Assessing the Population Genetic Structure of the Endangered Cucumber Tree (Magnolia Acuminata) in Southwestern Ontario Using Nuclear and Chloroplast Genetic Markers

Home , Liriodendron, Magnolia acuminata

ASSESSING THE POPULATION GENETIC STRUCTURE OF THE ENDANGERED CUCUMBER TREE (MAGNOLIA ACUMINATA) IN SOUTHWESTERN ONTARIO USING NUCLEAR AND CHLOROPLAST GENETIC MARKERS.

A Thesis Submitted to the Committee on Graduate Studies in Partial Fulfillment to the Requirements for the Degree of Master of Science in the Faculty of Arts and Science

Trent University Peterborough, Ontario, Canada © Copyright by Cara E. Budd 2014 Environmental and Life Sciences M.Sc. Graduate Program September 2014

Assessing the population genetic structure of the endangered Cucumber tree (Magnolia acuminata) in southwestern Ontario using nuclear and chloroplast genetic markers.

Cara E. Budd

ABSTRACT

Magnolia acuminata (Cucumber tree) is the only native Magnolia in Canada, where it is both federally and provincially listed as endangered. Magnolia acuminata in Canada can be found inhabiting pockets of Carolinian forest within Norfolk and Niagara regions of southwestern Ontario. Using a combination of nuclear and chloroplast markers, this study assessed the genetic diversity and differentiation of M. acuminata in Canada, compared to samples from the core distribution of this species across the United States. Analyses revealed evidence of barriers to dispersal and gene flow among Ontario populations, although genetic diversity remains high and is in fact comparable to levels of diversity estimated across the much broader range of M. acuminata in the USA. When examining temporal differences in genetic diversity, our study found that seedlings were far fewer than mature trees in Ontario, and in one site in particular, diversity was lower in seedlings than that of the adult trees. This study raises concern regarding the future viability of M. acuminata in Ontario, and conservation managers should factor in the need to maintain genetic diversity in young trees for the long-term sustainability of M. acuminata in Ontario.

Key words: conservation genetics, population genetic structure, endangered, Magnolia acuminata, Carolinian, habitat fragmentation, microsatellites, cpDNA, genetic diversity, genetic differentiation

ACKNOWLEDGEMENTS

First and foremost, I am undoubtedly extremely lucky to have parents whom have worked their whole lives to provide me with a life full of opportunity. If it wasn’t for their love, patience, and unlimited support, both emotionally and financially, I would not be half the person I am today. There are no words that could describe how grateful I am to have them, and I dedicate this project to them.

I would like to express my sincerest gratitude to my project supervisor, Dr. Joanna Freeland, for her patience, enthusiasm, and optimism from beginning to end. Even at what seemed to be the most trying times, she always managed to stay positive and see the light at the end of the tunnel. Thank you for the constant reminder that everything will not go as planned and that it’s okay… do no stress, accept the situation and adapt to it. In addition, a big thanks to Dr. Marcel Dorken and Bill Crins for their willingness to participate in my supervisory committee, as well as their invaluable guidance and expertise regarding plant genetics and evolutionary biology. Thank you for taking the time out of your busy schedules to provide me with the support I needed at all levels of this project.

I’d like to thank Graham Buck and Donald Kirk of the Ministry of Natural Resources for getting the ball rolling on this project. Without their instrumental assistance I would not have been able to acquire the necessary permission to collect samples from private landowners, nor have all the essential documents and background knowledge regarding the location and ecology of these beautiful trees. I must also acknowledge Danny Bernard of Big Creek Conservation Authority for providing me with the opportunity to visit and sample trees within the National Wildlife Area, as well as his vast knowledge regarding the natural heritage and history of the Long Point region. It was a truly an amazing and fun-filled research experience, and one I hope to have again. I cannot forget to thank Elizabeth Zimmer and Gabe Johnson at the Smithsonian Institute, for generously providing the Magnolia acuminata leaves and associated information from across the United States, allowing for the conservative comparison of diversity between the two regions.

III

A special thanks to Jennifer Paul for her help, accompaniment, and many laughs in the field and lab. To Marco Raponi, the genius behind my sampling equipment! Thank you for the drives, coffee runs, and what seemed to be regular “therapy” sessions. I really couldn’t have asked for a better friend to have taking this MSc. journey with me. Adam Wilford, thank you for being so tall and most importantly, my easy-going better half. Without you I wouldn’t have the leaves to complete this project nor hair on my head. It means the world to me that I was able to share this project, experience and the love of Ontario’s southwest with you.

To all those whom I failed to mention, your contributions have in no way gone unnoticed and I appreciate all you have done to make this project possible.

TABLE OF CONTENTS

ABSTRACT II

ACKNOWLEDGEMENTS III-IV

LIST OF FIGURES VI

LIST OF TABLES VII

GENERAL INTRODUCTION 1

INTRODUCTION 8

METHODS 21

RESULTS 32

DISCUSSION 43

REFERENCES 68

APPENDICIES 86 APPENDIX I- DISTRIBUTION AND NUMBER OF SAMPLED MAGNOLIA ACUMINATA ACROSS ONTARIO AND UNITED STATES 86 APPENDIX II- MAP OF THE DISTRIBUTION OF UNITED STATES MAGNOLIA ACUMINATA SAMPLES 87 APPENDIX III- RAW GENOTYPIC DATA: III-A. ONTARIO MAGNOLIA ACUMINATA 88-96 III-B. UNITED STATES MAGNOLIA ACUMINATA 97-99 APPENDIX IV- CONSERVATIVE COMPARISONS OF ALLELIC DIVERSITY BETWEEN: III-A. ONTARIO AND THE UNITED STATES 100-101 III-B. NORFOLK COUNTY AND THE MUNICIPALITY OF NIAGARA 102-103 III-C. NORFOLK COUNTY AND THE UNITED STATES 104-105 III-D. MUNICIPALITY OF NIAGARA AND THE UNITED STATES 106-107 APPENDIX V- CLUSTER ASSIGNMENT VALUES: IV-A. STRUCTURE POPULATION CLUSTER ASSIGNMENT 108 IV-B. STRUCTURE INDIVIDUAL CLUSTER ASSIGNMENT 109-113 IV-C. TESS POPULATION CLUSTER ASSIGNMENT 114 IV-D. TESS INDIVIDUAL CLUSTER ASSIGNMENT 115-119

LIST OF FIGURES

FIGURE 1: DISTRIBUTION OF COLLECTED M. ACUMINATA SAMPLES FROM NATURAL STANDS WITHIN ONTARIO AND THE UNITED STATES. 22

FIGURE 2: ∆K RESULTS BASED ON STRUCTURE ANALYSIS. ∆K CALCULATED BY STRUCTURE HARVESTER AND ACCORDING TO EVANNO ET AL. (2005) FOR K VALUES RANGING FROM 1 TO 8. 37

FIGURE 3: TESS CLUSTERING ANALYSIS OF M. ACUMINATA SAMPLED ACROSS 12 SITES IN ONTARIO.

CLUSTERING WAS COMPLETED USING (A) CAR AND (B) BYM MODELS FOR K VALUES RANGING FROM 2 TO 10. 38

FIGURE 4: DISTRIBUTION OF M. ACUMINATA CLUSTERS IN NORFOLK COUNTY BASED ON CLUSTER ANALYSIS RESULTS. 39

FIGURE 5: DISTRIBUTION OF M.ACUMINATA CLUSTERS IN THE MUNICIPALITY OF NIAGARA BASED ON CLUSTER ANALYSIS RESULTS. 40

FIGURE 6: POSTERIOR ESTIMATES OF ASSIGNMENT PROBABILITIES FOR (A) STRUCTURE AND

(B) TESS BYM MODEL, WHERE KMAX= 7. SITE AND ASSOCIATED CLUSTER NUMBERS ARE DEFINED IN TABLE 7. 41

LIST OF TABLES

TABLE 1: NUCLEAR MICROSATELLITE PRIMERS USED TO ASSESS THE POPULATION GENETIC STRUCTURE OF M. ACUMINATA IN ONTARIO AND THE UNITED STATES. 23

TABLE 2: CHLOROPLAST MINISATELLITE REGIONS FOR WHICH AMPLIFICATION PRIMERS WERE DESIGNED. 25

TABLE 3: GENETIC VARIABILITY OF FIVE MICROSATELLITE LOCI AVERAGED ACROSS ALL LOCI AND POPULATIONS OF M. ACUMINATA USING GENOTYPIC AND PHENOTYPIC DATA. 34

TABLE 4: SUMMARY OF GLOBAL F STATISTICS (FST) AND JOST D (D_EST) AMONG THE 7 IDENTIFIED CLUSTERS IN ONTARIO (K=7), NORFOLK COUNTY (K=5), AND THE MUNICIPALITY OF NIAGARA (K=2). 35

TABLE 5: ANALYSIS OF MOLECULAR VARIANCE CALCULATED USING PHENOTYPIC DATA FOR ONTARIO, NORFOLK COUNTY, AND THE MUNICIPALITY OF NIAGARA. 36

TABLE 6: PAIRWISE FST AND D_EST BETWEEN THE INFERRED CLUSTERS OF M. ACUMINATA IN ONTARIO. 36

TABLE 7: ONTARIO M. ACUMINATA SAMPLE SITES AS DEFINED BY GEOGRAPHIC LOCATION AND CLUSTER. CLUSTER ASSIGNMENTS BASED ON TESS BYM MODEL AND STRUCTURE RESULTS FOR K=7. 41

TABLE 8: ALLELE SIZES FOR ALL FIVE CHLOROPLAST REGIONS IDENTIFIED AMONG ALL INDIVIDUALS FROM ONTARIO AND US POPULATIONS OF MAGNOLIA ACUMINATA. 42

TABLE 9: GENETIC DIVERSITY OF CANOPY TREES AND SEEDLINGS OF ONTARIO M. ACUMINATA USING GENOTYPIC AND PHENOTYPIC DATA FROM FIVE MICROSATELLITE LOCI. 42

TABLE 10: GENETIC DIVERSITY OF CANOPY TREES AND SEEDLINGS OF ONTARIO M. ACUMINATA WITHIN SITES NWA AND FD USING GENOTYPIC AND PHENOTYPIC DATA. 43

VII

GENERAL INTRODUCTION

Deciduous forest dominated North America during the latter part of the

Mesozoic era (Braun, 1947). By the Eocene epoch, vegetation in the west had shifted in response to latitudinal climate shifts, resulting in the contraction of subtropical species and the expansion of those more temperate species found in the north, with dicot fossils from the Upper Cretaceous revealing the dominance of the broad-leaved forest across eastern North America. During this time, the Coastal Plain of the east and the Appalachian Highlands contained a large proportion of the existing genera of the contemporary deciduous forest (Braun, 1955) including one of the oldest genera of broad-leaved trees in North America, Magnolia (Magnoliaceae). While time of divergence among Magnolia spp. is not completely known, Magnolias were not only abundant but substantial diversification had occurred during the Miocene epoch when North America had a warm and humid, tropical climate (Goldenberg et al., 1990). Magnolia spp. were widely distributed, predominantly inhabiting riparian areas, extending southward from what is now the current latitude of Nunivak Island,

Alaska (Flannery, 2003). While repeated stochastic environmental events altered much of the landscape, southern United States remained a haven for a variety of taxa, especially during the Wisconsinan glaciation when northern species sought refugia south of the Laurentian Ice Sheet (Braun 1955; Wright 1964; Waldron 1993;

Flannery 2003). Among the many species which would have required southerly refugia due to the inhospitability of the Great Lakes Basin during full glaciation was the Cucumber tree (Magnolia acuminata), Canada’s only native Magnolia, which is

2 both federally and provincially listed as endangered (Species at Risk in Ontario,

2014).

During the Wisconsinan glaciation, approximately 8,500 to 11, 000 years ago, ice engulfed the Great Lakes Basin, covering all areas of southern Ontario. While glaciation created the extensive system of moraines, drumlins, and eskers which are prominent throughout the region, the creation of lake plains played the biggest role in the reestablishment of forests throughout southern Ontario (Maycock, 1962) after

3,500 BP when rich, moist soils were characteristic of the region (Waldron, 2003).

The Lake Erie basin was the first deglaciated region in Ontario (Lewis et al., 2008), where land was thought to have first appeared just northeast of the present Forks of the Thames in London, Ontario. The Thames was the first major river to develop in

Ontario, and during its early stages, it drained into the current Saugeen, Maitland, and Grand River watersheds, carrying melt water southwest towards the Mississippi

River (Murphy, 2010). As the Erie ice lobe melted and the glacial retreat advanced, the glacial lakes Warren, Grassmere, and Lundy (Murphy, 2010; UTRCA, 1998) expanded northward and acted as immediate barriers for northward migration by multiple taxa (Lewis et al., 2008). However, Ontario was readily accessible in the south from the non-glaciated Ohio Valley, notably one of the most important access points for movement into the province (Lewis et al., 2008); in addition, two alternative passageways into Ontario have also been described: one ran west of Lake

St. Clair and another through a narrow corridor of the Niagara Peninsula (Maycock,

1962).

Phylogeography is the study of factors that govern the distribution of species’ lineages over space and time. Over the last few decades, phylogeographic studies have begun to answer fundamental questions regarding the factors and processes which influence the distribution and population genetic structuring of species.

More specifically, phylogeographic studies can aid in inferring the relationship between geological history and historical patterns of gene flow on contemporary population genetic structure (Avise, 2000).

At the intraspecific level, Hewitt (1996), Taberlet et al. (1998), and Petit et al.

(2002) have shown that glacial refugia and post-glacial movement along migration pathways have strongly shaped the spatial genetic structuring of contemporary populations. Furthermore, gene flow and vicariance among populations can result in distinct genetic signatures and the phylogeography of a species (Nason et al.,

2002), which is further influenced by the degree of connectivity and restrictions to gene flow by seed dispersal and pollinator activity (Hewitt 2000). It is widely accepted that the distribution of extant populations had been largely influenced by stochastic processes and topographical barriers, but also the location of suitable habitat and the frequency and distance of dispersal during interglacial periods. In many plant species, the recurring periods of Pleistocene cooling resulted in repeated occasions of range contractions and expansions, thus generating geographical patterns of genetic structure as a product of long-lasting isolation and population bottlenecks (Hewitt, 1996). The availability of phylogeographic studies on dominant, wind-pollinated species is abundant, particularly those of the late

Pleistocene and Holocene (Jackson et al., 2000). For less common, insect-pollinated species like Magnolia acuminata, these studies are underrepresented and limited, leaving a critical gap in our knowledge regarding both phylogeography of these taxa and its effect on population genetic structures.

Phylogeographic and population genetic studies often used multiple techniques to not only elucidate potential refuge areas, but also to measure the extent of gene flow and dispersal among extant populations. Two complementary methods include the genotyping and sequencing of chloroplast DNA and genotyping of nuclear genetic markers. Multiple-approach studies are useful in molecular research because they may provide insights into species’ molecular evolutionary and phylogenetic potential which otherwise would be unattainable from a single data source. Chloroplast DNA can be used to infer deep historical patterns of persistence and dispersal via seed made possible due to the slow mutation rates and most often, maternal inheritance. Unlike chloroplast DNA, nuclear genetic markers, such as microsatellites, are co-dominant markers that are valued for their high levels of polymorphisms (Guichoux et al., 2011). Microsatellites are characterized by individual-based variations in the number of tandem repeats of a simple motif

(Kelkar et al., 2010). These variations provide measures of genetic diversity allowing one to make inferences about population-genetic processes including genetic structure, parentage, as well as pollen and seed dispersal of more contemporary populations.

It has been long presumed that temperate deciduous forest species of eastern

North America and Great Lakes region persisted in low-latitudinal glacial refugia during the Last Glacial Maximum (LGM) (Austin et al., 2002; Braun 1947, 1955;

Wright, 1964; Waldron, 1993; Flannery, 2003; Church et al., 2003; Hewitt, 2004;

Loehle, 2007). Preceding the LGM, the Coastal Plains and the Appalachian highlands contained a large proportion of the existing genera of the temperate deciduous forest and had a high number of species with a restricted range (Braun, 1955), many of which were found distributed along riparian habitat of the steep banks of

Virginia’s Blue Ridge, through the Carolinas, and into Georgia (Braun 1950, 1955) which is consistent with the current distribution of Magnolia acuminata across the eastern United States. Similarly, bluff habitats along major north-south trending streams in the southeast were also thought to have harbored temperate deciduous species (Delcourt and Delcourt 1975, 1977a, 1979). It is because of its rich floristic history and variety of soils, climates, and topographies, that the Appalachians comprise a variety of microsites, providing a wide range of suitable refugia for multiple taxa, including numerous species of aquatic and wetland hydrophytes

(Stuckey et al., 1993), Spring peeper (Pseudacris crucifer) (Austin et al., 2002), Jack pine (Pinus banksiana) (Wright, 1964; Yeatman, 1967; Delcourt, 1979), and birch spp. (Betula lutea and Betula lenta) (Potzger, 1946). More recently, genetic studies have not only disputed the vegetative assemblages during the LGM, but also the traditional far south refugia sites, suggesting the occurrence of a heterogeneous assortment of plant taxa at mid-latitudes (Loehle, 2007). In addition to the three known potential refuge sites for mixed mesophytic forest species and other

6 deciduous tree species, Nonconnah Creek in the bluff lands of southwestern

Tennessee (Delcourt and Delcourt, 1980), Goshen Springs in south-central Alabama, and Sheelar Lake, northern Florida (Watts and Hansen, 1994), genetic evidence more recently has identified cryptic refugia in the Appalachian basin of West

Virginia, Kentucky, and Tennessee (McLachlan et al., 2005; Soltis et al., 2006;

Gonzales et al., 2008).

Using chloroplast DNAto reconstruct the postglacial range expansion of

American beech (Fagus grandifolia) and Red maple (Acer rubrum) in Eastern North

America, rather than the traditional fossil pollen records, McLachlan et al. (2005) revealed the likely potential of low-density populations having persisted within 500 km of the Laurentian Ice Sheet. The diversity of haplotypes for both species found at mid-latitudes act as indicators of refuge location, areas where lineages were able to diversify through mutation (Hewitt, 1996; Petit et al., 2002). The identification of distinct northern and southern haplotypes and/or the absence of southern haplotypes from mid- to upper-latitudes seen among Liriodendron tulipifera (Sewell et al., 1996), Tellima grandiflora (Soltis et al., 1991), Quercus rubra (McDougal,

1984), and Dirca palustris (Peterson, 2013) suggest discontinuities in phylogeny and the presence of interior northerly refugia sites or microrefugia closer to the ice margin.

Hewitt (1993) first introduced the “leading-edge” hypothesis which suggests that populations expanded northward from a southern refugium via long-distance

7 dispersal, facilitated by individuals along the leading edge of recolonization. If the recolonization process is associated with population bottlenecks, expansion from a single refugium should result in a decline in genetic diversity as latitude increases following repeated dispersal and founding events, leading to the loss of alleles. The uniformity of genetic structure across a relatively broad spatial scale is expected to be maintained over long periods of time, especially among long-lived tree species

(Hewitt, 2004) such as M. acuminata, as the number of generations which have passed since recolonization may be insufficient for random mutation and/or local adaptation to have a substantial effect on population genetic structure. Magnolia acuminata likely possesses the ability for long-distance dispersal via water or animals, which historically may have played an important role in the reestablishment of populations in southern Ontario following the Last Glacial

Maximum. However, while hundreds of generations may have passed since recolonization of Magnolia acuminata in Ontario, the lack of genetic studies on this species makes it difficult to assess the impacts of geological history, dispersal, vicariance, and other ecological and evolutionary processes, including forest fragmentation, on the contemporary population genetic structure of this species.

Thus, in order to understand the processes which lead to this species endangerment, as well as the long-term viability of M. acuminata in southern Ontario, this study implemented complementary genetic approaches, aimed to provide the first conservation genetic study of this species, thereby providing information needed to ensure the recovery of this species in Canada.

INTRODUCTION

Despite the relatively large proportion of remaining natural areas in Canada, there has been a continuous increase in the number of threatened and endangered species especially among regions of high biodiversity. Over the last ten years, the number of species at risk in Canada has increased from 429 to 676 (COSEWIC, 2003 and 2013). With the loss of species having potential detrimental effects on biological, ecological, and evolutionary processes, there has been a heightened desire to better understand the implications of human land use on biodiversity and natural processes. A thriving human population has led to a subsequent increase in the demand and consumption of land, resources, and ecosystem services. Over the last few decades, considerable research has aimed to assess the effects of pollution, invasive species, overexploitation, as well as habitat loss, configuration, and fragmentation on population and community structures at various spatial scales

(Fahrig, 2003). While all can contribute to the decline of native species in Canada,

Kerr and Cihlar (2004) deem agricultural land use and agricultural pollution as the likely culprits of much of the endangerment. The conversion of natural areas into cropland notably results in long term habitat loss, leading to significant reductions in the resources needed to naturally maintain ecosystem dynamics and functions

(Kerr and Cihlar, 2004; Kerr and Deguise, 2004; Turner et al., 2003).

In Canada, species in southern Ontario are at the greatest risk of endangerment because they inhabit the most densely populated and agricultural regions in the country (Beardmore et al. 2006). In the mixedwood plains ecozone,

9 which spans the Quebec City- Windsor corridor, lives approximately half of the

Canadian human population alongisde extensive agricultural land use, with a large proportion of cropland dedicated to corn and soy production. Both of these crop types are linked to high species endangerment (Kerr and Deguise, 2004; Kerr and

Cihlar, 2004). The mixedwood plains is the smallest, yet most biodiverse ecozone in

Canada, covering approximately 1.5% of Canada’s land area (Bernhardt, McGill

University; Canadian Council on Ecological Areas, 1995) which is distributed among four ecoregions: St. Lawrence Lowlands, Frontenac Axis, Manitoulin- Lake Simcoe, and the Lake Erie lowland. The mixedwoods are predominantly characterized by temperate deciduous forest, and the Lake Erie lowland, more commonly known as the Carolinian zone, is home to the species-rich Carolinian forest. Described as the most endangered ecosystem in Canada, the Carolinian zone itself covers a mere

0.25% of Canada’s area, yet houses 25% of Canada’s human population (Jalava et al.,

2007, ERCA, 2009) and 160 species designated Species-at-Risk (Species at Risk in

Ontario, 2014).

Occurring in a region where land-use pressures are high, over the last three centuries much of the previously continuous forest has been converted for agriculture and urbanization. Land conversion in this region is believed to have begun as early as 600 A.D (Suffling et al., 2003), and since then, has reduced forest cover from approximately 80% to as low as 11% in some areas (Jalava et al., 2007).

The Carolinian zone is considered the most ecologically-degraded area of the Great

Lakes basin (ERCA, 2009), with less than 15% of its natural cover remaining (Jalava

10 et al., 2007). While a substantial decline in forest cover can have long-term, detrimental effects on the population and genetic structure of any tree species, 30 of the 73 tree species that inhabit the Carolinian forest occur nowhere else in the country, and many are at the northernmost limit of their range (Beardmore et al.

2006). Among those 30 marginal tree species is Magnolia acuminata, the endangered Cucumber tree.

Magnolia acuminata is a large and hardy forest canopy species found throughout the Appalachian range extending from its most northernmost distribution among the Carolinian forests of southern Ontario, through New York to

Georgia, and westward through Arkansas (Ambrose and Aboud, 1984). The

Cucumber tree is the only native magnolia species in Canada and is both federally and provincially listed as endangered, with fewer than 300 known individuals remaining. Limited to two disjunct regions in southern Ontario, Norfolk County and the Municipality of Niagara, M. acuminata is found sparsely distributed among the

Carolinian forest remnants. While historic populations predominantly grew in riparian habitat (Beresford-Kroeger, 2003), contemporary populations inhabit inland forested sites with rich, moist soil, with a preference for elevated sites in or adjacent to swamps (Ambrose and Aboud, 1984; COSEWIC, 2010). Cucumber trees are moderately shade-tolerant (Strobl and Bland, 2000; Smith, 1990), although the seedlings require forest openings for establishment. After establishment, growth is rapid, and mature trees can reach heights of 30 m. Magnolia acuminata has protogynous flowers: its flowers’ stigmas become receptive before pollen is released

by the anthers, which bloom in May (Ambrose and Aboud, 1984) for a period much shorter than most Magnolia spp. (Smith, 1990). Magnolia acuminata is insect- pollinated, and following successful fertilization, the Cucumber tree bears young green fruits, from which it gets its name. As the fruits ripen, changing from green to red, they begin to bear bright, red-orange seeds which are typically released beginning in October (Ambrose and Aboud, 1984). While Cucumber trees fruit as early as 20 years, optimum seed production extends beyond 50 years (Ambrose and

Aboud, 1984).

Knowledge regarding the pollen and seed dispersal agents of M. acuminata is limited. Species of Magnoliaceae are highly specialized for beetle pollination (Thien,

1974; Matsuki et al., 2008). While there is the potential for pollination to occur via alternative vectors, the mutualisms between Magnolia spp. and beetles arose from long-term coevolution (Bernhardt, 2000). Effective pollinators of Magnolia spp. include a variety of beetles: Cerambycidae, Chrysomelidae, Curculionidae,

Scarabaeidae, Scraptiidae, and Staphylinidae (Bernhardt, 2000). A preliminary study conducted by the Ontario Ministry of Natural Resources on the constraints on sexual reproduction in M. acuminata observed beetles carrying pollen, which included Gaurotes cyanipennis (Say) (Cerambycidae) and Cephaloon sp.

(Cephaloidae) (Kevan, 2002). Despite the known inefficiencies of beetle pollination

(Thien, 1974), recent studies have shown that while long-distance dispersal events by beetles may be rare (Setsuko et al., 2007, Setsuko et al. 2013) compared to other animal pollinators including Hymenoptera and Lepidoptera, beetles carry

12 proportionally more outcrossed-pollen and higher levels of genetic diversity, reflecting significant interplant movement (Englund, 1993; Matsuki et al. 2008).

However, interplant movement is stand-density dependent due to potential competition for pollinators (Lowe et al., 2005; Sork and Smouse, 2006; Isagi et al.,

2000, 2007; Levin and Kerster, 1974; Fenster, 1991; Setsuko et al. 2013).

Magnolia acuminata seeds have morphologically evolved for long-distance dispersal via water, thus seed design has been a product of its natural riparian habitats (Beresford-Kroeger, 2003): the flat top and bottom allow for movement with the smallest current, and its circular shape provides the ability to spin and rotate with minimal water flow (Beresford-Kroeger, 2003). Despite the historical significance of seed movement via waterways, with respect to the potential contributions to the reestablishment of M. acuminata populations in Ontario following the last glaciation, more contemporary modes of dispersal may be limited to dispersal by birds and small mammals due to the absence of waterways in most

Ontario M. acuminata locations. While there is no direct evidence suggesting that birds are in fact the primary dispersers of seeds, seed characteristics are also typical of bird dispersal (van der Pijl, 1969). Though it is not known what species are responsible for seed movement within and among populations of M. acuminata, frugivorous species such as Wild turkey (Meleagris gallopavo) or Cedar waxwing

(Bombycilla cedrorum) are thought to play a major role in both the dispersal and consumption of the high-energy,oily seeds (Beresford-Kroeger, 2003)

Much of the reduction in the extent of the Carolinian forest in Ontario did not occur until the early 1800s, when European settlement, followed by extensive agriculture, accelerated the exploitation of natural resources in this region

(Waldron, 1993). Over a period of less than 200 years, much of the forest was lost to development and industry, reducing the Carolinian forest to nothing more than scattered woodlots, and threatening the persistence of many species, including the

Cucumber tree. More recently, selective lumbering, tree-cutting and clearing by land owners further threatens the natural viability of M. acuminata (COSEWIC, 2010).

Early records of M. acuminata in Ontario predominantly place Cucumber trees below the Niagara Escarpment, whereas all recent records place it above

(Waldron, 1993). Early climatic oscillations could have contributed to the range shift of the Cucumber tree over the Escarpment (Hebda and Irving, 2004), but the demand for domestic timber as well as the presence of hundreds of lumber mills throughout the region (Waldron, 1993), particularly below the Escarpment, could have contributed significantly to the shift in distribution. During this time,

Cucumber trees found within riparian habitats would have been readily accessible for harvest, resulting in the scattered, inland forest distribution of current populations in southern Ontario. Its soft, durable, straight-grained wood has been previously described as being similar to yellow-poplar (Liriodendron tulipifera), and they often have been marketed together for use in pallets, crates, furniture, plywood, and special products (Smith, 1990).

Forest ecosystems are essential habitats providing not only socio-economic value, but also a number of biological and ecosystem services, including the maintenance of biodiversity through the creation of resource niches and habitat patches (Pautasso, 2008). As forest systems become increasingly compromised, not only do the dynamics within the ecosystem become disrupted, but as previously mentioned, the long-term viability of many tree species can become increasingly threatened, particularly those which are already at risk. In addition to reducing habitat heterogeneity, anthropogenic land use creates a patchwork effect within a formerly continuous landscape, resulting in a series of small, isolated forest fragments and populations that often suffer from limited gene flow, random genetic drift, and increased levels of interbreeding (Ellstrand and Elam, 1993; Bunnell et al.,

2004, Barbeta et al., 2011) . As local population sizes of trees decline and distances between patches increase, populations can become progressively more susceptible to the ecological and genetic consequences of fragmentation. However, the severity and rate of impact on a species are strongly dependent on a number of factors, including effective population size, patterns of genetic variability, and life history traits (Tremblay, 2001, 2003).

While some of the ecological consequences of fragmentation include the immediate loss of habitat, reconfiguration of remnant habitat patches and increased isolation, changes to forest systems also have the potential to alter plant-animal interactions. Since reproduction requires successful pollination and fertilization, increased distance between reproducing individuals could result in the disruption of

15 reproductive ecology through shifts in plant-pollinator interactions (Powell and

Powell, 1987; Sih and Baltus, 1987; Rathcke and Jules, 1993), leading to a reduction in the number of pollinator events (Anguilar et al., 2006; Isagi et al., 2007; Hirayama et al., 2007; Barbeta et al., 2011) and a subsequent decline in reproductive potential through reductions of fruit and seed set. Consequently, as forest patches become increasingly isolated, there is the potential for reduced out-crossing events as the degree of isolation, which is dependent on the hostility of the intervening environment and dispersal abilities, strongly influences the amount of gene flow between population fragments and ultimately the fate of species-at-risk. An immediate drop in reproductive output results in reduced individual fitness, and over time, a loss of genetic diversity. These plant-animal interactions are essential components of population resilience and are directly associated with landscape and genetic connectivity through the movement of pollen and seeds. Unfortunately, limited pollen and seed dispersal is a common symptom of spatial isolation, particularly among species with a high reliance on animals to maintain demographic and genetic connectivity within and between populations. Thus, changes in the activity, abundance, and species of pollinators and other dispersal agents can contribute to the subsequent modification of population and community structure dynamics.

Gene flow among populations can be essential for species survival as it reduces the speed at which genetic diversity is lost following genetic drift. A loss or reduction in gene flow, and hence genetic diversity, can compromise a species’

16 ability to adapt to changing environmental conditions and respond to stochastic events (Frankham, 1995, 2003). Therefore, increased isolation through habitat fragmentation, often accompanied by changes in plant-animal interactions as previously discussed, can have significant impacts on population genetic structure by reducing gene flow and thus, potentially reducing genetic variability and increasing inbreeding (Slatkin, 1987; Young et al., 1996). However, the extent to which reduced gene flow causes populations to become genetically differentiated from one another depends on the effective population size, as the number of breeding adults strongly influences the rate at which genetic drift will lead to the fixation of alternative alleles in different populations (Frankham, 1995; Tremblay,

2001). Therefore, populations which can be naturally sustained genetically are those capable of maintaining within-stand genetic variability, which can be augmented by gene flow.

Since many species-at-risk comprise relatively small populations, in the absence of gene flow, genetic drift can rapidly lead to the loss of alleles (Ueno, 2005;

Tremblay, 2001). This in turn will lead to an increase in inbreeding within a relatively short time, and over the long term will compromise the population’s ability to respond and adapt to stochastic events (Anguilar et al., 2008; Hirayama et al., 2007; Holderegger et al., 2010, Isagi et al., 2004, 2007; Young et al., 1996). For example, in light of global climate shifts, tree species not only run the risk of being maladapted to shifts in environmental conditions and climate events such as extreme drought and fire regimes, but also the potential range expansion and

17 introductions of pests, pathogens, and invasive species. The lack of extreme low winter temperatures across western North America is suspected to be one of the main causes for the decimation of Lodgepole pine (Pinus contorta), Ponderosa pine

(Pinus ponderosa), and Whitebark pine (Pinus albicaulis) populations in many western forests, as warming winters fail to reduce populations of the Mountain pine beetle (Dendroctonus ponderosae) (Hicke et al., 2006). The American chestnut

(Castanea dentata) was ecologically extirpated within a few decades after Chestnut blight (Cryphonectria parasitica Murr. Barr) was first introduced in the United States in the early 1890s (Clark et al., 2012). Resorting to asexual reproduction, the species has been able to persist via remnant root systems in some locations (Clark et al., 2012). The lack of natural resistance coupled with the species shift in mating system to clonal reproduction has failed to provide C. dentata with any evolutionary benefit, resulting in the continued decline of populations. Since then, rehabilitation techniques have primarily focused on producing and planting blight-resistant trees through backcrossing breeding techniques with resistant Asian parental species, however, this technique poses its own threats to the genetic integrity of our native species (Anagnostakis, 2001; Burnham et al., 1986).

Since all M. acuminata populations in southern Ontario are small, each having far fewer than 100 individuals, they have the increased risk of suffering from increased rates of genetic deterioration, more specifically the loss of genetic diversity following drift, and environmental stochasticity. Though it is expected that smaller populations experience proportionately greater effects on genetic

18 structure (Barrett and Kohn, 1991; Frankham et al., 2002), fragmentation effects on trees can be delayed because they are often characterized by extensive gene flow prior to fragmentation, a long life span, and delayed maturity (Kramer et al., 2008;

Austerlitz et al., 2000). These characteristics can counterbalance genetic drift and reduce the effects of bottlenecks, while maintaining genetic variability and minimizing among-population differentiation (Anguilar et al., 2008; Wagner et al.,

2011; Hamrick, 2004). Therefore, current levels of gene flow may not be reflected in levels of genetic differentation because the time since fragmentation may have been insufficient for genetic divergence among populations, a pattern particularly seen in long-lived species (Aldrich et al., 1998). In order to identify contemporary patterns of gene flow, which can be dependant on forest structure, adult density and seed dispersal methods (Hamrick et al., 1993), seedling data may provide a more accurate estimate. Combined, multiple-stage cohort data can be useful when inferring the direction of future population genetic structure (Nason et al., 1997).

Similar to the expectations for small, isolated populations, species at their range periphery, like M. acuminata in Ontario, in comparison to those within or near central populations, are traditionally thought to maintain lower levels of genetic diversity (Lewontin, 1974; Mayr, 1970) and have relatively low viability

(MacArthur and Wilson, 1967; Goel and Richter-Dyn, 1974). However, more contemporary studies assessing the persistence and distribution of genetic variability across species distributions have found these patterns to be inconsistently supported (Lesica and Allendorf, 1995, Garner et al., 2004; Channel,

2004, Mandák et al., 2005, Wagner et al., 2011). Channel (2004) suggests that the maintenance of genetic diversity, as well as the rate of extinction, would be similar among peripheral and core populations if populations at the range edge were not consistently smaller than those within and/or near the central populations.

The purpose of this study was to assess the consequences of forest fragmentation on the genetic diversity and population genetic strucutre of M. acuminata in southern Ontario. Furthermore, to address the hypotheses regarding peripheral populations being more genetically depauperate, we also compared the genetic diversity of the Ontario M. acuminata populations at their northernmost range limits to those across the species’ core range in the United States. Using both chloroplast and nuclear microsatellites, I addressed several questions regarding the genetic viability of Ontario’s natural populations of Cucumber trees: Is genetic variation in M. acuminata populations comparable in Norfolk County and the

Municipality of Niagara? 2) Are M. acuminata populations in Norfolk County genetically isolated from those in the Municipality of Niagara? 3) Is the “core” population in the United States more genetically variable than the peripheral populations in Ontario? 4) Does the delayed effect of habitat fragmentation, characteristic of woody-plant life-history features, mean that levels of genetic diversity and differentiation vary between cohorts?

The results of this study will be important in the recovery efforts for M. acuminata in southern Ontario, leading to other critical avenues of genetic research which should provide a better understanding of the reproductive and dispersal barriers and their influences on the population genetic structure and the persistence of Magnolia acuminata in southern Ontario. The long-term vision of this project is to inform policy makers and landowners of the implications for future management strategies of M. acuminata, as well as to provide the foundations for the recovery programs of other endangered Carolinian forest species, promoting the use of integrative conservation techniques as tools for conserving the genetic variability of rare and/or species at risk in Ontario.

METHODS

Study sites and sample collection- Clustered within two disjunct regions separated by approximately 200 km, a total of 12 M. acuminata sites in southwestern Ontario were sampled between May and September, 2011 and 2012 (Figure 1). Of those, 7 sites are located in Norfolk County (Appendix I) and the remainder in the

Municipality of Niagara (Appendix I). These sites are fragmented remnants of what once was a continuous forest spanning the Appalachian Mountain range, with southern Ontario being its northernmost limit of distribution. Sample sizes ranged from 1- 40 trees; this represented approximately 93- 100% of the trees expected at each site according to the 2010 COSEWIC Assessment and Status Report (COSEWIC,

2010). A total of 187 leaf samples and 6 bark samples were collected (Figure 1). All bark samples were from a single site in the Niagara Municipality, (RW), collected in

2011. Leaf samples from RW could not be acquired due to an unanticipated change in land ownership. GPS coordinates of each sampled tree were recorded using a handheld GPS device (Garmin eTrex H). Both leaf and bark samples were stored in individual, sealed plastic bags containing silica beads for desiccation. Upon returning to the lab, samples were stored at -20°C. In addition, 65 leaf samples collected from 16 sites across the Appalachian range in the USA were graciously provided by the Smithsonian Institution of Washington, DC.

Figure 1: Distribution of collected M. acuminata samples from natural stands within Ontario and the United States.

DNA extraction and genotyping- Approximately 50-100 mg of either leaf or bark tissue from each sample was ground using a Retsch MM300 mixer mill (Haan,

Germany), and genomic DNA was extracted using E.Z.N.A Plant DNA Mini Kits

(Omega Bio-Tek, USA) with a final resuspended volume of 100 µl. DNA from bark samples was extracted following the same protocol as leaf tissue with the following modifications from Rachmayanti et al. (2006): 2.5-3.1% (w/v) Polyvinylpyrrolidone was added to the E.Z.N.A. P1 buffer; 800 µl of the mixed P1 buffer solution and 8 µl of RNase was added to each sample, vortexed, and incubated overnight at 65°C; samples were inverted 2-3 times during this incubation period; and following

23 incubation, 260 µl of E.Z.N.A. P2 buffer was added and samples were centrifuged for

5 minutes at 20 000 g. After being centrifuged, samples were then placed in a freezer and further incubated for 15 minutes at -20 °C; lastly, 700 µl of DNA wash buffer was used. All samples were genotyped at 5 nuclear microsatellite loci (Isagi et al., 1999; Table 1).

Table 1: Nuclear microsatellite primers, taken from Isagi et al. (1999), used to assess the population genetic structure of M. acuminata in Ontario and the USA. Annealing temperatures (Ta) and fluorescent labels are also provided. Annealing Temperature Locus Primer Sequence Label (Ta) F: 5’- ACATGGATAGTCGTTGGATA -3’ M6D3 HEX 47.8 R: 3’- ACCCCACTGAAGACAAACAT -5’ F: 5’- CACCGTACCCTATCAGAACC -3’ M6D4 FAM 47.8 R: 3’- ATTTTCAGCATCATCAGTTG -5’ F: 5’- GTCTAGTGAGCCGCAAATGG -3’ M10D3 FAM 58.3 R: 3’- GTGAACAGCTTTCTTGTGAA -5’ F: 5’- CGACGACGAAACTACTAACA -3’ M10D6 HEX 51.5 R: 3’- TTAAGTTGAGGTGGAATGAC -5’ F: 5’- TGCTGCTCGAAGTTCTGAAT -3’ M17D5 FAM 51.5 R: 3’- CGTGCATGAAATCAGGATGT -5’

PCR reactions for each locus included 1X PCR reaction buffer (UBI Life

Sciences Ltd., Canada), 0.2 mM dNTPs, 2.0 mM MgSO4 (UBI Life Sciences Ltd.,

Canada), 0.12 U BSA, 0.5 µM of each forward and reverse primers, 0.1 U HP taq polymerase (UBI Life Sciences Ltd., Canada) and 2 µl of DNA in a 10 µl reaction.

Amplifications were completed using a Mastercycler proS thermal cycler

(Eppendorf, Canada).

Loci M6D3 and M6D4 were amplified beginning with a 2 minute denaturation at 94°C, followed by 35 cycles of 15 seconds denaturation at 94°C, 15 seconds annealing (Ta in Table 1), and 30 seconds extension at 72°C. A further extension period for 1 minute at 72°C completed the amplification process ensuring uniform

24 strands. M10D3 was amplified following a 3 minute denaturation at 94°C, 35 cycles of denaturation for 45 seconds at 94°C, 45 seconds annealing (Ta Table 1), and 1 minute and 30 seconds of extension at 72°C. Further extension periods lasted for 10 minutes at 72°C and 45 minutes at 60°C. M10D6 and M17D5 began with 2 minute denaturation at 94°C, followed by 35 cycles of 15 seconds denaturation at 94°C, 15 seconds annealing at 51.5°C, and 30 seconds extension at 72°C. A final extension step for 10 minutes occurred at 72°C.

Following successful amplification, reactions were diluted to 1 µl in 20 µl of water and genotyped using a Hitachi 3730 DNA Analyzer (Applied Biosystems, USA) with ROX 350 (Applied Biosystems, USA) as a size standard. The submitted reaction comprised of 0.7 µl diluted PCR product and 10 µl of ROX350/HiDi solution (4 µl

ROX350 + 1 mL HiDi). Genemarker software v. 1.6 (SoftGenetics, USA) was then used to size the fragments.

In addition to characterizing nuclear genotypes which are biparentally inherited, we also genotyped chloroplast haplotypes. Since chloroplast haplotypes are maternally inherited in pure Magnolia, not including congeneric hybrids which have shown traces of paternal leakage (Sewell et al., 1993), M. acuminata haplotypes are transmitted only via seeds (Corriveau and Coleman, 1988). A comparison of nuclear and chloroplast population genetic data can therefore provide insight into dispersal patterns of seeds versus pollen. Minisatellite repeats (repeat motifs of

>10 bp) are among the most variable regions in chloroplast genomes (Cozzolino et

25 al., 2003; Vachon and Freeland, 2011), and I therefore targeted regions of chloroplast DNA that potentially contained minisatellite repeats. This was done in two ways. First, I downloaded the entire chloroplast genome sequence of Magnolia kwangsiensis from the National Centre for Biotechnology Information (NCBI) (Kuang et al., 2011), and used mreps software (Kolpakov et al., 2003) to identify minisatellite repetitive sequences within that sequence. Second, I downloaded NCBI sequences of chloroplast regions that had each been obtained from at least two

Magnolia species, and for each chloroplast region aligned available Magnolia sequences using ClustalX (Larkin et al., 2007). From these interspecific alignments, I identified indels that had resulted from variable numbers of minisatellite repeats.

After regions containing minisatellite repeats had been identified, I designed primers that would anneal to the flanking regions using Primer 3 (Rozen and

Skaletsky, 2000) (Table 2).

PCR reactions for each cpDNA region consisted of 1X PCR reaction buffer

(UBI Life Sciences Ltd., Canada), 0.2 mM dNTPs, 1.5 mM MgSO4 (UBI Life Sciences

Ltd., Canada), 0.12 U BSA, 0.5 µM of each primer, and 0.1 U HP taq polymerase (UBI

Life Sciences Ltd., Canada). Amplifications were completed in a Mastercycler proS thermal cycler (Eppendorf, Canada) which began with a 2 minute denaturation at

94°C followed by 35 cycles of 45 seconds denaturation at 94°C, 45 seconds annealing, and 45 seconds extension at 72°C. Further extension occurred for 10 minutes at 72°C and 45 minutes at 60°C. Genotyping and scoring were done as for nuclear microsatellites.

Table 2: Chloroplast minisatellite regions for which amplification primers were designed. GenBank accession numbers refer to the sequences used to identify flanking regions for primer design. Annealing temperatures (Ta) and fluorescent labels are also provided.

Accession Ta Name Region Primer Sequences Repeat Motif Label No. (°C) AY727149.1 F: 5’-AAAGCCGGGTGTGGCCCAA-3’ (M. acuminata) Mini2 trnC-ycf6 TGTATCCAGA FAM 55 AY727148.1 R:5’-CACCCTATTCCATTCTTGGCGGGG-3’ (M. tripetala) AY727291.1 F: 5’-TGATCGGAATCAATACAAATGG-3’ (M. acuminata) Mini3 ycf6-psbM CTTTATACAA HEX 55 AY727290 R: 5’- TGAGTGGATTGGAGTCAGCA-3’ (M. tripetala) AY727107 F: 5’-CGGTGTTCCATTGGTCACAGACCC- (M. acuminata) 3’ TATTTATTGA Mini4 psbM-trnD HEX 55 AY727110 TTTTGTT (M. tripetala) R: 5’-AGTTACCCCGACAGCGACAGAGT-3’ DQ813516 F: 5’-CGGGACATGGATATTCGAGA-3’ GTTTTACTACT (M. acuminata) Mini5 trnT-psbD TCTCCATTGTT FAM 58 DQ813515.1 R: 5’-GCGTGGTCCAAGGAAATAAA-3’ CCA (M. tripetala) AY727314.1 F: 5’-GCTACTATGACCTTCCCAACCACGA- (M. acuminata) 3’ GGTATAAAAT Mini8 rpS12-rpL20 FAM 55 AY727308.1 A (M. tripetala) R: 5’- AGGGAAGGGGCTCCGGTGTA- 3’

Genetic Analyses- M. acuminata is tetraploid (2n=76), with each locus having one to four alleles. To account for the uncertainties in allelic dosage when analysing tetraploids, when four alleles were not present at any given locus, the genoytpe was considered incomplete and the unknown alleles were assigned as missing data.

Because of these uncertainties, analyses were completed using both genotype- and phenotype- based approaches, the latter suggested to be least affected by ploidy

(Obbard et al., 2006). Genotypic data were converted into phenotypic data using a presence/absence matrix treating each allele as a marker. Present alleles were represented as 1 and absent alleles as 0.

To assess genetic diversity and the amount of inbreeding within clusters using genotypic data, the number of alleles (NA), number of effective alleles (NE), gene diversity (He), and inbreeding coefficients (FIS) were calculated from genotypic

27 data using SpaGeDi 1.3 and 1.4 (Hardy and Vekemans, 2002), which assumes polysomic inheritance. Significant P values for one-sided permutation tests were calculated based on 20 000 random permutations of individuals, genes, and alleles.

Although we sampled most, if not all, trees from each site, the number of individuals per site varied within and among regions. As a default to account for these differences, SpaGeDi provides gene diversity corrected for sample size. For the purposes of calculating genetic diversity in the United States, all samples were pooled due to the limited number of samples from each site location. While this does not provide a realistic representation of “population” genetic structure in the

US, it does however provide a conservation comparison for Ontario samples. Like the diversity measurements calculated for Ontario clusters, number of alleles (NA), number of effective alleles (NE), gene diversity (He), and inbreeding coefficients (FIS) were calculated for the US. Hickory 1.1 (Holsinger and Lewis, 2003) was used for the calculation of within- stand diversity (Hs) using phenotypic data.

Using genotypic data, the assessment of genetic differentiation, within and among clusters, was calculated with SpaGeDi 1.4 (Hardy and Vekemans, 2002) which uses Weir and Cockerham’s (1984) ANOVA-based approach to estimate overall and pairwise F-statistics (FST ). GENODIVE 2.0b23 (Meirmans and van

Tienderen, 2004) was used to calculate Jost’s D (Dest) (Jost, 2008), an analogue of

Weir and Cockerham’s (1984) FST. This measurement uses effective number of alleles to estimate genetic differentiation as a proportion of allelic variation, which is assumed to increase linearly with an increase in equally frequent alleles (Jost, 2008;

Meirmans and Hedrick, 2011). Hickory 1.1 (Holsinger and Lewis, 2003) provided an unbiased theta (θII) estimate using phenotypic data. This measurement is directly comparable to Weir and Cockerham’s FST , which was calculated using MCMC methods under the f-free model according to the default parameters (burn-in=5000, sample=100 000, thin=20) (Holsinger and Lewis, 2003).

Spatial Analyses- We used two different Bayesian clustering methods to detect genetic structuring of M. acuminata across Ontario:

STRUCTURE 2.3.4 (Pritchard et al. 2000) uses a Markov chain Monte Carlo

(MCMC) method to simulate genotypes and predict the posterior probabilities that the genotypic data fits into a number of predefined clusters (K). The optimal number of K can be inferred according to Evanno et al. (2005) based on the posterior probabilities of each value of K (lnP(K)), as well as through the change in model likelihoods between the successive values of K (∆K). Using genotypically scored microsatellite data, we ran an admixture model with correlated allele frequencies for K values ranging from 1-10. 10 runs per K were carried out with a simulation and burn-in length of 300,000 and 200,000, respectively.

TESS 2.3 (Durand et al. 2009) is a spatial model which utilizes individual multilocus genotypes and the geographic location of each individual to determine the number of distinct genetic clusters. The use of sample locations in this model is used to identify an individual’s cell, which is needed to define the individual neighbourhood network. This method corresponds with the Potts statistical model where the state of each individual is inﬂuenced only by the states of its neighbours

(Francois and Durand, 2010). Bayesian hierarchical admixture models, the conditional autoregressive (CAR) and Besag-York-Moille (BYM) conditional autoregressive models, were used to develop spatial neighbourhood networks by applying Voronoi tesselation (Durand et al. 2009). Both models can be mathematically defined by a Gaussian Markov random field, in which individuals are independent of each other and are defined by individual spatial networks (Durand et al. 2009). Based on these models, we can assume that individual genomes are spatially autocorrelated and arise from the admixture of Kmax unobserved parental populations or genetic clusters (Durand et al., 2009). Therefore, under these assumptions, spatially proximal individuals or neighbours will be more genetically similar than those which are more distant.

Prior to running our admixture models using TESS, a no-admixture model was run to determine the potential maximum number of clusters or Kmax (Durand et al. 2009). K values were set from 2 to 20, with 10 runs per K, and the total number of sweeps and burn-in length were 30,000 and 10,000, respectively. The K values used in both admixture models, BYM and CAR, were then simulated based on the no- admixture results.

K value ranges for both models were set from 2 to 10, while the number of runs per K as well as total simulation- and burn-in lengths varied. The CAR method was simulated twice having 10 and 20 runs per K and total simulation- and burn-in lengths of 300,000 and 200,000, respectively. The BYM model was applied to our dataset three times. The first trial had 20 runs per K and total simulation- and burn-

30 in lengths of 30,000 and 10,000, respectively. The second and third simulations had

10 and 20 runs per K with total simulation- and burn-in lengths of 300,000 and

200,000. We averaged the DIC values for each run and plotted them against Kmax.

The point at which the DIC curve exhibits a plateau indicates the most likely number of clusters (Durand et al., 2009). The assignment probabilities and hard clustering of individuals for the inferred number of clusters was obtained from the “best” run which corresponds to the Kmax (K=) run with the lowest DIC value.

Identification of discrete populations based on sampling location is problematic, because the distribution of potential mates has likely varied over time, therefore, I used genetic clusters, as identified by STRUCTURE and TESS, as a surrogate for populations in our analyses of genetic diversity and differentiation.

Isolation by distance (IBD) (Slatkin, 1987) within Ontario and Norfolk

County, was assessed using a Mantel test (20,000 randomizations) (Mantel, 1967), implemented by the program IBD 1.52 (available online at http://www.bio.sdsu.edu/pub/andy/IBD.html) (Bohonak, 2002), based on untransformed and log-transformed data. The strength of the relationship was determined using a reduced major axis (RMA) regression implemented by IBD 1.52.

RMA, unlike the traditional least squares regression, takes into account sampling and measurement errors for both the dependent (Y-axis) and independent (X-axis) axes. Confidence limits for the RMA slope were generated by jackkniffing over populations, due to the small population sizes. We also applied a Mantel test (20,

000 randomizations) to matrices of genetic and geographic distances based on

31 phenotypic data. Genetic distances were calculated using GENALEX 6.3 based on a

Euclidean distance metric according to Huff et al., (1993). A spatial autocorrelation analysis, based on phenotypic data, was also completed using GENALEX 6.3. To assess the scale at which spatial autocorrelation was relevant, we varied the size of linear geographical distances between 5 km and 120 km in increments of 5km.

Statistical significance was tested using 9999 random permutations.

Temporal variation- 65 canopy trees, those which contributed to the forest overstory and reached a minimum height of 20 m and 63 seedlings, those which did not exceed a height of 1 m, were used to assess the differences in genetic diversity and differentiation within each of the two age classes for each Norfolk County and the Municipality of Niagara based on genotypic data. To account for differences in the number of sample sites represented in either cohort, genetic diversity and differentiation was also compared within two sites from either region: NWA from

Norfolk County and FD from the Municipality of Niagara, based on both genotypic and phenotypic data.

RESULTS

Nuclear Microsatellites

Part I: Regional Variation

Genetic diversity- All five of the loci analyzed were polymorphic, with the number of alleles per locus ranging from 12 to 22. The total number of alleles across all loci and populations, including the United States, was 94. Within Ontario M. acuminata, approximately 70.4% of the total number of alleles were found in both Norfolk and

Niagara, with percent per locus ranging from 53.3% (M17D5) to 91.7% (M6D3).

However, twenty-one alleles were found only in Norfolk and 3 only in Niagara. The frequencies of the alleles private to Norfolk County predominantly ranged from

0.172-2.73%, with the exception of 5 alleles: 220 (M10D3), 152 (M6D10), 170

(M6D10), 118 (M6D3), and 218 (M10D3), whose frequencies were 6.78%, 7.41%,

10.8%, 11.7%, and 23.7%, respectively. Those alleles which were private to the

Municipality of Niagara had frequencies ranging from 0.93-2.6%. Overall, nineteen of the 24 private alleles occurred at frequencies <3%. Seventy-one alleles (75.5% of the total number of alleles) were observed in both Ontario and the USA, with the percent of shared alleles per locus ranging from approximately 66.7% (M6D10 and

M17D5) to 90.5% (M10D3). Thirteen alleles were observed only within the US and

10 only in Ontario, with private allele frequencies ranging from 0.137-4.41%.

Twenty-two of the 23 alleles that were private to either the USA or Ontario were found at frequencies <3%. Magnolia acuminata from the USA shared a higher percentage of alleles with Norfolk County (74.2%) than with the Municipality of

Niagara (65.5%) (See Appendices III-A-D).

Table 3 provides a comparison of genetic diversity estimates across all loci and populations. At the national level, the effective number of alleles in Ontario and

USA were 8.2 and 11.0, respectively. Regionally, the effective number of alleles per locus was 7.8 in Norfolk County and 5.9 in the Municipality of Niagara, ranging from

4.0-7.3 within each of the seven genetic clusters. Gene diversity (He), was significantly higher in the USA (0.899) than Ontario (0.866), and similarly in Norfolk

(0.850) compared to Niagara (0.828). He values ranged from 0.714- 0.852 within the

Norfolk clusters, and from 0.754-0.861 within Niagara clusters. Inbreeding coeffiecients were negative at all loci indicating an excess of heterozygotes, or fixed heterozygosity, within each cluster and region (Table 3).

Table 3: Genetic variability of five microsatellite loci averaged across all loci and populations of M. acuminata using genotypic (SpaGeDi 1.4) and phenotypic (Hickory 1.1) data. Genotypic Data Phenotypic Data LOCATION CLUSTER NA NAe He FIS HS 1 12.2 6.8 0.852 -0.164 0.168 2 6.8 4.0 0.719 -0.359 0.132 3 9.4 6.6 0.841 -0.185 0.177 NORFOLK 4 10.0 5.7 0.767 -0.282 0.152 5 5.0 4.0 0.714 -0.395 0.138 COMBINED 15.6 7.8 0.850 -0.157 0.154 1 10.0 7.3 0.861 -0.165 0.237 NIAGARA 2 9.4 4.3 0.754 -0.301 0.174 COMBINED 12.0 5.9 0.828 -0.184 0.206 ONTARIO COMBINED 16.2 8.2 0.866 -0.139 0.148 USA COMBINED 17.0 11.0 0.899 -0.109 0.200 A= mean number of alleles; AE= effective number of alleles; He= gene diversity, corrected for sample size; FIS= inbreeding coefficient; HS= within-population genetic diversity

Phenotypically scored data yielded 82 polymorphic markers in Ontario; 78 in

Norfolk County, and 61 in the Municipality of Niagara. Within-population genetic diversity (Hs) values revealed a similar trend to genotypic data, however, values were consistently and substantially lower, ranging from 0.132- 0.237. However,

contrary to genotypic measures of gene diversity (He), the Municipality of Niagara was found to have greater overall genetic diversity (Hs) than Norfolk County.

Genetic Differentiation- Global FST estimates suggest that M. acuminata in Ontario is characterized by low to moderate levels of differentiation within and among regions

(Table 4). Global FST estimates were fairly low, ranging from 0.0102-0.2312 for each locus. Overall genetic differentiation within each of the two regions was highly significant (p< 0.0001; one-sided P values based on permutations on locations, individuals, and genes) for all loci with the exception of M10D3 among the

Municipality of Niagara clusters (p= 0.08). Jost’s measurement of differentiation

(Dest) produced a similar trend as global FST estimates, but most values were substantially higher, ranging from -0.0290- 0.7540 (p< 0.0001, with the exception of

M6D4 (p= 0.004) and M10D3 (p= 0.447) among the Municipality of Niagara clusters)

(Table 4).

Table 4: Summary of Global F statistics (FST) and Jost D (D_est) among the 7 identified clusters (k=7) in Ontario, as well as those identified within Norfolk County (k=5) and the Municipality of Niagara (K=2). All FST and D_est estimates were highly significant (p<0.0001) for all loci and regions with the exception of M6D4 (D_est : p=0.004) and M10D3 (FST : p=0.08; D_est : p= 0.447) in the Municipality of Niagara.

ONTARIO NORFOLK NIAGARA LOCUS FST D_est FST D_est FST D_est M6D3 0.0865 0.5240 0.0830 0.5680 0.0710 0.3670 M6D4 0.0444 0.3330 0.0380 0.2700 0.0436 0.3280 M6D10 0.0759 0.4910 0.0720 0.5270 0.1041 0.6670 M10D3 0.0711 0.5360 0.0669 0.5340 0.0102 -0.0290 M17D5 0.1716 0.4210 0.1233 0.2360 0.2312 0.7540 ALL LOCI 0.0886 0.4600 0.0755 0.4050 0.0932 0.4800

Phenotypic results (θII) reveal greater amounts of among-population differentiation than those detected using genotypic data when measured as FST, but not when measured as Dest. Based on phenotypically scored data, Norfolk clusters overall are more differentiated from one another [0.222 (95% credible interval

0.175-0.271)] than those within the Municipality of Niagara [(0.195±0.037 (95% credible interval 0.125-0.272)]. Overall, levels of differentiation among Ontario clusters based on phenotypic data are moderate [0.219 (95% credible interval

0.180-0.260)].

Analysis of molecular variance (AMOVA) based on phenotypically scored data revealed that 73.7% (p= 0.001) of the genetic variation in Ontario occurs within clusters, 19.9% (p= 0.001) among clusters, and only 6.3% (p= 0.001) among regions.

Similarly, when Norfolk County and the Municipality of Niagara were analyzed independently, each expressed greater variation within than among clusters (Table

5).

Table 5: Analysis of molecular variance calculated using phenotypic data for Ontario, Norfolk and Niagara. Where ** indicates a P-value of 0.001 .

df SS MS Est. Var. % Ontario

Among Regions 1 94.091 94.091 0.608 6.3** Among clusters 5 282.823 56.565 1.912 19.9** Within clusters 186 1314.070 7.065 7.065 73.7** Total 192 1690.984 9.584 100.0

Norfolk Among clusters 4 243.913 60.978 1.920 21.2 ** Within clusters 149 1065.801 7.153 7.153 78.8** Total 153 1309.714 9.073 100.0 Niagara

Among clusters 1 38.910 38.910 1.858 21.7** Within clusters 37 248.269 6.710 6.710 78.3** Total 38 287.179 8.568 100.0

Table 6: Pairwise FST (below; SpaGeDi) and Dest (above; GENODIVE) between the inferred clusters of M. acuminata in Ontario. D_est NOR-1 NOR-2 NOR-3 NOR-4 NOR-5 NIA-1 NIA-2 NOR-1 0.4568 0.2598 0.2545 0.3903 0.3038 0.4599 NOR-2 0.1006 0.4215 0.4136 0.4115 0.4181 0.6712 NOR-3 0.0460 0.1017 0.3579 0.4868 0.3333 0.4484 D_est FST NOR-4 0.0547 0.1184 0.0799 0.2651 0.3664 0.6394 NOR-5 0.0902 0.1437 0.1172 0.0865 0.4281 0.7916 NIA-1 0.0513 0.1017 0.0600 0.0818 0.1058 0.4103 NIA-2 0.0952 0.1832 0.1004 0.1584 0.2139 0.0932 FST Spatial Structure- Bayesian cluster analysis conducted using STRUCTURE revealed that the most likely number of genetic clusters in Ontario was 7 (K=7), following the method of Evanno et al. (2005) (Figure 2). Five of the genetic clusters occur within

Norfolk County and the remaining two clusters in the Municipality of Niagara

(Figure 2). Four of the seven sampled sites in Norfolk County were defined as their own cluster, while three sites were combined into one genetic cluster. Of the two clusters in the Municipality of Niagara, one comprised trees from a single site, and the other comprised trees from four sites. a)

Figure 2: ∆K results based on STRUCTURE analysis. ∆K calculated by STRUCTURE HARVESTER and according to Evanno et al. (2005) for K values 1-10.

37 a) b)

Figure 3: TESS clustering analysis of M. acuminata sampled across 12 sites in Ontario. Clustering was completed using CAR (a) and BYM (b) models for Kmax values ranging from 2 to 10. DIC values were computed and averaged across 20 runs per K. The plateau starting at Kmax=7 indicates that the most likely number of genetic clusters in Ontario is 7.

The results obtained from both models in TESS, CAR and BYM, are comparable to those achieved using STRUCTURE, as they each estimated seven genetic clusters using the DIC criterion (Figure 3 a&b.). Although the genetic clustering results are roughly consistent between the two programs, with respect to numbers and approximate distributions of clusters, the TESS results provide the clearest evidence of genetic subdivision among M. acuminata sites in Ontario (Figure

4 a&b). Both programs consistently identified individuals within sites CT (Cluster

2), SW (Cluster 3), NWA (Cluster 4), and FD (Cluster 7) as genetic clusters, however, the degree of admixture among the populations was generally stronger in

STRUCTURE. Unlike the STRUCTURE results, TESS results provided a more clear representation of the genetic groupings across Ontario: sites BT, ST, and VD were clustered together (Cluster 1); WATT was identified as a distinct population (Cluster

5), rather than being clustered with some of the Niagara sites as suggested by

STRUCTURE; and with the exception of FD, all Niagara sites were clustered together

(Cluster 6).

Figure 4: Distribution of M. acuminata clusters in Norfolk County based on cluster analysis results.

Mantel tests, using genotypic data, failed to show a significant relationship between either untransformed spatial and genetic distances among all clusters in

Ontario [FST: r= 0.1754, R2= 0.0308, p= 0.2808; Dest: r= 0.3206, R2= 0.1030, p=

0.1426 ] nor among Norfolk clusters [FST: r= -0.3128, R2= 0.0978, p= 0.7036; Dest: r=

-0.4748, R2= 0.2250, p= 0.8816] based on FST and Dest meaurements. Similarly, mantel tests also failed to show a significant relationship based on log-transformed distances among all clusters in Ontario (FST: r= 0.0735 , R2= 5.40e-03, p= 0.3059; Dest: r= 0.1761, R2= 3.10e-02, p= 0.1806) and Norfolk County (FST: r= -0.2812, R2= 0.0790, p= 0.6703 ; Dest: r= -0.4638 ,R2= 0.2150, p= 0.8928). Based on phenotypic data,

39 significant isolation by distance was detected across Ontario clusters using both untransformed (r= 0.5337, R2= 0.2850, p= 0.0130) and log-transformed distances

(r= 0.4323, R2= 0.1870, p= 0.0208), however, not among Norfolk clusters based on either untransformed (r= -0.0082, R2= 6.771e-05, p= 0.4272) nor log-transformed distances (r= 0.0526, R2= 2.77e-03, p= 0.4424). The strength of the relationship across Ontario was moderate; according to the RMA regression, geographic distance accounted for approximately 29% of the genetic variance across all Ontario clusters.

The spatial autocorrelation analysis based on phenotypic data indicated a strong positive correlation between genetic and geographic distance up to a geographic distance of 15 km (r= 0.031, p= 0.002), and strong negative correlations between the distances of 25-40 km (r= -0.034, p= 0.000), and 80-100 km (r= -0.103, p= 0.000).

Figure 5: Distribution of M. acuminata clusters in the Municipality of Niagara based on cluster analysis results.

40 a)

Figure 6: Posterior estimates of assignment probabilities for a) STRUCTURE and b) TESS BYM model, where Kmax= 7. Site and associated cluster numbers are defined in Table 7.

Table 7: Ontario M. acuminata sample sites as defined by geographic location and cluster. Cluster assignment is based on STRUCTURE and TESS BYM model results for Kmax=7. Site No. Sample Location Site No. of Samples Ontario Cluster Ontario Cluster Regional ID (STRUCTURE) (TESS) Cluster ID 1 BT 9 1 1 NOR-1 2 CT 17 2 2 NOR-2 3 ST 40 3 1 NOR-1 4 NOR SW 24 1 3 NOR-3 5 VD 5 1 1 NOR-1 6 NWA 51 4 4 NOR-4 7 WATT 8 5 5 NOR-5 8 SA 4 6 6 NIA-1 9 PL 1 6 6 NIA-1 10 NIA FD 26 7 7 NIA-2 11 RW 6 6 6 NIA-1 12 WD 2 4 6 NIA-1

CpDNA

Haplotype diversity- Based on the analysis of five chloroplast regions, we found no variability in M. acuminata haplotypes across Ontario and the USA (Table

8), despite characterizing haplotypes on the basis of the cpDNA regions that were specifically targeted for their potentially high variability.

Table 8: Allele sizes for all five chloroplast regions identified among all individuals from Ontario and US populations of M. acuminata. Mini 2 Mini3 Mini4 Mini5 Mini8 234 212 464 234 212

Part II: Variation within age cohorts

Table 9: Genetic diversity of canopy trees and seedlings of Ontario M. acuminata using genotypic (SpaGeDi) and phenotypic data (Hickory 1.1) from five microsatellite loci. Values averaged across all loci and individuals. Genotypic Phenotypic N # of sites A Ae He FIS Hs Norfolk Canopy 51 12 13 8 0.8570 -0.152 0.1829 Niagara Canopy 14 5 10 7 0.8641 -0.162 0.1835 Norfolk 44 3 11 6 0.7829 -0.239 0.1630 Seedlings Niagara 19 1 8 4 0.7321 -0.339 0.1626 Seedlings

N= sample size, A= mean number of alleles; AE= effective number of alleles; he= gene diversity, corrected for sample size; FIS= inbreeding coefficient, HS= within-population genetic diversity

Genetic diversity and differentiation- The number of M. acuminata seedlings in

Ontario is relatively low, compared to the number of adult trees. Genetic diversity was lower in Ontario seedlings compared to canopy trees, based on both genotypic and phenotypic data (Table 9). Furthermore, both genotypic and phenotypic measurements of genetic differentiation among regions revealed an increase in mean genetic differentiation between seedlings (FST= 0.1617; θII= 0.2566±0.0367

[0.1918-0.3342]) compared to the mean differentiation among clusters of canopy trees (FST= 0.0333; θII=0.1228±0.0267 [0.0766-0.1805]). Measures of diversity based on genotypic diversity were higher in the adult cohort within FD, compared to

NWA which exhibited slightly higher diversity within the seedling cohort (Table 10).

On the contrary, based on phenotypic data genetic diversity was higher in the adult cohorts compared to the seedling cohorts within both sites. While within- population differentiation based on genotypic data was found to be higher among

FD cohorts (pairwise FST= 0.0348) compared to NWA cohorts (pairwise FST=

0.0073), phenotypic estimates revealed within-population differentiation among

NWA and FD cohorts were both fairly low and comparable, 0.0038±0.0027 [0.0003-

0.0103] and 0.0032±0.0029 [0.0002-0.0104], respectively.

Table 10: Genetic diversity of canopy trees and seedlings of Ontario M. acuminata within sites NWA and FD using genotypic (SpaGeDi) and phenotypic (Hickory 1.1) data. Estimates based on genotypic data are averaged across all loci and individuals. Genotypic Phenotypic N A Ae He FIS Hs NWA Canopy 10 6.4 5.3 0.7580 -0.3140 0.2261 NWA Seedlings 41 9.6 5.8 0.7686 -0.2780 0.1650 FD Canopy 4 6 7.4 0.8848 -0.1310 0.4042 FD Seedlings 19 8 3.9 0.7321 -0.3370 0.1586

N= sample size, A= mean number of alleles; AE= effective number of alleles; he= gene diversity, corrected for sample size; FIS= inbreeding coefficient

DISCUSSION

Magnolia acuminata across North America: patterns of recolonization following glaciation

The uniformity of M. acuminata cpDNA haplotypes across Ontario and the

United States, consistent with the US cpDNA results produced by Wollaeger (2011), as well as the high proportion of shared nuclear microsatellite alleles between

Ontario and the USA, suggests that southern Ontario was recolonized by M. acuminata from a large, single southern refugium, which is consistent with the leading-edge hypothesis. The leading-edge model is typically characteristic of rapid, large-scale geographic expansions, and has been exhibited by a variety of taxa following the last glaciation, including Quercus ilex (Hampe et al., 2013), Poecile rufescens (Burg et al., 2006), and mustelids (Ibrahim et al. 1996). Alternatively, patterns of genetic diversity and differentiation, as discussed below, suggest that M. acuminata may have expanded into southern Ontario following the phalanx model

(Hewitt, 1996). Under the phalanx model, new populations formed at the expansion front are represented by a more even mixture of ancestral populations, or genotypes. Because the expanding populations consist of similar groups of genotypes, genetic differentiation can be as low as those characteristic of the leading-edge model, however genetic diversity is expected to be higher. The phalanx model has been described for populations with high densities, high gene flow, wide colonization fronts, and slow expansions (Hewitt, 1996). While the slow or gradual expansion by M. acuminata out of refugia may be the consequence of interglacial cooling-warming cycles, the cpDNA results suggest that expansion into southern

Ontario may have been facilitated through rare long-distance dispersal events at the onset of recolonization. Thus, to understand the congruent cpDNA pattern that was identified across the species range, it is important to consider all plausible methods of range expansion and seed dispersal mechanisms.

As previously mentioned, M. acuminata seeds have evolved for long-distance dispersal via water, a product of its natural riparian habitat (Beresford-Kroeger,

2003). Thus, historically, long-distance seed dispersal events by M. acuminata into the Lake Erie region could have been achieved via waterways, particularly after the

Wisconsinian glaciation. Following the last glacial retreat, approximately between

14,400-12,400 years ago, various high/low-water stages occurred in the Lake Erie

Basin (Herdendorf, 2013). These events resulted in the formation and collapse of proglacial lakes, shifting drainage patterns across the Niagara Escarpment, a consequence of isostatic-depression (Larson and Schaetz, 2001; Herdendorf, 2013).

In addition, ice damming the St. Lawrence River System and the Atlantic Ocean began to break, leading to a large influx of sea water from the Atlantic, moving into and out from the Great Lakes Region via the St. Lawrence Seaway and its outlets

(MacClintock and Stewart, 1965; Winslow, 2008). Over the next couple of thousand years, these regions entered a period of glacio-isostatic rebound, effectively causing the flow of fresh water from rivers and streams to push the more dense sea water northward, reversing flow (Winslow, 2008; Herdendorf, 2013). Therefore, surface water from glacial meltwater, coupled with the repeated occurrences of isostatic events leading to the alteration of flow patterns, could have resulted in the dispersal

45 of seeds from a southern refugium into the Great Lakes region, particularly within the Lake Erie Basin from what is now Ohio. More contemporarily, New, Ohio and

Mississippi Rivers provide key waterways for seed dispersal across the United

States, as well as the Welland and Niagara Rivers for dispersal of seeds into and throughout southern Ontario, similar to the dispersal routes of pawpaw proposed by

Keener and Kuhns (1997).

In conjunction with long-distance dispersal events, short-distance dispersal events, including movement via fauna, gravity, and water dispersal within low-flow systems, also contributed to the genetic structuring of M. acuminata in Ontario following range expansion and recolonization in the area. Magnolia acuminata seeds are extremely high in protein and, when available (October), could provide both mammals and birds with much needed food and energy, particularly for those which are migrating and hibernating. While the mutualistic relationship between fleshy-fruits and birds in North America has been widely studied, the relationship between mammals and fruits has been underestimated (Wilson, 1993).

Nevertheless, both should be considered likely methods of seed dispersal as, in addition to the number of fruit-eating birds, ungulates and small rodents, there are also four families among the nine ‘Carnivora’ orders in North America containing fruit-eating species found historically and/or presently within the range of M. acuminata: Ursidae, Procyonidae, Mustelidae, and Canidae. Evidence of mammal consumption has been previously observed in two of the largest seeds in North

America, Asimina spp. and Diospyros spp. These seeds have been known to be

46 consumed, excreted and rendered germinable by Raccoon (Procyon rotor), Red fox

(Vulpes vulpes) (Wilson, 1993; Murphy, 2001), and Opossum (Didelphis virginiana)

(Worth, 1975; Murphy, 2001). Black bear (Ursus americanus), Coyote (Canis latran),

Striped skunk (Mephitis mephitis), White-tailed deer (Odocoileus virginianus), and

Eastern gray squirrel (Sciurus carolinensis) (Murphy, 2001) also could have facilitated both historical, as well as contemporary, seed movement of M. acuminata into and within Ontario.

In addition, the seeds’ colouration, scent, and fleshiness are considered to be advertisements to known avian seed-dispersers, including Cedar waxwing

(Bombycilla cedrorum), Gray catbird (Dumetella carolinensis), Swainson’s thrush

(Catharus ustulatus), and Hermit thrush (Catharus guttatus) (Wilson et al. 1990), and Wild turkey (Meleagris gallopavo) (Cleland, 1966). Passenger Pigeon (Ectopistes migratorius) could have facilitated seed movement, both long and short distances, as they were prolific across southern Ontario during prehistoric and early historic times. These species likely contributed to frequent short-distance dispersal events within and between populations, in addition to dispersal from the maternal parent via gravity and flotation, accounting for the prevalence of a single haplotype across the species range.

Magnolia acuminata in Ontario: historic and contemporary gene flow

Despite substantial habitat fragmentation, genetic diversity in Ontario M. acuminata remains surprisingly high. In fact, estimates are not much lower than the

47 genetic diversity of M. acuminata compiled from numerous USA populations covering a much broader geographical range than the populations sampled from

Ontario. Since phenotypic analyses may be more appropriate for polyploid species

(Obbard et al., 2006), contrary to genotypic results which reveal greater amounts of diversity among clusters in Norfolk County than those in the Municipality of Niagara, phenotypic measurements of diversity (HS) reveal that the Municipality of Niagara, the northernmost region, has greater amounts of genetic diversity (0.206) than both

Norfolk County (0.154) and the United States (0.200), a pattern that contradicts the expectations of the leading-edge model. The discrepancy between the two data types may be due to the uncertainty of allele frequencies in genotypic estimates, the exclusion of monomorphic markers in phenotypic analyses, as well as the differences in sample sizes. However, sample sites in the Municipality of Niagara not only had fewer trees per site than those in Norfolk County, but also had a substantially smaller dataset overall. Furthermore, phenotypic analyses of the

Municipality of Niagara populations contained fewer polymorphic markers, so the occurrence or frequencies of rare alleles would not be contributing to this region’s gene diversity. If genetic diversity can be augmented by gene flow and the amount of gene flow is often influenced by patterns of pollen and seed dispersal, then higher genetic diversity in the Municipality in Niagara could be attributed to higher rates of gene flow between sample sites. Unlike the distance between most of the sites in

Norfolk County, all of the sampled sites in the Municipality of Niagara were no further than a few kilometers from one another. Coupled with low density at these sites, these features theoretically should promote gene flow via interplant

48 movement and outcrossing events, as well as potentially reduce the risks of self- pollination and pollination of closely related individuals (Eglund, 1993). Thus, it is likely through the maintenance of genetic connectivity between sites in Niagara that genetic diversity has been retained.

The measures of genetic variability in each region presented in this study more closely reflect colonization following the phalanx model. Under the phalanx model, populations are capable of retaining higher levels of genetic diversity because significant population bottlenecks occur less frequently, allowing many different genotypes to colonize available sites over shorter distances (Hewitt, 1996).

The gradual and continuous expansion of the range of a species produces populations with little genetic structuring, with the most recently founded populations being most similar to source populations (Hewitt, 1996). However, if M. acuminata did follow leading-edge colonization exclusively, then the high levels of diversity may be partially explained by initial colonizers acting as sinks, or by the suspected high rates of pollen flow which would have provided additional diversity in the gene pool as M. acuminata continued to expand across Ontario (Austerlitz and

Garnier-Gere, 2003). Furthermore, Hampe et al. (2013) found that leading-edge populations of Quercus ilex could rapidly restore levels of genetic diversity following recolonization through the interacting effects of long-distance pollen flow and the purging of inbred individuals during recruitment. Nonetheless, while it is possible that M. acuminata did expand following the leading-edge and through historical connectivity and the prevention of self-fertilization by its protogynous system, I

49 speculate that Magnolia acuminata underwent stratified northern dispersal characterized by both gradual diffusion and occasional long-distance dispersal by waterways and active vectors accounting for both the uniformity in cpDNA and relatively high measures of genetic diversity, as discussed below.

The comparable levels of genetic diversity seen across the species’ range may be attributed to a variety of factors, including historical levels of gene flow, as mentioned above, as well as genetic and life-history characteristics such as polyploidy, longevity, and protogyny. The pattern of genetic diversity presented in our results is not unusual for long-lived tree species growing in forest fragments which have formerly experienced much higher levels of connectivity (Rosas et al.,

2011; Burgos-Hernández et al., 2013). This would be especially true for a tetraploid species since the population size of its nuclear genes is higher than that of a diploid, and hence, its alleles will be lost at a relatively slow rate following genetic drift

(Haldane 1930; Moody et al., 1993; Obbard et al., 2006).

In Ontario, Magnolia acuminata lives on average around 80 years (Ambrose and Kirk, 2010, Appendix B). Of the natural stands that I sampled in this study (~89

% of sampled sites), and excluding the seedlings that were planted at the National

Wildlife Area, most of the trees are from older generations. Therefore, current patterns of genetic diversity could reflect previously high rates of pollen flow.

Furthermore, the long generation time of M. acuminata may also have acted as an intrinsic buffer against rapid loss of genetic diversity. As new individuals and or

50 cohorts had established in southern Ontario, those surviving until maturity may have potentially harbored different alleles or genotypes. Therefore, as individuals and cohorts continued to establish and mature, new alleles were subsequently introduced into and/or maintained within the gene pool, thus retaining or increasing genetic diversity within populations.

Former genetic connections among currently fragmented populations, as well as the dispersal capabilities of M. acuminata, are also suggested by the low occurrences and frequencies of alleles private to either Norfolk or Niagara, as well as the high proportion of shared alleles between each Ontario region. The aforementioned genetic patterns of diversity among Ontario M. acuminata further supports the phalanx model, as allelic variation is expected to decrease progressively as colonization proceeds northward. When comparing the diversity measurements across M. acuminata distribution, both the average number of alleles and effective number of alleles was the lowest in the Municipality of Niagara, intermediate in Norfolk County, and highest in the United States, although these measurements could be confounded by the scale of sampling. Taking this into consideration, it is possible that diversity in Ontario may potentially be as high as, or even higher than American M. acuminata since the United States dataset comprised samples from 16 locations across eastern and central United States. Even though the sample size from each location on average was only 4-5 individuals, the breadth of the dataset could be capturing a much larger proportion of alleles than those represented in the much smaller regions in Ontario. Contrastingly, the limited

51 number of sample sites within the Municipality of Niagara could be underestimating diversity measurements.

Population dynamics of M. acuminata may also be comparable across its range and account for the similarities in genetic diversity. It is highly unexpected that even in its most northernmost, peripheral populations, genetic diversity has been maintained. Parisod and Bonvin (2008), Herrera and Barzaga (2008), and Raffl et al. (2006), have all reported that across a mountainous landscape, marginal populations are capable of having similar, and potentially higher amounts of genetic diversity than core populations, due to a species’ dispersal abilities. Therefore, according to the expectations of population dynamics outlined by Channel (2004) and the aforementioned studies, it may be that the patchy distribution and small population sizes of M. acuminata (Smith, 1990) have enabled genetic connectivity by promoting pollen flow between clusters and regions, based on theoretical expectations on the relationship between gene flow and population density.

However, that is not to say that the occurrences of patches, or geographic connectivity, had not substantially declined with historical fragmentation.

Nonetheless, there is no evidence suggesting an overall reduction in genetic variability following a latitudinal cline, in range edge (northerly) versus central populations.

Within Ontario, seven genetic clusters of M. acuminata were identified: five in

Norfolk County and two in the Municipality of Niagara. While some of the clustering results may not be particularly clear in STRUCTURE, by utilizing both STRUCTURE

52 and TESS, we were able to infer the overall general pattern of population structure across Ontario, with some evidence of long distance gene flow, represented as admixture, between the two regions. However, the identified clusters do not entirely correspond to geographic locations as in some cases, trees from multiple sites were clustered together (i.e. Norfolk Cluster 1, Niagara Cluster 2). Although M. acuminata populations in Ontario are subdivided into seven genetic clusters, overall genetic differentiation remains fairly low according to FST values, but it is considerably higher based on either Jost’s D (mean= 0.4600) or phenotypic data

(mean= 0.2190). As noted in the results, FST values may be artificially reduced by high levels of heterozygosity, regardless of whether the same alleles are present at high frequencies in different populations (Obbard et al., 2006; Kisel et al., 2012).

Nevertheless, all of our analyses (Structure, TESS, FST, Dest, and phenotypic analyses) support a clear level of genetic distinction among the seven clusters in Ontario and although immature Cucumber trees may be expected to show higher levels of genetic differentiation than mature trees, as a product of contemporary forest fragmentation, this does not entirely explain the patterns found because four of the clusters comprised only mature trees. Thus, the clustering analyses suggest that current levels of gene flow in this species may in fact be largely constrained to relatively short distances, potentially leading to increased genetic isolation and the formation of distinct populations.

Low regional genetic differentiation, as presented in this study, is most commonly explained by dispersal capabilities, more specifically long-distance dispersal events by pollen and/or seed, which are generally indicative of

53 unrestricted gene flow prior to fragmentation. This pattern is particularly true among long-lived, outcrossing species, which often retain most of their diversity within populations (Hamrick and Godt, 1989), rather than among them. This pattern is evident in Ontario M. acuminata, as approximately 74% of the genetic variation occurs within clusters and only 6.3% between regions. Since M. acuminata possesses the ability for long-distance dispersal, high rates of historical gene flow within- and between M. acuminata populations in Ontario, as well as the potential contribution to the dispersal of nuclear DNA via seeds in addition to pollen (Setsuko et al., 2007) may have also reduced the rate of genetic drift and slowed the rate of population differentiation (O’Connell et al., 2008; Wilson and Traveset, 2000).

Testing for isolation by distance produced contrasting results when applying

Mantel tests to pairwise genetic and geographical distances based on genotypic and phenotypic data. Weak IBD was detected among the seven Ontario genetic clusters based on allelic phenotype data, but was not detected among the subset of clusters within Norfolk County, irrespective of data type. IBD among Ontario clusters reflects increased genetic differentiation at greater distances which is further corroborated by the spatial autocorrelation analysis which reveals across Ontario,

M. acuminata individuals are more alike at shorter distances. This could be attributed to temporal genetic differences based on adult and seedling cohorts (see below). The number of sampled seedlings in the Municipality of Niagara exceeds the number of sampled adults; therefore, regional differences in diversity and differentiation between the two cohorts could be influencing the correlation

54 between genetic and geographic distance at 80-100 km, representing the distances between Norfolk County and the Municipality of Niagara sample sites, <100 km. The lack of IBD based on genotypic data could be attributed to the underestimates of FST, especially since the strength of the relationship based on phenotypic data wasn’t particularly strong.

The increasing levels of differentiation, as suggested by our cohort data (see below), could be indicative of constraints to long-distance gene flow via contemporary topographic or reproductive barriers. For example, varying environmental conditions and more recent climatic shifts, above and below the

Niagara Escarpment, may result in changes to flower viability, the flowers’ capability of being fertilized, as well as pollinator availability and activity during flowering

(Setsuko et al., 2013). Theoretically, since interplant movement is stand-density dependent, the number of mature individuals as well as the number of viable flowers per site, could be influencing the quality and quantity of pollen being dispersed across the landscape. The number of naturally occurring Cucumber trees

I sampled per site in Ontario ranged from a single individual to as many as 40 trees.

Of the two regions, Norfolk County had on average both the largest and most dense sites. Of those sites sampled, based on the 2010 COSEWIC report, eight of the twelve sites contained fewer than ten mature, flowering adults. Kevan’s (2002) report on the reproductive constraints of M. acuminata found of the five study trees observed, all, with the exception of one tree having 50 flowers, had a total of 30 flowers. Of the total 170 flowers studied, only 25 (14.7 %) were identified as viable, capable of

55 being fertilized, and comprised only three of five trees in the study. If foraging behavior and interplant movement are dependent on stand-density, as well as the number of viable flowers per adult, based on the approximate population sizes and flowering data compiled by Kevan (2002), pollination of M. acuminata in Ontario should favour both interplant movement and outcrossing events (Eglund, 1993), which could also partially explain the greater estimates of genetic variability in the

Municipality of Niagara than within Norfolk County.

Evidence of genetic deterioration and increased differentiation in cohort data

Although current patterns of genetic diversity and differentiation do not at first seem to flag any potential threats to the viability of M. acuminata in Ontario, the picture changes when we divide our data into age cohorts, especially since contemporary genetic structuring following fragmentation is expected to be most prominent in younger age classes. Not only was the genetic diversity of pooled seedlings considerably lower than that of the pooled canopy trees, levels of genetic differentiation between the Norfolk and Niagara seedling cohort was considerably higher. However, when assessing diversity and differentiation among the two sites exclusively, this study produced conflicting results between data types: the genotypic data revealed that the adult cohort in NWA exhibited lower diversity compared with seedlings, comparative to the phenotypic data which suggests both adult cohorts from NWA and FD have higher levels of diversity than its seedlings.

The comparison of genetic diversity at the cohort level could be confounded by the heterogeneity of the adult cohort. Unlike seedling cohorts which were recently established, 1-3 years, adult M. acuminata cohorts comprise individuals from successive generations, when the gene pool was likely much larger and more diverse. Furthermore, by pooling individuals into regional cohort datasets (Norfolk and Niagara canopy/seedling datasets), measures of diversity are likely to be inflated compared to those for a single, within-population estimate. Since variation in M. acuminata is greater within populations than among them, pooling individuals from various sites can potentially result in the false representation of genetic variation or gene pool that otherwise may not occur as a consequence of limitations to gene flow. Similarly, estimates of diversity within seedling cohorts may not accurately reflect the future genetic stock of M. acuminata populations as not all seedlings will survive until maturity. Since genetic variation is expected to be higher among seedling cohorts compared with adult cohorts for this reason, this study raises some concern regarding the genetic viability of future M. acuminata, particularly within FD. Nonetheless, higher diversity in adult cohorts has been described among other tree species, including Quercus suber L. (Lorenzo et al., 2009;

Vranckx et al., 2014), Beccariophoenix madagascariensis (Shapcott et al., 2007), and

Medusagyne oppositifolia (Finger et al., 2011), which may be indicative of a general pattern of high diversity in adult cohorts among tree species with fragmented habitats as a function of their longevity and generation time.

The likely cause of increased differentiation between Norfolk and Niagara seedling cohorts is contemporary limitations to dispersal and lack of establishment sites. If gene flow is limited predominantly to within Norfolk County, and predominantly within the Municipality of Niagara, then each set of new recruits would typically reflect genotypes from a subset of adults from either region exclusively, leading to further differentiation with each new M. acuminata generation, as the rate of genetic drift and the magnitude of its effects increases.

Since gene movement is a sequential process, pollen followed by seed movement, disruptions to either or both dispersal mechanism can influence the genetic structure: limited seed dispersal can potentially lead to genetically distinct seed shadows, while limitations to pollen flow will influence the intensity of population structuring over subsequent generations (Hamrick and Nason, 1996).

Seedling establishment is critical for the natural regeneration of populations and the longterm maintenance of genetic diversity, and the lack of establishment sites can not only prevent establishment but also alter regeneration dynamics leading to increased differentiation. The presence of excessive leaf litter and herbaceous layers, as well as dense understory, can impede germination by hindering seed and soil contact and blocking necessary light for germination to occur (McConnell and Menges, 2002), thus, delaying or inhibiting seedling establishment and development (Caccia and Ballare, 1998). As mentioned above, regeneration dynamics within clusters, more specifically the occurrence of generational gaps between adult and seedling cohorts, has also likely influenced the

58 current patterns of differentiation reflected in Ontario M. acuminata. As new individuals fail to establish annually, there is an uneven representation of age classes over time. Similar to the restrictions to gene flow within regions, between generations, we may expect to see each new surviving generation represented by different subsets of reproducing adults, and likely different subsets of genotypes.

Thus, the apparent low natural recruitment within most clusters, with the exception of Cluster 7 (FD) where natural recruitment is apparently extensive, not only indicates a disruption to gene flow as well as the reproductive ecology of M. acuminata in southern Ontario, but also contributes to differentiation among seedling cohorts between Ontario regions caused by increasing rates of genetic drift with each new passing generation.

The observed patterns between genetic diversity, differentiation, and cohorts, as well as the lack of reproductive output in most Ontario M. acuminata clusters, could be attributed to a number of factors: decreases in adult population sizes (Aizen and Feinsinger, 1994; Aguillar and Galleto, 1994), deteriorating site conditions (Hirayama et al., 2007), pollen shortages caused by the disruption and/or reduction in the abundance, diversity, or shifts in foraging behaviours of pollinators

(Powell and Powell, 1987; Sih and Baltus, 1987; Rathcke and Jules, 1993), shifts in environmental and weather conditions (Kikuzawa and Mizui, 1990), and flowering phenology (Kevan, 2002). Variability in pollination can create fluctuating limitations within a given season, site, or year (Burd, 1994), for example, seed set in Magnolia hypoleuca depended partly on weather conditions on the day of anthesis (Kikuzawa

59 and Mizui, 1990). Compared to numerous congeneric species, Magnolia acuminata has both an early flowering season, as well as a shorter period of receptivity and pollen shedding, which could account for recent limitations to or year-to-year variability in natural regeneration. Since flowering in Ontario occurs in early spring, there is the potential for an increased risk of flower damage as well as suboptimal conditions on the day of anthesis, limiting flower viability as well as pollination through the alteration of pollinator abundance, type, activity, and foraging behaviours. Kevan (2002) revealed that approximately 85% of the M. acuminata flowers in his study were rendered inviable as flowers were damaged caused by inclement weather including frost and storms

The decline in diversity in seedling cohorts, particularly within FD and sites of similar characteristics, could also be partly caused by a shortage of effective dispersers during flowering (Kikuzawa and Mizui, 1990). If the frequency of visitation by beetles is reduced due to inclement weather conditions, then there is the potential for a reduction in beetle pollination, and potentially the decline of outcrossing events. The reduction of outcrossing events increases the risks of reduced seed production and seed set, as well as the potential of increased frequency of fruit abortion (Matsuki et al., 2008; Setsuko et al., 2013), a product of self-pollination. Kevan’s (2002) study revealed not only that fruiting was limited among the study trees (only one of the five trees produced fruit), but that seed set was extremely low, ranging from a single seed to four seeds per infructescence.

While the Kevan (2002) study represented a very small subset of the M. acuminata

60 populations in Ontario, it nevertheless raises some concerns regarding the frequency of self-pollination in Ontario populations of M. acuminata and its potential consequences to the future population genetic structure.

Even following successful fertilization, seed production, and seed set, M. acuminata seeds are particularly sensitive to temperature and moisture (Heit,

1975), with high susceptibility to freezing and sub-optimal conditions which hinder germination (Olson et al., 1974). Low germination rates, coupled with elevated levels of seed predation and seedling herbivory, could be further limiting M. acuminata reproduction in forests, and increasing the rate of differentiation.

Grackles (Quiscalus spp.) and blackbirds (Agelaius spp.) are known to eat the young, unripe fruit of Magnolia acuminata (McDaniel, 1974), and its twigs, leaves, and buds are frequently browsed by deer (Knierim et al., 1971; Smith, 1990; personal observation). Beginning around 1912, excessive deer browsing of most shrubs in and around the Pelee and Rondeau areas (Bartlett, 1958) has been documented as the likely cause for the decline in Dwarf cherry (Prunus pumila var. pumila) in the region (Catling and Larson, 1997) and may similarly impact M. acuminata. Within the National Wildlife Area specifically, the carrying capacity of deer is < 100 individuals (D. Bernard, Big Creek Conservation Area, personal communication).

Prior to the deer culls which began in 1989, much of the vegetation suffered from excessive overbrowsing by deer, when the population reached as high as 550-600 individuals (D. Bernard, Big Creek Conservation Area, personal communication).

This could have significant implications for the reproductive ecology of M.

61 acuminata, especially in southern Ontario as reproduction appears to be extremely limited. Only one site in Niagara (FD) contained a large proportion of naturally grown seedlings; however, while the seedlings were widely distributed within the site, they were also heavily clumped, an indication of limited seed dispersal (Gapare and Atkin, 2005).

The recovery of M. acuminata in Ontario: implementing appropriate management strategies

This study raises some concern regarding the future genetic viability of M. acuminata in southern Ontario. Without the application of adaptive management strategies, a continued pattern of decreased diversity and increased differentiation in young M. acuminata trees is expected to persist or increase within future populations in Ontario. Through the utilization of adaptive management, conservation programs are able to continuously improve their procedures and associated policies based on the outcomes of current and active management and monitoring practices.

Regarding M. acuminata specifically, managers need to consider strategies that will not only increase population sizes but also those which will promote and enhance reproductive success. Moreover, they should ensure the continued existence of target populations, while creating favorable conditions for the growth and viability of trees and natural regeneration (Rotach, 2005). Conservation efforts should first pursue the protection and maintenance of genetic diversitythrough in

62 situ management strategies to prevent further significant genetic differentiation between Norfolk County and the Municipality of Niagara supplmented by an appropriate, species-specific monitoring program. In situ management can be further complemented by ex situ methods, such as seed collection and transfer, if initial measures fail to produce satisfactory results. The measures utilized in the recovery of M. acuminata should most importantly take into account and ensure the genetic integrity of this species for long-term sustainability, in light of current environmental changes.

Site maintenance, for the purpose of promoting favorable conditions for seed germination and seedling establishment, as well as utilizing seed transfer techniques, are efficient and cost-effective approaches to enhancing genetic diversity and mitigating further genetic deterioration. The use of seed transfers in research and reforestation projects has a long history in Canada, particularly among commercially significant species including Douglas-fir (Pseudotsuga menziesii), Scots pine (Pinus sylvestris), Jack pine (Pinus banksiana), eastern white pine (Pinus strobus), and White spruce (Picea glauca). Putting aside the silvicultural motivations for seed transfer, this management technique, as mentioned above, may also be an effective method to conserve forest genetic resources.

Seed zones specific to M. acuminata, should be initially delineated across the species’ distribution, to be used as references for seed sources. Local valuable resource populations are those which will not compromise the species’ genetic structuring at the geographic level but also reduce the risk of outbreeding

63 depression with the intent to increase progeny fitness (Mijnsbrugge et al., 2010;

Aavik et al., 2012). Taking this into consideration, populations which show high levels of genetic diversity and low levels of differentiation should be targeted for use as seed sources (Hirayama et al., 2007; Frankham et al., 2010). The number of seeds collected and the number of individuals sampled should be proportional to the effective population size and the size of the crop. Furthermore, due to the potential for reduced seed set in M. acuminata and variability in seed crops between years and geographic locations (Smith, 1990), the collection of seeds in southern Ontario should be done during years of good crops which, on average, occur once every four to five years, although less frequently at range edges (Olson et al., 1974), when there is no apparent seed limitation nor subsequent risk to natural regeneration within sites. The transfer of seeds should occur annually, and initially target sites characterized by 1) lower levels of genetic diversity, 2) moderate levels of genetic differentiation, and 3) single to small parent populations.

In Norfolk County, there is high potential for the successful transfer of seeds and/or seedlings to suitable sites. WATT (Cluster 5) and CT (Cluster 2) should take priority as they meet the aforementioned criteria. While the size of the WATT property may provide more opportunity for seedling establishment than CT, canopy cover may limit its success by impeding solar radiation. Cluster 2 (CT) is located within a small valley surrounded by an adjacent cemetery, houses, and a corporate office, it is the most isolated population and has the lowest genetic diversity. Despite a relatively high number of trees present considering the size of the site, tree health

64 and site degradation appear to be of major concern. Site maintenance coupled with genetic assistance should aid in the long-term persistence of this site, even if it is just a small proportion of the original population. In light of the limited reproductive capacity of the remaining sites in Norfolk, managers should first manage the understory vegetation and leaf litter and monitor its effects on seedling establishment rates at all sites. If sites, particularly BT, ST, and SW, do not appear to be benefiting from site maintenance exclusively, then seed transfers may be enforced annually.

The implementation of a seed transfer program in the Municipality of Niagara would be difficult as many of the sites are privately owned and most of the population sizes are small. Of the sites I visited and sampled, only one (FD) has a potential for making significant contributions to the recovery of M. acuminata in this region, not only based on the number of trees on the lot, but also the apparent reproductive success, and co-operation and support by the landowners. There are a few other known sites in the Municipality of Niagara that home large number of

Cucumber trees, but without any sort of incentive to the landowners, the willingness of support and co-operation seems unlikely since permission to access the land to survey and/or monitor the status of these trees, more recently, has been repeatedly denied. Taking this into consideration, public outreach may be beneficial for the maintenance of geographic and genetic connectivity between known M. acuminata sites, having landowners voluntarily accept seed and/or seedling transfers. Genetic connectivity may also benefit from the reintroduction of seeds into extirpated sites

(Ambrose and Kirk, 2007), including riparian habitat along waterways and wet-

65 mesic forest fragments in both the Municipality of Niagara and Norfolk County. By introducing a small number of seeds annually, continuous immigration, even by a single seed or individual, can reduce the rate of genetic differentiation (Woodworth,

2002).

As mentioned above, a species-specific monitoring program should be developed and implemented in order to assess the status of the populations, effectiveness and success of site maintenance and/or seed(ling) transfers, as well as update the management plan accordingly. The program should be undertaken by a designated authority, who would then work collaboratively with relevant agencies, private landowners, NGOs, and any other interested external stakeholders. The

Long Point Region Conservation Authority (LPRCA), North American Native Plant

Society (NANPS), Environment Canada, Friends of Shorthills and Canada Parks all contribute to the management of sites I had visited, many of which are home to a significant proportion of the Ontario population of M. acuminata, with the exception of most sites within the Municipality of Niagara which were low-density and were located on private property.

Aside from the support and co-operation of the aforementioned governmental agencies, NGO’s, and private landowners, I believe that the success of a monitoring program would rely heavily on the assessment of populations utilizing both ecological and genetic approaches. According to Koskela et al. (2013), data collection should focus on natural regeneration and population sizes, taking note of

66 age and size class distributions, reproductive fitness, and regeneration abundance.

In addition, regular site visits are encouraged to allow for the detection of any potential issues that could influence the viability and reproduction of populations, as well as the success of the program, particularly damage to individuals and/or populations caused by natural and or man-made catastrophes such as selective or clear cutting, land conversion, storms, etc. Field inventories of M. acuminata should also take note of instances of fruit abortion (i.e., number of trees per cluster undergoing fruit abortion), number of reproducing individuals, and when applicable and possible, flowering and pollinator/dispersal data including flowering season and length of flowering period, number of flowers per effective individual, number of types and species of pollinators, interactions with other types and species of animals. Regarding the genetic component of the assessment, Aravanopolous

(2011) developed an approach to use three demographic and four genetic verifiers as indicators of natural selection, genetic drift, and gene flow- mating system.

Based on this method, genetic assessment should quantify 1) effective population size, 2) allelic richess, 3) genetic potential, and 4) outcrossing and actual inbreeding rates.

Conservation Implications and Conclusion

This study is the first to investigate the population genetic structure of M. acuminata in southern Ontario. Utilizing both chloroplast and nuclear markers we found that the seven identified clusters of Ontario M. acuminata, despite revealing uniformity in cpDNA haplotypes, comprise high levels of genetic diversity within

67 clusters, and moderate levels of genetic differentiation among clusters. Coupled with the absence of a latitudinal cline in genetic diversity, the results presented in this study suggest that M. acuminata may have followed a stratified dispersal model characterized by gradual recolonization of southern Ontario following the LGM, and occasional long-distance dispersal events, further facilitated expansion into the region during the onset of recolonization. The reduced diversity and increased differentiation seen in the young cohort reveal recent limitations to pollen and seed dispersal, which eventually could lead to further genetic and ecological disruptions through long-term isolation. Our results identify the need for species-specific adaptive management, including the development and implementation of an effective monitoring program. The management strategies proposed promote the long-term maintenance of genetic diversity within Ontario clusters, through the enhancement of seed germination and seedling establishment rates.

This project promotes the use of integrative approaches in conservation studies to elucidate the relationship between historical landscape changes and species viability. While many of these remnant populations of Magnolia acuminata have the potential to persist, without the application of co-operative conservation strategies, there is the potential for a further reduction in size and/or number of genetic reserves available for long-term population sustainability of M. acuminata in

Ontario.

REFERENCES

Aavik , T., Edwards, P.J., Holderegger, R., Graf, R., Billeter, R. 2012. Genetic consequences of using seed mixtures in restoration: A case study of a wetland plant Lychnis flos-cuculi. Biological Conservation 145: 195-204.

Afzal-Rafii, Z., and Dodd, R.S. 2007. Chloroplast DNA supports a hypothesis of glacial refuge over postglacial recolonization in disjunct populations of black pine (Pinus nigra) in western Europe. Molecular Ecology 16: 723-736.

Aguilar, R., and L. Galleto. 1994. Effects of forest fragmentation on male and female reproductive success in Cestrum parqui (Solanaceae). Oecologia 138: 513-520.

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. Molecular Ecology 17: 5177-5188.

Aizen M.A., and P. Feinsinger. 1994. Forest fragmentation, pollination, and plant reproduction in a Chaco dry forest, Argentia. Ecology 75: 330-351.

Altizer, S., Harvell, D., Friedle, E. 2003. Rapid evolutionary dynamics and disease threats to biodiversity. Trends in Ecology and Evolution 18: 589-596.

Ambrose, J.D. and D. Kirk. 2010. Recovery strategy for Cucumber tree (Magnolia acuminata) in Canada. Prepared for the Ontario Ministry of Natural Resources by the Cucumber tree Recovery Team, viii: 24 pp. + addenda.

Ambrose, J.D. and P.G. Kevan. 1990. Reproductive biology of rare Carolinian plants with regard to conservation management. In G. Allen (eds) Conserving Carolinian Canada, University of Waterloo Press, Waterloo, ON.

Ambrose, J.D. and S.W. Aboud. 1984. Status report on the Cucumber tree Magnolia acuminata in Canada. Committee on the Status of Endangered Wildlife in Canada, Canadian Wildlife Service, Ottawa, Unpublished report. 22 pp.

Anagnostakis, S.L. 2001. The effect of multiple importations of pests and pathogens on a native tree. Biological Invasions 3: 245–254.

Anguilar, R., Ashworth, L., Galetto, L., Aizen, M.A. 2006. Plant reproductive susceptibility to habitat fragmentation: review and synthesis through a meta- analysis. Ecology Letters 9: 968-980.

Aravanopolous, F.A. 2011. Genetic monitoring in natural perennial plant populations. Botany 89: 75-81.

Austerlitz F, Mariette S, Machon N, Gouyon P-H, Godelle B. 2000. Effects of colonization processes on genetic diversity: differences between annual plants and tree species. Genetics 154: 1309-1321.

Austin, J.D., Lougheed, S.C., Neidrauer, L., Chek, A.A., Boag, P.T. 2002. Cryptic lineages in a small frog: the post-glacial history of the spring peeper, Pseudacris crucifer (Anura: Hylidae). Molecular Phylogenetics and Evolution 25: 316-329.

Avise, J.C. 2000. Phylogeography: the history and formation of species. Harvard University Press, Cambridge, MA.

Barbeta, A., Penuelas, J., Ogayaa, R., Jump, A.S. 2011. Reduced tree health and seedling production in fragmented Fagus sylvatica forest patches in the Montseny Mountains (NE Spain). Forest Ecology and Management 261: 2029-2037.

Barrett, S.C.H, and J.R. Kohn. 1991. Genetic and evolutionary consequences of small population size in plants: implications for conservation. In Genetics and conservation of rare plants. Oxford University Press, New York, NY. 3-30pp.

Bartlett, C.O. 1958. A summary of some deer and forest relationships in Rondeau Park. Wildlife Series 7. Ontario Department of Lands and Forests, Toronto, ON.

Beardmore, T., Loo, L., McAfee, B., Malouin, C., Simpson, D. 2006. A survey of tree species of concern in Canada: the role for genetic conservation. The Forestry Chronicle 82: 351-363.

Beresford-Kroeger, D. 2003. Arboretum America: A Philosophy of the forest. University of Michigan Press, MI. 96pp.

Bernabo, J. C., and T. Webb III. 1977. Changing patterns in the Holocene pollen record of northeastern North America: a mapped summary. Quaternary Research 8: 64-96.

Bernhardt, P. 2000. Convergent evolution and adaptive radiation of beetle- pollinated angiosperms. Plant Systematics and Evolution 222: 293-320.

Bernhardt, T. Mixedwood Plains. Canada's Ecozones, Canadian Biodiversity project. McGill University, Redpath Museum. (http://canadianbiodiversity.mcgill.ca/english/ecozones/mixedwoodplains/mixed woodplains.htm). Braun, E.L. 1950. Deciduous forests of eastern North America. Blackburn Press, Caldwell, NJ.

Braun, E.L. 1947. Development of the deciduous forests of eastern North America. Ecological Monographs 17: 211-219.

Braun, E.L. 1955. The phytogeography of unglaciated eastern United States and its interpretation. Botanical Review 21: 297-375.

Brussard, P.F. 1984. Geographic patterns and environmental gradients: The central- marginal model in Drosophila revisited. Annual Review of Ecology and Systematics 15: 25-64.

Bunnell, F.L., Campbell, R.W., Squires, K.A. 2004. Conservation priorities for peripheral species: the example of British Columbia. Canadian Journal of Forest Restoration 34: 2240-2247.

Burd, M. 1994. Bateman’s principle and plant reproduction: the role of pollen limitation in fruit and seed set. The Botanical Review 60: 83-139.

Burg, T.M., Gaston, A.J., Winker, K., Friesen, V.L. 2006. Effects of Pleistocene glaciations on population structure of North American chestnut-backed chickadees. Molecular Ecology 15: 2409-2419.

Burgos-Hernández, M., Hernández, D., Castillo-Campos, G. 2013. Genetic diversity and population genetic structure of wild banana Musa ornata (Musaceae) in Mexico. Plant systematics and evolution 299: 1899-1910.

Burnham, C.R., Rutter, P.A., French, D.W. 1986. Breeding blight-resistant chestnuts. Plant Breeding Reviews 4: 347–397.

Caccia, F.D., and C.L. Ballare. 1998. Effects of tree cover, understory vegetation, and litter on regeneration of Douglas-fir (Pseudotsuga menziessii) in southwestern Argentina. Canadian Journal for Forest Restoration 28: 683-692.

Canadian Council of Ecological Areas. 1995. An introduction into Canadian ecozones. Canadian Council of Ecological Areas, Hull, Quebec.

Catling, P.M., and B.M.H Larson. 1997. The decline and current status of the dune race of dwarf cherry, Prunus pumila, var. pumila, on the Canadian shores of the lower Great Lakes. Canadian field-naturalist 111: 187-193.

Channel, R. 2004. The conservation value of peripheral populations: the supporting science. Proceedings of the Species at Risk 2004 Pathways to Recovery Conference. 1 March 2–6, 2004, Victoria, B.C. Species at Risk 2004 Pathways to Recovery Conference Organizing Committee, Victoria, B.C.

Chase, T.N., Pielke, R.A., Kittel, T.G.F., Nemani, R.R., Running, S.W. 1996. Sensitivity of a general circulation model to global changes in leaf area index. Journal of Geophysical Research 101: 7393–7408.

Church, S.A., Kraus, J.M., Mitchell, J.C., Church, D.R., Taylor, D.R. 2003. Evidence for multiple Pleistocene refugia in the postglacial expansion of the eastern tiger salamander, Ambystoma tigrinum tigrinum. Evolution 57: 372-383.

Clark, S., McNab, H., Loftis, D., Zarnoch, S. 2012. American chestnut growth and survival five years after planting in two silvicultural treatments in the southern Appalachians, USA. Forests 3: 1017-1033.

Cleland, C.E. 1966. The prehistoric animal ecology and ethnozoology of the Upper Great Lakes Region. Anthropological Papers 29. Museum of Anthropology, University of Michigan, Ann Arbor, MI.

Corriveau, J.L., and A.W. Coleman. 1988. Rapid screening method to detect potential biparental inheritance of plastid DNA and results for over 200 angiosperm species. American Journal of Botany 75: 1443-1458.

COSEWIC, 2003. Canadian Species at Risk, November 2003. Committee on the Status of Endangered Wildlife in Canada. 44 pp.

COSEWIC. 2010. COSEWIC assessment and status report on the Cucumber Tree Magnolia acuminata in Canada. Committee on the Status of Endangered Wildlife in Canada. Ottawa. x + 18 pp. (www.sararegistry.gc.ca/status/status_e.cfm).

COSEWIC. 2013. Canadian Wildlife Species at Risk. Committee on the Status of Endangered Wildlife in Canada. (http://www.cosewic.gc.ca/eng/sct0/rpt/rpt_csar_ e_2013.pdf).

Cozzolino, S., Cafasso, D., Pellegrino, G., Musacchio, A., Widmer, A. 2003a. Fine-scale phylogeographic analysis of Mediterranean Anacamptis palustris (Orchidaceae) populations based on chloroplast minisatellite and microsatellite vatiation. Molecular Ecology 12 : 2783-2792.

Davis, A.J., Jenkinson, L.S., Lawton, J.H., Shorrocks, B., Wood, S. 1998. Making mistakes when predicting shifts in species range response to global warming. Nature 391: 783-786.

Davis, M.B. 1983. Quaternary history of deciduous forests of eastern North America and Europe. Annals of the Missouri Botanical Garden 70: 550-563.

Davis, M.B.D. 1976. Pleistocene biogeography of temperate deciduous forests. Geoscience and Man 13 : 13-26.

Del Cerro, A., Lucas-Borja, M.E., Martinez García, E., López-Serrano, F.R., Andrés- Abellán, M., García-Morote, F.A., Navarro-López, R. 2009. Influence of stand density and soil treatment on the Spanish black pine (Pinus nigra Arn. ssp Salzmannii) regeneration in Spain. Invest. Agrar. Sist. Rec. For. 18: 167–180.

Delcourt H.R., and P.A. Delcourt. 1977. Presettlement magnolia-beech climax of the gulf coastal plain: quantitative evidence from the Apalachicola River Bluffs, north central Florida. Ecology 58: 1085-1093.

Delcourt, H.R. 1979. Late Quaternary history of the Eastern Highland Rim and adjacent Cumberland Plateau of Tennessee. Ecological Monographs 49: 255-280.

Delcourt, H.R., and P.A. Delcourt. 1975. The Blufflands: Pleistocene pathways into the Tunica Hills. American Midland Naturalist 94: 385-400.

Delcourt, H.R., and P.A. Delcourt. 1979. Late Pleistocene and Holocene distributional history of the deciduous forest in the southeastern United States. Verdffentlichungen des Geobotanischen Institutes der ETH, Stiftung Ruibel (Zurich), CH. 68pp.

Dick, C.W., Etchelecu, G., Austerlitz, F. 2003. Pollen dispersal of tropical trees (Dinizia excels: Fabaceae) by native insects and African honey bees in pristine and fragmented Amazonian rainforest. Molecular Ecology 12: 64-71.

Dobrowski, S.Z. 2011. A clamitic basis from microrefugia: the influence of terrain on climate. Global Change Biology 17: 1022-1035.

Durand, E., Jay, F., Gaggiotti, O.E. 2009. Spatial inference of admixture proportions and secondary contact zones. Molecular Biology and Evolution 26: 1963–1973.

Ellstrand, N.C. 1992. Gene flow among seed plant populations. New Forests 6: 241- 256.

Ellstrand, N.C., and D.R. Elam. 1993. Population genetic consequences of small population size: implications for plant conservation. Annual Review of Ecology and Systematics 24: 217-242.

Ellstrand, N.C., Devlin, B., Marshall, D.L. 1989. Gene flow by pollen into small populations: data from experimental and natural stands of wild radish. Proceddings of the National Academy of Science 86: 9044-9047.

Englund, R. 1993. Movement patterns of Cetoniabettles (Searabaeidae) among flowering Viburnum opulus (Caprifoliaceae). Option for long-distance pollen dispersal in a temperate shrub. Oecologia 94: 295-302.

ERCA. 2009. Essex forests and wetlands conservation action plan. Essex Region Conservation Authority Publication, Essex, ON. 64pp.

Evanno, G., Regnaut, S., Goudet, J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611–2620.

Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution, and Systematics 34: 487-515.

Fenster, C. B. 1991. Gene flow in Chamaecrista fasciculate (Leguminosae). I. gene dispersal. Evolution 45: 398-409.

Finger, A., Kettle, C.J., Kaiser-Bunbury, C.N., Valentin, T., Doudee, D., Matatiken, D., Ghazoul, J. 2011. Back from the brink: potential genetic rescue in a critically endangered tree. Molecular Ecology 20: 3773-3784.

Flannery, T. 2003. The eternal frontier: an ecological history of north american and its peoples. Grove Press, New York, NY. 432pp.

Frankham, R. 1995. Effective population size/adult population size ratios in wildlife: a review. Genetical Research 66:95-107.

Frankham, R. 2003. Genetics and conservation biology. Comptes Rendus Biologies 326: S22-S29.

Frankham, R. 2010. Challenges and opportunities of genetic approaches to biological conservation. Biological Conservation 143: 1919-1927.

Frankham, R., Ballou, J.D., Briscoe, D.A. 2002. Introduction to Conservation Genetics. Cambridge University Press, Cambridge, UK.

Gapare, W.J. and S.N. Atkin. 2005. Strong spatial genetic structure in peripheral but not core populations of Sitka spruce [Picea sitchensis (Bong.) Carr.]. Molecular Ecology 14: 2659-2667.

Gapare, W.J., Yanchuk, A.D., Aitken, S.N., 2007. Optimal sampling strategies for capture of genetic diversity differ between core and peripheral populations of Picea sitchensis (Bong.) Carr. Conservation Genetics 9: 411–418.

Garner, T.W.J., Pearman, P.B., Angelone, S. Genetic diversity across a verterate species’ range: a test of the central-peripheral hypothesis. Molecular Ecology 13: 1047-1053.

Geburek, T. 1997. Isosymes and DNA markers in gene conservation of forest trees. Conservation Biology 15: 1632-1644.

Glasgow, L.S., and G.R. Matlack. 2007. Prescribed burning and understory composition in a temperate deciduous forest, Ohio, USA. Forest Ecology and Management 238: 54-64.

Goel, N.S. and N. Richter-Dyn. 1974. Stochastic models in biology. Academic Press, New York, NY. 269 pp.

Goldenberg, E.M., Giannasi, D.E., Clegg, M.T., Smiley, C.J., Durbin, M., Henderson, D., Zurawski, G. 1990. Chloroplast DNA sequences from a Miocene Magnolia species. Nature 344: 656-658.

Gonzales, E., Hamrick, J.L., Change, S.M. 2008. Identification of glacial refugia in southeastern North America by phylogeographic analyses of a forest understory plant, Trillium cuneatum. Journal of Biogeography 35: 844- 852.

Guichoux, E., Lagache, L., Wagner, S., Chaumeil, P., Leger, P., Lepais, O., Lepoittevin, C., Malausa, T., Revardel, E., Salin, F., Petit, R.J. 2011. Current trends in microsatellite genotyping. Molecular Ecology Resources 11: 591-611.

Haldane, J.B.S. 1930. Enzymes. In R.H.A. Plimmer, D.Sc., and Sir F. G. Hopkins, M.A., M.B., D.Sc., F.R.S. M.A. (eds) Monographs on Biochemistry. Longmans, Green & Co., London, UK. vii+235pp.

Hampe, A., Pemonge, M.H., Petit, R.J. 2013. Efficient mitigation of founder effects during the establishment of a leading-edge oak population. Proceedings of the Royal Society of B: Biological Sciences 280: 1-7.

Hamrick, J.L. 2004. Response of forest trees to global environmental changes. Forest Ecology Management 197: 323-335.

Hamrick, J.L. and J.D. Nason. 1996. Consequences of dispersal in plants. In Rhodes, O.E., Chesser, R.K., Smith, M.H. (eds) Population Dynamics in Ecological Space and Time. University of Chicago Press, Chicago, IL. 203-236 pp.

Hardy O.J., and X. Vekemans. 2002. Spagedi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Molecular Ecology Notes 2: 618-620.

Hebda,R.J. and E. Irving. 2004. On the origin and distribution of Magnolias: tectonics, DNA, and climate change. In Timescales of the Paleomagnetic Field Geophysical Monograph Series (AGU) 145: 43-57.

Heiser, C.B. 1962. Some observations on pollination and compatibility in Magnolia. Proceedings of the Indiana Academy of Science 72: 259- 266.

Heit, C. E. 1975. Propagation from seed. Collecting, testing, and growing magnolia species. American Nurseryman 142: 10-11, 79-85.

Herdendorf, C.E. 2013. Research overview: Holocene development of Lake Erie. Ohio Journal of Science 112: 24-36.

Herrera, C.M., and P. Bazaga. 2008. Adding a third dimension to the edge of a species' range: altitude and genetic structuring in mountainous landscapes. Heredity 100: 275–285.

Hewitt G.M. 2004. Some genetic consequences of ice ages, and their role in divergence and speciation. Biological Journal of the Linnean Society 58: 247-276.

Hewitt, G. M. 1993. in Hybrid zones and Evolutionary Process(ed. Harrison, R. G.). Oxford University Press, Oxford, UK, 140–164pp.

Hewitt, G. M. 2000. The genetic legacy of the Quarternary ice ages. Nature 405: 907- 913.

Hirayama, K., Ishida, K., Setsuko, S., Tomaru, N. 2007. Reduced seed production, inbreeding, and pollen shortage in a small population of a threatened tree, Magnolia stellata. Biological Conservation 136: 315-323.

Hirayama, K., Ishida, K., Tomaru, N. 2005. Effects of pollen shortage and self- pollination on seed production of an endangered tree, Magnolia stellata. Annual Botany- London 95: 1009-1015.

Holsinger, K.E., and P.O. Lewis. 2003. Hickory: A package for analysis of population genetic data version 1.1. University of Connecticut, Storrs, Connecticut, USA. http://darwin.eeb.uconn.edu/hickory/hickory.html

Ibrahim, K.M., Nichols, R.A., Hewitt, G.M. 1996. Spatial patterns of genetic variation generated by different forms of dispersal during range expansion. Heredity 77: 282– 291.

Isagi, Y., Kanazashi, T., Suzuki, W., Tanaka, H., Abe, T. 1999. Polymorphic microsatellite DNA markers for Magnolia obovata Thunb. And their utility in related species. Molecular Ecology 8: 698-700.

Isagi, Y., Kanazashi, T., Suzuki, W., Tanaka, H., Abe, T. 2000. Microsatellite analysis on regeneration process of Magnolia obovata. Heredity 84: 143-151.

Isagi, Y., Kanazashi, T., Suzuki, W., Tanaka, H., Abe, T. 2004. Highly variable pollination patterns in Magnolia obovata revealed by microsatellite paternity analysis. International Journal of Plant Science 165: 1047-1053.

Isagi, Y., Tateno, R., Matsuki, Y., Hirao, A., Watanabe, S., Shibata, M. 2007. Genetic and reproductive consequences of forest fragmentation for populaions of Magnolia obovata. Ecological Restoration 22: 382-389.

Jackson, S.T., Webb, R.S., Anderson, K.H., Overpeck, T., Webb III, T., Williams, J.W., Hansen, B.C.S. 2000. Vegetation and environment in Eastern North America during the Last Glacial Maximum. Quaternary Science Reviews 19: 489-508.

Jalava, J.V., Sorrill, P.J., Henson, J., Brodribb, K. 2000. The Big Picture Project: Developing a natural heritage vision for Canada’s southernmost ecological region. Natural Heritage Information Centre, Ontario Ministry of Natural Resources, Peterborough. Manuscript. 13 pp.

Jost, L. 2008. GST and its relatives do not measure differentiation. Molecular Ecology 17: 4015-4026.

Keener, C. and E. Kuhns. 1997. The impact on Iroquoian populations on the northern distribution of pawpaws in the northeast. North American Archeologist 16: 327-342.

Keir, K.R., Bemmels, J.B., Aitken, S.N. 2011. Low genetic diversity, moderate local adaptation, and phylogeographic insights in Cornus Nuttallii (Cornaceae). American Journal of Botany 98: 1327-1336.

Kelkar, Y.D., Strubczewski, N., Hile, S.E. 2010. What is a microsatellite: a computational and experimental definition based upon repeat mutational behavior at A/T and GT/AC repeats. Genome Biology and Evolution 2: 620-635.

Kerr, J.T. and I. Deguise. 2004. Habitat loss and the limits to endangered species recovery. Ecology Letters 7: 1163-1169.

Kerr, J.T. and J. Cihlar. 2004. Patterns and causes of species endangerment in Canada. Ecological Applications 14: 743-753.

Kevan, P. 2002. Sexual reproduction as a constraint to recovery in Magnolia acuminata. Report to Ontario Ministry of Natural Resources, Ontario, Canada. 11pp. (Accessible via OMNR)

Kikuzawa, K. and N. Mizui. 1990. Flowering and fruiting phenology of Magnolia hypoleuca. Plant Species Biology 5: 255-261.

Kim, S., Soltis, D.E., Soltis, P.S., and Suh, Y. 1994. DNA sequences from Miocene fossils: an NDHF sequence of Magnolia Latahensis (Magnoliaceae) and an RBCL sequence of Persea Pseudocarolinensis (Lauraceae). American Journal of Botany 91: 615-620.

Kisel, Y., Moreno-Letelier, A.C., Bogarin, D., Powell, M.P., Chase, M.W., Barraclough, T.G. 2012. Testing the link between population genetic differentiation and clade diversification in Costa Rican orchids. Evolution 66: 3035-3052.

Knierim, P.G., Carvell, K.L., Gill, J.D. 1971. Browse in thinned oak and cove hardwood stands. Journal of Wildlife Management 35: 163-168.

Kolpakov, R., Bana, G., Kucherov, G. 2003. mreps: efficient and flexible detection of tandem repeats in DNA. Nucleic Acids Research 31: 3672-3678.

Konuma, A., Tsumura, Y., Lee, C.T., Okuda, T. 2000. Estimation of gene flow in the tropical-rainforest tree Neobalanocarpus heimii (Dipterocarpaceae), inferred from paternity analysis. Molecular Ecology 9: 1843-1852.

Koskela, J., Lefèvre, F., Schueler, S., Kraigher, H., Olrik, D.C., Hubert, J., Longauer, R., Bozzano, M., Yrjänä, L., Alizoti, P., Rotach, P., Vietto, L., Bordács, S., Myking, T., Eysteinsson, T., Souvannavong, O., Fady, B., De Cuyper, B., Heinze, B., von Wühlisch, G., Ducousso, A., Ditlevsen, B. 2013. Translating conservation genetics into management: Pan-European minimum requirements for dynamic conservation units of forest tree genetic diversity. Biological Conservation 157: 39-49

Kramer, A.T., Ison, J.L., Ashley, M.V., Howe, H.F. 2008. The paradox of forest fragmentation genetics. Conservation Biology 22: 878-885.

Kuang, D.Y., Wu, H., Wang, Y.L., Gao, L.M., Zhang, S.Z., Lu, L. 2011. Complete chloroplast genome sequence of Magnolia kwangsiensis (Magnoliaceae): implication for DNA barcoding and population genetics. Genome 54: 663-673. Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam, H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J., Higgins, D.G. 2007. Clustal W and Clustal X version 2. Bioinformatics 23: 2947-2948.

Larson, G., and R. Schaetzl. 2001. Review: origin and evolution of the Great Lakes. Journal of Great Lakes Research 27: 518-546.

Lei, T.T., Semones, S.W., Walker, J.F., Clinton, B.D., Nilsen, E.T. 2002. Effects of Rhododendron maximum thickets on tree seed dispersal, seedling porphology, and survivorship. Internation Journal of Plant Science 163: 991-1000.

Lesica, P., and Allendorf, F.W. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9: 753-760.

Levin, D. A., and H.W. Kerster. 1974. Gene flow in seed plants. Evolutionary Ecology 7: 139-220.

Levin, D.A. 1981. Dispersal versus gene flow in plants. Annals of the Missouri Botanical Garden 68: 233-253.

Lewis, C.F., Karrow, P.F., Blasco, S.M., McCarthy, F.M.G., King, J.W., Moore, Jr., T.C., Rea, D.K. 2008. Evolution of lakes in the Huron basin: Deglaciation to present. Aquatic Ecosystem Health and Management 11: 127-136.

Lewis, T. 1997. Thrips as crop pests. In Flight and dispersal, T. Lewis (ed.).CAB International, Wallingford, UK. 175-196pp.

Lewontin, R.C. 1974. The genetic basis of evolutionary change. Columbia University Press, N.Y. 346 pp.

Loehle, C. 2007. Predicting Pleistocene climate from vegetation in North America. Climate of the Past 3: 109-118.

Lorenzo, Z., Soto, A., Valladares, F., Gil, L. 2009. Elucidating the role of genetic drift and natural selection in cork oak differentiation regarding drought tolerance. Molecular Ecology 18: 3803-3815.

Lowe, A.J., Boshier, D., Ward, M., Bacles, C.F.E., Navarro, C. 2005. Genetic resource impacts of habitat loss and degradation; reconciling empirical evidence and predicted theory for neotropical trees. Heredity 95: 255-273.

Lucas-Borja, M.E., Fonseca, T., Parresol, B.R., Silva-Santos, P., García-Morote, F.A., Tíscar-Oliver, P.A. 2011. Modelling Spanish black pine seedling emergence: Establishing management strategies for endangered forest areas. Forest Ecology and Management 262: 195-202.

MacArthur, R.H., and E.O. Wilson. 1967. Island biogeography. Monographs in Population Biology 1: 1-203.

Mandák, B., Bímová, K., Plačková, I., Mahelka, V., Chrtek, J. 2005. Loss of genetic variation in geographicaly marginal populations of Atriplex tatarica (Chenopodiaceae). Annals of Botany 96: 901-912.

Matsuki, Y., Tateno, R., Shibata, M., Isagi, Y. 2008. Pollination efficiencies of flower- visiting insects as determined by direct genetic analysis of pollen origin. American Journal of Botany 95: 925-930.

Maxwell, J.A, and M.D. Davis. 1972. Pollen evidence of Pleistocene and Holocene vegetation on the Allegheny Plateau, Maryland. Quarternary Research 2: 506-530.

Maycock, P.F. 1963. The phytosociology of the deciduous forest of extreme southern Ontario. Canadian Journal of Botany 41: 379-438.

Mayr, E. 1970. Populations, species, and evolution. Harvard University Press, Cambridge, Massachusetts. 453pp.

McConnell, K., and E.S. Menges. 2002. Effects of fire and treatments that mimic fire on the Florida endemic scrub buckwheat (Eriogonum longifolium Nutt. Var. gnaphalifolium Gand.). Natural Areas Journal 22: 194-201.

McDaniel, J.C. 1963. Securing seed production in Magnolia acuminata and M. cordata. In Proceedings, Thirteenth Annual Meeting of International Plant Propagators Society, Eastern Region. 327pp.

McDaniel, Joseph C. 1974. Magnolia breeding possibilities. Plants and Gardens 30:74- 77.

McLachlan, J.S., Clark, J.S., Manos, P.S. 2005. Molecular indicators of tree migration capacity under rapid climate change. Ecology 86: 2088-2098.

Meirmans, P.G. and P.H. van Tienderen. 2004. GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes 4: 792-794.

Meirmans, P.G. and P.W. Hedrick. 2011. Assessing population structure: FST and related measures. Molecular Ecology Resources 11:5-18.

Mijnsbrugge, K. V., Bischoff, A., Smith, B. 2010. A question of origin: where and how to collect seed for ecological restoration. Basic and Applied Ecology 11: 300–311.

Moody, M.E., Muellert, L.D., Soltis, D.E. 1993. Genetic variation and random drift in autotetraploid populations. Genetics 154: 649-657.

Muir, G., Lowe, A.J., Fleming, C.C., Vogl, C. 2004. High nuclear genetic diversity, high levels of outcrossing and low differentiation among remnant populations of Quercus petraea at the margin of its range in Ireland. Annuals of Botany 93: 691-697.

Murphy, B. 2010. 2.0 Watershed geologic characterization. Natural Heritage Areas Inventory: 1-22.

Murphy, J.L. 2001. Pawpaws, persimmons, and ‘possums: on the natural distribution of pawpaws in the northeast. North American Archeologist 22: 93-115.

Nason, J.D., Aldrich, P.R., Hamrick, J.L. 1997. Dispersal and the dynamics of genetic structure in fragmented tropical tree populations. In Lawrence, W.F., Bierregaard Jr., O. (eds) Tropical forest remnants: ecology, management, and conservation of fragmented communities. University of Chicago Press, Chicago, IL. 304-320 pp.

Nason, J.D., Hamrick, J.L., Fleming, T.H. 2002. Historical vicariance and plostglacial colonization efects on the evolution of genetic structure in Lophocereus, a Sonoran Desert columnar cactus. Evolution 56: 2214-2226.

O’Connell, L.M., Ritland, K., Thompson, S.L. 2008. Patterns of post-glacial colonization by western redcedar (Thuja plicata, Cupressaceae) as revealed by microsatellite markers. Botany 86: 194-203.

Obbard, D.J., Harris, S.A., Pannell, J.R. 2006. Smple allelic-phenotype diversity and differentiation statistics for allopolyploids. Heredity 97: 296-303.

Ogden, J.G., III. 1966. Forest history of Ohio. I. Radiocarbon dates and pollen stratigraphy of Silver Lake, Logan County, Ohio. Ohio Journal of Science 66: 387- 400.

Olson, D. F., Jr., Barns, R.L., Jones, L. 1974. Magnolia L. In C. S. Schopmeyer (eds) Seeds of woody plants in the United States. U.S. Department of Agriculture, Washington, DC. 527-530pp.

Parisod, C., and G. Bonvin. 2008. Fine-scale genetic structure and marginal processes in an expanding population of Biscutella laevigata L. (Brassicaceae). Heredity 101: 536-542.

Parks, C.R., Wendel, J.F., Sewell, M.M. and Qiu, Y.L. 1994. The significance of allozyme variation and introregression in the Liriodendron tulipifera complex (Magnoliceae). American Journal of Botany 81: 878-889.

Pautasso, M. 2009. Geographical genetics and the conservation of forest trees. Perspectives in Plant Ecology, Evolution, and Systematics 11: 157-189.

Peterson, B.J. 2013. Phylogeography of eastern leatherwood (Dirca palustris L.) resolved by chloroplast sequencing and microsatellite genotyping. Graduate Thesis and Dissertations. Iowa State University, Ames, IA.

Petit, R.J., Brewer, S., Bordacs, S. 2002a. Identification of refugia and post-glacial colonisation routes of European white oaks based on chloroplast DNA and fossil pollen evidence. Forest Ecology Management 156: 49–74.

Petit, R.J., Csaikl, U.M., Bordacs, S. 2002b. Chloroplast DNA variation in European white oaks-phylogeography and patterns of diversity based on data from over 2600 populations. Forest Ecology Management 156: 5–26.

Petit, R.J., Pineau, E., Demesure,B., Bacilieri, R., Ducousso, A., Kremer, A., 1997. Chloroplast DNA footprints of postglacial recolonization by oaks. Proceedings of the National Academy of Sciences of the United States of America 94: 9996–10001.

Potzger, J.E. 1946. Phytosociology of the primeval forest in central-northern Wisconsin and upper-Michigan, and a brief post-glacial history of the Lake Forest formation. Ecological Monographs 16: 211-250.

Powell, A.H., and G.V.N. Powell. 1987. Population dynamics of male euglossine bees in Amazonian forest fragements. Biotropica 19: 176-179.

Prevec, R., and W.C. Noble. 1983. Historic Neutral Iroquois faunal utilization. Ontario Archaeology 39: 41-56.

Pritchard, J.K., Stephens, M., Donnelly, P. 2000a. Inference of population structure using multilocus genotype data. Genetics 155: 945–959.

Rachmayanti, Y., Leinemann, L., Gailing, O., Finkeldey, R. 2006. Extraction, amplification, and characterization of wood DNA from dipterocarpaceae. Plant Molecular Biology Reporter 24: 45-55

Raffl, C., Schoenswetter, P., Erschbamer, B. 2006. 'Sax-sess'—genetics of primary succession in pioneer species on two parallel glacier forelands. Molecular Ecology 15: 2433–2440.

Rathcke, B.J., and E.S. Jules, E.S. 1993. Habitat fragmentation and plant-pollinator interactions. Current Science 65: 273-277.

Raymond, M. and F. Rousset. 1995. GENEPOP (version 3.3): population genetics software for exact test and ecumenism. Journal of Heredity 86: 248-249.

Rehfeldt, G.E. 1983a. Seed transfer in the northern Rocky Mountains. In Proceedings of the USDA Forest Service Forest Genetics Workshop, Charlottetown, SC, 1-26.

Romanya J., Khanna, P.K., Raison, R.J. 1994. Effects of slash burning on soil phosphorus and sorption and desorption of phosphorus. Forest Ecology and Management 65 : 89-103.

Rosas-Guerrero, V., Quesada, M., Armbruster, W.S., Pérez-Barrales, R., De Witt, S.S. 2011. Influence of pollination specialization and breeding system on floral integration and phenotypic variation in Ipomoea. Evolution 65: 350–364.

Rosenberg, N.A. 2004. DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes 4: 137-138.

Rotach, P. 2005. In situ conservation methods. In Geburek, T., Turok, J. (eds.), Conservation and Management of Forest Genetic Resources in Europe. Arbora Publishers, Zvolen, SK, 535–565pp.

Rozen, S. and H. Skaletsky. 2000. Primer3 on the WWW for general users and for biologist programmers. In S. Krawetz, S. Misener (eds.), Bioinformatics methods and protocols: methods in molecular biology, Humana Press, Totowa, NJ, 365-386pp.

Seavey, S.R., and K.S. Bawa. 1986. Late-acting self-incompatibility in angiosperms. The Botanical Review 52: 195-219.

Setsuko, S., Ishida, K., Ueno, S., Tsumura, Y., Tomaru, N. 2007. Population differentiation and gene flow within a metapopulation of a threatened tree, Magnolia stellata (Magnoliaceae). American Journal of Botany 94: 128-136.

Setsuko, S., Nagamitsu, T., Tomaru, N. 2013. Pollen flow and effects of population structure on selfing rates and female and male reproductive successin fragmented Magnolia stellata populations. BMC Ecology 13: 1-12.

Sewell, M.M., Parks, C.R., Chase, M.W. 1996. Intraspecific chloroplast DNA variation and biogeography of North American Liriodendron L. (Magnoliaceae). Evolution 50: 1147-1154.

Sewell, M.M., Qui, Y., Parks, C.R., Chase, M.W. 1993. Genetic evidence for trace paternal transmission of plastids in Liriodendron and Magnolia (Magnoliaceae). American Journal of Botany 80: 854-858.

Shapcott, A., Rakotoarinivo, M., Smith, R.J., Lysaková, G., Fay, M.F., Dransfield, J. 2007. Can we bring Madagasgar’s critically endangered palms back from the brink? Using an understanding of genetics and ecology to guide a conservation and recovery

83 programme for the iconic and critically endangered palm Beccariophoenix madagascariensis. Botanical Journal of the Linnean Society 154: 589-608.

Sih, A. and M.S. Baltus. 1987. Patch size, pollinator behaviour, and pollinator limitation in catnip. Ecology 68: 1679-1690.

Slatkin, M. 1987. Gene flow and the geographic structure of natural populations. Science 236: 787-792.

Smith, H.C. 1990. Silvics of North America, vol. 2. USDA Forest Service bulletin 654. http://www.na.fs.fed.us/spfo/pubs/silvics_manual/Volume_2/magnolia/acuminata

Soltis, D.E., Morris, A.B., McLachlan, J.S., Manos, P.S., Soltis, P.S. 2006. Comparative phylogeography of unglaciated eastern North America. Molecular Ecology 15: 4262- 4293.

Sork, V.L. and P.E. Smouse. 2006. Genetic analysis of landscape connectivity in tree populations. Landscape Ecology 21: 821-836.

Species at Risk in Ontario. 2014. Species at risk in Ontario (SARO) list. (mnr.gov.on.ca/en/Business/Species/2ColumnSubpage/MNR_SAR_CSSR_SARO_LST _EN.html)

Stuckey, R.L. 1993. Phytogeographical outline of aquatic and wetland angiosperms in continental eastern North America. Aquatic Botany 44: 259-301.

Suffling, R., Evans, M., Perera, A. 2003. Presettlement forest in southern Ontario: Ecosystems measured through a cultural prism. The Forestry Chronicle 79: 485-501.

Thien, L.B. 1974. Floral biology of Magnolia. American Journal of Botany 61: 1037- 1045.

Tremblay, R.L. and J.D. Ackerman. 2001. Gene flow and effective population size in Lepanthes (Orchidacea): a case for genetic drift. Biological Journal of the Linnean Society 72: 47-62.

Tremblay, R.L. and M.J. Hutchings. 2003. Chapter 10: Population dynamics in orchid conservation: a review of analytical methods, based on the rare species Lepanthes eltoroensis. In Dixon, K., Kell, S., Barrett, R., Cribb, P. (eds) Orchid conservation. Natural History Publications (Borneo), Kota Kinabalu, Sabah, Malaysia, 183–204pp.

Turner, B.L., Kasperson, R.E., Matson, P.A., McCarthy, J.J., Corell, R.W., Christensen, L., Eckley, N., Kasperson, J.X., Luers, A., Martello, M.L., Polsky, C., Pulsiphera, A., Schiller,

A. 2003. A framework for vulnerability analysis in sustainability science. Proceedings of the National Academy of Sciences of the United States of America 100: 8074-8079.

Turner, M.G., Pearson, S.M., Bolstad, P., Wear, D.N. 2003. Effects of land-cover change on spatial pattern of forest communities in the Southern Appalachian Mountains (USA). Landscape Ecology 18: 449-464.

UTRCA. 1998. The Thames River watershed: A background study for nomination under the Canadian River Heritage Rivers system. Upper Thames River Conservation Authority Publication, London, ON. 1-162pp.

Vachon, N., and J.R. Freeland. 2011. Phylogeographic inferences from chloroplast DNA: quantifying the effects of mutations in repetitive and non-repetitive sequences. Molecular Ecology Resources 11: 279-285.

Vandermast, D.B., Moorman, C.E., Russell, K.R. 2004. Initial vegetation response to prescribed fire in some oak-hickory forests of the South Carolina piedmont. Natural Areas Journal 24: 216-222.

Vranckx, G., Jacquemyn, H., Mergeay, J., Cox, K., Kint, V., Muys, B., Honnay, O. 2014. Transmission of genetic variation from the adult generation to naturally established seedling cohorts in small forest stands of pendunculate oak (Quercus robar L.). Forest Ecology and Management 312: 19-27.

Wagner, V., Durka, W., Hensen, I. 2011. Increased genetic differentiation but no reduced genetic diversity in peripheral vs. Central populations of a steppe grass. American Journal of Botany 98 : 1173-1179.

Waldron, G. 1993. Trees of the Carolinian forest: A guide to species, their ecology, and uses. Boston Mills Press. Erin, Ontario. 274p.

Watts, W.A. and B.C.S. Hansen. 1994. Pre-holocene and Holocene pollen records of vegetation history form the Florida peninsula and their climatic implications. Palaeogeography, Palaeoclimatology, Palaeoecology 109: 163-176.

Wilson, M.F. 1993. Animals as seed-dispersal mutualists in North America. Oikos 67: 159-176.

Wilson, M.F., and A. Traveset. 2000. The ecology of seed dispersal. In Seeds: the ecology of regeneration in plant communities, 2nd ed., (ed. M. Fenner). CAB International.

Wilson, M.F., and C.J. Whelan. 1990. Variation in postdispersal survival of vertebrate- dispersed seeds: effects of density, habitat, location, season, and species. Oikos 57: 191-198.

Winslow, M. 2008. Lake Champlain: a natural history. Lake Champlain Committee Publication. Burlington, Vermont. 150p.

Woodworth, L.M., Montgomery, M.E., Briscoe, D.A., Frankham, R. 2002. Rapid deterioration in captive populations: causes and conservation implications. Conservation genetics 3: 277-288.

Worth, C.B. 1975. Virginia opossums (Didelphis virginiana) as disseminators of the common persimmon (Diospyros virginiana). Journal of Mammalogy 56: 517.

Wright, S. 1964. The distribution of self-incompatibility alleles in populations. Evolution 18: 609-619.

Yang, R.C., and F.C. Yeh. 1992. Genetic consequences of in-situ and ex-situ conservation of forest trees. The Forestry Chronicle 68: 720–729.

Yasukawa, S., Kato, H., Yamaoka, R., Tanaka, H., Arai, H., Kawano, S. 1992. Reproductive biology and pollination of magnolia and its allied genera (Magnoliceae). I. Floral volatiles of several Magnolia and Michella species and their roles in attracting insects. Plant Species Biology 7: 121-140.

Yeatman, C.W. 1967. Biogeography of Jack pine. Canadian Journal of Botany 45: 2201-2211.

Ying, C.C., and A.D. Yanchuk. 2006. The development of British Columbia’s tree seed transfer guidelines: Purpose, concept, methodology, and implementation. Forest Ecology and Management 227: 1-13.

Young, A., Boyle, T., Brown, A.H.D. 1996. The population genetic consequences of habitat fragmentation for plants. Trends in Ecological Evolution 11: 413-418.

APPENDIX I- Distribution and number of genotyped Magnolia acuminata across Ontario and the United States.

Region Sample Site Associated Site/Sample No. of Location Abbrev. Genotyped Individuals Baker Tract Norfolk BT 9 Bruce Smith Transport/ Norfolk CT 17 Adjacent Cemetery Smith Tract/ Smith Tract Norfolk ST 40 Roadside Shining Tree Woods Norfolk SW 24 Vandehei Property Norfolk VD 5 ONTARIO Watt Property Norfolk WATT 8 National Wildlife Area Norfolk NWA (1-4) 51 Floral Dimension Niagara FD 26 Wright Property Niagara RW 6 Welland Roadside Niagara WD 2 Peninsula Lakes G.C Niagara PL 1 St. Amand Property Niagara SA 4 Beaver Lake Nature Centre, New York NY 5 Baldwinsville Clear Creek, Hocking Ohio CC 5 County Pochahontas County W Virginia POCO 5 Rock Castle Creek, Patrick Virginia RCC 5 County Otto, Macon County N Carolina OTTO 2 Swain County N Carolina SWA 4 Pickens County S Carolina PICK 3 UNITED Sky Valley, Rabun County Georgia RAB (1-4) 4 STATES Rabun County Georgia RAB (5-9) 4 Cane Creek Canyon Nature Alabama COLB 4 Preserve, Colbert County Berrys Cove, Jackson Alabama JACK 5 County Lost Valley, Newton Arkansas NEW 5 Rich Mountain, Polk Arkansas POLK 5 Saline Arkansas SAL 5 Columbia Louisiana COLLA 5 Florida Florida FL 3

APPENDIX II- Map of the distribution of United States Magnolia acuminate samples

APPENDIX III-A: Raw Data: Genotyped Ontario Magnolia acuminata

APPENDIX III-A: CONTINUED.

APPENDIX III-B: Raw Data: Genotyped United States Magnolia acuminata

APPENDIX III-B CONTINUED.

100

APPENDIX IV-A: Comparing allelic diversity between Magnolia acuminata from Ontario and the United States.

APPENDIX IV-A: CONTINUED.

101

102

APPENDIX IV-B: Comparing allelic diversity between Magnolia acuminata from Norfolk County and the Municipality of Niagara.

103

APPENDIX IV-B: CONTINUED.

104

APPENDIX IV-C: Comparing allelic diversity between Magnolia acuminata from Norfolk County and the United States.

105

APPENDIX IV-C: CONTINUED.

106

APPENDIX IV-D: Comparing allelic diversity between Magnolia acuminata from the Municipality of Niagara and the United States.

107

APPENDIX IV-D: CONTINUED.

108

APPENDIX V- A: STRUCTURE Population cluster assignment values

Site No. 1 2 3 4 5 6 7 1 0.039 0.323 0.054 0.168 0.032 0.322 0.062 2 0.011 0.023 0.017 0.016 0.893 0.025 0.017 3 0.040 0.155 0.079 0.566 0.040 0.081 0.040 4 0.016 0.121 0.032 0.019 0.025 0.772 0.016 5 0.013 0.367 0.045 0.160 0.022 0.064 0.329 6 0.015 0.099 0.038 0.041 0.017 0.022 0.767 7 0.021 0.069 0.810 0.034 0.029 0.026 0.043 8 0.054 0.472 0.250 0.054 0.035 0.063 0.074 9 0.772 0.083 0.068 0.022 0.021 0.022 0.016 10 0.062 0.870 0.012 0.007 0.030 0.010 0.009 11 0.243 0.579 0.060 0.021 0.007 0.042 0.048 12 0.047 0.252 0.634 0.010 0.032 0.013 0.012

109

APPENDIX V-B: STRUCTURE individual cluster assignment values Sample Site Sample 1 2 3 4 5 6 7 No. No. 1 BT1 1 0.023 0.388 0.081 0.071 0.008 0.313 0.117 2 BT3 1 0.012 0.057 0.023 0.118 0.018 0.744 0.027 3 BT5 1 0.035 0.398 0.107 0.274 0.144 0.033 0.009 4 BT7 1 0.025 0.134 0.048 0.179 0.052 0.546 0.017 5 BT9 1 0.078 0.197 0.026 0.133 0.023 0.509 0.034 6 BT10 1 0.021 0.627 0.023 0.233 0.011 0.073 0.012 7 BT11 1 0.090 0.314 0.048 0.135 0.009 0.213 0.191 8 BT13 1 0.023 0.182 0.037 0.222 0.014 0.391 0.131 9 BT14 1 0.040 0.610 0.095 0.145 0.013 0.075 0.022 10 CT1 2 0.028 0.106 0.014 0.042 0.651 0.039 0.119 11 CT2 2 0.019 0.011 0.007 0.008 0.936 0.010 0.010 12 CT4 2 0.011 0.014 0.021 0.008 0.925 0.012 0.008 13 CT5 2 0.007 0.007 0.006 0.005 0.964 0.006 0.005 14 CT6 2 0.004 0.007 0.009 0.006 0.960 0.007 0.006 15 CT7 2 0.008 0.036 0.027 0.042 0.840 0.033 0.013 16 CT8 2 0.007 0.007 0.007 0.005 0.961 0.007 0.006 17 CT9 2 0.007 0.010 0.010 0.012 0.943 0.010 0.009 18 CT10 2 0.009 0.017 0.021 0.009 0.910 0.019 0.015 19 CT11 2 0.006 0.019 0.015 0.006 0.934 0.008 0.012 20 CT12 2 0.005 0.006 0.005 0.005 0.968 0.006 0.005 21 CT14 2 0.040 0.026 0.015 0.014 0.754 0.143 0.009 22 CT15 2 0.011 0.015 0.010 0.011 0.908 0.026 0.019 23 CT16 2 0.009 0.012 0.009 0.011 0.938 0.013 0.008 24 CT17 2 0.005 0.039 0.019 0.005 0.916 0.012 0.005 25 CT18 2 0.004 0.006 0.007 0.008 0.961 0.007 0.006 26 CT19 2 0.006 0.049 0.095 0.078 0.711 0.030 0.030 27 ST1 3 0.055 0.031 0.011 0.778 0.009 0.106 0.009 28 ST2 3 0.049 0.029 0.019 0.858 0.010 0.027 0.008 29 ST3 3 0.135 0.107 0.069 0.115 0.028 0.487 0.059 30 ST4 3 0.143 0.143 0.142 0.143 0.143 0.142 0.143 31 ST5 3 0.018 0.023 0.016 0.910 0.007 0.014 0.011 32 ST6 3 0.091 0.084 0.026 0.538 0.023 0.191 0.046 33 ST7 3 0.008 0.076 0.346 0.465 0.015 0.030 0.060 34 ST8 3 0.008 0.051 0.013 0.830 0.007 0.081 0.011 35 ST9 3 0.007 0.009 0.006 0.954 0.004 0.009 0.011 36 ST10 3 0.008 0.589 0.024 0.088 0.018 0.259 0.015 37 ST11 3 0.008 0.020 0.012 0.897 0.006 0.015 0.042 38 ST12 3 0.016 0.094 0.052 0.666 0.012 0.137 0.023 39 ST13 3 0.052 0.108 0.025 0.429 0.010 0.084 0.292

110

40 ST14 3 0.028 0.527 0.095 0.119 0.037 0.148 0.046 41 ST15 3 0.090 0.143 0.087 0.385 0.025 0.207 0.063 42 ST16 3 0.169 0.066 0.683 0.017 0.009 0.031 0.026 43 ST17 3 0.025 0.428 0.239 0.019 0.025 0.252 0.011 44 ST18 3 0.016 0.205 0.067 0.044 0.605 0.028 0.035 45 ST19 3 0.196 0.122 0.038 0.554 0.010 0.046 0.033 46 ST20 3 0.088 0.198 0.225 0.064 0.383 0.031 0.009 47 ST21 3 0.098 0.147 0.322 0.369 0.034 0.023 0.008 48 ST22 3 0.014 0.063 0.028 0.816 0.006 0.048 0.024 49 ST23 3 0.004 0.008 0.007 0.957 0.010 0.008 0.006 50 ST24 3 0.004 0.008 0.008 0.958 0.009 0.007 0.006 51 ST25 3 0.006 0.013 0.009 0.947 0.008 0.010 0.008 52 ST26 3 0.004 0.209 0.023 0.731 0.010 0.015 0.007 53 ST27 3 0.004 0.006 0.007 0.963 0.004 0.007 0.008 54 ST28 3 0.004 0.006 0.005 0.967 0.004 0.006 0.008 55 ST29 3 0.011 0.017 0.007 0.915 0.017 0.024 0.010 56 ST30 3 0.008 0.010 0.010 0.942 0.006 0.007 0.016 57 ST31 3 0.015 0.024 0.013 0.876 0.008 0.037 0.027 58 ST32 3 0.032 0.027 0.014 0.869 0.005 0.009 0.043 59 ST33 3 0.060 0.097 0.022 0.731 0.010 0.025 0.056 60 ST34 3 0.009 0.034 0.042 0.854 0.007 0.036 0.019 61 ST35 3 0.058 0.351 0.078 0.055 0.006 0.447 0.005 62 ST36 3 0.008 0.027 0.017 0.883 0.007 0.029 0.029 63 ST37 3 0.015 0.018 0.013 0.882 0.006 0.007 0.059 64 ST38 3 0.006 0.861 0.034 0.035 0.011 0.042 0.012 65 ST39 3 0.023 0.075 0.075 0.530 0.010 0.018 0.268 66 ST40 3 0.017 0.257 0.219 0.393 0.014 0.083 0.018 67 STR 3 0.008 0.908 0.018 0.015 0.008 0.019 0.024 68 SW1 4 0.012 0.015 0.008 0.012 0.006 0.938 0.010 69 SW2 4 0.018 0.010 0.007 0.005 0.009 0.942 0.008 70 SW3 4 0.010 0.023 0.020 0.016 0.006 0.913 0.012 71 SW4 4 0.015 0.027 0.018 0.088 0.005 0.814 0.033 72 SW5 4 0.012 0.019 0.011 0.014 0.005 0.928 0.011 73 SW6 4 0.012 0.023 0.012 0.005 0.022 0.918 0.008 74 SW7 4 0.011 0.022 0.024 0.010 0.011 0.908 0.013 75 SW8 4 0.085 0.018 0.011 0.012 0.009 0.854 0.011 76 SW9 4 0.011 0.137 0.162 0.014 0.018 0.641 0.018 77 SW10 4 0.005 0.599 0.064 0.021 0.013 0.270 0.028 78 SW11 4 0.011 0.018 0.017 0.007 0.010 0.931 0.006 79 SW12 4 0.008 0.044 0.010 0.015 0.028 0.887 0.008 80 SW13 4 0.008 0.073 0.011 0.015 0.008 0.875 0.010 81 SW14 4 0.014 0.016 0.013 0.076 0.006 0.865 0.010

111

82 SW15 4 0.008 0.054 0.018 0.019 0.021 0.868 0.011 83 SW16 4 0.006 0.008 0.006 0.008 0.009 0.956 0.006 84 SW17 4 0.011 0.008 0.006 0.005 0.009 0.954 0.006 85 SW18 4 0.014 0.038 0.013 0.014 0.009 0.896 0.015 86 SW19 4 0.008 0.894 0.009 0.013 0.054 0.017 0.005 87 SW20 4 0.021 0.017 0.010 0.011 0.008 0.920 0.013 88 SW21 4 0.023 0.090 0.078 0.027 0.317 0.452 0.011 89 SW22 4 0.031 0.250 0.068 0.016 0.007 0.535 0.092 90 SW23 4 0.013 0.013 0.006 0.018 0.006 0.937 0.007 91 SW24 4 0.017 0.488 0.159 0.012 0.005 0.300 0.020 92 VD_1 5 0.008 0.092 0.066 0.416 0.013 0.018 0.387 93 VD_2 5 0.022 0.605 0.045 0.126 0.020 0.098 0.085 94 VD_4 5 0.011 0.745 0.038 0.061 0.026 0.099 0.019 95 VD_5 5 0.006 0.025 0.023 0.050 0.016 0.020 0.860 96 VD_6 5 0.018 0.370 0.052 0.145 0.036 0.083 0.295 97 NWA1_1 6 0.027 0.733 0.054 0.014 0.022 0.128 0.021 98 NWA1_2 6 0.007 0.092 0.069 0.030 0.049 0.008 0.746 99 NWA1_3 6 0.008 0.078 0.016 0.031 0.011 0.032 0.824 100 NWA1_4 6 0.007 0.014 0.010 0.020 0.006 0.010 0.933 101 NWA1_5 6 0.005 0.012 0.021 0.011 0.011 0.008 0.933 102 NWA1_6 6 0.010 0.009 0.009 0.014 0.004 0.011 0.943 103 NWA1_7 6 0.005 0.013 0.009 0.006 0.006 0.005 0.956 104 NWA1_8 6 0.006 0.011 0.009 0.007 0.008 0.006 0.954 105 NWA1_9 6 0.016 0.568 0.027 0.085 0.024 0.020 0.260 106 NWA1_10 6 0.008 0.010 0.009 0.008 0.006 0.008 0.950 107 NWA1_11 6 0.006 0.013 0.011 0.011 0.012 0.008 0.938 108 NWA1_12 6 0.008 0.008 0.008 0.007 0.007 0.006 0.957 109 NWA1_13 6 0.010 0.013 0.012 0.007 0.007 0.012 0.938 110 NWA1_14 6 0.018 0.021 0.022 0.014 0.007 0.012 0.905 111 NWA1_15 6 0.018 0.021 0.012 0.029 0.010 0.021 0.889 112 NWA2_2 6 0.006 0.054 0.014 0.129 0.007 0.013 0.777 113 NWA3_1 6 0.129 0.118 0.030 0.023 0.007 0.060 0.633 114 NWA3_2 6 0.005 0.402 0.016 0.041 0.017 0.159 0.359 115 NWA3_3 6 0.016 0.172 0.100 0.187 0.189 0.020 0.315 116 NWA3_4 6 0.017 0.032 0.082 0.023 0.011 0.013 0.821 117 NWA3_5 6 0.019 0.167 0.078 0.202 0.010 0.012 0.512 118 NWA3_6 6 0.011 0.225 0.013 0.046 0.035 0.043 0.626 119 NWA3_7 6 0.010 0.513 0.010 0.008 0.019 0.009 0.430 120 NWA3_8 6 0.009 0.031 0.028 0.006 0.009 0.009 0.908 121 NWA3_9 6 0.008 0.026 0.011 0.008 0.104 0.006 0.837 122 NWA3_10 6 0.007 0.017 0.059 0.009 0.021 0.007 0.880 123 NWA3_11 6 0.006 0.074 0.060 0.042 0.006 0.012 0.800

112

124 NWA3_12 6 0.012 0.043 0.015 0.009 0.008 0.016 0.896 125 NWA4_1 6 0.031 0.048 0.126 0.023 0.011 0.034 0.727 126 NWA4_2 6 0.019 0.022 0.025 0.012 0.007 0.047 0.867 127 NWA4_3 6 0.027 0.843 0.052 0.013 0.008 0.014 0.044 128 NWA4_4 6 0.004 0.008 0.011 0.013 0.007 0.007 0.949 129 NWA4_5 6 0.009 0.015 0.011 0.058 0.007 0.010 0.889 130 NWA4_6 6 0.029 0.020 0.017 0.041 0.007 0.015 0.870 131 NWA4_7 6 0.006 0.012 0.009 0.009 0.009 0.011 0.945 132 NWA4_8 6 0.067 0.060 0.017 0.019 0.008 0.034 0.796 133 NWA4_9 6 0.007 0.014 0.009 0.022 0.006 0.011 0.932 134 NWA4_10 6 0.010 0.028 0.023 0.010 0.011 0.010 0.908 135 NWA4_11 6 0.008 0.012 0.018 0.015 0.007 0.007 0.933 136 NWA4_12 6 0.009 0.102 0.356 0.306 0.012 0.062 0.154 137 NWA4_13 6 0.008 0.009 0.009 0.014 0.006 0.009 0.944 138 NWA4_14 6 0.005 0.016 0.017 0.022 0.010 0.012 0.917 139 NWA4_15 6 0.009 0.021 0.038 0.055 0.013 0.020 0.843 140 NWA4_16 6 0.012 0.054 0.055 0.058 0.014 0.021 0.786 141 NWA4_17 6 0.014 0.057 0.045 0.075 0.013 0.020 0.776 142 NWA4_18 6 0.009 0.022 0.018 0.209 0.005 0.020 0.718 143 NWA4_19 6 0.011 0.023 0.028 0.006 0.008 0.009 0.914 144 NWA4_20 6 0.007 0.012 0.010 0.015 0.007 0.008 0.941 145 NWA4_21 6 0.022 0.118 0.063 0.036 0.043 0.077 0.642 146 NWA4_22 6 0.004 0.013 0.025 0.009 0.005 0.006 0.938 147 NWA4_23 6 0.012 0.042 0.130 0.037 0.008 0.012 0.760 148 WATT1 7 0.143 0.143 0.144 0.143 0.142 0.142 0.142 149 WATT2 7 0.008 0.028 0.901 0.031 0.010 0.010 0.013 150 WATT3 7 0.007 0.012 0.937 0.005 0.009 0.007 0.023 151 WATT4 7 0.007 0.014 0.926 0.011 0.010 0.011 0.021 152 WATT5 7 0.004 0.038 0.820 0.034 0.009 0.011 0.085 153 WATT6 7 0.007 0.039 0.865 0.046 0.009 0.009 0.026 154 WATT7 7 0.004 0.031 0.854 0.011 0.049 0.021 0.029 155 WATT8 7 0.006 0.013 0.927 0.011 0.008 0.010 0.024 156 WATT9 7 0.006 0.017 0.919 0.011 0.011 0.011 0.026 157 SA_1 8 0.103 0.305 0.342 0.068 0.015 0.152 0.015 158 SA_2 8 0.086 0.329 0.187 0.071 0.021 0.054 0.253 159 SA_3 8 0.020 0.556 0.342 0.042 0.008 0.019 0.014 160 SA_4 8 0.008 0.696 0.127 0.034 0.095 0.025 0.014 161 FD_1 9 0.764 0.076 0.071 0.040 0.006 0.016 0.026 162 FD_2 9 0.946 0.018 0.008 0.008 0.004 0.007 0.009 163 FD_3 9 0.311 0.298 0.025 0.254 0.008 0.053 0.051 164 FD_4 9 0.018 0.038 0.884 0.009 0.020 0.019 0.011 165 FD_5 9 0.949 0.009 0.011 0.007 0.005 0.012 0.008

113

166 FD_6 9 0.358 0.086 0.432 0.026 0.011 0.054 0.034 167 FD_7 9 0.954 0.008 0.006 0.007 0.008 0.011 0.006 168 FD_8 9 0.803 0.044 0.021 0.026 0.006 0.038 0.061 169 FD_9 9 0.680 0.055 0.013 0.013 0.133 0.093 0.012 170 FD_10 9 0.957 0.008 0.013 0.006 0.005 0.007 0.005 171 FD_11 9 0.904 0.015 0.017 0.011 0.009 0.026 0.018 172 FD_12 9 0.962 0.007 0.006 0.006 0.006 0.008 0.006 173 FD_13 9 0.951 0.009 0.006 0.007 0.011 0.010 0.006 174 FD_14 9 0.960 0.007 0.007 0.007 0.005 0.008 0.007 175 FD_15 9 0.951 0.009 0.006 0.007 0.010 0.009 0.007 176 FD_16 9 0.950 0.009 0.007 0.012 0.004 0.008 0.010 177 FD_18 9 0.963 0.006 0.006 0.006 0.006 0.009 0.005 178 FD_19 9 0.913 0.014 0.006 0.007 0.046 0.007 0.007 179 FD_20 9 0.935 0.021 0.008 0.008 0.006 0.017 0.007 180 FD_21 9 0.578 0.116 0.141 0.010 0.120 0.018 0.017 181 FD_23 9 0.826 0.033 0.009 0.022 0.060 0.029 0.020 182 FD_24 9 0.882 0.030 0.013 0.038 0.005 0.017 0.015 183 FD_25 9 0.938 0.012 0.017 0.007 0.006 0.011 0.009 184 FD_26 9 0.144 0.746 0.011 0.015 0.021 0.054 0.010 185 FD_27 9 0.603 0.338 0.020 0.010 0.012 0.011 0.006 186 FD_29 9 0.884 0.030 0.009 0.012 0.012 0.011 0.042 187 WD_1 10 0.084 0.832 0.015 0.007 0.041 0.011 0.010 188 WD_2 10 0.040 0.908 0.009 0.006 0.019 0.009 0.008 189 PL 11 0.243 0.579 0.060 0.021 0.007 0.042 0.048 190 RW_1 12 0.011 0.812 0.027 0.004 0.109 0.011 0.026 191 RW_2 12 0.228 0.596 0.059 0.018 0.056 0.032 0.011 192 RW_3 12 0.020 0.026 0.918 0.008 0.010 0.010 0.008 193 RW_4 12 0.008 0.019 0.945 0.006 0.006 0.008 0.008 194 RW_5 12 0.006 0.025 0.933 0.013 0.005 0.007 0.011 195 RN 12 0.009 0.031 0.920 0.013 0.008 0.012 0.008

114

APPENDIX V-C: TESS Population cluster assignment values

Site No. 1 2 3 4 5 6 7 1 0.1032 0.0197 0.7496 0.0405 0.0223 0.0544 0.0103 2 0.0178 0.0060 0.0233 0.0255 0.9117 0.0083 0.0073 3 0.0835 0.0096 0.7974 0.0354 0.0341 0.0149 0.0252 4 0.9020 0.0040 0.0231 0.0278 0.0209 0.0171 0.0051 5 0.0418 0.0101 0.6643 0.0554 0.0493 0.1691 0.0099 6 0.0166 0.0052 0.0114 0.0082 0.0045 0.9359 0.0182 7 0.0327 0.0065 0.0460 0.8506 0.0483 0.0037 0.0121 8 0.0138 0.2491 0.0591 0.0867 0.0312 0.0595 0.5006 9 0.0029 0.8726 0.0074 0.0052 0.0066 0.0074 0.0979 10 0.0022 0.4011 0.0085 0.0090 0.0168 0.0168 0.5455 11 0.0080 0.4254 0.0319 0.0157 0.0133 0.0415 0.4642 12 0.0028 0.1251 0.0072 0.0094 0.0127 0.0120 0.8307

115

APPENDIX V-D: TESS individual cluster assignment values

Sample Site Sample 1 2 3 4 5 6 7 No. No. 1 BT1 1 0.1280 0.0210 0.6154 0.0809 0.0140 0.1313 0.0094 2 BT3 1 0.2150 0.0141 0.6964 0.0191 0.0173 0.0307 0.0073 3 BT5 1 0.0421 0.0215 0.8066 0.0485 0.0511 0.0181 0.0120 4 BT7 1 0.1090 0.0131 0.7903 0.0227 0.0301 0.0263 0.0087 5 BT9 1 0.0854 0.0201 0.8053 0.0128 0.0253 0.0447 0.0064 6 BT10 1 0.0449 0.0148 0.8694 0.0242 0.0139 0.0215 0.0113 7 BT11 1 0.0824 0.0253 0.7805 0.0328 0.0139 0.0588 0.0064 8 BT13 1 0.1269 0.0220 0.6824 0.0429 0.0191 0.0918 0.0149 9 BT14 1 0.0947 0.0259 0.7002 0.0802 0.0161 0.0661 0.0168 10 CT1 2 0.0541 0.0156 0.0801 0.0807 0.7346 0.0257 0.0092 11 CT2 2 0.0140 0.0083 0.0145 0.0109 0.9400 0.0072 0.0051 12 CT4 2 0.0125 0.0052 0.0134 0.0201 0.9339 0.0056 0.0093 13 CT5 2 0.0062 0.0040 0.0075 0.0101 0.9648 0.0032 0.0041 14 CT6 2 0.0086 0.0026 0.0125 0.0143 0.9540 0.0042 0.0038 15 CT7 2 0.0304 0.0058 0.0557 0.0223 0.8623 0.0107 0.0127 16 CT8 2 0.0072 0.0040 0.0109 0.0086 0.9608 0.0041 0.0044 17 CT9 2 0.0106 0.0043 0.0231 0.0147 0.9349 0.0062 0.0062 18 CT10 2 0.0177 0.0058 0.0164 0.0217 0.9186 0.0093 0.0104 19 CT11 2 0.0126 0.0045 0.0237 0.0209 0.9263 0.0055 0.0065 20 CT12 2 0.0115 0.0065 0.0194 0.0175 0.9321 0.0069 0.0061 21 CT14 2 0.0360 0.0095 0.0192 0.0332 0.8884 0.0075 0.0062 22 CT15 2 0.0177 0.0064 0.0119 0.0223 0.9255 0.0097 0.0065 23 CT16 2 0.0208 0.0109 0.0209 0.0489 0.8769 0.0106 0.0110 24 CT17 2 0.0106 0.0030 0.0066 0.0253 0.9408 0.0037 0.0100 25 CT18 2 0.0087 0.0022 0.0149 0.0103 0.9553 0.0048 0.0037 26 CT19 2 0.0228 0.0037 0.0462 0.0525 0.8494 0.0160 0.0093 27 ST1 3 0.0687 0.0059 0.8975 0.0097 0.0104 0.0044 0.0034 28 ST2 3 0.0593 0.0035 0.9237 0.0029 0.0064 0.0017 0.0025 29 ST3 3 0.2134 0.0111 0.7062 0.0240 0.0256 0.0093 0.0104 30 ST4 3 0.1515 0.0033 0.8167 0.0094 0.0127 0.0030 0.0034 31 ST5 3 0.1885 0.0001 0.8104 0.0002 0.0005 0.0001 0.0001 32 ST6 3 0.1237 0.0039 0.8501 0.0088 0.0084 0.0029 0.0021 33 ST7 3 0.0308 0.0018 0.9272 0.0202 0.0083 0.0048 0.0069 34 ST8 3 0.0290 0.0022 0.9510 0.0061 0.0056 0.0032 0.0029 35 ST9 3 0.0211 0.0026 0.9577 0.0064 0.0046 0.0046 0.0030 36 ST10 3 0.0972 0.0108 0.7973 0.0203 0.0391 0.0117 0.0236 37 ST11 3 0.0291 0.0046 0.9268 0.0110 0.0081 0.0153 0.0051 38 ST12 3 0.1179 0.0075 0.8101 0.0289 0.0161 0.0122 0.0074

116

39 ST13 3 0.1135 0.0231 0.7381 0.0378 0.0242 0.0450 0.0183 40 ST14 3 0.1612 0.0211 0.4117 0.2248 0.0663 0.0392 0.0757 41 ST15 3 0.1264 0.0176 0.7242 0.0439 0.0402 0.0212 0.0264 42 ST16 3 0.1975 0.0657 0.2237 0.1343 0.0328 0.0515 0.2946 43 ST17 3 0.4023 0.0368 0.1147 0.1995 0.0594 0.0230 0.1644 44 ST18 3 0.0870 0.0145 0.2452 0.1640 0.3984 0.0353 0.0554 45 ST19 3 0.0888 0.0349 0.8099 0.0174 0.0180 0.0140 0.0170 46 ST20 3 0.0874 0.0108 0.6854 0.0416 0.1355 0.0068 0.0325 47 ST21 3 0.1005 0.0058 0.8458 0.0224 0.0164 0.0021 0.0070 48 ST22 3 0.0669 0.0074 0.8658 0.0189 0.0121 0.0161 0.0130 49 ST23 3 0.0136 0.0009 0.9730 0.0044 0.0049 0.0018 0.0013 50 ST24 3 0.0154 0.0020 0.9549 0.0076 0.0104 0.0051 0.0047 51 ST25 3 0.0173 0.0021 0.9592 0.0064 0.0065 0.0040 0.0045 52 ST26 3 0.0394 0.0031 0.8929 0.0299 0.0166 0.0069 0.0113 53 ST27 3 0.0111 0.0013 0.9741 0.0046 0.0030 0.0039 0.0020 54 ST28 3 0.0092 0.0011 0.9783 0.0034 0.0023 0.0038 0.0019 55 ST29 3 0.0238 0.0019 0.9597 0.0028 0.0066 0.0032 0.0020 56 ST30 3 0.0112 0.0017 0.9716 0.0034 0.0047 0.0046 0.0028 57 ST31 3 0.0342 0.0034 0.9398 0.0046 0.0065 0.0074 0.0042 58 ST32 3 0.0207 0.0015 0.9669 0.0022 0.0045 0.0029 0.0013 59 ST33 3 0.0343 0.0061 0.9256 0.0093 0.0098 0.0086 0.0062 60 ST34 3 0.0350 0.0047 0.9039 0.0122 0.0148 0.0172 0.0122 61 ST35 3 0.1337 0.0200 0.7801 0.0206 0.0130 0.0057 0.0270 62 ST36 3 0.0377 0.0030 0.9268 0.0079 0.0069 0.0138 0.0039 63 ST37 3 0.0125 0.0040 0.9537 0.0062 0.0061 0.0138 0.0036 64 ST38 3 0.0895 0.0069 0.7530 0.0758 0.0381 0.0133 0.0235 65 ST39 3 0.0358 0.0110 0.8034 0.0329 0.0240 0.0805 0.0123 66 ST40 3 0.1480 0.0082 0.6424 0.0777 0.0700 0.0207 0.0330 67 STR 3 0.1389 0.0177 0.3935 0.0864 0.1983 0.0660 0.0992 68 SW1 4 0.9188 0.0028 0.0357 0.0154 0.0058 0.0191 0.0023 69 SW2 4 0.9249 0.0045 0.0268 0.0090 0.0130 0.0184 0.0035 70 SW3 4 0.9162 0.0031 0.0386 0.0151 0.0063 0.0163 0.0042 71 SW4 4 0.9027 0.0037 0.0488 0.0116 0.0046 0.0246 0.0041 72 SW5 4 0.9582 0.0030 0.0124 0.0103 0.0032 0.0106 0.0023 73 SW6 4 0.9631 0.0031 0.0077 0.0083 0.0094 0.0060 0.0024 74 SW7 4 0.9492 0.0033 0.0134 0.0107 0.0072 0.0123 0.0039 75 SW8 4 0.9574 0.0052 0.0110 0.0087 0.0066 0.0087 0.0024 76 SW9 4 0.9119 0.0035 0.0153 0.0368 0.0095 0.0164 0.0066 77 SW10 4 0.7946 0.0039 0.0466 0.0720 0.0208 0.0458 0.0163 78 SW11 4 0.9611 0.0020 0.0110 0.0111 0.0068 0.0052 0.0027 79 SW12 4 0.9144 0.0022 0.0223 0.0299 0.0176 0.0117 0.0020 80 SW13 4 0.9412 0.0027 0.0127 0.0201 0.0086 0.0116 0.0032

117

81 SW14 4 0.9275 0.0027 0.0386 0.0125 0.0050 0.0107 0.0031 82 SW15 4 0.9102 0.0026 0.0368 0.0168 0.0158 0.0149 0.0030 83 SW16 4 0.9368 0.0021 0.0195 0.0113 0.0173 0.0105 0.0025 84 SW17 4 0.9607 0.0036 0.0088 0.0081 0.0105 0.0067 0.0016 85 SW18 4 0.9305 0.0032 0.0152 0.0186 0.0111 0.0167 0.0047 86 SW19 4 0.5796 0.0109 0.0354 0.1013 0.2454 0.0125 0.0149 87 SW20 4 0.9433 0.0056 0.0137 0.0116 0.0102 0.0125 0.0031 88 SW21 4 0.8376 0.0058 0.0227 0.0686 0.0440 0.0125 0.0087 89 SW22 4 0.8602 0.0078 0.0216 0.0272 0.0079 0.0635 0.0118 90 SW23 4 0.9443 0.0038 0.0220 0.0114 0.0072 0.0088 0.0025 91 SW24 4 0.8048 0.0047 0.0182 0.1203 0.0073 0.0347 0.0100 92 VD_1 5 0.0207 0.0055 0.7325 0.0521 0.0206 0.1617 0.0070 93 VD_2 5 0.0372 0.0158 0.7849 0.0372 0.0390 0.0759 0.0100 94 VD_4 5 0.0704 0.0147 0.6771 0.0947 0.0769 0.0506 0.0156 95 VD_5 5 0.0374 0.0060 0.4392 0.0508 0.0431 0.4150 0.0085 96 VD_6 5 0.0432 0.0084 0.6879 0.0423 0.0670 0.1425 0.0087 97 NWA1_1 6 0.2053 0.0273 0.0808 0.0091 0.0354 0.6102 0.0320 98 NWA1_2 6 0.0108 0.0036 0.0343 0.0142 0.0159 0.9085 0.0126 99 NWA1_3 6 0.0248 0.0023 0.0172 0.0023 0.0060 0.9436 0.0038 100 NWA1_4 6 0.0123 0.0023 0.0173 0.0018 0.0029 0.9600 0.0034 101 NWA1_5 6 0.0201 0.0022 0.0171 0.0041 0.0048 0.9476 0.0042 102 NWA1_6 6 0.0141 0.0026 0.0178 0.0012 0.0017 0.9594 0.0031 103 NWA1_7 6 0.0083 0.0014 0.0125 0.0012 0.0022 0.9711 0.0033 104 NWA1_8 6 0.0074 0.0022 0.0071 0.0032 0.0049 0.9698 0.0054 105 NWA1_9 6 0.0200 0.0064 0.0294 0.0042 0.0120 0.9063 0.0216 106 NWA1_10 6 0.0089 0.0032 0.0092 0.0024 0.0039 0.9681 0.0043 107 NWA1_11 6 0.0073 0.0015 0.0076 0.0022 0.0049 0.9731 0.0036 108 NWA1_12 6 0.0064 0.0027 0.0057 0.0021 0.0034 0.9758 0.0038 109 NWA1_13 6 0.0163 0.0031 0.0073 0.0029 0.0042 0.9615 0.0047 110 NWA1_14 6 0.0184 0.0026 0.0081 0.0021 0.0022 0.9624 0.0042 111 NWA1_15 6 0.0457 0.0051 0.0203 0.0009 0.0017 0.9216 0.0048 112 NWA2_2 6 0.0102 0.0018 0.0213 0.0043 0.0030 0.9552 0.0043 113 NWA3_1 6 0.0411 0.0044 0.0129 0.0023 0.0028 0.9302 0.0063 114 NWA3_2 6 0.1078 0.0013 0.0267 0.0037 0.0113 0.8433 0.0060 115 NWA3_3 6 0.0194 0.0034 0.0319 0.0150 0.0208 0.8920 0.0175 116 NWA3_4 6 0.0135 0.0035 0.0106 0.0033 0.0052 0.9518 0.0122 117 NWA3_5 6 0.0179 0.0058 0.0315 0.0028 0.0093 0.8995 0.0331 118 NWA3_6 6 0.0279 0.0024 0.0201 0.0033 0.0089 0.9311 0.0063 119 NWA3_7 6 0.0112 0.0046 0.0057 0.0050 0.0102 0.9315 0.0318 120 NWA3_8 6 0.0099 0.0019 0.0034 0.0040 0.0025 0.9707 0.0077 121 NWA3_9 6 0.0054 0.0015 0.0036 0.0018 0.0066 0.9741 0.0070 122 NWA3_10 6 0.0037 0.0009 0.0025 0.0018 0.0014 0.9859 0.0037

118

123 NWA3_11 6 0.0124 0.0014 0.0159 0.0046 0.0026 0.9560 0.0071 124 NWA3_12 6 0.0269 0.0010 0.0076 0.0027 0.0026 0.9541 0.0053 125 NWA4_1 6 0.0049 0.0106 0.0031 0.0114 0.0017 0.9568 0.0116 126 NWA4_2 6 0.0074 0.0035 0.0034 0.0026 0.0018 0.9756 0.0058 127 NWA4_3 6 0.0001 0.0642 0.0003 0.0004 0.0001 0.4123 0.5225 128 NWA4_4 6 0.0013 0.0019 0.0014 0.0049 0.0005 0.9864 0.0036 129 NWA4_5 6 0.0005 0.0048 0.0025 0.0025 0.0004 0.9827 0.0066 130 NWA4_6 6 0.0009 0.0064 0.0013 0.0012 0.0004 0.9837 0.0062 131 NWA4_7 6 0.0008 0.0037 0.0010 0.0030 0.0003 0.9858 0.0054 132 NWA4_8 6 0.0063 0.0247 0.0044 0.0039 0.0012 0.9493 0.0101 133 NWA4_9 6 0.0034 0.0017 0.0036 0.0031 0.0011 0.9833 0.0039 134 NWA4_10 6 0.0014 0.0028 0.0030 0.0063 0.0007 0.9753 0.0105 135 NWA4_11 6 0.0026 0.0024 0.0019 0.0339 0.0005 0.9552 0.0034 136 NWA4_12 6 0.0042 0.0042 0.0018 0.1909 0.0006 0.7906 0.0078 137 NWA4_13 6 0.0016 0.0054 0.0051 0.0020 0.0005 0.9804 0.0050 138 NWA4_14 6 0.0120 0.0014 0.0062 0.0054 0.0030 0.9683 0.0037 139 NWA4_15 6 0.0100 0.0030 0.0087 0.0044 0.0038 0.9642 0.0059 140 NWA4_16 6 0.0088 0.0040 0.0082 0.0045 0.0043 0.9552 0.0150 141 NWA4_17 6 0.0077 0.0025 0.0081 0.0038 0.0033 0.9676 0.0070 142 NWA4_18 6 0.0113 0.0025 0.0115 0.0045 0.0019 0.9617 0.0065 143 NWA4_19 6 0.0087 0.0012 0.0029 0.0022 0.0024 0.9783 0.0043 144 NWA4_20 6 0.0022 0.0019 0.0030 0.0042 0.0009 0.9838 0.0040 145 NWA4_21 6 0.0086 0.0041 0.0059 0.0049 0.0041 0.9651 0.0073 146 NWA4_22 6 0.0035 0.0026 0.0038 0.0054 0.0015 0.9769 0.0063 147 NWA4_23 6 0.0043 0.0021 0.0063 0.0048 0.0013 0.9766 0.0047 148 WATT1 7 0.0831 0.0231 0.1304 0.5864 0.1419 0.0052 0.0300 149 WATT2 7 0.0194 0.0049 0.0384 0.8445 0.0837 0.0021 0.0069 150 WATT3 7 0.0151 0.0041 0.0116 0.9292 0.0305 0.0027 0.0067 151 WATT4 7 0.0159 0.0045 0.0250 0.9021 0.0423 0.0030 0.0073 152 WATT5 7 0.0607 0.0055 0.0613 0.8135 0.0253 0.0067 0.0269 153 WATT6 7 0.0150 0.0045 0.0473 0.8968 0.0252 0.0034 0.0076 154 WATT7 7 0.0337 0.0036 0.0375 0.8658 0.0454 0.0036 0.0104 155 WATT8 7 0.0218 0.0036 0.0208 0.9288 0.0145 0.0042 0.0062 156 WATT9 7 0.0299 0.0050 0.0417 0.8886 0.0255 0.0022 0.0072 157 SA_1 8 0.0334 0.3378 0.0739 0.0593 0.0267 0.0401 0.4290 158 SA_2 8 0.0082 0.3161 0.0602 0.0248 0.0209 0.1332 0.4365 159 SA_3 8 0.0040 0.1490 0.0439 0.0272 0.0116 0.0283 0.7359 160 SA_4 8 0.0097 0.1936 0.0583 0.2353 0.0657 0.0365 0.4009 161 FD_1 9 0.0037 0.6820 0.0152 0.0079 0.0061 0.0123 0.2728 162 FD_2 9 0.0012 0.9482 0.0038 0.0021 0.0026 0.0044 0.0377 163 FD_3 9 0.0069 0.7784 0.0276 0.0089 0.0053 0.0203 0.1526 164 FD_4 9 0.0064 0.1332 0.0109 0.0101 0.0148 0.0146 0.8101

119

165 FD_5 9 0.0036 0.8903 0.0061 0.0052 0.0038 0.0069 0.0842 166 FD_6 9 0.0051 0.4782 0.0161 0.0174 0.0090 0.0165 0.4578 167 FD_7 9 0.0009 0.9570 0.0035 0.0026 0.0040 0.0030 0.0290 168 FD_8 9 0.0036 0.9548 0.0069 0.0062 0.0032 0.0133 0.0121 169 FD_9 9 0.0047 0.9457 0.0043 0.0042 0.0121 0.0069 0.0221 170 FD_10 9 0.0012 0.9639 0.0033 0.0028 0.0028 0.0023 0.0238 171 FD_11 9 0.0016 0.9677 0.0033 0.0028 0.0036 0.0051 0.0160 172 FD_12 9 0.0011 0.9784 0.0032 0.0020 0.0025 0.0027 0.0100 173 FD_13 9 0.0025 0.9735 0.0028 0.0032 0.0052 0.0033 0.0094 174 FD_14 9 0.0015 0.9769 0.0028 0.0019 0.0018 0.0030 0.0122 175 FD_15 9 0.0018 0.9734 0.0051 0.0022 0.0042 0.0040 0.0091 176 FD_16 9 0.0019 0.9743 0.0047 0.0024 0.0024 0.0036 0.0107 177 FD_18 9 0.0013 0.9791 0.0022 0.0018 0.0025 0.0025 0.0106 178 FD_19 9 0.0017 0.9662 0.0034 0.0026 0.0071 0.0033 0.0156 179 FD_20 9 0.0023 0.9505 0.0038 0.0023 0.0031 0.0037 0.0342 180 FD_21 9 0.0033 0.8111 0.0070 0.0088 0.0167 0.0081 0.1449 181 FD_23 9 0.0032 0.9033 0.0113 0.0063 0.0117 0.0122 0.0520 182 FD_24 9 0.0034 0.9241 0.0104 0.0055 0.0041 0.0083 0.0443 183 FD_25 9 0.0019 0.9264 0.0040 0.0039 0.0031 0.0048 0.0560 184 FD_26 9 0.0074 0.8277 0.0201 0.0105 0.0251 0.0152 0.0941 185 FD_27 9 0.0025 0.8755 0.0061 0.0069 0.0096 0.0048 0.0945 186 FD_29 9 0.0021 0.9470 0.0056 0.0043 0.0056 0.0063 0.0291 187 WD_1 10 0.0024 0.5038 0.0093 0.0104 0.0217 0.0186 0.4339 188 WD_2 10 0.0021 0.2983 0.0078 0.0075 0.0119 0.0151 0.6572 189 PL 11 0.0080 0.4254 0.0319 0.0157 0.0133 0.0415 0.4642 190 RW_1 12 0.0026 0.1247 0.0036 0.0128 0.0347 0.0150 0.8065 191 RW_2 12 0.0049 0.3914 0.0114 0.0098 0.0176 0.0114 0.5534 192 RW_3 12 0.0035 0.0594 0.0072 0.0096 0.0078 0.0116 0.9010 193 RW_4 12 0.0022 0.0575 0.0051 0.0080 0.0045 0.0097 0.9129 194 RW_5 12 0.0011 0.0592 0.0082 0.0085 0.0050 0.0128 0.9052 195 RN 12 0.0024 0.0588 0.0075 0.0080 0.0063 0.0117 0.9054