The Effect of Forest Fragment Quality on Cerambycid

Home , Acanthocinini, Acanthoderini, Aegomorphus, Aegomorphus modestus, Astylidius, Astylopsis

THE EFFECT OF FOREST FRAGMENT QUALITY ON CERAMBYCID

OCCUPANCY AND ABUNDANCE

Kaitlin Handley

A thesis submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Master of Science in Entomology

Spring 2014

UMI Number: 1562381

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.

UMI 1562381 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346

THE EFFECT OF FOREST FRAGMENT QUALITY ON CERAMBYCID

OCCUPANCY AND ABUNDANCE

Kaitlin Handley

Approved: ______Judith A. Hough-Goldstein, Ph.D. Professor in charge of thesis on behalf of the Advisory Committee

Approved: ______Jacob L. Bowman, Ph.D. Chair of the Department of Entomology and Wildlife Ecology

Approved: ______Mark W. Rieger, Ph.D. Dean of the College of Agriculture and Natural Resources

Approved: ______James G. Richards, Ph.D. Vice Provost for Graduate and Professional Education ACKNOWLEDGMENTS

I would like to acknowledge and thank my committee members, Judy Hough-

Goldstein, Vince D’Amico, Charles Bartlett, and Greg Shriver for all of their guidance and support over the last two years. Each of you brought different things to my project, and it would not have been possible without help from all of you. I want to thank Judy Hough-Goldstein in particular for always being around whenever I had a question and reading and editing my thesis about 100 times and getting it back to me as quickly as humanly possible.

I would also like to thank my excellent field technicians, Jake Shaner, Lily

Newton, Theresa Andrew, and Sam Nestory. You all did an excellent job and made my field work fun! I hope that you all have extremely successful futures, because you are all wonderful and deserve great things.

I also want to thank my fellow graduate students in the ENWC department.

These last two years would have been much less fun without all of you. I especially want to thank Scott Berg and David Gardner for all of their help, humor, and emotional support. I’ve enjoyed every minute spent with you two.

Finally, I want to thank my family. You have been there every step of the way and are interested in my work even if it is hard to understand and strange because I study beetles.

iii TABLE OF CONTENTS

LIST OF TABLES ...... v LIST OF FIGURES ...... vii ABSTRACT ...... viii

Chapter

1 DIVERSITY AND PHENOLOGY OF CERAMBYCID BEETLES IN DELAWARE ...... 1

1.1 Materials and Methods ...... 2

1.1.1 2012 ...... 2 1.1.2 2013 ...... 4 1.1.3 Statistical analysis ...... 5

1.2 Results ...... 6 1.3 Conclusion ...... 16

2 FACTORS INFLUENCING THE OCCUPANCY AND ABUNDANCE OF LONG-HORNED BEETLES (COLEOPTERA: CERAMBYCIDAE) IN FOREST FRAGMENTS IN NORTHERN DELAWARE ...... 23

2.1 Methods and Materials ...... 25

2.1.1 Tree Survey ...... 28 2.1.2 Statistical Analysis ...... 29

2.2 Results ...... 32 2.3 Conclusion ...... 47

REFERENCES ...... 56

iv LIST OF TABLES

Table 1.1 Taxa and abundance of Cerambycidae captured by cross-vein panel traps in Delaware during 2012-2013...... 9

Table 1.2 Taxonomy and numbers of beetle families that were captured by cross- vein panel traps in Delaware during 2012 and 2013...... 11

Table 2.1 Study sites used for the 2012 and 2013 field seasons...... 31

Table 2.2 Total cerambycid specimens and species captured in each trap in 2013 and 2012, arranged in rank order of abundance for 2013...... 33

Table 2.3 Differences in catch between traps at sites with two traps located 100 m to 180 m from each other (2013) ...... 34

Table 2.4 Models to predict Phymatodes amoenus abundance. The global model is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead))...... 36

Table 2.5 Models to predict Megacyllene caryae abundance. The global model is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead))...... 37

Table 2.6 Models to predict Xylotrechus colonus abundance. The percent sweetgum model is the best (abundance= βo+β1 (pct_sweetgum))...... 38

Table 2.7 Models to predict Prionus laticollis abundance. The global model is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead))...... 39

Table 2.8 Models to predict Graphisurus fasciatus abundance. The global model is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead))...... 40

v Table 2.9 Models to predict Anelaphus villosus abundance. The model that incorporates percentages of all dominant tree species is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech))...... 41

Table 2.10 Models to predict Cyrtophorus verrucosus abundance. The patch size model is the best (abundance= βo+β1 (patch_ha))...... 42

Table 2.11 Models to predict Euderces pini abundance. The percent tulip tree is the best model (abundance= βo+β1 (pct_tulip))...... 43

Table 2.12 Models to predict Neoclytus acuminatus abundance. The percent sweetgum model is the best (abundance= βo+β1 (pct_sweetgum))...... 44

Table 2.13 Models to predict Neoclytus mucronatus abundance. The global model is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead))...... 45

Table 2.14 Beta (±SE) estimates for best models for the top ten species of cerambycids captured in 2013...... 46

vi LIST OF FIGURES

Figure 1.1 Location of FRAME study sites in Newark, Delaware. Sites where traps were located for both the 2012 and 2013 field seasons are marked in black. Sites used only in 2013 are dark grey. Additional 2013 sites located in Wilmington and Hockessin are not depicted on this map have utm coordinates of 18S 456999 4400333 (Wilmington 1), 18S 455821 4401112 (Wilmington 2), 18S 444616 4404625 (Mount Cuba Interior), and 18S 441685 4404829 (Hough)...... 5

Figure 1.2 Species with total catch greater than 20 individuals in 2012 arranged chronologically by subfamily. Bars indicate range of activity period. Species in bold were captured in 2012 and 2013...... 12

Figure 1.3 Species with total catch greater than 20 individuals in 2013 arranged chronologically by subfamily. Bars indicate range of activity period. Species in bold were captured in 2012 and 2013...... 13

Figure 1.4 Mean (± SEM) number of individuals collected in traps with different lures. Letters indicate significant differences (Two-way ANOVA, Tukey’s Test)...... 14

Figure 1.5 Mean (± SEM) ) number of individuals collected in traps with different lures. Letters indicate significant differences (Two-way ANOVA, Tukey’s Test)...... 15

vii ABSTRACT

Cerambycidae are a large family of beetles that have been studied in detail in recent years. They are of particular interest because there are a number of species that are important pests of silviculture. Many of these species are not native in the area in which they are considered pests. Therefore recent research has developed synthesized aggregation pheromones that attract a broad range of cerambycid species that can be used in sentinel traps in ports to detect non-native beetles before they can establish so that eradication is possible.

These pheromone blends were used in this study to test these lures as possible sentinel traps for non-native cerambycids and to survey the cerambycids of Delaware.

Field bioassays were conducted in collaboration with the USDA Forest Service

FRAME project (Forest Fragments in Managed Ecosystems). The goal was to determine which species occurred in which forest fragments and predict what forest factors influenced abundance. Traps baited with pheromone blends and host plant volatiles were put out in 24 for sites for 22 weeks from April to September in 2012 and 2013. The catches in these traps were collected once a week. A total of 15,368 cerambycid beetles in 72 species were captured including one exotic species. The forest factors predicted to influence cerambycid abundance that were measured in each of the sites were tree species, patch size, and tree health. I tested these factors using

viii AIC values from the unmarked package in R. Tree species and patch size were significant factors in determining cerambycid abundance on a species by species basis.

Tree stress was not, but the visual cues that were used to determine health may have not been accurate enough to successfully predict cerambycid abundance.

ix Chapter 1

DIVERSITY AND PHENOLOGY OF CERAMBYCID BEETLES IN DELAWARE

Recent research has revealed considerable pheromonal parsimony within the beetle family Cerambycidae, with closely and more distantly related species producing similar pheromone components (Hanks and Millar 2013). This research has resulted in many experiments to develop the best blends of pheromone to attract the greatest number of cerambycid species. The majority of pheromones identified to date are produced by males and attract both sexes (Hanks and Millar 2013). Each of the synthesized pheromones that I used also attract a number of species in three subfamilies of cerambycids, Cerambycinae, Lamiinae, and Spondylidinae (Hanks and

Millar 2013). Cerambycidae are of particular interest because this family contains many species that are among the most important insect pests of woody plants in natural and managed ecosystems worldwide (Hanks et al. 2012). There has also been a detailed survey of the cerambycids of Pennsylvania, and I would like to compare the similarities or differences in the richness of communities between these close areas

(Hanks and Millar 2013). The sites used for this project are a part of the larger scale FRAME (Forest

Fragments in Managed Ecosystems) project (Rega 2012), and have been partially characterized by soil, understory vegetation species and density, and litter dwelling arthropods. The goal of the FRAME project is to develop strategies for land owners to manage their land in a way that maximizes ecosystem services. It is a multifaceted

1 project that is looking at a broad range of forest aspects, including native bee behavior and abundance, tick occupancy and density, bird populations in relation to non-native plants, small mammals, and meso-carnivores. In collaboration with these efforts, I investigated the cerambycid species that can be attracted using synthesized pheromone blends. I also tested the effectiveness of pheromone lures to attract non-native beetles that may come in from shipments from other countries. If effective, sentinel traps could be used at ports to detect non-native cerambycids, potentially allowing quick eradication of invasive species (Hanks et al. 2012). Therefore, the goals of this experiment were to develop a quantitative survey of the long-horned beetles that can be found in northern Delaware and to test the synthesized pheromone as a probe for detecting potentially detrimental non-native long-horned beetles.

1.1 Materials and Methods

1.1.1 2012 From April and through September 2012, 48 traps located in 12 FRAME forest fragments near Newark, DE (Figure 1.1) were collected once a week. The traps were black cross-vein panel traps (corrugated plastic, 1.2 m high by 0.3 m wide) from

Alpha Scents Inc. (West Linn, Oregon), coated in Fluon or Insect-a-Slip (Bioquip Products, Rancho Dominguez, California). Ethylene glycol was used as a preservative and killing agent. Traps were suspended from 2-m tall PVC pipe frames. Four traps were located at each site, baited with three different lures known to attract cerambycids, and a control (Hanks et al. 2012). The traps were randomly arranged 25 m apart in a square. The treatments were MelangeC, a synthesized cerambycid

2 aggregation pheromone consisting of 2,3-Hydroxyhexanone (50 mg), 2,3- Hydroxyhexanol (50 mg), Fuscumol (50 mg), Fuscumol acetate (50 mg), Monochamol (25 mg), and 2-Methybutanol (50 mg) in isopropanol; ethanol, which mimics host plant volatiles; a combination of MelangeC and ethanol as separate baits; and isopropanol as the control. The components of MelangeC are known attractants for cerambycids that should be native to this area based on field surveys of Pennsylvania, and that would be logical candidates for inclusion in multispecies lures (Hanks et al.

2012). Ethanol has been used for years as an attractant for woodboring beetles, especially bark beetles (Coleoptera: Curculionidae: Scolytinae) (Oliver and Mannion 2001; Redding et al. 2013). To prepare the pheromone lures, 0.275 ml of pure pheromone blend (prepared in the Hanks laboratory by Judy Mongold-Diers) was dissolved in 0.725 ml 91% isopropyl alcohol, and put into 5.0 cm by 7.6 cm re-sealable bags. These lures were changed every other week. The control lures consisted of 1 ml of isopropyl alcohol in a 5.0 cm by 7.6 cm re-sealable bag and were also replaced every other week. At the start of the season, 100 ml of ethanol was added to 10.2 cm by 15.3 cm bags, which lasted the whole field season. All of these lures were attached to the center of each trap by binder clips.

A sieve was used to filter the insects from the ethylene glycol, and the waste was either reused or disposed of if diluted with rain water. Insects were taken back to the laboratory where they were sorted by family. In addition to cerambycids, I saved other xylophagous species (scolytines, elaterids, bostrichids, buprestids, anobiids) and predators of xylophagous species (clerids). I also saved staphylinids as general predators, because they were captured in large numbers in my traps and I was

3 interested to see if there was a correlation. All xylophagous species and predators of xylophagous species were counted and saved in alcohol. All cerambycids were pinned and identified to species based on specimens sent to the Hanks lab for identification verification and according to the Illustrated Key to the Longhorned Woodboring Beetles of the Eastern United States (Lingafelter 2007).

1.1.2 2013 In 2013, the same traps and collecting intervals were used (22 weeks from April to September), except only one trap was used per site, with a pheromone blend in combination with an ethanol lure. In 2012 I wanted to see which lure was the most effective in attracting cerambycid beetles, while in 2013 I was more interested in the species richness of cerambycid beetles in urban forest fragments. In 2013 the pheromone blend contained MelangeC enhanced by the addition of citral (50 mg) and prionic acid (1 mg) to try to capture more species that are known to be attracted to these lure components (mainly Megacyllene caryae and Prionine species). The pheromone was prepared in the same manner as in 2012, but the concentration was increased. The new blend was prepared with 0.32 ml of pheromone mixed with 0.68 ml of 91% isopropanol. Traps were placed in 24 forest fragments, including all of the ones used in 2012 and 12 additional fragments (Figure 1.1). Of the 24 fragments, six of them had two traps, for a total of 30 traps. All cerambycids were identified to species and counted, but in 2013 they were stored in vials of alcohol. I also counted and kept the other taxa as in 2012.

1.1.3 Statistical analysis The mean number of individuals of cerambycids as a family and for each subfamily that were caught in each treatment over the entire 2012 field season was compared using a two-way ANOVA by trap type and site, followed by Tukey’s test. Results for other borers and predators were also compared using a two-way ANOVA by trap type and site and Tukey’s test. These analyses used R (R Core Team 2012).

1.2 Results A total of 15,368 cerambycid beetles in 72 species were collected in the traps, including 31 species in 13 tribes of the subfamily Cerambycinae, 22 species in seven tribes of Lamiinae, 13 species in two tribes of Lepturinae, one species in each of the Necydaliniae and Parandrinae, two species in the Prioninae, and two species in the Spondylidinae (Table 1.1; numbers for 2012 are totals for all four trap types). Only one of the captured species is known to be exotic: Phymatodes testaceus (L.), which is native to Europe (Swift and Ray 2010). Of the captured cerambycids, 33 species were represented by 1-5 specimens, 10 species were represented by 6-19 specimens, and 31 species were caught in numbers greater than 20 individuals (Table 1.1). The most abundant species was the cerambycine Xylotrechus colonus (F.) which was represented by 4,617 specimens. There were 37 species of cerambycids were caught in both 2012 and 2013. Of those

37 species, 13 of them were captured in numbers greater than 20 individuals. Fewer than 100 specimens total were captured for the subfamilies Necydalinae, Parandrinae, and Spondylidinae. The most abundant taxon of other wood-boring beetles collected were the scolytines, with more than 29,000 individuals captured in 2012 and 2013 together (Table 1.2). They were most active early in the season, with most caught between early May and late June. Eucnemids were active for the month of June, while elaterids and staphylinids came into the traps consistently all season. Buprestids, bostrichids, clerids, and anobiids were less abundant in my traps. I caught fewer than 300 individuals for each of these families over both seasons (Table 1.2).

Activity periods began in mid-April, with the cerambycines Phymatodes amoenus (Say) and Cyrtophorus verrucosus (Olivier) (Figure 1.2). Cerambycines that were best represented in the data set (catch greater than 20 for the season) varied greatly in their phenologies, ranging from a few weeks for Phymatodes amoenus (Say) and Gaurotes cyanipennis (Say), to five months for Xylotrechus colonus (F.) and Neoclytus a. acuminatus (F.) (Figure 1.2). Aside from Psenocerus supernotatus LeConte which flew during May and early June, many of the lamiine species were active over a more uniform time of late June through early July. Six species caught in 2013 but not in 2012 were Phymatodes varius, Megacyllene caryae, Neoclytus caprea, Euderces picipes, Euderces pini, and Prionus laticollis (Figure 1.3). Megacyllene caryae was the earliest cerambycid to emerge, with 644 individuals caught in the first week alone. Prionus laticollis was one of the latest species to emerge with its first appearance being in the last week of June, and not peaking until mid-July. Of the 13 repeat species that were caught in numbers greater than 20 in both 2012 and 2013, 12 of them emerged about two weeks later in 2013 than in 2012 (Figures 1.2 and 1.3). From the 2012 data, I tested the efficacy of each lure for attracting cerambycid beetles. Overall, traps baited with the pheromone blend with ethanol attracted significantly more cerambycids (F(3,11) = 159.0, p <0 .001) than the other treatments (Figure 1.4). Cerambycines were significantly attracted to the pheromone and ethanol combination over other treatments (F(3,11) = 41.546, p <0.001) (Figure 4). For Cerambycidae overall the pheromone alone attracted more individuals than ethanol alone or isopropyl alcohol (Figure 1.4). Lamiines were also significantly attracted to the combination of pheromone and ethanol, but were equally attracted to the

7 pheromone alone (F(3,11) = 30.308, p<0.001). The lepturines had a greater attraction to the ethanol as a lure alone, with or without the pheromone (F(3,11) = 6.214, p=0.002) (Figure 1.4). I did not catch the other subfamilies in large enough numbers for statistical analysis. There were also differences in treatment attractiveness in the other borers and predators. Staphylinids and bostrichids were more attracted to the pheromone and pheromone and ethanol combination than the ethanol alone or the control (F(3,11) =

13.181, p<0.001; F(3,11) = 16.515, p<0.001) (Figure 5). Clerids and scolytids were more prevalent in the lures that included ethanol (F(3,11) = 4.667, p=0.007; F(3,11) = 13.822, p<0.001). There was no difference in treatment on the numbers of elaterids or eucnemids (data not shown). An unexpected finding from 2012 was that traps baited with isopropyl alcohol were attractive to American carrion beetles, Necrophila americana (L.) (Coleoptera: Silphidae). The catches for the control traps at two sites, Glasgow Park and Folk Park, for the weeks of July 9 and July 16 combined had over 150 beetles each, while no American carrion beetles were caught in the traps with the other three lures. At the Glasgow Park fragment, 172 N. americana were caught, and at the Folk Park fragment, 337 N. americana were caught.

Table 1.1 Taxa and abundance of Cerambycidae captured by cross-vein panel traps in Delaware during 2012-2013.

Taxon 2012 2013 Total Cerambycinae Anaglyptini Cyrtophorus verrucosus (Olivier) 246 267 513 Bothriospilini Knulliana cincta cincta (Drury) 0 1 1 Callidiini Phymatodes aereus (Newman) 5 3 8 Phymatodes amoenus (Say) 629 1933 2562 Phymatodes testaceus (Linnaeus) 76 79 155 Phymatodes varius (Fabricius) 0 4 4 Clytini Clytoleptus albofasciatus (Castelnau & Gory) 2 0 2 Clytus ruricola (Olivier) 35 12 47 Megacyllene caryae (Gahan) 9 1877 1886 Neoclytus a. acuminatus (Fabricius) 452 146 598 Neoclytus caprea (Say) 0 77 77 Neoclytus jouteli jouteli Davis 9 0 9 Neoclytus jouteli simplarius Blatchley 4 0 4 Neoclytus mucronatus (Fabricius) 224 116 340 Neoclytus scutellaris (Olivier) 32 2 34 Sarosesthes fulminans (Fabricius) 15 4 19 Xylotrechus colonus (Fabricius) 2924 1693 4617 Xylotrechus s. sagittatus (Germar) 1 0 1 Curiini Curius dentatus Newman 2 0 2 Eburiini Eburia quadrigeminata (Linnaeus) 1 0 1 Elaphidiini Anelaphus parallelus (Newman) 39 0 39 Anelaphus villosus (Fabricius) 146 276 422 Elaphidion mucronatum (Say) 200 53 253 Enaphalodes hispicornis (Linnaeus) 0 2 2 Hesperophanini Tylonotus bimaculatus Haldeman 1 0 1 Methiini Tessaropa tenuipes (Haldeman) 8 2 10 Molorchini Molorchus b. bimaculatus Say 12 1 13 Obriini Obrium maculatum (Olivier) 38 7 45

Table 1 continued Stenopterini Callimoxys sanguinicollis (Olivier) 0 1 1 T Tillomorphini Euderces picipes (Fabricius) 0 112 112 Euderces pini (Olivier) 0 187 187 Lamiinae Acanthocinini Astylidius parvus Casey 1 0 1 Astylopsis macula (Say) 86 54 140 Astylopsis perplexa (Haldeman) 0 1 1 Graphisurus despectus LeConte 151 8 159 Graphisurus fasciatus (Degeer) 790 472 1262 Graphisurus triangulifer (Haldeman) 1 0 1 Leptostylus transversus (Gyllenhal) 0 1 1 Lepturges angulatus (LeConte) 16 13 29 Lepturges confluens (Haldeman) 1 0 1 Sternidius misellus (LeConte) 9 5 14 Sternidius variegatus (Haldeman) 8 0 8 Styloleptus biustus (LeConte) 173 100 273 Acanthoderini Aegomorphus modestus (Gyllenhal) 92 5 97 Aegomorphus morrisii (Uhler) 3 0 3 Aegomorphus quadrigibbus (Say) 1 1 2 Oplosia nubila (LeConte) 17 3 20 Urgleptes querci (Fitch) 16 9 25 Cyrtinini Cyrtinus pygmaeus (Haldeman) 1 0 1 Desmiphorini Eupogonius pauper LeConte 1 0 1 Psenocerus supernotatus LeConte 147 39 186 Lamiini Goes pulverulentus (Haldeman) 1 0 1 Monochamus scutellatus (Say) 0 1 1 Pogonocherini Ecyrus d. dasycerus (Say) 6 17 23 Saperdini Saperda lateralis Fabricius 2 1 3 Lepturinae Lepturini Bellamira scalaris (Say) 23 1 24 Brachyleptura rubrica (Say) 2 0 2 Guarotes cyanipennis (Say) 132 90 222 Judolia cordifera (Olivier) 1 0 1 Metacmaeops vittata (Swederus) 9 1 10 Strangalepta abbreviata Casey 1 0 1 Strophiona nitens (Forster) 2 0 2 Trachysida mutabilis (Newman) 2 0 2

Table 1 continued Trigonarthris minnesotara (Casey) 0 1 1 Trigonarthris proxima (Say) 2 1 3 Typocerus lugubris (Say) 11 4 15 Typocerus v. velutinus (Olivier) 2 0 2 Xylosteini Leptorhabdium pictum (Haldeman) 2 0 2 Necydalinae Necydalis mellita (Say) 1 0 1 Parandrinae Parandrini Neandra brunnea (Fabricius) 59 12 71 Prioninae Prionini Orthosoma brunneum (Forster) 20 1 21 Prionus laticollis (Drury) 6 748 754 Spondylidinae Asemini Arhopalus rusticus (Haldeman) 5 10 15 Asemum striatum (Linneaus) 1 0 1

Table 1.2 Taxonomy and numbers of beetle families that were captured by cross-vein panel traps in Delaware during 2012 and 2013.

Taxon 2012 2013 Total Curculionidae Scolytinae 9260 19742 29002 Cerambycidae 6971 8481 15452 Eucnemidae 1719 96 1815 Elateridae 1103 268 1371 Staphylinidae 935 401 1336 Bostrichidae 127 40 167 Cleridae 95 69 164 Anobiidae 127 7 134 Buprestidae 19 7 26

Figure 1.2 Species with total catch greater than 20 individuals in 2012 arranged chronologically by subfamily. Bars indicate range of activity period. Species in bold were captured in 2012 and 2013.

Figure 1.3 Species with total catch greater than 20 individuals in 2013 arranged chronologically by subfamily. Bars indicate range of activity period. Species in bold were captured in 2012 and 2013.

Figure 1.4 Mean (± SEM) number of individuals collected in traps with different lures. Letters indicate significant differences (Two-way ANOVA, Tukey’s Test).

Figure 1.5 Mean (± SEM) ) number of individuals collected in traps with different lures. Letters indicate significant differences (Two-way ANOVA, Tukey’s Test).

1.3 Conclusion

Hanks and Millar (2013) assessed the variety of cerambycid species attracted to pheromone components from 2009 until 2011 at 47 sites in Pennsylvania. In my study, the pheromone used in 2012 consisted of the same components as were used by

Hanks and Millar (2013). Citral and prionic acid were added to my lure in 2013.

Because Delaware and Pennsylvania are neighbors and the pheromone consisted of most of the same components, I expected the Delaware sites to have a similar composition to the sites in Eastern Pennsylvania. Along with the pheromone, host plant volatiles have been found to enhance cerambycid catches, and helped to catch large numbers of species simultaneously (Hanks et al. 2012, Hanks and Millar 2013).

When plants are stressed they produce ethanol. Kimmerer and Kozlowski (1982) found that when red pine (Pinus resinosa Aiton) and paper birch (Betula papyrifera

Marshall) seedlings were exposed to sulfur dioxide, they produced acetaldehyde and ethanol. Ginzel and Hanks (2005) showed that the species Neoclytus mucronatus and

Xylotrechus colonus were attracted by volatiles released by the stressed and dying trees that are hosts for their larvae. My results helped support this claim because as a whole, cerambycids had a significantly greater attraction to the pheromone and ethanol treatment over the other treatments (Figure 1.4).

Hanks and Millar (2013) covered a greater area and deployed traps over a longer period of time, and so caught a greater number of specimens and species than this study. During their three years of sampling, Hanks and Millar (2013) captured a

16 total of 15,438 cerambycid beetles of 134 species. Many of the species captured in

Pennsylvania were also captured in Delaware. Of the 72 species that I captured, 64 of them were also found in Pennsylvania (Hanks and Millar 2013). The unique species caught in my study were Knulliana cincta cincta, Neoclytus jouteli jouteli, Neoclytus jouteli simplarius, Enaphalodes hispicornis, Callimoxys sanguinicollis, Sternidius variegates, Cyrtinus pygamaeus, and Goes pulverulentus (Table 1.1). Knulliana cincta cincta, Callimoxys sanguinicollis and Cyrtinus pygmaeus are in unique tribes,

Bothriospilini, Stenopterini and Cyrtinini (Table 1.1). I found fewer than ten of these unique species each, between both field seasons.

Cerambycid species emerged between one and two weeks later in 2013 than they did in 2012. According to a study by Kenna and Moore (2010), Asian longhorned beetles need temperatures between 10˚ and 30˚C for successful development. In 2012 there were 184.5 growing degree days from March 1 to April 15. In 2013 there were only 84.5 growing degree days between March 1 and April 15

(http://www.weather.com/outdoors/agriculture/growing-degree-days/19716). Different species probably have different degree day requirements, but there are not a lot of data on most of the native cerambycid species’ developmental requirements. Insects in general develop more slowly at lower temperatures, and since 2013 was cooler than

2012, cerambycids emerged later.

The large richness of cerambycids captured in 2012 and 2013 shows the effectiveness of the pheromone blend in attracting a broad range of beetles. It also shows that there are similarities in the pheromones that different cerambycid species

17 produce. Multiple species can be attracted to the same pheromone components because of reproductive isolation due to the staggered emergence time of some species over the season, and difference in species that are crepuscular and species that are nocturnal (Hanks and Millar 2013). For example, if Phymatodes amoenus and

Phymatodes testaceus are attracted to the same components of the pheromone, they can still avoid each other because Phymatodes amoenus emerges and peaks one to two weeks before Phymatodes testaceus (Figures 1.2 and 1.3). Hanks and Millar (2013) found that the species C. dentatus could potentially be attracted to the pheromone emitted by male S. fulminans but C. dentatus is active much later in the season so they also avoid each other.

The different components in the pheromone in 2013 compared to 2012 attracted six new cerambycid species in large numbers. The citral component of the pheromone attracted large numbers of Megacyllene caryae (Mitchell et al. 2012) and the prionic acid attracted large numbers of Prionus laticollis (Barbour et al. 2011).

Other new species were Euderces pini, Euderces picipes, Neoclytus caprea, and

Phymatodes varius. They were all most likely attracted by the citral component, because they were found at the sites that were used in 2012 suggesting that they were present there and just not attracted to the pheromone blend used before. They are also all in the subfamily Cerambycinae and prionic acid is a component that is attractive to the cerambycids in the subfamily Prioninae.

Based on the significant difference between cerambycid attraction to the pheromone combined with ethanol lures, and the pheromone alone, the majority of

18 cerambycids are attracted to places where there are other beetles of the same species, if there are plant volatiles as well. Ginzel and Hanks (2005) found that mate location for certain cerambycine species involves three sequential behavioral stages: (1) both sexes are attracted to larval hosts by plant volatiles; (2) males attract females over shorter distances with pheromones; and (3) males recognize females by contact pheromones in their epiculticular wax layer. Adult Xylotrechus colonus of both sexes were attracted by volatiles emanating from cut hickory logs in olfactory bioassays, which suggest that mate location is mediated by plant volatiles, not long-range sex pheromones (Ginzel and Hanks 2005), even though my results show that it was also attracted to the pheromone blend alone. The only species that I captured that was more attracted to ethanol alone than the pheromone trap was Gaurotes cyannipennis. It is in the subfamily Lepturinae, so it probably was not as attracted to the pheromones that were geared towards cerambycines and lamiines. Cerambycines as a whole emerged earlier in the season than lamiines.

Based on the current study and past experiments on the other wood boring beetles, scolytines and clerids are highly attracted to ethanol. Field experiments using baited sticky stovepipe traps and Lindgren multiple funnel traps done in Ontario,

Canada showed that several species of conifer-feeding beetles were attracted to conifer monoterpenes or to monoterpenes and ethanol, including multiple species of scolytines

(Chenier and Philogene 1989). Clerids from this experiment in Ontario were found to be attracted to the conifer monoterpenes (Chenier and Phiilogene 1989). Schowalter

(2012) showed that under drought stress condition, loblolly pine (Pinus taeda L.) is

19 especially vulnerable to bark beetles because it produces volatiles that attract scolytines. My results support this because scolytines were more abundant in traps with ethanol which is a volatile produced by stressed plants (Kimmerer and Kozlowski

1982). There are a number of checkered beetle (clerid) species that are important predators of scolytines and have even been selected as potential predatory biological control agents for some of the particularly problematic scolytine species (Schrey et al.

2005). Therefore my findings and the findings of Chenier and Philogene (1986) suggest that clerids are attracted to the same volatile that their prey is attracted to.

Bostrichids and staphylinids were attracted to the pheromone blend. Bostrichids are also wood borers (Nardi and Zahradnik 2004) so they could either tune into the cerambycid pheromones to find acceptable host trees, or they may produce a similar pheromone and are also attracted to the components of the synthesized pheromone blend. Staphylinids are predators, so they may have been attracted to the pheromone because that is what they use to find their prey. Certain staphylinids are thought to be predators of xylophagous insects (Kennedy and McCullough 2002, Balog et al. 2008).

However, the staphylinids collected in the current study were not identified to species, and their actual prey is not known.

There were no significant differences in lure attraction for elaterids or eucnemids. Elaterids were abundant throughout the entire field season, and because I only identified related beetles to family, I did not take into consideration specific species of elaterids. Some elaterids are important because their larvae are agricultural pests, and some are important in silviculture because they are saproxylic species that

20 feed on wood decaying organisms (Majka and Johnson 2008). They could also be predators in woodland environments (Majka and Johnson 2008). Aside from this, little is known about elaterid ecology, even though it is the seventh most taxonomically diverse beetle family globally, with almost 1,000 species in North America (Majka and Johnson 2008). Therefore the different species of elaterids may or may not be attracted to my lures, and could just be abundant in the ecosystem. Eucnemids did not show any preference for lure. They were abundant in the field and were probably just incidental catches. Details of eucnemid biology are not well known, but they are thought to develop in wood infected by fungi that cause white rot (Webster et al.

2012). They do appear to be good indicators of diverse forest structure (Webster et al.

2012), so while I did not find any significance in eucnemids attraction to lures, I did find significant differences in the fragments in which they were found (see Chapter 2).

Sites that were in later successional stages and that had lower percentages of non- native plants had significantly more eucnemids than the sites that were in earlier successional stages and had greater percentages of non-native plants. They are also associated with fungi in wood, so they could also be associated with older forests because there is a greater amount of dead wood and fungi available (Webster et al.

2012). Therefore in more established sites, eucnemids are more abundant, and that is why so many were captured without having significant lure attractions.

The large number of Necrophila americana that were captured in the control traps supports the findings of Reut et al. (2010) that American carrion beetles are attracted to isopropanol. Reut et al. (2010) captured 22 carrion beetles in six traps over

21 three weeks and found that most of the carrion beetles were caught around the second week of July. I captured 509 carrion beetles in two traps over two weeks from July 2 to July 16, with none caught in nearby traps with other lures, which shows even stronger evidence for Reut et al.’s (2010) findings that American carrion beetles are attracted to isopropanol.

In conclusion, this study has demonstrated that the synthesized pheromone blend MelangeC effectively attracts a broad range of cerambycid species. The attractive ability of the pheromone can be strengthened by adding additional pheromone components (i.e., citral and prionic acid) and by adding a separate treatment of ethanol. These combinations can also be used to attract other borer and predator families. Even though my sites consisted of many fragmented and degraded forests, I was still able to catch large number of cerambycids and related beetles.

Chapter 2

FACTORS INFLUENCING THE OCCUPANCY AND ABUNDANCE OF LONG-HORNED BEETLES (COLEOPTERA: CERAMBYCIDAE) IN FOREST FRAGMENTS IN NORTHERN DELAWARE

Cerambycidae are one of the larger families of insects with more than 35,000 species in almost 4,000 genera (Rodriguez-del-bosque 2013). It is the third largest family of insects making up almost 4% of all insect species (Hodkinson and Parnell

2007). Larvae of almost all species of cerambycids bore within and feed upon plant tissues, the majority in decaying wood, but many will feed on woody and herbaceous living plants (Yanega 1996). Cerambycidae are of particular interest as native and nonnative forest pests, many of which can be brought to baited traps (Hanks et al.

2012).

I was interested in different forest aspects that affect cerambycid abundance and occupancy in different forest fragments. Since I was trying to determine how different forest factors influence cerambycids, I chose sites that had different compositions to try to determine an array of factors that affect cerambycids.

Many cerambycid species are extremely limited in their host associations and others seem capable of feeding on a broad diversity of conifers or hardwoods (Yanega

1996); however, there seems to be a general conclusion that the degree of host specificity decreases with increasing wood decay (Milberg et al. 2014). Therefore it is important to understand which tree species, in which condition are preferred by each

23 cerambycid species so that that a species presence in a patch of forest can be accurately predicted based on forest composition. There is little known about the majority of cerambycid beetles because most of them do not damage silviculture and are therefore not of economic significance. Cerambycids that are highly studied are usually introduced pests, like the Asian longhorned beetle (Anoplophora glabripennis

(Motschulsky)) and Japanese Pine Sawyer (Monochamus alternatus (Hope)), which jeopardize tree health by girdling the branches and vectoring nematodes that cause pine wilt disease (Kobakashi et al. 1984, Dodds and Orwig 2011). Lingafelter (2007) developed an illustrated key to the longhorned beetles of the eastern United States, in which he provides general information concerning the tree species in which each of the cerambycid species of the eastern US prefer to oviposit. However, most of these species have not been studied individually or in depth so the data in Lingafelter’s key show multiple genera as hosts for many of these species or no host data at all.

Past studies have found that woodboring insects were captured in greater numbers in areas with shelterwood versus the margins of clearcuts (Dodds et al. 2010).

Other studies have addressed the benefits of having coarse woody debris as an important part of increasing forest ecosystem species richness because dead wood provides a variety of microhabitats (Priewasser et al. 2012). Species requirements for dead wood quantity and quality vary considerably and the availability of downed and standing dead wood in different stages of decomposition is crucial for species composition and richness (Priewasser 2012). Therefore, because cerambycids are woodborers, I predicted that sites in my forest fragments with a greater amount of

24 available dead wood would have more cerambycid species and a greater abundance of cerambycids.

Between 2012 and 2013 there was an addition of two more aggregation pheromone components to the pheromone blend used, to maximize the amount of cerambycids attracted to my traps. By using a pheromone blend that is known to attract multiple species of cerambycids (Hanks et al. 2012) I was hoping to sample a large range of cerambycid species and predict what factors, specifically tree species, tree health, and forest fragment size, affect their abundance and occupancy in urban forest fragments.

2.1 Methods and Materials

As detailed in chapter one, starting in April and continuing through September

2012, all of the insects found in 48 traps baited with four different lures, located at 12 different forest fragments located in or near Newark, DE (Figure 1.1) were collected once a week. Similarly, from April through September 2013, all of the insects found in

30 traps baited with a single lure, a pheromone blend plus ethanol, were collected once per week. The lure used in 2013 was the same as in 2012 but with the addition of two components, citral (50 mg) and prionic acid (1 mg). The 2013 traps were located at 24 forest fragments located in or near Newark, DE, including the 12 used in 2012. Of the

24 sites, six of them had two traps. These sites were all in New Castle County and were dominated by hardwood trees (Figure 1.1, Table 2.1).

The 12 sites used in 2012 were chosen based on dominant trees in each fragment. I ranked 16 of the 21 FRAME sites based on the species of trees present. I then separated them into four categories based on average DBH (diameter breast height) of the trees in each fragment and dominant tree species, and randomly chose three of the four of those as the twelve original sites.

In 2013, in six of the original fragments I placed one trap at the same site as used in 2012, and an additional trap between 100 and 180 meters away. I put a second trap in the Chrysler Woods fragment because a section of the forest was cut down to create a drainage ditch for a new building that was built on that land. All dead plant material was removed. The second trap that I put there was to determine whether there would be a difference between the heavily disturbed part of the forest and the still intact part of the forest. I also put an additional trap in my Laird, Iron Hill 2 and Folk fragments because they captured a large number of beetles in 2012 and I wanted to see if the high number of beetles was localized or if there were a large number of beetles throughout the entire fragment. Similarly, I put extra traps in the Ecology Woods and

Christina Creek 1 fragments because those sites had very few beetles in 2012 and I wanted to see if that was localized.

For the remaining twelve sites, I chose an array of ecosystem types. Mount

Cuba, the Newark Reservoir, and Rittenhouse Park were more established sites because they had been forests for over 77 years (based on aerial photographs taken in

1937, [http://demac.udel.edu/data/aerial-photography]), while Christina Creek 2 and

Motorpool were less established sites and on the edge of infrastructure (Table 2.1).

Iron Hill 1, White Clay 2, and White Clay 3 were chosen because they are sites that are part of the same contiguous patch of forest as some of my sites from 2012. White

Clay 2 and White Clay 3 are unique because both of them have a very different composition than White Clay 1 even though they are part of a contiguous forest. Even though these sites are located in the same patch of forest within 1 km of each other, they have very different tree and shrub composition (Table 2.1). White Clay 1 has been an established forest for over 77 years because it is on a steep hill which is unsuitable for farming, the main use of the surrounding landscape. It consists of mainly beech trees (Fagus grandifolia Ehrlich). White Clay 2 on the other hand was used as a farm field, and was let go to return to forest. It is heavily invaded with invasive plants, primarily autumn olive (Elaeagnus umbellata Thunberg) and multiflora rose (Rosa multiflora Thunberg). White Clay 3 is a successional forest fragment consisting of mainly tulip trees (Liriodendron tulipifera L.). The Hough site is a small patch of trees in a neighborhood in a very suburban area. This was chosen to see if cerambycids would live in areas that are heavily inhabited by humans. My final sites were chosen because of their locations near places where there is transport of goods from other countries and other parts of the United States. Devon Park is adjacent to train tracks, and I had two sites near the Port of Wilmington. The rail line next to Devon Park is part of the CSX Transportation system, and is connected by rail lines to almost every state east of the Mississippi River. The Port of Wilmington imports goods from all over the world, especially fruit and juice concentrates from

Central and South America, but also automobiles, livestock, and break bulk (goods

27 that are loaded individually and not in large containers) and bulk cargoes including steel, forest products, dry bulk materials, and petroleum products

(http://www.portofwilmington.com). I put the traps in these locations to test their sentinel abilities in capturing non-native cerambycids coming into the country from wood imports. The age of these fragments were determined by aerial photographs over time from the Delaware Environmental Monitoring and Analysis Center

(http://demac.udel.edu/data/aerial-photography).

2.1.1 Tree Survey

I conducted tree surveys so that I would have a detailed record of all of the trees surrounding my traps. Using Forest Inventory and Analysis (FIA) standards set out by the Forest Service, I sampled all of the trees within 11.3 m of each of the traps

(http://www.fia.fs.fed.us).

To measure the distance at each site, I cut four pieces of 11.3 m double loop chain and placed them in the four cardinal directions surrounding the trap. The chains separated the circle around the trap into four quadrants that I sampled one by one. In each quadrant I took the bearings of each tree that I sampled using a compass and identified it to species. I also used a laser rangefinder (Laser Technology, Inc.,

Centennial, CO) to determine the height of each tree and measured the DBH of each tree. I quantified tree health overall for each tree using a scale from one to five. The rankings were: 1 = no large dead branches, full dense canopy, 2 = <15% of branches

28 dead, mostly full canopy, 3 = about half of branches dead, 4 = mostly dead branches, little vegetation, 5 = completely dead, no vegetation. Finally, I quantified ground cover of each quadrant based on tree cover (using the base of the trunk of each tree), shrub cover, and ground cover.

2.1.2 Statistical Analysis

Sites were compared qualitatively between traps within the same fragment for

2013 data. The top ten most abundant cerambycid species collected in 2013 were modeled against each of the dominant tree species found in the forest fragments

(Quercus spp., Acer rubrum, Nyssa sylvatica, Fagus grandifolia, Liriodendron tulipifera, and Liquidambar styraciflua), all of the trees species together, the health of the trees surrounding the trap, the size of the forest fragments, and all of these covariates combined in a global model (cerambycid abundance = βo + percent oak + percent red maple + percent blackgum + percent tulip tree + percent sweetgum + percent American beech + percent trees with stress level above 3 + size of patch + percent of dominate tree species combined + percent tree, percent stress, and patch size combined). Week of catch was used as the detection covariate because of the varying phenologies in the beetle species. Models were also run with no detection covariate. The ten species that were examined were the most abundant species captured: Xylotrechus colonus, Graphisurus fasciatus, Phymatodes amoenus,

Neoclytus a. acuminatus, Cyrtophorus verrucosus, Neoclytus mucronatus,

Megacyllene caryae, Prionus laticollis, Anelaphus villosus, and Euderces pini.

All trees were compared using single species except for species in the genus

Quercus which were combined because of their diversity in the sites. This analysis was done using the Akaike information criterion (AIC) in the unmarked package

(Akaike 1974; Fiske et al. 2014) in R (R Core Team 2012). I chose the unmarked package to run my analyses because I was trying to predict cerambycid abundance from repeated counts, and unmarked has a function, pcount, that helps predict abundance and fits models to see which covariates have the greatest effect on abundance (Fiske and Chandler 2011).

Table 2.1 Study sites used for the 2012 and 2013 field seasons.

Elaeagnusumbellata

Rosamultiflora

Large forest Large patch urban in area

Located next toLocated thenext Wilmington ofPort

Heavily invaded Heavily by

Establishedforest managed in ecosystem

Locatedalong edge forestof

Isolatedby suburbana landscape

Locatedalong edge forestof

Isolatedyears by over agriculturefor 100

Adjacenttotracks cargo railroad train

Clear cut between 2012 and cut 2013 between Clear 2012

Interesting CharacteristicsInteresting

Stand

Age of Ageof

1.17

5.95

4.55

5.53

20.32

11.15

50.81

53.34

62.09

75.58

16.56

25.57

12.71

50.81

Patch Patch

163.56

157.87

163.56

150.06

116.55

Size(ha)

Number Number

of Trees of

Species

Acerrubrum

Quercusspp.

Acernegundo

Dominant Tree Tree Dominant

Nyssasylvatica

Fagusgrandifolia

Carpinuscaroliniana

Robinia pseudoacacia Robinia

Liriodendron tulipifera Liriodendron

Liquidambar styraciflua Liquidambar

30.80

13.65

21.55

29.71

25.68

21.15

26.07

25.49

32.07

19.48

17.00

20.05

15.93

19.96

16.03

19.47

36.97

27.50

11.60

21.82

22.45

19.81

23.79

19.04

23.27

18.02

16.46

26.54

16.76

Average Average

DBH (cm) DBH

2013

Year Year

2012/2013

WC3

Wilm2

Wilm1

MP_I9

Devon

RH_F9

Hough

FO_G5

CD_F2

RE_A1

LA_B4

IH2_E7

IH1_F3

FO_A3

EW_E7

SL2_C4

EW_C5

WF_D4

IH2_A7

GG1_E1

CW_G5

CW_C3

CC1_E7

Trap ID Trap

LA_F10

GG2_C4

CC2_C3

MCI_C4

CC1_A2

WC1_E4

WC2_C2

Site

Laird

Folk Park Folk

Coverdale

Motorpool

WebbFarm

DevonPark

White Clay 3 White Clay

White Clay 2 White Clay

White Clay 1 White Clay

Wilmington 2 Wilmington

Wilmington 1 Wilmington

Sunset2 Lake

HoughHouse

Iron Hill Park 2 Park IronHill

Iron Hill Park 1 Park IronHill

Glasgow Park 2 Park Glasgow

Glasgow Park 1 Park Glasgow

EcologyWoods

Chrysler WoodsChrysler

Christina Creek 2 Creek Christina

Christina Creek 1 Creek Christina

RittenhousePark

Newark Reservoir Newark MountCubaInterior

2.2 Results

We captured a total of 6,914 beetles representing 62 cerambycid species during the 22 week period that the traps were deployed in 2012, and 8,454 beetles representing 48 cerambycid species, during the 22 week period that the traps were deployed in 2013 (Table 2.2). Numbers of individuals caught varied 5.2-fold in 2012 and 3.5-fold in 2013 between the sites that caught the most beetles and those that caught the fewest (Table 2.2).

Table 2.2 Total cerambycid specimens and species captured in each trap in 2013 and 2012, arranged in rank order of abundance for 2013.

Individuals Species Individuals Species Site Trap ID (2013) (2013) (2012) (2012) Reservoir RE_A1 473 23 Sunset Lake 2 SL2_C4 418 20 955 29 Chrysler Woods CW_C3 393 19 545 34 Iron Hill 2 IH2_A7 368 23 White Clay 2 WC2_C2 364 15 Glasgow 1 GG1_E1 358 20 685 30 Devon Park Devon 349 23 Laird LA_B4 342 14 Glasgow 2 GG2_C4 336 18 765 32 Laird LA_F10 336 17 613 29 Mount Cuba Interior MCI_C4 312 17 Motorpool MP_I9 309 19 Coverdale CD_F2 306 20 528 26 White Clay 3 WC3 297 13 Webb Farm WF_D4 293 22 581 25 Folk FO_G5 288 17 663 27 Iron Hill 2 IH2_E7 284 21 649 34 Wilmington 2 Wilm2 276 19 White Clay 1 WC1_E4 254 12 397 27 Ecology Woods EW_E7 237 17 Folk FO_A3 229 16 Wilmington 1 Wilm1 228 13 Christina Creek 1 CC1_A2 208 17 348 26 Rittenhouse RH_F9 204 15 Iron Hill 1 IHI_F3 198 19 Hough House Hough 179 15 Chrysler Woods CW_G5 177 16 Christina Creek 2 CC2_C3 164 15 Christina Creek 1 CC1_E7 138 15 Ecology Woods EW_C5 136 19 185 19

I looked at the difference between the traps that were located in the same fragment as another trap in 2013. The biggest difference in beetles captured at the same site was in Chrysler Woods. I captured almost 40% more cerambycids at the trap that was located at the clear-cut part of the woods as I did in the interior forest trap

(Table 2.3). The traps in Laird, Iron Hill 2, and Folk Park captured similar numbers of cerambycids, with less than a 15% difference between the two traps in all of those fragments (Table 2.3). There were however large differences between the traps in

Ecology Woods and Christina Creek 1, each with over a 20% difference (Table 2.3).

All traps except the traps at Laird were statistically different from each other using an exact binomial test of goodness of fit (Christina Creek 1: p< 0.001, Chrysler Woods: p< 0.001, Ecology Woods: p< 0.001, Folk: p= 0.01, Iron Hill 2: p< 0.001).

Table 2.3 Differences in catch between traps at sites with two traps located 100 m to 180 m from each other (2013)

Total Site Trap ID Individuals Difference Christina Creek 1 CC1_A2 208 Christina Creek 1 CC1_E7 138 20.2% Chrysler Woods CW_C3 393 Chrysler Woods CW_G5 177 37.9% Ecology Woods EW_C5 136 Ecology Woods EW_E7 237 27.1% Folk FO_A3 229 Folk FO_G5 288 11.4% Iron Hill 2 IH2_A7 368 Iron Hill 2 IH2_E7 284 12.9% Laird LA_B4 342 Laird LA_F10 336 0.9%

The other set of traps that I had in a contiguous forest fragment were the traps located in White Clay Creek State Park (WC1, WC2, and WC3). Even though these sites were located in the same forest fragment within 1 km of each other, they had different numbers of cerambycids. White Clay 2, which was the most invaded, had the most cerambycids captured with 364 individuals in 15 species. White Clay 3 had 297 individuals in 13 species, and White Clay 1 had the fewest beetles with 254 beetles being captured in 12 species (Table 2.2).

The traps I had in the urban and suburban settings (Wilmington 1, Wilmington

2, and Hough) captured few cerambycids. I captured 276 cerambycids in Wilmington

2, 228 cerambycids in Wilmington 1, and 179 beetles in Hough. These catches are in the lower half of total catches by site for the 2013 field season (Table 2.2). The trap at

Devon Park, which was located along railroad tracks, captured over 300 beetles in 23 species.

Based on my AIC models, a few species have more specialized needs in the forest fragments, but for most species all forest factors combined are important. Date was not significant as a detection variable. For Phymatodes amoenus, Megacyllene caryae, Prionus laticollis, Graphisurus fasciatus, and Neoclytus mucronatus, all of the covariates that I tested were important for predicting their abundance and occupancy in urban forest fragments (Tables 2.4, 2.5, 2.7, 2.8, and 2.13). Xylotrechus colonus and Neoclytus acuminatus were both affected by the percentage of sweetgum in the fragment (Tables 2.6 and 2.12). Anelaphus villosus was affected by total tree composition (Table 2.9), Euderces pini was affected by the percentage of tulip trees

(Table 2.11), and Cyrtophorus verrucosus was affected by the size of the fragment (Table 2.10). Table 2.14 shows the beta estimates for all of the best models for the top ten species of cerambycids captured in 2013.

Phymatodes amoenus Model K AICc Delta_AICc AICcWt amo.global 10 9266.84 0.00 0.93 amo.global.day 11 9271.93 5.09 0.07 amo.tree 8 9305.91 39.07 0.00 amo.tree.day 9 9310.06 43.21 0.00 amo.oak 3 9370.80 103.96 0.00 amo.oak.day 4 9373.48 106.63 0.00 amo.nyssa 3 9391.41 124.57 0.00 amo.tulip 3 9392.61 125.77 0.00 amo.nyssa.day 4 9394.09 127.24 0.00 amo.tulip.day 4 9395.29 128.45 0.00 amo.patch 3 9432.47 165.62 0.00 amo.patch.day 4 9435.01 168.16 0.00 amo.beech 3 9445.24 178.39 0.00 amo.beech.day 4 9447.91 181.07 0.00 amo.sweetgum 3 9452.22 185.38 0.00 amo.sweetgum.day 4 9454.90 188.06 0.00 Null 2 9455.70 188.86 0.00 amo.acer 3 9456.17 189.32 0.00 amo.stress 3 9456.62 189.77 0.00 null.day 3 9458.18 191.34 0.00 amo.acer.day 4 9458.84 192.00 0.00 amo.stress.day 4 9459.29 192.45 0.00

Megacyllene caryae Model K AICc Delta_AICc AICcWt car.global 10 8130.24 0.00 0.97 car.global.day 11 8137.48 7.24 0.03 car.patch 3 8152.20 21.97 0.00 car.patch.day 4 8154.85 24.61 0.00 car.stress 3 8176.05 45.81 0.00 car.sweetgum 3 8176.08 45.84 0.00 car.tree 8 8178.37 48.14 0.00 car.stress.day 4 8178.72 48.49 0.00 car.sweetgum.day 4 8178.76 48.52 0.00 car.tree.day 9 8182.51 52.28 0.00 car.acer 3 8187.55 57.31 0.00 car.acer.day 4 8190.23 59.99 0.00 car.nyssa 3 8191.88 61.64 0.00 Null 2 8192.32 62.08 0.00 car.tulip 3 8192.46 62.22 0.00 car.beech 3 8194.31 64.08 0.00 car.oak 3 8194.37 64.13 0.00 car.nyssa.day 4 8194.56 64.32 0.00 null.day 3 8194.80 64.56 0.00 car.tulip.day 4 8195.14 64.90 0.00 car.beech.day 4 8196.99 66.75 0.00 car.oak.day 4 8197.04 66.81 0.00

Table 2.6 Models to predict Xylotrechus colonus abundance. The percent sweetgum model is the best (abundance= βo+β1 (pct_sweetgum)).

Xylotrechus colonus Model K AICc Delta_AICc AICcWt xy.sweetgum 3 4028.41 0.00 0.61 xy.sweetgum.day 4 4031.09 2.68 0.16 xy.tulip 3 4032.90 4.49 0.06 xy.stress 3 4033.80 5.39 0.04 xy.global 10 4034.59 6.18 0.03 xy.acer 3 4035.53 7.12 0.02 xy.tulip.day 4 4035.58 7.17 0.02 xy.stress.day 4 4036.48 8.07 0.01 Null 2 4036.93 8.52 0.01 xy.tree 8 4037.18 8.76 0.01 xy.oak 3 4037.24 8.83 0.01 xy.patch 3 4037.52 9.11 0.01 xy.acer.day 4 4038.20 9.79 0.00 xy.beech 3 4038.73 10.32 0.00 xy.nyssa 3 4038.84 10.43 0.00 null.day 3 4039.41 11.00 0.00 xy.oak.day 4 4039.92 11.51 0.00 xy.patch.day 4 4040.20 11.79 0.00 xy.global.day 11 4040.39 11.98 0.00 xy.tree.day 9 4041.32 12.91 0.00 xy.beech.day 4 4041.41 12.99 0.00 xy.nyssa.day 4 4041.52 13.11 0.00

Prionus laticollis Model K AICc Delta_AICc AICcWt pri.global 10 3304.18 0.00 0.60 pri.tree 8 3305.02 0.85 0.40 pri.patch 3 3338.32 34.15 0.00 pri.acer 3 3342.22 38.05 0.00 pri.nyssa 3 3361.96 57.79 0.00 pri.oak 3 3368.76 64.59 0.00 pri.beech 3 3371.14 66.96 0.00 pri.stress 3 3374.30 70.12 0.00 pri.sweetgum 3 3377.69 73.51 0.00 Null 2 3392.51 88.34 0.00 pri.tulip 3 3394.24 90.07 0.00 pri.global.day 11 6445.51 3141.34 0.00 pri.tree.day 9 6445.99 3141.82 0.00 pri.stress.day 4 6428.59 3124.42 0.00 pri.patch.day 4 6418.44 3114.27 0.00 pri.beech.day 4 6428.59 3124.42 0.00 pri.sweetgum.day 4 6428.59 3124.42 0.00 pri.tulip.day 4 6428.59 3124.42 0.00 pri.nyssa.day 4 6428.59 3124.42 0.00 pri.acer.day 4 6428.59 3124.42 0.00 pri.oak.day 4 6428.59 3124.42 0.00 null.day 3 6425.92 3121.74 0.00

Graphisurus fasciatus Model K AICc Delta_AICc AICcWt fas.global 10 1750.24 0.00 0.90 fas.global.day 11 1755.32 5.09 0.07 fas.patch 3 1757.78 7.55 0.02 fas.patch.day 4 1760.46 10.22 0.01 fas.tree 8 1762.04 11.81 0.00 fas.tree.day 9 1766.19 15.95 0.00 fas.tulip 3 1776.68 26.44 0.00 fas.tulip.day 4 1779.36 29.12 0.00 Null 2 1781.73 31.50 0.00 fas.nyssa 3 1782.13 31.89 0.00 fas.beech 3 1782.80 32.56 0.00 fas.stress 3 1784.09 33.85 0.00 fas.sweetgum 3 1784.20 33.96 0.00 fas.oak 3 1784.21 33.97 0.00 fas.acer 3 1784.21 33.97 0.00 null.day 3 1784.21 33.97 0.00 fas.nyssa.day 4 1784.81 34.57 0.00 fas.beech.day 4 1785.48 35.24 0.00 fas.stress.day 4 1786.77 36.53 0.00 fas.sweetgum.day 4 1786.88 36.64 0.00 fas.oak.day 4 1786.89 36.65 0.00 fas.acer.day 4 1786.89 36.65 0.00

Table 2.9 Models to predict Anelaphus villosus abundance. The model that incorporates percentages of all dominant tree species is the best (abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5 (pct_sweetgum)+ β6 (pct_beech)).

Anelaphus villosus Model K AICc Delta_AICc AICcWt vil.tree 8 1564.12 0.00 0.93 vil.global 10 1569.44 5.32 0.07 vil.acer 3 1659.51 95.40 0.00 vil.beech 3 1721.71 157.59 0.00 vil.sweetgum 3 1745.01 180.89 0.00 vil.tulip 3 1746.31 182.19 0.00 vil.stress 3 1797.86 233.74 0.00 vil.patch 3 1800.12 236.00 0.00 null 2 1806.65 242.53 0.00 vil.oak 3 1808.63 244.51 0.00 vil.nyssa 3 1808.74 244.62 0.00 vil.global.day 11 4326.03 2761.91 0.00 vil.tree.day 9 4316.54 2752.42 0.00 vil.stress.day 4 4299.14 2735.02 0.00 vil.patch.day 4 4298.96 2734.84 0.00 vil.beech.day 4 4299.14 2735.02 0.00 vil.sweetgum.day 4 4299.14 2735.02 0.00 vil.tulip.day 4 4299.14 2735.02 0.00 vil.nyssa.day 4 4299.14 2735.02 0.00 vil.acer.day 4 4299.14 2735.02 0.00 vil.oak.day 4 4299.14 2735.02 0.00 null.day 3 4296.46 2732.34 0.00

Table 2.10 Models to predict Cyrtophorus verrucosus abundance. The patch size model is the best (abundance= βo+β1 (patch_ha)).

Cyrtophorus verrucosus Model K AICc Delta_AICc AICcWt ver.patch 3 1274.79 0.00 0.99 ver.global 10 1284.76 9.98 0.01 ver.sweetgum 3 1298.34 23.55 0.00 ver.tree 8 1300.06 25.27 0.00 ver.tulip 3 1300.84 26.05 0.00 ver.beech 3 1304.64 29.85 0.00 ver.oak 3 1305.17 30.38 0.00 null 2 1306.13 31.34 0.00 ver.nyssa 3 1306.37 31.58 0.00 ver.stress 3 1306.56 31.77 0.00 ver.acer 3 1306.57 31.78 0.00 null.day 3 2042.92 768.13 0.00 ver.patch.day 4 2044.70 769.91 0.00 ver.acer.day 4 2045.60 770.81 0.00 ver.beech.day 4 2045.60 770.81 0.00 ver.tulip.day 4 2045.60 770.81 0.00 ver.nyssa.day 4 2045.60 770.81 0.00 ver.stress.day 4 2045.60 770.81 0.00 ver.oak.day 4 2045.60 770.81 0.00 ver.sweetgum.day 4 2045.60 770.81 0.00 ver.tree.day 9 2063.00 788.21 0.00 ver.global.day 11 2071.77 796.98 0.00

Table 2.11 Models to predict Euderces pini abundance. The percent tulip tree is the best model (abundance= βo+β1 (pct_tulip)).

Euderces pini Model K AICc Delta_AICc AICcWt pin.tulip 3 1293.89 0.00 0.96 pin.global 10 1301.18 7.30 0.03 pin.tree 8 1302.49 8.61 0.01 pin.nyssa 3 1319.42 25.53 0.00 pin.beech 3 1319.95 26.07 0.00 pin.stress 3 1320.50 26.62 0.00 pin.patch 3 1320.92 27.04 0.00 null 2 1321.02 27.14 0.00 pin.oak 3 1321.90 28.02 0.00 pin.sweetgum 3 1322.14 28.26 0.00 pin.acer 3 1322.63 28.75 0.00 null.day 3 2440.37 1146.49 0.00 pin.patch.day 4 2442.88 1149.00 0.00 pin.acer.day 4 2443.05 1149.16 0.00 pin.tulip.day 4 2443.05 1149.16 0.00 pin.stress.day 4 2443.05 1149.16 0.00 pin.nyssa.day 4 2443.05 1149.16 0.00 pin.beech.day 4 2443.05 1149.16 0.00 pin.sweetgum.day 4 2443.05 1149.16 0.00 pin.oak.day 4 2443.05 1149.16 0.00 pin.tree.day 9 2460.45 1166.56 0.00 pin.global.day 11 2469.95 1176.06 0.00

Table 2.12 Models to predict Neoclytus acuminatus abundance. The percent sweetgum model is the best (abundance= βo+β1 (pct_sweetgum)).

Neoclytus acuminatus Model K AICc Delta_AICc AICcWt ac.sweetgum 3 814.60 0.00 0.38 ac.oak 3 816.53 1.93 0.15 null 2 816.84 2.24 0.13 ac.patch 3 817.23 2.63 0.10 ac.acer 3 818.29 3.69 0.06 ac.tulip 3 818.37 3.78 0.06 ac.nyssa 3 818.80 4.20 0.05 ac.beech 3 819.24 4.65 0.04 ac.stress 3 819.26 4.66 0.04 ac.tree 8 825.68 11.08 0.00 ac.global 10 832.31 17.71 0.00 null.day 3 908.90 94.30 0.00 ac.patch.day 4 911.56 96.96 0.00 ac.acer.day 4 911.58 96.98 0.00 ac.nyssa.day 4 911.58 96.98 0.00 ac.beech.day 4 911.58 96.98 0.00 ac.stress.day 4 911.58 96.98 0.00 ac.tulip.day 4 911.58 96.98 0.00 ac.sweetgum.day 4 911.58 96.98 0.00 ac.oak.day 4 911.58 96.98 0.00 ac.tree.day 9 928.98 114.38 0.00 ac.global.day 11 938.63 124.03 0.00

Neoclytus mucronatus Model K AICc Delta_AICc AICcWt muc.global 10 695.79 0.00 0.91 muc.tree 8 702.34 6.55 0.03 muc.tulip 3 703.27 7.48 0.02 muc.beech 3 705.35 9.56 0.01 muc.sweetgum 3 705.43 9.64 0.01 muc.acer 3 705.52 9.73 0.01 muc.stress 3 705.69 9.90 0.01 muc.patch 3 707.98 12.19 0.00 null 2 708.18 12.39 0.00 muc.oak 3 708.34 12.55 0.00 muc.nyssa 3 710.32 14.53 0.00 null.day 3 1084.53 388.74 0.00 muc.patch.day 4 1087.20 391.41 0.00 muc.beech.day 4 1087.20 391.42 0.00 muc.tulip.day 4 1087.20 391.42 0.00 muc.acer.day 4 1087.20 391.42 0.00 muc.nyssa.day 4 1087.20 391.42 0.00 muc.stress.day 4 1087.20 391.42 0.00 muc.sweetgum.day 4 1087.20 391.42 0.00 muc.oak.day 4 1087.20 391.42 0.00 muc.tree.day 9 1104.60 408.82 0.00 muc.global.day 11 1114.26 418.48 0.00

Table 2.14 Beta (±SE) estimates for best models for the top ten species of cerambycids

captured in 2013.

-1.83 (1.35)

-0.50 (0.63)

-0.29 (0.43)

-0.48 (0.26)

0.15 (0.21)

Pct_Stressed

-0.01 (0.002)

0.007 (0.001)

-0.01 (0.001)

0.001 (0.001)

0.00 (0.001)

0.003 (0.001)

Patch Size

2.89 (0.87)

2.38 (0.55)

-1.03 (0.41)

0.86 (0.31)

-0.28 (0.18)

0.44 (0.16)

Pct_Beech

-0.67 (1.88)

-2.19 (1.13)

-7.37 (2.00)

-1.62 (0.60)

-0.41 (0.55)

-0.87 (0.28)

-0.82 (0.32)

0.17 (0.29)

Pct_Sweetgum

4.90 (1.44)

2.81 (0.51)

5.54 (0.89)

-3.13 (0.64)

0.44 (0.48)

-0.71 (0.28)

0.36 (0.24)

Pct_Tulip

3.57 (1.01)

3.47 (0.69)

-1.65 (0.44)

-1.45 (0.31)

-0.40 (0.19)

-0.59 (0.19)

Pct_Blackgum

1.80 (1.50)

-4.81 (1.14)

-1.92 ( 0.62)

-0.91 (0.49)

0.31 (0.22)

0.10 (0.20)

Pct_Maple

0.98 (2.12)

-1.47 (1.20)

-0.17 (0.87)

0.15 (0.69)

-1.47 (.47)

-3.73 (.51)

Pct_Oak

0.49 (0.91)

2.54 (0.47)

3.37 (0.23)

2.65 (0.53)

2.28 (0.56)

4.84 (0.39)

3.85 (0.29)

4.60 (0.07)

4.77 (0.12)

4.85 (0.12)

Intercept

Neoclytus mucronatus

Neoclytus acuminatus

Euderces pini

Cyrtophorus verrucosus

Anelaphus villosus

Graphisurus fasciatus

Prionus laticollis

Xylotrechus colonus

Megacyllene caryae

Phymatodes amoenus Species

2.3 Conclusion

The unmarked package that I used to analyze my data is not usually used in the insect world. I used a point count function which is usually used to analyze bird distributions. It is however a useful tool for this project because it uses repeated counts to study population trends in response to different factors (Fiske and Chandler 2011).

Chrysler Woods had significantly different cerambycid captures between the two traps placed in the fragment in 2013. The trap placed in the interior of the patch was less than 120 meters away from the trap placed at the edge of the disturbance, but it captured 216 fewer cerambycids, almost a 40% difference (Table 2.3). The trap at the edge of disturbance was surrounded by stressed trees, which probably impacted the high abundance of cerambycids, but it was also located in an area where it was downwind from a great area of flat land and upwind from the forest fragment. Most of my traps were located in the interior of forest fragments, so the pheromone spread is probably not very affected by wind because of the tree breaks. The clear-cut site was in perfect position for the wind to aid in the spread of the pheromone further into the forest, which also could have helped attract more cerambycids.

There was also a significant difference in the trap catches between the two traps in Christina Creek 1 and Ecology Woods (Table 2.3). The difference in the catch between the two traps at Christina Creek 1 was due to flooding, and not to forest covariates. CC1_E7 was located along a stream which flooded multiple times during the 2013 field season, washing out my catches for that week. Due to this, I cannot

47 draw any conclusions about the differences in catches between these two traps. The

Ecology Woods trap differences are probably due to the difference in tree diversity surrounding each trap. At the EW_E7 trap, there were 11 species of trees represented in the surrounding area, consisting mostly of oak, red maple, and pagoda dogwood

(Cornus alternifolia L.f.). The EW_C5 trap however only had 7 species of tree with the most abundant tree being red maple. Both traps had 21 trees in the surveyed area and were part of the same sized fragment, so diversity in trees was probably in the main factor in determining the number of cerambycids found in the two traps. The traps at Folk Park and Iron Hill 2 also were significantly different. Each of these sites had a trap placed on the edge of the site and a trap placed in the interior. The trap on the edge of Iron Hill 2 was along a road, and would therefore probably have more stress on the trees along the edge, so that may have helped lead to the greater number of cerambycids captured in that trap. The trap in Folk that was on the edge was on the edge of a field and there was low disturbance there, so that area is probably less attractive to cerambycids which is why we captured fewer beetles in that trap than the trap in the interior of the forest.

The traps at Laird were the only traps that did not show a significant difference between catches. Both traps at Laird were surrounded by black locusts (Robinia pseudoacacia L.). The similar composition of beetles captured between the traps at these sites helps to support my conclusion that tree species has a significant effect on cerambycid occupancy and abundance. It also shows that local factors are important to

48 cerambycid abundance and may show that the pheromone blend does not attract beetles over long distances.

While all three traps from White Clay Creek State Park were located within 1 km of each other in the same contiguous fragment, they had very different cerambycid compositions. White Clay 2 had the greatest number of beetles captured of all three traps with 364 cerambycids. Out of the 15 species represented from that trap though, over 75% of the species were either M. caryae or P. amoenus. M. caryae is extremely attracted to citral (Mitchell et al. 2012), which was a new component in my pheromone blend in 2013, and appeared at every site in large numbers early in the season. P. amoenus on the other hand was captured in the largest numbers at White

Clay 2, but it was also captured in great numbers at Coverdale, Laird_B4, and

Chrysler Woods_C3. The best AIC model for Phymatodes amoenus was the global model (Table 2.4), so its abundance was influenced by tree species, tree health, and patch size. All of these sites range from 16% to 42% red maple and 0% to 35% tulip tree, which according to my AIC results have large affect on P. amoenus occupancy.

White Clay 2 consisted of over 75% of red maple and tulip tree. All of these sites are also invaded with non-native plants and most of my traps are close to edges that are along human development. Phymatodes amoenus is known to be a grape vine borer

(Vitis spp.) (Lingafelter 2007), and grape thrives under these conditions. Therefore sites that are highly disturbed probably have a greater number of stressed trees and a greater amount of grape, so sites with a lot of stressed trees would support more P. amoenus.

My trap at White Clay 3 caught a comparable number of cerambycids to my trap at White Clay 1. The only difference in species composition between each of these traps was that the White Clay 3 trap captured more than double the number of P. amoenus that were captured in White Clay 1, but White Clay 1 captured more than double the number of P. laticollis that were captured in White Clay 3. If those species are not included, White Clay 1 captured 152 individuals and White Clay 3 captured

153 species, so the difference between these sites may be due to the species of trees that are attractive to P. amoenus and P. laticollis. White Clay 3 is over 50% tulip tree and red maple, which is attractive to P. amoenus, and White Clay 1 is over 75% oak,

American beech, and blackgum which are the most attractive tree species to P. laticollis according to my study and Lingafelter (2007). The results from this single forest fragment show that tree species in localized areas may be influential in cerambycid abundance in these fragments.

The three traps in my urban and suburban (Wilmington 1, Wilmington 2, and

Hough) fragments captured few cerambycids compared to the traps in my other sites

(Table 2.2). All three of these sites are extremely isolated by development and infrastructure, so emigration to these sites is most likely difficult for cerambycid species. I see similar trends with the traps in Ecology Woods, which has been isolated by agriculture for over 100 years. Wilmington 1 is in the area inside the middle of an exit ramp off of a major highway. It is composed of primarily sweet gum, but has nine other tree species as well. Sweet gum prefers areas that are somewhat swampy, it grows quickly, and is very intolerant of shade (Taber 2012). For these reasons, it is an

50 early successional tree and exists in young forests in areas that have been recently disturbed. Therefore it makes sense that few cerambycids would be found at

Wilmington 1 because of how young and isolated it is.

Wilmington 2 has been established for over 77 years

(http://demac.udel.edu/data/aerial-photography) so there is high tree diversity and high average DBH (>30cm). It is however very isolated, which may be the cause of the low number of cerambycids. It did have 50% of all of the Neoclytus caprea that I captured, and is one of the only sites that had any ash species (Fraxinus americana L.), which, according to Lingafelter (2007), is the main host plant for N. caprea. Hough is also very isolated. It is located in a suburban community and is only 4.55 hectares. Because of its suburban location and the use of non-native plants in a lot of landscaping, it is dominated by an ornamental cherry tree species (40%) (Reichard and White 2001).

This species most likely does not have any cerambycid pests because it is not native to the area and therefore may not be a good host for cerambycids.

The trap at Devon Park, along the railroad tracks, collected more cerambycids than my Wilmington site on the port. Over 300 beetles were captured in site, which is the seventh greatest number out of all 30 sites (Table 2.2). It was the only site that was dominated by large oak trees (>25% with an average DBH of 45.4cm), which shows that it has been established for a long time because oaks are slow growing, long-lived trees that usually live in climax forests (Taber 2012). It is located close to other fragments, so emigration to this patch would be more likely for cerambycids, and it

51 has been established for over 77 years (http://demac.udel.edu/data/aerial-photography) so there has been time for populations to establish.

Based on the best models for each of my species, the global model

(abundance= βo+β1 (pct_oak)+ β2 (pct_acer)+ β3 (pct_nyssa)+ β4 (pct_tulip)+ β5

(pct_sweetgum)+ β6 (pct_beech)+ β7 (patch_ha)+ β8 (pct_stressed/dead)) was the best predictor of abundance for Phymatodes amoenus, Megacyllene caryae, Prionus laticollis, Graphisurus fasciatus, and Neoclytus mucronatus (Tables 2.4, 2.5, 2.7, 2.8, and 2.13). As discussed before, P. amoenus is a grape vine borer, so all of the factors that went into my global model affect species of animals in forests, but they also affect species of plants in forests. The majority of sites where P. amoenus was found in the largest numbers were young sites (<50 years old). It is not uncommon to find large numbers of grapevines growing in young stands of regeneration where there is ample light coming through the canopy, which is why there are large amounts of grape in my younger sites (Pennsylvania Bureau of Forestry). Therefore P. amoenus was found in the greatest numbers in the young, disturbed fragments.

The best model for Megacyllene caryae was also the global model (Table 2.5).

I found M. caryae in all of my traps. According to my beta estimates, it only had a positive association with maple trees. Only Wilmington 1 had fewer than 20 of this species, and only Christina Creek E7 and Iron Hill 2 E7 had no red maple within an

11.3 radius of the trap. Since the tree species that M. caryae is associated with is located in almost all of my forest fragments, it makes sense that I found M. caryae in all of my traps. Graphisurus fasciatus was another abundant species found in my traps

52 with the global model as the best model for predicting its abundance (Table 2.8).

According to Lingafelter (2007) the main host for G. fasciatus is pine (Pinus spp.).

There are no pine trees in the areas surrounding any of my traps and there are few pine trees in the fragments as a whole and all of the beta estimates for this species were negative. Based on these observations and the global model being the best model, pine is probably not the only host tree, and there are probably more forest factors that influence the abundance of this species.

The other species’ abundance that were best modeled by the global model were

Prionus laticollis and Neoclytus mucronatus (Tables 2.7 and 2.13). Both of these species were found in larger sites that had been established for over 70 years. Oak and hickory are the main hosts of these species (Lingafelter 2007). Therefore they most likely prefer older, larger fragments that have a higher percentage of oak, beech and hickory, which are climax tree species (Snyder 2006). Prionus laticollis also had a strong association with tree species (Table 2.7). According to the beta values, if has positive associations with oak and beech and negative associations maple and sweetgum (Table 2.14). Older sites also tend to have a greater amount of dead trees and coarse woody debris which also support more cerambycids Dodds et al 2010).

Xylotrechus colonus and Neoclytus acuminatus abundance was best modeled by percent sweetgum (Tables 2.6 and 2.12). Each of these cerambycid species has a broad range of hosts, but they consist mainly of climax tree species (oaks, hickory, and ash) (Lingafelter 2007). The host trees usually occur in older forest stands because they are shade tolerant and live a long time (Snyder 2006). Sweetgum does not usually

53 occur in these forest situations because it is fast growing and usually just exists in early successional situations. Therefore sites with a large percentage of sweetgum probably do not contain a large percentage of the tree species that these cerambycid species select as their hosts and therefore have a significant negative effect on abundance of X. colonus and N. acuminatus. Neoclytus acuminatus was also positively associated with oak (Table 2.14).

Euderces pini and Anelaphus villosus abundance was also best modeled by tree species. Euderces pini was affected by percent tulip tree (Table 2.11). This species, like Xylotrechus colonus and Neoclytus acuminatus, is thought to prefer mainly climax trees like hickories and beech as their hosts (Lingafelter 2007). The sites that this species was captured in the greatest numbers only had a few tulip trees, but they were all very large (average basal area= 422.45 cm2). Tulip trees are typically successional trees, but they can persist in climax forests because they grow very tall and can still get sunlight even after the canopy has fill in.

Anelaphus villosus abundance was affected by all of the dominant tree species as a whole (β= 2.280, SE= 0.562). Larvae of this species are live twig pruners of most eastern hardwoods and shrubs (Lingafelter 2007). It occurred in the highest numbers at my Glasgow 1 site, which contained the largest number of trees of any of my sites (62 stems). Therefore I conclude that this cerambycid species is a generalist and will be in places where there are a lot of a lot of trees available.

The only species that had patch size as the best model for predicting its abundance was Cyrtophorus verrucosus (β= 2.649, SE= 0.533). It was captured in the

54 largest number in the fragments that were greater than 150 hectares. More forested area allows for more trees, more dead trees, and more cover from predators, so it is reasonable that larger fragments will have positive effects on C. verrucosus abundance.

Based on many past studies on woodboring beetles, I would have predicted that tree stress would have been a significant predictor for most of the species of cerambycids that I captured (Dodds et al. 2010; Priewasser 2012); however, it was not the best model for any of the top ten cerambycid species I captured. My evaluation of tree health was based on visual characteristics, so trees that were recently stressed (as in my Chrysler Woods clear-cut site) may have not been showing any signs yet, although they could have been producing stress volatiles that attracted the large number of cerambycids. In order to fully evaluate tree stress impact on cerambycid abundance, chemical evaluations will have to be conducted.

In conclusion, from my analyses thus far, most cerambycid species are influenced by a large number of forest factors. A few species have stronger relationships with certain trees or size of fragment, but there are other influences in those relationships. Tree stress also appears to have an impact on cerambycid abundance, but was not shown in my results due my data collection methods.

REFERENCES

Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic Control 19(6), 716–723, doi:10.1109/TAC.1974.1100705, MR 0423716.

Balog, A., Marko, V. and Adam, L. (2008). Rove beetles (Coleoptera: Staphylinidae) collected during the long term ecological research in Hungarian oak forest. Journal of Environmental Biology, 29(March), 263–266.

Barbour, J. D., Millar, J. G., Rodstein, J., Ray, A. M., Alston, D. G., Rejzek, M., and Hanks, L. M. (2011). Synthetic 3,5-Dimethyldodecanoic acid serves as a general attractant for multiple species of Prionus (Coleoptera: Cerambycidae). Annals of the Entomological Society of America, 104(3), 588–593. doi:10.1603/AN10182

Chenier, J.V.R., and B.J.R. Philogene. (1989). Field responses of certain forest Coleoptera to conifer monoterpenes and ethanol. Journal of Chemical Ecology, 15(6), 1729-1745.

Dodds, K.J., G.D. Dubois, and E.R. Hoebeke. (2010). Trap type, lure placement, and habitat effects on Cerambycidae and Scolytinae (Coleoptera) catches in the Northeastern United States. Journal of Economic Entomology, 103(3), 698-707.

Dodds, K. J., and D. A., Orwig. (2011). An invasive urban forest pest invades natural environments — Asian longhorned beetle in northeastern US hardwood forests. Canadian Journal of Forest Research. 41(9): 1729–1742. doi:10.1139/X11-097.

Fiske, I. J. and R. B. Chandler. (2011). unmarked : An R Package for Fitting Hierarchical Models of Wildlife Occurrence and Abundance. Journal of Statistical Software, 43(10). http://www.jstatsoft.org/v43/i10/.

Fiske, I., Chandler, R., Miller, D., Royle, A., and Kery, M. (2014). Unmarked: Models for Data from Unmarked Animals. R package version 0.10-3. (http://cran.at.r- project.org/web/packages/unmarked/unmarked.pdf).

Ginzel, M. D., and L. M. Hanks. (2005). Role of host plant volatiles in mate location for three species of longhorned beetles. Journal of Chemical Ecology, 31(1), 213–217. doi:10.1007/s10886-005-6735-6

Hanks, L. M. and J. G. Millar. 2013. Field bioassays of cerambycid pheromones reveal widespread parsimony of pheromone structures, enhancement by host plant volatiles, and antagonism by components from heterospecifics. Chemoecology, 23, 21-44.

Hanks, L. M., J. G. Millar, J. A. Mongold-Diers, J.C.H. Wong, L. R. Meier, P. F. Reagel, and R. F. Mitchell. 2012. Using blends of cerambycid beetle pheromones and host plant volatiles to simultaneously attract a diversity of cerambycid species. Canadian Journal of Forest Research. 42, 1050-1059.

Hodkinson, T.R., J.A.N. Parnell, eds. Reconstructing the Tree of Life: Taxonomy and Systematics of Species Rich Taxa. Boca Raton: CRC Press, 2007.

Keena, M.A., and P.M. Moore. (2010). Effects of temperature on Anoplophora glabripennis (Coleoptera: Cerambycidae) larvae and pupae. Environmental Entomology, 39(4), 1323–35. doi:10.1603/EN09369.

Kennedy, A.A., and D.G. Mccullough. (2002). Phenology of the Larger European Pine Shoot Beetle Tomicus piniperda (L .) (Coleoptera: Scolytidae) in Relation to Native Bark Beetles and Natural Enemies in Pine Stands. Environmental Entomology, 31(2), 261–272.

Kimmerer, T. W. and T. T. Kozlowski. (1982). Ethylene, Ethane, Acetaldehyde, and Ethanol Production by Plants under Stress. Plant Physiology, 69(4), 840–7.

Kobakashi, F., Yamane, A., and Ikeda, T. (1984). The Japanese pine sawyer beetle as the vector of pine wilt disease. Annual Review of Entomology, 29, 115–135.

Lingafelter, S.. 2007. Illustrated Key to the Longhorned Woodboring Beetles of the Eastern United States. The Coleopterists Society, special publication #3. North Potomac, MD. 206 pp.

Linsley, E. (1959). Ecology of cerambycidae. Annual Review of Entomology. http://www.annualreviews.org/doi/pdf/10.1146/annurev.en.04.010159.000531

Majka, C. G. and P. J. Johnson. (2008). The Elateridae (Coleoptera) of the Maritime Provinces of Canada: faunal composition, new records, and taxonomic changes. Zootaxa, 33, 1–2.

Milberg, P., K.O. Bergman, H. Johansson, and N. Jansson. (2014). Low host-tree preferences among saproxylic beetles: a comparison of four deciduous species. Insect Conservation and Diversity, doi:10.1111/icad.12074

Mitchell, R. F., Hughes, D. T., Luetje, C. W., Millar, J. G., Soriano-Agatón, F., Hanks, L. M., & Robertson, H. M. (2012). Sequencing and characterizing odorant receptors of the cerambycid beetle Megacyllene caryae. Insect biochemistry and molecular biology, 42(7), 499–505. doi:10.1016/j.ibmb.2012.03.007

Nardi, G. and P. Zahradník. (2004). Bostrichidae and Anobiidae (Coleoptera). Invertebrati di una foresta della Pianura Padana, Bosco della Fontana. Secondo contributo. Conservazione Habitat Invertebrati 2, 125-139.

Oliver, J.B. and C.M. Mannion. (2001). Ambrosia beetle (Coleoptera: Scolytidae) species attacking chestnut and captured in ethanol-baited traps in middle Tennessee. Environmental Entomology, 30(5), 909-918.

Priewasser, K., P. Brang, H. Bachofen, H. Bugmann, and T. Wohlgemuth. (2012). Impacts of salvage-logging on the status of deadwood after windthrow in Swiss forests. European Journal of Forest Research, 132(2), 231–240. doi:10.1007/s10342-012-0670-1

R Core Team. (2012). R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/

Reding, M.E., J.B. Oliver, P.B. Schultz, C.M. Ranger, and N.N. Youssef. (2013). Ethanol injection of ornamental trees facilitates testing insecticide efficacy against ambrosia beetles (Coleoptera: Curculionidae: Scolytinae). Journal of Economic Entomology, 106(1), 289-298.

Rega, C. (2012). Impacts of soil calcium availability and non-native plant invasions on an urban forest bird community. Master’s thesis University of Delaware, Newark.

Reichard, S. H. and P. White. (2001). Horticulture as a pathway of invasive plant introductions in the United States. BioScience, 51(2), 103-113.

Reut, M., Cowell, B., and Pszczolkowski, M. A. (2010). Traps baited with isopropanol attract the American Carrion Beetle, Necrophila americana (L.) (Coleoptera: Silphidae). Coleopterists Bulletin, 64(3), 230–234. doi:10.1649/0010-065X- 64.3.230.8.

Rodríguez-del-bosque, A. L. A. (2013). Feeding and survival of Oncideres pustulata ( Coleoptera : Cerambycidae ) adults on Acacia farnesiana and Leucaena leucocephala (Fabaceae). Southwestern Entomologist, 38(3), 487–498.

Schrey, N. M., Reeve, J. D., and Anderson, F. E. (2005). Mitochondrial DNA analysis of the bark beetle predator Thanasimus dubius F. (Coleoptera: Cleridae) reveals regional genetic differentiation. Molecular Ecology, 14(11), 3317–24. doi:10.1111/j.1365-294X.2005.02684.x

Schowalter, T. D. (2012). Ecology and Management of Bark Beetles (Coleoptera: Curculionidae: Scolytinae) in Southern Pine Forests. Journal of Integrated Pest Management, 3(2), 1–7. doi:10.1603/IPM11025.

Snyder, M. (2006) What is a climax forest? Northern Woodlands, 50, 19.

Swift, I. P. and A. M. Ray. (2010). Nomenclatural changes in North American Phymatodes Mulsant. Zootaxa, 52, 35–52.

Taber, W.. 2012. Delaware Trees: A Guide to the Identification of the Native Tree Species. Dover: The Delaware Department of Agriculture.

Webster, R. P., Sweeney, J. D., and Demerchant, I. (2012). New Coleoptera records from New Brunswick, Canada: Eucnemidae. ZooKeys, 91(179), 77–91. doi:10.3897/zookeys.179.2492

Yanega, D. 1996. Field Guide to Northeastern Longhorned Beetles (Coleoptera: Cerambycidae). Illinois Natural History Survey, Champaign.