<<

Conserving Ash (Fraxinus) Populations and Genetic Variation in Forests Invaded

by Using Large-scale Insecticide Applications

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Erin Margaret O’Brien

Graduate Program in Entomology

The Ohio State University

2017

Dissertation Committee:

Daniel A. Herms, Advisor

Mary M. Gardiner

P. Charles Goebel

Andrew Michel

Copyrighted by

Erin Margaret O’Brien

2017

Abstract

Emerald ash borer (EAB) has drastically affected the population demographics of ash species in North America, having killed hundreds of millions of since its accidental introduction more than 20 years ago. Near the epicenter of this invasion

(southeastern Michigan), nearly all mature ashes have died, reproduction has ceased, and the seed bank depleted, leaving only an “orphaned” cohort of established seedlings and saplings too small to be susceptible to EAB. Because of the very high mortality, it is possible that populations of seedlings that established recently may have lower genetic variation than those that established before the EAB invasion. Insecticides, such as emamectin benzoate, can successfully protect ash trees from EAB attack, and clusters of treated ash trees may slow the progression of ash mortality by reducing the surrounding density of EAB. Therefore, it may be possible to use insecticides to protect treated trees and provide associational protection to untreated trees. If high enough densities of trees can be protected, then ash reproduction, regeneration, and genetic variation may be conserved.

In 2011, Five Rivers Metroparks (FRMP) in and near Dayton, Ohio began a large- scale insecticide management program to treat ash trees with emamectin benzoate

(-äge, Arborjet, Inc.) insecticide every two years. The objectives of this dissertation were to quantify the effects of density of this treatment program on (1) direct protection

ii of treated trees, (2) associational protection of untreated trees, (3) ash reproduction and regeneration, and (4) conservation of genetic variation by comparing genetic variation of ash seedling populations near the epicenter of the EAB invasion in southeastern Michigan and FRMP.

Six parks were sampled in Ohio that varied in the degree to which they have been impacted by EAB; Cox Arboretum, Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek MetroParks, and five parks were sampled in the Upper Huron River

Watershed in Michigan for genetic analysis only; Highland, Island Lake, and Proud Lake

State Recreation Areas, as well as Mills and Indian Springs Metroparks. From

2014 – 2016, survival was higher and there were fewer signs of EAB infestation on treated green, white, black, and pumpkin ash trees than untreated conspecifics. However, survival of blue ash was high for both treated and untreated individuals, few of which showed signs of EAB infestation. The extent of green and white ash mortality differed between parks, with survival higher at Sugarcreek, Englewood, and Germantown

Metroparks (low EAB impact) than at Cox Arboretum, Taylorsville, and Twin Creek

(high EAB impact).

The total amount of ash phloem area in a stand provides a better measure of EAB carrying capacity than ash density, as phloem area is a function of tree size as well as density. Hence, the percentage of ash phloem area treated with insecticide provides a better estimate of the potential impact of the treatment on the EAB population. From

2014 – 2016, I found that survival of untreated green-white ashes increased with percentage of ash phloem area treated, but only in parks with low EAB impact.

iii

Additionally, survival of untreated trees was higher when the nearest treated ash was within 100 m, percentage of ash phloem treated was high, and EAB impact was low.

However, this pattern was not observed in parks with high EAB impact. These results suggest that treating ash trees with insecticide may slow the progression of ash mortality if the program is initiated when ash mortality is still low. Further testing is required to determine if treating at a higher density than treating in this study can conserve ash when

EAB impact is already high.

There were more flowering green and white ash trees (treated and untreated) in plots with higher percentage ash phloem treated. Density of flowering green and white ash that were untreated was higher in low impact than high impact parks. Although blue ash was less abundant than green and white ash in all parks, there was no difference in the density of flowering blue ash in parks with low and high impact. In parks with high

EAB impact, seedling density was low and was not affected by insecticide treatment. In parks with low EAB impact, seedling densities increased with percentage of ash phloem treated. Density of blue ash and green or white ash seedlings increased with density of seed bearing ash and decreased with height of herbaceous understory vegetation, which may suppress seedling population through competition for resources.

Patterns of genetic variation of seedling and sapling populations differed in

Michigan and Ohio. In Michigan, established seedlings and saplings had similar allelic richness, but seedlings had fewer than expected heterozygotes. There was also less genotypic variation in the larger seedlings and saplings than in smaller seedlings that probably established more recently. In Ohio, newly germinated and small established

iv seedlings had higher allelic richness than larger established seedlings, but each population had similar number of effective alleles (high frequency alleles). Larger seedlings had the least genotypic variation and newly established seedlings had the most.

Collectively, these results are consistent with a loss of genetic variation in Michigan, but no loss of genetic variation in Ohio. EAB-induced ash mortality and density of treated ash trees also had no effect on genetic variation in populations of ash seedlings, but the impact of the EAB invasion may intensify as ash mortality increases over time.

v

Dedication

I dedicate this dissertation to my parents, who taught me the value of hard work and

perseverance in order to achieve my dreams.

In memory of my mother, Leslie O’Brien, who introduced me to the beauty of nature, taught me how to be a strong and intelligent woman, and inspired me to enjoy all of life’s

moments.

vi

Acknowledgments

I would first and foremost like to thank my advisor, Daniel Herms, for his support and guidance through my journey of becoming a scientist. He has been a great teacher and mentor throughout my program and I look forward to continuing this relationship on the next chapter of my journey. I also thank my committee: Mary Gardiner, Charles Goebel, and Andy Michel for their advice and guidance. Much of my field and laboratory work could not have been completed without the help of the 2014 to 2017 Herms lab: Diane

Hartzler, Wendy Klooster, Laurel Haavik, Amilcar Vargas Loyo, Christian Bonilla, Cali

Granger, Andrea Wade, Nicole Sharpe, Megan Zerrer, Kara Taylor, Jalyn Devereaux,

Meredith Mann, and Adriana Alfaro Inocente. Research support provided by state and federal funds appropriated by The Ohio State University, Ohio Agricultural Research and

Development Center (Graduate Student SEEDS award).

I would also like to thank my fellow entomology graduate students for your support and friendship and to Lori Jones for being a calming voice during the stressful times of school. I thank my family for their unending support and encouragement. To my cat May who was constantly determined to sit on my keyboard. Lastly, I thank my future wife and partner in life, Tammy Bundy. You have kept me well fed and well rested, you lifted my spirits and made me laugh, and you are my best friend. I love you and I look forward to starting the next chapter of our lives together.

vii

Vita

2002...... B.S. Biology, St. Michael’s College

2010...... M.S. Biology, Eastern Kentucky University

2013 to 2016 ...... Graduate Research Associate, Department

of Entomology, The Ohio State University

2016 to 2017 ...... Graduate Teaching Associate, Center for

Life Sciences Education, The Ohio State

University

Publications

O’Brien, E.M., and G. Ritchison, 2011. Non-breeding ecology of Loggerhead Shrikes in

Kentucky. Wilson Journal of Ornithology 123:360-366.

Fields of Study

Major Field: Entomology

viii

Table of Contents

Abstract ...... ii Dedication ...... vi Acknowledgments...... vii Vita ...... viii Publications ...... viii Fields of Study ...... viii Table of Contents ...... ix List of Tables ...... xi List of Figures ...... xiv Chapter 1: Literature Review ...... 1 Natural Forest Dynamics ...... 1 Invasive Insects and Forest Dynamics ...... 3 The Emerald Ash Borer Invasion ...... 4 Insecticide Management for EAB ...... 8 Impact of EAB on Ash Reproduction ...... 11 Impacts of EAB on Genetic Variation ...... 13 Chapter 2: Effects of emerald ash borer and emamectin benzoate insecticide applications on populations of five species of ash in forests of southwestern Ohio ...... 26 Abstract ...... 26 Introduction ...... 27 Methods...... 30 Results ...... 35 Discussion ...... 51 Literature cited ...... 55

ix

Chapter 3: Relationship between survival of untreated green and white ash and local abundance of insecticide treated ash trees: Treated trees provide associational protection to untreated ash trees ...... 58 Abstract ...... 58 Introduction ...... 59 Methods...... 62 Results ...... 66 Discussion ...... 77 Literature cited ...... 83 Chapter 4: Preserving future ash populations: Using insecticides to maintain ash reproduction and recruitment in the understory ...... 86 Abstract ...... 86 Introduction ...... 87 Materials and Methods ...... 90 Results ...... 95 Discussion ...... 109 Literature cited ...... 114 Chapter 5: The effect of widespread ash mortality and density of treated ash trees on conserving genetic variation in ash population ...... 117 Abstract ...... 117 Introduction ...... 118 Methods...... 124 Results ...... 130 Discussion ...... 152 Literature cited ...... 158 Chapter 6: Summary, conclusion and future research ...... 162 Conclusions and future direction ...... 167 Literature cited ...... 170 Literature Cited ...... 172

x

List of Tables

Table 2.1: Summary of tree communities and ash importance within each park surveyed in 2014 within the Five Rivers MetroParks in Dayton, OH. Data is shown as mean ± standard error...... 38 Table 2.2: Ordination of forest communities and environmental variables using non- metric multidimensional scaling (NMDS) analysis. Samples collected from forest communities at Five Rivers MetroParks in Dayton, OH from June to August 2016. NDMS statistical summary: R2=0.96, Stress = 0.195, P<0.0001...... 39 Table 2.3: Ash species abundance and size of untreated and treated ash trees sampled in 2014 from six parks within the Five Rivers MetroParks in Dayton, OH. Data represent the total trees sampled in this study...... 43 Table 2.4: Abundance and size of blue ash and green-white ash trees (treated and untreated) in 2014 at six parks from the Five Rivers MetroParks in Dayton, OH. Data for green and white ash trees were combined due to similar canopy health and survival...... 48 Table 3. 1: Comparisons of stand density of ash and treated ash trees from six parks within the Five Rivers Metroparks near Dayton, Ohio. Data collected from 2014 to 2016. Data is shown as mean ± SE for each park. 68 Table 3.2: Summary of the density and total phloem area of treated ash trees within parks at the Five Rivers Metroparks near Dayton, OH. Parks were categorized as low or high EAB impact based on percent ash survival at the beginning of this study (2014). All treated ash trees within 400 m radius of each plot (50 ha area) were included. Summary of metrics calculated to assess the local abundance of treated ash trees. Percentage of ash phloem area treated was tested to determine the significance of this variable as a predictor for percentage survival of untreated ash trees...... 69 Table 3.3: Statistical summary of the best fit model for assessing the relationship of ash survival in 2014, 2015, and 2016 using GLM with binomial distribution (logit link). Model: Survival ~ EAB impact * % phloem...... 72 Table 3.5: Statistical summary of the best fit model predicting percentage survival of green-white ash trees in 2015. Model was calculated using GLM with binomial distribution (logit link). Model: Survival ~ Distance * EAB impact + % phloem + year, pseudo-R2 = 0.31, Goodness of fit: P=0.12; dispersion = 1.4...... 75 Table 4. 1: Descriptions of ash demography at Five Rivers MetroParks in Dayton, OH. .93 Table 4.2: Density (ash ha-1) and percentage inflorescence of treated and untreated blue ash from forests at Five Rivers Metroparks (FRMP) in southwestern, OH, according to percentage ash phloem area treated and EAB impact. Parks within FRMP were characterized as low or high EAB impact based on the percentage survival in 2014; low EAB impact indicates parks with ≥ 75% survival and high EAB impact parks with ≤ 25% xi survival. Phloem area of each treated ash tree within 200 m (12.5 ha) of each plot was estimated to calculate percentage ash phloem treated within 12.5 ha of each plot, then plots were grouped by mean percentage ash phloem (0, 1, 2.5, or 4.5%). Statistical significance was calculated using generalized linear mixed models with negative binomial distribution; P < 0.1 “.”; P < 0.05 “*”...... 98 Table 4.3: Density (ash ha-1) and percentage inflorescence of treated and untreated green- white ash from forests at Five Rivers Metroparks (FRMP) in southwestern, OH, according to percentage ash phloem area treated and EAB impact. See table 4.2 for description of percentage ash phloem and EAB impact. Statistical significance was calculated using generalized linear mixed models with negative binomial distribution; P < 0.0001 “***”...... 99 Table 4.4: Seedling densities (ash ha-1) in forests at Five Rivers MetroParks in southwestern Ohio, according to EAB impact (based on percentage survival in 2014) and percentage ash phloem area treated within 200 m (12.5 ha) of each plot. Statistical analyses were from generalized linear model with negative binomial distribution of the relationship between EAB impact and % phloem area treated...... 103 Table 4.5: Summary of statistical output for generalized linear models to quantify the effects of percentage ash phloem treated, EAB impact, and vegetation height on density of first-year seedlings from 2014 to 2016 at Five Rivers Metroparks in southwestern Ohio...... 105 Table 4. 6: Summary of statistical output for generalized linear models to quantify the effects of seed bearing treated and untreated ash and habitat characteristics on density of conspecific seedlings in 2016 at Five Rivers Metroparks in southwestern Ohio. Flowering blue ash and green-white ash with female flowers and/or seeds in 2015 were considered seed bearing, since they have the potential of producing seeds if pollination occurs. ... 107 Table 5.1: Population genetics of green ash seedlings and saplings from five state and metroparks in the Upper Huron Watershed in southeastern Michigan. Samples were collected in 2014, 12 years after emerald ash borer (EAB) was first detected in this area. Seedlings were divided into size classes: class 2 (stems ≤ 0.25 m tall) and class 3 (stems > 0.25 m, < 1.5 m tall); saplings were > 1.5 m tall, but unsusceptible to EAB (< 2.54 cm trunk diameter)...... 134 Table 5.2: Population genetics of green ash from five state and metroparks in the Upper Huron Watershed in southeastern Michigan. Samples were collected from seedlings and saplings in 2014 from plots established by Smith (2006)...... 136 Table 5.3: Population genetics of ash seedlings from the Five Rivers MetroParks (FRMP) in southwestern Ohio. Samples were collected in 2014, 7 years after emerald ash borer (EAB) was first detected in this area and 4 years after FRMP began an insecticide program to protect mature ash trees. Seedlings were divided into three size classes: class 1 (stems ≤ 0.25with cotyledons), class 2 (stems ≤ 0.25 m tall without cotyledons) and class 3 (stems > 0.25 m, < 1.5 m tall)...... 141

xii

Table 5.4: Population genetics of green ash from six parks within the Five Rivers Metroparks (FRMP) in southwestern Ohio. Samples were collected from green ash seedlings in 2014...... 143 Table 5.5: Population genetics of green ash seedlings and saplings from the Upper Huron Watershed in southeast Michigan and seedlings Five Rivers Metroparks (FRMP) in southwestern Ohio. Samples were collected in 2014 and analyzed with five microsatellite loci...... 149

xiii

List of Figures

Figure 2.1: Non-metric multidimensional scaling analysis of the tree communities and the corresponding environmental variables within natural forests at Five Rivers MetroParks in Dayton, OH. Species data was collected in 2014. Vegetation height and cover, slope, and aspect were collected in 2016. (A) Community characteristics with environmental variables. (B) Tree species with environmental variables. See Table 2.2 for descriptions of environmental variables (vectors). NDMS statistical summary: R2=0.96, Stress = 0.195, P<0.0001...... 40 Figure 2.2: Canopy decline and survival of untreated ash (left) and treated ash (right) at the Five Rivers MetroParks in Dayton, OH. (A-B) Show average canopy decline (mean ± SE) from 1 to 5; 1 indicated healthy ash, 2-4 indicated varying degrees of canopy decline, and 5 indicated dead ash. (C-D) Show percent survival of ash (mean ± SE)...... 44 Figure 2.3: Ash canopy ratings for untreated blue ash (A), treated blue ash (B), untreated green-white ash (C), and treated green-white ash (D) from six parks at Five Rivers MetroParks in Dayton, OH. CA: Cox Arboretum, EN: Englewood, GT: Germantown, SC: Sugarcreek, TA: Taylorsville, and TC: Twin Creek Metroparks. Ratings were from 1-5; 1 for healthy canopies, 2-4 for varying degrees of canopy openness, and 5 for dead ash. Data shown as mean canopy rating ± SE per plot...... 49 Figure 2.4: Percent ash survival for untreated blue ash (A), treated blue ash (B), untreated green-white ash (C), and treated green-white ash (D) from six parks at Five Rivers MetroParks in Dayton, OH. CA: Cox Arboretum, EN: Englewood, GT: Germantown, SC: Sugarcreek, TA: Taylorsville, and TC: Twin Creek Metroparks. Data shown as mean percent survival ± SE per plot...... 50 Figure 3.1: (A) Quadrat design with 4 replicated plots (18-m radius) and (B) Transects established within each quadrat to quantify the effect of local abundance of treated ash trees on survival of untreated green-white ash trees...... 64 Figure 3.2: Percent survival of untreated green-white ash trees as a function of percent treated ash phloem during 2014 (A), 2015 (B), and 2016 (C). Black circles represent low EAB impact plots and white triangles represent high EAB impact plots. See Table 3.3 for statistical output...... 73 Figure 3.3: Relationship between percent survival of untreated green-white ash trees and distance to the nearest treated tree in plots from parks with low EAB impact (A & C) and high EAB impact (B & D). Green-white ash survival was collected in 2015 (Top) and 2016 (Bottom). Data was analyzed using GLM with binomial distribution (logit link) and solid lines indicate the significant model: Survival = Distance*EAB impact + % phloem area treated; dotted lines indicate model ± 1.5% phloem area treated. See Table 3.4 for statistical output...... 76

xiv

Figure 4.1: Quadrat and plot layout for each study site. (A) Distribution of 4 replicated plots per quadrat. (B) 18-m radius plot subdivided into 4 microplots (4m2) and a subplot (8 m radius) used to sample seedlings (microplot), saplings (subplot), and immature trees (subplot). Mature trees were quantified within the plot boundary (18 m radius)...... 93 Figure 4.2: Generalized linear model (GLM) to estimate the effect of percentage ash phloem area treated, EAB impact, and vegetation height on density of first-year seedlings (class 1) in (A) 2014, (B) 2015, and (C) 2016 at Five Rivers Metroparks in southwestern Ohio. Model equations: class 1 seedlings ~ % phloem treated * EAB impact + vegetation height (cm). See Table 4.5 for summary of statistical output. Solid black line indicated predicted model for low EAB impact; solid gray line indicated predicted model for high EAB impact; and dashed lines indicated the effect of +/- standard deviation of vegetation height (cm) for low EAB impact only...... 106 Figure 4.3: Generalized linear model to estimate predictor variables for density of (A) blue ash seedlings and (B) green-white ash seedlings Model equations: (A) blue seedlings = untreated + treated flowering blue ash + herbaceous vegetation height + woody vegetation height; (B) green-white seedlings = untreated + treated female ash + herbaceous height + slope. Solid line indicated predicted model; dashed lines indicated the effect of +/- standard deviation of herbaceous vegetation height. See Table 4.6 for the statistical summary...... 108 Figure 5. 1: Principal coordinate analysis of genetic variation collected from green ash seedlings and saplings from the Upper Huron River Watershed in southeastern Michigan. (A) Differentiation of the genetic structure of class 2 seedlings (CL2), class 3 seedlings (CL3), and saplings (SAP). (B) Differentiation of the genetic structure for each state or metropark; Highland (HL), Hudson Mills (HM), Island Lake (IL), Indian Springs (IS), and Proud Lake (PL). Ellipses indicate 95% confidence intervals for each group...... 138 Figure 5.2: Principal coordinate analysis on the genetic variation of green ash seedlings in southwestern Ohio at Five Rivers MetroParks. (A) Differentiation among seedling size classes (Classes 1-3) and (B) Differentiation among metroparks (Cox Arboretum, Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek). Ellipses indicate 95% confidence intervals for each group...... 146 Figure 5.3: Principal coordinate analysis on the genetic variation of green ash seedlings in southwestern Ohio at Five Rivers MetroParks. (A) Differentiation among low and high EAB impact sites and (B) Differentiation among density of treated ash trees. Ellipses indicate 95% confidence intervals for each group...... 147 Figure 5.4: Principal coordinate analysis of ash seedlings and saplings from Michigan and Ohio. (A) Differentiation of Michigan and Ohio populations, size classes were pooled. Ellipses represent 95% confidence interval of each group...... 150 Figure 5.5: Principal coordinate analysis of ash seedlings and saplings from Michigan and Ohio. (A) Comparison of size classes in Michigan vs. all of Ohio. (B) Comparison of size classes in Ohio vs. all of Michigan...... 151

xv

Figure 6.1: Conceptual model of factors affecting population demography of green and white ashes at Five Rivers MetroParks in Dayton, Ohio. Solid lines indicate significant relationships quantified in this study, dashed lines indicate predicted variables that were not quantified or not supported from the study...... 169

xvi

Chapter 1: Literature Review

Natural Forest Dynamics

Forests are complex ecosystems that support diverse biotic communities. Tree communities sequester and store large amounts of carbon, recycle water and other nutrients (e.g., nitrogen and phosphorus) and provide valuable (aesthetical and economic) resources and services to humans (Spurr and Barnes, 1980). Forests develop during late stages of primary succession (Clements, 1916). However, secondary succession and instances of disturbance are very important for maintaining the ecological processes and biodiversity in forests (Bray, 1956). Natural disturbances in forests include fire, canopy gap formation, and understory browsing (Spurr and Barnes, 1980). Studies have shown that incidences of natural disturbance reduce overcrowding, increase diversity, and stimulate seed germination (Watt, 1947; Harper, 1977).

Canopy gaps are defined as an opening in the forest canopy. When gaps are created light penetrates mid and understory forest levels (Canham, 1988) causing temperature and moisture to change on and above the forest floor (Gray et al., 2002).

Accumulation of dead woody debris can alter nitrogen (Lovett et al., 2002; Scharenbroch and Bockheim, 2008a; Rubino et al., 2015) and carbon cycling (Scharenbroch and

Bockheim, 2008b). Gaps range from a single dead tree (standing) to multiple dead trees

1

(standing or fallen) (Whitmore, 1989). Gaps are an important stage of the forest growth cycle, because they increase plant diversity by creating opportunities for suppressed, shade intolerant seedlings and saplings to grow (Whitmore, 1989).

The intermediate disturbance hypothesis states that biodiversity is highest when disturbances are intermediate in frequency, size, and/or time since the last disturbance

(Connell, 1978). In forests, gaps represent intermediate disturbances. In the absence of disturbance, tree communities typically becomes dominated by fewer species and individuals (Runkle, 1982). Alternatively, large-scale disturbance events, such as tornado blowdowns, that kill many trees alter the forest floor through substantial increases in accumulation of coarse woody debris, which can drastically change the structure of the forest community (Clinton and Baker, 2000; Phillips et al., 2008). Additionally, forest fauna, including soil arthropods, birds, and mammals become displaced, experience population declines, or cannot survive in this new habitat (Hirao et al., 2008; Thorn et al.,

2016).

When small to medium size gaps occur in forests, shade-intolerant understory saplings and trees can respond to the increased light by growing vertically towards the opening in the canopy (Runkle, 1990). These pockets of secondary succession following gap formation support populations of shade-intolerant tree species such as black cherry

(Prunus serotina), tulip poplar (Liriodendron tulipifera), and black walnut (Juglans nigra), and mid-tolerant species such as ashes (Fraxinus spp.), white oak (), red oak (Quercus rubra), and elm (Ulmus spp.) (Braun, 1961).

2

Invasive Insects and Forest Dynamics

Currently, some forests in North America are being heavily disturbed by invasions of exotic species, which are among the most devastating causes of disturbance and biodiversity loss in forest systems (Lovett et al., 2006). In particular, the invasion of emerald ash borer (EAB), Agrilus planipennis Fairmaire (Coleoptera: Buprestidae), has killed millions of ash (Fraxinus sp.) trees in North America since its accidental introduction more than 20 years ago (Siegert et al., 2014).

Other consequences of biological invasion that cause widespread tree mortality are impacts to the animal communities that depend on these tree species. For example, black-throated green warbler, blackburnian warbler, Acadian flycatcher, and hermit thrush are strongly associated with intact forests (Tingley et al., 2002). However, following widespread hemlock mortality by hemlock woolly adelgid, abundance of these species declined and the avian community shift to species associated with edge habitat

(Tingley et al., 2002). The same trend has been reported following ash mortality. As the ash population shifted from mostly healthy to mostly dead trees, bird species richness and diversity increased, prevalence of woodpeckers and other bark gleaning birds and bird species associated with forest edge habitat increased (Long, 2013). Many interior bird species are declining as a result of forest fragmentation, and tree mortality is further isolating an already disturbed ecosystem. The arthropod community is also at risk due to widespread ash mortality. In particular, 282 native and exotic arthropods were reported to be associated with ash species in North America, 25% of which were categorized as moderate to high risk of extinction (Gandhi and Herms, 2010). As ash continues to die

3 within forests, populations of arthropods, small mammals, and birds that are associated with ash may decline.

The Emerald Ash Borer Invasion

Emerald ash borer (EAB, Agrilus planipennis) was introduced to North America near Detroit, Michigan more than 20 years ago. EAB was first detected in 2002 and dendrochronological analyses of tree mortality patterns indicate that EAB was killing ash trees at least by the early 1990’s (Siegert et al., 2014). As of March 2017, EAB was known to have spread to 30 states and 2 Canadian provinces (emeraldashborer.info) and has killed hundreds of millions of ash trees throughout much of eastern North America

(Herms and McCullough, 2014).

Ash populations in southeastern Michigan and northwestern Ohio have been impacted by EAB the longest, because of their proximity to the epicenter of the invasion

(Knight et al., 2013; Klooster et al. 2014). Long-term studies of the ash population in these areas reported on the relationships between canopy condition, degree of EAB infestation, and the subsequent death of infested ash trees. There was a non-linear positive relationship between density of EAB exit holes on the lower bole (i.e., exit holes below the tree canopies) and severity of canopy decline (Smith, 2006; Smith et al., 2015).

Another study showed a direct correlation between density of EAB larvae and canopy dieback (Flower et al., 2014). These relationships suggest that canopy dieback can be used as an index of degree of EAB infestation of an individual tree. As the population densities of EAB increases within an area, the mortality rate of ash trees also increases.

4

Near the epicenter of the EAB invasion, in 2005 (three years after EAB was detected) approximately 40% of ash had died from EAB (Smith et al., 2015). By 2009, 99.7% of susceptible ash were dead (Klooster et al., 2014). Similarly, in northwestern Ohio (near

Toledo, OH), stands transitioned from mostly healthy trees to mostly dead trees within six years (Knight et al., 2013). Knight et al. (2013) also found that mortality rates varied depending on the position of ash trees in the canopy. Specifically, dominant and co- dominant ash trees survived longer than intermediate and suppressed trees.

EAB can infest ash trees with stems as small as 2.5 cm in diameter and EAB often kills ash trees before they reach sexual maturity (McCullough and Siegert, 2007).

Overall, mesic sites have the greatest ash regeneration than xeric or hydric sites. Within mesic sites, viable ash seeds declined from approximately 23/m2 in 2005 to 0/m2 in 2007

(Klooster et al., 2014). Additionally, they reported that newly germinated seedlings (with cotyledons) rapidly declined from 2,266 ± 869/ha in 2008 to 0 in 2010. These data suggest that the ash seedbank was depleted two to three years following peak ash mortality. Furthermore, as ash trees died from EAB infestation, reproduction and new regeneration ceased Since EAB can infest and kill trees before they reach sexual maturity, it threat to extirpate ash throughout its range of invasion.

Emerald ash borer is native to Asia and has co-evolved with Asian ash species, such as Manchurian ash (Fraxinus mandshurica) and Chinese ash (F. chinensis) (Villari et al., 2016), and F. rhyncophylla. In its native range, EAB is considered a secondary pest that colonizes stressed or dying ash trees (Liu et al. 2003). However, in Asia EAB is a primary pest of North American ash when they are planted there as ornamental and shade

5 trees, including green ash (F. pennsylvanica), white ash (F. americana), and velvet ash

(F. velutina) (Liu et al. 2003).

All North American species of ash encountered by EAB thus far are susceptible to varying degrees. The susceptibility of ash species depends on (1) likelihood of oviposition and (2) successful larval development. Oviposition may differ by species as indicated by host preference of adults for green ash > white ash > black ash >blue ash >

European ash > Manchurian ash (Pureswaran and Poland, 2009; Tanis and McCullough,

2015). Studies have also reported a hierarchy in larval survival on green ash > black ash

> white ash > blue ash > Manchurian ash (Anulewicz et al., 2008; Tanis and

McCullough, 2015). Differential survival of green, white, and blue ash in forests where these species naturally occur have also been reported. Specifically, white ash mortality increased after most green ash trees had been killed by EAB (Anulewicz et al. 2007), and in the same study sites, blue ash mortality increased after most white ash trees had been killed by EAB (Tanis and McCullough, 2012). Therefore, in North America, green ash and black ash may be more likely to be killed by EAB than white ash and blue ash, and white ash may be more likely to be killed than blue ash.

Composition of forest stands had no effect on rate of ash mortality (Smith, 2006;

Smith et al., 2015). This suggests that all forest types with ash present are susceptible to

EAB infestation. In particular, Smith et al. (2015) found that black ash declined initially at a faster rate than green or white ash. Black ash is a hydric species that is typically found in bogs, swamps, along streams, or in poorly drained soils. Within these habitats, black ash can exist in nearly pure stands (MacFarlane and Meyer, 2005). Therefore,

6 although all stands are equally susceptible to EAB attack, stands with higher proportions of ash trees will undergo more extreme changes to the forest structure.

Ashes are important tree species throughout North America (MacFarlane and

Meyer, 2005). In particular, ashes represent approximately 2.5% of the aboveground carbon biomass across the U.S. and 5.5 – 9% of aboveground carbon in forests of

Michigan and Ohio (Flower et al., 2013). Unlike its coevolved host which EAB colonizes when stressed, EAB infests healthy and stressed North American ash trees. Furthermore, studies show that healthy populations of ash can die two to three years of first signs of

EAB infestation (Knight et al., 2013). As a result, forest populations undergo widespread and simultaneous ash mortality, creating large canopy gaps (Klooster et al., 2014). The formation of canopy gaps changes the microclimate of the understory by increasing light

(Canham, 1988), temperature and moisture (Gray et al., 2002). Additionally, the accumulation of fine and coarse woody debris can alter nutrient cycling (Lovett et al.,

2002; Scharenbroch and Bockheim, 2008a). Canopy gaps can also be invaded by non- native and other disturbance tolerant species (Hausman et al., 2010). Plant invasions can further change soil conditions and outcompete native vegetation for light and other recourses (Hausman et al., 2010). Lastly, as the forest plant community changes, the animal community also changes (Niesenbaum, 1992; Redman and Scriber,

2000; Ulyshen et al., 2011). The native arthropod community may be the most at risk from the EAB invasion. In particular, 282 native and exotic arthropods were reported to be associated with ash species in North America, 25% of which were categorized as moderate to high risk of extinction (Gandhi and Herms, 2010).

7

Widespread ash mortality also causes major economic losses. Ash is commonly used for baseball bats, furniture, cabinets, and tool handles. Additionally, ashes are common street trees. Municipalities often replaced elm trees that died from

Dutch elm disease with ash trees (Parker and Leopold, 1983). Thus, the EAB invasion is impacting property values. It has been estimated that economic impact of EAB could exceed $10 billion by 2019 (Vannatta et al., 2012). The expenses associated with tree inspections, dead tree removal and disposal, and new tree replacement in urban environments could be as much as $7.6 billion in Ohio (Sydnor et al., 2007). Vannatta et al. (2012) compared costs associated with various management plans. Their simulation concluded that treating an urban ash population with emamectin benzoate insecticide applications every two years was able to protect 59% of the trees over 20 years, and was cheaper than a “do nothing” (3% of ash remaining) and removal and/or replacement (0% of ash remaining) management scenarios.

Insecticide Management for EAB

Several insecticides have been tested for efficacy of EAB control in urban and natural forests. Imidacloprid and emamectin benzoate are the two most common insecticides used to protect ash trees against EAB (Smitley et al., 2010; McCullough et al., 2011). Imidacloprid is a systemic neo-nicotinoid insecticide that is applied either by soil drench or direct trunk injection. Emamectin benzoate is an avermectin applied by trunk injection, which causes insect paralysis by disrupting neurotransmitters (US EPA,

2009).

8

Imidacloprid is a broad spectrum systemic insecticide that has been extensively used to control various insect pests. In order to maintain effective doses of this insecticide within the trees, treatments have to be administered every year (Smitley et al., 2010).

Imidacloprid effectively kills adult insects feeding on the foliage of plants as well as larvae feeding on vascular tissues.

Large-scale applications of imidacloprid have been used as a conservation tool to protect eastern hemlock trees from hemlock woolly adelgid at Great Smokey Mountain

National Park (Benton et al., 2015). Imidacloprid treatments have also been used in Asian longhorned beetle eradication programs (Ugine et al., 2011). Most recently, studies have been conducted to investigate the integration of imidacloprid applied at reduced rates with biological control (Davidson and Rieske, 2016), and found support that lower rates may slow development of EAB larvae and increase their window of vulnerability to parasitism.

Emamectin benzoate is another commonly used, broad spectrum, systemic insecticide that also impacts EAB adults and larvae (McCullough et al., 2011). Unlike imidacloprid, toxic levels of emamectin benzoate can remain present in trees for two to four years (Smitley et al., 2010; McCullough et al., 2011). However, emamectin benzoate cannot effectively protect ash trees that are already experiencing greater than 50% canopy decline at the time of treatment (Flower et al., 2015).

A few studies have looked at the effectiveness of large-scale insecticide treatments within urban and natural forests. Of the insecticides use against EAB, emamectin benzoate may be the best choice since trees do not need to be treated every

9 year to remain protected. Additionally, if funding is available to treat annually, treatments can be staggered to treat a proportion of the ash population. Simulation models indicated that staggering treatments of 20% of the ash trees annually, could result in nearly 100% ash protection over a 10 year period (McCullough and Mercader, 2012). These models also suggest that treated ash trees that could provide associational protection to untreated neighboring trees by decreasing the surrounding density of EAB. Furthermore, as the proportion of treated trees increases, the magnitude of the effect of associational protection is predicted to increase (McCullough and Mercader, 2012).

The concept of associational protection is similar in some ways to the theory of associational resistance of plants to herbivory that was proposed by Tahvanainen and

Root (1972), who found that insect herbivory was lower in areas with higher plant diversity. In particular, they found evidence that some plants may release allelochemicals that interfere with an herbivore’s ability to detect their host plant. Additionally, when plant diversity increased, predators and parasitoids of crop pests also increased

(Tahvanainen and Root, 1972). These findings have been supported by other studies. For example, the aromatic sweet gale (Myrica gale), disrupted the ability for herbivores to locate purple loosestrife (Lythrum salicaria) (Hamback et al., 2000), and ponderosa pine (Pinus ponderosa) and Juniperus spp decrease bark beetle abundance on P. edulis due to volatile terpenes that interfered with host detection (Sholes, 2008).

Proposed mechanisms of associational resistance include decreased plant apparency and increased natural enemies (Tahvanainen and Root, 1972). Other studies have suggested that certain plants alter the microclimate which releases host plant

10 defensive chemicals (Quiroz et al., 1997; Karban and Maron, 2002; Ayres et al., 2007).

Lastly, some plants disrupt the ability of an herbivore to locate the host plant (Hamback et al., 2000; Sholes, 2008; Jactel et al., 2011; Castagneyrol et al., 2013).

A study conducted in the Upper Peninsula in Michigan, (Mercader et al., 2015), tested whether insecticide treatments may provide associational protection to untreated ash trees by reducing the surrounding density of EAB. They treated 587 trees from 2009 to 2010 with emamectin benzoate and found that EAB larval densities decreased as density of treated ash trees increased. They also found that areas with more ash phloem area (i.e., the food source for larvae) have the ability to produce more EAB adults and may be more attractive to females for oviposition sites. These findings indicate that treating large ash trees (i.e., ash trees with more phloem) with insecticide may have a greater effect on reducing the surrounding EAB density than treating smaller trees, and therefore provide stronger associational protection to neighboring untreated trees.

The objectives of Chapters 2 and 3 were to investigate whether insecticides provide direct protection to five North American ash species (green ash, white ash, black ash, pumpkin ash, and blue ash) and to investigate whether treated ash trees provide associational protection to neighboring untreated ash trees.

Impact of EAB on Ash Reproduction

In areas where EAB has been established for at least a decade, few living reproductive ash trees remain, which has caused ash reproduction and new regeneration to decline precipitously and even cease (Kashian and Witter, 2011; Klooster et al., 2014).

11

Furthermore, the surviving reproductive ash trees have become increasingly isolated. In

2005, ash mortality in southeastern Michigan near the epicenter of EAB invasion was approximately 40% and ash seed density in the soil was 1.5 million ash seeds per hectare

(ha) in mesic sites (Klooster et al., 2014). By 2007, ash mortality reached 92% and no seeds were recovered in mesic sites (Klooster et al., 2014). In 2008, density of first year ash seedlings in the same sites was approximately 2,300 plants per ha, which rapidly decreased to 143 stems/ha in 2009 and 0 stems in 2010 (Klooster et al., 2014). Over the same time period and sites, density of established seedlings remained high (>200,000 stems/ha), but decreased from 2008 to 2010. Density of saplings (stems > 1.5 m tall and less than 2.5 cm diameter, which is too small to be susceptible to EAB attack) was estimated at 75, 287, and 274 per ha from 2008 to 2010, respectively. Natural mortality rates of seedlings are typically high at early seedling stages and decrease rapidly once individuals reach a certain age or size (Hett and Loucks, 1968; Kobe et al., 1995). These reports indicate that (1) viable seeds do not survive in the soil seedbank for more than two years, and that as mature ash trees die, new regeneration rapidly slows to a stop; and

(2) in the absence of new germination, most established ash seedlings die before they become saplings.

Cessation of ash reproduction and new regeneration in forests invaded by EAB jeopardizes the future of ash. Since natural mortality is high for younger seedlings, long- term persistence of trees requires continuous reproduction and regeneration (Plumptre,

1995). However, it may be possible to protect mature ash trees in forests using insecticide treatments that maintain ash reproduction and new regeneration. Additionally, by

12 maintaining mature ash trees, ecological impacts of ash mortality on forest communities may be reduced. In Chapter 4, I tested whether treating ash trees maintains ash reproduction and population demographics of various size classes as the impact of EAB intensifies, and whether this effect is dependent on the density of treated trees.

Impacts of EAB on Genetic Variation

The effect of widespread ash mortality on genetic variation in ash populations is also a concern. Small populations are more susceptible to inbreeding and genetic drift, which decreases genetic variation (Nei et al., 1975; Frankham, 1996). Specifically, when a population undergoes a sudden decrease in population size, the genetic variation of that population goes through a bottleneck that decreases the number and frequency of different alleles per loci (Nei et al., 1975). Additionally, as the population size decreases, frequency of mating between related individuals increases, which increases the proportion of homozygotes. A major problem associated with inbreeding is the higher chance that deleterious recessive alleles will (1) have increased prevalence in the population, and (2) be expressed in homozygous recessive individuals (Ouborg et al.,

2006). Deleterious recessive alleles can be expressed as traits that lower fitness.

However, individuals need two copies of that allele in order to express the phenotype

(i.e., must be homozygous recessive for that trait). Therefore, if smaller populations have more individuals with homozygous recessive genotypes for deleterious alleles, they have a higher risk of extinction (Nei et al., 1975; Frankham, 1996).

13

Another effect of small population size is genetic drift, which is defined as random change in allele frequency from one generation to the next (Ellstrand and Elam,

1993). In larger populations, episodes of random mortality have little effect on allele frequencies (Frankham, 1995). In small populations, however, random mortality can rapidly change allele frequencies (Ellstrand and Elam, 1993). Additionally, in small populations, there is a greater chance that random mortality will increase the frequency of deleterious alleles (i.e., traits associated with lower fitness) through genetic drift

(Pautasso, 2009; Ilves et al., 2013), thus making the population more prone to extinction.

The effects of genetic erosion (i.e., the loss of genetic variation) in plant populations isolated by habitat fragmentation, and its direct consequences on survival and fecundity of off spring have been reported in several studies. A study of fragmented and continuous populations of European beech (Fagus sylvatica) found that there were fewer alleles per loci (allelic richness), higher incidence of inbreeding, fewer heterozygotes, and more bottleneck events in fragmented forests (Jump and Penuelas, 2006). Similarly, low genetic variation and higher rates of self-pollination were observed in the endangered perennial herb (Lingularia sibirica) following severe population declines caused by deforestation (Ilves et al., 2013). Meta-analyses of the relationship between fragmentation and genetic erosion of various plant species concluded that fragmented habitats resulted in lower frequency heterozygotes, fewer alleles per loci, lower percentages of polymorphic loci (loci with more than one allele), and lower outcrossing rate (Aguilar et al., 2008). They observed no overall effect of inbreeding, but when comparing progeny to adults, inbreeding was more evident within the progeny generation

14

(Aguilar et al., 2008). Vranckx et al. (2012) also reported that genetic variation of tree species was lower in fragmented than contiguous forests, but that inbreeding was not associated with fragmentation. They suggested that changes in allelic richness occurred more rapidly than changes to heterozygosity following fragmentation. They also found that adults and progeny became more differentiated over time in fragmented forests

(Vranckx et al., 2012).

Genetic variation increases with range size (Frankham, 1996). Therefore, isolated populations typically have lower levels of genetic variation. Additionally, inter-connected populations are characterized by continuous gene flow as individuals move between them. Fragmentation of populations with a larger range sizes 1) causes an extreme reduction in population size, 2) isolates that population and disrupts gene flow, and 3) reduces genetic variation of the surviving individuals (Frankham, 1996). If fragmentation persists over several generations, inbreeding and genetic drift will continue to decrease genetic variation (Keller and Waller, 2002), which can lead to inbreeding depression as frequency of homozygous recessive deleterious alleles increases. Furthermore decreased genetic variation can constrain evolutionary responses to future environmental changes

(Bensch et al., 1994; Husband and Schemske, 1996; Amos et al., 2001; Vranckx et al.,

2012).

Negative impacts have been documented for populations with high rates of inbreeding, which provides empirical evidence of inbreeding depression. Husband and

Schemske (1996) compared seed production, germination, and survival to reproduction of various self-fertilizing and outcrossing Angiosperm and Gymnosperm species. Self-

15 fertilizing individuals had higher inbreeding coefficients than outcrossing individuals.

Additionally, the self-fertilizing Angiosperms grew slower, produced fewer seeds, had fewer successful seed germinations, and lower reproduction than outcrossing individuals

(Husband and Schemske, 1996). Bensch et al. (1994) investigated the great reed warbler

(Acrocephalus arundinaceus) to determine the effects of inbreeding on offspring fitness and found a linear decline in the number of eggs hatched from parents who shared more than 40% of the genetic markers tested. However, they did not observe mating between siblings or parents and hypothesized that the high degree of relatedness may have been due to (1) distant relatives having similar banding patterns for the loci used, and/or (2) a recent genetic bottleneck. Finally, a meta-analysis of effects of inbreeding by three large vertebrates on their offspring found that parents that were more closely related were more likely to produce offspring that did not survive to maturity (Amos et al., 2001).

Habitat fragmentation caused by humans has directly impacted many plant populations, resulting in a loss of genetic variation (Knapp et al., 2001; Sork et al., 2002).

The method of pollination among plants affects the susceptibility to genetic erosion following fragmentation (Vranckx et al., 2012). Gene flow of plants that are pollinated by birds is less restricted by distance between plants than gene flow in wind or insect pollinated plants, suggesting that severe fragmentation may negatively impact wind and insect pollinated plants (Vranckx et al., 2012). For example, Knapp et al. (2001) found that the number of acorns produced by blue oak (Quercus douglasii) increased as the number of pollen producing trees within 60 m increased. They also suggested that when there is less connectivity between neighborhoods of blue oak trees, the rate of seedling

16 recruitment will not be high enough to balance the rate of seedling mortality, which could have long-term implications for the conservation of this species (Knapp et al., 2001).

Sork et al. (2002) found that pollen dispersal in a population of California valley oak (Q. lobata) that has been declining for the past 200 years was localized, and that successful fertilization decreased with distance from pollen source to the focal tree. They predicted that fragmentation will continue to cause this population to decline because of the low rate of successful pollination (Sork et al., 2002).

Generation time can also determine how habitat fragmentation affects genetic variation. When fragmentation results in genetic erosion, plant species with shorter generation times (herbaceous species) will complete more generations in the fragmented habitat, further increasing inbreeding and magnifying effects of genetic drift (Young et al., 1996). However, studies have shown that progeny of trees within fragmented habitats have lower genetic variation than conspecific adults (Vranckx et al., 2012) and higher rates of inbreeding (Aguilar et al., 2008; Vranckx et al., 2012), especially as time after fragmentation increased (>50 years).

As EAB-induced ash mortality eliminated reproductive ash trees and regeneration

(Klooster et al., 2014), genetic variation may have been reduced in the declining seedling population. Specifically, as ash mortality increases, reproductive ash trees may become increasingly isolated which may restrict gene flow to those few individuals remaining within close proximity. Since degree of relatedness increases as distance between trees decreases (Heuertz et al., 2001), reproduction may increasingly occur between related individuals (i.e., higher rates of inbreeding). Hence, EAB-induced mortality may

17 decrease genetic variation and fitness of ash progeny, making them less resilient to future biotic and abiotic disturbances (Hamrick, 2004).

Effects of EAB on the genetic structure of ash populations in North America have not been studied. Population genetics of green ash (F. pennsylvanica) in northwest Ohio was quantified in 2005 by Hausman et al. (2014) before these populations were impacted by EAB, and they observed high allelic richness and no evidence of inbreeding.

However, it is possible that EAB-induced mortality has decreased the effective population size of ash in southeastern Michigan where ash mortality exceeded 99% in

2009 (Klooster et al., 2014), creating higher rates of inbreeding and a genetic bottleneck in the surviving “orphaned” seedling population. Based on age approximations of seedlings (1-7 years) and saplings (7-20 years) (Burns and Honkala, 1990), smaller seedlings would be offspring of the few mature ashes surviving during the period of peak mortality, while saplings would have established prior to the EAB invasion. Therefore, comparisons of the genetic structure of seedling and sapling populations could indicate whether EAB-induced ash mortality has produced a genetic bottleneck. If so, the ash population could be further threatened with extirpation as inbreeding depression lowers fitness (Hu et al., 2010; Potter et al., 2012).

Managers of ash populations should consider the effects of genetic variation to avoid the negative effects similar to those associated with captive breeding programs. In particular, preserving genetic variation of ash requires understanding effects of pollination and seed dispersal, and degree of genetic variation present prior to the EAB invasion. Most ash species, including green and white ash, are dioecious (separate male

18 and female trees) and wind pollinated (Burns and Honkala, 1990). Multiple males can fertilize flowers from a single female ash tree, therefore genetic variation is determined by the number of males present (Heuertz et al., 2001). Additionally, female flowers can be fertilized by males up to 200 m away (Heuertz et al., 2003). Hausman et al. (2014) reported that green ash in northwestern Ohio had high allelic variation and low rates of inbreeding and concluded that 200 seeds need to be collected from at least five mother trees in order to capture 99% of the genetic variation. This suggests that females need to be protected to maintain seed production, but protecting males may be more important for preserving genetic variation. Protecting male and female ash trees with insecticides may enable managers to conserve genetic variation of ash during the EAB invasion.

In Chapter 5, I quantified the population genetic parameters of allele richness, heterozygosity, inbreeding, and genetic differentiation of ash progeny at two sites with varying levels of mortality caused by EAB: (1) southeastern Michigan where nearly

100% of mature ashes have died from EAB and no ash protection strategies have been implemented and (2) southwestern Ohio where forest managers initiated an insecticide program in 2011 to protect ash and preserve the ash gene pool.

In Chapter 6, I have summarized the results from Chapters 2 through 5 into a conceptual model of factors that affect in survival of untreated ash, seedling densities, and overall demography of the ash population. Additionally, I have highlighted future research required to improve the understanding of managing tree populations in response to invasions by species.

19

Literature cited

Aguilar, R., Quesada, M., Ashworth, L., Herrerias-Diego, Y., Lobo, J. 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Mol. Ecol. 17: 5177-5188. Amos, W., Wilmer, J.W., Fullard, K., Burg, T.M., Croxall, J.P., Bloch, D., Coulson, T. 2001. The influence of parental relatedness on reproductive success. Proc. R. Soc. B 268: 2021-2027. Anulewicz, A.C., McCullough, D.G., Cappaert, D.L. 2007. Emerald ash borer (Agrilus planipennis) density and canopy dieback in three North American ash species. Arboric Urban For. 33: 338-349. Anulewicz, A.C., McCullough, D.G., Cappaert, D.L., Poland, T.M. 2008. Host range of the Emerald ash borer (Agrilus planipennis Fairmaire) (Coleoptera: Buprestidae) in North America: Results of multiple-choice field experiments. Environ. Entomol. 37: 230-241. Ayres, E., Dromph, K.M., Cook, R., Ostle, N., Bardgett, R.D. 2007. The influence of below-ground herbivory and defoliation of a legume on nitrogen transfer to neighboring plants. Funct. Ecol. 21: 256-263. Bensch, S., Hasselquist, D., Vonschantz, T. 1994. Genetic similarity between parents predicts hatching failure: Nonincestuous inbreeding in the great reed warbler? Evolution 48: 317-326. Benton, E.P., Grant, J.F., Webster, R.J., Nichols, R.J., Cowles, R.S., Lagalante, A.F., Coots, C.I. 2015. Assessment of imidacloprid and its metabolites in foliage of eastern hemlock multiple years following treatment for hemlock woolly adelgid, Adelges tsugae (Hemiptera: Adelgidae), in forested conditions. J. Econ. Entomol. 108: 2672-2682. Braun, E. L. 1950. forests of eastern North America. Hafner, New York, New York, U.S.A. Bray, J.R. 1956. Gap phase replacement in a maple-basswood forest. Ecology 37: 598- 600. Burns, R.M., Honkala, B.H. 1990. Silvics of North America: 2, hardwoods. Agriculture handbook 654, U.S. Department of Agriculture, Washington, D.C. Canham, C.D. 1988. An index for understory light levels in and around canopy gaps. Ecology 69: 1634-1638. Castagneyrol, B., Giffard, B., Pere, C., Jactel, H. 2013. Plant apparency, an overlooked driver of associational resistance to insect herbivory. J. Ecol. 101: 418-429. Clements, F.E. 1916. Plant Succession: An analysis of the development of vegetation. Carnegie Inst., Washington. Clinton, B.D., Baker, C.R. 2000. Catastrophic windthrow in the southern Appalachians: Characteristics of pits and mounds and initial vegetation responses. For. Ecol. Manag. 126: 51-60. Connell, J.H. 1978. Diversity in tropical rain forests and coral reefs: High diversity of trees and corals is maintained only in a non-equilibrium state. Science 199: 1302- 1310. 20

Davidson, W., Rieske, L.K. 2016. Establishment of classical biological control targeting emerald ash borer is facilitated by use of insecticides, with little effect on native arthropod communities. Biol. Control 101: 78-86. Ellstrand, N.C., Elam, D.R. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217-242. Flower, C.E., Dalton, J.E., Knight, K.S., Brikha, M., Gonzalez-Meler, M.A. 2015. To treat or not to treat: Diminishing effectiveness of emamectin benzoate tree injections in ash trees heavily infested by emerald ash borer. Urban For. Urban Greening 14: 790-795. Flower, C.E., Knight, K.S., Gonzalez-Meler, M.A. 2013. Impacts of the emerald ash borer (Agrilus planipennis Fairmaire) induced ash (Fraxinus spp.) mortality on forest carbon cycling and successional dynamics in the eastern United States. Biol. Invasions 15: 931-944. Flower, C.E., Long, L.C., Knight, K.S., Rebbeck, J., Brown, J.S., Gonzalez-Meler, M.A., Whelan, C.J. 2014. Native bark-foraging birds preferentially forage in infected ash (Fraxinus spp.) and prove effective predators of the invasive emerald ash borer (Agrilus planipennis Fairmaire). For. Ecol. Manag. 313: 300-306. Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327. Frankham, R. 1996. Relationship of genetic variation to population size in wildlife. Conserv. Biol. 10: 1500-1508. Gandhi, K.J.K., Herms, D.A. 2010. North American arthropods at risk due to widespread Fraxinus mortality caused by the Alien emerald ash borer. Biol. Invasions 12: 1839-1846. Gray, A.N., Spies, T.A., Easter, M.J. 2002. Microclimatic and soil moisture responses to gap formation in coastal Douglas-fir forests. Can. J. For. Res. 32: 332-343. Hamback, P.A., Agren, J., Ericson, L. 2000. Associational resistance: Insect damage to purple loosestrife reduced in thickets of sweet gale. Ecology 81: 1784-1794. Hamrick, J.L. 2004. Response of forest trees to global environmental changes. For. Ecol. Manag. 197: 323-335. Harper, J.L. 1977. Population biology of plants. Academic Press, London. Hausman, C.E., Bertke, M.M., Jaeger, J.F., Rocha, O.J. 2014. Genetic structure of green ash (Fraxinus pennsylvanica): implications for the establishment of ex situ conservation protocols in light of the invasion of the emerald ash borer. Plant Genet. Resour. 12: 286-297. Hausman, C.E., Jaeger, J.F., Rocha, O.J. 2010. Impacts of the emerald ash borer (EAB) eradication and tree mortality: Potential for a secondary spread of invasive plant species. Biol. Invasions 12: 2013-2023. Herms, D.A., McCullough, D.G. 2014. Emerald ash borer invasion of North America: History, biology, ecology, impacts, and management. Annu. Rev. Entomol. 59: 13- 30. Hett, J.M., Loucks, O.L. 1968. Application of life-table analyses to tree seedlings in Quetico Provincial Park, Ontario. For. Chron. 44: 29-32.

21

Heuertz, M., Hausman, J.F., Tsvetkov, I., Frascaria-Lacoste, N., Vekemans, X. 2001. Assessment of genetic structure within and among Bulgarian populations of the common ash (Fraxinus excelsior L.). Mol. Ecol. 10: 1615-1623. Heuertz, M., Vekemans, X., Hausman, J.F., Palada, M., Hardy, O.J. 2003. Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash. Mol. Ecol. 12: 2483-2495. Hirao, T., Murakami, M., Iwamoto, J., Takafumi, H., Oguma, H. 2008. Scale-dependent effects of windthrow disturbance on forest arthropod communities. Ecol. Res. 23: 189-196. Hu, L.-J., Uchiyama, K., Saito, Y., Ide, Y. 2010. Contrasting patterns of nuclear microsatellite genetic structure of Fraxinus mandshurica var. japonica between northern and southern populations in Japan. J. Biogeogr. 37: 1131-1143. Husband, B.C., Schemske, D.W. 1996. Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54-70. Ilves, A., Lanno, K., Sammul, M., Tali, K. 2013. Genetic variability, population size and reproduction potential in Ligularia sibirica (L.) populations in Estonia. Conserv. Genet. 14: 661-669. Jactel, H., Birgersson, G. Andersson, S., Schlyter, F., 2011. Non-host volatiles mediate associational resistance to the pine processionary moth. Oecologia 166: 703-711. Jump, A.S., Penuelas, J. 2006. Genetic effects of chronic habitat fragmentation in a wind- pollinated tree. Proc. Natl. Acad. Sci. U.S.A. 103: 8096-8100. Karban, R., Maron, J. 2002. The fitness consequences of interspecific eavesdropping between plants. Ecology 83: 1209-1213. Kashian, D.M., Witter, J.A. 2011. Assessing the potential for ash canopy tree replacement via current regeneration following emerald ash borer-caused mortality on southeastern Michigan landscapes. For. Ecol. Manag. 261: 480-488. Keller, L.F., Waller, D.M. 2002. Inbreeding effects in wild populations. Trends Ecol. Evolut. 17: 230-241. Klooster, W.S., Herms, D.A., Knight, K.S., Herms, C.P., McCullough, D.G., Smith, A., Gandhi, K.J.K., Cardina, J. 2014. Ash (Fraxinus spp.) mortality, regeneration, and seed bank dynamics in mixed hardwood forests following invasion by emerald ash borer (Agrilus planipennis). Biol. Invasions 16: 859-873. Knapp, E.E., Goedde, M.A., Rice, K.J. 2001. Pollen-limited reproduction in blue oak: implications for wind pollination in fragmented populations. Oecologia 128: 48- 55. Knight, K.S., Brown, J.P., Long, R.P. 2013. Factors affecting the survival of ash (Fraxinus spp.) trees infested by emerald ash borer (Agrilus planipennis). Biol. Invasions 15: 371-383. Kobe, R.K., Pacala, S.W., Silander, J.A., Canham, C.D. 1995. Juvenile tree survivorship as a component of shade tolerance. Ecol. Appl. 5: 517-532. Liu, H., L.S. Bauer, R. Gao, T. Zhao, T.R. Petrice, R.A. Haack. 2003. Exploratory survey for the emerald ash borer, Agrilus planipennis (Coleoptera: Buprestidae), and its natural enemies in China. Gt. Lakes Entomol. 36: 191-204.

22

Long, L. 2013. Direct and indirect impacts of emerald ash borer on forest bird communities. In, Entomology. The Ohio State University, OhioLINK Electronic Theses and Dissertations Center, p. 165. Lovett, G.M., Canham, C.D., Arthur, M.A., Weathers, K.C., Fitzhugh, R.D. 2006. Forest ecosystem responses to exotic pests and pathogens in eastern North America. Bioscience 56: 395-405. Lovett, G.M., Christenson, L.M., Groffman, P.M., Jones, C.G., Hart, J.E., Mitchell, M.J. 2002. Insect defoliation and nitrogen cycling in forests. Bioscience 52: 335-341. MacFarlane, D.W., Meyer, S.P. 2005. Characteristics and distribution of potential ash tree hosts for emerald ash borer. For. Ecol. Manag. 213: 15-24. McCullough, D.G., Mercader, R.J. 2012. Evaluation of potential strategies to SLow Ash Mortality (SLAM) caused by emerald ash borer (Agrilus planipennis): SLAM in an urban forest. Int. J. Pest Manage. 58: 9-23. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. McCullough, D.G., Siegert, N.W. 2007. Estimating potential emerald ash borer (Coleoptera : Buprestidae) populations using ash inventory data. J. Econ. Entomol. 100: 1577-1586. Mercader, R.J., McCullough, D.G., Storer, A.J., Bedford, J.M., Heyd, R., Poland, T.M., Katovich, S. 2015. Evaluation of the potential use of a systemic insecticide and girdled trees in area wide management of the emerald ash borer. For. Ecol. Manag. 350: 70-80. Nei, M., Maruyama, T., Chakraborty, R. 1975. Bottleneck effect and genetic variability in populations. Evolution 29: 1-10. Niesenbaum, R.A. 1992. The effects of light environment on herbivory and growth in the dioecious shrub Lindera benzoin (Lauraceae). Am. Midl. Nat. 128: 270-275. Ouborg, N.J., Vergeer, P., Mix, C. 2006. The rough edges of the conservation genetics paradigm for plants. J. Ecol. 94: 1233-1248. Parker, G.R., Leopold, D.J. 1983. Replacement of Ulmus americana L. in a mature east- central Indiana . Bull. Torrey Bot. Club 110: 482-488. Pautasso, M. 2009. Geographical genetics and the conservation of forest trees. Perspect. Plant Ecol. Evol. Syst. 11: 157-189. Phillips, J.D., Marion, D.A., Turkington, A.V. 2008. Pedologic and geomorphic impacts of a tornado blowdown event in a mixed pine-hardwood forest. Catena 75: 278- 287. Plumptre, A.J. 1995. The importance of “seed trees” for the natural regeneration of selectively logged tropical forest. Commonw. Forest. Rev. 74: 253-258. Potter, K.M., Jetton, R.M., Dvorak, W.S., Hipkins, V.D., Rhea, R., Whittier, W.A. 2012. Widespread inbreeding and unexpected geographic patterns of genetic variation in eastern hemlock (Tsuga canadensis), an imperiled North American conifer. Conserv. Genet. 13: 475-498.

23

Pureswaran, D.S., Poland, T.M. 2009. Host Selection and Feeding Preference of Agrilus planipennis (Coleoptera: Buprestidae) on Ash (Fraxinus spp.). Environ. Entomol. 38: 757-765. Quiroz, A., Pettersson, J., Pickett, J.A., Wadhams, L.J., Niemeyer, H.M. 1997. Semiochemicals mediating spacing behavior of bird cherry-oat aphid, Rhopalosiphum padi feeding on cereals. J. Chem. Ecol. 23: 2599-2607. Redman, A.M., Scriber, J.M. 2000. Competition between the gypsy moth, Lymantria dispar, and the northern tiger swallowtail, Papilio canadensis: Interactions mediated by host plant chemistry, pathogens, and parasitoids. Oecologia 125: 218-228. Rubino, L., Charles, S., Sirulnik, A.G., Tuininga, A.R., Lewis, J.D. 2015. Invasive insect effects on nitrogen cycling and host physiology are not tightly linked. Tree Physiol. 35: 124-133. Runkle, J.R. 1982. Patterns of disturbance in some old-growth mesic forests of eastern North America. Ecology 63: 1533-1546. Runkle, J.R. 1990. Gap dynamics in an Ohio Acer-Fagus forest and speculations on the geography of disturbance. Can. J. For. Res. 20: 632-641. Scharenbroch, B.C., Bockheim, J.G. 2008a. The effects of gap disturbance on nitrogen cycling and retention in late-successional northern hardwood-hemlock forests. Biogeochemistry 87: 231-245. Scharenbroch, B.C., Bockheim, J.G. 2008b. Gaps and soil C dynamics in old growth northern hardwood-hemlock forests. Ecosystems 11: 426-441. Sholes, O.D.V. 2008. Effects of associational resistance and host density on woodland insect herbivores. J. Anim. Ecol. 77: 16-23. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Smitley, D.R., Doccola, J.J., Cox, D.L. 2010. Multiple-year protection of ash trees from emerald ash borer with a single trunk injection of emamectin benzoate, and single-year protection with an imidacloprid basal drench. Arboric. Urban For. 36: 206-211. Sork, V.L., Davis, F.W., Smouse, P.E., Apsit, V.J., Dyer, R.J., Fernandez, J.F., Kuhn, B. 2002. Pollen movement in declining populations of California Valley oak, Quercus lobata: Where have all the fathers gone? Mol. Ecol. 11: 1657-1668. Spurr, S.H., Barnes, B.V. 1980. Forest Ecology. Wiley, New York, NY. Sydnor, T.D., Bumgardner, M., Todd, A. 2007. The potential economic impacts of emerald ash borer (Agrilus planipennis) on Ohio, US, communities. Arboric. Urban For. 33: 48-54. 24

Tahvanainen, J.O., Root, R.B. 1972. The influence of vegetational diversity on the population ecology of a specialized herbivore, Phyllotreta cruciferae (Coleoptera: Chrysomelidae). Oecologia 10: 321-346. Tanis, S.R., McCullough, D.G. 2012. Differential persistence of blue ash and white ash following emerald ash borer invasion. Can. J. For. Res. 42: 1542-1550. Tanis, S.R., McCullough, D.G. 2015. Host resistance of five Fraxinus species to Agrilus planipennis (Coleoptera: Buprestidae) and effects of paclobutrazol and fertilization. Environ. Entomol. 44: 287-299. Thorn, S., Bussler, H., Fritze, M.A., Goeder, P., Muller, J., Weiss, I. Seibold, S. 2016. Canopy closure determines arthropod assemblages in microhabitats created by windstorms and salvage logging. For. Ecol. Manag. 381: 188-195. Tingley, M.W., Orwig, D.A., Field, R., Motzkin, G. 2002. Avian response to removal of a forest dominant: Consequences of hemlock woolly adelgid infestations. J. Biogeogr. 29: 1505-1516. Ugine, T.A., Gardescu, S., Hajek, A.E. 2011. The effect of exposure of imidacloprid on Asian longhorned beetle (Coleoptera: Cerambycidae) survival and reproduction. J. Econ. Entomol. 104: 1942-1949. Ulyshen, M.D., Klooster, W.S., Barrington, W.T., Herms, D.A. 2011. Impacts of emerald ash borer-induced tree mortality on litter arthropods and exotic earthworms. Pedobiologia 54: 261-265. United States Environmental Protection Agency. 2009. Memorandum: Ecological risk assessment for emamectin benzoate use as a tree injection insecticide to control arthropod pests. PC Code 122806. US EPA, Washington, DC, USA. Vannatta, A.R., Hauer, R.H., Schuettpelz, N.M. 2012. Economic analysis of emerald ash borer (Coleoptera: Buprestidae) management options. J. Econ. Entomol. 105: 196-206. Villari, C., Herms, D.A., Whitehill, J.G.A., Cipollini, D., Bonello, P. 2016. Progress and gaps in understanding mechanisms of ash tree resistance to emerald ash borer, a model for wood-boring insects that kill angiosperms. New Phytol. 209: 63-79. Vranckx, G., Jacquemyn, H., Muys, B., Honnay, O. 2012. Meta-analysis of susceptibility of woody plants to loss of genetic diversity through habitat fragmentation. Conserv. Biol. 26: 228-237. Watt, A.S. 1947. Pattern and process in the plant community. J. Ecol. 35: 1-22. Whitmore, T.C. 1989. Canopy gaps and the two major groups of forest trees. Ecology 70: 536-538. Young, A., Boyle, T., Brown, T, 1996. The population genetic consequences of habitat fragmentation for plants. Trends Ecol. Evol. 11: 413-418.

25

Chapter 2: Effects of emerald ash borer and emamectin benzoate insecticide

applications on populations of five species of ash in forests of southwestern Ohio

Abstract

The introduction of emerald ash borer (EAB) has drastically affected the population demographics of ash species in North America. EAB has killed hundreds of millions of ashes since its introduction nearly 20 years ago. Insecticides, such as emamectin benzoate, can successfully protect ash trees from EAB attack. The objectives of this study were to determine the effects of insecticide treatments on health and survival of green, white, black, pumpkin, and blue ash in forests of southwestern Ohio. This study was conducted at Five Rivers MetroParks (FRMP) near Dayton, Ohio, where a large-scale insecticide program was initiated in 2011 to treat ash trees with emamectin benzoate

(TREE-äge, Arborjet, Inc.) every two years. Survival of untreated green and white ash differed between parks, being higher (>75%) at Sugarcreek, Englewood, and

Germantown Metroparks than at Cox Arboretum, Taylorsville, and Twin Creek (<25%).

From 2014 – 2016, survival was higher and there were fewer signs of EAB infestation on insecticide treated green, white, black, and pumpkin ash trees than untreated conspecifics.

Survival of blue ash, however, was high for both treated and untreated individuals, and few trees showed signs of EAB infestation, which was a reflection of its higher resistance

26 to EAB. However, blue ash survival was lower and more variable at Cox Arboretum and

Twin Creek Metroparks, which was likely due to low survival of green and white ash at these parks.

Introduction

Invasions by exotic species are among the most devastating causes of disturbance and biodiversity loss in forest systems (Lovett et al., 2006). In particular, the invasion of emerald ash borer (EAB), Agrilus planipennis Fairmaire (Coleoptera: Buprestidae), has killed millions of ash (Fraxinus sp.) trees in North America since its introduction over 20 years ago (Siegert et al., 2014).

Emerald ash borer is native to Asia and has coevolved with Asian ash species, such as Manchurian ash (Fraxinus mandshurica) and Chinese ash (F. chinensis) (Liu et al. 2003). In its native range, EAB is considered a secondary pest and colonizes stressed or dying ash trees (Liu et al. 2003). However, in Asia EAB is a primary pest of exotic ornamental ash species that are native to North America, such as green ash (F. pennsylvanica), white ash (F. americana), and velvet ash (F. velutina) (Liu et al. 2003).

All North America ash species encountered by emerald ash borer thus far are susceptible (Villari et al., 2016). However, studies suggest a hierarchy of EAB preference for maturation feeding by adults on green ash > white ash > black ash (F. nigra) >blue ash (F. quadrangulata) > European ash (F. excelsior) > Manchurian ash (Anulewicz et al., 2008; Pureswaran and Poland, 2009). Preference towards a certain species is likely attributed to volatile emissions of secondary metabolites. In particular, Pureswaran and

27

Poland (2009) found lowest volatile emission from green ash (i.e., most preferred species) and highest in Manchurian ash (i.e., the least preferred species). Survival on different ash species also indicates a hierarchy. Specifically, EAB adults survive longest when feeding on black ash, intermediate on white, green, and Manchurian ash, and least on blue ash (Tanis and McCullough, 2015).

Unlike in its native range, EAB is able to infest healthy ash trees in North

America (Liu et al. 2003). Furthermore, studies show that after EAB has been detected, stands that consist of primarily healthy ashes can become mostly dead stands within six years (Knight et al., 2013). As a result, forest populations undergo widespread and relatively simultaneous ash mortality and gap formation (Gandhi and Herms, 2010a). The formation of canopy gaps increases light and dead woody debris on the forest floor, changing nutrient cycling and microclimate conditions (Perry and Herms, 2016a, b). The impacts of plant invasions, changes in nutrient cycling and light availability, can further alter the plant and animal communities that are supported by the forest (Niesenbaum,

1992; Ulyshen et al., 2011). The native arthropod community is also at risk due to widespread ash mortality. In particular, 282 native and exotic arthropods were reported to be associated with ash species in North America, 25% of which were categorized as moderate to high risk of extinction (Gandhi and Herms, 2010b). As ash populations continue to die within forests, arthropods, small mammals, and birds that are either associated with ash or interior forest are at risk of population declines.

Ash are important tree species throughout North America (MacFarlane and

Meyer, 2005). In particular, ashes represent approximately 2.5% of the aboveground

28 carbon biomass across the U.S. and 5.5 – 9% aboveground carbon in midwestern forests of Michigan and Ohio (Flower et al., 2013). Therefore the impacts of EAB can devastate forest communities in areas where ash comprises a large portion of the forests.

The use of systemic insecticides, such as emamectin benzoate and imidacloprid, are highly effective at protecting individual ash trees. Imidacloprid and emamectin benzoate are the two most common insecticides used to protect ash trees against emerald ash borer (Smitley et al., 2010; McCullough et al., 2011). Imidacloprid is a systemic neo- nicotinoid insecticide that is applied either by soil drench or direct trunk injection.

Emamectin benzoate is an avermectin, which causes insect paralysis by disrupting neurotransmitters (US EPA, 2009). Emamectin benzoate is capable of killing larvae within the trees and can remain effective for two to three years after a single treatment

(Smitley et al., 2010; McCullough et al., 2011).

Large-scale applications of insecticides in forest habitats are currently being implemented against emerald ash borer. In particular, the Five Rivers MetroParks

(FRMP) system in Dayton, Ohio started an insecticide program in 2011 to treated 600 mature ash trees with emamectin benzoate every two years. These parks consist of natural forests with five species of ash; green ash, white ash, blue ash, black ash, and pumpkin ash (F. profunda). Managers at FRMP selected trees from all five species to treat with

TREE-äge (Arborjet, Inc). This provides an opportunity to measure the effects of EAB on different ash species in a natural environment and the efficacy of a large-scale insecticide program for conserving ash in a natural forest. The objectives of this study were to (1) determine the extent of canopy thinning and percent survival (termed impact of EAB) for

29 five ash species at FRMP, (2) determine whether insecticide treatments reduce EAB- induced ash mortality on each species and (3) if the efficacy of insecticide treatment changed as the impact of EAB increases.

Methods

Study Sites

Five Rivers MetroParks (FRMP) are a series of 18 parks in Montgomery, Greene, and Warren counties (southwestern Ohio). Our study was conducted at six of the metroparks, each with naturally regenerating, contiguous forests. Four parks were located in Montgomery County; Englewood, Taylorsville, Cox Arboretum, and Germantown.

Twin Creek was split between southern Montgomery County and northwestern Warren

County. Lastly, Sugarcreek was located in Greene County (eastern border with

Montgomery County). Forest habitat covered approximately 35 ha (50%) of Cox

Arboretum, 93 ha (17%) of Taylorsville, 218 ha (87%) of Sugarcreek, 315 ha (80%) of

Twin Creek, and 495 ha (82%) of Germantown Metroparks. Germantown and Twin

Creek Metroparks mainly consisted of mid to late successional forests along steep ridges of the Twin Creek valley: soils were mainly xeric (dry, upland). Cox Arboretum and

Taylorsville Metroparks consisted of mid-successional forests; soils were a mix of mesic to xeric. Englewood Metropark was early to mid-successional forest with a large swamp

(“Pumpkin Ash Swamp”) covering approximately 7% of the park. Lastly, Sugarcreek

Metropark was the youngest forest and mainly consisted of early-successional species.

30

From June to July 2011, FRMP treated 600 ash trees with TREE-äge (emamectin benzoate 4% ME, Arborjet, Woburn, MA, USA). All ash trees were trunk-injected using the Arborjet Quik-Jet system. FRMP retreated all trees in 2013 and 2015 to maintain efficacy of insecticide treatments. Most treated trees were within 10 m of hiking trails and many were located in close proximity to other treated ash trees. Treated trees ranged from 10 – 200 cm trunk diameter at breast height (dbh) and included individuals from all five ash species. The ash population in southwestern Ohio consisted of white ash, green ash, blue ash, pumpkin ash, and black ash. White, green, and blue ash trees were present in all six metroparks, but pumpkin and black ash occurred only in Englewood Metropark in the “Pumpkin Ash Swamp”. EAB was detected in Warren and Miami Counties in 2006 and Montgomery Counties in 2007 (emeraldashborer.info). EAB was first detected in

2006 south of Twin Creek MetroPark in northern Warren County and east of Taylorsville

MetroPark in southwestern Miami County. EAB was detected in Montgomery County in

2007 at Cox Arboretum (emeraldashborer.info). EAB detection is less clear at the remaining parks, because new detections were not reported after EAB was detected in each county.

In 2014, I established 24 quadrats (100 m X 100 m; 1 ha) at the six metroparks.

The number of quadrats ranged from three to six per park depending on the forested area of the park. Quadrats were randomly located within each park using ArcMap 10.4 and a measured grid. Within each quadrat, I located four 0.1 ha circular plots; two proximal plots were within 50 m of the quadrat center and two distal plots were established 150 m away from the center on a randomly selected azimuth (degree bearing). The second distal

31 plot was established at approximately 90 degrees clockwise from the first distal plot; also

150 m from the quadrat center. Each plot was marked by a center tree (non-ash tree ≥ 10 cm diameter at breast height; dbh) and contained at least two mature ash trees (≥ 10 cm dbh).

Habitat characterization

Within each plot, I measured the following community variables: tree community composition, species richness, total tree density, total ash density, total ash basal area cm per m2, relative density, relative dominance, and ash importance. I counted and identified all trees ≥ 10 cm dbh within each plot. Diameter at breast height was used to calculate basal area of each tree and total basal area for each species or genus per m2 (Equation 1).

Additionally, I calculated relative density, relative dominance, and importance value for each species or genus per ha (Equation 2-4).

휋 퐷퐵퐻 퐵퐴 푐푚/푚2 = ∑푛 ( )2 Equation 1 𝑖=1 10000 2

푇표푡푎푙 푑푒푛푠𝑖푡푦 표푓 푠푝푒푐𝑖푒푠 𝑖 푝푒푟 ℎ푎 푅푒푙푎푡푖푣푒 퐷푒푛푠푖푡푦 = ∗ 100 Equation 2 푇표푡푎푙 푡푟푒푒 푑푒푛푠𝑖푡푦 푝푒푟 ℎ푎

푇표푡푎푙 퐵퐴 표푓 푠푝푒푐𝑖푒푠 𝑖 푅푒푙푎푡푖푣푒 퐷표푚푖푛푎푛푐푒 = ∗ 100 Equation 3 푇표푡푎푙 퐵퐴 표푓 푎푙푙 푡푟푒푒푠 푝푒푟 ℎ푎

푅푒푙.퐷푒푛.+푅푒푙.퐷표푚 퐼푉 = Equation 4 2

Where i indicated each species or genus and n indicated the total number of species or genus per plot.

32

I also estimated ash ground cover (%), herbaceous (non-woody) vegetation cover

(%), herbaceous and woody vegetation height (m), slope, and aspect. I limited vegetation cover and vegetation height to stems ≤ 1.5 m tall (i.e., breast height). Woody vegetation height was estimated for all woody or seedlings ≤ 1.5 m tall. I estimated herbaceous and woody vegetation height by recording the median height of all stems inside of four microplots within each plot. Slope was categorized from 0-2, 2-5, 6-9, 10-

14, and >25 based on steepness and aspect of the slope was measured using a compass.

Aspect degrees were converted to radians and then transformed to reflect site productivity

(Beers et al. 1966).

Canopy health and mortality of ash trees

Ash trees were identified to species and by treatment (treated or untreated). I recorded species and counted all untreated ash trees per 0.1 ha plot and treated ash trees within each park. Additionally, I measured the dbh, recorded signs of EAB infestation, and assessed the canopy health of each ash tree. Ash trees were identified to species using bark characteristics and/or seeds, if present. Many white and green ash trees had extensive woodpecker damage from EAB infestation and were identified as Fraxinus spp.

Signs of EAB infestation included woodpecker damage (predation holes and patches of bark scraped off), D-shaped exit holes, and epicormic branches. Canopy health was recorded using a rating scale from 1 to 5: 1 indicated a healthy trees with a full canopy, 2-

4 were increasing degrees of canopy thinning, and 5 for a dead tree (Smith, 2006; Smith et al., 2015).

33

Statistical analyses

I conducted a non-metric multidimensional scaling (NMDS) analysis to compare habitat characteristics and tree community data of each plot. Environmental variables included in this analysis were species richness, total tree density, total basal area, presence of standing water (hydric), slope, and herbaceous vegetation height. Plots were categorized into groups (presence/absence) based on total basal area; < 80, 80-175, >175 cm/m2, slope: 0-2, 2-5,6-9, 10-24, >25, and herbaceous vegetation height; ≤ 10, 11-20,

21-30, ≥ 31 cm. Additionally, I tested the NMDS ordination using the adonis function to determine whether the forest communities differed by park. These analyses were conducted using R 3.2.3 and significant differences of each group were defined as P <

0.05.

Data was recorded during the ash growing season (June to August) from 2014 to

2016 to document change in ash canopy health and percent mortality over time. I compared this pattern by treatment (insecticide treated vs. untreated) for each species. I calculated average dbh, canopy rating, and percent survival for each ash species by treatment. Percent survival per plot was calculated as the number of living ash trees

(canopy rating 1 – 4) divided by the total number of individuals of that species. Pumpkin and black ashes were present only in one metropark, so I did not conduct statistical analyses on the differences between treated and untreated individuals. Additionally, I pooled data for green ash and white ashes, because a large proportion of trees were indistinguishable due to extensive woodpecker damage and because the pattern of decline

34 and mortality was similar for both species. I compared average canopy rating and percent ash survival of treated and untreated blue ash and green-white ash. I used non-parametric

ANOVA-type statistics (ATS) for ordinal scale longitudinal data (2014 – 2016) to compare ratings, as described by Brunner et al. (2002). ATS statistics were conducted using nparLD package in R 3.2.3 (Noguchi et al., 2012). For comparing percent ash survival of treated and untreated ashes, I used generalized linear models with binomial distribution with the MASS package in R 3.2.3. Statistical significance was set at P ≤

0.05.

Results

Five ash species were present in these parks (F. pennsylvanica; green ash, F. americana; white ash, F. quadrangulata; blue ash, F. nigra; black ash, and F. profunda; pumpkin ash). Green ash, white ash, and blue ash trees were found in all six metroparks, but black ash and pumpkin ash were isolated to Englewood Metropark in the Pumpkin

Ash Swamp.

Ash trees were abundant and important components of the tree communities across the study sites. Ash density was highest in Cox Arboretum, Englewood, and

Sugarcreek Metroparks (Table 2.1). Similarly, relative density of ash was high (>20%) in

Cox Arboretum, Englewood, and Sugarcreek Metroparks (Table 2.1). Relative dominance was high (>23%) in all parks except for Taylorsville Metropark (Table 2.1).

Species richness was similar across parks, ranging on average from 9 to 12 species per plot.

35

The tree community at Five Rivers MetroParks consisted of primarily ashes

(Fraxinus, 96 plots), maples (Acer, 91 plots), and oaks (Quercus, 62 plots). The genus

Acer (maples) had the highest relative density (39.5 ± 2.4) and relative dominance (32.0

± 2.1). Acer saccharum was the most common maple species and was present in 88 plots.

Fraxinus was the second most important genus, with high relative density (19.6 ± 1.3) and relative dominance (23.6 ± 1.2). The most common ash species were F. pennsylvanica and F. americana (92 plots) and F. quadrangulata (37 plots). Quercus

(oaks) had lower relative density (7.0 ± 1.0) and relative dominance (9.9 ± 1.3), but were the third most important genus. Of the oak species, Q. rubra, Q. macrocarpa, and Q. alba were the most common (present in 35, 26, and 19 plots; respectively). Other common tree species included Prunus serotina, Aesculus glabra, Celtis occidentalis, Ulmus americana, and Carya spp.

A significant three-dimensional NMDS of forest community composition

(R2=0.96, Stress = 0.195, P<0.0001) indicated strong correlations for sites characterized as hydric, high species richness (S), high tree density (Treeden), low and high basal area

(Low.BA and High.BA, respectively), and 0 slope (Slope0) (Table 2.2). Forest communities at Englewood, Sugarcreek, and Taylorsville Metroparks had high similarity within each park (Figure 2.1a). Sugarcreek consisted of sites characterized with high tree density, flat terrain (slope = 0), and tall vegetation (Figure 2.1a). Additionally, Acer negundo, pomifera, and Robinia pseudoacacia were commonly found at

Sugarcreek Metropark (Figure 2.1b). Taylorsville Metropark consisted primarily of sites with low basal area and higher slope (≥ 25°) (Figure 2.1a). Englewood Metropark

36 consisted of two distinct habitat types, a hydric habitat (the Pumpkin Ash Swamp) and habitat similar to Taylorsville (Figure 2.1a). In the Pumpkin Ash Swamp of Englewood, there was high species richness and the forest community was dominated by pumpkin ash and black ash. The second Englewood habitat consisted of primarily green ash, white ash, and blue ash. The remaining parks, Germantown, Twin Creek, and Cox Arboretum

Metroparks, consisted of forest communities with less within park similarity, but were similar across parks. Overall, habitat characteristics and tree communities differed across the six metroparks (F = 2.94, df = 5, R2= 0.14, P=0.0001).

37

Table 2.1: Summary of tree communities and ash importance within each park surveyed in 2014 within the Five Rivers MetroParks in Dayton, OH. Data is shown as mean ± standard error.

Species Total Relative Relative # of plots Ash/ha IV richness trees/ha Density Dominance Cox 12 11.8 ± 0.7 477 ± 22 111 ± 17 22.9 ± 3.1 25.3 ± 2.8 24.1 ± 3.0 Arboretum Englewood 12 10.2 ± 0.6 528 ± 38 141 ± 35 25.2 ± 4.8 28.6 ± 4.5 26.9 ± 4.6

38

Germantown 24 10.0 ± 0.6 420 ± 31 83 ± 11 18.5 ± 1.7 23.7 ± 2.1 21.1 ± 1.8

Sugarcreek 12 9.5 ± 0.7 465 ± 23 113 ± 34 22.8 ± 6.0 24.0 ± 4.5 23.4 ± 5.1 Taylorsville 16 8.8 ± 0.5 399 ± 29 57 ± 10 14.4 ± 2.7 17.2 ± 2.6 15.8 ± 2.5 Twin Creek 20 10.4 ± 0.5 440 ± 21 79 ± 11 17.2 ± 1.9 23.7 ± 2.4 20.4 ± 2.1

38

Table 2.2: Ordination of forest communities and environmental variables using non-metric multidimensional scaling (NMDS) analysis. Samples collected from forest communities at Five Rivers MetroParks in Dayton, OH from June to August 2016. NDMS statistical summary: R2=0.96, Stress = 0.195, P<0.0001.

Environmental NMDS1 NMDS2 R2 P variables Tree basal area (BA) Low BA 0.460 -0.888 0.08 0.03 * Mid BA -0.964 0.267 0.04 0.13

High BA 0.344 0.939 0.07 0.04 * Hydric (plots with -0.412 -0.909 0.07 0.03 * standing water) Tree density -0.978 0.210 0.12 0.002 ** (Treeden) Species richness (S) -0.333 0.943 0.40 0.001 *** Slope 0-1 (Slope0) -0.972 -0.234 0.14 0.001 *** 2-5 0.960 0.279 0.01 0.77

6-9 0.999 -0.036 0.003 0.87

10-24 0.657 0.754 0.02 0.35

≥ 25 (Slope25) 0.926 -0.367 0.01 0.53

Herbaceous veg height (cm) hh.10 0.709 0.705 0.03 0.23 hh.20 0.925 -0.380 0.01 0.55 hh.30 0.458 0.889 0.003 0.86 hh.40 -0.911 -0.412 0.11 0.005 **

39

A B

40

Figure 2.1: Non-metric multidimensional scaling analysis of the tree communities and the corresponding environmental variables within natural forests at Five Rivers MetroParks in Dayton, OH. Species data was collected in 2014. Vegetation height and cover, slope, and aspect were collected in 2016. (A) Community characteristics with environmental variables. (B) Tree species with environmental variables. See Table 2.2 for descriptions of environmental variables (vectors). NDMS statistical summary: R2=0.96, Stress = 0.195, P<0.0001.

40

Ash health and survival of each species

I sampled 1,145 ash trees from six metroparks within the Five Rivers MetroParks

(FRMP). Of these, 251 were treated with emamectin benzoate and 894 trees received no insecticide treatments. The species distribution of untreated ash consisted of green ash

(45%), white ash (10%), unidentified Fraxinus spp. (26%), blue ash (16%), pumpkin ash

(2.2%), and black ash (0.8%) (Table 2.3). The species distribution of treated ash trees consisted of green ash (30%), white ash (25%), unidentified Fraxinus spp. (13%), blue ash (14%), pumpkin ash (11%), and black ash (6%) (Table 2.3). Green ash, white ash, and blue ash were the most common species and were found in all parks. The other two species, black ash and pumpkin ash, were only present at Englewood Metropark in the

Pumpkin Ash Swamp. Due to the restricted distributions of these species, FRMP managers treated more than half of the pumpkin and black ash trees within this area.

Overall species distributions of treated and untreated ash trees were similar. On average, the treated ash trees were larger than untreated ash trees. In particular, treated green ash, white ash, and blue ash were 9 – 16 cm larger than untreated conspecifics (Table 2.3).

This may indicate that FRMP was biased towards protecting larger ash trees.

Untreated green ash, white ash, and pumpkin ash declined from 2014 to 2016, mostly declining (rating =3) in 2014 to dying or dead (rating = 4 or 5) in 2016 (P<

0.0001) (Figure 2.2a). Canopy ratings of untreated black ash remained between 4 and 5 throughout this study (Figure 2.2a). Unlike the other species, blue ash remained healthy

(ratings = 1 or 2) from 2014 to 2016 (Figure 2.2a). Overall, untreated blue ash had significantly lower canopy ratings than all other species (P<0.001). Canopy ratings of

41 treated ash trees remained healthy throughout the study, with no differences among species (Figure 2.2b).

Survival of untreated ash trees followed a trend similar to that of canopy health.

Untreated green ash and white ash survival decreased from approximately 50% ash living in 2014 to < 20% ash living in 2016 (Figure 2.2c). Pumpkin ash survival declined rapidly from 92% to 8% from 2014 to 2016, respectively (Figure 2.2c). Untreated black ash survival remained low throughout the study. Lastly, blue ash survival remained high

(>90%) from 2014 to 2016 (Figure 2.2c).

All species of treated ash trees had high survival from 2014 to 2016. No differences were detected among species of treated ash (Figure 2.2d). Additionally, survival of treated and untreated blue ash trees was similarly high throughout the study

(Figure 2.2c&d). Survival of treated green ash, white ash, pumpkin ash, and black ash trees was higher than untreated conspecifics (Figure 2.2c&d).

42

Table 2.3: Ash species abundance and size of untreated and treated ash trees sampled in 2014 from six parks within the Five Rivers MetroParks in Dayton, OH. Data represent the total trees sampled in this study.

No. % DBH Species trees abundance (± SE) Untreated ash F. pennsylvanica 411 46% 24.9 ± 0.6 Green ash F. americana 86 10% 45.7 ± 2.1 White ash Fraxinus spp. 230 26% 30.4 ± 1.1 Green or white ash F. quadrangulata 140 16% 27.1 ± 1.3 Blue ash F. profunda 20 2% 24.7 ± 2.6 Pumpkin ash F. nigra 7 1% 22.5 ± 3.5 Black ash Total 894 100%

Treated ash F. pennsylvanica 76 30% 36.6 ± 1.5 Green ash F. americana 63 25% 54.5 ± 1.9 White ash Fraxinus spp. 32 13% 46.7 ± 2.3 Green or white ash F. quadrangulata 36 14% 43.3 ± 2.6 Blue ash F. profunda 28 11% 31.3 ± 1.9 Pumpkin ash F. nigra 16 6% 26.6 ± 2.1 Black ash Total 251 100%

43

Untreated ash Treated ash

5 A B Green ash White ash Green/white ash Blue ash 4 Pumpkin ash Black ash

3

2

Ash canopy rating Ash canopy

1

0 C 2014 2015 2016 D 2014 2015 2016 100 Year Year

80

60

40

Percentage of ash survival 20

0

2014 2015 2016 2014 2015 2016 Year Year

Figure 2.2: Canopy decline and survival of untreated ash (left) and treated ash (right) at the Five Rivers MetroParks in Dayton, OH. (A-B) Show average canopy decline (mean ± SE) from 1 to 5; 1 indicated healthy ash, 2-4 indicated varying degrees of canopy decline, and 5 indicated dead ash. (C-D) Show percent survival of ash (mean ± SE).

44

Impact of EAB at Five Rivers MetroParks

I compared treated and untreated ash trees at the six metroparks to determine the extent of the EAB impact across FRMP. For this analysis, I excluded pumpkin ash and black ash, because these species were present at only Englewood Metropark. I combined green ash, white ash, and unidentified Fraxinus spp, (collectively termed “green-white ash”) since there were no overall differences between their mean canopy ratings and percent mortality.

Blue ash trees were present in all six metroparks. However, this species was most abundant in Englewood and Taylorsville Metroparks with 11 total treated ashes, and approximately 50 untreated ashes in both parks (Table 2.4). Blue ash was frequent, but at lower densities in Germantown, Cox Arboretum, and Twin Creek. Blue ash was least common at Sugarcreek, with only one tree untreated tree reported (Table 2.4).

Green-white ash trees were common throughout the study plots. In particular, the density of green-white ash was highest in Sugarcreek and Cox Arboretum, with ≥ 100 trees per ha (Table 2.4). Green-white ash densities were lowest at Taylorsville Metropark

(30.0 ± 4.4 ash/ha) (Table 2.4).

Size of treated trees for both blue ash and green-white ash, were larger than untreated ashes (Table 2.4). FRMP selected primarily ash trees along trails to treat with

TREE-äge. Thus, treated ash trees were not a random subset of the population. The size and species composition of untreated ash trees were more reflective of the ash population within each park.

45

Blue ash trees showed little EAB-induced decline or mortality throughout the study. Canopy ratings of treated and untreated blue ashes did not differ and remained healthy (rating=1 or 2) from 2014 to 2016 (Figure 2.3a&b). In 2015 and 2016, the untreated blue ash trees at Cox Arboretum had mean canopy ratings above 2 (i.e., healthy, but showing some signs of EAB infestation), however these differences were not significantly different from the other parks.

Untreated green-white ash trees showed high levels of EAB-induced decline and mortality by 2016. Specifically, Cox Arboretum, Taylorsville, and Twin Creek were mostly dying or dead in 2014 (Figure 2.3c). At Sugarcreek, Englewood, and

Germantown, untreated green-white ashes significantly declined from 2014 to (P <

0.0001) (Figure 2.3c). Sugarcreek was the only park with mostly living ash trees in 2016, however mean canopy rating was 3 (i.e., thinning canopies and signs of EAB infestation).

Treated green-white ashes were healthier than untreated green-white ash and did not differ among parks from 2014 to 2016 (Figure 2.3d).

Percentage of ash survival of blue ash remained high throughout the study for both untreated and treated trees (Figure 2.4a&b). However, there were some differences in the survival of untreated blue ashes at each park. Specifically, at Cox Arboretum mean percentage survival was 95 ± 5% in 2014, and 76 ± 14% in both 2015 and 2016, and at

Twin Creek, percentage survival was 83 ± 17% from 2014 to 2016 (Figure 2.4b). At the remaining four parks, mean percentage survival was > 95% from 2014 to 2016. This suggests that survival of blue ash was more variable at these Cox Arboretum and Twin

46

Creek. All treated ash trees were alive throughout the study and there were no differences across parks.

Survival of treated green-white ashes was higher than for untreated trees and did not change across parks (P<0.0001) (Figure 2.4c&d). However, survival of untreated green-white ashes considerably varied by park (Figure 2.4c). Specifically, in 2014 survival of untreated green-white ashes was < 25% at Cox Arboretum, Taylorsville, and

Twin Creek, and was > 70% at Englewood, Germantown, and Sugarcreek (Figure 2.4c).

By 2016, untreated green-white ash survival was ≤ 15% at Cox Arboretum, Englewood,

Germantown, Taylorsville, and Twin Creek (Figure 2.4c). Survival at Sugarcreek

Metropark decreased at a slower rate, from 91% in 2014 to 53% in 2016 (Figure 2.4c).

Overall, the pattern of ash survival corresponded with the average canopy ratings for these parks. However, when I began this study three of the parks (Englewood,

Germantown, and Sugarcreek) consisted of mostly living green-white ash trees that were showing some signs of EAB infestation, while the other three parks (Cox Arboretum,

Taylorsville, and Twin Creek) consisted of mostly dead green-white ash trees with the few living trees showing severe canopy decline. This suggests that EAB impact was higher at Cox Arboretum, Taylorsville, and Twin Creek Metroparks, and the EAB impact was lower at Englewood, Germantown, and Sugarcreek.

47

Table 2.4: Abundance and size of blue ash and green-white ash trees (treated and untreated) in 2014 at six parks from the Five Rivers MetroParks in Dayton, OH. Data for green and white ash trees were combined due to similar canopy health and survival.

Diameter at breast height Relative Density of Total treated (mean ± SE cm) # of plots frequency ash/ha / park (%) (mean ± SE) Treated Untreated

Blue ash Cox Arboretum 12 58.3 8 18.6 ± 3.4 47.3 ± 7.2 29.9 ± 4.7 Englewood 12 66.7 11 48.8 ± 16.6 37.9 ± 2.7 25.0 ± 2.2 48 Germantown 24 25 4 30.0 ± 8.2 32.7 ± 3.7 28.8 ± 2.7

Sugarcreek 12 8.3 0 10.0 ± 0 - 15.6 Taylorsville 16 62.5 11 54.0 ± 12.8 49.7 ± 5.2 27.7 ± 2.1 Twin Creek 20 30 2 13.3 ± 3.3 41.8 ± 3.3 26.3 ± 6.5

Green-white ash Cox Arboretum 12 100 13 99.2 ± 18.0 44.5 ± 2.2 27.5 ± 1.6 Englewood 12 100 24 76.7 ± 33.7 43.1 ± 2.8 26.4 ± 1.5 Germantown 24 100 59 74.2 ± 10.4 41.0 ± 2.0 31.4 ± 1.3 Sugarcreek 12 100 18 113.3 ± 24.0 52.1 ± 4.7 21.7 ± 1.1 Taylorsville 16 75 18 30.0 ± 4.4 54.7 ± 4.9 36.7 ± 2.8 Twin Creek 20 100 39 75.0 ± 10.9 44.8 ± 2.2 34.7 ± 1.2

48

5 A B CA EN GT 4 SC TA TC 3

(Blue ash) (Blue 2

Ash canopy rating Ash canopy 1

5 C D

4

3

2

(Green-white ash) (Green-white

Ash canopy rating Ash canopy 1

2014 2015 2016 2014 2015 2016

Year Year

Figure 2.3: Ash canopy ratings for untreated blue ash (A), treated blue ash (B), untreated green-white ash (C), and treated green-white ash (D) from six parks at Five Rivers MetroParks in Dayton, OH. CA: Cox Arboretum, EN: Englewood, GT: Germantown, SC: Sugarcreek, TA: Taylorsville, and TC: Twin Creek Metroparks. Ratings were from 1-5; 1 for healthy canopies, 2-4 for varying degrees of canopy openness, and 5 for dead ash. Data shown as mean canopy rating ± SE per plot.

49

100 A B CA EN GT 80 SC TA TC 60

(Blue ash) (Blue 40

20

Percentage ash survival Percentage 0 100 C D

80

60

40

(Green-white ash) (Green-white 20

Percentage ash survival ash survival Percentage 0 2014 2015 2016 2014 2015 2016

Year Year

Figure 2.4: Percent ash survival for untreated blue ash (A), treated blue ash (B), untreated green-white ash (C), and treated green-white ash (D) from six parks at Five Rivers MetroParks in Dayton, OH. CA: Cox Arboretum, EN: Englewood, GT: Germantown, SC: Sugarcreek, TA: Taylorsville, and TC: Twin Creek Metroparks. Data shown as mean percent survival ± SE per plot.

50

Discussion

Population structure

Five species of ash were found at FRMP. White ash, green ash, and blue ash were widespread and found in all six parks. Black ash and pumpkin ash were limited to a small swamp habitat in Englewood MetroPark. The percentage of treated and untreated ash trees was similar for each species, suggesting that FRMP selected a similar proportion of individuals from each species to be treated with TREE- äge. Treated ash trees were, on average, larger than untreated ash trees, indicating a possible bias towards selecting larger trees to treat. Furthermore, trees selected for treatments were easily accessible, often along hiking trails, and thus not a random sample of the ash population. Furthermore, since these trees are near hiking trails, they had more access to light than trees in the interior of the forest, possibly resulting in faster growth (Koch et al., 2004). Although the selection of these trees was not random, comparisons of ash health and survival between treated and untreated ash over time provides an indication of whether insecticides protect ash trees, the differential susceptibility of ash species to EAB, and the change overall change in the health and survival of the ash population from 2011 to 2016.

Ash species were common throughout the forests of the Five Rivers MetroParks

(FRMP) and were consistently dominant or co-dominant throughout the study sites.

Maples (Acer spp.) were the most common and were co-dominant with ashes. Overall, ash was an important canopy tree throughout FRMP, which suggests that 1) available resources for EAB were high and 2) the effects of widespread EAB-induced mortality could create major changes to forest community dynamics.

51

The insecticide treatments protected green ash, white ash, pumpkin ash, and black ash trees. Specifically, in 2016 most untreated green, white, pumpkin, and black ashes were either dead or dying from EAB infestation and nearly all treated conspecifics were healthy and alive. These results confirm the efficacy of TREE-äge for providing direct protection of ash trees against EAB (Smitley et al., 2010; McCullough et al., 2011).

On the other hand, blue ashes remained healthy and living throughout the study with or without insecticidal protection. Other studies have reported that blue ash may be more resistant or less susceptible to EAB infestation than green, white, and black ash

(Anulewicz et al. 2007, Carson 2013, Tanis and McCullough 2012; Tanis and

McCullough 2015, Herms 2015). Anulewicz et al. (2007) reported a hierarchy of EAB colonization between green ash, white ash, and blue ash. They found that EAB colonization in white ashes increased as green ash mortality increased and colonization of blue ash increased as white ash died (Anulewicz et al. 2007). Herms (2015) also observed that blue ash declined as other species died. Additionally, Tanis and McCullough (2012) resurveyed the plots from Anulewicz et al. and found that after more than 10 years of

EAB infestation 16% and 0% of white ashes were still alive, but 62% and 81% of blue ashes had survived at the two forest sites. Furthermore, they reported that 87% of blue ash trees had evidence of EAB colonization, but in some cases calluses formed in response to larval feeding (Anulewicz et al. 2007, Tanis and McCullough, 2012).

EAB is capable of feeding, surviving, and developing on blue ash. In no-choice tests, there were no differences in the number of larval galleries among black ash, green ash, white ash, and blue ash logs (Anulewicz et al. 2006). In a common garden study,

52 larval densities were highest in green ash and black ash, moderate in white ash, and lowest on blue ash and Manchurian ash (Tanis and McCullough, 2015).

Impact of EAB at Five Rivers MetroParks

Survival and canopy decline of green-white ash differed between treated and untreated individuals. Most treated green-white ashes were still alive and healthy by

2016. However untreated green-white ashes declined overtime and were mostly dead or dying by 2016. There were also different patterns of ash decline and survival across parks. Survival at Cox Arboretum, Twin Creek, and Taylorsville was < 15% in 2014, with the surviving trees showing signs of severe EAB infestation. Survival at Englewood,

Germantown, and Sugarcreek in 2014 were much higher (≥ 75%), but most trees showed signs of EAB infestation. This indicates that the impact of the EAB invasion differed across metroparks. Specifically, this pattern suggests that EAB had been established longer at Cox Arboretum, Taylorsville, and Twin Creek than the other three parks.

Blue ash survival was high and most trees remained healthy throughout the study.

However, at Cox Arboretum and Twin Creek, more untreated blue ash trees showed signs of EAB infestation (i.e., higher canopy ratings) and percentage survival decreased, especially in 2015 and 2016. Previous studies indicated that the likelihood of EAB colonizing blue ash trees increases when most of the green ash and white ash trees are killed by EAB (Anulewicz et al. 2007, Tanis and McCullough, 2012). Therefore, the variation in blue ash survival and canopy decline may be a reflection of the percentage mortality of green-white ash in those parks. Since < 5% of green-white ashes were alive

53 in Cox Arboretum and Twin Creek, the only resources available to EAB would have been blue ash trees.

The different patterns of survival and canopy decline of green-white ash trees may also be an indication of the extent of EAB infestation when the insecticide program started in 2011. Studies show that in areas with known infestations of EAB, near complete ash mortality can occur within six years (Knight et al., 2013; Klooster et al.,

2014). Although EAB was detected in 2007 in Montgomery County, OH, EAB populations had likely been established for much longer. In particular, EAB populations had likely been established longest at Cox Arboretum, Taylorsville, and Twin Creek than at Englewood, Germantown, and Sugarcreek. This is shown with both survival and canopy decline of green-white ash trees and blue ash trees in these parks.

Conclusion

This study showed different susceptibility of five species of North American ash to EAB infestation and the protection that insecticide treatments can provide to these species. Green ash, white ash, pumpkin ash, and black ash had low survival and exhibited obvious signs of EAB infestation throughout this study. However, blue ash survival remained high and trees showed few signs of EAB infestation. Green ash, white ash, pumpkin ash, and black ash trees that had been treated with TREE-äge were healthy and alive throughout the study, indicating that emamectin benzoate can be used to directly protect these species against EAB infestation.

54

Additionally, we found that the impact of EAB differed among parks. This was indicated by lower percentage survival and high canopy ratings in 2014 at Cox

Arboretum, Taylorsville, and Twin Creek than at Englewood, Germantown, and

Sugarcreek Metroparks. Furthermore, canopy ratings and survival of blue ash trees varied within Cox Arboretum and Twin Creek, which was likely due to the loss of green and white ash within these parks.

Literature cited

Anulewicz, A.C., McCullough, D.G., Cappaert, D.L. 2007. Emerald ash borer (Agrilus planipennis) density and canopy dieback in three North American ash species. Arboric Urban For. 33: 338-349. Anulewicz, A.C., McCullough, D.G., Miller, D.L. 2006. Oviposition and development of emerald ash borer (Agrilus planipennis) (Coleoptera: Buprestidae) on hosts and potential hosts in no-choice bioassays. Gt Lakes Entomol. 39: 99-112. Anulewicz, A.C., McCullough, D.G., Cappaert, D.L., Poland, T.M. 2008. Host range of the Emerald ash borer (Agrilus planipennis Fairmaire) (Coleoptera: Buprestidae) in North America: Results of multiple-choice field experiments. Environ. Entomol. 37: 230-241. Beers, T.W., Dress, P.E., Wensel, L.C. 1966. Aspect transformation in site productivity research. J. For. 64: 691. Brunner, E, Domhof, S., Langer, F. 2002. Nonparametric analysis of longitudinal data in factorial experiments. John Wiley & Son, New York, NY. Carson, S.L.E. 2013. Spatiotemporal dynamics and host selection of emerald ash borer, Agrilus planipennis (Coleoptera: Buprestidae), at Point Pelee National Park, Ontario, Canada. The University of Guelph, Dissertation. p. 190. Flower, C.E., Knight, K.S., Gonzalez-Meler, M.A. 2013. Impacts of the emerald ash borer (Agrilus planipennis Fairmaire) induced ash (Fraxinus spp.) mortality on forest carbon cycling and successional dynamics in the eastern United States. Biol. Invasions 15: 931-944. Gandhi, K.J.K., Herms, D.A. 2010a. Direct and indirect effects of alien insect herbivores on ecological processes and interactions in forests of eastern North America. Biol. Invasions 12: 389-405. Gandhi, K.J.K., Herms, D.A. 2010b. North American arthropods at risk due to widespread Fraxinus mortality caused by the Alien emerald ash borer. Biol. Invasions 12: 1839-1846.

55

Herms, D.A. 2015. Host range and host resistance. In: Van Driesche, R.G., Reardon, R., eds. Biology and control of emerald ash borer, Technical Bulletin FHTET 2014- 09. Morgantown, WV, USA: USDA Forest Service, 65-73. Klooster, W.S., Herms, D.A., Knight, K.S., Herms, C.P., McCullough, D.G., Smith, A., Gandhi, K.J.K., Cardina, J. 2014. Ash (Fraxinus spp.) mortality, regeneration, and seed bank dynamics in mixed hardwood forests following invasion by emerald ash borer (Agrilus planipennis). Biol. Invasions 16: 859-873. Knight, K.S., Brown, J.P., Long, R.P. 2013. Factors affecting the survival of ash (Fraxinus spp.) trees infested by emerald ash borer (Agrilus planipennis). Biol. Invasions 15: 371-383. Koch, G.W., Sillett, S.C., Jennings, G.M., Davis, S.D. 2004. The limits to tree height. Nature 428: 851-854. Liu, H., L.S. Bauer, R. Gao, T. Zhao, T.R. Petrice, R.A. Haack. 2003. Exploratory survey for the emerald ash borer, Agrilus planipennis (Coleoptera: Buprestidae), and its natural enemies in China. Gt. Lakes Entomol. 36: 191-204. Lovett, G.M., Canham, C.D., Arthur, M.A., Weathers, K.C., Fitzhugh, R.D. 2006. Forest ecosystem responses to exotic pests and pathogens in eastern North America. Bioscience 56: 395-405. MacFarlane, D.W., Meyer, S.P. 2005. Characteristics and distribution of potential ash tree hosts for emerald ash borer. For. Ecol. Manag. 213: 15-24. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. Niesenbaum, R.A. 1992. The effects of light environment on herbivory and growth in the dioecious shrub Lindera benzoin (Lauraceae). Am. Midl. Nat. 128: 270-275. Noguchi, K., Gel, Y.R., Brunner, E., Konietschke, F. 2012. nparLD: An R software package for the nonparametric analysis of longitudinal data in factorial experiments. J. Stat. Softw. 50: 1-23. Perry, K.I., Herms, D.A. 2016a. Response of the forest floor invertebrate community to canopy gap formation caused by early stages of emerald ash borer-induced ash mortality. For. Ecol. Manag. 375: 259-267. Perry, K.I., Herms,D.A. 2016b. Short-term responses of ground beetles to forest changes caused by early stages of emerald ash borer (Coleoptera: Buprestidae)-induced ash mortality. Environ. Entomol. 45: 616-626. Pureswaran, D.S., Poland, T.M. 2009. Host Selection and Feeding Preference of Agrilus planipennis (Coleoptera: Buprestidae) on Ash (Fraxinus spp.). Environ. Entomol. 38: 757-765. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. 56

Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Smitley, D.R., Doccola, J.J., Cox, D.L. 2010. Multiple-year protection of ash trees from emerald ash borer with a single trunk injection of emamectin benzoate, and single-year protection with an imidacloprid basal drench. Arboric. Urban For. 36: 206-211. Tanis, S.R., McCullough, D.G. 2012. Differential persistence of blue ash and white ash following emerald ash borer invasion. Can. J. For. Res. 42: 1542-1550. Tanis, S.R., McCullough, D.G., 2015. Host resistance of five Fraxinus species to Agrilus planipennis (Coleoptera: Buprestidae) and effects of paclobutrazol and fertilization. Environ. Entomol. 44: 287-299. Ulyshen, M.D., Klooster, W.S., Barrington, W.T., Herms, D.A. 2011. Impacts of emerald ash borer-induced tree mortality on leaf litter arthropods and exotic earthworms. Pedobiologia 54: 261-265. United States Environmental Protection Agency. 2009. Memorandum: Ecological risk assessment for emamectin benzoate use as a tree injection insecticide to control arthropod pests. PC Code 122806. US EPA, Washington, DC, USA. Villari, C., Herms, D.A., Whitehill, J.G.A., Cipollini, D., Bonello, P. 2016. Progress and gaps in understanding mechanisms of ash tree resistance to emerald ash borer, a model for wood-boring insects that kill angiosperms. New Phytol. 209: 63-79.

57

Chapter 3: Relationship between survival of untreated green and white ash and local abundance of insecticide treated ash trees: Treated trees provide associational

protection to untreated ash trees

Abstract

The introduction of emerald ash borer (EAB) has drastically affected ash populations in

North America. EAB has killed hundreds of millions of ashes since its introduction nearly 20 years ago. Insecticides, such as emamectin benzoate, can successfully protect ash trees from EAB attack. Additionally, large clusters of insecticide treated trees may slow the progression of ash mortality by reducing the surrounding density of EAB.

Therefore, insecticide treatments may provide associational protection to untreated ashes.

The objectives of this study were to determine the effect of density of treated ash trees on survival of untreated green and white ashes. This study was conducted at Five Rivers

MetroParks (FRMP) in Dayton, Ohio, where a large-scale insecticide program was initiated in 2011 to treat ash trees with emamectin benzoate (TREE-äge, Arborjet, Inc.) every two years. I tested the effect of percentage of ash phloem area treated and EAB impact on percent survival of green and white ash (“green-white ash”). From 2014 –

2016, I found that survival of untreated green-white ashes increased with percentage of ash phloem area treated, but only in plots from parks with low EAB impact. Additionally,

58

I tested the effect of distance to the nearest treated ash on survival of untreated green- white ashes and found that survival was higher when treated ash trees were closer. This relationship strengthened in plots from parks with low EAB impacts with higher percentage of ash phloem area treated. This indicated that associational protection was measureable in plots nearby (within 100 m) treated ash trees and with higher percentage of ash phloem area treated, but only in plots from parks with low EAB impact. Therefore,

EAB densities may have been too high in the parks with high EAB impact to provide associational protection to neighboring untreated ash trees. Alternatively, this may also indicate higher percentages of ash phloem need to be treated in order to increase the effectiveness of providing associational protection.

Introduction

Large-scale disturbances cause wide-spread tree mortality and significant losses of biodiversity and invasions by exotic species are among the most devastating causes of disturbance in forest systems (Lovett et al. 2006). Emerald ash borer (Agrilus planipennis

Fairmaire; EAB), has killed hundreds of millions of ash trees (Fraxinus spp.) throughout eastern North America (Herms and McCullough, 2014). EAB was first detected in North

America in 2002 near Detroit, Michigan (Cappaert et al., 2005), but dendrochronological evidence indicates that the beetle had established by at least the early 1990’s and has been killing ash trees in eastern North American for more than 20 years (Siegert et al.,

2014).

59

Wide-spread ash mortality caused by emerald ash borer creates large, contiguous canopy gaps that can change plant and animal communities (Lovett et al., 2006). Large canopy gaps increase light and temperature, and alter soil properties (Scharenbroch and

Bockheim, 2008a). As a result, disturbance and shade-intolerant plants, arthropods

(Ulyshen et al., 2011; Gandhi et al., 2014; Perry and Herms, 2016a), and nesting birds

(Long, 2013) became more abundant Most notably, Gandhi and Herms (2010) reported

282 phytophagous arthropods were associated with ash trees and 25% of the host specific species were at moderate to high risk of extinction due to emerald ash borer-induced ash mortality. This suggests that the emerald ash borer invasion could cause cascading effects to the plant and animal community that will change the forest structure throughout the invaded range.

Insecticides can effectively protect ash trees from emerald ash borer infestation.

Imidacloprid, a systemic neo-nicotinoid insecticide, and emamectin benzoate (TREE-äge,

Arborjet, Inc.), a systemic avermectin insecticide, are highly effective against emerald ash borer (Smitley et al., 2010a,b; McCullough et al., 2011). Imidacloprid and emamectin benzoate primarily target larvae, but are also effective against adults feeding on foliage

(McCullough et al., 2011). Additionally, a single application of emamectin benzoate can protect trees for 2–3 years (Smitley et al., 2010a).

Studies have suggested that EAB densities are limited by the amount of ash phloem available and if this carrying capacity can be reduced, managers may be able to slow the rate of ash mortality (SLAM program) and the subsequent spread of EAB

(Mercader et al., 2011b; McCullough and Mercader, 2012; Mercader et al., 2015).

60

Simulations conducted by McCullough and Mercader (2012) indicated that in an urban forest, treating 20% of ash trees annually resulted in nearly 100% ash survival over a ten year period (McCullough and Mercader, 2012). On average, EAB can travel up to 400 m per year (Siegert et al., 2010; Taylor et al., 2010). Therefore, as EAB fly into an area with treated ash trees, each treated tree that an adult comes into contact with will be toxic to adults and larvae. Mercader et al. (2015) tested the effect of density of treated ash on larval density from untreated ash, and found that larval densities of EAB decreased in plots that had higher densities of treated ash trees within 800 m. These studies suggest that clustered treatments of ash trees may reduce the EAB density in the surrounding environment and slow the progression of ash mortality in a given area. In other words, ash trees protected by insecticides may provide associational resistance (associational protection) to neighboring untreated ash trees.

The theory of associational resistance states that plants can be protected from herbivory by proximity to other plants (Tahvanainen and Root, 1972). Mechanisms for associational resistance include decreased ability of herbivores to locate their host due to the mixing of plant volatiles of host and non-host plants (Hamback et al., 2000; Sholes,

2008; Jactel et al., 2011), or host plant dilution when non-host plants are more abundant or larger than host plants (Castagneyrol et al., 2013), and perhaps increased production of defensive compounds by host plants in response to soil changes induced by non-host plants (Quiroz et al., 1997; Karban and Maron, 2002; Karban et al., 2006; Ayres et al.,

2007)

61

Studies have suggested correlations between treated ash trees and survival and health of untreated ash trees (McCullough and Mercader, 2012; Mercader et al., 2015).

However, few studies have tested the effect that abundance of treated ash trees has on survival of untreated ash. Additionally, studies have not quantified the spatial relationship

(proximity) and abundance of treated ash trees on the survival of untreated neighboring ash trees. To test this hypothesis that clusters of treated green and white ash trees can provide associational protection of untreated ash trees, I collaborated with the Five Rivers

MetroParks (FRMP) in Montgomery County, Ohio, who began a large-scale insecticide program in 2011. I predicted that sites where treated ashes are more abundant will have higher survival rates of untreated ashes. Additionally, I predicted that this relationship would weaken as distance between treated and untreated ashes increases.

Methods

In 2011, Five Rivers Metroparks (FRMP) near Dayton, OH began an on-going program to treat 600 mature ash trees with TREE-äge insecticide (emamectin benzoate,

4% ME, Arborjet, Woburn, MA, USA). Each treated tree was trunk-injected using the

Arborjet Quik-Jet system. Trees were retreated in 2013 and 2015. I conducted this study at six of the parks within the FRMP network; Cox Arboretum, Englewood, Germantown,

Sugarcreek, Taylorsville, and Twin Creek. Parks were characterized as low or high EAB impact based on percentage survival of untreated green and white ash at the start of this study (2014). Survival at Sugarcreek, Englewood, and Germantown was greater than

75%, therefore I characterized the plots in these parks as low EAB impact. Initial ash

62 survival at the remaining parks, Cox Arboretum, Taylorsville, and Twin Creek, was approximately 25%, so the plots in these parks were identified as high EAB impact.

In 2014, I established 24 quadrats (100 m X 100 m; 1 ha) in six of the metroparks

(Cox Arboretum, Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek).

Quadrats were randomly located within each park using ArcMap 10.4 and a measured grid. Once located, quadrats were categorized as having zero, low, medium, and high density of treated ash trees (0, 2-3, 4-6, ≥ 7 treated ash trees / ha, respectively). Because parks had different areas of forest cover, I established three to six quadrats per park. I also established one control quadrat in each park with no treated trees. Within each quadrat, I located four 0.1 ha circular plots (18-m radius); two proximal plots were within

50 m of the quadrat center and two distal plots 150 m from the center with the first located on a random azimuth and the second approximately 90 degrees clockwise from the first (Figure 3.1a). Each plot consisted of a non-ash center tree (≥ 10 cm diameter at breast height; dbh) and at least two mature ash trees (≥ 10 cm dbh). Replicate plots were independent experimental units.

In 2015, I established five transects per quadrat (Figure 3.1b). All transects were oriented north to south and six points (mapped by GPS) were surveyed along each transect at 20 m intervals (0, 20, 40, 60, 80, and 100 m). At each point, the number and species of ash trees present within 5 m radius were recorded. These data were used to calculated ash density and total ash phloem area per ha. Both ash density and phloem area estimate the total available food resources available to EAB. However, ash phloem area provides a better estimation of resources, because it is a function of the size of ash trees.

63

Therefore, larger ash trees have greater phloem area and therefore can support more EAB larvae. Total living ash phloem area was calculated by converting the average diameter at breast height (dbh) of all living ash trees per quadrat to phloem area (m2) using the following equation (McCullough and Siegert, 2007):

Ash phloem area (m2) = 0.024*dbh2 – 0.307*dbh + 2.63

I estimated the percentage of ash phloem area treated with insecticide by converting the dbh of each treated ash tree to phloem area using the equation above from

McCullough and Siegert (2007). Total treated ash phloem/ha was divided by total ash phloem/ha to estimate percentage of ash phloem area treated.

Figure 3.1: (A) Quadrat design with 4 replicated plots (18-m radius) and (B) Transects established within each quadrat to quantify the effect of local abundance of treated ash trees on survival of untreated green-white ash trees.

64

Survival of untreated green-white ash

I recorded the degree of canopy thinning of all untreated green-white ash trees (≥

10 cm dbh) within each 0.1 ha plot. Canopy thinning was assessed from 2014 to 2016 using a rating scale from 1 to 5 (healthy to dead) (Smith, 2006; Smith et al., 2015). Ash trees rated as 5 were dead and we did not include ash trees with signs of extensive decomposition, since these trees likely died prior to the EAB invasion. Signs of decomposition included absence of branches in the canopy, extensive bark peeling, and wind/lightning damage. I estimated average canopy rating and percentage survival per plot. Percentage ash survival was calculated based on the percentage of ash trees rated as

1 to 4 (i.e., living ash).

I tested the effect of EAB impact and percentage of ash phloem area treated on the percentage ash survival. I tested competing generalized linear models with binomial distribution models with one or both predictor using second-order Akaike’s Information

Criterion (AICc) (Sugiura 1978). Predictor variable were EAB impact (low or high impact) and percentage of ash phloem area treated. The best fit model was selected based on the lowest AICc score. Additionally, I confirmed the fit and significance of the model by calculating pseudo-R2, goodness of fit, and dispersion parameters. Models were tested separately each year to determine whether the pattern of survival changed. Each plot was an independent replicate since the treated ashes were outside of the plot boundaries and there was no interaction between plots in the same quadrat. Statistical analyses were conducted using the Mass package in R 3.2.3. Statistical significance was accepted if P <

0.05.

65

I also tested the effect of distance to the nearest treated ash tree, percentage ash phloem area treated, and EAB impact on survival of untreated green-white ash. To test this, I calculated survival of ash trees within 5 m radius of each transect point (Figure

3.1b). Percentage survival was estimated using the methods described above. Using

ArcMap 10.4, I estimated the distance to the nearest treated ash tree and percentage of ash phloem treated within 400 m from each point. Points without ash trees were excluded from analysis. I tested competing generalized linear models with binomial distribution that included one to three predictor variables using second-order Akaike’s Information

Criterion (AICc) (Sugiura 1978). Predictor variable included distance to nearest treated ash tree, percentage of ash phloem area treated, and EAB impact. The best fit model was selected based on the lowest AICc score. Additionally, I confirmed the fit and significance of the model by calculating pseudo-R2, goodness of fit, and dispersion parameters. Models were tested separately each year (2015 and 2016). Each transect point was an independent sample since distance to treated trees and percentage ash phloem area treated changed based on the location of each point. Statistical analyses were conducted using the Mass package in R 3.2.3. Statistical significance was accepted if P <

0.05.

Results

Density of ash trees varied among parks, with the highest density (trees/ha) at

Englewood, Sugarcreek, and Cox Arboretum Metroparks (Table 3.1). Treated ash trees within 50 ha of each plot were similar among Germantown, Sugarcreek, Cox Arboretum,

66

Taylorsville, and Twin Creek. However, there was substantially larger number of treated ash trees at Englewood Metropark (68 ± 4 trees/50 ha) (Table 3.1), due to the high density of pumpkin and black ash trees that were treated in the Pumpkin Ash Swamp.

Total ash phloem varied among parks, but percentage of ash phloem area treated was similar among parks (Table 3.1). Survival of ash at Englewood, Germantown, and

Sugarcreek Metroparks (i.e., low EAB impact) was higher than Cox Arboretum,

Taylorsville, and Twin Creek Metroparks (i.e., high EAB impact) from throughout the study (2014 – 2016) (Table 3.1). However, in 2015 and 2016, survival at Sugarcreek

Metropark was higher than all other parks (Table 3.1).

Density and percent ash phloem of treated ash trees was similar among parks, but were on a gradient across individual plots. Percentage of ash phloem area treated ranging from 0-3.1% and the distance between plot center and the nearest treated ash tree ranged from 2.3 – 721 m, with the farthest distance for plots located in parks characterized as high EAB impact (Table 3.2).

.

67

Table 3. 1: Comparisons of stand density of ash and treated ash trees from six parks within the Five Rivers Metroparks near Dayton, Ohio. Data collected from 2014 to 2016. Data is shown as mean ± SE for each park.

Ash Treated Treated Total ash % phloem % ash mortality Park density/ha trees/50 ha trees/1 ha phloem area area treated 2014 2015 2016 Low EAB impact 76.3 ± 36.1 ± 5.6 ± Englewood 141 ± 35 68 ± 4 3.8 ± 2.4 2414 ± 130 1.57 ± 0.16 7.9 8.7 4.3

68 74.9 ± 36.1 ± 39.2 ± Germantown 83 ± 11 15 ± 2 2.0 ± 0.5 1058 ± 130 1.22 ± 0.15 7.9 8.7 7.0 90.9 ± 80.0 ± 52.7 ± Sugarcreek 113 ± 34 11 ± 1 1.6 ± 0.7 1080 ± 148 1.40 ± 0.15 2.2 6.1 8.3

High EAB impact Cox 24.6 ± 6.0 ± 3.0 ± 111 ± 17 12 ± 1 1.8 ± 0.5 1581 ± 140 0.68 ± 0.03 Arboretum 8.1 2.4 1.6 27.4 ± 11.9 ± 10.0 ± Taylorsville 57 ± 10 11 ± 2 1.1 ± 0.4 1400 ± 173 1.00 ± 0.26 9.7 6.5 5.6 19.6 ± 8.3 ± 4.7 ± Twin Creek 79 ± 11 11 ± 1 6.1 ± 1.1 1129 ± 84 0.79 ± 0.13 5.9 4.0 2.4

68

Table 3.2: Summary of the density and total phloem area of treated ash trees within parks at the Five Rivers Metroparks near Dayton, OH. Parks were categorized as low or high EAB impact based on percent ash survival at the beginning of this study (2014). All treated ash trees within 400 m radius of each plot (50 ha area) were included. Summary of metrics calculated to assess the local abundance of treated ash trees. Percentage of ash phloem area treated was tested to determine the significance of this variable as a predictor for percentage survival of untreated ash trees.

Low EAB impact High EAB impact

Range Mean Median Range Mean Median

69 Treated ash trees / 50 ha 0 - 87 27.2 ± 3.8 15 0 - 26 11.4 ± 0.8 11

Distance to nearest 4 – 330 76.7 ± 8.0 43.2 2.3 - 721 116.8 ± 17.8 44.1 treated ash tree

Total treated ash 0 - 2616 908 ± 96 737 0 - 1979 544 ± 58 514 phloem area (cm2/ha)

% ash phloem area 0 - 3.1 1.4 ± 0.1 1.2 0 - 3.1 0.8 ± 0.1 0.7 treated

69

Effect of percentage ash phloem treated on associational protection

I tested various competing models to determine whether there was a relationship between survival of untreated green-white ash trees and abundance of treated ash trees. Each model included EAB impact (low and high), year (2014 to 2016), and one of the metrics of local abundance of treated ash trees (density or percent treated ash phloem within 50 ha). I also included a null model (1 + year) to ensure that the model with predictor variables was significantly better than a no effect model.

The best fit model for predicting survival of untreated green-white ashes included the interaction between EAB impact and percent phloem of treated ash trees within 50 ha provided the best fit (AICc = 967.2, ΔAICc = 0, wAICc = 0.92; Table 3.4). I confirmed this model by testing Hosmer and Lemeshow’s goodness of fit (χ2 = 5.15, df = 8, P=0.74) and coefficient of determination (pseudo-R2=0.49). Statistical analysis indicated survival of untreated green-white ash trees significantly decreased each year (Z value = -13.67, P <

0.0001; Table 3.5, Figure 3.3). High EAB impact sites had significantly lower ash survival than low impact sites (Z value = -3.28, P = 0.001; Table 3.5, Figure 3.3). However, percent survival of untreated green-white ashes increased in plots with higher percent treated ash phloem (Z value = 5.86, P < 0.0001; Table 3.5 Figure 3.3).

There was much variation of percent survival from 2014 to 2016 for both EAB impacts. This variation was explained by the best fit model. In low EAB impact, percent survival of untreated green-white ashes increased as the percent treated ash phloem increased (Figure 3.4 A-C). However, the opposite pattern was explained in high EAB

70 impact plots (Figure 3.4 A-C). In other words, percent survival of untreated green-white ashes was lowest in areas with higher percent treated ash phloem (Figure 3.4 A-C). This suggests that treated ashes may not be able to protect untreated ashes in areas with well- established EAB populations.

Overall, the model indicates that survival of untreated green-white ashes has decreased over time in both low and high EAB impacts. However, the chance of survival for untreated ashes increases when the percent treated ash phloem increases. Additionally, this model indicates that treated ash trees up to 400 m away can reduce the impact of EAB on the surrounding area by protecting untreated green-white ashes. However, when the infestation of EAB is higher in an area, increased prevalence of treated ash trees may make untreated trees more susceptible to EAB attack. Whereas, when EAB infestation pressure is low, more untreated green-white ash trees may receive some protection in areas with > 3% treated ash phloem.

71

Table 3.3: Statistical summary of the best fit model for assessing the relationship of ash survival in 2014, 2015, and 2016 using GLM with binomial distribution (logit link). Model: Survival ~ EAB impact * % phloem.

Estimate Std. Error Z value P <

2014 Intercept 1.36 0.67 2.0 0.041 * EAB impact -0.98 0.41 -2.4 0.017 * % Phloem 2.21 0.58 3.8 0.0001 *** EAB impact * % -1.63 0.42 -3.8 0.0001 *** phloem Pseudo-R2 = 0.59, Goodness of fit: P =0.9, Dispersion = 1.8

2015 Intercept 0.09 0.67 0.1 0.89 EAB impact -0.81 0.48 -1.7 0.09 . % Phloem 2.31 0.68 3.4 0.0007 *** EAB impact * % -1.82 0.60 -3.0 0.002 ** phloem Pseudo-R2 = 0.40, Goodness of fit: P =0.49, Dispersion = 2.5

2016 Intercept -0.90 0.77 -1.2 0.24 EAB impact -0.53 0.56 -1.0 0.34 % Phloem 2.12 0.80 2.6 0.008 ** EAB impact * % -1.69 0.72 -2.4 0.019 * phloem Pseudo-R2 = 0.23, Goodness of fit: P =0.33, Dispersion = 2.6

72

2014 2015 2016

100 Low EAB A B High EAB C Model low Model high 80

60

73

40

green-white ash green-white

Percent survival of Percent survival 20

0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Percent ash phloem treated with insecticide

Figure 3.2: Percent survival of untreated green-white ash trees as a function of percent treated ash phloem during 2014 (A), 2015 (B), and 2016 (C). Black circles represent low EAB impact plots and white triangles represent high EAB impact plots. See Table 3.3 for statistical output.

73

Effect of proximity to treated trees on associational protection

The model that best explained survival of untreated ash included distance to nearest treated ash tree, percentage ash phloem area treated, and EAB impact (Table 3.4).

In 2015, survival was significantly higher when treated trees were closer (P = 0.006) and percentage ash phloem treated was high (P = 0.005), and lower in the parks with high

EAB impact (P < 0.0001) (Table 3.4). In 2016, I did not find a relationship between survival of untreated ash and distance to the nearest treated ash tree. However, survival was significantly higher when percentage ash phloem treated was high (P = 0.0001) and in the parks with low EAB impact (P < 0.0001) (Table 3.4). Overall, the relationship between survival of untreated ash and distance to the nearest treated ash was detected in

2015 and was strongest in parks with low EAB impact (Figure 3.3a-b). However, for both low and high EAB impact, the chance of survival increased when both treated ash trees were closer and there was higher percentage ash phloem treated within 50 ha.

74

Table 3.4: Statistical summary of the best fit model predicting percentage survival of green-white ash trees in 2015. Model was calculated using GLM with binomial distribution (logit link). Model: Survival ~ Distance * EAB impact + % phloem + year, pseudo-R2 = 0.31, Goodness of fit: P=0.12; dispersion = 1.4.

Estimate Std. Error Z value P ≤

2015 Intercept 0.611 0.61 1.01 0.31

Distance -0.011 0.004 -2.76 0.006 ** EAB Impact -1.344 0.23 -5.86 <0.0001 *** % Phloem 0.900 0.32 2.80 0.005 ** Distance * EAB imp 0.004 0.00 2.09 0.037 * pseudo-R2 = 0.31, Goodness of fit: P=0.12; dispersion = 1.4

2016 Intercept -0.737 0.62 -1.19 0.23

Distance -0.004 0.004 -1.06 0.29

EAB Impact -1.125 0.24 -4.60 <0.0001 *** % Phloem 1.25 0.33 3.80 0.0001 *** Distance * EAB imp 0.002 0.00 0.95 0.34

pseudo-R2 = 0.36, Goodness of fit: P=0.36; dispersion = 1.4

75

Low EAB impact High EAB impact 1.0 Low impact A High impact B GLM model GLM model +/- 1.5% phloem +/- 1.5% phloem 0.8

0.6

0.4

green-white ash green-white 0.2

% survival of untreated survival % 0.0 1.0 C D

0.8

0.6

0.4

green-white ash green-white 0.2

% survival of untreated survival % 0.0 0 100 200 300 400 0 200 400 600 Distance to nearest Distance to nearest treated ash tree (cm) treated ash tree (cm)

Figure 3.3: Relationship between percent survival of untreated green-white ash trees and distance to the nearest treated tree in plots from parks with low EAB impact (A & C) and high EAB impact (B & D). Green-white ash survival was collected in 2015 (Top) and 2016 (Bottom). Data was analyzed using GLM with binomial distribution (logit link) and solid lines indicate the significant model: Survival = Distance*EAB impact + % phloem area treated; dotted lines indicate model ± 1.5% phloem area treated. See Table 3.4 for statistical output.

76

Discussion

By 2016, most ashes were dead or dying, except for ashes at Sugarcreek

MetroPark. Survival of untreated green-white ashes increased with density of treated ash trees at low EAB impact sites. At high EAB impact sites, survival decreased with abundance of treated ashes. Similarly, more untreated green-white ashes were healthy in low impact sites with larger abundance of treated ashes.

I tested two different metrics for quantifying local abundance, percent treated ash phloem and density. The differences with regard to percent treated ash phloem were that this estimate included size of treated ash trees and the stand density of ash. Therefore, the relative abundance (percentage) of treated ash phloem accounted for the proportion of food resources that have become toxic to EAB. Analyses showed that percent treated ash phloem within 400 m (50 ha) explained untreated survival better than density of treated ashes within 50 ha. However, this was only significant for low EAB impact sites.

Ashes from low impact sites were healthier (lower average canopy decline rating) than ashes from high impact sites in 2014 and 2015. Most ashes were dead or dying in

2016, except for trees at Sugarcreek Metropark, where EAB has probably not been established as long (see chapter 2, Figure 2.7). However, when Sugarcreek Metropark was removed from the analysis, ashes were still healthier in plots that had higher percent treated ash phloem within 50 ha.

Parks were characterized as low or high EAB impact based on percent survival in

2014. Percent survival was > 60 % at Englewood, Germantown, and Sugarcreek and was

< 20% at Cox Arboretum, Taylorsville, and Twin Creek. Previous studies reported that it

77 takes approximately six years after EAB is detected for a stand with mostly healthy ashes to become a stand with mostly ashes (Knight et al., 2013; Klooster et al., 2014).

Therefore, when FRMP began their insecticide treatments in 2011, the EAB population was likely more established at Cox Arboretum, Taylorsville, and Twin Creek than at the other three parks. This indicates that the timing of insecticide treatments may be important in determining whether treated ash trees provide protection to untreated ashes.

These results suggest that clusters of treated ash trees can slow the rate of ash mortality through associational effects with untreated neighboring ash at early stages of the EAB invasion. However at later stages of the EAB invasion when populations are higher, the low percentage of treated ash phloem in this study provided little associational protection.

Species richness, species diversity, ash density, and total stand density did not affect EAB-induced ash mortality (Smith et al., 2015). Therefore, I propose that areas where EAB is less established (low EAB impact) higher densities or percentage of ash phloem of treated ashes may decrease the probability that EAB to will oviposit on untreated ashes. In other words, as EAB immigrates into a new stand with treated and untreated ash, when more of the ash population has been treated with insecticide, there will be greater chances that EAB will find a treated ash tree to either feed or oviposit.

Since the ash trees treated with emamectin benzoate will likely kill EAB adults and neonate larvae that come into contact with it (McCullough et al., 2011), the population of

EAB will decline within that neighborhood and indirectly reduce herbivory on untreated ashes.

78

Another possible mechanism for this association is that treated ashes were on average healthier and larger than untreated ashes. Thus, treated ash trees may be more apparent to EAB as individuals are immigrating into new area. This has been documented with the pine processionary moth (Thaumetopoea pityocampa), where stands with hedgerows on the edge of forest patches decreased the total number of nests on pine trees located in the interior of the patch (Dulaurent et al., 2012). Similarly, herbivory on oak saplings by various trophic guilds of herbivores (leaf chewers, skeletonizers, and leaf miners), showed that the extent of damage increased with sapling height (Castagneyrol et al., 2013). They concluded that this was due to the detectability of taller saplings

(Castagneyrol et al., 2013). Since EAB use both visual and olfactory cues to locate ashes and mates (Rodriguez-Saona et al., 2006; Crook et al., 2009; Crook and Mastro, 2010), it is likely the larger and healthier ashes are more noticeable to EAB adults. Therefore, areas with higher abundances of treated ashes will contain large and healthy ashes that will attract and kill adults that feed on the foliage.

The associational protection of ash from large-scale insecticide treatments were first proposed as a tactic for EAB management in the SLow Ash Mortality (SLAM) project (McCullough and Mercader, 2012; McCullough et al., 2015). In a model simulation of urban forests, they concluded that if 20% of the ash population was treated annually with emamectin benzoate, nearly 100% of the ash population could be protected

(McCullough and Mercader, 2012). They suggested that treated ash trees may create a protective buffer and as the density of treated ash trees increases, this radius may protect their untreated neighbors. This study was tested in the Upper Peninsula of Michigan,

79 where they found larval densities decreased and ashes exhibited healthier canopies in areas with more treated ash within 800 m (Mercader et al., 2015). They also reported that the addition of girdled or stressed ashes further decreased larval densities. Girdled trees are more attractive to EAB, because they release plant volatiles that signal that the tree is undergoing stress due to a disruption of resources (Rodriguez-Saona et al., 2006;

McCullough et al., 2009). These support my findings and suggest that treated and girdled ashes reduce the likelihood that EAB will locate untreated ashes. In particular, this suggests that in order for larval densities to be lower, more adult EAB must have been killed by treated ashes. I did not quantify larval density. However, there is a direct relationship between larval density and the canopy ratings used in this study (Flower et al., 2013).

I only found this relationship at sites with a lower EAB impact. This was likely due to the extent of the EAB infestation when the insecticide treatments began. If EAB was already established and showing signs of infestation, then the surrounding density of

EAB may have been too large to see any associational effects. FRMP selected only healthy ashes to treat with insecticides. Therefore, if EAB was already established, that means fewer individuals will be immigrating into these areas. If food sources are available, most EAB adults disperse within 100 m of the tree that they emerged from

(Mercader et al., 2009). Thus, there may be less of a chance that EAB will encounter a treated ash due to their dispersal patterns after emergence, especially in areas undergoing high rates of ash mortality. Additionally, when densities of EAB are high, the chances of an ash tree escaping infestation may be random or due to a genetic resistance mechanism.

80

In 2015 and 2016, survival of untreated ashes was higher as percentage of treated ash phloem within 400 m. This relationship was strongest when the nearest treated ash tree was < 100 m away. However, this pattern was only evident in the low EAB impact sites. Additionally, survival was similar in 2015 and 2016, which resulted in approximately the same linear model.

Overall, the chance of survival was highest in plots where the closest treated ash was < 100 m and when percent treated ash phloem exceeded > 1.5%. This indicates that treated ashes provided associational protection to untreated ashes and this effect was strongest in close proximity to treated ash.

Spatial patterns of associational effects between neighboring plants have varied.

Many studies found strong support that herbivory decreased as the distance between host and non-host decreased (Karban et al., 2006; Graff et al., 2007; Karban, 2007; Talamo et al., 2015; Champagne et al., 2016). For example, sagebrush (Artemisia tridentata) releases volatiles when clipped and the presence of these clippings up to 60 cm away from a plant reduce herbivory (Karban et al., 2006). Additionally, the presence of sagebrush reduced feeding mule’s ears (Wyethia mollis), but the strength of this relationship increased as the distance between sagebrush and mule’s ears decreased

(Karban, 2007). A meta-analysis found that associational effects (herbivore damage) decreased with plot size (0.1 to 100 m2) and distance between neighbors (0 to 2 m)

(Champagne et al., 2016). Other studies, however, found that distance between neighboring plants was not an important indicator of associational resistance (Stokes and

Stiling, 2013). Conflicting results of studies testing associational resistance show that

81 much needs to be understood about the mechanisms of these relationships. Additionally, it is likely that it may be difficult to make general conclusions about plant community dynamics since this depends on competition between plants, microsite conditions (i.e., soil chemistry, water, and light availability), and the herbivore community.

I conclude that large-scale insecticide management may be able to provide direct and indirect protection to the ash community. Specifically, I found evidence that larger abundances of treated ashes reduce the effects of emerald ash borer on neighboring untreated ashes. Furthermore, I found evidence that this relationship strengthens when treated ash are closer to untreated ash. The associational effects were not documented in the sites that have been affected by the EAB invasion the longest, which indicates that there is an optimal window of time to begin a large-scale treatment program with the goal of also protecting untreated neighbors. Additionally, the highest percent treated ash phloem was approximately 3% of the total ash population within 50 ha. Studies that test the effects of treating a larger percentage of ash phloem are required to further test this relationship. Lastly, most of the ashes were dying or dead at the end of this study, with the exception of plots at Sugarcreek Metropark. This indicates that this large-scale insecticide program has not reduced EAB populations enough to slow the rate of ash mortality. However, testing higher percentages of treated ash phloem may improve the ability to slow the spread of EAB. Additionally, if insecticides are used with girdled trap trees, the associational protection of treated ashes may increase and more effectively slow the rate of ash mortality.

82

Literature cited

Ayres, E., Dromph, K.M., Cook, R., Ostle, N., Bardgett, R.D. 2007. The influence of below-ground herbivory and defoliation of a legume on nitrogen transfer to neighboring plants. Funct. Ecol. 21: 256-263. Cappaert, D., McCullough, D.G., Poland, T.M., Siegert, N.W. 2005. Emerald ash borer in North America: A research and regulatory challenge. Am. Entomol. 51: 152-165. Castagneyrol, B., Giffard, B., Pere, C., Jactel, H. 2013. Plant apparency, an overlooked driver of associational resistance to insect herbivory. J. Ecol. 101: 418-429. Champagne, E., Tremblay, J.P., Cote, S.D. 2016. Spatial extent of neighboring plants influences the strength of associational effects on mammal herbivory. Ecosphere 7: e01371. Crook, D.J., Francese, J.A., Zylstra, K.E., Fraser, I., Sawyer, A.J., Bartels, D.W., Lance, D.R., Mastro, V.C. 2009. Laboratory and field response of the emerald ash borer (Coleoptera: Buprestidae), to selected regions of the electromagnetic spectrum. J. Econ. Entomol. 102: 2160-2169. Crook, D.J., Mastro, V.C. 2010. Chemical ecology of the emerald ash borer Agrilus planipennis. J. Chem. Ecol. 36: 101-112. Dulaurent, A.M., Porte, A.J., van Halder, I., Vetillard, F., Menassieu, P., Jactel, H. 2012. Hide and seek in forests: Colonization by the pine processionary moth is impeded by the presence of nonhost trees. Agric. For. Entomol. 14: 19-27. Flower, C.E., Knight, K.S., Rebbeck, J., Gonzalez-Meler, M.A. 2013. The relationship between the emerald ash borer (Agrilus planipennis) and ash (Fraxinus spp.) tree decline: Using visual canopy condition assessments and leaf isotope measurements to assess pest damage. For. Ecol. Manag. 303: 143-147. Gandhi, K.J.K., Herms, D.A. 2010. North American arthropods at risk due to widespread Fraxinus mortality caused by the Alien emerald ash borer. Biol. Invasions 12: 1839-1846. Gandhi, K.J.K., Smith, A., Hartzler, D.M., Herms, D.A. 2014. Indirect effects of emerald ash borer-induced ash mortality and canopy gap formation on epigaeic beetles. Environ. Entomol. 43: 546-555. Graff, P., Aguiar, M.R., Chaneton, E.J. 2007. Shifts in positive and negative plant interactions along a grazing intensity gradient. Ecology 88: 188-199. Hamback, P.A., Agren, J., Ericson, L. 2000. Associational resistance: Insect damage to purple loosestrife reduced in thickets of sweet gale. Ecology 81: 1784-1794. Herms, D.A., McCullough, D.G. 2014. Emerald ash borer invasion of North America: History, biology, ecology, impacts, and management. Annu. Rev. Entomol. 59: 13- 30. Jactel, H., Birgersson, G., Andersson, S., Schlyter, F. 2011. Non-host volatiles mediate associational resistance to the pine processionary moth. Oecologia 166: 703-711. Karban, R. 2007. Associational resistance for mule's ears with sagebrush neighbors. Plant Ecol. 191: 295-303. Karban, R., Maron, J. 2002. The fitness consequences of interspecific eavesdropping between plants. Ecology 83: 1209-1213. 83

Karban, R., Shiojiri, K., Huntzinger, M., McCall, A.C. 2006. Damage-induced resistance in sagebrush: Volatiles are key to intra- and interplant communication. Ecology 87: 922-930. Long, L. 2013. Direct and indirect impacts of emerald ash borer on forest bird communities. In, Entomology. The Ohio State University, OhioLINK Electronic Theses and Dissertations Center, p. 165. Lovett, G.M., Canham, C.D., Arthur, M.A., Weathers, K.C., Fitzhugh, R.D. 2006. Forest ecosystem responses to exotic pests and pathogens in eastern North America. Bioscience 56: 395-405. McCullough, D.G., Mercader, R.J. 2012. Evaluation of potential strategies to SLow Ash Mortality (SLAM) caused by emerald ash borer (Agrilus planipennis): SLAM in an urban forest. Int. J. Pest Manage. 58: 9-23. McCullough, D.G., Mercader, R.J., Siegert, N.W. 2015. Developing and integrating tactics to slow ash (Oleaceae) mortality caused by emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 349-358. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. McCullough, D.G., Poland, T.M., Cappaert, D. 2009. Attraction of the emerald ash borer to ash trees stressed by girdling, herbicide treatment, or wounding. Can. J. For. Res. 39: 1331-1345. McCullough, D.G., Siegert, N.W. 2007. Estimating potential emerald ash borer (Coleoptera : Buprestidae) populations using ash inventory data. J. Econ. Entomol. 100: 1577-1586. Mercader, R.J., McCullough, D.G., Storer, A.J., Bedford, J.M., Heyd, R., Poland, T.M., Katovich, S. 2015. Evaluation of the potential use of a systemic insecticide and girdled trees in area wide management of the emerald ash borer. For. Ecol. Manag. 350: 70-80. Mercader, R.J., Siegert, N.W., Liebhold, A.M., McCullough, D.G. 2009. Dispersal of the emerald ash borer, Agrilus planipennis, in newly-colonized sites. Agric. For. Entomol. 11: 421-424. Mercader, R.J., Siegert, N.W., Liebhold, A.M., McCullough, D.G. 2011. Simulating the effectiveness of three potential management options to slow the spread of emerald ash borer (Agrilus planipennis) populations in localized outlier sites. Can. J. For. Res. 41: 254-264. Perry, K.I., Herms, D.A. 2016. Response of the forest floor invertebrate community to canopy gap formation caused by early stages of emerald ash borer-induced ash mortality. For. Ecol. Manag. 375: 259-267. Quiroz, A., Pettersson, J., Pickett, J.A., Wadhams, L.J., Niemeyer, H.M. 1997. Semiochemicals mediating spacing behavior of bird cherry-oat aphid, Rhopalosiphum padi feeding on cereals. J. Chem. Ecol. 23: 2599-2607. Rodriguez-Saona, C., Poland, T.M., Miller, J.R., Stelinski, L.L., Grant, G.G., de Groot, P., Buchan, L., MacDonald, L. 2006. Behavioral and electrophysiological 84

responses of the emerald ash borer, Agrilus planipennis, to induced volatiles of Manchurian ash, Fraxinus mandshurica. Chemoecology 16: 75-86. Scharenbroch, B.C., Bockheim. J.G., 2008. Gaps and soil C dynamics in old growth northern hardwood-hemlock forests. Ecosystems 11: 426-441. Sholes, O.D.V. 2008. Effects of associational resistance and host density on woodland insect herbivores. J. Anim. Ecol. 77: 16-23. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Siegert, N.W., McCullough, D.G., Williams, D.W., Fraser, I., Poland, T.M., Pierce, S.J. 2010. Dispersal of Agrilus planipennis (Coleoptera: Buprestidae) from discrete epicenters in two outlier sites. Environ. Entomol. 39: 253-265. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Smitley, D.R., Doccola, J.J., Cox, D.L. 2010a. Multiple-year protection of ash trees from emerald ash borer with a single trunk injection of emamectin benzoate, and single-year protection with an imidacloprid basal drench. Arboric. Urban For. 36: 206-211. Smitley, D.R., Rebek, E.J., Royalty, R.N., Davis, T.W., Newhouse, K.F. 2010b. Protection of individual ash trees from emerald ash borer (Coleoptera: Buprestidae) with basal soil applications of imidacloprid. J. Econ. Entomol. 103: 119-126. Stokes, K., Stiling, P. 2013. Effects of relative host plant abundance, density and inter- patch distance on associational resistance to a coastal gall making midge, Asphondylia borrichiae (Diptera: Cecidomyiidae). Fla. Entomol. 96: 1143-1148. Sugiura, N., 1978. Further analysts of the data by akaike’s information criterion and the finite corrections. Commun. Stat. Theory Methods 7: 13-26. Tahvanainen, J.O., Root, R.B. 1972. The influence of vegetational diversity on the population ecology of a specialized herbivore, Phyllotreta cruciferae (Coleoptera: Chrysomelidae). Oecologia 10: 321-346. Talamo, A., Barchuk, A., Cardozo, S., Trucco, C., Maras, G., Trigo, C. 2015. Direct versus indirect facilitation (herbivore mediated) among woody plants in a semiarid Chaco forest: A spatial association approach. Austral Ecol. 40: 573-580. Taylor, R.A.J., Bauer, L.S., Poland, T.M., Windell, K.N. 2010. Flight performance of Agrilus planipennis (Coleoptera: Buprestidae) on a flight mill and in free flight. J. Insect Behav. 23: 128-148. Ulyshen, M.D., Klooster, W.S., Barrington, W.T., Herms, D.A. 2011. Impacts of emerald ash borer-induced tree mortality on leaf litter arthropods and exotic earthworms. Pedobiologia 54: 261-265.

85

Chapter 4: Preserving future ash populations: Using insecticides to maintain ash

reproduction and recruitment in the understory

Abstract

Emerald ash borer (EAB) has drastically affected the population demographics of ash species in North America. EAB has killed hundreds of millions of ashes since its introduction nearly 20 years ago. Near the epicenter of this invasion (southeastern

Michigan), mature ashes died, which stopped reproduction and depleted the seed bank and all that remains is an “orphaned” cohort of established seedlings and saplings that are too small to be susceptible to EAB attack. Insecticides, such as emamectin benzoate, can successfully protect ash trees from EAB attack. If higher densities of treated ash trees protect more mature ash trees, then ash reproduction, regeneration, and genetic variation can be conserved. The objectives of this study were to determine whether higher densities of treated ash trees can maintain ash reproduction and regeneration. This study was conducted at Five Rivers MetroParks (FRMP) in Dayton, Ohio. FRMP began a large- scale insecticide program in 2011 to treat ash trees with emamectin benzoate (TREE-äge,

Arborjet, Inc.) every two years. I tested the effects of percent treated ash phloem and

EAB impact on densities of seedlings and flowering ash trees. In high EAB impact plots, seedling densities remained low and did not change with percent treated ash phloem. In

86 low EAB impact plots, seedling densities increased with percent treated ash phloem.

There were more flowering green-white ashes (treated and untreated) in plots with higher percent treated ash phloem. In low impact plots, density of flowering untreated green- white ashes was higher due to higher survival. Blue ashes were less abundant throughout parks, but there were no differences in the density of flowering blue ash between low and high impact plots. I tested the effect of density of seed bearing ashes and habitat characteristics on density of seedlings. I found seedlings increased with density of seed bearing ash and decreased with herbaceous vegetation height. This indicated that higher densities of treated ash trees, especially seed bearing ashes, directly impacts reproduction and new regeneration. Additionally, this indicated that habitat characteristics may also be important in determining the suitability of a site for seed germination and establishment.

Introduction

The emerald ash borer invasion (Agrilus planipennis Fairmaire) has caused widespread ash (Fraxinus spp.) mortality throughout eastern North America (Herms and

McCullough, 2014). Emerald ash borer (EAB) was first detected in 2002 near Detroit,

Michigan (Cappaert et al., 2005), but dendrochronological reconstruction shows that

EAB was killing ash trees by the early 1990’s (Siegert et al., 2014).

In southeast Michigan, widespread ash mortality caused a rapid decrease in ash reproduction and new regeneration. Kashian and Witter (2011) found that seedling densities decreased as percentage mortality increased. Similarly, Klooster et al. (2014) reported that reproduction and new regeneration ceased during peak EAB-induced ash

87 mortality. Currently the ash population in southeast Michigan consists of an “orphaned” cohort of established seedlings and saplings (Klooster et al., 2014). Natural mortality rates of tree populations are typically highest at early seedling stages and rapidly decrease once individuals reach a certain age or size (Hett and Loucks, 1968), which suggests that most established ash seedlings and saplings will die before they reach maturity.

Cessation of ash reproduction and new regeneration jeopardizes the future of the ash population throughout the EAB invasion range. Since natural mortality is high for younger seedlings, long-term recruitment and persistence of tree populations requires continuous reproduction and new regeneration to replace the seedlings that have died

(Plumptre, 1995). Therefore, protection of mature ash trees may be a strategy for preserving ash populations in areas affected by EAB.

Insecticides can effectively protect ash trees from emerald ash borer infestation by killing adults and larvae (McCullough et al., 2011). In particular, a single application of emamectin benzoate protects trees for 2 – 3 years (Smitley et al., 2010). Due to its multi- year efficacy, studies indicate that treating multiple ash trees with emamectin benzoate may provide associational protection to neighboring ash trees and slow the spread of

EAB (Mercader et al., 2011; McCullough and Mercader, 2012; Mercader et al., 2015).

These studies were all a part of the SLow Ash Mortality (SLAM) project to evaluate the use of emamectin benzoate on population growth of emerald ash borer and the rate of ash mortality. They reported that signs of EAB infestation and ash mortality decreased with density of treated ash trees within 800 m of a focal tree (Mercader et al., 2015). I tested this model in Chapter 3 and found that untreated green and white ash survival was higher

88 in areas with higher densities of treated ash trees within a 400 m radius. In addition to providing associational protection of neighboring untreated ash trees, insecticide treated ash trees may preserve ash reproduction and regeneration, but this hypothesis has not been tested.

Most ash species are dioecious (separate male and female trees) and wind pollinated (Burns and Honkala, 1990). In eastern North America all species are wind pollinated, but green (F. pennsylvanica), white (F. americana), black (F. nigra), and pumpkin ash (F. profunda are dioecious, and blue ash (F. quadrangulata) is monoecious

(male and female flowers on the same individual) (Burns and Honkala, 1990). Since multiple males can fertilize flowers from a single female ash tree, protecting both male and female trees with insecticides is important for preserving successful reproduction.

Ash pollen can fertilize female flowers up to 200 m away (Heuertz et al., 2003).

Therefore, treated ash trees within 200 m may play a role in fertilization and seed production.

The objectives of this study were to 1) quantify ash demography in southwestern

Ohio to determine how the ash population changed as ash trees died from EAB infestation, 2) compare the effect of density of treated ash trees on density of flowering ash and seedling regeneration, and 3) determine how abundance of treated ash trees impacted ash demography. If protecting mature ash trees with insecticide treatments can maintain ash reproduction and regeneration, it will provide a tool that forest managers can use to conserve reproductive ash trees and maintain regeneration, while minimizing ecological impacts of ash mortality.

89

Materials and Methods

This study was conducted at Five Rivers MetroParks (FRMP) in southwestern

Ohio. FRMP consists of a network of parks located throughout Montgomery County and in southern Miami County, western Greene County, and northern Warren County. In

2011, FRMP began an on-going insecticide program to treat 600 mature ash trees with

TREE-äge (emamectin benzoate 4% ME, Arborjet, Woburn, MA, USA). Each treated tree was trunk-injected using the Arborjet Quik-Jet system in 2011, 2013, and 2015.

In 2014, I established 24 quadrats (100 m X 100 m; 1 ha) in six of the metroparks

(Cox Arboretum, Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek), which were categorized based on degree to which they were impacted by EAB when the study was initiated in 2014. Survival of untreated ash trees was lowest at Cox Arboretum

(CA), Taylorsville (TA), and Twin Creek (TC) Metroparks (~ 25% living ash), which were categorized as high EAB impact. Survival of untreated ash trees was highest at

Sugarcreek (SC), Englewood (EN), and Germantown (GT) (>75% living ash), which I categorized as low EAB impact.

Quadrats were randomly located within each park using ArcMap 10.4 and a measured grid. Plots were independent measurements of ash demography within a cluster of treated ashes. The cluster sizes of treated ash trees ranged from 0 to 19 trees per ha, but treated ash were not present within the boundaries of each plot. I established three to six quadrats per park depending on their area of forest cover.

Four 0.1 ha circular plots (18-m radius) were established within each quadrat; two proximal plots within 50 m of the quadrat center and two distal plots 150 m from the

90 center. One distal plot was located on a randomly selected azimuth (degree bearing). The second distal plot was established approximately 90 degrees clockwise from the first

(Figure 4.1). Each plot consisted of a non-ash center tree (≥ 10 cm diameter at breast height; dbh) and at least two mature ash trees (≥ 10 cm dbh). Plots were further differentiated into an 8-m radius subplot and four microplots (4m2) following methods of

Smith (2006) and Smith et al. (2015) (Figure 4.1).

Ash demography

From June to August 2014 – 2016, I quantified density of seedlings and mature ash trees within each plot. Seedlings were divided into three height classes (classes 1-3)

(Table 4.1). Class 1 seedlings were first-year ash seedlings with cotyledons, class 2 seedlings were ash seedlings ≤ 0.25 m tall and class 3 seedlings were ash seedlings 0.25 to 1.5 m tall. Trees were identified as mature if they were ≥ 10 cm diameter at breast height (dbh) (Table 4.1).

Density of seedlings was quantified annually from 2014-2016 in the four microplots (4 m2) within each plot. Seedling counts from the microplots were pooled for each plot and density per ha was estimated. In 2016, I identified seedlings as green-white ash (Fraxinus pennsylvanica or F. americanum) and blue ash (F. quadrangulata) to determine the density of each species.

Density of living mature ash trees was quantified annually from 2014-2016 for each plot. Additionally, in 2015, I surveyed each mature ash tree for presence of inflorescence and determined their sex. Ash pollen and seeds can disperse up to 200 m

91 per year (Heuertz et al., 2003). I surveyed density, presence of inflorescence and determined the sex of treated ash trees within 200 m (12.5 ha) of each plot, because ash pollen can disperse up to 200 m away (Heuertz et al., 2003). Lastly, mature ash trees were divided by species group into blue ash and green-white ash due to similar EAB susceptibility of green and white ash and lower EAB susceptibility of blue ash

(Anulewicz et al., 2007; Tanis and McCullough, 2012).

Percent treated ash phloem area

As described in Chapter 3, I estimated the amount of ash phloem area for each treated ash tree using the second order polynomial equation from McCullough and

Siegert (2007). Total ash phloem area per ha and percentage of ash phloem area treated within 200 m of each plot were then grouped based on average percent treated ash phloem within 200 m (0, 1, 2.5, and 4.5% treated ash phloem).

92

8 m

18 m

Figure 4.1: Quadrat and plot layout for each study site. (A) Distribution of 4 replicated plots per quadrat. (B) 18-m radius plot subdivided into 4 microplots (4m2) and a subplot (8 m radius) used to sample seedlings (microplot), saplings (subplot), and immature trees (subplot). Mature trees were quantified within the plot boundary (18 m radius).

Size class Description Class 1 seedlings First-year seedlings with cotyledons Class 2 seedlings < 0.25 m tall, without cotyledons Class 3 seedlings ≥ 0.25 m to 1.5 m tall Saplings ≥ 1.5 m tall, < 2.5 cm stem diameter Immature tree 2.5 – 9.9 cm diameter at breast height (dbh) Mature tree ≥ 10 cm dbh

Table 4.1: Descriptions of ash demography at Five Rivers MetroParks in Dayton, OH.

93

Habitat characteristics

Within each plot, I characterized the plant community and habitat characteristics.

These included ash ground cover (%), canopy cover (%), slope, and aspect. Canopy cover was measuring using a spherical densiometer. Densiometer measurements were recorded by counting the number of squares with light penetrating at each position. Slope was categorized from 0-2, 2-5, 6-9, 10-14, and >25 based on steepness. Aspect of the slope was measured using a compass, then converted to radians and transformed to reflect site productivity (Beers et al. 1966). Additionally, I estimated vegetation cover (%) of stems

≤ 1.5 m tall that were growing in each microplot. In 2014 and 2015, I estimated median vegetation height for all stems ≤ 1.5 m tall growing in the microplots, and in 2016, I divided vegetation into woody and herbaceous (non-woody).

Statistical analyses

Mature ash were divided into insecticide treated and untreated trees, and by species group (blue ash and green-white ash). For each group, I compared the effects of the percentage of ash phloem area treated and EAB impact on the total density and density of flowering individuals using generalized linear models (GLM) with negative binomial distribution (log link) using proc GLIMMIX and least square mean with

Statistical Analysis Software (SAS 9.4). Additionally, I tested the effect of percentage ash phloem area treated and EAB impact on the percentage of flowering using GLM with a binomial distribution with the MASS package in R 3.2.3. Significance was accepted when P < 0.05.

94

Effects of percentage of ash phloem area treated and EAB impact on the density of class 1, class 2, and class 3 seedlings was tested using GLM with negative binomial distribution (log link) using SAS 9.4 (proc GLIMMIX). Separate tests were conducted for each year (2014-2016). Additionally, I tested the effects of percentage of ash phloem area treated, EAB impact, and habitat characteristics on the density of class 1 seedlings using

GLM with negative binomial distribution (log link) with the MASS package in R 3.2.3.

Lastly, I tested the relationship between density of flowering blue ash or female green- white ash (treated and untreated) and habitat characteristics on the density of conspecific seedlings using GLM with negative binomial distribution (log link) with the MASS package in R 3.2.3. Significance was accepted if P < 0.05.

Results

Population of mature ash

Green, white, and blue ash trees were the most common species of ash at FRMP.

As reported in Chapter 3, survival of blue ash remained high, whereas green and white ash survival declined throughout this study. In particular, percentage survival of green and white ash at the parks characterized as high EAB impact (Cox Arboretum,

Taylorsville, and Twin Creek Metroparks) decreased from approximately 25% in 2014 to

<10% in 2016, but survival of blue ash was >75% from 2014 to 2016. The parks with low

EAB impact had variable patterns of survival for green and white ash, specifically

Englewood and Germantown Metroparks decreased from 75% in 2014 to <15% in 2016, and Sugarcreek Metropark decreased slightly from 91% in 2014 and 53% in 2016.

95

Survival of blue ash at all parks with low EAB impact was >95% from 2014 to 2016.

Survival of treated green, white, and blue ash was nearly 100% throughout the study.

Density of blue ash was low for both untreated and insecticide treated populations. I found no differences among density of untreated blue ash by either percentage ash phloem area treated or EAB impact (Table 4.2). Percentage of flowering for untreated blue ash was high and was not affected by EAB impact or percentage ash phloem treated (Table 4.2). For treated blue ash, density was higher in plots with higher percentage ash phloem treated (i.e., > 0%). Overall, I found significantly more flowering treated blue ash in the plots with 2.5% ash phloem treated (F = 3.91, df= 2, P = 0.03)

(Table 4.2).

Density of untreated green and white ash differed by both EAB impact and percentage ash phloem area treated. There were significantly more untreated trees in parks with low EAB impact (F = 4.49, df = 1, P = 0.04), and in plots with 0 and 1% ash phloem treated (F = 3.15, df = 3, P = 0.03) (Table 4.3). Percentage of flowering for untreated green and white ash was highest in the parks with low EAB impact, especially in the plots with at least 2.5% treated ash phloem (Table 4.3), however these differences were not significant. The percentage of female untreated ash trees varied by plot, but no differences were found by EAB impact or percentage ash phloem treated (Table 4.3).

Insecticide treated green and white ash increased by percentage of ash phloem area, but did not differ by EAB impact. Total density of treated trees increased with percentage ash phloem area (F = 34.1, df = 3, P <0.0001) (Table 4.3). Plots with higher percentage ash phloem treated also had significantly higher densities of treated trees that

96 were flowering within 12.5 ha (F = 27.1, df = 3, P < 0.0001). Percentage flowering was high for treated ash trees and did not differ by percentage ash phloem area treated.

Overall, there were higher percentages of treated females in parks with low EAB impact when 4.5% ash phloem was treated (i.e., higher densities of treated trees). However, in the parks with high EAB impact, the percentage of females was similar for plots with 1,

2.5, and 4.5% ash phloem treated (Table 4.3).

97

Table 4.2: Density (ash ha-1) and percentage inflorescence of treated and untreated blue ash from forests at Five Rivers Metroparks (FRMP) in southwestern, OH, according to percentage ash phloem area treated and EAB impact. Parks within FRMP were characterized as low or high EAB impact based on the percentage survival in 2014; low EAB impact indicates parks with ≥ 75% survival and high EAB impact parks with ≤ 25% survival. Phloem area of each treated ash tree within 200 m (12.5 ha) of each plot was estimated to calculate percentage ash phloem treated within 12.5 ha of each plot, then plots were grouped by mean percentage ash phloem (0, 1, 2.5, or 4.5%). Statistical significance was calculated using generalized linear mixed models with negative binomial distribution; P < 0.1 “.”; P < 0.05 “*”.

Low EAB impact High EAB impact

0% 1% 2.5% 4.5% P 0% 1% 2.5% 4.5% P

98 Untreated blue ash

-1 Total (ha ) 8 ± 8 16 ± 9 17 ± 10 9 ± 4 18 ± 9 19 ± 12 17 ± 6 4 ± 2 Flowering 6 ± 5 16 ± 8 9 ± 6 11 ± 5 15 ± 6 13 ± 8 14 ± 6 4 ± 2 (ha-1)

% Flowering 62.5 86.7 ± 13.3 65.7 ± 14.8 82.7 ± 9.2 93.9 ± 6.1 75.0 ± 8.3 70.4 ± 12.9 100

Treated blue ash Total (12.5 0 a 3 ± 2 b 5 ± 2 b 3 ± 1 b . 0 a 2 ± 0 b 4 ± 1 b 4 ± 2 b ha-1) . Flowering 0 a 3 ±2 b 5 ± 2 c 3 ±1 b * 0a 2 ± 0 b 4 ± 1 c 2 ±1 b * (12.5 ha-1) % Flowering 0 100 88.9 ± 11.1 100 0 100 95.8 ± 4.2 86.1 ± 13.9

98

Table 4.3: Density (ash ha-1) and percentage inflorescence of treated and untreated green-white ash from forests at Five Rivers Metroparks (FRMP) in southwestern, OH, according to percentage ash phloem area treated and EAB impact. See table 4.2 for description of percentage ash phloem and EAB impact. Statistical significance was calculated using generalized linear mixed models with negative binomial distribution; P < 0.0001 “***”.

Low EAB impact High EAB impact

0% 1% 2.5% 4.5% P 0% 1% 2.5% 4.5% P

Untreated green-white ash

Total (ha-1) 131 ± 41 146 ± 74 75 ± 14 49 ± 7 52 ± 16 62 ± 10 68 ± 14 73 ± 23

99 Flowering 45 ± 21 36 ± 34 21 ± 7 18 ± 4 (ha-1) 2 ± 1 4 ± 2 6 ± 2 3 ± 3 % Flowering 29.0 ± 5.5 11.1 ± 7.6 37.1 ± 9.8 40.6 ± 7.7 14.2 ± 10.1 7.0 ± 4.5 13.4 ± 5.1 2.6 ± 2.6

% Female 49.3 ± 13.7 17.6 ± 17.6 47.7 ± 13.1 32.1 ± 11.1 66.7 ± 33.3 50.0 ± 28.9 40.5 ± 17.0 0

Treated green-white ash Total (12.5 0 a 6 ± 1 b 14 ± 4 c 17 ± 4 d *** 1 ± 0 a 3 ± 1 b 5 ± 1 c 11 ± 1 d *** ha-1) Flowering 0 a 5 ± 1 b 10 ± 2 c 12 ± 2 d *** 0 a 3 ± 1 b 5 ± 0 c 9 ± 1 d *** (12.5 ha-1) % Flowering 100 81.8 ± 7.3 79.0 ± 4.0 85.7 ± 4.0 66.7 ± 23.6 86.1 ± 5.6 90.4 ± 3.0 80.2 ± 7.7

% Female 0 33.9 ± 13.3 35.5 ± 5.6 52.8 ± 4.6 0 47.6 ± 6.0 52.4 ± 4.7 49.8 ± 4.4

99

Population of ash seedlings

Densities of class 1 seedlings (first-year seedlings with cotyledons) were higher in parks with low EAB impact than high EAB impact, however these differences were only significant in 2015 (F=62.7, df=1, P<0.0001) and 2016 (F=6.52, df=1, P=0.012).

Comparisons by percentage ash phloem area treated (0, 1, 2.5, and 4.5% ash phloem) showed densities of first-year seedlings were significantly higher in plots with 4.5% ash phloem treated in 2014 (F=4.29, df = 3, P=0.007), 2015 (F=4.39, df=3, P=0.006), and

2016 (F=2.27, df=3, P=0.09) (Table 4.4). Density of first-year seedlings was especially high in 2015 in parks with low EAB impact, which may indicate a mast seed year the previous year. However, this increase was not measured in the parks with high EAB impact. Overall, densities of first-year seedlings in parks with high EAB impact decreased from 2014 to 2016 and no relationship was detected with percentage ash phloem treated (Table 4.4).

Since there was much variation in mean densities of first-year seedlings within each group, I tested the response of seedlings to determine which factors are important predictors of seed germination. In 2014, density of first-year seedlings significantly increased with percentage ash phloem area treated (Z = 3.21, P = 0.001) and decreased with understory vegetation height (Z = -3.97, P < 0.0001) (Table 4.5, Figure 4.2a).

Additionally, there was a negative interaction between percentage ash phloem treated and

EAB impact, indicating that treated phloem did not affect densities of first-year seedlings in parks with high EAB impact (Z = -2.73, P = 0.006) (Table 4.5, Figure 4.2a). A similar relationship was detected in 2015 and 2016 (Table 4.5, Figure 4.2b-c).

100

From 2014 to 2016, density of class 2 seedlings (smaller established seedlings) increased in parks with low EAB impact, but did not change in parks with high EAB impact (F=3.76, df=2, P=0.025). Specifically, in parks with low EAB impact, density of class 2 seedlings was highest in plots with at least 2.5% ash phloem area treated in 2014

(F=2.8, df=3, P=0.04), 2015 (F=3.38, df=3, P=0.02), and 2016 (F=3.23, df=3, P=0.03)

(Table 4.4). Overall, class 2 seedlings increased in areas with low impact, but this increase was pronounced in plots with 2.5% and 4.5% ash phloem treated (Table 4.4). In parks with high EAB impact, percentage ash phloem treated had no effect on density of class 2 seedlings (Table 4.4).

Density of class 3 seedlings (larger established seedlings) did not differ by EAB impact or percentage ash phloem area treated (Table 4.4). Overall, there were more class

3 seedlings in the parks with high EAB impact, but all plots indicated an increase in density of class 3 seedlings from 2014 to 2016.

Lastly, comparisons of total seedlings by species showed higher density of green- white ash seedlings in parks with low EAB impact than high EAB impact (F=8.39, df=1,

P=0.005), but density of blue ash seedlings did not differ by EAB impact. I tested the relationship between habitat characteristics and either flowering blue ash or female green-white ash (treated and untreated individuals) on the density of conspecific seedlings. Blue ash seedlings significantly increased with density of flowering untreated blue ash (Z = 3.79, P < 0.0001) and treated blue ash (Z = 2.17, P = 0.03), and decreased with herbaceous vegetation height (Z = -3.74, P < 0.0001) (Table 4.5, Figure 4.3a).

Similarly, green-white ash seedlings increased with density of treated ash trees (Z = 3.06,

101

P = 0.002), and decreased with herbaceous vegetation height (Z = -1.94, P = 0.05) and steeper slope (Z = -2.42, P = 0.02) (Table 4.5, Figure 4.3b).

102

Table 4.4: Seedling densities (ash ha-1) in forests at Five Rivers MetroParks in southwestern Ohio, according to EAB impact (based on percentage survival in 2014) and percentage ash phloem area treated within 200 m (12.5 ha) of each plot. Statistical analyses were from generalized linear model with negative binomial distribution of the relationship between EAB impact and % phloem area treated.

% ash EAB phloem 2014 (ha-1) 2015 (ha-1) 2016 (ha-1) Impact treated Class 1 seedlings (first-year seedlings with cotyledons) Low 0% 23,250 ± 12,422 a 113,000 ± 42,356 a 13,500 ± 9,244 1% 5,500 ± 3,742 a 47,000 ± 23,041 a 12,000 ± 8,344 2.5% 34,000 ± 9,334 a 138,833 ± 31,704 a 10,667 ± 4,201 4.5% 110,417 ± 43,145 b 345,556 ± 96,047 b 47,778 ± 24,365 P 0.007 0.006 0.09

High 0% 29,231 ± 13,269 20,577 ± 8,441 12,308 ± 6,408 1% 36,500 ± 27,237 21,000 ± 10,443 6,750 ± 3,376 2.5% 19,306 ± 8,872 9,583 ± 2,244 5,556 ± 2,301 4.5% 17,143 ± 9,313 7,857 ± 4,831 2,500 ± 1,336 P N.S. N.S N.S

Class 2 seedlings (stems < 0.25 m tall) Low 0% 6,250 ± 4,104 a 16,750 ± 5,579 a 36,250 ± 11,845 a 1% 18,000 ± 9,301 a 29,000 ± 11,848 ab 52,500 ± 23,888 ab 2.5% 18,333 ± 6,088 ab 54,500 ± 10,821 b 121,000 ± 22,949 bc 4.5% 36,528 ± 14,989 ab 86,667 ± 20,455 c 208,056 ± 53,281 c P 0.045 0.022 0.026

High 0% 50,769 ± 18,293 83,846 ± 20,883 84,615 ± 28,803 1% 25,500 ± 8,818 50,000 ± 16,099 62,000 ± 19,902 2.5% 57,917 ± 23,401 80,972 ± 22,610 79,583 ± 25,116 4.5% 28,571 ± 12,148 44,286 ± 21,552 48,571 ± 25,459 P N.S. N.S. N.S. (Continued)

103

(Table 4.4: Continued)

% ash EAB phloem 2014 (ha-1) 2015 (ha-1) 2016 (ha-1) Impact treated Class 3 seedlings (stems between 0.25 – 1.5 m tall) Low 0% 5,000 ± 2,958 5,000 ± 3,184 5,750 ± 3,802 1% 2,000 ± 935 4,500 ± 2,151 10,500 ± 4,138 2.5% 10,833 ± 8,290 11,000 ± 8,950 16,167 ± 11,639 4.5% 2,778 ± 1,105 3,194 ± 922 4,028 ± 1,264 P N.S. N.S. N.S.

High 0% 15,962 ± 6,377 16,539 ± 7,755 16,923 ± 6,293 1% 4,000 ± 1,190 4,750 ± 1,512 9,250 ± 3,334 2.5% 10,000 ± 2,651 14,583 ± 5,287 20,694 ± 6,222 4.5% 7,500 ± 4,364 5,714 ± 3,260 10,000 ± 6,050 P N.S. N.S. N.S.

104

Table 4.5: Summary of statistical output for generalized linear models to quantify the effects of percentage ash phloem treated, EAB impact, and vegetation height on density of first-year seedlings from 2014 to 2016 at Five Rivers Metroparks in southwestern Ohio.

Predictors Estimate Std. Error Z value P < 2014 Intercept 2.74 0.91 3.00 0.003 ** % Phloem 0.83 0.26 3.21 0.001 ** EAB Impact 0.61 0.47 1.29 0.20 Vegetation -0.07 0.02 -3.97 0.0001 *** height % Phloem * -0.45 0.16 -2.73 0.006 ** EAB impact Pseudo-R2 = 0.20, Goodness of fit: P =0.09, Dispersion = 1.3

2015 Intercept 6.45 0.77 8.42 0.0001 *** % Phloem 0.65 0.22 2.96 0.003 ** EAB Impact -1.65 0.40 -4.12 0.0001 *** Vegetation -0.058 0.02 -3.87 0.0001 *** height % Phloem * -0.37 0.14 -2.60 0.009 ** EAB impact Pseudo-R2 = 0.48, Goodness of fit: P =0.07, Dispersion = 1.3

2016 Intercept 5.30 1.22 4.35 0.0001 *** % Phloem 0.89 0.29 3.06 0.002 ** EAB Impact -0.22 0.56 -0.40 0.69 Herbaceous veg -0.12 0.02 -5.88 0.0001 *** height Woody veg -0.03 0.01 -2.69 0.007 ** height % Phloem * -0.54 0.19 -2.89 0.004 ** EAB impact Pseudo-R2 = 0.38, Goodness of fit: P =0.42, Dispersion = 0.98 Note: % Phloem: percentage of ash phloem area treated within 12.5 ha of each plot; EAB impact was either low impact (≥75% ash survival) or high impact (≤25% ash survival) in 2014. In 2014 and 2015 vegetation height was measured as all understory vegetation < 1.5 m tall in the microplots used to quantify seedling densities, and in 2016 vegetation height was divided into woody and herbaceous (non-woody) vegetation.

105

A Low EAB impact 300 High EAB impact Model - low impact Model - high impact +/- SD veg height 200

100

(counts per plot) per (counts

First-year seedlings First-year

0 B 500

400

300

200

(counts per plot) per (counts

First-year seedlings First-year 100

0 C 150

100

50

(counts per plot) per (counts

First-year seedlings First-year

0 0 2 4 6 8 10

% ash phloem area treated

Figure 4.2: Generalized linear model (GLM) to estimate the effect of percentage ash phloem area treated, EAB impact, and vegetation height on density of first-year seedlings (class 1) in (A) 2014, (B) 2015, and (C) 2016 at Five Rivers Metroparks in southwestern Ohio. Model equations: class 1 seedlings ~ % phloem treated * EAB impact + vegetation height (cm). Solid black line indicated predicted model for low EAB impact; solid gray line indicated predicted model for high EAB impact; and dashed lines indicated the effect of +/- standard deviation of vegetation height (cm) for low EAB impact only.

106

Table 4.6: Summary of statistical output for generalized linear models to quantify the effects of seed bearing treated and untreated ash and habitat characteristics on density of conspecific seedlings in 2016 at Five Rivers Metroparks in southwestern Ohio. Flowering blue ash and green-white ash with female flowers and/or seeds in 2015 were considered seed bearing, since they have the potential of producing seeds if pollination occurs.

Predictors Estimate Std. Error Z value P < Blue ash Intercept 2.97 0.63 4.69 0.000 *** Untreated 0.35 0.09 3.79 0.0001 *** flowering Treated 0.21 0.09 2.17 0.03 * flowering Herbaceous veg. -0.06 0.02 -3.74 0.0001 *** height (cm) Woody veg 0.003 0.01 0.31 0.76 height (cm) Pseudo-R2 = 0.28, Goodness of fit: P =0.20, Dispersion = 0.90

Green-white ash Intercept 3.97 0.37 10.6 0.0001 *** Untreated -0.06 0.11 -0.59 0.56 females Treated females 0.15 0.05 3.06 0.002 ** Herbaceous veg. -0.02 0.01 -1.94 0.05 * height (cm) Slope -0.04 0.02 -2.42 0.02 * Pseudo-R2 = 0.11, Goodness of fit: P =0.05, Dispersion = 0.94

107

A Low impact 300 High impact Model +/- 25 cm herb. height 200

(counts per plot) (counts 100

Blue ash seedlings ash Blue

0 0 2 4 6 8 10 12 14 16 Flowering blue ash per 12.5 ha

300 B 250

200

150

100

(counts per plot) (counts 50

Green-white ash seedlings ash Green-white 0 0 2 4 6 8 10 12 14 16

Female green-white ash per 12.5 ha

Figure 4.3: Generalized linear model to estimate predictor variables for density of (A) blue ash seedlings and (B) green-white ash seedlings Model equations: (A) blue seedlings = untreated + treated flowering blue ash + herbaceous vegetation height + woody vegetation height; (B) green-white seedlings = untreated + treated female ash + herbaceous height + slope. Solid line indicated predicted model; dashed lines indicated the effect of +/- standard deviation of herbaceous vegetation height. See Table 4.6 for the statistical summary.

108

Discussion

Density of mature ash trees was similar among low and high EAB impact plots and by percentage ash phloem area treated. However, low impact plots had more flowering green-white ash than high impact plots. Density of blue ash was low throughout the study sites, but both survival and percent flowering was high in parks with low and high EAB impact. Previous studies have similarly reported that blue ash growing with green and white ash survived longer and were not colonized by EAB until much of the green and white ash population had been killed by EAB (Anulewicz et al., 2007,

Tanis and McCullough, 2012). Density of treated ash trees corresponded with percent treated ash phloem, thus there were more treated ash trees in areas with higher percentage ash phloem area treated. Most of the treated ash trees were green or white ash and the density of these trees aligned directly with percentage treated ash phloem. There were fewer blue ash treated and density of this species did not change with percentage ash phloem treated.

Densities of seedlings changed over time as ash trees continued to die from EAB infestation. However, the response was most severe in parks with high EAB impact.

First-year seedling density remained higher in the parks with low EAB impact, and was increased by the presence of treated trees within 12.5 ha. Similarly, established class 2 seedlings increased with percentage ash phloem treated, but this was only measured in parks with low EAB impact. Insecticide treatment had no effect in parks with high EAB impact. The pattern detected in the parks with high EAB impact was similar to studies that have been conducted near the epicenter of the EAB invasion (southwestern

109

Michigan). Specifically, studies show that as EAB-induced ash mortality increases, densities of all seedling size classes decreased over time (Kashian and Witter, 2011;

Klooster et al., 2014). Kashian and Witter (2011) suggested that both the insufficiency in seed bearing trees and natural seedling mortality caused densities of new seedlings to decrease, which then decreased the seedling recruitment to larger size classes. In this study, seedling densities increased with percentage ash phloem in parks with low EAB impact, indicating that the protecting trees with insecticides conserves ash reproduction and maintains regeneration. However, I did not find the same relationship with percent treated ash phloem in the high impact plots. Therefore, either more ash trees are needed to be treated or untreated ash survival is necessary to effectively maintain ash reproduction.

In this study, density of established class 2 seedlings increased when density of first-year seedlings was high and did not change when densities of new seedlings was low during the previous year. However, class 3 seedlings were not affected by EAB impact, percent treated ash phloem, or density of class 2 seedlings. Trees typically follow a type

III survivorship curve, with high probability of mortality at younger stages that rapidly decrease once individuals have reached a certain age or size (Hett and Loucks, 1968;

Kobe et al., 1995). As a result, the class 1 and 2 seedlings have a higher probability of mortality than the class 3 seedlings. In order for tree populations to maintain long-term survival, continuous regeneration is necessary due to seedling mortality (Desteven, 1994;

Plumptre 1995). For larger seedlings, mortality is lower, but growth rate varies based on environmental conditions such as light, water, and soil nutrients (Kubo et al., 2004).

110

Therefore, the effects of ash mortality and percentage of ash phloem area treated may not be apparent for larger seedlings for several years.

Studies on forest regeneration suggest that when seedlings are not present in a site either (1) the conditions are not favorable for seed germination or seedling establishment

(“safe sites”) (Grubb, 1977), (2) the conditions are favorable but seeds are absent (seed availability), or (3) the conditions are unfavorable and seeds are absent (Harper 1977).

However, studies show that gaps increase light (Canham, 1988), temperature and moisture (Gray et al., 2002), alter soil nitrogen (Lovett et al., 2002; Scharenbroch and

Bockheim, 2008a; Rubino et al., 2015) and carbon cycling (Scharenbroch and Bockheim,

2008b), which could cause that habitat to become unsuitable to certain plants.

Additionally, increased light cause many shade intolerant saplings and other vegetation to respond with rapid growth and expansion (Runkle, 1990), creating light competition between neighboring plants (Harper, 1977).

Since established seedlings (classes 2 and 3) were abundant in parks with low and high EAB impact, the conditions within these plots were suitable for seedling growth and survival. However, I found that some sites may have been unsuitable for seed germination due to competition for light with taller herbaceous vegetation. Therefore, seed germination was dependent on suitable conditions (i.e., “safe sites”) (Grubb, 1977).

Additionally, I found that first-year seedlings increased with percentage ash phloem treated and in areas with low EAB impact. This indicates that new seed germination was also limited by the presence of seed sources (i.e., living and reproductive ash) (Harper,

111

1977). This also shows that treated ash trees within 12.5 ha contribute to reproduction and seed germination.

Due to the differential survival and incidence of flowering for blue ash and green- white ash, I evaluated the effect of seed bearing ash trees on the density of conspecific seedlings. The density of green-white ash seedlings increased with the density of female treated green-white ash, and decreased with herbaceous vegetation height. These results support my previous conclusions that habitat characteristics make sites suitable for seed germination and seedling establishment (Grubb, 1977), and that seed sources (i.e., female ash trees) are needed to maintain populations of seedlings (Harper, 1977). These results also confirm that treated ash trees within 12.5 ha (200 m) of a plot can contribute to seed production. Similar to green-white ash, density of blue ash seedlings was predicted by density of flowering blue ash (treated and untreated ash trees) and herbaceous vegetation height.

Density of untreated female green-white ash trees was higher in the parks with low EAB impact than high EAB impact, which corresponded with higher densities of green-white ash seedlings. For blue ash, density of untreated flowering trees was similar for low and high EAB impact. This suggests that the population of reproductive ash trees differs between blue ash green-white ash populations. Specifically, reproductive success of blue ash may not be impacted by the EAB invasion, whereas reproductive success of green and white ash was limited to treated ash and surviving untreated ash trees.

Additionally, blue ash trees are monoecious (mature individuals produce both male and female flower structures), and green and white ash are dioecious (separate male and

112 female trees) (Wallander, 2008). Therefore all flowering blue ash can produce seeds, whereas only female green and white ash trees are capable of producing seeds. This suggests that seedling densities of blue ash may increase as the impact of EAB increases.

Seed dispersal of most wind dispersed seeds, like ash, follow a leptokurtic curve, with a sharp peak in seed densities closer to the seed source and then tapering off as distance increases away from the seed source (Okubo and Levin, 1989). For ash, seed dispersal is typically within 20 m of a seed bearing tree (Clark et al., 1998; Heuertz et al.,

2003). Treated ashes were along hiking trails and were outside of the plot boundary. The distance between treated ash and the plot centers were variable, but ranged between 18 to

200 m. Thus, surviving untreated female green-white ash trees were likely the primary seed sources for each plot. In parks with high EAB impact, most plots contained no surviving female green-white ash and the nearest treated female was ≥ 20 m from the plot. Thus most female treated trees were may have been too far for seed dispersal into the plot. However, the plots from parks with low EAB impact typically contained at least one female tree (i.e., within seed dispersal distance). Additionally, as percentage of ash phloem area treated increased, there was a greater chance that both males and females were protected with insecticides. Therefore, treated ash trees were likely important pollen sources and thus increase reproductive success for surrounding female trees (both treated and untreated).

I conclude that insecticide management can be used to maintain continuous ash reproduction and regeneration in forests invaded by EAB. I also found that ash regeneration was not affected by insecticide management in areas that have been

113 impacted by EAB more severely. However, this may indicate the value of starting a large-scale insecticide program during early stages of EAB establishment. The differential response between low and high EAB impact may also indicate that either more trees need to be treated (i.e., higher percentage ash phloem area treated) or the location of the plots were too far from the treated ash trees to measure an impact of treated trees on density of seedlings. Further research needs to be conducted to determine if treating higher densities of trees more effectively maintains ash reproduction in areas where EAB is well established. Lastly, the differences in survival and incidence of flowering of blue ash and green-white ash suggest that as green and white ash are killed by EAB, density of blue ash in the understory may increase. Therefore, future studies should investigate the effects of the EAB invasion on blue ash population dynamics.

Literature cited

Anulewicz, A.C., McCullough, D.G., Cappaert, D.L., 2007. Emerald ash borer (Agrilus planipennis) density and canopy dieback in three North American ash species. Arboric Urban For. 33: 338-349. Beers, T.W., Dress, P.E., Wensel, L.C., 1966. Aspect transformation in site productivity research. J. For. 64: 691. Burns, R.M., Honkala, B.H. 1990. Silvics of North America: 2, hardwoods. Agriculture handbook 654, U.S. Department of Agriculture, Washington, D.C. Canham, C.D. 1988. An index for understory light levels in and around canopy gaps. Ecology 69: 1634-1638. Cappaert, D., McCullough, D.G., Poland, T.M., Siegert, N.W. 2005. Emerald ash borer in North America: A research and regulatory challenge. Am. Entomol. 51: 152-165. Clark, J.S., Macklin, E., Wood, L. 1998. Stages and spatial scales of recruitment limitation in southern Appalachian forests. Ecol. Monogr. 68: 213-235. Desteven, D. 1994. Tropical tree seedling dynamics: Recruitment patterns and their population consequences for 3 canopy species in Panama. J. Trop. Ecol. 10: 369- 383. Gray, A.N., Spies, T.A., Easter, M.J. 2002. Microclimatic and soil moisture responses to gap formation in coastal Douglas-fir forests. Can. J. For. Res. 32: 332-343. 114

Grubb, P.J. 1977. Maintenance of species richness in plant communities: Importance of regeneration niche. Biol. Rev. Camb. Philos. Soc. 52: 107-145. Harper, J.L. 1977. Population biology of plants. Academic Press, London. Herms, D.A., McCullough, D.G. 2014. Emerald ash borer invasion of North America: History, biology, ecology, impacts, and management. Annu. Rev. Entomol. 59: 13- 30. Hett, J.M., Loucks, O.L. 1968. Application of life-table analyses to tree seedlings in Quetico Provincial Park, Ontario. For. Chron. 44: 29-32. Heuertz, M., Vekemans, X., Hausman, J.F., Palada, M., Hardy, O.J. 2003. Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash. Mol. Ecol. 12: 2483-2495. Kashian, D.M., Witter, J.A. 2011. Assessing the potential for ash canopy tree replacement via current regeneration following emerald ash borer-caused mortality on southeastern Michigan landscapes. For. Ecol. Manag. 261: 480-488. Klooster, W.S., Herms, D.A., Knight, K.S., Herms, C.P., McCullough, D.G., Smith, A., Gandhi, K.J.K., Cardina, J. 2014. Ash (Fraxinus spp.) mortality, regeneration, and seed bank dynamics in mixed hardwood forests following invasion by emerald ash borer (Agrilus planipennis). Biol. Invasions 16: 859-873. Kobe, R.K., Pacala, S.W., Silander, J.A., Canham, C.D. 1995. Juvenile tree survivorship as a component of shade tolerance. Ecol. Appl. 5: 517-532. Lovett, G.M., Christenson, L.M., Groffman, P.M., Jones, C.G., Hart, J.E., Mitchell, M.J. 2002. Insect defoliation and nitrogen cycling in forests. Bioscience 52: 335-341. McCullough, D.G., Mercader, R.J. 2012. Evaluation of potential strategies to SLow Ash Mortality (SLAM) caused by emerald ash borer (Agrilus planipennis): SLAM in an urban forest. Int. J. Pest Manage. 58: 9-23. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. McCullough, D.G., Siegert, N.W. 2007. Estimating potential emerald ash borer (Coleoptera : Buprestidae) populations using ash inventory data. J. Econ. Entomol. 100: 1577-1586. Mercader, R.J., McCullough, D.G., Storer, A.J., Bedford, J.M., Heyd, R., Poland, T.M., Katovich, S. 2015. Evaluation of the potential use of a systemic insecticide and girdled trees in area wide management of the emerald ash borer. For. Ecol. Manag. 350: 70-80. Mercader, R.J., Siegert, N.W., Liebhold, A.M., McCullough, D.G. 2011. Simulating the effectiveness of three potential management options to slow the spread of emerald ash borer (Agrilus planipennis) populations in localized outlier sites. Can. J. For. Res. 41: 254-264. Okubo, A., Levin, S.A. 1989. A theoretical framework for data analysis of wind dispersal of seeds and pollen. Ecology 70: 329-338. Plumptre, A.J. 1995. The importance of “seed trees” for the natural regeneration of selectively logged tropical forest. Commonw. Forest. Rev. 74: 253-258. 115

Rubino, L., Charles, S., Sirulnik, A.G., Tuininga, A.R., Lewis, J.D. 2015. Invasive insect effects on nitrogen cycling and host physiology are not tightly linked. Tree Physiol. 35: 124-133. Runkle, J.R. 1990. Gap dynamics in an Ohio Acer-Fagus forest and speculations on the geography of disturbance. Can. J. For. Res. 20: 632-641. Scharenbroch, B.C., Bockheim, J.G. 2008a. The effects of gap disturbance on nitrogen cycling and retention in late-successional northern hardwood-hemlock forests. Biogeochemistry 87: 231-245. Scharenbroch, B.C., Bockheim, J.G. 2008b. Gaps and soil C dynamics in old growth northern hardwood-hemlock forests. Ecosystems 11: 426-441. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Smitley, D.R., Doccola, J.J., Cox, D.L. 2010. Multiple-year protection of ash trees from emerald ash borer with a single trunk injection of emamectin benzoate, and single-year protection with an imidacloprid basal drench. Arboric. Urban For. 36: 206-211. Tanis, S.R., McCullough, D.G. 2012. Differential persistence of blue ash and white ash following emerald ash borer invasion. Can. J. For. Res. 42: 1542-1550. Wallander, E. 2008. Systematics of Fraxinus (Oleaceae) and evolution of dioecy. Plant Syst. Evol. 273: 25-49.

116

Chapter 5: The effect of widespread ash mortality and density of treated ash trees

on conserving genetic variation in ash population

Abstract

The introduction of emerald ash borer has drastically affected the population demography of ash species in North America. Emerald ash borer (EAB) has killed hundreds of millions of ashes since its introduction nearly 20 years ago. As more ash trees were killed by EAB, fewer ash trees were contributing to reproduction. Therefore, it is likely that seedlings produced during this time may have lower genetic variation than ash that established before the EAB invasion. Insecticides, such as emamectin benzoate, can successfully protect ash trees from EAB attack. If higher densities of treated ash trees protect more mature ash trees, then ash reproduction and genetic variation can be conserved. The objectives of this study were to quantify whether widespread ash mortality caused a loss of genetic variation and determine whether insecticides can conserve genetic variation. This study was conducted at Five Rivers MetroParks (FRMP) in Dayton, Ohio. FRMP began a large-scale insecticide program in 2011 to treat ash trees with emamectin benzoate (TREE-äge, Arborjet, Inc.) every two years. Six parks were samples that were part of FRMP; Cox Arboretum, Englewood, Germantown, Sugarcreek,

Taylorsville, and Twin Creek MetroParks. To quantify genetic variation in Michigan, I

117 sampled ash seedlings and saplings from five parks in the Upper Huron River Watershed;

Highland, Island Lake, and Proud Lake State Recreation Areas, and Hudson Mills and

Indian Springs Metroparks. I quantified genetic variation of seedling and sapling populations in Michigan and Ohio. In Michigan, I found that class 2 and 3 seedlings, and saplings had similar allele richness, but class 2 and 3 seedlings had fewer than expected heterozygotes. According to principle component analysis, there was more genotypic variation in the class 3 seedlings and saplings than the class 2 seedlings. In Ohio, I found that class 1 and 2 seedlings had higher allele richness than class 3 seedlings, but each population had similar number of effective alleles. Principle coordinate analysis showed that class 3 seedlings had the least variation in genotype combinations and class 1 seedlings had the most variation. This indicated that there was a loss of genetic variation in Michigan, but no evidence of a genetic bottleneck in Ohio. I also compared genotypes by EAB impact and density of treated ash trees, but found no differences among populations. These samples were collected in 2014, therefore it is likely that the effects of

EAB were not extensive enough to see an impact to genetic variation in Ohio.

Introduction

Large-scale disturbances from invasive herbivorous insects cause wide-spread tree mortality and significant biodiversity losses (Lovett et al., 2006). In particular, emerald ash borer (Agrilus planipennis Fairmaire) is an invasive insect from Asia that has killed hundreds of millions of ashes (Fraxinus sp.) since its introduction over 20 years ago (Siegert et al., 2014). Ashes are common canopy trees in eastern deciduous forests of

118

North America and many animals rely on ash for food and shelter (Burns and Honkala,

1990). Gandhi and Herms (2010) reported 282 herbivorous arthropods are associated with ash species in North America, 25% of which were host-specific categorized as moderate to high risk of extinction. Therefore, widespread ash mortality caused by the emerald ash borer invasion can drastically alter forest dynamics, threaten biodiversity loss of many native fauna and cause extirpation of Fraxinus species in North America.

Emerald ash borer was first detected in southeastern Michigan in 2002.

Dendrochronological analyses of dead ash trees estimate that emerald ash borer had established by the early 1990’s (Siegert et al., 2014). Therefore forests in southeastern

Michigan have been affected by emerald ash borer for more than 20 years. Studies conducted in Michigan quantified the pattern of ash mortality and the impact on ash regeneration (Klooster et al., 2014). Klooster et al. (2014) reported that from 2005 to

2009 mortality of all susceptible ash trees (stems ≥ 2.54 cm diameter at breast height; dbh) rapidly increased from 40% to 99.7%. Simultaneously, as mature ash trees died, reproduction and new ash regeneration stopped (Klooster et al., 2014). Currently the ash population in this area consists only of established seedlings and saplings (Klooster et al.,

2014). Ash seedlings have a high rate of natural mortality (Good and Good, 1972).

Therefore the likelihood that an individual seedling will reach sexual maturity is low.

Additionally, emerald ash borer populations are persisting at low densities (Herms personal communication) and may colonize ash saplings as individuals grow large enough to become susceptible to attack (McCullough and Siegert, 2007).

119

Populations that have undergone significant reductions in size are more susceptible to inbreeding and genetic drift. Mating between close relatives (i.e., inbreeding) decreases genetic variation (Frankham, 1996). Additionally, inbreeding increases the chance that alleles with lower survival or fitness remain in the population

(Frankham, 1995). Furthermore, in small populations alleles are often removed from the gene pool through random mortality (i.e., genetic drift), further reducing genetic variation

(Frankham, 1995). If fragmentation occurs for several generations, inbreeding and genetic drift continue, which further decreases genetic variation (Keller and Waller,

2002). Low genetic variation can also lead to lower fitness and survival (Husband,

Schemske, 1996; Pautasso, 2009; Ilves et al., 2013). Thus, small populations may be more prone to local extinction.

The loss of genetic variation in ash populations as a result of the emerald ash borer invasion is also a concern. In particular, long-term impact of emerald ash borer in forests may decrease genetic variation of the surviving ash population by reducing the effective population size (i.e., individuals contributing to reproduction). As emerald ash borer kills trees, populations of mature ash trees may become fragmented and gene flow reduced as fewer individuals contribute to mating (Potter et al., 2012; Ilves et al., 2013).

Additionally, neighboring trees are more likely to be closely related (Heuertz et al.,

2001), thus increasing the occurrence of inbreeding.

The effects of population decline on genetic variation have been studied in animal and plant communities. For example, populations of endangered bird species have low genetic variation, high rates of inbreeding, and deviation from Hardy-Weinberg due to

120 breeding populations being too low when management was initiated (Ramstad et al.,

2013; Woolaver et al., 2013). Populations of a critically endangered perennial herb

(Lingularia sibirica) have low genetic variation and high rates of self-pollination due to human disturbance and habitat fragmentation (Ilves et al., 2013). Fragmented landscapes also cause losses of alleles, decreased heterozygosity, and evidence of inbreeding in various plant populations (Aguilar et al., 2008; Vranckx et al., 2012).

Loss of genetic variation has occurred within forest tree populations following disturbances. Secondary succession in forests is initiated by shade intolerant pioneer species (Runkle, 1984; Brokaw, 1985). In intact forests, pioneer species are typically present at low numbers and have suppressed growth until canopy gaps are created that increase availability of light (Runkle, 1984). As a result, many pioneer species have low allelic diversity, because individuals that colonize gaps in localized areas tend to be closely related (Davies et al., 2010).

Hemlock woolly adelgid (Adelges tsugae) is an invasive hemipteran insect native to Asia and western North America that is causing widespread mortality of hemlock trees

(Tsuga spp.) throughout eastern United States (McClure et al., 2003). Potter et al. (2012) studied the effects of widespread hemlock mortality on genetic variation and found allelic variation, proportion of heterozygote genotypes, and inbreeding of mature hemlock trees to be similar between sites with and without hemlock woolly adelgid. However, they suggested that the pattern may be different among seed and seedling populations.

To date, no studies have investigated the effects of emerald ash borer on the genetic structure of ashes in North America. Population genetics of green ash (F.

121 pennsylvanica) in northeast Ohio was quantified in 2005 by Hausman et al. (2014), before EAB had caused widespread mortality of ash. Hausman et al. (2014) reported that green ash had high allelic richness and no evidence of inbreeding. In southeastern

Michigan, however, where ash mortality exceeded 99% by 2009 (Klooster et al. 2014), it is possible that emerald ash borer-induced mortality has decreased the effective population size, creating higher rates of inbreeding and a genetic bottleneck in the surviving “orphaned” seedling population. Based on age approximations of seedlings (1-

7 years) and saplings (7-20 years) (Burns and Honkala, 1990), saplings established prior to the emerald ash borer invasion and seedlings were progeny of the few mature ash trees that continued to survive during peak mortality. The relationship between size and age of seedlings and saplings is complex, because seedling growth of shade intolerant species is often reduced until a gap is formed in the canopy (Bray, 1956; Runkle and Yetter, 1987;

Runkle, 1990). However, height can be used to estimate age of seedlings and saplings.

Therefore, comparisons of population genetic structure of different size classes of ash could indicate whether there has been a loss of genetic variation in the population (i.e., genetic bottleneck). If a genetic bottleneck did occur in Michigan, the surviving ash population in Michigan may be further threatened by potentially lower fitness caused by inbreeding depression (Hu et al., 2010a; Potter et al., 2012).

Managers of ash populations should consider the effects of genetic variation to prevent deleterious effects of inbreeding that are sometimes the consequences of captive breeding programs (Frankham, 1995; Wee et al., 2012, Ramstad et al., 2013; Roberts et al. 2013; Woolaver et al. 2013). In particular, preserving genetic variation of ash requires

122 understanding the reproductive ecology including pollination and seed dispersal, as well as magnitude of background genetic variation in the population prior to the emerald ash borer invasion. Most ash species, including white ash (F. americana), green ash (F. pennsylvanica), and black ash (F. nigra) are dioecious (separate male and female trees) and wind pollinated (Burns and Honkala, 1990; Wallander, 2008). It has been estimated that ash pollen can be dispersed up to 200 m (Heuertz et al., 2003). Since multiple males can fertilize flowers from a single female tree, genetic variation increases when density of males is high and individuals are close enough to successfully pollinate female trees

(Heuertz et al., 2001). Hausman et al. (2014) reported that 200 seeds from at least five mother trees captured 99% of the genetic variation in a population. This suggests that females need to be protected to maintain seed production, but protecting males may be more important for preserving genetic variation.

Systemic insecticides can effectively protect ash trees. In particular, emamectin benzoate is a highly effective, trunk-injected, systemic insecticide that kills emerald ash borer adults and neonate larvae (McCullough et al., 2011). In forests, large-scale insecticide treatments may protect enough ash trees to preserve the ash gene pool.

Furthermore, increasing the density of treated ash trees may increase the amount of genetic variation conserved within the next generation. The Five Rivers MetroParks

(FRMP) in Montgomery County, in southwestern Ohio initiated a large-scale insecticide program in 2011 to treat 600 mature ash trees every two year with the goal of conserving the ash population within their parks. The objective of this study was to investigate whether this management strategy effectively conserves genetic variation in green ash

123 seedlings, and determine whether there was a relationship between density of treated ash trees and amount of genetic variation in the seedlings.

In this study, I compared genetic variation in green ash seedling and sapling populations in southeastern Michigan and FRMP to (1) determine whether widespread emerald ash borer-induced ash mortality caused a detectable bottleneck in the smallest size class of ash seedlings, and (2) determine whether insecticide treatments can maintain genetic variation. Specifically, I hypothesized that in Michigan where mortality of reproductive trees exceeded 99% (Klooster et al. 2014), the smallest seedlings would have lower genetic variation and higher rates of inbreeding than the larger size classes. In southwestern Ohio, where mortality of ash is lower because EAB has not been established as long, I hypothesized that (1) the first-year ash seedlings would have lower genetic variation and higher rates of inbreeding than the larger size classes, and (2) that treating higher densities of ash trees would result in greater genetic variation and lower rates of inbreeding in the seedling population. Finally, I compared the population structure of Ohio and Michigan to determine differentiation at different stages of the emerald ash borer invasion.

Methods

Study Sites

In 2004, transects were established by Smith (2006) at five parks within the Upper

Huron River Watershed in southeastern Michigan: Highland, Island Lake, and Proud

Lake State Recreation Areas, and Hudson Mills and Indian Springs Metroparks.

124

Transects consisted of three non-overlapping replicate plots located 5 to 25 m apart. Plots were 18-m radius (0.1 ha) that contained an 8-m subplot in the center and four microplots

(4m2), each located 8 m from the plot center along one of the four cardinal compass directions. Each plot had at least two mature ash trees (≥10 cm diameter at breast height, dbh). Transects were extensively surveyed from 2004 to 2010 to monitor EAB-induced ash decline and mortality, and to quantify ash population demographics (densities of seedlings, saplings, and immature trees) in the understory (Smith, 2006; Klooster et al.,

2014; Smith et al., 2015).

In 2014, I established 24 quadrats (100 m X 100 m; 1 ha) at six metroparks within

FRMP. Quadrats were located randomly within each park using ArcMap 10.4. Within each quadrat, I located four 0.1 ha plots; two proximal plots within 50 m of the quadrat center and two distal plots 150 m from the center, with one on a randomly selected azimuth (degree bearing) and the other approximately 90 degrees (clockwise) from the first. Plots were established following the design used in southeastern Michigan (Smith,

2006; Smith et al. 2015). At the beginning of the study, approximately 75% of green and white ash trees were still alive at Englewood, Germantown, and Sugarcreek Metroparks, and approximately 25% were still alive at Cox Arboretum, Taylorsville, and Twin Creek

Metroparks. Therefore, I identified quadrats at Englewood, Sugarcreek, and Germantown as low EAB impact and quadrats at Cox Arboretum, Taylorsville, and Twin Creek as high EAB impact.

FRMP began a large-scale treatment program in 2011 to biannually treat 600 ash trees with TREE-äge (emamectin benzoate 4% ME, Arborjet, Woburn, MA, USA). Trees

125 were retreated in 2013 and 2015. Treated ash trees ranged in size from 10 to 200 cm trunk diameter at breast height (dbh) and included individuals from all five ash species present in the parks (green ash, white ash, blue ash, pumpkin ash, and black ash). Plots were grouped based on the density of treated ashes within 200 m (12.5 ha) of each plot (<

1, 2-4, 5-7, and > 7 treated ash/12.5 ha).

Seedling and sapling sampling protocol

Green ash seedlings and saplings were sampled from five parks in Michigan

(Highland, Island Lake, and Proud Lake State Recreation Areas, and Hudson Mills and

Indian Springs Metroparks) and six FRMP metroparks in Ohio (Cox Arboretum,

Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek Metroparks).

Seedlings were categorized according to the following size classes: newly germinated ash seedlings with cotyledons (class 1), ash stems ≤ 25 cm tall without cotyledons (class 2), ash stems > 25 cm tall and ≤ 1.5 m tall (class 3), and ash stems ≥ 1.5 m tall and less than

2.5 cm diameter at breast height (dbh) (saplings). Seedlings were sampled in 1m2 quadrats located 8 m from the plot center, adjacent to each 4m2 microplot. Within each 1 m2 quadrat, each class 1 and class 2 ash seedlings were sampled by clipping their stems at ground level. Each class 3 seedling in the quadrats was sampled by collecting four per plant. Each sapling within the subplot (8 m radius) of the plots was sampled by collecting foliage. All samples were stored at -20oC until they were identified to species, then stored in individual microcentrifuge tubes at -80oC.

126

DNA Extraction and fragment analysis

I randomly selected 64 green ash seedlings (classes 2-3) and saplings from

Michigan and 111 green ash seedlings (classes 1 to 3) from Ohio for genetic analysis. I selected one individual per size class per plot to limit sample bias of related individuals.

In Michigan, I conducted genetic analysis on 29 class 2 seedlings, 23 class 3 seedlings, and 12 saplings from Highlands (8 plots), Hudson Mills (7 plots), Island Lake (8 plots),

Indian Springs (9 plots), and Proud Lake (8 plots). Since reproduction and regeneration has ceased in southeast Michigan, class 1 seedlings could not be analyzed from these plots. In Ohio, I conducted genetic analysis on 34 class 1 seedlings, 57 class 2 seedlings, and 20 class 3 seedlings from Cox Arboretum (9 plots), Englewood (9 plots),

Germantown (15 plots), Sugarcreek (7 plots), Taylorsville (11 plots), and Twin Creek (14 plots). The density of saplings was low from Ohio and most of the saplings were blue ash. Therefore, I did not analyze saplings from Ohio.

I extracted 50-80 mg of nuclear DNA using the DNeasy Plant Mini Kit

(QIAGEN). Leaf samples were homogenized using liquid nitrogen and a sterilized mortar and pestle. After homogenization, I added 500 ul of AP1 buffer that was pre-heated to

65oC. Samples were vortexed and then placed on a shaker for 15 minutes (Al-Saghir

2009). The remaining procedure followed the kit recommendations. Extracted ash DNA was stored at -20oC.

Five nuclear microsatellites were used to quantify population genetics parameters including M2-30, 1.19, and 3.1(Brachet et al., 1999), and FEMSATL1and FEMSATL19

(Lefort et al., 1999). Forward primers for each microsatellite contained a fluorescent

127

Well-Red (D2, D3, or D4) that was used for fragment analysis. Polymerase chain reaction (PCR) was used to amplify microsatellite fragments for each sample. Reactions contained 10ul of GoTaq Green Master Mix (Promega Co., Madison, WI), 0.5 ul each of forward and reverse primers, 8 ul of molecular grade H2O, and 1 ul of DNA sample.

Reaction protocol included initial denaturation for 4 minutes at 94oC, followed by 35 cycles of denaturation (94oC for 40 seconds, annealing (45 – 56oC for 40 seconds), and elongation (72oC for 45 seconds). Annealing temperature differed by microsatellite; 1.19

(45oC), M2-30 and 3.1 (52oC), and FEMSATL1 and FEMSATL19 (56oC). PCR amplifications were confirmed using gel electrophoresis (3% agarose) to detect the presence of bands. DNA fragment were analyzed using a CEQ8800 Beckman Coulter genotype analyzer.

Statistical analysis

The following were recorded for microsatellite loci: allelic richness (Na; the number of different alleles), effective population size (Ne; the number of individuals contributing to reproduction), observed and expected heterozygosity (Ho and HE; respectively), and Wright’s inbreeding coefficient (FIS). Additionally, deviation from

Hardy-Weinberg equilibrium (HWE) was tested by comparing observed and expected heterozygosity using Chi-Square analysis. Lastly, I estimated the occurrence of null alleles to determine whether there may have been alleles that failed to amplify during

PCR. These data were used to compare population genetics for each population.

Population genetic information and deviation from Hardy-Weinberg were conducted

128 using GenAlEx 6.503 (Peakall and Smouse, 2006) and occurrence of null alleles was conducted using Micro-Checker (van Oosterhout et al. 2003).

Seedlings and saplings were sampled in five parks in Michigan and seedlings were sampled from six parks in Ohio. Populations were defined according to state

(Michigan or Ohio), park, and size class. Analyses for state and park were conducted by pooling individuals for each size class into a single population (e.g., class 1 seedlings from all parks were compared to class 2 seedlings from all parks). Lastly, analyses for size class were conducted by designating both state and size class (i.e., MI-class 2, MI- class 3, MI-sap, OH-class 1, OH-class 2, OH-class 3). Genetic differentiation between populations was estimated using FST values. These estimates were conducted using analysis of molecular variance (AMOVA) with 999 pairwise permutations. All calculations were conducted using GenAlEx 6.503 (Peakall and Smouse, 2006).

Significant differences were designated at P < 0.05.

Ordination using principal component analysis (PCA) was used to compare the spatial genetic structure of each population. I conducted separate analyses on samples from Ohio and Michigan to determine the within state genetic structure. For within-state analyses, I compared populations from each size class and park. In Ohio, I also compared the genetic structure by EAB impact and density of treated ashes. Lastly, I analyzed the genetic structure of all samples to compare Ohio to Michigan. Differences between populations were evaluated using the adonis function to calculate statistical differences.

All analyses were conducted with package Vegan in R 3.2.3. Significant differences were accepted in P < 0.05.

129

Results

I collected genetic data on 171 green ash seedlings and saplings from Michigan and Ohio using five nuclear microsatellite loci. Total samples per locus ranged from 107

(locus 3.1) to 171 (locus M2-30). Allelic richness was highest for microsatellites M2-30

(NA =20 alleles) and FEMSATL1 (NA=17 alleles) and moderate for FEMSATL19 (NA

=11 alleles) and 1.19 (NA =10 alleles). Only three alleles were present for locus 3.1 in both Michigan and Ohio populations. The number of effective alleles (NE) was highest for FEMSATL1 (NE = 11.5), moderate for M2-30 (NE =5.8), FEMSATL19 (NE =5.0), and

1.19 (NE =4.8), and lowest for 3.1 (NE =2.1 alleles). Observed heterozygosity was similar to expected for M2-30, FEMSATL1, FEMSATL19, and 3.1 (HO=0.708, 0.708, 0.603, and 0.430; respectively), but lower than expected for 1.19 (0.395 vs. 0.790; HO vs. HE).

Genetic variation in Michigan

I conducted genetic analysis on 64 individual samples from five parks in

Michigan. Samples were categorized based on their size class; class 2 (N=29), class 3

(N=23), and saplings (N=12). Overall, loci M2-30 and FEMSATL1 had the highest allelic richness (NA=17 and 14 alleles; respectively) and loci 3.1 the lowest (NA=3 alleles). Number of effective alleles (NE) indicated the number of higher frequency alleles

(i.e., alleles that will be more likely to be present in the next generation). I estimated that there were 9 effective alleles for both M2-30 and FEMSATL1, and FEMSATL19 and

1.19 had 5 and 4 effective alleles (respectively). Locus 3.1, which had the lowest allelic richness, only had two effective alleles.

130

Genetic variation of class 2 seedlings was variable across different microsatellite loci. Allelic richness was highest for loci M2-30 and FEMSATL1 (NA =12 and 11 alleles; respectively) and lowest for locus 3.1 (NA =2). Similarly, the number of effective alleles was highest for M2-30 and FEMSATL1 (NE =7) (Table 5.1). Observed heterozygosity was significantly lower than expected for three of the five microsatellite loci

(FEMSATL1, FEMSATL19, and 1.19), indicating that allelic distribution for these loci significantly deviate from Hardy-Weinberg equilibrium (HWE) (Table 5.1). However, there was evidence of null alleles for loci FEMSATL1, FEMSATL19, and 1.19.

Therefore, it is likely that these loci were in HWE.

Genetic variation of the class 3 seedlings was also variable by loci. Similar to the class 2 seedlings, allelic richness was highest for loci M2-30 and FEMSATL1 (NA=12 and 11 alleles; respectively) and lowest for loci 3.1 (NA=3 alleles). The number of effective alleles was highest for FEMSATL1 and M2-30 (NE =8.2 and 7.0 alleles; respectively) (Table 5.1). Heterozygosity was lower than expected for loci FEMSATL19,

M2-30, and 1.19. However, this difference was only significant for loci M2-30 and 1.19, indicating significant deviation from HWE (Table 5.1). I found evidence of null alleles with loci M2-30 and 1.19, therefore may not be deviating from HWE. The sample size was low for FEMSATL19 due to low amplification for this locus. This was likely the reason why I did not detect a significant deviation from HWE for this locus.

There was high amplification of each locus for the sapling population. Allelic richness was highest for M2-30 and FEMSATL1 (NA =14 and 9 alleles; respectively).

The number of effective alleles was highest for M2-30 (NE =11.5) and lowest for 3.1 (NE

131

= 2) (Table 5.1). Heterozygosity was high for loci FEMSATL1, FEMSATL19, and M2-

30 and was not significantly different from expected heterozygosity. Additionally, loci

FEMSATL19 and M2-30 had an excess of heterozygotes. Locus 1.19 had lower than expected heterozygosity, however there was evidence of null alleles for this locus.

Overall, no loci significantly deviated from HWE in the sapling population (Table 5.1).

Overall, the results indicated different patterns of genetic variation across size classes in Michigan. Mean allelic richness and effective alleles were similar across all size classes (Table 5.1). Class 2 and class 3 seedlings had lower than expected heterozygotes as indicated by the fixation index (FIS=0.213 ± 0.089 and 0.163 ± 0.070), whereas observed heterozygosity of saplings was similar to expected (Table 5.1). The likelihood of null alleles in class 2 and class 3 seedlings may have impacted the proportions of observed heterozygotes. These results indicate that class 2 and class 3 seedlings may be deviating from HWE, however testing of additional alleles may be needed to strengthen these conclusions.

I also compared genetic variation within each park with size classes pooled.

Allelic richness was highest at Indian Springs (IS) and Proud Lake (PL), although mean allelic richness did not differ significantly (Table 5.2). Effective alleles was highest at IS

(5.4 ± 1.5 alleles), but seedlings and saplings from the remaining parks averaged between

4 and 5 effective alleles. Heterozygosity was lower than expected at Island Lake (IL), IS, and PL, as indicated by inbreeding coefficient (FIS) (Table 5.2). At Highlands (HL) and

Hudson Mills (HM), observed and expected heterozygosity was similar. This indicates

132 that inbreeding may have been occurring at IL, IS, and PL, due to the loss of heterozygosity.

133

Table 5.1: Population genetics of green ash seedlings and saplings from five state and metroparks in the Upper Huron Watershed in southeastern Michigan. Samples were collected in 2014, 12 years after emerald ash borer (EAB) was first detected in this area. Seedlings were divided into size classes: class 2 (stems ≤ 0.25 m tall) and class 3 (stems > 0.25 m, < 1.5 m tall); saplings were > 1.5 m tall, but unsusceptible to EAB (< 2.54 cm trunk diameter).

Null Loci N NA NE alleles HO HE FIS P = (Y/N) Class 2 sdlgs FEMSATL1 21 11 6.8 Y 0.667 0.854 0.219 <0.0001 *** FEMSATL19 18 7 4.6 Y 0.556 0.781 0.289 0.001 *** 134 M2-30 29 12 7.1 N 0.862 0.859 -0.004 0.84 ns

1.19 28 7 3.5 Y 0.357 0.716 0.501 0.001 ** 3.1 11 2 1.9 N 0.455 0.483 0.060 0.84 ns Mean 21.4 ± 3.3 7.8 ± 1.8 4.8 ± 1.0 0.579 ± 0.087 0.738 ± 0.069 0.213 ± 0.089 Class 3 sdlgs FEMSATL1 21 11 8.2 N 0.810 0.878 0.078 0.59 ns FEMSATL19 7 5 3.8 N 0.571 0.735 0.222 0.49 ns M2-30 23 12 7.0 Y 0.652 0.856 0.238 <0.0001 *** 1.19 23 8 4.7 Y 0.522 0.788 0.338 0.05 * 3.1 18 3 2.1 N 0.556 0.523 -0.062 0.72 ns Mean 18.4 ± 3.0 7.8 ± 1.7 5.1 ± 1.1 0.622 ± 0.052 0.756 ± 0.063 0.163 ± 0.070 (Continued)

134

(Table 5.1: Continued)

Null Loci N NA NE alleles HO HE FIS P = (Y/N) Saplings FEMSATL1 11 9 5.6 N 0.818 0.822 0.005 0.79 ns FEMSATL19 10 6 3.6 N 0.800 0.720 -0.111 0.87 ns M2-30 12 14 11.5 N 1.000 0.913 -0.095 0.23 ns 1.19 12 6 4.0 Y 0.333 0.750 0.556 0.12 ns 3.1 10 2 2.0 N 0.600 0.500 -0.200 0.53 ns

135 Mean 11.0 ± 0.5 7.4 ± 2.0 5.3 ± 1.6 0.710 ± 0.114 0.741 ± 0.069 0.031 ± 0.135

N = Number of samples, NA = Allelic richness, NE = Effective alleles, HO = Observed heterozygosity, HE = Expected heterozygosity, FIS = Wright’s inbreeding coefficient. P value indicates significant deviation from Hardy-Weinberg equilibrium for each locus.

135

Table 5.2: Population genetics of green ash from five state and metroparks in the Upper Huron Watershed in southeastern Michigan. Samples were collected from seedlings and saplings in 2014 from plots established by Smith (2006).

Park N N N H H F A E O E IS Highlands (HL) 10.2 ± 1.0 6.0± 1.2 4.6 ± 0.8 0.689 ± 0.056 0.737 ± 0.069 0.007 ± 0.169

Hudson Mills (HM) 6.2 ± 0.8 5.8 ± 1.1 4.7 ± 1.0 0.695 ± 0.130 0.727 ± 0.070 0.046 ± 0.141

136 Island Lake (IL) 8.6 ± 1.2 6.0 ± 1.5 4.7 ± 1.5 0.610 ± 0.077 0.719 ± 0.065 0.119 ± 0.151

Indian Springs (IS) 11.0 ± 0.9 7.4 ± 1.9 5.4 ± 1.5 0.575 ± 0.130 0.734 ± 0.081 0.218 ± 0.137 Proud Lake (PL) 14.8 ± 2.2 7.2 ± 1.6 4.2 ± 1.0 0.620 ± 0.099 0.708 ± 0.064 0.136 ± 0.115

N = Number of samples, NA = Allelic richness, NE = Effective alleles, HO = Observed heterozygosity, HE = Expected heterozygosity, FIS = Wright’s inbreeding coefficient. P value indicates significant deviation from Hardy-Weinberg equilibrium for each locus.

136

There was significant genetic differentiation between all size classes. Population genetic structure of class 2 seedlings was significantly different than class 3 seedlings

(FST = 0.047, P = 0.001) and saplings (FST = 0.043, P = 0.002). Additionally, class 3 seedlings were significantly differentiated from saplings (FST = 0.035, P = 0.004).

Differentiation of the genetic structure of each population was further supported by principal component analysis (PCA). Although there was some overlap for each size class, the genetic structure of class 2 seedlings was more tightly clustered and distinct from both the class 3 seedlings and saplings (F = 8.39, R2 = 0.20, P < 0.0001) (Figure

5.1a). Class 3 seedlings were also comprised of individuals with similar genotypes, which were represented by a smaller ellipse with 95% confidence interval than the sapling population (Figure 5.1a). Overall, the class 2 seedlings were the most similar to each other, but shared genotypes with class 3 seedlings and saplings. However, there were unique genotypes that were only detected in the class 3 seedling and sapling populations.

I also compared the genetic structure of each of the five parks. This was conducted by pooling individuals from each size class and only comparing by park. There were no differences between genetic structures of ash from each park (F = 0.52, R2 =

0.02, P = 0.9) (Figure 5.1b). However, Indian Springs (IS) and Island Lake (IL) contained unique genotypes that were not found in the other parks. Furthermore, the unique genotypes were from class 3 seedlings and saplings as indicated by triangle and square shaped symbols, respectively (Figure 5.1b).

137

138

A B

Figure 5.1: Principal component analysis of genetic variation collected from green ash seedlings and saplings from the Upper Huron Watershed in southeastern Michigan. (A) Differentiation of the genetic structure of class 2 seedlings (CL2), class 3 seedlings (CL3), and saplings (SAP). (B) Differentiation of the genetic structure for each state or metropark; Highland (HL), Hudson Mills (HM), Island Lake (IL), Indian Springs (IS), and Proud Lake (PL). Ellipses indicate 95% confidence intervals for each group.

138

Genetic variation in Ohio (class 1 to 3 seedlings)

For seedlings sampled from FRMP in Ohio, microsatellite loci FEMSATL1 and

M2-30 had the highest allelic richness (NA=16 and 14 alleles; respectively) and locus 3.1 had the lowest allelic richness (NA=3 alleles). FEMSATL1 had the most number of effective alleles (NE=10 alleles) and 3.1 had the fewest effective alleles (NE=2 alleles).

The remaining microsatellite loci had approximately the same number of effective alleles

(Ne=4 to 5 alleles).

Genetic variation of class 1 seedlings was variable across different microsatellite loci. Allelic richness was highest for loci FEMSATL1 and M2-30 (NA =15 and 13 alleles; respectively) and lowest for locus 3.1 (NA =2). The number of effective alleles was highest for FEMSATL1 (NE =9) and second highest for FEMSATL19 (NE =5) (Table

5.2). The proportion of heterozygotes was lower than expected for all microsatellite loci

(Table 5.2). Differences between observed and expected heterozygosity were only significant for FEMSATL19, M2-30, and 1.19 (P< 0.05), which were loci that likely had null alleles. Therefore, it is not clear whether the class 1 seedlings were deviating from

HWE.

Genetic variation of class 2 seedlings was also variable across different microsatellite loci. Allelic richness was highest for FEMSATL1 and M2-30 (NA =13 alleles for each locus). However, the number of effective alleles was higher for

FEMSATL1 (NE =11 alleles). Loci M2-30, FEMSATL19, and 1.19 had approximately 4 to 5 effective alleles and locus 3.1 only had 2 effective alleles (Table 5.2). The proportion of heterozygotes was lower than expected for all microsatellite loci and null alleles were

139 likely for all loci, except for 3.1. Thus, it is unclear whether class 2 seedlings were deviating from HWE.

Lastly, genetic variation of class 3 seedlings was more similar across the different microsatellite loci. Allelic richness ranged from 2 alleles (3.1) to 9 alleles (FEMSATL1).

FEMSATL1 had the most effective alleles (NE =7 alleles) and 3.1 had the fewest effective alleles (NE =2 alleles) (Table 5.2). Observed heterozygosity was lower than expected for each microsatellite loci, however only 1.19 significantly deviated from

HWE (P = 0.001) (Table 5.2). Null alleles were likely for locus 1.19, but not for the other loci in this population. However, based on the significant tests for HWE, this suggests that the class 3 seedlings were in HWE.

Overall, allelic richness varied by size class, but the number of effective alleles and heterozygosity was similar across all size classes. Specifically, allelic richness was higher for class 1 (NA=9.4 ± 2.3) and class 2 (NA=9.0 ± 1.9) seedlings then the class 3 seedlings (NA=6.4 ± 1.2) (Table 5.2). The proportion of heterozygotes within each population was lowest for class 2 seedlings, but both observed and expected heterozygosity was similar among all size classes. As a result, the fixation indices (FIS) ranged from 0.25 to 0.3 for each size class, showing that there were fewer than expected heterozygotes for all size classes of seedlings in Ohio.

140

Table 5.3: Population genetics of ash seedlings from the Five Rivers MetroParks (FRMP) in southwestern Ohio. Samples were collected in 2014, 7 years after emerald ash borer (EAB) was first detected in this area and 4 years after FRMP began an insecticide program to protect mature ash trees. Seedlings were divided into three size classes: class 1 (stems ≤ 0.25with cotyledons), class 2 (stems ≤ 0.25 m tall without cotyledons) and class 3 (stems > 0.25 m, < 1.5 m tall).

Null Loci N NA NE alleles HO HE FIS P = (Y/N) Class 1 sdlgs FEMSATL1 26 15 8.9 Y 0.692 0.888 0.220 0.19 ns FEMSATL19 31 10 5.0 N 0.677 0.800 0.153 <0.0001 *** M2-30 33 13 3.6 Y 0.576 0.724 0.204 0.03 *

141 1.19 23 7 3.4 Y 0.522 0.703 0.258 0.05 *

3.1 20 2 2.0 N 0.300 0.495 0.394 0.08 ns Mean 26.6 ± 2.4 9.4 ± 2.3 4.6 ± 1.2 0.553 ± 0.071 0.722 ± 0.065 0.246 ± 0.041 Class 2 sdlgs FEMSATL1 37 13 10.6 Y 0.622 0.906 0.314 0.03 * FEMSATL19 50 8 4.0 Y 0.540 0.749 0.279 <0.0001 *** M2-30 55 13 4.3 Y 0.673 0.765 0.121 0.92 ns 1.19 50 8 4.6 Y 0.340 0.784 0.567 <0.0001 *** 3.1 36 3 2.2 N 0.417 0.554 0.248 <0.0001 *** Mean 45.6 ± 3.8 9.0 ± 1.9 5.1 ± 1.4 0.518 ± 0.062 0.752 ± 0.057 0.306 ± 0.073 (Continued)

141

(Table 5.3: Continued)

Null Loci N NA NE alleles HO HE FIS P = (Y/N) Class 3 sdlgs FEMSATL1 15 9 6.8 N 0.733 0.853 0.141 0.18 ns FEMSATL19 15 6 4.1 N 0.600 0.753 0.204 0.08 ns M2-30 19 8 3.7 N 0.684 0.729 0.061 0.48 ns 1.19 16 7 5.4 Y 0.313 0.816 0.617 0.001 *** 3.1 12 2 1.9 N 0.333 0.486 0.314 0.28 ns Mean 15.4 ± 1.1 6.4 ± 1.2 4.4 ± 0.8 0.533 ± 0.088 0.728 ± 0.064 0.267 ± 0.097

142 N = Number of samples, NA = Allelic richness, NE = Effective alleles, HO = Observed heterozygosity, HE = Expected heterozygosity, FIS = Wright’s inbreeding coefficient. P value indicates significant deviation from Hardy-Weinberg equilibrium for each locus.

142

Table 5.4: Population genetics of green ash from six parks within the Five Rivers Metroparks (FRMP) in southwestern Ohio. Samples were collected from green ash seedlings in 2014.

N N N H H F A E O E IS

Cox Arboretum 9.6 ± 1.0 7.0 ± 1.6 4.5 ± 1.3 0.667 ± 0.063 0.703 ± 0.081 0.006 ± 0.123 (CA) Englewood (EN) 13.8 ± 0.6 6.0 ± 0.9 4.1 ± 0.9 0.502 ± 0.121 0.709 ± 0.058 0.323 ± 0.138

Germantown (GT) 21.6 ± 1.8 7.8 ± 1.7 5.0 ± 1.1 0.498 ± 0.080 0.740 ± 0.077 0.317 ± 0.087

143

Sugarcreek (SC) 8.2 ± 1.4 4.8 ± 0.8 3.2 ± 0.6 0.411 ± 0.073 0.641 ± 0.060 0.375 ± 0.075

Taylorsville (TA) 18.2 ± 0.8 7.6 ± 1.4 4.9 ± 1.1 0.523 ± 0.088 0.748 ± 0.060 0.311 ± 0.083

Twin Creek (TC) 16.0 ± 1.9 6.4 ± 1.3 4.3 ± 0.8 0.610 ± 0.083 0.725 ± 0.061 0.132 ± 0.138

N = Number of samples, NA = Allelic richness, NE = Effective alleles, HO = Observed heterozygosity, HE = Expected heterozygosity, FIS = Wright’s inbreeding coefficient. P value indicates significant deviation from Hardy-Weinberg equilibrium for each locus.

143

I also compared genetic variation within each metropark in Ohio. Allelic richness was lowest at Sugarcreek (SC, 4.8 ± 0.8 alleles) and highest at Germantown (GT, 7.8 ±

1.7 alleles) and Taylorsville (TA, 7.6 ± 1.4 alleles) (Table 5.4). Effective alleles were lowest at SC (3.2 ± 0.6 alleles), but similar at all other parks ranging from four to five alleles. Observed heterozygosity was lower than expected at Englewood (EN), GT, SC, and TA with mean inbreeding coefficients (FIS) > 0.3 (Table 5.4). Observed heterozygosity was similar to expected at Cox Arboretum (CA) and slightly lower than expected at Twin Creek (TC) (Table 5.4).

Seedling size classes in Ohio were genetic differentiated. Class 1 seedlings were significantly differentiated from class 2 seedlings (FST = 0.010, P = 0.019). However, no differences were detected between class 1 and class 3 seedlings (FST=0.000, P = 0.43).

Also, no differences were found between class 2 and class 3 seedlings (FST=0.005, P =

0.180). This suggests that the genetic structure of class 1 and class 3 seedlings were more similar than class 1 and class 2. Finally, there was more genetic differentiation between class 1 and class 3 than between class 2 and class 3 seedlings.

Principal component analysis was used to compare genetic structure of seedlings by size class, park, EAB impact, and density of treated ash trees. I found that the genetic structure of green ash varied by size class. Specifically, class 3 seedlings had the lowest genetic variation as indicated by the tight ordination clustering (Figure 5.2a). Class 1 seedlings were more diverse group, with widespread clustering (Figure 5.2a). However, there were no significant differences among the three size classes (F=1.20, df=2,

R2=0.02, P=0.15).

144

Comparisons of seedlings from different parks little differences in genetic structure in their ash populations (F=1.01, df=5, R2=0.05, P=0.44) (Figure 5.2b). Within parks, seedlings were most similar at Englewood Metropark (Figure 5.2b). All other parks had higher within park diversity, but similar combinations of genotypes.

Genetic structure did not differ among seedlings from parks with low and high

EAB impacts (F=0.95, df=1, R2=0.009, P=0.53) (Figure 5.3a). Similarly, genetic structure did not differ by density of treated ash trees (F=1.39, df=1, R2=0.01, P=0.09).

However, there were more unique genotype combinations from plots with higher densities of treated ash trees (Figure 5.3b).

145

146 A B

Figure 5.2: Principal component analysis on the genetic variation of green ash seedlings in southwestern Ohio at Five Rivers MetroParks. (A) Differentiation among seedling size classes (Classes 1-3) and (B) Differentiation among metroparks (Cox Arboretum, Englewood, Germantown, Sugarcreek, Taylorsville, and Twin Creek). Ellipses indicate 95% confidence intervals for each group.

146

147

A B

Figure 5.3: Principal component analysis on the genetic variation of green ash seedlings in southwestern Ohio at Five Rivers MetroParks. (A) Differentiation among low and high EAB impact sites and (B) Differentiation among density of treated ash trees. Ellipses indicate 95% confidence intervals for each group.

147

Michigan vs. Ohio

Genetic variation in ash seedling population was similar in Michigan and Ohio. In particular, Michigan and Ohio had similar allelic richness (NA = 10.2 ± 2.4 and 10.6 ±

2.6; respectively) and number of effective alleles (NE=5.8 ± 1.3 and 5.1 ± 1.5; respectively) (Table 5.4). Observed heterozygosity was lower than expected in Ohio (HO

= 0.534 ± 0.066, HE = 0.755 ± 0.061) and Michigan (HO = 0.629 ± 0.072, HE = 0.771 ±

0.070) (Table 5.4). Genetic variation of loci FEMSATL1, M2-30, and 1.19 significantly deviated from HWE (P < 0.002) in Michigan and all loci deviated from HWE (P< 0.001) in Ohio (Table 5.4). Since null alleles were likely for loci FEMSATL1 and 1.19 in

Michigan, genotypes for these loci may have been in HWE. Similarly, since all loci likely had null alleles in Ohio, I cannot make conclusions on deviation from HWE.

I found genetic differentiation between the Michigan and Ohio populations.

Specifically, there was significant genetic differentiation (FST = 0.04, P = 0.001) and distinct genetic structure of the two ash populations (F=4.83, df=1, R2=0.03, P<0.0001)

(Figure 5.4).

Additionally, significant genetic structure distinguished seedling size classes and site (F=1.49, df=4, R2=0.03, P=0.0009) (Figure 5.5a-b). In Michigan, class 3 seedlings and saplings captured all genetic structure unique to Michigan (Figure 5.5a). However, class 2 seedlings contained only genotypes that were shared between Ohio and Michigan.

In Ohio, all size classes contained individuals with genotypes that were unique to Ohio

(Figure 5.5b).

148

Table 5.5: Population genetics of green ash seedlings and saplings from the Upper Huron Watershed in southeast Michigan and seedlings Five Rivers Metroparks (FRMP) in southwestern Ohio. Samples were collected in 2014 and analyzed with five microsatellite loci.

Null Loci N NA NE alleles HO HE F P = (Y/N) Michigan FEMSATL1 53 14 8.8 Y 0.755 0.886 0.148 0.002 ** FEMSATL19 35 8 5.3 Y 0.629 0.810 0.224 0.12 ns M2-30 64 17 8.7 N 0.813 0.885 0.082 <0.0001 ***

149 1.19 63 9 4.2 Y 0.413 0.763 0.459 <0.0001 *** 3.1 39 3 2.0 N 0.538 0.509 -0.059 0.71 ns Mean 50.8 ± 6.0 10.2 ± 2.4 5.8 ± 1.3 0.629 ± 0.072 0.771 ± 0.070 0.171 ± 0.086 Ohio FEMSATL1 79 18 10.8 Y 0.671 0.907 0.261 <0.0001 *** FEMSATL19 99 10 4.7 Y 0.606 0.788 0.231 <0.0001 *** M2-30 107 14 4.1 Y 0.645 0.755 0.146 0.001 *** 1.19 91 8 4.8 Y 0.385 0.790 0.513 <0.0001 *** 3.1 69 3 2.1 Y 0.362 0.533 0.320 <0.0001 *** Mean 89.0 ± 6.8 10.6 ± 2.6 5.3 ± 1.5 0.534 ± 0.066 0.755 ± 0.061 0.294 ± 0.062

N = Number of samples, NA = Allelic richness, NE = Effective alleles, HO = Observed heterozygosity, HE = Expected heterozygosity, FIS = Wright’s inbreeding coefficient. P value indicates significant deviation from Hardy-Weinberg equilibrium for each locus.

149

Figure 5.4: Principal component analysis of ash seedlings and saplings from Michigan and Ohio. (A) Differentiation of Michigan and Ohio populations, size classes were pooled. Ellipses represent 95% confidence interval of each group.

150

151

A B

Figure 5.5: Principal component analysis of ash seedlings and saplings from Michigan and Ohio. (A) Comparison of size classes in Michigan vs. all of Ohio. (B) Comparison of size classes in Ohio vs. all of Michigan.

151

Discussion

Loss of heterozygosity and allelic variation are indications of a genetic bottleneck

(Frankham, 1996). Genetic bottlenecks often occur when a population has undergone rapid decline in size and are often characteristic of endangered species. For example, isolated populations of the little spotted kiwi (Apteryx owenii), an endangered bird to

New Zealand, have lower allelic richness, fewer rare alleles, and lower proportions of heterozygosity than the source population (Ramstad et al., 2013). Similarly, genetic variation of the endangered Asteraceae (Ligularia sibirica) decreased as the population size decreased (Ilves et al., 2013). Additionally, low genetic variation and evidence of inbreeding was correlated with reduced seed production and germination success.

Genetic variation can be maintained in small populations if there is continuous gene flow between populations. This was observed when comparing isolated and continuous forest populations of the tree Vateria inidica, which had similar allelic variation, heterozygosity, and inbreeding (Ismail et al., 2014). Furthermore, they tested both pollen and seed dispersal distances and found that gene flow can occur between these populations. This suggests that genetic erosion (i.e., loss of genetic variation) can be prevented if gene flow is maintained between smaller populations.

The ash seedlings and saplings in this study were analyzed to compare genetic variation of different generations of ash regeneration. In Michigan, allelic variation and the number of effective alleles were similar for all seedling and saplings. The number of effective alleles was approximately half of the total number of alleles per loci. This indicates that many of the alleles were at low frequencies and with low probability of

152 contributing to reproduction in the next generation (Kimura and Crow, 1963). Class 2 and class 3 seedlings had fewer than expected heterozygotes and deviated from Hardy-

Weinberg Equilibrium (HWE) for three and two loci, respectively. Since low heterozygosity is an indication of inbreeding (Ellstrand and Elam, 1993), the class 2 and class 3 seedlings may have been the progeny of related individuals. However, there was also indication that some alleles were undetected (i.e., null alleles), giving false homozygote genotypes. Additionally, allelic variation was similar across size classes.

Typically loss of allelic variation occurs before the loss of heterozygotes, which is often attributed to mating between closely related individuals after a sudden decrease in population size (Nei et al., 1975; Ellstrand and Elam, 1993; Frankham, 1996).

In Ohio, allelic richness was higher for class 1 and 2 seedlings than class 3 seedlings. However, the number of effective alleles was similar for all size classes. This indicates that many of the low frequency alleles have been lost in the class 3 seedlings, perhaps due to natural seedling mortality. Tree mortality is highest at younger ages, but decreases once a tree reaches a certain age (Hett and Loucks, 1968). Therefore, the class

3 seedlings have probably experienced random genetic drift and/or natural selection as individuals died, which has shaped the genetic structure of this population.

Genetic variation in ash has been studied in F. excelsior in Europe (Heuertz et al.,

2001; Heuertz et al., 2004; Fussi and Konnert, 2014; Beatty et al., 2015), F. mandshurica in Asia (Hu et al., 2010a-b), and green ash in northwestern Ohio (Hausman et al., 2014).

Overall, European and Asian ash populations have high genetic variation and low incidences of inbreeding (Heuertz et al., 2001; Heuertz et al., 2004; Hu et al., 2010a-b;

153

Fussi and Konnert, 2014; Beatty et al., 2015). Hausman et al. (2014) quantified genetic variation of green ash in 2005 in early stages of the EAB invasion and found mean allelic richness 16.4 ± 5.2 (10 – 24 alleles per locus), which was higher than reported in my study for Michigan (10.2 ± 2.4 alleles per locus) and Ohio (10.6 ± 2.6 alleles per locus).

They also reported more effective alleles (7.9 ± 4.3) (Hausman et al., 2014) than found here (5.8 ± 1.3 in Michigan, and 5.3 ± 1.5 in Ohio). A comparison of results of Hausman et al. (2014) from early stages of the EAB invasion and results reported here from mid- and late stages of the EAB invasion, provides evidence that genetic variation of green ash may have been negatively impacted by widespread ash mortality

Heterozygosity and incidence of inbreeding reported in this study also differed from results of Hausman et al. (2014), who found an excess of heterozygotes at each locus. However, I found differences among size classes in both Michigan and Ohio. In particular, the smallest size classes (class 2 in Michigan and class 1 and 2 in Ohio) had significantly fewer heterozygotes than expected at three loci. The largest size classes

(saplings in Michigan and class 3 seedlings in Ohio) had similar proportions of heterozygotes than expected for at least four loci. However, due to the likelihood of null alleles, it is unclear whether these populations have deviated from HWE or whether there is evidence of inbreeding. Overall, these findings indicate that inbreeding is not common for green ash in the Ohio-Michigan area, but the smaller size classes of green ash may have experienced higher levels of inbreeding in both Michigan and Ohio after EAB- induced ash mortality. More microsatellites will need to be tested to be able to make

154 stronger conclusions on the loss of heterozygosity and occurrence of inbreeding in these populations.

Genetic structure of green ash in Michigan and Ohio

The genetic structure of seedlings and saplings in Michigan differed by size class, whereas there were no differences among size classes in Ohio. Specifically, class 2 seedlings (i.e., smallest/youngest seedlings) were more similar and had lower genotypic variation than class 3 seedlings and saplings. In Ohio, the genetic structure of each size class of seedlings overlapped, but the class 3 seedlings (i.e., the largest/oldest seedlings) were most similar and had lower genotypic variation. The class 3 seedlings were the survivors of a larger pool of seedlings (Hett and Loucks, 1968). This may explain why there was greater genetic variation in class 1 seedlings. In a natural tree population that has not been impacted by widespread mortality or a genetic bottleneck, more seeds are produced than germinate and more seedlings germinate than survive to maturity (Hett and

Loucks, 1968). Therefore, some individuals will survive due to selection and others survive due to random chance (Ellstrand and Elam, 1993; Frankham, 1995). However, in

Michigan, the youngest individuals had the lowest genotypic variation, which was the opposite pattern as in Ohio. This suggests that when the class 2 seedling cohort reaches class 3 or sapling sizes, natural mortality will continue to reduce genetic variation (Hett and Loucks, 1968; Ellstrand and Elam, 1993). This also suggests that the class 3 seedlings and saplings were progeny of a larger pool of mature ash trees than the class 2

155 seedlings (Ellstrand and Elam, 1993), perhaps because of extensive mortality caused by

EAB.

Density of treated ash trees or EAB impact had no effect on genetic structure of seedlings in Ohio. These suggest that EAB has not yet negatively impacted the genetic structure of green ash in Ohio, as it appears to have in Michigan. Furthermore, this suggests that the impact of EAB has not been great enough to detect a positive impact of treated ash trees on conserving genetic variation. However, I predict that as ash trees continue to die, the population genetics of class 1 seedlings will reflect a loss of genetic variation similar to the class 2 seedlings in Michigan.

The different genetic structures of the size classes in Michigan and Ohio reflect differential patterns of widespread ash mortality. Specifically the Michigan population has experienced greater mortality than the Ohio population. Distinct patterns of genetic variation characterized the Michigan and Ohio populations, with some overlap of shared genotypes, possibly reflecting geographic isolation and post-glacial establishment of green ash in these locations (Heuertz et al., 2004; Matos et al., 2013; Su et al., 2017).

However within each location, the genetic structure should be similar for each size class since they are from the same parent population (Malysheva-Otto et al., 2006).

In Ohio, the genetic structures of seedlings were similar and each seedling size class contained genotypes unique to the Ohio population. This indicates that Ohio has not experienced a genetic bottleneck from EAB-induced ash mortality. The lower genetic variation in class 3 seedlings suggests that genetic variation declined over time due to high seedling mortality (Hett and Loucks, 1968). However, unlike the Michigan

156 population, all size classes of seedlings shared similar genotypes. Therefore, no negative impact of EAB was detected on genetic variation of green ash at FRMP in Ohio.

In Michigan, the genotypic variation of class 3 seedlings and saplings were similar, but class 2 seedlings had less genotypic variation and did not share the genotypes that were unique to Michigan. Class 3 seedlings and saplings are the survivors of a once larger cohort (Hett and Loucks, 1968), and perhaps because they possess traits that better adapt them to their environment than the individuals that did not survive. Additionally, the lower genetic variation of class 2 seedlings may make them less resilient to environmental change (Ellstrand and Elam, 1993; Ouborg et al., 2006). Plant populations with low genetic variation and high rates of inbreeding have lower fecundity and reduced likelihood of germination and seedling establishment (Husband and Schemske, 1996;

Keller and Waller, 2002). In particular, lower genetic variation has attributed to smaller seeds (size and mass), and malformation of structures important for dispersal (Donohue,

1999). Hence, decreased genetic variation may decrease the chances of conserving ash in forests impacted by EAB.

It should be noted that these conclusions were based on analyses of only five microsatellite loci. Analyses using more may reflect a more complete understanding of the population genetics of these populations. Additionally, microsatellites are neutral markers. To understand the evolutionary significance of EAB-induced ash mortality will require investigation of variation in genes related to survival and/or fecundity.

In summary, a genetic bottleneck may have occurred in ash seedling populations in southeastern Michigan, where mortality of reproductive ash trees exceeded 99%. This

157 was evidenced by higher inbreeding, lower allelic variation, and reduced genotypic variation in comparison to the larger size classes. In Ohio, there was no evidence of a genetic bottleneck as the genetic structure of seedlings was similar to the parent population. However, the class 1 and 2 size classes had higher allelic and genotypic variation than the class 3 seedlings. I predict that a genetic bottleneck may emerge over time as EAB-induced ash mortality increases. Additionally, future analyses from FRMP may document whether higher densities of treated ash trees can conserve genetic variation in the seedling populations.

Literature cited

Aguilar, R., Quesada, M., Ashworth, L., Herrerias-Diego, Y., Lobo, J. 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Mol. Ecol. 17: 5177-5188. Al-Saghir, M.G. 2009. Rapid and efficient method of genomic DNA extraction from pistachio trees (Pistacia vera L.). Res. J. Bot. 4: 70-73. Beatty, G.E., Brown, J.A., Cassidy, E.M., Finlay, C.M.V., McKendrick, L., Montgomery, W.I., Reid, N., Tosh, D.G., Provan, J. 2015. Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: implications for restorative planting. Tree Genet. Genomes 11: 53. Brachet, S., Jubier, M.F., Richard, M., Jung-Muller, B., Frascaria-Lacoste, N. 1999. Rapid identification of microsatellite loci using 5 ' anchored PCR in the common ash Fraxinus excelsior. Mol. Ecol. 8: 160-163. Bray, J.R. 1956. Gap phase replacement in a maple-basswood forest. Ecology 37: 598- 600. Brokaw, N.V.L. 1985. Gap-phase regeneration in a tropical forest. Ecology 66: 682-687. Burns, R.M., Honkala, B.H. 1990. Silvics of North America: 2, hardwoods. Agriculture handbook 654, U.S. Department of Agriculture, Washington, D.C. Davies, S.J., Cavers, S., Finegan, B., Navarro, C., Lowe, A.J. 2010. Genetic consequences of multigenerational and landscape colonization bottlenecks for a neotropical forest pioneer tree, Vochysia ferruginea. Trop. Plant Biol. 3: 14-27. Donohue, K. 1999. Seed dispersal as a maternally influenced character: Mechanistic basis of maternal effects and selection on maternal characters in an annual plant. Am. Nat. 154: 674-689. 158

Ellstrand, N.C., Elam, D.R. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217-242. Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327. Frankham, R. 1996. Relationship of genetic variation to population size in wildlife. Conserv. Biol. 10: 1500-1508. Fussi, B., Konnert, M. 2014. Genetic analysis of European common ash (Fraxinus excelsior L.) populations affected by ash dieback. Silvae Genet. 63: 198-212. Gandhi, K.J.K., Herms, D.A. 2010. North American arthropods at risk due to widespread Fraxinus mortality caused by the Alien emerald ash borer. Biol. Invasions 12: 1839-1846. Good, N.F., Good, R.E. 1972. Population dynamics of tree seedlings and saplings in a mature eastern hardwood forest. Bull. Torrey Bot. Club 99: 172-178. Hausman, C.E., Bertke, M.M., Jaeger, J.F., Rocha, O.J. 2014. Genetic structure of green ash (Fraxinus pennsylvanica): implications for the establishment of ex situ conservation protocols in light of the invasion of the emerald ash borer. Plant Genet. Resour. 12: 286-297. Hett, J.M., Loucks, O.L. 1968. Application of life-table analyses to tree seedlings in Quetico Provincial Park, Ontario. For. Chron. 44: 29-32. Heuertz, M., Hausman, J.F., Hardy, O.J., Vendramin, G.G., Frascaria-Lacoste, N., Vekemans, X. 2004. Nuclear microsatellites reveal contrasting patterns of genetic structure between western and southeastern European populations of the common ash (Fraxinus excelsior L.). Evolution 58: 976-988. Heuertz, M., Hausman, J.F., Tsvetkov, I., Frascaria-Lacoste, N., Vekemans, X. 2001. Assessment of genetic structure within and among Bulgarian populations of the common ash (Fraxinus excelsior L.). Mol. Ecol. 10: 1615-1623. Heuertz, M., Vekemans, X., Hausman, J.F., Palada, M., Hardy, O.J. 2003. Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash. Mol. Ecol. 12: 2483-2495. Hu, L.-J., Uchiyama, K., Saito, Y., Ide, Y. 2010a. Contrasting patterns of nuclear microsatellite genetic structure of Fraxinus mandshurica var. japonica between northern and southern populations in Japan. J. Biogeogr. 37: 1131-1143. Hu, L.-J., Uchiyama, K., Shen, H.-L., Ide, Y. 2010b. Multiple-scaled spatial genetic structures of Fraxinus mandshurica over a riparian-mountain landscape in Northeast China. Conserv. Genet. 11: 77-87. Husband, B.C., Schemske, D.W. 1996. Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54-70. Ilves, A., Lanno, K., Sammul, M., Tali, K. 2013. Genetic variability, population size and reproduction potential in Ligularia sibirica (L.) populations in Estonia. Conserv. Genet. 14: 661-669. Ismail, S.A., Ghazoul, J., Ravikanth, G., Kushalappa, C.G., Shaanker, R.U., Kettle, C.J. 2014. Fragmentation genetics of Vateria indica: Implications for management of forest genetic resources of an endemic dipterocarp. Conserv. Genet. 15: 533-545. Keller, L.F., Waller, D.M. 2002. Inbreeding effects in wild populations. Trends Ecol. Evolut. 17: 230-241. 159

Kimura, M., Crow, J.F. 1963. Measurement of effective population numbers. Evolution 17: 279-288. Klooster, W.S., Herms, D.A., Knight, K.S., Herms, C.P., McCullough, D.G., Smith, A., Gandhi, K.J.K., Cardina, J. 2014. Ash (Fraxinus spp.) mortality, regeneration, and seed bank dynamics in mixed hardwood forests following invasion by emerald ash borer (Agrilus planipennis). Biol. Invasions 16: 859-873. Lefort, F., Brachet, S., Frascaria-Lacoste, N., Edwards, K.J., Douglas, G.C. 1999. Identification and characterization of microsatellite loci in ash (Fraxinus excelsior L.) and their conservation in the olive family (Oleaceae). Mol. Ecol. 8: 1088- 1090. Lovett, G.M., Canham, C.D., Arthur, M.A., Weathers, K.C., Fitzhugh, R.D. 2006. Forest ecosystem responses to exotic pests and pathogens in eastern North America. Bioscience 56: 395-405. Malysheva-Otto, L.V., Ganal, M.W., Roder, M.S. 2006. Analysis of molecular diversity, population structure and linkage disequilibrium in a worldwide survey of cultivated barley germplasm (Hordeum vulgare L.). BMC Genet. 7:6 10.1186/1471-2156-7-6. Matos, E.L.S., Oliveira, E.J., Jesus, O.N., Dantas, J.L.L. 2013. Microsatellite markers of genetic diversity and population structure of Carica papaya. An. Appl. Biol. 163: 298-310. McClure, M.S., Salom, S.M., Shields, K.S. 2003. Hemlock woolly adelgid. Forest Health Technology Enterprise Team, U.S. Dept. of Agriculture, Forest Service, Morgantown, p 14. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. McCullough, D.G., Siegert, N.W. 2007. Estimating potential emerald ash borer (Coleoptera : Buprestidae) populations using ash inventory data. J. Econ. Entomol. 100: 1577-1586. Nei, M., Maruyama, T., Chakraborty, R. 1975. Bottleneck effect and genetic variability in populations. Evolution 29: 1-10. Ouborg, N.J., Vergeer, P., Mix, C. 2006. The rough edges of the conservation genetics paradigm for plants. J. Ecol. 94: 1233-1248. Pautasso, M. 2009. Geographical genetics and the conservation of forest trees. Perspect. Plant Ecol. Evol. Syst. 11: 157-189. Peakall, R., Smouse, P.E. 2006. GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6: 288-295. Potter, K.M., Jetton, R.M., Dvorak, W.S., Hipkins, V.D., Rhea, R., Whittier, W.A. 2012. Widespread inbreeding and unexpected geographic patterns of genetic variation in eastern hemlock (Tsuga canadensis), an imperiled North American conifer. Conserv. Genet. 13: 475-498. Ramstad, K.M., Colbourne, R.M., Robertson, H.A., Allendorf, F.W., Daugherty, C.H. 2013. Genetic consequences of a century of protection: Serial founder events and 160

survival of the little spotted kiwi (Apteryx owenii). Proc. R. Soc. B 280: 20130576. Roberts, J.H., Angermeier, P.L., Hallerman, E.M. 2013. Distance, dams and drift: what structures populations of an endangered, benthic stream fish? Freshwater Biol. 58: 2050-2064. Runkle, J.R. 1984. Development of woody vegetation in treefall gaps in a beech-sugar maple forest. Holarct. Ecol. 7: 157-164. Runkle, J.R. 1990. Gap dynamics in an Ohio Acer-Fagus forest and speculations on the geography of disturbance. Can. J. For. Res. 20: 632-641. Runkle, J.R., Yetter, T.C. 1987. Treefalls revisited: Gap dynamics in the southern Appalachians. Ecology 68: 417-424. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Su, Z.H., Richardson, B.A., Zhuo, L., Jiang, X.L. 2017. Divergent population genetic structure of the endangered Helianthemum (Cistaceae) and its implication to conservation in northwestern China. Front. Plant Sci. 7: 2010. doi: 10.3389/fpls.2016.02010. van Oosterhout, C., Hutchison, W.F., Wills, D.P.M., Shipley, P.F. 2003. Micro-Checker User Guide. The University of Hull. Vranckx, G., Jacquemyn, H., Muys, B., Honnay, O. 2012. Meta-analysis of susceptibility of woody plants to loss of genetic diversity through habitat fragmentation. Conserv. Biol. 26: 228-237. Wallander, E. 2008. Systematics of Fraxinus (Oleaceae) and evolution of dioecy. Plant Syst. Evol. 273: 25-49. Wee, A.K.S., Chunhong, L., Dvorak, W.S. 2012. Genetic diversity in natural populations of Gmelina arborea: implications for breeding and conservation. New Forests 43: 411-428. Woolaver, L.G., Nichols, R.K., Morton, E.S., Stutchbury, B.J.M. 2013. Population genetics and relatedness in a critically endangered island raptor, Ridgway's Hawk Buteo ridgwayi. Conserv. Genet. 14: 559-571.

161

Chapter 6: Summary, conclusion and future research

The main goal of this study was to investigate the effects that a large-scale insecticide program with emamectin benzoate trunk injections has on conserving ash populations and genetic variation in forests invaded by EAB. To meet this goal, I addressed the following questions: (1) Do insecticide treatments protect ash trees in invaded forests? (2) Do treated ash trees provide associational protection to untreated ashes? (3) Can insecticide treatments maintain ash reproduction and seedling recruitment? (4) Does widespread emerald ash borer-induced ash mortality cause a genetic bottleneck and can genetic variation be preserved by protecting ash trees with insecticides?

Question 1: Do insecticide treatments protect ash trees in invaded forests?

I found that treated green, white, black, and pumpkin ashes were healthier than untreated conspecifics. Additionally, most of the treated green, white, black and pumpkin ashes were alive at the end of the study, whereas most untreated trees were dead. The results of this study demonstrate that biannual applications of emamectin benzoate effectively protect ash in natural forests over an extended period.

162

Blue ash may not require insecticidal protection, as treated and untreated trees remained healthy and survival was high even in plots with low survival of green and white ash. Other studies have also found blue ash to have higher survival than green and white ash (Anulewicz et al. 2007,Tanis and McCullough, 2012).

Question 2: Do treated ash trees provide associational protection to untreated ashes?

Percentage survival of green and white ash decreased from 2014 – 2016.

However, parks with low EAB impact had higher survival when percentage of ash phloem area treated with emamectin benzoate was highest and when treated ash trees were within 100 m of untreated ash trees. This suggests that the probability of survival of untreated ash trees increases when located in close proximity to a cluster of treated ash trees. However, in parks with high EAB impact, there was no relationship between survival of untreated ash trees and percentage of ash phloem area treated or proximity to treated ash trees. This suggests that in areas with lower EAB impact, larger clusters of treated ash trees may delay the progression of EAB infestation, which is consistent with previous studies that have shown that clusters of ash trees treated with emamectin benzoate can slow the progression of ash mortality (SLAM project) (McCullough and

Mercader, 2012; Mercader et al., 2015). Since I did not observe a relationship with ash survival in areas with higher EAB impact, either more ash trees may need to be treated

(i.e., higher percentage ash phloem) or the trees in these areas were treated too late to provide associational protection towards neighboring untreated ash trees.

163

Question 3: Can insecticide treatments maintain ash reproduction and seedling recruitment?

Density of treated ash trees flowering increased with percentage ash phloem area treated and did not differ between low and high EAB impact. The number of untreated flowering trees was similar for blue and green-white ash in low EAB impact plots.

However, there were more flowering blue ash than green-white ash in the high EAB impact plots. Total flowering ash trees (treated and untreated) increased with percentage ash phloem area treated and were overall higher in parks with low EAB impact.

New ash regeneration decreased in the parks with high impact and was not affected by percentage ash phloem area treated. However, seedling density was higher in parks with low EAB impact and increased with percentage ash phloem treated.

Insecticide treatment increased the density of green-white ash seedlings in the low impact plots, but remained low in the high EAB impact plots. Density of blue ash seedlings was low in all plots and was not affected by ash mortality. Overall, the density of seedlings

(blue and green-white ash) was highest where the density of seed-bearing ash was highest. For green-white ash, density of female ash increased with percent treated ash phloem, suggesting that the insecticide treatments can maintain reproduction and regeneration of these species. Density of seed bearing blue ash was not affected by insecticide treatment because of the low impact of EAB on this species. Therefore, insecticides were not required to protect blue ash, or maintain reproduction and regeneration. As ash trees die from EAB infestation, reproduction declines (Kashian and

164

Witter, 2011), I have shown that the use of insecticide management can maintain seed production and ash regeneration.

Question 4: Does widespread emerald ash borer-induced ash mortality cause a genetic bottleneck and can genetic variation be preserved by protecting ash trees with insecticides?

In southeastern Michigan, where mortality of ash exceeded 99%, allelic richness was similar in ash seedlings and saplings. However, heterozygosity was lower than expected in the class 2 and class 3 seedlings, indicating a loss of genetic variation and evidence of inbreeding (Ellstrand and Elam, 1993; Husband and Schemske, 1996).

Additionally, different size classes of seedlings (as indices of age) were genetically differentiated, which indicates that these cohorts may have been progeny from a different pool of mature ash (Malysheva-Otto et al., 2006). The three size classes of seedlings shared similar combinations of genotypes. However, the larger seedlings and saplings contained more variations of genotypes than did smaller (and probably younger) seedlings. This indicates that the smaller seedlings were progeny of mature ash that shared a smaller gene pool. Furthermore, genetic variation of this seedling population was lower than a population studied in northwest Ohio prior to experiencing high mortality from EAB (Hausman et al., 2014). The Michigan ash seedlings and saplings also had less genetic variation than ash populations studied in Europe and Asia (Heuertz et al., 2004; Hu et al., 2010; Beatty et al., 2015).

165

In ash populations at Five Rivers Metropark (FRMP) in Ohio, allelic richness was higher in the smallest (younger) seedling size classes than in larger (older) seedlings.

However, the number of effective alleles was similar among all size classes. This indicates that the largest seedlings had fewer low frequency alleles than the smaller seedlings. Perhaps low frequency alleles were eliminated from the population over time due to high seedling mortality (Hett and Loucks, 1968). Heterozygosity was lower than expected for each size class, indicating that there was lower than expected variation, and perhaps some degree of inbreeding (Ellstrand and Elam, 1993; Husband and Schemske,

1996). There was no evidence of genetic differentiation between size classes, which shows that all classes of ash seedlings at FRMP were progeny of parents that shared the same gene pool. However, the population of smaller seedlings had more genetic variation.

There was no difference in the genetic variation of seedlings from parks with low and high EAB impact. There was more genotypic variation (P=0.1) in seedling populations in the vicinity of higher densities of insecticide treated ash. The effect was only marginally significant, but suggests that protecting reproductive ash trees with insecticide treatments may contribute to conservation of genetic variation.

There was evidence of genetic differentiation between seedling populations from

Michigan and Ohio. In Michigan, smaller seedlings consisted of a restricted pool of genotypes that did not include the genotypes unique to larger seedlings and saplings. In

Ohio, all three size classes of seedlings shared the genotypes that were unique to Ohio.

This provides evidence of a loss of genetic variation in smaller seedlings in Michigan,

166 where EAB mortality was very high. However, seedling populations in Ohio showed no evidence of a loss of genetic variation.

This study was limited by the number of microsatellite loci used to quantify genetic variation, and additional microsatellites are needed to confirm the loss of genetic variation in Michigan. In Ohio, ash mortality may not yet have been sufficient to decrease genetic variation. It is likely, however, that as more ash trees die, the effects of both

EAB-induced mortality and insecticide treated ash trees will become more apparent.

Conclusions and future direction

In summary, insecticide treatments effectively protected mature ash from EAB

(objective 1), there was evidence that treated ash trees provide associational protection to untreated green and white ashes (objective 2), ash reproduction and seedling recruitment was higher in areas with higher density of treated ash (objective 3), and widespread ash mortality caused a genetic bottleneck in the smallest (youngest) seedlings in Michigan

(objective 4). There was no evidence of a genetic bottleneck in Ohio, therefore, whether insecticide can conserve genetic variation remains an open question.

Based on these results, I have constructed a conceptual model of the interacting factors that contribute to ash population demography (Figure 6.1). I found that densities of seedlings and saplings vary based on habitat characteristics. The impact of EAB on the ash community was a main driver in determining whether insecticide treated ash provide associational protection to untreated trees and increase density of seedlings, which

167 occurred when initial percentage ash mortality was low, but not high. Lastly, density of treated ash was an important predictor of survival of untreated ash and seedling densities

This model can be used to construct a structured equation model of these complex relationships that increases understanding of factors that are most important for maintaining a stable population structure of ash, and predict the number and spatial arrangement of ash trees that must be treated with insecticide to conserve ash populations and genetic variation in forests invaded by EAB.

A greater understanding of the relationship between the spatial arrangement and density of treated ash trees and seedling regeneration is required. I hypothesize that seedling densities would be higher in areas with more treated ashes, but will decrease with distance away from treated female trees. To test this hypothesis, one could quantify seedling densities along transects radiating away from individual treated female trees along a gradient of percentage of ash phloem area treated.

There is potential to incorporate ash conservation strategies based on insecticides with other management approaches for EAB, classical biological control. Three species of parasitoid wasps that are EAB specialists have been released and established in North

America (Bauer et al., 2008; Bauer et al., 2015). There is limited evidence that populations that have been established for at least five years are affecting EAB density and survival (Duan et al., 2013). Therefore, combining insecticide management with classical biological control allows for the protection of ash trees while populations of biological control agents are establishing in an area.

168

Figure 6.1: Conceptual model of factors affecting population demography of green and white ash in response to insecticide treatments in forests invaded by emerald ash borer. Solid lines indicate significant relationships detected in this study, dashed lines indicate predicted variables that were not quantified during this study.

169

Literature cited

Anulewicz, A.C., McCullough, D.G., Cappaert, D.L. 2007. Emerald ash borer (Agrilus planipennis) density and canopy dieback in three North American ash species. Arboric Urban For. 33: 338-349. Bauer, L.S., Duan, J.J., Gould, J.R., Van Driesche, R. 2015. Progress in the classical biological control of Agrilus planipennis Fairmaire (Coleoptera: Buprestidae) in North America. Can. Entomol. 147: 300-317. Bauer, L.S., Liu, H., Miller, D., Gould, J. 2008. Developing a classical biological control program for Agrilus planipennis (Coleoptera: Buprestidae), an invasive ash pest in North America. In, Newsletter of the Michigan Entomological Society. Michigan Entomological Society, East Lansing, MI, pp. 38-39. Beatty, G.E., Brown, J.A., Cassidy, E.M., Finlay, C.M.V., McKendrick, L., Montgomery, W.I., Reid, N., Tosh, D.G., Provan, J. 2015. Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: implications for restorative planting. Tree Genet. Genomes 11: 53. Duan, J.J., Bauer, L.S., Abell, K.J., Lelito, J.P., Van Driesche, R. 2013. Establishment and abundance of Tetrastichus planipennisi (Hymenoptera: Eulophidae) in Michigan: Potential for success in classical biocontrol of the invasive emerald ash borer (Coleoptera: Buprestidae). J. Econ. Entomol. 106: 1145-1154. Ellstrand, N.C., Elam, D.R. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217-242. Hausman, C.E., Bertke, M.M., Jaeger, J.F., Rocha, O.J. 2014. Genetic structure of green ash (Fraxinus pennsylvanica): implications for the establishment of ex situ conservation protocols in light of the invasion of the emerald ash borer. Plant Genet. Resour. 12: 286-297. Hett, J.M., Loucks, O.L. 1968. Application of life-table analyses to tree seedlings in Quetico Provincial Park, Ontario. For. Chron. 44: 29-32. Heuertz, M., Hausman, J.F., Hardy, O.J., Vendramin, G.G., Frascaria-Lacoste, N., Vekemans, X. 2004. Nuclear microsatellites reveal contrasting patterns of genetic structure between western and southeastern European populations of the common ash (Fraxinus excelsior L.). Evolution 58: 976-988. Hu, L.-J., Uchiyama, K., Saito, Y., Ide, Y. 2010. Contrasting patterns of nuclear microsatellite genetic structure of Fraxinus mandshurica var. japonica between northern and southern populations in Japan. J. Biogeogr. 37: 1131-1143. Husband, B.C., Schemske, D.W. 1996. Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54-70. Malysheva-Otto, L.V., Ganal, M.W., Roder, M.S. 2006. Analysis of molecular diversity, population structure and linkage disequilibrium in a worldwide survey of cultivated barley germplasm (Hordeum vulgare L.). BMC Genet. 7:6 10.1186/1471-2156-7-6.

170

McCullough, D.G., Mercader, R.J. 2012. Evaluation of potential strategies to SLow Ash Mortality (SLAM) caused by emerald ash borer (Agrilus planipennis): SLAM in an urban forest. Int. J. Pest Manage. 58: 9-23. Mercader, R.J., McCullough, D.G., Storer, A.J., Bedford, J.M., Heyd, R., Poland, T.M., Katovich, S. 2015. Evaluation of the potential use of a systemic insecticide and girdled trees in area wide management of the emerald ash borer. For. Ecol. Manag. 350: 70-80. Tanis, S.R., McCullough, D.G. 2012. Differential persistence of blue ash and white ash following emerald ash borer invasion. Can. J. For. Res. 42: 1542-1550.

171

Literature Cited

Aguilar, R., Quesada, M., Ashworth, L., Herrerias-Diego, Y., Lobo, J. 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Mol. Ecol. 17: 5177-5188. Al-Saghir, M.G. 2009. Rapid and efficient method of genomic DNA extraction from pistachio trees (Pistacia vera L.). Res. J. Bot. 4: 70-73. Amos, W., Wilmer, J.W., Fullard, K., Burg, T.M., Croxall, J.P., Bloch, D., Coulson, T. 2001. The influence of parental relatedness on reproductive success. Proc. R. Soc.B 268: 2021-2027. Anulewicz, A.C., McCullough, D.G., Miller, D.L. 2006. Oviposition and development of emerald ash borer (Agrilus planipennis) (Coleoptera: Buprestidae) on hosts and potential hosts in no-choice bioassays. Gt Lakes Entomol. 39: 99-112. Anulewicz, A.C., McCullough, D.G., Cappaert, D.L. 2007. Emerald ash borer (Agrilus planipennis) density and canopy dieback in three North American ash species. Arboric Urban For. 33: 338-349. Anulewicz, A.C., McCullough, D.G., Cappaert, D.L., Poland, T.M. 2008. Host range of the Emerald ash borer (Agrilus planipennis Fairmaire) (Coleoptera: Buprestidae) in North America: Results of multiple-choice field experiments. Environ. Entomol. 37: 230-241. Ayres, E., Dromph, K.M., Cook, R., Ostle, N., Bardgett, R.D. 2007. The influence of below- ground herbivory and defoliation of a legume on nitrogen transfer to neighboring plants. Funct. Ecol. 21: 256-263. Bauer, L.S., Duan, J.J., Gould, J.R., Van Driesche, R. 2015. Progress in the classical biological control of Agrilus planipennis Fairmaire (Coleoptera: Buprestidae) in North America. Can. Entomol. 147: 300-317. Bauer, L.S., Liu, H., Miller, D., Gould, J. 2008. Developing a classical biological control program for Agrilus planipennis (Coleoptera: Buprestidae), an invasive ash pest in North America. In, Newsletter of the Michigan Entomological Society. Michigan Entomological Society, East Lansing, MI, pp. 38-39. Beatty, G.E., Brown, J.A., Cassidy, E.M., Finlay, C.M.V., McKendrick, L., Montgomery, W.I., Reid, N., Tosh, D.G., Provan, J. 2015. Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: implications for restorative planting. Tree Genet. Genomes 11: 53. Beers, T.W., Dress, P.E., Wensel, L.C. 1966. Aspect transformation in site productivity research. J. For. 64: 691.

172

Bensch, S., Hasselquist, D., Vonschantz, T. 1994. Genetic similarity between parents predicts hatching failure: Nonincestuous inbreeding in the great reed warbler? Evolution 48: 317-326. Benton, E.P., Grant, J.F., Webster, R.J., Nichols, R.J., Cowles, R.S., Lagalante, A.F., Coots, C.I. 2015. Assessment of imidacloprid and its metabolites in foliage of eastern hemlock multiple years following treatment for hemlock woolly adelgid, Adelges tsugae (Hemiptera: Adelgidae), in forested conditions. J. Econ. Entomol. 108: 2672- 2682. Brachet, S., Jubier, M.F., Richard, M., Jung-Muller, B., Frascaria-Lacoste, N. 1999. Rapid identification of microsatellite loci using 5 ' anchored PCR in the common ash Fraxinus excelsior. Mol. Ecol. 8: 160-163. Braun, E. L. 1950. Deciduous forests of eastern North America. Hafner, New York, New York, U.S.A. Bray, J.R. 1956. Gap phase replacement in a maple-basswood forest. Ecology 37: 598-600. Brokaw, N.V.L. 1985. Gap-phase regeneration in a tropical forest. Ecology 66: 682-687. Brunner, E, Domhof, S., Langer, F. 2002. Nonparametric analysis of longitudinal data in factorial experiments. John Wiley & Son, New York, NY. Burns, R.M., Honkala, B.H. 1990. Silvics of North America: 2, hardwoods. Agriculture handbook 654, U.S. Department of Agriculture, Washington, D.C. Canham, C.D. 1988. An index for understory light levels in and around canopy gaps. Ecology 69: 1634-1638. Cappaert, D., McCullough, D.G., Poland, T.M., Siegert, N.W. 2005. Emerald ash borer in North America: A research and regulatory challenge. Am. Entomol. 51: 152-165. Carson, S.L.E. 2013. Spatiotemporal dynamics and host selection of emerald ash borer, Agrilus planipennis (Coleoptera: Buprestidae), at Point Pelee National Park, Ontario, Canada. The University of Guelph, Dissertation. p. 190. Castagneyrol, B., Giffard, B., Pere, C., Jactel, H. 2013. Plant apparency, an overlooked driver of associational resistance to insect herbivory. J. Ecol. 101: 418-429. Champagne, E., Tremblay, J.P., Cote, S.D. 2016. Spatial extent of neighboring plants influences the strength of associational effects on mammal herbivory. Ecosphere 7: e01371. Clark, J.S., Macklin, E., Wood, L. 1998. Stages and spatial scales of recruitment limitation in southern Appalachian forests. Ecol. Monogr. 68: 213-235. Clements, F.E. 1916. Plant Succession: An analysis of the development of vegetation. Carnegie Inst., Washington. Clinton, B.D., Baker, C.R. 2000. Catastrophic windthrow in the southern Appalachians: Characteristics of pits and mounds and initial vegetation responses. For. Ecol. Manag. 126: 51-60. Connell, J.H. 1978. Diversity in tropical rain forests and coral reefs: High diversity of trees and corals is maintained only in a non-equilibrium state. Science 199: 1302-1310. Crook, D.J., Francese, J.A., Zylstra, K.E., Fraser, I., Sawyer, A.J., Bartels, D.W., Lance, D.R., Mastro, V.C. 2009. Laboratory and field response of the emerald ash borer (Coleoptera: Buprestidae), to selected regions of the electromagnetic spectrum. J. Econ. Entomol. 102: 2160-2169. 173

Crook, D.J., Mastro, V.C. 2010. Chemical ecology of the emerald ash borer Agrilus planipennis. J. Chem. Ecol. 36: 101-112. Davidson, W., Rieske, L.K. 2016. Establishment of classical biological control targeting emerald ash borer is facilitated by use of insecticides, with little effect on native arthropod communities. Biol. Control 101: 78-86. Davies, S.J., Cavers, S., Finegan, B., Navarro, C., Lowe, A.J. 2010. Genetic consequences of multigenerational and landscape colonization bottlenecks for a neotropical forest pioneer tree, Vochysia ferruginea. Trop. Plant Biol. 3: 14-27. Desteven, D. 1994. Tropical tree seedling dynamics: Recruitment patterns and their population consequences for 3 canopy species in Panama. J. Trop. Ecol. 10: 369- 383. Donohue, K. 1999. Seed dispersal as a maternally influenced character: Mechanistic basis of maternal effects and selection on maternal characters in an annual plant. Am. Nat. 154: 674-689. Duan, J.J., Bauer, L.S., Abell, K.J., Lelito, J.P., Van Driesche, R. 2013. Establishment and abundance of Tetrastichus planipennisi (Hymenoptera: Eulophidae) in Michigan: Potential for success in classical biocontrol of the invasive emerald ash borer (Coleoptera: Buprestidae). J. Econ. Entomol. 106: 1145-1154. Dulaurent, A.M., Porte, A.J., van Halder, I., Vetillard, F., Menassieu, P., Jactel, H. 2012. Hide and seek in forests: Colonization by the pine processionary moth is impeded by the presence of nonhost trees. Agric. For. Entomol. 14: 19-27. Ellstrand, N.C., Elam, D.R. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217-242. Flower, C.E., Dalton, J.E., Knight, K.S., Brikha, M., Gonzalez-Meler, M.A. 2015. To treat or not to treat: Diminishing effectiveness of emamectin benzoate tree injections in ash trees heavily infested by emerald ash borer. Urban For. Urban Greening 14: 790-795. Flower, C.E., Knight, K.S., Gonzalez-Meler, M.A. 2013a. Impacts of the emerald ash borer (Agrilus planipennis Fairmaire) induced ash (Fraxinus spp.) mortality on forest carbon cycling and successional dynamics in the eastern United States. Biol. Invasions 15: 931-944. Flower, C.E., Knight, K.S., Rebbeck, J., Gonzalez-Meler, M.A. 2013b. The relationship between the emerald ash borer (Agrilus planipennis) and ash (Fraxinus spp.) tree decline: Using visual canopy condition assessments and leaf isotope measurements to assess pest damage. For. Ecol. Manag. 303: 143-147. Flower, C.E., Long, L.C., Knight, K.S., Rebbeck, J., Brown, J.S., Gonzalez-Meler, M.A., Whelan, C.J. 2014. Native bark-foraging birds preferentially forage in infected ash (Fraxinus spp.) and prove effective predators of the invasive emerald ash borer (Agrilus planipennis Fairmaire). For. Ecol. Manag. 313: 300-306. Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327. Frankham, R. 1996. Relationship of genetic variation to population size in wildlife. Conserv. Biol. 10: 1500-1508. Fussi, B., Konnert, M. 2014. Genetic analysis of European common ash (Fraxinus excelsior L.) populations affected by ash dieback. Silvae Genet. 63: 198-212. 174

Gandhi, K.J.K., Herms, D.A. 2010a. Direct and indirect effects of alien insect herbivores on ecological processes and interactions in forests of eastern North America. Biol. Invasions 12: 389-405. Gandhi, K.J.K., Herms, D.A. 2010b. North American arthropods at risk due to widespread Fraxinus mortality caused by the Alien emerald ash borer. Biol. Invasions 12: 1839- 1846. Gandhi, K.J.K., Smith, A., Hartzler, D.M., Herms, D.A. 2014. Indirect effects of emerald ash borer-induced ash mortality and canopy gap formation on epigaeic beetles. Environ. Entomol. 43: 546-555. Good, N.F., Good, R.E. 1972. Population dynamics of tree seedlings and saplings in a mature eastern hardwood forest. Bull. Torrey Bot. Club 99: 172-178. Graff, P., Aguiar, M.R., Chaneton, E.J. 2007. Shifts in positive and negative plant interactions along a grazing intensity gradient. Ecology 88: 188-199. Gray, A.N., Spies, T.A., Easter, M.J. 2002. Microclimatic and soil moisture responses to gap formation in coastal Douglas-fir forests. Can. J. For. Res. 32: 332-343. Grubb, P.J. 1977. Maintenance of species richness in plant communities: Importance of regeneration niche. Biol. Rev. Camb. Philos. Soc. 52: 107-145. Hamback, P.A., Agren, J., Ericson, L. 2000. Associational resistance: Insect damage to purple loosestrife reduced in thickets of sweet gale. Ecology 81: 1784-1794. Hamrick, J.L., 2004. Response of forest trees to global environmental changes. For. Ecol. Manag. 197: 323-335. Harper, J.L. 1977. Population biology of plants. Academic Press, London. Hausman, C.E., Bertke, M.M., Jaeger, J.F., Rocha, O.J. 2014. Genetic structure of green ash (Fraxinus pennsylvanica): implications for the establishment of ex situ conservation protocols in light of the invasion of the emerald ash borer. Plant Genet. Resour. 12: 286-297. Hausman, C.E., Jaeger, J.F., Rocha, O.J. 2010. Impacts of the emerald ash borer (EAB) eradication and tree mortality: Potential for a secondary spread of invasive plant species. Biol. Invasions 12: 2013-2023. Herms, D.A. 2015. Host range and host resistance. In: Van Driesche, R.G., Reardon, R., eds. Biology and control of emerald ash borer, Technical Bulletin FHTET 2014-09. Morgantown, WV, USA: USDA Forest Service, 65-73. Herms, D.A., McCullough, D.G. 2014. Emerald ash borer invasion of North America: History, biology, ecology, impacts, and management. Annu. Rev. Entomol. 59: 13- 30. Hett, J.M., Loucks, O.L. 1968. Application of life-table analyses to tree seedlings in Quetico Provincial Park, Ontario. For. Chron. 44: 29-32. Heuertz, M., Hausman, J.F., Hardy, O.J., Vendramin, G.G., Frascaria-Lacoste, N., Vekemans, X. 2004. Nuclear microsatellites reveal contrasting patterns of genetic structure between western and southeastern European populations of the common ash (Fraxinus excelsior L.). Evolution 58: 976-988. Heuertz, M., Hausman, J.F., Tsvetkov, I., Frascaria-Lacoste, N., Vekemans, X. 2001. Assessment of genetic structure within and among Bulgarian populations of the common ash (Fraxinus excelsior L.). Mol. Ecol. 10: 1615-1623. 175

Heuertz, M., Vekemans, X., Hausman, J.F., Palada, M., Hardy, O.J. 2003. Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash. Mol. Ecol. 12: 2483-2495. Hirao, T., Murakami, M., Iwamoto, J., Takafumi, H., Oguma, H. 2008. Scale-dependent effects of windthrow disturbance on forest arthropod communities. Ecol. Res. 23: 189-196. Hu, L.-J., Uchiyama, K., Saito, Y., Ide, Y. 2010a. Contrasting patterns of nuclear microsatellite genetic structure of Fraxinus mandshurica var. japonica between northern and southern populations in Japan. J. Biogeogr. 37: 1131-1143. Hu, L.-J., Uchiyama, K., Shen, H.-L., Ide, Y. 2010b. Multiple-scaled spatial genetic structures of Fraxinus mandshurica over a riparian-mountain landscape in Northeast China. Conserv. Genet. 11: 77-87. Husband, B.C., Schemske, D.W. 1996. Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54-70. Ilves, A., Lanno, K., Sammul, M., Tali, K. 2013. Genetic variability, population size and reproduction potential in Ligularia sibirica (L.) populations in Estonia. Conserv. Genet. 14: 661-669. Ismail, S.A., Ghazoul, J., Ravikanth, G., Kushalappa, C.G., Shaanker, R.U., Kettle, C.J. 2014. Fragmentation genetics of Vateria indica: Implications for management of forest genetic resources of an endemic dipterocarp. Conserv. Genet. 15: 533-545. Jactel, H., Birgersson, G., Andersson, S., Schlyter, F. 2011. Non-host volatiles mediate associational resistance to the pine processionary moth. Oecologia 166: 703-711. Jump, A.S., Penuelas, J. 2006. Genetic effects of chronic habitat fragmentation in a wind- pollinated tree. Proc. Natl. Acad. Sci. U.S.A. 103: 8096-8100. Karban, R. 2007. Associational resistance for mule's ears with sagebrush neighbors. Plant Ecol. 191: 295-303. Karban, R., Maron, J. 2002. The fitness consequences of interspecific eavesdropping between plants. Ecology 83: 1209-1213. Karban, R., Shiojiri, K., Huntzinger, M., McCall, A.C. 2006. Damage-induced resistance in sagebrush: Volatiles are key to intra- and interplant communication. Ecology 87: 922-930. Kashian, D.M., Witter, J.A. 2011. Assessing the potential for ash canopy tree replacement via current regeneration following emerald ash borer-caused mortality on southeastern Michigan landscapes. For. Ecol. Manag. 261: 480-488. Keller, L.F., Waller, D.M. 2002. Inbreeding effects in wild populations. Trends Ecol. Evolut. 17: 230-241. Kimura, M., Crow, J.F. 1963. Measurement of effective population numbers. Evolution 17: 279-288. Klooster, W.S., Herms, D.A., Knight, K.S., Herms, C.P., McCullough, D.G., Smith, A., Gandhi, K.J.K., Cardina, J. 2014. Ash (Fraxinus spp.) mortality, regeneration, and seed bank dynamics in mixed hardwood forests following invasion by emerald ash borer (Agrilus planipennis). Biol. Invasions 16: 859-873. Knapp, E.E., Goedde, M.A., Rice, K.J. 2001. Pollen-limited reproduction in blue oak: implications for wind pollination in fragmented populations. Oecologia 128: 48-55. 176

Knight, K.S., Brown, J.P., Long, R.P. 2013. Factors affecting the survival of ash (Fraxinus spp.) trees infested by emerald ash borer (Agrilus planipennis). Biol. Invasions 15: 371-383. Kobe, R.K., Pacala, S.W., Silander, J.A., Canham, C.D. 1995. Juvenile tree survivorship as a component of shade tolerance. Ecol. Appl. 5: 517-532. Koch, G.W., Sillett, S.C., Jennings, G.M., Davis, S.D. 2004. The limits to tree height. Nature 428: 851-854. Kubo, M., Sakio, H., Shimano, K., Ohno, K. 2004. Factors influencing seedling emergence and survival in Cercidiphyllum japonicum. Folia Geobot. 39: 225-234. Lefort, F., Brachet, S., Frascaria-Lacoste, N., Edwards, K.J., Douglas, G.C. 1999. Identification and characterization of microsatellite loci in ash (Fraxinus excelsior L.) and their conservation in the olive family (Oleaceae). Mol. Ecol. 8: 1088-1090. Liu, H., L.S. Bauer, R. Gao, T. Zhao, T.R. Petrice, R.A. Haack. 2003. Exploratory survey for the emerald ash borer, Agrilus planipennis (Coleoptera: Buprestidae), and its natural enemies in China. Gt. Lakes Entomol. 36: 191-204. Long, L. 2013. Direct and indirect impacts of emerald ash borer on forest bird communities. In, Entomology. The Ohio State University, OhioLINK Electronic Theses and Dissertations Center, p. 165. Lovett, G.M., Canham, C.D., Arthur, M.A., Weathers, K.C., Fitzhugh, R.D. 2006. Forest ecosystem responses to exotic pests and pathogens in eastern North America. Bioscience 56: 395-405. Lovett, G.M., Christenson, L.M., Groffman, P.M., Jones, C.G., Hart, J.E., Mitchell, M.J. 2002. Insect defoliation and nitrogen cycling in forests. Bioscience 52: 335-341. MacFarlane, D.W., Meyer, S.P., 2005. Characteristics and distribution of potential ash tree hosts for emerald ash borer. For. Ecol. Manag. 213: 15-24. Malysheva-Otto, L.V., Ganal, M.W., Roder, M.S. 2006. Analysis of molecular diversity, population structure and linkage disequilibrium in a worldwide survey of cultivated barley germplasm (Hordeum vulgare L.). BMC Genet. 7:6 10.1186/1471-2156-7-6. Matos, E.L.S., Oliveira, E.J., Jesus, O.N., Dantas, J.L.L. 2013. Microsatellite markers of genetic diversity and population structure of Carica papaya. An. Appl. Biol. 163: 298-310. McClure, M.S., Salom, S.M., Shields, K.S. 2003. Hemlock woolly adelgid. Forest Health Technology Enterprise Team, U.S. Dept. of Agriculture, Forest Service, Morgantown, p 14. McCullough, D.G., Mercader, R.J. 2012. Evaluation of potential strategies to SLow Ash Mortality (SLAM) caused by emerald ash borer (Agrilus planipennis): SLAM in an urban forest. Int. J. Pest Manage. 58: 9-23. McCullough, D.G., Mercader, R.J., Siegert, N.W. 2015. Developing and integrating tactics to slow ash (Oleaceae) mortality caused by emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 349-358. McCullough, D.G., Poland, T.M., Anulewicz, A.C., Lewis, P., Cappaert, D. 2011. Evaluation of Agrilus planipennis (Coleoptera: Buprestidae) control provided by emamectin benzoate and two neonicotinoid insecticides, one and two seasons after treatment. J. Econ. Entomol. 104: 1599-1612. 177

McCullough, D.G., Poland, T.M., Cappaert, D. 2009. Attraction of the emerald ash borer to ash trees stressed by girdling, herbicide treatment, or wounding. Can. J. For. Res. 39: 1331-1345. McCullough, D.G., Siegert, N.W. 2007. Estimating potential emerald ash borer (Coleoptera : Buprestidae) populations using ash inventory data. J. Econ. Entomol. 100: 1577- 1586. Mercader, R.J., McCullough, D.G., Storer, A.J., Bedford, J.M., Heyd, R., Poland, T.M., Katovich, S. 2015. Evaluation of the potential use of a systemic insecticide and girdled trees in area wide management of the emerald ash borer. For. Ecol. Manag. 350: 70-80. Mercader, R.J., Siegert, N.W., Liebhold, A.M., McCullough, D.G. 2009. Dispersal of the emerald ash borer, Agrilus planipennis, in newly-colonized sites. Agric. For. Entomol. 11: 421-424. Mercader, R.J., Siegert, N.W., Liebhold, A.M., McCullough, D.G. 2011. Simulating the effectiveness of three potential management options to slow the spread of emerald ash borer (Agrilus planipennis) populations in localized outlier sites. Can. J. For. Res. 41: 254-264. Nei, M., Maruyama, T., Chakraborty, R. 1975. Bottleneck effect and genetic variability in populations. Evolution 29: 1-10. Niesenbaum, R.A. 1992. The effects of light environment on herbivory and growth in the dioecious shrub Lindera benzoin (Lauraceae). Am. Midl. Nat. 128: 270-275. Noguchi, K., Gel, Y.R., Brunner, E., Konietschke, F. 2012. nparLD: An R software package for the nonparametric analysis of longitudinal data in factorial experiments. J. Stat. Softw. 50: 1-23. Okubo, A., Levin, S.A. 1989. A theoretical framework for data analysis of wind dispersal of seeds and pollen. Ecology 70: 329-338. Ouborg, N.J., Vergeer, P., Mix, C. 2006. The rough edges of the conservation genetics paradigm for plants. J. Ecol. 94: 1233-1248. Parker, G.R., Leopold, D.J. 1983. Replacement of Ulmus americana L. in a mature east- central Indiana woods. Bull. Torrey Bot. Club 110: 482-488. Pautasso, M. 2009. Geographical genetics and the conservation of forest trees. Perspect. Plant Ecol. Evol. Syst. 11: 157-189. Peakall, R., Smouse, P.E. 2006. GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6: 288-295. Perry, K.I., Herms, D.A. 2016a. Response of the forest floor invertebrate community to canopy gap formation caused by early stages of emerald ash borer-induced ash mortality. For. Ecol. Manag. 375: 259-267. Perry, K.I., Herms,D.A. 2016b. Short-term responses of ground beetles to forest changes caused by early stages of emerald ash borer (Coleoptera: Buprestidae)-induced ash mortality. Environ. Entomol. 45: 616-626. Phillips, J.D., Marion, D.A., Turkington, A.V. 2008. Pedologic and geomorphic impacts of a tornado blowdown event in a mixed pine-hardwood forest. Catena 75: 278-287.

178

Plath, M., Dorn, S., Riedel, J., Barrios, H., Mody, K. 2012. Associational resistance and associational susceptibility: specialist herbivores show contrasting responses to tree stand diversification. Oecologia 169: 477-487. Plumptre, A.J. 1995. The importance of “seed trees” for the natural regeneration of selectively logged tropical forest. Commonw. Forest. Rev. 74: 253-258. Potter, K.M., Jetton, R.M., Dvorak, W.S., Hipkins, V.D., Rhea, R., Whittier, W.A. 2012. Widespread inbreeding and unexpected geographic patterns of genetic variation in eastern hemlock (Tsuga canadensis), an imperiled North American conifer. Conserv. Genet. 13: 475-498. Prevosto, B., Gavinet, J., Ripert, C., Fernandez, C. 2015. Identification of windows of emergence and seedling establishment in a pine Mediterranean forest under controlled disturbances. Basic Appl. Ecol. 16: 36-45. Pureswaran, D.S., Poland, T.M. 2009. Host Selection and Feeding Preference of Agrilus planipennis (Coleoptera: Buprestidae) on Ash (Fraxinus spp.). Environ. Entomol. 38: 757-765. Quiroz, A., Pettersson, J., Pickett, J.A., Wadhams, L.J., Niemeyer, H.M. 1997. Semiochemicals mediating spacing behavior of bird cherry-oat aphid, Rhopalosiphum padi feeding on cereals. J. Chem. Ecol. 23: 2599-2607. Ramstad, K.M., Colbourne, R.M., Robertson, H.A., Allendorf, F.W., Daugherty, C.H. 2013. Genetic consequences of a century of protection: Serial founder events and survival of the little spotted kiwi (Apteryx owenii). Proc. R. Soc. B 280: 20130576. Redman, A.M., Scriber, J.M. 2000. Competition between the gypsy moth, Lymantria dispar, and the northern tiger swallowtail, Papilio canadensis: Interactions mediated by host plant chemistry, pathogens, and parasitoids. Oecologia 125: 218-228. Roberts, J.H., Angermeier, P.L., Hallerman, E.M. 2013. Distance, dams and drift: what structures populations of an endangered, benthic stream fish? Freshwater Biol. 58: 2050-2064. Rodriguez-Saona, C., Poland, T.M., Miller, J.R., Stelinski, L.L., Grant, G.G., de Groot, P., Buchan, L., MacDonald, L. 2006. Behavioral and electrophysiological responses of the emerald ash borer, Agrilus planipennis, to induced volatiles of Manchurian ash, Fraxinus mandshurica. Chemoecology 16: 75-86. Root, R.B. 1973. Organization of a plant-arthropod association in simple and diverse habitats: Fauna of collards (Brassica oleracea). Ecol. Monogr. 43: 95-120. Rubino, L., Charles, S., Sirulnik, A.G., Tuininga, A.R., Lewis, J.D. 2015. Invasive insect effects on nitrogen cycling and host physiology are not tightly linked. Tree Physiol. 35: 124-133. Runkle, J.R. 1982. Patterns of disturbance in some old-growth mesic forests of eastern North America. Ecology 63: 1533-1546. Runkle, J.R. 1984. Development of woody vegetation in treefall gaps in a beech-sugar maple forest. Holarct. Ecol. 7: 157-164. Runkle, J.R. 1990. Gap dynamics in an Ohio Acer-Fagus forest and speculations on the geography of disturbance. Can. J. For. Res. 20: 632-641. Runkle, J.R., Yetter, T.C. 1987. Treefalls revisited: Gap dynamics in the southern Appalachians. Ecology 68: 417-424. 179

Scharenbroch, B.C., Bockheim, J.G. 2008a. Gaps and soil C dynamics in old growth northern hardwood-hemlock forests. Ecosystems 11: 426-441. Scharenbroch, B.C., Bockheim, J.G. 2008b. The effects of gap disturbance on nitrogen cycling and retention in late-successional northern hardwood-hemlock forests. Biogeochemistry 87: 231-245. Sholes, O.D.V. 2008. Effects of associational resistance and host density on woodland insect herbivores. J. Anim. Ecol. 77: 16-23. Siegert, N.W., McCullough, D.G., Liebhold, A.M., Telewski, F.W. 2014. Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Divers. Distrib. 20: 847-858. Siegert, N.W., McCullough, D.G., Williams, D.W., Fraser, I., Poland, T.M., Pierce, S.J. 2010. Dispersal of Agrilus planipennis (Coleoptera: Buprestidae) from discrete epicenters in two outlier sites. Environ. Entomol. 39: 253-265. Smith, A. 2006. Effects of community structure on forest susceptibility and response to the emerald ash borer invasion of the Huron River Watershed in southeastern Michigan. In, Entomology. The Ohio State University, Columbus, OH, p. 122. Smith, A., Herms, D.A., Long, R.P., Gandhi, K.J.K. 2015. Community composition and structure had no effect on forest susceptibility to invasion by the emerald ash borer (Coleoptera: Buprestidae). Can. Entomol. 147: 318-328. Smitley, D.R., Doccola, J.J., Cox, D.L. 2010a. Multiple-year protection of ash trees from emerald ash borer with a single trunk injection of emamectin benzoate, and single- year protection with an imidacloprid basal drench. Arboric. Urban For. 36: 206-211. Smitley, D.R., Rebek, E.J., Royalty, R.N., Davis, T.W., Newhouse, K.F. 2010b. Protection of individual ash trees from emerald ash borer (Coleoptera: Buprestidae) with basal soil applications of imidacloprid. J. Econ. Entomol. 103: 119-126. Sork, V.L., Davis, F.W., Smouse, P.E., Apsit, V.J., Dyer, R.J., Fernandez, J.F., Kuhn, B. 2002. Pollen movement in declining populations of California Valley oak, Quercus lobata: Where have all the fathers gone? Mol. Ecol. 11: 1657-1668. Spurr, S.H., Barnes, B.V. 1980 Forest Ecology. Wiley, New York, NY. Stokes, K., Stiling, P. 2013. Effects of relative host plant abundance, density and inter-patch distance on associational resistance to a coastal gall making midge, Asphondylia borrichiae (Diptera: Cecidomyiidae). Fla. Entomol. 96: 1143-1148. Su, Z.H., Richardson, B.A., Zhuo, L., Jiang, X.L. 2017. Divergent population genetic structure of the endangered Helianthemum (Cistaceae) and its implication to conservation in northwestern China. Front. Plant Sci. 7: 2010. doi: 10.3389/fpls.2016.02010. Sugiura, N. 1978. Further analysts of the data by akaike’s information criterion and the finite corrections. Commun. Stat. Theory Methods 7: 13-26. Sydnor, T.D., Bumgardner, M., Todd, A. 2007. The potential economic impacts of emerald ash borer (Agrilus planipennis) on Ohio, US, communities. Arboric. Urban For. 33: 48-54. Tahvanainen, J.O., Root, R.B. 1972. The influence of vegetational diversity on the population ecology of a specialized herbivore, Phyllotreta cruciferae (Coleoptera: Chrysomelidae). Oecologia 10: 321-346. 180

Talamo, A., Barchuk, A., Cardozo, S., Trucco, C., Maras, G., Trigo, C. 2015. Direct versus indirect facilitation (herbivore mediated) among woody plants in a semiarid Chaco forest: A spatial association approach. Austral Ecol. 40: 573-580. Tanis, S.R., McCullough, D.G. 2012. Differential persistence of blue ash and white ash following emerald ash borer invasion. Can. J. For. Res. 42: 1542-1550. Tanis, S.R., McCullough, D.G. 2015. Host resistance of five Fraxinus species to Agrilus planipennis (Coleoptera: Buprestidae) and effects of paclobutrazol and fertilization. Environ. Entomol. 44: 287-299. Taylor, R.A.J., Bauer, L.S., Poland, T.M., Windell, K.N. 2010. Flight performance of Agrilus planipennis (Coleoptera: Buprestidae) on a flight mill and in free flight. J. Insect Behav. 23: 128-148. Thorn, S., Bussler, H., Fritze, M.A., Goeder, P., Muller, J., Weiss, I. Seibold, S. 2016. Canopy closure determines arthropod assemblages in microhabitats created by windstorms and salvage logging. For. Ecol. Manag. 381: 188-195. Tingley, M.W., Orwig, D.A., Field, R., Motzkin, G. 2002. Avian response to removal of a forest dominant: Consequences of hemlock woolly adelgid infestations. J. Biogeogr. 29: 1505-1516. Ugine, T.A., Gardescu, S., Hajek, A.E. 2011. The effect of exposure of imidacloprid on Asian longhorned beetle (Coleoptera: Cerambycidae) survival and reproduction. J. Econ. Entomol. 104: 1942-1949. Ulyshen, M.D., Klooster, W.S., Barrington, W.T., Herms, D.A. 2011. Impacts of emerald ash borer-induced tree mortality on leaf litter arthropods and exotic earthworms. Pedobiologia 54: 261-265. United States Environmental Protection Agency. 2009. Memorandum: Ecological risk assessment for emamectin benzoate use as a tree injection insecticide to control arthropod pests. PC Code 122806. US EPA, Washington, DC, USA. Vannatta, A.R., Hauer, R.H., Schuettpelz, N.M. 2012. Economic analysis of emerald ash borer (Coleoptera: Buprestidae) management options. J. Econ. Entomol. 105: 196- 206. van Oosterhout, C., Hutchison, W.F., Wills, D.P.M., Shipley, P.F. 2003. Micro-Checker User Guide. The University of Hull. Villari, C., Herms, D.A., Whitehill, J.G.A., Cipollini, D., Bonello, P. 2016. Progress and gaps in understanding mechanisms of ash tree resistance to emerald ash borer, a model for wood-boring insects that kill angiosperms. New Phytol. 209: 63-79. Vranckx, G., Jacquemyn, H., Muys, B., Honnay, O. 2012. Meta-analysis of susceptibility of woody plants to loss of genetic diversity through habitat fragmentation. Conserv. Biol. 26: 228-237. Wallander, E. 2008. Systematics of Fraxinus (Oleaceae) and evolution of dioecy. Plant Syst. Evol. 273: 25-49. Watt, A.S. 1947. Pattern and process in the plant community. J. Ecol. 35: 1-22. Wee, A.K.S., Chunhong, L., Dvorak, W.S. 2012. Genetic diversity in natural populations of Gmelina arborea: implications for breeding and conservation. New Forests 43: 411- 428.

181

Whitmore, T.C. 1989. Canopy gaps and the two major groups of forest trees. Ecology 70: 536-538. Woolaver, L.G., Nichols, R.K., Morton, E.S., Stutchbury, B.J.M. 2013. Population genetics and relatedness in a critically endangered island raptor, Ridgway's Hawk Buteo ridgwayi. Conserv. Genet. 14: 559-571. Young, A., Boyle, T., Brown, T. 1996. The population genetic consequences of habitat fragmentation for plants. Trends Ecol. Evol. 11: 413-418.

182