Citation
Spatial and Temporal Effects of Fire on Insect Herbivore Community Structure

Material Information

Title:
Spatial and Temporal Effects of Fire on Insect Herbivore Community Structure
Creator:
KIM, TANIA N.
Copyright Date:
2008

Subjects

Subjects / Keywords:
Community structure ( jstor )
Ecology ( jstor )
Fire damage ( jstor )
Herbivores ( jstor )
Herbivory ( jstor )
Infestation ( jstor )
Insect communities ( jstor )
Insects ( jstor )
Leaves ( jstor )
Species ( jstor )
City of Lake Wales ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Tania N. Kim. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
6/30/2007
Resource Identifier:
649810214 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

1 SPATIAL AND TEMPORAL EFFE CTS OF FIRE ON INSECT HERBIVORE COMMUNITY STRUCTURE By TANIA N. KIM A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

2 Copyright 2006 by Tania N. Kim

PAGE 3

3 ACKNOWLEDGMENTS First, I thank my committee members, Bob Holt, Scott Robinson, and Emilio Bruna, for reading my thesis and providing critical feedback at every stage of my research study. I also thank the scientists at Archbold Biological Statio n, particularly Hilary Swain, Eric Menges, Carl Weekley, Mark Deyrup, Patrick Bohlen, Adam Pete rson, Jenny Schafer, and Marcia Rickey, for logistical support and encouragement. In additio n, I thank the Bruna lab for allowing me to use their facilities while on-campus and I thank Pedro Mendez for aiding me with laboratory work. This project was funded by various sources, su ch as the Garden Cl ub of America, the Florida Native Plant Society, a nd the Department of Zoology. Finally, I thank my family and friends for thei r love and encouragemen t, and of course, I thank Brian J. Spiesman and Reginald P. Doggi e for their constant love, patience, humor, and emotional support over the past 6 years.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3 LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES................................................................................................................ .........8 CHAPTER 1 INTRODUCTION................................................................................................................. .13 Conceptual Model............................................................................................................... ....15 The Lake Wales Scrub: A Fi re-Dominated Ecosystem..........................................................16 Study Site..................................................................................................................... ...........18 Study Organisms................................................................................................................ .....19 Insect Herbivores.............................................................................................................. ......20 Overview....................................................................................................................... ..........20 2 EFFECTS OF FIRE ON INSECT HERBIVORE COMMUNITY STRUCTURE................23 Introduction................................................................................................................... ..........23 Methods........................................................................................................................ ..........27 Plot Establishment and Sampling Regime......................................................................28 Leaf traits and Tannin Analyses......................................................................................29 Vegetation Sampling.......................................................................................................30 Fire Regime and Landscape Heterogeneity.....................................................................30 Arthropod Identification..................................................................................................31 Data Analysis.................................................................................................................. .31 Differences in insect community structure...............................................................31 Determinants of insect community structure...........................................................32 Results........................................................................................................................ .............33 Mixed Models..................................................................................................................3 3 Discriminant Analyses.....................................................................................................35 Principal Component Analyses.......................................................................................35 Path Analysis.................................................................................................................. .36 Effects of fire on leaf, plant, and plot architecture...................................................36 Effects of fire on insect communities.......................................................................37 Discussion..................................................................................................................... ..........37 Herbivores of Q. chapmanii ............................................................................................39 Herbivores of Q. inopina .................................................................................................40 Herbivores of Q. geminata ..............................................................................................42 Concluding Comments....................................................................................................42 3 SPATIAL AND TEMPORAL EFFECTS OF FIRE ON PLANT DAMAGE IN THE FLORIDA SCRUB.................................................................................................................8 1

PAGE 5

5 Introduction................................................................................................................... ..........81 Methods........................................................................................................................ ..........85 Plot Establishment and Sampling Regime......................................................................86 Data Analysis.................................................................................................................. .88 Mapping damage patterns........................................................................................88 Path analysis.............................................................................................................88 Results........................................................................................................................ .............89 Inverse Distance Weighting Maps...................................................................................91 Factors Influencing Leaf Tissue Damage........................................................................92 Discussion..................................................................................................................... ..........92 4 CONCLUSIONS.................................................................................................................. 117 LIST OF REFERENCES............................................................................................................. 119 BIOGRAPHICAL SKETCH.......................................................................................................128

PAGE 6

6 LIST OF TABLES Table page 2-1 Variables used in PCA for path analysis............................................................................44 2-2 The number of arthropods by order sustaine d on the 3 species of oaks dominant in the Florida scrub.............................................................................................................. ..46 2-3 Number of arthropod herbivores found on the three dominant oak species in the Florida scrub.................................................................................................................. ....47 2-4 Mixed model results comparing insect herbivore abundance on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub...................................49 2-5 Mixed model results comparing insect herbivore richness on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub.........................................50 2-6 Mixed model results comparing SimpsonÂ’s diversity of insect herbivores on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub...........51 2-7 MANOVA results examining differences in the abundance of herbivores as a function of time since fire, season, and oak species..........................................................52 2-8 Results from discriminant analysis of insect species grouped according to oak species........................................................................................................................ ........54 2-9 Structure matrix from discriminant analysis......................................................................55 2-10 Chi-square statistic results comparing de fault model with a fu lly saturated model..........57 2-11 Direct and indirect effects of fire on insect herbivore abundance.....................................58 3-1 Differences in leaf damage as a functi on of seasonal stage and time-since-fire...............98 3-2 Differences in the average number of leaves consumed per plant as a function of seasonal stage and time-since-fire.....................................................................................99 3-3 Multi-scale patterns of plant da mage as a function of stand age.....................................100 3-4 Variation in the magnitude of leaf damage and spatial extent of damage as a function of stand age................................................................................................................... ...100 3-5 Path analysis results for the direct an d indirect effects of fire on leaf damage...............101 3-6 Path analysis results for the direct and i ndirect effects of fire on the number of leaves damaged per plant............................................................................................................10 1

PAGE 7

7 3-7 Path analysis results for the direct and i ndirect effects of fire on the plants damaged per plot....................................................................................................................... ......101 3-9 Within-plant variation and between -plant variation in leaf traits....................................102

PAGE 8

8 LIST OF FIGURES Figure page 1-1 Conceptual model of the direct and indi rect effects of fire on insect herbivore communities and host plant damage..................................................................................21 1-2 Florida scrub habitat...................................................................................................... ....22 2-1 Five main habitat associations at Archbold Biological Station.........................................59 2-2 Locations of 30 meter by 30 meter resear ch plots in scrubby flatwoods habitat at Archbold Biological Station..............................................................................................60 2-3 Variation in insect herbivore abunda nce among oak species and across season...............61 2-4 Variation in insect herbivore abundan ce among oak species and across time since fire........................................................................................................................... ...........61 2-5 Variation in insect herbivore ric hness among oak species and across season...................62 2-6 Variation in insect herb ivore richness among oak species and across time since fire.......62 2-7 Variation in insect herb ivore diversity (SimpsonÂ’s di versity) among oak species and across season.................................................................................................................. ....63 2-8 Variation in insect herb ivore diversity (SimpsonÂ’s di versity) among oak species and across time since fire......................................................................................................... .63 2-9 Variation in insect abundan ce according to insect order on Q. chapmanii, Q. geminata , and Q. inopina ...................................................................................................64 2-10 Discriminant analysis scatterplot of in sect herbivore commun ities found on the three oak species.................................................................................................................... .....65 2-11 Variation in arthropod abundance of th e significant species from discriminant analyses among the th ree oak species................................................................................66 2-12 Path analyses results for all oak species combined............................................................67 2-13 Path analyses results for Quercus chapmanii ....................................................................68 2-14 Path analyses results for Quercus geminata ......................................................................69 2-15 Path analyses results for Quercus inopina .........................................................................70 2-16 Changes in habitat structure following fire........................................................................71 2-17 Average number of Collembola found on each oak plant species.....................................72

PAGE 9

9 2-18 Average number of Hemiptera found on each oak species................................................72 2-19 Differences in average leaf area among oak species.........................................................73 2-20 Relationship between plant density and plant species richness.........................................74 2-21 Variation in arthropod predat or and parasite abundances as a function of vegetation......75 2-22 Discriminant function analysis of herbivore communities found on Q. chapmanii grouped according to time-since-fire.................................................................................76 2-23 Mean abundance of herbivores on Q. chapmanii , Q. geminata, and Q. inopina as a function of fire return interval............................................................................................77 2-24 Mean richness of herbivore species on Q. chapmanii , Q. geminata, and Q. inopina as a function of fire return interval.........................................................................................78 2-24 Mean abundance of herbivores on Q. chapmanii , Q. geminata, and Q. inopina as a function of habitat richness................................................................................................79 2-25 Mean richness of herbivore species on Q. chapmanii , Q. geminata, and Q. inopina as a function of habitat richness.............................................................................................80 3-1 Variation in leaf chemistry and physical traits among oak species. Leaf density contributed most to the varia tion in leaf physical traits...................................................103 3-2 Average plant height of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub.............................................................................................104 3-3 Average number of branches from prim ary stem of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub.............................................105 3-4 Average number of leaves of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub.....................................................................105 3-5 Variation in the average amount of leaf damage of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina )........................................................................106 3-6 Average leaf damage across seasonal stage for Q. chapmanii , Q. geminata , and Q. inopina .............................................................................................................................10 7 3-7 Average leaf tissue damage across time-since fire for Q. chapmanii , Q. geminata , and Q. inopina ..................................................................................................................108 3-8 Variation in leaf tissue damage for all oak species combined.........................................109 3-9 Average number of leaves consumed per plant across time-since-fire for all three oak species Q. chapmanii , Q. geminata , and Q. inopina combined.......................................110

PAGE 10

10 3-10 Percentage of marked plants affected by herbivore damage as a function of timesince-fire and seasonal stage............................................................................................111 3-11 Leaf damage patterns as a function of seasonal stage......................................................112 3-12 Number of leaves consumed per pl ant as a function of seasonal stage...........................113 3-13 Proposed default model (Model 1) describi ng the effects of fire, plant, and insect variables on leaf tissue damage........................................................................................114 3-14 Proposed default model (Model 2) descri bing the effects of fire and plant level variables on plant damage................................................................................................115 3-15 Spatio-temporal variation in inse ct herbivore abundances per plant...............................116

PAGE 11

11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science SPATIAL AND TEMPORAL EFFE CTS OF FIRE ON INSECT HERBIVORE COMMUNITY STRUCTURE By Tania N. Kim December 2006 Chair: Robert D. Holt Major Department: Zoology In fire-dependent habitats, th e effects of fire on plant comm unities have been extensively studied. However, relatively fe w studies have examined how fi re affects other trophic levels such as insect herbivores. In this study, I exam ined the direct and indirect effects of fire on insect herbivore communities found on three co-occurring oak species ( Quercus chapmanii, Quercus inopina , and Quercus geminata ) in Florida scrub habitat. I also investigated how changes in herbivore communities scaled up to influence damage patterns on the host plant species. The results showed that for herbivores found on Q. inopina , the direct effects of fire may play a large role in herbivore community organi zation. Plants in stands with high fire-return intervals had higher insect he rbivore abundance, richness, and diversity. For herbivore communities found on Q. chapmanii , changes in plot architecture following fire influenced herbivore communities, such that dense stands sustained lower insect numbers and species richness. Finally, for in sect herbivores found on Q. geminata , the surrounding heterogeneity in landscape structure influenced herbivore commun ity structure. Here, stands surrounded by a higher number of different habitat types sustaine d higher abundances and herbivore species.

PAGE 12

12 Changes in herbivore communities in response to fire may in turn influence damage patterns experienced by the host plan t species. The results of my study indicate that variation in vegetation density in response to fire played a large role in determining insect community composition and the subsequent damage they impos ed onto host plant species. For example, densely vegetated plots experienced more plants damaged and more leaf tissue damage per plant. This study showed that insect herbivore assemb lages found on three closel y related plant species responded differently to fire. For land-managers and restoration biologists interested in using prescribed fires as a restorati on tool in fire-dependent habi tat, these results may further complicate restoration efforts because fire regi mes may need to incorporate different specieslevel responses to fire (as well as different ta xon-level) in order to effectively increase local species diversity.

PAGE 13

13 CHAPTER 1 INTRODUCTION The influence of fire in structuring plant communities has been extensively studied in both natural and human-modified systems. However, the effects of fire on higher trophic levels have been less well studied, especially for insect herbivores. Fire can in fluence insect herbivore communities directly thr ough mortality, by movement out of pa tches by individuals fleeing the fire, and by the loss of eggs buried in soil. In direct effects may occur through changes in host plant quality and abundance, and via shifts in hab itat level characteristics following fire (Swengel 2001). Ultimately, both direct and indir ect effects of fire affect herbivore community structure by influencing both rate s of recolonization into newl y burned habitat and population dynamics following such colonization. The indirect effects of fire are mediated in part through changes in host plant abundance and quality, and can operate at multiple spatial scales. At the leaf level, variation in macronutrient and phenolic content may occur fo llowing fire, with recently burned habitat having higher leaf quality in re sponse to increased nutrients re leased into soils (Whelan 1995, Harrison et al. 2003). This increase in food quality could increase the grow th rate and fecundity of herbivorous insects, with cascading effects up the food web. At the plant level, fires can indirectly affect insect herbivore communities through changes in host plant shape and size. Fires can consume entire plants or simply scor ch leaves and lower branches, thereby changing food abundance and the architecture of host plan ts which may lead to increased exposure to predators and environmental stress. These chan ges in microhabitat may influence the abundance and types of insect herbivores that colonize pl ants, and shift the array of cues available to predators searching for prey. At the habitat le vel, variation in plant community composition can occur as pioneer species are replaced by late successional plants. Changes in host plant

PAGE 14

14 abundance through species replacement may influen ce insect herbivore communities, especially if herbivores are host specific. With changes in plant community composition, variation in the three dimensional structure of the habitat ma y occur as dominant plants overshadow other species, and shade tolerant sp ecies grow in the understory (Siemann et al. 1997, Swengel 2001, Collett 2003, Fay 2003). Changes in the three-di mensional structure of the habitat may also impede or facilitate herbivore movement, there by influencing rates of co lonization and herbivore community composition. These effect s can interact at different leve ls. For instance, higher plant quality may attract herbivores to recently burne d habitat, and in some cases their high numbers may negatively affect plant communities through increases in plant damage, thereby offsetting the beneficial effects of increased nutrient lo ading following fire (Radho-Toly et al. 2001). As herbivore communities change in respons e to fire, the subsequent damage they impose on plant communities may vary as well. Selective feeding following fire may lead to increased plant mortality, reduced growth, and lo wer fecundity, which can potentially influence rates of succession (Mills 1983), alter plant hab itat distribution (van Langevelde et al. 2003, Archibald et al. 2005), and shif t the outcome of plant compe tition (Belsky 1992, Fuhlendorf and Engle 2004). These processes are better studied for vertebrate then inverteb rate herbivores. For example, preferential feeding by mammalian herbivores on competitively dominant Ceanothus species in Californian chaparral decreased suc cessional rates by allowing competitively inferior chamise seedlings ( Adenostoma fasciculatum ) to establish in experi mental plots (Mills 1983). Similarly, selective grazing by ungulates on shru b seedlings in recently burned African grasslands reduced shrub encroachment from adjacent unburned habitat (Roques et al. 2001). Finally, selective feeding on competitively dominant plant species following fire can also alter

PAGE 15

15 plant community composition by allowing competitivel y inferior species to continue to exist, thereby increasing plant species divers ity (Belsky 1992, Harr ison et al. 2003). Most studies examining the interactive effect s of fire and herbivory have focused on the effects of mammalian herbivores on landscape patterns and plant community dynamics (van Langevelde et al. 2003, Mills and Fey 2005). However, the interactive effect s of fire and insect herbivory is receiving increased attention in the ecological litera ture (Alfaro et al. 1999, Carson and Root 2000, Santoro et al. 2001, Bishop 2002). In applied ecology, there was a traditional focus on fire as a means for insect control (Lemonnier-Darcemont 2003, Vermeire et al. 2004), but there is now increasing intere st in understanding how diversit y patterns of insects change following fire, and how this knowledge can bene fit conservation and restoration efforts (Di Giulio et al. 2001, Swengel 2001). The overall goal of this thesis is to examin e the spatial and temporal effects of fire on insect herbivore communities, and the reciprocal damage patterns insect herbivores impose on their host plants. Oak species ( Quercus sp.) provide excellent model organisms to study influences on insect community structure, as th ey house a large suite of insect herbivores. In central Florida, several species of oak are the dominant plant species in fire-dependent scrub habitat, providing a readily avai lable setting to examine the eff ects of fire on insect community organization. Conceptual Model A conceptual model illustrating the direct and i ndirect effects of fire on insect herbivore community structure and then plan t damage can be seen in Figure 1-1. Again, the direct effects of fire on herbivore communities can be mortality, m ovement out of habitat, and/or loss of eggs buried in soil. Indirect effects are mediat ed through host plant quality and abundance, and habitat level characteristics. Pl ant quality at the leaf level includes palatabi lity of leaves through

PAGE 16

16 variation in nutrient levels, and chemical and phy sical defenses. Variation of plant quality at the entire plant level includes suitability of host plan t for protection against predators and against the stresses of the physical environment. Habitat-le vel characteristics can also influence herbivore community structure, by facilita ting or restricting movement be tween and within patches, or through host plant availability. Finally, lands cape-level variation can influence herbivore communities by serving as source habitats for both insect herbivores and their natural enemies, and by proving allochthonous nutri ents to focal habitats thro ugh runoff and other processes (Polis et al. 1997, Vanni et al. 2004). These direct and indirect effects of fire on herbivore community st ructure can lead to spatial and temporal variation in plant damage. For example, gall damage may dominate earlier in the growing season and on younger plants with softer and highly nutritional leaves (Fonseca et al. 2006). On the other hand, extensive chewing damage by grasshoppers may occur later in the growing season, when nymphs emerge from so ils (J. Capinera, personal communication). Changes in herbivore community structure can influence the amount and type of damage experienced on a leaf, as well as th e number of plants affected by he rbivory in a particular area. These chains of effects on the insect herbivore co mmunity can ultimately feed back to influence the dynamics of the plant communities. Because the following study took place within a short time scale, such feedbacks into plant communities, habitats, and landscapes, can only be addressed as speculative hypotheses for future work. The Lake Wales Scrub: A Fire-Dominated Ecosystem An understanding of the interpla y of fire, plant ecology, and in sect herbivory is relevant to many habitats of conservation concern. Centra l Florida is considered a hotspot for biological diversity, endemism, and species endangerment, with the third highest number of federally threatened and endangered species in the Un ited States, following Hawaii and California

PAGE 17

17 (Dobson et al. 1997, USFWS 1999). Florida is also ranked the third highest in the nation for population change from 2000-2004 (8.8% versus th e national average of 3.8% (USBC 2004)). As a result, natural habitats such as the Florida scrub are quickly diminishing in size due to rapid urbanization, agriculture (primarily citrus produ ction), and fire suppression. Today, less than 40% of pre-settlement scrub habitat remains in Fl orida, mostly in isolated fragments (see Figure 1-2A; Abrahamson 1984, Myers 1990). The result ing competition for space between humans and the natural environment has resulted in a di re need for conservation management plans to protect the remaining scrub habitat from further species loss and habitat degradation. Ideally, such management plans would be based upon a firm ecological understanding of the dynamics of these complex ecosystems. The Florida scrub is considered to be a shr ub habitat characterized by evergreen oaks and rosemary shrubs, with a sparse overstory of pine tr ees (Myers 1990). It is a fire-dependent habitat, where fire frequencies range from 1-100 years depending many factor s such as elevation, soil moisture, and landscape configurations (M enges and Hawkes 1998). One of the largest, continuous patches of Florida scr ub is located on the Lake Wale s Ridge in Polk and Highlands counties, Florida (Figure 1-2A). The Lake Wale s Ridge is a series of relic sand dunes located 300 feet above sea level, and prior to the Pleist ocene, when much of the state was submerged, this chain of islands running down central Florida remained above water (Myers 1990). As a result of this long period of ge ographic disjunction, a high level of endemism occurs within the ridge. A substantial and ongoing challenge for conservation biol ogists and landowners is to preserve and restore the remaining 30% of pre-se ttlement scrub habitat on the Lake Wales Ridge (Myers 1990).

PAGE 18

18 The Florida scrub is home to a wide suite of endemic plants and animals such as the gopher tortoise ( Gopherus polyphemus ), Florida mouse ( Podomys floridanus ), pygmy fringetree ( Chionaznthus pgymaeus ), and the short-leaved rosemary ( Conradina brevifolia ), (Abrahamson 1984, Myers 1990). The two most widely reco gnized endemic taxon of the Florida scrub are the Florida scrub jay ( Aphelocoma c. coerulescens) and Garrett’s Ziziphus ( Ziziphus celata ). Over the past few decades, the federally enda ngered Florida scrub jay has experienced a population decline of over 90%, l eaving only 4000 breeding pairs worldwide (Woolfenden et al. 1996). The main cause of the decline is habita t loss due to citrus production, urbanization, and fire suppression (Bowman and Wool fenden 2002). Garrett’s Ziziphus is an extremely rare shrub and was considered extinct in the mid-1950’s until it was rediscovered in 1987 (Myers 1990). Ziziphus is a self-incompatible , clonal shrub, and loss of ge netic diversity through habitat fragmentation is responsible fo r its major decline (Weekley et al. 2002). There are many other species that have suffered the same fate as well. Study Site The research reported in this thesis was c onducted at Archbold Biological Station (ABS) in 2005. ABS is an 8000acre natural preserve lo cated at the southern tip of the Lake Wales Ridge near Lake Placid, Florid a (27º11’N, 81º21’W, Highlands Co., FL). There are 5 main habitat associations th at make up the Florida scrub at AB S (Figure 1-2B, (Abrahamson 1984)). Each main habitat association is further divided into sub-catego ries according to the dominant plant species found in the main habitat associatio n. The dominant habitat association at ABS is scrubby flatwoods, which provides the focal habita t of this research. Scrubby flatwoods are characterized by low growing shrubs (1-2m) in terspersed with slash pines (Abrahamson 1984, Myers 1990). The shrubs are largely comprised of several species of oaks, but other species cooccur with the oaks in lower numbers (e.g. saw palm ettos, staggerbushes, and blueberry shrubs).

PAGE 19

19 Fire return intervals in scr ubby flatwoods are between 5-20 years (Menges and Hawkes 1998). The oaks are the dominant plant in this commun ity and provide the focal taxa for this study. Study Organisms There are three dominant oak spec ies in the scrubby flatwoods: Quercus inopina Ashe (Archbold oak), Q. chapmanii Sargent (Chapman oak), and Q. geminata Small (sand live oak)(Abrahamson 1984). The distribution of Q. inopina is limited to north-central peninsular Florida, whereas Q. geminata and Q. chapmanii occur throughout the stat e, in high and frequent occurrences, respectively (Wunderlin 1998). All th ree shrub species are low to moderate in stature (typical heights are 1-2 meters), flower in spring (March-May), and produce fruit in the fall (September-November). Quercus geminata , Q. inopina , and Q. chapmanii co-exist only in the scrubby flatwoods. Large-scale factors determining patterns of species co-existence for an assembly of oak species were studied in north-central Fl orida by Cavender-Bares et al. ( 2004). In this large-scale study, the authors described how 17 species of oaks co exist regionally. Variation in soil moisture, soil fertility, and fire regime we re responsible for habitat pa rtitioning among the congeners. However questions remained as to how different species of oa k could coexist locally. After constructing a phylogeny, it was discovered that dist antly related species were most likely to coexist at the local scale than were closely relate d species. The authors s uggested various possible mechanisms of local co-existence, one of which was partitioning herbivores. It was suggested that variation in susceptibility to herbivores could promote the coexistence of distantly related species, through the impact of vari ation in leaf traits such as secondary chemistry and phenology on herbivore preference and performance.

PAGE 20

20 Insect Herbivores Oak species are known to house a large suite of insect herbivores. Studies at ABS have focused on gall forming insects, with the oak species housing as many as 28 gall species in a single growing season (Price et al. 2004). To my knowledge, no prior studies have looked at relationships of generalist herbivores, such as grasshoppers, to oak species. However grasshoppers have been seen feeding on oak leaves during insect surveys as part of the Florida scrub jay long-term demographic study (R. Bowma n, personal communication). Studies in scrub habitat at the Kennedy Space Center (Brevard Co ., FL) have also observed leaf-mining moths on all three oaks species, such as Stilobosis sp., Stigmella sp., Cameraria sp., and Buccalatrix sp. (Stiling et al. 2003). It is an open question as to how insect herbivore assemblages on these oak species are influenced by factors such as season, time-since-fire, and landscape structure. Overview In this thesis, I will describe the spatial and temporal effects of fire on insect herbivore communities in the Florida scrub. Chapter 2 will de scribe factors that influence insect herbivore community structure and composition, and how th ese factors differ between three oak species ( Q. chapmanii , Q. geminata , and Q. inopina ). Chapter 3 will descri be spatial and temporal patterns of insect damage on oak foliage as well as investigate the factors that influence damage. Chapter 4 will then summarize results from chapte rs 2 and 3, and provide concluding comments.

PAGE 21

21 Figure 1-1. Conceptual model of the direct and indirect effect s of fire on insect herbivore communities and host plant damage. The dash ed arrows depict feedbacks for insect herbivory to different aspects of pl ant ecology, and the in sects themselves.

PAGE 22

22 Figure 1-2. Florida scrub habitat. A. The distribution of remaining scrub habitat in Florida. B. The main vegetation association at Archbol d Biological Station in Highlands, Co., FL (Abrahamson 1984).

PAGE 23

23 CHAPTER 2 THE EFFECTS OF FIRE ON INSECT HERBIVORE COMMUNITY STRUCTURE Introduction Understanding factors that regulate community structure is a centr al theme in ecology (Dayton 1971, Krebs 2001, Paine 1974). For insect herbivores, many studie s have investigated determinants of abundance and community compos ition, at multiple spatial scales (Price 1997). At the gene level, variation in host plant ge notypes can influence se lection by individual herbivorous insects for leaves and oviposition sites (Cronin and Abrahamson 2001, Wimp et al. 2005). Similarly, at the leaf leve l, variation in nutrient conten t and leaf shape can influence herbivore communities through selective feedi ng and oviposition habitat selection (Underwood 2004, Boege 2005, Helms and Hunter 2005). At larger spatial scales, host plant density can influence insect herbivore communities such that plants found in higher densities may support higher numbers of insect herbivores, simply because these plants are easier to find (Wiens et al. 1991, Coley and Barone 1996, Marques et al. 2000). Similarly, host plant range can influence insect herbivore communities in that widely di stributed plants may sustain higher numbers of insect herbivore species due to larger re gional species pool (C ornell and Lawton 1992). Although the relative strengths of these determin ants on insect herbivore communities may vary according to habitat and feeding guilds, the ge neral consensus appears to be that insect community structure is regulated by multiple factors, and these factors vary or interact with each other at different scal es (Strong et al. 1984). In fire-maintained systems, the effects of fi re can influence insect herbivore community structure, and these effects can be direct or indirect. Di rect and indirect effects of fire can also vary temporally, and therefore be further catego rized as immediate or long-term effects (Swengel 2001). Direct, immediate effects of fire on indi vidual insects include movement out of burned

PAGE 24

24 habitats, direct mortality due to fire, or through loss of eggs buried in soils. The indirect effects are mediated through host plants, and immediate, i ndirect effects include st arvation due to loss of host plant species and exposure to pr edators (Warren et al. 1987). Long-term effects of fire on insect herbivore communities may also be direct or indirect. Direct effects included variation in fire intensity, area and extent, and fire return interval, all of which may influence the type of herbivore specie s that colonize into a habitat. For example, sites that experience frequent di sturbance events may favor highly mobile, generalist species, and the frequency of such events may prevent other species from reaching and establishing themselves at that same site (Swengel 2001). Studi es have shown that frequent fire events can decrease insect diversity (Wa rren et al. 1987, Panzer and Schw artz 2000, Swengel 2001). This has raised concern among insect biologists, as the us e of prescribed fires is increasingly used as a restoration tool in fire-dependent ecosystems. Indirect, long term effects of fire on insect herbivore communities are mediated through impacts on host plant species. Through sele ctive feeding, microhabi tat preferences, and dispersal to burned patches, these effects may influence insect recol onization and operate at multiple spatial scales. At the leaf level, variat ion in nutrient and chemical content of leaves may differentially attract insect herb ivores to recently burned patche s. There are numerous studies that have demonstrated increased nitrogen content in leaves follo wing fire, due to the increased nutrient availability in soils (Whelan and Main 1979, Radho-Toly et al. 2001). Higher leaf quality may then lead to increased insect a bundance and richness. Af ter severe fires which eliminate local insect populations, early insect communities shortly after fires may be largely dominated by generalists or highl y mobile insects, such as gr asshoppers and certain butterfly species (Swengel 1996, 2001).

PAGE 25

25 At the plant level, variation in plant size and shape may al so influence the abundance and richness of insect herbivores (Mopper et al. 2000, Boege 2005). As plants recover from fire, changes in their shapes and sizes will occur, thereby resulting in variation in microhabitat environments. Immediately after a fire, foliage will be essentially non-existent, so individual insects will be exposed to full array of ambient light and wind conditions. As foliage recovers, greater opportunities aris e for microhabitat sele ction to avoid harsh physical conditions. Preferences of insect herbivores for microhabita t environments may then lead to changes in insect herbivore communities as plants continue to grow and fill into the surrounding space. At the habitat level, as time passes after a fi re, variation in host pl ant abundance, density, and the three-dimensional structure of the habita t will occur, all of which can then influence herbivore community composition. As plants quickly occupy available space following fire, plant density will increase and may influence herbivore community structure (Price 1991, Stein et al. 1992). For example, herbivores favoring en closed habitats, or shade leaves, may increase as foliage density increases. Likewise, cha nges in plant species composition through succession may alter host plant density, qualit y, and suitability and these ch anges in turn may lead to changes in herbivore abundance (Strong et al. 198 4). For example, as overall plant density increases, insect herbivore feeding on rare plant species may simply bypass some patches or have difficulties locating their host plant species. Habitat-level featur es may also influence herbivore communities by impeding or facilita ting insect movement into and around habitat (Basset et al. 2001, Swengel 2001). If fires are intense, local ex tinctions may occur and in thes e cases, herbivore community structure may also depend on the proximity of in sect herbivores to s ource habitats. Surrounding unburned habitat may serve as refuges for insect herbivores during fire, and colonization back

PAGE 26

26 into the burned habitat may depend on proximity to habitat edges (Bishop 2002, Knight and Holt 2005). For example, Knight and Holt (2005) found higher herbivor e abundances along edges of burned habitat compared to interior burned habita t. This resulted in higher plant damage along edges; however, this spatial gradient in damage dissipated through time, as insect herbivores made their way towards the habitat interior. Ther efore, the amount of edge habitat or proximity to source habitats may influence the amount and type of insect herbivores found at a site, and this dependency may shift with time since a major fire. Compared to other taxa in fire-dominated sy stems, relatively few studies have looked at the effects of fire on insect community compos ition (but see Swengel 2001). Most studies have investigated the effects of fire as a means of insect control, for ex ample, the use of fire to control grasshoppers in prairie systems (Vermeire et al . 2004) or to prevent sp ruce budworm infestation in northern boreal forests (Anders on et al. 1987). However, as prescribed fires are becoming increasingly accepted as a restoration tool in fire-m aintained systems, an interest in the effects of fire on insect diversity is be ginning to grow in the ecological literature (Warren et al. 1987, Schwartz 1994, Panzer and Schwartz 2000, Di Giulio et al. 2001, Cook and Holt 2006). The goal of this study is to characterize a number of factors that influence insect herbivore community composition in fire-maintained Florida scrub. Specifically, the goals are (1) to determine whether there are differences in insect herbivore community structure among three dominant oak species ( Quercus geminata , Q. inopina , and Q. chapmanii ); (2) to describe how fire influences insect communities, and whethe r these effects are direct or indirect; and (3) to characterize how fire effects change as a function of stand age (t ime-since-fire), and throughout the growing season.

PAGE 27

27 Oaks have a long evolutionary history with their insect herbi vores providing ample opportunity for specialization (Stron g et al. 1984). The three conge ners are also phylogenetically distantly related (Caven der-Bares et al. 2004), th erefore I predicted that insect herbivore communities would differ between the three oak species (Goal 1). Sin ce the direct, immediate effects of fire via mortality and movement are t ypically seen shortly afte r fire and can dissipate within a single year (Knighst and Holt 2005, Swenge l 1996), I predicted that the effects of fire on these herbivore communities to have been larg ely indirect and long-term effects, reflecting changes in leaf quality, plant arch itecture, and habitat-level featur es (Goal 2). In addition, since many studies have described increased insect abundances following fire (Radho-Toly et al. 2001), I predicted that insect herb ivore communities would vary as a function of time-since-fire, with higher abundances in recently burned pa tches, but higher richness in older patches (Southwood et al. 1979, Price 1991). Fi nally, I predicted seasonal vari ation in insect community structure with abundances decrea sing throughout the growing season in response to decreases in temperatures and humidity (Goal 3). Methods Research was conducted at Ar chbold Biological Station (ABS) in 2005. ABS is an 8000acre natural preserve and is located at the southern tip of the Lake Wales Ridge near Lake Placid, Florida (27º11’N, 81º21’W, Highla nds Co., FL). The Lake Wales Ridge is an endangered shrub habitat and is home to a wide suite of endemic pl ants and animals, such as the Florida Scrub Jay ( Aphelocoma c. coerulescens ), gopher tortoise ( Gopherus polyphemus ), and short-leaved rosemary ( Conradina brevifolia ). This habitat is quickly di minishing in size due to rapid urbanization, citrus production, and fire suppressi on, and efforts exist to re store and maintain the remaining 30% of pre-settlement scr ub found on the Lake Wales Ridge.

PAGE 28

28 There are 5 main habitat associations that make up the Florida scrub at ABS (Abrahamson 1984). The dominant association is the scrubby flatwoods which provides the focal habitat of this research (Figure 2-1). Scrubby flatwoods is a xeromorphic shrub habitat, characterized by low growing shrubs (1-2 m) interspersed with slash pines, Pinus elliottii (Abrahamson 1984, Myers 1990). Shrubs by biomass are largely comprised of oaks, but other species of shrubs also co-occur in lower numbers (e.g., Vaccinium sp ., Sabal sp ., and Lyonia sp . ). Prescribed fires are conducted routinely at AB S as a restoration and management tool. Plot Establishment and Sampling Regime Fifteen 30 meter by 30 meter plots were establ ished in scrubby flat woods habitat along a chronosequence of time-since-fire (TSF) (Figure 2-2) . Plots were establishe d near fire lanes to facilitate access; however, they were placed 20 meters from edges to reduce edge effects on insect communities. Locations were chosen base d on size and proximity to other habitat types. For example, sites needed to be large enough to accommodate a 30m by 30m plot and at least 20m from other habitat types so as to minimize the effects of insect herbivores from other habitat types on herbivore communities in the scrubby flat woods. Five TSF intervals were used that span the natural range of fire frequencies of th e scrubby flatwoods (1, 4, 6, 11, 19 years since last fire). Three 30m by 30m plots were established within each time-since-fire interval. Within each plot, 45 individuals of each oak species ( Quercus geminata, Q. inopina, and Q. chapmanii ) were randomly selected and marked for repeated sampling. On each plant, 10 newly flushed leaves were randomly selected and marked with inde lible ink. All three species of oak are evergreen. A short period exists wh ere old and new leaves co-occur on plants (approximately 2 weeks). New leaves were easily distinguishable from old leaves in both texture and color. Plots were established in May 2005. Plant height, diameter at base, number of

PAGE 29

29 branches from main stem, and number of leaves we re measured in an effo rt to characterize plant shape and size (hereafter collectively refe rred to as “plant architecture”). Plants and insects were sampled in July, September, and November 2005, (hereafter referred to as “seasonal stage”). During each se asonal stage, 5 of the marked plants were randomly selected per plot and sampled for insect s using a sweep net (15 inches in diameter). Twenty sweeps across the entire pl ant were conducted. The insect abundance data consisted of the combined set of 20 sweeps per plant. From these plants, 10 marked leaves were removed, and leaf area, density, toughness, thickness, mass, and total tannin concentration were measured in an effort to characterize “leaf traits”. This was done for each plant species, resulting in a total of 15 plants sampled per plot pe r seasonal stage (15 plants x 15 time-since-fire plots x 3 seasonal stages = 675 plants sampled). Since I did not sample the same plants across seasonal stages, genetic differences in leaf and plant architecture were not cont rolled. This method of using different plants across seasonal stages was chos en, because sweeping and removal of leaves for analyses were destructive methods, and repeated sampling of the same individual plants would bias sequential censuses. Leaf traits and Tannin Analyses The following physical characteristics of l eaves were measured: thickness, toughness, density, area, and mass. Leaf toughness was measur ed using a penetrometer (Imada Inc. Force Gauges, series FB). Penetrometers measure the amount of force required to penetrate a leaf by a rod of determined size, and are frequently used as a measure of leaf toughness in analyses of herbivory (Sanson et al. 2001). The rods used in this study were 2mm in diameter, which was small enough to pass through the veins of leaves . Leaf area was measured using a LICOR 3000 leaf area meter; density was measured by simp ly dividing leaf volume by dry leaf mass.

PAGE 30

30 The total amount of tannin per leaf (mg of tannin/100 mg of tissue) was determined using the radial diffusion method desc ribed by Hagerman (1987). Th is method involves extracting tannins from leaves using methanol at a ratio of 0.5 milliliters of metha nol per 0.1 gram of plant tissue. Three 3 aliquots of 8 mi croliters of tannin extract were pl aced into wells (4 millimeters in diameter) in agar plates that were infused with a protein (e.g., bovine serum albumin). As the tannin extract diffuse across the plate, tannins bi nd to proteins and a circular precipitate forms around the wells. The amount of ta nnin is correlated to the area of the precipitate. A standard curve with known quantities of tannin (mg) was established us ing tannic acid, and used to determine quantities of tannin in leaf samples. The radial diffusion method is a simple, highly accurate, and cost effective method for determini ng total tannin concentration, and the technique is generally used in experiments with large sample sizes (Hagerman 1987). Vegetation Sampling In September, vegetation within plots wa s sampled using the point-intercept method (Wilson 1960). For this method, four equally spaced 30 meter transects were established within each plot, and each plant touching a 0.5 cm in diam eter rod placed every 2 meters was recorded. Plant species and height of interception were re corded. These data were used to calculate vegetation abundance, vegetation height, vegetation density, plan t species richness, and host plant abundance as predictor variables for the analyses reported below. Fire Regime and Landscape Heterogeneity In addition to time-since-fire, average fire retu rn interval was also determined using fire history data provided by ABS (available at http://www.archboldstation.org/abs/inde x.htm). Fire return interval is defined as the time between two successive fire events. Surrounding habitat may also influence herbivore communities, in that the surrounding la ndscape provides source habitat for cross-system flows of herbivores, predators and nutrients (Polis et al. 1997).

PAGE 31

31 Landscape heterogeneity was determined using detailed vegetation maps provided by ABS in ArcView 3.3 (ESRI 1999). Circles, referred to as “buffers”, of fixed radii (ranging from 20m to 200m in size at 25m increments) were centered at each plot, and used to establish habitat composition of surrounding areas. The type of hab itat, amount of differe nt habitat, length of edge, and a shape index of the focal habitat were determined for each buffer size-class. These landscape metrics are frequently used in landscape eco logy to describe as pects of landscape complexity (Riitters et al. 1995). Buffers of 100m in size were used in the following analysis for they proved to exhibit the grea test variance in landscape heterogeneity between the TSF intervals. Arthropod Identification All arthropods were identified to order, a nd herbivores were iden tified to the species level. Specimens were verified with voucher sp ecimens harbored at the Florida State Arthropod Collection at the Division of Plant Industry (G ainesville, FL), and with specimens at the Archbold Biological Station Arth ropod Collection at Archbold Bi ological Station (Lake Placid, FL). A voucher collection has been assembled a nd curated by the author. All analyses were performed using only arthropod herbivores. Data Analysis Differences in insect community structure Mixed models were used to compare differences in insect herbivore abundance among the three oak species, among seas onal stage, and across time-sinc e-fire (TSF). Fixed effects were oak species; time-since-fire ; and seasonal stage, whereas ra ndom effects were plots nested in time-since-fire. Mixed models were also used to compare insect herbivore richness and Simpson’s diversity index. Herbivore abundanc e was square-root transformed, and diversity was log-root transformed, so as to meet the assumption of normality. Mu ltiple analysis of

PAGE 32

32 variance (MANOVA) was performed to compare the abundance of herbivores belonging to different insect orders among the three oak sp ecies, time-since-fire, and seasonal stage. Discriminant analysis using herbivore species da ta was performed to determine dissimilarities between insect herbivore comm unities. All analyses were done using SPSS v.11.5 (SPSS 2002). Determinants of insect community structure Path analysis was used to describe the f actors that influenced herbivore community structure for all three oak species, both combined and separately. Path analysis is a form of structural equation modeling which provides a conve nient tool for using multiple regression data to predict causal relationships between many vari ables. Path analysis compares observational data to causal hypotheses and distinguishes relati onships as being direct or indirect (Mitchell 2001). Relationships indicated by one-way a rrows imply causation, whereas double-headed arrows indicate unanalyzed patterns of correlation. Path coefficients (b eta weights in regression analyses) indicate the strength of re lationships. Error terms are also assigned to variables, so as to account for variation that is not e xplained by the measured variables. In order to test the fit of a proposed model (hereafter referred to as the “default” model), path analysis compares the default model to a fully saturated model (a model that fits any dataset), and an independent model (a conser vative model where variables are treated as completely unrelated). A 2 goodness of fit test is performed to compare the fit of the default model to the fully saturated model. Default models that are not significantly different from the fully saturated model have P-values greater th an 0.05, and indicate a good fit (Shipley 2000). Path analysis was performed in AMOS v5.0 (Arbuckle 1999). Because 24 predictor variables we re measured at various scales (e.g., leaf, plant, plot, and landscape levels), principal component analyses (PCA) were used to reduce the number of

PAGE 33

33 variables for the default model. PCA reduced vari ables at the leaf-level to form one predictor variable referred to as “leaf traits”; one predictor variable to represent plant level traits (hereafter referred to as “plant architecture”); one to represen t plot-level traits (herea fter referred to as “plot architecture”); and one for landscape-level traits (hereafter referred to as “landscape heterogeneity”). PCA reduced the original 24 vari ables to 4 manageable variables. In total, seven predictor variables were used in the path analyses which included leaf traits, plant architecture, plot architecture, landscape heterogeneity, time-sincefire, fire return interval, and seasonal stage. With a sample of 225 plants per species (75 plants per species per seasonal stage), this analysis met the requirement of a 5: 1 ratio of observations to predictor variables for path analysis (Shipley 2000). PCA was pe rformed in SPSS v.11.5 (SPSS 2002). Variables used in the PCA analyses are compiled in Table 2-1. Results The arthropod sampling protocol resulted in 3559 specimens: 26.89% Collembola (1 species), 22.79% Hemiptera (21 species), 14.16% Acari (1 family), 11.86% Araneae, 10.20% Diptera (10 families), 5.09% Hymenoptera (6 families), and 4.33% Coleoptera (16 species), 2.78% Orthoptera (7 species), 1.12% Lepidoptera (3 familes), and 10 other orders, each consisting of less than 1% of the arthropod total (T able 2-2). Forty-four insect herbivore species were identified to the species level; these included external-feeding herbivores such as beetles, grasshoppers and their allies, springtails, and spider mites. Internal-feeding herbivores included gall-makers and leaf miners. The following analyses focused on phytophagous (foliage-feeding) insects only (Table 2-3). Mixed Models Mixed model results indicate significant differences in insect abundance among oak species (F 2,560 = 5.35, p=0.000), among seasonal stages (F 2,560 = 37.84, p = 0.000), by time-

PAGE 34

34 since-fire (F4,70 = 4.93, p=0.000), and interactions among thes e factors (Table 2-4). The highest average abundances occurred during July and Se ptember (Figure 2-3). The relative ranking of each oak species seems to shift with seasonal stage. On average, Q. chapmanii showed the highest abundance, followed by Q. inopina , and then Q. geminata . Q. chapmanii seemed to show the clearest pattern of abundance related to time-since fire, with an initial increase, followed by a decrease in the older plots (Figure 2-4). Variation in insect richness was al so significant among oak species (F 2,630 = 3.094, p=0.046), among seasonal stage (F 2,630 = 28.402, p = 0.000), and by time-since-fire (F4,630 = 6.872, p=0.000). Interactions were not significant (T able 2-5). Again, July and September had the highest insect herbivore ri chness, and intermediate time-si nce-fire plots (TSF 4 and 6) exhibited high species richness as well (Figures 2-5 and 2-6). Ther e were significant differences in SimpsonÂ’s index of diversit y between seasonal stages (F 2,560 = 18.4, p=0.000), and timesince-fire (F 2,560 = 5.83, p=0.000), (Table 2-6). Once again, insect diversity was higher in July and September, and also in intermediate time-s ince-fire plots, TSF 4 and 6, (Figures 2-7 and 28). MANOVA results comparing differences in in sect abundance according to insect order, between oak species, time-since-fi re, and seasonal stage are pres ented in Table 2-8. Seasonal stage had significant effects on a ll insect orders except for Le pidoptera; time-since-fire had significant effects only on Collembola and Diptera; and oak species had a significant effect on herbivore abundance for Collembola, Coleoptera, and Hemiptera. Interaction effects varied and are reported in Table 2-7. Figur e 2-9 shows the differences in herbivore abundance as a function of insect orders among the three oak species.

PAGE 35

35 Discriminant Analyses The discriminant analysis results showed significant overlap among the herbivore communities found on the three oak sp ecies (Table 2-8). Despit e significant overlap between herbivore communities (Figure 2-10), clear differences in insect composition existed (Table 2-8). (Table 2-9). For example, Hysteropterum fuscomaculosus (Issidae: Hemiptera) , Neochlamisus insularis (Chrysomelidae: Coleoptera), Metachroma anaemicum (Chrysomelidae: Coleoptera), Cedusa sp. (Cicadellidae: Hemiptera), Tetranychidae (Acari), Jikadia melanota (Cicadellidae: Hemiptera); Dicyrtoma atra (Collembola), and Anchastus asper (Elateridae: Coleoptera) showed the greatest variation between the three oak species (Table 210). Variation in abundance of these species between the three oak speci es can be seen in Figure 2-11. Principal Component Analyses Variables used for PCA analyses and their c ontributions to variati on are shown in Table 2-1 for all oak species combined and separately. For all oak species combined, the first PCA axis for leaf traits explained 47.7% of the variation in leaf level trai ts, and leaf area contributed the most to this axis. The first PCA axis for pl ant architecture explained 76.3% of the variation, with the number of leaves havi ng contributed to the majority of this variation. The first PCA axis for plot architecture explai ned 46.9% of variation in plot level features, and plot density contributed to the majority to this varia tion. Finally, the first PCA axis for landscape heterogeneity explained 70.73% of variation in landscape levels features, with patch richness (the number of different habitat types within a 100m buffer) contributing mo st to the variation. The contributions of each measured variable to the first PCA axis varied with oak species and are summarized in Table 2-1.

PAGE 36

36 Path Analysis Figure 2-12 shows the path diagram for the pr oposed effects of pr edictor variables on herbivore response variables (that is, herbivore abundance, richness, and diversity) for all three oak species combined. Chi-square tests showed default models as no t being significantly different than fully saturated models, thus havi ng indicated a good fit (Tab le 2-10). Significant pathways are shown by thick arrows. Values above rectangular boxes indicate the percent of variation explained by predictor variables. Ta ble 2-11 and Figures 2-13 through 2-15 illustrate the direct, indirect, and total e ffects of predictor va riables on herbivore response variables for each oak species. Effects of fire on leaf, plant, and plot architecture Fire had significant influence at the leaf, plan t, and plot-level arch itecture (Figures 2-12 to 2-15). Specifically, time-since-fi re had significant effects on plan t and plot architecture for all three oak species, and also in fluenced leaf traits for Q. geminata . Fire return interval had a strong influence on plant and plot architect ure for all three oak species, and landscape heterogeneity influenced plant architectur e for all three oak species as well. The influence of fire on habitat structure wa s significant, with both time-since-fire and fire return interval having strong influences on pl ot architecture. Plot ar chitecture was greatest in intermediate time-since-fire plots (TSF 4, 6, 11), versus younger (TSF 1) and older (TSF 19) plots. Space is a limiting factor in many terres trial plant communities, so as habitats mature following fire disturbance, plants quickly o ccupy available space until they become dense in structure (Abrahamson 1984). Beyond this point, die-backs occur as plants experience selfthinning, thus opening up habitat, leaving a mixture of tall and short plants at the site. Figure 216 shows the transition from an open habitat to a dense, intermediate-aged habitat, back to an older, more open habitat. Younger plots were more open, and plants were also shorter in height.

PAGE 37

37 Intermediate time-since-fire plots were denser, with taller plants that showed little variation in height. Older plots had on average taller vegetation with greater va riation in plant height, and so there was more small scale heterogeneity. Plot de nsity decreased in older plots, as plants died back due to senescence or crow ding (Figure 2-16). These ar chitectural changes might be expected to be reflected in changes in the insect herbivore communities found on these oaks. Effects of fire on insect communities Using the combined dataset for insects found on all three species of oaks, seasonal stage was found to strongly influence in sect herbivore abundance, richne ss, and SimpsonÂ’s diversity. Leaf traits also contributed to herbivore abundance and richness, but not to diversity (Figure 212). For herbivore communities found on Q. chapmanii , the influence of fire was indirect, and mediated through plot architecture for herbivor e richness and diversity. The total amount of variation in herbivore abundance, richness and diversity that wa s explained with the significant response variables were 11%, 12%, and 10%, respectively (Figure 2-13). For herbivore communities found on Q. inopina , fire-return interv al had a direct influence on all three insect response variables. Total vari ation explained was 19%, 14%, and 9% for herbivore abundance, richness, and dive rsity respectively (Figur e 2-15). For insect herbivores found on Q. geminata , landscape heterogeneity contribu ted significantly to insect herbivore communities. Total variation in herbivore abundance, richness, and diversity, explained by significant response variables were 11%, 9% and 8% respectively (Figure 2-14). Discussion According to the assembly rules of commun ity organization, the interplay of habitat filtering and species traits plays a large role in determining community composition (Diamond and Case 1986). Habitats serve as filters, elimin ating species whose traits are not suitable for a particular environment. For insect herbi vore communities found on different oak species, the

PAGE 38

38 palatability of leaves, their ease of mastication, and th e suitability of leaves as oviposition sites can be considered as filters that influence herbivore abundance and community composition. When combining all three oak spec ies together in the path analys is reported above, leaf traits significantly explained herbivore abundance and richness. Collembola was the dominant insect herbivore found on the thin, tannic leaves of Q. chapmanii . The tiny mandibles of these insects may explain why higher abundances we re found on the thin leaves of Q. chapmanii, versus the thicker leaves of Q. inopina and Q. geminata (Figure 2-17). On the other hand, high numbers of Hemiptera were found on Q. inopina , where leaves had lower concen trations of tannins and were less dense than Q. geminata leaves (Figure 2-18). Although studies examining the effects of leaf chemistry and physical structure on hemipteran feeding strategies are scant in the ecological literature, studies examining leaf selection for ovipositing sites can provide some insight into why high numbers were observed on Q. inopina . Female Hemiptera lay eggs inside leaf veins and typically prefer leaves with large vein s (Sharma and Singh 2002, Lokesh and Singh 2005). Q. inopina had larger leaves, (and perhaps larger veins), which may explain the higher abundances of Hemiptera observed on Q. inopina plants (Figure 2-19). Insect herbivore abundance a nd richness were lowest on Q. geminata leaves, which had the lowest amount of tannins. However, this sp ecies also had extremely dense and tough leaves, and these structural defenses may have served as filters in limiting the number and types of insect herbivores found on these plants. When analyzing herbivore community data separately according to oak species, the direct and indirect effects of fire influenced herb ivore communities differently across oak species, despite strong similarities in community compos ition. Below, I describe how direct and indirect factors may have influenced the compositi on of insect herbivore communities.

PAGE 39

39 Herbivores of Q. chapmanii For insect herbivores found on Q. chapmanii , the indirect effects of fire on herbivore community structure were mediated through plot ar chitecture. Plant dens ity contributed greatly to plot architecture. When pl otted as a function of time-sin ce-fire, a distinct hump-shaped pattern emerged, with the greate st density occurring in TSF 6 and 11. The relationship between plant density and herbivore abundance or richness may in part be due to heterogeneity of plant species composition (Tahvanainen and Root 1972). For example, plants found in dense monospecific stands typically e xperience higher numbers of insect herbivores and plant damage; this has been a premise for using polyculture cr ops in agricultural systems (Letourneau 1995). The ability of plants to escape herbivory in polyspecific stands is called “associational resistance” (Tahvanainen and Root 1972, Rand 1999, Hamback et al. 2000). Since plant species richness in the scrubby flatwoods increased as a function of plant density (Figure 2-20), perhaps the observed lower insect herbivor e abundances and richness found on Q. chapmanii was a result of associational resistance. Therefore, Q. chapmanii plants may have been able to escape herbivory by simply being in a habitat with many other plant species. Alternatively, plot density may have influen ced the abundance of predators and parasites which in turn, may have influenced herbivor e abundance and composition (Price 1997, Strong et al. 1994). Dense plots had slightly higher numbers of predators which may have resulted in low herbivore richness and diversity. Figure 2-21 show s a slight increase in predator numbers as a function on plant density, although th is pattern was not statistically significant. However, since scrub jays prefer to feed in dense stands of oaks where acorn abundance are highest (Bowman and Woolfenden 2002, Fleck and Woolfenden 1997), perhaps scrub jays played a role in regulating herbivore abundance and richness in th ese stands, as well as arthropod predators.

PAGE 40

40 The effects of access to focal habitat a nd movement around habitat may also have influenced insect herbivore community com position (Tilman 1994, Basset et al. 2001). For example, since plant density and insect richne ss and diversity exhibite d a negative relationship, perhaps dense plots impeded insect movement. Figure 2-22 is a scatterplot of a discriminant analysis illustrating how herbivore communities found on Q. chapmanii differ according to timesince-fire. Plants that experienced fire six years prior (TSF 6) harbor ed a distinct insect herbivore community, compared to the other plots. Results from a structure matrix showed that planthoppers ( Hysteropterum fuscomaculosum) and microlepidopter ans (Gracillariidae) characterized herbivore communities of Q. chapmanii at TSF 6. Planthoppers appeared in lower numbers than expected in TSF 6 plots than in other TSF intervals, whereas micrlepidopterans appeared in higher numbers than expected. For weak fliers such as microlepidopterans, dense vegetation may have facilitated movement be tween plants in effect by providing “stepping stones”. In a study examining butterfly and mo th recovery following fire (Powell 1995), the rates of colonization for microlepidopterans were extremely low compared to strong fliers such as the orange sulfur ( Colias eurytheme ) and the acmon blue ( Plebejus acmon ). However, locations that experienced less in tense fires had a slightly higher rate of colonization, suggesting that perhaps surviving plants se rved as refuge for lepidopterans during fire, and/or facilitated movement back into newly burned habitat. Herbivores of Q. inopina For herbivore communities found on Q. inopina , the effects of fire were direct. More specifically, the average fire-ret urn interval exhibited a positi ve relationship with herbivore abundance, richness, and diversity. The time betw een disturbance events may have influenced community structure by preventing certain herbivore species from successfully establishing into frequently burned plots (Southwood et al. 1979, Siemann et al. 1999). For example, studies

PAGE 41

41 examining the role of ploughing on insect comm unities in grasslands have found that long, undisturbed grasslands supported higher abundances and numbers of insects compared to frequently ploughed fields (Corbet 1995). Si milarly, studies by Siem ann et al. (1999) found higher herbivore richness in olde r successional fields compared to young fields. However, this study attributed higher insect richness to higher plant diversity, rather than to the direct effects of disturbance in and of itself, as a source of variat ion in insect richness. Hence, the relationship between insect species richness and dist urbance frequency is still unclear. For herbivores feeding on Q. inopina , plots with longer fire-re turn intervals had higher herbivore numbers and herbivore species. However, this pattern was only seen in Q. inopina (Figures 2-23 and 2-24). For this species, plots with longer fire -return intervals had more species and abundance of Hemipterans. Frequent distur bance events may have prevented Hemipteran species from successively establishing at these sites. This result contrasts with the findings of a study conducted in a tropical eucal ypt forests in Australia, which examined the effects of high fire frequency on the abundance of phytophagous insects (Fensham 1994). Their results indicated that forests with higher fire freque ncy supported higher numbers of Hemipteran and Orthopteran species. However, th e Australian study used forest he ight as an indicator of fire frequency and so did not discriminate the effects of fire on herbivore communities that are direct from those that are indirectly mediated thr ough vegetation height. The results of Fensham (1994) study simply show that tall forests ha d higher abundance of Hemipteran; hence, the question of how fire acts as a disturbance to influence insect communities remains open. Perhaps the discrepancy between my results and t hose of Fensham (1994) may be in part due to the types of Hemipterans found on plant species. For example, generalists may prefer to feed

PAGE 42

42 and oviposit in frequently burned habitat, whereas specialists (or less gene ral insect herbivores) may prefer long-unburned stands. The role of disturbance on species richness is a growing concern in insect grassland communities. As prescribed fires are becoming mo re prevalent as a mean for habitat restoration, concerns over their effects on inse ct diversity is growing in the ecological literature (Warren et al. 1987, Swengel 1996, 2001, Cook and Holt 2006). My study provides evidence that although prescribed fire may have beneficial effects on pl ant diversity in fire-maintained systems, these benefits may not translate up to insect herbivores. Herbivores of Q. geminata For insect herbivores of Q. geminata , landscape heterogeneity had a strong, positive effect on herbivore communities (Figures 2-24 a nd 2-25). Landscape heterogeneity was largely driven by habitat richness (that is, the number of different hab itat types within a 100m buffer), therefore Q. geminata located in plots surrounded by high numbers of habitat types also supported greater number of insect herbivore specie s. Because no species of insect herbivore showed strong preference for Q. geminata (Figure 2-9), perhaps herbivore community structure on this oak species was driven by proximity to di fferent habitat types. Proximity to source habitat s has been shown to influence community stru cture in island biogeography, where habitats located close to source populat ions experience higher rates of colonization and thus elevated species richness (MacArthur and Wilson 1967). If different habitat types support different suites of herbivore species, this may explai n why insect herbivore community composition on Q. geminata changed as a function of landscape heterogeneity. Concluding Comments In fire-maintained systems, the effects of fi re on herbivore insect communities can vary according to the specific identity of host plant species. In this study, habita t filtering through

PAGE 43

43 variation in leaf quality was obs erved and predicted to have strong influences on community structure. However, other factors such as plot architecture, fire-retur n interval, and landscape heterogeneity were also shown to have a great influence on comm unity organization. Specifically, fire-return interval influe nced insect herbivore communities found on Q. inopina , and the indirect effects of fire on vegetation density and landscape heterogeneity influenced herbivore assemblages on Q. chapmanii and Q. geminata , respectively. Although insect herbivore communities on the th ree congeners were similar, there were enough dissimilarities in community structure to perhaps explain why different patterns in community organization emerged. Additionally, the three oak species are distantly related therefore any species-level differenc es in leaf-level and plant-level traits may have influenced the mechanisms of community organization. As prescr ibed fires are becoming increasingly used as a restoration tool in fire-dependent habitats, th e impact of such management strategies on plants and higher trophic levels must be considered for effective habitat management. In addition, to fully understand determinants of community struct ure following fire, it is important to examine the rules of assembly at multiple scales, from th e leaf, to the plot, and to the entire landscape levels.

PAGE 44

44Table 2-1. Variables used in principal com ponent analyses (PCA) for path analyses. Variables with the greatest contributions to predictor variable (PCA axis 1) are show with an asterisk. All oak species combined Q. chapmanii Q. geminata Q. inopina Predictor variable (PCA axis 1) Variables used in PCA Contributions to predictor variable % variation explained by predictor variable Contributions to predictor variable % variation explained by predictor variable Contribution s to predictor variable % variation explained by predictor variable Contributions to predictor variable % variation explained by predictor variable Leaf traits Leaf area (cm2) 0.949* 47.70% 0.886 38.15% 0.83 36.45% 0.903 36.86% Average width (cm) 0.768 0.863 0.935* 0.913* Thickness (mm) -0.743 0.362 0.089 0.205 Maximum width (cm) 0.721 0.866 0.931 0.893 Toughness (kg) -0.686 0.242 -0.232 0.149 Density (cm3/g) -0.67 0.24 0.099 0.175 Length (cm) 0.576 0.203 -0.328 0.107 Amount of tannin (mg/100mg of leaf tissue) 0.494 0.251 0.117 0.187 Mass (g) 0.486 0.895* 0.811 0.853 Plant architecture Number of leaves 0.932* 76.33% 0.943* 80.34% 0.954* 82.24% 0.952* 80.70% Number of branches 0.899 0.93 0.867 0.929 Plant diameter at base 0.862 0.857 0.95 0.873 Plant height 0.812 0.858 0.876 0.823

PAGE 45

45Table 2-1. Continued All oak species combined Q. chapmanii Q. geminata Q. inopina Predictor variable (PCA axis 1) Variables used in PCA Contributions to predictor variable % variation explained by predictor variable Contributions to predictor variable % variation explained by predictor variable Contribution s to predictor variable % variation explained by predictor variable Contributions to predictor variable % variation explained by predictor variable Plot architecture Vegetation density 0.942* 46.89% 0.923* 47.74% 0.948* 53.21% 0.940* 46.51% Vegetation abundance 0.942* 0.923* 0.948* 0.940* Openness -0.872 -0.839 -0.892 -0.868 Vegetation height 0.658 0.695 0.598 0.668 Variance in vegetation height 0.517 0.564 0.469 0.528 Oak abundance 0.202 -0.352 0.744 0.048 Vegetation richness -0.073 -0.081 -0.004 -0.08 Landscape heterogeneity Habitat richness 0.939* 70.73% 0.939* 70.73% 0.939* 70.73% 0.939* 70.73% Total edge 0.932 0.932 0.932 0.932 Diversity index of habitat types 0.906 0.906 0.906 0.906 Number of patches 0.775 0.775 0.775 0.775 Mean shape index 0.602 0.602 0.602 0.602

PAGE 46

46 Table 2-2. The number of arthropods by order su stained on the 3 dominant oak species in the Florida scrub. Order Quercus chapamanii Quercus geminata Quercus inopina Total % of total Acari 192 187 125 504 14.16% Araneae 151 138 133 422 11.86% Blattodea 1 3 12 16 0.45% Coleoptera 34 48 72 154 4.33% Collembola 414 229 314 957 26.89% Dermaptera 0 0 1 1 0.03% Diptera 127 114 122 363 10.20% Hemiptera 243 221 347 811 22.79% Hymenoptera 56 74 51 181 5.09% Isoptera 1 0 0 1 0.03% Lepidoptera 16 13 11 40 1.12% Odonata 0 0 2 2 0.06% Orthoptera 36 29 34 99 2.78% Phasmatodea 0 0 1 1 0.03% Pscoptera 2 0 0 2 0.06% Pseudoscorpiones 1 0 0 1 0.03% Thysanoptera 0 2 0 2 0.06% Tricoptera 0 0 1 1 0.03% Zoraptera 0 1 0 1 0.03%

PAGE 47

47 Table 2-3. Number of arthr opod herbivores found on the thre e dominant oak species in the Florida scrub. Species Name Order Quercus chapmanii Quercus geminata Quercus inopina Total Anchastus asper Hebard Coleoptera 0 5 1 6 Aptenopedes hubbelli Hebard Orthoptera 17 7 8 32 Aptenopedes nigropicta Hebard Orthoptera 9 10 7 26 Aptenopedes sphenarioides Scudder Orthoptera 1 0 2 3 Balcutha sp . Hemiptera 0 0 1 1 Brachys areosus Melsheimer Coleoptera 1 0 3 4 Cecidomyidae* Diptera 65 64 48 177 Cedusa sp . Hemiptera 10 1 2 13 Cerambycidae sp . Coleoptera 0 1 1 2 Chortophaga australiur Rehn and Hebbard Orthoptera 0 0 1 1 Ciidae* Coleoptera 0 0 1 1 Coleoptera (unknown larvae) Coleoptera 8 3 7 18 Deltocephalis obtectus Osborn and Ball Hemiptera 0 0 1 1 Dendrocoris fruticicola Bergroth Hemiptera 3 5 8 16 Dicyrtoma atra Linnaeus Collembola 414 229 314 957 Epicauta heterodera Horn Coleoptera 0 0 1 1 Erythroneura comes Say Hemiptera 36 33 54 123 Eutettix nitens Van Duzee Hemiptera 3 12 12 27 Excultanus excultus Uhler Hemiptera 1 0 0 1 Flatoidinus punctatus Walker Hemiptera 4 2 2 8 Gracillariidae* Lepidoptera 9 8 6 23 Gyponana fastige Delong Hemiptera 13 5 5 23 Homoeolabus analis Illiger Coleoptera 1 0 0 1 Hymenorus fuscipennis Fall Coleoptera 0 1 0 1 Hysteropterum fuscomaculosum Deering Hemiptera 67 102 148 317 * Individuals identified to family level; ** Individuals identified to order level.

PAGE 48

48 Table 2-3. Continued Species Name Order Quercus chapmanii Quercus geminata Quercus inopina Total Idioderma virescens Van Duzee Hemiptera 2 1 0 3 Jikradia melanota Spanberg Hemiptera 93 56 110 259 Lepidoptera (unknown larvae)** Lepidoptera7 3 5 15 Lycaenidae* Lepidoptera0 1 0 1 Melanoplus forcipatus Hubbell Orthoptera 0 1 2 3 Metachroma anaemicum Fall Coleoptera 3 8 17 28 Microcentrus perditus Amyot and Seville Hemiptera 1 0 0 1 Neochlamisus insularis Shaeffer Coleoptera 3 16 17 36 Notolomus basilis LeConte Coleoptera 5 5 6 16 Odontoxiphidium apterum Morse Orthoptera 3 3 1 7 Orocharis luteolira Walker Orthoptera 6 8 13 27 Pachybrachis conformis Suffrain Coleoptera 3 2 0 5 Pelitropis rotulata Van Duzee Hemiptera 3 1 0 4 Polana quadrinotata Spanberg Hemiptera 2 0 2 4 Scathophagidae* Diptera 1 0 0 1 Tetranychidae* Acari 192 187 125 504 * Individuals identified to family level; ** Individuals identified to order level.

PAGE 49

49 Table 2-4. Mixed model results comparing insect herbivore a bundance on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub. Results indicate significant differences in herbivore abunda nces across time-since-fire (TSF), among oak species, and across seasonal stage. In teractions of TSF and oak species, and seasonal stage and oak species were also significant. Source Numerator df Denominator df F P Intercept 1 70 948.17 0.00 Seasonal stage 2 560 37.84 0.00 TSF 4 70 4.92 0.00 Species 2 560 5.35 0.00 Seasonal stage * TSF 8 560 0.82 0.59 Seasonal stage * Species 4 560 2.74 0.03 TSF * Species 8 560 2.22 0.02 Seasonal stage * Species * TSF 16 560 1.08 0.38

PAGE 50

50 Table 2-5. Mixed model results comparing insect herbivore richness on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub. Results indicate significant differences in herbivore richne ss across time-since-fire (TSF); among oak species, seasonal stage, and the inte raction of TSF with oak species. Source Numerator df Denominator df F P Intercept 1 630 1927.73 0.00 Seasonal stage 2 630 28.40 0.00 TSF 4 630 6.87 0.00 Species 2 630 3.09 0.05 Seasonal stage * TSF 8 630 0.86 0.55 Seasonal stage * Species 4 630 0.27 0.90 TSF * Species 8 630 1.75 0.08 Seasonal stage * Species * TSF 16 630 1.20 0.26

PAGE 51

51 Table 2-6. Mixed model results comparing SimpsonÂ’s diversity of insect herbivores on three dominant oaks (Q. chapmanii, Q. geminata, and Q. inopina ) in the Florida scrub. Results indicate significant differences in herbivore diversity acr oss time-since-fire (TSF) and seasonal stage. Source Numerator df Denominator df F P Intercept 1 70 2133.51 0.00 Seasonal stage 2 560 18.40 0.00 TSF 4 70 5.83 0.00 Species 2 560 1.87 0.16 Seasonal stage * TSF 8 560 1.19 0.30 Seasonal stage * Species 4 560 0.23 0.92 TSF * Species 8 560 1.23 0.28 Seasonal stage * Species * TSF 16 560 1.27 0.21

PAGE 52

52 Table 2-7. MANOVA results examin ing differences in the abundance of herbivores according to order as a function of time-since-fire (TSF), seasonal stage, and oak species (Species) . Results indicate great variation in the effects of time-since-fire, oak species, and seasonal stage and their inter actions on the different arthropod orders. Source Dependent Variable Type III Sum of Squares df Mean Square F P Coleoptera 3.28 2 1.64 6.92 0.00 Collembola 167.34 2 83.67 25.76 0.00 Diptera 1.99 2 0.99 3.01 0.05 Hemiptera 71.65 2 35.82 21.49 0.00 Lepidoptera 0.10 2 0.05 0.73 0.48 Seasonal stage Orthoptera 2.05 2 1.02 6.16 0.00 Coleoptera 1.85 5 0.37 1.56 0.17 Collembola 39.61 5 7.92 2.44 0.03 Diptera 19.86 5 3.97 12.01 0.00 Hemiptera 15.31 5 3.06 1.84 0.10 Lepidoptera 0.21 5 0.04 0.64 0.67 TSF Orthoptera 1.82 5 0.36 2.19 0.05 Coleoptera 2.14 2 1.07 4.53 0.01 Collembola 50.48 2 25.24 7.77 0.00 Diptera 0.78 2 0.39 1.18 0.31 Hemiptera 49.08 2 24.54 14.72 0.00 Lepidoptera 0.03 2 0.01 0.21 0.82 Species Orthoptera 0.22 2 0.11 0.67 0.51

PAGE 53

53 Table 2-7. Continued Source Dependent Variable Type III Sum of Squares df Mean Square F P Coleoptera 1.91 10 0.19 0.81 0.62 Collembola 31.92 10 3.19 0.98 0.46 Diptera 16.03 10 1.60 4.85 0.00 Hemiptera 28.78 10 2.88 1.73 0.07 Lepidoptera 0.40 10 0.04 0.62 0.80 Seasonal stage * TSF Orthoptera 1.33 10 0.13 0.80 0.63 Coleoptera 0.28 4 0.07 0.30 0.88 Collembola 41.90 4 10.47 3.23 0.01 Diptera 0.70 4 0.18 0.53 0.71 Hemiptera 2.76 4 0.69 0.41 0.80 Lepidoptera 0.36 4 0.09 1.38 0.24 Seasonal stage * Species Orthoptera 1.22 4 0.31 1.83 0.12 Coleoptera 2.65 10 0.27 1.12 0.35 Collembola 52.56 10 5.26 1.62 0.10 Diptera 6.85 10 0.69 2.07 0.03 Hemiptera 25.34 10 2.53 1.52 0.13 Lepidoptera 0.99 10 0.10 1.52 0.13 TSF * Species Orthoptera 1.83 10 0.18 1.10 0.36 Coleoptera 5.60 20 0.28 1.18 0.26 Collembola 96.41 20 4.82 1.48 0.08 Diptera 6.50 20 0.33 0.98 0.48 Hemiptera 30.43 20 1.52 0.91 0.57 Lepidoptera 1.28 20 0.06 0.99 0.47 Seasonal stage * TSF * Species Orthoptera 2.43 20 0.12 0.73 0.80

PAGE 54

54 Table 2-8. Results from discrimi nant analysis of insect herbi vore species grouped according to oak species. Predicted Group Membership Species Q. chapmanii Q. geminata Q. inopina Total Original Count Q .chapmanii 110 73 42 225 Q. geminata 47 132 46 225 Q. inopina 50 71 103 224 % Q. chapmanii 48.9 32.4 18.7 100 Q. geminata 20.9 58.7 20.4 100 Q. inopina 22.3 31.7 46 100

PAGE 55

55 Table 2-9. Structure matrix from discriminant analysis indicati ng insect herbivore species with the greatest variation betw een the three oak species (s hown with an asterisk). Discriminant Function Herbivore 1 2 Hysteropterum fuscomaculosus -.409(*) 0.15 Neochlamisus insularis -.244(*) -0.112 Metachroma anaemicum -.243(*) 0.119 Cedusa sp . .215(*) 0.135 Tetranychidae .208(*) -0.207 Eutettix nitens -.186(*) -0.085 Pelitopis rotulata .168(*) 0.001 Gyponana fastige .163(*) 0.077 Aptenopedes nigropicta .156(*) -0.082 Aptenopedes hubbelli .155(*) 0.073 Orocharis luteolira -.137(*) 0.084 Dendrocoris fruticicola -.126(*) 0.053 Melanoplus forcipatus -.125(*) 0.032 Idioderma virescens .125(*) -0.031 Homoeolabus analis .119(*) 0.056 Excultanus excultus .119(*) 0.056 Microcentrus perditus .119(*) 0.056 Scathophagidae .119(*) 0.056 Pachybrachis conformis .118(*) -0.062 Cerambycidae -.085(*) -0.039 Flatoidinus punctatus .085(*) 0.04 Gracillariidae .063(*) -0.033

PAGE 56

56 Table 2-9. Continued Discriminant Function Herbivore 1 2 Jikradia melanota -0.046 .460(*) Dicyrtoma atra 0.278 .400(*) Anchastus asper -0.088 -.295(*) Brachys areosus -0.085 .194(*) Erythroneura comes -0.118 .178(*) Polana quadrinotata 0.023 .167(*) Lycaenidae -0.024 -.166(*) Hymenorus fuscipennis -0.024 -.166(*) Aptenopedes sphenarioides -0.042 .160(*) Coleoptera (unknown larvae) 0.042 .143(*) Cecidomyidae 0.115 -.120(*) Lepidoptera (unknown larvae) 0.074 .116(*) Epicauta heterodera -0.096 .110(*) Chortophaga australiur -0.096 .110(*) Manomera sp -0.096 .110(*) Deltocephalis obtectus -0.096 .110(*) Ciidae -0.096 .110(*) Balcutha sp . -0.096 .110(*) Odontoxiphidium apterum 0.073 -.083(*) Notolomus basilis -0.025 .028(*)

PAGE 57

57 Table 2-10. Chi-square statisti c results comparing default model with a fully saturated model. Since P is greater than 0.05, default models used for path analyses indicated a goodfit. Chi-square df P Herbivore abundance 0.08 5 1.00 Herbivore species richness 0.081 5 1.00 Herbivore Simpson's Diversity 0.082 5 1.00

PAGE 58

58 Table 2-11. Direct and indirect effects of fire on insect herbivore abundance (A); species richness (B), and SimpsonÂ’s diversity (C) for Q. chapmanii , Q. geminata , and Q. inopina combined. A. Direct Indirect Total Seasonal stage -0.273 -0.014 -0.286 p=0.000 Landscape heterogeneity 0.004 0.007 0.011 Fire return interval 0.066 0.006 0.073 Time-since-fire -0.038 0.002 -0.036 Leaf traits -0.095 0.001 -0.094 p=0.013 Plot architecture -0.013 0.013 0 Plant architecture 0.008 0 0.008 B. Direct Indirect Total Seasonal stage -0.248 -0.008 -0.256 p=0.000 Landscape heterogeneity -0.001 0.005 0.004 Fire return interval 0.06 0.023 0.083 Time-since-fire -0.022 -0.01 -0.031 Leaf traits -0.055 0.001 -0.055 Plot architecture -0.038 0.008 -0.03 Plant architecture 0.013 -0.004 0.009 C. Direct Indirect Total Seasonal stage -0.2 -0.004 -0.204 p=0.000 Landscape heterogeneity 0.006 0.002 0.008 Fire return interval 0.046 0.041 0.087 Time-since-fire -0.022 -0.016 -0.038 Leaf traits -0.03 0.001 -0.028 Plot architecture -0.068 0.004 -0.064 Plant architecture 0.032 -0.009 0.023

PAGE 59

59 Figure 2-1. Five main habitat associations at Arc hbold Biological Station (modified from Abrahamson 1984).

PAGE 60

60 Figure 2-2. Locations of 30 meter by 30 meter re search plots in scrubby flatwoods habitat at Archbold Biological Station.

PAGE 61

61 0 1 2 3 4 5 6 7 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean herbivore abundance per plant July September November Figure 2-3. Variation in insect herbivore abundance among oak sp ecies and across season. Error bars are +/1 SE. Time-since-fire0 1 2 3 4 5 6 7 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean herbivore abundance per plant 1 4 6 11 19 Figure 2-4. Variation in insect herbivore abundance among oak sp ecies and across time since fire. Error bars are +/1 SE.

PAGE 62

62 0 0.5 1 1.5 2 2.5 3 3.5 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean herbivore richness per plant July September November Figure 2-5. Variation in insect herbivore richness among oak sp ecies and across season. Error bars are +/1 SE. Time-since-fire0 0.5 1 1.5 2 2.5 3 3.5 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean herbivore richness per plant 1 4 6 11 19 Figure 2-6. Variation in insect herbivore richness among oak speci es and across time since fire. Error bars are +/1 SE.

PAGE 63

63 0 0.5 1 1.5 2 2.5 3 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean Simpson's diversity per plant July September November Figure 2-7. Variation in insect herbivore diversity (SimpsonÂ’s diversity) among oak species and across season. Error bars are +/1 SE. Time-since-fire0 0.5 1 1.5 2 2.5 3 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean Simpson's diversity per plant 1 4 6 11 19 Figure 2-8. Variation in insect herbivore diversity (SimpsonÂ’s diversity) among oak species and across time since fire. Error bars are +/1 SE.

PAGE 64

64 0 0.5 1 1.5 2 2.5 Q. chapmaniiQ. geminataQ. inopina Oak speciesMean Herbivore Abundance Acari Coleoptera Collembola Diptera Hemiptera Lepidoptera Orthoptera Figure 2-9. Variation in insect a bundance according to insect order on Q. chapmanii, Q. geminata , and Q. inopina. Error bars are +/1 SE.

PAGE 65

65 Figure 2-10. Discriminant analysis scatterplo t of insect herbivore communities found on the three oak species. Each point represents herbivore communities found on one individual plant. Although overlap exists in herbivor e assemblages found on the oak species, there are also considerable ar eas of non-overlap between the herbivore communities, indicating some levels of community dissimilarity.

PAGE 66

66 0 0.5 1 1.5 2 2.5 Q. chapmaniiQ. geminataQ. inopinaOak speciesMean abundance per plant Anchastus asper Cedusa sp. Dicyrtoma atra Hysteropterum fuscomaculosum Jikradia melanota Metachroma anaemicum Neochlamisus insularis Tetranychidae Figure 2-11. Variation in arth ropod abundance of the significant species from discriminant analyses among the three oak species.

PAGE 67

67 Figure 2-12. Path analyses results for all oak species combined. Significant negative relationships are indicated by bold, dashed arrows, whereas significant positive relationships are indicated by bold, solid arrows. Numbers located on top of boxes indicate the amount of vari ation explained by predicto r variables. Amount of variation explained by predic tor variables for herbivore abundance, richness, and diversity is .10, .08, .05, respec tively. Seasonal stage and leaf traits are the only predictor variables affecting all three insect response variables.

PAGE 68

68 Figure 2-13. Path analyses results for Quercus chapmanii . Significant negative relationships are indicated by bold, dashed arrows, whereas significant positive relationships are indicated by bold, solid arrows. Numbers located on top of boxes indicate the amount of variation explained by predictor variable s. A. Amount of variation explained by predictor variables for herb ivore abundance is 0.11 and only seasonal stage affected herbivore abundance. B. Amount of va riation explaining herbivore richness and diversity are 0.12 and 0 .10, respectively. Season al stage and plot architecture affect herbivore richness and diversity.

PAGE 69

69 Figure 2-14. Path analyses results for Quercus geminata . Significant negative relationships are indicated by bold, dashed arrows, whereas significant positive relationships are indicated by bold, solid arrows. Numbers located on top of boxes indicate the amount of variation explained by predictor vari ables. Amount of variation explained by predictor variables for herb ivore abundance, richness, and diversity is .11, .09, .06, respectively. Seasonal stage and lands cape heterogeneity ar e the only predictor variables affecting all three insect response variables.

PAGE 70

70 Figure 2-15. Path analyses results for Quercus inopina . Significant negative relationships are indicated by bold, dashed arrows, whereas significant positive relationships are indicated by bold, solid arrows. Numbers located on top of boxes indicate the amount of variation explained by predictor vari ables. Amount of variation explained by predictor variables for herb ivore abundance, richness, and diversity is .19, .14, .09, respectively. Seasonal stage and fire return interval are the only predictor variables affecting all three insect response variables.

PAGE 71

71 Figure 2-16. Changes in habitat structure fo llowing fire. Panel on far left shows a re cently burned plot; middle panel shows a plot 6 years following fire; far right panel shows a long unburned plot (19 years post-fire). As plots age, average plant height and variation in height increases, however, plant density exhibi ts a hump-shaped pattern.

PAGE 72

72 0 0.5 1 1.5 2 2.5 3 Q. chapmaniiQ. geminataQ. inopina Oak speciesAverage number of Collembola per pla n Figure 2-17. Average number of Collembola found on each oak plant species. Error bars are +/1 SE. 1 1.05 1.1 1.15 1.2 1.25 1.3 Q. chapmaniiQ. geminataQ. inopina Oak speciesAverage number of Hemiptera per plant Figure 2-18. Average number of Hemiptera found on each oak species. Error bars are +/1 SE.

PAGE 73

73 0 1 2 3 4 5 6 7 8 Q. chapmaniiQ. geminataQ. inopina Oak speciesAverage leaf area (cm2) Figure 2-19. Differences in average leaf area (cm2) among oak species. Error bars are +/1 SE.

PAGE 74

74 Time-since-fire4 6 8 10 12 14 16 18 11.522.5 Vegetation densityPlant species richness 1 4 6 11 19 Figure 2-20. Relationship between plant density and plant species richne ss. In most time-sincefire plots, there was a positive relationship with plant density and species richness.

PAGE 75

75 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 11.21.41.61.822.22.4 Vegetation densityMean abundance of arthropods per plant arthropod predators arthropod parasites Figure 2-21. Variation in ar thropod predator and parasite abundances as a function of vegetation.

PAGE 76

76 Time-since fire -30 -20 -10 0 10 20 30 -50050100150 Function 1Function 2 1 4 6 11 19 Figure 2-22. Discriminant function anal ysis of herbivore communities found on Q. chapmanii grouped according to time-since-fire (T SF). Herbivore assemblages found on Q. chapmanii in TSF 6 were very different in herbivore community composition compared to assemblages in other time-since-fire plots.

PAGE 77

77 r2 = 0.0038 r2 = 0.0053 r2 = 0.4889 0 1 2 3 4 5 6 7 8 24681012141618 Fire-return intervalMean herbivore abundance per plant Q. chapmanii Q. geminata Q. inopina Figure 2-23. Mean abundance of herbivores on Q. chapmanii , Q. geminata, and Q. inopina as a function of fire return interval.

PAGE 78

78 r2 = 0.4767 r2 = 0.0303 r2 = 0.3893 0.5 1 1.5 2 2.5 3 3.5 4 24681012141618 Fire-return intervalMean herbivore richness per plant Q. chapmanii Q. geminata Q. inopina Figure 2-24. Mean richne ss of herbivore species on Q. chapmanii , Q. geminata, and Q. inopina as a function of fire return interval.

PAGE 79

79 r2 = 0.4361 r2 = 0.9237 r2 = 0.6291 0 1 2 3 4 5 6 7 8 345678910 Surrounding habitat richnessMean herbivore abundance per plant Q. chapmanii Q. geminata Q. inopina Figure 2-24. Mean abundance of herbivores on Q. chapmanii , Q. geminata, and Q. inopina as a function of habitat richness (the number of different habitat types within a 100m buffer centered in the middle of the plot).

PAGE 80

80 r2 = 0.4059 r2 = 0.8369 r2 = 0.7302 1 1.5 2 2.5 3 3.5 4 345678910 Surrounding habitat richnessMean herbivore abundance per plant Q. chapmanii Q. geminata Q. inopina Figure 2-25. Mean richne ss of herbivore species on Q. chapmanii , Q. geminata, and Q. inopina as a function of habitat richness (the number of different habitat types within a 100m buffer centered in the middle of the plot).

PAGE 81

81 CHAPTER 3 SPATIAL AND TEMPORAL EFFE CTS OF FIRE ON PLANT DAMAGE IN THE FLORIDA SCRUB Introduction Disturbances have long been recognized as playing a major role in shaping both animal and plant communities (White 1979, Sousa 1984). Since the 1980Â’s, the field of disturbance ecology has gained tremendous momentum as ecologists recognize the important impact disturbances have on community organization (Resh et al. 1988). Disturba nces can affect plant and animal communities by influencing compe titive interactions (Dayton 1971), species diversity (Connell 1979), and su ccession (Knowlton 1992), and disturbances may also affect abiotic processes such as nutrient cy cling and energy flow (Sousa 1984). Fire and herbivory are two important dist urbances that influence plant community structure (White 1979, Sousa 1984). In a recen t paper, Bond and Keeley (2005) describe similarities of herbivory and fi re in structuring plant communities. Both processes consume plant matter and convert this matter to orga nic compounds, and both can be destructive to individual plants. However, entire populations and communities of plants may benefit from both processes. For example, in the case of fire, nut rients released into the soils following fire can greatly benefit plant communities by increasing pl ant growth and promoting seed germination from underground seedbanks (Keeley 1987, Bell et al. 1993). In some systems, plants may even be selected to foster fire as a weapon of interspecific competit ion. In these systems, flammable plants sacrifice themselves and neighboring plants thereby crea ting gaps in the canopy for their offspring (Schwilk and Kerr 2002). In addition, when there is competition for space, increased availability of space following fire can create opportunities fo r population expansion (Roques et al. 2001). Although plants are consumed in fire , the long-term benefits of fire may include

PAGE 82

82 increased plant recruitment, diversity, and distribution (Clark 1991, Hawkes and Menges 1996, Harrison et al. 2003). Herbivore effects also vary and range from being harmful to beneficial (Maschinski and Whitham 1989, Ritchie and Olff 199 9). Harmful effects include loss of photosynthetic material, mortality, and reduction in growth (Huntly 1991). These effects may scale up to influence plant reproduction, as energy used to produce seeds or runners may be diverted to repair damaged plant parts (Ritchie and Olff 1999). However, herbivores have also at times been shown to have no effect (or minimal effect) on plant growth a nd reproduction because of the capacity of plants for compensatory growth (Belsky 1986, Hawkes and Sullivan 2001). However, compensation in growth and overcompensation in growth may be limited to nutrient-rich areas, where nutrients are readily available for growth and repa ir (Belsky 1986, Hawkes and Sullivan 2001). Herbivory can also have larg e-scale effects on plant commun ities such as changes in plant abundance (Bailey and Whitham 2002, He ssl and Graumlich 2002), plant species composition (Harrison et al. 2003, Davies et al . 2005), and rates of succession (Seabloom and Richards 2003, van Langevelde et al. 2003). For example, selective foraging on competitively dominant plant species can increase plant speci es richness and some cases reduce invasion by non-native species (DÂ’antonio 1993, Harrison et al. 2003). Because multiple disturbance events can operate concurrently, many studies have investigated the interactive e ffects of fire and herbivory on plant communities. Interactive effects included changes in successional rates (M ills 1983), species distribution (Archibald et al. 2005), and species composition (Harris on et al. 2003). For example, the interactive effect of fire and grazing has been proposed as a mechanism of tree-grass co-existen ce in African savannas (van Langevelde et al. 2003, Archibald et al. 2005). Following fire, grasses and trees are

PAGE 83

83 consumed by grazers and browsers respectively. This not only redu ces the intensity of fire for future events, but prevents landscapes from be ing completely dominated by one life form thus promoting tree and grass coexistence (van Langevelde et al. 2003). Numerous studies have examined the interact ive effects of fire and herbivory on plant communities, however, few studies have looked at their interactive effects in creating heterogeneity in plant damage. Alone, fire creates heterogeneity in plant damage through variation in intensity, severit y, frequency and rate of spread (Turner et al. 1994, Turner and Romme 1994) and its effect can be seen at multip le levels. At the individual plant level, differences in fire intensity can create vari ation in damage throughout the plant where tree canopies remain intact at the expens e of lower branches. At larger spatial scales, variation in fire intensity and rate of spread can lead to heterogeneity in habitat structure, leaving a mosaic of burned and unburned habitat in the landscape (Turner et al. 1997). Similarly, herbivores alone, can create heter ogeneity in plant damage which can also be seen at multiple levels. At the leaf level, va riation in feeding strate gies of herbivores and nutrient levels within leaves can create heter ogeneity in damage (Basset 1991, Coley and Barone 1996, Mopper et al. 2000, Van Zandt and Agra wal 2004a, Agrawal 2004). For example, monarch caterpillars feedi ng on milkweed plants ( Asclepias syriaca ) notch leaves at the petiole and then proceed to feed on leaf tips where latex content is lowe r. This feeding behavior and damage pattern differs from other insect herb ivores (e.g. beetles and aphids) that feed on milkweed leaves as well (Van Zandt and Agra wal 2004b). Similarly, vari ation in plant quality both within plants and between plants have been shown to influence damage patterns (Strong et al. 1984, Huntly 1991, Van Zandt and Agrawal 2004a).

PAGE 84

84 The coupled effects of fire and herbivory on the heterogeneity of plant damage can further increase levels of heterogeneity. For ex ample, pockets of habita t left unburned after fire may serve as refuge for insect herbivores which ma y lead to variation in plant damage such that plants near habitat edges experien ce more damage than plants in th e interior of burned patches. Knight and Holt (2005) examined the role of fire in creating spatial gradient s of insect herbivory. In their study, recently burned patches exhibite d variation in leaf da mage as a function of proximity to habitat edge. Slow colonization of flightless grasshoppers into the newly burned patches was responsible for spatia l variation in leaf damage pa tterns. Similarly, following the 1980 Mount St. Helens eruption, lupine located al ong edges of the burned habitat consistently experienced greater leaf damage by Lepidopteran herbivores, even after 10 years following the eruption (Fagan et al. 2004). This gradient of pl ant damage left by insect herbivores scaled up to influence reproduction where lupine populations along edges remained in low densities, thereby reducing the spread of lupine across the landscape. Although variation in plant damage resulting fr om fire may be tran sient (Knight and Holt 2005), at broader temporal and spatial scales , variation in damage may persist between individual plants, and among plants in stands of va rying ages (Fagan et al . 2004). As stands age, plant and habitat architecture changes, acco mpanied by variation in insect herbivore communities and leaf damage. Chapter 2 described how variation in stand age (time-since-fire), fire-return interval, and landscap e heterogeneity can directly or indirectly influence insect herbivore community structure. These effects can translate up to influence damage patterns found on plants. In this study I first describe the interactive effects of in sect herbivory and fire on leaf and plant damage, and how patterns di ffer between three co-existing oak species ( Q. chapmanii , Q. geminata , and Q. inopina ). Secondly, I describe how fire can influence leaf and

PAGE 85

85 plant damage by use of path analyses. Lastly, I describe spatial and temporal patterns of herbivory, as a function of stand age and time during the growing season. Since oak species differ greatly in their leaf-l evel characteristics (Figure 3-1), I predicted (1) that variation in leaf damage between the three oak species would be observed. Additionally, the three oak species differ in plant-le vel traits, such as height, shape, and number of leaves, which may be important for microhabita t selection by insects (Figures 3-2 to 3-4). Therefore plant-level traits may also contribute to variation in herbi vore damage. Since the shortest time-since-fire interval was 1 year since fire, I predicted (2) that fire effects would be largely indirect, as the impacted direct effects to fire, such as mortality and emigration, may dissipate within a single year (S wengel 1996, Knight and Holt 2005). I predicted fire to influence plant damage by increasing nutrient content in leaves (Radho-Toly et al. 2001) and opening up habitats , thereby facilitating herbivore movement and the spread of herbivore damage (Knight and Holt 2005). However, I predicted (3) a decrease in leaf damage as a function of stand age because nutr ient content in leaves typically decreases as a function of plant age where nutrients are reabsorb ed and transferred to other parts of the plant (Wright and Westoby 2003). Methods The following study was conducted in scr ubby flatwoods habita t at the Archbold Biological Station (ABS) located at the southern tip of the Lake Wales Ridge in south-central Florida (27º11’N, 81º21’W, Highla nds Co., FL). Scrubby flatwoods are characterized by low to moderate growing shrubs inte rspersed with slash pine, Pinus elliottii (Wright and Westoby 2003). Scrubby flatwoods habitat is home to a wide array of endemic plants and animals, such as the Florida scrub-jay ( Aphelocoma c. coerulescens ), gopher tortoise ( Gopherus polyphemus ), Florida mouse ( Podomys floridanus ), and pygmy fringetree ( Chionaznthus pgymaeus ). Scrubby

PAGE 86

86 flatwoods are a fire-dependent habitat where fire return in tervals are between 5-20 years (Abrahamson 1984, Myers 1990). Plot Establishment and Sampling Regime Fifteen 30 meter by 30 meter plots were establ ished in scrubby flat woods habitat along a chronosequence of time-since-fire (Figure 2-2 in Chapter 2). Plots were established near fire lanes to facilitate access however; they were placed 20m from edge to reduce edge effects on insect communities. Locations were chosen base d on size and proximity to other habitat types. For example sites needed to be large enough to accommodate a 30m by 30m plot and at least 20m from other habitat types as to minimize the e ffects of insect herbivores from other habitat types on plant damage patterns in the scrubby flatwoods. Five tim e-since-fire (TSF) intervals were used to span the natural range of fire fr equency events in the sc rubby flatwoods (1, 4, 6, 11, 19 years since last fire). Three 30m x 30m plot s were established within each TSF interval. Within each plot, 25 individuals of each oak species ( Quercus geminata, Q. inopina, and Q. chapmanii ) were marked for repeated sampling, resulti ng in a total of 75 i ndividual plants per plot. On each plant, 10 newly flushed leaves we re randomly selected and marked with indelible ink. Plots were established in May 2005, and plan t conditions such as plant height, diameter at base, number of branches, and number of leaves, were recorded. These plant conditions were collectively referred to as “plant architecture” . In addition to plant conditions, spatial coordinates for individual plants we re recorded for spatial analyses. Plants were sampled in July, September, a nd November, 2005 (herea fter referred to as “seasonal stage”), where the amount of leaf tissu e damage was recorded. Leaf tissue damage was the percent tissue loss due to herbivory and was estimated using 2mm by 2mm grid paper. Leaves on unmarked plants were traced ont o grid paper, and th e number of 2mm by 2mm squares missing from the estimated outline of th e intact leaf was counted. This number was

PAGE 87

87 divided by the total number of squares that made up the entire leaf area to give an estimation of leaf damage. Since tracing leaves can be destructive to the leaves and plants in general, practice runs were conducted on unmarked leaves until I fe lt comfortable with damage estimates (usually between 10-15 trials). Trials we re done prior to sampling new plots and new oak species. The type of damage was also recorded and categor ized as being chewing damage, skeletonized damage, necrosis/sclerosis, mine damage, leaf roll, and gall damage (Labandeira 2002). Tannin concentration was also c onducted, using different leaves than those used in leaf damage surveys. Tannin concentration was dete rmined using the radial diffusion method (see Chapter 2 for details). Leaf architecture was also determined by measuring leaf area, mass, thickness, toughness, width, and density (see Chapte r 2). In addition, ins ect communities were also sampled using a sweep net (15 inches in diameter). Tannin analyses, leaf architecture measurements, and insect sampling were conducted on 45 additional plants marked in the plots. Because I did not sample the same plants as t hose used in the leaf damage surveys, genetic differences between plants were not controlled. This method of using different plants was chosen because sweeping and removal of leaves for analyses were destructive methods, and revisiting the same individuals may have influenced future census results with repeated sampling of the same individual plants. Vegetation sampling was conducted using th e point intercept method (Wilson 1960). Four equally spaced 30 meter transects were esta blished within each plot, and plants touching a 0.5 cm diameter rod placed ever y 2 meters were recorded. Plant species and height of interception were recorded. These data were used to calculate vegeta tion abundance, vegetation height, vegetation density, plant species richness, and host plant abundance, in an effort to characterize “plot architecture”.

PAGE 88

88 Landscape level characteristics were also de scribed using 100m buffe rs centered on the plot. Landscape-level characteri stics included the num ber of habitat patc hes, the number of different habitat types, total ed ge, and habitat shape, and were measured using vegetation maps provided by Archbold Biologi cal Station. Fire return interval was also determined using fire history data at ABS that is available at http://www.archboldstation.org/abs/index.htm. For further description of each of these procedures , refer to the Methods section in Chapter 2. Data Analysis Data were analyzed using mixed effect mode ls with leaf damage (% total tissue removed per leaf) at each seasonal stag e as a repeated measure. Fixe d effects were time-since-fire intervals, and seasonal stage. Th e random effects were plants nested in plots and plots nested in time-since-fire. Similarly, the number of leaves damaged per plant and the number of plants damaged per plot were also analyzed using mi xed effect models. Analyses were performed using SPSS v11.5 (SPSS 2002). Mapping damage patterns Inverse-distance weighted interpolations (IDW) were performed in ArcView 3.3 (ESRI 1999), in an effort to visually describe spatial and temporal variation in damage on plants across plots. Inverse distance weighting is a simple interpolation method where weighted damage from neighboring plants determines the amount of damage on unmarked plants, surrounding focal plants (Zimmerman et al. 1999). The number of leaves consumed per plant and the number of plants consumed per plot were also described using IDW. Path analysis Path analysis was also used to determine the effects of fire on herbivore damage and leaf consumption. Two default models, similar to those used in Chapter 2, were used in the analyses. One model included insect herbivor e data from Chapter 2 (“Model 1” where leaf damage simply

PAGE 89

89 added to the end of the path sequence), and a second model, Model 2, where insect herbivores were excluded from the analyses. Two models were chosen due to the insect sampling method and time of sampling. Since sweep netting was hi ghly biased towards external feeding insects, and carried out during the day (the reby excluding internal feeders and nocturnal feeders from the datasets), a disparity between he rbivore community structure and leaf damage could occur, as leaf damage data included damage done by all insect herbivores (e.g. miners and gall makers). Plant and plot architecture variables were calcu lated using principal component analyses for plant traits and plot characteristics, respectively (T able 2-1 from Chapter 2). Path analyses were performed using all plant data, and separately for each oak species using AMOS v5 (Arbuckle 1999). Results Chewing was by far, the dominant form of damage, followed by skeletonized damage, and gall damage (Figure 3-5). However this may not imply that chewing insects were the dominant insect herbivores. Since chew damage may have been preceded by any other damage type (e.g. galls or leaf mines) in between the seasonal stages, these othe r types of damage may not have been fully recorded. Therefore, tota l leaf tissue damage (per cent leaf area removed) was used instead in the following analyses. Variation in Leaf Damage Species-level differences in l eaf damage were apparent. Q. chapmanii experienced the greatest overall damage, followed by Q. inopina , and then Q. geminata (Figure 3-6). For Q. inopina and Q. chapmanii , total leaf damage showed no si gnificant difference with time-sincefire, but significant differences occurred acro ss seasonal stage and ther e was a significant interaction between seasonal stage and time-since-fire (Table 3-1). By contrast, leaf damage on Q. geminata was significant for time-since-fire, seasona l stage, and their interaction. Leaf

PAGE 90

90 damage was relatively low earlier in the season, with the greatest change in leaf damage occurring later on in the season (Figure 3-7). A distinct hump-shaped pa ttern was observed later in the growing season as a function of time-since-fi re, with plants in the intermediate time-sincefire intervals, (TSF 4, 6, and 11) exhibiting the highe st amount of leaf damage. Variation in leaf damage among plants was also hi ghest in the intermediate plot s versus the recently burned and older plots (Figure 3-8). Differences in the amount of l eaf damage between the three species were significant, with the highest amount of damage on Q. chapmanii leaves, followed by Q .inopina , and Q. geminata . Variation in the Number of Leaves Damaged per Plant The number of leaves damaged per plant in creased throughout the growing season, with the greatest change occurring between July and Se ptember (Table 3-2). The average number of leaves damaged per plant exhibited a U-shaped pa ttern as a function of time since fire for all three seasonal stages (Figure 3-9). Species-level differences were seen with the highest number of leaves damaged on Q. chapmanii , followed by Q. inopina , and then Q. geminata . The Ushaped pattern indicates that although the average num ber of leaves damaged increased through time, there were more leaves damaged in recently burned and older plots (TSF 1 and 19), compared to intermediate time-since fire plots (TSF 4,6,11). Variation in the Number of Plants Damaged per Plot The total number of plants damaged per pl ot increased throughout the growing season, with the greatest number of plants being consum ed in the intermediate time-since fire plots (Figure 3-10). This indicates that although more plants became damaged through time on every plot, more were being damaged in the intermediate time-since fire plot, than in either the recently burned plots and older plots (F4,3018 = 220.48, p = 0.000).

PAGE 91

91 Combining all three types of damage data (e .g. leaf damage, number of leaves damaged per plant and number of plants damaged per plot) shows that at the leaf level, plants in the intermediate time-since-fire plot experienced mo re damage; at the plant level, plants in the younger and older plots experienced more damage (that is, had more leaves affected by herbivory), and at the plot leve l, intermediate time-since-fire plots experienced more damage, with a greater number of plants affected by herb ivory than in younger and older plots. Table 3-3 shows a summary of these results. Inverse Distance Weighting Maps Maps plotting results from inverse distance weighting illustrate how damage pattern changed through time, and how these changes differed between and within plots. Since all plant species exhibited the same qualitative patterns as a function of time-since-fire, they were analyzed together to increase th e accuracy of Inverse Distance Weighting (IDW) analyses. At the beginning of the growing season , all plants started out with sim ilar levels of leaf damage, but intermediate time-since-fire plots showed great er increases in leaf damage as the season progressed (Figure 3-11). Figure 3-11 also indicates that the spatio-temporal distribution of damage between plants vary, such that damage is more spread out across plots at intermediate ages, compared to young and old plots, where high or low leaf damage is concentrated in certain areas. The spatial patterning of herbivory t hus varies with time-s ince fire disturbance. Figure 3-12 illustrates how the number of leaves damaged per plant changed throughout the growing season, in recently burned plots, intermediate time-sincefire plots, and long, unburned plots, respectively. Fewer leaves were damaged per plant in intermediate time-sincefire plots, whereas more leaves were damaged in younger and older plots. To summarize, IDW maps show that damage was greater in the intermediate time-sincefire plots, but damage was more spread out w ithin the plot, versus localized damage in the

PAGE 92

92 recently burned and older plots. However at the plant level, damage was localized to only a few leaves in plants in intermediate aged plots vers us plants in recently bu rned and older stands. Plants in these plots experienced a greater spread in damage th roughout the entire plant. Table 3-4 summarizes these results. Factors Influencing Leaf Tissue Damage Two default path models were constructed and show in Figure 3-13 (Model 1, using insect herbivore data from Chapter 2) and Fi gure 3-14 (Model 2, where insect herbivore data were not used). A disparity between insect herbivore data and da mage patterns occurred, resulting in a poorly fit model (2=680.95, d.f = 12, p=0.001). As a result, the following analyses were conducted using the second model, where ins ect responses were omitted from the analyses. The second model showed a strong fit with the fully sa turated model (2=0.0068, d.f = 5, p=1.00). The path analysis results demonstrate that all predictor variables (except landscape heterogeneity) were significant in explaining 22% of the total variation in leaf damage (Table 35). However, seasonal stage, plot architecture and leaf morphology had larg er influences on leaf damage, indicated by larger path coefficients. The number of leaves damaged was affected by leaf morphology, plant architecture, and seasonal stage, explaining 13% of variation (Table 3-6). Total number of plants damaged per plot wa s strongly affected by seasonal stage, plot architecture, and heteroge neity, explaining 48% of va riation (Table 3-7). Discussion The results indicate the existence of great spatial variation in leaf damage among oak species, among stands of varying ages, a nd throughout the growing season. Although heterogeneity in leaf damage is in itself not surprising, the effects of fire on patterns of

PAGE 93

93 heterogeneity of leaf damage are noteworthy and have not been previously documented. Q. chapmanii showed the greatest overa ll leaf damage, followed by Q. inopina , and then Q. geminata (Figure 3-6). Species-level differences in damage may reflect the chemical and physical properties of leaves. Q. chapmanii had the most tannin but lack ed structural defenses. On the other hand, Q. geminata had the least amount of tannins but its leaves we re thicker and tougher (Figures 3-1 and 3-6). Perhaps in this syst em, structural defenses play a larger role in protection against herbivore damage than do ch emical defenses. When comparing insect herbivore abundance data from Chapter 2 to l eaf damage data, note that there were more herbivores found on Q. chapmanii , followed by Q. inopina , and then Q. geminata , suggesting that greater herbivore abundances contributed to the observed pattern of species-level differences in damage. Path analyses show that leaf ar chitecture indeed played a significant role in explaining the 22% varia tion in leaf damage. Temporal variation in leaf damage was also observed with an increase in cumulative damage as the season progressed. The greatest ch ange in damage occurred later in the growing season; however note that insect herbivore abun dance dropped during this period (Figure 2-3 of Chapter 2) Therefore, the type of insects oc curring late in the grow ing season may explain why higher damage patterns occurred even though fe wer insects were present. For example, orthopterans can impose greater damage effects compared to some of the other herbivores observed (e.g., Collembola and beetles). Orthoptera ns become more active and larger in the fall, as they develop into adults (J. Capinera, person al communication). Perhaps this increase in orthopteran activity and body size contributed to increased le af damage patterns found in November. However, when plotting orthopteran abundance as a function of seasonal stage, a decrease in abundance is observed. The disparity in insect herbivore and damage data may have

PAGE 94

94 instead been due to the chosen sampling method because sampling was conducted during the day and not at night. Since many ort hopteran species are nocturnal in their feeding, there may have actually been more herbivores with in the plots, than indicated by samples. Further studies using a variety of sampling regimes would be desirabl e to characterize the entire insect herbivore community. Spatial variation was also observed, in that plan ts in the intermediate time-since-fire plots exhibited the greatest am ount of leaf damage, compared to ot her plots. Moreover, the variation in leaf damage within plants and between plan ts was highest in intermediate time-since plots (Figure 3-8). This suggests that plants w ithin the intermediate plots showed greater heterogeneity in plant damage compared to plants in both younger and older plots. Path analyses showed that plot architectur e played a significant role in leaf damage variation, and that a po sitive relationship was observed. Si nce vegetation density contributed most to plot architecture, the pl ots with the greatest density also showed the greatest amount of leaf damage. Variation in plot architecture wa s shown in Figure 2-16 of Chapter 2, with denser plots occurring in intermediate time-since fire stands. Theoretically according to the Resource Concentration Hypothesis, plants in dense patches are able to su stain higher numbers of insect herbivores because these plants are easier to locate (Root 1973). Perhaps dense plots in this study attracted insect herbivores which may have translated into increased leaf damage. Although herbivore data collected in November showed only a slig ht increase in abundance in intermediate aged plots (Figure 3-15), pe rhaps sweep netting during the day may have underestimated the actual number of insect herbivores feeding on the plants. Inverse distance weighting interpolations s howed an increase in damage with growing season for all plot types, however, higher damage levels were observed in intermediate time-

PAGE 95

95 since fire plots. In addition, a greater extent in damage in intermediate time-since fire plots was observed in that more plants were being dama ged through time, compared to younger and older plots. Possible mechanisms for the increased ex tent in damage through time may be that insect herbivores were moving between plants more th an herbivores in the younger and older plots. Intermediate aged plots had higher plant density and this may have f acilitated small-scale herbivore movement between plants, especially fo r flightless and immature individuals. Another possible mechanism may be that there are new he rbivores entering the intermediate plots through time and feeding on different plants. Therefore the increase in the number of damaged plants was not necessarily due to the in ternal movement but rather due to an influx of insects from outside the patch, leading to a broader dispersion of feeding he rbivores among plants. Marking and tracking herbivore movements may help cl arify the mechanism underlying how damage is moving across the plots. Conversely, younger and older plots experien ced lower overall damage, and a lower extent in damage. Again, vegetation density may have played a role in the amount of damage experienced by plants in these plots. Less dense plots may have attracted fewer insect herbivores, leading to an overall lower abundanc e of insect herbivores. Similarly, open plots may not have facilitated movement between plants, thereby reducing the extent of damage within the research plots. Variation in plant quality be tweenand within-plants may help determine the extent of damage throughout research plots (Hannune n and Ekbon 2002, Nykanen and Koricheva 2004, Underwood 2004). In all time-since-fire plots, va riation in tannin and physical defenses of leaves were greater between-plant s than within-plants (Table 39). This would suggest higher between-plant herbivore m ovement compared to within-plant m ovement as herbivores search for

PAGE 96

96 better quality plants (Nykanen and Koricheva 2004, Underwood 2004). Since intermediate timesince-fire plots had denser vegetation, this may have facilitated herbivor e movement and led to a greater extent in damage thr ough time, compared to recently bur ned and long-unburned plots. However, once again, marking herbivores and tr acking their movement will determine which of these mechanisms are responsible for these patterns of herbivory. Greater leaf damage in dens er plots is not surprising, a nd many earlier studies have observed similar effects (e.g. Karban 1997, Carson et al. 2004). Likewise, increased extent of leaf damage in denser plots is not surprising, in th at plants that are closer together may facilitate herbivore movement by serving as stepping ston es, especially for flightless and immature herbivores (though to my knowledge this pattern has not been directly documented). However the indirect effect of fire on l eaf damage patterns was surprising, in that indirect effects were mediated through plot architecture versus leaf traits. Similarly, the increase in leaf damage patterns later in the growing season was not predicted. This study de monstrates that the effect of fire on insect herbivore damage is variable and unpredictable, and may depend on the habitat where fire occurs. Previous studies examining damage effects following fire have shown greater damage in recently burned plots due to increased nutrient content in leaves (Radho-Toly et al. 2001), and proximity to habitat edge (Knight a nd Holt 2005). However, this study showed that damage was greatest in intermediate burned plots, where density in plants was greatest. Previous studies examining herbivore damage were done in grasslands (Swengel 2001, LemonnierDarcemont 2003), where habitat stru cture is relatively low. Perh aps in habitats with strong spatial structure, the effects of fire on herb ivory are largely mediated through changes in the three dimensional structure of the habitat as opposed to sh ifts in leaf quality. Further studies in

PAGE 97

97 fire dependent habitats with st rong spatial structure are required to make general conclusions about the interactive effects of fire an d herbivory on plant communities.

PAGE 98

98 Table 3-1. Differences in leaf damage (% leaf tissue removed) as a function of seasonal stage and time-since-fire (TSF) for Q. chapmanii (A ), Q. geminata (B), and Q. inopina (C). A. Source Numerator df Denominator df F p Intercept 1 9.62 303.61 0 SEASONAL STAGE 2 894.22 139.22 0 TSF 4 9.46 1.16 0.39 SEASONAL STAGE* TSF 8 894.14 4.92 0 B. Source Numerator df Denominator df F p Intercept 1 9.92 184.31 0 SEASONAL STAGE 2 931.89 139.78 0 TSF 4 9.79 8.06 0 SEASONAL STAGE* TSF 8 931.76 20.25 0 C. Source Numerator df Denominator df F p Intercept 1 10.04 208.18 0 SEASONAL STAGE 2 1013.38 153.4 0 TSF 4 10.03 2.24 0.14 SEASONAL STAGE* TSF 8 1013.35 14.53 0

PAGE 99

99 Table 3-2. Differences in the average number of leaves consumed per plant as a function of seasonal stage and time-since-fire (TSF) for Q. chapmanii (A) , Q. geminata (B), and Q. inopina (C). A. Source Numerator df Denominator df F p Intercept 1 10.11 312.45 0.00 SEASONAL STAGE 2 605.46 213.57 0.00 TSF 4 9.65 1.35 0.32 SEASONAL STAGE* TSF 8 605.15 7.93 0.00 B. Source Numerator df Denominator df F p Intercept 1 10.00 181.18 0.00 SEASONAL STAGE 2 652.63 159.79 0.00 TSF 4 9.85 7.93 0.00 SEASONAL STAGE* TSF 8 648.28 23.32 0.00 C. Source Numerator df Denominator df F p Intercept 1 10.03 203.02 0.00 SEASONAL STAGE 2 690.17 204.71 0.00 TSF 4 10.02 2.17 0.15 SEASONAL STAGE* TSF 8 689.62 19.42 0.00

PAGE 100

100 Table 3-3. Multi-scale patterns of plant damage as a function of stand age (time-since-last fire). Young stands Intermediate aged stands Older stands (1 year since fire) (6 years since fire) (19 years since last fire) Leaf-level damage (percent leaf area damaged) low high low Plant-level damage (number of leaves damaged on plant) high low high Plot-level damage (number of plants damaged per plot) low high low Table 3-4. Variation in the magn itude of leaf damage (percent leaf area damaged) and spatial extent of damage as a function of stand age (time-since-fire). Young stands Intermediate aged stands Older stands (1 year since fire) (6 years since fire) (19 years since last fire) Plot-level damage (between plant variation in damage) low leaf damage and localized in space high leaf damage and spread out in space low leaf damage and localized in space Plant-level damage (within plant variation in damage) low leaf damage and spread out in space high leaf damage and localized in space low leaf damage and spread out in space

PAGE 101

101 Table 3-5. Path analysis results for the direct and indirect effects of fire on leaf damage (% leaf area removed). Most predictor variables are significant; however, season, leaf architecture, and plot arch itecture explained a higher am ount of variation in leaf damage than other predictor variables. Direct Indirect Total Sig. Season 0.422 0.015 0.437 p = 0.000 Landscape heterogeneity -0.091 0.017 -0.073 p = 0.000 Fire return interval 0.073 -0.096 -0.023 p = 0.002 Time-since-fire -0.096 0.039 -0.057 p = 0.000 Leaf architecture 0.123 0.001 0.123 p = 0.000 Plot architecture 0.133 0.012 0.145 p = 0.000 Plant architecture -0.011 -0.012 -0.022 p = 0.559 Table 3-6. Path analysis results for the direct and indirect effect s of fire on the number of leaves damaged per plant. Most predictor variab les are significant; however, time-since-fire, leaf architecture, and plant architecture explained a higher amount of variation in damage than other predictor variables. Direct Indirect Total Sig. Season 0.082 0.03 0.113 p = 0.000 Landscape heterogeneity 0.04 8 0.037 0.084 p = 0.010 Fire return interval -0.051 -0.002 -0.053 p = 0.033 Time-since-fire 0.206 -0.034 0.172 p = 0.000 Leaf architecture 0.248 0.007 0.254 p = 0.000 Plot architecture 0.005 0.024 0.03 p = 0.823 Plant architecture -0.177 -0.002 -0.18 p = 0.000 Table 3-7. Path analysis results for the direct and indirect effect s of fire on the plants damaged per plot. Most predictor variables are significant; however, season, landscape heterogeneity, and plot architecture explai ned a higher amount of variation in leaf damage than other predictor variables. Direct Indirect Total Sig. Season 0.583 0.005 0.588 p = 0.000 Landscape heterogeneity -0.127 0.031 -0.097 p = 0.000 Fire return interval 0.044 -0.223 -0.179 p = 0.017 Time-since-fire 0.08 0.064 0.144 p = 0.000 Leaf architecture 0.038 0.002 0.041 p = 0.004 Plot architecture 0.324 0.004 0.328 p = 0.000 Plant architecture -0.04 -0.026 -0.067 p = 0.006

PAGE 102

102 Table 3-9. Within-plant variation and between-pla nt variation in leaf tr aits in recently burned plots (TSF 1); intermediate time-since-fire plots (TSF 6); and older plots (TSF 19). Recently burned plots Intermediate aged plots Long unburned plots Leaf trait Within plant variation Between plant variation Within plant variation Between plant variation Within plant variation Between plant variation density (cm3/g) 0.0007 0.0002 0.0000 0.0000 0.0000 0.0000 area (cm2) 2.7745 4.4086 1.9408 3.5245 2.7108 4.3296 length (cm) 0.6710 0.4706 0.5620 0.3062 0.5738 0.3129 mass (g) 0.0010 0.0009 0.0006 0.0008 0.0008 0.0009 max. width (cm) 0.6562 0.6360 0.5591 0.4294 0.6838 0.5923 tannin (mg/100mg of tissue) 0.0604 0.1141 0.0573 0.1181 0.0610 0.1278 thick (mm) 0.0018 0.0196 0.0017 0.0220 0.0030 0.0188 tough (kg) 0.0425 0.1082 0.0423 0.0754 0.0380 0.0852 width (cm) 0.3188 0.3343 0.3455 0.2342 0.3738 0.3066

PAGE 103

103 Figure 3-1. Variation in leaf chemistry and physical traits among oak species. Leaf density contributed most to the variati on in leaf physical traits.

PAGE 104

104 0 10 20 30 40 50 60 70 80 90 100 Q. chapmaniiQ. geminataQ. inopinaAverage plant height (cm) Figure 3-2 Average plant height (cm) of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub. Error bars are +/1 SE.

PAGE 105

105 9.4 9.6 9.8 10 10.2 10.4 10.6 10.8 11 11.2 11.4 Q. chapmaniiQ. geminataQ. inopinaAverage number of branches Figure 3-3. Average number of branches from primary stem of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub. Error bars are +/1 SE. 0 50 100 150 200 250 300 350 400 Q. chapmaniiQ. geminataQ. inopinaAverage number of leaves Figure 3-4. Average number of leaves of three dominant oak species ( Q. chapmanii, Q. geminata , and Q. inopina ) in Florida scrub. E rror bars are +/1SE.

PAGE 106

106 0 2 4 6 8 10 12 14 16 18 20ChewGallMineNecrosisLeaf rollSkeletonizedDamage type July September November 0 2 4 6 8 10 12 14 16 18 20ChewGallMineNecrosisLeaf rollSkeletonizedDamage typeAverage leaf area removed (%) per plant July September November 0 2 4 6 8 10 12 14 16 18 20ChewGallMineNecrosisLeaf rollSkeletonizedDamage typeAverage leaf area removed (%) per plant July September November Figure 3-5. Variation in the average amount of leaf damage (% leaf area removed per leaf) and type of damage on Q. chapmanii (A); Q. inopina (B); and Q. geminata (C). A B C

PAGE 107

107 0 2 4 6 8 10 12 JulySeptemberNovember Phenological stageAverage leaf damage (%) Q. chapmanii Q. geminata Q. inopina Figure 3-6. Average leaf damage (percent l eaf area removed) across seasonal stage for Q. chapmanii , Q. geminata , and Q. inopina. Error bars are +/1 SE.

PAGE 108

108 0 2 4 6 8 10 12 14 16 1461119 Time-since-fireAverage leaf damage (%) July September November 0 2 4 6 8 10 12 14 16 1461119Time-since-fireAverage leaf damage (%) July Sepember November 0 2 4 6 8 10 12 14 16 1461119Time-since-fireAverage leaf damage (%) July Sepember November Figure 3-7. Average leaf tissue dama ge (percent leaf area removed) for Q. chapmanii (A), Q. geminata (B), and Q. inopina (C) as functions of time-since-fire. Error bars are +/1 SE. A B C

PAGE 109

109 0 20 40 60 80 100 120 140 160 180 200 1461119 Time-since-fireVariance in leaf damage among plants July September November 0 20 40 60 80 100 120 140 160 180 200 1461119 Time-since-fireVariance of leaf damage within plants July September November Figure 3-8. Variation in leaf tis sue damage for all oak species combined. A. Between plant variation in leaf tissue dama ge (% leaf tissue removed per leaf). B. Within plant variation in leaf tissue damage . Error bars are +/1 SE. A B

PAGE 110

110 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 1461119Time-since-fireAverage number of leaves consumed per plant July September November Figure 3-9. Average number of leaves consumed per plant across time-since-fire for all three oak species Q. chapmanii , Q. geminata , and Q. inopina combined. Error bars are +/1 SE.

PAGE 111

111 0 20 40 60 80 100 120 1461119 Time-since-fireNumber of plants consumed per plot (%) July September November Figure 3-10. Percentage of marked plants aff ected by herbivore damage as a function of timesince-fire and seasonal stage. Error bars are +/1 SE.

PAGE 112

112 Figure 3-11. Leaf damage patterns as a function of seasonal stage, for (A) recently burned plots, (time-since-fire 1), (B) intermediate time-si nce-fire plots, time-since-fire 6, and (C) long unburned plots, (tim e-since-fire 19).

PAGE 113

113 Figure 3-12. Number of leaves consumed per plant as a function of seasonal stage, for (A) recently burned plots, (time-since-fire 1), (B ) intermediate time-since-fire plots, timesince-fire 6, and (C) long unburned pl ots, (time-since-fire 19).

PAGE 114

114 Figure 3-13. Proposed default model (Model 1) desc ribing the effects of fire, plant, and insect variables on leaf tissue damage. Pa th analyses indicate a poor fit ( 2=680.95, d.f = 12, p=0.001).

PAGE 115

115 Figure 3-14. Proposed default model (Model 2) de scribing the effects of fire and plant level variables on plant damage. Path analyses indicate a good fit ( 2=0.0068, d.f = 5, p=1.00). Bold, solid arrows depict significant positive relationships whereas bold, dashed arrows depict signifi cant negative relationships. Th e relative magnitude of the relationships are indicated by the width of arrows.

PAGE 116

116 0 1 2 3 4 5 6 7 1461119 Time-since-fireHerbivore abundance July September November Figure 3-15. Spatio-temporal variation in insect herbivore abundances per plant. Early and midseason abundances are high a nd not significantly different from one another, however decreases in abundances occur later in the growing season.

PAGE 117

117 CHAPTER 4 CONCLUSIONS The overall objectives of this study were to de scribe how fire influenced insect herbivore community organization and to describe the subs equent damage they imposed on their host plant species. I was also interested in characteri zing spatial and temporal variation in herbivore community structure and plant damage. The effect s of fire were both dir ect and indirect, where the direct effect was time between disturbance ev ents, and indirect effects were largely mediated through plot architecture. Res earch plots that experienced fre quent fire events may have influenced herbivore community organization by preventing colonization of insect herbivores, and plots that had high plot architecture may have influen ced community organization by attracting herbivores and facili tating movement between and w ithin the research plots. Landscape structure was also important in that different habitat types surrounding the focal habitat served as source habitats for insect herbivores. Therefore local species richness increased with increases in the number of surrounding habitat types. Scale was an important theme is in this st udy when describing factors that influenced herbivore communities and plant damage. Variati on in leaf-level, plant-level, plot-level, and surrounding landscape-level traits were characterized to determine how they influenced herbivore community structure. All levels were shown to have direct or indirect effects on herbivore assemblages found on the three oak species. Scale was also important when describing vari ation in damage patterns. The amount of damage per leaf (leaf-level), th e number of leaves damaged pe r plant (plant-level), and the number of plants damaged per plot (plot-level) were characterized. Tem poral variation at all three levels were observed in that damage incr eased as a function of growing season. Spatial

PAGE 118

118 variation was also observed such that plants w ithin dense plots had highe r leaf-level and plotlevel damage. There are intriguing differences in how fire influenced community organization among herbivores found on the three oa k species. Herbivores found on Q. inopina were largely influenced by fire-return in terval, whereas herbivores on Q. chapmanii and Q. geminata were indirectly affected by fire through changes in loca l and large scale habitat structure, respectively. Because these oaks species are phylogenetical ly distantly related, perhaps species-level differences in leaf-level and plan t-level traits may have attribut ed to differences in community structure and organization. Oak species are often overlooke d as useful study organisms in conservation studies of the Florida scrub. They are overlooked in preference for endemic plant and animal species such as the gopher tortoise, Florida scrub jay, and Garre ttÂ’s Ziziphus. The three oak species provide ideal settings to study fundamental ecological qu estions, such as determ inants of community organization, and may also be used to answer ap plied questions such as factors influencing plant damage and spatial and temporal spread of he rbivore damage on host plants. Oak species comprise up to 85% of the Florida scrub habita t (Abrahamson 1984) and are important resources for the Florida scrub jay and the gopher tortoise . Therefore a holistic, ecosystem approach focused on these foundation species, rather th an a single-species (or even multi-species) approach is needed in order fully understand and protect the remaining scrub habitat.

PAGE 119

119 LIST OF REFERENCES Abrahamson, W. G. 1984. Post-fir e recovery of Florida Lake Wa les Ridge vegetation. American Journal of Botany 71 :9-21. Alfaro, R. I., S. Taylor, G. Brown, and E. Wegwitz. 1999. Tree mortality caused by the western hemlock looper in landscapes of central British Columbia. Forest Ecology and Management 124 :285-291. Anderson, L., C. E. Carlson, and R. H. Waki moto. 1987. Forest fire frequency and western spruce budworm outbreaks in western Montan a. Forest Ecology and Management 22 :251-260. Archibald, S., W. J. Bond, W. D. Stock, and D. H. K. Fairbanks. 2005. Shaping the landscape: Fire-grazer interactions in an Afri can savanna. Ecological Applications 15 :96-109. Bailey, J. K., and T. G. Whitham. 2002. Interact ions among fire, aspen, and elk affect insect diversity: Reversal of a community response. Ecology 83 :1701-1712. Basset, Y. 1991. Leaf production of an overstory rain-for est tree and its effects on the temporal distribution of associated in sect herbivores. Oecologia 88 :211-219. Basset, Y., E. Charles, D. S. Hammond, and V. K. Brown. 2001. Short-term effects of canopy openness on insect herbivores in a rain forest in Guyana . Journal of Applied Ecology 38 :1045-1058. Bell, D. T., J. A. Plummer, and S. K. Tayl or. 1993. Seed-germination ecology in Southwestern Western-Australia. Botanical Review 59 :24-73. Belsky, A. J. 1986. Does herbivory benefit plants a review of the evidence. American Naturalist 127 :870-892. Belsky, J. A. 1992. Effects of grazing, competiti on, disturbance and fire on species composition and diversity in grassland communities. Journal of Vegetation Science 3 :187-200. Bishop, J. G. 2002. Early primary succession on Mount St. Helens: Impact of insect herbivores on colonizing lupines. Ecology 83 :191-202. Boege, K. 2005. Herbivore attack in Casearia niti da influenced by plant ontogenetic variation in foliage quality and plant architecture. Oecologia 143 :117-125. Bond, W. J., and J. E. Keeley. 2005. Fire as a gl obal “herbivore”: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution 20 :387-394. Bowman, R., and G. E. Woolfenden. 2002. Nest site selection by Florida Scrub-Jays in natural and human-modified habitats. Wilson Bulletin 114 :128-135.

PAGE 120

120 Carson, W. P., J. P. Cronin, and Z. T. Long. 2004. A General Rule for Predicting When Insects Will Have Strong Top-Down Effects on Pl ant Communities: On the Relationship Between Insect Outbreaks and Host Concentration. Pages 193-211 in Ecological Studies. Springer-Verlag, Berlin. Carson, W. P., and R. B. Root. 2000. Herbivory and plant species coex istence: Community regulation by an outbreaking phytopha gous insect. Ecological Monographs 70 :73-99. Cavender-Bares, J., D. D. Ackerly, D. A. Baum, and F. A. Bazzaz. 2004. Phylogenetic overdispersion in Floridian oak communities. American Naturalist 163 :823-843. Clark, J. S. 1991. Disturbance and tree life-hist ory on the shifting mosaic landscape. Ecology 72 :1102-1118. Coley, P. D., and J. A. Barone. 1996. Herbivory a nd plant defenses in tropical forests. Annual Review of Ecology and Systematics 27 :305-335. Collett, N. 2003. Short and long-term effects of pres cribed fires in autumn and spring on surfaceactive arthropods in dry scle rophyll eucalypt forests of Vi ctoria. Forest Ecology and Management 182 :117-138. Connell, J. H. 1979. Intermediate-disturbance hypothesis. Science 204 :1344-1345. Cook, W. M., and R. D. Holt. 2006. Fire freque ncy and mosaic burning effects on a tallgrass prairie ground beetle assemblage . Biodiversity and Conservation 15 :2301-2323. Corbet, S. A. 1995. Insects, plants and succession advantages of long-term set-aside. Agriculture Ecosystems & Environment 53 :201-217. Cornell, H. V., and J. H. Lawton. 1992. Species in teractions, local and regional processes, and limits to the richness of ecological communitie s a theoretical perspective. Journal of Animal Ecology 61 :1-12. Cronin, J. T., and W. G. Abrahamson. 2001. Gold enrod stem galler preference and performance: effects of multiple herbivores and plant genotypes. Oecologia 127 :87-96. DÂ’antonio, C. M. 1993. Mechanisms controlling invasion of coastal plant-communities by the alien succulent carpobrotus edulis . Ecology 74 :83-95. Davies, K. F., P. Chesson, S. Harrison, B. D. Inouye, B. A. Melbourne, and K. J. Rice. 2005. Spatial heterogeneity explains the scale dependence of the native-exotic diversity relationship. Ecology 86 :1602-1610. Dayton, P. K. 1971. Competition, disturbance and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecological Monographs 41 :351-389.

PAGE 121

121 Di Giulio, M., P. J. Edwards, and E. Meister. 2001. Enhancing insect di versity in agricultural grasslands: the roles of management and la ndscape structure. Journal of Applied Ecology 38 :310-319. Diamond, J. M., and T. J. Case. 1986. Commun ity Ecology. Harper and Row, New York. Dobson, A. P., J. P. Rodrigues, W. M. R oberts, and D. S. Wilcove. 1997. Geographic distribution of endangered species in the United States. Science 275 :550:553. ESRI. 1999. ArcView version 3.3. Environmental Syst ems Research Institute (ESRI), Redlands, California, USA. Fagan, W. F., J. G. Bishop, and J. D. Schade. 2004. Spatially structured herbivory and primary succession at Mount St Helens: field surveys and experimental growth studies suggest a role for nutrients. Ecological Entomology 29 :398-409. Fay, P. A. 2003. Insect diversity in two burned and grazed grasslands. Environmental Entomology 32 :1099-1104. Fensham, R. J. 1994. Phytophagous insect woody spr out interactions in tr opical eucalypt forest. Australian Journal of Ecology 19 :189-196. Fleck D. C. and G. E. Woolfenden. 1997. Can acorn tannin predict scr ub-jay caching behavior? Journal of Chemical Ecologsy 23 :793-806. Fonseca, C. R., T. Fleck, and G. W. Fernande s. 2006. Processes driving ontogenetic succession of galls in a canopy. Biotropica 38 :514-521. Fuhlendorf, S. D., and D. M. Engle. 2004. Applicati on of the fire-grazing interaction to restore a shifting mosaic on tallgrass prairi e. Journal of Applied Ecology 41 :604-614. Hagerman, A. E. 1987. Radial diffu sion method for determining tannin in plant-extracts. Journal of Chemical Ecology 13 :437-449. Hamback, P. A., J. Agren, and L. Ericson. 2000. Associational resistance: Insect damage to purple loosestrife reduced in th ickets of sweet gale. Ecology 81 :1784-1794. Hannunen S. and B. Ekbon. 2002. Within species va riation in host plant quality and movement behavior of Lygus rugulipennis numphs. Entomologia Experimentalis et Applicata 104 :95-101. Harrison, S., B. D. Inouye, and H. D. Safford. 200 3. Ecological heterogeneity in the effects of grazing and fire on grassland diversity. Conservation Biology 17 :837-845. Hawkes, C. V., and E. S. Menges. 1996. The relationship between open space and fire for species in a xeric Florida shrubland. Bulletin of the Torrey Botanical Club 123 :81-92.

PAGE 122

122 Hawkes, C. V., and J. J. Sullivan. 2001. The impact of herbivory on plants in different resource conditions: A meta-analysis. Ecology 82 :2045-2058. Helms, S. E., and M. D. Hunter. 2005. Variation in plant quality and the population dynamics of herbivores: there is nothing av erage about aphids. Oecologia 145 :197-204. Hessl, A. E., and L. J. Graumlich. 2002. Interac tive effects of human activities, herbivory and fire on quaking aspen ( Populus tremuloides ) age structures in western Wyoming. Journal of Biogeography 29 :889-902. Huntly, N. 1991. Herbivores and the dynamics of communities and ecosystems. Annual Review of Ecology and Systematics 22 :477-503. Karban, R. 1997. Neighbourhood affects a plant's ri sk of herbivory and subsequent success. Ecological Entomology 22 :433-439. Keeley, J. E. 1987. Role of fire in seed-germina tion of woody taxa in California chaparral. Ecology 68 :434-443. Knight, T. M., and R. D. Holt. 2005. Fire generate s spatial gradients in herbivory: An example from a Florida sandhill ecosystem. Ecology 86 :587-593. Knowlton, N. 1992. Thresholds and multiple stable states in coral reef community dynamics. American Zoologist 32 :673-682. Krebs, C. J. 2001. Ecology: The Experimental Analysis of Distri bution and Abundance. Benjamin Cummings, San Francisco. Labandeira, C. C. 2002. The histor y of associations between pl ants and animals. Pages 26-74 in C. M. Herrera and O. Pellmyr, editors. Plant-Animal Interactions: An Evolutionary Approach. Blackwell Publishing, Malden. Lemonnier-Darcemont, M. 2003. Effects of controll ed fire on grazed grassland Orthoptera communities in medium mountains (Alpes-M aritimes, France). Revue D Ecologie-La Terre Et La Vie 58 :283-292. Letourneau, D. K. 1995. Associational susceptibi lity: effects of cropping pa ttern and fertilizer on malawian bean fly levels . Ecological Applications 5 :823-829. Lokesh, and R. Singh. 2005. Influence of leaf ve in morphology in Okra genotypes (Malavaceae) on the oviposition of the leafhopper species Amrasca biguttula (Hemi : Cicadellidae). Entomologia Generalis 28 :103-114. MacArthur, R. H., and E. O. Wilson. 1967. Th e Theory of Island Biogeography. Princeton University Press, Princeton. Marques, E. S. D., P. W. Price, and N. S. Cobb. 2000. Resource abundanc e and insect herbivore diversity on woody fabaceous desert plants. Environmental Entomology 29 :696-703.

PAGE 123

123 Maschinski, J., and T. G. Whitham. 1989. The con tinuum of plant-response s to herbivory: The influence of plant-association, nutrient av ailability, and timing. American Naturalist 134 :1-19. Menges, E. S., and C. V. Hawkes. 1998. Interactive effects of fire and microhabitat on plants of Florida scrub. Ecologi cal Applications 8 :935-946. Mills, A. J., and M. V. Fey. 2005. Interactive re sponse of herbivores, so ils and vegetation to annual burning in a South African savanna. Austral Ecology 30 :435-444. Mills, J. N. 1983. Herbivory and seedling esta blishment in post-fire southern California chaparral. Oecologia 60 :267-270. Mitchell, R. J. 2001. Path Analysis: Pollination. Pages 217-234 in S. M. Scheiner and J. Gurevitch, editors. Design and Analysis of Ec ological Experiments. Oxford University Press, Oxford. Mopper, S., P. Stiling, K. Landau, D. Simberlo ff, and P. Van Zandt. 2000. Spatiotemporal variation in leafminer populat ion structure and adaptati on to individual oak trees. Ecology 81 :1577-1587. Myers, R. L. 1990. Scrub and High Pine. Pages 150-193 in R. L. Myers and J. J. Ewel, editors. Ecosystems of Florida. Universi ty of Central Florida, Orlando. Nykanen, H. and J. Koricheva. 2004. Damage-indu ced changes in woody plants and the effects on insect herbivore performance: A meta-analysis. Oikos 104 :246-268. Paine, R. T. 1974. Intertidal community structure: Experimental studies on relationship between a dominant competitor and its principal predator. Oecologia 15 :93-120. Panzer, R., and M. Schwartz. 2000. Effects of management burning on prairie insect species richness within a system of small, highly fragmented reserves. Bi ological Conservation 96 :363-369. Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an integration of landscape and food web ecology: The dynamics of spatially s ubsidized food webs. Annual Review of Ecology and Systematics 28 :289-316. Powell, J. A. 1995. Recovery of Lepidoptera (Moths and Butterflies) Following a Wildfire at Inverness Ridge in Central Coastal California. National Park Service, U.S. Department of the Interior, Point Reyes National Seashore. Price, P. W. 1991. The plant vigor hy pothesis and herbivore attack. Oikos 62 :244-251. Price. P. W. 1997. Insect Ecology. Wiley, New York. Price, P. W., W. G. Abrahamson, M. D. Hunt er, and G. Melika. 2004. Using gall wasps on oaks to test broad ecological concepts. Conservation Biology 18 :1405-1416.

PAGE 124

124 Radho-Toly, S., J. D. Majer, and C. Yates. 2001. Impact of fire on leaf nutrients, arthropod fauna and herbivory of native and e xotic eucalypts in Kings Par k, Perth, Western Australia. Austral Ecology 26 :500-506. Rand, T. A. 1999. Effects of environmenta l context on the susceptibility of Atriplex patula to attack by herbivorous beetles. Oecologia 121 :39-46. Resh, V. H., A. V. Brown, A. P. Covich, M. E. Gurt z, H. W. Li, G. W. Minshall, S. R. Reice, A. L. Sheldon, J. B. Wallace, and R. C. Wissmar. 1988. The role of disturbance in stream ecology. Journal of the North Am erican Benthological Society 7 :433-455. Riitters, K. H., R. V. Oneill, C. T. Hunsaker, J. D. Wickham, D. H. Yankee, S. P. Timmins, K. B. Jones, and B. L. Jackson. 1995. A factor-a nalysis of landscape pattern and structure metrics. Landscape Ecology 10 :23-39. Ritchie, M. E., and H. Olff. 1999. Herbivore diversity and plant dynamics: compensatory and additive effects. Pages 175-204 in H. Olff, V. K. Brown, and R. H. Drent, editors. Herbivores: Between Plants and Pred ators. Blackwell Science, Oxford. Root, R. B. 1973. Organization of a plant-arthropod association in simple and diverse habitats: Fauna of collards ( Brassica oleracea ). Ecological Monographs 43 :95-120. Roques, K. G., T. G. O'Connor, and A. R. Watk inson. 2001. Dynamics of shrub encroachment in an African savanna: relative influences of fire, herbivory, rainfall and density dependence. Journal of Applied Ecology 38 :268-280. Sanson, G., J. Read, N. Aranwela, F. Clissol d, and P. Peeters. 2001. Measurement of leaf biomechanical properties in studies of herbivory: Opportuni ties, problems and procedures. Austral Ecology 26 :535-546. Santoro, A. E., M. J. Lombardero, M. P. Ayres, and J. J. Ruel. 2001. Interactions between fire and bark beetles in an old growth pine forest. Forest Ecology and Management 144 :245254. Schwartz, M. W. 1994. Conflicting goals for conservi ng biodiversity: Issues of scale and value. Natural Areas Journal 14 :213-216. Seabloom, E. W., and S. A. Richards. 2003. Multiple stable equilibria in grasslands mediated by herbivore population dynamics a nd foraging behavior. Ecology 84 :2891-2904. Sharma, A., and R. Singh. 2002. Oviposition preferen ce of cotton leafhopper in relation to leafvein morphology. Journal of Applied En tomology-Zeitschrift Fur Angewandte Entomologie 126 :538-544. Shipley, B. 2000. Cause and Correlation in Biology: A User's Guide to Path Analysis, Structural Equations and Causal Inference. Camb ridge University Press, Cambridge.

PAGE 125

125 Siemann, E., J. Haarstad, and D. Tilman. 1997. S hort-term and long-term effects of burning on oak savanna arthropods. Am erican Midland Naturalist 137 :349-361. Siemann, E., J. Haarstad, and D. Tilman. 1999. D ynamics of plant and arthropod diversity during old field succession. Ecography 22 :406-414. Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15 :353-391. Southwood, T. R. E., V. K. Brown, and P. M. Reader. 1979. Relationships of plant and insect diversities in succession. Biological Journal of the Linnean Society 12 :327-348. SPSS. 2002. SPSS for Windows, release 11.5.0. SPSS Inc. Stein, S. J., P. W. Price, W. G. Abrahamson, and C. F. Sacchi. 1992. The Effect of Fire on Stimulating Willow Regrowth and Subseque nt Attack by Grasshoppers and Elk. Oikos 65 :190-196. Stiling, P., D. C. Moon, M. D. Hunter, J. Colson , A. M. Rossi, G. J. Hymus, and B. G. Drake. 2003. Elevated CO2 lowers relative and absolute herbivore density ac ross all species of a scrub-oak forest. Oecologia 134 :82-87. Strong, D. R., J. H. Lawton, and T. R. E. Southwood. 1984. Insects on Plants: Community Patterns and Mechanisms. Harvard University Press, Cambridge. Swengel, A. B. 1996. Effects of fire and hay management on abundance of prairie butterflies. Biological Conservation 76 :73-85. Swengel, A. B. 2001. A literature review of in sect responses to fire, compared to other conservation managements of open hab itat. Biodiversity and Conservation 10 :1141-1169. Tahvanainen, J. O., and R. B. Root. 1972. Infl uence of vegetational diversity on population ecology of a specialized herbivore, Phyllotreta cruciferae (Coleoptera-Chrysomelidae). Oecologia 10 :321-&. Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75 :2-16. Turner, M. G., W. W. Hargrove, R. H. Gardne r, and W. H. Romme. 1994. Effects of fire on landscape heterogeneity in Yellowstone-Natio nal-Park, Wyoming. Journal of Vegetation Science 5 :731-742. Turner, M. G., and W. H. Romme. 1994. Landscape dynamics in crown fire ecosystems. Landscape Ecology 9 :59-77. Turner, M. G., W. H. Romme, R. H. Gardner, and W. W. Hargrove . 1997. Effects of fire size and pattern on early succession in Yellowst one National Park. Ecological Monographs 67 :411-433.

PAGE 126

126 Underwood, N. 2004. Variance and skew of the distri bution of plant quality influence herbivore population dynamics. Ecology 85 :686-693. USCB (United States Census Bureau). 2004. Cumula tive Estimates of Population Change for the United States-Table NST-EST2004-02. U.S. Census Bureau-Population Division, Washington. USFWS (United States Fish and Wildlife Serv ice). 1999. Endangered and Threatened Wildlife and Plants (50 CFR 17.11 and 1712). U. S. Fish and Wildlife Service. van Langevelde, F., C. van de Vijver, L. Kumar, J. van de Koppel, N. de Ridder, J. van Andel, A. K. Skidmore, J. W. Hearne, L. Stroosn ijder, W. J. Bond, H. H. T. Prins, and M. Rietkerk. 2003. Effects of fire and herbivor y on the stability of savanna ecosystems. Ecology 84 :337-350. Van Zandt, P. A., and A. A. Agrawal. 2004a. Community-wide impacts of herbivore-induced plant responses in milkweed ( Asclepias syriaca ). Ecology 85 :2616-2629. Van Zandt, P. A., and A. A. Agrawal. 2004b. Specifi city of induced plant responses to specialist herbivores of the common milkweed Asclepias syriaca. Oikos 104 :401-409. Vanni, M. J., D. L. DeAngelis, D. E. Schindler , and G. R. Huxel. 2004. Overview: Cross-Habitat Flux of Nutrients and Detritus. Pages 3-11 in G. A. Polis, M. E. Power, and G. R. Huxel, editors. Food Webs at the Landscape Level. Th e University of Chicago Press, Chicago. Vermeire, L. T., R. B. Mitchell, S. D. Fuhlendor f, and D. B. Wester. 200 4. Selective control of rangeland grasshoppers with prescribed fire. Journal of Range Management 57 :29-33. Warren, S. D., C. J. Scifres, and P. D. Teel . 1987. Response of grassland arthropods to burning a review. Agriculture Ecosystems & Environment 19 :105-130. Weekley, C. W., T. L. Kubisiak, and T. M. Race. 2002. Genetic impoverishment and crossincompatibility in remnant genotypes of Ziziphus celata (Rhamnaceae), a rare shrub endemic to the Lake Wales Ridge, Flor ida. Biodiversity and Conservation 11 :2027-2046. Whelan, R. J. 1995. The Ecology of Fire. Ca mbridge University Press, Cambridge. Whelan, R. J., and A. R. Main. 1979. Insect gr azing and post-fire plant succession in southwest Australian woodland. Austra lian Journal of Ecology 4 :387-398. White, P. S. 1979. Pattern, Process, and Natural Disturbance in Vegetation. 45 :229-299. Wiens, J. A., R. G. Cates, J. T. Rotenbe rry, N. Cobb, B. Vanhorne, and R. A. Redak. 1991. Arthropod dynamics on sagebrush ( Artemisia tridentata ): Effects of plant chemistry and avian predation. Ecological Monographs 61 :299-321. Wilson, W. 1960. Inclined point quadrats. New Phytologist 59 :1-8.

PAGE 127

127 Wimp, G. M., G. D. Martinsen, K. D. Floate , R. K. Bangert, and T. G. Whitham. 2005. Plant genetic determinants of arthropod community structur e and diversity. Evolution 59 :6169. Wright, I. J., and M. Westoby. 2003. Nutrient concentr ation, resorption and lifes pan: leaf traits of Australian sclerophyll sp ecies. Functional Ecology 17 :10-19. Wunderlin, R. P. 1998. Guide to the Vascular Plants of Florida. University Press of Florida, Gainesville, FL. Zimmerman, D., C. Pavlik, A. Ruggles, a nd M. P. Armstrong. 1999. An experimental comparison of ordinary universal kriging and inverse distance weighting. Mathematical Geology 31 :375-390.

PAGE 128

128 BIOGRAPHICAL SKETCH Tania N. Kim was born in Tor onto, Canada, but raised in Mo ntreal, Canada. She attended McGill University where she completed her Bachelor of Science in the Department of Natural Resource Conservation. After colle ge, she worked in various fiel ds of ecology, such as marine invertebrate research, invasive species management, and food-we b ecology. After three years away from academia, she enrolled in the Master of Science program at the University of Florida where she completed her degree in 2006. She is currently pursuing her PhD at Florida State University.