USF Libraries
USF Digital Collections

Herbivory by leaf-miners on Florida scrub oaks

MISSING IMAGE

Material Information

Title:
Herbivory by leaf-miners on Florida scrub oaks
Physical Description:
Book
Language:
English
Creator:
Cornelissen, Tatiana
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Host quality
Enemies
Quercus laevis
Quercus geminata
Quercus myrtifolia
Dissertations, Academic -- Biology -- Doctoral -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: This study investigated effects of plant quality and natural enemies on the abundance and survivorship of several leaf miner species on Florida scrub oaks over several ecological scales. Three oak species (Quercus laevis, Q. geminata, and Q. myrtifolia) and four leafminer species (Acrocercops albinatella, Brachys tesselatus, Stilbosis quadripustulatus, and Cameraria sp. nova) were the main focus of five separate studies, addressing effects of bottom-up and top-down factors at regional, local, and individual scales. At the regional scale, it was observed that Cameraria sp. nova was aggregated into sites, and sites closer to each other exhibited similar densities of mines than sites farther apart. None of the bottom-up and top-down factors studied were spatially structured, but did influence the variation in Cameraria abundance over the range of the host plant Q. myrtifolia. At the local scale, all leaf miners studied were aggregated between and within plants, and variation^ in bottom-up factors among individual plants explained variation in abundance for some of the leaf miners studied. Intra-specific competition was identified as an important factor influencing mine survivorship, but inter-specific competition among leaf miners and gall-formers did not shape the community structure of oak herbivores. Experimental manipulation of bottom-up and top-down factors via fertilization and natural enemy removal showed that bottom-up effects were important determinants of leaf miner abundance, as fertilized plants supported 2 to 5-fold more herbivores than control plants. The removal of natural enemies, on the other hand, did not significantly impact the abundance and/or the survivorship of leaf miners and other guilds studied. At individual scales, it was demonstrated that two leaf miner species responded to random variations in leaf morphology, by increasing in abundance in individual host plants with more asymmetric leaves and/or higher levels of fluctuating as ymmetry. These results offered support for the plant stress hypothesis and differences in host plant quality were again partially responsible for the results found.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Tatiana Cornelissen.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 155 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001790119
oclc - 144457643
usfldc doi - E14-SFE0001486
usfldc handle - e14.1486
System ID:
SFS0025805:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Herbivory by Leaf Miners on Florida Scrub Oaks by Tatiana Cornelissen A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Biology College of Arts and Sciences University of South Florida Major Professor: Peter Stiling, Ph.D. Anthony Rossi, Ph.D. Henry Mushinsky, Ph.D. Susan Bell, Ph.D. Date of Approval: April 3, 2006 Keywords: Host quality, Enemies, Quercus laevis, Quercus geminata Quercus myrtifolia Copyright 2006, Tatiana Cornelissen

PAGE 2

DEDICATION “Eu no sou daqui tambm marinheiro Mas eu venho de longe E ainda do lado de trs da terra alm da misso cumprida Vim s dar despedida Filho de sol poente Quando teima em passear desce de sal nos olhos doente da falta de voltar Filho de sol poente Quando teima em passear desce de sal nos olhos doente da falta que sente do mar vim s dar despedida vim s dar despedida” This dissertation is dedicated to my parents Anthonius Whilhelmus Gerardus Cornelissen and Wanda Garabini Cornelisse n. Their unquestionable love, friendship and constant support inspired each and every one of my days over the past years.

PAGE 3

ACKNOWLEDGMENTS My first and special “thank you” goes to my major professor, Pe ter Stiling, for his guidance, friendship and enthusiasm. He provided the encouragement and necessary advice to start and finish all my projects. Special thanks to my committee members, Dr. Bell, Dr. Huxel and Dr. Rossi, for their s upport and helpful s uggestions. Many people have inspired, guided, helped, cried and laughed with me during the five years I spent at USF. Special thanks to the Stiling lab, pa st and present members: Tere Albarracin, Amanda Baker, Laura Altfled, Mark Barrett Rebecca Forkner, Heather Jezorek, Kerry Bohl, and Sylvia Lukanewic. I was very luc ky to meet Tere and es tablish such a strong friendship with her, which I hope it will last for the years to come. I could not have asked for a better friend. I also thank Celina Bellanc eau, Daniela Schiopu a nd Katie Basiotis for their friendship and their ability to keep thi ngs fun. I was also very fortunate to meet Rebecca Forkner on my last doctoral year and sh e inspired me over this last year. I owe a huge debt of gratitude to my parents, my brother Willy a nd sister Mariana and Miguel. Their encouragement and understanding in deal ing with all the chal lenges I have faced are greatly appreciated. I also thank my in-laws, Adao, Va l, Karinne, Fabio, Dani, Lucas and Gabriel for their love and support. Finally, all my gratitude and special thanks to my husband, Andrey Castro, for making my past 5 years so special and bright. His support, encouragement, and understanding made my days special and nothing in a simple paragraph can express the love and admiration I have for him. After all, I think we both have learned that graduate sc hool is a challenging experience, but rewarding. I love you.

PAGE 4

TABLE OF CONTENTS LIST OF TABLES iv LIST OF FIGURES v ABSTRACT viii CHAPTER 1 – INTRODUCTION 1 Leaf Miners 1 Bottom-up and To-down effects on insect herbivores 3 CHAPTER 2 – SMALL VARIATIONS OVER LARGE SCALES: A TEST OF THE ABUNDANT-CENTER hypothesis 6 Synopsis 6 Introduction 7 Study System 10 Methods 12 Data Collection 12 Data Analysis 14 Results 18 Discussion 28 CHAPTER 3 – CLUMPED DISTRIBUTION OF OAK LEAF MINERS BETWEEN AND WITHIN PLANTS 35 Synopsis 35 Introduction 36 Study Systems 39 Methods 41 Data Collection 41 Data Analysis 43 Results 46 Discussion 57 CHAPTER 4 – RESPONSES OF DIFFERENT HERBIVORE GUILDS TO NUTRIENT ADITION AND NATURAL ENEMY EXCLUSION 61 Synopsis 61 Introduction 62 i

PAGE 5

Study Systems 64 Methods 65 Data Collection 65 Data Analysis 68 Results 71 Treatment Effectiveness 71 Treatment effects on host plant quality 71 Treatment effects on the abundance of leaf miners and herbivores 72 Treatment effects on leaf miner survivorship 73 Strength of bottom-up forces 74 Discussion 81 CHAPTER 5 – DOES LOW NUTRITIONAL QUALIT Y ACT AS A PLANT DEFENSE? AN EXPERIMENTAL TEST OF THE SLOW-GROWTH, HIGH-MORTALITY HYPOTHESIS 85 Synopsis 85 Introduction 86 Study System 89 Methods 90 Data Collection 90 Data Analysis 93 Results 94 Treatment effectiveness 94 Treatment effects on host plant quality 95 Treatment effects on leaf miner performance 96 Discussion 101 CHAPTER 6 – PERFECT IS BEST: LOW LEAF FLUCTUATING ASYMMETRY REDUCES HERBIVORY BY LEAF MINERS 105 Synopsis 105 Introduction 106 Study Systems 109 Methods 112 Data Collection 112 Data Analysis 116 Results 119 Test for asymmetry on Q. laevis and Q. geminata 119 Fluctuating asymmetry and l eaf quality 120 Fluctuating asymmetry between indivi duals and herbivory 120 Fluctuating asymmetry within indivi duals and herbivory 122 Fluctuating asymmetry and mine survivorship 123 Do herbivores cause asymmetry? 125 Discussion 133 CONCLUSION 138 ii

PAGE 6

LIST OF REFERENCES 140 APPENDICES 154 Appendix 1Some leaf and stem gall-formers sampled on Quercus myrtifolia and Q. chapmanii over the range of their distribution in Florida 155 ABOUT THE AUTHOR End Page iii

PAGE 7

LIST OF TABLES Table 2.1 List of sites where Cameraria sp. nova mines were collected in Florida. 21 Table 2.2 Results of partia l regression analyses for Cameraria densities and the combined effects of biotic and sp atial variables. 22 Table 3.1 C-score indices of the ra ndomized and observed matrices for leaf-miners on Q.laevis and Q. geminata and gall-formers on Q. geminata 49 Table 4.1 Results from multivariate analyses of variance for mean herbivore density on Q. geminata and Q. laevis. 75 Table 4.2 Effects of treatments (b ottom-up and top-down manipulations) on herbivore abundance on Q. geminata and Q laevis 76 iv

PAGE 8

LIST OF FIGURES Figure 2.1 Cameraria sp. nova mines on Quercus myrtifolia. 23 Figure 2.2 Distribution of Cameraria sp. nova mines on 40 sites sampled in scrub patches in Florida. 24 Figure 2.3 Spatial correlogram of the abundance of Cmaeraria mines from across the geographical range of Q. myrtifolia 25 Figure 2.4 Spatial correlograms of bo ttom-up factors that might influence Cameraria abundance over the range of Q. myrtifolia. 26 Figure 2.5 Spatial correlograms of de mographic rates and top-down factors that might influence Cameraria abundance over the range of Q. myrtifolia. 26 Figure 3.1 Examples of some of the herbivores on Quercus geminata 50 Figure 3.2 Examples of some of the herbivores on Quercus laevis 51 Figure 3.3 Temporal variation on the concentration of foliar nitrogen, tannins, leaf toughness a nd leaf water content for Q. geminata (solid circles) and Q. laevis (open circles). 52 Figure 3.4 Relationship between the abundance of Stilbosis mines and variation in Q. geminata nutritional quality. 53 v

PAGE 9

Figure 3.5 Relationship between the abundance of Brachys mines and variation in Q. laevis nutritional quality. 54 Figure 3.6 Relationship between the abundance of Acrocercop mines and variation in Q. laevis nutritional quality. 55 Figure 3.7 Co-occurrence patterns of leaf-miners and gall-formers at the plant scale. 56 Figure 4.1 Treatment effects on the c oncentration of tannins and foliar nitrogen of Quercus geminata (left panels) and Q. laevis (right panels) over the season. 77 Figure 4.2 Treatment effects on the abundance of different herbivore guilds on Q. geminata 78 Figure 4.3 Treatment effects on the abundance of different herbivore guilds on Q. laevis 79 Figure 4.4 Strength of bottom-up ma nipulations on the abundance of herbivores on both Q. geminata and Q. laevis in the absence of natural enemies. 80 Figure 5.1 Treatment effects on the concentration of foliar nitrogen, tannin concentration, leaf wa ter and foliar toughness of Quercus laevis 98 Figure 5.2 Treatment effects on the size and development of the leaf miners Acrocercops albinatella and Brachys tesselatus 99 Figure 5.3 Frequency of occurrence of mortality factors for leaf miners growing under four different treatments. 100 vi

PAGE 10

Figure 6.1 Schematic representation (not to scale) of measurements used to define fluctuating asymmetry in A) Quercus laevis and B) Q. geminata 126 Figure 6.2 Differences in (a) tannin c oncentration and (b) nitrogen content between symmetric and asymmetric leaves of Q, laevis and Q geminata 127 Figure 6.3 Relationship betw een the abundance of a) Brachys mines and the percentage of asymmetric leaves on Q. laevis (r2=0.279, P<0.005) and b) Stilbosis mines and the percentage of asymmetric leaves on Q. geminata (r2=0.318, P<0.005). 128 Figure 6.4 Relationship between th e abundance of mines caused by a) Brachys (r2=0.475, P<0.001), b) Stilbosis (r2=0.394, P<0.001) and c) Acrocercops (r2=0.018, P>0.05) and the levels of relative asymmetry (FA Index 2) on host plants. 129 Figure 6.5 Differences in a) fluctuati ng asymmetry between mined and unmined leaves attacked by Acrocercops Brachys and Stilbosis and b) frequency of occurrence of combinat ions of asymmetric and symmetric leaves on pairs of mined-unmined leaves. 130 Figure 6.6 Differences in a) mine size and b) mine growth rate between symmetric and asymmetric leaves attacked by Acrocercops, Brachys and Stilbosis leaf miners. 131 Figure 6.7 Differences in a) leaf miners developmental time and b) survivorship in symmetric and asymmetric leaves of oak species. 132 vii

PAGE 11

Herbivory by Leaf-miners on Florida Scrub Oaks Tatiana Cornelissen ABSTRACT This study investigated effects of pl ant quality and natural enemies on the abundance and survivorship of several leaf miner species on Florida scrub oaks over several ecological scales Three oak species ( Quercus laevis, Q. geminata, and Q. myrtifolia ) and four leafminer species ( Acrocercops albinatella Brachys tesselatus Stilbosis quadripustulatus and Cameraria sp. nova) were the main focus of five separate studies, addressing effects of bottom-up and top-down factors at regional, local, and individual scales. At the regional scale, it was observed that Cameraria sp. nova was aggregated into sites, and sites closer to ea ch other exhibited sim ilar densities of mines than sites farther apart. None of the bottom-up and top-down factors studied were spatially structured, but did influence the variation in Cameraria abundance over the range of the host plant Q. myrtifolia. At the local scale, all leaf miners studied were aggregated between and within plants, and variation in bottom-up factors among individual plants explained va riation in abundance for some of the leaf miners studied. Intra-specific competition was identified as an important factor influencing mine survivorship, but inter-speci fic competition among leaf miners and gall-formers viii

PAGE 12

did not shape the community structure of oa k herbivores. Experime ntal manipulation of bottom-up and top-down factors via fertilizati on and natural enemy removal showed that bottom-up effects were important determinants of leaf miner abundance, as fertilized plants supported 2 to 5-fold more herbivores than control plants. The removal of natural enemies, on the other hand, did not signifi cantly impact the a bundance and/or the survivorship of leaf miners and other guilds studied. At individual scales, it was demonstrated that two leaf miner species responded to random variations in leaf morphology, by increasing in abundance in indivi dual host plants with more asymmetric leaves and/or higher levels of fluctuati ng asymmetry. These results offered support for the plant stress hypothesis and differences in host plant quality were again partially responsible for the results found. ix

PAGE 13

1 Chapter 1 INTRODUCTION Leaf Miners Leaf mining is a feeding habit defi ned by consuming "live foliage while simultaneously dwelling inside it" (Connor & Taverner 1997). In practice, a leaf mine can be distinguished from most other forms of herbivory by the pr esence of at least partially intact epidermal layers on both surfaces of a leaf at the site of damage (Hering 1951). Leaf mines are usually visible on the ex terior of the leaf as serpentine paths, blotches, or other characterist ic shapes of discolored ti ssue. Mines may be occupied throughout an insect's feeding li fe, or may be abandoned for ot her feeding habits at some point in development. Some leaf miners, es pecially larger leaf-mining lepidopterans, excavate more than one mine during the course of development. Most, however, develop completely inside a single mine. In comparis on to external folivores, leaf miners are relatively small insects, physically constraine d by the thickness and area of leaves they occupy. The number of generati ons per year varies widely from univoltine species to facultative and obligate multivoltine ones. With in the insects, there are approximately 10,000 described species of leaf miners (F aeth 1991), but density estimates and demographic data are available for on ly 1% of these species (Auerbach et al. 1995). The habit is known from taxa in 51 families, but only in larvae of the holometabolous orders Coleoptera, Diptera, Hymenoptera, a nd Lepidoptera (Connor & Taverner 1997).

PAGE 14

2 Lepidoptera include more leaf mining fam ilies and species than any other order (Hespenheide 1991) and leaf mining appears to be derived from other feeding habits, both external and internal forms of herbivory. Leaf mining insects attack herbaceous a nd woody plants in both terrestrial and aquatic habitats (Auerbach et al. 1995) and are important pe sts of agricultural crops, greenhouse plants and orchard tr ees. Approximately 25% of the described species have been observed at high, eruptive densities but most species typically occur at low densities. Several ecological hypotheses have been proposed to explain the adaptive significance of leaf mining ha bit (reviewed by Connor & Tave rner 1997), such as escape from natural enemies (such as predators, pa rasites and pathogens), protection from the physical environment (desiccation, UV ra diation, dislodgment by weather), and avoidance of plant defenses (selective feedi ng to maximize intake of most nutritious/least noxious tissues). Recent analyses of mortalit y for leaf miner species revealed that parasitoids and, in early life stages, plant qual ity effects can be the most important factors that influence leaf miner abundance and su rvivorship (Cornell & Hawkins 1995, Hawkins et al 1997). Leaf miners are ideal for the study of the effects of bottom-up and top-down factors on insect abundance and survivorship. Th e fact that these herb ivores live out most of their life histories within the confines of a mine, as well as the persistence of the record of the mine, allows an ecologist to observe and reconstruct their lif e history, to measure the effect of host plant, and to observe associ ated faunas such as predators and parasitoids (Hespenheide 1991). The ease with which leaf mines can be studied has made them a

PAGE 15

3 popular system for ecological research and fo r addressing questions related to population dynamics and community structure. Bottom-up and Top-Down effect s on insect herbivores Understanding what influences herbivore a bundance is a major challenge for insect ecologists. For insect herbivores, their occu rrence and feeding is influenced by plant geographical range and local abundance (Doak 2000, Carson & Root 2000, Kery et al. 2001), size and structural complexity (Strong et al. 1984, Tinney et al. 1998, Masumoto et al. 2000), nutritiona l quality (Cooke et al. 1984, Fischer & Konrad 2000, Denno et al. 2000), secondary chemistry (Bernays & Chapman 1994, Johnson et al 1996, Lindroth et al. 2000) and phenology (L eather 2000, Jarzomski et al. 2000, Masters et al. 2001), and influence of competitors (Norris 1997, Fisher et al. 2000) and higher trophic levels, such as predators and parasitoids (Hawkins et al. 1997). The effects of plant quality and natural enemies on the attack rates of herbivor ous insects on plants have been extensively studied and many hypotheses have been propos ed to explain both within and between population-level variations of herbivory rates among and between different plant species. Studies of herbivory on natural plant populat ions have figured prominently in the ecological literature for decades and have contributed to a general framework of ecological theory (Hairston et al 1960, Oksanem et al. 1981, Polis & Strong 1996). Although much information has accumulated for insect-plant interactions over the past years, many questions remain unanswered such as the interactions of plant quality, abiotic factors and effects of the third tr ophic levels on herbivores and plants. For

PAGE 16

4 example, are species commonly limited by the same factors over all parts of their ranges? Do bottom-up and top-down factors inter act to shape responses of herbivore communities? If so, do all spec ies respond equally or can we detect interand intra-guild variation in responses? Most species are embedded in complex f ood webs and studying a single guild in related plant species at differe nt scales can lead to a more complete understanding of how different factors influence the abundance, surv ivorship and interacti ons of species with other members of the community. Although mu ch has been done to understand what factors influence insect herbivore occurren ce and survival, to my knowledge there are very few studies in which multiple scales of observation have been used in ways to better understand what governs insect herbivore feed ing, abundance and survivorship. I believe that the use of multiple levels of organiza tion create the opportunity to better understand plant-herbivore interact ions and to investigate how factor s that influence the survival and performance of herbivorous insects are affect ed by scale. The purpose of this study was 1) to determine the influence of spatial scal e on the occurrence of a leaf miner over its host plant range and examine how differences in plant quality and demographic rates influence the abundance of this particular inse ct when the whole rang e of distribution is considered; 2) to examine how natural va riations in plant quality influence the distribution of leaf miners a nd other herbivores both between and within host plants and how mine survivorship is affected by the presence of intraand inter-specific competition; 3) to examine how experimenta lly induced variation in bottom-up and topdown factors influence both the abundance and the survivorship of leaf miners and to detect any species-specific variation in respons e to altered plant qual ity and third trophic

PAGE 17

5 level effect, and 4) to examine how indivi dual variation in leaf morphology influences leaf miner abundance and survivorship. These questions were addressed in five separate chapters.

PAGE 18

6 Chapter 2 LARGE VARIATIONS OVER LARGE SCALES: BOTTOM-UP AND TOP-DOWN EFFECTS ON THE ABUNDANCE OF AN OAK-LEAF MINER. SYNOPSIS Many plant and animal species have hi gher densities at th e centre of their distribution, with a gradual decline in abundance towards the edge of the range, though reasons for this pattern is not well known. We examined the abundance structure of the leaf miner Cameraria sp. nova over the range of its host plant Quercus myrtifolia in Florida and addressed how bottom-up and t op-down factors varied over its whole distribution. Leaf miner densities as well as plant quality and effects of natural enemies on mine survivorship were evaluated in 40 site s that covered the whole distribution of the plant. Spatial indices of intr aspecific aggregation and spatia lly structured models were used to determine the effects of spatial location on the abundance of Cameraria and effects of both bottom-up (tannin concentrati on, foliar nitrogen, soil nitrogen, and leaf area) and top-down (larvae pa rasitism and predation) on abundance and survivorship. Cameraria mines were, on average, three times more abundant on coastal compared to inland sites and did not suppor t the hypothesis of higher abu ndance on the centre of the distribution. Differences in pl ant quality, larvae parasitism and successful emergence of mines on edge versus central sites might be partially responsible for this finding. Correlograms constructed for the abundance da ta indicated a significant spatial structure

PAGE 19

7 of Cameraria mines over the range: mines were positively spatially autocorrelated at small distances ( 122Km), indicating that sites close to each other tend to have similar mine densities compared to sites further apar t. Partial regression an alyses indicated that only a small variation on Cameraria abundance was explained by variation in bottom-up and top-down effects after the effects of spatial position were taken into account. INTRODUCTION It has been widely demonstrated that species tend to vary in abundance across their distributional ranges: many plant and an imal species have higher densities at the centre of their distribution, w ith a gradual decline in abunda nce towards the edge of the range (e.g., Whittaker 1971, Hengeveld & Haeck 1982, Brown 1984, Curnutt et al. 1996, Mehlman 1997, Sorte & Hofmann 2004). Many bi ogeography texts have described this ‘abundant centre’ distribution and it has even been called a ‘general rule’ of biogeography (Sagarin & Gaines 2002). Seve ral mechanisms have been proposed to explain the abundant centre distribution, and most of them are variations of the idea that species abundance distributions are coupled to environmental gradients. According to Brown (1984), habitat quality for a species is determined by a combination of many abiotic and biotic factors and local abundance is a reflection of how well a particular site meets the needs of a species along many niche axes. These axes include the physiological characteristics of the species (e.g., temperat ure tolerance, soil quality) as well as ecological factors (e.g., response to competitors and mortality imposed by predators and parasites). Brown (1984) and Brown et al. (1995) assumed that these parameters are

PAGE 20

8 spatially autocorrelated and increasing the di stance from an optimal site decreases the probability of a site meeting the multidimensi onal needs of a species, with a consequent decline in population abundance. The processes behind these patterns are controversial, and some exceptions such as higher abundance on the edges of the distribution have been found (e.g., Prince et al. 1985, Blackburn et al. 1999), but the pattern holds true for some species (examples in Sagarin & Gaines 2002) However, among the 22 studies reviewed by Sagarin & Gaines (2002) that directly addressed the abundant centre distribution hypothesis, only 8 studies include d data collected throughout the entire range of the species, raising concerns as to how well other studies reflect the patterns of change in local abundance across entire geographical ranges (Brewer & Gaston 2002). Little is known about the relationship between spat ial variation and abundance of insects throughout the host plant range, except for the extensive studies conducted for the Holly leaf miner Phytomiza ilicis (Diptera: Agromyzidae) in Europe (Brewer & Gaston 2002, 2003, Klok et al. 2003, Gaston et al. 2004). Leaf miner insects can be ideal models to address the abundant centre distribution hypothesi s, because they are generally restricted to a single host plant species and their sessile habit allows sampling in different parts of the range as well as the collection of life-history data in the field. The edges of the range of the host plant sp ecies are supposed to be subject to the most stressful conditions, and environmental st resses such as water deficit and/or nutrient imbalances that affect host plant physiol ogy and quality can be quite important to herbivores, as proposed by the Plant Stress Hypothesis (White 1984). The Plant Stress Hypothesis argues that herbivor e abundance is higher on stress ed host plants due to an increased availability of nut rients, a decreased concentr ation of defensive compounds

PAGE 21

9 and/or changes in the ra tio of nutrients to chemical defens es. In a recent meta-analysis of the effects of water-stress on insect he rbivores, Huberty & Denno (2004) found that stressed plants showed a tendency to exhibi t elevated foliar nitrog en and there was a nonsignificant trend for leaf miners (mainly Le pidoptera and Diptera) to achieve higher densities on stressed plants over control plants. As argued by Huberty & Denno (2004) leaf miners are thought to benefit from wa ter-stressed plants because they can take advantage of elevated leaf nitrogen and yet avoiding feeding on compartmentalized allelochemicals, and thus benefit from elevat ed nitrogen arising from intermittent stress. In this sense, stressed plants at the edges of the range might be more heavily attacked and exhibit larger populations of insect herbivores Also, herbivores might develop faster in stressed plants, in accordance with pred ictions of the slow-growth, high-mortality hypothesis (Clancy & Price 1987), decreasing th us the mortality pressure exerted by natural enemies, such as predators and parasitoids. How effects of plant quality and natural enemies interact to determine herbivore abundance and performance is well understood, but virtually all efforts to quantify the relative contributions of top-down and bottom-up regulatory factors on herbivore populations have been perf ormed over relatively local scales, with very few exceptions (e.g., Brewer & Gaston 2002, Gaston et al. 2004). Plant quality features such as the concentration of secondary compounds as well as the abundance and composition of natural enemy communities might vary accordin g to environmental factors such as soil type, temperature and rainfall patterns (Price 1997). As a consequence, both the quality of herbivore host plants and th e pressure exerted by the third trophic level might differ between different parts of the distri butional range of the species (Gaston et al. 2004).

PAGE 22

10 Previous studies of oak leaf miners have revealed high variation in density both between and within plants (e.g., Sato 1991, Brown et al. 1997, Forkner & Hunter 2000) and similar results were found for scrub oaks in Florida at local spatial scales (e.g., Stiling et al. 1987, Cornelissen & Stiling 2005, Corneli ssen & Stiling 2006a,b). In this study we investigated the abundance and su rvivorship of the leaf miner moth Cameraria sp. nova throughout the range of its host plant distribution Quercus myrtifolia (Fagaceae). This system provides an ideal model to te st both the abundant cen tre hypothesis as well as the effects of top-down a nd bottom-up factors at larger scales: the sessile larva is easily identified by the shape of the mine, larval abundance and development might be strongly influenced by aspects of plant qual ity and mines leave a permanent record of larval fate, which provides insights on their population dynamics and effects of the third trophic level. Moreover, Q. myrtifolia has a geographic range 99.8% contained within the state of Florida (Nixon 1997, Price et al. 2004). This enabled us to sample the full distributional range of Cameraria mines and contrast abundance, survivorship, and demographic features of this leaf miner in all parts of its range. STUDY SYSTEM The Florida scrub is one of its most di stinctive ecosystems found in coastal and ancient inland dunes throughout the state. The major scrub groupi ngs in Florida are in the coastal panhandle, coastal peninsula and inland peninsula (for details see Marshall et al. 2000). Quercus myrtifolia (Fagaceae) is a semi-deciduous oa k that often grows in dense stands in scrub ecosystems. Leaves are dark green, shiny and have a leathery appearance.

PAGE 23

11 Plants are found in dry, sandy soils usually associated with sand dunes, sand hills and scrubs in Florida (Wunderlin & Hansen 2000). Although present in small, patchy distributions in Alabama, Georgia and South Carolina (Nixon 1997), most of the distribution of Q. myrtifolia is in Florida, although the plan t is absent in approximately 21 Florida counties in the northern edge -of-range (Wunderlin & Hansen 2000). Cameraria sp. nova (Lepidoptera: Gracillariidae) is the most common leaf miner on Q myrtifolia and mines are abundant in late spring and early summer. Mating, oviposition and egg hatch occur between Ap ril and May and the upper surface blotch mines appear on Q. myrtifolia leaves by mid-May and ear ly-June (Figure 2.1). Mines might reach 4cm in length and occupy 50-80% of the upper leaf surf ace (T Cornelissen, pers. obs.). The larvae go through 5 instar s, taking up to 4 weeks to complete development and pupae form into a silk en cocoon inside the mine. Although Cameraria sp. nova is the most common leaf miner on Q. myrtifolia similar mines occur in low densities on the chapman oak Q. chapmanii and very low densities on the sand live oak Q. geminata though whether they belong to the same species is not known. However, we suggest that Cameraria’s geographical range is primarily limited by the availability of myrtle oaks, as densities on both chapman a nd sand live oaks are extremely low (average SD number of Cameraria mines on Q chapmanii 0.058 0.06 per 200 leaves and on Q. geminata 0.015 0.02 per 200 leaves).

PAGE 24

12 METHODS Data Collection Between July 1st and September 5th of 2005, data on the abundance and demographic parameters of Cameraria sp. nova as well as data on aspects of plant quality were determined at 40 sites in Florida, encompassing the whole distribution of Q. myrtifolia its primary host. Sites included many state park s, state forests and re serves (Table 2.1), and sites were selected based on the presence of pure scrub patches, consisting mainly of Q. myrtifolia Q.chapmanii Q geminata and Q. inopina In a few sites (3 out of 40) the turkey oak Quercus laevis was also present, but not samp led in this study. Plants were sampled in 36 Florida counties, covering approximately 170,000 Km2. At each site, we recorded geographic position (latitude-longitude coordina tes expressed as decimal degrees) and elevation using a GPS positioned where the first individual of Q. myrtifolia was found. Leaf miner density was estimated at each site by counting the number of Cameraria sp. nova mines on 200 randomly chosen leaves of 15 Q. myrtifolia individual plants per site (n=3,000 leaves per s ite). Leaves were sampled randomly, all around the canopy, and leaf miners were identified a nd recorded. We also recorded the abundance of other leaf miners (mainly Stigmella sp., Bucculatrix sp., Stilbosis sp. and Brachys sp.) and other herbivores belonging to several guilds, such as stem-gallers, leaf-gallers, leaf-tiers, chewers and leaf rollers (Appendix 1). Brown (1984) argued that local abundance is a reflection of how well a particular site meets the needs of a species along many ni che axes and plant quality features such as

PAGE 25

13 the concentration of secondary compounds a nd foliar nitrogen, as well as the abundance and the composition of the natural enemy community, might vary according to how environmental factors vary at different parts of the host plant range. Differences in plant quality were assessed by collecting 50 undama ged leaves (collected from at least 10 trees) on each site. Leaves were immediately pl aced on ice, returned to the laboratory and dried in an oven at 50oC for 3-4 days. Digital pictures of all leaves were taken in the lab to estimate mean leaf area of Q. myrtifolia on each site. Leaves were then milled to a fine powder and tannin and foliar nitrogen concentrat ion were determined from 10 leaves per site. Tannins were extracted from 50 mg of dry tissue, and tannin concentration was quantified using the radial diffusion assay with three replicates per l eaf (for details see Hagerman, 1987). Nitrogen content was dete rmined using a CE Instruments NC2100 CN Analyzer (CE Elantech, Incorporated, Lakew ood, New Jersey, USA). We also collected 3 soil samples at each site, by taking 3 random co res near 3 different plants and sampling soil 5-10 cm deep. Soil samples were analyzed for the percent nitr ogen and carbon using a CHN Analyzer. Because top-down factors, such as the pr essure exerted by natural enemies, might also influence leaf miner abundance over the range, and studies have shown that there can be systematic changes in particular de mographic rates across a species’ geographical range (e.g., Taylor et al. 1980, Sanz 1997), at each site we also collected between 50 and 100 mined leaves to assess survivorship ra tes and sources of mortality (n= 3,026 mines collected). Successfully emerged larvae of Cameraria exit the mine as flying adults by cutting the edges of the s ilky cocoon. Mines that were preyed upon are usually found open and the larva is missing. Larval parasitism is easily recognized by the presence of

PAGE 26

14 parasitoids inside the mine, by the presence of parasitoid hole s on the external surface of mines and by the presence of parasitoid pupae at tached to the leaf miner larvae. Larvae killed by host plant resistance are usually found dead inside th e mine, but intact and flat and mine walls are usually blackened. Demogra phic rates for this particular leaf miner consist only of larval survi vorship and mortality data, be cause assessment of oviposition for Cameraria sp. nova is not easily done, as no scar is left after female oviposition and eggs were no longer found when censuses started. Data Analysis Although data on several other leaf mi ners and other herbivore guilds on Q. myrtifolia were collected during this study, species association, community composition as well as competitive interactions over the host plant range will be addressed in another study. This study focused on the spatial structure of Cameraria sp. nova across its distributional range and effects of variation in host plant qua lity and pressure exerted by natural enemies on its distribution a nd abundance at the regional scale. Spatially indices of intraspecific ag gregation were calculated using SADIE techniques (Spatial Analysis by Distance In dices) (Perry 1995). Briefly, SADIE operates by comparing the spatial arrangement of th e observed sample with other arrangements derived from it, such as those where the indi viduals are as crowded as possible, those in which they are arranged at random and those in which they are as regularly spaced as possible. Specifically, intra-sp ecific aggregation indices co mpare the spatial arrangement of the observed distance to regularity (the total number of moves which individuals in each sample must make so that all sample units have the same number of individuals)

PAGE 27

15 with the permuted distances to regularity derived from randomization procedure (Perry 1995, Perry et al. 1999). An index of aggregation (Ia) and randomization tests were calculated as Ia = 1, indicating a random distribution and Ia >1 an aggregated distribution of individuals across sample units. Additionally, the index v provides a measure of clustering for each sampling unit (site), with subscript i for patches and subscript j for gaps. A cluster is defined as a region of relatively large c ounts close to one another in a two-dimensional space (i.e., a patch) or of relatively small counts (i .e., a gap). Significant positive mean vi values indicate spatial clustering of leaf mines into patches, whereas significant negative mean vj values indicate the presence of gaps in the spatial distribution of the species. Details of SADIE techniques can be found in Perry et al (1999). These indices were then tested unde r the null hypothesis of random arrangement of observed densities using form al randomization tests (Perry et al. 1999, McGeoch & Price 2004). All indices were calculate d using SADIEShell version 1.22 The analysis of spatial pattern is of pr ime interest to ecologists because most ecological phenomena investigated by sa mpling geographic space are structured by forces that have spatial components (Lege ndre & Legendre 1998). To determine whether local population densities of Cameraria exhibit any spatial struct ure across the range of Q. myrtifolia spatial correlation analysis was performed using the mean density of Cameraria mines at each site on Q myrtifolia Spatial structure was determined by calculating Moran’s I spatial autocorrelation statistics as: I (d) = n i i n h n i i h hiy y n y y y y w W1 2 11) ( / 1 ) )( ( / 1 for h i

PAGE 28

16 where yh’s and yi’s are the values of the observed variables at sites h and i Before computing spatial autocorrelation coefficients a matrix of geographic distances D = [Dhi] was created using Geodesic distan ces, i.e., distances that ta ke into account the curvature of the earth’s surface. Distance values were then grouped into distance classes following Sturge’s rule (Nclasses=1+3.3logn, Legendre & Legendre 1998) and Moran’s I was calculated for 7 equally spaced classes. Spatial correlograms we re then created by plotting autocorrelation coefficients for th e various distance cla sses d. The weights whi in the above equation are Kronecker deltas (L egendre & Legendre 1998) of a binary form, i.e., the weights take the value whi = 1 when sites h and i are at distance d (same class) and whi= 0 when sites h and I are at different classes. Finally, W is the sum of the weights whi for the given distance class, i.e., the nu mber of pairs used to calculate the coefficient. Moran’s I usually takes valu es in the interval [-1,+1] and positive autocorrelation in the data translate into pos itive values of I and negative autocorrelation produces negative values. Spatial autocorrelatio n coefficients were tested for significance by calculating confidence intervals and the si gnificance of the overa ll correlograms were assessed using Bonferroni’s corre ction for multiple comparisons. Because not only leaf miner abundance might be spatially autocorrelated, but also aspects of plant quality and mortality rates might show spatial structur e, i.e., closer sites are more similar than more distant sites, we also used Moran’s I to describe any spatial structure in bottom-up (tannin concentration, foliar nitrogen, soil nitrogen and leaf area) as well as top-down (mortality imposed by predators and mortality imposed by parasitoids) factors. All analyses were conducted using PASSAGE (Pattern analysis, spatial statistics, and geographic exeges is) software version 1.0 (Rosenberg 2001).

PAGE 29

17 The use of partial regression analysis is of great interest when one wants to partition the effects of biotic and abiotic factors on the a bundance of a species and to model data showing any type of spatial dependence. Following Legendre & Legendre (1998), Brewer & Gaston (2002) and McGeoch & Price (2004) we used partial regression analysis to estimate how much variation in Cameraria abundance can be attributed to bottom-up and top-down factors once the effects of spatial location have been taken into account. We first ran a multiple stepwise regr ession to determine the best-fit combination of spatial and non-spatia l variables that contributed to si gnificantly explai n the variability of Cameraria densities. Thereafter, significant terms were retained to construct a matrix of biotic (bottom-up and top-down factors) a nd abiotic (latitude, longitude, elevation) variables to be used in subsequent an alyses. The combined effects of both topdown/bottom-up and spatial variables on Cameraria densities were calculated by multiple regression of Cameraria abundance onto both sets of predictive variables combined. Environmental variables and abio tic variables were then removed at two separate steps and the explanatory potential of the biotic variables, after correcting for spatial dependence, was calculated by measur ing the change in deviance accounted for the regression model after the biotic variab les were removed. Statistical significance of these fractions was tested using F-tests. At the end of partial regression analyses, variation in Cameraria abundance was then partitioned into 4 fractions (a, b, c, d) representing: a) no n-spatial environment (fraction that can be explained by biotic vari ables independent of sp atial structure); b) spatially structured environmen t (spatial structure shared by Cameraria and biotic variables measured); c) non-environm ental spatial (spatial structure in Cameraria

PAGE 30

18 densities not explained by the measured biot ic variables), and d) unexplained residual variation. Original data were log (for linear measurements) or angular transformed (for percentages) to meet normality assumptions. All analyses were conducted using Systat 9.0 for Windows. RESULTS Cameraria mines were found in all 40 sites sa mpled in Florida, although large variations were observed at larger scales: mines were, on average, three times more abundant on coastal compared to inland sites (mean number of Cameraria mines on coastal sites 99.1 8.99, inland sites 33.65 4.99, F1,38=41.58, P<0.00001, Figure 2.2). As a consequence of Florida’s peninsular char acteristic, sites on the edges of the range of Q. myrtifolia coincided with most coastal sites a nd exhibited significantly higher numbers of mines than sites at the center of the di stribution of the host pl ant (mean number of Cameraria mines on edge sites 95.9 8.07, center sites 26.23 2.84, F1,38=53.27, P<0.00001). Cameraria mines were significantly intra-spec ifically aggregated on sites, as demonstrated by a significant index of aggregation (Ia = 1.936, P <0.005). The clustering index v also indicated significant spatial clustering of mines into patches (vi=1.708, P=0.015) and significant gaps in the spat ial distribution of this species (vj= 2.358, P=0.013). Correlograms constructed for the abundan ce data indicated a significant spatial structure of Cameraria mines over the range: mines were positively spatially

PAGE 31

19 autocorrelated at small distances ( 122Km), indicating that site s close to each other tend to have similar mine densities compared to sites further apart (P< 0.05 after Bonferroni’s correction, Figure 2.3). The size of the patches, i.e., the distance be tween zones of high and low densities, is indicated in the corre logram by the distance in which the first negative autocorrelation was found, at appr oximately 487Km. A significant positive autocorrelation was also observed between sites separated by longest distance ( 850Km), which might represent coasta l sites at opposite edges of the geographical range. No spatial structure was detected on the bottomup (Figure 2.4) as well as top-down factors (Figure 2.5). The only demographic parameter that showed spatial structure was mine successful emergence (Figure 2.5), indicating that sites near to each other exhibited similar levels of mine survivorship. At ve ry long distances, successful emergence was also negatively spatially autoco rrelated, indicating that sites very far apart are dissimilar, although these results should be interpreted with caution, as in the last distance classes Moran’s I is calculated from a relatively low num ber of pairs of sites (9 in this case). Successful emergence showed a similar pattern of autocorrelation to Cameraria abundances at small distances, i. e., sites that were apart in 122km were more similar in the number of mines and rates of mine su rvivorship than sites farther apart. Regression analyses indicated that both bi otic and abiotic variables contributed to variation in the density of Cameraria mines over the range: 6 7.6% of the variation in Cameraria abundance was explained by the combined effects of latitude, longitude, elevation, foliar nitrogen, successful emergence of mines and parasiti sm rates of larvae (r2=0.676, P<0.0001). Partial regression analyses using these significant terms revealed that, once spatial positio n was taken into account (fracti on c on model), only 23.1% of the

PAGE 32

20 variation in Cameraria abundance could be attributed to biotic variation alone (fraction a on model: r2=0.231, P<0.05). The combined effects of biotic variables and spatial position (fraction b) explaine d 13.3% of the variation in Cameraria abundance over the range (Table 2.2). Although onl y 23.1% of the variation in Cameraria abundance was explained by environmental variation after taking into account the effects of spatial variables, this small fraction was still signifi cant, indicating that a significant effect of bottom-up (successful emergence and foliar nitrogen) as well as top-down (larvae parasitism) factors was detected.

PAGE 33

21 TABLE 2.1List of sites where Cameraria sp. nova mines were coll ected in Florida. (St= State). Site Number Site Name County 1 Fort Cooper St Park Citrus 2 Withlacoochee St Forest Citrus 3 Oscar Scherer St Park Sarasota 4 Little Manatee River St Park Hillsborough/Manatee 5 Paines Creek Historic St Park Hardee 6 Silver River St Park Marion 7 Alligator Creek Preserve Charlotte 8 UCF Arboretum Orange 9 Archbold Biological Station Highlands 10 Lake June-in-winter St Park Polk 11 Lake Wales Ridge St Forest Polk 12 Hobart Park Indian River 13 Bluefield Ranch St Lucie 14 Jonathan Dickson St Park Martin 15 Yamato Scrub Palm Beach 16 Crystal Lake Sand Scrub Nature Reserve Palm Beach 17 Rookery Bay Preserve Collier 18 Estero Scrub Preserve Lee 19 Lake Manatee St Park Manatee 20 Blue Springs St Park Osceola 21 Washington Oaks St Park St Johns 22 UNF Scrub Duval 23 Etoniah Creek St Forest Putnam 24 Jay B. Starkey Wilderness Park Hernando 25 Enchanted Forest Brevard 26 Big Lagoon St Park Escambia 27 Naval Live Oaks Nature Trail Santa Rosa 28 Top Sail St Park Walton 29 St Andrews St Park Bay 30 St Joseph Peninsula St Park Gulf 31 Tate’s Hell St Forest Franklin 32 Route 51 Scrub Steinhatchee Taylor 33 Cedar Key Scrub St Preserve Levy 34 Swift Creek Conservation Area Columbia 35 Arcadia Scrub DeSoto 36 Ridge Manor Scrub County Park Hernando 37 USF Ecoarea Hillsborough 38 Paynes Prairie St Park Alachua 39 Tenoroc Fish management Area Polk 40 Boyd Hill Preserve Pinellas

PAGE 34

22 Table 2 2 Results of partial regression analyses for Cameraria densities and the combined effects of biotic and spatial variab les. (* indicates stat istically significant differences at P=0.05 and ** at P=0.001 after F-tests). Dependent Variable % Variation in mine density explained by Total a (env) b (env x space) c (space) d (residual) Cameraria abundance 67.6 23.1* 13.3 31.2** 32.4

PAGE 35

23 Figure 2.1 Cameraria sp nova mines on Quercus myrtifolia

PAGE 36

24 Figure 2.2Distribution of Cameraria sp nova mines on 40 sites sampled in scrub patches in Florida. Sites were plotted using latitude and longitude coordinates converted into decimal degrees and the size of sy mbols is proportional to the abundance of Cameraria (average number of individuals per 200 leaves on 15 plants) at each site.

PAGE 37

25 Figure 2.3Spatial correlogram of the abundance of Cameraria mines from across the geographical range of Q myrtifolia Solid circles identify autocorrelation statistics that remain significant after progressi ve Bonferroni’s correction ( = 0.05) and empty circles are non-significant values. Distance (Km) Moran's I -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 122.0244.0366.0487.9609.9731.9 854.0

PAGE 38

26 Figure 2.4Spatial correlograms of bottomup factors that might influence Cameraria abundance over the range of Q. myrtifolia All correlograms were statistically nonsignificant after Bonferroni’s correction (all P>0.05) indicating that va riation in bottomup factors over the range of Q. myrtifolia are independent of sp atial structure. Distance Classes Moran's I -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 Distance Classes Moran's I -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 Distance Classes Moran's I -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 Distance Classes Moran's I -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 Tannins P>0.05 Foliar Nitrogen P>0.05 Leaf Area P>0.05 Soil Nitrogen P>0.05122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0

PAGE 39

27 Figure 2.5Spatial correlograms of demographic rates and top-down factors that might influence Cameraria abundance over the range of Q. myrtifolia Successful emergence was the only demographic parameter spatially structured after Bonf erroni’s correction, indicating that mine survivorship is spatially structured (P<0.05). Distance Classes Moran's I -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 Distance Classes Moran's I -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 Distance Classes Moran's I -0.15 -0.10 -0.05 0.00 0.05 0.10 Distance Classes Moran's I -0.4 -0.3 -0.2 -0.1 0.0 0.1 Succesful Emergence P<0.05 Predation P>0.05 Plant Resistance P>0.05 Larvae Parasitism P>0.05122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0 122.0244.0366.0487.9609.9731.9854.0

PAGE 40

28 DISCUSSION Almost without exception, indi vidual species exhibit prof ound spatial variation in the local densities which they attain (Blackburn et al. 1999). They are entirely absent from most places, at relatively low densities in the majority of those at which they occur, and at relatively high densities in a small proportion of occupied sites. A significant spatial structure was observed for Cameraria mines over the range of Q. myrtifolia and positive spatial autocorrelations indicated that sites in closest proximity to each other were similar in both the abundance of mine s as well as successful emergence, an indicator of mine survivorship. This pa tchy structure was also corroborated by the aggregation and clustering indices, a form of distribution typically resulting from one or more dispersal, intraor inter-specific intera ctions, biotic or abioti c environmental factors (Legendre 1993). However, none of the top-dow n and bottom-up factors analyzed in this study were spatially structured, as de monstrated by the nonsignificant spatial correlograms. Similar results were found by Gaston et al. (2004) addressi ng variation in quality of Ilex aquifolium leaves over its range in Europe. Values of Moran’s I for that study were very low and non-signi ficant, indicating little simila rity in most measures of host-plant quality used. Our data suggests th at variation in mine survivorship (i.e., successful emergence) across sites might contribute to the variation in Cameraria densities over the range. Our analyses of tannin concentration over the range of Q. myrtifolia indicated that tannins were not spatially structured a nd did not contribute to the variation in Cameraria abundance. Genotypic differences among individuals that are not influenced by spatial position might explain these results. The am ounts of foliar nitrogen

PAGE 41

29 were also not spatially structured, but did cont ribute to a small amount of variation in leaf miner abundance, a finding previously reco rded in other plant-herbivore systems (reviewed by Waring & Cobb 1992, Kyto et al. 1996). The lack of sp atial structure in the bottom-up and top-down factors analyzed in our study and the presence of spatial structure in Cameraria abundance indicates that there is no simple covariance between these two (Gaston et al 2004). Although we have shown th at there is variation in the amount of tannins and foliar nitrogen among sites, as well as the amount of larvae mortality inflicted by natural enemies, there is little evidence that this variation assumes any marked spatial structure. Although factors that influence host-plant quality such as light and soil nutrients might vary with environmental conditions, there may be no consistent trends in these f actors with changes in latitude and/or longitude (Gaston et al. 2004), as shown in the present study. Also, othe r aspects of plant quality not addressed in this study, as well as climatic conditions such as temperature and ra diation, might affect leaf miner abundance and might expl ain the spatial structure found for Cameraria population. Although the bottom-up and top-dow n factors analyzed in this study were not spatially structured, some factors still contributed for a fraction of the variation in Cameraria abundance over the range of Q. myrtifolia independently of spatial position, a finding previously recorded for other sessile insects (e.g., McGeoch & Price 2004, Gaston et al. 2004). In previous studies, we have de monstrated that leaf miners on oaks respond to both natural as well as experimenta lly elevated levels of foliar nitrogen, and mine mortality tends to be relatively low and an important determinant of mine abundance in subsequent seasons (Corne lissen & Stiling 2005, 2006a,b).

PAGE 42

30 Contrary to what was predicte d by the abundant-centre hypothesis, Cameraria densities did not peak in the center of the distribution of Q. myrtifolia : higher abundances were actually observed at the edges of the di stribution, coinciding with coastal sites in Florida. It has been suggeste d that the most favorable c onditions are found at the centre of a species’ distribution, and these conditions in tur n, support the highest population density across the distributiona l range. Our results do not su pport this hypothesis, as central sites exhibited lower densities. Brown et al. (1995) suggested that spatial correlograms of species typi cally show a bowl-shape patt ern, with positive spatial autocorrelations at both short and ve ry long distances, and argued that high autocorrelations at very long lags ( 850km in our study) might indicate the similarly low levels of density found at opposing range edges. Although the correlogram for Cameraria shows the typical bowl-shape, we suggest that strong and significant positive autocorrelations found at the la rgest distance cl ass actually indicates edge sites with high density of mines, at the opposite edges of th e distribution, such as the Florida Panhandle and the Atlantic coast. A comparison of plant quality features between central and edge sites revealed that edge sites exhibi ted significantly highe r foliar nitrogen (F1,38=4.834, P=0.034) and a tendency for reduced tannin concentration (F1,38=3.248, P=0.08). These results are in accordance with predictions from the Plant Stress Hypothesis, and reinforce the idea that leaf miner populations might be nefit from increased nitrogen concentration in stressed plants and might peak in abunda nce on the distribution edges, instead of center or interior populations. Although a recent meta-analytical review (Huberty & Denno 2004) showed limited support for the plant stress hypothesis, leaf miners were an underrepresented group, and these authors suggest ed that future stud ies addressing this

PAGE 43

31 hypothesis should focus on herbivore guilds for whom substantial examples were lacking, such as leaf miners and stem borers. Our st udy provided evidence that, at the largest scale of distribution of this particul ar leaf miner species, plant stre ss seems to be an important determinant of the spatial structure of this species. Organisms’ abundance and fitness levels ar e predicted to peak in the species’ range center and decline towards its range edges when environmental optimums are found at the centre of the di stribution, which doesn’t seem to be the case for Cameraria populations. Few studies have addressed e dge effects on intra-specific abundance and demographic rates of populations, although di fferential survivorsh ip, oviposition rates and effects of natural enemies on overall mort ality might determine the spatial structure of some populations (e.g., Fagan et al. 1999, McGeoch & Gaston 2000). McGeoch & Gaston (2000), however, found higher a bundance of the holly leaf miner P. ilicis on edge compared to interior habitats in England. Explanations for this pattern included lower bird predation and pupal parasitism in edge versus interior populat ions. In the present study, parasitism rates of Cameraria larvae on centre sites were approximately 35% higher than parasitism rates on edge site s (mean parasitism centre sites: 21.2% 2.5, edge sites: 15.3% 1.9), although no significant differe nces were observed for mine predation (mean larvae pred ation on centre sites 27.9% 2.4, edge sites: 25.2% 1.2). Overall, larvae mortality was higher on centre (62.0% 3.2) compared to edge populations (57.1% 3.3) indicating that demographic parameters that determine species abundance and population dynamics might di ffer between peripheral and central populations of Cameraria

PAGE 44

32 Recent studies have pointed out the importa nce of using spatially explicit models when studying population dynamics, especially when nearby populations share more similarity than would be expected by chance. Scrub oaks are critic ally endangered habitat in Florida, and our results ha ve shown that coastal populatio ns support higher densities of leaf miners and leaf mine survivorship in these coastal areas is greatest, indicating the importance of the edges of the distribution for the population dynamics of this species. The Florida peninsula is fam ous for being flat (Marshall et al. 2000) but because of its low-lying topography, Florida underwent a se ries of inundations during the glacial period, and during interglacial events, cu rrents deposited sand along what were shorelines, creating inland sand ridges. These ridges represent the highest altitudes in Florida, and their existence has been shown to be important for several other species, such as spiders (Marshall et al. 2000), beetles (Hubbell 1954), and grasshoppers (Deyrup 1996). Our results indicate th at central populations of Q. myrtifolia which coincide with Florida’s most prominent ridges (e.g., Br ooksville Ridge, Lake Wales Ridge, DelandCrescent City Ridge) and highest elevations (41 to 65 meters) do not support large populations of Cameraria suggesting that either isola tion, elevation or even genotypic differences between host plants on center scrub islands might have influenced the patterns found. The limited dispersa l ability of leaf miners mi ght be responsible for the existence of spatial autocorrelation in Cameraria populations, as demonstrated for other insects with limited dispersa l that exhibit metapopulat ion dynamics (e.g., GonzalezMegias et al. 2005). After controlling for spat ial location, the amounts of foliar nitrogen and the percentage of larvae parasitism were statistically significant predictors of Cameraria

PAGE 45

33 abundance over the range of its host plant. Th e importance of foliar nitrogen for insect herbivores has been extensively discussed (Mattson 1980) and higher foliar nitrogen on coastal edge plants, where mine densities were higher, offer additional support for the plant stress hypothesis and its effects on herbi vores. Drought stress has been shown in a number of studies to significantly imp act the top-down and bottom-up influences on insect herbivores (Mattson & Haack 1987, Fay et al. 1993, Levine & Paige 2004) and coastal barrier islands in Florida exhibit c onsiderably low rainfall than mainland areas (U.S Department of Interior, Florida). Para sitoids have also be en long suggested as potential forces in the dynamics of leaf miner populations, as miners tend to support richer parasitoid communities then free-li ving herbivores (Connor & Taverner 1997). We are unaware of studies that a ddressed levels of natural en emy attack over the range of other leaf miner species and their importance to mine a bundance and survivorship. Lower parasitism rates of Cameraria larvae on edge, coastal sites also offer support to the slowgrowth, high-mortality hypothesis (Clancy & Price 1987). According to this hypothesis, herbivores feeding on plants of low nutritiona l quality do not necessarily increase damage on their host by overcompensatory feeding if increased development time due to poor host quality increases the window of vulnerability of herbivores to natural enemy attack. In this sense, leaf miners developing on e dge populations where plant quality was higher might develop faster and decrease the wi ndow of vulnerability to the attack of parasitoids. Future studies evaluating mine development tim e, growth and survivorship on edge versus central populations will addre ss this point and offer a better test of the slow-growth, high-mortality hypothesis.

PAGE 46

34 The abundance structure of a species is clearly the compound outcome of several processes, such as climate, history, resource quality and biotic interactions (McGeoch & Price 2004). If species have population dynamics driven strongly by resource quality, then abundance structures are likely to be determined by th e distribution and quality of resources across the landscape. Although at local scales it has been demonstrated that leaf miners respond strongly to bottom-up factors (Moon & Stiling 2004, Cornelissen & Stiling 2006b), resources themselves were not spatially structured in our study, indicating that at larger spatial scales, other factors might dictate the abundance structures found for the Cameraria population studied. Host plant quality and natural enemy effects may be strong and identifiable only at fine, and not landscape, scales (Bevers & Flather 1999), whereas dispersal abilities and ot her abiotic variables, such as latitude and altitude, might determine leaf miner abundance and structure at the landscape scale. Within patches, however, plant quality might be a strong dete rminant of mine density, as demonstrated by our previous studies in similar oak systems.

PAGE 47

35 Chapter 3 CLUMPED DISTRIBUTION OF OAK LEAF MINERS BETWEEN AND WITHIN PLANTS SYNOPSIS Leaf miners typically sh ow non-random distributions both between and within plants. We tested the hypothesis that leaf miners on two oaks species were clumped on individual host trees and individual branches and a ddressed whether clumping was influenced by aspects of plant quality and how clumping and/or inte ractions with other oak herbivores affected leaf miner survivorship. Null models were used to test whether oak herbivores and different herbivore gu ilds co-occur at the plant scale. Twenty individual Q. geminata plants and 20 Q. laevis plants were followed over the season for the appearance of leaf miners and other herbivores, and foliar nitrogen, tannin concentration, leaf toughness and leaf wate r content were evaluated monthly for each individual tree. The survivorship of the most common leaf miners was evaluated by following the fate of marked mines in several combinations that invo lved intraand interspecific associations. We observed that all leaf miners studied were clumped at the plant and branch scale, and the abundance of most leaf miner species was influenced by plant quality traits. Mines that occurred single on leaves exhibited significantly higher survivorship than double and triple mines and le aves that contained a mine and a leaf gall and a mine and damage by chewers exhibited lowest survivorship. Although leaf miners were clumped at individual host trees, null m odel analyses indicated that oak herbivores

PAGE 48

36 do not co-occur significantly less than expected by chance and there was no evidence for biological mechanisms such as inter-sp ecific competition determining community structure at the plant scale. Thus, despite co -occurrence resulting in reduced survivorship at the leaf scale, such competition was not strong enough to structure separation of these oak herbivore communities. INTRODUCTION Patterns of distribution of insects on plan ts are strongly determ ined by host plant variation. A question of great in terest in herbivory-related studies is how plant traits affect attack rates by phytopha gous insects and how variation in plant quality affects the distribution and performance of herbivores both within a nd among host plants. Recently, it has become clear that differences betw een individuals can have profound effects upon the kind of dynamics exhibited by herbivorous populations, their stab ility and their mean levels of abundance (Crawley & Akhter uzzaman, 1988) and the study of individual variation an its consequences for populat ion dynamics has become a priority. For folivorous insects, it is not unusual for populations to be ag gregated on their host plants (Stanton, 1983; Faeth, 1990). For leaf mini ng insects, oviposition site selection by females may be highly influenced by variati on in leaf structure (Reavey & Gaston, 1991), leaf age and size (Faeth, Mopper, & Simb erloff, 1981; Faeth, 1991), leaf chemistry (Stiling, Brodbeck, & Strong, 1982) and effects of the third trophic level (reviewed by Connor & Taverner, 1997) and leaf miners usually show non-random distributions among plants and among leaves with in an individual tree (Stiling, Simberloff, & Anderson,

PAGE 49

37 1987; Shibata, Ishida, Soeya, Morino, Yosh ida et al., 2001). One recurring explanation for clumped patterns of leaf miners in particular is that insect distribution reflects variation in nutritional qual ity and/or secondary chemistry among and within the host plants (Faeth, 1990). Because l eaf quality is a major dete rminant of host choice by many herbivores (Strong, Lawton, & Southwood, 1984), va riations in leaf quality are expected to influence leaf miner distribu tion, abundance and survivorship. Besides variations in leaf quality, intera ctions with other he rbivores, including external and other internal feeders may affect the distri bution and abundance of leafmining insects. The distribution of endophagous insects, such as gall-formers and leaf miners, is typically aggregated and when sele ction for tissues of better nutritional quality does occur, competition between insects that u tilize the same type of resource can arise. The role of intraand inter-specific comp etition among insects in ecological theory has changed throughout the years, from the argument that competition was weak and infrequent on phytophagous insect communities in the early 80’s, to the resurrection of the importance of competition between phytophagous insects in the 90’s (Denno, McClure, & Ott, 1995; Reitz & Trumble, 2002) For leaf miners, competition may arise from the presence of con-specifics and/or from the presence of other feeding guilds, such as gall-formers and free-feeding herbivores. The quality of resources available to leaf miners, and hence their performance, may be modified by the feeding of other folivores (Faeth, 1992). Leaf chewers, in particular, can affect the folia ge on which leaf miners are feeding by changing plant quality and by altering or redirecting plant resources (Shibata et al., 2001). Damage caused by leaf chewers can, for example, lead to an increase in secondary compounds, such as phenolics, and a decrease in nitr ogen concentration

PAGE 50

38 (Hartley & Lawton, 1987). Gall-formers also ha ve the potential to m odify plant quality by physically and/or chemically modifying plan t vasculature, architecture and nutritional quality (Johnson, Mayhew, Douglas, & Ha rtley, 2002), and some studies have demonstrated that galled leaves exhibit hi gher nutritional quality than neighboring nongalled leaves (e.g., Abrahamson & Weis, 1986, but see Hartley & Lawton, 1992). For sessile insects and free-feeding herbivores in ge neral, most of the studies that indirectly addressed interspecific competition by analyzing species co-occurrence have used presence-absence distributional data as a surrogate for competition (e.g., Stiling, Rossi, Catell, & Bowdish, 1999; Kagata & Ohgushi, 2001, and examples in Denno et al., 1995), although more recent studies advocate for the us e of more refined st atistical te sts based on random distributions such as null models (e.g., Gotelli & Graves, 1996; Ribas & Schoereder, 2002; Zwolfer & Stadler, 2004). The purpose of this study was to inve stigate how differences in plant phenology and nutritional quality influence the dist ribution and abundance of leaf miners on Quercus geminata (Fagaceae) and Q. laevis both between and within plants. We also aimed to examine the effects of other herbi vores co-occurring on th e same host plants on both the abundance and survi vorship of the most comm on leaf miners on both oak species. The specific aims of this study were : 1) to examine variation in the spatial distribution of leaf miners among plants, w ithin plants and according to canopy position, 2) to determine how leaf miner abundance wa s affected by plant nut ritional quality in terms of foliar nitrogen, water content, tannin concentration and l eaf toughness, 3) to observe how the presence of other herbivore guilds such as chew ers and gall-formers affect the abundance and survivorship of l eaf miners on oaks, and 4) to determine

PAGE 51

39 whether herbivore guilds and he rbivore species co-occur significantly less or more than expected on oak host plants using null models. STUDY SYSTEMS The sand live oak, Quercus geminata (Fagaceae), is a semi-evergreen oak and, typically, old leaves abscise and new leaves appear in late April or early May, reaching full size in approximately 2 weeks. Stilbosis quadripustulatus (Lepidoptera: Cosmopterygidae) is a moth whose larvae induce mines on the adaxial surfaces of Q. geminata S. quadripustulatus is a univoltine species, whose adults emerge in early summer (from May to June) from pupae that overwinter in soil a nd litter. Oviposition occurs approximately in early June, when fema les oviposit at the junction of the midvein and a major lateral vein. Larvae take from 60 to 90 days to complete their 5 instars and mines may reach 3.0 cm in length (Simberlo ff & Stiling, 1987). Many other herbivores compose the insect community associated with Q. geminata (Figure 3.1). Leaves are frequently found damaged by chewing inse cts such as the eastern buck moth Hemileuca maia (Lepidoptera: Saturniidae), and at least 4 cynipid species (Hyme noptera: Cynipidae) of galling insects are commonly observed on sand live oak leaves and stems: Andricus quercusfoliatus Disholcaspis quercussuccinipes Callirrhytis quercusbatatoides and Belonocnema quercusvirens. A. quercusfoliatus induces white flow er-like galls on sand live oak stems, whereas D. quercussuccinipes wasps induce clusters of 5-20 yellowish brown galls usually crowded around a terminal oak twig. C. quercusbatatoides wasps induce abrupt swellings of twigs, varying in form and size and B. quercusvirens induces tan, globular pea-like galls on the underside of Q. geminata leaves. Galls are unilocular

PAGE 52

40 and occur in large numbers dur ing the fall. Eyespot galls (Diptera: Cecidomyiidae) are recognized as circular spots, usually 8-10 mm in diameter. The adults emerge from the soil in the spring and lay eggs in the upper leaf surface. As the larva grows, the leaf tissue surrounding it swells slig htly and red rings are seen ar ound the galls. Larvae complete their development in 8-12 days and pupate in the soil. This is the most common gall found on sand live oak leaves, often reachi ng densities of 5 galls per leaf. The turkey oak Quercus laevis is one of the characterist ic trees associated with the sand hill community over much of Florida. Q. laevis is a moderately fast to fastgrowing tree and presents deci duous simple leaves, alternatel y arranged with usually 5 lobes, although this number may vary from 3 to 7. Acrocercops albinatella (Lepidoptera: Gracillaridae) is a microlepidopteran species whose larval stages feed on young leaves, creating distinct linear-blotch mines on the lower surface of Q. laevis leaves. Larvae typically feed on the palisade parenchyma cells and deposit frass throughout the mine, completing their development in approximately 10 days. Larvae emerge from the blotch mine and usually pupate on the same leaf from which they emerge (T Cornelissen, pers. obsv.). Brachys tesselatus (Coleoptera: Buprestidae) is a univoltine species that also forms distinct blotch mines in Q. laevis leaves. The adults emerge in Mid-March to MidApril, coinciding with budburst of turkey oa k. Adults initially feed on the early leaves and flowers until mating and oviposition. Eggs are deposited singly on the upper surface of the leaves and after hatc hing the larvae mine into the mesophyll creating distinct, characteristic damage. Brachys in our study sites go through tw o generations and the first mines appear in early April a nd remain active until late June, when larvae complete their development and exit mines to pupate on the so il. New adults emerge in early July and

PAGE 53

41 oviposit to form new Brachys mines that remain active until September-October. Pupation and overwintering of this second genera tion occurs within th e leaves after they have senesced and abscised from the tree. New adults emerge from the leaf litter in the following spring (Waddell, Fox, White, & Mou sseau, 2001). Turkey oak leaves are also attacked by a vast array of herbivores (Figure 3.2), such as the leaf roller weevil Homoeolabus analis (Coleoptera: Atellabidae) the eastern buck moth H. maia the white tussock moth Orgyia leucostigma (Lepidotpera: Lymantriidae) and other leaf miners such as Stigmella (Lepidoptera: Nepticulidae) and Cameraria (Lepidoptera: Gracillariidae). Eyespot galls and an unidentified cynipid gall are the most common galls observed in turkey oak leaves in our field sites and no stem galls have been recorded. METHODS Data Collection This study was conducted between Febr uary and November of 2003 at the University of South Florida Botanical Gard en. To examine effects of plant phenology and nutritional quality on the abundance of leaf mine rs and other herbivor es, 20 individuals of Q. geminata and 20 individuals of Q. laevis were marked in February, just before budbreak and leaf flush. Q. geminata trees ranged from 1.9 to 2.5 m in height and Q. laevis trees ranged from 2.5 to 3.2 m in height. On each individual plant, 5 upper-canopy (above 2.0m) and 5 lower-canopy (below 1.5m) br anches were selected and all leaves on each branch were individually numbered using a permanent marker. A total of 6,489 marked leaves on Q. geminata and 2,243 marked leaves on Q. laevis were followed over the season. Leaves were marked at the time of l eaf flush when all leav es were still intact,

PAGE 54

42 and examined monthly for the appearance of leaf miners, gall-formers, damage by chewers, and leaf abscission. To assess variation in host plant qual ity among trees, water content, foliar nitrogen concentration, tanni n concentration and leaf t oughness were evaluated monthly for each individual plant. On each collec tion date, 10 undamaged leaves (5 from the upper-canopy and 5 from the lower-canopy) were sampled from each tree, placed immediately on ice, and leaf toughness wa s evaluated using an Effegi FT-011 penetrometer (International Ripening Co, Italy). Water content was quantified by the difference between leaf wet and dry weight s and leaves were fu rther oven-dried and milled to a fine powder. Tannins were extr acted from 50 mg of dry tissue, and tannin concentration was quantified using the radial di ffusion assay with three replicates per leaf (for details see Hagerman, 1987). Nitroge n content was determined using a CE Instruments NC2100 CN Analyzer (CE Elant ech, Incorporated, Lakewood, New Jersey, USA). To assess the effects of conspecifics and other herbivores on the survivorship of the most common leaf miners on both oak spec ies, on each individual plant, we noted leaves that exhibited each of the following combinations: Quercus geminata : 1) one Stilbosis mine, 2) two Stilbosis mines, 3) three or more Stilbosis mines, 4) one Stilbosis one Brachys mine, 5) one Stilbosis and one or more eyespot galls, 6) one Stilbosis mine and leaf damage by chewers. Although Brachys mines occur mainly on Q, laevis some mines have been observed on Q, geminata where both plant species co-occur (T Cornelissen, pers. obs). Acrocercops was never recorded in Q. geminata in our field sites. For Q. laevis the following combinations were investigated: 1) one Acrocercops mine, 2)

PAGE 55

43 two or more Acrocercops mines, 3) one Brachys mine, 4) two or more Brachys mines, 5) one or more Acrocercops and one or more Brachys mine, 6) one or more Acrocercops mine and damage by chewers, 7) one or more Brachys mine and damage by chewers. The fate of leaf miners on each combination was followed over the season, with observations conducted at bi-weekly interval s (n=598 marked leaves for Q. geminata and 708 leaves for Q. laevis ). Leaf miner survivorship and mortality, as well as date of leaf abscission were scored for each leaf combination thr oughout the season. Leaf miners offer a great opportunity to assess population survivorship and mortality factors si nce a record of the miner success is clearly observed on the leaves : parasitized mines have tiny circular exit holes on mine’s surface and/or pupae within the mine, and predated mines are usually found ripped open and the larva is abse nt. Successfully emerged larvae of Acrocercops cut open the mines and pupate usually on the same leaf where the mine developed. Brachys larvae pupate inside mines a nd/or cut circular holes on the underside of the leaf. Stilbosis mines cut semi-circular holes at the mi ne edge and larvae ex it to pupate in the soil. Data Analysis To examine the spatial distribution of leaf miners both between and within individual plants, the distribut ion of leaf miners were comp ared to random (Poisson) and clumped (negative binomial ) distributions using a X2 analysis (Ludwig & Reynolds, 1988), with individual plants a nd branches within plants used as sampling units. We also calculated indexes of dispersi on (variance-to-mean ratio) a nd Green’s modified index of clumping (Green, 1966) for each data set. The variance and mean are equal in a theoretical Poisson distribution and the i ndex of dispersion was calculated as ID= s2/ x,

PAGE 56

44 where s2 and x are the variance and sample mean, re spectively. Significant departures of ID from the value of 1.0 were tested using a X2 statistic with n-1 degrees of freedom. Green’s index of clumping wa s calculated as GI = [(s2/ x) – 1/ n-1], where n = sample size. GI varies between 0 (for ra ndom) and 1 (for maximum clumping). To test for differences in nutritional quali ty among plants and between leaves at different canopy positions, we averaged toughne ss, water, nitrogen and tannin contents for each plant over the season and regressed th ese values with the abundance of the most common leaf miners and other herbivores, summed over the season. Differences in leaf quality and herbivore abundance between low and high canopy were tested using a OneWay ANOVA. All the variables analyzed were first submitted to Lilliefor’s test for data normality and log transformations were employed to stabilize variances and normalize the data. However, for the sake of clar ity, figure axes and means (+1SEM) show untransformed data. To examine the community effects of herb ivores on oaks, we first analysed data on mine survivorship among all the leaf categ ories as described in the methods. Single (n=457 leaves) and double mines (n= 377) we re the most frequent combinations, followed by mines and damage by chewers (n=228), and mines and galls (n= 149). Differential survivorship among cat egories was tested using a X2 and proportional survivorship among classes was tested usi ng an ANOVA with individual plants as replicates. Because previous studies have shown non-random distribut ion of leaf miners and other herbivores on oaks, we also asked whether clumped distributions among individual plants differed am ong herbivores and among guilds, i.e., if a plant already

PAGE 57

45 heavily occupied by an herbivore in part icular is avoided or preferred by other herbivores. To test whether herbivores on oaks co -occur significantly more or less than expected at the plant scale we used a null models analysis. Here, observed patterns are randomly generated and a null model is then used to randomize the occurrence of species and to compare the patterns in these “ar tificial” communities with those in real communities (Ribas & Schoereder, 2002). Data on species distribution among the oak plants were transformed into presence/absence matrices, in which columns are individual plants and rows are herbivore species. Analyses were conducted separately for Q. geminata and Q. laevis We used the C-score index (Stone & Roberts, 1990) as a metric to quantify the pattern of co -occurrence of leaf miners and gall-formers within a presence-absence matrix, as follows: C = (ri-S)(rj-S), where ri and rj are the row totals, and S is the number of sites occupied by bot h species. The C-score measures the average number of “checkerboard units” (Gotelli & Entsminger, 2001) a nd is an index negatively correlated to species co-occurren ce. The null hypothesis in this case is that the presence of a given herbivore species does not influence the occurrence of other species and if the index of co-occurrence falls within the 95% frequency distribution of the randomized matrices, the null hypothesis is accepted a nd the hypothesis of biological mechanisms conditioning the species co-occu rrence is rejected (Ribas & Schoereder, 2002). To test these distributions against randomized matrices we used a fixed-fixed model, with 5,000 iterations, in which the row and column sums of the original matrix are preserved. This algorithm was chosen for the fact that it has a low frequency of Type I and Type II errors (Gotelli & Ellison, 2002) and random matrices were created using a swapping algorithm,

PAGE 58

46 in which the original matrix is shu ffled through repeated swapping of random submatrices. All analyses were conducted using EcoSim (Gotel li & Entsminger, 2001). Analyses were conducted only for leaf miners and ga ll-formers due to their sessile habit and for the fact that leaf chewers move fr eely among plants and should be less influenced by the clumped distribution of endophagous herbivores. RESULTS Testing the data against the null hypothe sis of a Poisson (r andom) distribution indicated that all leaf mine r species studied were not ra ndomly distributed among plants ( Acrocercops : X2=52.68, df=11, P>0.05; Brachys : X2=43.86, df=10, P>0.05; Stilbosis : X2=68.45, df=13, P>0.05) and within bran ches on individual plants ( Acrocercops : X2=711.72, df=5, P>0.05; Brachys : X2=135.11, df=5, P>0.05; Stilbosis : X2=744.3, df=6, P>0.05). The distribution of a ll leaf miners, however, did fit a negative binomial distribution, suggesting clumped di stributions both among plants ( Acrocercops : X2=52.68, df=11, P>0.05; Brachys : X2=43.86, df=10, P>0.05; Stilbosis : X2=68.45, df=13, P>0.05) and within branch es on individual plants ( Acrocercops : X2=14.88, df=14, P<0.05; Brachys : X2=7.11, df=9, P<0.05; Stilbosis : X2=9.37, df=13, P<0.05). Indices of dispersion and Green’s index corroborated th e clumped distribution of leaf miners on both Q. laevis ( Acrocercops : ID=8.03, X2=152.58, P>0.05, GI=0.741; Brachys : ID=4.40, X2=83.62, P>0.05, GI=0.853) and Q. geminata ( Stilbosis : ID=5.19, X2=1008.3, P>0.05, GI=0.727). No significant differences in nutritional quality were observed between leaves from the lowerand upper-canopy for both plant species (all P>0.05), although Q.

PAGE 59

47 geminata leaves tended to be softer in th e lower canopy (average toughness lower-canopy leaves: 0.71 0.023, average toughness high canopy leaves: 0.84 0.072), and Q. laevis upper-canopy leaves tended to exhibit high er tannin concentration (average tannin concentration 0.2860.091) than lower-canopy le aves (average tannin concentration 0.2230.11). Although leaf miner ab undance tended to be higher in lower-canopy leaves compared to upper-canopy leaves for both oak species, these differences were not statistically signi ficant (all P>0.05). For both oak species, we observed seas onal trends in plant quality, with a decrease in nitrogen concen tration over the season, as well as an increase in toughness and tannin concentration (Figure 3.3). For Q. geminata we observed higher Stilbosis density in plants with more nitrogen and softer leaves (Figure 3.4), whereas Q. laevis plants with higher nitrogen and lower tann in concentration supported significantly higher densities of the first generation of Brachys (Figure 3.5). None of the plant quality variables analyzed in this st udy affected the abundance of Acrocercops mines on Q. laevis (Figure 3.6), as well as eyespot galls, Andricus galls and chewed leaves on Q. geminata For Q. laevis however, a higher percentage of damage by chewers was observed in individual plan ts with softer leaves (r2=0.21, P=0.04). Survivorship of leaf miners was high est when mines were single on leaves, compared to double mines or mines occurri ng on leaves that we re also chewed (X2= 14.69 to 19.55, all P<0.05). Lowest survivorship rates were observed for Stilbosis mines occurring on Q. geminata leaves with one or more eyespot galls (F1,18= 16.44, P=0.032). For the community of herbivores on both oak spec ies, null models indicated that both leaf miners and gall-formers co-occurred at th e plant scale (Figure 3.7) and there was no

PAGE 60

48 evidence for competitive exclusion between a nd within guilds, as indicated by C-score indices falling within the 95% limits of freque ncy distribution of the randomized matrices (Table 3.1).

PAGE 61

49 Table 3.1C-score indices of the randomised and observed matrices for leaf miners on Q. laevis and Q. geminata and gall-formers on Q. geminata Table shows the minimum and maximum values of the indices calculated for 5,000 randomised matrices pre data set, together with the observed index and P-values in two-tailed tests (Obs. Observed, Exp. Expected). Guild Randomized matrix Obs. matrix P-values MinimumMaximum Obs.>Exp. Obs.
PAGE 62

50 Figure 3.1 – Examples of some of the herbivores on Quercus geminata (from left to right): Hemileuca maia (Lepidoptera: Saturniidae), Leaf rollers on tips of leaves, old Disholcaspis (Hymenoptera: Cynipidae) galls on stems, new Disholcaspis galls on stems associated with ants, Andricus quercusfoliatus (Hymenoptera: Cynipidae) galls on stems, Stilbosis quadripustulatus (Lepidoptera: Cosmopterygidae) mines on expanded leaves, Phigalia sp. (Lepidoptera: Geometridae), banded tussock moth Halysidota tesselaris (Lepidoptera: Arctiidae), and eyespot galls on leaves.

PAGE 63

51 Figure 3.2 – Examples of some of the herbivores on Quercus laevis (from left to right): unidentified Geometridae (Lepi doptera), Leaf rollers on ne w leaves, unidentified stem gall, Homoeolabus analis (Coleoptera: Attelabidae), Brachys tesselatus (Coleoptera: Buprestidae) mines, Acrocercops albinatella (Lepidoptera: Gracill aridae) mines, adult Brachys on leaves, Tischeria sp. mine (Lepidoptera: Tischeriidae), and whitemarked tussock moth Orgyia leucostigma (Lepidoptera: Lymantriidae).

PAGE 64

52 Figure 3.3Temporal variation on the concentra tion of foliar nitrog en, tannins, leaf toughness and leaf water content for Q. geminata (solid circles) and Q. laevis (open circles). Data are means (+1SE) of 10 undamage d leaves per individu al tree, with lowerand upper-canopy leaves combined. Foliar nitrogen (%) 1.00 1.05 1.10 1.15 1.20 1.25 Tannin concentration (mg/g) 0.1 0.2 0.3 0.4 0.5 0.6 Leaf toughness (lbs./mm2) 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Water content (mg) 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 AprMayJunJulAugSepOct NovAprMayJunJulAugSepOct Nov

PAGE 65

53 Foliar nitrogen concentration 0.91.01.11.21.31.41.5 Stilbosis density 0 10 20 30 40 50 60 70 r2=0.27 P=0.019 Tannin Concentration 0.300.350.400.450.50 Stilbosis density 0 10 20 30 40 50 60 70 P>0.05 Water content 0.150.200.250.300.350.400.450.50 Stilbosis density 0 10 20 30 40 50 60 70 P>0.05 Leaf toughness 0.40.50.60.70.80.91.0 Stilbosis density 0 10 20 30 40 50 60 70 r2=0.38 P=0.003 Figure 3.4Relationship between the abundance of Stilbosis mines and variation in Q. geminata nutritional quality. Data on Stilbosis abundance are summed over the season for each individual plant and data on plant nutriti onal quality are averages over the season.

PAGE 66

54 Tannin concentration 0.100.150.200.250.300.35 Brachys density 0 5 10 15 20 25 30 35 Nitrogen concentration 0.80.91.01.11.21.31.4 Brachys density 0 5 10 15 20 25 30 35 Water content 0.60.70.80.91.01.11.2 Brachys density 0 5 10 15 20 25 30 35 Leaf toughness 0.250.300.350.400.450.50 Brachys density 0 5 10 15 20 25 30 35 r2=0.255 P=0.023 r2=0.143 P=0.058 P>0.05P>0.05 Figure 3.5Relationship between the abundance of Brachys mines and variation in Q. laevis nutritional quality. Data on Brachys abundance are summed over the season for each individual plant and data on plant nutriti onal quality are averages over the season.

PAGE 67

55 Tannin concentration 0.100.150.200.250.300.35 Acrocercops density 0 20 40 60 80 100 120 Nitrogen concentration 0.80.91.01.11.21.31.4 Acrocercops density 0 20 40 60 80 100 120 Water content 0.60.70.80.91.01.11.2 Acrocercops density 0 20 40 60 80 100 120 Leaf toughness 0.250.300.350.400.450.50 Acrocercops density 0 20 40 60 80 100 120 P>0.05 P>0.05 P>0.05 P>0.05 Figure 3.6Relationship between the abundance of Acrocercops mines and variation in Q. laevis nutritional quality. Data on Acrocercops abundance are summed over the season for each individual plant and data on plant nutritional quality are averages over the season.

PAGE 68

56 Figure 3.7Co-occurrence patterns of l eaf miners and gall-formers at the plant scale. The histograms give the frequencies of simulate d C-scores using a fixed-fixed model. Ranking of C-scores are frequency classes of randomized matrices represented by numbers to facilitate scaling. Refer to Ta ble 1 for minimum and maximum values on each class. Arrows represent the observed C-score indices for each data set. Indices falling within the 95% limits of frequency di stribution of randomized matrices indicate co-occurrence, whereas higher C-scores represent smaller co-occurrence than expected by chance and lower C-scores indicate higher co-occurrence than expected by chance. Ranking of C-score 1234567891011 Frequency of simulated matrices 0 200 400 600 800 1000 1200 1400 1600 Leaf-miners Q. geminata observed index Ranking of C-score 12345678910111213 Frequency of simulated matrices 0 200 400 600 800 1000 1200 1400 observed index Leaf-miners Q. laevis

PAGE 69

57 DISCUSSION Our data showed that leaf miners on both oak species were not randomly distributed, but clumped among host plants a nd within branches on individual plants. Several factors could potentially select for non-random distribution of leaf miners. Bottom-up factors such as foliar nitrogen c oncentration and secondary chemistry have been frequently invoked as potential factors aff ecting the distribution, abundance and survivorship of phytophagous insects and l eaf miners in particular (e.g., Stiling et al. 1982; Faeth, 1991; Eber, 2004). Our data corrobora tes the hypothesis that leaf miners are strongly affected by bottom-up f actors, as plants with hi gher nitrogen concentration, lower leaf toughness and lower tannin concen tration exhibited signi ficantly more mines than plants with lower nutritional quality. Variation in nitrogen le vels, especially, both among and within plants has been demonstr ated to affect leaf miners choice for oviposition sites and larval development (M attson, 1980; Scheirs, DeBruyn, & Verhagen, 2001, 2002). Faeth (1990), on the other hand, obs erved that larvae of the leaf miner Cameraria sp. on Quercus emoryi were highly clumped at va rious spatial scales among and within trees, but his st udy did not support the hypothesis that leaf miners cluster because of variation in plant nutritional quality. Although our study showed that leaf miners respond to bottom-up factors, responses to plant quality va ried among the leaf miners studied. Stilbosis mines were mainly affected by foliar nitrogen co ncentration and leaf toughness, whereas Brachys abundance was influenced by nitrogen and tannin concentration, and none of the plant quali ty variables affected the abundance of Acrocercops These differential responses might be explained by differences in lifehistory traits of these leaf miners. Acrocercops albinatella causes relatively small linear-

PAGE 70

58 blotch superficial mines just under the l eaf epidermis of turkey oak leaves and development times do not exceed 10 days. These mines are unlikely to be strongly affected by variations in plant quality due to their fast develo pment rates and the fact that they create limited depth mines in young leav es with higher nitrog en content and lower concentrations of defensive chemicals. Brachys and Stilbosis mines, on the other hand, are more likely to be affected by host qual ity, having longer developmental times and full depth mines. Another factor operating at the leaf scale that potentially selects for non-random distribution of leaf miners is resource or interference competition. Insects that are relatively immobile seem particularly suscep tible to competitive influences because they cannot easily escape from ne ighboring individuals (Stiling et al ., 1987) and results of competitive interactions should then be manife sted in their distribution patterns. In a previous study of the distribution of Stilbosis mines on Q. geminata and Q. nigra Stiling et al. (1987) have shown that fewer mines were found together on the same side of the leaf mid-vein than expected by chance. Intr a-specific competition among leaf miners has mostly been neglected as a regulatory m echanism on their populat ion dynamics (Eber, 2004), even though leaf miners ar e restricted to small “res ource islands” represented by individual leaves (Janzen, 1968). Our results have shown that mines usually occurred single on leaves, but lower survivorship was observed on leaves with double or triple mines. In these categories, 34% of the leaf mines dissected exhibited dead and dry larvae inside the mine. Similar results have b een found for other systems (e.g., Bultman & Faeth, 1986; Auerbach & Simberloff, 1989) a nd interference competition or indirect exploitative competition with conspecifics ha s been suggested as a dominant source of

PAGE 71

59 mortality for many leaf miner species (re viewed by Auerbach, Connor, & Mopper, 1995). We also observed that leaf miners developing on leaves that were also damaged by chewers experienced lower survivorship than mines developing on intact leaves. Previous or concurrent feeding by other phytophages may alter physical and chemical aspects of the leaf or reduce leaf size so that insuffici ent area remains for development, especially for sessile herbivores such as leaf miners and gall-formers. Faeth (1985), for example, observed that Stilbosis juvantis mines developing on leaves th at were artificially an/or naturally damaged by chewer herbivores also experi enced significantly lower survivorship than did miners on intact leav es due to increased parasitism levels. The higher attack rate of parasitoids on mines that were developing on damaged leaves was probably attributable to phys ical and/or chemical alterations caused by chewing herbivores. The exact mechanism by which damage and intensified parasitism levels interact is unclear, although physical, visual, and chemical cu es associated with damaged leaves may influence parasitoid searching behavior and oviposition preferences (Faeth, 1985). Variation and changes in resource quali ty can lead to different patterns of distribution of insect damage, and herbivores might become positively or negatively associated, both intraand inter-specifi cally (Fisher, Hartle y & Young, 1999). For the silver birch Betula pendula several studies have demonstr ated a negative association between generalist herbivores and the leaf miner Eriocrania, possibly due to direct interference and resource removal (e.g., Bylund & Tenow, 1994; Valladares & Hartley, 1994). In our systems, although intraspecifi c competition was an important source of mortality for the leaf miners studied, we did not find evidence suggesting that

PAGE 72

60 interspecific competition structured distributional patterns on individual host trees. Although leaf miners were clumped among host trees, and at the le af scale competition might reduce leaf mine survivorship, at the pl ant scale both leaf mi ners and gall-formers co-occur. These results might be explained by the fact that although leaf miners were clumped among plants, they occurred in low densities in our field sites (mean 1SE per 200 leaves in 35 plants: Acrocercops : 36.7 3.49; Brachys 1st generation: 21.1 2.03; Brachys 2nd generation: 8.55 1.91; Stilbosis : 35.64 3.25) and patterns of inter-specific repulsion might be detected only during outbreak seasons. In conclusion, our results indicated that leaf miners show no n-random patterns of distribution both among and with in plants and these differe ntial distribution might be determined by variation in several aspects of plant quality. Intra-specific competition is suggested as a regulatory mechanism in the population dynamics of the species studied at the leaf scale, although at the plant scale inter-sp ecific associations do not seem to be important mechanisms determining the comm unity structure of these oak herbivores.

PAGE 73

61 Chapter 4 RESPONSES OF DIFFERENT HERBIVORE GUIL DS TO NUTRIENT ADDITION AND NATURAL ENEMY EXCLUSION SYNOPSIS We experimentally investigat ed the effects of plant qua lity and natural enemies on the abundance of different herbivore guilds on oak trees. Two oak species ( Quercus laevis and Q. geminata ) and four guilds of leaf herbivores (leaf miners, gall-formers, leafrollers and chewers) were studied usi ng a factorial design that manipulated predation/parasitism pressure and plant nutriti onal quality. Forty plan ts of each species were divided into 4 treatments: 1) contro l plants (nutrients and natural enemies unaltered); 2) nutrients added, natural enemie s unaltered; 3) nutrients unaltered, natural enemies removed; and 4) nutrients added and natural enemies excluded. Fertilized plants exhibited significantly higher fo liar nitrogen for both oak sp ecies and tannins tended to increase over time and decrease w ith fertilization, but only for Q. geminata this trend was significant. Fertilized plants supported significantly higher densities of all herbivore guilds than unfertilized plants, but exclusi on of natural enemies did not significantly affect herbivore abundance for any guild stud ied. Our results demonstrated that all herbivores on oaks, regardless of guild t ype, respond more strongly to bottom-up effects such as host plant quality, and less to top-down effects caused by natural enemies.

PAGE 74

62 INTRODUCTION A central question in community ecology is the degree to which populations are limited by both top-down and bottom-up forces (Hunter, 2001). Previ ous studies have suggested that, for herbivores, top-down and bottom-up forces commonly interact to influence herbivore populations, but bottomup forces set the stage on which top down forces act, in a way that enemy pressure wi ll vary with plant growth and quality (e.g., Hunter & Price, 1992; Price, 2002). Although individual studies investigating the importance of top-down or bottom-up factors on insect herbivores have been extensively performed, relatively few studies have atte mpted simultaneous manipulations of plant quality and natural enemies pressure in terrest rial plant systems. So me previous studies recognize that herbivore ident ity and life history could be an important variable in determining the relative strength of t op-down and bottom-up forces (e.g., Forkner & Hunter, 2000; Moon & Stiling, 2002; Denno et al., 2002). For ectophagous lepidopterans in upland forest communities, for example, it has been demonstrated that bottom-up forces dominate oak-herbivore-natural enemy interactions and top-down forces such as predation by birds on chewers and several he rbivore guilds are rela tively weak (Forkner & Hunter, 2000). In salt marsh communities, however, both plant quality and natural enemies can significantly impact the abunda nce of gall-makers and sap-suckers of Borrichia frutescens (Stiling & Moon, 2005) and vege tation complexity mediate the impact of natural enemie s of plant hoppers in Spartina alterniflora (Denno et al ., 2002). However, Gruner (2004) suggests that, in genera l, few terrestrial studies have attempted to partition the relative impacts of top-dow n and bottom-up forces among trophic levels or feeding guilds.

PAGE 75

63 The aim of this research was to investig ate the combined effects of plant quality and natural enemies on the abundance of different guilds of herbivores utilizing the same host plants by using a factorial manipulation of predation/parasitism pressure and plant nutritional quality. Invertebrate predation and attack by parasitoids we re chosen as focal top-down effects in our system and manipulation of plant nutritional quality by fertilization was chosen as the bottom-up effect based on previous studies of the importance of nitrogen content and plant s econdary chemistry to insect herbivores (Strong et al., 1984). To our knowledge, this is the first study simultaneously manipulating top-down and bottom-up forces for several oak herb ivore guilds with within-guild variation in species life history patterns. For leaf miners, previous studies have stressed the importance of plant qual ity on abundance and survivorship (Faeth & Simberloff, 1989; Connor, 1991), but have also shown how indirect effects of plant quality can affect natural en emy performance (Bultman & Faeth, 1986) and we expected strong bottom-up effects for this oak guil d. For gall-makers, the nutrition hypothesis (Stone & Schnrogge, 2003) stat es that gall-formers shoul d be less influenced by the nutritional status of the host pl ant, as they have the ability to manipulate galled tissues to make them more nourishing and less well defe nded than non-galled tissues on the same plant and we expected top-down effects to be strong. For chewers, experimental studies (e.g., Strauss, 1987) have demonstrated hete rogeneous responses to increased plant quality caused by fertilization depending on the herbivore species, but based on Forkner and Hunter’s (2000) and Gruner’s (2004) recent studi es, we expected bottom-up effects to be strong.

PAGE 76

64 Specifically, we aimed to investigate th e following questions: 1) What are the impacts of plant quality a nd natural enemy pressure on Quercus herbivore densities?; 2) How do plant quality and natura l enemies interact to affect the abundance of different guilds?; 3) Does the strength of top-down fo rces change with plant quality, i.e., are parasitism rates higher/lower in fertilized pl ants?, and 4) Does the strength of top-down forces change with herbi vore guild and identity? STUDY SYSTEMS Sand live oak, Quercus geminata (Fagaceae), is a semi-evergreen oak which, typically, supports many diffe rent herbivore species. Stilbosis quadripustulatus (Lepidoptera: Cosmopterygidae) is a univoltine leafminer whose larvae induce mines on the adaxial leaf surfaces and leaves are also frequently chewed by the eastern buck moth Hemileuca maia (Lepidoptera: Saturniidae). At l east 4 cynipid galling insects (Hymenoptera: Cynipidae) are commonly obs erved on sand live oak leaves and stems: Andricus quer cusfoliatus Disholcaspis quercussuccinipes Callirrhytis quercusbatatoides and Belonocnema quercusvirens. Andricus quercusfoliatus induces white flower-like galls on sand live oak stem s and it the most common stem gall in our study site. Eyespot galls (Diptera: Cecidom yiidae) are also very abundant and are recognized as circular spots on the leaves, us ually 8-10 mm in diameter. This is the most common foliar gall found on sand live oaks, ofte n reaching densities of 5 galls per leaf. Turkey oak, Quercus laevis has deciduous simple leaves, alternately arranged with usually 5 lobes, although this number may vary from 3 to 7. Acrocercops albinatella

PAGE 77

65 (Lepidoptera: Gracillaridae) is a microlepidopteran species w hose larval stages feed on young leaves, creating distinct linear-blotch mi nes on the lower leaf surface, with larvae completing their development in approximately 10 days. Brachys tesselatus (Coleoptera: Buprestidae) is a leaf mine r species that also forms distinct blotch mines in Q. laevis leaves and goes through two generations in our study sites. Turkey oak leaves are also attacked by an array of other herbivor es, such as the leaf rolling weevil Homoeolabus analis (Coleoptera: Atellabida e), the eastern buck moth H. maia the white tussock moth Orgyia leucostigma (Lepidotpera: Lymantriidae) a nd other leaf miners such as Stigmella spp. (Lepidoptera: Nepticulidae), Bucculatrix spp (Lepidoptera: Bucculatricidae) and Cameraria spp. (Lepidoptera: Gracillariidae). Generalist predatory ants (Hymenoptera : Formicidae) and the green lynx spider Peucetia viridans (Araneae: Oxyopidae) are among the most common arthropod predators on both oak species in our study sites. The leaf mine rs studied are also attacked by several parasitoid species, including Zagrammosoma multilineatum (Hymenoptera: Eulophidae) and Chrysonotomyia sp. (Hymenoptera: Eulophidae). METHODS Data Collection This study was conducted between Februa ry and November of 2003 on natural stands of scrub oak vegetati on in Tampa, Florida. Forty Quercus laevis trees ranging between 2.0 and 2.5 meters in height were ma rked at the University of South Florida ECOAREA and forty Q. geminata trees ranging between 1.7 to 2.0 meters in height were marked at the USF Botanical Garden. We i nvestigated effects of plant quality and

PAGE 78

66 pressure exerted by natural enemies on the abundance of three different leaf miners ( Acrocercops albinatella Brachys tesselatus Stilbosis quadripustulatus ), on the abundance of one leaf roller on Q. geminata and on the abundance of chewing herbivores and gall-formers on both Q. laevis and Q. geminata Using a 2 x 2 factorial design, the forty trees of each plant species we re randomly divided into 4 treatments: 1) – F,+P: control plants with nut rients and natural enemies una ltered ; 2) +F,+P: nutrients added, natural enemies present (unaltered); 3) –F,-P: nutrients unaltered, natural enemies excluded, and 4) +F,-P: nutrients added and natural enemies excluded. This design allowed 10 replicates per treatment combination per plant species. Plant quality was manipulated by adding, bi-weekly from April to June, 150 g of 46:0:0 NPK fertilizer to assigned Q. laevis trees and 100 g of the same fertilizer to assigned Q. geminata trees. A previous study with the same systems indicate d that predation by spid ers was a negligible mortality factor for the l eaf miners studied (Cornelissen & Stiling, 2005), and only mortality caused by predatory ants and pa rasitism caused by microhymenopterans were manipulated as the top-down factor in our systems. Ants were removed by placing masking tape covered with Tanglefoot around the base and twigs of assigned oak trees at the beginning of the experiments and by handpicking ants alrea dy present on the trees. Parasitoid removal was performed by addi ng between 12 and 15 yellow 13X8 cm Sticky Strip traps (Gempler’s, Belleville, Wisconsi n, USA) per assigned tree. These traps are plastic cards coated with Ta nglefoot adhesive and hymenopter an parasitoids are attracted to the yellow color and get stuck in the adhe sive, effectively reduci ng parasitism levels (Moon & Stiling, 2002). Previous studies conducted in salt marsh and hammock communities have shown the efficiency of yellow sticky traps to reduce parasitoid

PAGE 79

67 abundance and parasitism rate s of gall-formers and leaf -miners (e.g., Moon & Stiling, 2002; 2004). Traps were placed all around th e tree canopy using binder clips and replaced bi-weekly for Q. laevis and every three weeks for Q. geminata Old traps were individually wrapped in plasti c and refrigerated for furthe r analyses under a microscope, where the densities of all parasitic microhym enopteran and other i nvertebrate species were determined by counting the number of microhymenopterans and other invertebrates caught on the traps on three 4 X 4 cm quadrats on 2 traps per plant, at the beginning and at the end of the season (n= 240 sample s per treatment per plant species). To assess variation in host plant qua lity among treatments, tannin concentration and foliar nitrogen concentration were evalua ted monthly for each plant by haphazardly sampling 8 undamaged leaves from each tree, all around the canopy. Leaves were ovendried and milled to a fine powder and tannins were extracted from 50 mg of dry tissue. Tannin concentration was analyzed as foliar protein binding capacity using the radial diffusion assay with three replic ates per leaf (for details see Hagerman, 1987) and tannic acid was used as a standard. Nitrogen conten t was determined using a CE Instruments NC2100 CN Analyzer (CE Elantech, Inco rporated, Lakewood, New Jersey, USA). Variation in herbivore densiti es among treatments were quantified monthly on each plant by haphazardly counting the number of mines, chewed leaves, gall-formers, and other herbivores on 200 leaves of Q. laevis and 300 leaves of Q. geminata on each sampling date. All leaf miners including Acrocercops, Brachys, Stilbosis, Cameraria, Stigmella and Bucculatrix were identified and counted. Andricus galls and leaf rollers on Q. geminata were counted at the beginning and end of the season. To test for effects of treatments on herbivore mortality, we assessed survivorship of selected leaf miner species

PAGE 80

68 using 5 mines of each of the 3 most common leaf miner species, namely Acrocercops, Brachys, and Stilbosis Mines were permanently marked (n=50 mines per species per treatment combination) on each in dividual plant using a Sharpie pen as soon as the eggs hatched and larvae initiated mine forma tion. After mine termination, all leaves were inspected under a stereomicroscope for a ssessment of leaf miner survivorship and identification of mortality fact ors. Parasitized mines have ti ny circular exit holes left by the parasitoid on mine’s surface and/or pup ae within mine, and preyed upon mines are usually found ripped open and the larva is absent. Successfully emerged larvae of Acrocercops cut open the mines and pupate usually on the same leaf where mine developed, Brachys larvae pupate inside mines and/or cut circular holes on the underside of the leaf and Stilbosis mines cut semi-circular exit hole s on the mine underside towards the apex (Simberloff & Stiling, 1987; Stiling et al., 1999). Data Analysis All the variables analyzed were first subm itted to Lilliefor’s test for data normality and Levene’s test for homogeneity of variances. Transformations (angular, logtransformation and arcsine square rooted) we re employed when necessary to stabilize variances and normalize the data. To test fo r differences in plant nutritional quality among treatments two-way repeated measures Analysis of Variance (ANOVA) was used to assess effects of treatment on plant qual ity. Between-subject factors were fertilizer addition (yes, no), natural enemies exclusion (yes, no) and their in teraction, and time (6 sampling dates) was the within-subject fact or. The sphericity a ssumption of repeatedmeasure designs is less likely to hold for experiments with more than two-treatments and

PAGE 81

69 univariate F-tests were adjusted us ing the Greenhouse-Geisser epsilon ( )(Quinn & Keough, 2002). To test for differences in th e abundance of leaf miners and other herbivores among treatments we performed tw o-way multivariate analysis of variance (MANOVA) based on the mean density of each herbivore over the season, for each oak species separately. For both MANOVAs, F-te sts were based on values using Wilk’s lambda, provided by the GLM proce dure on SPSS 12.0.2 (SPSS, 2003). Significant differences in herbivore abundance among treatments detected with MANOVAs were further analyzed using two-way repeated-m easures ANOVA for each herbivore species, with F-tests based on type III sum of squares. Differences in leaf miner survivorship among treatments were tested using two-way ANOVAs. Because we detected strong bottom-up eff ects for most herbivores studied (see results), we examined the magnitude of fer tilization effects on the abundance of different herbivore species and different herbivore guilds, by calculati ng effect sizes using the log of the response ratio (Hedges et al., 1999). The response ratio is the ratio of some measure of outcome in the experimental group to that of the cont rol group (Rosemberg et al., 2000), and it has the advantage of estimati ng the effect as a proportionate change resulting from experimental manipulation. We obtained least square means and standard deviations by averaging monthly counts for each herbivore species and used the 10 plants per treatment as replicates to calculate eff ect sizes. We contrast ed herbivore abundances under the treatment +F,-P vs. herbivore a bundance under the treatment –F,-P, which gives effect size on herbivore densities of bottom-up effects in the absence of natural enemies. We calculated the natural log of the response ratio as lr=ln( X+F-P/ X-F-P), where X+F-P is the mean herbivore abundance on plants that were fertilized and traps were

PAGE 82

70 added and X-F-P is the mean herbivore abundance on plants that were trapped only. Negative proportional changes indicate a decr ease in herbivore abundance compared to trapped plants and positive values indicate an increase in abundance due to fertilization. We also contrasted herbivore abundances under the treatment –F,+P vs. –F,-P, which gives the difference in herbivore densities ow ing to predation in th e absence of a bottomup effect. However, because we did not dete ct significant effects of natural enemy exclusion on herbivore abundan ce on both oak species (see re sults), these analyses are not shown To estimate the cumulative effect size (E++) for the 10 replicates per treatment combination, effect sizes were combined using a weighted average (Rosemberg et al., 2000) and we used a mixed-model effect to examine the effects of herbivore guild and dietary breath in shaping responses to bo ttom-up effects in the absence of natural enemies. Herbivores were grouped into guilds (l eaf-miners, gall-formers, leaf rollers, and chewers) and further classified into specia lists and generalist ba sed on the literature. Acrocercops Brachys and Stilbosis mines, as well as Andricus galls were considered as specialists, whereas leaf rollers, chewers and eyespot galls were considered generalists. It is important to point out, howev er, that we are using the percentage of chewed leaves as a surrogate for the abundance of chewers on both oak species. We calculated the total heterogeneity (QT) as well as heterogeneity within (QW) and between groups (QB) and the significance of these statistics was evaluated using a Chi-squa re distribution. Effect sizes were considered significan t if bootstrapped (3,000 iterati ons) 95% confidence intervals did not overlap zero.

PAGE 83

71 RESULTS Treatment Effectiveness Application of Tanglefoot on assigned oak trees proved effective in keeping ants away from treated trees compared to c ontrol trees at the begi nning (control trees: 10.3 1.36 ants, treated trees: 0.71 0.12 ants, F1,28 = 87.1, P<0.001) and end of the season (control trees: 18.90 0.84 ants, treated trees: 1.13 0.33 ants, F1,28 = 91.3, P<0.001 ). Sticky traps also proved efficient in capturing microhymenopterans and other small invertebrates. Although individual para sitoid identification was not possible, we observed that microhymenopterans were much more abundant on the traps than other insects such as ants, ladybugs, and flies (microhymenopterans: 32.3 1.56, other invertebrates: 11.24 0.19) and no significant differe nces were observed in the abundance of parasitoids caught on traps from plants treated with traps only (29.66 0.75 parasitoids per trap) and plants that were fertilized and trapped (29.47 1.05, F1,38=0.22, P=0.74). However, the number of parasitoids caught per trap signifi cantly increased from the beginning (mean parasitoid number 29.47 0.73) to the end (mean parasitoid number 31.61 0.65) of the season (paire d t-test, t=-2.96, P=0.033). Treatment effects on host plant quality Fertilization tended to decrease tannin concentration for both species, but only for Q. geminata was this decrease significant (F3,36 = 3.07, P<0.05, Figure 4.1). For both plant species, we observed a significant effect of time on tannin concentration ( Q. geminata F5,180= 108.75, P = 0.0001; Q. laevis F5,180= 40.49, P = 0.0001), indicating a

PAGE 84

72 significant increase in tannin c oncentration from the beginnin g to the end of the season, but no time x treatment interaction was observed for either tree species ( Q. geminata : P=0.106 and Q. laevis : P = 0.370). Fertilization also signif icantly increased the amount of foliar nitrogen for Q. geminata by approximately 18.5% compared to control plants and by 10.1% for fertilized Q. laevis plants ( Q. geminata F3.36=46.27, P =0.0001, Q. laevis F3.36=3.05, P =0.041), but foliar nitrogen concentr ation significantly decreased over the season for both plant species (significant time effects, P = 0.0001 for both plant species, Figure 3.1). For Q. laevis there was also a significant time x treatment interaction (F4,12=6.08, P<0.001), indicating that changes in n itrogen concentrati on over the season varied among treatments. For both oak species, natural enemy rem oval did not affect host plant quality (natural enemies effect, all P>0.05) and no si gnificant interactions between fertilization and natural enemy exclusion were observed fo r the host plant parame ters investigated. Treatment effects on the abundance of le af miners and other herbivores Fertilization significantly increa sed the abundance of herbivores on Q. geminata trees (Table 4.1), although diffe rent species were impacted at different magnitudes. Stilbosis mines, for example, increased by approxim ately 5-fold on fertilized plants and by 4-fold on fertilized and trapped plants compared to control plants, whereas other mines such as Cameraria and Stigmella increased by approximately 2-fold on fertilized compared to control plants (Figure 4.2). Leaf rollers were th e only herbivores not significantly affected by fertiliz ation and chewers were the on ly herbivores significantly affected by top-down effects (Table 4.2). For all herbivores studied on Q. geminata we

PAGE 85

73 observed a significant time effect (all P = 0.0001) indicating seasonal changes in herbivore density. For Stilbosis and other mines, we also observed a significant time x bottom-up effects, indicating that the eff ects of fertilizer ch anged over the season. For herbivores on Q. laevis we also observed significant bottom-up effects on herbivore densities (Table 4.1) although univariate analyses indicated that only leaf miners increased in density with th e addition of fertilizer (Table 4.2). Brachys mines, for example, approximately doubled in density on fe rtilized or fertilized and trapped plants compared to control plants (Figure 4.3), and even stronger impacts of fertilization were observed for the second generation of Brachys mines on turkey oaks. Acrocercops mines increased by approximately 40% on fertilized compared to non-ferti lized plants, but the abundance of eyespot galls, chewers and othe r leaf miners were not affected by the addition of fertilizer (all P > 0.05). None of the herbivores on turkey oaks were affected by the removal of natural enemies or the combined effects of bottom-up and top-down manipulations (Table 4.2, all P >0.05), but we observed a significant time effect for all herbivores studied, indicating seasonal changes in herbivor e densities from April to September. Treatment effects on leaf miner survivorship Neither bottom-up nor top-down manipul ations significantly impacted mine survivorship, as no significant differences in the proportion of mines that survived and/or were killed by natural enemies (predators and parasitoids combined) were observed among treatments (Two-Way ANOVAs, all P>0.05) Survivorship rates of the leaf miners studied were relatively high (range 46.3 to 76.0%) and mortality caused by natural enemies was relatively low (range 3.5 to 24.0%), regardless of experimental

PAGE 86

74 manipulation. Parasitism rates for Brachys mines, for example, developing on control plants averaged 9.2 0.02%, whereas parasitism rate s on mines developing on trapped plants averaged 7.8 0.06%. For Acrocercops mines, predation rates by ants on mines developing in control plants averaged 30.1 0.08%, whereas predation by ants on mines developing on trapped plants averaged 24.6 0.09%. Strength of bottom-up forces Quantitative synthesis of our results using the log of the response ratio revealed strong and positive responses of herbivore de nsity to plant fertilization, although different responses were observed both between and w ithin guilds. Among leaf miners, stronger bottom-up effects were observed for Stilbosis mines (E++ = 1.32, bootstrapped CI= 0.837 to 1.83, Figure 4.4A), followed by mines induced by the second generation of Brachys Grouping herbivores into guilds revealed that bottom-up manipulations caused changes in density that significantly differed among guilds (QB=27.97, P = 0.0001) and increased abundance caused by fertilization was stronge r for leaf miners (E++=0.780, bootstrapped CI = 0.406 to 0.946) compared to other herb ivore guilds. Leaf rollers were not significantly affected by fer tilization and weakest bottom-up effects were observed for chewers (Figure 4.4B). Specialists were significantly more strongly influenced by bottom-up effects than generalists (QB=34.09, P = 0.0001), as specialists were 95.8% more abundant on plants that were ferti lized and trapped (E++=0.9582, bootstrapped CI = 0.661 to 1.25, n=45), whereas generalists in creased in abundance by only 28.7% on the same experimental plants (E++ = 0.2 877, bootstrapped CI = 0.122 to 0.47, n=63).

PAGE 87

75 Table 4.1Results from multivariate analyses of variance for mean herbivore density on Q. geminata and Q. laevis F-tests were based on Wilk’s lambda and bottom-up effects refer to fertilization and top-down effects refer to natural enemy removal. Bottom-up x Top-down refers to the combined effects of fertilization and na tural enemy exclusion. Plant species Source df Wilk’s F P Q. geminata Bottom-up 6,31 0.266 14.25 <0.001 Top-down 6,31 0.768 1.56 0.191 Bottom-up x Topdown 6,31 0.692 2.29 0.06 Q. laevis Bottom-up 5,32 0.327 13.19 <0.001 Top-down 5,32 0.981 0.123 0.986 Bottom-up x Topdown 5,32 0.867 0.982 0.444

PAGE 88

76 Table 4.2Effects of treatments (bottom-up and top-down ma nipulations) on herbivore abundance on Q. geminata and Q. laevis Bottom-up effect refers to fertilizer addition, top-down effect refers to natural enemy rem oval and top-down x bottom-up effect refers to combined effects of fertili zation and natural enemy exclusion. Time effect refers to subsequential samplings ( Andricus galls and leaf rollers on Q. geminata were counted only at the beginning and end of season and analyzed using a Two-Way Anova, with no time effect). Table shows (Greenhouse-Geisser Epsilon) corrected P values for withinsubject factors and their interactions. Herbivore sp Bottom-up effect Top-down effect Bottom-up x Top-down effect Time Time x bottom-up effects Time x topdown effects Q. geminata F1,36 P F1,36 P F1,36 P F5,180 P F5,180 P F5,180 P Stilbosis 73.63 0.001 0.65 0.425 2.01 0.165 113.93 0.001 37.13 0.0001 0.54 0.599 Leaf rollers 0.77 0.384 1.08 0.304 2.32 0.137 Eyespot galls 17.51 0.001 0.002 0.960 1.55 0.220 48.80 0.001 0.41 0.615 1.04 0.343 Andricus galls 8.69 0.006 0.39 0.845 5.80 0.021 Chewed Leaves 4.20 0.048 5.05 0.031 5.54 0.024 52.49 0.001 0.95 0.384 1.08 0.337 Other mines 5.37 0.026 1.71 0.200 2.99 0.092 49.12 0.001 3.32 0.041 1.13 0.333 Q. laevis F1,36 P F1,36 P F1,36 P F5,180 P F5,180 P F5,180 P Brachys 1st 15.20 0.0001 0.009 0.978 1.57 0. 217 136.94 0.0001 14.30 0.0001 1.53 0.227 Brachys 2nd 21.30 0.0001 0.17 0.682 0.12 0. 728 78.94 0.0001 17.37 0.0001 1.27 0.281 Acrocercops 7.25 0.011 0.19 0.666 0.66 0.422 93.58 0.0001 0.83 0.376 0.013 0.920 Eyespot galls 1.12 0.298 0.34 0.564 0.55 0.465 71.40 0.0001 4.47 0.018 4.30 0.020 Chewed leaves 1.95 0.170 0.15 0.700 0.99 0.326 3.66 0.038 0.05 0.925 0.16 0.813 Other mines 0.11 0.980 0.12 0.734 0.28 0.602 31.44 0.0001 0.26 0.697 0.54 0.523

PAGE 89

77 Figure 4.1Treatment effects on the concentrati on of tannins and foliar nitrogen of Quercus geminata (left panels) and Q. laevis (right panels) over the season. See text for explanation of treatment symbols. Tannin concentration (mg/g) 0.30 0.35 0.40 0.45 0.50 0.55 Control +F, +P -F, -P +F, -P MayJuneJulyAugustSeptemberOctober Tannin concentration (mg/g) 0.05 0.10 0.15 0.20 0.25 0.30 Control +F, +P -F, -P +F, -P AprilMayJuneJulyAugustSeptember Foliar Nitrogen (%) 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Control +F, +P -F, -P +F, -P JuneJulyAugustSeptemberOctober Foliar Nitrogen (%) 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Control +F, +P -F, -P +F, -P MayJuneJulyAugustSeptember

PAGE 90

78 Figure 4.2Treatment effects on the abundance of different herbivores guilds on Q. geminata Bars ( 1SE) show average of monthly c ounts for all herbivores belonging to the same guild. Density (counts per 300 leaves) 0 5 10 15 20 25 Density(counts per 300 leaves/stems) 0 10 20 30 40 50 60 Density (counts per 200 leaves) 0 2 4 6 8 10 Chewed leaves (%) 0 20 40 60 80 100 Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-PControl+F,+P-F,-P+F,-P Gall-formersLeaf-miners ChewersLeaf-rollers

PAGE 91

79 Figure 4.3Treatment effects on the abundance of different herbivores guilds on Q. laevis. Bars ( 1SE) show average of monthly count s for all herbivores belonging to the same guild. Density (counts per 200 leaves) 0 5 10 15 20 25 Density (counts per 200 leaves) 0 5 10 15 20 25 Chewed leaves (%) 0 10 20 30 40 50 60 Gall-formersLeaf-miners Chewers Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P

PAGE 92

80 Figure 4.4A) Strength of bottom-up manipulati ons on the abundance of herbivores on both Q. geminata and Q. laevis in the absence of natural enemies. Bottom-up effects refer to fertilizer addition and herb ivore densities under the trea tments –F,-P and +F,-P were contrasted. The cumulative effect size (ln ra tio) is reported for each herbivore with its 95% confidence interval ( Q.g = Quercus geminata and Q.l = Quercus laevis ); B) Strength of bottom-up manipulations according to herbivore feeding guild. Numbers in parentheses indicate the number of independe nt comparisons for each effect and effects are significant if confidence inte rvals do not overlap with zero. Proportional change in density -0.4-0.20.00.20.40.60.81.01.21.41.61.8 Acrocercops Brachys 1st gen Brachys 2nd gen Stilbosis Eyespot on Ql Eyespot on Qg Andricus Leaf rollers Chewers on Qg Chewers on Ql Other mines on Qg Other mines on QlA. Proportional change in density -0.4-0.20.00.20.40.60.81.01.2 Leaf miners Gall formers Leaf rollers Chewers (53) (26) (9) (20) B.

PAGE 93

81 DISCUSSION Using field experiments, the abundance of different herbivore guilds and the survivorship of oak leaf miners were comp ared among different combinations of natural enemy pressure and host plant quality. Our re sults show that fertilization successfully increased the availability of nitrogen for he rbivores feeding on both plant species. We also observed a decrease in tannins on fe rtilized trees, which has been previously demonstrated for oaks (e.g., Forkner & Hunt er, 2000) and for other unrelated plant species (Waring & Price, 1988; reviewed by Koricheva et al., 1998, Haukioja et al., 1998). Overall, fertilization tended to increase the quality of oak foliage in our studied systems, and variation in host quality among a nd within plants has been demonstrated to affect different herbivore guilds (reviewed by Waring & Cobb, 1992; Kyto et al. 1996). Higher host quality of ferti lized plants offer support for the higher densities of leaf miners and most other herbivores on fert ilized compared to c ontrol plants. Increases in the biomass and nutritional quality of fertilized vegetation may enhance the overall abundance of herbivorous insects and, as a consequence, could result in increased densities of natural enemies and increased rates of predation a nd parasitism. In our systems, however, natural enemy pressure did not increase with productivity, as no significant differences in the proportion of mines killed by na tural enemies were observed among treatments. Also, analyses of natura l enemies caught on sticky traps showed no difference in the abundance of microhymenopt erans on plants that were trapped and plants that were fertilized and trapped. Th ese results are contrary to other studies in which predation and parasitism rates were hi gher in fertilized pl ants (e.g., Stiling &

PAGE 94

82 Rossi, 1997; Moon & Stiling, 2004), but in agreem ent with other stud ies with oaks (e.g., Forkner & Hunter, 2000) and are in accordance with the suggestion that top-down factors such as natural enemy pressure might have weak impacts on the abundance and performance of the oak herbivores studied. Contrary to our results, other field st udies have shown the strong effect of parasitoids in leaf miner su rvivorship and performance (Auerbach & Simberloff, 1988; Hawkins, 1994). In a factorial manipulation of plant quality and parasitism pressure for five different plant species and 15 herbi vores including 7 leaf miner species, Moon & Stiling (2004) found that fert ilization strongly in creased the density of the most common herbivores and fertilization also increased the effects of parasitism for herbivores of two of the host plants studied. Despite the evid ence for the importance of natural enemies on herbivore abundance and performance (reviewed by Hawkins et al., 1997), removing predators and parasitoids in our systems di d not strongly increase herbivore abundance on oak trees. Low rates of egg parasitism and high rates of egg survivorship have been demonstrated for some of the leaf miners studied (e.g., Mopper et al., 1984; 1995) and negligible effects of predator s and parasitoids on egg mortality might be responsible for the absence of a positive response by herbivor es to the removal of natural enemies. Cornell & Hawkins (1995) reviewed the evid ence for survival patterns and mortality sources of herbivorous insects and found th at natural enemies emerged as the most frequent source of mortality, although endophytic species such as leaf miners and gall formers sustained higher survival rates in th e latest stages than free-feeding herbivores. Additionally, although enemies were the most fre quent cause of death at all life stages, their effects are smallest in the egg and early larval stag es (Cornell & Hawkins, 1995;

PAGE 95

83 Price, 1997). Stilbosis females, for example, deposit eggs on the lower leaf surface, at the junction of a midvein and ma jor lateral vein. Mopper et al. (1995) have shown that Stilbosis eggs and early stage larvae usually suffer little natural enemy attack, resulting in low mortality rates, as females place eggs among the dense trichomes on the ventral leaf surface. For Brachys mines, eggs are deposited on the upper surface of turkey oak leaves, but secured and protected by a waxy secre tion (Waddell & Mousseau, 1996). Data on mortality rates of Acrocercops eggs caused by predation or parasitism are not available, but fast egg hatching and fast larval development might reduce the window of vulnerability to the attack of natural enemies. Alternatively, although application of Tanglefoot and use of sticky traps have been efficient in reducing parasitism rates in other plant-herbivore-natural enemy systems, it might have had limited efficiency in our oak trees, especially for leaf miners. The fact that the percentage of chewed leaves significantly increased in plants that were fertilized and natural enemies were excluded suggest that reduction of ant abundance might have influenced the abundance of chewers on these plants, increasing leaf consumpti on, as demonstrated by other studies (e.g., Marquis & Whelan, 1994). Moran & Scheidler (2002) manipulated both top-down and bottom-up forces in a successional plant community and observed that, although fertilizer addition caused changes in plant biomass, many herbivore species were unaffected by experimental manipulation. This result stresses the im portance of examining many components of communities when addressing trophic interactions, as some species might not strongly respond to either bottom-up or top-down proces ses, as suggested by studies in terrestrial diverse communities such as successional fields (Moran & Scheidler, 2002) and

PAGE 96

84 temperate forests (Forkner & Hunter, 2000). Our results are in accordance with this suggestion, as the strength of bottom-up forces varied among and within guilds. Speciesspecific and differential guild responses to bottom-up and top-down manipulations might be explained by differences in life-history tr aits, such as growth and development, of each herbivore and leaf miner species studied. While our results show that the strength of bottom-up forces varied among and within guilds, bottom-up effects on herbivore abundance were usually strong, and strongest effects of bo ttom-up manipulations were observed for leaf miners and gall-formers. In a similar analysis for herbivores inhabiting Spartina marshes, Denno et al. (2003) observed that the re lative effect of nitrogen addition was greater than the impact of spid er predation on 5 out of 6 sap-feeders and, overall, bottom-up effects dominated over topdown impacts. Our results also showed that amongst the leaf miners, Stilbosis exhibited the greatest change in abundance caused by bottom-up effects such as increased plant qua lity. This particular mine species may be most likely to be affected by host quality because Stilbosis mines have the longest developmental times, full depth mines, a nd, therefore, a highe r likelihood of being affected by spatial and seasonal vari ation in host plant quality. Our study has simultaneously manipulated top-down and bottom-up factors for a community of herbivores on scrub oaks and our analyses of the relative strengths of plant quality and natural enemies on herbivore a bundance and performance have shown that bottom-up forces dominate our oak-herbivores-natural enemies system and top-down effects such as the impact of predators and parasitoids do not significantly impact herbivore abundance and the performance and survivorship of leaf miners.

PAGE 97

85 Chapter 5 DOES LOW NUTRITIONAL QUALITY ACT AS A PLANT DEFENCE? AN EXPERIMENTAL TEST OF THE SLOW-GROWTH, HIGH-MORTALITY HYPOTHESIS SYNOPSIS The slow-growth-high-mortality hypothesis was experimentally tested in this study by investigating the eff ects of plant quality and na tural enemies on leaf miner growth, performance and survi vorship. Two leaf miners ( Acrocercops albinatella and Brachys tesselatus ) occurring on the turkey oak Quercus laevis were studied using a factorial design that manipulated predation/parasitism pressure and plant nutritional quality. Forty trees were randomly divided in to four treatments: 1) control plants (nutrients and natural enemies unaltered); 2) nutrients added, natural enemies unaltered; 3) nutrients unaltered, natural enemies reduc ed, and 4) nutrients added and natural enemies reduced. Water content, leaf toughness, tannin concentration, and foliar nitrogen were quantified monthly for each plant and mi ne growth and survivorship were assessed by tracing mines on a 2 to 3-day interval and by following the fates of 50 mines per species per treatment combination. Fertili zed plants exhibite d significantly higher amounts of nitrogen, but no significant diffe rences among treatments were observed for water content, leaf toughness and tannin c oncentration. These resu lts only partially support the slow-growth-high-mortality hypothesi s, as mines were significantly smaller and developed faster on fertil ized plants, but neither fe rtilization nor natural enemy

PAGE 98

86 exclusion significantly affected mine surv ivorship or mortality caused by natural enemies. INTRODUCTION Resources and natural enemies, as well as the interaction between them, have the potential to impact herbivore abundance, dist ribution, performance a nd survivorship. The effects of both resources and natural enemie s on herbivory rates e xperienced by plants have been widely discussed (e.g., Moran & Hamilton, 1980; Augner, 1995; Williams et al ., 2001) and low plant quality can affect insect performance directly, by reducing growth rate, fecundity and survival (Slansky, 1993; Haggstrn & Larsson, 1995) or indirectly by affecting the risk of morta lity caused by natural en emies (Feeny, 1976; Price et al., 1980; Clancy & Price, 1987). The interac tion between variati on in host plant quality and risk of attack by natural enemies was formalized into the slow-growth, highmortality hypothesis (hereafte r SGHMH; Clancy & Price 1987). According to this hypothesis, herbivores feeding on plants of low nutritional quality (e.g., low nitrogen, high levels of secondary compounds, high toughness and/or lignin content) do not necessarily increase damage on their host by overcompensatory feeding if increased development time due to poor host quality in creases the window of vulnerability of herbivores to natural enemy attack. Specifically, the SGHMH proposes that the nutritional quality, allelochem istry, and/or morphology of th e host plant can act as a sublethal plant defence by prolonging de velopment of herbivorous insects and subsequently increasing mortality inflicte d by predators and parasitoids (Benrey &

PAGE 99

87 Denno, 1997). Clancy and Price (1987) wrote, “many more individual case studies are needed to determine if the SGHMH should be generally rejected or accepted”. Since then, relatively few tests of the SGHMH have been performed, with mixed results. In a factorial manipulation of plant quality a nd predation pressure, Loader and Damman (1991) found that Pieris rapae larvae growing on low-nitr ogen plants developed more slowly and were more likely to be killed by predators than consp ecifics growing on highnitrogen plants. Parasitism rates, however, were higher in herbivores feeding on highnitrogen plants. For galling sawflies on a rroyo willows, Clancy and Price (1987) also observed higher parasitism rates in fast-developing Pontania galls and slow-growing galls were less vulnerable to attack from parasitoids. Benrey and Denno (1997), on the other hand, observed that sl ow-developing larvae of Pieris rapae were more heavily parasitized by Cotesia glomerata than fast-developing larvae r eared on artificial diets, but slow-growth did not translate into increased parasitism when variation in larval growth was achieved with the use of natural variati on of plant quality caused by interspecific differences. Williams (1999) reviewed the evidence for the SGHMH, and found that, usually, slow-growing, surface-feeding herbivores were at less risk from parasitism but greater risk from predation, than faster-g rowing herbivores. For concealed herbivores, such as gall-formers and leaf miners, the S GHMH was rejected in approximately half of the cases reviewed, regardless of natural en emies being predators or parasitoids. Most of the data used to test the SGHM H has been observational in nature, and only a few studies have experimentally ma nipulated aspects of plant quality that potentially affect herbivore growth and development and mi ght consequentially impact natural enemy attack (see Loader & Da mman, 1991; Benrey & Denno, 1997; Lill &

PAGE 100

88 Marquis, 2001). Furthermore, among the 41 studies reviewed by Williams (1999), only five studies were conducted with leaf-miners, and none of thes e studies were specifically designed to test the SGHMH, as most of them assessed leaf miner survivorship under natural variations in host plant quality (i.e., differences in host plant species, within-host variation in quality, effects of plant hybrids) and none of these studies manipulated natural enemies to assess their effects on le af miner survivorship and performance. The current study aimed to test the SGHMH for tw o very distinct speci es of leaf miners, Acrocercops albinatella (Lepidoptera: Gracillaridae) and Brachys tesselatus (Coleoptera: Buprestidae), feeding upon the turkey oak Quercus laevis (Fagaceae), with a factorial manipulation of both plant quality and natura l enemy pressure. Mani pulation of the third trophic level was achieved by reducing invert ebrate predators and parasitoids and manipulation of plant nutritional quality was achieved by plant fertilization. Specifically, the following predictions were tested: 1) Low tissue quality cause herb ivores to feed for longer periods of time and lengthens larval development time. Therefore, leaf miners feeding on fertilized plants should exhibit sm aller mines and shorter development times than leaf miners on control plants; 2) Lengthened development time or increased feeding results in increased mortality caused by natura l enemies. Therefore, leaf miners feeding on fertilized plants should exhibit high er survivorship/lower mortality caused by predators and parasitoids than leaf miners on contro l plants; 3) If natural enemies are a significant source of mortality for leaf miners on turkey oaks, mines feeding on plants in which natural enemies were reduced should exhibit higher survivorship than mines on control plants. Because both plant quality and natural enemies were manipulated in the factorial design, here it was also predicted that 4) leaf mine rs feeding on fertilized plants

PAGE 101

89 in which natural enemies were excluded, s hould exhibit the highest survivorship amongst all treatments. STUDY SYSTEM The turkey oak, Quercus laevis is one of the characteristic trees associated with the sand hill community throughout Florida. Q. laevis is a moderately fast to fast-growing tree and presents deciduous glab rous leaves, alternately arrang ed with usually five lobes, although this number may vary from thr ee to seven (Nixon, 1997). Although a common tree in Florida native vegetati on, there are relatively few studies concerning herbivory in this plant species. Acrocercops albinatella (Lepidoptera: Gracillaridae) is a microlepidopteran species whose larval st ages feed on young leaves creating distinct linear-blotch mines on the lower surface of Q. laevis leaves. Larvae typically appear in early spring (late February or early March) and feed on the palisade parenchyma cells, completing their development in approximately ten days. Larvae emerge from the blotch mine and usually pupate on the same leaf from which they emerge (T. Cornelissen, pers. obs.). Brachys tesselatus (Coleoptera: Buprestidae) is a beetle species that also forms distinct blotch mines in Q. laevis leaves. The adults emerge from mid-March to midApril, coinciding with budburst of turkey oa k. Adults initially feed on the early leaves until mating and oviposition. Eggs are deposited singly on the upper surface of the leaves and after hatching the larvae mine into the mesophyll creating distinct, characteristic damage. Contrary to what happe ns in South Carolina (Waddell et al ., 2001) Brachys in the study sites here go through two generations instead of just one. The first mines appear in early April and remain active unt il late June, when larvae complete their

PAGE 102

90 development and pupate inside the mines. New adults emerge in early July and oviposit to form new Brachys mines that remain active until September-October. Pupation and overwintering of this second ge neration occurs within the leav es after they have senesced and abscised from the tree. New adults emerge from the leaf litter in the following spring (Waddell et al ., 2001). Turkey oak leaves are also attacked by many other herbivores, such as the l eaf roller weevil Homoeolabus analis (Coleoptera: Atellabidae), the eastern buck moth Hemileuca maia the white-marked tussock moth Orgyia leucostigma (Lepidotpera: Lymantriidae) and, less commonly, other leaf miners such as Stigmella spp. (Lepidoptera: Nepticulidae) and Cameraria spp. (Lepidoptera: Gracillariidae). Generalist predatory ants (Hymenoptera : Formicidae) and the green lynx spider Peucetia viridans (Araneae: Oxyopidae) are among the most common arthropod predators in these study sites. The leaf mi ners studied are also attacked by several microhymenopteran parasitoid species, including Zagrammosoma multilineatum (Hymenoptera: Eulophidae) and Chrysonotomyia sp. (Hymenoptera: Eulophidae). METHODS Data collection This study was conducted between Febr uary and November of 2003 on natural stands of scrub oak vegetati on in Tampa, Florida. Forty Quercus laevis trees ranging between 1.5 and 2.5 meters in height were mark ed at an unburned patch at the University of South Florida ECOAREA (for a descri ption of the study s ite see Mushinsky et al., 2003).

PAGE 103

91 The effects of plant quality and pres sure exerted by natural enemies on the abundance, performance, and survivorsh ip of two different leaf miners ( Acrocercops albinatella and Brachys tesselatus ) on Q. laevis were investigated us ing a 2 X 2 factorial design. The forty trees were randomly divi ded into four treatments, allowing ten replicates of each treatment, as follows: 1) –F,+P: control plants with nutrients and natural enemies unaltered ; 2) +F,+P: nut rients added, natural enemies present (unaltered); 3) –F,-P: nutri ents unaltered, natural enemie s reduced, and 4) +F,-P: nutrients added and natural enemies redu ced. Plant quality was manipulated by the addition of 150 g of 46:0:0 NPK fertilizer to assigned Q. laevis trees bi-weekly from April to June. Unfertilized plants had so il around the tree slightly disrupted, but no granular fertilizer was added. Plants did not significantly di ffer in height, number of leaves on ten shoots and leaf area (five leaves per tree) before treatments were assigned (One-Way ANOVAs, all P > 0.05). A previous study with the same system indicated that predation by spiders was a neg ligible mortality factor for th e leaf miners (Cornelissen & Stiling, 2005), hence only mortality caused by predatory ants and parasitism caused by microhymenopterans were manipulated. A pr eliminary count of green spiders on all marked plants in March of 2003 showed that spider abundance was very low (average 0.8 0.2) and did not differ among plants assi gned to the four treatments. Ants were excluded by placing masking tape covered with Tanglefoot around the base and twigs of assigned oak trees at the beginning of the experiments and by handpicking ants already present on the trees. Efficiency of ant tr apping was checked monthly by counting ants during five minutes per tree on all treatment combinations. Tanglefoot was reapplied on assigned turkey oak trees in May 2003. Pa rasitoid reduction was achieved by adding

PAGE 104

92 between 12 and 15 yellow 13X8 cm Sticky Strip traps (Gempler’s, Belleville, Wisconsin, U.S.A.) per assigned tree. Previous studies with gall-formers and leaf miners have shown that yellow traps are efficient at capturing pa rasitoids in other systems such as salt marshes and oak hammock communities (e.g., Moon & Stiling, 2002, 2004). Traps were placed throughout the tree canopy using binder clips and were replaced bi-weekly. Old traps were collected and indivi dually wrapped in plastic and frozen for further analyses. To assess sticky trap effi ciency, the number of mi crohymenopterans and other invertebrates caught on the traps were counted on three 4 X 4 cm quadrats on two traps per plant at the beginning and at the end of the season. To assess variation in host plant quality among treatment s, water content, foliar nitrogen concentration, tanni n concentration and leaf t oughness were evaluated monthly for each plant, between April and Septem ber of 2003. On each sampling date, eight undamaged leaves were collected from each tree and placed immedi ately on ice. Leaf toughness was evaluated using an Effegi FT011 penetrometer (International Ripening Co, Italy) and water content was quantified by the difference between leaf wet and dry weights. Leaves were further oven-dried a nd milled to a fine powder. Tannins were extracted from 50 mg of dry tissue, and tannin concentra tion was quantified using the radial diffusion assay with tannic acid as a standard (for details see Hagerman, 1987). Nitrogen content was determined using a CE Instruments NC2100 CN Analyzer (CE Elantech, Incorporated, Lakewood, New Jersey, U.S.A.). To test for differences in mine growth and survivorship among treatments, five mines of each leaf miner species (n = 50 mi nes per species per tr eatment combination) were permanently marked on each plant using a permanent black ink marker as soon as

PAGE 105

93 the eggs hatched and larvae initiated mine formation. Acrocercops mines were measured at 2-day intervals by tracing the num bered mines using acetate sheets. Brachys mines were traced at 3-day intervals. At the e nd of the experiment, mine drawings were digitized and mine size was measured us ing the software UTHSCSA Image Tool (University of Texas, USA), with digital pictures calibra ted to the nearest 0.01mm. We compared final mine size (cm2), developmental time (days to pupation) and mine growth rate among treatments for each leaf miner sp ecies. Mine growth rate on each leaf was calculated as Growth rate = (Final mine size – Initial mine size) / number of days mine was growing. After mine termination, all leav es were inspected under a stereomicroscope to assess leaf miner survivorship and identifi cation of mortalit y factors. Leaf miners offer a great opportunity to assess population survivor ship and mortality factors since a record of the miner success is clearly observed on the leaves: parasitized mines have tiny circular exit holes on mine’s surface and/or pupae within the mine, and predated mines are usually found ripped open and the larva is absent. Successfully emerged larvae of Acrocercops cut open the mines and pupate usually on the same leaf where the mine developed. Brachys larvae pupate inside mines and/or cut circular holes on the underside of the leaf. Data analysis All the variables analyzed were first submitted to Lilliefor’s test for data normality and transformations (log-transformation and arcsine square rooted) were employed to stabilize variances and normalize the data. However, for the sake of clarity, figure axes and means (+1SEM) show untransfo rmed data. Seasonal variation in turkey

PAGE 106

94 oak quality was described elsewhere (Corne lissen & Stiling, 2006b) and on this study only variation in plant quality among treatments is emphasized. To test for differences in plant nutritional quality among treatments, we first averaged tannin concentration, nitrogen content, water, and toughness using all eight leaves sampled for each plant on each collection date and two-way ANOVAs were used to assess effects of treatment on host plant quality. A paired t -test was used to assess differences in leaf area of Q. laevis before and after fertilization and Pears on correlations were used to examine the relationship between tannin and nitrogen c oncentration and between water and toughness on each individual plant. To test for differen ces in leaf miner growth and survivorship among treatments, differences in mine size, days to pupation, and growth rate of mines that survived to pupation were analyzed us ing a Two-Way ANOVA with fertilization and natural enemies as main factors. Differentia l survivorship among treatments as well as the proportion of mines killed by natural enemies (parasitoids and predators) and proportion killed by unknown fact ors (e.g., plant resistance, larvae dead inside mine) on each treatment were also analyzed using a Two-Way ANOVA. All statistical analyses were performed using Systat 9.0 for Windows (Wilkinson, 1999). RESULTS Treatment effectiveness Application of Tanglefoot on assigned turkey oak tress proved effective in keeping ants away from treat ed trees compared to cont rol trees (cont rol trees: 14.6 1.1 ants, treated trees: 0.92 0.22 ants, F1,28 = 137.6, P < 0.001). Sticky traps also proved

PAGE 107

95 efficient in capturing microhymenopter ans and other small invertebrates. Microhymenopterans were much more abundant on the traps than other insects such as ants, ladybugs, and flies (microhymenopterans 32.3 1.56, other invertebrates 11.24 0.19). No significant differences were observe d in the abundance of parasitoids caught on traps from plants treated with traps only (29.66 0.75 parasitoids per trap) versus plants that were fertilized and had sticky traps added (29.47 1.05, F1,38 = 0.22, P = 0.74). Treatment effects on host plant quality Fertilization significantly increase d the amount of foliar nitrogen on Q. laevis trees ( F1,36 = 8.993, P = 0.005) and fertilized trees s howed a tendency for decreased tannin concentration (Figure 5.1), although diffe rences in tannins among treatments were not statistically significant (Fertilizer effect: F1,36 = 2.82, P = 0.092). No significant differences among treatments were observed for leaf toughness and leaf water content (all P > 0.05). Linear regressions revealed that, for Q. laevis no significant relationship was observed between tannin and nitrogen c oncentration per individual plant ( P = 0.361), or between water and leaf toughness ( P = 0.688). Paired t -tests revealed no significant difference in leaf area before and after fertilization ( t = -0.778, P = 0.441). As expected, natural enemy removal did not affect host plant quality (natural enemies effect, all P > 0.05) and no significant interac tions between treatments were observed.

PAGE 108

96 Treatment effects on leaf miner performance Fertilization significantly decreased mine size of Acrocercops ( F1,126 = 12.71, P = 0.001; Fig. 5.2a) and of both generations of Brachys (1st generation F1,96 = 21.74, P < 0.0001; 2nd generation: F1,66 = 4.85, P = 0.031; Fig. 5.2d). Analysis conducted only for mines that survived showed that Acrocercops mines on fertilized plants were approximately 56% smaller than mine s growing on control plants, whereas Brachys mines on fertilized plants were approximatel y 16% smaller than mines on control plants. Removal of natural enemies did not significantly impact the size of Acrocercops mines ( F1,126 = 2.97, P =0.09) or the size of Brachys mines (1st generation: F1,96 = 2.23, P = 0.138; 2nd generation: F1,96 = 31.30, P = 0.583). Fertilization also significantly affected the development of leaf miners. Acrocercops mines growing on fertilized plants pupated earlier ( F1,126 = 9.69, P = 0.002) and developed faster ( F1,126 = 16.74, P < 0.001) than mines growing on control plants (Fig. 5.2b, 5.2c). For the first generation of Brachys mines, significant effects of fertilizer on performance was also observed, as mines required fewer days to pupate ( F1,96 = 4.19, P = 0.043, Fig. 5.2e) and developed faster ( F1,96 = 5.83, P = 0.018, Fig. 5.2f) than mines developing on control plants. For the second generation of Brachys mines, fertilization significantly decreased the number of days required to pupation ( F1,66 = 20.15, P = 0.001), but no significant effects were observed on mine growth rate ( F1,66 = 2.35, P = 0.07). No significant interactions between fer tilization and removal of natural enemies were observed for either leaf miner (all P > 0.05). In general, high survivorship was observe d for all the leaf miners studied (Fig. 5.3) and for both Acrocercops and Brachys, survivorship was not affected by

PAGE 109

97 experimental manipulation (fertilizer and natural enemy effects: all P > 0.05). For Acrocercops however, miners experienced lower mortality inflicted by predators when developing on fertilized plants compared to control and/or trapped plants ( F1,36 = 12.72, P < 0.001), but no significant effects of fertilizer were observed when natural enemies were parasitoids ( F1,36 = 2.34, P = 0.159). For both generations of Brachys no significant effects of fertilizer addition or natural enem y removal were observed on the proportion of mines killed by predators or parasitoids (all P > 0.05).

PAGE 110

98 Figure 5.1Treatment effects on the concentr ation of foliar nitrogen, tannin concentration, leaf wate r and foliar toughness of Quercus laevis Bars shown mean (+1SE) of averaged monthly samplings (Con trol: plants with nutrients and natural enemies unaltered; +F,+P: nutrients added, na tural enemies present (unaltered); –F,-P: nutrients unaltered, natural enemies reduced; +F ,-P: nutrients adde d and natural enemies reduced). Tannin concentration (mg/g) 0.00 0.05 0.10 0.15 0.20 0.25 Foliar nitrogen (%) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Water content (mg) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Leaf toughness (lbs/mm2) 0.0 0.1 0.2 0.3 0.4 0.5 Control+F,+P-F,-P+F,-PControl+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P

PAGE 111

99 Figure 5.2Treatment effects on the size and development of the leaf miners Acrocercops albinatella and Brachys tesselatus For Brachys bars to the left indicate data for the first generation and bars to the right indicate data for the second generation. Bars show means ( 1SE) and treatment le gends are as on Fig. 1. b. c. Days to pupation 0 2 4 6 8 Mine growth rate 0.0 0.2 0.4 0.6 0.8 A. b. c. a. Mine size (cm2) 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Control +F,+P -F,-P +F,-P Control +F,+P -F,-P +F,-P Control +F,+P -F,-P +F,-PAcrocercops a. Acrocercops Acrocercops Mine size (cm2) 0 5 10 15 20 25 30 Days to pupation 0 10 20 30 40 Control+F,+P-F,-P+F,-P Control+F,+P-F,-P+F,-P Mine growth rate 0.0 0.2 0.4 0.6 0.8 Control+F,+P-F,-P+F,-Pd. e. f. Brachys Brachys Brachys

PAGE 112

100 Figure 5.3Frequency of occurrence of mortality f actors for leaf miners growing under four different treatments. Bars show means ( 1SE) and treatment legends are as in Fig.1. Acrocercops Frequency of occurrence 0.0 0.2 0.4 0.6 0.8 1.0 survived preyed upon plant resistance parasitized unknown factorsAcrocercops Frequency of ocurrence 0.0 0.2 0.4 0.6 0.8 Brachys 1st generation Control +F +P -F, -P +F -P

PAGE 113

101 DISCUSSION Herbivorous insects often suffer reduced growth rates when feeding upon suboptimal hosts and, although poor diets are not lethal in them selves, it is assumed that they nevertheless increase the mortality ra te by prolonging the vulnerable stages of herbivores (Rostas & Hilker, 2003), which might lead to higher levels of enemy attack (Moran & Hamilton, 1980; Clancy & Price, 1987). Plant nitrogen concentration influences important intera ctions between herbivorous insects and plants (Kyto et al., 1996) and low nitrogen supplies might result in increased total plant consumption through increased consumption rates and/or pr olonged periods of feeding, digestion, and development (Mattson, 1980). Kyto et al. (1996) reviewed the evid ence for the effects of soil fertilization on phytophagous insects and concluded that enhanced nitrogen availability benefited herbivores by impr oving host plant quality, but also affected population-regulating processes such as pr edation, parasitism, and competition. Our results have demonstrated the be neficial effects of plant fertil ization in terms of increased availability of foliar nitrogen and a tendenc y for decreased concentration of tannins. Previous studies in the same oak system i ndicated that several herbivore guilds respond to improved plant quality cause d by fertilization with an incr ease in density (Cornelissen & Stiling, 2006b). In this study, it has been shown that leaf miners also respond to improved plant quality by altering developmen t and reducing the leng th of the larval stage. Mines were significantly smaller on fe rtilized plants, developed faster and pupated earlier than mines growing on control plants. Similar respons es to improved plant quality have been reported for othe r leaf miners (e.g, DeBruyn et al., 2002,), gall-formers (e.g.,

PAGE 114

102 Clancy & Price, 1987; Stiling & Moon, 2005) and free-feeding herbivores (e.g., Myers, 1985; Loader & Damman, 1991). Although the experiments here have dem onstrated that enhanced nutritional quality caused by plant fertilization allowed mine s to develop faster and pupate earlier, no significant effects of improving plant quali ty and/or removing natural enemies were observed on mine survivorship and mortal ity imposed by the third trophic level. Therefore, the shorter window of larval vul nerability di d not translate into higher survivorship or escape from natural enem y pressure. Interestingly, these results demonstrated high survivorship rates for both leaf miners studied, regardless of experimental treatments. Other studies with leaf miners occurring in oaks (e.g, Faeth, 1980; Auerbach & Simberloff, 1988; Stiling & Simberloff, 1989; Connor & Beck, 1993) have reported much lower survi vorship and emergence rates (range: 0.6% to 42%) and higher mortality inflicted by natural enemies (range: 21% to 38%). Adding sticky traps and treating experimental plants with Tanglefoot proved efficient at removing natural enemies on experimental trees, although e ffects of the third trophic level on mine performance and survivorship do not seem to be relevant in our studied systems. Alternatively, the results reported here might also show that, although application of Tanglefoot and use of sticky traps have been efficient in reducing natural enemy impact in other plant-herbi vore-parasitoid systems, (e.g., M oon & Stiling, 2004), it might have had limited efficiency in our oak trees, comp ared to other systems such as salt marshes (e.g., Moon & Stiling, 2002). Alt hough Hawkins (1994) reported high parasitism rates for leaf miners compared to other feeding guild s in biological contro l studies, the results reported here suggest that, under natural c onditions, the concealment provided by the

PAGE 115

103 leaf-mining habit might actually offer an advantage to larval stages of Acrocercops and Brachys The hypothesis that the leaf mine might be adaptive is part of the broader concept that concealed feeding strategies serve as defences against natural enemies (for a review see Connor & Taverner, 1997) and seem to hold true at least for the leaf miners studied here. Variation in host plant nutrition and ch emistry is often a primary cause of differences in feeding efficiency, growth rate and pupal mass of herbivores feeding on different host plants and di fferent parts within plants These results generally demonstrated that leaf mine rs feeding on sub-optimal hos ts tended to consume more tissue and develop for longer periods of tim e, although attack rates by natural enemies and survivorship did not differ among optim al and sub-optimal hosts. Lower predation rates of Acrocercops mines on optimal hosts suggest that faster development did promote escape from natural enemies such as predatory ants. For Brachys mines, however, these results reinforce the idea that sub-lethal pl ant defences remain a paradox (Clancy & Price, 1987; Leather & Walsh, 1993). Brachys mines develop much slower than Acrocercops mines, may exhibit mines that damage more th an 70% of the leaf in which they develop (T. Cornelissen, pers. obs.) and are, therefor e, “apparent” to natural enemies for longer periods of time. Moran and Hamilton (1980) suggested two other scenarios in which poor nutritive quality of foliage could evolve as an adaptation to insect herbivory: 1) if herbivores are able to detect differences in th e nutritive quality of i ndividual plants and if they preferentially feed upon mo re nutritious host individuals then low nutritive quality of leaves is advantageous and 2) if successi ve herbivore generations tend to feed upon the same host individual, then low nutritive quality will prevent future build up of herbivore

PAGE 116

104 numbers, thereby increasing plant fitness. Prev ious studies with leaf miners have shown high degrees of discrimination among hosts w ith varying degrees of nutritional quality (e.g, Faeth et al., 1981; Auerbach & Alberts, 1992; Fox et al,. 1997) and high fidelity to individual natal hosts (e.g., Mopper et al., 1995; Mopper et al., 2000), corroborating both scenarios proposed as explanations for the ad aptive significance of low nutritional quality of hosts. Alternatively, low densities of both Acrocercops and Brachys mines in our study site (mean 1SE per 200 leaves: Acrocercops : 25.9 2.26; Brachys 1st generation: 11.1 1.02; Brachys 2nd generation: 7.71 0.97) might indicate that the detrimental effects of leaf miners on plant fitness are realized only during outbreak years. Previous studies have demonstrated th at the support for the SGHMH is mixed, varying among taxa and natural enemy gu ilds (Williams 1999; Fordyce & Shapiro, 2003). In this study, low nutritional quality does not act as a plant defence and our results reinforce the idea that sub-le thal plant defences remain a paradox, although alternative explanations such as those provided by Mora n and Hamilton (1980) might be sustained in the studied system.

PAGE 117

105 Chapter 6 PERFECT IS BEST: LOW LEAF FLUCTUATING ASYMMETR Y REDUCES HERBIVORY BY LEAF MINERS SYNOPSIS Fluctuating asymmetry (FA) represents small, random variation from symmetry and can be used as an indicator of plant sus ceptibility to herbivor y. We investigated the effects of FA of two oaks species, Quercus laevis and Q. geminata and the responses of three herbivore guilds: leaf mi ners, gallers, and chewers. To examine differences in FA and herbivory between individuals, 40 leav es from each tree were collected, and FA indices were calculated. To examine differences in FA and herbivory within-individuals, we sampled pairs of mined and unmined leaves for asymmetry measurements. Differences in growth of leaf miners between leaf types were determined by tracing 50 mines of each species on symmetric leaves and asymmetric leaves. Asymmetric leaves contained significantly lower concentrations of tannins and higher concentrations of nitrogen than symmetric leaves for both plant species. Both frequency of asymmetric leaves on plants and levels of asymmetr y positively influenced the abundance of Brachys, Stilbosis and other leaf miners, but no signifi cant relationship between asymmetry and herbivory was observed for Acrocercops. Brachys and Stilbosis mines were smaller on asymmetric leaves, but differences in mi ne survivorship between symmetric and asymmetric leaves were observed only for Stilbosis mines. This study indicated that leaf

PAGE 118

106 miners might use leaf FA as a cue to plant quality, although differential survivorship among leaf types was not observed for all sp ecies studied. Reasons for the different results between guilds are discussed. INTRODUCTION Studies on the effects of plant quality on the attack rates of herbivorous insects have been performed extensively and many hypotheses have been proposed to explain within and between variations in herbi vory rates among different plant species. A frequently invoked factor influe ncing herbivory levels is stre ss, since stressors may affect plant nutritional quality for herbivores. Th e plant stress hypothesis (PSH) proposed by White (1984) argues that herbi vore abundance is higher on stre ssed host plants due to an increased availability of nut rients, a decreased concentr ation of defensive compounds and/or changes in the ratio of nutrients to chemical defenses. Evidence supporting the prediction that moderate stress benefits herb ivores due to increased nutritional quality are abundant (e.g., McClure 1980, Lewis 1984, Mattson & Haack 1987) and positive relationships between insect herbivory and plant stress have been found for some tree species, crops and herbaceous plants (e.g., Mattson & Haack 1987, Heinrichs 1988). Nevertheless, some authors claim that expe rimental tests of the PSH have generated conflicting results (e.g., Bultman & F aeth 1987, Louda & Collinge 1992, Koricheva et al. 1998, DeBruyn et al. 2002), and many authors (e.g., Larson 1989, Koricheva et al 1998) have championed the abandonment of the PSH and the search for alternative hypotheses. DeBruyn et al (2002) argued that a major cause of the inconsistent support for the PSH is an inconsistency in the measurements of stress used. Frequent measures of stress in

PAGE 119

107 plants include different estimat es of productivity, plant grow th, biomass, shoot:root ratios and physiological parameters, such as leaf water deficit and plan t secondary chemistry. As suggested by Moller (1995), an objective definition of environmental stress would advance our understanding of the relationshi p between plant stre ss and herbivory. During recent years, it has been claimed th at developmental instability reflects the inability of organisms to control developmental processes during ontogeny and to achieve a predetermined phenotypic optimal expres sion (Moller & Swaddle 1997). One measure of developmental instability is fluctuating as ymmetry (FA) that represents small, random variations from symmetry in otherwise bi laterally symmetrical characters. Leaf fluctuating symmetry has been used as an objective measurement of the effects of environmental stress on plants (e.g., Martel et al. 1999, Roy & Stanton 1999, Alados et al. 2001). Individualand populationlevels of bilateral FA have been related to several biotic and abiotic stresses, including e nvironmental factors, such as nutrition, temperature, radiation, and pollution, as we ll as genetic factors, such as mutation, inbreeding, and hybridisa tion. FA is then suggested as a reliable stress estimator and measures of FA could thus represent se nsitive indicators of the developmental performance of organisms in their environm ent and biomonitors of how organisms are able to deal with deviant envi ronmental and genetic conditions. In addition to being an indicator of plant stress, some studies have shown correlations between FA and insect herbivore attack, suggesting leaf FA can be used not only as an indicator of plant stress, but also plant susceptibility to herbivory (e.g., Wiggins 1997, Zvereva et al 1997). Plants with more asymme tric leaves or higher levels of leaf asymmetry should exhibit increased leve ls of herbivory due to higher nutritional

PAGE 120

108 quality of asymmetric leaves compared to symmetric leaves (Sakai & Shimamoto 1965, Lempa et al. 2000). Little is known as to how deve lopmental disorders are connected to plant metabolism and the associated biochemi cal changes exhibited by asymmetric leaves (Lempa et al. 2000), but, since the left and the righ t sides of a bilaterally symmetrical trait develop under the control of the same genes, minor deviations from perfect symmetry actually represent developmental instability and may be responsible for differences in nutritional quality or sec ondary chemistry between asymmetric and symmetric leaves. Positive corr elations between FA and herb ivory indicate either that plants with asymmetric leaves are, on averag e, more susceptible to attack by herbivores, and/or that herbivory itself acts as a stressor and directly increases the level of leaf asymmetry. Although some authors favour the idea that herbivores themselves can act as stressors increasing leaf asymmetry (e.g., Zvereva et al. 1997), correlatio ns between leaf FA and herbivory are not always likely to be causal (Lempa et al. 2000). Instead, chemical and nutritional differences between symmetric and asymmetric leaves may influence leaf selection by herbivores, whic h leads to positive correlations between herbivory and FA. This study aimed to examine the relations hip between leaf fl uctuating asymmetry and herbivores on Quercus geminata and Q. laevis. We addressed the relationship between herbivory and FA by examining the community of herbivor es attacking these two oak species and how they respond to random variations in leaf morphology. Leaf miners ( Stilbosis quadripustulatus Brachys tesselatus and Acrocercops albinatella ), leaf gallers (eyespot galls and Belonocnema quercusvirens ) and leaf chewers ( Hemileuca maia, Orgyia leucostigma ) were studied. The followi ng hypotheses were tested: 1)

PAGE 121

109 Fluctuating asymmetry, plant stress and herbiv ory: fluctuating asymmetry in otherwise symmetrical bilateral traits is a surrogate of plant stress and asymmetrical leaves should differ in nutritional quality and herbivore sus ceptibility compared to symmetrical leaves; 2) Fluctuating asymmetry between-individuals and frequency of herbivory: if FA in leaves predicts plant susceptibility to herbivor es, plants with more asymmetric leaves or higher degrees of asymmetry should be subj ect to higher levels of herbivory than individual plants with a lower incidence of foliar asymmetry; 3) Fluctuating asymmetry within-individuals and frequency of herbivory: frequenc y of herbivory in asymmetrical leaves should be higher than frequency of herb ivory in symmetrical leaves or leaves with lesser degrees of asymmetry; 4) Fluctu ating asymmetry and the slow-growth, highmortality hypothesis (Clancy & Price 1987): he rbivores feeding on leaves with lower nutritional quality or digestib ility should take longer to develop and would be more susceptible to natural enemies. Therefore, in sects feeding on asymmetrical leaves should exhibit higher survivorship than inse cts feeding on symmetrical leaves. STUDY SYSTEMS Sand live oak, Quercus geminata (Fagaceae), is a semi-evergreen oak and, typically, old leaves abscise and new leaves appear in late April or early May, reaching full size in approximately 2 weeks. The leav es are rounded and pers istent with deeply revolute, conspicuous impressed veins on the underside and the base s and sides of the leaves are observed to be asymmetric in many instances. Stilbosis quadripustulatus (Lepidoptera: Cosmopterygidae) is a moth whose larvae induce mines on the adaxial surfaces of Q. geminata S. quadripustulatus is a univoltine species, whose adults emerge

PAGE 122

110 in early summer (from May to June) from pupae that overwinter in soil and litter. Oviposition occurs approximately in early June when females oviposit at the junction of the midvein and a major lateral vein. Larvae ta ke from 60 to 90 days to complete their 5 instars and mines may reach 3.0 cm in leng th (Simberloff & Stiling 1987). Many other herbivores compose the insect community associated with Q. geminata Leaves are frequently found damaged by chewing inse cts such as the eastern buck moth Hemileuca maia (Lepidoptera: Saturniidae), and at least 4 cynipid species (Hyme noptera: Cynipidae) of galling insects are commonly observed on sand live oak leaves and stems: Andricus quercusfoliatus Disholcaspis quercussuccinipes Callirrhytis quercusbatatoides and Belonocnema quercusvirens. A. quercusfoliatus induces white flow er-like galls on sand live oak stems, whereas D. quercussuccinipes wasps induce clusters of 5-20 yellowish brown galls usually crowded around a terminal oak twig. C. quercusbatatoides wasps induce abrupt swellings of twigs, varying in form and size and B. quercusvirens induces tan, globular pea-like galls on the underside of Q. geminata leaves. Galls are unilocular and occur in large numbers dur ing the fall. Eyespot galls (Diptera: Cecidomyiidae) are recognized as circular spots, usually 8-10 mm in diameter. The adults emerge from the soil in the spring and lay eggs in the upper leaf surface. As the larva grows, the leaf tissue surrounding it swells slig htly and red rings are seen ar ound the galls. Larvae complete their development in 8-12 days and pupate in the soil. This is the most common gall found on sand live oak leaves, often reachi ng densities of 5 galls per leaf. The turkey oak Quercus laevis is one of the characterist ic trees associated with the sand hill community over much of Florida. Q. laevis is a moderately fast to fastgrowing tree and presents deci duous simple leaves, alternatel y arranged with usually 5

PAGE 123

111 lobes, although this number may vary from 3 to 7. Although a common tree in Florida native vegetation, there are relatively few studi es concerning herbivory in this plant species. Acrocercops albinatella (Lepidoptera: Gracillaridae ) is a microlepidopteran species whose larval stages feed on young leav es, creating distinct linear-blotch mines on the lower surface of Q. laevis leaves. Larvae typically feed on the palisade parenchyma cells and deposit frass throughout the mi ne, completing their development in approximately 10 days. Larvae emerge from the blotch mine and usually pupate on the same leaf from which they emerge (T Cornelissen, pers. obsv.). Brachys tesselatus (Coleoptera: Buprestidae) is a un ivoltine species that also forms distinct blotch mines in Q. laevis leaves. The adults emerge in Mid-Marc h to Mid-April, coinciding with budburst of turkey oak. Adults initia lly feed on the early leaves and flowers until mating and oviposition. Eggs are deposited singly on the upper surface of the leaves and after hatching the larvae mine into the mesophyll creating distinct, characteristic damage. Contrary from what happens in South Carolina (Waddell et al 2001) Brachys in our study sites go through two generations, instead of just one. The first mines appear in early April and remain active until late June, wh en larvae complete their development and pupate inside the mines. New adults emerge in early July and oviposit to form new Brachys mines that remain active until Septem ber-October. Pupation and overwintering of this second generation occurs within the l eaves after they have senesced and abscised from the tree. New adults emerge from the l eaf litter in the following spring (Waddell et al 2001). Turkey oak leaves are also attacked by a vast array of herb ivores, such as the leaf roller weevil Homoeolabus analis (Coleoptera: Atellabid ae), the eastern buck moth H. maia the white tussock moth Orgyia leucostigma (Lepidotpera: Lymantriidae) and

PAGE 124

112 other leaf miners such as Stigmella (Lepidoptera: Nepticulidae) and Cameraria (Lepidoptera: Gracillariidae). METHODS Data Collection Patterns of leaf asymmetry, leaf qualit y and herbivory were examined for 30 individuals of Q. geminata and 30 individuals of Q. laevis from March to October 2002 at the University of South Florida Botanica l Garden, Tampa, Florida. To verify the relationship between FA and leaf quality and to examine the frequency of occurrence of asymmetric leaves and levels of leaf asym metry on each plant, 40 leaves were sampled from each individual plant in April 2002. Because herbivores themselves may act as plant stressors, these leaves were sampled before the beginning of mine initiation and before leaves were damaged by free-f eeding herbivores. To quantify Quercus geminata fluctuating asymmetry, widths of all leaves we re measured on both the right and the left side, from the leaf edge to the midrib, at the middle point of the leaf, which usually coincides with the wide st part of the leaf. Q. laevis exhibited some variation in the number of leaf lobes, but 83% of the leaves we sampled exhibited 3 pairs of lobes and measurements were taken between the first and second pairs of lobes. These distances were measured after photographing each leaf with a digital camera at a standard distance of 30 cm in the laboratory and analyzing leaf length, leaf area, and right and left widths using the software UTHSCSA Image Tool (Uni versity of Texas, US A). All the digital pictures were calibrated to the nearest 0.01mm before measurements were taken and the resolution set to the software did not allow measurement errors greater than 1.0%.

PAGE 125

113 Absolute asymmetry in leaf width was defi ned as the unsigned difference between right width (RW) and left width (LW) of a particular leaf as FAwidth = |RW – LW| (Figure 6.1). The absolute value of right-left traits is a good estimator of variance in FA among leaves assuming that there is no directional asymmetry (consistently larger left or right side) or antisymmetry (consistent lack of symmetry, but in no particular direction). To examine differences in nutritional quality between symmetric and asymmetric leaves, all the leaves sampled from each plant were analyzed for water, nitrogen content, and tannin concentration. Water content was quantified by the difference between leaf wet and dry weights after leaves were oven-dried. Leaves were then milled to a fine powder. Tannins were extracted from 50 mg of dry tissue, and tannin concentrati on was quantified using the radial diffusion assay (for details see Hagerman 1987). The average of three replicates per leaf was used for statis tical analysis. Nitrogen conten t was determined using a CHN analyzer. To verify the relationship between plan t fluctuating asymmetry and herbivory between-individuals we used data on asymmetry from the 40 leaves collected from each plant to calculate two indices of FA (Palmer & Strobeck 1986): N IndexL Ri i 1 N IndexL i R i i L i R 2 ) (2 where Ri is the value of the right side, Li is the value of the left side and N is the number of measurements taken. Index 1 is the absolute fluctuating asymmetry and it is the most intuitive asymmetry measurement (R oy & Stanton 1999). Index 2 is size-scaled,

PAGE 126

114 calculated as the absolute value of right (Ri) minus left (Li) sides divided by the average (Ri + Li) /2, to correct for the fact that asymmetry may be si ze-dependent. These indices were then correlated with the density of leaf miners, galls, and chewed leaves recorded on each individual plant. Quercus plants were monitored for th e occurrence of herbivores and the number of Acrocercops, Stilbosis, and Brachys mines were quantified in October 2002 by recording the number of mines in 200 leaves on each plant. Leaf galls were counted in 200 leaves on each plant and stem galls were quantified by counting 100 twigs on each individual in September 2002. We also r ecorded the number of other leaf miners and chewed leaves on both plant species by counting 100 leaves of each plant in October 2002. To examine the relationship between FA and herbivory within -individuals, 40 mined leaves from each leaf miner species were collected from each plant and the 40 nearest neighbouring leaves without mines were collected from the same individual plant. Leaves were oven-dried between sheets of filter pa per, mounted as herbarium specimens and asymmetry measurements of mined and unmin ed leaves were calculated for each leaf miner species as described before. If the di fference between right and left widths (RWLW) was different from zero (either positive or negative) leaves were categorized as asymmetric, and if the difference between right and left widths was equal to zero leaves were categorized as symmetric. However, since we used only 2 decimal places when categorizing FA values and calculating FA indices, leaves with FA measurements ranging between 0.001 and 0.09 mm were rounded to zero and categorized as symmetric, in a more conservative approach. Leaves yiel ding absolute FA values equal or greater than 0.1mm (either positive or negative) were all classified as asymmetric since we

PAGE 127

115 believe the resolution of the equipment we used is quite accurate, as revealed by the small error range (1.00% or 0.001 mm) and high valu es obtained for the Index of Repeatability for both plant species (see Results). To examine herbivore di stribution between symmetric and asymmetric leaves, for each spec ies of leaf miner each pair of leaves collected on each plant was categorized as 1=mined leaf, symmetric: unmined, symmetric, 2= mined leaf, symmetric: unmined, asymmetric, 3= mined leaf, asymmetric: unmined, asymmetric and 4 = mined leaf, asymmetric: unmined, symmetric. To test the relationship between FA within plants and l eaf gallers, we sample d 20 leaves galled by Belonocnema and 20 leaves with eyespot galls an d the nearest 20 non-galled leaves and each pair of leaves was placed in one of the 4 categories as described above. To test for differences in mine survivorship between symmetric and asymmetric leaves, 50 mines of each leaf miner species were marked in asymmetric leaves and another 50 in symmetric leaves. Mined leav es were classified as symmetric or asymmetric after photographing 300 mined leaves in the field and taking measurements as described before. To account for individual variation in leaf miner development related to individual oak plants, no mo re than 6 mines (3 on symmetric leaves, 3 on asymmetric leaves) were marked on each pl ant. All mines (n = 300) were permanently marked using a Sharpie pen as soon as the eggs hatched a nd larvae initiated mine formation. Acrocercops mines were measured at 2-day in tervals by tracing the numbered mines using acetate sheets, whereas Brachys mines were traced at 3-day intervals and Stilbosis mines were traced at 5-day intervals. Mine dr awings were also digitalized and mine size was measured using the software UTHSCSA Im age Tool, with digital pictures calibrated to the nearest 0.01mm. We compared mine size, developmental time (days to pupation)

PAGE 128

116 and mine growth rate on symmetric and asymme tric leaves for each one of the leaf miner species. Mine growth rate on each leaf type wa s calculated as Growth rate = (Final mine size – Initial mine size) / numb er of days mine was growing. After mine termination, all leaves were inspected under a stereomicrosc ope to assess leaf miner survivorship and identification of mortality f actors. Leaf miners offer a great opportunity to assess population survivorship and mortality factor s since a record of the miner success is clearly observed on the leaves: parasitized mine s have tiny circular exit holes left by the parasitoid on mine’s surface and predated mines are usually found ripped open. Successfully emerged larvae cut semi-circula r exit holes on the mine underside towards the apex (Simberloff & Stiling 1987). Data Analysis Leaf characters demonstrate FA if signe d right-minus-left values are normally distributed with a mean value of zero, refl ecting randomly directed deviations from the optimal symmetrical phenotype. One sample t-te sts and Lilliefor’s tests (Wilkinson 1999) were used to test whether mean values of signed right-minus-left values differed significantly from zero. Asymmetry was calculated as the absolute difference between right and left widths of a pa rticular leaf. However, asym metry of undamaged leaves of both plant species increa sed with leaf size ( Q. laevis : r=0.412, P<0.01; Q. geminata : r=0.361, P<0.05) and so measurements of asymmetry on leaf width were further corrected for leaf size ac cording to the formula FAwidth=2*|RW-LW| / (RW+LW). According to Palmer (1996), an important consideration in asymmetry studies is measurement error, since errors may look lik e asymmetry, requiring that either the symmetry differences measured are larger than the measurement error, or that subsequent

PAGE 129

117 measurements taken on the same leaf are highly correlated. We estimated measurement error by remeasuring 10% of the leaves collected from each plant species 10 days after the first measurements were taken and the two measurements were correlated using an index of repeatability (Falconer 1981). All the variables analysed were first submitted to Lilliefor’s test for data normality and transformations (angular, l og-transformation and centering) were employed to stabilize variances and normalize th e data. However, for the sake of clarity, figure axes and means (+1SEM) show untransfo rmed data. To test for differences in nutritional quality between symmetric and asymme tric leaves, we averaged water, tannin, and nitrogen content of all symmetric and asymme tric leaves within each individual plant (n = 30 for each species) and differences betw een the two leaf types were examined using one-way ANOVAs. To examine the relationshi p between plant fluctuating asymmetry and herbivory between-individuals we regressed the percenta ge of asymmetric leaves and the two FA indices calculated for each plan t species with the density of leaf miners ( Brachys, Acrocercops, Stilbosis, other mines), galls, and percen tage of leaves attacked by folivores. Other mines included Cameraria, Buccalatrix and Stigmella mines that were present on the leaves, but in low abunda nce compared to the other leaf miners. A stepwise interactive multiple regression (Wilkinson 1999) was used to examine which factors predicted leaf miner abundance and pred ictors with low tolerance values (<0.10) and high collinearity were excluded from the model. To verify the relationship between FA and herbivory within -individuals we used One-Way AN OVAs to test for differences in FA between mined and unmined leaves of Q. laevis and Q geminata Pairs of minedunmined leaves and galled-non-galled leaves were classified into 4 categories as

PAGE 130

118 described before (data collection) and differe nces in the frequency of these categories were tested using a chi-square test. Pref erence for each leaf type was assumed if mines/galls were found more frequently in a pa rticular leaf type than would have been expected as a result of a simple chance enc ounter with leaves of both types (symmetric, asymmetric). To examine the relationship betw een leaf types and herbivore survivorship, differences in mine growth, days to pupa tion and growth rates of leaf miners in symmetric and asymmetric leaves were tested using One-Way ANOVAs. Some authors have stated that herbivores themselves may cause leaf asymmetry due to their feeding activities. We hypothesized that if leaf miners themselves cause asymmetry, we would expect that mined side s of leaves would be larger/wider than unmined sides when the entire mine is encounter ed on a particular side (right or left) of the leaf. To test this hypothes is, we performed paired t-test s over two scales: plants and leaves. For the leaf scale, we conducted a one -tailed paired t-test with all the mines sampled that were located on a pa rticular side of the leaf of Q. laevis (n = 314 Brachys ) and Q. geminata (n = 1017 Stilbosis mines). For the plant scale, we also performed a paired t-test comparing measurements of asymmetry of 20 undamaged leaves collected before herbivory (April 2002) with 20 undama ged leaves collected from each plant after herbivory and after the sec ond and partial flush of new leaves in July of 2002. To verify whether increased herbivory in one year influences asymmetry in the following season, we compared levels of FA of 30 sand live oaks measured in 2002 with levels of FA of these same plants in 2003. Thirty individuals between 0.7 and 1.3 m in height were monitored from April 2002 to J une 2003. Twenty undamaged leaves were sampled from each plant in August 2002 fo r FA measurements and plants were

PAGE 131

119 monitored for Stilbosis occurrence from May to October 2002, when we recorded the number of Stilbosis mines on 200 leaves on each plant. In June of 2003 we again sampled twenty new but fully developed undamaged leaves from each plant for fluctuating asymmetry measurements as described before. These measurements were further compared with leaf FA and herbivory ra tes recorded for the previous year. RESULTS Tests for asymmetry on Q. laevis and Q. geminata Q. laevis and Q. geminata demonstrated similar patterns of leaf asymmetry before herbivory, as signed right-minus-left char acter values were normally distributed (Lilliefor’s tests, P=0.14-0.61) and did not devi ate significantly from zero in all data sets tested (One-sample t-tests, P= 0.29-0.74) showing no evidence of antisymmetry or directional asymmetry. The data set used to test the relationship between FA and herbivory within individuals also exhibit the normal distri bution with a mean that does not significantly deviate from zero, except for leaves mined by Acrocercops ( Brachys : mean RW-LW = 0.017, t=0.598, P= 0.55; Stilbosis : mean RW-LW = 0.029, t=1.016, P= 0.31; Acrocercops: mean RW-LW = -0.065, t=-2.777, P= 0.006). The mean repeatability of FA measurements was high for both Q. laevis (Index of repeatability = 0.905; F119,120 = 19.46, P <0.0001) and Q. geminata (Index of repeat ability = 0.966; F119,120 = 21.57, P <0.0001), indicating small measurement errors and the reliability of FA measurements. Q. laevis leaves were, on average, more asymmetric than Q. geminata leaves. Values of FA index 1, for example, ranged be tween 0.065 and 0.882 (average: 0.469 0.045) for turkey oaks and between 0.100 and 0.576 (average: 0.266 0.020) for sand live oaks.

PAGE 132

120 Fluctuating asymmetry and leaf quality Asymmetric leaves of both plant species exhibited better nutritional quality for herbivores than symmetric leaves within the same individual plant (Figure 6.2). Asymmetric leaves contained significantly lower concentrations of tannins ( Q. laevis : F1,58=18.19; P<0.0001; Q. geminata : F1,58=12.14; P<0.001) and higher nitrogen content ( Q. laevis : F1,58=4.50; P<0.05; Q. geminata : F1,58=4.79; P<0.05) than symmetric leaves. No differences in water content were obs erved between symmetric and asymmetric leaves of either plant species (P>0.05). Wh en we regressed tannin concentration and FA indices calculated for each plant, we observe d that variation in FA index 2 explained 13.0% of the variation in tannin concentration of Q. geminata (F1,28=8.12, P<0.05), and 16% of the variation in tannins of Q. laevis leaves (F1,28=5.49, P<0.05), but no significant relationship was observed between the variation in FA indices and nitrogen for both plant species (P>0.05). Fluctuating Asymmetry betw een individuals and herbivory Herbivore abundance on the two oak species exhibited a tendency to vary with both the frequency and the levels of asymmetry on individual plants. Q. laevis plants with a higher percentage of asymmetric leav es were significantly more attacked by Brachys (r2=0.279, F1,28=10.84, P<0.005; Figure 6.3) and other leaf miners (r2=0.259, F1,28=9.81, P<0.005). Similar patterns were obs erved for herbivores attacking Q. geminata and the percentage of asymmetric leaves ex plained 31.8% of the variation of Stilbosis (F1,28=13.06, P<0.001; Figure 6.3) and other mines (r2=0.125, F1,28=11.46, P<0.005). No significant relationship was observed between th e percentage of asymmetric leaves on

PAGE 133

121 turkey oaks and the abundance of Acrocercops mines and between the percentage of asymmetric leaves and chewers and stem gallers for both oak species (all P>0.05). Variation in levels of asymmetry between plants also influenced leaf miners abundance. Plants with higher FA in dices exhibited higher densities of Brachys and Stilbosis mines (Figure 6.4). Variation In FA inde x 2 (size-scaled) expl ained 38.8% of the variation in Brachys abundance between plants (F1,28=17.72, P<0.001) and 37.8% of the variation in Stilbosis mines among individual sand live oak plants (F1,28=17.04, P<0.001). Variations in levels of asymmetry also in fluenced the abundance of other leaf miners mines – mainly Cameraria and Stigmella in both plant species ( Q. geminata : r2 = 0.303, F1,28=12.16, P<0.005; Q. laevis : r2 = 0.263, F1,28=9.96, P<0.005). No significant relationship was observed between variation in FA indices an d variation in densities of Acrocercops mines in Q. laevis, eyespot galls on Q. geminata and chewed leaves on both plant species (all P>0.05). Although variation in both th e percentage of asymmetric leaves and levels of asymmetry of Q. geminata tended to positively influence Belonocnema galls (r2=0.10, P=0.071), no significant relationshi p was observed between varia tion in levels of FA and variations in densities of other leaf and stem galling species studied ( Andricus: r2=0.09, P=0.11; Disholcaspis : r2=0.06, P=0.32; Callirhytis : r2=0.10, P=0.09; eyespot galls: r2=0.04, P=0.42). A Pearson correlation matrix revealed high collinearity among pr edictors of leaf miner abundance between plants, such as the percentage of symmetric and asymmetric leaves in a plant ( Q. laevis : r= 0.98, Q. geminata : r=0.97) and the two FA indices used ( Q. laevis : r= 0.974, Q. geminata : r=0.913). These correlat ions generated multiple

PAGE 134

122 regressions coefficients with low tolera nce and all correlated predictors were standardized by centering each value to gene rate a mean of zero for each predictor (Quinn & Keough 2002). An interact ive stepwise multiple regre ssion revealed that 67.4% of the variation in Brachys abundance among plants was explained by variation in FA index 1, the amount of tannins in leaves and variation in Q. laevis leaf area (r2=0.674, F5,24=9.902, P<0.0001). Variation in Stilbosis abundance was also influenced by variation in asymmetry among plants, since almost 62% of the variation was explained by variation in FA index 1, the percentage of asymmetric leaves in a plant and the presence of other leaf miners (r2=0.619, F4,25=10.16, P<0.0001). For Acrocercops mines asymmetry was not important because 38% of the variation among plants was explained by tannin concentration and the amount of nitrogen on leaves alone (r2=0.380, F2,27=8.288, P<0.005). Fluctuating asymmetry with in individuals and herbivory Leaves attacked by both Brachys and Stilbosis were, on average, more asymmetric than unmined leaves within the sa me individual plant (Figure 6.5a). Leaves mined by Brachys were, on average, 4.3 times mo re asymmetric than neighboring unmined leaves (F1,58= 39.67, P<0.0001) while leaves mined by Stilbosis were approximately 2.6 times more asymmetric than unmined leaves (F1,58= 43.39, P<0.0001). However, no significant differences in as ymmetry were observe d between neighbouring mined and unmined leaves attacked by Acrocercops (average absolute FA of mined leaves: 0.265mm 0.027, average absolute FA of unmined leaves: 0.276mm 0.035).

PAGE 135

123 Leaf miners were more frequently encounter ed in asymmetric leaves compared to symmetric ones in both plant species ( Q. laevis : 2 = 65.84, P < 0.0001; Q. geminata: 2 = 60.31, P < 0.0001; Figure 6.5b), although freque ncy of occurrence of categories of leaf types was different for each one of the leaf miner species. For Brachys mines, for example, in approximately 61% of the cases the mined leaf was asymmetric and the unmined leaf in the pair was symmetric, whereas for Acrocercops mines, a more even distribution of the mines was observed among the four possible cate gories (Figure 6.5b). For Stilbosis mines, in 61.4% of the cases, the mined leaf was asymmetric and the unmined leaf was symmetric, and in approximately 30% of the cases, both leaves in a pair were asymmetric. In cases where both leaves were asymmetric, for both Brachys and Stilbosis mean asymmetry of the mined leaf wa s significantly higher than asymmetry of the nearest unmined leaf ( Brachys : F1,58 =11.78; Stilbosis : F1,58 =14.63; both P < 0.005). For both Belonocnema and eye spot galls, we observed no significant differences in frequency of occurrence among the 4 categories of galled-nongalled leaves ( Belonocnema: X2=2.78, P>0.05; eyespot galls: X2= 4.31, P>0.05) and no differences in asymmetry were found between the galled and the nearest non-galled leaf ( Belonocnema: average FA galled leaf = 0.213mm 0.019, average FA of non-galled leaf =0.209mm 0.026; eyespot galls: averag e FA galled leaf = 0.236mm 0.023, average FA of nongalled leaf =0.239mm 0.025; all P>0.05). Fluctuating asymmetry and mine survivorship Mines of all three species were, on average, smaller on asymmetric leaves than on symmetric ones ( Brachys : F1,98 = 5.90, P<0.05, Stilbosis : F1,98 = 32.77, P<0.0001,

PAGE 136

124 Acrocercops: n.s.; Figure 6.6a) and Brachys and Stilbosis mines on asymmetric leaves exhibited significantly slower growth rate s, consuming less plant tissue than mines growing on symmetric leaves (Brachys: F1,49=13.89, P<0.001; Stilbosis : F1,22=15.601, P<0.001; Figure 6.6b). Although Brachys and Stilbosis mines were smaller and consumed less tissue on asymmetric leaves, no significant differences in the number of days necessary to pupation were observed betw een the two leaf types (Figure 6.7a, all P>0.05). Also, no significant differences in mine size, growth rates or days to pupation were observed for Acrocercops mines growing on symmetric and asymmetric leaves of Q. laevis. Mortality imposed by top-down or bo ttom-up factors differed among the leaf miners studied. Acrocercops mines exhibited the highest survivorship am ong the species studied, with 68% of the mines developi ng until pupation, compared to 51% of survivorship of Brachys mines and 49% of survivorship for Stilbosis mines. Parasitism by micro hymenopterans was responsib le for mortality of 23% of Brachys mines, while predation and plant resistance (larvae found dead inside th e mine) accounted for 38% of the mortality of Stilbosis mines. For the Stilbosis mines that survived to pupation, we observed a significantly higher survivorship of mines on asymmetric compared to symmetric leaves ( X2 = 24.51, P<0.05; Figure 6.7b). No significant differences in survivorship between symmetric and asymmetric leaves were observed for Brachys or Acrocercops mines (P>0.05).

PAGE 137

125 Do herbivores cause asymmetry? At the plant scale, paired t-tests revealed that the mean difference in relative FA before and after herbivory did not de part significantly from zero for both Q. laevis (mean difference before and after herbivory: 0. 003 mm, 95% CI: = -0.005 to 0.011, t=0.689, P=0.469) and Q. geminata (mean difference before and af ter herbivory: 0.002 mm, 95% CI: = -0.001 to 0.006, t=1.271, P=0.214). Paired t-tests conducted at the leaf scale also revealed no significant differences in leaf width between mined and unmined sides of Q. laevis leaves attacked by Brachys (mean width of mined side of leaf = 7.429; mean width of unmined side of leaf = 7.412; t = 0.604, P = 0.550) or between mined and unmined leaves of Q. geminata leaves attacked by Stilbosis (mean width of mined side of leaf = 2.612; mean width of unmined side of leaf = 2.587; t = 1.192, P = 0.243) These results suggest that there was no dire ct relationship between the pr esence of these leaf miners and changes in width of the mined side of the leaves. The number of Stilbosis mines per 200 leaves on the 30 plants studied between 2002 and 2003 ranged between 7 and 48 mines (mean: 26.23 2.27) but higher attack rates in some plants did not influence FA measurements in the following year, since we observed a high correlation between FA m easurements in 2002 and 2003 for each one of the 30 plants studied (r2= 0.86, n =30). These results demonstrated that although herbivory rates varied among individuals, plants with higher number of mines in one year did not exhibit higher FA in the following y ear, demonstrating a consistency of variation in FA among individual plan ts during two years.

PAGE 138

126 Figure 6.1Schematic representation (not to scale) of measurements used to define fluctuating asymmetry in A) Quercus laevis and B) Q. geminata RW = right width and LW = left width. A) Q uercus B ) Q g eminata

PAGE 139

127 Leaf nitrogen (%) 0.0 0.5 1.0 1.5 Tannins (mg) 0.00 0.05 0.10 0.15 0.20 Symmetric leaves Asymmetric leaves a. b.Q. laevisQ. geminata Figure 6.2Differences in (a) tannin concentra tion and (b) nitrogen content between symmetric and asymmetric leaves of Q laevis and Q. geminata Bars indicate mean 1SE.

PAGE 140

128 Figure 6.3-. Relationship between the abundance of a) Brachys mines and the percentage of asymmetric leaves on the host plant Q.laevis (r2=0.279, P<0.005) and b) Stilbosis mines and the percentage of asymmetric leaves on Q. geminata (r2=0.318, P<0.001). Mine abundance was determined by counting th e number of mines of each leaf miner species on 200 leaves per plant. Percentage of asymmetric leaves Q. laevis 0.10.20.30.40.50.60.70.80.9Brachys abundance (no.per 200 leaves) 0 5 10 15 20 25 30 35 40 Percentage of asymmetric leaves Q. geminata 0.10.20.30.40.50.60.7Stilbosis abundance (no. per 200 leaves) 10 20 30 40 50 60 70 a. b.

PAGE 141

129 Figure 6.4Relationship between the a bundance of mines caused by a) Brachys (r2=0.475, P<0.001), b) Stilbosis (r2=0.394, P<0.001), and c) Acrocercops (r2=0.018, P>0.05) and the levels of relative asymmetry (F A Index 2) on host plants. FA index 2 refers to the asymmetry index that is size-scal ed, calculated as the absolute value of right (Ri) minus left (Li) sides divided by its average (Ri + Li) /2. 0.000.020.040.060.080.100.120.140.160.18Brachys abundance (no. per 200 leaves) 0 5 10 15 20 25 30 35 40 0.000.010.020.030.040.050.060.07Stilbosis abundance (no. per 200 leaves) 10 20 30 40 50 60 70 FA Index 2 (mm) 0.000.020.040.060.080.100.120.140.160.18Acrocercops abundance (no. per 200 leaves) 10 20 30 40 50 60 70 a. b. c.

PAGE 142

130 Figure 6.5Differences in a) fluctuating asymme try between mined and unmined leaves attacked by Acrocercops, Brachys and Stilbosis and b) frequency of occurrence of combinations of asymmetric and symmetric l eaves on pairs of mined and unmined leaves for each species studied (ms-us= mined leaf symmetric: unmined, symmetric, ms-ua= mined leaf, symmetric: unmined, asymmetric, ma-ua= mined leaf, asymmetric: unmined, asymmetric and ma-us= mined leaf, asymme tric: unmined, symmetric). Bars indicate mean 1SE. Fluctuating asymmetry (mm) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Mined Leaves Unmined Leaves AcrocercopsBrachysStilbosis a. b. Frequency (%) 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Acrocercops Brachys Stilbosis ms-usms-uama-uama-us Combination of leaf mining and leaf symmetry b.

PAGE 143

131 Figure 6.6Differences in a) mine size and b) mine growth rate between symmetric and asymmetric leaves attacked by Acrocercops, Brachys and Stilbosis leaf miners Bars indicate mean 1SE. Mine size (cm2) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 Symmetric Leaves Asymmetric Leaves a. Mine growth rate (mm2 d-1) 0.0 0.2 0.4 0.6 0.8 AcrocercopsBrachysStilbosis b.

PAGE 144

132 Figure 6.7Differences in a) leaf miners developmental time and b) survivorship in symmetric and asymmetric leaves of oak species. Bars indicate mean 1SE. Days to pupation 0 10 20 30 40 50 60 70 Symmetric Leaves Asymmetric Leaves a. Survivorship (%) 0.0 0.2 0.4 0.6 0.8 AcrocercopsBrachysStilbosisb.

PAGE 145

133 DISCUSSION Although literature relating th e effects of plant quality to variation in herbivore abundance between and within plants is pl entiful, few studies have related random variations in leaf morphology and its effects on leaf quality to variation in herbivory rates in individual plants (but see Wiggins 1997, Martel et al. 1999). We suggest that leaf fluctuating asymmetry may be used by herb ivores as a predictor of plant quality, positively influencing insect abundance. Se veral findings of our study reinforce the hypothesis that herbivores may use asymmetry as a cue to plant quality: 1) Asymmetric leaves offered better nutritional quality for he rbivores, such as lower tannin concentration and higher nitrogen content; 2) plants with more asymmetric leaves or higher levels of asymmetry were attacked more by both Brachys and Stilbosis leaf miners; 3) FA indices calculated before herbivory were reasonabl e predictors of leaf miners abundance at the end of the season; 4) Within a plant, l eaf miners were more frequently found in asymmetric leaves and mined leaves exhibite d higher levels of FA than unmined leaves; 5) Mines were smaller on asymmetric leaves compared to symmetric leaves, and Stilbosis mines exhibited higher survivorship on asymmetr ic leaves. Most of our findings reinforce the hypothesis that herbivores are not responsible for as ymmetry and the relationship between herbivory and asymmetry is not causal: 1) individual oak plan ts were consistent in their levels of asymmetry before and after herbivory; 2) there was no direct relationship between the presence of a mine and changes on the width of the side of the leaf where the mine had developed; 3) na turally increased levels of herbivory on Q. geminata plants did not increase asymmetry in th e following season. Since we have used

PAGE 146

134 natural variation in asymmetry between and w ithin individuals, our study did not address the sources of physiological a nd developmental stress in the Quercus species studied, but it has been suggested that FA in leaves may be influenced by several abioti c and biotic factors, such as nutritional deficiencies, water shortage, pollution and plant competition (Palmer & Strobeck 1986). Higher attack rates on plants with more asymmetric leaves a nd higher levels of asymmetry may be attributed to the observed differences in nutritional quality between symmetric and asymmetric leaves, although it is not known how these differences arise. Since the left and the right sides of a partic ular bilaterally symmetr ical trait develop under the control of the same genes, deviations from perfect symmetry actually represent developmental instability, and the ability to develop symmetrical traits may be related to the ability to produce defensive chemical s (Moller 1995), if re source allocation to developmental control competes with alloca tion to production of defensive chemicals. Also, FA may be determined by the same genes as those affecting resistance to herbivores or genes giving rise to elevated levels of FA may have pleiotropic effects on plant resistance (Moller & Swaddle 1997). Although it is not well known how FA is associated with biochemical changes and pl ant metabolism, differences in nutritional quality between symmetric and asymmetric leaves arise and may be responsible for differential attack rates in these two leaf t ypes. This was first demonstrated by Sakai & Shimamoto (1965) studying tobacco plan ts and further supported by Lempa et al (2000) studying several chemical compounds in birch plants. For birch plants, it was observed that plants with higher levels of FA contained significantly lower amounts of hydrolysable tannins, gallic acid, and flavonoid-glycosides. Labor atory feeding trials also

PAGE 147

135 showed that Epirrita autumnata consumed more from birch leaves collected from high FA trees compared to low FA trees (Lempa et al. 2001). We are unaware of other studies that have tested differences in nutritiona l quality between symmetric and asymmetric leaves and how these affect herbivore pr eference and performance. Differences in nutritional quality between asymmetric and sy mmetric leaves reinforce the idea that the relationship between asymmetry and herbivor y is not causal, i.e., herbivory does not cause asymmetry in our study system. Instead, herbivores may use asymmetry as a cue to plant quality and suitable oviposition sites. Other st udies that found a positive relationship between fluctuati ng asymmetry and herbivory ra tes have sampled only after herbivory has occurred, eliminating the possibi lity to assess whether asymmetric leaves were present before herbivory and whether he rbivores preferentially attack these leaves. With our sampling design, we demonstrated th at asymmetric leaves were present in both Q. laevis and Q. geminata before leaf miners attacked the leaves. Our results also showed that mines caused by Brachys and Stilbosis were smaller in asymmetric leaves compared to symmetric leaves and this may be explai ned by the lower consumption rates observed in asymmetric leaves compared to symmetri c ones. Our results par tially support the slowgrowth, high-mortality hypothesis since Brachys and Stilbosis mines growing on symmetric leaves with reduced nutritional quality exhibited higher consumption rates resulting in bigger mines at the end of the season. Neve rtheless, we found only limited support for higher mortality rates caused by natural enemies on mines growing on symmetric leaves, since survivorship of Brachys and Acrocercops mines did not differ between symmetric and asymmetric leaves of Q. laevis and only Stilbosis mines exhibited higher survivorship on asymmetric leaves of Q. geminata These results

PAGE 148

136 indicate that although bottomup factors, such as asymmetry and plant quality may influence the choice of oviposition sites by insects, the positive effects of better nutritional quality on mine survivorship are no t always realized when top-down factors, such as predation and parasitism, are strong and other studies with leaf miners have demonstrated the strength of top-down pressure s on insect survivorship (e.g., Auerbach & Simberloff 1988, Mopper et al. 1995, Hawkins et al. 1997, Forkner & Hunter 2000, reviewed by Connor & Taverner 1997). Although variation in herbivory rates in both Q. laevis and Q. geminata may be related to fluctuating asymmetry and its conse quential changes in plan t quality, our study has also demonstrated idiosyncratic respons es to FA both between and within guilds. Leaf chewers and leaf galls were not influen ced by levels of FA in both plant species. Galling insects may not be as influenced by aspe cts of plant quality as leaf miners are, since their special mode of feeding within the gall allows them to manipulate plant characteristics, and possibly avoid defensiv e strategies of the host and circumvent differences in nutritional quality betw een symmetric and asymmetric leaves. Manipulation of host plants by gall-formers may extend to control over the chemical composition of gall tissues, and galling herbi vores may alter the physiological state of host tissues, especially the tissues ne arest to the developing larvae (Price et al. 1987, Shorthouse & Rohfritsch 1992, Hartley 1998). Am ong leaf miners, strong responses were found for both Stilbosis and Brachys but Acrocercops mines on turkey oaks seem not to be influenced by variation in asymmetry between and within plants. This variation in leaf miner response to fluctuating asymmetry may be explained by differences in life-history traits of the species studied. Acrocercops cause relatively small linear-blotch superficial

PAGE 149

137 mines just under the leaf epidermis of turk ey oak leaves and development times do not exceed 10 days. These mines are unlikely to be strongly affected by variations in plant quality due to their fast development rates a nd the fact that they create limited depth mines in young leaves with hi gher nitrogen content and smaller concentration of defensive chemicals. Brachys and Stilbosis mines may be more likely to be affected by host quality, having long developmental times full depth mines, and a higher likelihood to be affected by spatial and s easonal variation in host quality.

PAGE 150

138 CONCLUSION Our study of leaf miners on Florida scrub oaks has demonstrated the diversity and peculiarity of responses of a single guild to effects of plant quality and natural enemies when several scales of organization are consider ed. At the largest scale, leaf miners were clustered into sites (Chapter 2) and sites separated by smalle r distances were more similar in the abundance of mines than sites farther apart. Howeve r, at the regional, landscape scale, we observed that bottom-up and topdown factors were not spatially structured, demonstrating that other factors were pr obably more relevant in structuring the abundance of leaf miners when spatial position is taken into account. At local scales, we have demonstrated that leaf miners tended to be clustered into indivi dual plants, and into individual branches within plants (Chapter 3). Natural inter-individual variation in bottom-up factors such as the concentration of secondary defenses and foliar nitrogen is partially responsible for the aggregation patt erns found. Leaf miners also responded to experimentally elevated levels of foliar nitrogen, as fertiliz ed plants supported significantly more herbivores than plants with unmanipulated levels of nitrogen (Chapter 4). Our studies have demonstrated that, although bottom-up factors affected both the abundance and performance of some leaf mine r species, top-down factors tended to be weak and non-significant. Removing natural en emies from the plants, for example, did not change the abundance (Chapter 4) and/or the survivorship (Chapter 5) of leaf miners, indicating that, if topdown effects on leaf miners do occur, they were not detected with our experimental designs. At the individua l scale, our studies of fluctuating asymmetry

PAGE 151

139 have indicated that leaf miners responded to natural and random variation in leaf morphology, an indicator of plant stress (Cha pter 6). Individual plants with higher percentage of asymmetric leaves and/or higher levels of asymmetry supported higher densities of some leaf miner species and differences in plant quality between symmetric and asymmetric leaves partially explained th is result. At the indi vidual scale, however, we again did not detect top-down effects on the survivorship of leaf miners, as no differences in survivorship and mortality rates were observed between symmetric and asymmetric leaves. Although species-specific vari ation in leaf miner res ponse to plant quality and natural enemies were observed, bottom-up fact ors tended to be stronger than top-down factors, and significantly impact ed the abundance, performance, and survivorship of leaf miners, especially at local and individual scales.

PAGE 152

140 LIST OF REFERENCES Abrahamson WG & Weis AE (1986) Nutritiona l ecology of arthropod gall-makers. In: F. Slansky & J.G. Rodriguez (Eds.), The nutritional ecology of insects, mites and spiders (pp. 235-258). New York: Wiley. Alados CL, Navarro T, Escos J, Cabez udo B & Emlen JM (2001) Translational and fluctuating asymmetry as tools to detect stress in stress-adapted and nonadapted plants. International Journal of Plant Sciences 162: 607-616. Auerbach M & Simberloff D (1988) Rapid leaf miner colonization of introduced trees and shifts in sources of herbivore mortality. Oikos 52: 41-50. Auerbach MJ & Simberloff DJ (1989) Ovipositio n site preference and larval mortality in a leaf-mining moth. Ecological Entomology 14: 131-140. Auerbach M & Alberts JD (1992) Occurren ce and performance of the aspen blotch miner, Phyllonorycter salicifoliella on three host-tree species. Oecologia 89: 1-9. Auerbach MJ, Connor EF & Mopper S ( 1995) Minor miners and major miners: population dynamics of leaf-mining insects. In: N. Cappuccino & P.W. Price (Eds.), Population dynamics – ne w approaches and synthesis (pp. 83-110). New York: Academic Press. Augner M (1995) Low nutritive qua lity as a plant defence: eff ects of herbivore-mediated interactions. Evolutionary Ecology 9: 605-616. Benrey B & Denno RF (1997) The slow-growt h-high-mortality hypothesis: a test using the cabbage butterfly. Ecology 78: 987-999. Bernays EA & Chapman RF (1994) Host-p lant selection by phyt ophagous insects. Chapman & Hall, 312 p. Bevers M & Flather CH (1999) The distribution and abundan ce of populations limited at multiple spatial scales. Journal of Animal Ecology 68:976-987. Blackburn TM, Gaston KJ, Quinn RM & Gre gory RD (1999) Do local abundances of British birds change with proximity to range edge? Journal of Biogeography 26: 493505.

PAGE 153

141 Brewer AM & Gaston KJ (2002) The geographi cal range structure of the holly leaf miner. I – Population density. Jour nal of Animal Ecology 71: 99-111. Brewer AM & Gaston KJ (2003) The geographi cal range structure of the holly leaf miner. II – Demographic rates. Journal of Animal Ecology 72: 82-93. Brown JH (1984) On the relationship betw een abundance and distribution of species. American Naturalist 124: 255-279. Brown JH, Mehlman DW & Stevens GC (1995) Spatial variation in abundance. Ecology 76: 2028-2043. Brown JL, Vargo S, Connor EF & Nichols MS ( 1997) Causes of vertical stratification in the density of Cameraria hamadryadella Ecological Entomology 22: 16-25. Bultman TL & Faeth SH (1986) Experimental evidence for intraspecific competition in a lepidopteran leaf mine r. Ecology 67: 442-448. Bultman TL & Faeth SH (1986) Impact of ir rigation and experimental drought stress on leaf-mining insects on Emory oak. Oikos 48: 5-10. Bylund H & Tenow T (1994) Long-term dynamics of leaf miners, Eriocrania sp. on mountain birch: alternative year fl uctuations and interactions with Epirrita autumnata Ecological Entomology 19: 310-318. Carson WP & Root RB (2000) Herbivory and plant specie s coexistence: community regulation by an outbreaking phytophagous insect. Ecological Monographs 70: 7399. Clancy KM & Price PW (1987) Rapid herbi vore growth enhances enemy attack: sub lethal plant defenses remain a paradox. Ecology 68:736-738 Connor EF (1991) Colonization, surviv al, and the causes of mortality of Cameraria hamadryadella (Lepidoptera: Gracillariidae) on four species of host plants. Ecological Entomology 16: 315-322. Connor EF & Beck MW (1993) Density-related mortality on Cameraria hamadryaella (Lepidoptera: Gracillaridae) at epidemic and endemic densities. Oikos 66: 515-525. Connor EF & Taverner MP ( 1997) The evolution and adaptiv e significance of the leafmining habit. Oikos 79: 6-25. Cooke FP, Brown JP & Mole S (1984) Herbiv ory, foliar enzyme inhibitors, nitrogen and leaf structure of young and matu re leaves in a tropical forest. Biotropica 16: 257-263.

PAGE 154

142 Cornelissen T & Stiling P (2005) Perfect is be st: low leaf fluctuating asymmetry reduces herbivory by leaf miners. Oecologia 142: 46-56. Cornelissen T & Stiling P (2006a) Clumped dist ribution of oak leaf miners between and within plants. Basic and A pplied Ecology (in review). Cornelissen T & Stiling P (2006b) Responses of different herbivore guilds to nutrient addition and natural enemy excl usion. Ecoscience 13: 66-74. Cornelissen T & Stiling P (2006c) Does low nutr itional quality act as a plant defense? An experimental test of the slow-growth, high-mortality hypothesis. Ecological Entomology 30: 1-9. Cornell HV & Hawkins BA (1995) Survival patterns and mortality sources of herbivorous insects: some demographic trends. The American Naturalist 145: 563593. Crawley MJ & Akhteruzzaman M (1988) Indivi dual variation in the phenology of oak trees and its consequence for herbivorous insects. Functional Ecology 2: 409-415. Curnutt JL, Pimm SL & Maurer BA (1996) Population variabili ty of sparrows in space and time. Oikos 76: 131-144. DeBruyn L, Scheirs J & Verhagen R (2002) Nutrient stress, host plant quality and herbivore performance of a leaf-mining fly on grass. Oecologia 130: 594-599 Denno RF, McClure MS & Ott JR (1995) Inte rspecific interactions in phytophagous insects: competition re-examined and re surrected. Annual Review of Ecology and Systematics 40: 297-331. Denno RF, Peterson MA, Gratton C, Cheng J, Langelloto GA, Huberty AF & Finke DC (2000) Feeding-induced changes in plant quality mediate interspecific competition between sap-feeding herbi vores. Ecology 81: 1814-1827. Denno RF, Gratton C, Peterson MA, Langello tto GA, Finke DL & Huberty AF (2002) Bottom-up forces mediate natural-enem y impact in a phytophagous insect community. Ecology 83: 1443-1458. Denno RF, Gratton C, Dobel H & Finke DL (2003). Predation risk affects relative strength of top-down and bottom-up impact s on insect herbivores. Ecology 84: 10321044. Deyrup MA (1996) Two new grasshoppers from relict uplands of Florida (Orthoptera: Acrididae). Transactions of the American Entomological Society 122: 199-211.

PAGE 155

143 Doak P (2000) Habitat patc hiness and the distributi on, abundance and population dynamics of an insect herbivore. Ecology 81: 1842-1857. Eber S (2004) Bottom-up density re gulation in the holly leaf-miner Phytomyza ilicis Journal of Animal Ecology 73: 948-958. Faeth SH (1980) Invertebrate predation of leaf miners at low densities. Ecological Entomology 5: 111-114. Faeth, SH, Mopper S & Simberloff D (1981) Abundances and diversity of leaf-mining insects on three oak host species: effect s of host-plant phenology and nitrogen content of leaves. Oikos 37: 238-251. Faeth SH (1985) Host leaf selection by l eaf miners: interactio ns among three trophic levels. Ecology 66: 870-875. Faeth SH & Simberloff D (1989) Populati on regulation of a leaf-mining insect, Cameraria sp., at increased field de nsities. Ecology 62: 620-624. Faeth SH (1990) Aggregation of a leafminer, Cameraria sp. nov. (Davis): consequences and causes. Journal of Animal Ecology 59: 569-586. Faeth SH (1991) Effect of oak leaf size on abundance, dispersion, and survival of the leafminer Cameraria sp. (Lepidoptera: Gracillariid ae). Environmental Entomology 20: 196-204. Faeth SH (1992) Interspecific and intraspecific interactio ns via plant-responses to folivory – an experimental fi eld-test. Ecology 73: 18021813. Fagan WF, Cantrell RS & Cosner C ( 1999) How habitat edges change species interactions. American Naturalist 153: 165-182. Falconer DS (1981) Introduction to quantitative genetics. 2nd edition, Longman, New York Fay PA, Hartnett DC & Knapp AK (1993) Incr eased photosynthesis and water potential in Silphium integrifolium galled by cynipid wasps. Oecologia 93: 114-120. Feeny P (1976) Plant apparenc y and chemical defence. Biochemical interaction between plants and insects (ed. by J.W Wallace and R.L. Mansell), pp. 1-40 Plenum Press, New York. Fischer K & Konrad F (2000) Re sponse of the copper butterfly Lycaena tityrus to increased leaf nitrogen in natural f ood plants: evidence ag ainst the nitrogen limitation hypothesis. Oecologia 124: 235-241.

PAGE 156

144 Fisher AEI, Hartley SE & Young M (1999) Be havioral responses of the leaf-chewing guild to the presence of Eriocrania mines on silver birch ( Betula pendula ). Ecological Entomology 24: 156-162. Fisher AEI, Hartley SE & Young M (2000) Dire ct and indirect competitive effects of foliage feeding guilds on the perf ormance of the birch leaf-miner Eriocrania Journal of Animal Ecology 69: 165-176. Fordyce JA & Shapiro AM (2003) Anothe r perspective on the slow-growth/highmortality hypothesis: chi lling effects on swallowtail larvae. Ecology 84: 263-268. Forkner RE & Hunter MD (2000) What goes up must come down? Nutrient addition and predation pressure on oak he rbivores. Ecology 81: 1588-1600. Fox CW, Waddell KJ, Groeters FR & Mou sseau TA (1997) Variation in budbreak phenology affects the distribution of a leaf mining beetle ( Brachys tessellatus ) on turkey oak ( Quercus laevis ). Ecoscience 4: 480-489. Gaston KJ, Genney DR, Thurlow M & Hartley SE (2004) The geographical range of the Holly leaf miner. IV – Effects of variati on in host-plant quality. Journal of Animal Ecology 73: 911-924. Gonzalez-Megias A, Gomez JM & Sanchez-Pi nero F (2005) Consequences of spatial autocorrelation for the analysis of metapopulation dynamics. Ecology 86: 32643271. Gotelli NJ & Graves GR (1996) Null models in ecology Washington D.C.: Smithsonian Institution Press. Gotelli NJ & Entsminger GL (1999) EcoSim. Null models software for ecology Version 3.0. Acquired Intelligence Incorporated, and Kesey-Bear. [Online: {http://homepages.together.net/~gentsmin/ecosim.htm}]. Gotelli NJ & Ellison AM (2002) Assembly rules for New England ant assemblages. Oikos 99: 591-599. Green RH (1966) Measurement of non-randomne ss in spatial distributions. Researches Population Ecology 8: 1-7. Gruner DS (2004). Attenuation of top-down and bottom-up forces in a complex terrestrial community. Ecology 85: 3010-3022. Hagerman AN (1987) Radial di ffusion method for determining tannin in plant extracts. Journal of Chemical Ecology 13: 437-449.

PAGE 157

145 Haggstrn H & Larsson S (1995) Slow larval growth on suboptimal willow results in high predation mortality in the leaf beetle Galerucella lineola Oecologia 104: 308315. Hairston NG, Smith EE & Slobodkin LB ( 1960) Community structure, population control, and competition. American Naturalist 94:421-425. Hartley SE (1998) The chemical composition of plant galls: are levels of nutrients and secondary compounds controlled by the gall-former? Oecologia 113: 492-501 Hartley SE & Lawton JH (1987) Host-plant manipulation by ga ll-insects: a test of the nutrition hypothesis. Journal of Animal Ecology 61: 113-119. Haukioja E, Ossipov V, Koricheva J, H onkanem T, Larsson S & Lempa K (1998) Biosynthetic origin of ca rbon-based secondary com pounds: cause of variable responses of woody plants to fertilization? Chemoecology, 8: 133-139. Hawkins BA (1994) Pattern and process in host-parasitoid interactions Cambridge Press, Cambridge. Hawkins BA, Cornell HV & Hochberg ME (1997) Predators, parasitoids, and pathogens as mortality agents in phytophagous insect populations. Ecology 78: 2145-2152. Hedges LV, Gurevitch J & Curtis PS (1999). Th e meta-analysis of response ratios in experimental ecology. Ecology 80: 1150-1156. Heinrichs EA (1988) Plant stress-insect in teractions. John Wiley & Sons, New York, New York, USA Hengeveld R & Haeck J (1982) The distri bution of abundance. I – Measurements. Journal of Biogeography 9: 303-316. Hering EM (1951). Biology of the le af miners, Springer, Berlin. Hespenheide HA (1991) Bionomics of leaf -mining insects. Annual Review of Entomology 36: 535-560. Hubbell T (1954) Relationships a nd distribution of the genus Mycrotrupes Miscellaneous publications of the Museum of Zoology, University of Michigan, 84: 39-51. Huberty AF & Denno RF (2004) Plant water st ress and its consequences for herbivorous insects – a new synthesis. Ecology 85: 1383-1398.

PAGE 158

146 Hunter MD & Price PW (1992) Playing chutes and ladders: he terogeneity and the relative roles of bottom-up and top-down forces in natural communities. Ecology 73: 724732. Hunter MD (2001) Multiple approaches to estimating the relativ e importance of topdown and bottom-up forces on insect populati ons: experiments, life-tables, and timeseries analysis. Basic and Applied Ecology 2: 295-309. Janzen DH (1968) Host plants as islands in evolutionary and contemporary time. American Naturalist 104: 501-528. Jarzomski CM, Stamp NE & Bowers MD (20 00) Effects of plant phenology, nutrients and herbivory on growth and defensive chemistry of plantain, Plantago lanceolata Oikos 88: 371-379. Johnson KS, Scriber JM & Nair M (1996) Phenylpropenoid phenolics in Sweetbay Magnolia as chemical determinants of host use in satu rniid silkmoths ( Callosamia ). Journal of Chemical Ecology 22: 1955-1967. Johnson SN, Mayhew PJ, Douglas AE & Hartley SE (2002) Insects as l eaf engineers: can leaf-miners alter leaf structure for bi rch aphids? Functional Ecology 16: 575-584. Kagata H & Ohgushi T (2001) Preference and performance lin kage of a leaf-mining moth ( Paraleucoptera sinuella ) on different Salicaceae species. Population Ecology 43: 141-147. Kery M, Matthies D & Fischer M (2001) The effect of plant population size on the interactions between the rare plant Gentiana cruciata and its specialized herbivore Maculinea rebeli Journal of Ecology 89: 418-427. Klok CJ, Chown SL & Gaston KJ (2003) The ge ographical range structure of the Holly leaf miner. IIICold hardiness physiology. Functional Ecology 17: 858-868. Koricheva J, Larsson S, Haukioja E (1998) Insect performance on experimentally stressed woody plants: a meta-analysis. Annual Review of Entomology 43: 192-216. Koricheva J, Larsson S, Haukioja E & Ke inanen M (1998) Regulation of woody plant secondary metabolism by resource availabi lity: hypothesi s testing by means of metaanalysis. Oikos 83: 212-226. Kyto M, Niemela P & Larsson S (1996) Insects on trees: popul ation and individual response to fertilization. Oikos 75:148-159. Larson S (1989) Stressful times for the plant stress-insect performance hypothesis. Oikos 56: 277-283

PAGE 159

147 Leather, S.R. & Walsh, P.J. (1993) Sub-le thal plant defences: the paradox remains. Oecologia 93: 153-155. Leather SR (2000) Herbivory, phenology, mor phology and the expression of sex in trees: who is the driver’s se at? Oikos 90: 194-196. Legendre P (1993) Spatial autocorrelation: trouble or new paradigm? Ecology 74: 16591673. Legendre P & Legendre L (1998) Numerical Ec ology – Developments in Environmental Modeling. Elsevier, Amsterdam, The Netherlands, 853 p. Lempa K, Martel J, Koricheva J, Haukioja E, Ossipov V, Ossipova S, Pihlaja K (2000) Covariation of fluctuating asymmetry, herbivory and chemistry during birch leaf expansion. Oecologia 122: 354-360 Levine MT & Paige KN (2004) Direct and in direct effects of drought on compensation following herbivory in scarle t gilia. Ecology 85: 3185-3191. Lewis AC (1984) Plant quality and grasshopper feeding: effects of sunflower condition on preference and performance in Melanoplus differentialis Ecology 65:836-843. Lill JT & Marquis RJ (2001) The effects of leaf quality on herbivore performance and attack from natural enemie s. Oecologia 126: 418-428. Lindroth RL, Hofman RW, Campbell BD, McNabb WC & Hunt DY (2000) Population differences in Trifolim repnes L. response to ultra-violet radiation: foliar chemistry and consequences for two Lepidopteran herbivores. Oecologia 122: 20-28. Loader C & Damman H (1991) Nitrogen cont ent of food plants and vulnerability of Pieris rapae to natural enemies. Ecology 72:1586-1590. Louda SM & Collinge SK (1992) Plant resistance to insect herbivores: a field test of the environmental stress hypothesis. Ecology 73: 153-169. Ludwig JA & Reynolds JF (1988) Statistical Ecology. A primer on methods and computing New York: John Wiley & Sons. Marquis R J & Whelan CJ (1994) Insectivor ous birds increase growth of white oak through consumption of leaf-chewi ng insects. Ecology 75: 2007–2014. Marshall SD, Hoeh WR & Deyrup MA ( 2000) Biogeography and conservation biology of Florida’s Gelycosa wolf spider: threatened spiders in endangered ecosystem. Journal of Insect Conservation 4: 11-21.

PAGE 160

148 Martel J, Lempa K, Haukioja E (1999) Effect s of stress and rapid growth on fluctuating asymmetry and insect damage in birch leaves. Oikos 86:208-216 Masters GJ, Jones TH & Rogers M (2001) Host-p lant mediated effect s of root herbivory on insect seed predators and their parasitoids. Oecologia 127: 246-250. Masumoto T, Sunahara T & Suzuki N (2000) Effects of non-host and host-plants on insect herbivory covarying with pl ant size in the cruciferous plant Turritis glabra Population Ecology 42: 145-152. Mattson WJ (1980) Herbivory in relation to pl ant nitrogen content. Annual Review of Ecology and Systematics 11: 119-161. Mattson WJ & Haack J (1987) The role of drought stress in provoking outbreaks of phytophagous insects. In: Barbosa P, Schultz J (eds) Insect outbreaks: ecological and evolutionary perspectives. Academic pre ss, Orlando, Florida, USA, pp 365-407. McClure MS (1980) Foliar nitrogen: a basis for host suitability for elongate hemlock scale, Fiorinia externa (Homoptera: Diaspidi dae). Ecology 61: 72-79 McGeoch MA & Gaston KJ (2000) Edge effects on the prevalence and mortality factors of Phytomyza ilicis (Diptera, Agromyzidae) in a suburban woodland. Ecology Letters 3: 23-29. McGeoch MA & Price PW (2004) Spatial abundan ce structures in an assemblage of gallforming sawflies. Journal of Animal Ecology 73: 506-516. Mehlman DW (1997) Change in avian a bundance across the geographic range in response to environmental change. Ecological Applications 7: 614-624. Moller AP & Erickson M (1994) Patterns of fluctuating asymmetry in flowers: implications for sexual selection in plants Journal of Evolutionary Biology 7:97113. Moller AP & Swaddle JP (1997) Asymmetry, developmental stability and evolution. Oxford University Press, Oxford. Moller AP & Shykoff JA (1999) Morphological developmental stability in plants: patterns and causes. Interna tional Journal of Plant Sciences 160: S135-S146. Moon DC & Stiling P (2002) The influence of species identity and herbivore feeding mode on top-down and bottom-up effects in a salt marsh system. Oecologia 133: 243-253. Moon DC & Stiling P (2004) The influence of sa linity and nutrient gr adient on coastal vs. upland tritrophic complexes. Ecology 85: 2709-2716.

PAGE 161

149 Mopper S, Faeth SH, Boecklen WJ & Simberlo ff DS (1984) Host-spe cific variation in leafminer population dynamics: effects on dens ity, natural enemies, and behavior of Stilbosis quadricustatella (Lepidoptera: Cosmopterigidae). Ecological Entomology 9: 169-177. Mopper S, Beck M, Simberloff D & Stiling P (1995) Local adaptation and agents of selection in a mobile ins ect. Evolution 49: 810-815. Mopper S, Stiling P, Landau K, Simberlo ff D & Van Zandt P (2000) Spatiotemporal variation in leaf miner popul ation structure and adaptati on to individual oak trees. Ecology 81: 1577-1587. Moran N & Hamilton WD (1980) Low nutritive quality as a defence against herbivores. Journal of Theoretical Biology 86: 247-254. Moran M & Scheidler AR (2002) Effects of nut rients and predators on an old-field food chain: interactions of top-down and bo ttom-up processes. Oikos 98: 116-124. Mushinsky HR, Stilson TA & McCoy ED (2003) Diet and dietary preference of the juvenile gopher tortoise ( Gopherus polyphemus ). Herpetologica 59: 475-483. Myers JH (1985) Effect of the physiologi cal condition of the host plant on the ovipositional choice of th e cabbage white butterfly, Pieris rapae Journal of Animal Ecology 54: 193-204. Nixon KC (1997) Fagaceae Dumortier – Beech Family. Pages 436-506 in NR Morin (ed.). Flora of North America: north of Me xico. Volume 3, Oxford University Press, New York. Oksanem L, Fretwell SD, Arruda J & Niem ala P (1981) Exploitation ecosystems along gradients of primary productivity. Am erican Naturalist 118: 240-261. Palmer AR (1996) Waltzing with as ymmetry. Bioscience 46: 518-532. Palmer RA & Strobeck C (1986) Fluctuati ng asymmetry: measurem ent, analysis, and patterns. Annual Review of Ecology and Systematics 17:391-421. Perry JN (1995) Spatial analysis by distance indices. Journal of Animal Ecology 64: 303314. Perry JN, Winder L, Holland JM & Alston RD (1999) Red-blue plots for detecting clusters in count data. Ecology Letters 2: 106-113.

PAGE 162

150 Polis GA & Strong D (1996) Food web comple xity and community dynamics. American Naturalist 147:813-846. Price PW (1997) Insect Ecology. 3rd edition, Wiley, New York, 514 p. Price PW (2002) Resource-driven terrestrial interaction webs. Ecol ogical Research 17: 241-247. Price PW, Bouton CE, Gross P, McPheron BA, Thompson JN & Weis AE (1980) Interactions among tri-trophic levels – infl uence of plants on in teractions between insect herbivores and natural enemies. Annual Review of Ecology and Systematics 11: 41-65. Price PW, Fernandes GW & Waring GL (1987) Adaptive nature of insect galls. Environmental Entomology 16:15-24. Price PW, Abrahamson WG, Hunter MD & Meli ka G (2004) Using gall wasps on oaks to test broad ecological concepts Conservation Biology 18: 1405-1416. Prince SD, Carter RN & Dancy KJ (1985) The geographical dist ribution of prickly lettuce ( Lactuca serriola ). II – characteristics of popul ations near its distribution limits in Britain. Journal of Ecology 73: 39-48. Quinn GP & Keough MJ (2002) Experimental de sign and data analysis for biologists. Cambridge University Press, UK Reavey D & Gaston KJ (1991) The importance of leaf structure in oviposition by leafmining microlepidopte ra. Oikos 61: 19-28. Reitz SR & Trumble JT (2002) Competitiv e displacement among insects and arachnids. Annual Review of Entomology 47: 435-465. Ribas CR & Schoereder JH (2002) Are all ant mosaics caused by competition? Oecologia 131: 606-611. Rosenberg MS (2001) PASSAGE. Pattern anal ysis, spatial statis tics, and geographic exegesis. Version 1.0. Department of Biol ogy, Arizona State Univ ersity, Tempe, AZ. Rosenberg MS, Adams DC & Gurevitch J (2000) MetaWin: statistical software for metaanalysis. Version 2.1.3.4, Sinauer Asso ciates, Sunderland, Massachusetts, USA. Rostas M & Hilker M (2003) Indirect inte ractions between a phytopathogenic and an entomopathogenic fungus. Naturwissenschaften 90: 63-67.

PAGE 163

151 Roy BA & Stanton ML (1999) As ymmetry of wild mustard, Sinapis arvensis (Brassicaceae), in response to severe physio logical stresses. Journal of Evolutionary Biology 12: 440-449. Sagarin RD & Gaines SD (2002) The ‘abundant cen tre’ distribution: to what extent is it a biogeographical rule? Ec ology Letters 5: 137-147. Sakai KI & Shimamoto Y (1965) Developmenta l instability in leaves and flowers of Nicotinia tabacum Genetics 51:801-813. Sanz JJ (1997) Geographic variation in breed ing parameters of the pied flycatcher Ficedula hypoleuca Ibis 139: 107-114. Sato SH (1991) Differential resource utiliza tion and co-occurrence of leaf miners on oak ( Quercus dentata ). Ecological Entomology 16: 105-113. Scheirs J, DeBruyn L & Verhagen R (2001) Nu tritional benefits of the leaf-mining behavior of two grass miners: a test of the selective feeding hypothesis. Ecological Entomology 26: 509-516. Scheirs J, DeBruyn L & Verhagen R (2002) S easonal changes in leaf nutritional quality influence grass miner performance. Ecological Entomology 27: 84-93. Shibata S, Ishida TA, Soeya F, Morino N, Yoshida K, Sato H & Kimura, M.T. (2001) Within-tree variation in density and survival of leaf miners on oak Quercus dentata Ecological Research 16: 135-143. Shorthouse JD & Rohfritsch O (1992) Biology of insect-induced galls. Oxford University Press, New York. Simberloff D & Stiling P (1987) Larval dispersion and surv ivorship in a leaf-mining moth. Ecology 68: 1647-1657. Slansky FJ (1993) Nutritional ecology: the fundamental quest for nutrients. Caterpillars: Ecological and evolutionar y constraints on foraging (ed. by N.E. Stamp and T.M. Casey), pp. 29-91. Chapman & Hall, New York. Sorte CJB & Hofmann GE (2004) Changes in latitude, changes in aptitudes: Nucella canaliculata (Mollusca: Gastropoda) is more st ressed at its range edge. Marine Ecology Progress Series 274: 263-268. SPSS Inc. 2003. SPSS Base 12.0.2 for windows user’s guide. SPSS Inc., Chicago IL. Stanton M. (1983) Spatial patt erns in the plant community and their effects upon insect research. In: S. Ahmad (Ed.), Host-seeking behavior and mechanisms (pp. 125-157). New York: Academic Press.

PAGE 164

152 Stiling P, Brodbeck BV & Strong DR (1982) Foliar nitrogen and larval parasitism as determinants of leafminer distribution patterns on Spartina alterniflora Ecological Entomology 7: 447-452. Stiling PD, Simberloff D & Anderson LC ( 1987) Non-random distri bution patterns of leaf miners on oak trees. Oecologia 74: 102-105. Stiling P & Simberloff D (1989) Leaf absci ssion: induced defence against pests or response to damage? Oikos 55: 43-49. Stiling P & Rossi AM (1997) Experimental manipulation of top-down and bottom-up factors in a tri-trophic system. Ecology 78: 1602-1606. Stiling P, Rossi AM, Catell M, & Bowdish TI (1999) Weak competition between coastal insect herbivores. Florida Entomologist 82: 599-608. Stiling P, Rossi AM, Hungate B, Dijkstra P, Hinkle CR, Knott WM & Drake B (1999) Decreased leaf-miner abundance in elevated CO2: reduced leaf quality and increased parasitoid attack. Ecological Applications 9: 240-244. Stiling P & Moon D (2005) Quality or quantity: the direct and indir ect effects of host plants on herbivores and their natu ral enemies. Oecologia 142: 413-420. Stone GN & Schnrogge K (2003) The adaptive significance of insect gall morphology. Trends in Ecology and Evolution 18: 512-522. Stone L & Roberts A (1990) The checkerboard score and species distributions. Oecologia 85: 74-79. Strauss SY (1987) Direct and indirect effect s of host-plant fertil ization on an insect community. Ecology 68; 1670-1678. Strong DR, Lawton JH & Southwood R (1984) In sects on plants: comm unity patterns and mechanisms. Blackwell, Oxford. Taylor LR, Woiwod IP & Perry JM (1980) Vari ance and the large scal e spatial stability of aphids, moths and birds. Journal of Animal Ecology 49: 831-854. Tinney GW, Hatcher PE, Ayres PG, Paul ND & Whittaker JB (1998) Interand intraspecies difference in plants as hosts to Tyria jacobaea Entomologia Experimentalis et Applicata 88: 137-145. Valladares GR & Hartley SE (1994) Effect of scale on detecting in teractions between Coleophora and Eriocrania leaf miners. Ecological Entomology 19: 257-262.

PAGE 165

153 Waddell KJ & Mousseau TA (1996) Ov iposition preference hierarchy of Brachys tessellatus (Coleoptera: Buprestidae). Envir onmental Entomology, 25: 63-67. Waddell KJ, Fox CW, White KD & Mousseau TA (2001) Leaf abscission phenology of a scrub oak: consequences for growth and survivorship of a leaf-mining beetle. Oecologia 127: 251-258. Waring GL & Cobb NS (1992) The impact of plant stress on herbivore population dynamics. In: Bernays E (ed) Insect-Plant in teractions, volume IV CRC Press, Boca Raton, FL, 167-226. Waring RH & Price PW (1988) Consequences of host plant chemical and physical variability to an associated herbiv ore. Ecological Re search 3: 205-216. White TCR (1984) The abundance of invertebrate herbivory in relation to the availability of nitrogen in stressed food plants. Oecologia 63: 90-105. Whittaker JB (1971) Population changes in Neophilaenus lineatus (Homopetra: Cercopidae) in different pa rts of its range. Journal of Animal Ecology 40: 425-443. Wiggins DA (1997) Fluctuating asymmetry in Colophospermum mopane leaves and oviposition preference in an African silk moth Imbrasia belina Oikos 79: 484-488 Wilkinson L (1999) SYSTAT: the system for st atistics. Version 9.0. SYSTAT, Evanston, Illinois. Williams IS (1999) Slow-growth, high-mor tality – a general hypothesis, or is it? Ecological Entomology 24: 490-495. Williams IS, Jones TH & Hartley SE (2001) The role of resources and natural enemies in determining the distribution of an in sect herbivore population. Ecological Entomology 26: 204-211. Wunderlin RP & Hansen BF (2000) Atlas of Florida Vascular Plan ts [S M Lry and K N Campbell (application development). Flor ida Center for Community Design and Research, University of South Florida, http://wwwplantatlasusfedu/. Zvereva E, Kozlov M & Haukioja E (1997) Stress responses of Salix borealis to pollution and defoliation. Journal of Applied Ecology 34:1387-1396. Zwolfer H & Stadler B (2004) The organization of phytophagous guilds in Cardueae flower heads: conclusions from null m odels. Evolutionary Ecology Research 6: 1201-1218.

PAGE 166

154 APPENDICES

PAGE 167

155 Appendix 1 – Some leaf and stem gall-formers sampled on Quercus myrtifolia and Q. chapmanii over the range of their distribution in Florida.

PAGE 168

ABOUT THE AUTHOR Tatiana G. Cornelissen, daughter of An tonius W.G. Cornelissen and Wanda G. Cornelissen was born on August 1st 1975 in Belo Horizonte, Brazil. Between 1994 and 2000 she obtained a Bachelor’s Degree in Ec ology and a M.Sc. degree in Ecology and Conservation from the Federal University of Minas Gerais, Brazil. She moved to Florida in August of 2001, where she began a Ph.D. program in Ecology, under the supervision of Dr.Peter Stiling. Her work has been published on Global Change Biology, Oecologia, Oikos and Ecological Entomology. She is expe cted to obtain her Ph.D. degree in May of 2006.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001790119
003 fts
005 20070613131749.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070613s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001486
040
FHM
c FHM
035
(OCoLC)144457643
049
FHMM
090
QH307.2 (ONLINE)
1 100
Cornelissen, Tatiana.
0 245
Herbivory by leaf-miners on Florida scrub oaks
h [electronic resource] /
by Tatiana Cornelissen.
260
[Tampa, Fla] :
b University of South Florida,
2006.
3 520
ABSTRACT: This study investigated effects of plant quality and natural enemies on the abundance and survivorship of several leaf miner species on Florida scrub oaks over several ecological scales. Three oak species (Quercus laevis, Q. geminata, and Q. myrtifolia) and four leafminer species (Acrocercops albinatella, Brachys tesselatus, Stilbosis quadripustulatus, and Cameraria sp. nova) were the main focus of five separate studies, addressing effects of bottom-up and top-down factors at regional, local, and individual scales. At the regional scale, it was observed that Cameraria sp. nova was aggregated into sites, and sites closer to each other exhibited similar densities of mines than sites farther apart. None of the bottom-up and top-down factors studied were spatially structured, but did influence the variation in Cameraria abundance over the range of the host plant Q. myrtifolia. At the local scale, all leaf miners studied were aggregated between and within plants, and variation^ in bottom-up factors among individual plants explained variation in abundance for some of the leaf miners studied. Intra-specific competition was identified as an important factor influencing mine survivorship, but inter-specific competition among leaf miners and gall-formers did not shape the community structure of oak herbivores. Experimental manipulation of bottom-up and top-down factors via fertilization and natural enemy removal showed that bottom-up effects were important determinants of leaf miner abundance, as fertilized plants supported 2 to 5-fold more herbivores than control plants. The removal of natural enemies, on the other hand, did not significantly impact the abundance and/or the survivorship of leaf miners and other guilds studied. At individual scales, it was demonstrated that two leaf miner species responded to random variations in leaf morphology, by increasing in abundance in individual host plants with more asymmetric leaves and/or higher levels of fluctuating as ymmetry. These results offered support for the plant stress hypothesis and differences in host plant quality were again partially responsible for the results found.
502
Dissertation (Ph.D.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 155 pages.
Includes vita.
590
Adviser: Peter Stiling, Ph.D.
653
Host quality.
Enemies.
Quercus laevis.
Quercus geminata.
Quercus myrtifolia.
690
Dissertations, Academic
z USF
x Biology
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1486