses and transplanted into the field. In all cases phenotypes were assessed prior to experimentation and only cyanogenic plants showing the strongest reaction used (Dirzo & Harper 1982b). Even so, there is still a lot of heterogeneity in the response of herbivores to cyanogenic and acyanogenic morphs. Ellis, Keymer & Jones (1977b) have shown that there is a degree of plasticity in the intensity of cyanogenic reaction in different cyanogenic morphs of L. corniculatus and that some clones are phenotypically unstable (i.e. show differences in phenotype expression at different times). Such phenotype instability has not been found in T. repens (Dirzo & Harper 1982b), although the intensity of cyanogenic response has been shown to vary at different times of the day and between days (de Waal 1942). Not all the variation in the acceptability of T. repens can be attributed to the plant and variation within the herbivore cannot be discounted. Even if within-species variation is ignored, differences in the acceptability of cyanogenic and acyanogenic morphs have been noted between species, since every mollusc species is not always selective (Crawford-Sidebotham 1972, Angseesing 1974). Therefore, tolerance to cyanide cannot be ruled out in these herbivores, although Dirzo & Harper (1982a) have shown that slugs fed on cyanogenic leaves made slightly smaller live-weight gains or lost weight faster than those given a diet of acyanogenic leaves. Thus direct feeding cannot be accepted as a measure of resistance. Specialist herbivores, notably insects, have been shown to feed exclusively on cyanogenic plants of both T. repens and L. corniculatus. The existence of detoxification mechanisms have been proposed by Parsons & Rothschild (1964) in Polyommatus icarus 127

Rott., since the larvae contain rhodanese (one of the enzymes responsible for detoxification, see Beesley, Compton & Jones 1985). This is perhaps why this species of butterfly shows no selective egg-laying and larval feeding on its host plant, L. corniculatus (Lane 1962). In the light of this evidence, it is somewhat surprising to find that few studies have been undertaken to evaluate the selective pressure of insects on the maintenance of the polymorphism for cyanogenesis. Dritschilo et al. (1979) investigated the effect of colonizing insects on cyanogenic and acyanogenic morphs of T. repens. They found, that with the exception of aphids, insect herbivores were infrequent visitors and failed to discriminate between morphs. The aphids, on the other hand, were two to three times more abundant on acyanogenic plants. Protection against aphids and other kinds of colonizing insects must be important, particularly at the seedling stage, since these insects possess the ability to find new clover stands. Cyanide maybe an effective deterrent in raderal sites, since there is a tendency for the colonizing herbivores to be highly mobile and highly fecund generalist feeders (Brown 1986), and thus may not be tolerant to this plant defence (Rhoades & Cates 1976). The susceptibility of the seedling stage of T. repens has been suggested by Crawford-Sidebotham (1972), Jones (1972) and Dirzo & Harper (1982a), although Miller et al. (1975) failed to find any evidence of differential seedling mortality between cyanogenic and acyanogenic morphs. However, Dritschilo et al. (1979) reported that seedling populations of T. repens had lower cyanogenesis frequencies than the surrounding adult populations. One important group of herbivores so far not mentioned are the vertebrates. Cyanide tolerance within this group of herbivores is found particularly in the mammals, although tolerance may vary at different times depending on the amount eaten, level of starvation or time to acquire resistance (see Jones 1966, 1972). Selection against the cyanogenic morph may be further complicated, since Cahn & Harper (1976b) suggested that leaf mark polymorphism of T. repens might be important in the visual selection of diet by sheep. Other reports suggest that cyanogenesis deters rabbits from grazing T. repens (Corkill 1952), M. agrestis from eating L. corniculatus (Jones 1966) and deer and sheep from browsing Pteridium aquilinum (L.) Kuhn (Cooper-Driver & Swain 1976). However, in the case of M. agrestis, selection was only suitably demonstrated when an adequate alternative food supply was available (Jones 1966). Selection in T. repens by other herbivores (e.g. molluscs) might also be altered especially when the herbivore is surrounded by acceptable neighbouring plants (Dirzo & Harper 1982b). Thus, although cyanogenesis is not totally effective against herbivores that normally include T. repens in their diet it may, however, serve to deter many potential herbivores and pathogens from attacking the plant (Jones 1972). 128

The defensive role of cyanogenesis is not without its costs. Dirzo & Harper (1982b) suggested that these costs may be described in terms of partitioning of resources (Cody 1966). There are three costs which contribute to the individual’s fitness: (i) reproduction, (ii) competitive ability and (iii) defence. As resources are limited, all three costs must balance so that a disproportionate expenditure on one is made at the expense of the others. Dirzo (1984) reported the superior competitive ability of the acyanogenic morphs of T. repens when both cyanogenic and acyanogenic morphs are grown in mixture, but competitive equilibrium is only attained when the acyanogenic morphs are selectively defoliated. Competitive ability of susceptible plants has been shown to be better when the pathogen/predator is absent, while the performance of the resistant plants is always better when the pathogen/predator is present (Burdon & Olivers 1977, Windle & Franz 1979). All these experiments involve intraspecific competition. However, there are few studies that incorporate other variables which operate in natural conditions. Interspecific competition with other plant species (e.g. grasses) has been studied (Turkington et al. 1979, Turkington & Harper 1979a), although how this is related to the maintenance of the polymorphism for cyanogenesis is not clear. Hence, studies into the existence and maintenance of polymorphisms in plant populations are only really valid by detailed ecological investigations into the performance of a plant species at the small rather than the geographic scale (Cahn & Harper 1976a). In this chapter, the polymorphism for cyanogenesis is investigated in a field study in which the intraspecific variation of the acceptability of cyanogenic and acyanogenic morphs of T. repens to insects is assessed by monitoring the performance of these morphs under natural and reduced levels of insect herbivory. The effects of herbivory on the vegetative and reproductive characteristics of the four morphs in respect of cyanogenesis in field sites of different successional age are reported.

5.2. Materials and Methods 52.1. Detecting Cyanide using the Guignard Picrate Paper Test This method involved making a stock solution of sodium picrate, by dissolving 2.5g of picric acid (Hopkin & Williams, general purpose reagent) in 0.1M solution of sodium carbonate. The sodium carbonate solution was prepared by dissolving 10.6g of anhydrous sodium carbonate (B.D.H. general purpose reagent) in 200ml of distilled water. Carbon 129 dioxide was released in the process and the sodium picrate formed was not very soluble, so the contents of the vessel were warmed gently over a bunsen flame until all the precipitate had dissolved. Test strips, made from Whatman No.l filter paper cut into 3mmx35mm strips, were soaked in sodium picrate solution. Excess sodium picrate solution was blotted off using a clean piece of filter paper. The presence of hydrogen cyanide (HCN) was found by grinding up about 1cm2 of freshly obtained leaf material with 3 drops of toluene (sulphur free) in a glass vial (50x12mm) using the end of a glass pasteur pipette as a pestle. A clean pipette was used with every leaf tested. Once the leaf material was ground up, a freshly prepared moist test strip was inserted into the vial. Care was taken to prevent the strip contacting the vial contents by sealing the vial with a plastic stopper and thereby wedging the strip in the top of the vial. The vial was left in a warm (20-35°C), dark place for 24 hours. A colour change from yellow to brick red indicated the presence of HCN and the AcLi phenotype. No colour change indicated one of three acyanogenic phenotypes present.

52.2. Indentifying each Acyanogenic Phenotype The determination of cyanogenic and acyanogenic morphs was first undertaken by the Guignard sodium picrate paper test. Once the AcLi phenotype has been identified, the three acyanogenic phenotypes can be separated with the aid of crude cyanoglucoside and glucosidase extracts. The methods of extraction and a small quantity of cyanoglucoside solution were kindly given to me by Monica Hughes, Newcastle University. The cyanoglucoside extract was made by picking enough leaves to fill a 200ml flask. Sufficient absolute ethanol was added to cover the leaves, and the flask was placed in a water bath set at 80°C and the contents boiled for five hours. Frequent checks were made to ensure that the contents did not boil dry. The resultant green liquid was strained from leaves and evaporated to concentrate the solution. This was the crude cyanoglucoside extract. Identification of the acLi phenotype could now be carried out. This was done by testing an acyanogenic leaf with four drops of cyanoglucoside extract and three drops of toluene and carrying out the Guignard picrate test in the normal way. A positive reaction to this test indicated the presence of the acLi phenotype, while a negative score indicated that either the Acli or acli phenotype were present. 130

Once the acLi phenotype has been determined, a crude extract of glucosidase could be prepared from its leaves. This was done by placing a handful of leaves and a small quantity of acid washed silver sand in a mortar. The leaves were covered with 0.15M citrate phosphate buffer pH 5.0, 0.2M sodium chloride and thoroughly ground to a puree with a pestle. The crude enzyme extract was strained through muslin and stored in a refrigerator. The buffer was made by preparing 0.5M stock solutions of citric acid (B.D.H. general purpose reagent) and disodium hydrogen orthophosphate (Sorensen’s salt, B.D.H. general purpose reagent). To make 200ml of buffer it was necessary to mix together 41.2ml solution of Sorensen’s salt, 19.4ml of citric acid solution and 139.6ml of distilled water. Finally, 2.3g of sodium chloride (B.D.H. general purpose reagent) was added to the buffer, because the enzyme is ionically bound to cell membranes and a high ionic gradient is needed to release it. The remaining phenotypes can now be identified by carrying out two Guignard picric paper tests simultaneously on one leaf. In test A, four drops of cyanoglucoside and test B, four drops of glucosidase extract were added at the same time as the toluene and the picrate paper test was carried out in the normal way. In the case of test B, all vials were left for 48-72 hours, since the reaction was much slower. A positive reaction in both tests indicated the AcLi phenotype present, while negative reactions in both of the tests indicated the acli phenotype present. A negative reaction to test A and a positive reaction to test B indicated the Acli phenotype present, while the exact opposite indicated the presence of the acLi phenotype.

53.3. Survey of Cyanogertesis In this survey, ten leaves of T. repens (when present) were collected from each of the 45 subplots in the 1984 site, 1983 site, 1980 site and 1971 site; and from 28 subplots in the 1982 site. The leaves were individually tested by Guignard picric paper test within six hours of collection. Leaf samples were taken from these sites in July and September 1984 131 and February 1985 and cyanogenic and acyanogenic morphs recorded. The frequency of each morph in each site was calculated from: _ , , „ number of leaves in each morph x 100 Morph Frequency = ------total------number------of leaves—------sampled—---- In addition, leaves were sampled from L. corniculatus in the 1983 and 1971 sites in September 1984 and the frequency of cyanogenesis determined. In March 1985, a new site was created (referred to as "1985 site") with 20 3mx3m subplots, of which ten were treated with Malathion-60 (insecticide-treated) and the remainder with an equal volume of water (control). Twenty subplots in the 1984 and 1983 sites were also treated in the same way. Details of sites and treatments are given in Chapter 4 and Brown et al. (1987). Leaf samples, consisting of one leaf per stolon, were taken in May 1985 from previously marked stolons from both treatments (up to 130 stolons in control and 50 in insecticide-treated subplots) in 1985, 1984 and 1983 sites. In the 1982 site, T. repens was very rare and leaf samples were taken from only 13 stolons, while in the 1971 site, 105 stolons were sampled from 45 subplots. In these latter two sites, no insecticide treatment was begun in March 1985, and results are therefore included only for comparison with the previous year’s data. The acyanogenic leaves in the May 1985 sample were investigated further to determine the actual phenotypes using the method described in section 5.2.2. of this chapter.

52.4. Monitoring the Performance of Cyanogenic and Acyanogenic T. repens The performance of T. repens was followed in the 1985, 1984, 1983 and 1971 sites. Details of sites, treatments and protection against mammals and molluscs are given in Chapter 4 and Brown et al. (1987). Identification of the phenotypes was carried out in May 1985 as described in the previous section. Stolons were tested again in August and October 1985 as a precaution against phenotype plasticity (see Ellis, Keymer & Jones 1979b). Any phenotype that showed a different expression to that originally tested was removed from the analysis. However, the occurrence of phenotype plasticity could not be confirmed because the number of cases were small and stolon misidentification could not be ruled out. Table 5.1. gives the number of stolons of each phenotype that were marked and monitored in each site and treatment. The performance of individual stolons were monitored every month from June to October. The following plant characteristics were recorded on each sampling occasion: maximum height, total leaf number, number of 1985 site 1984 site 1983 site 1971 site Treatment I C IC I C I C Phenotype Cyanogenic (AcLi) 18 27 27 25 15 25 25 Cyanoglucoside only (Acli) 21 9 12 18 9 20 15 Cyanoglucosidase only (acLi) 4 4 2 12 5 9 5 Homozygous acyanogenic (acli) 4 6 2 10 5 5 3

Table 5.1. The number of stolons of each phenotype involved in tests for cyanogenesis in T. repens marked in sites of different successional age. Symbols: I = insecticide-treated; C = control. 133 leaves damaged by insects, number of inflorescences, number of seedheads and stolon mortality. Stolon height and leaf number were analysed by a 4-way analysis of variance, while reproductive characteristics were analysed by a 3-way analysis of variance. The fate of individual leaves and the amount of insect damage present was assessed as described in Chapter 4 and Brown et al. (1987). Leaf turnover was assessed by calculating life expectancy (see Southwood 1978), using the number of marked leaves which were alive on each sampling occasion, irrespective of damage. Assessment of insect damage and stolon mortality was monitored again in March and May 1986. In mid-August 1985 ripe seedheads were collected from the marked stolons of each phenotype and the number of florets per seedhead, seeds per floret and seeds per seedhead ascertained. The seeds were then oven dried and weighed individually on a Cahn 28 automatic electrobalance. Due to the small sample size (because of low abundance in the field), the acyanogenic phenotypes were combined and analysed by a 3-way analysis of variance.

5.3. Results 5.3.1. Frequency of Cyanogenesis in Succession The frequency of cyanogenesis varies seasonally and with site (Figure 5.1.). The occurrence of cyanogenesis was uniformly more abundant in the 1982 and 1980 sites, whereas in other sites the proportion varied from 40-60%. In July and September 1984 seasonal trends were quite consistent in the 1983, 1982 and 1971 sites but in February 1985 the frequency of cyanogenesis decreased although this was reversed by May 1985 (control plots only). In the 1984 site, the frequency of the cyanogenic and acyanogenic morphs was similar and varied very little during September 1984, February and May 1985 (control plots only). The 1980 site was only sampled once, since T. repens was relatively scarce in this site and as a result the number of leaves sampled was low. In the 1971 site, the frequency of acyanogenic morphs was consistently higher than the cyanogenic morph in each sample. In September 1984, the frequency of cyanogenesis in L. corniculatus was assessed in the 1983 and 1971 sites where it was sufficiently abundant. It was found that the populations in these two sites were over 99% cyanogenic (Figure 5.1.e) Frequency of cyanogenic/acyanogenic morphs Figure 5.1. Variation in the frequency of cyanogenic and and cyanogenic of frequency the in Variation 5.1. Figure acyanogenic morphs at different times of the season in sites of of sites in season the of times different at morphs acyanogenic different successional age, for for age, successional different (b) September 1984, (c) February 1985, (d) May 1985; and and 1985; May (d) 1985, February (c) 1984, September (b) insect herbivory. Symbols: (I) insecticide-treated; (C) control or or control (C) insecticide-treated; (I) of Symbols: levels herbivory. natural or insect reduced either experienced (d) in those while natural levels of insect herbivory; (CD) cyanogenic morph; (E3) (CD)morph; cyanogenic herbivory; insect of levels natural herbivory, invertebrate of levels natural to subjected were (e) and all 3 acyanogenic morphs combined. morphs acyanogenic 3 all corniculatus : (e) September 1984. Leaf samples in (a), (b), (c) (c) (b), (a), in samples Leaf 1984. September (e) : Trifolium repens Trifolium Site 95 94 93 92 1971 1982 1983 1984 1985 : (a) July 1984, July (a): Lotus

134 135

The effect of reducing insect herbivory on the frequency of cyanogenesis is shown in Figure 5.1.(d). In the 1985 site, comparison of the insecticide and control treatments indicated that the frequency of occurrence of the cyanogenic morph was lower when insect herbivory was reduced. A similar trend occurred in the 1983 site, but the increase in the acyanogenic morph was not as marked. The 1984 site was anomalous, since the frequency of the cyanogenic morph was higher when insect herbivory was reduced. In May 1985, the frequency of the three acyanogenic phenotypes was estimated (Figure 5.2.). In all other sites except the 1982 site, the Acli phenotype was the most frequent while the acLi and acli phenotypes occurred in lower but roughly equal frequencies. However, in the 1982 site, the Acli phenotype was the only acyanogenic morph represented. The frequency of the Acli phenotype in the 1985 site was higher in the insecticide-treated plots, but there was no difference in the other two phenotypes. In contrast, the frequency of the acLi and acli phenotypes in the 1983 site was higher in the insecticide-treated plots while there was no difference in the Acli phenotype. In the 1984 site the frequency of each acyanogenic phenotype was consistently higher in the control treatment.

5.3.2. Effects of Cyanogenesis on Vegetative Characteristics (i). Stolon survival There was differential stolon survival between sites and treatments during the season (Figure 5.3.). In the 1985 site, no mortality occurred in either treatment between June and October 1984. However, in the following spring mortality was detected in both treatments in the AcLi phenotype and the insecticide treatment of the Acli and acli phenotypes. The highest mortality was recorded in the acli phenotype where, unfortunately, only four stolons could be monitored (see Table 5.1.). Thus the loss of only one stolon resulted in a 25% increase in mortality in May 1985. In the 1984, 1983 and 1971 sites stolon survival steadily decreased during the season. The effect of reducing insect herbivory tended to increase stolon survival in most cases. However, winter mortality became apparent in the spring of 1985 and affected all phenotypes in both treatments. Survival of the AcLi phenotype in control treatments was poor in the 1984 and 1983 sites although in the 1971 site over 50% of the marked stolons were still present in May 1985. By comparison, survival of the Acli phenotype was higher in the 1984 and 1983 sites. Mortality in the control plants of the 1971 site was generally high. The pattern of survival in the AcLi and acLi phenotypes was very similar. In the acLi phenotype 60% of the stolons were still present in May 1985 in the 1971 site whereas in the 1984 site only 10% remained and in the 1983 site complete mortality Figure 5.2. Changes in the frequency of each acyanogenic morph of of morph acyanogenic each of frequency the in Changes 5.2. Figure levels of insect herbivory; (Acli) cyanoglucoside only; (acLi) enzyme only; only; enzyme (acLi) only; cyanoglucoside (Acli) herbivory; insect of levels T. repens following sites : (a) 1985 site; (b) 1984 site; (c) 1983 site; (d) 1982 (d) site; 1983 (c) site; 1984 (b) site; 1985 (a) : sites following (acli) homozygous acyanogenic. homozygous (acli) successional age. All samples were taken during May 1985 from the the from 1985 May during taken were samples All age. successional site; (e) 1971 site. Symbols : (I) insecticide-treated; (C) natural natural (C) insecticide-treated; (I) : Symbols site. 1971 (e) site; Frequency of each acyanogenic morph under natural and reduced levels of insect herbivory with with herbivory insect of levels reduced and natural under ci ci acli acLi Acli Acyanogenic morph and treatment and morph Acyanogenic I C I C 136 (b)i. (c)i. (d)i.

li.

in.

IV. IV.IV. iv. 100 r 80 : 60 : 40 1 - . 20 - 0L ql6\% 0\^ 6F<£Fv9\\V 'V ^ Date of sample Figure 5.3. Stolon survival of the (i) cyanogenic, (ii) glucoside only, (iii) enzyme only and (iv) homozygous acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in: (a) 1985 site; (b) 1984 site; (c) 1983 site; (d) 1971 site. All plants were sampled between June 1985 and May 1986. Symbols: (■ ) insecticide-treated; ( li) control. 137 138 occurred (control plots only). In the insecticide-treated plots, stolons were still present although mortality was still high. Stolon survival of the acli phenotype was good in the 1984 site, with 50-60% of the stolons remaining in both treatments by May 1985. However, this trend was not reflected in the 1983 or 1971 sites where 80 and 100% mortality occurred respectively.

(ii). Stolon Height and Leaf Number The maximum height of stolons was reached in July in all phenotypes (Table 5.2.a). There was no significant difference in height between phenotypes, although stolons in the control treatment attained a greater height overall (F=33.30, P<0.001) (Table 5.2.b). There was a significant treatment x phenotype interaction (F = 3.27, p <0.05), indicating that in some of the treatments, differences in height occurred between phenotypes (Table 5.2.b). Leaf number tended to increase during the season and this was most marked in the 1985 site (Table 5.3.a). There was no difference in leaf number between phenotypes, but stolons exposed to natural levels of herbivory had significantly more leaves than those where insect herbivory was reduced (Table 5.3.b).

(Hi). Leaf Life Expectancy and Insect Damage The percentage of leaves damaged on each stolon for each phenotype varied between sites and during the season (Figures 5.4., 5.5., 5.6. & 5.7.). Lower levels of damage were experienced by insecticide-treated plants, although the trend for a higher proportion of leaves to be damaged early in the season was still found. In the 1985 site, damage increased during June and July, decreased in August and increased again in September and October. Damage decreased steadily during the season in the 1984, 1983 and 1971 sites and the level of damage was similar in each site. There was no difference in the percentage damage between cyanogenic and acyanogenic phenotypes in each site or treatment. Comparison of all three acyanogenic phenotypes in the control treatment in each site (Figure 5.4.c, 5.5.C, 5.6.c & 5.7.c) also showed no discernible difference in the percentage of leaves damaged. The leaf life expectancy of individually marked leaves of each phenotype of T. repens at different times during the season are given in Table 5.4. Reducing the level of insect herbivory does not have any clear or consistent impact on leaf life expectancy in any phenotype. Comparison of the leaf life expectancy of cyanogenic and three acyanogenic phenotypes also fails to show any clear general trends. However in the control plots, 139

Morph Site June 1984 July August early late October September September AcLi 1985 I 6.6±1.6 17.1±1.5 16.5±1.5 15.0±1.5 14.8±1.5 15.3±1.5 C 7.1±1.4 18.3±1.3 19.9±1.3 18.9±1.3 17.0±1.3 16.6±1.3 1984 I 19.7±1.2 28.7±1.2 25.4±1.2 17.2±1.2 14.4±1.3 12.7±1.5 C 20.2±1.4 39.7±1.4 24.1±1.4 16.6±1.4 16.5±1.6 12.0±1.7 1983 I 14.1±1.6 27.3±1.6 21.1±1.6 14.6±1.6 13.3±1.7 12.1±1.7 C 12.4±1.4 25.3±1.4 21.3±1.4 15.4±1.4 13.7±1.4 10.4±1.5 1971 I .. _ ... C 10.6±1.5 18.8±1.5 13.8±1.5 12.4±1.5 10.9±1.6 9.7±1.6 Acli 1985 I 6.1±1.6 14.1±1.4 17.4±1.4 16.5±1.4 15.0±1.4 15.3±1.4 C 7.9±2.5 19.9±2.1 20.0±2. 19.8±2.1 17.8±2.1 16.1dt2.1 1984 I 17.2±2.5 28.3±2.5 21.1±2.5 18.2±2.8 16.0±2.8 13.0±3.6 C 17.7±1.6 35.4±1.7 24.4±1.6 17.5±1.6 12.8±1.7 16.8±2.0 1983 I 12.0±2.1 22.6±2.2 20.3±2.1 14.3±2.1 11.7±2.1 11.2±2.2 c 14.1±1.8 26.3±1.8 19.7±1.8 14.5±1.8 14.6±2.0 13.3±2.0 1971 I .. c 13.1±1.8 20.7±1.9 15.1±1.9 13.7±1.9 10.0±1.9 11.6±2.2 acLi 1985 I 6.7±6.2 13.0±3.1 11.5±3.1 12.1±3.1 13.1±3.1 13.4±3.1 c 6.7±6.2 13.0±3.1 12.0±3.1 9.6±3.1 11.2±3.1 11.0±3.1 1984 I 14.6±4.4 30.3±6.2 20.3±4.4 12.4±4.4 6.0±6.2 5.3±6.2 c 18.5±2.1 34.3±2.1 28.4±2.1 16.1±2.1 13.1±2.1 13.1±2.2 1983 I 13.4±2.8 21.7±2.8 20.1±2.8 19.4±2.8 15.5±3.1 11.4±2.8 c 14.4±2.8 33.1±2.8 24.1±2.8 14.9±2.7 19.1±3.1 14.2±4.4 1971 I «« _ c 9.3±3.2 14.4±3.6 17.4±3.2 16.2±3.2 16.5±3.2 17.2±3.6 acli 1985 I 6.9±4.4 11.0±3.1 13.3±3.1 12.2±3.1 14.3±3.1 13.6±3.1 c 7.4±3.1 20.2±2.8 21.9±2.8 17.8±2.8 16.2±2.8 17.5±2.8 1984 I 19.4±3.6 37.5±3.6 33.0±3.6 22.3±3.6 8.8±4.4 10.2±3.6 c 18.4±2.1 40.6±2.2 28.4±2.1 21.8±2.2 16.3±2.1 14.8±2.3 1983 I 11.7±3.1 17.3±3.1 15.7±3.1 11.3±3.1 9.2±3.1 8.1±3.1 c 13.7±2.8 33.1±2.8 24.7±2.8 17.5±2.8 16.0±2.8 12.5±2.8 1971 I . - . c 9.0±4.5 23.0±4.5 15.8±4.5 11.4±4.5 12.2±4.5 9.6±6.3 Table 5.2.a Mean stolon height (±1 S.E.) of each cyanogenic/acyanogenic morph (phenotype)T. repens in of sites of different succession^ age. Symbols: I = insecticide-treated; C = control (natural levels of insect herbivory); - = no data available. The height of each stolon was measured in centimeters (cm). 140

(0 Factor d.f. F P Site 2 124.55 *** Phenotype 3 1.32 N.S. Treatment 1 33.30 *** Sample 5 147.40 *** SitePhenotype 6 4.96 *** Site.Treatment 2 0.65 N.S. Site.Sample 10 35.58 *** Phenotype.Treatment 3 3.27 * Phenotype.Sample 15 0.64 N.S. T reatment. S ample 5 4.14 Hck* Total 1437

0*0 Factor d.f. F P Site 3 75.72 *** Phenotype 3 2.23 N.S. Sample 5 112.30 *** SitePhenotype 9 3.60 *** Site.Sample 15 15.64 *** Phenotype.Sample 15 0.94 N.S. SitePhenotype.Sample 45 0.65 N.S. Total 985

Table 5.2.b Results of analysis of variance of mean height of each cyanogenic/acyanogenic phenotype of Trifolium repens in sites of different successional age. (/) 1985, 1984 & 1983 sites (insecticide and control treatments); (ii) 1985, 1984, 1983 and 1971 sites (control treatments only). Symbols: *** = p < 0.001; ** = /?< 0.01; * = p < 0.05; N.S. = not significant. 141

Morph Site June 1984 July August early late October September September AcLi 1985 I 11.0±14.3 33.0±13.9 61.5±13.9 127.9±13.9 158.3±13.9 141.7±13.9 C 11.0±12.8 35.3±11.7 73.0±11.7 155.1±11.7 229.4±11.7 278.0±11.7 1984 I 13.3±11.0 12.5±11.3 13.4±11.0 22.4±11.5 23.3±12.0 19.2±13.9 C 11.1±12.8 12.0±12.8 13.7±12.8 13.9±12.8 20.2±12.8 15.1±14.0 1983 I 17.3±14.8 15.8±14.8 15.9±14.8 18.7±14.8 18.5±15.3 15.9±15.3 c 14.6±12.8 20.9±12.8 24.9±12.8 28.8±12.8 19.3±12.8 14.3±13.9 1971 I -__ c 11.1±13.3 8.9±13.3 10.6±13.3 8.6±13.3 8.1±14.1 7.4±14.5 Acli 1985 I 16.0±14.3 37.7±12.8 17.1±12.8 118.0±12.8 187.9±12.8 190.6±13.2 c 12.0±23.4 39.6±19.1 72.1±19.1 181.9±19.1 262.8±19.1353.4±19.1 1984 I 19.2±23.4 18.3±23.4 17.6±25.7 23.0±25.7 15.3±23.4 31.3±33.1 c 13.9±14.8 16.1±15.3 19.3±14.8 24.6±14.8 25.6±15.3 31.9±18.1 1983 I 13.9±19.1 16.6±20.3 12.4±19.1 12.1±19.1 11.6±19.1 14.1±20.3 c 13.3±16.6 11.0±16.6 13.5±16.6 16.0±16.6 18.0±18.1 16.1±18.1 1971 I . - - . c 14.0±16.2 9.5±17.0 12.9±17.0 10.3±17.0 5.9±17.0 5.9±19.9 acLi 1985 I 5.0±57.4 24.8±28.7 59.3±28.7 174.5±28.7 249.8±28.7 211.5±28.7 c 3.0±57.4 27.5±28.7 69.3±28.7 134.7±28.7 224.0±28.7 403.0±28.7 1984 I 21.0±40.6 3.0±57.4 6.5±40.6 13.0±40.6 6.0±57.4 5.0±57.4 c 12.4±19.1 11.4±19.1 9.8±19.1 16.0±19.1 17.3±19.1 13.5±20.3 1983 I 9:4±25.7 13.2±25.7 11.4±25.7 12.4±25.7 13.0±28.7 15.4±25.7 c 16.8±25.7 14.6±25.7 17.2±25.7 13.0±25.7 20.3±28.7 24.0±40.6 1971 I .. c 19.3±28.1 11.5±28.1 18.8±28.1 20.5±28.1 25.8±28.1 11.7±32.5 acli 1985 I 5.0±40.6 8.0±28.7 18.0±28.7 48.3±28.7 91.0±28.7 85.8±28.7 c 11.5±28.7 34.0±25.7 75.4±25.7 129.6±25.7 169.8±25.7 312.8±25.7 1984 I 14.0±33.1 18.3±33.1 18.7±33.1 23.7±33.1 23.0±40.7 46.0±40.6 c 14.9±19.1 22.6±20.3 18.9±19.1 33.3±20.3 30.6±19.1 27.7±21.7 1983 I 11.5±28.7 17.8±28.7 22.5±28.7 21.5±28.7 21.3±28.7 22.0±28.7 c 17.2±25.7 33.0±25.7 28.0±25.7 16.6±25.7 16.6±25.7 10.4±25.7 1971 I -_ _ c 8.5±39.8 8.0±39.8 7.0±39.8 6.5±39.8 9.5±39.8 7.0±56.3 Table 5.3.a Mean leaf number per stolon (±1 S.E.) of each cyanogenic/acyanogenic morph (phenotype)T. of repens in sites of different successional age. Symbols: I = insecticide-treated; C = control (natural levels of insect herbivory); - = no data available. 142

(0 Factor d.f. F P Site 2 49.75 *** Phenotype 3 1.16 N.S. Treatment 1 29.56 *** Sample 5 78.20 *** Site.Phenotype 6 4.32 *** Site.Treatment 2 28.03 *** Site.Sample 10 62.31 *** Phenotype.Treatment 3 0.84 N.S. Phenotype.Sample 15 0.51 N.S. Treatment. S ample 5 6.96 *** Total 1437

(») Factor d.f. F P Site 3 302.30 *** Phenotype 3 0.96 N.S. Sample 5 46.16 *** Site.Phenotype 9 1.46 N.S. Site.Sample 15 41.92 *** Phenotype.Sample 15 0.55 N.S. Site.Phenotype.Sample 45 0.46 N.S. Total 985

Table 5.3.b Results of analysis of variance of mean leaf number per stolon of each cyanogenic/acyanogenic phenotype of Trifolium repens in sites of different successional age. (i) 1985, 1984 & 1983 sites (insecticide and control treatments); (ii) 1985, 1984, 1983 and 1971 sites (control treatments only). Symbols: *** = p <0.001; ** = p<0.01; * = p<0.05; N.S. = not significant. 143

Figure 5.4. Seasonal variation in the percentage of damaged leaves recorded on cyanogenic/acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in the 1985 site, (a) cyanogenic morph; (b) all acyanogenic morphs combined; (c) each acyanogenic morph under natural levels of herbivory. Values are means (arcsine square root transformed) ± 1 S.E. Symbols represented are: (•) insecticide-treated; (o) control; (□ ) cyanoglucoside only (Acli); ( a) enzyme only (acLi); (a) homozygous acyanogenic (acli). 144

Figure 5.5. Seasonal variation in thepercentage of damaged leaves recorded on cyanogenic/acyanogenic morphs of T. repens under natural and reduced levels of insect herbivory in the 1984 site, (a) cyanogenic morph; (b) all acyanogenic morphs combined; (c) each acyanogenic morph under natural levels of herbivory. Values are means (arcsine square root transformed) ±1 S.E. Symbols represented are: (•) insecticide-treated; (o) control; (□ ) cyanoglucoside only (Acli); (a) enzyme only (acLi); (a) homozygous acyanogenic (acli). 145

Figure 5.6. Seasonal variation in the percentage of damaged leaves recorded on cyanogenic/acyanogenic morphs of 1 . repens under natural and reduced levels of insect herbivory in the 1983 site, (a) cyanogenic morph; (b) all acyanogenic morphs combined; (c) each acyanogenic morph under natural levels of herbivory. Values are means (arcsine square root transformed) ±1 S.E. Symbols represented are; (•) insecticide-treated; (o) control; (□ ) cyanoglucoside only (Acli); ( a) enzyme only (acLi); ( a) homozygous acyanogenic (acli). 146

a> Oh

Figure 5.7. Seasonal variation in the percentage of damaged leaves recorded on cyanogenic/acyanogenic morphs of T. repens under natural levels of insect herbivory in the 1971 site, (a) cyanogenic morph; (b) all acyanogenic morphs combined; (c) glucoside and enzyme/homozygous acyanogenic morphs. Values are means (arcsine square root transformed) ±1 S.E. Symbols represented are: (o) control; (□ ) cyanoglucoside only morph (Acli); (■ ) enzyme only and homozygous acyanogenic (acLi & acli) morphs combined. 147

1985 site Cyanogenic Acyanogenic (AcLi) (all morphs) (Acli) (acLi) (acli) I C I C I C I C I C Date 3/6 1.28 1.50 1.50 1.33 1.38 1.50 2.50 - 1.84 1.50 11/7 1.67 2.08 2.22 2.22 2.31 2.17 2.25 2.25 1.75 2.30 9/8 2.38 2.12 2.35 2.50 2.26 3.06 2.25 2.00 2.20 1.90 1/9 1.91 2.23 1.64 2.76 1.64 2.39 1.75 2.25 1.50 2.33 16/9 1.89 1.39 1.40 1.45 1.36 1.50 1.50 1.50 1.50 1.33 1984 site Cyanogenic Acyanogenic (AcLi) (all morphs) (Acli) (acLi) (acli) I C I C I C I C I C Date 5/6 1.13 0.75 1.50 1.04 1.05 0.89 0.50 1.08 1.17 1.27 15/7 1.19 1.10 1.07 1.28 1.06 1.15 1.00 1.40 1.17 1.50 13/8 1.54 1.18 1.25 1.36 0.93 1.27 1.50 1.70 1.50 1.17 3/9 1.71 1.94 1.30 1.53 1.70 1.37 1.00 1.63 1.17 1.70 20/9 0.94 1.13 1.10 1.15 1.00 1.14 1.50 1.38 1.17 0.94 1983 site Cyanogenic Acyanogenic (AcLi) (all morphs) (Acli) (acLi) (acli) I C I C I C I C I C Date 9/6 1.10 0.98 1.19 1.03 0.94 1.13 1.70 0.88 1.10 0.90 18/7 1.17 1.50 1.56 1.54 1.39 1.63 1.90 1.36 1.75 1.50 15/8 1.30 1.70 1.39 1.88 1.28 2.00 1.70 1.67 1.25 1.90 5/9 1.70 1.61 1.78 1.71 1.61 1.90 2.10 1.10 2.00 2.10 23/9 1.30 1.19 1.28 1.00 1.28 1.17 1.30 1.00 1.13 0.90 1971 site Cyanogenic Acyanogenic (AcLi) (all morphs) (Acli) (acLi & acli) I C I C I C I C Date 12/6 - 1.14 - 1.17 - 1.03 - 1.50 23/7 - 1.22 - 0.89 - 0.75 - 1.17 19/8 - 1.22 - 1.38 - 1.14 - 1.83 9/9 - 1.72 - 1.32 - 1.14 - 1.67 25/9 - 0.88 - 0.97 - 0.96 - 1.00 (See Southwood 1978) Table 5.4. Leaf life expectancy A of individual leaves of each cyanogenic/acyanogenic morph (phenotype) of T. repens in sites of different successional age. Symbols: I = insecticide-treated; C = control; - = no sample; Acli = cyanoglucoside only; acLi = enzyme only; acli = homozygous acyanogenic. 148 leaves of cyanogenic stolons tend not to live as long as acyanogenic leaves on at least three dates in every site. Unfortunately, the dates when leaf life expectancy of acyanogenic leaves was higher than cyanogenic were not consistent in every site and therefore no general trends can be seen regarding the relationship between leaf life expectancy and insect herbivory. The amount of leaf area removed, as a direct result of insect herbivory, from cyanogenic and all acyanogenic phenotypes is shown in Figure 5.8. and from each acyanogenic phenotype in Figure 5.9. The reduction of natural levels of insect herbivory tended to decrease the amount of lamina removed from each leaf during the season, although over 30% of the leaves sampled were moderately to heavily eaten by insect herbivores. Comparison of the three acyanogenic phenotypes indicated that the acli phenotype was more heavily eaten in the 1985 and 1984 sites (insecticide treatment), however, there were no general trends in the acyanogenic phenotypes in control plots. Under natural levels of insect herbivory, peak leaf damage occurred in the 6-25% damage category in both cyanogenic and acyanogenic morphs in the 1985,1984 and 1983 sites. In the 1971 site peak damage occurred in the 1-5% category, although in this site more leaves were totally removed than in any other site. Differences between the cyanogenic and acyanogenic morphs occurred, but tended to be smaller than expected and were not consistent between sites. For instance, the percentage of cyanogenic leaves not eaten was greater in the 1985 and 1983 sites, but in the 1984 and 1971 sites there were 2-4% more acyanogenic leaves not damaged. In the other damage categories similar differences occurred between cyanogenic and acyanogenic morphs. However, when the ratios of untouched to eaten leaves were compared (Table 5.5) there was no significant difference. A closer examination of the eaten leaves (Table 5.5.) showed that both cyanogenic and acyanogenic morphs were equally "nibbled" (<5% removed) and "heavily eaten" (6-100% removed) in the 1985, 1983 and 1971 sites. In the 1984 site cyanogenic leaves were significantly more "heavily eaten".

53,3. Effects of Cyanogenesis on Reproductive Characteristics The mean number of inflorescences and seedheads produced by stolons of different phenotypes under natural and reduced levels of insect herbivory are given in Table 5.6. and 5.7. respectively. There appears to be no significant difference between the number of inflorescences and seedheads produced by each phenotype in either treatment. Thus, insect herbivores do not seem to be influencing these reproductive characteristics. There was, however, a treatment x phenotype interaction for inflorescence number which indicated that in some of the treatments there was a difference in the number of Percentage of leaves in each damage category Figure 5.8. The percentage of leaves recorded in each damage damage each in recorded leaves of percentage The 5.8. Figure aeoybtenJn 95adMy18 rmcaoei ■§) (■ cyanogenic 1986 from May and 1985 June between category from (a) insecticide-treated and (b) control treatments in (i) 1985, (i) in treatments control (b) and insecticide-treated (a) from and all acyanogenic (□ ) morphs of of morphs ) (□ acyanogenic all and (ii) 1984, (iii) 1983 and (iv) 1971 sites. 1971 (iv) and 1983 (iii) 1984, (ii) Estimated leaf area removed removed area leaf Estimated (%) T. repens T. Estimated leaf area removed (%) removed area leaf Estimated . Leaves were sampled sampled were Leaves . 149 Percentage of leaves in each damage category 40 60 40 20 40 60 60 20 20 0 0 0 Estimated leaf area removed (%) removed area leaf Estimated between June 1985 and May 1986 in each acyanogenic morph of of morph acyanogenic each in 1986 May 1985and June between Figure 5.9. Percentage of leaves recorded in each damage category category damage each in recorded leaves of Percentage 5.9. Figure only (acLi) morph; (®0 homozygous acyanogenic (acli) morph. (acli) acyanogenic homozygous (®0 morph; (acLi) only sites. Symbols: (■ ) glucoside only (Acli) morph; (□ ) enzyme enzyme ) (□ morph; (Acli) only glucoside ) (■ Symbols: sites. T. repens. T. (b) control treatments in (i) 1985, (ii) 1984, (iii) 1983 and (iv) 1971 (iv) and 1983 (iii) 1984, (ii) 1985, (i) in treatments control (b) 5 V * Leaves were sampled from (a) insecticide-treated and and insecticide-treated (a) from sampled were Leaves n. in. & Estimated leaf area removed (%) removed area leaf Estimated 150 Damage cyanogenic acyanogenic untouched (0% removed) 13 3 1985 site eaten (1-100% removed) 110 84 X2 = 2.73 p = r t3 . untouched (0% removed) 4 13 1984 site eaten (1-100% removed) 51 111 X2 = 0.16p =rt.s. untouched (0% removed) 11 5 1983 site eaten (1-100% removed) 68 87 X2 = 2.68 p = ns. untouched (0% removed) 13 14 1971 site eaten (1-100% removed) 65 58 XI = 0.05p = rt-s. nibbled (<5% removed) 38 29 1985 site eaten (6-100% removed) 72 55 X2 — O.OOp = rt.s. nibbled (<5% removed) 1 36 1984 site eaten (6-100% removed) 44 75 X2 = 5.35p= 0.021 nibbled (<5% removed) 18 29 1983 site eaten (6-100% removed) 50 58 X2 = 0.5 6p = n s . nibbled (<5% removed) 34 23 1971 site eaten (6-100% removed) 31 35 X2= 1.50p = n.s. Table 5.5. Contingency x2 for the partition of eating scores between cyanogenic and acyanogenic stolons in sites of different successional age (control treatment only). Symbols: n.s.= not significant. All x2 values have been calculated using Yate’s correction for continuity. 1985 site 1984 site 1983 site

Phenotype I C I C I C Factor d.f. P P Nean 0.56 1.08 1.33 1.75 2.07 1.24 Phenotype 3 0.95 N.S. A c L l S. B. 0.31 0.26 0.25 0.29 0.34 0.26 Treatment 1 0.18 N.S. (18) (26) (27) (20) (15) (25) Site 2 11.38 I l l PhenolTreat 3 1.55 N.S. Nean 1.05 0.89 1.50 1.78 1.69 0.55 PhenolSite 6 1.86 N.S. A c l i S. B. 0.29 0.44 0.38 0.31 0.44 0.29 TreatlSite 2 3.72 1 (21) (9) (12) (18) (9) (20) PhenalTreatlSlte 6 4.63 III Error 262 Nean 0 0.50 6.00 1.00 0.20 0.57 Total 285 a c L i S.B. 0 0.65 0.92 0.38 0.59 0.49 (4) (4) (2) (12) (5) (7) Nean 0 0.83 2.50 2.30 1.40 0.80 a c l i S.B. 0 0.53 0.92 0.41 0.59 0.58 (4) (6) (2) (tO) (5) (51

Table 5.6. The effect of reducing levels of insect herb ivory and site age on mean Inflorescence number in 4 phenotypes of T. repens. Figures in parenthesis represent the number of replicates. Symbols: I = Insecticide-treated; C = control; *** = p<0.001; O = p<0.01; * = p<0.05; N.S. = p>0.05 (not significant). 1965 site 1984 site 1963 site

Phenotype I C I CI C Factor d. f. F P Kean 0.89 1.04 1.67 1.85 1.73 0.68 Phenotype 3 1.41 N.S. A cL 1 S.B. 0.33 0.28 0.27 0.32 0.36 0.28 Treatment 1 0.00 N.S. (18) (26) (27) (20) (15) (25) Site 2 12.63 I l l PhenolTreat 3 0.64 H.S. Mean 1.00 0.67 0.75 1.89 1.33 0.65 PhenofSite 6 1.06 N.S. A c l i S.B. 0.31 0.47 0.41 0.33 0.47 0.32 TreatlSite 2 2.09 N.S. (21) (9) (12) (161 (9) (20) PhenolTreatlSite 6 2.33 1 Brror 262 Kean 0.50 0.25 3.50 1.33 0.20 0.86 Total 285 a c L i S. B. 0.70 0.70 1.00 0.41 0.63 0.53 (4) (4) (2) (12) (5) (7) Kean 0 1.33 3.50 2.70 0.80 0.80 a c l i S.B. 0 0.58 1.00 0.45 0.63 0.63 (4) (6) (2) (10) (5) (5)

Table 5.7. The effect of reducing levels of insect herb ivory and site age on mean seedhead number in 4 phenotypes of T. repens. Figures in parenthesis represent the number of replicates. Symbols: I = insecticide-treated; C = control; *#* = p<0.001; ft = p<0.01; t = p<0.05; N.S. = p>0.05 (not significant). 153 154 inflorescences produced by different phenotypes. There was also a phenotype x treatment x site interaction for both inflorescence and seedhead number indicating differences between phenotypes which were consistent between treatments in all sites. However, there were in general no clear trends in inflorescence or seedhead number. Table 5.8. gives details of the effect of reducing insect herbivory on a number of seedhead characteristics. Here, the number of replicates was low, so all the acyanogenic phenotypes have been pooled. Even so, the number of replicates were still too small for significant trends to be established. For a comparison of the differences between treatment and site for T. repens as a species, see Chapter 4.

5.3.4. Successional Trends All four phenotypes share similar successional trends in both vegetative and reproductive characteristics. The frequency of cyanogenesis is initially determined by the proportion of seedlings of each phenotype germinating from the seed bank in a ruderal site. Seedling recruitment after the ruderal phase is limited (Cahn & Harper 1976a), thus the frequency of cyanogenesis in the older sites is dependent on clone survival and vegetative propagation. Figure 5.3. indicates that stolon survival in the 1985 site was initially high, but in older sites survival was lower. Height of the leaf canopy was greatest in the 1984 site and lowest in the 1985 and 1971 sites (Table 5.2.a). Leaf number was significantly greater in the 1985 site, but in the older sites due to stolon fragmentation, leaf number was much lower especially in the 1971 site (Table 5.3.a & b). Leaf life expectancy (Table 5.4.) was generally higher and the percentage of leaves damaged lower in the youngest site. In the older sites, there tended to be no noticeable difference in leaf life expectancy and percentage of leaves damaged. Changes in the reproductive characteristics of each phenotype in control treatments were apparent during succession. Peak inflorescence and seedhead number was reached in the 1984 site, after which production decreased with successional age (Table 5.9.). A similar trend was seen in the number of florets per seedhead (all phenotypes) and seeds per seedhead (cyanogenic phenotype). However, in the acyanogenic morph the largest number of seeds per seedhead was recorded in the 1985 site, but seed number subsequently declined along the successional gradient (Table 5.10). 1089 site 1911 site 1083 site

lorpk I C I C I C Factor d. f. P p

l e u 51.40 S3. SO 11.71 61.00 16.00 5Q.50 lorpk l 1.06 I.S. Cycaogeiic S.B. 0.19 9.08 l.lt 6.71 7.18 0.01 Treatment I 1.03 I.S. la. of (SI (61 (71 (III (1) (61 Site 2 2.92 1.3. florets/ Error 60 seedkead K e u 51.10 ‘ 16.50 51.00 67.50 31.00 51.00 Total 73 Acyaiojealc S.B. M S 11.11 15.73 11.12 0.09 1.71 (5) (11 (21 (11 (51 (111

l e u M l l. It 1.30 l.ll 1.30 0.71 lorpk 1 0.06 1.3. Cyuogesle S.B. 0.33 0.31 0.21 0.23 0.28 0.31 Treatuit 1 0.50 I.S. la. of (5) (II (71 (111 (1) (61 Site 2 1.33 1.3. seeds/ Error 69 floret l e u 1.11 1.33 1.30 0.17 0.72 0.60 Total 73 Aeyuogeiie S.B. 0. 33 0.37 0.53 0.37 0.33 0.23 (51 (1) (21 (1) (51 (111

l e u 61.10 60.13 21.11 01.18 76.25 52.00 Marpk l 1.20 I.S. Cyaiogeiie S.B. 11.70 11.93 20.17 11.19 10.53 22.55 Treatneit I 0.10 I.S. Seed (51 (61 (71 (til (6) (61 Site 2 2.11 I.S. 10tier/ Error 69 seedketd u 61. 60 60.15 01.00 56.50 29.10 13.73 Total 73le Acyaiajeuic S.B. 11.70 17.11 30.05 27.11 21.70 16.65 (5) (11 (21 (11 (51 (111

l e u 0.15 0.51 0.12 0.11 0.11 0.36 lorpk 1 0.51 I.S. lidlridoil Cyaiojenic S.B. 0.06 0.06 0.09 0.01 0.05 0.06 Treatneot l 1.57 1.3. seed (51 (II (71 (III (61 (61 Site 2 0.55 I.S. w l fit/ Error 69 seedkeid l e u 0. 36 0. 15 0.31 0.17 0.33 0.15 Total 73 (■gl Acyaiojealc S. E. 0.01 0.07 0.10 0.07 0.08 0.06 (51 (11 (21 (11 (51 (111

Table 5.8. The effect of reducing the levels of Insect herb Ivory on reproductive characteristics of cyanogenic and acyanogenla morphs of T. repena in a ruderal site and two early suocesslonal sites. Figures in parenthesis represent the number of replicates. 155 Symbols: I a Insecticide-*treated; C a control; N.S. a p>0.05 (not significant). 156

1985 1984 1983 1971 Pheiotype site site site site Pactor d.f. F P Keai 1.08 1.75 1.24 0.68 Pieiotype 3 1.53 I.S. A c L l S.E. 0.25 0.28 0.25 0.25 Site 3 8.24 Iff (26) (20) (25) (25) Pkeiotypefsite 9 0.60 I.S. firrOr 194 leai 0.89 1.78 0.55 0.40 Total 209 Iiflor- A c l I S.E. 0.42 0.30 0.26 0.33 (9) (18) (20) (15) lean 0.50 1.00 0.57 0.80 Iniher a c L i S.E. 0.63 0.36 0.48 0.56 (4) (12) (7) (5) leas 0.83 2.30 0.80 0.67 a c l i S.E. 0.51 0.40 0.56 0.73 (6) (10) (5) (3) leai 1.04 1.65 0.68 0.64 Plteiotype 3 2.13 I.S. A cL i S.E. 0.25 0.29 0.25 0.25 Site 3 12.49 Iff (26) (20) (25) (25) Pkeaotypefsite 9 0.46 I.S. Brror 194 leai 0.67 1.89 0.65 0.47 Total 209 A c l i S.E. 0.42 0.30 0.29 0.33 Seediead (9) (16) (20) (15) limber lean 0.25 1.33 0.86 0.40 a c L i S.B. 0.64 0.37 0.46 0.57 (4) (12) (7) (5) leai 1.33 2.70 0.80 0.33 a c l i S.E. 0.5! 0.40 0.57 0.74 (6) (10) (5) (3)

Table 5.9. The effect of site age on maximum inflorescence and seedhead number in 4 phenotypes of T. repens (control plots only). Figures in parenthesis represent the number of replicates. Symbols; «-* = p<0.001; ** = p<0.01; * = p<0.05; N.S. = p>0.05 (not significant). 157

1985 1984 1983 1971 lorph s ite site site site Pactor d. f . F P

le a i 53.50 62.09 50.50 34.25 Morph. 1 0.42 8.S. Cyaiogeiic S.B. 7.63 5.63 7.62 6.60 Site 3 7.08 11 la . of (6) (11) (6) (8) Morphiaite 3 0.37 H.S. flo rets/ Error 47 seedhead le a i 46.50 67.50 51.00 25.60 Total 54 Acyaiogeiic S.E. 9.33 9.33 5.63 8.34 (4) (4) (It ) (5)

le a i 1 .12 1.41 0.72 0.54 Morph 1 0.11 N.S. Cyaiogeiic S.E. 0.29 0.21 0.29 0.25 Site 3 2.54 N.S. la . of (6) (It ) (6) (8) Morphiaite 3 0.79 N.S. seeds/ Error 47 flo ret le a i 1.33 0.87 0.89 0.74 Total 54 Acyaiogeiic S.B. 0.35 0.35 0.21 0.31 (4) (4) (It ) (5)

le a i 60.83 94.18 52.00 18.75 Morph 1 1.05 I.S. Cyaiogeiic S.E. 20.04 14.80 20.04 17.36 Site 3 4.38 11 Seed (6) (11) (6) (8) Morphiaite 3 0.62 N.S. limber/ Brror 47 seedhead le a i 80.25 56.50 43.73 17.20 Total 54 Acyaiageiic S.E. 24.55 24.55 14.80 21.96 (4) (4) ( t l) (5)

le a i 0.51 0.44 0.38 0.42 Morph 1 0.13 N.S. Individn&l Cyaiogeiic S.E. 0.05 0.04 0.05 0.44 Site 3 0.84 N.S. seed (6) (11) (6) (8) Morphlsite 3 0.55 N.S. w ight/ Brror 47 seedkead le a i 0.45 0.47 0.45 0.43 Total 54 (■$) Acyaiageiic S.E. 0.06 0.06 0.04 0.06 (4) (4) (11) (5)

Table 5.10. The effect of site a ge on reproductive characteristics of cyanogenic and acyanogenic morphs of T. repens (control plots only). Figures in parenthesis represent the number of replicates. Symbols; ** = p<0.01; N.S. = p>0.05 (not significant). 158

5.4. Discussion Populations of T. repens have a high level of genetic diversity which is maintained by the variable nature of the biotic environment and general physical factors (Burdon 1980a). Changes in both these biotic and abiotic components take place during secondary succession and, as a result, the composition of phenotypes within different polymorphisms may be altered. Changes in the frequency of occurrence of cyanogenesis in T. repens take place during succession. However, these changes do not follow specific successional trends, because factors affecting the distribution of cyanogenic and acyanogenic phenotypes will vary from site to site. One such factor is the distribution of cyanogenic L. corniculatus. Jones (1968) found that, when both T. repens and L. corniculatus occur in the same habitat, the frequency of cyanogenesis in T. repens was lower than when T. repens occurred by itself. Jones (1968) suggested that the factors limiting the distribution of L. corniculatus may be responsible for the higher frequency of cyanogenesis in non-mixed populations of T. repens. Thus, the high frequency of cyanogenesis in L. corniculatus in the 1983 and 1971 sites may confer selective advantage to the acyanogenic phenotypes of T. repens. L. corniculatus occurred at very low frequencies in the 1982 and 1980 sites (see Chapter 2) and this correlates with higher frequencies of cyanogenesis in T. repens in both sites. The contention of Daday (1954a, b, 1965) that low winter temperature may be disadvantageous to the cyanogenic phenotype of T. repens is supported in part by the lower frequency of cyanogenic phenotypes sampled during February 1985 in the 1983, 1982 and 1971 sites. However, the marginally higher frequency of cyanogenesis in the 1984 site does question Daday’s suggestion that low winter temperature is a major factor responsible for the polymorphism. The increased frequency in cyanogenesis in the 1983, 1982 and 1971 sites from February to May 1985 (control sites) may be the result of factors, responsible for the selection of the cyanogenic phenotype, being encouraged by increased mean temperature. Jones (1970) suggested that the activity of herbivores might be one of these factors. The expected outcome of reducing natural levels of insect herbivory would be to favour the more palatable acyanogenic phenotypes, whereas selective advantage of the cyanogenic phenotype would be maintained under natural levels of herbivory. In the 1985 and 1983 sites, insecticide application increased the frequency of the acyanogenic morphs, with the effect being most marked in the former site, presumably as a result of increased seedling survival of the acyanogenic morphs. These findings tend to contradict the results of Miller et al. (1975) but support those of Dritschilo et a/.(1979). An elegant 159 bioassay performed by Woodhead & Bemays (1977) showed that HCN released at suitable concentrations into the preoral cavity of Locusta migratoria (L.) inhibited feeding. They concluded that the cyanide released from Sorghum bicolor was an effective defence against these acridid herbivores. Cyanogenesis in seedlings of T. repens in the 1985 site may well provide protection from colonizing generalist insect herbivores. The major constituent of the acyanogenic morph is the Acli phenotype. The reason may be due to the Ac locus being associated with other fitness traits other than cyanogenesis. It was originally thought that B-glucosidases in the gut of herbivores may hydrolyse the ingested cyanoglucoside resulting in the release of HCN which may deter the herbivore from feeding. Subsequent studies have shown that HCN is liberated within the herbivore’s gut. However, no adverse effects have been reported for molluscs (Dirzo & Harper 1982a), insects (Bemays 1983) and vertebrates (Jones 1972). The ability to detoxify cyanide released within these animals has therefore been postulated, since herbivores from these groups show no selective feeding between the Acli and other acyanogenic phenotypes. Selection of the Ac locus has also been linked with environmental factors (Daday 1965, Foulds & Grime 1972b). However, in the case of moisture stress, T. repens shows a genetic variability for root length and drought tolerance (Ennos 1985), but whether this characteristic is linked to cyanogenesis is not known. The survival of all four phenotypes was greater in the 1985 site and this may well be associated with the low structural and architectural diversity of the vegetation (Southwood et al. 1979) resulting in low interspecific competition. As succession proceeds, competition becomes more intense especially from the grasses (Brown & Southwood 1987) resulting in increased mortality of all four morphs in the 1984 and 1983 sites. Mortality is further accentuated by winter conditions. Whittaker (1979) suggested that the outcome of competition might be mediated or modified by invertebrate grazing. Thus, reducing insect herbivory may have increased the competitive ability of each phenotype and this is reflected by a better survivorship during the season. The effect of reducing insect herbivory in the 1971 site was not measured, hence, the impact of insect herbivores cannot be independently assessed. However, it is interesting that by spring 1985, marked stolons of the acLi and acli phenotypes in the 1983 and 1971 sites respectively were eliminated under natural levels of insect herbivory. This suggests that acyanogenic phenotypes may suffer mortality during winter conditions especially when stressed by herbivores. There was no significant correlation between cyanogenesis and vegetative characteristics (i.e. height and leaf number). Interpretation in terms of partitioning of resources (Cody 1966) would infer that the cyanogenic phenotype would be at an 160 advantage under differential grazing and as a result stolon height and leaf number should be greater. The exact opposite would be expected when the levels of herbivory are reduced. However, studies investigating the variation of growth rate within a clover population (herbivore-free) revealed that, while most individuals had similar growth rates, a few individuals had very slow growth rates giving rise to a skewed size distribution (Burdon & Harper 1980). The reason for the occurrence of these few anomalous individuals was thought to be due to possible infection by viral pathogens. Thus, in natural populations of T. repens variability in height and leaf number may not be detected between cyanogenic and acyanogenic phenotypes, since factors other than insect herbivory may be involved. Such factors include the size, age and identity of immediate neighbours (Turkington & Harper 1979a), differential attack from fungal pathogens (Burdon 1980b), ability to withstand moisture stress (Ennos 1985), and so on. Burdon (1980a) considered T. repens within the concept of sisyphean fitness, since genotypes sampled in the field represent successful clones which have achieved near maximum fitness via an interaction of momentarily effective combination of genes. The experimental manipulation of insect populations in the successional sites demonstrated that height and leaf number were reduced by insecticide application. Suggestions that changes in the level of insect herbivory might alter the balance between vegetative growth and sexual reproduction seem unlikely, since insecticide application does not significantly increase inflorescence or seedhead number. There was also no effect on floret number, seed yield or individual seed weight between treatments, although the number of replicates was low. As these effects were consistent between phenotypes, it seems probable that differences in height and leaf number between treatments may be a result of changes in the height and structure of the surrounding vegetation (Brown, Jepsen & Gibson 1988). However, successional trends in both vegetative and reproductive characteristics were apparent, although there were no obvious trends between phenotypes. These general changes are discussed in Chapter 4. Leaf turnover was highest during June and at the end of the season in all sites. A number of studies have shown that leaf longevity declines with increasing density and shading (White 1979) and this may explain the higher life expectancy of leaves in the 1985 site where the vegetation is less dense. During August and early September, leaves in all sites have the longest life expectancy and this may be linked to the mid seasonal lull in grass growth. There were no consistent differences in leaf life expectancy between treatments and phenotypes. Turkington (1983) found no difference in average lifespan of leaves of T. repens when grown in the presence of different grass neighbours. Thus, leaf age in T. 161 repens may not be correlated simply to single factors and considerations of possible interactions may have to be made. The reduced effectiveness of a contact insecticide associated with T. repens was discussed in Chapter 4. Changes in the proportion of leaves damaged during the season in each site were probably a reflection of herbivore abundance. In the 1985 site, the initial increase in the proportion of leaves damaged was probably associated with immigration and colonization of insect herbivores (mainly generalists) from the surrounding area. Whereas, the increase in the proportion of leaves damaged during the latter half of the season was probably associated with an increase in the immigration and abundance of specialist herbivores and may be a second generation of the initial colonizers (see Brown 1985). In the 1984, 1983 and 1971 sites, the pattern of damage was similar throughout the season. At the beginning of June, there was a high percentage of leaves damaged probably by the previously established herbivores emerging from their overwintering sites on or near the host plant. The decrease in damage during July and August probably reflects their demise, whereas the increase later in the season may reflect either the emergence of a second generation or the effects of other herbivore species. There was no difference in the percentage damage received by cyanogenic and acyanogenic plants and this was matched by no significant difference between eaten/uneaten leaves of each morph in any site. However, when nibbled and heavily eaten leaves were compared, the 1984 site was an exception, since more cyanogenic leaves were recorded in the latter category, although this was only marginally significant. Thus, the defensive role of cyanogenesis against insect herbivores was not clearly defined. Insects are physiologically very tolerant of cyanide, compared with mammals (Bemays 1983), and those specializing on cyanogenic plants should therefore be even more tolerant since selection pressures will be greater. For example, all life stages of the bumet moth (Lepidoptera: Zygaenidae) are themselves cyanogenic and this is not related to whether the larvae feed on cyanogenic or acyanogenic phenotypes of L. corniculatus or T. repens, since they appear to be able to synthesize cyanogenic substances de novo (Jones 1972). Consequently, such insects will be indiscriminate of the cyanogenic morphs, because of their ability to tolerate high concentrations of cyanide. Other insects which specialize on cyanogenic food plants may also have similar detoxification mechanisms, both physiological and behavioural. Several weevil species are known to be associated with L. corniculatus (Jones 1973); however, only one ( Hypera plantaginis (DeGeer)) is known to be cyanide tolerant (Parsons & Rothschild 1964). Weevils were also shown to be non-selective between cyanogenic and acyanogenic morphs of T. repens (Dirzo & Harper 1982b) and this further * substantiates the tolerance of specialist insect herbivores to cyanide. The crop plant cassava, Manihot escunlenta Crantz., is extremely cyanogenic 162 with relatively few pest insects, except Zonocerus variegatus (L.) in West Africa (Bemays et al. 1977). Even so, this specialist herbivore can inflict high levels of damage on the plant. Thus, much of the insect damage recorded on T. repens may well have been the result of its repertoire of specialist insect herbivores. Therefore, cyanogenesis may only be an effective deterrent against herbivores that include T. repens only as an occasional dietary item. Such generalist herbivores will be influenced by the amount of cyanide released, alternative food plants available and, as a result of the latter, level of hunger. No definite evidence is presented here of selection by generalist herbivores, but it is interesting that in sites where interspecific competition was high (1982 and 1980 sites), the frequency of cyanogenesis was also high. In the 1971 site, light grazing by rabbits enabled this light-demanding plant to persist (Burdon 1983), since interspecific competition was reduced (see Chapter 2). The incidence of the acyanogenic morph was also higher in this site, although the frequency of whole leaves removed (most likely by rabbits) in this morph was greater than in the cyanogenic morph (Figure 5.7.). The reasons for the differences in the frequency of cyanogenesis in the 1982 and 1971 sites may be explained in terms of changes in the balance of defence and regrowth. This hypothesis was suggested by van der Meijden, Wijn & Verkaar (1988) as alternative strategies in the plant’s struggle against its herbivores, although there are similarities to Cody’s allocation of resources (Cody 1966). However, when T. repens is under intense interspecific competition, its regrowth capacity may be seriously impaired, hence selection for cyanogenic (defence) phenotypes may take place as protection is afforded from generalist herbivores. Such protection was substantiated by the personal observation of a quicker and more intense reaction with sodium picrate paper when leaves from the 1982 site were tested. The regrowth strategy is most likely to occur when interspecific competition is reduced, such as in swards grazed by vertebrates (e.g. 1971 site). Here T. repens may be subjected to different selective pressures, e.g. leaf mark (Cahn & Harper 1976a, b). Under these circumstances, trade-offs between regrowth and defence, will select for the ability to recover rapidly from the effects of defoliation. As a result, acyanogenic (regrowth) phenotypes may be favoured since they are more competitive. However, as other selective agents are also acting at the same time in both sites, a polymorphism results. The effects of differential grazing by different herbivores has been investigated by Georgiadis and McNaughton (1988). In their study the grass, Cynodon plectostachys (K.Schum) Piliger, survives best under conditions of heavy grazing and trampling by large vertebrate herbivores. However, when defoliation by Spodoptera exempta Walker is particularly intense, cyanogenesis is induced (although this may need substantiating) in C. 163 plectostachys. The high levels of cyanide in the leaves do not deter S. exempta which feeds preferentially on Cynodon species even though they are sufficient to kill domestic cattle. Such cases of mortality are rare and it is more likely that cyanide acts as a deterrent to grazers (Compton and Jones 1985). What is apparent from many of the recent studies into the maintenance of cyanogenesis in T. repens is that no single factor is solely responsible. Insect herbivory is therefore only one of a host of factors affecting the polymorphism. Even so, interactions between insect herbivory and other factors do occur and these will vary depending on the conditions imposed by a particular habitat. Hence, "different plants in different habitats may respond to selection in entirely different ways and therefore contrary explanations of the role of cyanogenesis within species, let alone between species are to be expected " (Jones, Keymer & Ellis 1978). Further studies of within-species and even within-clonal differences to changes in the reaction to competing neighbours (Turkington & Harper 1979a), changes in the responses to physical factors of the environment, changes in the differential susceptibility to pathogens (e.g. Burdon 1980b) and to differences in grazing by different herbivores are essential at the local scale for a fuller understanding of the forces which drive this polymorphism. The interplay between cyanogenesis and the herbivore is two sided, since the herbivore can alter the balance of cyanogenesis and the effects of cyanogenesis can alter the dynamics of the herbivore. The role of the specialist insect herbivore in this reaction will be investigated experimentally in the next chapter. Chapter 6

The Effect of the Plant on Preference, Fecundity and Survival of the Insect Herbivore

6.1 Introduction

Herbivory has been shown to reduce plant fitness in terms of growth and survival (e.g. Brown et al. 1987, Gange et al. in press, Hendrix 1988, Waloff & Richards 1977). In a bid to overcome the disadvantage imposed by herbivory, plants have evolved an armoury of physical defences (e.g. spines, hairs,leaf toughness) and chemical defences (e.g. alkaloids, glucosides, non-essential amino acids, tannins). Herbivores range from specialist to generalist and, as a consequence, plants have adapted their chemical defences to counteract both strategies. Rhoades and Cates (1976) put forward the idea that ephemeral, non-apparent plants will tend to have toxic divergent (i.e. dissimilar) chemical defences that deter generalist herbivores but not specialists and their escape from herbivory is determined by heterogeneity in time and space. Apparent plants (see Feeny 1976), on the other hand, cannot escape in time or space and have evolved convergent (i.e. similar), digestibility-reducing compounds effective against both specialist and generalist herbivores. In this hypothesis, it is assumed that plants characteristic of early successional habitats should support specialist herbivores and that generalism should be limited. However, these ideas do not take account of plants which have a chemical defence that is polymorphic. Two of the legume species (T. repens and L. corniculatus) found in the sites (see Chapter 2) were polymorphic for the character of cyanogenesis and as a result their susceptibility to herbivory can vary. Generalist gastropod molluscs have been shown to preferentially eat acyanogenic L. corniculatus in the laboratory (Crawford-Sidebotham 1972, Keymer & Ellis 1978) and in natural populations (Compton, Beesley & Jones 1983). The same selection has also been shown to occur in small mammals (Jones 1962, 1966, Compton, Newsome & Jones 1983) and generalist insects (Compton & Jones 1985). Preference for acyanogenic T. repens has been demonstrated in molluscs (Crawford-Sidebotham 1972, Angseesing & Angseesing 1973) and insects (Dritschilo et al. 1979). Specialist herbivores may gain marginal benefits from eating acyanogenic morphs through a reduction in detoxification costs (Feeny 1975), but the time spent searching for these plants (as well as other factors) may reduce the selective advantage. Therefore, it probably pays these herbivores to be non-selective 165 towards cyanogenic plants. Such "adapted" insects have been shown to have little or no preference by Compton & Jones (1985), Dirzo & Harper (1982b), Lane (1962) and Jones (1966). However, preference exhibited by generalist and specialist herbivores may not be rigidly fixed, but may be determined by the relative density of a particular morph or phenotype in the vegetation. Although it was suggested by Harper(1969) that this sort of selection may exist between species, the idea can be extended towards phenotypes of a single species since variation in herbivory within a species is often greater than between species (Jones 1959). Such differential selection has been termed frequency-dependent selection and has been demonstrated in predator-prey systems (Murdock 1969, Murdock & Oaten 1975, Popham 1941, 1943), in vertebrates (Allen 1972, 1976, Fullick & Greenwood 1979, Horsley et al. 1979) and invertebrates (Chandra & Williams 1983, Cottam 1985). All these studies have involved prey types of different species (of plants or animals) or baits varying in colour and size. No study to date has included a single species polymorphic in one or more characters, even though it has been shown that frequency dependent selection is capable of maintaining polymorphism in a population (Clark 1962, Clark & O’Donald 1964). The effects of frequency-dependent selection can be either stabilizing or destabilizing. When a species or phenotype in a community is selectively grazed only when the species or phenotype is common, the predator (or herbivore) will increase diversity in the community, so selection is destabilizing and is termed pro-apostatic. However, selection against species or phenotypes which are rare will increase stability and consequently reduce diversity, in this case selection is termed anti-apostatic. There is also a third component of selection, whereby a species or phenotype is always preferred and selected for whether rare or common, this type of selection is called frequency-independent selection. This selection may be important in promoting diversity when the preferred species is a dominant competitor (Tansley & Adamson 1925). Host plant quality may also be important particularly when a preference is indicated for a certain species or phenotype, but there is no correlation between preference, survival and fecundity. Chandra & Williams (1983) demonstrated that Schistocerca gregaria (Forskal) readily ate Cyperus L. but only sustained poor growth and survival. Lemon grass (Cymbopogon citratus Staph.) was not readily accepted and was proved to be an unsuitable host, whereas wheat was a preferred host sustaining growth and survival. It seems likely that a continuum of preference and suitability exists with an unsuitable host being perceived at one extreme and at the other, a plant which is suitable being readily devoured. Suitability of the host will fluctuate, since changes in host plant quality with age have been noted by Feeny (1970) and these differences are associated with changes 166 in plant toughness, water content, nitrogen levels and concentration of secondary chemicals. Differences in host plant quality within individuals of the same species may also be important particularly in sedentary herbivores (Edmunds & Alstad 1978). The work presented in this chapter reports the responses of nine species of Curculionoidea to leaves of four species of legume and four cyanogenic/acyanogenic phenotypes of T. repens. An investigation into the possible frequency-dependent grazing behaviour of Sitona lineatus on cyanogenic/acyanogenic T. repens was carried out using the technique of Cottam (1985), whose test was based on presenting the animal with a choice of at least two food types in replacement series (de Wit 1960) and comparing the amounts of each food type eaten at each relative frequency. Finally, the effect of host plant quality for each cyanogenic/acyanogenic phenotype of T. repens on the fecundity, rate of increase and longevity of Acyrthosiphon pisum was evaluated.

6.2. Materials and Methods 62.1. Stock Plants Four species of stock plants were grown from seed in an unheated greenhouse. The seeds of T. repens (variety S.184) and T. pratense (variety "Sabtoron") were obtained from the Welsh Plant Breeding Station (Aberystwyth), while Medicago lupulina and Lotus corniculatus seeds were obtained from B and S Weed Seed Suppliers (Whatton in the Vale, Nottingham). A total of 20 plants of T. pratense, M. lupulina and L. corniculatus were established in 15F Optipots (Congleton Plastic Company Ltd.) filled with John Innes No.2 compost at the beginning of September 1984. Similarly, 30 plants of T. repens were established at the same time, and their cyanogenic/acyanogenic phenotypes determined using the same procedure as for field plants (see Chapter 5). Further sodium picrate paper tests were carried out in May and September 1985 and May 1986 to evaluate the stability of each phenotype (Ellis, Keymer & Jones 1977b). No change in phenotype expression was found.

6.2.2. Choice Experiments The choice chamber consisted of a 9cm diameter plastic petri dish, in which a moistened Whatman No.l filter paper was placed. The moist filter paper was necessary to keep the leaves turgid and prevent the desiccation of the test insects. The leaves for each feeding trial were obtained from the stock plants and only healthy, undamaged, young leaves were used. In the first feeding trial, all four species of legume ( T. repens, T. 167 pratense, M. lupulina and L. corniculatus) were compared but in this trial only homozygous acyanogenic leaves from T. repens were used. In the second feeding trial the preference between the cyanogenic/acyanogenic phenotypes of T. repens was evaluated. The experimental design of both trials was the same, only two species/phenotypes were compared in each choice chamber. As a result a total of six experiments had to be carried out in each trial so that each species/phenotype could be compared. Each experiment consisted of four leaves, two from each species/phenotype, being placed in opposite pairs around the edge of the petri dish. The number of times each experiment was replicated was dependent on the number of Curculionoidea available, since five individuals were needed for each replicate (see Tables 6.1 & 6.2. for the exact number of replicates in each experiment ). The following species of Curculionoidea were obtained by sweeping T. repens and T. pratense plants in areas close to the succession plots: Apion apricans Herbst, A. assimile Kirby, A. dichroum Bedel, A. trifolii Bach, Sitona hispidulus (Fabricius), S. lineatus (L.) and S. sulcifrons (Thunberg). Two further species, Hyper a postica (Gyllenhal) and S. humeralis Stephens were obtained by Martyn Jepsen by sweeping M. lupulina in pasture at Wytham Wood, Oxfordshire. All insects were starved of food (but not water) for 24 hours prior to the initiation of each experiment. The duration of each experiment in the Hypera and Sitona species was 24 hours, but in the Apion species this was extended to 48 hours on account of their small size and low food intake. All experiments were carried out under diffuse lighting conditions at room temperature between 17th June and 24* August 1985. Individual insects were used in no more than two experiments in each trial. In addition to these experiments, three choice experiments comparing young, intact leaves from three phenotypes of T. repens plants naturally established in the field were also undertaken for S. lineatus and H. postica. The leaf area of each leaf was estimated by measuring the length (1) and breadth (b) of the middle leaflet of every leaf exposed to the herbivores. A note was made of any leaf that was not damaged. The areas of leaves for each species of legume were measured using an electronic planimeter. This consisted of a Hitachii C.C.T.V. camera which recorded an image and this image was translated (after calibration), by a Digithurst Microeye image analysis system linked to an Amstrad PCI512 computer, into cm2 units. The necessary software to operate this system was obtained from its writer Dr.P.W. Mueller. Additionally, leaf areas of 120 intact leaves of T. pratense and over 200 intact leaves of T. repens, M. lupulina and L. corniculatus were also measured together with the 1 and b of the middle leaflets, since leaf area of an intact leaf varies linearly with the 168

Interspecific choice between leaves of different species of legume Herbivore Tr.Tp Tr.ML Tr.Lc Tp.Ml Tp.Lc Ml.Lc Apion apricans 10 10 9 10 10 10 Apion assimile n.d. 10 10 10 10 10 Apion dichroum 10 7 10 10 10 10 Apion trifolii 6 6 4 4 2 5 Hypera postica 14 10 15 11 16 22 Sitona hispidulus n.d. 3 4 n.d. 3 4 Sitona humeralis 14 9 10 14 10 10 Sitona lineatus 5 10 10 10 10 10 Sitona sulcifrons 4 4 5 4 7 4 Table 6.1. Number of replicates in each pair-wise preference test comparing four species of legume. Symbols are: Lc = Lotus corniculatus; Ml = Medicago lupuliita; Tp = Trifolium pratense; Tr = Trifolium repens; n.d. = no data available. Intraspecific choice between leaves of four phenotypes of cyanogenic/acyanogenic Trifolium repens Leaves obtained from: greenhouse plants wild plants Herbivore Cy.Ac Cy.Gl Cy.En Gl.En Gl.En Ac.En Cy.Gl Cy.En Gl.En Apion apricans n.d. 10 10 10 10 10 n.d. n.d. n.d. Apion assimile 10 10 10 9 10 10 n.d. n.d. n.d. Apion dichroum 10 10 10 10 10 5 n.d. n.d. n.d. Apion trifolii 2 3 2 6 8 7 n.d. n.d. n.d. Hypera postica 20 10 10 10 17 12 10 10 10 Sitona hispidulus 6 5 3 n.d. 3 3 n.d. n.d. n.d. Sitona humeralis 10 10 10 10 10 10 n.d. n.d. n.d. Sitona lineatus 14 10 10 10 10 10 10 10 10 Sitona sulcifrons 5 4 6 n.d. 4 4 n.d. n.d. n.d. Table 6.2. Number of replicates for each pair-wise preference test comparing four phenotypes of cyanogenic/acyanogenic Trifolium repens. Symbols are: Cy = cyanogenic; G1 = cyanoglucoside only; En = enzyme only; Ac = homozygous acyanogenic; n.d. = no data available. 170

product of 1 x b. Linear regressions for each species gave correlation coefficients of between 0.95 and 1.00 and, as a result of the good fit, the slope for each species could be used to estimate the whole leaf area of each damaged leaf from: Whole leaf area = slope x 1 x b The leaf area removed was then obtained by subtracting the measured damaged leaf area from the estimated whole leaf area. The estimated area removed for each species/phenotype in each experiment was pooled and the means compared using a two sample Student’s t Test, the results of which are tabulated in Tables 6.3. and 6.4.

62.3. A Test of Frequency-Dependent Grazing by S. lineatus on Cyanogenic and Acyanogenic Phenotypes ofT. repens Each experiment was run in a 12cm diameter petri dish in which a moistened disc of Whatman No.l filter paper (9cm diameter) was placed in the centre. Ten similar-sized leaves of T. repens were placed in the dish, eight around the edge of the filter paper and two in the centre. All leaves were obtained from predetermined phenotypes of the stock plants. Detached leaves from only two phenotypes were compared in each trial in the following combinations: Experiment no. 123456789 Phenotype 1 :987654321 number of leaves Phenotype 2 : 123456789 In order to compare all four phenotypes with each other, a total of six trials were carried out, each consisting of nine experiments. Each experiment was replicated ten times. The leaves in each trial were arranged so that the phenotype with the least number of leaves always had a neighbour that was a different phenotype. Identifying phenotypes at the end of each experiment was facilitated by the petioles of one phenotype being marked with a small drop of "Tipp-Ex". S. lineatus used in these feeding trials were collected in June and July 1986 by pooting the young foliage of which was grown in a number of large plots at Silwood Park. All individuals were starved for 24 hours prior to the start of each experiment and ten insects were used in each replicate. Individuals were used in no more than three experiments, one replicate in each. The duration of each experiment was 24 hours and all trials were carried out in diffuse lighting at room temperature between June and August 1986. 171

The leaf area consumed for every leaf was estimated in the same manner as for the choice experiments. Estimates of the area removed for each leaf were pooled for each phenotype in each replicate so an estimate of the total amount eaten per phenotype per replicate was obtained. In a very few cases slight error resulted in the estimated whole leaf area being less than the measured damaged leaf area. Such negative values of leaf area removed were given arbitrary values of zero. A simple descriptive model of apostatic selection developed by Elton & Greenwood (1970) and Greenwood & Elton (1979) was used to test for frequency-dependent selection in S. lineatus. The procedure uses the following parameters: e,: amount of food type 1 eaten e2: amount of food type 2 eaten A,: number of individuals of food type 1 available A2: number of individuals of food type 2 available The values e, and e2 represent the total leaf area consumed for each pair of phenotypes in each replicate, while At and \ correspond to the number of leaves of the two phenotypes in each replicate. Any replicate with zero leaf area removed was assigned a value of 0.01 cm to avoid zero values when using logarithms. Values of b^A/Aj) and ln(e,/e2) were calculated for each replicate, since logarithmic transformation linearizes the model to a simple regression equation: ln(e,/e2) = a + 6 In (A/Aj) Fitting the data of each trial to the equation gave estimates of constants, a and B (Table 6.5.) which are measures of frequency-independent and frequency-dependent selection, respectively. When B >1 the commoner food type is eaten at a greater frequency than expected and selection is termed pro-apostatic, but when the rarer food type is eaten at a greater frequency than expected, then B <1, and selection is anti-apostatic. Frequency-independent preference is inferred for food type 1 when a >0 and for food type 2 when a <0. There is no selection for either food type when B = 1 and a = 0. In this case each food type is eaten in the frequency it is presented, hence, frequency dependent selection (Cottam 1985). Graphical representation of the model for each trial is given in Figures 6.1. & 6.2. The diagonal represents the line of no selection i.e. e,/e2 = A,/A2. If the plotted line does not cross the diagonal then selection is frequency-independent, which side of the diagonal the plotted line lies depends on which food type is preferred. In each of these cases B=l, and only the a value varies. When the line crosses the diagonal, then selection is frequency-dependent and both pro-apostatic and anti-apostatic components can be 172 recognized. This was termed "switching" by Murdock (1969) since the attention of the predator shifts from one food type to another as their relative frequencies change. It was noted by Greenwood (1985) that the direction of selection does not depend entirely upon abundance since there may be a frequency-independent component present. Thus when selection switches from one food type to another, it does not necessarily occur when the frequencies of both morphs or phenotypes are equal. One important point that should be made about this model is that it treats frequency-independent and frequency-dependent selection as mathematically distinct, a fact which Greenwood and Elton (1979) admit might not be quite so clear cut when viewed biologically.

62.4. A Study of Fecundity and Longevity in Acyrthosiphon pisum Reared on Cyanogenic and Acyanogenic Phenotypes ofT. repens Cuttings from the stock plants of T. repens were taken at the beginning of May 1986, ten from each of the four phenotypes. Each cutting was prepared by removing a 4 cm section of stolon with two leaves and a terminal bud. To reduce within-phenotype variation, all ten cuttings were taken from the same plant. The cut end of each stolon was dipped in "Strike" rooting compound (May & Baker) and planted in a 7.5cm diameter peat pot filled with John Innes No.2 potting compost. The cuttings were then propagated in an unheated greenhouse on plastic trays filled with water to ensure that the compost was kept moist to encourage rooting. After 4 weeks the cuttings had become established and they were then transplanted to 15F Optipots. The plants were left a further week to recover. Each plant was carefully labelled, as there was no visible distinctions between phenotypes. Sodium picrate tests (see Chapter 5) were carried out at the beginning and end of the experiment to confirm that no change or mistaken identification of phenotype had taken place. The aphids were retained in clip cages, the design of which was slightly modified from that of Van Emden (1972) and Noble (1958). Each clip cage was made from a stainless steel hair curl clip and two transparent perspex rings (each 19mm diameter and 5mm in height). A layer of muslin was cut to size and stuck onto the outside rim and a ring of plastic foam stuck on the inside rim with "Bostik". The muslin was necessary for free air flow and access of light and the foam provided a seal to prevent the captive aphid escaping and to cushion the petiole. Since each clip cage was too heavy to be supported by the leaf petiole, an external support was provided in the form of a 16cm long, 0.1cm diameter piece of steel wire. 173

The aphid A. pisum was obtained as a virus-free culture from Rothamsted Experimental Station. The culture had been reared continually on V. faba and represented the bean biotype of this species. Establishment of the aphid on T. repens was carried out by caging one apterous adult or fifth instar on a leaf petiole just behind the leaflets. This was done on newly-expanded leaves; mature and senescent leaves were ignored. The three first newly deposited offspring were removed while the fourth nymph was left to grow to maturity. However, it was necessary to move the later instars of this nymph to younger foliage using a soft camel-haired brush, although the early instars were left undisturbed as these stages are very sensitive. This procedure was continued until at least the third generation so that acclimatization to the plant could occur and the influence of the original host plant could be minimized. The aphids were reared in constant conditions at 20±0.5°C and 73% relative humidity. Sixteen hours of light was provided each day by a bank of five fluorescent tubes with an intensity of 2800-4200 lux. Three aphids were caged (one per cage) on each plant giving initially 30 individuals per phenotype, although this was reduced to 28, 29 and 27 on the Acli, acLi and acli phenotypes respectively. This reduction in numbers was due to factors such as escape and accidental death and was no way connected with the effects of the plant. In some cases, where death was inexplicable as on two of the cyanogenic (AcLi) plants, the replicate was repeated at least six times, to ascertain the causes of death. When third generation aphids had been reared, the life history of individual aphids were observed. Observations took place daily and the time of offspring and death of adult was recorded. The number of offspring was recorded every two days. All offspring produced were removed and adult aphids were frequently moved to young petioles, because it was observed that uncaged individuals tended to congregate on these parts of the plant. From the data, the average length of life, mean pre-reproductive, reproductive and post-reproductive periods as well as the mean fecundity for each phenotype was calculated. The age-specific survival (lx) and age-specific fecundity (mx) for each age interval (x) of two days was calculated by the method described in Southwood (1978). Using the lx and mx data, the rate of increase (r) was estimated by iterative substitution of values of r into equation: Z eTlx.mx = 1 Once r was known for each phenotype, the following statistics were calculated: finite rate of increase (X), doubling time, generation time (T) and net reproductive rate (Ro) (for details, see Southwood 1978). All statistics have been tabulated in Table 6.6. 174

6.3. Results 6.3.1. Choice Experiments There was a very varied response by the Curculionoidea to the different species of legume and to the four phenotypes of T. repens (Tables 6.3. & 6.4.). The Apion species showed no clear preference in the laboratory. Apion apricans only ate appreciable quantities of T. repens when paired with M. lupulina, while A. assimile ate T. repens in preference to M. lupulina and L. corniculatus although the differences were not significant. No test between T. repens and T. pratense was carried out. In all other choice tests for these two species the amount eaten was not significantly different from zero. Both A. dichroum and A. trifolii showed no preference whatsoever for any species and the amount eaten did not vary significantly from zero. The low number of replicates for A. trifolii may be the causal factor for the high variance, although the same cannot be said for A. dichroum. Preference between the four phenotypes of T. repens was demonstrated in a number of cases. Three Apion species, ate significantly more of the glucoside only phenotype when paired with the cyanogenic phenotype, while A. trifolii significantly ate more of the enzyme only phenotype when paired with the cyanogenic phenotype. A. apricans also preferred the enzyme only phenotype when paired with the homozygous acyanogenic phenotype. Preference between all other phenotype pairs were not significantly different from each other. The remaining species of Curculionoidea were much larger in size and capable of removing a greater leaf area. However, only a few species showed conclusive preference for a particular species of legume, and the picture of preference between the phenotypes of T. repens was also unclear. H. postica tended to select T. repens and M. lupulina when paired with L. corniculatus and T. pratense. However, selection in T. pratense was not significant. There was virtually no intraspecific selection between phenotypes of T. repens, the exception being the selection for the glucoside only phenotype when paired with the cyanogenic phenotype for leaves taken from plants established naturally in the field. S. hispidulus showed a similar pattern of preference to H. postica, but in this species there was also an indication of a selection for T. repens when paired with M. lupulina, however, this was not significant. As with H. postica there was no preference for any of the phenotypes of T. repens, although only leaves from greenhouse plants were tested. Both S. humeralis and S. lineatus preferred M. lupulina and T. repens when paired with L. corniculatus and T. pratense but S. humeralis showed no preference between M. lupulina and T. repens, whereas S. lineatus ate significantly more M. lupulina. Selection against Apion apricans Apion assimile Apion dichroumApion trifolii Sitona hispidulusSitona humeralis Sitona sulcifrons Sitona lineatus Hypera postica Tr -0.0110.05 N.S. - 0.02±0.05 N.S. 0.05±0.04 N.S. - 0.2110.05 ** 0.0810.09 N.S. 0.1810.09 *** 0.0810.03 N.S. Tp -0.02±0.06 - 0.04±0.03 -0.03±0.07 - 0.0110.03 -0.0310.06 -0.4110.11 0.0210.02 Tr 0.12±0.05 0.8210.05 0.28±0.08 0.08±0.04 0.1310.10 N.S. 0.1710.04 N.S. 0.1610.06 N.S. -0.0110.03 ** 0.0810.04 N.S. Ml 0 0 0 0 -0.0310.07 0.1510.04 0.0210.06 0.2310.06 0.1710.06 Tr 0.0610.04 0.1410.04 0.04±0.05 0.0810.06 0.0910.05 * 0.2910.04 *** 0.1110.05 N.S. 0.2010.04 *** 0.1910.06 ** Lc 0 0 0 0 -0.0410.02 -0.0110.02 0.0110.03 -0.3810.02 0.0110.01 Tp 0 0 -0.05±0.03 0 - -0.0110.03 *** -0.1410.08 N.S. -0.1510.11 * 0.1110.11 N.S. Ml 0 0.11 ±0.07 0 0 - 0.1910.03 -0.0410.03 0.1110.03 0.1510.09 Tp -0.06±0.05 0 -0.03±0.06 0.0510.17 0.1410.08 N.S. 0.0210.03 N.S. 0.0510.05 N.S. -0.0910.09 N.S. 0 Lc 0 0 0 0 0.0810.07 0.0310.02 -0.0210.03 0.0710.02 0.0110.01 Ml 0 0 0.04±Q.02 0 0.1010.04 ** 0.2410.04 *** -0.0510.03 *** 0.1410.04 N.S. 0.2410.06 *** Lc -0.01±0.01 0 0 0 -0.1010.05 -0.0510.02 0.1110.03 0.0710.03 0.0310.01

Table 6.3. Mean leaf area removed (cm1) ±1 S.E. from each leaf in pair-wise preference tests comparing four species of legume and nine species of phytophagous Curculionoidea. The data were analysed by two sample t test. Symbols : Lc - Lotus corniculatus; Ml = Medicago lupulina; Tp = Trifolium pratense; Tr = Trifolium repens; *** = <0.001;p ** = p <0.01; * = p <0.05; N.S. = not significant. Apion apricans Apion assimile Apion dichroum Apion trifoliiSitona hispidulus Sitona humeralis Sitona sulcifrons Sitona lineatus Hypera postica

cy • 0.2410.05 N.S. 0.1110.04 N.S. 0.0810.08 N.S. 0.1910.06 N.S. 0.0710.06 ♦ 0.0710.06 N.S. 0.2310.04 N.S. 0.1810.03 N.S. Ac - 0.1110.04 0.1910.06 0.0310.12 0.1410.04 0.20i0.03 0.0210.07 0.2810.05 0.2210.04 Cy 0.1210.05 ** 0.0310.03 * 0.1010.06 * 0.0410.04 N.S. 0.1710.06 N.S. 0.2110.04 N.S. 0.1410.06 N.S. 0.3110.04 * 0.3110.07 N.S. G1 0.3410.05 0.1910.06 0.3210.06 -0.0410.10 0.1610.05 0.2510.04 0.2710.09 0.1810.04 0.2410.05 Cy 0.1310.05 N.S. 0.1710.07 N.S. 0.2310.04 N.S. 0.2310.04 *♦ 0.0710.07 N.S. 0.1110.03 * 0.14i0.03 N.S. 0.3010.04 N.S. 0.0110.01 N.S. En 0.1210.06 0.1710.05 0.1010.07 0.1010.08 -0.0110.07 0.2610.05 0.2310.04 0.2010.04 0.0310.03 G1 0.22±0.06 N.S. 0.1410.05 N.S. 0.0610.07 N.S. 0.1610.10 N.S. - 0.1510.04 N.S. - 0.2410.04 N.S. 0.2110.07 N.S. En 0.2010.05 0.0310.04 0.1010.05 0.2110.07 - 0.1610.04 - 0.2610.04 0.1010.04 G1 0.1210.04 N.S. 0.1610.04 N.S. 0.1610.04 N.S. 0.1010.05 N.S. 0.1410.08 N.S. 0.1810.06 N.S. 0.2710.07 ** - 0.0810.02 N.S. Ac 0.2310.04 0.1310.04 0.1010.04 0.0610.05 0.1110.07 0.1510.05 -0.0110.01 - 0.0810.04 Ac 0.0410.03 * 0.1410.05 N.S. 0.3510.07 N.S. -0.0210.05 N.S. 0.3210.05 N.S. 0.2010.04 N.S. 0.0810.08 N.S. 0.3010.02 N.S. 0.0910.02 N.S. En 0.2110.07 0.1510.04 0.3610.07 0.0 110.05 0.1510.07 0.1310.05 0.0810.05 0.2110.03 0.0910.04 Leaves Cy 0.0510.07 *** 0.0410.02 *** obtained G1 0.4710.05 0.3810.06 from wild Cy 0.1910.03 N.S. 0.2310.04 N.S. plants En 0.1710.03 0.2510.05 G1 0.2510.04 N.S. 0.0910.04 N.S. En 0.3210.04 0.02db0.03

Table 6.4. Mean leaf area removed (cm1) ±1 S.E. from each leaf in pair-wise preference tests comparing four phenotypes of cyanogenic/acyanogenic Trifolium repens and nine species of phytophagous Curculionoidca. The data were analysed by two sample t test Symbols: Cy = cyanogenic; G1 = cyanoglucoside only; En = enzyme only; Ac = homozygous acyanogenic; *** = p <0.001; ** = p <0.01; * = p <0.05; N.S. — not significant. 177 the cyanogenic phenotype of T. repens by S. humeralis was clearly demonstrated by a significant preference for the enzyme only and homozygous acyanogenic phenotypes, however, no preference was shown between cyanogenic and glucoside only phenotypes. By contrast, a larger leaf area was removed from the cyanogenic phenotype by S. lineatus when paired with the glucoside only phenotype, but this preference was reversed (and at a significantly higher level) when the same phenotypes of "wild” leaves were compared. All other choices between phenotypes for S. humeralis and S. lineatus were not significantly different. Finally, S. sulcifrons chose L. corniculatus in preference to M. lupulina, although the larger leaf area removed from T. repens indicated some selection for this species. Selection between phenotypes was shown only between the glucoside only and homozygous acyanogenic phenotypes with a significant preference for the former. There was an indication of the cyanogenic phenotype being avoided in comparison with the glucoside only and enzyme only phenotypes, but the results were not significant.

6.3.2. Frequency-Dependent Grazing Figures 6.1. and 6.2. shows that selection varies not only within each frequency of presented leaves but also between frequencies. This may be attributed to the multiple use of some of the herbivores, since their behaviour in later trials would be influenced by their experience gained in previous trials. This was taken into account when carrying out these experiments and as a result the same animal was only used in one replicate in one experiment, although some animals had unavoidably to be used in up to three experiments. Thus, experience may be a contributory factor to the variance within one frequency. However, Greenwood (1985) pointed out that using different animals in each experiment will not necessarily reduce the variation and analogies may result simply because of the natural variance between individual animals. As a result of this variation, only general trends will be considered here. In Table 6.5. a and B are given for each of the trials. Figure 6.1.a shows an example of frequency-dependence (in cyanogenic and homozygous recessive phenotypes) whereby each phenotype is eaten significantly more when common and switching from one phenotype to another occurs when the frequency of each food type is equal. This selection is termed pro-apostatic. In the cyanogenic/ glucoside only trial (Figure 6.1.b) there was some indication of pro-apostatic selection, but the slope of the regression was not significandy different from the line of no preference (Table 6.5.). Hence, each phenotype in this trial was eaten at the frequency presented. The cyanogenic/enzyme only trial (Figure 6.1.c) showed a highly significant frequency-independent preference for the 178

Proportion of each food type presented [ln(A,/A2)] Figure 6.1. A test of frequency dependent selection between cyanogenic and acyanogenic morphs of Trifolium repens by Sitona lineatus. The morphs compared are (a) cyanogenic (cy)/ homozygous acyanogenic (ac); (b) cyanogenic (cy)/glucoside only (gl); (c) cyanogenic (cy)/enzyme only (en). The broken line represents the line of no selection (i.e. B = 1 and a = 0) and the diagonal solid line represents the fitted regression line calculated using the method of Greenwood and Elton (1979). 179

Figure 6.2. A test of frequency dependent selection between acyanogenic morphs of T. repens by S. lineatus. The morphs compared are (a) glucoside only (gl)/enzyme only (en); glucoside only (gl)/homozygous acyanogenic (ac); (c) enzyme only (en)/homo -zygous acyanogenic (ac). The broken line represents the line of no selection (i.e. 6=1 and a = 0) and the diagonal solid line represents the fitted regression line calculated using the method of Greenwood and Elton (1979). 180

Significance Significance Trial phenotypes a S.E. from zero B S.E. from unity Cy/Ac -0.04 ±0.13 N.S. 1.40 ±0.10 *** Cy/Gl 0.05 ±0.12 N.S. 1.16 ±0.10 N.S. Cy/En -0.83 ±0.14 *** 1.38 ±0.11 *** Gl/En -0.34 ±0.14 * 1.19 ±0.10 N.S. Gl/Ac 0.34 ±0.13 ** 1.27 ±0.10 ** En/Ac -0.66 ±0.16 *** 1.26 ±0.13 * Table 6.5. Regression parameters a and B (±1 S.E.) for Sitona lineatus feeding on six pair-wise combinations of four phenotypes of cyanogenic/acyanogenic T. repens. Tests for significance from zero for a and unity for B were calculated by Student’s t test. Symbols: Cy = cyanogenic; G1 = cyanoglucoside only; En = enzyme only; Ac = homozygenous acyanogenic; *** = p <0.001; ** = p <0.01; * = p <0.05; N.S. = not significant. 181

enzyme only phenotype, although selection was pro-apostatic. This meant that each phenotype was eaten at a higher frequency than expected when common but shifts in the frequency-independent component caused the enzyme only phenotype to be preferred at much lower relative frequencies than the cyanogenic phenotype. In the glucoside only/enzyme only trial there was evidence for frequency-independent preference for the enzyme only phenotype, although this was only marginally significant. However, there was no evidence for frequency-dependent selection (Table 6.5.). Figure 6.2.a shows that the enzyme only phenotype is preferred at high relative frequencies, but at the lowest frequency there is no preference and both phenotypes are eaten at the frequency presented. Individual cases do suggest that pro-apostatic grazing does occur but this was not significant statistically. Finally, in the last two trials (glucoside only/homozygous acyanogenic and enzyme only/homozygous acyanogenic; Figures 6.2.b and 6.2.C respectively) both frequency-independent and frequency-dependent pro-apostatic grazing occurred at a significant level. In both trials the homozygous acyanogenic phenotype was the least preferred, however, it is eaten when presented at high relative frequencies.

6.3.3. Fecundity and Longevity in A. pisum Reared on Cyanogenic and Acyanogenic Phenotypes ofT. repens There was a clear difference in age specific survival and number of births per original female surviving per 2 day interval between aphids feeding on cyanogenic and acyanogenic phenotypes (Figures 6.3. & 6.4.). However, this distinction was not so clear between the three acyanogenic phenotypes. The reason for the sharp drop in the survival of aphids on the cyanogenic plants (Figure 6.3.a) in the first two days of life was because there were two plants on which no aphids would survive. Survival on the other cyanogenic plants was better and very little mortality occurred until the 18* day, after which survival dramatically decreased up to the 45* day, when the last aphid died. The pattern of survival on the enzyme only and glucoside only plants was superficially similar (Figure 6.3.b & c) but with only a slight decrease in survival in the first 24 days of life and no sudden decrease in the first two days. Survival sharply declined after 20-24 days and the last aphid on both phenotypes died on the 47* day. The aphids feeding on the homozygous acyanogenic phenotype had the highest survival rates with no mortalities until the 27* day, while the last aphid died on the 50* day. reared on each cyanogenic/acyanogenic morph of of morph cyanogenic/acyanogenic each of on survival reared specific age the in Variation 6.3. Figure cyanogenic = Acli; acyanogenic = Acli (cyanoglucosides only), acLi acLi only), (cyanoglucosides Acli = acyanogenic Acli; = cyanogenic Conventions for each morph used in the following figures are: figures following the in used morph each for Conventions (glucosidase only), acli (neither glucoside or enzyme present), present), enzyme or glucoside (neither acli only), (glucosidase n = number of replicates (aphids). replicates of number = n Age specific survival (lx) Trifolium repens. Trifolium Acyrthosiphon pisum Acyrthosiphon

182 Number of progeny bom per aphid in each 2 day age interval (lx.mx) Figure 6.4. Changes in the number of progeny/aphid/2 day age age day progeny/aphid/2 of number the in Changes 6.4. Figure interval for for interval T. repens. T. A. pisum A. on each cyanogenic/acyanogenic morph of of morph cyanogenic/acyanogenic each on 183 184

The effects of the phenotypes on the reproductive performance of the aphids showed a similar pattern to their survival, with the poorest performance on the cyanogenic phenotype and the best on the homozygous acyanogenic phenotype (Figure 6.4.). The time of first progeny production was similar in all phenotypes, although the time of peak production does vary. On the cyanogenic phenotype, peak progeny production of 1.366 births per original female surviving per 2 day interval was reached after only 19 days from birth and this declines to a very low level after a further 10 days. However, a few individuals continued to reproduce until the 42nd day from birth. The peak progeny production on the acyanogenic phenotypes was much higher but was attained later than in the cyanogenic phenotype i.e. 26 days after birth. Aphids on the glucoside only and enzyme only phenotypes showed similar peak values of around 2 births per original female surviving per 2 day interval, while on the homozygous acyanogenic phenotype, a peak progeny production of 2.48 births per original female surviving per 2 day interval was reached. There was little variation in the time of last offspring on the acyanogenic phenotypes, with reproduction terminating on the 44th day on the glucoside only and homozygous acyanogenic phenotypes and on the 45th day on the enzyme only phenotype. It is interesting that on all phenotypes peak progeny production was not only followed by a reduction in reproductive rate but also survival. The cumulative lx.mx curves are shown in Figure 6.5. The shape of each curve was similar for each phenotype and after an initial period of no reproduction (development period), the curve rises sharply to a plateau where reproduction terminates. It is the upper value which represents the maximum number of females produced by a female in that generation i.e. the net reproductive rate. The Zlx.mx curve for the cyanogenic phenotype reached a plateau comparatively quickly and most reproduction had taken place after 27 days. However, the plateau was not reached until much later on the acyanogenic phenotypes. As expected, the mean reproductive period on each acyanogenic phenotype (Table 6.6.) was significantly longer than on the cyanogenic phenotype. Since the mean pre-reproductive and post-reproductive periods do not vary significantly between any of the phenotypes, the significant difference in mean length of life between aphids feeding on cyanogenic and acyanogenic phenotypes was solely due to the difference in the length of the reproductive period. Aphids on the cyanogenic phenotype had the lowest rate of increase and the highest percentage of individuals not reproducing, consequently, it is hardly surprising that aphids on this phenotype have a significantly lower mean fecundity. The highest rate of increase was recorded on the homozygous acyanogenic phenotype, but in this case both the mean reproductive period and mean fecundity was not Cumulative progeny production (S(lx.mx)) A. A. pisum Figure 6.5. The cumulative progeny production/female/day for for production/female/day progeny cumulative The 6.5. Figure on each cyanogenic/acyanogenic morph of of morph cyanogenic/acyanogenic each on T. repens. T. 185 186

Statistic AcLi Acli acLi acli Pre-reproductive period (days)1 15.30+0.78 16.30±0.73 16.93+0.48 16.04+0.55 Reproductive period (days)1 9.45+1.46 14.52±1.39 15.76±1.15 18.11±1.21 Post-reproductive period (days)1 3.05+0.34 4.22±0.46 3.52±0.27 3.37±0.41 Life expectancy (days)1 22.10±2.21 34.17+1.36 36.17±1.36 37.59+1.09 Fecundity 13 5.60±1.25 16.36+2.41 15.97+1.53 20.33+1.87 Net reproductive rate (Ro)2 5.48 16.11 15.95 21.18 Rate of increase (rm)3 0.09 0.12 0.11 0.13 Finite rate of increase (X)3 1.10 1.12 1.12 1.13 Doubling time (days) 7.62 5.92 6.25 5.55 Generation time (T) (days) 18.70 23.76 24.95 24.42 Percentage producing no progeny 33.33 3.57 0.00 0.00 Table 6.6. Demographic and biological statistics of A. pisum reared on four phenotypes of cyanogenic/acyanogenic T. repens. Symbols: AcLi = cyanogenic; Acli = cyanoglucoside only; acLi = enzyme only; acli = homozygous acyanogenic; 1 = mean ±1 S.E.; 2 = female/ female/ generation; 3 = female/ female/ day. 187 significantly greater than on the other two acyanogenic phenotypes. Aphids reared on glucoside only and enzyme only phenotypes were barely distinguishable by their demographic and biological statistics.

6.4. Discussion The role of herbivorous insects upon plants can be identified as one of the biotic forces of the environment that shape the ecology and evolution of plants in a population (Dirzo 1985). Harper (1977) stated that if an animal plays a critical role in determining the distribution and abundance of plants, it is not likely to be obvious and suggested the use of experiments as the only way to understand fully the role that the herbivores may play in the population dynamics of plants. Seven species of Curculionoidea tested in the choice experiments were common in the successional sites and were taken by Univac suction sampling (see Chapter 3) from all four legume species. H. postica and S. humeralis were comparatively rare and suggests that other factors other than availability of suitable hosts restricted their numbers, since these two species were extremely abundant at Wytham Wood. In the laboratory, four Apion species showed little preference for any legume species although there was an indication that T. repens was favoured. The Apion genus is predominantly seed-feeding and the species used in this study feed, at least in part, on the flowers and fruits of legumes (P.C.Hyman pers. comm.). However, leaf feeding did occur, but may be related to hunger. Indeed, Bemays & Chapman (1970b) found that by extending the period of starvation, Chorthippus parallelus Zetterstedt would accept food plants that it would normally reject. It must also be pointed out that the leaves used in the choice experiments were chosen by the experimenter and may not necessarily represent the choice of the insect tested. Hence, although damaged leaves similar to those used in the experiment were observed in the field, they may have been attacked at a different stage to the ones presented in the choice chamber. Only one species, S. sulcifrons showed a preference for L. corniculatus, this species being avoided by the other weevil species. All Sitona species and H. postica tended to prefer T. repens and M. lupulina which is hardly surprising since some of them are recognized pests (particularly during the larval stage) of clovers and lucerne, a close relative to M. lupulina. Previous conditioning is also important in preference testing (Wiseman & McMillan 1980). Two of the species ( H. postica and S. humeralis) were collected from M. lupulina, while the rest were collected from T. pratense and T. repens although the results do not indicate that this is important. However, Johansson (1951) 188 found that fifth instar larvae of Pier is brassicae (L.) preferred plants from which they had been collected in the field. Such induced preferences have also been demonstrated in other species of Lepidoptera (see Hanson 1983). The effect of a neighbour may also influence choice since H. postica and S. hispidulus both significantly preferred M. lupulina and T. repens when paired with L. corniculatus, but this response was not equalled when paired with T. pratense or with each other. The defensive role of cyanogenesis was not conclusively demonstrated. It would be expected that cyanogenic phenotypes would be rejected by generalist herbivores, while no preference should be exhibited by specialists. The herbivores in these tests do show some degree of specialism for T. repens, but only two species (H. postica and S. hispidulus) show no preference for any phenotype of the stock plants. A close relative to one of the above herbivores, H. plantaginis is able to detoxify HCN (Parsons & Rothschild 1964) and a similar mechanism may exist in other related species. All the Apion species and S. humeraliSy on other hand, show some preference against the cyanogenic phenotype. This seems a little surprising, since Beesley, Compton and Jones (1985) demonstrated that Apion species have a higher than average level of rhodanase, an enzyme capable of detoxifying HCN. S. lineatus shows a preference for the cyanogenic leaves from the stock plants in one of the tests, but when the same test was carried out using wild leaves, the preference was reversed. A similar differential response was recorded for H. postica and clearly indicates that cyanogenic plants grown in the greenhouse, although giving a positive response to the sodium picrate paper test, sometimes differ from the plants established naturally in the field. All plants chosen for the tests had stable phenotype expression and only strongly cyanogenic plants were used. However, de Waal (1942) found that the cyanogenic response varied during the day and between days and it is possible that this variation could account for the differences in preference, although all possible precautions in the form of sequential testing were taken. L. corniculatus has been shown to contain other noxious defensive chemicals (Forde & Lautour 1978) which may also be polymorphic and inversely related to mean cyanide content (Ross & Jones 1983). Whether similar compounds exist in T. repens is not known but this could explain the differential selection against the homozygous acyanogenic phenotype by A. apricans and S. sulcifrons when compared with another acyanogenic phenotype. However, other changes in host plant quality (such as nitrogen levels) cannot be mled out. In the frequency-dependent grazing experiment, S. lineatus exhibited either a pro-apostatic, frequency-dependent response or no frequency-dependent selection. Thus it would appear that a phenotype was eaten at the frequency presented or eaten at at a higher frequency than expected when common regardless of whether it was cyanogenic or acyanogenic. Thus, an "adapted" herbivore may help promote polymorphism by feeding 189 pro-apostatically, thereby maintaining an otherwise rare phenotype in the population. In contrast, if feeding was anti-apostatic then feeding would select against the rare phenotype, eventually eliminating it from the population. However, the pro-apostatic selection by S. lineatus further substantiates the concept that frequency-dependent grazing is an optimal foraging strategy (Hubbard et al. 1982), since it allows the herbivore to maximize its intake of a preferred food when present in sufficient quantity. In addition to pro-apostatic frequency-dependent selection, there is also an underlying frequency-independent preference between some of the phenotypes. It is interesting that a preference for a particular phenotype was strongest when two acyanogenic phenotypes were compared. The lack of preference for the acyanogenic phenotypes (glucoside only and homozygous acyanogenic) when paired with the cyanogenic phenotype indicates that HCN may not inhibit feeding by this species. However, this was not conclusive since the enzyme only phenotype was preferred. The choice exhibited by S. lineatus in the species preference tests differs from the preference exhibited in the frequency-dependent grazing experiments and illustrates the problem that insects reared on different foods may have different responses to test stimuli (Stadler & Hanson 1978). Thus the role of conditioning to a particular food type and previous experience is important in determining the preference that each individual insect portrays. It should also be mentioned that sex and age may influence feeding preference, since this was found to be important in grasshopper feeding on different grass species (Bemays & Chapman 1970a). The time of year was shown to influence the selectivity in grazing sheep, switching from acyanogenic morphs in summer to cyanogenic morphs in late autumn (Dirzo & Harper 1982b). This shift in selection indicates that a complex interaction may exist. Even so, Jones (1966) demonstrated that Microtis agrestis preferred acyanogenic L. corniculatus, but was only selective when an adequate alternative food supply was available. Thus care must be exercised in drawing conclusions since factors affecting selection (whether frequency-dependent or not) in generalist and specialist herbivores can vary depending on conditions. The impact of pro-apostatic grazing by an "adapted" herbivore on T. repens in a natural sward is difficult to predict. One might expect that heavy grazing of the common phenotype would not lead to its depletion in the population, since the herbivore would switch to another phenotype when its frequency fell below some threshold level. Pro-apostatic grazing might even serve to counter balance the selective pressure exerted by generalist herbivores, thereby contributing to the maintenance of the polymorphism. 190

The effect of a polymorphic chemical defence of a plant on the reproductive rate and longevity of A. pisum is described for T. repens. Variation in plant susceptibility to aphids was described by Kennedy & Booth (1951), although in this work the variation was correlated with physiological changes such as leaf growth and senescence. It was later shown by van Emden and Bashford (1969, 1971) that the polyphagous aphid, Myzus persicae (Sulzer), showed a greater sensitivity to host plant secondary chemistry and leaf amino acid levels than the monophagous Brevicoryne brassicae (L.). Although plant nitrogen levels were not investigated in my experiments, it was assumed that by keeping A. pisum on petioles of expanding leaves (preferred feeding site of uncaged individuals), the soluble nitrogen levels would be similar in each treatment. However, A. pisum had the shortest longevity, smallest rate of increase and lowest mean fecundity on the cyanogenic phenotype. The performance on the acyanogenic morphs did not differ significantly, but there was a tendency for the performance to be slightly better on the homozygous acyanogenic phenotype. This result seems rather surprising when one considers A. pisum to be a monophagous aphid specializing on Leguminosae. However, comparative data with other aphid species are now required, together with the subspecies and biotypes of A. pisum known to vary in their performance on different host species and varieties as well as their behaviour and colour, see Cammell and Way (1983). It was demonstrated by Frazer (1972b) that different adaptive biotypes of A. pisum existed on four hosts { scoparius (L.) Link, M. sativa, T. repens and V. faba) and their adaptation to a particular host may limit the acceptability of other hosts. Biotypes develop in the field each spring when fundatrices and their alate offspring migrate from their primary legume hosts to secondary legume hosts and the succeeding generations of viviparous virginoparae become better adapted to their host plants by selection. However, different varieties of host plants (Tambs-Lyche & Kennedy 1958) and biotypes of aphid species (Cartier 1959) have been shown to produce different fecundity and survival data. The performance of the bean biotype in this experiment was much poorer on T. repens than on its host V.faba (see Frazer 1972a for data on V.faba). However, the variation in performance on each phenotype of T. repens would lead to the prediction that biotypes developing on cyanogenic plants would have different fecundity and survival rates to those developing on acyanogenic plants. Potential effects of the aphid on the plant were apparent four weeks after the last caged aphid had died. In the case of the homozygous acyanogenic phenotype, all plants were dead or dying while the three other phenotypes were still green and healthy. 191

It is not known for certain whether the cyanide liberated during feeding directly reduces fecundity and survival through toxicity. Dirzo and Harper (1982a) demonstrated that the slug, Agriolimax reticulatus (Muller) fed on monotonous diets of cyanogenic or acyanogenic T. repens leaves attained a slightly smaller live-weight or lost weight on the former morph compared with the latter and attributed this to a reduced food intake rather than the effects of poisoning. Acyrthosiphon spartii (Koch), a close relative of A. pisum, but feeding on C. scoparius, was observed to change its feeding sites and these changes were coincident with the distribution of the alkaloid sparteine (Smith 1966). High concentrations of sparteine are associated with the most nutritious tissues and this aphid appears to use the alkaloid as a feeding stimulant, performing better when its concentration is artificially raised. By contrast, A. pisum feeding on T. repens does not appear to use cyanoglucosides or 6-glucosidases as feeding cues, even though the most nutritious parts of the plant contain the highest concentrations (Hughes 1968). This was evident as individuals perform better on the homozygous acyanogenic plants where cyanoglucosides and 6-glucosidases are not detected. A. pisum may respond to other chemical stimuli, not measured in these experiments, and this may be an adaptation to accepting a wider range of legume hosts than A. spartii which only feeds on broom. In summary, insect herbivores that occur on a wide range of legume species should tend towards a narrow choice of preferred hosts which may be related to induced preferences gained by previous feeding experience. Even within a single species of legume, factors such as cyanogenesis may influence the choice and this may be further complicated by grazing patterns that are frequency-dependent. Finally, the presence of secondary chemicals may not directly deter feeding, but their presence might influence the population dynamics of the herbivore especially if the species is not monophagous. Chapter 7 General Discussion

This discussion is presented as a general synthesis and summary of the main conclusions reached in previous chapters. Although described in the context of a community study, only two trophic levels were studied in detail (higher plants (or angiosperms) and insect herbivores). However, other trophic levels, such as predators and pathogens were taken into consideration when making these conclusions. One part of the system that has largely been ignored is the symbiotic bacterial association. Tuikington (1985) noted, in his studies on T. repens, that interactions between plant and nitrogen-fixing Rhizobium are complex involving plant specific strains. If such relationships occur in other species of legume, this may have serious implications for plant-plant and insect-plant relationships. However, the plant -Rhizobium interaction was not measured in this study and as a result remains largely undetermined. In the legumes, the annual life-form persists much later in succession than is generally seen in other groups (Brown, Hendrix & Dingle 1987). This persistence is dependent on the regular creation of gaps in the canopy, caused either by disturbance, death of individual plants or by litter accumulation. The perennial legumes tend to colonize the site from the seedbank in the ruderal phase, although establishment later in succession does occur but is very rare. Seeds of each legume species accumulate in the seedbank, however, Chapman and Anderson (1987) found that in New Zealand hill pastures the half life of buried seeds of T. repens was about one year or less, although other workers have found viable T. repens seeds approximately 600 years old (Harrington 1972). Such variation in the longevity of seed stresses the importance of a knowledge of previous land-use history (see Stinson 1983). In this study, two sites (1981 and 1971) had different histories to the rest and as a result, anomalies in plant composition were evident. Once the plant community has established, the influence of the seedbank declines and the composition of neighbouring species become increasingly important, as the effects of competition begin to increase. Other factors such as herbivory, disturbance and site attributes mediate their effects on the legumes through changes in the structure and composition of the plant community. Of all the plant groups, perennial grasses appear to have the greatest effect in reducing the abundance of the legumes. 193

The density of insects on L. corniculatus, T. repens, V. hirsuta and V. sativa decreased with successional age, thereby supporting the prediction of Lawton and McNeill (1979) that early successional plants will have a greater absolute abundance of insect herbivores than late successional plants. Moreover, in this case, differences in insect density along the successional gradient occurred within single plant species. In addition to changes in insect density within a species, there was also between species variation with the lowest insect density recorded on the Vicia species and the highest on L. corniculatus and Trifolium species. These findings tend to lend support to the plant apparency theory (Feeny 1976), since the most apparent plants (in time) have the highest density of insect herbivores, while the less apparent annuals have the lowest insect density. However, within species changes in insect herbivore density tend to conflict with the assumptions of apparency theory, since it appears that L. corniculatus, T. repens, V. hirsuta and V. sativa were most apparent (in space) in early succession and least apparent later in succession. These differences were probably due to other factors, such as the abundance of natural enemies, since the ordination technique suggested that the composition of the insect community varied very little within species during succession (except for perhaps the raderal site). However, differences in the composition of the insect herbivore community were associated with differences in life-form and growth form of the legumes. The effects of herbivory on each species of legume tend to reflect changes in insect herbivore density and composition. Leaf turnover was highest in the annuals and lowest in the perennials, which infers that the former were the most palatable, least defended by secondary chemicals and least apparent in time (and space). Such differences were correlated with insect herbivore density and may further substantiate plant apparency theory. However, trends in leaf damage were not consistent, since leaves were most likely damaged in both Vicia species (in all sites) and Trifolium species (in the 1985 and 1983 sites) when they were least apparent, whereas in L. corniculatus, M. lupulina and T. repens (1984 site) insect damage increased with leaf apparency. The effect of herbivory on reproduction, whether direct (i.e. flower and seed predation) or indirect (i.e. altering the carbohydrate balance) varied between species, but in all cases seed production was increased when insect herbivory was reduced. Seed weight was increased following insecticide application in T. pratense, V. hirsuta and M. lupulina, remained unchanged in V. sativa, and was reduced in T. repens. The consequences of variation in seed size have been discussed in Chapter 4. In addition, it has been found in T. repens that in the early seedling stage, plants derived from light-weight seed have fewer, smaller nodules and take longer to nodulate than do plants derived from heavier seed (Mytton 1973). It is not known whether this effect occurs in other species but if it does, it will have important consequences on plant establishment in the legumes. 194

However, variations in seed size and seedling establishment probably have their greatest effect on the annual legumes, since these have to colonize the site each year. Hence, reductions in seed size as a result of insect herbivory might influence the size of gap needed for successful establishment. Thus, insect herbivory may cause plant populations to be seed limited (Louda 1982), even though large numbers of seeds are produced, because few will be able to germinate, since the optimum gap size (or microsite) may not be available. On the other hand, if microsites are heterogeneously distributed in space (and time), it may be beneficial to increase seed production as well as possess some sort of dormancy mechanism. However, there will be trade-offs between seed size and seed number but these will change during succession as the plants try to maximize their number of offspring. Even when seeds reach suitable microsites, insect herbivory (both above and below ground) may impair seedling recruitment and plant establishment, which may in turn alter the direction and rate of succession (Brown 1982a, Gibson, Brown & Jepsen 1987, Gibson etal. 1987). Changes in the palatability of different plants as a result of differences in the levels of secondary metabolites were investigated in the field within a single species ( T. repens) known to be polymorphic for cyanogenesis. A number of factors were found to decrease the frequency of cyanogenesis confirming previous studies. These were: the presence of L. corniculatus (Jones 1968); winter conditions (Daday 1954a, b, 1965) even though trends were not consistent in every site; and reducing insect herbivory (in the 1985 and 1983 sites). In the 1985 site, reduced insect herbivory was probably related to increased seedling survival of the acyanogenic morphs and it is probably at this stage that insect herbivores have their greatest impact. What is now needed are field experiments in which T. repens is sown in insecticide-treated and control plots and monitored over a number of years to see if the frequency of cyanogenesis is affected. The effects of insect herbivory on the vegetative characteristics of each phenotype polymorphic for cyanogenesis were difficult to detect in the field due to other compounding factors such as susceptibility to drought, pathogens and plant competition. What was established was that a reduction in insect herbivory does not necessarily increase plant size (in terms of height and leaf number). Such differences in height were initially attributed to changes in the balance between sexual reproduction and vegetative growth in Chapter 4. However, it seems probable that reduced plant size was the result of increased plant competition, since insecticide application was found to increase the height and structure of the vegetation overall (Brown, Jepsen & Gibson 1988). In the current study, there was no discernible difference in the level of damage experienced by all four phenotypes polymorphic for cyanogenesis. This result may at first seem unexpected but 195 until the balance between generalist and specialist herbivores is determined, further explanations can only be speculative. Generally, it seemed that cyanogenesis is most effective against herbivores that feed on T. repens occasionally. In the laboratory, stenophagous legume herbivores (Curculionoidea) preferred T. repens (and M. lupulina) while L. corniculatus was selected the least. Preference for acyanogenic phenotypes in T. repens was not conclusively determined in these herbivores, although variation in plants grown in different conditions and under different selection pressures was apparent. However, in an "adapted" insect herbivore ( S. lineatus) selection was found to be frequency-dependent. This behaviour is probably common in other stenophagous insects and even generalist herbivores (Burgess & Ennos 1987), but whether it is environmentally or genetically determined is uncertain. The effect of cyanogenesis on survival and fecundity was experimentally determined using the pea aphid, A. pisum. Both characteristics were reduced on the cyanogenic morph which is not surprising, since Compton & Jones (1985) suggest that in addition to acting as a feeding inhibitor, cyanide may also influence the population dynamics of the herbivore. In natural conditions, the effects of natural enemies may accentuate the effects of the cyanogenic morph (Starks et al. 1972) Finally, it was generally recognized that insect herbivory was most pronounced in early succession, whereas plant competition was most effective later in succession. The legumes in this study were characteristic of early to mid-successional habitats, and it is therefore not surprising that competition becomes more intense later in succession. The effect of insect herbivory probably accentuates the process of succession by decreasing the competitive ability of the plant. This may in turn affect the balance of polymorphism for cyanogenesis, since herbivory at the same time as intense interspecific competition may select for cyanogenesis. Whereas when interspecific competition is low, a higher level of herbivory might be tolerated and the acyanogenic morphs will be selected for because of their increased competitive ability. In efforts to understand the complex nature of the processes which govern the composition of insect and plant communities, it is therefore essential to undertake manipulative field and laboratory experiments to complement descriptive field studies. Such experiments are necessary so that theories which provide the general principles of plant-plant, plant-insect and insect-plant relationships can be tested. Whether these theories are accepted or refuted will depend on the accumulation of evidence which encompasses all aspects of community ecology. This thesis is presented in an effort to contribute to the body of knowledge which may lead to a greater understanding of community processes. However, this study does highlight that work at the species level in 196 community ecology may not be a fine enough scale. Thus, in order to understand fully the changes in the plant and insect communities during succession, future studies should incorporate phenotypic variation within plant species. 197 References

Aarssen, L.W., Turkington, R. & Caver, P.B. (1979). Neighbour relationships in grass-legume communities, n. Temporal stability and community evolution. Canadian Journal of Botany, 57,2695-2703. Abrahamson, W.G. (1975). Reproductive strategies in dewberries. Ecology, 56, 721-726. Allan, P.B.M. (1948). Larval food plants. A Vade-mecum for the Field Lepidopterist. Watkins & Doncaster, London. Allen, J.A. (1972). Evidence for stabilizing and apostatic selection by wild blackbirds. Nature, 237, 348-349. Allen, J.A. (1976). Further evidence for apostatic selection by wild passerine birds - 9:1 experiments. Heredityf 36,173-180. Allen, T.C. & Cassida, J.E. (1951). Criteria for evaluating insecticidal phytotoxicity - aerial growth. Journal of Economic Entomology, 44,737-740. Angseesing, J.P.A. (1974). Selective eating of the cyanogenic form of Trifolium repens. Heredity, 32, 73-83. Angseesing, J.P.A. & Angseesing, W.J. (1973). Field observations on the cyanogenesis polymoiphism in Trifolium repens. Heredity, 31, 276-282. Antonovics, J. (1976). The input from population genetics: "the new ecological genetics". Systematic Botany, 1,233-245. Auerbach, M.J. & Strong, D.R. (1981). Nutritional ecology of Heliconia herbivores: experiments with plant fertilization and alternative hosts. Ecological Monographs, 51,63-83. Bach, C.E. (1980). Effects of plant density and diversity on the population dynamics of a specialist herbivore, the striped cucumber beetle, Acalymma vittata (Fab.). Ecology, 61,1515-1530. Banks, C.J. & Macaulay, E.D.M. (1967). Effects of Aphis fabae and its attendant ants and insect predators on yields of field beans (Vicia faba L.). Annals of Applied Biology, 60,445-453. Barnes, H.F. (1946). Gall Midges of Economic Importance. Vol. II: Gall Midges of Fodder Crops. Crosby Lockwood, London. Bazzaz, F.A. (1975). Plant species diversity in old field successional ecosystems in southern Illinois. Ecology, 56,485-488. 198

Becker, P. (1983). Effects of insect herbivory and artificial defoliation on the survival of Shorea seedlings. In Tropical Rain Forest: Ecology and Management (Ed. by S.L. Sutton, T.C. Whitmore & AC Chadwick), pp.241-252. Blackwell Scientific Publications, Oxford. Beesley, S.G., Compton, S.G. & Jones, D.A. (1985). Rhodanese in insects. Journal of Chemical Ecology, 11,45-50. Beinhart, C. (1963). Effects of environment on meristematic development, leaf area and growth of white clover. Crop Science, 3, 209-213. Belsky, A.J. (1986). Does herbivory benefit plants? A review of the evidence. American Naturalist, 127, 871-892. Benson, R.B. (1952). : Symphyta. Handbooks for the identification of British insects, 6(2b), 1-88. Benson, R.B. (1958). Hymenoptera: Symphyta. Handbooks for the identification of British insects, 6(2c), 1-114. Bentley, B.L. (1977). Extrafloral nectaries and protection by pugnacious body guards. Annual Review of Ecology and Systematics, 8,407-427. Bentley, S., Whittaker, J.B. (1979). Effects of grazing by a chrysomelid beede Gastrophysa viridula, on competition between Rumex obtusifolius and Rumex crispus. Journal of Ecology, 67,79-90. Bentley, S., Whittaker, J.B. & Mai loch, A.J.C. (1980). Field experiments on the effects of grazing by a chrysomelid beetle ( Gastrophysa viridula) on seed production and quality in Rumex obtusifolius and Rumex crispus. Journal of Ecology, 68, 671-674. Bernays, E.A. (1983). Nitrogen in defence against insects. In Nitrogen as an Ecological Factor (Ed. by J.A. Lee, S. McNeill & I.H. Rorison), pp. 321-344. Symposium of the British Ecological Society, 22. Blackwell Scientific Publications, Oxford. Bernays, E.A. & Chapman, R.F. (1970a). Food selection by Chorthippus parallelus (Zetterstedt)(Orthoptera: Acrididae) in the field. Journal of Animal Ecology, 39, 383-394. Bernays, E.A. & Chapman, R.F. (1970b). Experiments to determine the basis of food selection by Chorthippus parallelus (Zetterstedt) (Orthoptera: Acrididae) in the field. Journal of Animal Ecology, 39,761-776. Bernays, E.A., Chapman, R.F., Leather, EM, McCafferv, A.R. & Modder, W.W.D. (1977). The relationship of Zonocerus variegatus (L.) (Acridoidea: Pyrogomorphidae) with cassava ( Manihot esculenta). Bulletin of Entomological Research, 67,391-404. Biswell, H.H. (1974). Effects of fire on chaparral. In Fire and Ecosystems (Ed. T.T. Kozlowski & C.E. Ahlgren), pp. 321-365. Academic Press, London. 199

Blackman, A. (1974). Aphids. Ginn, London. Bradley, J.D., Tremewan, W.G. & Smith, A. (1973). British Tortricoid Cochylidae and : Tortricinae. Ray Society, London. Bradley, J.D., Tremewan, W.G. & Smith, A. (1979). British Tortricoid Moths Cochylidae and Tortricidae: Olethreutinae. Ray Society, London. Brown, L.C., Cathey, G.W. & Lincoln, C. (1962). Growth and development of cotton as affected by toxaphene-DDT, methyl partition and calcium arsenate. Journal of Economic Entomology, 55, 398-401. Brown, V.K. (1982a). The phytophagous insect community and its impact on early successional habitats. Proceedings of the 5th International Symposium on Insect Plant Relationships, Wageningen (Ed. by J.H. Visser & A.K. Minks), pp. 205-213. Pudoc, Wageningen. Brown, V.K. (1982b). Size and shape as ecological discriminants in successional communities of Heteroptera. Biological Journal of the Linnean Society,18, 279-290. Brown, V.K. (1984). Secondary succession: insect-plant relationships. Bioscience, 34, 710-716. Brown, V.K. (1985). Insect herbivores and succession. Oikos, 44, 17-22. Brown, V.K. (1986). Life cycle strategies and plant succession. In The Evolution of Insect Life Cycles (Ed. by F. Taylor & R. Kasban), pp. 105-124. Springer Verlag, Berlin. Brown, V.K., Gange, A.C., Evans, I.M. & Storr, A.L. (1987). The effect of insect herbivory on the growth and reproduction of two annual Vida species at different stages in plant succession. Journal of Ecology, 1173-1189. 75, Brown, V.K., Gange, A.C. & Gibson, C.W.D. (1988). Effects of insect herbivory on vegetational structure. Proceedings of the International Symposium on Vegetational Structure. (Ed. by M. Werger), in press. Utrecht, The Netherlands. Brown, V.K., Hendrix, S.D. & Dingle, H. (1987). Plants and insects in early old-field succession: comparison of an English site and an American site. Biological Journal of the Linnean Society, 31, 59-74. Brown, V.K. & Hyman, P.S. (1986). Successional communities of plants and phytophagous Coleoptera. Journal of Ecology, 74, 963-975. Brown, V.K., Jepsen, M. & Gibson, C.W.D. (1988). Insect herbivory: effects on early old field succession demonstrated by chemical exclusion methods. Oikos, 52, 293-302. 200

Brown, V.K., Leijn, M. & Stinson, C.S.A. (1987). The experimental manipulation of insect herbivore load by the use of an insecticide (Malathion): The effect of application on growth. Oecologia, 72, 377-381. Brown, V.K. & Llewellyn, M. (1985). Variation in aphid weight and reproductive potential in relation to plant growth form. Journal of Animal Ecology, 54, 651-661. Brown, V.K. & Southwood, T.R.E. (1983). Trophic diversity, niche breath and generation times of exopterygote insects in a secondary succession. Oecologia, 56, 220-225. Brown, V.K. & Southwood, T.R.E. (1987). Secondary succession: patterns and strategies. In Colonization, Succession and Stability (Ed. by A.J. Gray, M.J. Crawley & P.J. Edwards), pp. 315-338. Symposia of the British Ecology Society, 26. Blackwell Scientific Publications, Oxford. Burdon, J.J. (1980a). Intraspecific diversity in a natural population of Trifolium repens. Journal of Ecology, 68, 717-735. Burdon, J.J. (1980b). Variations in disease resistance within a population of Trifolium repens. Journal of Ecology, 68, 717-736. Burdon, J.J. (1983). Biological Flora of the British Isles No. 154. Trifolium repens L. Journal of Ecology, 71, 307-330. Burdon, J.J. & Chivers, G.A. (1977). The effect of barley mildew on barley and wheat competition in mixtures. Australian Journal of Botany, 22, 103-114. Burdon, J.J. & Harper, J.L. (1980). Relative growth rates of individual members of a plant population. Journal of Ecology, 68, 953-957. Burgess, R.S.L. & Ennos, R.A. (1987). Selective grazing of acyanogenic white clover: Variation in behaviour among populations of the slug Deroceras reticulatum. Oecologia, 73,432-435. Burks, B.D. (1957). A new Bruchophagus from a liliaceous plant with a host plant list for the genus. Proceedings of the Entomological Society of Washington, 59, 273-277. Cahn, M.G. & Harper, J.L. (1976a). The biology of the leaf mark polymorphism in Trifolium repens L.1. Distribution of phenotypes at local scale. Heredity, 37, 309-325. Cahn, M.G. & Harper, J.L. (1976b). The biology of the leaf mark polymorphism in Trifolium repens L. 2. Evidence for the selection of leaf marks by rumen fistulated sheep. Heredity, 37, 327-333. Cammed, M.E. & Way, M.J. (1983). Aphid pests. In The Faba bean (Vida faba L.) a basis for improvement (Ed. by P.D. Hebblethwaite), pp.315-346. Butterworths, London. 201 Cantlon, J.E. (1969). The stability of natural populations and their sensitivity to technology. In Diversity and Stability of Ecological Systems (Ed. by G.M. Woodwell & H.H. Smith), pp.197-205. Brookhaven Symposia in Biology, 22. Springfield, Va. National Bureau of Standards, U.S. Dept, of Commerce Cartier, J.J. (1959). Recognition of three biotypes of the pea aphid from southern Quebec. Journal of Economic Entomology, 52,293-294. Chandra, S. & Williams, G. (1983). Frequeny-dependent selection in the grazing behaviour of the desert locust Schistocerca gregaria. Ecological Entomology, 3, 13-21. Chapman, D.F. & Anderson, C.B. (1987). Natural re-seeding and Trifolium repens demography in grazed hill pastures. I. Flowerhead appearance and fate, and seed dynamics. Journal of Animal Ecology, 24,1025-1035. Churchfield, S. & Brown, V.K. (1987). The trophic impact of small mammals in successional grasslands. Biological Journal of the Linnean Society,31, 273-290. Clapham, A.R., Tutin, T.G. & Warburg, E.F. (1981). Excursion Flora of the British Isles. Third Edition. Cambridge University Press, Cambridge. Clark, B. (1962). Balanced polymorphism and the diversity of sympatric species. In and Geography (Ed. by D. Nichols), pp. 47-70. Systematics Association, Oxford. Clark, B. & O’Donald, P. (1964). Frequency-dependent selection. Heredity, 19,201-206. Clements, F.E. (1916). Plant succession. An analysis of the development vegetation. Carnegie Institute, Publications 242, Washington D.C. Cody, M.K. (1966). A general theory of clutch size. Evolution. International Journal of Organic Evolution, Lancaster, 20,174-184. Pa., Coley, P.D. (1980). Effects of leaf age and plant life history patterns on herbivory. Nature, 284, 545-546. Compton, S.G. (1983). Studies of insects associated with Lotus corniculatus Ph.D. L. Thesis. University of Hull. Compton, S.G., Beesley, S.G. & Jones, D.A. (1983). On the polymorphism of cyanogenesis in Lotus corniculatus IX. Selective herbivory in natural populations at Porthdafarch, Anglesey. Heredity, 51, 537-547. Compton, S.G. & Jones, D.A. (1985). An investigation of the responses of herbivores to cyanogenesis in Lotus corniculatus L. Biological Journal of the Linnean Society, 26,21-38. Compton, S.G., Newsome, D. & Jones, D.A. (1983). Selection for cyanogenesis in the leaves and petals of Lotus corniculatus L. at high latitudes. Oecologia, 60, 353-358. 202

Connell, J.H. (1972). Community interactions on marine rocky intertidal shores. Annual Review of Ecology and Systematics, 3, 169-192. Connell, J.H. & Slatyer, R.O. (1977). Mechanisms of succession in natural communities and their role in community stability and organisation. American Naturalist, 111, 1119-1144. Coop, I.E. (1940). Cyanogenesis in white clover ( Trifolium repens L.). HI. A study of linamarase, the enzyme which hydrolyses lotaustralin. New Zealand Journal of Science and Technology, 22B, 71-83. Cooper-Driver, G.A. & Swain, T. (1976). Cyanogenic polymorphism in bracken in relation to herbivore predation. Nature, 260, 604. Corkill, L. (1940). Cyanogenesis in white clover ( Trifolium repens L.) I. Cyanogenesis in single plants. New Zealand Journal of Science and Technology, 22B, 65-67. Corkill, L. (1942). Cyanogenesis in white clover ( Trifolium repens L.). V. The inheritance of cyanogenesis. New Zealand Journal of Science and Technology, 23B, 178-193. Corkill, L. (1952). Cyanogenesis in white clover {Trifolium repens L.). VI. Experiments with high-glucoside and glucoside-free strains. New Zealand Journal of Science and Technology, 34 A, 1-16. Cottam, D.A. (1985). Frequency-dependent grazing by slugs and grasshoppers. Journal of Ecology, 73, 925-934. Crawford-Sidebotham, T.J. (1972). The role of slugs and snails in the maintenance of cyanogenesis polymorphisms of Lotus corniculatus and Trifolium repens. Heredity, 28,405-411. Crawley, M.J. (1988). Herbivores and plant population dynamics. In Plant Population Biology (Ed. by A.J. Davy, M.J. Hutchings & A.R. Watkinson), pp.367-392. Blackwell Scientific Publications, Oxford. Crawley, M.J. (1983). Herbivory: the Dynamics of Animal-Plant Interactions. Blackwell Scientific Publications, Oxford. Crawley, M.J. (1986). The structure of plant communities. In Plant Ecology (Ed. by M.J. Crawley), pp. 1-50. Blackwell Scientific Publications, Oxford. Crawley, M.J. & Nachapong, M. (1985). The establishment of seedlings from primary and regrowth seeds of Ragwort {Senecio jacobaea). Journal of Ecology, 73, 255-261. Cromartie, W.J. (1975). The effect of stand size and vegetational background on the colonization of cruciferous plants by herbivorous insects. Journal of Applied Ecology, 12,517-533. 203

Daday, H. (1954a). Gene frequencies in wild populations of Trifolium repens L. I. Distribution by latitude. Heredity, 8, 61-78. Daday, H. (1954b). Gene frequencies in wild populations of Trifolium repens L. n. Distribution by altitude. Heredity, 8, 377-384. Daday, H. (1965). Gene frequencies in wild populations of Trifolium repens L. IV. Mechanisms of natural selection. Heredity, 20, 355-365. Davis, B.N.K. (1973). The Hemiptera and Coleoptera of stinging nettle ( Urtica dioica L.) in East Anglia. Journal of applied Ecology, 10, 213-237. Dawson, C.D.R. (1941). Tetrasomic inheritance in Lotus corniculatus L. Journal of Genetics, 42,49-72. Dirzo, R. (1984). Herbivory: A phytocentric overview. Perspectives on Plant Population Ecology (Ed. by R. Dirzo & J. Sarukh&n), pp. 141-165. Sinauer Associates, Sunderland, Massachusetts. Dirzo, R. (1985). The role of the grazing animal. In Studies on plant demography: A Festschrift for John . LHarper, pp. 343-355. Academic Press, London. Dirzo, R. & Harper, J.L. (1982a). Experimental studies on slug-plant interactions, in. Differences in the acceptability of individual plants of Trifolium repens L. to slugs and snails. Journal of Ecology, 70, 101-117. Dirzo, R. & Harper, J.L. (1982b). Experimental studies on slug-plant interactions. IV. The performance of the cyanogenic and acyanogenic morphs of Trifolium repens L. in the field. Journal of Ecology, 70,119-138. Dritschilo, W., Krummel, J., Nafus, D. & Pimentel, D. (1979). Herbivorous insects colonizing cyanogenic and acyanogenic Trifolium repens. Heredity, 42,49-56. Drury, W.H. & Nisbet, I.C. (1973). Succession. Journal of the Arnold Arboretum, 54, 331-368. Edmunds, G.F. & Alstad, D.N. (1978). Coevolution in insect herbivores and conifers. Science, 199, 941-945. Egler, F.E. (1954). Vegetation science concepts. I. Initial floristic composition - a factor in old field development. Vegetatio, 4,412-417. Ellis, W.M., Keymer, R.J. & Jones, D.A. (1977a). On the polymorphism of cyanogenesis in Lotus corniculatus L. Vffl. Ecological studies in Anglesey. Heredity, 39,45-66. Ellis, W.M., Keymer, R.J. & Jones, D.A. (1977b). The effect of temperature on the polymorphism of cyanogenesis in Lotus corniculatus L. Heredity, 38, 339-347. Ellison, L. (1960). Influence of grazing on plant succession of rangelands. Botanical Review, 26, 1-78. 204

Elton, R.A. & Greenwood, J.J.D. (1970). Exploring apostatic selection. Heredity, 25, 629-633. Emmet, A.M. (Ed.) (1980). A Field Guide to the Smaller British Lepidoptera. British Entomological and Natural History Society, London. Ennos, R.A. (1985). The significance of genetic variation for root growth with a natural population of white clover (Trifolium repens). Journal of Ecology, 73, 605-614. Feeny, P. (1970). Seasonal changes in oak leaf tannins on the hydrolysis of proteins by trypsin. Phytochemistry, 8, 2119-2126. Feeny, P. (1975). Biochemical coevolution between plants and their insect herbivores. In Coevolution of animals and plants (Ed. by L.E. Gilbert & P.H. Raven), pp. 3-19. Texas University Press, Autin and London. Feeny, P. (1976). Plant apparency and chemical defense. Recent Advances in Phytochemistry, 10,1-40. Fisher, R.A., Corbet, A.S. & Williams, C.B. (1943). The relation between the number of species and the number of individuals in a random sample of an animal population. Journal of Animal Ecology, 12,42-58. Fishpool, L.D.C. (1982). The ecology of the West African grasshopper fauna of the Sahel and Sudan sahvanna with special reference to Oedoleus senezalensis (Krauss 1877). Ph.D. Thesis, University of London. Fitter, R., Fitter, A. & Blarney, M. (1978). The Wild Flowers of Britain and Northern Europe. 3rd edn. Collins, London. Ford, E.B. (1940). Polymorphism and taxonomy. In The New Systematics (Ed. by J. Huxley), pp. 493-513. Clarendon Press, Oxford. Forde, M.B. & de Lautour, G. (1978). Classification of Lotus introductions. New Zealand Journal of Experimental Agriculture, 6, 293-297. Foulds, W. (1977). The physiological response to to moisture supply of cyanogenic and acyanogenic phenotypes of Trifolium repens L. and Lotus corniculatus L. Heredity, 39, 219-234. Foulds, W. & Grime, J.P. (1972a). The influence of soil moisture on the frequency of cyanogenic plants in populations of Lotus corniculatus and Trifolium repens. Heredity, 28,143-146. Foulds, W. & Grime, G.P. (1972b). The response of cyanogenic and acyanogenic phenotypes of Trifolium repens to soil moisture supply. Heredity, 28, 181-187. Foulds, W. & Young, L. (1977). Effect of frosting, moisture stress and potassium cyanide on the metabolism of cyanogenic and acyanogenic phenotypes of Lotus corniculatus L. and Trifolium repens L. Heredity, 38,19-24. 205

Fowler, W.W. (1890). The Coleoptera of the British Islands. Lamellicornia-Serricornia-Longicornia-Phytophaga. Vol. IV. L. Reeve & Co., London. Fowler, W.W. (1891). The Coleoptera of the British Islands. Heteromera-Rhynchophora-Abnorma Coleoptera. Vol. V. L. Reeve & Co., London. Fraenkel, G.S. (1959). The raison d’etre of secondary plant substances. Science, 129, 1466-1470. Frazer, B.D. (1972a). Life tables and intrinsic rates of increase of apterous Black Bean Aphids and Pea Aphids on Broad Bean (Homoptera: Aphididae). Canadian Entomologist, 104,1717-1722. Frazer, B.D. (1972b). Population dynamics and recognition of biotypes in the Pea Aphid (Homoptera: Aphididae). Canadian Entomologist, 104,1729-1733. Freude, H., Harde, K.W. & Lohse, G.A. (1981). Die Kafer Mitteleuropas Vol. 10: Bruchidae, Anthribidae, Scolytidae, Platypodidae, Curculionidae. Goecke & Evers, Krefeld. Fullick, T.G. & Greenwood, J.J.D. (1979). Frequency-dependent food selection in relation to two models. American Naturalist, 113, 762-765. Gange, A.C., Brown, V.K., Evans, I.M. & Storr, A.L. (in press). Reduction of fitness in Trifolium pratense by insect herbivory during plant succession. Journal of Ecology. Gauch, H.G. (1982). Multivariate Analysis in Community Ecology. Cambridge University Press, Cambridge. Gauch, H.G., Whittaker, R.H. & Wentworth, T.R. (1977). A comparative study of reciprocal averaging and other ordination techniques. Journal of Ecology, 65, 157-174. Geiger, D.R. (1976). Effects of translocation and assimilate demand on photosynthesis. Canadian Journal of Botany, 54, 2337-2345. Georgiadis, N.J. & McNaughton, S.J. (1988). Interactions between grazers and a cyanogenic grass, Cynodon plectostachyus. Oikos, 51, 343-350. Gibson, C.W.D. & Brown, V.K. (1985). Plant Succession: theory and applications. Progress in Physical Geography, 9, 473-491. Gibson, C.W.D., Brown, V.K. & Jepsen, M. (1987). Relationships between the effects of insect herbivory and sheep grazing on seasonal changes in an early successional plant community. Oecologia, 71, 245-253. 206

Gibson, C.W.D., Dawkins, H.C., Brown, V.K. & Jepsen, M. (1987). Spring grazing by sheep: effects on seasonal change during early Old Field succession. Vegetatio, 70, 33-43. Gilbert, L.E. (1975). Ecological consequences of a coevolved mutualist between butterflies and plants. In Coevolution of Animals and Plants (Ed. by L.E. Gilbert & P.H. Raven), pp.210-214. University of Texas Press, Austin. Godfray, H.C.J. (1982). Leaf mining insects and their parasitoids in relation to plant succession. Ph.D. Thesis, University of London. Godfray, H.C.J. (1985). The absolute abundance of leaf miners on plants of different successional stages. Oikos, 45, 17-25. Greenwood, J.J.D. (1985). Frequency-dependent selection by seed-predators. Oikos, 44, 195-210. Greenwood, J.J.D. & Elton, R.A. (1979). Analysis experiments on frequency-dependent selection by predators. Journal of Animal Ecology, 48, 721-738. Greig-Smith, P. (1983). Quantitative Plant Ecology. Edn. 3rd Blackwell Scientific Publications, Oxford. Grime, J.P. (1979). Plant Strategies and Vegetation Processes. John Wiley, Chichester. Guppy, J. C. (1958). Insect surveys of clovers, and birdsfoot trefoil. Canadian Entomologist, 90, 523-531. Hairston, N.G., Smith, F.E. & Slobodkin, L.B. (1960). Community structure, population control and control. American Naturalist, 94,421-425. Hanson, F.E. (1983). The behavioural and neurophysiological basis of food-plant selection by Lepidopterous larvae. In Herbivorous insects, host seeking behaviour and mechanisms (Ed. by Sami Ahmad), pp.3-23. Academic Press, London. Harper, J.L. (1969). The role of predation in vegetational diversity. In Diversity and Stability of Ecological Systems (Ed. by G.M. Woodwell & H.H. Smith), pp.48-62. Brookhaven Symposia in Biology, 22. Springfield, Va. National Bureau of Standards, U.S. Dept, of Commerce Harper, J.L. (1977). Population Biology of Plants.Academic Press, London. Harper, J.L., Lovell, P.H. & Moore, K.G. (1970). The shapes and sizes of seeds. Annual Review of Ecological Systematics, 1, 327-356. Harrington, J.F. (1972). Seed storage and longevity. In Seed Biology, Vol. III. Insects and seed collection, Storage, Testing and Certification (Ed. by T.T. Kozlowski), pp. 145-245. Academic Press, New York. Haukioja, E. (1980). On the role of plant defences in the fluctuation of animal populations. Oikos, 35, 202-213. 207

Havlickova, H. (1982). Different responses of alfalfa plants to artificial defoliation and to feeding by pea leaf weevil {Sitona lineatus). Experimentia, 38, 569-570. Heath, J. (Ed.) (1976). The Moths and Butterflies of Great Britain and Ireland, Vol. 1. Micropterigidae-Heliozelidae. Blackwell Scientific Publications and Curwen Press, Oxford. Henderson, I.F. & Whittaker, T.M. (1977). The efficiency of an insect suction sampler in grassland. Ecological Entomology, 2, 57-60. Hendrix, S.D. (1979). Compensatory reproduction in a biennial herb following insect defloration. Oecologia, 42,107-118. Hendrix, S.D. (1984). Reactions of Heracleum lanatum to floral herbivory by Depressaria pastinacella. Ecology, 65, 191-197. Hendrix, S.D. (1988). Herbivory and its impact on plant reproduction. In Plant Reproductive Ecology: Patterns and Strategies (Ed. by J. Lovett-Doust & L. Lovett-Doust), pp. 246-263. Oxford university Press, New York. Hendrix, S.D., Brown, V.K. & Dingle, H. (1988). guild structure during early old field succession in a new and old world site. Journal of Animal Ecology, 57,1053-1065. Hendrix, S.D., Brown, V.K. & Gange, A.C. (in press). Effects of insect herbivory on early plant succession: comparison of an English and an American site. Biological Journal of the Linnean Society. Higgins, L.G. & Riley, N.D. (1970). A Field Guide to the Butterflies of Britain and Europe, Collins, London. Hill, M.O. (1973a). Reciprocal averaging: an eigenvector method of ordination. Journal of Ecology, 61, 237-249. Hill, M.O. (1973b). The intensity of spatial pattern in plant communities. Journal of Ecology, 61, 225-235. Hill, M.O. (1979). DECORANA-A FORTRAN program for detrended correspondence analysis and reciprocal averaging. Ecology and Systematics. Cornell University, Ithaca. New York 14850. 1-52. Hill, M.O. & Gauch, H.G. (1980). Detrended correspondence analysis: an improved ordination technique. Vegetatio, 42, 47-58. Hils, M.H. & Vankat, J.L. (1982). Species removals from a first year old field plant community. Ecology, 63,705-711. Hoffman, A. (1945). Faune de France: Coleopteres Curculionides, 52. Lechevalier, Paris. Hoffman, A. (1958). Faune de France: Coleopteres Curculionides, 62. Lechevalier, Paris. 208

Horn, H.S. (1976). Successions. In Theoretical Ecology, 1‘* edn. (Ed. R.M. May), pp. 187-204. Blackwell Scientific Publications, Oxford. Horsley, D.T., Lynch, B.M., Greenwood, J.J.D., Hardman, B. & Mosely, S. (1979). Frequency-dependent selection by birds when the density of prey is high. Journal of Animal Ecology, 48,483-490. Hubbard, C.E. (1984). Grasses: A Guide to their Structure, Identification, Uses and Distribution in the British Isles. 3*edn. Pelican Books, London. Hubbard, S.F., Cook, R.M., Glover, J.G. & Greenwood, J.J.D. (1982). Apostatic selection as an optimal foraging strategy. Journal of Animal Ecology, 51, 625-633. Hughes, M.A. (1968). Studies on the B-glucosidase system of Trifolium repens. Journal of Experimental Botany, 19,427-434. Jackson, J.B.C. (1981). Interspecific competition and species distributions: the ghosts of theories and data past. American Zoologist, 21, 889-902. Janzen, D.H. (1977). Variation in seed size within a crop of a Costa Rican Mucana andreana (Leguminosae). American Journal of Botany, 64, 347-349. Janzen, D.H. (1979). New horizons in biology of plant defenses. In Herbivores: Their Interaction with Secondary Plant Metabolites (Ed. by G.A. Rosenthal & D.H. Janzen), pp. 331-350. Academic Press, New York. Johansson, A.S. (1951). The food plant preference of the larvae of Pier is brassicae L. Norsk entomologisk tidsskrift, B8, 187-195. Jones, D.A. (1962). Selection eating of the acyanogenic form of the plant Lotus corniculatus L. by various animals. Nature, 193,1109-1110. Jones, D.A. (1966). On the polymorphism of cyanogenesis in Lotus corniculatus. I. Selection by animals. Canadian Journal of Genetics and Cytology, 8, 556-567. Jones, D.A. (1968). On the polymorphism of cyanogenesis in Lotus corniculatus L. II. The interaction with Trifolium repens L. Heredity, 23, 435-455. Jones, D.A. (1970). On the polymorphism of cyanogenesis in Lotus corniculatus L. IQ. Some aspects of selection. Heredity, 25, 633-641. Jones, D.A. (1972). Cyanogenic glucosides and their function. In Phytochemical Ecology (Ed. by J.B. Harbome), pp. 103-124. Academic Press, London. Jones, D.A. (1973). Coevolution and cyanogenesis. In Taxonomy and Ecology, (Ed. by V.H. Heywood), pp. 213-242. Academic Press, London. Jones, D.A., Keymer, R.L. & Ellis, W.M. (1978). Cyanogenesis in plants and animal feeding. Biochemical Aspects of Plain and Animal Coevolution (Ed. by J.B. Harboume), pp. 21-34. Academic Press, London. Jones, D.A. & Turkington, R. (1986). The biological flora of the British Isles: Lotus corniculatus L. Journal of Ecology, 74, 1185-1212. 209

Jones, E.W. (1959). Quercus L. Journal of Ecology, 47, 169-222. Joy, N.H. (1932). A practical handbook of British Beetles. H.F. & G. Witherby (1932). (Reprinted 1976). E.W. Classey Ltd., Faringdon, Oxon. Kendall, D.G. (1971). Seriation from abundance matrices. In Mathematics in the archeological and historical sciences. (Ed. by F.R. Hodson, D.G. Kendall & P. Tautu), pp. 215-252. Edinburgh University Press, Edinburgh. Kennedy, J.S. & Booth, C.O. (1951). Host alternation in Aphis fabae Scop. I. Feeding preferences and fecundity in relation to the age and kind of leaves. Annals of Applied Biology, 38, 25-64. Kevan, D.K. (1962). The British species of the genus Haltica Geofffay (Col., Chrysomelidae). Entomologist’s Monthly Magazine, 98,189-196. Keymer, R. & Ellis, W.M. (1978). Experimental studies on plants of Lotus corniculatus L. from Anglesey polymorphic for cyanogenesis. Heredity, 40,189-206. Kinsman, S. & Platt, W.J. (1984). The impact of a herbivore upon Mirabilis hirsuta, a fugitive praire plant. Oecologia, 65, 2-6. Koptur, S. & Lawton, J.H. (1988). Interactions among vetches bearing extrafloral nectaries, their biotic protective agents and herbivores. Ecology, 69, 278-283. Krieger, R.J., Feeny, P.P. & Wilkinson, C.F. (1971). Detoxification enzymes in the guts of catepillars: an evolutionary answer to plant defence. Science, 172, 579-581. Lane, C. (1962). Notes on the Common Blue (. Polyommatus icarus R.). Egg-laying and feeding on the cyanogenic strains of birdsfoot trefoil ( Lotus corniculatus). Entomologist*s Gazette, 13,112-116. Lawton, J.H. (1976). The structure of the arthropod community on bracken. Biological Journal of the Linnean Society, 73, 187-216. Lawton, J.H. (1978). Host-plant influence on insect-diversity: effects of space and time. In Diversity of Insect Faunas (Ed. by L.A. Mound & N. Waloff), pp. 105-125. Symposium of the Royal Entomology Society of London, 9. Blackwell Scientific Publications, Oxford. Lawton, J.H. & McNeill, S. (1979). Between the devil and the deep blue sea: on the problem of being a herbivore. In Population Dynamics (Ed. by R.M. Anderson, B.D. Turner & L.R. Taylor), pp. 223-244. Symposium of the British Ecological Society, 20. Blackwell Scientific Publications, Oxford. Lawton, J.H. & Price, P.W. (1979). Species richness of parasites on hosts: agromyzid flies on the British Umbelliferae. Journal of Animal Ecology, 48, 619-637. Lawton, J.H. & Schroder, D. (1977). Effects of plant type, size of geographical range and taxonomic isolation on a number of insect species associated with British plants. Nature, 265,137-140. 210

Lee, T.D. & Bazzaz, F.A. (1980). Effect of defoliation and competition on growth and reproduction in the annual plant Abutilon theophrasti. Journal of Ecology, 68, 813-822. Lemen, C. (1981). Elm trees and elm leaf beetles: patterns of herbivory. Oikos, 36, 65-67. Louda, S.M. (1982). Limitation of the recruitment of the shrub Haplopappus squarrosus (Asteraceae) by flower-feeding and seed-feeding insects. Journal of Ecology, 70, 43-54. Louda, S.M. (1983). Seed predation and seedling mortality in the recruitment of a shrub Haplopappus venetus (Asteraceae), along a climatic gradient. Ecology, 64, 511-521. Malone, C.R. (1969). Effects of diazinon contamination on an old-field ecosystem. American Midland Naturalist, 82,1-27. Margalef, R. (1968). Perspectives in Ecological Theory. University of Chicago Press, Chicago. Martin, H. & Worthing, C.R. (1976). Insecticide and fungicide Handbook. Blackwell Scientific Publications, Oxford. Maun, M.A. & Caver, P.B. (1971). Seed production and dormancy in Rumex crispus. 1. The effects of removal of cauline leaves at anthesis. Canadian Journal of Botany, 49,1123-1130. MacArthur, R.H. & Wilson, E.O. (1967). The Theory of Island Biogeography. Princeton University Press, Princeton, New Jersey. McBrien, H., Harmsen, R. & Crowder, A. (1983). A case of insect grazing affecting plant succession. Ecology, 64,1035-1039. McIntosh, R.P. (1980). The relationship between succession and the recovery process in ecosystems. In The Recovery Process in Damaged Ecosystems (Ed. by J. Cairns Jr.), pp. 11-62. Ann Arbor Science, Ann Arbor, Michigan. McNaughton, S.J. (1983). Compensatory plant growth as a response to herbivory. Oikos, 40, 329-336. McNaughton, S.J., Wallace, L. & Coughenour, M.B. (1983). Plant adaptation in an ecosystem context: effects of defoliation, nitrogen and water on growth of an African C4sedge. Ecology, 64, 307-318. Melville, F. & Doak, B.W. (1940). Cyanogenesis in white clover ( Trifolium repens L.). II. Isolation of the glucosidal constituents. New Zealand Journal of Science and Technology, 22B, 67-71. Miller, J.D., Gibson, PB, Cope, W.A. & Knight, W.E. (1975). Herbivore feeding on cyanogenic and acyanogenic white clover seedlings. Crop Science, 15, 90-91. 211

Mound, L.A., Morrison, G.D., Pitkin, B.R. & Palmer, J.M. (1976). Thysanoptera. Handbooks for the identification of British insects. 1 (11), 1-79 Morton, A.J. & Bates, J.W. (1985). ECOUB: a package of multivariate ordination methods. Imperial College, University of London. Murdock, W.W. (1969). Switching in general predators: experiments on predator specificity and stability of prey populations. Ecological Monographs, 39, 335-354. Murdock, W.W. & Oaten, A. (1975). Predation and population stability. Advances in Ecological Research, 9,1-131. Mytton, L.R. (1973). The effects of seed weight on the early growth and nodulation of white clover. Annals of Applied Biology, 73,329-338. Nicholson, S.A. & Monk, C.D. (1974). Plant species diversity in old-field succession on the Georgia Piedmont. Ecology, 55,1075-1085. Noble, I.R. & Slatyer, R.O. (1980). The use of vital attributes to predict successional changes in plant communities subject to recurrent disturbances. Vegetatio, 43, 5-21. Noble, M.D. (1958). A simplified clip cage for aphid investigations. Canadian Entomologist, 90, 760. Odum, E.P. (1969). The strategy of ecosystem development. Science, 164, 262-270. Parker, M.A. & Salzman, A.G. (1985). Herbivore exclosure and competitor removal: effects on juvenile survivorship and growth in the shrub Gutierrezia microcephala. Journal of Ecology, 73,903-913. Parsons, J. & Rothschild, M. (1964). Rhodanese in the larva and pupa of the common blue butterfly, Polyommatus icarus (Rott.) (Lepidoptera). Entomologist’s Gazette, 15, 58-59. Parvone, L.V. & Reader, R.J. (1982). The dynamics of seedbank size and seed state of Medicago lupulina. Journal of Ecology, 70, 537-547. Parvone, L.V. & Reader, R.J. (1985). Effects of microtopography on the survival and reproduction of Medicago lupulina. Journal of Ecology, 73, 685-694. Peet, R.K. & Christensen, N.L. (1980). Succession: a population process. Vegetatio, 43, 131-140. Pemadasa, M.A. & Lovell, P.H. (1974). Factors affecting the distribution of some annuals in the dune systems at Aberffraw, Anglesey. Journal of Ecology, 62, 379-402. Pickett, S.T.A. (1982). Population patterns through twenty years of old field succession. Vegetatio, 49,45-59. 212

Popham, E.J. (1941). The variation in the colour of certain species of Arctocorisa (Hemiptera. Corixidae) and its significance. Proceedings of the Zoological Society, London, A lll, 135-172. Popham, E.J. (1943). Further experimental studies of the selective action of predators. Proceedings of the Zoological Society of London, A112,105-117. Popov, T. (1976). New and little known thrips in our country. Rastitelna Zashchita, 24, 35-36. Putwain, P.D. & Harper, J.L. (1968). Components and regulation of a natural population of Rumex acetosella L. Journal of Ecology,421-431. 56, Rabinowitz, D. (1979). Bimodal distributions of seedling weight in relation to density of Festuca paradoxa Desv. Nature, 211, 297-298. Rackham, O. (1980). Ancient woodland. Edward Arnold, London. Rai, J.P.N. & Tripat hi, R.S. (1985). Effect of herbivory by the slug Mariaelle dussumieri and certain insects on growth and competition success of two annual weeds. Agriculture, Ecosystems and Environment, 13, 125-138. 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.D. (1981). The effect of native vegetation on the susceptibility of Aristolochia reticulata (Aristolochiaceae) to herbivore attack. Ecology, 62, 1187-1195. Rausher, M.D. & Feeny, P. (1980). Herbivory, plant density and plant reproductive success: the effect of Battus philenor on Aristolochia reticulata. Ecology, 61, 905-917. Reader, P.M. & Southwood, T.R.E. (1981). The relationship between palatability to invertebrates and successional status of a plant. Oecologia, 51, 271-275. Rhoades, D.F. (1979). Evolution of chemical defense against herbivores. In Herbivores: Their Interaction with Secondary Plant Metabolites (Ed. by G.A. Rosenthal & D.H. Janzen), pp. 3-54. Academic Press, New York. Rhoades, D.F. & Cates, R.G. (1976). Toward a general theory of plant antiherbivore chemistry. Recent Advances in Phytochemistry, 10,168-213. Rice, E.L. (1974). Allelopathy. Academic Press, New York. Richards, O.W. & Davies, R.G. (1977). Immes General Textbook of Entomology. 10th edn., Vol. 2. Chapman & Hall, London. Risch, S. (1980). The population dynamics of several herbivorous beetles in a tropical agroecosystem: the effects of intercropping com, beans and squash in Costa Rica. Journal of Applied Ecology, 17, 593-612. 213

Rockwood, L.L. (1973). The effect of defoliation on seed production of six Costa Rica tree species. Ecology, 54, 1363-1369. Root, R.B. (1973). Organization of plant-arthropod association in simple and diverse habitats: the fauna of collards ( Brassica oleracea). Ecological Monographs, 43, 95-124. Ross, M.D. & Jones, W.T. (1983). A genetic polymorphism for tannin production in Lotus corniculatus and its relationship to cyanide polymorphism. Theoretical and Applied Genetics, 64, 263-268. Sagar, G.R. & Harper, J.L. (1960). Factors affecting the germination and early establishment of plantains ( Plantago lanceolata, P. media, P. major). In Biology of Weeds (Ed. by J.L. Harper), pp. 236-245. Symposium of the British Ecological Society, 1. Blackwell Scientific Publications, Oxford. Salisbury, E.J. (1952). " Downs and Dunes". Bell, London. Scheiner, S.M. (1987). Size and fecundity hierarchies in an herbaceous perennial. Oecologia, 74,128-132. Scherf, H. (1964). Die Entwicklungsstadien der mitteleuropaischen Curculioniden (Morphologie, Bionomie, Okiologie). Abhandlungen herausgegeben, von der Senckenbergischen naturforschenden Gessellschaft, 506, 1-335. Scorer, A.G. (1913). The Entomologist’s Log Book. Routledge, London. Shure, D.J. (1971). Insecticide effects on early succession in an old field ecosystem. Ecology, 52, 271-279. Singh, J.S., Hadley, H.H. & Bernard, R.L. (1971). Morphology of pubescence in and its relationship to plant vigour. Crop Science, 11, 13-16. Smith, B.D. (1966). Effect of the plant alkaloid sparteine on the distribution of the aphid Acyrthosiphon spartii. Nature, 212, 213-214. Smith, J.G. (1976). Influence of crop background on aphids and other phytophagous insects on brussel sprouts. Annals of Applied Biology, 83,1-13. Snaydon, R.W. (1962). Micro-distribution of Trifolium repens L. and its relation soils factors. Journal of Ecology, 50, 133-143. Solbrig, O.T. (1981). Studies on the population biology of the genus Viola, n. The effect of plant size on fitness in Viola sororia. Evolution, 35, 1080-1093. South, R. (1961). The Moths of the British Isles. Wame, London. Southwood, T.R.E. (1960). The abundance of Hawaiian trees and the number of their associated insect species. Proceeding of the Hawaiian Entomological Society, 17, 299-303. Southwood, T.R.E. (1961a). The number of species of insect associated with various trees. Journal of Animal Ecology, 30, 1-8. 214

South wood, T.R.E. (1961b). The evolution of the insect host tree relationship - a new approach. Proceedings of the XIth International Congress on Entomology, Vienna 1960, pp. 651-654. Southwood, T.R.E. (1978). Ecological Methods. 2nd edn. Chapman and Hall, London. Southwood, T.R.E., Brown, V.K. & Reader, P.M. (1979). The relationship of plant and insect diversities in succession. Biological Journal of the Linnean Society,12, 327-348. Southwood, T.R.E., Brown, V.K. & Reader, P.M. (1983). Continuity of vegetation in space and time: A comparison of insects’ habitat templet in different successional stages. Research on Population Ecology, 3, 61-74. Southwood, T.R.E., Brown, V.K. & Reader, P.M. (1986). Leaf palatability, life expectancy and herbivore damage. Oecologia, 70, 544-548. Southwood, T.R.E., Brown, V.K., Reader, P.M. & Green, E.E (1986). The use of different stages of a secondary succession by birds. Bird Study, 33, 159-163. Southwood, T.R.E. & Leston, D. (1959). Land and Water Bugs of the British Isles. Frederick Wame. London. Spencer, K.A. (1972). Diptera: Agromyzidae. Handbook for the identification of British insects, 10(5g), 1-136. Stadler, E. & Hanson, F.E. (1978). Food discrimination and induction of preference for artificial diets in the tobacco homworm, Manduca sexta. Physiological Entomology, 3,121-133. Starks, K.J., Muniappan, R. & Eikenbary, R.D. (1972). Interaction between plant resistance and parasitism against the greenbug on barley and sorghum. Annals of the Entomological Society of America, 650-655. 65, Stinson, C.S.A. (1983). Effects of insect herbivores on early successional habitats. Ph.D. Thesis, University of London. Stinson, C.S.A. & Brown, V.K. (1983). Seasonal changes in the architecture of natural plant communities and its relevance to insect herbivores. Oecologia, 56, 67-69. Strong, D.R., Lawton, J.H. & Southwood, T.R.E. (1984). Insects on plants: community patterns and mechanisms. Blackwell Scientific Publications, Oxford, London. Stroyan, H.L.G. (1972). Additions and amendments to the check list of British aphids (Homoptera: Aphididae). Transactions of the Royal Entomological Society of London, 124,37-79. Stroyan, H.L.G. (1977). Hemiptera - Homoptera: Aphidoidea Chaitophoridae and Callaphididae. Handbook for the identification of British insects, 2,(4a.), 1-130. 215

Stroyan, H.L.G. (1984). Aphids-Pterocommatinae and Aphidinae (Aphidini) (Homoptera, Aphididae). Handbook for the identification of British insects, 2(6), 1-232. Tambs-Lyche, H. & Kennedy, J.S. (1958). Relationship between growth pattern and resistance to Aphis fabae Scopoli in three varieties of field bean {Vida faba L.). Entomologia Experimentalis et Applicata, 1, 223-239. Tansley, A.G. & Adamson, R.S. (1925). Studies of the vegetation of the English Chalk. IQ. The chalk grasslands of the Hampshire-Sussex border. Journal of Ecology, 13,177-223. Tahvanainen, J.O. & Root, R.B. (1972). The influence vegetational diversity on the population ecology of a specialized herbivore Phyllotreta cruciferae (Coleoptera: Chrysomelidae). Oecologia, 10, 321-346. Thompson, J.N. & Price, P.W. (1977). Plant plasticity, phenology and herbivore dispersion: wild parsnip and parsnip webworm. Ecology, 58, 1112-1119. Tinker, M.A.H. (1930). The effect of cutting to ground level upon the growth of established plants of Dactylis glomerata and Phleum pratense. Welsh Journal of Agriculture, 6, 182-198. Tor mala, T. (1982). Structure and dynamics of reserved field ecosystem in central Finland. Biological Research Reports from the University of Jyvaskyla, Finland, 8, 1-58. Tramer, E.J. (1975). The regulation of plant species diversity on an early successional old-field. Ecology, 56, 905-914. Turnipseed, S.G. & Kogan, M. (1976). Soybean entomology. Annual Review of Entomology, 21, 247-282. Turkington, R. (1979). Neighbour relationships in grass-legume communities. IV. Fine scale biotic differentiation. Canadian Journal of Botany, 57, 2711-2716. Turkington, R. (1983). Leaf and flower demography of Trifolium repens L. I. Growth in mixture with grasses. New Phytologist, 93, 599-616. Turkington, R. (1985). Variation and differences in populations of Trifolium repens in permanent pastures. In Studies on Plant Demography: A Festschrift for John L. Harper, pp. 69-82. Academic Press, London. Turkington, R., Cahn, M.A., Vardv, A. & Harper, J.L. (1979). The growth, distribution and neighbour relationships of Trifolium repens L. in permanent pasture. ILL The establishment and growth of Trifolium repens in natural and perturbed sites. Journal of Ecology, 67, 231-243. 216

Turkington, R. & Cavers, P.B. (1979). Neighbour relationships in grass-legume communities. EH. Development of pattern and association in artificial communities. Canadian Journal of Botany, 2704-2710. 57, Turkington, R., Cavers, P.B. & Aarssen, L.W. (1977). Neighbour relationships in grass-legume communities. I. Interspecific contacts in four grassland communities near London, Ontario. Canadian Journal of Botany, 55, 2701-2711. Turkington, R. & Harper, J.L. (1979a). The growth, distribution and neighbour relationships of Trifolium repens L. in permanent pasture. IV. Fine-scale biotic differentiation. Journal of Ecology, 67, 245-254. Turkington, R. & Harper, J.L. (1979b). The growth, distribution and neighbour relationships of Trifolium repens L. in permanent pasture, n. Inter- and intraspecific contact. Journal of Ecology, 67, 219-230. Turkington, R. & Harper, J.L. (1979c). The growth, distribution and neighbour relationships of Trifolium repens L. in permanent pasture. I. Ordination, pattern and contact. Journal of Ecology, 67, 201-218. Usher, M.B. (1987). Modelling successional processes in ecosystems. In Colonization, Succession and Stability (Ed. by A.J. Gray, M.J. Crawley & P.J. Edwards), pp.31-55. Symposia of the British Ecology Society, 26. Blackwell Scientific Publications, Oxford. van der Maarel, E. (1980). On the interpretability of ordination diagrams. Vegetatio, 42, 43-45. van der Meijden, E., Wijn, M. & Verkaar, H.J. (1988). Defence and regrowth, alternative plant strategies in the stmggle against herbivores. Oikos, 51, 355-363. van Emden, H.F. (1972). Aphid Technology. Academic Press, London. van Emden, H.F. & Bashford, M.A. (1969). A comparison of the reproduction of Brevicoryne brassicae and Myzus persicae in relation to soluble nitrogen concentration and leaf age (leaf position) in the Bmssels sprout plant. Entomologia Experimentalis et Applicata, 12, 351-364. van Emden, H.F. & Bashford, M.A. (1971). The performance of Brevicoryne brassicae and Myzus persicae in relation to plant age and amino acids. Entomologia Experimentalis et Applicata, 14, 349-360. Waal, D de (1942). Het cyanophore karacter van witte klaver, Trifolium repens L. Thesis, Landbouwhoogeschool, Wageningen. The Netherlands. Waller, D.M. (1985). The genesis of size hierarchies in seedling populations of Impatiens capensis Meerb. New Phytologist, 100, 243-260. 217

Waloff, N. & Richards, O.W. (1977). The effect of insect fauna on growth, mortality and natality of broom, Sarothamnus scoparius. Journal of Applied Ecology, 14, 787-798. Walsh, G.B. & Dibb, J.R. (Eds.) (1975). A Coleopterist’s Handbook. 2Dd edn. The Amateur Entomological Society, Felham, Middlesex. Walter, H. (1973). Vegetation of the Earth in relation to climate and the ecophysiological conditions. English Universities Press, London. Ward, L.K. (1973). Thysanoptera occurring in flowers of a chalk grassland. Entomologist, 106, 97-113. Ward, L.K. & Lakhani, K.H. (1977). The conservation of juniper: the fauna of food-plant islands sites in southern England. Journal of Applied Ecology, 14,121-135. Weiner, J. & Solbrig, O.T. (1984). The meaning and measurement of size hierarchies in plant populations. Oecologia, 61, 334-336. Weis, I.M. (1982). The effects of propagule size on the germination and seedling growth in Mirabilis hirsuta. Canadian Journal of Botany, 60, 1868-1874. Werner, P.A. & Platt, W.J. (1976). Ecological relationships of co-occurring golden rods (Solidago: Compositae). American Naturist, 110, 959-971. White, J. (1979). The plant as a metapopulation. Annual Review of Ecology and Systematics, 10,109-145. Whitmore, T.C. (1985). Forest Succession. Nature, 315, 692. Whittaker, J.B. (1979). Invertebrate grazing, competition and plant dynamics. In Population Dynamics (Ed. by R.M. Anderson, B.D. Turner & L.R. Taylor), pp. 207-222. Symposium of the British Ecological Society, 20. Blackwell Scientific Publications, Oxford. Whittaker, J.B. (1982). The effect of grazing by a chrysomelid beetle, Gastrophysa viridula, on growth and survival of Rumex crispus on a shingle bank. Journal of Ecology, 70, 291-296. Whittaker, R.H. (1965). Dominance and diversity in land plant communities. Science, 147, 250-260. Whittaker, R.H. (1969). Evolution of diversity in plant communities. In Diversity and Stability in Ecological Systems. (Ed. by G.M. Woodwell & H.H. Smith), pp. 178-196. Brookhaven Symposium in Biology, 22. Springfield, Va. National Bureau of Standard, U.S. Dept, of Commerce. Whittaker, R.H. (1975). Communities and Ecosystems. 2nd edn. Macmillan, London. Wilson, F. (1964). The biological control of weeds. Annual Review of Entomology, 9, 225-244. 218

Windle, P.N. & Franz, E.H. (1979). The effects of insect parasitism on plant competition: greenbugs and barley. Ecology, 60, 521-529. Wiseman, B.R. & McMillan, W.W. (1980). Feeding preferences of Heleothis zea larvae preconditioned to several host crops. Journal of the Georgian Entomological Society, 15,449-453. Wit, C.T. de (1960). On competition. Verslagen van het landbouwkundig onderzoek in Nederland, 66,1-82. Woodhead, S. & Bernays, E.A. (1977). Changes in the release rates of cyanide in relation to palatability of Sorghum to insects. Nature, 270, 235-236 219

Appendix 1. Importance Values of Each Plant Species Recorded in Sites of Different Successional Age. 220

1984 site Importance value Plant species May June July Aug Sept Oct Agropyron repens Beauv. 1.2 1.5 2.8 1.9 Agrostis capillaris L. 1.5 1.4 2.7 1.2 2.0 A. stolonifera L. 3.6 5.5 6.5 12.2 18.4 Anagallis arvensis L. 0.2 Anchusa arvensis L. 0.2 Aphanes arvensis L. 0.2 Capsella bursa-pastoris Medicus 14.1 11.3 5.9 5.3 4.9 4.8 Carex L. spp. 0.5 0.9 0.9 1.3 1.3 Cerastiumfontanum Baumg. 0.8 0.3 1.0 Chenopodium alba L. 38.8 11.7 10.7 8.1 7.0 6.5 C. rubrum L. 2.4 4.0 1.4 Cirsium arvense Scop. 0.5 C. vulgare Ten. 0.3 Conyza canadensis Cronq. 1.1 0.5 0.2 2.3 0.7 1.1 Crepis capillaris Wallr. 0.4 0.4 Equisetum palustre L. 1.0 3.5 5.5 7.6 9.0 8.7 Fallopia convolvulus A. Love 36.2 12.1 12.0 11.0 0.6 2.1 Galinsoga parviflora Cav. 7.0 9.0 6.9 9.5 8.5 13.3 Geranium dissectum L. 0.3 0.6 0.2 G. molle L. 0.3 G. pusillum L. 1.1 0.4 0.2 Gnaphalium uliginosum L. 0.3 1.4 2.3 4.0 3.2 Holcus lanatus L. 0.4 0.9 1.8 Hypericum perforatum L. 0.3 0.2 Hypochaeris radicata L. 0.6 Lactuca serriola L. 0.2 Lamium album L. 0.6 0.2 L. purpureum L. Leontodon autumnalis L. 0.6 1.1 Lotus corniculatus L. 4.1 0.2 Matricaria matricariodes Porter 5.9 1.7 0.4 1.0 0.2 0.4 Medicago lupulina L. 1.3 4.9 5.0 6.5 7.3 Myosotis arvensis Hill 0.2 221

1984 site (continued) Importance value Plant species May June July Aug Sept Oct Papaver dubium L. 1.4 1.7 0.4 0.9 0.4 0.9 Plantago major L. 0.3 1.4 0.7 0.6 0.7 P. media L. 0.4 2.7 4.1 3.6 Poa annua L. 3.2 8.4 3.6 2.7 1.4 1.5 Polygonum aviculare L. 26.2 7.2 17.4 25.8 26.5 25.3 P. persicaria L. 33.9 30.1 30.8 28.5 24.4 24.0 Ranunculus repens L. 0.6 0.2 0.2 Raphanus raphanistrum L. 10.4 12.0 16.4 17.9 21.4 25.1 Rumex acetosa L. 0.5 0.2 R. crispus L. 0.3 0.6 0.4 0.8 R. obtusifolius L. 0.4 7.8 0.2 Senecio vulgaris L. 0.3 Sisymbrium officinale Scop. 0.8 0.6 0.8 Sonchus asper Hill 2.5 4.9 7.0 5.9 8.3 S. oleraceus L. 2.8 2.1 5.9 4.8 6.7 6.4 Spergula arvensis L. 72.5 98.7 94.1 84.4 79.9 69.1 Stachys palustris L. 0.6 1.0 0.3 0.5 Stellaria media Vill. 15.1 21.1 2.7 4.3 1.4 5.6 Taraxacum officinale Dahlst. 0.4 0.8 Trifolium hybridum L. 1.7 2.2 2.7 2.8 T. pratense L. 0.3 0.7 0.7 1.2 2.0 T. repens L. 1.0 1.8 2.1 1.4 2.3 Tripleurospermum inordorum Schultz Bip. 13.1 19.7 23.1 27.1 24.7 22.5 Urtica dioica L. 0.3 0.2 Verbascum nigrum L. 0.2 Veronica persica Poiret 3.5 2.7 1.5 1.6 1.9 2.1 Vida hirsuta S.F. Gray 8.0 20.6 30.1 18.4 21.4 17.1 V. sativa L. 3.6 6.4 0.7 0.4 0.3 0.2 V. tetrasperma Schreber 0.6 Viola tricolor L. 1.1 0.2 222

1983 site 1982 site 1981 site 1980 site 1979 site 1977 site 1971 site Plant species MayJuly May July MayMay July JulyMay July May JulyMayJuly Achillea millefolium L. 7.0 0.7 4.4 9.1 19.2 23.7 2.2 2.2 6.0 1.9 3.9 2.1 3.5 3.7 Agropyron repens Beauv. 9.1 3.8 42.0 13.4 4.3 15.4 9.2 10.1 7.8 3.3 Agrostis canina L. 1.0 A. capillaris L. 81.5 98.4 35.1 94.0 49.8 81.5 32.0 49.1 44.0 37.3 38.1 57.5 18.7 17.0 A. stolonifera L. 3.2 11.6 21.5 6.6 Anthoxanthum odoratum L. 1.0 1.0 6.0 6.1 Arrthenatherum elatius Beauv. 4.2 2.3 1.0 1.6 1.8 1.7 1.1 1.6 13.2 28.6 ex J & CPresl Betula pendula Roth 2.7 2.1 Bromus mollis L. 8.6 15.6 B. steralis L. 4.5 Cerastiumfontanum Baumg. 2.1 4.7 1.3 1.8 Chamerion angustifolium J. Holub 2.1 5.2 0.6 Cirsium arvense Scop. 2.3 2.7 2.2 8.7 4.1 5.9 42.1 40.3 18.7 29.4 13.2 11.3 5.2 6.4 Convolvulus arvensis L. 3.5 Crepis capillaris Wallr. 4.8 10.9 6.1 1.4 3.2 3.3 Cytisus scoparius Link 7.7 11.6 Dactylis glomerata L. 3.6 8.8 5.9 3.4 3.8 3.7 4.9 55.4 43.3 26.1 26.0 Equisetum palustre L. 31.4 Fallopio convolvulus A. Love 1.3 2.7 Festuca rubra L. 1.5 0.6 3.9 0.6 1.5 Galium aparine L. 1.0 Geranium dissectum L. 2.8 1.1 8.1 0.6 1.4 0.7 Glechoma hederacea L. 5.0 0.5 Holcus lanatus L. 108.4 42.5 110.5 41.3 122.2 43.2 63.4 40.0 61.3 47.3 64.6 59.1 41.3 25.3 H. mollis L. 26.6 12.3 0.7 3.2 Hypericum perforation L. 6.3 2.8 0.7 Hypochaeris radicata L. 12.3 22.7 3.5 13.6 2.2 4.5 4.0 4.9 7.3 J uncus effusus L. 0.7 0.6 0.7 1.6 5.2 Leontodon autumnalis L. 2.0 1.3 Lotus comiculatus L. 2.3 3.3 1.4 1.6 1.3 2.8 5.4 5.3 11.5 Luiula L. spp. 1.7 0.7 29.3 43.2 Medicago lupulina L. 29.2 48.8 20.2 46.8 1.0 2.2 0.6 Phleum pratense L. 8.2 3.1 0.9 Pinus sylvestris L. 0.8 Plantago lanceolata L. 6.9 9.4 20.4 40.6 44.4 66.4 1.6 3.7 5.6 2.5 8.7 12.0 23.8 P. major L. 2.2 0.7 223

1983 site 1982 site 1981 site 1980 site 1979 site 1977 site 1971 site Plant species MayJuly MayJulyMayJulyMay July MayJulyMay JulyMayJuly Poa annua L. 4.8 P. pratensis L. 0.7 1.0 2.0 31.9 19.8 P. trivalis L. 2.4 4.2 16.9 0.7 Pulcaria dysenterica Bemh. 3.7 1.4 1.2 Quercus robur L. 3.4 9.4 1.0 Ranunculus repens L. l.i 6.5 4.0 41.4 25.1 13.2 9.3 2.9 0.8 17.0 4.1 Raphanus raphanistnun L. 1.3 2.8 2.9 5.0 Rubus fruticosus L. 1.7 2.1 3.7 3.4 0.8 2.9 Rumex acetosa L. 8.6 1.3 2.1 0.7 1.3 1.4 R. acetosella L. 0.8 0.9 R. crispus L. 0.9 0.7 1.7 Senecio jacobaea L. 1.0 3.0 1.0 2.8 4.2 0.9 0.5 Stachys syhatica L. 0.8 0.8 Stellaria media Vill. 1.4 S. graminea L. 0.7 1.2 21.4 34.1 26.0 30.5 7.7 12.3 8.2 8.4 18.0 23.2 Taraxacum officinale D ahlst 11.3 2.3 4.2 0.7 9.1 3.7 7.9 1.3 Trifolium dubium Sibth. 0.7 1.0 0.7 T. hybridum L. 1.4 T. pratcnse L. 37.1 46.4 18.1 18.6 1.0 2.3 0.8 0.6 T. repens L. 17.3 22.0 31.2 13.9 0.9 1.0 0.9 0.8 0.6 1.3 13.7 14.2 Tussilago farafara L. 0.5 Urtica dioica L. 0.8 Veronica chamaedrys L. 2.1 3.4 8.7 V. offinalis L. 0.5 V. persica Poirct 3.2 Vida hirsula S i\ Gray 2.2 4.0 1.0 1.0 3.6 17.2 27.0 16.8 16.8 22.4 2.0 V. saliva L. 2.6 13.1 1.2 13.8 15.2 29.9 17.5 21.0 12.6 16.7 16.5 224

Appendix 2. British Species of Phytophagous Insects Recorded as Occurring on Leguminosae (particularlycorniculatus, L. M. lupulina, T. pratense, T. repens, V. hirsuta & V. sativa), Ordered Alphabetically; Symbols in Footnotes. 225

Reference Herbivore Host Plant source

COLEOPTERA A pionidae Apion aethiops H erbst V. sativa 13 Vida spp. 14 A. apricans H erbst T. pra tense 13, 14, 26 A. assimile K irby T. pratense 13, 26 Trifolium spp. 14 A. craccae

Reference Herbivore Host Plant source

GOLEOPTERA (Contd.) H. post lea (Gyllenhal) V. satlva 26 Leguminosae 13 H. venusta L. corniculatus 26 Lotus spp. 13 Legum inosae 13 Miccotrogus picirostris (F.) T. pratense 13, 26 Phyllobius roboretanus G redler L. corniculatus 10 P. viridiaeris (Laicharting) L. corniculatus 10 Sitona hlspidulus (F.) M. lupulina 26 Medicago spp. 14 T. pratense 20 T. repens 26 Tr I folium spp. 13, 14 5. humeral is Stephens Lotus spp. 13 M. lupulina 20, 26 Medicago spp. 14 T. pratense 22 T. repens 14 Trifolium spp. 13, 21 V ida spp. 13 5. lepidus Gyllenhal M. lupulina 26 T. pratense 14, 26 Tri folium 13 Leguminosae 28 S. lineatus

COLLEMBOLA Sminthur idae Bourletiella signata (Nic.) Tri folium spp. 24 Sminthurus viridus

DERMAPTERA Forf iculidae Forficula auricularia (L.) L. corniculatus 10 227

Reference Herbivore Host Plant source

DIFTERA Agromyzidae Agromyza fron tel la Rondani M. lupulina 30 A. nana Meigen M. lupulina 30 T. pratensc 30 T. repens 30 A. vicifoliae H ering V ida spp. 30 Liriomyza congests (Becker) L. corniculatus 30 T. repens 30 V. hirsute 30 V. sativa 30 Vicia spp. 30 Melanagromyza cuntans <. Meigen; L. corniculatus 30 Ophiomyia orbiculata (Hendel) Vicia spp. 30 L. corniculatus 30 Cec idomy i Idae Anabremia massalongi K ieffer Vicia spp. 3 A. medicaginls RUbsaamen Medicago spp. 3 A. tro ttcri K ieffer V ida spp. 3 A. viciae K ieffer 7/oia spp* 3 Asphondylia ervi RUbsaamen V ida spp. 3 A. lupullnae K ieffer Medicago spp. 3 A. melanopus K ieffer Lotus spp. 3 Campylomyza ornerodi (K ieffer) Trifolium spp. 3 Contarinia barbicbei (K ieffer) Lotus spp. 3 C. craccae K ieffer V ida spp. 3 C. lo ti (DeGeer) Lotus spp. 3 C. medicaginis K ieffer Medicago spp. 3 Dasyneura axillaris (K ieffer) Trifolium spp. 3 D. floscuiorum (K ieffer) Trifolium spp. *JO D. ignorata (W acbtl) Medicago spp. 3 D. leguminicola (L intn er) Tri folium spp. 3 D. loewiana RUbsaamen V ida spp. 3 D. lo ti (K ieffer) Lotus spp. 3 D. lupulinae (K ieffer) Medicago spp. 3 D. spadicea RUbsaamen V ida spp. 3 D. trifolii (F. Loew) Tri folium spp. 3 D. viciae (K ieffer) Vicia spp. 3 Hadrobremia longiventris (K ieffer) Trifolium spp. 3 Jaapiella japiana (RUbsaamen) Medicago spp. 3 /. loticola (RUbsaamen) Lotus spp. 3 J. medicaginis (RUbsaamen) Medicago spp. 3 Watchtliella dalmatica RUbsaamen Medicago spp. 3

HETEROPTERA Berytinidae Berytinus minor (Herricb-Scbaeffer) T. repens 29 Leguminosae 29 B. montivagus (Meyer) Medicago spp. 29 B. signoreti (Fieber) L. corniculatus 29 228

Reference Herbivore Host Plant source

HETEROPTERA (Contd.) Coreidae Batbysolen nub Hus (Fallen) M. lupulina 29 Ceraleptus lividus Stein T. pratense 29 Trifolium spp. 29 Coriomeris denticulatus (Scopoii) M. lupulina 29 Trifolium spp. 29 Leguminosae 29 Miridae Adelphocoris lineolatus (Goeze; L. corniculatus 29 Tri folium sop. 29 Leguminosae 29 A. seticomis (F.) L. corniculatus 29 Leguminosae 29 Calocoris norvegicus cGmelin) L. corniculatus 29 Tri folium spp. 29 C. roseomacuiatus (De G.) L. corniculatus 29 Trifolium spp. 29 Leguminosae 29 Capsodes sulcatus (Fieber) Leguminosae 29 Chlamydatus pullus (Reuter) M. lupulina 29 T. repcns 29 Trifolium spp. 29 Leguminosae 29 C. saltitans (Fallen) M. lupulina 29 T. repcns 29 Tri folium spp. 29 Leguminosae 29 Conostetbus roseus (Fallen) Trifolium spp. 29 Globiceps flavomaculatus ^F.) Leguminosae 29 Halticus apterus (L.) Leguminosae 29 Lygus maritimus Wagner Trifolium spp. 29 L. pratensis (L.) L. corniculatus 29 L. rugulipennis Poppins L. corniculatus 29 Tri folium spp. 29 Pbytocoris varipes Bobeman Trifolium spp. 29 Piezodorus lituratus (L.) Leguminosae 29 Plagiognathus chrysantbemi iWolff) L. corniculatus 16 M. lupulina 29

HOMOPTERA Aph ididae Acyrtbosipbon loti (Theobald) L. corniculatus 6 A. pisum (Harris) L. corniculatus 6 Medicago spp. 31, 33 T. pratense 31, 33 T. repens 31, 33 Apbis craccae (L.) Vida spp. 31, 33 .4. craccivora Koch L. corniculatus 31, 33 Trifolium sop. 31, 33 Vida spp. 31, 33 .4. fabae Scopoii T. pratense 31, 33 T. repens 31, 33 229

Reference Herbivore Host Plant source

HOMOPTHRA (Contd.) Aphis lo ti Kaltenbach L. corniculatus 31, 33 A. lot irad id s S tro y an L. corniculatus 31, 33 A. scalial Del Guerclo H. lupulina 31, 33 Trifolium spp. 31, 33 Aulacorthum solani (Kaltenbach.) T. pratensc 6 T. repens 6 Brachycaudus helichrysi (Kaltenbach) T. pratense 6 T. repens 6 Macrosiphuia euphorbiae (Thomas) T. pratense 6 T. repens 6 Megvura viciae Buckton Vida spp. 6 Myxus persicae (Sulzer) T. pratense 6 T. repens 6 Callaphidldae Therio aphis luteola (BBrner) T. pratense 32 T. trifolii (Monell) L. corniculatus 32 M. lupulina 32 T. repens 32

HTMEHOPTERA Eurytomidae Eurytoma platyptera (W alker) L. corniculatus 9 N em atinae myosotidis Fallen Trifolium spp. 5 Tenthred in Idae Tenthredo acerrima Benson L. corniculatus 4 T. arcuata F o rste r L. corniculatus 4 T. repens 4 T. perk in si (M orice) T. repens 4

LEPIDOPTERA A rc tiid a e Eilema caniola (HUbner) L. corniculatus 27 T. repens 27 2s. complana (L.) L. corniculatus 27 Trifolium spp. 27 Spilosoma luteum (Hufnag-el) Vida spp. 27 Coleophoridae Coleophora discordella Z eller L. corniculatus 11 Cosmopt er ig-inae Cosmopterix schmidiella Frey V icia spp. 15 anthyllidella (Htibner) L. corniculatus 15 Trifolium spp. 15 ononidis (Zeller) Trifolium spp. 15 Syncopaoma cinctella (C lerck) L. corniculatus 11 5. larseniella (Gozmany) L. corniculatus 11 5. sangiella (Stalnton) L. corniculatus 11 S. taeniolella (Z eller) L. corniculatus 11 Trifolium spp. 11 Reference Herbivore Host Plant source

LKPIDOPTKRA (Contd. > Geometridae Ematurga atomaria (L.) L. corniculatus 27 Trifolium 27 Idaea ochrata (Scolopi) L. corniculatus 27 Scopula cmutarla (HUbner) L. corniculatus 28 S. rub igin ata (Hufnagel) TrIfolium spp. 28 Scotopteryx bipunctaria (Den. & Sch.) L. corniculatus 28 Trifolium spp. 27 5. chcnopodiata (L.) L. corniculatus 1 Tri folium spp. 27 Vida spp. 27 Selidosema cricetaria (V illers) L. corniculatus 27 Hesperiidae Erynnis tages (L.) L. corniculatus 27 Las iocamp idae Laaiocaapa trifolii (Den. & Sch.) L. corniculatus 28 Lithocollet inae Phyllonoryctera insignitella (Zeller) Tri folium spp. 15 P. nigrescentclla (Logan) Tri folium spp. 15 Lycaenidae Everes argiades (P allus) L. corniculatus 27 Lyaandra coridon (Poda) L. corniculatus 18 Plebejus argus (L.) L. corniculatus 18 Polyommatus icarua (Rottemburg) L. corniculatus 18, 27 lotella (Stainton) L. corniculatus 11 Kepticul idae Trifurcula cryptella (Stalnton) L. corniculatus 17 T. eurema (T u tt) L. corniculatus 17 N octuidae Argot Is clavis (Hufnagel) Trifolium spp. 27 (C lerck) Tri folium spp. 27 Egira conapicillaris (L.) L. corniculatus 27 Euclid la glypbica (L.) L. corniculatus 27 M. lupulina 27 Trifolium spp. 27 Euxoa cursoria (Hufnagel) L. corniculatus 27 E. nigricans

Reference Herbivore Host Plant source

LHPIDOFTERA (Contd.) Tortricidae viburnana (Den. & Sch.) L. corniculatus 7, 8 Clcpsis scncdonana (HUbner) L. corniculatus 7, 8 Cnepbasia communana (Herr.-Sch.) L. corniculatus 7, 8 C. incertana (Treitschke) L. corniculatus 7, 8 Cydia compositclla (F.) L. corniculatus 7, 8 C. stephensiana (Doubleday) L. corniculatus 7, 8 Paraclepsis cinctana (Den. & Sch.) L. corniculatus 7, 8 Philedonc gcrningana (Den. & Sch.) L. corniculatus 7, 8 Zygaenidae Zygaena exulans (Hohenworth) L. corniculatus 27 Trifolium spp. 27 Z. fllipendulae (L.) L. corniculatus 27 Trifolium spp. 27 Z. lonicerae (Scheven) L. corniculatus 27 Trifolium spp. 27 Z. lo ti (Den. & Sch.) L. corniculatus 27 Z. purpuralls (Brtlnnich) L. corniculatus 27 Z. trifolii (E sper) L. corniculatus 27 Z. viviac (Den. & Sch.) L. corniculatus 27 T. rcpens 27

THTSAHOPTERA Aelothripae Aclothrips ericae Bagnall Leguminosae 23 A. intcrmedlus Bagnall Leguminosae 23 A. tcnuicormis Bagnall Legurainosae 23 T h rlp id ae Ondothrips biuncus John Vida spp. 23 0. lo ti (Haliday) Lotus spp. 23 0. phalaratus (Haliday) Vida spp. 23 Sericothrips abnormis (K arny) L. corniculatus 23 S. gracilicormis W illiam s Vida spp. 23 Tbrips atratus H aliday L. corniculatus 35 T. pbysapus L. L. corniculatus 35 T. vulgatissimus H aliday L. corniculatus 35

Reference Source (complete references are provided in bibliography). 1 Allan, P.B.M. (1948) 2 Allen, A.A. (1952) 3 Barnes, H.F. (1946) 4 Benson, R.B. (1952) 5 Benson, R.B. (1958) 6 Blackman, A. (1974) 7 Bradley, J.D., Tremewan, W.G. & Smith, A. (1973) 8 Bradley, J.D., Tremewan, W.G. & Smith, A. (1979) 9 Burks, B.D. (1957) 232

Reference Source. (Contd.) 10 Compton, S.G . k Jones, D.A. (1985) 11 Emmet, A.M. (1980) 12 Fowler, V.V. (1890) 13 Fowler, W.W. (1891) 14 Freude, H., H arde, K.W. k Lohse, G.A. (1981) 15 Godfrey, H.C.J. personal communication 16 Guppy, J.C. (1958) 17 Heath, J. (1976) 18 Higgins, L.G. k R iley, N.D. (1970) 19 Hoffmann, A. (1945) 20 Hoffmann, A. (1958) 21 Joy, N.H. (1932) 22 Kevan, D.K. (1967) 23 Hound, M.A., Morlson, G.D., Pitkin, B.R. & Palmer, J.M. (1976) 24 Popov, T. (1976) 25 R ich ard s, O.W. k Davies, R.G. (1977) 26 S cherf, H. (1964) 27 Scorer, A.G. (1913) 28 South, R. (1961) 29 Southwood, T.R.E. k L eston, D. (1959) 30 Spencer, K.A. (1972) 31 Stroyan, H.L.G. (1972) 32 Stroyan, H.L.G. (1977) 33 Stroyan, H.L.G. (1984) 34 Walsh, G.B. k Dibb, J.R. (1975) 35 Ward, L.K. (1973) 233

Appendix 3. Number of Phytophagous Insects Recorded by Univac Suction Sampling onL. corniculatus, M. lupulina, T. pratense, T. repens, V. hirsuta and V. sativa in Sites of different Successional Age. 234

Lot. us C am i cul a t us

1983 site 1971 site

29/5 19/8 8/8 29/8 27/9 16/10 6/6 4/7 8/8 29/8 27/9 16/10

COLEOPTERA Aplot aprleans t U 14 29 18 7 Aploa assiMile 2 2 ll 18 9 13 Apioa eraeeae 2 Apioa dlekrooa 4 6 4 3 2 l 4 4 2 l l Apioa loti 8 10 4 4 3 4 3 4 13 10 2 2 Apioa trttolil 2 7 to 5 3 Apioa vireas 2 2 5 l 3 Braekas rnflpes l Bypera nigrlrostfs l l 2 Miceotrogas picrirostis 2 2 1 t Sltoaa klspidalos 2 4 2 2 1 Sitoaa iepidos l l 3 1 Sltoaa liaeatus 3 3 Sltoaa soleifroas l t 12 25 12 9 Tyckins pnsiilos l

HBTBIOPTEIA Adelpkocoris llaeolatas 12 2 Berytiaos Minor 1 l 14 3 3 3 2 4 4 Berytiaos sigaoreti t l l t 5 3 6 l tl 19 6 7 Berytiaos lyiphs l 40 3 24 33 7 7 Ceralepos lividos t Pkytoeoris varipcs 9 2 10 l 3 Plaglopatkas ekrysaatkesi132 69 IB 52 52 1

HOKOPTERA Acyrtkosipkoasum pi 2 4 9 9 12 to Tkeoapkis trltolli l 2

LEPIDOPTERA Arctla eaja 1 l Antograpka gaaaa 2 2 l 3 4 SileMa eoMplaaa l Moetaa proaoka 1 2 Pkyloppkora Metical osa 1 PolyoMMatos iearns 2 3 l Scotopteryi ekeoopodiata 1 3 2 Iaatkorkoe soataaota 3 l lygaena fllipeadolae 2 18 4 1 2 7 4 3

THYSASOPTERA Aeolotkrips internedins Fraakliniella iatoasa Serieotkrips akaorais Serieotkrips nymphs 235

Meet I cag-o 1 upzil 1 ua.

1979 site 1980 site

5/6 3/7 9/8 28/6 28/9 5/6 3/7 9/8 28/6 26/9

COLEOPTBHA Apiot aprleans 6 3 7 3 Apiot asJtile l t 1 Apia eraeeae t l Apia dickroot 1 1 Apioi loti l 3 3 l L 3 2 l Apiot virens l Brackes loti l Breekas rnfipes 2 1 Bypen post in l Mleeotrogos plerirostis 3 Si tots klspidaias l l l 1 l 2 SI ton koae rails 5 2 It 10 1 4 It Si ton Hiatus 7 5 2 l 7 8 3 Si ton sal citrons l 1

HETEIOPTEIA Adelpkoearis linealatas 4 1 lerytinos Minor 1 Beryttnns lyiphs 2 Cenlepas liridas 1 Cor later is deiticolates t hyps regal ipenis 1 Phytoeoris nripcs 14 6 1 3 18 9 4 2 Plsgiognatias arbastorea 1 12 Plagiopatkes ckrysantkeai 156 11 93 10 t PI agiopatkus nyiphs 161 70

BOKOPTERA Acyrtkosipkon pisot 2 10 t S 8 1 12 5 Mepera viciae l Tkeoapkis tritolii 1 7 6 1

LBPIDOPTEM Arctic caja t t Aetopapha p a n l Scotopteryr ekenopodiata l lypena tillpendelae l UiideatIfled l Uiideatifled l 1 l OnldentIf led 1

3TMPHTTA Meautes ayosotidis 3 1

THYSAIOPTBRA Aelotkrips interaedies t Prankliniella intonsa 2 1 1 3 1 Serieotkrips aknorais 1 7 5 1 2 2 15 236

Trifal 1 um p ra t. ense

1983 site 1982 site

24/3 19/6 8/6 29/8 27/9 IB/10 30/5 28/6 8/8 29/8 29/9

C0LB0PTB1A Apiot tprictts 13 115 92 215 87 51 129 5 99 116 84 Apiotm s s Jaile 32 71 44 48 4 4 149 4 10 12 10 Apiot diekroas 1 5 l l 16 Apias loti l Apios tritolll 4 23 25 17 2 j 5 3 3 2 2 Apiot rim s 17 2 t 2 14 5 4 Bypert sigrirostis l 5 l Miccotrogns piertrostis U *i 3 2 3 Si ton kispiialns 2 1 t 3 5 2 4 4 Si tomm knertlis 1 2 l SitoiM lepiins SitOMM UlCMtOS 1 l l l 6 2 Sitou sol citrons 13 3 2 ll 21 29 13 12 22 44

HBTBIQPTB1A Adeipkoeoris UtcoiMtas 3 l 2 Berjtitos titor 34 2 1 2 4 18 Berytiias lympks 14 Lygts regal ipettis 2 2 fkytocoris rtripes It 1 l 1 12 13 1 ficiodotos iitarttos 1 Pltgiopttkns Mrbastorot l fltgiopotkos ckrpMttkcti l 33 5 59 69 21

I0V0PTB1A Acyrtkosipkoi pisot 1 l ll 2 5

LBPIS0PTB1A Ellen COMpiMIM 1 JhctoM protakt l fkyioppkort scticaiosM 1 l lypctt filipcidaiMe 2 Oildeitltled l

STMP1TTA Besutus syosotidis l t l l l

THT5AI0PTERA Aelotkrips ittcrscdias 4 4 2 1 FrMtklitiellM intoism 3 8 Sericotkrips Mbtortis 13 28 63 50 197 270 4 6 51 69 603 237

Tr I fol i u m r opens

1914 s i t e 1013 s i t e 1981 s i t e 1971 s i t e

11/9 11/10 30/9 30/1 1/1 29/1 17/9 17/10 30/5 30/6 8/8 29/6 26 /9 6/6 4/7 7/6 11 /1 17/9 5/1 1

COLEOPIB1A Aflat oprlcots l l l U 19 20 8 5 32 3 7 17 l l Aflat ass lolls 7 82 27 4 l 55 l 5 9 9 Aflat erteete l Aflat ilekroaa 1 11 17 32 l l 2 18 7 7 7 2 5 14 14 5 Aflat loti 1 1 l l l Aflat tipltsrse l Aflat trltoltl 31 21 4 2 5 l Aflat tints 1 l 1 2 2 l 13 12 7 6 3 7 l B/fart ilplnstts l l 2 ileeotraps plerlrostis l 33 5 5 4 l 3 3 2 2 4 5 l 2 l St tot* caai rices l l Sltoit kisptAtlas to 30 1 l 4 2 10 4 5 l 5 4 2 l l 3 l St tots leplits 1 7 l l 8 2 1 2 t 6 4 l l 3 Sttan lisettus 3 l l l 3 2 4 l SI ton sti elf rats 4 1 1 1 4 8 8 5 3 19 16 T/eilts ftslilts l l 3 3 2 2 4 H r n i o r r m Aielfkaeoris Jlieolatas l 2 5 l 4 5 l Berpthas altar 3 UO 25 2 l 18 119 52 26 6 22 2 13 4 Ser/thas slpantt l l l l 2 l 3 lerptltas ly ap k s l 108 17 Carolepas llrlias 1 l Cjrtss aelotaeefkolas l Ljgts rvpl tpenis 11 6 Piptoeoris roripes l 2 4 3 2 2 l Flo flaps tins ckr/sastkesi 10 47 5 119 55 8 32 31 I0 IQ R B U Asjrrtiosipioi pissa 17 51 3 25 34 33 7 3 3 8 3 3 ll 4 19 2 10 troei/eosias kellckr/st l Tkeetpils trItel 11 l LSm OPTElA Aitepopia pass 3 3 l l 9 3 l l l l 3 Blorsio rtil l Bfiro eetspleillorls l l Blleao eeaploso l Btelldo fi/piles l l Etna tlpletts l Pk/ioppkaro tetiealass 1 l l l l Polpaaaotss ieorts l 2 Seotopter/r ekeiapodiots 2 l l l 6 8 loitiarkae aottoioto l I/peto tlllpeideloe 4 2 238

T r ib a llu m repens (contd. >

1914 site 1985 site 1983 s i t e 1971 s i t e

18/9 IS/10 30/5 30/8 8/8 39/8 37/9 17/10 30/5 30/8 8/8 39/8 38/9 8/6 4/7 7/8 38/8 37/9 5/U

LHH0FTE1A contd. Unidentified 1 l 1 Unidentified 1 Unidentified l Unidentified l Unidentified 3 m n m BemttMS ujosotlils l T IT S U O m iA Frukliilellt ittatst : i 4 5 l t 3 SerlcatMrtps liion is U 7 80 13 183 599 318 1093 357 5 85 383 344 1437 9 37 35 49 69 19 Serleetkrlps grteiiieerMis l SerteotMrips iy v fU 6 147 88 455 801 33 8 118 30 U 7 588 36 11 8 17 31 19 3 239

Vic ia hlrstxta Vlcia sativa

1171 1977 im 1961 1964 1971 1977 1979 I960 t9 S t s i t e s i t e site site site s i t e s i t e s i t e s i t e s i t e

6/1 1/7 31/5 1/7 4 /6 3/7 3Q/5 4/7 3 1 /5 4/7 6 /8 7 /7 4 /6 3/7 6 /6 3/7 6 /6 3 /7 3 1 /5 30 /6

COLEOKE1A Aflat tprlctts 3 5 2 13 5 3 Apt at tssialle l Aplat crteete l Aflat Heines l 3 Aflat loti 1 t Aflat rim s l t 1 1 Andes stoairles 1 1 1 Andes nflpes 1 l 1 l 2 3 Mteeotnges plcrlnstis S 2 9 2 2 * SI ten Unites t l l l 5 5 6 3 14 n r n o m u Aielpiecoris llieeittes 2 1 Aerptlns i j q k i 2 1 l 5 Ijgts ngtllfault 1 Fkjteeerls wipes 2 3 2 4 6 3 10 l Fhgi apt ties trArstent l 1 4 6 7 FJtgi apt Lias d rjsu tie tl14 18 44 26 97 34 19 11 27 44 6 124 41 86 69 85 20 IOIOFTHA Ae/rtioslpkoi fine 6 25 2 4 2 15 l l 118 2 3 l 11 l l 8 Cenopls ml tent t l l 1 tepen riche 5 2 l l 4 5 l 1 4 l l 6 1 IS F IB O m U ii ioetet pntebt l i Intkarioe totUuts l : i THTSAIOfTBlA ! Aeletkrlps htenedies i 1 Fniklhlelli It taut J I t 1 2 13 1 22 I 3 l 6 1 10 7 Serlcatkrips dtartis l Reprinted from

THE JOURNAL OF ECOLOGY VOL. 75

BLACKWELL SCIENTIFIC PUBLICATIONS OXFORD LONDON EDINBURGH BOSTON PALO ALTO MELBOURNE Journal of Ecology (1987), 75, 1173-1189 THE EFFECT OF INSECT HERBIVORY ON THE GROWTH AND REPRODUCTION OF TWO ANNUAL VICIA SPECIES AT DIFFERENT STAGES IN PLANT SUCCESSION V. K. BROWN, A. C. GANGE, I. M. EVANS a n d A. L. STORR Imperial College at Sihvood Park, Ascot, Berks SL5 7PY

SUMMARY

(1) In natural plant communitiesVicia saliva is attacked mainly by chewing insects and V. hirsuta by sap-feeding insects. (2) The performance ofV. hirsuta and V. saliva was examined under natural and reduced levels of insect herbivory at three different points along a successional gradient. (3) Under natural levels of insect herbivory (controls),V. saliva produced more leaves and V. hirsuta fewer leaves than when herbivory was reduced. In both species, leaves from control plants had fewer leaflets, resulting in a reduction in total leaf biomass. (4) In V. saliva, herbivory caused a reduction in the number of seed pods per plant and the number of seeds per pod. No effect on individual seed weight was observed.V. In hirsuta herbivory had no effect on pod or seed number but caused a reduction in seed weight. (5) The effect of insect herbivory on the establishment and competitive ability of the two species in secondary succession is discussed.

INTRODUCTION It is generally accepted that herbivory plays an important role in the organization of plant communities (Harper 1977) as well as being a selective force in the evolution of plant secondary chemicals and plant morphology (Feeny 1970). The effects of vertebrate grazing on plant abundance are well documented in respect of mammals (Watt 1981; Gibson, Brown & Jepsen 1987; Gibsonet al. 1987), birds (Patton & Frame 1981) and reptiles (Merton, Bourn & Hnatiuk 1976). However, the impact of invertebrate grazing in natural plant communities has received far less attention (e.g. slugs, see Lutman 1978). Indeed Hairston, Smith & Slobodkin (1960) consider that natural communities are rarely seen to suffer insect damage. This latter view has been disputed by several workers (e.g. Brown 1982, 1985; Stinson 1983) who have shown, by manipulation experiments, that natural levels of insect herbivory can have substantial effects on species richness, plant cover and seedling establishment as well as an influence on the growth, survival and reproduction of individual species (e.g. Brown 1985). From these effects it has been suggested that insect herbivory can influence both the rate and direction of secondary succession. McBrien, Harmsen & Crowder (1983) found that the rate of succession was accelerated by outbreaks of a phytophagous beetle, while Brown (1982,1984) and Stinson (1983) showed that natural levels of herbivory can reduce the rate of early succession and that the direction of succession may be altered by the differential impact of herbivory on different plant growth forms (Gibson, Brown & Jepsen 1987; Gibsonet al. 1987). 1173 1174 Insect herbivory and Vicia performance Examples of the effects of insect herbivory on single plant species have generally been restricted to investigation of monophagous insects on perennials. Herbivory on such plants may reduce the number and size of seeds produced (Bentley, Whittaker & Malloch 1980; Kinsman & Platt 1984; Louda 1982; Waloff & Richards 1977), stimulate vegetative growth (McNaughton 1983; McNaughton, Wallace & Coughenour 1983) or reduce the ability of a species to compete (Rai & Tripathi 1985). It was suggested by Whittaker (1979), and shown by Lee & Bazzaz (1980) and Whittaker (1982), that even if invertebrate damage to a plant were not sufficient to affect performance, the impact could be considerable if it occurred at the same time as other adverse effects such as interspecific competition. In studies of this type, the damage done to plants by chewing insects may be obvious, but that done by sap-feeding species is difficult to quantify. Manipulation experiments, involving the application of insecticide, can be used to indicate the overall effect of herbivory, but do not quantify the effects of the separate guilds of species. Simulated herbivory by leaf clipping has produced similar results to natural defoliation (Rockwood 1973), but as demonstrated by Havlickova (1982), artificial defoliation may not mimic insect herbivory. This paper compares the performance of two annual, early successional plant species, Vicia hirsuta S.F. Gray (hairy tare) andV. sativa L. (common vetch) under natural and experimentally reduced levels of insect herbivory. In nature, different levels of plant competition and abundance of insect herbivores are seen at various stages of early succession (e.g. Brown & Southwood 1987). This paper reports the effects of herbivory on vegetative and reproductive characters of the plants in field sites of different successional age.

MATERIALS AND METHODS Vicia sativa and V. hirsuta are annual plants frequently found growing together in a range of different plant communities, from ruderal weed communities to established young fields (Southwood, Brown & Reader 1979). They are particularly amenable to field experimentation, because, despite being closely related, they show a number of differences in their strategies. V. hirsuta produces large numbers of leaves which have a rapid turnover and large numbers of pods each containing twoV. seeds. sativa produces fewer, longer-lived leaves and fewer seed pods which contain a variable number of seeds. At Silwood Park, Berkshire, a series of sites of known successional age has been developed by the natural colonisation of bare ground (see Southwood, Brown & Reader 1979). Sites have been created annually since 1977 and the vegetation and insect fauna closely monitored (Southwood, Brown & Reader 1979, 1983; Brown 1982, 1985). In addition, the insect fauna associated withVicia hirsuta and V. sativa was assessed by regular Univac suction sampling during 1984. Forty-five plants were sampled in mid-June and July on sites of one, four, six, eight and fourteen years of age. Each plant was sampled for 30 seconds. The sites used in this study, determined byVicia abundance, were those created in 1985 (first year site), 1984 (second year) and 1979 (seventh year). Each site was 405 m2 and within it twenty 3 m x 3 m plots were defined. Each plot was separated from its neighbour by 3 m (second and seventh year sites) or 2 m (first year site) and systematically allocated to a treatment: ten plots were sprayed with 45 ml water (‘control’) and ten with 45 ml of Malathion-60 solution (‘insecticide-treated’). The 45 ml of insecticide contained 1 • 134 ml active ingredient, conforming to the standard agricultural rate of 1 -26 kg a.i. ha~1 (Martin V. K. Brown et al. 1175 T able 1. Relative abundance of herbivorous insects associatedV. hirsuta with and V. sativa as assessed by ‘Univac’ sampling. Values are total numbers taken from forty-five plants in each of five sites in mid June and July 1984 at Silwood Park, Ascot, Berks. Feeding site V. hirsuta V. sativa Chewing insects Coleoptera (adults) Apion apricans Herbst flowers 10 5 Apion craccae L. fruits 1 Apion dichroum Bedel flowers 1 3 Apion loti Kirby flowers 2 Apion virens Herbst flowers 5 Bruchus atomaria L. fruits 2 Bruchus rufipes Herbst fruits 1 8 H yper a spp. fruits 2 Miccotrogus picirostis Fabricius fruits 18 Sitona lineatus L. leaves 3 24 Lepidoptera (larvae) Noclita pronuba L. leaves 1 Xanthorhoe monlanata (Schifiermueller) leaves 2 Total leaf-feeding insects 3 27 Sap-feeding insects Heteroptera (nymphs and adults) Adelphocoris lineolatus (Goeze) flowers 3 Berytinus minor (Herrich-Schaeffer) leaves 3 5 Phylocoris varipes Boheman leaves 11 10 Plagiognathus arbustorum (Fabricius) flowers 6 Plagiognathus chrysanthemi (Wolff) flowers 261 222 Homoptera (nymphs and adults) Acyrthosiphon pisum (Harris) leaves/stems 185 25 Cercopis vulnerata Illiger leaves 1 1 Megoura viciae (Buckton) leaves 18 11 Mvzus ornatus Laing leaves 4 Thysanoptera (nymphs and adults) Aeolothrips intermedius Bagnall flowers 1 Frankliniella intonsa (Trybom) flowers 37 22 Total leaf and stem-feeding insects 222 52

& Worthing 1976). Applications were made at ten-day intervals with a hand-held ultra- low volume sprayer (the ‘Micron ulva’, Micron Sprayers, Bromyard). Spraying took place in calm conditions at dawn in order to prevent drift. Treatment commenced in late April and continued until October, when allVicia plants were dead. Manipulation experiments, which reduce the natural levels of herbivory by the application of insecticide, incorporate the assumption that the insecticide has no direct effect on the vegetation (but see Allen & Casida 1951; Malone 1969; Shure 1971). A test of this assumption must therefore be an integral part of any such study of herbivory. Details of tests undertaken on Malathion-60, a non-persistent contact insecticide commonly used in such studies, are given elsewhere (Brown, Leijn & Stinson 1987). In summary, applications of the compound at the above dose and frequency were found to have no significant effects on a wide range of early successional species. Tests were undertaken in the field on natural plant communities and on selected plant species under controlled conditions. In field tests, above and below-ground biomass were recorded for dominant plant species and groups (e.g. annual herbs, perennial herbs, grasses), while under controlled conditions plant height, leaf number, number of reproductive structures, and the biomass of vegetative and reproductive structures for individual species were recorded. 1176 Insect herbivory and Vicia performance The sites were protected from rabbits by fencing. Small mammals were present on the sites, but the damage they caused toVicia plants was distinctive (stems being severed at the base) and not common. Although mollusc numbers were generally very low on the sandy soil in early succession, pellets of a molluscicide (‘Mifa slug’) were applied around each plant on each sampling occasion. When present five plants ofV. hirusta and V. sativa were marked (by plastic labels around the base of the stem) in each of the twenty plots. Details of replication are given in Table 2. Monitoring of plant performance occurred at intervals of approximately two weeks from late May throughout the growing season. On each occasion the following were recorded: plant height, total number of leaves, number of leaves damaged by insects and number of reproductive structures. Height and leaf number of the two species in different sites under the two treatments were analysed according to dates by a split-plot analysis of variance. Total flower numbers were analysed by a three-way analysis of variance. On each sampling occasion the youngest expanding leaf was marked by a piece of coloured cotton around the petiole and its longevity (in days), and extent of damage recorded subsequently. T urnover of leaves was assessed by calculating life expectancy (see Southwood 1978) using the number of leaves which were alive on each sampling occasion, irrespective of damage. Damage was recorded independently on the following scale: Damage rating Estimated leaf area removed (%) 0 0 1 1-5 2 6-25 3 26-50 4 51-75 5 76-99 6 Total removal (only petiole remaining) At regular intervals throughout the season 100 leaves of each species were collected at random from plants in each treatment on each site and damage recorded on the above scale. The number of leaflets forming each leaf was recorded and leaves were oven-dried for biomass measurements. Leaflet numbers were analysed by factorial chi-square and biomass by three-way analysis of variance. Leaf areas were not measured; the shape of Vicia leaves precludes simple measurement in the field. In mid-August a sample of 100 mature pods of each species was collected at random from each treatment on each site. The length of each pod was measured and the number of seeds recorded. Seeds were oven dried and weighed individually. Regression relationships were obtained between pod length and number of seeds per pod and pod length and total seed weight. Also in mid-August, the lengths of twenty mature pods on each marked plant were recorded and the number of seeds per pod and total seed weight per pod estimated from the regressions. The relationship between the maximum leaf number on a plant and reproductive capacity was also investigated. In late April of the following year (1986) the germination of seedlings in an area 500 cm2 around each of the marked plants was recorded. This was repeated in late May when the distinction was made between seedlings and established plants. V. K. Bro w n e t al. 1177

RESULTS Insect herbivores associated with V ic ia sp p . Several insect herbivore species were found on both host plants, but amongst the leaf and stem feeders there was an overall trend for sap-feeding insects to be m ore comm on on Vicia hirsuta and chewing insects on V . s a tiv a (Table 1).

Effects o f insect herbivory on vegetative characteristics B o th V ic ia species reached their maximum height in July. There was no significant difference between the height of V . sa tiv a plants in control and insecticide-treated plots, although insecticide-treated plants of V. h ir s u ta were significantly taller in the 2- and 7- year sites ( F = 1 0 -5 , P c O - O l) . The mean total num ber of leaves per plant at the height of the growing season (early- mid July) showed a consistent inter-specific difference (Table 2). In V . sa tiv a there were more leaves produced on control plants, while the converse occurred in V. hirsuta. T h e num ber of leaflets per leaf was consistently higher in insecticide-treated plants of V. s a tiv a (X 2 = 1 0 3 -4 , P < 0 001) and V . h ir s u ta ( / 2 = 1 1 H , P < 0 001) and there was also a more pronounced seasonal peak (Table 3). As expected, this trend was reflected in leaf biomass (F ig . 1) (V . s a tiv a , F— 1 0 8 9 -9 , P < 0 -0 0 1 ; V . h ir s u ta, F = 6 5 3 -7 , P < 0-001). However, there

Vicia sativa

0)o V/cia hirsuta

F ig 1. The effect of insect herbivory on the leaf biomass of Vicia sativa and V. hirsuta in sites of different successional age at Silwood Park, Ascot, Berks. Values are means+ 1 S.E. Symbols: (• ) = insecticide-treated; ( 0) = Control (natural levels of herbivory). (a) and (d) 1-year site; (b) and (e) 2-year site; (c) and (f) 7-year site. 1178 Insect herbivory and Vicia performance

T able 2. Maximum number of leaves per plant (mean + 1 S.E) in two V icia sp p . at Silwood Park, Ascot, Berks. I: insecticide-treated, C: control (natural levels of insect herbivory), n: number of marked plants. All comparisons signficant at PcO -O O l. I-year site 2-year site 7-year site I C I C I C V. saliva 25-10±2-65 45-94±3-87 14-60 + 0-74 20-36+1-10 9-98±0-55 16-71 ±0-91 (/i = 20) (n=17) pi = 20) (/i = 25) (« = 50) (« = 50) V. hirsute 83-40±4-06 52-12±4-58 78-72±4-02 51-84 + 3-19 71-00±4-01 46-04 + 4-06 (/, = 40) (/i = 45) (// = 50) (u = 50) (« = 48) (i; = 50)

were no significant differences in the biomass of individual leaflets or petioles. Thus total leaf biomass increased on insecticide-treated plants solely as a result of each leaf having m ore leaflets.

Leaf persistence and damage by chewing insects In b o th V ic ia species control plants started to lose their leaves and die before insecticide-treated plants. By monitoring the fate of individual leaves produced at different times during the season, their life expectancy (irrespective of damage) was determined (Table 4). Leaves of V. sa tiv a were generally longer lived than those of V. h irsu ta . In both species the turnover rate of leaves of control plants was greater than that of insecticide-treated plants. Differences observed were consistent between sites. Table 4 also lists the percentage reduction in leaf-life expectancy between the two treatm ents as a measure of a ‘herbivory effect’. This effect was greatest at the beginning of the season in the 7-year site, while on the youngest site the peak was in mid-season. Figure 2 shows the relationship between the life expectancy of a Vicia sativa le a f a t th e 1-year site and its final damage rating. Early season leaves, with a relatively high life expectancy (i.e. low turnover rate), suffered less damage than did later produced leaves. Other relationships were also calculated as follows:

V. sa tiv a , 2-yr site: y = — 3 -3 8 jc H- 4 *8 0 , r = —0-620, d.f. = 98, P < 0 001 V. sa tiv a , 7-yr site: j’ = — 2-66.X-I-4-04, r — —0-672, d.f. = 148, P < 0-001 V. hirsuta, 1-yr site: y = — 1-9 8 .Y-F3 -0 2 , r = —0-640, d.f. = 198, P < 0-001 V. hirsuta, 2-yr site: y = — 2-1 l x + 2 -9 8 , r = — 0 -5 8 6 , d .f. = 2 4 8 , P < 0-001 V. hirsuta, 7-yr site: y = — 1-6 7 .X + 3 -4 4 , r — — 0 -5 9 1 , d .f .= 1 9 8 , P c O -O O l The proportion of the leaves of a control plant that were damaged (Fig. 3) was generally high but varied during the season and between the species. Overall levels of damage were h ig h e r in V. s a tiv a especially towards the end of the season and were greatest on the 1 -year site. This can be related to the predominance of chewing insects on this species (Table 1) which are particularly abundant later in the season. To demonstrate the efficacy of the insecticide treatm ent, the proportion of leaves damaged on insecticide-treated plants is also given in Fig. 3. The proportion was generally less than 5%; only in V. s a tiv a on the 1- year site did the level exceed 8% on one sampling occasion. T a ble 3. The effect of reducing levels of insect herbivory on leaflet number per leaf (mean ± 1 S.E.) in Vicia spp., analysed by factorial chi- square. I = Insecticide-treated plants, C = Control (natural levels of insect herbivory), *** = P < 0-001 Mean leaflet number per leaf 1-year site 2-year site 7-year site Date 1 C 1 C I C Effect d.f. X2

(a) V. saliva B . K . V 1/6 14-61 ±0 16 12-32 + 0-13 11-04 ±0-18 10-74 ±0-12 8-8 ±0-18 8-38 ±0-12 Sites 14 540-26*** Treatments 7 103-39*** 16/6 16-81 +0-11 13-33 + 0-12 11-71 ±0-19 9-87 ±0-11 9 73±0 16 8-41 ±0-14 Dates 28 281-83*** 28/6 16-33 + 0-14 13-30 ±0-11 11 05 ±0-16 9-84±0-10 9-76±0-10 8-61 ±0-09 Sites x treatments 14 67-37*** n w o r Sites x dates 56 292-75*** 29/7 15-00 ±0-13 12-12 ± 0-12 9-82 ±0-13 8-91 ±0-09 8-89 ±0-11 7-36 ±0-13 Treatments x dates 28 113-32***

16/8 10-11 ±0-13 9-06 ±0-11 9-64 ±0-11 7-32 ±0-09 8-03 ±0-09 7-11 ±0-09 Sites x treatments x dates 56 134-43*** l. a t e Total 203 1533-35

(b) V. hirsuta 9 7 1 1 1/6 12-91 ±0-13 11 -88 ± 0-10 14-18 ± 0-16 13-79 ±0-12 13-63 ±0-13 12-97 ±0-12 Sites 16 159-86*** Treatments 8 111-13*** 16/6 13-11+0-13 11 -89 ±0-12 16-26 + 0-17 14-21 +0-12 14-04 + 0-12 13-06 + 0-14 Dates 32 176-82*** 28/6 16-41 ±0-14 12-34 ±0-15 17-27 ±0-19 14-25 ±0-11 14-41 ±0-13 12-66 ±0-13 Sites x treatments 16 85-74*** Sites x dates 64 196-38*** 29/7 1500±0-14 11-94 ±0-16 15-41 ±0-13 13-06 ± 0* 13 12-66 ± 0-12 10-18 ±0-11 Treatments x dates 32 126-23*** 16/8 11-61 +0-11 11 -14 ±0-11 14-76±0-13 12-98 ±0-12 11 -29 ±0-12 8-85±0-10 Sites x treatments x dates 64 167-54*** Total 232 1023-70 1180 Insect herbivory and Vicia performance

T a b le 4. Life expectancy of leaves produced by two V icia spp. at Silwood Park, Ascot, Berks, throughout the season and herbivory effect expressed as the percentage reduction in leaf life expectancy between the two treatments. Values at the end of the season are given in parenthesis since sample size is small. I: insecticide-treated, C: control (natural levels of insect herbivory). 1-year site 2-year site 3-year site Herbivory effect Herbivory effect Herbivory effect Sample I C (%) I C (%) IC (%)

V. sativa 1. May 27 3-60 2-38 33-89 2 14 1-12 47-66 2-00 0-98 51 2. June 20 2-80 1-74 37-86 1 70 1-06 37-65 1-40 0-82 414 3. July 10 200 1-03 48-50 1-01 0-94 6-93 0-60 0-50 10 4. July 26 1-35 0-55 59-26 0-54 0-50 7-4 0-50 0 (100) 5. Aug. 13 0-70 0-50 28-75 0-50 0 ( 100) 0 0 6. Aug. 30 0-55 0 ( 100) 0 0 0 0 7. Sept. 11. 0-50 — (100) ———— 8. Sept. 25. 0 —— ——— V. hirsuta 1. May 27 1-75 1-28 26-86 1-86 1-06 43-01 1-78 0-92 48-31 2. June 20 2-20 1-43 35 2-13 1-18 44-60 1 82 0-94 48-35 3. July 10 1-48 0-85 42-5 1-05 0-84 20-00 0-84 0-54 35-71 4. July 26 0-78 0-54 30-77 0-57 0-54 5-26 0-57 0-50 12-28 5. Aug. 13 0-57 0-50 12-28 0-50 0-50 (0) 0-50 0 (100) 6. Aug. 30 0-50 0 ( 100) 0 0 0 — 7. Sept. 11 0 —————

Fig 2. The relationship between the insect damage sustained by leaves of control plants of Vicia saliva and their life expectancy at the 1 year site at Silwood Park, Ascot, Berks. Figures indicate numbers of observations. Fitted line: y = — 1-30.x + 3-98, r= —0-711, d.f. =66 , P<0-001.

Effects of insect herbivory on reproductive characteristics In b o th V ic ia species, significantly more flowers were produced on the insecticide- treated plants (Fig. 4). To avoid destructive sampling of marked plants, pods were collected from adjacent plants. Regression relationships were obtained between pod length and seed num ber per pod in V. s a tiv a (the m ajority of pods of V. h irsu ta contained only two seeds) and between pod length and total seed weight per pod in both species. The regressions were significant at P < 0 001. In V . s a tiv a , insecticide-treated plants produced V. K. Bro w n e t a l. 1181

Vicia sativa IOOr (a) I00r (b) 100r (O a R A / V 50 K 50 5 0 -

___[___L M J J A S M J J A S J J A S

B Vicia hirsuta £ lOOr- (d) lOOr (e) 1001— (f) 9 /ft \ 50 / \ 50 5 0 - b /l \ A —<^'6 o-6i qi —f *"«=♦ _!__ M J J A S M J J A S 0 M J J A __ S Month

F ig 3. Percentage of leaves of Vicia sativa and V. hirsuta damaged by insect herbivory in sites of different successional age at Silwood Park, Ascot, Berks. Values are means (arcsine square root)+l S.E. Symbols: ( • ) = insecticide-treated; (O) = control (natural levels of herbivory). (a) and (d) 1-year site; (b) and (e) 2-year site; (c) and (f) 7-year site. more and longer pods than control plants and although these contained significantly m ore seeds, the individual seed weight did not vary. In V. hirsuta, however, there were no differences in pod size or seed number per pod, but individual seed weight was significantly higher in pods from insecticide-treated plants (Table 5). The differences between the reproductive characteristics of the two species under the two treatm ents may be a simple effect of plant size. Thus, for all m arked plants, total pod num ber and total seed weight per pod, and for V . s a tiv a seed num ber per pod were plotted against the maximum num ber of leaves produced during the season (as a measure of plant size). Estimates of seed weight and number were obtained from the regression relationships. Covariance analyses (Zar 1974) were performed upon the regressions obtained (Table 6). In both species, larger plants produced more pods. In comparisons between the relationships of leaf and pod num ber in V . sa tiv a there were no differences between the slopes of the lines but there were between the elevations, indicating that in this species plants of comparable size within a site produce more pods when treated with insecticide. In V . h ir s u ta there were no differences between slopes or elevations, suggesting that treatm ent with insecticide does not affect pod number. However, these data had a very high variance. There were no significant relationships between leaf num ber and total seed weight per pod or seed number per pod in V. sa tiv a , nor between leaf number and total seed weight 1182 Insect herbivory and Vicia performance

Vicia sativa I6 |— (b)

—jM --- J Lb S

O Vicia hirsuta 7 0 r (f)

Month F ig 4. The effect of insect herbivory on the number of flowers on plants of Vicia sativa and V. hirsuta in sites of different successional age at Silwood Park, Ascot, Berks. Values are means ±1 S.E. Symbols: ( • ) = insecticide-treated; ( 0) = Control (natural levels of herbivory). (a) and (d) 1-year site; (b) and (e) 2-year site; (c) and (f) 7-year site.

per pod in V. hirsuta. Therefore, the number and size of seeds produced does not vary according to the size of the plant and any differences within a site m ust be a direct effect of herbivory. In V . sa tiv a there were no differences between the slopes of these lines, but there were between the elevations, indicating that within a site plants of comparable size produced pods containing more seeds when treated with insecticide so that the total seed weight was also increased. In V . h ir s u ta the difference in elevations of the relationships between leaf num ber and seed weight indicate that within a site plants of comparable size produce a greater total seed weight per pod when treated with insecticide. This m ust be the result of an increase in individual seed weight, since seed num ber per pod seldom varies. The results of these analyses therefore validate the conclusions reached in Table 5. Pods of both species were examined for seed predators. Samples of 100 mature pods collected from both treatments revealed that those of V . sa tiv a on control plants contained a significantly higher proportion of damaged seeds (1-year site: x 2 = 9 -5 4 , .PcO-Ol; 2-year: %2 = 16-28, PcO-OOl; 7-year: x2 = 33-21, .PcO-OOl). Seeds were mainly damaged by larvae of the pea m oth Cydia nigricana (Fabr.). Seeds of V . h ir s u ta re m a in e d virtually undamaged in control plants and there were no differences between treatments (1-year site: x2= 139, P >0 05; 2-year: x2= T 44, P > 0-05; 7-year: x2= l-54, P > 0-05). The germination of seeds around each plant was monitored the following spring. The num ber of seedlings and established plants is given in Table 7 and is used as a relative measure of seed viability in the field. In each site, the num ber of seeds germinating was T a ble 5. Measured and estimated (*) reproductive variables of two Vicia spp. at Silwood Park, Ascot, Berks. I: insecticide-treated, C: control (natural levels of insect herbivory). 1-year site 2-year site 7--year site I C I C I C B . K . V jc±S .E . x + S .E . P x + S.E. x + S.E. P x + S.E. x ± S .E . P

V. sativa Number of pods per plant 43-4 0 + 5-40 25-0 + 3-25 0-001 17-4 4 + 1-10 12-84± 1-00 0-01 10-68 ± 1-07 6-25 + 0-70 0-001 n w o r Pod length (mm) 55-53 + 0-41 52-41+0-45 0-001 52-38 ± 0-43 49-20 + 0-56 0 001 40-14 + 0-45 37-54 + 0-55 0-001

•Number of seeds per pod 10-6 + 0-19 9-8 + 0-24 0-05 8-3 + 0-22 7-4 + 0-26 0-01 8-4 + 0-24 7-1+ 0-26 0-001 •Individual seed weight (mg) 15-8 + 0-10 15-3 + 0-24 0-05 12-1+ 0-19 12-1+0-09 0-05 7-7 + 0-12 7-9 + 0-13 0-05 l. a t e

V. hirsuta 1183 Number of pods per plant 117-49+ 14-83 96-05 ± 21-66 0-05 69-5 ± 9-96 48-24 ± 7-28 0-05 39-56 + 6-47 26-88 + 7-63 0-05 Pod length (mm) 11-18 + 0-12 10-95 + 0-12 0-05 10-51 ± 0-09 10-54 ± 0-15 0-05 9-23 + 0-07 9-03 + 0-10 0-05 •Number of seeds per pod 1-9 + 0-02 1-9 + 0-03 0-05 1-9 ± 0-03 1-8 + 0-04 0-05 1-8 + 0-04 1-8 + 0-04 0-05 •Individual seed weight (mg) 9-3 + 0-09 8-2 + 0-13 0-001 7-5 + 0-10 6-4 + 0-10 0-001 5-2 + 0-04 4-8+0-04 0-001 1184 Insect herbivory and Vicia performance

T able 6. The effect of insect herbivory on the relationships between plant size and reproductive capacity in two Vicia spp. at Silwood Park, Ascot, ***:/>< 0-001; N.S.: not significant at />=0 05. Insecticide-treated Control slopesDifference betweenintercepts Regression Relationship r Relationship r t = / = (a) V. sativa Total1-year leaf site number (a): total pod number (y) v = 0- 59 a +14-45 0-699*** v = 0-39.v +8-32 0-707*** 1-206 N.S. 4-430*** r. total seed weight per pod (y) y = 004a + 130-86 r. seed content per pod O') y = 0-001.v + 8-88 0-1480-155 N.S. N.S. yv = = 0-015.v+5-080-14.v-+82-07 0-1960-104 N.S. N.S. 0-7660-330 N.S. 6-489***4-984*** Total2-year leaf site number (a): total pod number O') v a v a v. total seed weight per pod O’) i- = 0-620-l 5v+ + 7-29110-5 0-601***0-203 N.S. f = 0-650-23.v +80-84+0-46 0-1350-880*** N.S. 0-2230- N.S. 1166-609***6-007*** N.S. v. seed content per pod O') y = 0 01 a + 7-33 0-096 N.S. y = 0-12.v+ 2-99 0-115 N.S. 1- 793 N.S. 6-224*** Total7-year leaf site number (a): total pod number O’) y = 0-67.v+ 7-88 0-789*** v = 0-68.v +0-72 0-895*** 0-003 N.S. 6-291*** v. total seed weight per pod O') v = 2-1 6a + 79-63 0-255 N.S. y= 202a+55-65 0-219 N.S. 0-161 N.S. 3-920*** v.r. seed content per pod O') y = 0-03a+ 8-26 0-255 N.S. y = 0 07a+ 6-65 0-268 N.S. 0-804 N.S. 3-809*** (b) V. hirsuta Total1-year leaf site number (a): total pod number O') v = 0-93.v+79-7l 0-853*** )•= 1-37.V +29-07 0-753*** 1-987 N.S. 1-125 N.S. v. total seed weight per pod O') y = 0 005a + 17-57 0-198 N.S. y = 0 002a + 13-31 0-103 N.S. 1-725 N.S. 26-13*** Total2-year leaf site number (a): v. totaltotal seedpod number weight perO') pod O') yv = = 0-00Lv+14-19 0-84a+6-37 0-0440-678*** N.S. y = 001.v0-96.v + + 4-36 8-06 0-2460-876*** N.S. 0-1- 292 N.S. 21-02*** 510 N.S. 1 -474 N.S. Total7-year leaf site number (a): total pod number O') v=l-22.v+21-23 0-578*** v= 1-27a+ 33-984 0-807*** 0-105 N.S. 2-00 N.S. v. total seed weight per pod O') y = 0005a+ 9-17 0-347 N.S. y = 0009 a + 6-261 0-360 N.S. 0-936 N.S. 14-90*** small. However, a reduction in herbivory in V. s a tiv a resulted in an increase in the num ber of established plants (2-5, 3-0 and 9-0 times as many in one-, two-, and seven-year sites respectively). In V. h ir s u ta the num ber of established plants showed the same trend (2-2, 9-7 and 11-0 times as many in the one-, two-, and seven-year sites respectively).

Successional trends in V ic ia sp e c ie s A number of successional trends is apparent in the vegetative and reproductive characters of the two species. All plants in the one-year site germinated in the spring, whilst many of those in the older sites germinated the previous autum n and were thus more advanced phenologically at the start of the experiment. Plants of both species achieved their maximum height earlier in the older sites and flowered earlier. There was a strong successional trend in the maximum num ber of leaves per plant with more leaves being produced in the youngest site ( V. sativa: F= 11-6, P < 0 -0 0 1 ; V. hirsuta: F = 1 7 1 , P <0 001). A similar trend was seen in individual leaf biomass (V. sativa: F = 8 9 -6 , P < 0 -0 0 1 ; V. hirsuta: F= 22-4, P < 0-001) with leaves from the youngest site being largest. This was also reflected in the leaflet number (Table 3). The number of flowers produced decreased with successional age (V. sativa: / r = 1 5 -3 , P < 0 -0 0 1 ; V. hirsuta: F= 13-2, P < 0-001) as did the num ber of seed pods produced (Table 6). There were also successional patterns in pod characteristics, e.g. a decrease in pod length (V . s a tiv a , F = 3 7 1 -0 , P < 0 -0 0 1 ; V. hirsuta: F= 103-6, P < 0-001) and in individual seed T a ble 7. Seed germination and plant establishment of Vicia spp. under natural and reduced levels of insect herbivory. Values are mean number of plants in 500 cm2 around masked plants + 1 S.E. I: insecticide-treated, C: control, S: seedling, E: established plant. B . K . V 1-year site 2-year site 7-year site I C I C I c V. saliva SESE SESESESE n w o r

29/4/8630/5/86 1-40+0 0-31 0-50 ±006 0-80 ±0040 0-20 ±0-001 0-700-06 ±0-03±0-20 0-60±0-16 0-36±0-140 0-20 ±0-08 0-540-04 ±0-002±0-12 0-72±0-12 0-10±0-050-08 ±0-004 0-08 ±0-004 l. a t e

V. hirsuta 1185 30/5/8629/4/86 010±00011-50 ±0-41 l-30±0-50 200 +0 0-44 0-60 ±0-008 0-163-44 ±0-89±0-008 2-90 ±0-52 0-020-76±0-19 ±0-002 0-30±0-l 0-141 -14 ±±0-06 0-25 1 -54 ± 0-27 0-12±0-050-30 ±0-09 0-14±0-05 1186 Insect herbivory and Vicia performance w e ig h t ( V . sa tiv a , F = 1 6 3 *4 , P c 0 -0 0 1 ; V. hirsuta: F= 62-4, PcO-OOl) with increasing successional age.

DISCUSSION Insect herbivory has a considerable effect upon a number of characteristics in both th e s e V ic ia species. The guild structure of the insect herbivores associated with the two species differs: insects feeding on the leaves of V . s a tiv a are mainly chewers, i.e. leaf removers, while sap feeders predom inate on V. hirsuta. This m ay explain the differences in the response of vegetative structures of the two plant species to herbivory. In V . s a tiv a , leaf production appears to be stimulated by herbivory. This effect was most m arked on the one-year site, where the weevil, Sitona lineatus, was particularly abundant. Compensatory leaf growth as a response to leaf removal is not uncommon and may even increase plant fitness if the defoliation is moderate (M cNaughton 1983). On the other hand, the reduction in leaf num ber in V . h irsu ta by herbivory may be a reflection of the effect of sap removal. Sap-feeding insects have been shown to reduce leaf area in sycamore (Dixon 1971) and to affect tree growth (W hittaker & W arrington 1985). Leaf life expectancy (irrespective of the damage the leaf receives) is reduced by natural levels of herbivory. This ‘herbivory effect’ is highest early in the growing season in the 2- and 7-year sites and in mid season in the 1-year site. This difference may be related to the insect fauna associated with the sites at particular times of the year. At the start of the growing season there was an established fauna present upon the older sites, many species of which would have overwintered. However, on the youngest site very few insect species would have colonized the site and their effect would be relatively small. By mid-season more species would have invaded (Southwood, Brown & Reader 1979) and many of these would be generalist feeders (Brown 1986). The chemical defences of the two V ic ia sp e c ie s are currently being explored since, although both species have a relatively fast leaf turnover (see Southwood, Brown & Reader 1986), the fewer longer-lived leaves of V. s a tiv a may indicate a higher level of defence than that occurring in V . h ir s u ta (F e e n y 1976). In addition, leaves of V . s a tiv a produced early in the season are less likely to be damaged than later-produced leaves, suggesting a degree of protection. Differences in reproductive strategies between the species are substantial. Flower num ber was considerably reduced by herbivory in both species on all sites and in V. s a tiv a led to a significant reduction in the num ber of pods produced per plant. Pod num bers were very variable in V. h irsu ta but the results suggested a reduction in this species also. Seed size is not invariable (Harper 1977) and has been found to be indirectly reduced by defoliation, whether artificial (M aun & Cavers 1971) or by insects (Bentley, W hittaker & Malloch 1980; Kinsman & Platt 1984). In V . s a tiv a the response to a reduction in herbivory was to produce more seeds, while keeping size constant. In V . h irsu ta th e se e d num ber is more rigidly fixed (two) and grazed plants produced smaller seeds. It is known that smaller seeds generally germinate as well as and often more rapidly than large ones (Bentley, W hittaker & Malloch 1980; Maun & Cavers 1971). However, seedling establishment from small seeds may be seriously impaired (Parker & Salzman 1985). Herbivory has been demonstrated to have a deleterious effect on plant establishment through seed quality and quantity in a number of cases (Bentley, W hittaker & M alloch 1980; Kinsman & Platt 1984; Louda 1982; Parker & Salzman 1985; Stamp 1984). In this study seed germination and plant establishment were generally enhanced by a reduction in herbivory. A lowering of seedling success by herbivory may well be one of the m ajor V. K. Bro w n e t a l. 1187 factors which modifies the direction and rate of succession (e.g. Brown 1982; Gibson, Brown & Jepsen 1987; Gibson e t a l. 1 9 8 7 ). Many authors emphasize the importance of herbivory when linked with plant competition (e.g. Lee & Bazzaz 1980; Parker & Salzman 1985; W hittaker 1982; W indle & Franz 1979). Indeed, W hittaker (1979) suggested that the outcome of plant competition may only be explicable in term s of the invertebrate grazing occurring at the same time, and demonstrated such a case with Rumex crispus L ., R. obtusifolius L. and the beetle Gastrophysa viridula De Geer (Bentley & W hittaker 1979; W hittaker 1982). In the current study, when herbivory was reduced plant growth and reproduction was greatest on the youngest site and decreased with successional age. Indeed, the magnitude of the differences in vegetative and reproductive characteristics of the two species was greatest on the youngest site, indicating that the effect of insect herbivory is more pronounced early in succession. This agrees with the earlier findings of Brown (1982). As succession proceeds there are changes in plant taxonomic diversity and architecture (Southwood, Brown & Reader 1979) and competition experienced by annuals, such as the two V ic ia species, is likely to become m ore severe. It remains to be seen how herbivory m ay affect the performance of the two V ic ia species when interspecific and intraspecific com petition are experimentally m anipulated. This is the subject of current work at Silwood Park.

ACKNOWLEDGMENTS We are grateful to Charlotte Ford, Deborah Proctor and Janet Pryse for assistance with fieldwork, to Dr C. W. D. Gibson for guidance with the statistical analysis and to Dr P. Hyman and Dr P. Kirby for information on the feeding sites of Coleoptera and Heteroptera, respectively, and to the Natural Environment Research Council for financial support.

REFERENCES

Allen, T. C. & Casida, J. E. (1951). Criteria for evaluating insecticidal phytoxicity-aerial growth. Jo u rn a l o f Economic Entomology, 44, 737-740. Bentley, S. & Whittaker, J. B. (1979). Effects of grazing by chrysomelid beetle, Gastrophysa viridula, on competition between Rumex obtusifolius and R. crispus. Journal o f Ecology, 67, 79-90. Bentley, S., Whittaker, J. B. & Malloch, A. J. C. (1980). Field experiments on the effect of grazing by a chrysomelid beetle ( Gastrophysa viridula) on seed production and quality in Rumex obtusifolius and R um ex crispus. Journal of Ecology, 68, 671-674. Brown, V. K. (1982). The phytophagous insect community and its impact on early successional habitats. Proceedings o f the 5th International Symposium on Insect Plant Relationships, Wageningen, 1982 (Ed by J. H. Visser & A. K. Minks), pp. 205-213. Pudoc, Wageningen. Brown, V. K. (1984). Secondary succession: insect-plant relationships. BioScience, 34, 710-716. Brown, V. K. (1985). Insect herbivores and plant succession. O ikos, 44, 17-22. Brown, V. K. (1986). Life cycle strategies and plant succession. The Evolution of Insect Life Cycles (Ed by F. Taylor & R. Kasban), pp. 105-124. Springer Verlag, Berlin. Brown, V. K., Leijn, M. & Stinson, C. S. A. (1987). The experimental manipulation of herbivore load by the use of an insecticide (malathion): the effect of application on plant growth. Oecologia (Berlin), 72, 377-381. Brown, V. K. & Southwood, T. R. E. (1987). Secondary succession: patterns and strategies. In Colonization, Succession and Stability (Ed by A. J. Gray, M. J. Crawley & P. J. Edwards), pp. 315-338. Symposia of the British Ecological Society, 26. Blackwell Scientific Publications, Oxford. Dixon, A. F. G. (1971). The role of aphids in wood formation 1. The effect of the sycamore aphid Drepanosphum platanoidis (Schr.) (Aphididae) on the growth of sycamore; Acer pseudoplatanus (L.). Journal of Applied Eco lo g y, 8, 165-179. Feeny, P. P. (1970). Seasonal changes in oak leaf tannins and nutrients as a cause of spring feeding by winter moth caterpillars. Eco lo g y, 51, 565-581. Feeny, P. P. (1976). Plant apparency and chemical defense. Recent Advances in Phytochemistry, 10, 1-40. 1188 Insect herbivory and Vicia performance

Gibson, C. W. D., Brown, V. K. & Jepsen, M . (1987). Relationships between the effects of insect herbivory and sheep grazing on seasonal changes in an early successional plant community. Oecologia (Berlin), 71,245- 253. Gibson, C. W. D., Dawkins, H. C., Brown, V. K.& Jepsen, M. (1987). Spring grazing by sheep: effects on seasonal change during early Old Field succession. Vegetatio, 70, 33-43. Hairston, N. G., Smith, F. E. & Slobodkin, L. B. (1960). Community structure, population control and competition. American Naturalist, 94, 421-425. Harper, J. L. (1977). Population Biology o f Plants. Academic Press, London. Havlickova, H. (1982). Different responses of alfalfa plants to artificial defoliation and to feeding by pea leaf weevil (Sitona lineatus). Experientia (Basel), 38, 569-570. Kinsman, S. & Platt, W. J. (1984). The impact of a herbivore upon Mirabilis hirsuta, a fugitive prairie plant Oecologia (Berlin), 65, 2-6. Lee, T. D. & Bazzaz, F. A. (1980). Effect of defoliation and competition on growth and reproduction in the annual plant Abutilon theophrasti. Journal o f Ecology, 68, 813-822. Louda, S. M. (1982). Limitation of the recruitment of the shrub Haplopappus squarrosus (Asteraceae) by flower­ feeding and seed-feeding insects. Journal o f Ecology, 70, 43-54. Lutman, J. (1978). The role of slugs in anAgrostis-Festuca grassland. InProduction Ecology o f British Moors and Montane Grasslands (Ed by O. W. Heal & D. F. Perkins), pp. 332-348. Springer-Verlag, Berlin. McBrien, H., Harmsen, R. & Crowder, A. (1983). A case of insect grazing affecting plant succession. Eco lo g y, 64, 1035-1039. McNaughton, S. J. (1983). Compensatory plant growth as a response to herbivory.O ikos, 40, 329-336. McNaughton, S. J., Wallace, L. & Coughenour, M. B. (1983). Plant adaptation in an ecosystem context: effects of defoliation, nitrogen and water on growth of an African C 4 sedge. Eco lo g y, 64, 307-318. Malone, C. R. (1969). Effects of diazinon contamination on an old-field ecosystem. American Midland Naturalist, 82, 1-27. Martin, H. & Worthing, C. R. (1976). Insecticide and Fungicide Handbook. Blackwell Scientific Publications, Oxford. Maun, M. A. & Cavers, P. B. (1971). Seed production and dormancy in R u m ex crispus. 1. The effects of removal of cauline leaves at anthesis. Canadian Journal o f Botany, 49, 1123-1130. Merton, L. F. H., Bourn, D. M. & Hnatiuk, R. J. (1976). Giant tortoise and vegetation interactions on Aldabra atoll. Part 1: inland. Biological Conservation, 9, 293-316. Parker, M. A. & Salzman, A. G. (1985). Herbivore exclosure and competitor removal: effects on juvenile survivorship and growth in the shrubGulierrezia microcephala. Journal o f Ecology, 73, 903-913. Patton, D. L. H. & Frame, J. (1981). The effect of grazing in winter by wild geese on improved grassland in West Scotland. Journal of Applied Ecology, 18, 311-325. Rai, J. P. N. & Tripathi, R. S. (1985). Effect of herbivory by the slug M ariaella dussum ieri and certain insects on growth and competitive success of two sympatric annual weeds. Agriculture, Ecosystems and Environment, 13,125-138. Rockwood, L. L. (1973). The effect of defoliation on seed production of six Costa Rica tree species. Ecology, 54, 1363-1369. Shure, D. J. (1971). Insecticide effects on early succession in an old-field ecosystem. E co lo g y, 52, 271-279. Southwood, T. R. E. (1978). Ecological Methods. Methuen, London. Southwood, T. R. E., Brown, V. K. & Reader, P. M. (1979). The relationships of plant and insect diversities in succession. Biological Journal o f the Linnean Society, 12, 327-348. Southwood, T. R. E., Brown, V. K. & Reader, P. M. (1983). Continuity of vegetation in space and time: a comparison of insects’ habitat templet in different successional stages. Researches on Population Ecology. Supplement No. 3, 61-74. Southwood, T. R. E., Brown, V. K. & Reader, P. M. (1986). Leaf palatability, life expectancy and herbivore damage. O ecologia, 70, 544-548. Stamp, N. E. (1984). Effect of defoliation by checkerspot caterpillars ( Euphydryas phaeton) and larvae (Macrophya nigra and Tenthredo gradis) on their host plants (Chelone spp.). Oecologia (Berlin), 63, 275- 280. Stinson, C. S. A. (1983).Effects of insect herbivores on early successional habitats. Ph.D thesis, University of London. Waloff, N. & Richards, O. W. (1977). The effect of insect fauna on growth, mortality and natality of broom Sarothamnus scoparius. Journal o f Applied Ecology, 14, 787-798. Watt, A. S. (1981). A comparison of grazed and ungrazed grassland in East Anglian Breckland.Jo u rn a l o f Eco lo g y, 69, 499-508. Whittaker, J. B. (1979). Invertebrate grazing, competition and plant dynamics. Population Dynamics (Ed by R. M. Anderson, B. D. Turner & L. R. Taylor), Symposia of the British Ecological Society, 20. Blackwell Scientific Publications, Oxford. Whittaker, J. B. (1982). The effect of grazing by a chrysomelid beetle, Gastrophysa viridula, on growth and survival ofRumex crispus on a shingle bank. Journal of Ecology, 70, 291-296. V. K. Bro w n e t al. 1189 Whittaker, J. B. & Warrington, S. (1985). An experimental field study of different levels of insect herbivory induced by Form ica rufa predation on sycamore (Acer pseudoplatamis). III. Effects on tree growth. Jou rn al o f Applied Ecology, 22, 797-811. Windle, P. N. & Franz, E. H. (1979). Effects of insect parasitism on plant competition: greenbugs and barley. E co lo g y , 60, 521-529. Zar, J. H. (1974). Biostatistical Analysis. Prentice Hall, London.

(Received 12 January 1987)