RECREATIONAL TRAIL IMPACTS ON THE PLANT COMMUNITIES OF CASTLE AND CASTLE WILDLAND PROVINCIAL PARKS IN SOUTHERN ALBERTA TRINITAS CHISHOLM Bachelor of Science, University of Lethbridge, 2018 A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE in BIOLOGICAL SCIENCE Department of Biological Sciences University of Lethbridge LETHBRIDGE, ALBERTA, CANADA © Trinitas Evette Chisholm, 2022 RECREATIONAL TRAIL IMPACTS ON THE PLANT COMMUNITIES OF CASTLE AND CASTLE WILDLAND PROVINCIAL PARKS IN SOUTHERN ALBERTA TRINITAS CHISHOLM Date of Defense: November 1, 2022 Dr. J.L. McCune Assistant Professor Ph.D. Thesis Supervisor Dr. J.F. Bain Professor Emeritus Ph.D. Thesis Examination Committee Member Dr. J. Lee-Yaw Assistant Professor Ph.D. Thesis Examination Committee Member Dr. R. Laird Associate Professor Ph.D. Thesis Examination Committee Member Dr. C. Goater Professor Ph.D. Chair, Thesis Examination Committee DEDICATION To anyone reading this. With seemingly insurmountable struggles comes undeniable success in pursuing your passion. Keep trudging up that mountain so you can enjoy the view. iii ABSTRACT I measured the effect of recreational trails on plant species richness, community composition, and the presence of exotic and rare species in the Castle Provincial Parks of Alberta, Canada, by surveying 142 transects adjacent to or far from trails. I also characterized the habitat of species from the rare genus Botrychium Swartz and tested a species distribution model (SDM) to identify suitable Botrychium habitat. Plant communities near trails had higher species richness, shifts in composition, and greater occurrence of exotic plant species. These effects extended farther from off-highway vehicle (OHV) trails than from footpaths, but only in mixed/broadleaf and shrubland vegetation. The SDM was not a strong predictor of Botrychium presence, but I found Botrychium in 29% of surveyed sites. To minimize trail effects such as colonization by exotic species, managers should prioritize closing trails to OHVs or limiting OHV traffic, particularly in mixed/broadleaf and shrubland vegetation. iv ACKNOWLEDGMENTS To my family who have been patiently waiting for me to finish my degrees and are always there to give me helpful words of wisdom. To my partner, Kyle Bischke, who constructed the quadrats used in my project, adopted my dog during field season, and most importantly, provided me with the mental encouragement to keep pushing through those difficult moments. To my friends who, entertained my outspoken thoughts, passion for Botrychium and botanizing on camping and hiking trips, even though they still may not know exactly what my research entails. To my mentors: Jenny Burke, who inspired me to find that same passion for native plants that she does; her excitement and love for botany got me curious and wanting to know more about the fascinating world of plants and herbaria. Dr. John F. Bain, my favourite LEA Herbarium curator who can climb up steep scree slopes like no other in hopes of finding those rare plants! And to Joanne Golden who instilled confidence in my plant skills and encouraged me to pursue grad school. These three are all equally responsible for my botanical craze and have inspired me to continue this journey which has led me to climb many mountains and enjoy many views. To my funding partners, Alberta Conservation Association (ACA) who provided financial support for my research which helped enormously in making this project happen. To Alberta Environment & Parks (AEP) for allowing me to conduct my research in Castle. To Wonnita Andrus for providing location data of exotic plant control areas and the keys for accessing gated trails. To my committee members, Dr. John F. Bain, Dr. Julie Lee-Yaw, and Dr. Robert Laird, who provided constructive feedback on my proposal, presentations, and thesis. To my lab members who listened through practice presentations and provided me with feedback and advice on my project. To my field assistants, Kirsty McFadyen, David Musk, Dani Jakovljevic, and Cooper Hayward, who v endured long field hours, unpredictable mountain weather, and sometimes uncomfortable tenting trips. And last but certainly not least, a very special thank-you to my supervisor, Dr. Jenny McCune. Jenny provided advice during difficult moments and guided me in the statistical analyses world which allowed me to explore my ecological curiosities about the plant communities. She is always encouraging and supportive of her students along their journey, motivating us to put our best foot forward in everything we do. Throughout my entire project, Jenny has been patient from start to finish, and she inspires me to continue finding those little moments to still play the piano. Thank-you. Without any of you, this project would not have been possible! vi TABLE OF CONTENTS DEDICATION…………………………………………………………………………..iii ABSTRACT……………………………………………………………………………..iv ACKNOWLEDGMENTS……………………………………………………………....v LIST OF TABLES………………………………………………………………………………....ix LIST OF FIGURES………………………………………………………………………………xiv CHAPTER 1: INTRODUCTION……………………………………...…………...…..1 1.1 Background…………………………………………………………………..1 1.2 Objectives…………………………………………………………………….5 1.3 Thesis Organization…………………………………………………………5 CHAPTER 2: RECREATIONAL TRAIL IMPACTS ON THE PLANT COMMUNITIES OF CASTLE AND CASTLE WILDLAND PROVINCIAL PARKS (CCWPP).............................................................................................................8 2.1 Abstract………………………………………………………….…………...8 2.2 Introduction……………………………………………….………………....9 2.3 Methods………………………………………………………….………….18 2.3.1 Study area……………………………………………………….…18 2.3.2 Study Design……………………………………………………….22 2.3.3 Data Collection…………………………………………………….24 2.3.4 Statistical Analyses………………………………………………...27 2.4 Results……………………………………………………………………….34 2.4.1 Species richness……………………………………………………37 2.4.2 Community composition…………………………………………...41 2.4.3 Presence of at least one exotic species……………………………..44 2.4.4 Presence of at least one rare species……………………………….48 2.5 Discussion…………………………………………………………………...51 2.5.1 Trail impacts on species richness………………………………….51 2.5.2 Trail impacts on plant community composition…………………...55 2.5.3 Trail impacts on the presence of exotic species……………………57 vii 2.5.4 The effect of trails on the presence of exotic species at higher elevations………………………………………………………………...61 2.5.5 Trail impacts on the presence of rare species……………………...63 2.6 Conclusion…………………………………………………………………..65 CHAPTER 3: HABITAT CHARACTERISTICS OF KNOWN BOTRYCHIUM OCCURRENCES AND QUANTITATIVE ANALYSIS OF ITS ASSOCIATION WITH TRAILS………………………………………………………………………....66 3.1 Abstract……………………………………………………………………..66 3.2 Introduction………………………………………………………………...67 3.3 Methods……………………………………………………………………..73 3.3.1 Assessing Botrychium habitat……………………………………..73 3.3.2 Testing a Species Distribution Model……………………………..74 3.3.3 Data Collection…………………………………………………….79 3.3.4 Statistical Analyses………………………………………………...81 3.4 Results……………………………………………………………………….82 3.4.1 Assessing factors associated with Botrychium occurrences in CCWPP…………………………………………………………………..82 3.4.2 SDM as a predictor for off-trail Botrychium occurrences…………86 3.4.3 Quantifying differences in community composition of plots with and without Botrychium……………………………………………………...89 3.5 Discussion…………………………………………………………………...91 3.6 Conclusion…………………………………………………………………..99 CHAPTER 4: CONCLUSION……………………………………………………….100 4.1 Limitations and future directions………………………………………..101 4.2 Concluding Statement…………………………………………………….102 REFERENCES………………………………………………………………………..104 APPENDICES……………………………………………………………………...…116 1- Chapter 2 Supplementary Materials…………………………………….116 2- Chapter 3 Supplementary Materials…………………………………….134 viii LIST OF TABLES Table 1. Number of control and on-trail transects surveyed within each vegetation type in each watershed in CCWPP………………………………… …………………………...23 Table 2. Predictor variables used in mixed models for each response variable…….……30 Table 3. Results of the model for species richness………………………………………38 Table 4. Results of the model for species richness of footpaths and OHV trails only………………………………………………………………………………………40 Table 5. Results of the model for shifts in community composition (Bray-Curtis dissimilarity) compared to the 10m quadrat………………………..……………………42 Table 6. Results of the model for the probability of at least one exotic species present………...………………………………………………………………………….45 Table 7. Results of the model for the probability of at least one rare species present……………………………………………………………………………………49 Table 8. Results of the model for the probability of at least one rare species present for footpaths and OHV trails only…………………………………………………………...50 Table 9. Environmental predictors used to build the Botrychium SDM…………………77 Table 10. Table of Botrychium species (rows) identified from photographed occurrences within each vegetation type (columns)…………………………………………………..85 Table 11. Table of associated species identified in Botrychium photographs (n = 81) and the number of times each species was listed as an associate……………………….……86 Table 12. Characteristics of the 24 off-trail plots surveyed in the summer of 2021, in order of increasing predicted habitat suitability according to a species distribution model………………………………………………………………….…………………88 Table 13. Predictor variables included in the model for Botrychium habitat suitability...........................................................................................................................89 Table 14. Results of PERMANOVA pairwise tests comparing species composition in plots with Botrychium present and absent.………………………………………….…...90 Table 15. Significant indicator species of plots with or without Botrychium according to an indicator species analysis with 9,999 permutations…………………………………..91 Table A1.1. Final locations of trailside and off-trail transects surveyed in June to August of 2020 and 2021……………………………………………………………………….116 Table A1.2. Results of pairwise comparisons of estimated marginal mean maximum height of vegetation of each trail type compared between distances from trail.……….121 ix Table A1.3. Results of pairwise comparisons of estimated marginal mean soil compaction values at each distance compared between trail types……...…………………………..122 Table A1.4. Results of pairwise comparisons of estimated marginal mean (emmean) soil compaction values for each trail type between distances from trail.….………………..123 Table A1.5. List of exotic species (n = 35) recorded during two survey seasons, from June to August of 2020 and 2021…………………………………………..……….….124 Table A1.6. List of rare species (n = 15) recorded during two survey seasons, from June to August of 2020 and 2021……………………………………………………………125 Table A1.7. Results of pairwise comparisons of estimated marginal mean species richness at each distance for each vegetation type….……….…………………………………..126 Table A1.8. Results of pairwise comparisons of mean species richness (emmean) at each distance from trail for each trail type.……………………………………….…………127 Table A1.9. Results of pairwise comparisons of the slope of the relationship between species richness and distance from trail (emtrends) for the subset data of only footpaths and OHV trails within each vegetation type...................................................................128 Table A1.10. Results of pairwise comparisons of community dissimilarity within each vegetation type..………………………………………………………………………..128 Table A1.11. Results of pairwise comparisons of mean Bray-Curtis dissimilarity values (emmean) between each trail type at each distance from trail………………...……….129 Table A1.12. Results of pairwise comparisons of mean Bray-Curtis dissimilarity values (compared to the 10m quadrat) for footpaths and OHV trails (emmean) for each vegetation type…………………...…………………………………………………….129 Table A1.13. Results of pairwise comparisons of the probability of exotic species presence within each vegetation type……………………………………………….…130 Table A1.14a. Results of pairwise comparisons of the probability of exotic species present (emmean) at each distance from trail for each trail type.……………………...130 Table A1.14b. Results of pairwise comparisons of the probability of exotic species present (emmean) for each trail type at each distance…………………………………131 Table A1.15. Results of estimated marginal means of linear trends comparing the slope of the relationship between the probability of at least one exotic species and distance from trail (trend) for the subset data of only footpaths and OHV trails within each vegetation type……………………………………………………………….…………………….132 Table A1.16. Results of pairwise comparisons (emmeans) of the mean probability of at least one exotic species per transect compared between each vegetation type…...……132 x Table A1.17. Results of pairwise comparisons (emtrends) of the slope of the relationship between the probability of at least one exotic species per transect and elevation compared between each trail type. …………….…………………………………………………..133 Table A1.18. Results of pairwise comparisons of the probability of rare species present (emmean) compared between each trail type………………………..…………...……..133 Table A2.1. Results of the indicator species analysis for plots with or without Botrychium according to an indicator species analysis with 9,999 permutations.…………………..134 xi LIST OF FIGURES Figure 1. Photograph taken in the summer of 2021 from the summit of Table Mountain, showing the main gravel road traversing through coniferous and wetland habitats in Castle Provincial & Wildland Provincial Parks, heading towards Beaver Mines Lake in Alberta, Canada……………………………………………………………………..……11 Figure 2. Map of the study area, outlining Castle Provincial Park (light orange) and Castle Wildland Provincial Park (pink) and official trails (brown)…...…………………19 Figure 3. A schematic diagram of a trailside transect. An 11m transect (red line) runs perpendicular to the trail…………………………………………………..……….…….26 Figure 4. The maximum height of understory vegetation (A-B) and soil compaction (C- D) with increasing distance from trails for (A & C) vegetation types and (B & D) trail use types……………………………………………………………………………………...35 Figure 5. Partial regression plots showing the effect of a) elevation, b) northness, c) the interaction of distance from trail and vegetation type, and d) the interaction of distance from trail and trail type on the species richness within each quadrat……………………39 Figure 6. Partial regression plots showing the effect of distance from trail on species richness for footpaths versus OHV trails within each vegetation type. …………………41 Figure 7. Partial regression plots showing the effect of a) vegetation type and b) the interaction of distance from trail and trail type on Bray-Curtis dissimilarity of each quadrat compared to the 10 m quadrat…………………………………………………..42 Figure 8. Partial regression plots showing the effect of a) the interaction between distance from trail and vegetation type b) the interaction of distance from trail and trail type, and c) the interaction between vegetation type and trail type on Bray-Curtis dissimilarity of 0 m, 2 m, and 5 m quadrats compared to the 10 m quadrat in each transect, for footpath and OHV trail transects only………………………………………………………………...43 Figure 9. Partial regression plots showing the effect of a) elevation, b) northness, c) vegetation type, and d) the interaction between distance from trail and trail type on the probability of occurrence of one or more exotic species within each quadrat………….46 Figure 10. Partial regression plots showing the 3-way interaction between distance from trail, vegetation type, and trail type on the presence of exotic species within each quadrat for footpaths (blue) and OHV trails (orange). ………………………………………….47 Figure 11. Partial regression plot showing the effect of a) vegetation type, and b) the interaction between elevation and trail type on the presence of exotic species within each transect………………………………………………………………………………….48 Figure 12. Partial regression plots showing the effect of a) elevation, b) distance from trail, c) vegetation type, and d) trail type on the probability of at least one rare species………………………………………………………………………………….50 xii Figure 13. The diversity of Botrychium species found along two popular hiking trails in CCWPP, Alberta, Canada ………….…,,,,………………………………………………70 Figure 14. Inset: shows model extent (grey shaded) and location of Castle Provincial Park and Castle Wildland Provincial Park in the southwest corner of Alberta, Canada. Main map: CCWPP, coloured based on predicted habitat suitability for Botrychium according to a species distribution model (SDM)………………………………….……………….78 Figure 15. A schematic diagram of a 50 m x 50 m plot…………………………………80 Figure 16. The frequencies of a) moonwort species; and b) vegetation types of the 85 occurrence records……………………………………………………………………….84 Figure 17. Histograms showing the frequency of occurrences of Botrychium…………..86 Figure 18. NMDS (non-metric multidimensional scaling) ordination of all 50 m x 50 m sites in species space……………………………………………………………………..90 xiii CHAPTER 1: INTRODUCTION 1.1 Background Protected parks have been established in Canada since the 1800s, primarily to support the economy through recreational activities, and secondarily, to protect wildlife and the wilderness (Foster, 1998). As the popularity of parks and the human population continue to grow, both the number of recreational trails and the intensity of their use will increase, placing more impact on the surrounding ecosystem (Debarbieux et al., 2014). Globally, research on the impact of recreational trails on plant communities has increased (Sumanapala & Wolf, 2019); however, most studies focus on parks and wilderness reserves throughout the United States (e.g., Cole, 1978; Benninger-Truax et al., 1992; Gibson et al., 2000; Dickens et al., 2005) and Australia (e.g., Scherrer & Pickering, 2006; Ngugi et al., 2014; Pickering & Norman, 2017). Although some studies of trail impacts on plant communities have been done in Canada (Price, 1985; Parikesit et al., 1995; Thurston & Reader, 2001; Nepal & Way, 2007; Crisfield et al., 2012; Trip & Wiersma, 2015; Grenke et al., 2018), only a handful of published studies have been conducted in Alberta’s Rocky Mountains (Price, 1985; Crisfield et al., 2012; Grenke et al., 2018) even though parks in this region (including Banff, Jasper, Yoho, and Waterton Lakes National Parks) have the highest attendance rates among Canada’s National Parks (Parks Canada, 2021). Understanding the impact of recreational trails is an important consideration for all natural areas and protected parks in balancing recreation and conservation. Many studies on trail impacts have focused on large mammals like bears (Kasworm & Manley, 1990; Benn & Herrero, 2002), wolves (Whittington et al., 2005; Naylor et al., 2009; Rogala et al., 2011), and elk (Naylor et al., 2009; Rogala et al., 2011). These animals tend to avoid high traffic trails and roads (e.g., Kasworm & Manley, 1990; 1 Whittington et al., 2005; Rogala et al., 2011), and change their daily routine of resting, feeding, and travel to avoid high traffic trails (Rogala et al., 2011). Trails also affect birds, with increased nest predation (Miller et al., 1998), lower nest success (Yoo and Koper, 2017), and lower bird densities near roads and trails (Thompson, 2015). Trails can also positively impact some animals. For example, salamanders are associated with microhabitats found near low traffic trails (Davis, 2007; Smith et al., 2017). The effect of recreational trails on vegetation has been studied since the 1930s, with studies noting how trailside conditions can favour certain growth forms. For example, Bates (1935) found that species with prostrate, low-growing lifeforms were more likely to survive trailside than species with upright, brittle-stems. Trails can also affect the species richness of a plant community. For example, some studies have reported higher plant species richness near trails than away from trails, likely due to higher light availability trailside (Bates, 1935; Dale & Weaver, 1974; Tyser & Worley, 1992). Others report that intermittent disturbance along trails prevents dominance by the strongest competitor (Larson, 2002; Dickens, 2005). Increased richness near trails is also facilitated by the introduction of seeds of exotic species and disturbance-adapted native species via animals (Campbell & Gibson, 2001), hikers’ boots and clothing (Mount & Pickering, 2009), or vehicles (von der Lippe & Kowarik, 2007; Yang et al., 2021). Trails also alter environmental conditions, which in turn affects the kinds of species growing within a community. Therefore, the composition of plant communities near trails is often different than those away from trails (Müllerová et al., 2011; Benninger-Traux et al., 1992). 2 Finally, the presence or absence of rare plant species can be affected by nearby trails. Small populations of rare plant species in popular recreation areas are susceptible to trampling by hikers and OHVs (off-highway vehicles), especially when these activities stray from designated trails (e.g., Kerbiriou et al., 2008) or occur in sensitive alpine habitats (e.g., Rossi et al., 2009). However, the presence of trails can sometimes benefit rare plants, perhaps because intermittent disturbance reduces competition for light or nutrients (e.g., Catling & Kostuik, 2011; Wedegärtner et al., 2022). In Canada, only a handful of recreational trail studies have focused on plant communities (Parikesit et al., 1995; Nepal & Way, 2007; Crisfield et al., 2012; Trip & Wiersma, 2015). In Ontario, plant species richness was highest along trails with intermediate use levels (Parikesit et al, 1995), consistent with the intermediate disturbance hypothesis (Connell, 1978). A study in Newfoundland found that dry boreal forests were less resistant to changes in species composition caused by trails compared to heath or bog sites (Trip & Wiersma, 2015). In British Columbia, Nepal & Way (2007) found that two backcountry trails had significantly higher species richness of herbaceous species trailside compared to off-trail. In contrast, alpine vegetation in the northern Rocky Mountains of Alberta had lower species richness along trails compared to the undisturbed or naturally disturbed tundra (Crisfield et al., 2012). Differences in trail impacts across Canada and between different vegetation types indicate a need for more studies assessing the effect of recreational trails on plant communities. Castle Provincial Park and Castle Wildland Provincial Park (CCWPP) were established in 2017 (Alberta Environment & Parks, 2018). For over 10,000 years this region has been home to the Piikani nation, members of the Blackfoot confederacy, who 3 hunt and fish the land (Alberta Wilderness Association, 2022). The Castle region, situated in the southwest Rocky Mountain range of Alberta, has 106 species of plants that are provincially tracked due to their rarity, more than twice as many as Banff or Jasper (Farr et al., 2017). The region is also a world centre of diversity for a small, cryptic fern genus called moonwort (Botrychium Sw.; Wagner et al., 1983; Williston, 2001). CCWPP was originally part of Waterton Lakes Dominion Park (now Waterton Lakes National Park), established in 1895 as Canada’s fourth National Park after Banff, Glacier, and Yoho National Parks (Lothian, 1987). However, with the enactment of the Dominion Forest Reserves and Parks Act in 1911, the Castle region was removed from the National Park (Doherty, 2012). This resulted in the Castle region’s re-designation to forest reserve status, which limited public use, but still allowed for livestock grazing and timber harvest (Gillis & Roach, 1986). The CCWPP region was designated as a Provincial Game Reserve in 1921, providing extended areas of pasture for ranchers. However, this status was removed in 1954, reverting the region’s status to Provincial Crown Land, which permitted managed cattle grazing, timber harvest, extraction of oil and gas, as well as unregulated recreational use (Castle-Crown Wilderness Coalition (CWCC), 2022). In 1974, a Government of Alberta study recommended that the area should be protected (CCWC, 2022), but the CCWPP were not established as provincial parks until 43 years later (Alberta Environment and Parks, 2018). After designation as provincial parks, many trails in the southern region of Castle Provincial Park were closed to OHV users; however, most trails in the northwest area of CCWPP remain open to OHV use. The provincial government planned to close 130 km of OHV trails by the end of 2021; however, strong pushback against the trail closures by local OHV users resulted 4 in the delay of this phase-out (Bellefontaine, 2019). Currently, there are an estimated 2,000 km of linear features, which include roads and recreational trails, in CCWPP (Farr et al. 2017). Although there have been studies on the trail impacts on animals such as bears (Lee & Hanneman, 2011; Proctor et al., 2020), and elk (Ciuti et al., 2012; Paton et al., 2017), how trails affect plant communities in this area has not yet been studied. 1.2 Objectives The overall goal of my research was to quantify the effects of recreational trails on plant communities in the Castle and Castle Wildland Provincial Parks and determine how these effects vary among different trail types and different vegetation types. My specific objectives were: 1. To establish a set of plots along trails in CCWPP which can be resurveyed for future research. 2. To measure the effect of recreational trails on species richness and community composition, and to quantify how far this effect extends away from the trail. 3. To measure the effect of recreational trails on the presence of exotic and rare species. 4. To test how trail use type, vegetation type, or interactions between them, influence the effect of trails on species richness, composition, and the presence of exotic and rare species. 5. To characterize the habitats associated with Botrychium occurrences and test a habitat suitability model as a predictor of Botrychium occurrences away from trails in CCWPP. 1.3 Thesis Organization 5 In Chapter 2, I test the effect of recreational trails on i) the species richness of vascular plants, ii) the composition of plant communities, iii) the presence of exotic plants, and iv) the presence of critically imperiled (S1) and imperiled (S2) provincially tracked rare plants in CCWPP. I surveyed 118 trail, and 24 off-trail transects throughout the two provincial parks, recording the abundance of all vascular plants within a 1 m x 1 m survey area directly adjacent to the trail edge (0 m), and 2 m, 5 m, and 10 m along the transect perpendicular to the trail or off-trail starting point. I then used these data to test the effects of trail type, distance from trail edge, and vegetation type on vascular plant richness, the shift in plant community composition, and the presence of exotic and rare species moving along the transect. In Chapter 3, I focus on the genus Botrychium (moonwort), which includes 21 species found in Alberta, 15 of which are considered by NatureServe to be provincially rare (NatureServe, 2022). CCWPP is part of the region designated as the world’s centre of Botrychium diversity (Wagner et al., 1983; Williston, 2001). Botrychium are small ferns that have an aboveground stem divided into two axes: one that is sterile and leaf- like, the other with small round clusters of fertile sporangia which house spores (Farrar, 2011). Observations that several species of moonwort are often found near trail edges support the idea that moonwort may benefit from trails (Müllerová et al., 2011). I assess habitat characteristics of Botrychium occurrences to determine potential predictors of species distribution. Then, using a species distribution model built using all known records of species in the Botrychium genus in Alberta, I visited 24 sites at least 100 m away from any recreational trail in CCWPP that varied in their predicted species distribution. I recorded all vascular plant species present, while carefully searching for 6 moonwort presence. I then used these data to test the species distribution model as a predictor for moonwort occurrences away from trails and to characterize the plant community composition of sites with versus without Botrychium present. Chapter 4 summarizes the findings of Chapters 2 and 3 and highlights the implications for trail management in CCWPP. The effects of trails on plants have been studied in many places, but we know very little about how trails affect the plant communities in CCWPP, a hotspot for plant diversity. My research provides the foundation for ongoing ecological research on the plant communities in the two parks, and their response to disturbance associated with trails. With potential for more and more visitors as the park gains popularity and interest, it is important to understand how plant communities are affected by trails. My research will help managers make decisions about trail closures by revealing which vegetation types are most sensitive to trail-use and identifying trail types that are most likely to serve as conduits for invasion by exotic plant species. In addition, my research contributes information on habitat preferences for understudied, rare species like moonwort. Understanding how roads and trails affect plant communities in CCWPP is integral to balancing recreational use and conservation. 7 CHAPTER 2: RECREATIONAL TRAIL IMPACTS ON THE PLANT COMMUNITIES OF CASTLE AND CASTLE WILDLAND PROVINCIAL PARKS (CCWPP) 2.1 Abstract In protected parks, whose mandates balance recreation and conservation, it is important to understand how trails affect plant communities. I investigated the impacts of recreational trails on plant communities in Castle Provincial and Castle Wildland Provincial Parks (CCWPP), a provincial hotspot of plant diversity, including many rare species. I surveyed plant communities in transects extending 11 m from trails that vary in trail use and vegetation type, and in control transects distant from any trail. I tested the effects of trails on species richness, community composition, and the presence of exotic and rare species. I predicted that communities adjacent to trail edges would have higher species richness, shifts in community composition, and greater occurrence of exotic species compared to communities several meters from a trail or control transects. I also predicted that the magnitude of these differences would vary for different trail types and different vegetation types. For example, I predicted that the effects of OHV (off-highway vehicle) trails would extend farther from the trail edge compared to footpaths and that the effect of trails would be stronger in coniferous forests, which have restricted light availability compared to other vegetation types. My results showed the predicted patterns in most cases, with increased species richness, shifts in community composition, and increased probability of finding exotic species within five metres of trails or roads. However, vegetation type and trail type influenced the magnitude and extent over which these changes occurred. In grasslands, there were no significant increases in species 8 richness or shifts in composition near trail edges, likely because light availability is similar whether near or far from the trail. Grasslands had nearly 100% probability of exotic species occurrence up to 10 m away from trails, whereas in coniferous forests the probability of exotic species occurrences decreased dramatically 10 m away from the trail. Exotic species had a higher likelihood of occurring beyond 2 m from OHV trails than from footpaths, but only in mixed/broadleaf or shrubland vegetation. Although rare species were slightly more likely to occur in control transects away from trails, we found 15 different species near trails with either a provincial conservation status of critically imperiled (S1) or imperiled (S2). As the first study to look at trail impacts on plant communities in CCWPP, results of this study highlight that the effect of recreational trails depends on the type of trail and the type of vegetation it goes through. Limiting OHV trails through shrubland and mixed forest vegetation could reduce spread of exotic species from the trailside into these vegetation types. 2.2 Introduction As the human population continues to increase, more habitat is converted to human land use, and the remaining protected areas face increased numbers of recreational visitors (Wittemyer et al., 2008; Debarbieux et al., 2014; Monz et al., 2021). Protected areas often have two goals: conserving ecosystems and providing a space for outdoor recreational activities. For example, under Alberta’s Protected Parks Act, the purpose of establishing provincial parks is for preservation of natural heritage, and for the enjoyment of outdoor recreation (Government of Alberta, 2017). For this reason, it is important to understand how recreational activities affect ecological communities in protected areas. An important question is how roads and trails - which provide access for recreation - 9 affect plant communities. Parks are often criss-crossed by many kilometres of roads and trails (Figure 1). The construction, maintenance, and continued use of roads and trails increases soil compaction and erosion (Webb et al., 1978; Ballantyne & Pickering, 2015; Marion et al., 2016), decreases soil moisture (Webb et al., 1978; Ballantyne & Pickering, 2015), and increases light and disturbance levels (Watkins et al., 2003; Avon et al., 2010). It is important to quantify how these altered conditions affect plant community composition and diversity, how far these effects extend away from trails and roads, and whether impacts are greater for different types of trails or different vegetation types. The species richness of a plant community can be affected by trails. Many studies have reported higher plant species richness near trails than away from trails, likely due to increased light availability trailside coupled with the fact that fewer species are able to tolerate the low light conditions below dense canopies of forest interiors (Bates, 1935; Dale & Weaver, 1974; Tyser & Worley, 1992). Others suggest that the intermittent disturbance along trails promotes species richness by preventing dominance by the strongest competitor (Larson, 2002; Dickens, 2005). Increased richness near trails could also be caused by the introduction of seeds of exotic species and disturbance-adapted native species via animals (Campbell & Gibson, 2001), hikers’ boots and clothing (Mount & Pickering, 2009), or vehicles (von der Lippe & Kowarik, 2007; Yang et al., 2021). Plant species richness near trails also varies depending on the vegetation type that a trail traverses. For example, in the Rocky Mountain ranges of Colorado, USA, Wells et al. (2013) determined that native species richness along trails was significantly higher in aspen forests, riparian areas, and meadows compared to evergreen forests. Species richness near trails can also vary depending on the level of trail use. For example, 10 Benninger-Truax et al. (1992) determined that species richness was significantly higher along light and moderately used trails compared to heavily used trails. Similarly, Parikesit et al. (1995) found higher species richness along trails with intermediate disturbance compared to heavily used trails or undisturbed sites. Most studies have found that the effect of trails on plant species richness extends no more than 5m from roads or trails (Watkins et al., 2003; Godefroid & Koedam, 2004; Benninger-Truax et al., 1992). However, the extent of a trail effect on plant species richness could vary for different regions. Figure 1. Photograph taken in the summer of 2021 from the summit of Table Mountain, showing the main gravel road traversing through coniferous and wetland habitats in Castle Provincial & Wildland Provincial Parks, heading towards Beaver Mines Lake in Alberta, Canada. 11 The composition of plant communities is also altered by the presence of nearby roads and trails. This follows from the fact that some species prefer trailside conditions, while others are sensitive to them. For example, one study in the alpine tundra of the Czech Republic found that highly competitive species tolerant of human disturbance were dominant close to roads whereas less competitive stress-tolerant species became dominant farther from the road (Müllerová et al., 2011). Another study in the coniferous forests of Rocky Mountain National Park, USA found that in addition to significantly higher species richness trailside, there was a shift in community composition towards higher abundance of species with disturbance tolerant traits, such as ground-level leaves, or below ground stems (Benninger-Traux et al., 1992). In contrast, undisturbed communities away from trails were dominated by shade-loving species. The degree to which plant community composition changes near trails can be affected by the vegetation type that a trail traverses. For example, Trip & Wiersma (2015) found that forested habitats had a sharper contrast in species composition moving from intact vegetation towards a trail compared to open bogs or heaths. The authors suggest that species of open habitats are adapted to high-light conditions and more resistant to disturbance because they include stoloniferous/rhizomatous grasses that are resistant to trampling, whereas many forest species are not (Trip & Wiersma, 2015). Therefore, the species composition moving towards a trail in open habitats is more similar to the composition far from the trail. Roads and trails can also facilitate the introduction and sometimes invasion of exotic plants. The increased prevalence of exotic plants near roads and trails is often attributed to their ability to adapt to disturbances. For example, fast growth and large 12 production of easily dispersed seeds are traits associated with exotic plants found in disturbed areas (Baker, 1974; Lake & Leishman, 2004; Van Kleunen et al., 2015). Exotic plants are dispersed into native plant communities by vectors associated with roads and trails such as tires of vehicles (von der Lippe & Kowarik, 2007; Yang et al., 2021), boots and clothing (Campbell & Gibson, 2001; Mount & Pickering, 2009), and the hooves and fur of animals (Campbell & Gibson, 2001; Gower, 2008). Many studies have focused on the impact of roads and trails on the presence of exotic species found trailside, however, the distance that exotic species can spread into intact vegetation away from different corridors varies. For example, a study from the Great Lakes area of Minnesota, USA, found that increased exotic species richness and cover did not extend more than 1 m from the trail edge (Dickens et al., 2005). However, in another region of the Great Lakes in Wisconsin, USA, Watkins et al. (2003) found that exotic species were prevalent up to 15 m from roads. Tyser & Worley (1992) found that although most exotic species were limited to within 1-2 m of grasslands along primary and secondary roads, common dandelion (Taraxacum officinale F.H. Wiggers) and two exotic grasses, timothy grass (Phleum pratense Linnaeus) and Kentucky bluegrass (Poa pratensis Linnaeus) occurred as far as 100 m from backcountry trails. The type and intensity of trail use can also affect the presence and abundance of exotic species found along trails. For example, Benninger-Truax et al. (1992) found significantly higher richness of exotic plants along moderately used trails compared to lightly used trails. Potito & Beatty (2005) found significantly higher exotic species cover when comparing heavily used trails to newly established trails. Different vegetation types can also affect richness or frequency of exotic species trailside. Larson et al. (2001), for 13 example, found higher numbers and frequencies of exotic species near trails in mesic compared to drier mixed grass vegetation in North Dakota. Similarly, in the Central Grasslands and Colorado Rockies (USA), Stohlgren et al. (1999) found significantly higher exotic frequency in wetter aspen and meadow sites compared to drier coniferous forest sites. How far exotic species spread into the surrounding vegetation, and whether trail types or vegetation types are influencing this extent, should be studied in other natural areas impacted by roads and trails. Most studies in montane regions report a steady decrease in exotic species richness with increasing elevation (Becker et al., 2005; Pauchard et al., 2009). Several authors have suggested that this pattern is a result of the harsher climate, and lower propagule pressure at high elevations: fewer exotic species are able to survive high elevation conditions, and fewer seeds are able to reach these areas (Becker et al., 2005; Averett et al., 2016). With changing climate and more visitors reaching higher elevations, exotic species invasions at higher elevations are becoming more frequent in many areas of the world (Becker et al., 2005; Pauchard et al., 2009; Averett et al., 2016; Medvecká et al., 2018, Liedtke et al., 2020); however, I know of no studies that have examined whether recreation is facilitating the spread of exotic plants to higher elevations within protected areas in Canada’s montane regions. Protected areas are often home to rare plant species, many of which require intact, relatively undisturbed ecosystems to thrive. Disturbance via human recreation is a leading threat to rare plants in Canada (McCune et al. 2013), and around the world (e.g., Ballantyne & Pickering, 2013; Hernandez-Yanez et al. 2016). For example, trampling, recreational activities, OHVs, and horse riding are all listed as reported threats to rare 14 plant species and communities in Australia (Kelly et al., 2003). Although rare plant species in popular recreation areas are susceptible to trampling by hikers and OHVs, especially when these activities stray from designated trails, the presence of trails can sometimes benefit rare plants. Taylor and Raney (2013) found increased abundance of the rare thread-leaved sundew (Drosera filiformis Rafinesque) near OHV trails, a result of micro-habitats created by the tire tracks that favour the thread-leaved sundew and decrease competition for resources with other bog species. Similarly, Catling & Kostuik (2011) found that a native orchid, Calypso bulbosa var. americana (R. Brown) Luer, was more abundant within 1 m of recreational trails compared to 1 m to 3 m beyond the trail, which the authors suggest is a result of reduced competition from trampling of neighboring disturbance intolerant species. Protected areas rich in plant diversity and popular for outdoor recreation must be studied at local scales to understand the impacts of roads and trails on rare plant species. Castle Provincial Park (49.4314°N, 114.3933°W) and Castle Wildland Provincial Park (49.241°N, 114.244°W) in southwestern Alberta, Canada (CCWPP) are ideal study sites to test the effects of trails on plant communities in the Rocky Mountains. Together, these two parks encompass over 105,000 hectares of protected land (Alberta Environment and Parks, 2018). CCWPP has a long history of industrial and recreational use, including oil and gas extraction, community grazing, hunting, fishing, camping, and trail use by OHVs, equestrians, and hikers. As a result, it is estimated that approximately 2,000 km of linear features criss-cross the Parks (Farr et al., 2017). Nearly half of all Alberta’s vascular plant species are found in the CCWPP region, with over 100 species that are provincially and/or nationally rare (Alberta Environment & Parks, 2018), making this 15 region a hotspot for plant diversity. In addition, CCWPP has long been a popular off- highway vehicle destination for many locals and people from surrounding communities such as the Crowsnest Pass (Alberta Environment & Parks, 2015). After designating CCWPP as provincial parks in 2017, the province moved to decommission over 130 km of OHV trails, but this was met with strong opposition from local OHV advocates (Bellefontaine, 2019). Without quantitative data on trail effects, managers do not have the evidence they need to make decisions about trail closures or expansions. Although a few studies have examined the effect of roads and trails on large mammals within CCWPP (Lee & Hanneman, 2011; Ciuti et al., 2012; Paton et al., 2017; Proctor et al., 2020), no study has quantified the effects of trails on plant communities. Indeed, only a handful of such studies have ever been done in Canada (Price, 1985; Parikesit et al., 1995; Thurston & Reader, 2001; Nepal & Way, 2007; Crisfield et al., 2012; Trip & Wiersma, 2015; Grenke et al., 2018). In this study, I test the effect of recreational trails and roads on plant communities in CCWPP. Specifically, I investigate how trail use type, vegetation type, or interactions between them influence the effect of trails on plant species richness, community composition, and the presence of exotic and provincially rare species. My prediction are as follows: (1) I predict that species richness will increase near trail edges as more exotic species and disturbance-tolerant native species are found trailside, increasing the number of species present. In addition, I predict that the strength of this effect will vary with vegetation and trail type. For example, I expect that the increase in species richness near trails will be smaller in open light habitats such as grasslands and higher in dense canopy 16 coniferous forests where high light availability is restricted to trail edges. I also expect the increase in species richness moving from intact vegetation towards trails to be greater near OHV trails compared to footpaths because of more propagule pressure from OHV tires, which likely carry more seeds compared to foot traffic. In addition, I predict that interactions between vegetation type and trail type could affect the degree of increased species richness near trails compared to farther away. For example, I predict that the greater impact from OHV trails compared to footpaths will be less drastic in grasslands and more pronounced in coniferous forests. (2) I predict that community composition will shift near trails, as trailside disturbance and altered abiotic conditions favour a different suite of species. Coniferous forests with densely shaded understories often support few shade-tolerant species and so I predict that shifts in composition moving towards trails will be greater in coniferous forests compared to grasslands, where light levels near and far from trails are similar. I also predict shifts in composition with increasing proximity to footpaths will be less drastic compared to roads or OHV trails due to the increased disturbance frequency and intensity associated with wider trails and increased propagule pressure from vehicles bringing in more exotic seeds and dispersing them farther into the vegetation. (3) I predict that exotic species will be found most frequently directly beside trails and that the probability of their presence will decline when moving away from trails; however, this effect will differ between vegetation types and trail types. I predict that the presence of exotic species will be higher and extend farther in more open vegetation types such as grasslands. Grasslands are more open habitats with a more even distribution of light, allowing shade-intolerant exotic species to grow farther out from trails compared 17 to coniferous forests with little light penetrating through the dense canopy of the forest interior. Therefore, the decline in occurrence of exotic species moving away from trails will be steeper in coniferous forests. I also predict that exotic species will have a higher likelihood of occurring near OHV trails and roads compared to footpaths, and that exotics will be more likely to occur farther away from OHV trails and roads because they have higher frequency of use and vehicle tires bring in more seeds than foot traffic (Pickering & Mount, 2010). If trails are facilitating exotic species spread to higher elevations, I predict that the probability of finding exotic species at higher elevations will be greater along OHV trails compared to footpaths, and lowest on control transects. However, if the occurrence of exotic species declines at higher elevations simply because of the harsh climatic conditions, I do not expect to see any differences between trail types in the relationship between exotic species presence and elevation. (4) In this study, I define rare species as those ranked critically imperiled (S1) or imperiled (S2) in Alberta, as designated by NatureServe (NatureServe, 2022). It is unclear whether this group of species will be favoured by conditions near trails. If most rare species are somewhat disturbance tolerant, and benefit from increased light levels near trails, I predict that rare species will be more likely to occur closer to trails. However, if most rare species are sensitive to trail disturbance, I predict that rare plants will occur more frequently farther from trails. 2.3 Methods 2.3.1 Study area The Rocky Mountain ecoregion of southwestern Alberta, which includes CCWPP, is a hotspot of vascular plant diversity in the province, with over half of all 18 Alberta’s native plant species growing in the region (Kershaw, 2008). Bordered by British Columbia to the west, Crowsnest Pass in the North, and Waterton Lakes National Park to the South, CCWPP is part of the ‘Crown of the Continent’ ecosystem which houses important watersheds and habitats for many plants and animals, including species that are nationally and/or provincially rare (Alberta Parks, 2020; Figure 2). Figure 2. Map of the study area, outlining Castle Provincial Park (light orange) and Castle Wildland Provincial Park (pink) and official trails (brown). Circles depict trail transect sites, coloured by vegetation type; black diamonds depict off-trail transect sites. CCWPP is within the Rocky Mountain natural region, which includes the montane, subalpine, and alpine natural subregions of Alberta (Alberta Parks, 2020). These subregions have mean annual temperatures of 2.3°C, -0.1°C, and -2.4°C, mean frost-free periods of 64, 55, and 40 days, and growing season precipitation of 382 mm, 419 mm, and 472 mm, respectively (Natural Regions Committee, 2006). Although 19 CCWPP are two of the smaller protected parks in Alberta, they include a large elevation range, from 1,336 meters above sea level (a.s.l.) to 2,640 m a.s.l. on Loaf Mountain, the highest point in CCWPP. This large range in elevation results in a diversity of vegetation types. The montane subregion (825 m to 1,850 m a.s.l.) includes montane grasslands dominated by species such as mountain rough fescue (Festuca campestris Rydberg), Idaho fescue (Festuca idahoensis Elmer), and Parry oatgrass (Danthonia parryi Scribner) (Willoughby et al., 2008). Pine reed grass (Calamagrostis rubescens Buckley), buffalo berry (Shepherdia canadensis (Linnaeus) Nuttall), and bearberry (Arctostaphylos uva- ursi (Linnaeus) Sprengel) are understory species of well-drained open forest or mixed stand sites, whereas common understory species such as thimbleberry (Rubus parviflorus Nuttall) and white meadowsweet (Spiraea lucida Douglas ex Greene) are found in moister nutrient-rich coniferous forest sites (Natural Regions Committee, 2006). The subalpine subregion (1,300 m to 2,300 m a.s.l) is dominated by coniferous forests, with many young lodgepole pine (Pinus contorta Douglas ex Loudon) stands in low elevation post-fire sites, and subalpine fir (Abies lasiocarpa (Hooker) Nuttall) with occasional white-bark pine (Pinus albicaulis Engelmann) populations at higher elevations. Common understory species include buffalo berry, false azalea (Menziesia ferruginea J.E. Smith), and white-flowered rhododendron (Rhododendron albiflorum Hooker) shrubs (Natural Regions Committee, 2006). The alpine subregion vegetation (1,900 m to 3,650 m a.s.l.), includes alpine meadows and windswept barren tundra. Although vegetation is relatively sparse, some low-growing cushion species do occur, including white mountain avens 20 (Dryas drummondii Richardson ex Hooker) and moss campion (Silene acaulis (Linnaeus) Jacquin) in the ridgetops and shallow snow areas (Natural Regions Committee, 2006). The natural disturbance regime of the Rocky Mountain ecoregion includes biotic, geomorphic, and hydrological processes such as wildfires, windfall, wildlife grazing, flooding, drought, avalanches, insect infestations, forest pathogens, and beaver activity (Alberta Environment and Parks, 2018). In CCWPP, the 2003 Lost Creek Fire burned nearly 19,000 hectares of the two parks, resulting in large stands of young lodgepole pine in the northern regions of the parks (Farr et al., 2017). Mountain pine beetle (Dendroctonus ponderosae Hopkins) outbreaks have been reported in areas north and east of CCWPP and are likely to impact CCWPP in the future (Powell, 1966; Robertson et al., 2009; Taylor et al., 2006). After the replacement of bison with cattle introduced by European settlers, much of the area was used for unregulated grazing, resulting in increased frequency of invasive and agronomic plant species in the grasslands (Willoughby et al., 2008; Alberta Environment and Parks, 2018). CCWPP is part of the territory of the Blackfoot people. For over 10,000 years the Blackfoot people (Niitsitapi) have used the land for hunting, fishing, and sacred ceremonies (Alberta Wilderness Association, 2022). CWPP was originally included as part of Waterton Dominion Park (now Waterton Lakes National Park), which was established in 1895 as Canada’s fourth National Park after Banff, Glacier, and Yoho National Parks (Lothian, 1987). In 1911 the Castle region was removed from the National Park designation and ten years later was designated as a Provincial Game Reserve, which provided extended areas of pasture for ranchers. In 1954, the area became Provincial Crown Land with unregulated recreational activities including hiking, hunting, and off- 21 highway vehicle use, as well as extractive resource industries including logging, gravel extraction, and oil and gas wells (Castle- Crown Wilderness Coalition, 2022). Currently, there are over 2,000 km of linear features – including roads and trails - in CWPP (Farr et al., 2017). Trail types include long-standing foot trails for hiking, biking, and equestrian use, heavy use off-highway vehicle trails (OHV), and industry size gravel roads used for access to oil and gas wells and some trailheads. 2.3.2 Study Design To measure the effects of trails on plant communities, from June to August of 2020 and 2021, my field assistants and I surveyed 118 transects near trails and 24 transects at least 100 m away from any recreational trail, to serve as controls (Figure 2, Table 1). To select trailside sites to survey, I first stratified by watershed and vegetation type to ensure representation of all vegetation types and all watersheds throughout the study area. I delineated the watershed boundaries from the Hydrologic Unit Code (HUC) Watersheds of Alberta vector data produced by the Government of Alberta (Alberta Environment and Parks, 2017). I used a 2010 map of Alberta’s land cover produced by ABMI (Alberta Biodiversity Monitoring Institute) based on Landsat imagery to determine the vegetation types (coniferous forest, broadleaf forest, mixed forest, grassland, and shrubland; Castilla et al., 2014). Unfortunately, I was not able to obtain an official trail layer from Alberta Environment and Parks, or any information on usage levels of trails in the area. Therefore, I used a trail layer that was georeferenced and digitized by hand from the 2018 version of the Castle Provincial Park summer trails map and overlaid it on the watershed and vegetation layers using ArcMap version 10.2.1. I used the “Create Random Points” function in ArcMap to randomly select 20 points along 22 a trail within each watershed and vegetation type combination. I set the minimum distance between points to 50 meters. Table 1. Number of control and on-trail transects surveyed within each vegetation type in each watershed in CCWPP. Watershed Vegetation Control Trailside Grand Total transects transects Carbondale broadleaf 0 3 3 River coniferous 4 5 9 grassland 0 3 3 mixed 0 7 7 shrubland 5 2 7 Drywood Creek broadleaf 2 4 6 coniferous 0 1 1 grassland 0 6 6 mixed 0 5 5 shrubland 1 2 3 Middle Castle broadleaf 0 3 3 River coniferous 1 7 8 grassland 1 2 3 mixed 0 8 8 shrubland 1 4 5 Mill Creek broadleaf 0 0 0 coniferous 1 6 7 grassland 0 4 4 mixed 0 3 3 shrubland 0 2 2 Upper Castle broadleaf 1 1 2 River coniferous 0 4 4 grassland 0 1 1 mixed 1 3 4 shrubland 1 4 5 Upper broadleaf 0 0 0 Crowsnest River coniferous 1 6 7 grassland 0 0 0 mixed 0 3 3 shrubland 1 2 3 West Castle broadleaf 0 1 1 River coniferous 2 5 7 grassland 0 2 2 mixed 1 5 6 shrubland 0 4 4 Grand Total 24 118 142 23 Following this, I received spatial coordinates from Alberta Environment and Parks of trailside treatments of exotic species by Parks staff in 2018 and 2019 (hand pulling or herbicide application). I excluded any of the randomly chosen points within 100 m of these areas to avoid potential effects of weed treatments on the presence of exotic species. I chose trailside transects to survey from the randomly selected survey points based on the goal of surveying a similar number of different vegetation types within the different watersheds represented in CCWPP. Logistical constraints related to the feasibility of hiking from access points limited the number of sites surveyed. I chose the sites of control transects based on sampling sites that I used to test a species distribution model for Botrychium (see Chapter 3). This allowed me to efficiently obtain a sample of transects located at least 100 m from any trail (mean distance = 394 m ± 43 m) as a control for trailside transects, while also being able to collect data for Chapter 3. At each of the 24 50 m x 50 m plots that I surveyed to test the Botrychium model, I also set up a transect identical to the trailside transects running due north for consistency, and recorded the same data precisely as collected for the trailside transects. Because these 24 sites were chosen from a set of randomly selected plots with a range of predicted habitat suitability for Botrychium, they were not stratified by vegetation type or watershed. However, the surveyed control transects do represent all watersheds, and all vegetation types. 2.3.3 Data Collection I used a Garmin eTrex® 20 handheld GPS to navigate to each trailside and control transect site. Because I was unable to acquire an accurate shapefile of trails in CCWPP, the trail layer I used was subject to digitizing error, so the final locations of the 24 trailside transects occasionally had to be moved such that they began directly on a trail. Therefore, I took GPS coordinates of the final locations of all trailside transects so they can be precisely re-located and resurveyed for future comparisons (Appendix 1, Table A1). At each trailside location, I placed a transect on the side of the trail closest to the GPS coordinates. I determined the start (0 meters) of the transect based on the point where vegetation was visibly more continuous in contrast to the trampled trail surface while looking down the length of the trail. I then laid out an 11m transect perpendicular to the trail (Figure 3). For control transects, I used the GPS coordinates of the center of the 50 m x 50 m survey area as the start of the transect (0 meters). I recorded elevation of each surveyed transect from the GPS unit and measured the aspect and slope using a compass and clinometer. For trailside transects, I measured trail width and depth (depth of deepest part of trail surface relative to ground directly beside the trail) using a measuring tape. I took soil compaction measurements (kg/cm2) in the center of the trail and any ruts located on the trail, using a pocket spring-operated soil penetrometer (5DPJ8 Humboldt). Wider trails with greater soil compaction and deeper ruts generally have higher frequency and/or intensity of use (Dale & Weaver 1974; Trip & Wiersma, 2015). I took photos from the 0 m point looking towards the end of each transect, as well as images of the trail from both directions for future reference and research. For each trailside and control transect, I chose which side of the transect to lay the quadrats based on which side had the fewest obstructions (i.e., deadfall, boulders, or large standing trees which would skew the representation of the understory species present in that area) by visually looking down the transect line. I then used a 1 m x 1 m quadrat 25 (constructed of light-weight PVC pipe) to sample the plant community at 0 m (directly trailside), 2 m, 5 m, and 10 m along the transect, placing all quadrats on the same side. I selected 10 m as the most distant sampling point because most studies of trail effects on plant communities indicate no significant trail impact on species richness beyond 5 m from roads or trails (Watkins et al., 2003; Godefroid & Koedam, 2004; Dickens et al., 2005; Ngugi et al., 2014). Figure 3. A schematic diagram of a trailside transect. An 11 m transect (red line) runs perpendicular to the trail. Each 1 m x 1 m quadrat (black squares) represents a sample plot placed at 0 m, 2 m, 5 m, and 10 m away from the start of the transect. In each quadrat, I took measurements of maximum understory vegetation height (vegetation under 2 m tall) and soil compaction (using the same pocket penetrometer as above). I recorded the presence of each vascular plant understory species (less than 2 m tall) and following methods outlined in Stohlgren et al. (1999), I estimated the percent cover of each vascular plant species, bare ground, and moss or lichen in each of the quadrats by training myself and my field assistant to recognise 1% of a 1 m x 1 m area and estimate to the nearest percent. I took photographs of each quadrat from above. I also 26 collected plant samples and took photos of species not easily identifiable in the field for later identification in the lab. I identified all vascular plant species (excluding sedges) using ‘Vascular Flora of Alberta’ (Kershaw & Allen, 2020) and ‘Flora of Alberta’ (Moss & Packer, 1983) and sedge species using ‘Field guide to Intermountain sedges’ (Hurd et al., 1998). Nomenclature of identified species is based on Canadensys’ online Database of Vascular Plants of Canada (Brouillet et al. 2010+) and NatureServe explorer 2.0 (NatureServe, 2022). I determined the origin (exotic or native) and provincial conservation rank (S- rank) of each species using Alberta Conservation Information Management System (September 2018 version: ACIMS, 2018). Exotic species, also referred to as non-native species, are defined by NatureServe as species found outside their native range, whose presence in a natural ecosystem is due to direct or indirect human intervention (Morse et al., 2004). The conservation ranking of each species is based on NatureServe’s conservation status assessment methodology which focuses on the rarity, threats, and trends of a particular species (Faber-Langendoen et al., 2012). The S-rank refers to ‘subnational’ conservation ranks where S1 defines species as critically imperiled provincially with a very high risk of extirpation; S2 refers to imperiled, high risk of extirpation; S3 refers to vulnerable, moderate risk of extirpation; S4 refers to apparently secure; and S5 refers to species that are secure, with little to no risk of extirpation (NatureServe, 2022). For this study, I define provincially rare species as any S1- or S2- ranked species. 2.3.4 Statistical analyses 27 I used measurements of trail width to define trail type. I classified trails less than 1m wide as footpaths; trails observed to have tire ruts and ranging in width from 1.1 m to 3.9 m as OHV trails; and trails greater than 3.9 m as roads. Width requirements for single lane roads in Alberta are a minimum of 4 m (Alberta Infrastructure and Transportation, 1996). While I stratified site selection by a GIS layer of vegetation types provided by ABMI (Castilla et al., 2014), I found that the actual vegetation type at the selected transect sites sometimes differed from this layer. Therefore, I used my on-the-ground assessment of vegetation type in all analyses. After completing the surveys, I found that we were not able to survey any footpath transects in broadleaf vegetation; therefore, I lumped the vegetation types ‘broadleaf’ and ‘mixed’ into one category – called ‘mixed’ – for analyses. The ‘mixed’ vegetation type therefore included ‘mixed’ vegetation, where neither coniferous nor broadleaf trees are more than 75% dominant in the canopy, and ‘broadleaf’ vegetation, where broadleaf trees are more than 75% dominant. To test the effect of trail proximity on vegetation height and soil compaction, I compared the mean maximum height of vegetation and mean soil compaction at each distance category using estimated marginal means. To determine whether different trail types had different effects on vegetation height and soil compaction, I conducted pairwise comparisons of the estimated marginal mean maximum vegetation height and mean soil compaction values, respectively, between trail types at each distance and between distances for each trail type using the Tukey adjustment for multiple tests (Wright, 1992). I used mixed models with transect as the random effect to test the effects of distance from trail, trail use type, and vegetation type on species richness, community composition, the presence of at least one exotic species, and the presence of at least one 28 S1 or S2 provincially tracked species. I also tested the effects of interactions between distance from trail and trail type, and distance from trail and vegetation type. I built one model for each response variable: species richness, community composition, the presence of at least one exotic species, and the presence of at least one rare species. Species richness is the total number of vascular plant species recorded in each quadrat. As a measure of changes in community composition near the trail compared to away from the trail, I calculated the Bray-Curtis dissimilarity between the 10 m quadrat and each of the other three quadrats on the same transect. I calculated the Bray-Curtis dissimilarity using the square-root transformed abundance of each species in each quadrat. A Bray-Curtis dissimilarity value of 0 indicates identical community composition between quadrats, whereas a value of 1 indicates no species in common (Bray & Curtis, 1957). For example, if there was very little difference in community composition comparing the 10 m quadrat to the 0 m quadrat, the Bray-Curtis dissimilarity between these two quadrats would be close to zero. Quadrats with at least one exotic species received a ‘1’ for exotic species presence, while those with no exotic species received a ‘0’. Similarly, quadrats with at least one rare species recorded received a ‘1’ for rare species presence, while those with no rare species present received a ‘0’. All response and predictor variables are listed in Table 2. In each model, I included covariates that might affect the response variable in addition to the predictors of interest. For the species richness model, I included elevation as a covariate because higher elevations tend to have fewer species due to harsh climatic conditions and because lower elevations tend to have more human disturbance which introduces more species compared to less disturbed, high elevation areas (Pauchard et al., 29 2009). For exotic species presence, I included elevation because the occurrence of exotic species tends to decline with elevation (e.g., Becker et al., 2005; Pauchard et al., 2009; Medvecká et al., 2018). Table 2. Predictor variables used in mixed models for each response variable. Response Predictor Variables Variable type Variable Species Elevation Continuous (meters) richness Northness Continuous (index) Vegetation type Categorical (grassland, shrubland, mixed, coniferous) Distance from trail Continuous (meters) Trail type Categorical (control, footpath, OHV, road) Distance from trail X Vegetation Continuous X Categorical type Distance from trail X Trail type Continuous X Categorical Bray-Curtis Distance from trail Continuous (meters) Dissimilarity Trail type Categorical (control, footpath, OHV, road) Vegetation type Categorical (grassland, shrubland, mixed, coniferous) Distance from trail X Trail type Continuous X Categorical Distance from trail X Vegetation Continuous X Categorical type Probability Elevation Continuous (meters) of at least Northness Continuous (index) one exotic Vegetation type Categorical (grassland, shrubland, species mixed, coniferous) Distance from trail Continuous (meters) Trail type Categorical (control, footpath, OHV, road) Distance from trail X Vegetation Continuous X Categorical type Distance from trail X Trail type Continuous x Categorical Probability Elevation Continuous (meters) of at least Distance from trail Continuous (meters) one rare Vegetation type Categorical (grassland, shrubland, species mixed, coniferous) Trail type Categorical (control, footpath, OHV, road) Distance from trail X Vegetation Continuous X Categorical type Distance from trail X Trail type Continuous X Categorical 30 I also included aspect in both models because within plant communities studied in southern Alberta, species richness, regardless of species origin, is reported to be higher on north-facing slopes which tend to retain more moisture than south-facing slopes (Lieffers & Larkin-Lieffers, 1986). I transformed the predictor variable aspect to a linear variable, ‘northness’ index (Equation 1), where north-facing transects have a value of 1, south-facing transects have a value of 0, and east and west are equally counted as 0.5. I did not include elevation or aspect as covariates in the model for Bray-Curtis dissimilarity because the magnitudes of shifts in community composition over space is not expected to differ with elevation or aspect. I included only elevation as a covariate in the model for presence of rare species because some studies report increased presence of rare species at higher elevations due to reduced competition and less disturbance (Lomolino, 2001; Vetaas & Grytnes, 2002; Pauchard et al., 2009). Equation 1 To determine trail effects on species richness and community composition, I used linear mixed-effects models (LMMs). I first confirmed that there were no strong correlations between any of the predictors using Pearson’s correlation coefficients. The strongest correlation was a positive correlation between elevation and northness (r = 0.14). Then, for each response variable, I built a LMM with transect as a random factor to account for non-independence of quadrats from the same transect. I standardized the continuous predictors by subtracting the mean and dividing by two standard deviations. I used diagnostic plots to ensure that model assumptions were met for each model and constructed spatial correlograms to ensure no spatial autocorrelation in the residuals (Bjørnstad & Falck, 2001). To test the effect of each predictor or interaction while 31 accounting for the other predictors, I used a drop1 test to perform marginal fitting of terms. The drop1 test compares a model without the predictor of interest to the full model. If interactions were not significant, I re-fit the model without them and re-ran the drop1 test. If distance from trail was a significant predictor, I re-fit the model with distance as a categorical predictor and used post-hoc pairwise Tukey tests to determine which pairs of distances differed significantly (Wright, 1992). As a measure of the variance explained by each model, I calculated the percent null deviance explained using Equation 2, where null deviance is the deviance of the intercept-only plus random effects model: Equation 2 To determine trail effects on the presence of exotic species and provincially tracked species, I used the same approach as for species richness and community composition (above), except I used generalized linear mixed models (GLMMs) with a logit link because exotic species and rare species presence/absence are binomial responses. I checked for model specification errors and overdispersion using scaled residuals for each model (Hartig & Hartig, 2017) and again confirmed no spatial autocorrelation in model residuals using correlograms (Bjørnstad & Falck, 2001). When I included all the two-way interactions in the model for rare species presence/absence, the model failed to converge. Therefore, I included each interaction one at a time and used a drop1 test to determine if it was a significant predictor, then built the final model with only the significant interactions. As a measure of the variance explained by each model, I once again calculated the percent null deviance using Equation 2. 32 For all response variables, I predicted that there could be an interaction between vegetation type and trail type, such that the effect of trail type could vary depending on the vegetation type. I also predicted that the effect of distance from trail on the response variables could vary depending on the trail type and vegetation type combination. However, my sampling did not achieve replication in all trail type by vegetation type combinations. Therefore, I used the subset of data including only OHV trails and footpaths – which were the most common trail types - to test for an interaction between vegetation type and trail type, and for a 3-way interaction between vegetation type, trail type, and distance from the trail. I followed the same approach above to determine whether any of these interactions were significant. If there was a significant 3-way interaction, I used estimated marginal means of linear trends to test for significant differences in the slope of the relationship between the response variable and distance from trail for different combinations of vegetation type and trail type. To test whether trail use is facilitating the colonization of higher elevation sites by exotic species, I calculated the presence or absence of at least one exotic species on each transect by lumping the data for all four quadrats on each transect. I then built a generalized linear model (GLM) with a logit link to model presence/absence of at least one exotic plant per transect based on the predictors: elevation, vegetation type, trail type, and interactions between elevation and vegetation type as well as elevation and trail type. If the interaction between elevation and trail type is significant, it would suggest that some trail types are facilitating the colonization of higher elevation sites by exotic species more than others. As above, I ensured that there were no model specification issues and 33 no spatial autocorrelation in model residuals. Once again, I used a drop1 test to determine which predictors were significant, while accounting for all other predictors in the model. I carried out all statistical analyses using the statistical software R version 4.0.3 (R Core Team, 2020). I used the packages ‘lmer4’ (Bates et al., 2015), and ‘glmmTMB’ (Magnusson et al., 2017) to build GLMMs, ‘ncf’ to build spatial correlograms (Bjørnstad & Falck, 2001), ‘DHARMa’ to test for model misspecification (Hartig & Hartig, 2017), ‘arm’ to standardize predictors (Gelman et al., 2013), ‘emmeans’ to carry out post-hoc pairwise tests (Lenth et al., 2019), ‘vegan’ to calculate Bray-Curtis dissimilarity (Oksanen et al., 2013), ‘ggplot2’ to create graphs of soil compaction and maximum vegetation height, and ‘visreg’ to create partial regression plots for visualizing significant effects in each model (Breheny and Burchett, 2017). 2.4 Results The average on-trail soil compaction was 3.9 ± 1.7 kg/cm2 on footpaths, 4.1 ± 1.4 kg/cm2 on OHV trails, and 4.9 ± 0.4 kg/cm2 on roads. The maximum height of understory vegetation was higher at 10m compared to the trail edge for shrublands and mixed vegetation but declined with distance from trail in coniferous forests and did not change in grasslands (Figure 4A). Although vegetation height tended to increase moving away from trails, only for roads was the average maximum height of vegetation significantly higher at 10 m compared to 0 m (Figure 4B, Appendix 1, Table A1.2). Soil compaction declined moving away from trails in all vegetation types, with the lowest levels of soil compaction in coniferous forests (Figure 4C). The average soil compaction beside roads was higher at 0 m and 2 m compared to other trail types and control transects (Figure 4D). There was no significant difference in soil compaction between footpaths and OHV 34 trails within 0 m quadrats. At 2 m, soil compaction was significantly higher for roads than all other trails and control quadrats (Appendix 1, Table A1.3). There was no change in soil compaction moving from 0 m to 10 m on control transects. On footpaths, only the 0 m quadrat was significantly higher than the other distances. On OHV trails and roads, soil compaction was significantly elevated in the 0 m and 2 m quadrats compared to 5 m and 10 m (Appendix 1, Table A1.4) Figure 4. The maximum height of understory vegetation (A-B) and soil compaction (C- D) with increasing distance from trails for (A & C) vegetation types and (B & D) trail use types. Control transects are excluded from A and C. Points represent mean values; error bars show +/- 1 standard error. We recorded 388 plant species in 568 quadrats within 142 transects. 35 species were exotic; one species was ranked S1, and 14 species were S2 provincially tracked rare species (see Appendix 1; Table A1.5 for a list of all exotic species, and Table A1.6 for a list of all rare species). I could not identify five specimens to species. The most frequent species were Fragaria virginiana Mill. (wild strawberry, 245 quadrats), Taraxacum 35 officinale (221 quadrats), Achillea millefolium L. (Common yarrow, 217 quadrats) and Phleum pratense (202 quadrats). The most frequent native species found in quadrats were Fragaria virginiana, Achillea millefolium, Spiraea lucida (192 quadrats), Symphyotrichum laeve (L.) Á. Löve & D. Löve (smooth blue aster, 180 quadrats), and Galium boreale (L.) (northern bedstraw, 164 quadrats). The most frequently recorded exotic species in trailside transects were Taraxacum officinale, Phleum pratense, and Poa pratensis (88 quadrats). These species were also recorded in 15, 9, and 15 of the 24 off- trail control quadrats, respectively. The exotic species Plantago major (L.) (Nipple-seed plantain), Poa annua (L.) (Annual bluegrass), Alyssum alyssoides (L.) L. (Pale alyssum), Matricaria discoidea DC. (Pineapple-weed chamomile), and Echium vulgare L. (Common viper’s-bugloss) were only found directly trailside (0 m), and never occurred in quadrats 2 m or more from a trail or in any of the control transects. Among the 35 recorded exotic species, 21 were never recorded in control transects. Although Verbascum thapsus L. (Common mullein) was found only at intermediate distances from a single trail (2 m and 5 m quadrat), it was also found in one control transect. I recorded exotic species at elevations ranging from 1,338 m a.s.l to 1,914 m a.s.l, with Taraxacum officinale having the widest range (1,338 m a.s.l to 1,914 m a.s.l). I recorded one provincially tracked S1 species, Microsteris gracilis (Hook.) Greene (slender phlox) in two quadrats of different trail transects. The most frequently recorded provincially tracked S2 species was Melica subulata (Griseb.) Scribn. (Alaska oniongrass, 22 trail and 4 control quadrats), followed by Festuca occidentalis Hook. (Western fescue, 8 trail and 2 control quadrats), Paxistima myrsinites (Pursh) Raf. (Oregon boxleaf, 7 trail and 3 control quadrats), and Carex geyeri Boott. (Geyer’s sedge, 36 6 trail and 4 control quadrats). Overall, I found at least one rare species in one quadrat on a roadside transect (at the 10 m distance; 2% of all road quadrats), 16 quadrats along footpaths (18% of all footpath quadrats), 22 control quadrats (23% of all control quadrats), and 41 OHV trail quadrats (13% of all OHV quadrats). The number of species in a quadrat ranged from zero to 29. Two quadrats had no species present. Both were at the 10 m distance; one was a rocky, dried streambed, whereas the other was the interior of an old coniferous forest. The number of exotic species in a quadrat ranged from zero to nine, with a mean of 1.7 species. Only one quadrat had nine exotic species, and it was the 2 m quadrat of a transect located in the vicinity of a popular campsite where RVs (recreational vehicles) and OHVs were present. Most quadrats had zero or one rare species. However, I recorded three provincially tracked S2 species in a quadrat 5 m from a footpath trail in a mixed forest. Of all quadrats surveyed, 14% had at least one S1 or S2 provincially tracked rare species present. Of all transects surveyed, 23% had at least one rare species present. 2.4.1 Species richness Significant predictors of species richness included elevation, the interaction between vegetation type and distance from trail, and the interaction between trail type and distance from trail (Table 3; Figure 5). Species richness declined significantly with increasing elevation (Figure 5a), whereas the effect of distance from trail on species richness depended on vegetation type (Figure 5c; Appendix 1, Table A1.7). In grassland and broadleaf/mixed forests, species richness did not change significantly with distance. In shrublands, species richness in the 2 m and 5 m quadrats was significantly higher than in the 10 m quadrat. In coniferous forests, species richness in the 0 m and 2 m quadrats 37 was significantly higher than in the 10 m quadrat. The effect of distance from trail on species richness also varied with trail type (Figure 5d; Appendix 1, Table A1.8). In control transects more than 100 m from any trail, there was no significant change in species richness from 0 m to 10 m. On footpath transects, species richness was significantly higher at 0 m and 2 m compared to 10 m. On OHV transects, species richness was significantly higher at 0 m, 2 m, and 5 m compared to 10 m. On roadside transects, the 0 m quadrat had significantly lower species richness than 2 m, 5 m, or 10 m away. The model for species richness explained only 3.4% of the null deviance. Table 3. Results of the model for species richness. I included transect as a random effect to account for non-independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value** (Intercept) 11.77 1.42 3379.5 n/a elevation -2.89 0.79 3391.3 <0.001 aspect (northness) -0.63 0.76 3378.3 0.408 vegetation type (shrubland) 2.24 1.32 n /a n/a vegetation type (mixed) 1.17 1.23 vegetation type (coniferous) -0.83 1.25 distance 3.41 1.21 n/a n/a trail type (footpath) 2.03 1.30 n /a n /a trail type (OHV) 1.31 1.04 trail type (road) -0.45 1.58 vegetation type (shrubland) x -3.91 1.16 distance vegetation type (mixed) x distance -2.68 1.03 3390.9 <0.001 vegetation type (coniferous) x -4.30 1.08 distance trail type (footpath) x distance -4.38 1.14 trail type (OHV trail) x distance -2.93 0.90 3410.6 <0.001 trail type (road) x distance 2.34 1.34 *AIC values include all factors in the model except the one being tested. In marginal fitting of terms, main effects also included in an interaction were excluded. ** P-values indicate the significance of each predictor using a drop1 test. I did not drop individual predictors that were also in a significant interaction. Predictors with p < 0.05 are indicated in bold text. 38 Figure 5. Partial regression plots showing the effect of a) elevation, b) northness, c) the interaction of distance from trail and vegetation type, and d) the interaction of distance from trail and trail type on the species richness within each quadrat. Northness was not a significant predictor according to the drop1 test. For all panels, all other variables are held at the median value (for continuous variables) or the most common category (for categorical variables). For the subset including footpaths and OHV trails only, significant predictors of species richness were elevation and the three-way interaction between distance from trail, vegetation type, and trail type (Table 4). Species richness once again declined significantly with increasing elevation, as in the full dataset. The effect of distance from trail on species richness depended on both vegetation type and trail type (Figure 6). Results of estimated marginal means for linear trends (Appendix 1, Table A1.9) indicate significantly steeper change in species richness moving from 10 m to 0 m for footpaths 39 compared to OHV trails in shrublands (p = 0.008) and mixed vegetation types (p = 0.025). In grasslands and coniferous forests, the slopes of species richness with distance from footpaths and OHV trails were not significantly different (p = 0.647 and p = 0.087, respectively). The model explained 4.3% of null deviance in species richness. Table 4. Results of the model for species richness of footpaths and OHV trails only. I included transect as a random effect to account for non-independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value** (Intercept) 13.89 1.19 2469 n/a elevation -2.24 0.91 2473.4 0.016 aspect (northness) 0.03 0.86 2467 0.974 vegetation type (shrubland) 1.23 1.51 n/a n/a vegetation type (mixed) 0.91 1.44 n/a n/a vegetation type (coniferous) -1.53 1.38 n/a n/a distance 0.69 1.06 n/a n/a trail type 1.40 2.33 n/a n/a vegetation type (shrubland) x distance -3.50 1.38 n/a n/a vegetation type (mixed) x distance -3.47 1.23 n/a n/a vegetation type (coniferous) x distance -5.18 1.24 n/a n/a distance x trail type 0.97 2.11 n/a n/a vegetation type (shrubland) x trail type -1.36 3.07 n/a n/a vegetation type (mixed) x trail type -2.16 3.11 n/a n/a vegetation type (coniferous) x trail type -4.43 3.00 n/a n/a vegetation type (shrubland) x distance x 4.03 2.81 trail type vegetation type (mixed) x distance x trail 3.12 2.78 type 2475.4 0.007 vegetation type (coniferous) x distance x -3.90 2.71 trail type *AIC values include all factors in the model except the one being tested. In marginal fitting of terms, main effects also included in an interaction were excluded. ** P-values indicate the significance of each predictor using a drop1 test. I did not drop individual predictors that were also in a significant interaction. Predictors with p < 0.05 are indicated in bold text. 40 Figure 6. Partial regression plots showing the effect of distance from trail on species richness for footpaths versus OHV trails within each vegetation type. Asterisks indicate significant differences (p < 0.05) observed between slopes for footpath compared to OHV transects based on a test of the estimated marginal means for linear trends. For all panels, all other variables are held at the median value. 2.4.2 Community composition Significant predictors of the difference in community composition (as measured by Bray-Curtis dissimilarity values) of 0 m, 2 m, and 5 m quadrats compared to the 10 m quadrat included vegetation type and the interaction between distance from trail and trail type (Table 5). The mean community dissimilarity of all quadrats compared to the 10 m quadrat was significantly lower in grasslands compared to all other vegetation types (Figure 7a; Appendix 1, Table A1.10). The effect of trail type on community composition varied with distance from trail (Figure 7b; Appendix 1, Table A1.11). At 0 m, the shift in composition from 10 m was significantly greater for roads and OHV trails compared to control transects. At 2 m and at 5 m, the shift in composition was not significantly greater than the control transects for any type of trail transect. The model explained 45.1% of deviance in community composition shifts. 41 Table 5. Results of the model for shifts in community composition (Bray-Curtis dissimilarity) compared to the 10 m quadrat. I included transect as a random effect to account for non-independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value** (Intercept) 0.42 0.05 -419.67 n/a distance 0.00 0.01 n/a n/a trail type (footpath) 0.11 0.05 trail type (OHV) 0.18 0.04 n /a n/a trail type (road) 0.24 0.06 vegetation type (shrubland) 0.17 0.05 vegetation type (mixed) 0.13 0.04 -409.53 <0.001 vegetation type (coniferous) 0.16 0.04 distance x trail type (footpath) -0.02 0.01 distance x trail type (OHV) -0.04 0.01 -395.87 <0.001 distance x trail type (road) -0.05 0.01 *AIC values include all factors in the model except the one being tested. In marginal fitting of terms, main effects also included in an interaction were excluded. ** P-values indicate the significance of each predictor using a drop1 test. I did not drop individual predictors that were also in a significant interaction. Predictors with p < 0.05 are indicated in bold text. Figure 7. Partial regression plots showing the effect of a) vegetation type and b) the interaction of distance from trail and trail type on Bray-Curtis dissimilarity of each quadrat compared to the 10 m quadrat. For all panels, all other variables are held at the median (for continuous variables) or the most common category (for categorical variables). For the data subset including footpaths and OHV trails only, significant predictors of shifts in community composition included the interactions between distance from trail 42 and vegetation type, distance from trail and trail type, and vegetation type and trail type, but not the 3-way interaction. Shifts in community composition moving from 10 m quadrats towards trails were smallest in grassland plant communities, whereas shrublands had the steepest shift in community composition moving towards the trail (Figure 8a). There was a significantly greater shift in community composition moving from 10 m to 0 m for OHV trails compared to footpaths (Figure 8b). In grasslands, mixed forests, and coniferous forests, transects on OHV trails had greater shifts in community composition compared to footpaths (Figure 8c). Interestingly, in shrublands, this pattern was reversed (Figure 8c). However, pairwise tests comparing mean dissimilarity of plant communities on OHV transects compared to footpaths within each vegetation type indicate significant differences between footpaths and OHV trails only in grasslands (Appendix 1, Table A1.12). Figure 8. Partial regression plots showing the effect of a) the interaction between distance from trail and vegetation type b) the interaction of distance from trail and trail type, and c) the interaction between vegetation type and trail type on Bray-Curtis dissimilarity of 0 m, 2 m, and 5 m quadrats compared to the 10 m quadrat in each transect, for footpath and OHV trail transects only. For all panels, all other variables are held at the median (for continuous variables) or the most common category (for categorical variables). 43 2.4.3 Presence of at least one exotic species Significant predictors of the presence of at least one exotic species included elevation, vegetation type, and the interaction between distance from trail and trail type (Table 6). The probability of finding at least one exotic species in a quadrat declined with increasing elevation (Figure 9a). Grassland and shrubland quadrats had the highest probability of having at least one exotic species, at nearly 100%. Grasslands also had significantly higher probability of having exotic plants than mixed/broadleaf forests (Appendix 1, Table A1.13). Coniferous forests had significantly lower probability than all other vegetation types (Figure 9c). In control and on roadside transects, the probability of at least one exotic did not change significantly with distance (Figure 9d). Control transects had about 40% probability of supporting exotic species whereas transects near roads had a uniformly high probability of exotic species occurrence, with over 70% probability even at the 10 m quadrat. On footpath transects, the probability of at least one exotic was significantly higher at 0 m and 2 m quadrats compared to 10 m as well as at 0 m compared to 5 m. On OHV transects, exotic plants were more likely to occur at 0 m than 2 m, and at 2 m than 5 m, but there was no significant difference between 5 m and 10 m quadrats in the probability of at least one exotic plant being present (Appendix 1, Table A1.14a). At 10 m, no trail types had significantly higher probability of finding an exotic species than control transects, whereas at 0 m, OHV trails, roads and footpaths had a higher predicted probability of finding an exotic species compared to control transects, although pairwise tests suggest this difference was significant only for OHV trails and footpaths (Appendix 1, Table A1.14b). The model explained 29.4 % of the null deviance. 44 Table 6. Results of the model for the probability of at least one exotic species present. I included transect as a random effect to account for non-independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value (Intercept) 2.74 1.45 476.01 n/a elevation -4.10 0.90 505.9 <0.001 aspect (northness) 0.09 0.66 474.03 0.896 distance -0.06 0.78 n/a n/a vegetation type (shrubland) -2.70 1.33 512.42 <0.001 vegetation type (mixed) -4.28 1.35 vegetation type (coniferous) -7.09 1.53 trail type (footpath) 1.37 1.18 n/a n/a trail type (OHV) 3.11 1.03 n/a n/a trail type (road) 3.13 1.56 n/a n/a distance x trail type (footpath) -4.87 1.41 488.06 <0.001 distance x trail type (OHV) -3.42 0.96 distance x trail type (road) -2.00 1.38 *AIC values include all factors in the model except the one being tested. In marginal fitting of terms, main effects also included in an interaction were excluded. ** P-values indicate the significance of each predictor using a drop1 test. I did not drop individual predictors that were also in a significant interaction. Predictors with p < 0.05 are indicated in bold text. 45 Figure 9. Partial regression plots showing the effect of a) elevation, b) northness, c) vegetation type, and d) the interaction between distance from trail and trail type on the probability of occurrence of one or more exotic species within each quadrat. Northness was not a significant predictor according to the drop1 test. For all panels, all other variables are held at the median (for continuous variables) or the most common category (for categorical variables). For the subset including footpaths and OHV trails only, significant predictors for the probability of having at least one exotic species were elevation (as seen with the full data set) and the 3-way interaction. Pairwise tests of the mean probability at each distance indicate high probability of finding exotic species in grasslands regardless of distance with no significant differences between trail types (Figure 10a). In shrublands, OHV trails have significantly higher probability of supporting exotic species than footpaths at 10 m (Figure 10b), whereas in mixed/broadleaf vegetation, OHV trails have significantly higher probability of exotics at both 5 m and 10 m (Figure 10c). In coniferous forests, at 0 m, the probability of exotic species is higher for OHV trails, 46 however, after 2 m, the probability of exotic species is higher for footpaths (Figure 10d). Results of estimated marginal means for linear trends (Appendix 1, Table A1.15) indicate that the probability of exotic occurrence with distance from the trail is significantly different for OHV trails compared to footpaths in shrublands, mixed forests, and coniferous forests. Figure 10. Partial regression plots showing the 3-way interaction between distance from trail, vegetation type, and trail type on the presence of exotic species within each quadrat for footpaths (blue) and OHV trails (orange). For all panels, all other variables are held at the median. The probability of having at least one exotic species present in a transect (all quadrats lumped) included vegetation type and the interaction between elevation and trail type. In line with the quadrat level results, there was higher probability of exotic presence in grassland transects than all other vegetation types (Figure 11a; Appendix 1, Table 47 A1.16). The probability of finding at least one exotic species in a transect declined with elevation, but the slope of this decline varied for different trail types (Figure 11b). The decline in the probability of finding exotic species with increasing elevation was significantly steeper for footpaths than OHV trails (Appendix 1, Table A1.17). The model explained 47.9% of null deviance. Figure 11. Partial regression plots showing the effect of a) vegetation type, and b) the interaction between elevation and trail type on the presence of exotic species within each transect. Three outliers at high elevations were removed prior to building the model. For all panels, all other variables are held at the median (for continuous variables) or the most common category (for categorical variables). 2.4.4 Presence of at least one rare species The only significant predictor for the probability of at least one rare species was trail type (Table 7; Figure 12). The probability of finding at least one rare species was higher for control transects than for trailside transects, however, post-hoc pairwise tests showed no significant differences between the different trail types (Appendix 1, Table A1.18). The model explained 4.0% of null deviance. 48 For the subset including footpaths and OHV trails only, no individual variable was a significant predictor of the probability of finding at least one rare species once all the other predictors had been accounted for (Table 8). The model explained 1.8% of null deviance. Table 7. Results of the model for the probability of at least one rare species present. I included transect as a random effect to account for non-independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value** (Intercept) -3.15 1.13 408.79 n/a elevation -0.61 0.56 407.99 0.275 distance 0.40 0.30 408.61 0.178 vegetation type (shrubland) 0.80 1.07 vegetation type (mixed) 1.02 1.05 406.64 0.279 vegetation type (coniferous) 1.73 1.02 trail type (footpath) -0.51 0.87 trail type (OHV) -1.31 0.70 410.7 0.048 trail type (road) -3.30 1.48 *AIC values include all factors in the model except the one being tested. ** P-values indicate the significance of each predictor using a drop1 test. Predictors with p < 0.05 are indicated in bold text. 49 Figure 12. Partial regression plots showing the effect of a) elevation, b) distance from trail, c) vegetation type, and d) trail type on the probability of at least one rare species. Non-significant predictors are indicated by dashed lines; trail type was the only significant predictor in the model according to the drop1 tests. For all panels, all other variables are held at the median (for continuous variables) or the most common category (for categorical variables). Table 8. Results of the model for the probability of at least one rare species present for footpaths and OHV trails only. I included transect as a random effect to account for non- independence of quadrats within the same transect. Predictor Coefficient SE AIC* P-value** (Intercept) -4.11 1.35 303.33 n/a elevation -0.24 0.36 301.76 0.5133 distance 0.07 0.18 301.5 0.6815 vegetation type (shrubland) 0.91 1.35 vegetation type (mixed) 1.62 1.33 300.38 0.3839 vegetation type (coniferous) 1.96 1.27 trail type -0.89 0.79 302.55 0.2691 *AIC values include all factors in the model except the one being tested. In marginal fitting of terms, main effects also included in an interaction were excluded. ** P-values indicate the significance of each predictor using a drop1 test. 50 2.5 Discussion My results show that trails are affecting plant communities in CCWPP. I found higher species richness, shifts in community composition, and increased probability of exotic species presence near trails. These patterns, although relatively consistent, varied in different vegetation types and with different trail types. The trail effect was less prominent in grasslands than in other vegetation types, suggesting that vegetation types are affected differently by trails. There were greater shifts in plant community composition near OHV trails than footpaths, and a higher probability of exotic presence 10 m away, suggesting that OHV trails facilitate the spread of exotics out from trails more than footpaths. Importantly, the magnitude and extent of the effect of trails on plant communities sometimes depended on interactions between vegetation type and trail type. For example, OHV trails are associated with elevated species richness and a higher probability than footpaths of exotic species presence 10 m away from the trail edge, but only in mixed/broadleaf and shrubland vegetation, not in grasslands or coniferous forests. These findings show that currently, grasslands are highly invaded even 11 m away from trails, regardless of the trail type. In contrast, the presence of exotic species in mixed/broadleaf vegetation and shrublands beyond 2 m seemed to be facilitated by OHV trails but not by footpaths. If management goals are to reduce the spread of exotic species, managers should prioritize limiting or prohibiting OHV traffic through mixed/broadleaf forests and shrubland vegetation. 2.5.1 Trail impacts on species richness As predicted, I found that species richness generally increased moving toward trails. Across all trail types and vegetation types, on average there were significantly 51 more species at 0m (14.7 ± 5.8) compared to at 10m (11.9 ± 5.6). This increase in species richness moving toward trails could be attributed to one or more factors including higher light levels near trails compared to far from trails (e.g., Bates ,1935; Dale & Weaver, 1974; Tyser & Worley, 1992), increased disturbance near trails preventing competitive dominance by a few species (Larson, 2002; Dickens, 2005), or increased seed supply near trails due to seed dispersal on clothing or fur (Campbell & Gibson, 2001; von der Lippe & Kowarik, 2007; Mount & Pickering, 2009). I did not directly measure trail use intensity or seed availability, however, significantly higher soil compaction levels directly beside all trails compared to control transects provides evidence of greater trail disturbance via trampling near trails. My results align with findings of Benninger-Truax et al. (1992) who found significantly higher species richness at the trail edge compared to their ‘interior’ plot at 5 m, attributing their findings to more disturbance tolerant species near trails. This is contrary to Crisfield et al. (2012) who found lower species richness near trails in alpine meadows, due to fewer alpine species tolerant of trampling. This suggests that an increase in species richness near trails occurs in some vegetation types but not others. In my study, the change in species richness moving towards trails did in fact depend on the vegetation type. The increase in species richness moving from 10 m towards a trail was smaller for grasslands than for all other vegetation types. Different vegetation types have different gradients of light or water availability and consist of plants with traits adapted to such conditions (Dale & Weaver, 1974; Cole, 1978; Hall & Kuss, 1989; Stohlgren et al., 1999; Hill & Pickering, 2009; Meryem et al., 2009). For example, grasslands and meadows with no tree canopy consist mainly of rhizomatous 52 grasses and herbaceous species tolerant of full sun exposure and rainfall. Therefore, in grasslands, more species can establish far from trails where sun and water availability are comparable to conditions found directly beside trails. In contrast, coniferous forests are generally associated with less light and brittle, woody species that must compete for light and water which are limited below dense canopies. In my study, I found that grasslands had nearly no change in the number of species found near trails compared to 10 m away. Mixed/broadleaf vegetation also showed no statistically significant changes in species richness with distance from trails, likely because the canopy is much more open than in coniferous forests, and mottled sunlight can penetrate areas away from trails. The effect of trails on species richness also extended beyond 5 m in shrublands, whereas in coniferous forests, the effect on species richness extended no more than 2 m, suggesting the role of increased light availability near the more open trailside. Patterns of species richness also varied depending on the trail type, indicating the importance of trail disturbance. I found that directly beside trails, species richness was significantly higher for OHV trails and footpaths compared to roads or controls. Additionally, relative to 10 m from trails, species richness increased closer to OHV trails and footpaths and declined closer to roadsides. These results suggest that directly roadside, conditions are too harsh for most species to survive (Wolf & Croft, 2014) likely because soil compaction reduces water availability for seeds (Marion et al., 2016) and hardens the soil surface making it too dense for seeds to penetrate and begin germinating (Alessa & Earnhart, 2000). In my study, soil compaction was highest directly beside roads where species richness was lowest, relative to footpaths and OHV trails where species richness was highest. This matches other studies which found significantly 53 greater species richness along light and moderate use trails compared to heavy use trails (Benninger-Truax et al., 1992), and lightly used or recently abandoned trails compared to high-use trails or undisturbed sites (Parikesit et al., 1995). My results support the idea of an intermediate disturbance effect suggested by Benninger-Truax et al. (1992) and Parikesit et al. (1995), whereby the highest species richness is observed beside footpaths and OHV trails compared to undisturbed or highly disturbed trails, and species richness is lowest directly beside roadsides, which have the highest disturbance levels. Together, these results provide evidence that too much disturbance or compaction negatively impacts species richness directly trailside. While many studies do not find a trail effect on species richness extending beyond 5 m from trails, in CCWPP this effect extends past 5 m from OHV trails but not from footpaths. Therefore, in addition to the effect of trails through different vegetation types on species richness, increased trailside disturbance can also influence the extent to which increased species richness can be found away from trails. My study shows that vegetation type and trail type interact to influence the extent of trail impacts on species richness. For example, I found that in shrublands and mixed or broadleaf forests, the decline in species richness when moving away from footpaths towards intact vegetation is steeper compared to OHV trails. At 10 m, OHV transects had elevated species richness compared to footpaths, suggesting that the trail effect extends farther out from OHV trails in shrublands and mixed/broadleaf forests. In contrast, the slope of the decline in species richness with distance from the trail did not differ between footpaths and OHV trails in grasslands or coniferous forests. Myers & Harms (2009) suggest that disturbance can influence species richness by opening space, increasing the 54 success of propagules arriving to plant communities, and by providing opportunity for propagules from the soil seed bank to establish. Higher propagule pressure from OHVs compared to hikers, in addition to moderate light availability in shrublands and mixed/broadleaf vegetation may result in a larger zone of trail influence for OHV trails in these vegetation types. This is not evident in grasslands—where high light availability promotes more even distribution of all species—nor coniferous forests—where low light availability dramatically reduces the number of propagules able to establish in more shaded environments farther from the trail edge. Species able to colonize quickly and tolerate disturbance are likely the main contributors to the increased species richness observed trailside. Species richness does not reveal which species are contributing to species richness along trails. Therefore, it is important to consider measures of community composition to fully understand how trails affect plant communities. 2.5.2 Trail impacts on plant community composition In my study, recreational trails also affect plant community composition in CCWPP. I predicted that there would be shifts in community composition near trails compared to vegetation 10 m away from the trail edge and that shifts would be smaller in grasslands compared to other vegetation types due to grassland communities having more species adapted to disturbance and high light conditions (Dale & Weaver, 1974; Cole, 1978; Hall & Kuss, 1989; Hill & Pickering, 2009). I found that the change in species composition at 0 m, 2 m, or 5 m compared to 10 m away was lower in grasslands (mean Bray-Curtis dissimilarity of 0.53) than all other vegetation types. The shifts in composition moving towards trails in grasslands was no more than expected from natural variability that occurs in undisturbed vegetation, as measured in control transects. These 55 results suggest that shifts in community composition may be lower for trails in open habitats with high light availability such as grasslands compared to forested habitats. Alternatively, these results may suggest the compositional differences associated with trails have extended farther than 10 m in grasslands in CCWPP. I found that the extent of trail impacts on community composition was also affected by trail type, as suggested by the significant interaction between distance from trail and trail type. Roads and OHV trails were associated with the greatest change in composition at 0 m; the mean community dissimilarity at 0m was 0.79 for roads and 0.74 for OHV trails, 0.66 for footpaths, and 0.55 for control transects more than 100 m from trails. Although roads and OHV trails exhibit a significantly greater shift in composition than expected just based on a shift in distance, this difference is only significant at 0 m. It seems increased disturbance associated with roads and OHV trails supports the establishment of disturbance tolerant species, however, the extent of compositional shifts varies with trail type. My results show that shifts in community composition extend a shorter distance from the trail edge along footpaths compared to OHV trails. For all trail types, the community dissimilarity at 5 m compared to 10 m was no different than control transects. For footpaths, the shift in composition was higher than 5 m only at 0 m, suggesting the trail effect does not extend much past the immediate trail edge. For OHV trails, the community dissimilarity at both 0 m and 2 m were significantly greater than dissimilarity at 5 m, suggesting that changes in composition extend to 2 m but not beyond 5 m from trails. Therefore, trail effects on community composition are not observed at 5 m, indicating that shifts in composition occur less than 5 m from trails. Other studies have noted similar patterns. Wolf & Croft (2014) found that shifts in community 56 composition extended a greater distance from high use trails than from low use trails. The authors attribute these findings to trailside disturbance and increased soil compaction which facilitates the spread of species tolerant of disturbance—a trait commonly associated with exotic species—farther away from higher use trails (Wolf & Croft, 2014). My results indicate that not only is the degree of compositional shifts greater near roads and OHV trails than footpaths, but they extend farther away. In CCWPP, to reduce the severity and extent of compositional shifts in vegetation along trails, managers should limit the number of roads and OHV trails. I also found the interaction between vegetation type and trail type to be a significant predictor of shifts in community composition near trails. Of all vegetation types, grassland showed the lowest shifts in community composition near trails. However, OHV trails through grasslands are associated with significantly greater compositional shifts compared to footpaths. Perhaps this reflects increased dispersal and establishment of disturbance tolerant propagules associated with OHV use relative to footpaths. If Parks managers are concerned with changes to the composition of plant communities near trails, they should minimize OHV trails through grasslands. 2.5.3 Trail impacts on the presence of exotic species As expected, in CCWPP, the probability of exotic species occurrence increases moving toward trails. This falls in line with most studies that have found significantly greater occurrence or abundance of exotics near trails compared to away (e.g., Tyser & Worley, 1992; Potito & Beatty, 2005; Dickens et al., 2005; Lake & Leishman, 2004). These studies associate increased exotic species occurrence near trails with the ability of this group to adapt to disturbance. In my study, I found that compared to trails and 57 undisturbed areas away from trails, roads showed a high likelihood of exotic species 10 m away, suggesting that for roads, I was unable to capture the trail effect threshold and that it likely extends some distance beyond 10 m. Although most studies indicate the effects of park roads and trails on exotic occurrence are within 15 m (Tyser & Worley, 1992; Watkins et al., 2003; Dickens et al., 2005; Gower, 2008) some species can still be found far away from trails. For example, along high-traffic highways and railways near Banff National Park, Hansen & Clevenger (2005) found high frequency of exotics occurring up to 25 m away, suggesting that higher-use roads and trails promote invasion of exotic species well beyond trail edges. In my study, three exotic species also occurred with high frequencies in control transects at least 100 m from trails: Poa pratensis, Taraxacum officinale, and Phleum pratense. These are the same species identified by Tyser & Worley (1992) that occur more than 100 m from backcountry trails in Glacier National Park just south of the border from CCWPP. The authors note that seeds of P. pratensis and P. pratense were likely brought in the 1800s during road construction – giving them centuries to spread. Previous studies have found positive correlations between resident time of exotic species and their spread across the landscape (Castro et al., 2005; Harris et al., 2007; Ahern et al., 2010; Phillips et al., 2010). They are also preferred species among native grazers like deer or elk, who probably assisted the dispersal of their seeds via their dung or fur (Tyser & Worley 1992). The lower elevation grasslands in CCWPP are also subjected to community cattle grazing. Additionally, Phleum pratense and Poa pratensis are listed as good forage value in Alberta’s rangelands (Tannas, 2003). As a result, unintentional spread by cattle may be 58 an additional factor that has contributed to the spread of these two grass species away from trails in CCWPP. I also found that the likelihood of finding at least one exotic species near trails depends on the type of vegetation. On average, grasslands had nearly a 100% probability of at least one exotic occurring, followed by shrublands (98%), mixed/broadleaf (92%), and then coniferous forests (42%). Environmental conditions such as light (McDougall et al., 2018), increased soil pH and decreased nitrates (Gilbert & Lechowicz, 2005) which are found to promote native species presence also promotes exotic species presence (Lonsdale, 1999; Stohlgren et al., 1999; Seabloom et al., 2003; Gilbert & Lechowicz, 2005; McDougall et al., 2018). This matches my results that show the likelihood of exotic species occurring 10 m away from trails is highest in grasslands, where light and water availability is also high, and lowest in coniferous forests, where environmental conditions limit plant growth. In CCWPP, grassland plant communities not only have the lowest compositional shifts regardless of distance from trails, but they also have the highest likelihood of exotic species present 10 m from trails, suggesting that the grasslands surveyed may already be highly invaded by exotic species. In protected parks, the increased presence of exotic species, which has been correlated to increased number of visitors (Lonsdale, 1999), has important implications for mitigating potential exotic species invasions, especially along higher use trails. Trail type also influenced the probability of finding at least one exotic species away from trails. Road transects had the highest likelihood (about 78%) of exotic species occurring at 10 m, followed by OHV trails (40%), and then footpaths (3%). Interestingly, the probability of exotic presence in control sites 100 m away from any trail remained 59 relatively consistent at about 32% even at the 10 m distance which was significantly higher than the probability of exotics occurring 10 m away from footpaths. This supports the idea that a subset of the exotic species observed away from trails are likely dispersed by means other than trail-associated vectors or have been residents for a long time and therefore have had more time to disperse farther from trails than newly introduced species. As an example, both P. pratense and T. officinale have been recorded in Alberta more than 100 years ago and in conjunction with P. pratensis which was found in Alberta as early as 1856 (GBIF, 2022), these three species have the highest frequencies 10 m from trails compared to all other exotics surveyed in CCWPP. Although all trails had significantly higher likelihood of exotics 0 m from the edge than 10 m away, wide, or heavier-use trails may exert a greater disturbance effect on the surrounding vegetation and soils compared to narrow or lighter-use trails (Tyser & Worley, 1992; Potito & Beatty, 2005; Hochrein, 2008; Törn et al., 2009; Zhou et al., 2020), which promotes the establishment of more disturbance tolerant exotics. These patterns can be attributed to increased propagule pressure associated with wider or heavier use trails that can facilitate the wind-mediated dispersal of exotic propagules farther away from the trailside. In my study, additional evidence of a disturbance effect from trails can be seen with soil compaction values that decline with distance from trails. Compared to consistently low compaction values observed across distance at undisturbed (control) sites, all trail types had significantly greater soil compaction directly beside trails, with roads having the greatest compaction. Heavily used roads for example, are often devoid of any vegetation directly roadside, indicating extreme disturbance effects (Hansen & Clevenger, 2005). My results show that some exotic species can tolerate highly compact, disturbed habitats 60 that occur directly beside roads and OHV trails, and that higher use trail types can disperse exotic propagules farther than footpaths. I also found that the extent of the trail effect on the likelihood of finding an exotic species depended on the interaction between vegetation type and trail type. Compared to footpaths, OHV trails had a greater likelihood of supporting exotic species farther from the trail edge, but this was significant only for shrublands and mixed/broadleaf vegetation. This result mirrors the result for species richness, where species richness was elevated farther out from OHV trails than footpaths, but only in these same two vegetation types. In shrublands and mixed/broadleaf vegetation, adequate light far from OHV trails combined with higher propagule pressure are facilitating the spread of exotic species farther than footpaths. Parendes & Jones (2000) found the greatest exotic species prevalence in areas of high light availability and high-use trails compared to sites that had lower light availability or were less disturbed. The authors suggest that in addition to characteristics associated with different vegetation types (e.g., light, water, nutrients), exotic invasions may also depend on characteristics associated with the propagules being dispersed (e.g., dispersal mechanisms, seed morphology, germination requirements; Parendes & Jones, 2000). While the closure of all OHV trails in shrubland or mixed/broadleaf vegetation may not be possible, managers could focus exotic species monitoring near trails through these vegetation types and encourage users to stay on trails. Additionally, managers should consider limiting OHV traffic to reduce the spread of exotic species. 2.5.4 The effect of trails on the presence of exotic species at higher elevations 61 I examined the probability of finding at least one exotic species at the transect level to determine whether trails are facilitating the spread of exotic species upwards in elevation. I found that the decline in probability of finding exotics with increasing elevation was less steep in transects near OHV trails relative to footpaths, which is consistent with the hypothesis that OHV trails are facilitating exotic species spread to higher elevations. Trails not only transport propagules, but they can also channel human disturbance to higher elevations (Barros & Pickering, 2014). Similarly, Pauchard et al. (2009) note that the rate of exotic invasions at higher elevations can be attributed to changes in climatic and nutrient regimes as well as increased propagule pressure from increased access and development of montane regions for recreational use. Seeds can attach to humans, and they can also cling to vehicles and vehicle tires, especially if weather permits muddy trail conditions causing additional seed retention on vehicles (Taylor et al., 2012) and facilitate long-distance dispersal of exotic species (von der Lippe & Kowarik, 2007). In my study, footpaths may not be exerting enough propagule pressure or disturbance to result in significant changes to exotic presence at higher elevations. The higher soil compaction farther out from OHV trails, and the fact that OHV trails have a higher probability of exotic species farther away from the trail support the idea that OHV trails are facilitating exotic spread to higher elevations, whereas footpaths may not exert enough disturbance or propagule pressure to facilitate the spread of exotics to higher elevations. My study is the first in North America to show evidence that the spread of exotics to higher elevations is being facilitated by recreational trails, particularly OHV trails. If park management is concerned with exotic species spreading to higher elevations, exotic 62 control efforts at higher elevations should be directed towards OHV trails. Prevention efforts of cleaning vehicles prior to trail use could also reduce the propagule pressure exerted by OHV traffic. Additionally, if closure of trails is not possible or is too unpopular, limiting the amount of traffic along OHV trails at higher elevations would help reduce the spread of exotics upward. 2.5.5 Trail impacts on the presence of rare species The protection and conservation of rare species is crucial in regions where hotspots of biodiversity intersect with recreation. In CCWPP, 14% of quadrats and 23% of transects had at least one S1 or S2 provincially tracked rare species. The likelihood of rare species across vegetation types ranged from 1% in grasslands to 6% in coniferous forests, although vegetation type was not a significant predictor. Among the most frequent rare species in CCWPP, two are grasses (Melica subulata and Festuca occidentalis), one is a shrub (Paxistima myrsinites), and one is a sedge (Carex geyeri), all of which are upright, perennial species. Based on the habitat descriptions in Moss & Packer (1983) and Kuijt (1982), none of these species are associated with disturbance. Festuca occidentalis and Carex geyeri both exhibited higher frequencies at intermediate distances (2 m and 5 m), whereas Melica subulata and Paxistima myrsintes had increasing frequencies farther from the trail edge. These data suggest that the rare species surveyed occupy a diverse range of niches and vegetation types. The only significant predictor in the likelihood of finding at least one rare species was trail type. Rare species occurred slightly more often in transects more than 100 m from trails (11%), followed by footpaths (7%), OHV trails (3%) then roads, with nearly 0% probability of rare species occurring. Catling & Kostiuk (2011) found higher density 63 of some orchids—Calypso bulbosa var. americana (R. Brown) Leur on trails in Waterton Lakes National Park as well as Epipactis hellborine (L.) Crantz and Goodyera oblongifolia Raf. on trails in Ontario—within 1.5 m of trails, suggesting that some rare plants can tolerate light trampling and compact soil. In CCWPP, I found that footpaths are nearly as likely as sites beyond 100 m from trails to have a rare species present, and they have lower compaction relative to roads but not OHV trails. These data suggest that the rare species I observed can tolerate some level of disturbance beyond the immediate trailside. Within trailside transects only, I did not find evidence of the effect of distance from trails influencing rare species presence; the 10 m quadrat was no more likely to have a rare species present than the 0 m quadrat. Although greater shifts in composition and greater probability of exotics occur at 0 m compared to 10 m for most trails, the likelihood of finding a rare species was the same, regardless of distance from trails. Of the rare species that were found along trails, 6 of 15 (40%) occurred at the trail edge (0 m) in frequencies equal to or greater than frequencies found at 2 m. McIntyre & Lavorel (1994) found a significant negative correlation between the number of rare species and the proportion of exotic species, which was independent of the contrasting effects of habitat factors, suggesting a competitive nature between the two groups of species. These vegetation responses imply that both can withstand some disturbance, and rare species can still establish in vegetation near trails where exotic species are present. Along recreational trails in CCWPP, I found occurrences of both exotic and rare species. Although there is no evidence to suggest that rare species are being outcompeted near trail edges by exotic species, my analyses focused on occurrence rather than abundance. It could be that exotic species do outcompete rare native species, but only if 64 the exotic species are high in abundance. To maintain occurrences of rare species, managers should reduce the level of disturbance associated with recreational trails and avoid implementing new roads. Future studies should consider the abundance of exotic species relative to the abundance of rare species near trails to determine whether higher abundances of exotic species negatively affect the presence or abundance of rare species. 2.6 Conclusion In CCWPP, recreational trails are indeed affecting plant communities. Not only are OHV trails shown to affect the number of species and the likelihood of finding exotic species in plant communities relative to footpaths, but these effects also become more pronounced, depending on the type of vegetation the trail traverses. Overall, for OHV trails compared to footpaths, the increased species richness and exotic species probability observed directly beside trails extends farther in mixed/broadleaf and shrubland vegetation, and community composition is more dissimilar in grasslands. Although the probability of finding at least one rare species was lower near all trails relative to sites 100 m away, it seems they are slightly more likely to occur near footpaths where less disturbance occurs compared to the other trail types. Together, my results indicate that if park management is concerned with recreational trail impacts on plant communities in CCWPP, the number of OHV trails should be reduced, particularly through shrubland and mixed/broadleaf vegetation and additionally, at higher elevations. To improve the likelihood of rare species occurrences and subsequently decrease the chances of exotic species presence, reduce the level of disturbance associated with trails and avoid implementing new roads. 65 CHAPTER 3: Habitat characteristics of known Botrychium occurrences and quantitative analyses of its association with trails 3.1 Abstract Moonwort (Botrychium Swartz) is a genus of ferns which can be found in high diversity in the Rocky Mountains of southwestern Alberta. Observation records of these small, cryptic species may be subject to bias towards well-travelled areas near trails. Castle Provincial Park and Castle Wildland Provincial Park (CCWPP) are two recently established protected areas within this global hotspot of Botrychium diversity. However, prior to 2018, provincial and international databases included fewer than 15 occurrence records of Botrychium throughout CCWPP. Therefore, their frequency in the parks and their habitat preferences were not well understood. I used the pre-2018 occurrence records plus 73 georeferenced photos of Botrychium occurrences noted as part of the Castle Flora project, and an additional 8 georeferenced photos from my own surveys conducted in 2021 to characterize habitat preferences of Botrychium species found throughout CCWPP. I also tested the ability of a species distribution model (SDM) to successfully predict the presence of Botrychium species in CCWPP. I visited 24 sites at least 100 m away from official trails that varied in their predicted habitat suitability and carried out full plant community surveys. I found that most Botrychium occurrences were on south-facing slopes, in grassland vegetation, 10 m-100 m away from trails. I discovered 7 new occurrences in the off-trail surveys. Although 6 of the 7 new off-trail occurrences were found at sites with greater than 40% suitability, the species distribution model was not a significant predictor of Botrychium occurrence. My results show that Botrychium occur across a wide range of vegetation types, topographic conditions, and proximity to trails and there are likely many undiscovered populations in CCWPP. To 66 maximize discoveries of new occurrences, surveys should focus on grassland areas. Additionally, SDMs built for individual species could prove more useful in finding new records of targeted species. 3.2 Introduction Species within the genus Botrychium Swartz (commonly, moonwort), are cryptic, inconspicuous ferns belonging to the family Ophioglossaceae. There are 50 species recognized globally, and over 30 of these occur in North America (Flora of North America (FNA) Ed. Comm., 1993; Farrar, 2011). Moonwort are small in stature, reaching no more than 15 cm tall and often only noticed after a thorough survey of the ground- level vegetation (Wagner & Wagner, 1981; Figure 13d). They are distinguishable by their single, upright green leaf that is divided into two stalks, a sterile leaf-like ‘trophophore’ and a ‘sporophore’ that bears tiny clusters of spherical sporangia which release spores upon maturation (Farrar, 2011; Figure 13b, c). Moonwort spores require a dark environment (below ground) and mycorrhizal associations to germinate and produce individual gametophytes that have both male and female reproductive structures (Whittier, 1973). Once fertilization has occurred, the gametophytes provide nutrients for the below-ground sporophyte (Johnson-Groh et al., 2002b). Eventually, the sporophyte will create its own mycorrhizal associations, allowing moonwort to persist underground – sometimes for several years – until conditions are favourable for the sporophyte to emerge aboveground and photosynthesize (Johnson-Groh et al., 2002b). The diverse morphology and unpredictable belowground period of moonwort has prompted systematic and molecular analyses to investigate the various lineages of species belonging to this genus (Hauk, 1995; Farrar, 2011; Dauphin et al., 2014; Stensvold & 67 Farrar, 2011; Dauphin et al., 2017). Although much is known about the life history of Botrychium species, less is known about their precise geographic distributions and habitat requirements. The geographic distribution of many Botrychium species may be severely underestimated due to lack of observations (Williston, 2001). Limited ranges of some Botrychium species in North America have resulted in them being listed as species of conservation concern. Currently, there are 34 species found in North America with 18 listed as globally vulnerable (G3), imperiled (G2), or critically imperiled (G1) and one (B. subbifoliatum Brack. from Hawaii) as possibly extinct (NatureServe, 2022). In Alberta, 20 of these species occur, 5 which are ranked provincially as vulnerable (S3), 2 as imperiled (S2), and 7 as critically imperiled (S1) (NatureServe, 2022). B. pseudopinnatum W.H. Wagner (false northwestern moonwort) for example, is globally (G1) and provincially critically imperiled (S1) in Ontario, endemic to the northern shore of Lake Superior (NatureServe, 2022). B. x watertonense W.H. Wagner (GNA S1, Waterton moonwort) is endemic to Waterton Lakes National Park, and has greater than expected abundance for a sterile hybrid species – as most hybrid species occur in low abundances (Farrar, 2011). Species distribution maps can help identify gaps in the distribution which may be a factor of geographical barriers (e.g., mountains, dry flat plains) rather than lack of survey effort. However, more observations are needed to determine accurate distributions of Botrychium species. Because of the large morphological variation within single species and their tendency to hybridize, taxonomic differentiation of moonwort species is difficult. Taxonomists rely on a combination of morphological characteristics, chromosome 68 number, and spore size to differentiate species belonging to Botrychium (Hauk, 1995; Farrar, 2011). B. paradoxum W.H. Wagner (G3 S1, peculiar moonwort) for example, is morphologically identified by a stalk that is divided into two identical sporophores (Figure 13a). B. x watertonense - a hybrid between B. hesperium (Maxon & R.T. Clausen) W.H. Wagner & Lellinger (G4 S3, western moonwort) and B. paradoxum – has a trophophore that also bears sporangia, a characteristic unique among moonwort (FNA Ed. Comm., 1993; Hauk, 1995; Lesica & Ahlenslager, 1996; Farrar, 2011). More observations of Botrychium occurrences are needed to fully understand how their morphological complexity is related to genetic diversity, and to correctly differentiate between members of Botrychium and determine species distribution ranges. The habitats of Botrychium are diverse, ranging from open meadows to moist shaded woods at low elevations to alpine meadows at high elevations (Moss & Packer, 1983; FNA Ed. Comm., 1993; Fryer et al., 2022). Although all known species of Botrychium have been described in floras, the precise habitat requirements are often vague and limited to conditions at local occurrences which may not necessarily reflect conditions throughout the species’ entire range. Most species of Botrychium are usually described as preferring some level of disturbance. For example, nearly all descriptions of rare Botrychium species in Alberta refer to ‘roadside’, ‘ditches’, or ‘trailside’ habitats (Fryer et al., 2022). However, these descriptions of habitats could be reflecting the fact that these small statured plants are easier to spot in more open habitats like trail edges, as well as bias due to opportunistic collecting or botanizing near roads or trails. It is uncertain if some Botrychium species indeed require such disturbances, or whether sampling bias is at play. 69 Figure 13. The diversity of Botrychium species found along two popular hiking trails in CCWPP, Alberta, Canada: a) B. paradoxum (S1) showing two fertile sporophores; b) B. lunaria (S5) showing one leafy trophophore growing behind the tall sporophore; c) B. lanceolatum (S4) showing mature yellow sporangia on the sporophore and dentate margins of the trophophore - a, b, and c were all found along South Drywood Creek in the same day- and d) a small B. lunaria no higher than 2 cm in height found along North Drywood Creek. Globally, there are three areas of high moonwort diversity: the Alps in Europe (Dauphin et al., 2014 as cited in Dauphin et al., 2017), the Great Lakes region in Ontario, Canada, and the Rocky Mountains of southern Alberta, Canada (Hauk et al., 2012 as cited in Dauphin et al., 2017). Of these locations, the world’s centre of moonwort 70 diversity is in Alberta’s southern Rocky Mountains (Wagner et al., 1983 as cited in Williston, 2001; Wagner & Wagner, 1994). Two recently established parks are located within this Botrychium hotspot: Castle Provincial Park and Castle Wildland Provincial Park (CCWPP). There are over 2,000 km of roads and trails traversing the Castle region, an area that has a long history of public land use for recreational activities, logging, industrial extraction of oil and gas, and community grazing (Farr et al., 2017). As a popular off-highway vehicle (OHV) destination that will likely see more visitors, and a provincial biodiversity hotspot, it is important to study the effects of trails on the many rare plant species within CCWPP. Despite CCWPP being part of a global hotspot of Botrychium species diversity, a systematic analysis of the habitat preferences of these species in CCWPP has yet to be conducted. Until recently, there was very little floristic work done within CCWPP. In the summers of 2018 and 2019, Dr. John Bain designed and led a vascular plant inventory project for the newly designated parks. His team surveyed sites from all 9 watersheds throughout the two parks, collecting plant specimens from each surveyed site, and taking photos of all Botrychium plants observed during the surveys. As a member of the inventory team, I photographed 73 different occurrences of more than seven Botrychium species found in 5 different watersheds within CCWPP or just outside the official park boundaries (Upper Crowsnest River, Carbondale River, West Castle River, Pincher Creek, and Drywood Creek). I collected 12 different specimens for vouchers that were deposited as part of the CCWPP collection within the University of Lethbridge (LEA) Herbarium. Prior to this inventory project, known occurrences of Botrychium in CCWPP were limited to fewer than 15 localities total from GBIF (Global Biodiversity Information 71 Facility) and ACIMS (Alberta Conservation Information Management System), all in the southwest and southeast corner of the parks. During surveys conducted in 2021 as part of a trail impact analysis on plant communities in CCWPP (Chapter 2), I found an additional 8 occurrences (1 near trails and 7 off-trail) of Botrychium. In total, these 81 records provide a much larger sample of occurrences and allow for a more accurate characterization of Botrychium habitat preferences in CCWPP. My research aims to improve the current limited knowledge of Botrychium species in this biodiverse area by characterizing the environmental conditions and plant communities in CCWPP associated with Botrychium occurrences. In addition, I test the efficacy of a species distribution model (SDM) to target sites with a high likelihood of suitable habitat for Botrychium occurrences. SDMs are predictive models that use occurrence data of target species and associated environmental variables to predict species distributions (habitat suitability or probability of occurrence of target species) within a specified region (Guisan & Zimmerman, 2000). SDMs have been used effectively to target rare plant surveys in other regions (Williams et al., 2009; Gogol- Prokurat, 2011; McCune, 2016). By investigating the habitat preferences of Botrychium occurrences in CCWPP, we can improve our understanding of their potential distribution across different habitats. The specific objectives of my research are: 1) To assess factors associated with Botrychium occurrences in CCWPP, including associated species, vegetation type, presence of disturbance, proximity to trails, and environmental variables such as elevation, aspect, and slope. 72 2) To test a species distribution model as a predictor of Botrychium occurrences away from recreational trails in CCWPP and, if any species of Botrychium is found, to quantify differences in plant communities at sites with and without Botrychium present. I expect that the species associated with Botrychium occurrences will consist mainly of other species that prefer open vegetation with some disturbance, including for example Fragaria virginiana, Achillea millefolium, and Taraxacum officinale. If Botrychium species do prefer disturbed habitats, I expect that most of the 81 occurrences will be near trails with some disturbance. If Botrychium species are less frequent away from human trails, I expect to rarely find them at sites 10 m or more from trails. If SDMs can efficiently predict habitat preferences, then I expect Botrychium occurrences to be found more often at sites with higher predicted suitability than lower suitability sites. 3.3 Methods 3.3.1 Assessing Botrychium habitat To characterize habitat characteristics of Botrychium species in CCWPP, I used the 73 georeferenced photographs I took from 2017-2020, 8 georeferenced photographs of new occurrences from 2021 (1 of the 8 was from a trailside survey as part of Chapter 2; 7 were from off-trail surveys), and 4 georeferenced records without photographs for a total of 85 different occurrences. First, I compiled information on associated species (except for the 4 records without photographs), proximity of trails, vegetation type, elevation, aspect, slope, evidence of disturbance noted (if any), and their dates of observation. 73 I developed a list of all associated species by examining each photograph and identifying any vascular plant species growing near the Botrychium species. To classify occurrences based on the vegetation type they were found in, I used a Geographic Information System (GIS) to determine the land class within which each occurrence is located based on the 2010 Wall-to-Wall Land Cover Inventory layer from the Alberta Biodiversity Monitoring Institute (ABMI; Castilla et al., 2014). I then calculated the total number of occurrences of each Botrychium species within each vegetation type. I used a GIS to measure the distance to the nearest trail from each Botrychium occurrence record to (1) official trails in CCWPP as of 2018, or (2) a layer with all southern Alberta trails as of 2021 which included updated CCWPP trails based on new usage regulations. I used a raster of a digital elevation model with 25 m resolution to determine the elevation, aspect, and slope of each georeferenced point. If occurrences had site descriptions, I noted whether disturbance was mentioned, and the type of disturbance. Only 23 of the 85 occurrence records had site descriptions associated with them. I also recorded the date of each occurrence record. Using these data, I compiled the top 10 most common associated species of all photographed Botrychium occurrences, developed a list of species and the vegetation types they were found in, assessed the range of proximity to trails, median elevation, elevation range for each species and for the genus, the mean slope and aspect, and date of observation for each occurrence. 3.3.2 Testing a Species Distribution Model To test the ability of a species distribution model (SDM) to predict habitat suitability for Botrychium at the generic (genus-only) level, I used an SDM that was built using a total of 148 previous georeferenced occurrences of any Botrychium species 74 throughout the province of Alberta. These records were gathered from the Alberta Conservation Information Management System (ACIMS; n = 39), herbarium records of the area harvested from Global Biodiversity Information Management System (GBIF; n = 13), records of Botrychium species from the University of Lethbridge herbarium prior to 2018 (LEA; n = 7), observations by Jed Lloren during his research in Waterton Lakes National Park (Lloren, 2021; n = 3), observations from Parks Canada ecologist, Robert Sissons (n = 17), observations from the Castle inventory project (n = 69), and 5 occurrences noted during my trailside transects in the summer of 2020 (Chapter 2). Our research technician, Olivia Gauthier, built the SDM using the program MaxEnt (Phillips et al., 2006). This SDM was built as part of a larger project to build SDMs for 42 plant species ranked S1, S2, or S3 in Alberta by NatureServe. MaxEnt is a machine learning program that predicts habitat suitability of individual cells across a region using presence- only data and environmental features (Phillips et al., 2006; Elith et al., 2011). We used 13 environmental predictors (Table 9) and occurrence records with 100 m or less accuracy to build two models for predicting Botrychium habitat suitability. The predictors represent climatic, topographic, soil, and land cover conditions often used to predict plant species distributions. We built two SDM versions: one with the regularization setting at the default of 1, and one with regularization set at 0.5. The regularization parameter determines how strict models are with respect to overfitting; the first model, with a regularization parameter of 1 allowed for a more inclusive fit, whereas the second model with a regularization parameter of 0.5 was more conservative (Phillips et al., 2006). The model resolution was 50 m by 50 m grid cells, which we chose because this is an area that can be thoroughly surveyed in one day of fieldwork. Because some of 75 the 148 Botrychium records occurred within the same 50 m grid cell, this resulted in 111 unique records used by MaxEnt. The model extent included only the province of Alberta, and in addition was restricted to the natural subregions in which Botrychium is known to occur. These include Athabasca plain, upper boreal highlands, alpine, subalpine, montane, upper foothills, foothills parkland, dry mixed grass, foothills fescue, northern fescue, and mixed grass (Figure 14, inset map). For each model, we excluded 25% of the observations to use as test data. We set MaxEnt to repeat this procedure 10 times and take the average prediction from these 10 replicates. We used the cumulative model output, which avoids assumptions about the species’ prevalence (Phillips et al., 2006). This resulted in a raster layer in which each 50 m x 50 m grid cell receives a value ranging from zero to 100, with 100 indicating the highest predicted relative habitat suitability (Figure 14). 76 Table 9. Environmental predictors used to build the Botrychium SDM. Predictor Source Original Data Type Aspect Alberta Provincial 25m Raster Digital raster Elevation Model (2017) Elevation Alberta Provincial 25m Raster Digital raster Elevation Model (2017) Slope Alberta Provincial 25m Raster Digital raster Elevation Model (2017) Land Use/Land ABMI Wall-to-wall Land Cover Map 2010 polygon Cover Version 1.0 (ABMIw2wLCV2010v1.0) Surficial Surficial Geology of Alberta, 1:1,000,000 scale polygon Geology (Alberta Geological Survey) NDVI in Alberta W2W Normalized Difference raster October 2016 Vegetation Index (NDVI) (Alberta Biodiversity (an index of Monitoring Institute, 2014) 'greenness') Climate Climate Data For Alberta (monthly climate raster Moisture Deficit normals from 1961-1990; ABMI) Mean Annual Climate Data For Alberta (ABMI) raster Precipitation Mean Annual Climate Data For Alberta (ABMI) raster Temperature Mean Summer Climate Data For Alberta (ABMI) raster Precipitation Mean Warm Climate Data For Alberta (ABMI) raster Month Temperature Number of Climate Data For Alberta (ABMI) raster Frost-free Days Precipitation as Climate Data For Alberta (ABMI) raster Snow 77 Figure 14. Inset: shows model extent (grey shaded) and location of Castle Provincial Park and Castle Wildland Provincial Park in the southwest corner of Alberta, Canada. Main map: CCWPP, coloured based on predicted habitat suitability for Botrychium according to a species distribution model (SDM). Brown lines indicate official trails. Dark grey circles indicate Botrychium occurrences used to build the SDM that were found within the park boundaries (n = 69) or just beyond them (n = 8). The 24 50 m x 50 m off-trail plots are indicated by stars, including plots with no Botrychium species found (white, n = 17), and plots where a Botrychium species was found (red, n = 7). The yellow triangle indicates a new on-trail occurrence found during 2021 trailside surveys. Using ArcMap, I imported the habitat suitability raster for Botrychium from the SDM with regularization set to 0.5. I decided to use this SDM as it provides a more conservative estimate of the extent of suitable habitat throughout CCWPP. I then clipped this raster to include only cells more than 100 m but less than 1,000 m from official trails. I then stratified grid cells into 10 categories of relative suitability (0-10%, 10-20%, etc.). I used the sampling package in R to randomly select 10 cells from each of the first 7 strata and 20 from each of the highest 3 strata for a total of 130 potential survey sites. I chose 78 sites to survey from these 130 randomly chosen cells with the goal of ensuring replication within each stratum and across both park areas, as well as avoiding surveying two sites in the same suitability stratum that were near each other. I successfully surveyed 24 of these off-trail 50 m x 50 m sites with at least two sites in each stratum of predicted suitability (Figure 14). 3.3.3 Data Collection My field assistants and I surveyed the 24 off-trail 50 m x 50 m plots, which also served as a control for trail transects (see Chapter 2). I used a Garmin eTrex® 20 handheld GPS to navigate to the coordinates for the centre of each plot. Logistical constraints based on our ability to hike to distant sites nowhere near trails limited the number of sites I surveyed. At the GPS coordinates, I ran a transect 25 m in each cardinal direction to delineate 4 quadrants within the plot (Figure 15). My field assistant and I then carefully searched each quadrant in turn by using a compass to walk parallel transects approximately 3 m from each other. We looked carefully for Botrychium while also recording all vascular plants observed and estimating their abundance based on a coarse abundance scale with 5 classes: ‘very rare’ (1-2 individuals present), ‘rare’ (2-10 individuals), ‘infrequent’ (>10 individuals but not common throughout entirety of plot), ‘common’ (seen throughout plot but not a dominant species) or ‘dominant’ (dominant species throughout the plot) in each 50 m x 50 m survey site. We also determined the vegetation type of each plot based on our observation of the entire 50 m x 50 m area. If the plot seemed to us a mixture of vegetation types, we assessed it according to the ABMI land class in which it was located. Each survey took two people 2 to 4 hours, depending on the terrain and diversity of species present. I took photographs from each GPS point 79 facing in each cardinal direction. I also photographed difficult to identify species and each Botrychium encountered, noting any species growing in the direct vicinity of the plant. I also took samples of grasses and sedges for identification in the lab. Figure 15. A schematic diagram of a 50 m x 50 m plot. The red dashed lines delineate four quadrants originating from the target coordinates (black circle). In each quadrant, we systematically searched for Botrychium and all other vascular plant species. I identified Botrychium species from both the 50 m x 50 m plots and the floral inventory project by first using a synthesized key specifically for the Botrychiaceae (now Ophioglossaceae) of Alberta (Williston, 2001). I then confirmed all identifications using a more recent dichotomous key ‘Vascular Flora of Alberta’ (Kershaw & Allen, 2020). I also used this key as well as ‘Flora of Alberta’ (Moss & Packer, 1983) to identify all other identifiable vascular plant species from the plots, except for sedges. I identified all sedge species using the ‘Field guide to Intermountain sedges’ (Hurd et al., 1998). I used Canadensys’ online Database of Vascular Plants of Canada (Brouillet et al. 2010+) and 80 NatureServe explorer 2.0 (NatureServe, 2022) for currently accepted nomenclature of identified species. 3.3.4 Statistical Analyses To determine whether the species distribution model is a reliable method of predicting Botrychium species occurrences using the 24 surveyed plots, I built a binomial generalized linear model (GLM) with Botrychium presence (1) or Botrychium absence (0) as the response variable and habitat suitability as the predictor variable. I then tested whether predicted habitat suitability was a significant predictor using a drop1 test. As a measure of the variance explained, I calculated the percent null deviance explained using Equation 1, where null deviance is considered the deviance of the intercept-only model: Equation 1 To test whether the plant community composition differed between plots with versus without Botrychium present, I first created a NMDS (non-metric multidimensional scaling) ordination to visualize the distribution of all 24 50 m x 50 m plots in species space. I set the maximum number of starts for finding a stable solution to 999 and the number of dimensions to three. I ran the NMDS using the Bray-Curtis dissimilarity values between each pair of plots, based on the raw abundance class of each species in each plot. A Bray-Curtis dissimilarity value of 0 indicates identical community composition between plots, whereas a value of 1 indicates complete difference in species found between plots (Bray & Curtis, 1957). The ordination represents these differences in two dimensions, such that more similar plots are closer together in the ordination graph, whereas more dissimilar plots are far apart in the ordination graph. 81 I then used PERMANOVA (permutational multivariate analysis of variance) with 9,999 permutations to test for significant differences in community composition between plots with Botrychium present and with Botrychium absent. I also used PERMDISP (multivariate homogeneity of group dispersions) with 9,999 permutations to test whether beta diversity (degree of compositional variation) within the group with Botrychium present was significantly different than beta diversity within the group with Botrychium absent. To determine whether any of the recorded species in the 24 surveyed plots were significant indicators of the presence of Botrychium, I used an indicator species analysis (ISA). Indicator species analysis uses permutation to determine which species are significantly more frequent and/or abundant in one group of sites compared to another (Dufrêne & Legendre, 1997; McCune & Grace, 2002). I first defined groups based on Botrychium presence or absence and then conducted an ISA with 9,999 permutations to determine whether any species were indicators of plots with Botrychium or plots without Botrychium. I carried out all statistical analyses using the statistical software R version 4.0.3 (R Core Team, 2020). I used the package ‘vegan’ to construct the NMDS and conduct PERMANOVA and PERMDISP analyses (Oksanen et al., 2020), and ‘labdsv version 2.0-1’ for the ISA (Roberts, 2019). 3.4 Results 3.4.1 Assessing factors associated with Botrychium occurrences in CCWPP 82 Of the 85 georeferenced occurrences of Botrychium, 77 are within CCWPP boundaries – 4 of which had no photo for identification to species – whereas 8 are just outside the boundary along an access trail. I identified 12 different Botrychium species in the 81 georeferenced photographs (Figure 16a, Table 10). Most Botrychium records occurred in grasslands or coniferous forests: 35% of all occurrences were in grasslands, followed by 22% in shrublands and coniferous forests, 7% in developed sites as well as rock/talus sites, 4% in mixed forests, and 2% in broadleaf vegetation (Figure 16b, Table 10). It should be noted that ‘developed’ sites, based on ABMI documentation, may be overestimated based on exaggerated minimum road widths of 60 m (Castilla et al., 2014) and as such, I found that the Botrychium species found within these sites coincided with well used OHV trails or were in the vicinity of Sartoris road, an access point for OHV users entering CCWPP from Crowsnest Pass. Among the 23 records with site descriptions, most mentioned some form of disturbance, for example either being near ‘roads’ or ‘trailside’ or in ‘rocky exposure’, ‘burned’, or ‘weedy’ areas. Observation dates ranged from as early as June 7th to as late as August 12th. The median occurrence date was July 18th. The most frequently occurring species was Botrychium lanceolatum (S.G. Gmel.) Ångstr. (23 occurrences), closely followed by B. lunaria (L.) Sw. (22 occurrences). The least frequent species were B. pallidum W.H. Wagner, B. paradoxum, B. michiganense W.H. Wagner ex A.V. Gilman, Farrar, & Zika (2 occurrences each) and B. spathulatum W.H. Wagner (1 occurrence; Table 10). Based on the nearby species identifiable from photo records, 67 different species were noted to grow in the vicinity of Botrychium species. The most common associated species were Achillea millefolium (identified near 22 different Botrychium occurrences), followed by Fragaria virginiana 83 and Taraxacum officinale (each identified near 21 different Botrychium occurrences; Table 11). One of the least frequent species, B. pallidum had no identifiable associated species; both occurrences appear to be in dense duff of forest understories. The single occurrence of B. spathulatum was also growing in dense duff near a trail and had one associate that is likely Osmorhiza sp., although the individual was too young to confirm. The elevation of occurrences ranged from 1,430 m to 2,245 m above sea level (a.s.l; Figure 17a). 41% of occurrences were found on east to southeast-facing slopes, as 35 occurrences had an aspect around 100 degrees (Figure 17b). The highest frequency of Botrychium occurrences were in areas where the slope was just above 10 degrees (Figure 17c). The highest frequencies of Botrychium occurrences were found greater than 10m from trails (Figure 17d). Figure 16. The frequencies of a) moonwort species; and b) vegetation types of the 85 occurrence records. Note: ‘sp.’ refers to the 4 records where no photograph was available for species identification. The ‘developed’ vegetation type includes industrial sites and roads, although based on ABMI documentation, this category is greatly overestimated (Castilla et al., 2014). 84 Table 10. Table of Botrychium species (rows) identified from photographed occurrences within each vegetation type (columns). Vegetation type Specific epithet ascendens 0 3 0 0 0 0 0 3 campestre 0 1 3 3 0 2 1 10 crenulatum 1 5 0 0 1 0 0 7 lanceolatum 0 2 0 15 0 4 2 23 lunaria 0 3 0 8 0 8 3 22 michiganense 0 0 1 1 0 0 0 2 minganense 1 4 0 0 0 0 0 5 pallidum 0 0 0 0 2 0 0 2 paradoxum 0 0 0 0 0 2 0 2 pinnatum 0 0 1 0 0 2 0 3 sp.* 0 1 0 2 0 1 0 4 spathulatum 0 0 1 0 0 0 0 1 virginianum** 0 0 0 1 0 0 0 1 Total 2 19 6 30 3 19 6 85 * ‘sp.’ refers to the 4 records where no photograph was available for species identification. ** Botrychium virginianum now taxonomically accepted as Botrypus virginianus (L.) Michx. (Hauk et al., 2003). 85 Total talus shrubland mixed grassland developed coniferous broadleaf Table 11. Table of associated species identified in Botrychium photographs (n = 81) and the number of times each species was listed as an associate. Note: Antennaria sp. is total count of all Antennaria species combined as identification to species is difficult without all flowering components. Associated Species Count Achillea millefolium 22 Anaphalis margaritacea 7 Antennaria sp. 8 Fragaria virginiana 21 Galium boreale 6 Medicago lupulina * 8 Packera indecora 8 Penstemon confertus 10 Symphyotrichum leave 9 Taraxacum officinale * 21 * Indicates species exotic to Alberta. Figure 17. Histograms showing the frequency of occurrences of Botrychium (n = 85) across a) elevation (m); b) aspect (degrees); c) slope (degrees); and d) the proximity (m) to any trail recognized under new usage regulations in CCWPP as of 2021. 3.4.2 SDM as a predictor for off- trail Botrychium occurrences In total I documented 350 vascular plant species in the 24 off-trail plots surveyed between June 7th and July 22nd of 2021. I also documented 4 different species of 86 Botrychium in 7 plots ranging in elevation from 1,430 m to 2,000 m a.s.l (Table 12). Botrychium lunaria (common moonwort, S5), was the most often observed (3 plots), followed by B. campestre (prairie moonwort, S3, 2 plots). I observed B. minganense (Mingan moonwort, S4), and B. pallidum (pale moonwort, S2) each only once. The mean percent suitability of sites where Botrychium was present was 53.48 ± 25.97 %, whereas for sites where Botrychium was absent, the mean percent suitability was 41.88 ± 31.45 %. B. pallidum was unexpectedly observed at a low suitability site (habitat suitability = 8.25%). All but two of the off-trail occurrences were in shrubland vegetation. Habitat suitability according to the SDM was not a significant predictor for the presence of Botrychium (Table 13). The explained deviance of the model including SDM predicted suitability was only 2.7%. 87 Table 12. Characteristics of the 24 off-trail plots surveyed in the summer of 2021, in order of increasing predicted habitat suitability according to a species distribution model. Plots in bold are those in which Botrychium was found (‘Present?’). Plot Suitability Elevation Watershed Vegetation Present? Species (%) (m) o3 2.78 1492 Middle grassland no n/a Castle o4 3.45 1429 Middle shrubland no n/a Castle o10 8.25 1430 Upper mixed yes B. pallidum Castle o13 12.38 1374 Carbondale coniferous no n/a o18 18.56 1513 Middle coniferous no n/a Castle o19 19.04 1649 West Castle coniferous no n/a o21 24.58 1773 Upper broadleaf no n/a Castle o23 25.31 1492 Mill Creek coniferous no n/a o25 27.91 1703 Carbondale coniferous no n/a o31 32.54 1632 West Castle coniferous no n/a o33 34.58 1745 Carbondale coniferous no n/a o37 37.07 1591 Upper shrubland no n/a Castle o42 42.78 1479 Drywood broadleaf yes B.minganense Creek o45 43.15 1834 Carbondale shrubland yes B. campestre o54 53.00 1612 Carbondale shrubland yes B. lunaria o57 54.76 1559 West Castle mixed no n/a o68 63.26 1801 Carbondale shrubland no n/a o70 64.27 2000 Drywood shrubland yes B. lunaria Creek o73 71.40 1646 Upper shrubland no n/a Crowsnest o82 74.53 1891 Carbondale shrubland yes B. lunaria o105 88.40 1522 Carbondale shrubland yes B. campestre o106 88.41 1844 Upper coniferous no n/a Crowsnest o123 96.72 1720 Drywood broadleaf no n/a Creek o126 99.19 1762 Carbondale coniferous no n/a 88 Table 13. Predictor variables included in the model for Botrychium habitat suitability. Predictor Coefficient SE AIC* P-value** (Intercept) -1.53 0.90 32.199 n/a suitability 0.01 0.02 30.975 0.379 *AIC values include all factors in the model except the one being tested. ** P-values refer to the significance of SDM suitability based on a drop1 test. 3.4.3 Quantifying differences in community composition of plots with and without Botrychium In the NMDS ordination, plots with Botrychium present appear less clustered together in species space compared to plots with Botrychium absent (Figure 18). Results of the PERMANOVA indicate that plots with Botrychium present do not have significantly different community composition than plots where Botrychium was absent (Table 14). The average dispersion of plots with Botrychium present was 0.497 ± 0.022, whereas for the group of plots with Botrychium absent it was 0.447 ± 0.019. However, results of the PERMDISP analysis indicate that the variance (dispersion) within the group (beta diversity) of plots with Botrychium present does not differ significantly from the variance within the group of plots with Botrychium absent (p = 0.139). 89 Figure 18. NMDS (non-metric multidimensional scaling) ordination of all 50 m x 50 m sites in species space. Sites are coloured based on groupings of any species of Botrychium present (yes; red) or absent (no; black). Ellipses indicate ordination confidence intervals (90%). Table 14. Results of PERMANOVA pairwise tests comparing species composition in plots with Botrychium present and absent. F.model indicates the pseudo-F ratio of between group variance to within group variance. R2 refers to the proportion of variance observed in species composition for the presence/absence of Botrychium that is explained by the model. P-value is based on 9,999 permutational tests using the pseudo-F ratios. Pairs F.Model R2 P-value Absent vs. Present 1.53607 0.065265 0.092 The results of the ISA indicate 4 significant indicator species. One species, Agoseris glauca (pale agoseris), was a significant indicator of plots with Botrychium. Three species, Acer glabrum (Rocky Mountain maple), Lonicera utahensis (Utah honeysuckle), and Maianthemum racemosum (large false Solomon’s seal) were significant indicators of plots without Botrychium (Table 15). Indicator values for all species are tabulated in Table 2B (Appendix 2). 90 Table 15. Significant indicator species of plots with or without Botrychium according to an indicator species analysis with 9,999 permutations. Values include the relative frequency and average relative abundance of species occurring in plots with Botrychium absent or present as well as the group each species has maximum indicator value for Botrychium (Occurrence- present or absent). Scientific Relative Relative Relative Relative Occurrence p- Name frequency frequency abundance abundance value (absent) (present) (absent) (present) Agoseris 0.118 0.571 0.171 0.829 present 0.029 glauca Acer glabrum 0.588 0.143 0.908 0.092 absent 0.049 Lonicera 0.588 0 1 0 absent 0.019 utahensis Maianthemum 0.882 0.286 0.753 0.247 absent 0.019 racemosum 3.5 Discussion My results show that within CCWPP - two recently established parks in an area that has been dubbed the world’s centre for moonwort diversity - at least 13 of the 20 species of Botrychium known in Alberta (NatureServe, 2022) occur. This study contributes more occurrences than were recorded prior to 2018. Botrychium pallidum (S2- imperiled) for example, has only been recorded in 2 areas of the park; my study adds one more occurrence to these previous observations in addition to several more occurrences of other Botrychium species. Because these species are often no larger than 15 cm tall with no showy flowers, being able to find them often relies on going to areas where they have been found previously. Additionally, surveys from this study found Botrychium species in the central and northwest areas of the two parks where there were no previous occurrence records. This suggests that greater survey effort is needed to better understand Botrychium species distribution and frequency in the parks. 91 The habitat assessment shows that in CCWPP, most occurrences of Botrychium are found in grasslands. However, as a genus they do not appear to be negatively impacted by shade, as suggested by the high frequency of Botrychium occurrences in coniferous forests. The high number of occurrences in grasslands matches habitat descriptions of the two most common species (B. lanceolatum and B. lunaria) which are noted as being found in ‘open fields’ (FNA Ed. Comm., 1993). In addition, Achillea millefolium, Fragaria virginiana, and Taraxacum officinale are not only common, weedy species in open habitats such as grasslands and the among the most frequent species found near trails in CCWPP (Chapter 2), as expected, they are the three most common species associated with species of Botrychium. Others have similarly found Fragaria virginiana commonly growing alongside B. paradoxum (Zika, 1992; Vanderhorst, 1993), suggesting that these associations in open grassland sites may be linked to mycorrhizal interactions between the two plants (Vanderhorst, 1997). Although my study did not consider the variation in composition of soil biota in different vegetation types, disturbance associated with soil compaction and exotic species presence could be impacting mycorrhizae in the soil, subsequently affecting Botrychium species presence. In disturbed habitats, the presence of exotic species can cause shifts in the composition of mycorrhizal communities (Hawkes et al., 2006; Jordan et al., 2012; Sielaff et al., 2019), an important factor in growth for all members of Botrychium. If exotic species are shifting mycorrhizal communities to more non-mycorrhizal species (Hawkes et al., 2006) or have higher colonization rates of mycorrhizal fungi than native species (Sielaff et al., 2019), then it is likely that the presence of exotic species can negatively influence Botrychium presence. Interactions with non-mycorrhizal fungi have 92 been shown to reduce plant growth in contrast to the positive effects observed in the presence of mycorrhizal fungi (Klironomos, 2002). Such effects on the composition of soil biota may be more important in rare native plants such as those belonging to Botrychium, which rely on mycorrhizal associations for normal plant growth. Interestingly, in CCWPP, grasslands have the highest likelihood of exotic species occurring (Chapter 2) and they have the highest number of Botrychium occurrences. This suggests that more common Botrychium species can tolerate open, disturbed habitats where exotic species are highly likely to occur. However, the prevalence of Botrychium relative to exotic prevalence and abundance of mycorrhizal communities present in these grasslands has yet to be studied. Most Botrychium occurrences were located on south-facing slopes with a relatively flat incline, suggesting that most Botrychium occurrences are associated with lower soil moisture levels (Lieffers & Larkin-Lieffers, 1987). This may also be related to availability of mycorrhizal fungi. Botrychium sporophyte growth requires mycorrhizal associations, and a study comparing differences in north and south-facing slopes found increased arbuscular mycorrhizal species richness associated with south-facing slopes (Chai et al., 2018). I also found that most Botrychium occurrences were beyond the immediate vicinity of trails. In addition to impacts on vegetation (Chapter 2), soil disturbance associated with trails can impact mycorrhizal communities by decreasing diversity and changing community composition (Amalia et al., 1968). Although Botrychium species are often noted to be found near trails, this is likely due to bias towards occurrences being recorded only from well-travelled areas (Ingegno, 2015). Rather, it seems that in CCWPP, most Botrychium occurrences are away from trails, and 93 this may reflect their close association with mycorrhizal communities which are negatively impacted by trail presence. To obtain a better understanding of Botrychium distribution throughout CCWPP, more field surveys targeting Botrychium species should be conducted in grasslands and coniferous forests in regions away from trails and other known occurrences. Thorough field searches for single species are costly. Therefore, finding a more efficient method of locating rare species is often sought out. SDMs have been used successfully to increase the efficiency of surveys for rare plants (Boetsch et al., 2003; Bourg et al., 2005; McCune, 2016). Our generic SDM for Botrychium presence was not a good predictor of Botrychium occurrences, likely because species belonging to the same genus often have different specific habitat requirements. However, there was evidence of a trend towards higher predicted suitability at sites where species of Botrychium were found. It is possible that a larger sample of plots would have shown significant results. Unexpectedly, one of the 7 sites with Botrychium present was predicted to have less than 10% suitability. All other sites with Botrychium present ranged from 43% to 88%. A larger sample size could help determine whether the unexpected occurrence is an outlier or whether Botrychium occur more often than expected at sites with low predicted suitability, which would confirm that the SDM was not useful in this case. The failure of the SDM to predict Botrychium occurrences could be because the model was built at the generic level – not for each species. Botrychium species are known to occur in very different habitats. For example, of the species present in both grasslands and coniferous forests, species with more occurrences in grasslands (B. campestre, B. lanceolatum, B. lunaria, and B. michiganense) or coniferous forests (B. ascendens, B. 94 crenulatum, and B. minganense) had lower frequencies in the other habitat, suggesting some species occur at higher rates in certain habitats. Because of this, SDMs built using occurrences from the entire genus may not accurately predict suitable habitat for rare Botrychium species only found in a particular habitat. For example, Johnson-Groh et al. (2002a) found that a prairie species (B. gallicomontanum) must overcome more risks associated with open, exposed prairies by emerging much earlier than a forest species (B. mormo) which is less exposed to extreme changes in temperature and moisture. Future studies should use SDMs which are species-specific rather than generic to test whether some species of Botrychium occur in certain habitats over others. It is possible that the SDM could be a good indicator of suitable habitat, but survey efforts missed occurrences in some high suitability plots. Although I was able to detect occurrences in coniferous forests, later emergence rates and other barriers such as dense leaf litter (Johnson-Groh et al., 2002a) could contribute to missed occurrences of more sporadic or difficult to detect Botrychium species such as B. pallidum or B. paradoxum (FNA Ed. Comm., 1993) in these habitats. Wagner & Wagner (1981) note open meadows are particularly subject to similar looking or dense herbaceous cover that make it difficult to see Botrychium individuals. Repeating surveys at different times of the growing season would ensure true absence of Botrychium from plots. It is also possible that the SDM was not a good predictor of suitable habitat because it did not include important environmental predictors. Botrychium require mycorrhizal associations for successful establishment and as such, soil characteristics - such as soil pH and nutrient levels - that are associated with mycorrhizal communities may be necessary to accurately predict Botrychium occurrences (Lilleskov & Parrent, 95 2007). In addition to topography, climate, geology, and soil factors, the landscape context may also be playing a role in accurately predicting suitable habitat. For example, in testing SDMs for 8 rare species in Ontario, McCune (2016) found that proximity to the nearest known occurrence of a rare species also influenced the likelihood of finding a new occurrence, in addition to predicted habitat suitability. One other study has used SDMs to predict habitat for Botrychium (Ingengo, 2015). In this study, elevation, aspect, slope, soils, geology, mean May precipitation, mean June temperature, and land cover were the predictor variables, however, the predictions of this SDM were not tested with independent surveys (Ingengo, 2015). SDMs predicting suitable habitat for Botrychium in Alberta should be re-built, provided data are available regarding these other potential predictors. I did not find any significant differences in the plant community composition within 50 m x 50 m plots with versus without detections of Botrychium. In contrast, McCune (2016) found that regardless of the breadth of habitat for a target species, sites where the target species was found were more clustered in terms of plant community composition compared to all other suitable plots. This again could be due to lumping all Botrychium species together. The different Botrychium species we surveyed have different habitat preferences, and members of the genus occupy a diverse range of plant communities even within the same vegetation type. Although Botrychium as a genus was not found to be limited to a restricted subset of plant communities, more records of individual species may produce results that support specific plant community associations for individual species. 96 Although community composition did not differ significantly between sites with and without Botrychium present, there were significant indicator species of Botrychium presence or absence. The indicator species of Botrychium absent sites were Lonicera utahensis S.Watson (Rocky Mountain honeysuckle, S3), Maianthemum racemosum (L.) Link (Solomon’s plume, S5), and Acer glabrum Torr. (Rocky Mountain maple, S3) which are characteristic species of shaded, often shrubby habitats of mixed or broadleaf forest understories (FNA Ed. Comm., 1993). This aligns with the assessment of the vegetation types of the 85 georeferenced occurrences of Botrychium: a total of only 4% of those records occurred in mixed and broadleaf forests, the fewest of any vegetation type. Agoseris glauca (Pursh) Raf. (Pale agoseris, S5) was significantly more frequent and abundant in the seven plots with Botrychium present. Agoseris glauca is a perennial species that produces a rosette of pubescent basal, pale green leaves from a thick taproot and a single showy, yellow flower at the tips of flowering stems (FNA Ed. Comm., 1993) making it much more conspicuous than Botrychium. Interestingly, Agoseris glauca was not noted as an associated species for any of the 81 photo records, or for any documented occurrences of Botrychium deposited at the University of Lethbridge (LEA) herbarium. This could be attributed to the fact that associated species in photographs were limited to the direct vicinity of the specimen, whereas the ISA considered a much larger 50 m x 50 m area. Agoseris glauca and Botrychium lunaria have been described together in plant communities in Alberta’s northern Rocky Mountains (Russel & Roi, 1986), and they both occur in similar vegetation types—including moist to dry habitats, alpine meadows, montane forests, as well as gravelly or rocky soils—and elevation ranges (FNA Ed. Comm., 1993). Although both species are found in a wide variety of habitats, the more 97 conspicuous flowers of Agoseris glauca could be useful in pinpointing sites where smaller Botrychium plants are also found. The habitats of Botrychium in CCWPP range from disturbed sites such as rocky talus or sites near roads and trails to coniferous forests, with grasslands being the most frequent vegetation type. Botrychium species are known to occur in very diverse habitats (Wagner & Wagner, 1981; FNA Ed. Comm., 1993; Williston, 2002). Of all the species identified from the 85 CCWPP records, Botrychium lanceolatum was the most common species, and it also occurred most frequently in grasslands. Although I expected to see more Botrychium directly beside trails compared to areas distant from trails, most occurrences (22) of moonwort recorded in CCWPP are found greater than 10 m but less than 100 m from trails, and there were 21 occurrences more than 1,000 m away from the nearest trail, which does not provide evidence that Botrychium species prefer disturbed habitats directly beside trails (less than 10 m) compared to off-trail habitats. The tendency for Botrychium to occur at least 10 m from trails follows what Amalia et al. (1968) found regarding negative trail impacts on mycorrhizal communities. Especially given the fact that Botrychium require mycorrhizal interactions to complete their life cycle (Rayner, 1927; Whittier, 1973; Winther & Friedman, 2007), abiotic or biotic factors affecting the mycorrhizae microbiome could influence the success of the host Botrychium species (Sandoz et al., 2020). Although my assessment of the georeferenced photos indicates that Botrychium was not very frequent beside recreational trails, they nonetheless were found near other linear disturbances such as deer trails and scree slopes. To maximize efficiency, surveys to find new occurrences of Botrychium should target grassland and 98 coniferous forest communities away from trails, particularly in the Middle and West Castle watersheds – where no occurrences have been recorded prior to this project. 3.6 Conclusion Altogether, these results indicate that Botrychium is relatively frequent throughout CCWPP. Although Botrychium species are often associated with trails, they also occur frequently away from trails. Surveys for additional Botrychium occurrences should focus on grassland and coniferous forests away from trails at elevations around 1,800 m a.s.l, along more southerly slopes, and avoid shaded mixed or broadleaf forest sites. Areas with Agoseris glauca may also be a good indicator of Botrychium habitat. Additionally, more extensive surveys should be conducted in areas where my study has uncovered new occurrences not previously recorded, including the Middle Castle River and West Castle River watersheds. Areas which have been under-surveyed that could also prove insightful in terms of Botrychium occurrences include the Mill Creek and Upper (South) Castle River watersheds, where no occurrences have been recorded in the parks. Although our SDM was not a great predictor of Botrychium at the genus level, new occurrences from this project provide better representation of Botrychium species distribution throughout CCWPP. These data could be used to build species-level SDMs to predict habitat suitability for individual species, allowing targeted searches for species that are thought to be rare. 99 CHAPTER 4: CONCLUSION As more people visit protected areas for recreation, understanding the impacts that trail users have on the plant communities is vital for maintaining sustainable use of recreational trails. In parks that have been recently established such as CCWPP, assessing trail impacts within different vegetation types and along different trail types could help mitigate unwanted changes to the native plant communities, especially at higher elevations. My study shows that the effects of recreational trails have already impacted grasslands in this region and shrublands and mixed/broadleaf vegetation are showing trends that may lead to similarly greater impacts over time. Although coniferous forests do not seem to be impacted by exotic species, this vegetation type could see similar trends in the future as more people visit the parks. Additionally, relative to footpaths, OHV trails were found to impact not only the number of species present, but the likelihood of exotic species presence, an effect that was also found at higher elevations. Barros et al. (2020) indicated exotic species favoured by off-trail disturbances can negatively impact alpine vegetation, calling for limited off-trail use at higher elevations. My results have important implications for CCWPP. In particular, the number of OHV trails (and roads) should be reduced, particularly through shrubland and mixed/broadleaf vegetation. Additionally, although it may not be feasible to close OHV trails at higher elevations, my results suggest that implementing tactics to reduce the spread of exotic species and limiting the amount of OHV traffic that reaches higher elevations is key to sustainable trail use within montane parks. CCWPP is a hotspot for rare plants in Alberta, especially Botrychium. Prior to the recent floral inventory project and my present research, the distribution and habitat 100 preferences of Botrychium in CCWPP were poorly understood. My study shows that Botrychium is not limited to trailside vegetation as was previously assumed. Botrychim occur in many areas nowhere near trails, in various vegetation types—grasslands and coniferous forests in particular—and across a wide elevational range. Surveys should be conducted in areas where I found new occurrences, as well as areas which are currently still under-surveyed. The small stature and cryptic presence of these species make searches difficult and time consuming. More common species such as Agoseris glauca may be a good indicator of Botrychium habitat which could be used to improve the efficiency of field searches for these tiny ferns. Additionally, SDMs built for individual Botrychium species may prove useful in targeting specific species distributions throughout the parks. As suggested by my results of the likelihood of rare species presence (Chapter 2), park management should limit the amount of disturbance associated with trails to promote the presence of rare species such as Botrychium. 4.1 Limitations and future directions My study has some limitations. In Chapter 2, I categorized trail types based on trail widths alone and used soil compaction as a proxy for trailside disturbance. As a result of the new Trails Act recently put in place by the government of Alberta (Government of Alberta, 2022), updated trail use maps have been constructed for the Castle region, which may not reflect the trail types I categorized as ‘OHV trail’. Although the measures I used are likely accurate indicators of trail use intensity over the past few decades, future studies should assess these trail types accordingly to determine their impact on plant communities and provide further insight into the sustainability of current trail use in CCWPP. 101 Measures of trail use intensity were also not available for my study. Although the government of Alberta has attempted to capture visitor rates of most provincial parks, data are either unavailable or incomplete for some parks (Alberta Tourism, 2006) including Castle Provincial and Castle Wildland Provincial Parks. Future studies could additionally use camera traps or trail counters to measure actual trail use intensity of different trails. Some studies have used such methods to monitor and quantify wildlife and human-based activities along trails (Kays et al., 2009; Miller et al., 2017; Abildso et al., 2021). Similar approaches could be used in CCWPP which already utilizes camera traps for wildlife monitoring. In Chapter 3, the sample size of off-trail Botrychium surveys was limited to 24 plots. Provided more time and resources, surveying more plots could provide a more precise indication of the utility of the SDM used. In addition, the SDM used in this study was built at the generic level, which may not reflect specific habitat preferences for all Botrychium species. Additional occurrence records from my study could be used to develop species-level analyses to improve our understanding of specific Botrychium species and their distributions throughout CCWPP. 4.2 Concluding Statement Dozens of studies have investigated the impacts of recreations trails on plant communities throughout the world. However, very few have been carried out in Canada’s Rocky Mountain Parks, where the number of visitors and recreational trail use is continuing to increase. I showed that the effect of trails on plant communities varies with different vegetation types and different trail types. I also showed that the presence of some rare species like Botrychium are more prevalent than previously recorded. More 102 studies of recreational trail impacts on plant communities and more field surveys of species whose distribution is poorly understood are needed, especially in southern Alberta’s Rocky Mountains. 103 REFERENCES Abildso, C. G., V. Haas, S. M. Daily, and T. K. Bias. 2021. Field Test of a Passive Infrared Camera for Measuring Trail-Based Physical Activity. Frontiers in Public Health 9:584740. Ahern, R. G., D. A. Landis, A. A. Reznicek, and D. W. Schemske. 2010. Spread of exotic plants in the landscape: the role of time, growth habit, and history of invasiveness. Biological Invasions 12:3157-3169. Alberta Biodiversity Monitoring Institute (ABMI). [website]. Climate Data for Alberta. Accessed: August 31, 2022. https://www.abmi.ca/home/data-analytics/da-top/da-product- overview/Other-Geospatial-Land-Surface-Data/Climate-Variables.html Alberta Biodiversity Monitoring Institute (ABMI). 2014. Alberta W2W Normalized Difference Vegetation Index (NDVI). Alberta Biodiversity Monitoring Institute, Edmonton, Alberta. Alberta Environment and Parks. 2018. Castle Management Plan: Castle Provincial Park and Castle Wildland Provincial Park. Alberta Environment and Parks, Edmonton. Alberta Parks. [website]. Park Research & Management 2020. Accessed: March 28, 2022. https://www.albertaparks.ca/parks/south/castle-pp/park-research-management/ Alberta Wilderness Association. [website]. Castle. Accessed: March 28, 2022. https://albertawilderness.ca/issues/wildlands/areas-of- concern/castle/#parentHorizontalTab4 Amalia, A., N. Dewi, A. Heriyanti, F. Daeni, and R. Atunnisa. 2021. A review on arbuscular mycorrhizal fungal communities in response to disturbance. Page 012001 in Journal of Physics: Conference Series. IOP Publishing. Averett, J. P., B. McCune, C. G. Parks, B. J. Naylor, T. DelCurto, and R. Mata-Gonzalez. 2016. Non-native plant invasion along elevation and canopy closure gradients in a middle Rocky Mountain ecosystem. PloS one 11:e0147826. Avon, C., L. Bergès, Y. Dumas, and J.-L. Dupouey. 2010. Does the effect of forest roads extend a few meters or more into the adjacent forest? A study on understory plant diversity in managed oak stands. Forest Ecology and Management 259:1546-1555. Ballantyne, M., and C. M. Pickering. 2015. The impacts of trail infrastructure on vegetation and soils: Current literature and future directions. J Environ Manage 164:53- 64. Barros, A., and C. M. Pickering. 2014. Non-native plant invasion in relation to tourism use of Aconcagua Park, Argentina, the highest protected area in the Southern Hemisphere. Mountain Research and Development 34:13-26. 104 Barros, A., V. Aschero, A. Mazzolari, L. A. Cavieres, and C. M. Pickering. 2020. Going off trails: How dispersed visitor use affects alpine vegetation. J Environ Manage 267:110546. Bates, D., M. Mächler, B. Bolker, and S. Walker. 2015. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software 67:1 - 48. Bates, G. H. 1935. The Vegetation of Footpaths, Sidewalks, Cart-Tracks and Gateways. Journal of Ecology 23:470-487. Becker, T., H. Dietz, R. Billeter, H. Buschmann, and P. J. Edwards. 2005. Altitudinal distribution of alien plant species in the Swiss Alps. Perspectives in Plant Ecology, Evolution and Systematics 7:173-183. Bellefontaine, M. August 4, 2019. Government puts hold on Castle Parks trail closure plan. Retrieved from: https://www.cbc.ca/news/canada/edmonton/government-puts-hold- on-castle-parks-trail-closure-plan-1.5234811 Benn, B., and S. Herrero. 2002. Grizzly bear mortality and human access in Banff and Yoho National Parks, 1971-98. Ursus:213-221. Benninger-Truax, M., J. L. Vankat, and R. L. Schaefer. 1992. Trail corridors as habitat and conduits for movement of plant species in Rocky Mountain National Park, Colorado, USA. Landscape ecology 6:269-278. Bjørnstad, O. N., and W. Falck. 2001. Nonparametric spatial covariance functions: estimation and testing. Environmental and Ecological Statistics 8:53-70. Boetsch, J. R., F. K. Van Manen, and J. D. Clark. 2003. Predicting Rare Plant Occurrence in Great Smoky Mountains National Park, USA. Natural Areas Journal 23:229-237. Bourg, N. A., W. J. McShea, and D. E. Gill. 2005. Putting a CART before the search: successful habitat prediction for a rare forest herb. Ecology 86:2793-2804. Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27:326-349. Breheny, P., and W. Burchett. 2017. Visualization of regression models using visreg. R J. 9:56. Brouillet, L., F. Coursol, S.J. Meades, M. Favreau, M. Anions, P. Bélisle & P. Desmet. 2010+. VASCAN, the Database of Vascular Plants of Canada. http://data.canadensys.net/vascan/(consulted on 2022-03-15) Castilla, G., J. Hird, R. J. Hall, J. Schieck, and G. J. McDermid. 2014. Completion and updating of a landsat-based land cover polygon layer for Alberta, Canada. Canadian Journal of Remote Sensing 40:92-109. Castle- Crown Wilderness Coalition. [website]. History of the Castle. Accessed: March 28, 2022. http://ccwc.ab.ca/the-castle/history/ 105 Castro, S. A., J. A. Figueroa, M. Muñoz‐Schick, and F. M. Jaksic. 2005. Minimum residence time, biogeographical origin, and life cycle as determinants of the geographical extent of naturalized plants in continental Chile. Diversity and Distributions 11:183-191. Catling, P. M., and B. Kostiuk. 2011. Some Wild Canadian Orchids Benefit from Woodland Hiking Trails – and the Implications. Chai, Y., S. Jiang, W. Guo, M. Qin, J. Pan, A. Bahadur, G. Shi, J. Luo, Z. Jin, Y. Liu, Q. Zhang, L. An, and H. Feng. 2018. The effect of slope aspect on the phylogenetic structure of arbuscular mycorrhizal fungal communities in an alpine ecosystem. Soil Biology and Biochemistry 126:103-113. Cole, D. N. 1978. Estimating the Susceptibility of Wildland Vegetation to Trailside Alteration. The Journal of applied ecology 15:281-286. Connell, J. H. 1978. Diversity in Tropical Rain Forests and Coral Reefs. Science (American Association for the Advancement of Science) 199:1302-1310. Crisfield, V. E., S. E. Macdonald, and A. J. Gould. 2012. Effects of Recreational Traffic on Alpine Plant Communities in the Northern Canadian Rockies. Arctic, Antarctic, and Alpine Research 44:277-287. Ciuti, S., J. M. Northrup, T. B. Muhly, S. Simi, M. Musiani, J. A. Pitt, and M. S. Boyce. 2012. Effects of humans on behaviour of wildlife exceed those of natural predators in a landscape of fear. PloS one 7:e50611. Dauphin, B., D. R.Farrar, A. Maccagni, and J. R. Grant. 2017. A Worldwide Molecular Phylogeny Provides New Insight on Cryptic Diversity Within the Moonworts (Botrychium s. s., Ophioglossaceae). Systematic Botany 42:620-639. Dauphin, B., J. Vieu, and J. R. Grant. 2014. Molecular phylogenetics supports widespread cryptic species in moonworts (Botrychium ss, Ophioglossaceae). American Journal of Botany 101:128-140. Davis, A. K. 2007. Walking trails in a nature preserve alter terrestrial salamander distributions. Natural Areas Journal 27:385-389. Debarbieux, B., M. Varacca, and G. Rudaz. 2014. Challenges and opportunities for tourism development in mountain regions. Tourism in Mountain Regions. Department of Geography and Environment. University of Geneva Search in. Dickens, S. J. M., F. Gerhardt, and S. K. Collinge. 2005. Recreational Portage Trails as Corridors Facilitating Non‐Native Plant Invasions of the Boundary Waters Canoe Area Wilderness (USA). Conservation Biology 19:1653-1657. Doherty, J. 2012. American conservation and the expansion of Waterton Lakes Park 1910-1914. Alberta History 60:17+. 106 Dufrêne, M., and P. Legendre. 1997. Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecological monographs 67:345-366. Elith, J., S. J. Phillips, T. Hastie, M. Dudík, Y. E. Chee, and C. J. Yates. 2011. A statistical explanation of MaxEnt for ecologists. Diversity and Distributions 17:43-57. Faber-Langendoen, D., J. Nichols, L. Master, K. Snow, A. Tomaino, R. Bittman, G. Hammerson, B. Heidel, L. Ramsay, and A. Teucher. 2012. NatureServe conservation status assessments: methodology for assigning ranks. NatureServe, Arlington, VA. Farr, D., A. Braid, A. Janz, B. Sarchuk, S. Slater, A. Sztaba, D. Barrett, G. Stenhouse, A. Morehouse, and M. Wheatley. 2017. Ecological Response to Human Activities in Southwestern Alberta: Scientific Assessment and Synthesis. Government of Alberta. Farrar, D. 2011. Systematics and taxonomy of genus Botrychium. Website [accessed 15 December 2011] http://www. public. iastate. edu/~ herbarium/botrychium/Moonwort- Systematics. pdf. Flora of North America Editorial Committee. 1993. Botrychium. In: Flora of North America North of Mexico [Online]. 2: Pteridophytes and Gymnosperms. Oxford University Press on Demand. Botrychium in Flora of North America @ efloras.org (consulted on 2022-04-13). Foster, J. 1998. Working for Wildlife: The Beginning of Preservation in Canada. University of Toronto Press. Fryer, G. L., Jane; Ottenbreit, Kimberly; Metke, Christina; Cherniawsky, Donna; Griffiths, Amy; Foreman, Kristen; Mischkolz, Jenalee. 2022. Rare vascular plants of Alberta, 2nd Edition. 2 edition. The Alberta Native Plant Council, Edmonton, Alberta. GBIF. 2022. GBIF Occurrence Download- Poa pratensis L. GBIF.org, https://doi.org/10.15468/dl.acg54n. GBIF. 2022. GBIF Occurrence Download- Phleum pratense L. GBIF.org, https://doi.org/10.15468/dl.p343t6. GBIF. 2022. GBIF Occurrence Download- Taraxacum officinale Weber ex Wiggins. GBIF.org, https://doi.org/10.15468/dl.7mgygy. Gelman, A., Y.-S. Su, M. Yajima, J. Hill, M. G. Pittau, J. Kerman, T. Zheng, V. Dorie, and M. Y.-S. Su. 2013. Package ‘arm’. Data Analysis Using Regression and Multilevel/Hierarchical Models. Gibson, D. J., E. D. Adams, J. S. Ely, D. J. Gustafson, D. McEwen, and T. R. Evans. 2000. Eighteen years of herbaceous layer recovery of a recreation area in a mesic forest. Journal of the Torrey Botanical Society:230-239. Gilbert, B., and M. J. Lechowicz. 2005. Invasibility and abiotic gradients: the positive correlation between native and exotic plant diversity. Ecology 86:1848-1855. 107 Gilman, A. V., and V. Marshfield. 2003. Botrychium lunaria (L.) Sw. Moonwort. Conservation and Research Plan for New England. Approved, Regional Advisory Council:19. Godefroid, S., and N. Koedam. 2004. The impact of forest paths upon adjacent vegetation: effects of the path surfacing material on the species composition and soil compaction. Biological conservation 119:405-419. Gogol-Prokurat, M. 2011. Predicting habitat suitability for rare plants at local spatial scales using a species distribution model. Ecological Applications 21:33-47. Gower, S. T. 2008. Are horses responsible for introducing non-native plants along forest trails in the eastern United States? Forest Ecology and Management 256:997-1003. Guisan, A., and N. E. Zimmermann. 2000. Predictive habitat distribution models in ecology. Ecological modelling 135:147-186. Harris, C. J., B. R. Murray, G. C. Hose, and M. A. Hamilton. 2007. Introduction history and invasion success in exotic vines introduced to Australia. Diversity and Distributions 13:467-475. Hartig, F., and M. F. Hartig. 2017. Package ‘DHARMa’. R package. Hauk, W. D. 1995. A molecular assessment of relationships among cryptic species of Botrychium subgenus Botrychium (Ophioglossaceae). American Fern Journal:375-394. Hauk, W. D., C. R. Parks, and M. W. Chase. 2003. Phylogenetic studies of Ophioglossaceae: evidence from rbcL and trnL-F plastid DNA sequences and morphology. Molecular phylogenetics and evolution 28:131-151. Hauk, W. D., L. Kennedy, and H. M. Hawke. 2012. A phylogenetic investigation of Botrychium ss (Ophioglossaceae): evidence from three plastid DNA sequence datasets. Systematic Botany 37:320-330. Hawkes, C. V., J. Belnap, C. D’Antonio, and M. K. Firestone. 2006. Arbuscular Mycorrhizal Assemblages in Native Plant Roots Change in the Presence of Invasive Exotic Grasses. Plant and Soil 281:369-380. Hochrein, H. 2008. Corridors and plant invasions: a comparative study of the role of roadsides and hiking trails on plant invasions in Moorea, French Polynesia. Hurd, E. G., N. L. Shaw, J. Mastrogiuseppe, L. C. Smithman, and S. Goodrich. 1998. Field guide to Intermountain sedges. Gen. Tech. Rep. RMRS-GTR-10. Ogden, UT: US Department of Agriculture, Forest Service, Rocky Mountain Research Station. 282 p. 10. Ingegno, A. 2015. Acknowledging Sampling Bias in Species Distribution Modelling: Predicting Botrychium sp. Habitat in Lincoln County, Montana. Johnson‐Groh, C. L., and J. M. Lee. 2002a. Phenology and demography of two species of Botrychium (Ophioglossaceae). American Journal of Botany 89:1624-1633. 108 Johnson-Groh, C., C. Riedel, L. Schoessler, and K. Skogen. 2002b. Belowground distribution and abundance of Botrychium gametophytes and juvenile sporophytes. American Fern Journal 92:80-92. Kasworm, W. F., and T. L. Manley. 1990. Road and trail influences on grizzly bears and black bears in northwest Montana. Bears: Their Biology and Management:79-84. Kays, R., B. Kranstauber, P. Jansen, C. Carbone, M. Rowcliffe, T. Fountain, and S. Tilak. 2009. Camera traps as sensor networks for monitoring animal communities. Pages 811- 818 in 2009 IEEE 34th Conference on Local Computer Networks. IEEE. Kelly, C. L., C. M. Pickering, and R. C. Buckley. 2003. Impacts of tourism on threatened plant taxa and communities in Australia. Ecological management & restoration 4:37-44. Kershaw, L. J., and L. Allen. 2020. Vascular flora of Alberta: an illustrated guide. Linda Kershaw and Lorna Allen. Kershaw, R. 2008. Exploring the Castle: discovering the backbone of the world in southern Alberta. Rocky Mountain Books, Surrey, B.C. Kerbiriou, C., I. Leviol, F. Jiguet, and R. Julliard. 2008. The impact of human frequentation on coastal vegetation in a biosphere reserve. Journal of environmental management 88:715-728. Klironomos, J. N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature 417:67-70. Kuijt, J. 1982. A flora of Waterton Lakes National Park. University of Alberta Press, Edmonton. Lake, J. C., and M. R. Leishman. 2004. Invasion success of exotic plants in natural ecosystems: the role of disturbance, plant attributes and freedom from herbivores. Biological conservation 117:215-226. Lee, P., and M. Hanneman. 2011. Castle Area Forest Land Use Zone (Alberta)–Linear disturbances, access densities, and grizzly bear habitat security areas. Edmonton, Alberta: Global Forest Watch Canada 1st Publication for International Year of Forests:5-6. Lenth, R., H. Singmann, J. Love, P. Buerkner, and M. Herve. 2019. Estimated marginal means, aka least-squares means. R package version 1.3. 2. Lesica, P., and K. Ahlenslager. 1996. Demography and life history of three sympatric species of Botrychium subg. Botrychium in Waterton Lakes National Park, Alberta. Canadian Journal of Botany 74:538-543. Liedtke, R., A. Barros, F. Essl, J. J. Lembrechts, R. E. M. Wedegärtner, A. Pauchard, and S. Dullinger. 2020. Hiking trails as conduits for the spread of non-native species in mountain areas. Biological Invasions 22:1121-1134. 109 Lieffers, V. J., and P. Larkin-Lieffers. 1987. Slope, aspect, and slope position as factors controlling grassland communities in the coulees of the Oldman River, Alberta. Canadian Journal of Botany 65:1371-1378. Lilleskov, E. A., and J. L. Parrent. 2007. Can We Develop General Predictive Models of Mycorrhizal Fungal Community-Environment Relationships? The New Phytologist 174:250-256. Lloren, J. I. 2021. Quantifying plant community change at Waterton Lakes National Park over the past 25 years. University of Lethbridge (Canada). Lomolino, M. V. 2001. Elevation gradients of species‐density: historical and prospective views. Global Ecology and Biogeography 10:3-13. Magnusson, A., H. Skaug, A. Nielsen, C. Berg, K. Kristensen, M. Maechler, K. van Bentham, B. Bolker, M. Brooks, and M. M. Brooks. 2017. Package ‘glmmtmb’. R Package Version 0.2. 0. Marion, J. L., Y.-F. Leung, H. Eagleston, and K. Burroughs. 2016. A Review and Synthesis of Recreation Ecology Research Findings on Visitor Impacts to Wilderness and Protected Natural Areas. Journal of Forestry 114:352-362. McCune, B., and J. Grace. 2002. Indicator species analysis. Analysis of ecological communities:198-204. McCune, J. 2016. Species distribution models predict rare species occurrences despite significant effects of landscape context. Journal of Applied Ecology 53:1871-1879. Medvecká, J., M. Zaliberová, J. Májeková, D. Senko, and I. Jarolímek. 2018. Role of infrastructure and other environmental factors affecting the distribution of alien plants in the Tatra Mts. Folia Geobotanica 53:349-364. Miller, A. B., Y.F. Leung, and R. Kays. 2017. Coupling visitor and wildlife monitoring in protected areas using camera traps. Journal of outdoor recreation and tourism 17:44-53. Miller, S. G., R. L. Knight, and C. K. Miller. 1998. Influence of recreational trails on breeding bird communities. Ecological Applications 8:162-169. Morse, L. E., J. M. Randall, N. Benton, R. Hiebert, and S. Lu. 2004. An invasive species assessment protocol: Evaluating non-native plants for their impact on biodiversity, Version 1. Moss, E. H., and J. G. Packer. 1983. Flora of Alberta: a manual of flowering plants, conifers, ferns and fern allies found growing without cultivation in the province of Alberta, Canada. 2nd edition. University of Toronto Press, Toronto. Mount, A., and C. M. Pickering. 2009. Testing the capacity of clothing to act as a vector for non-native seed in protected areas. Journal of environmental management 91:168- 179. 110 Müllerová, J., M. Vítková, and O. Vítek. 2011. The impacts of road and walking trails upon adjacent vegetation: Effects of road building materials on species composition in a nutrient poor environment. Science of The Total Environment 409:3839-3849. Natural Regions Committee. 2006. Natural regions and subregions of Alberta. Compiled by D.J. Downing and W.W. Pettepiece. Government of Alberta, Edmonton, Alberta NatureServe. 2022. NatureServe Explorer [Microsoft Edge]. NatureServe, Arlington, Virginia. Available https://explorer.natureserve.org/. (Accessed: February 4, 2022). Naylor, L. M., M. J. Wisdom, and R. G. Anthony. 2009. Behavioral responses of North American elk to recreational activity. The Journal of Wildlife Management 73:328-338. Nepal, S. K., and P. Way. 2007. Comparison of vegetation conditions along two backcountry trails in Mount Robson Provincial Park, British Columbia (Canada). J Environ Manage 82:240-249. Ngugi, M. R., V. J. Neldner, and R. Dowling. 2014. Non-native plant species richness adjacent to a horse trail network in seven National Parks in southeast Queensland, Australia. Australasian Journal of Environmental Management 21:413-428. Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P.R., O'Hara, R.B., Simpson, G.L., Solymos, P., Stevens, M.H.M., Szoecs, E., and Wagner, H. 2020. vegan: Community Ecology Package. R package version 2.5-7. https://CRAN.R-project.org/package=vegan Parendes, L. A., and J. A. Jones. 2000. Role of light availability and dispersal in exotic plant invasion along roads and streams in the HJ Andrews Experimental Forest, Oregon. Conservation Biology 14:64-75. Parikesit, P., D. W. Larson, and U. Matthes-Sears. 1995. Impacts of trails on cliff-edge forest structure. Canadian Journal of Botany 73:943-953. Paton, D. G., S. Ciuti, M. Quinn, and M. S. Boyce. 2017. Hunting exacerbates the response to human disturbance in large herbivores while migrating through a road network. Ecosphere 8:e01841. Pauchard, A., C. Kueffer, H. Dietz, C. C. Daehler, J. Alexander, P. J. Edwards, J. R. Arévalo, L. A. Cavieres, A. Guisan, and S. Haider. 2009. Ain't no mountain high enough: plant invasions reaching new elevations. Frontiers in Ecology and the Environment 7:479-486. Peck, J. H., C. J. Peck, and D. R. Farrar. 1990. Influences of life history attributes on formation of local and distant fern populations. American Fern Journal 80:126-142. Phillips, M. L., B. R. Murray, M. R. Leishman, and R. Ingram. 2010. The naturalization to invasion transition: Are there introduction‐history correlates of invasiveness in exotic plants of Australia? Austral ecology 35:695-703. 111 Phillips, S. J., R. P. Anderson, and R. E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological modelling 190:231-259. Pickering, C., and A. Mount. 2010. Do tourists disperse weed seed? A global review of unintentional human-mediated terrestrial seed dispersal on clothing, vehicles and horses. Journal of Sustainable Tourism 18:239-256. Pickering, C., and P. Norman. 2017. Comparing impacts between formal and informal recreational trails. J Environ Manage 193:270-279. Powell, J. 1966. Distribution and outbreaks of Dendroctonus ponderosae in forests of Western Canada. Information Report AX-2, Forestry Research Laboratory, Canadian Department of Forestry, Calgary, Alberta. Proctor, M. F., B. N. McLellan, G. B. Stenhouse, G. Mowat, C. T. Lamb, and M. S. Boyce. 2020. Effects of roads and motorized human access on grizzly bear populations in British Columbia and Alberta, Canada. Ursus 2019:16-39. Rayner, M. C. 1927. Mycorrhiza (continued). The New Phytologist 26:22-45. R Core Team. 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. Roberts, D. 2019. labdsv 2.0-1: ordination and multivariate analysis for ecology. Robertson, C., T. A. Nelson, D. E. Jelinski, M. A. Wulder, and B. Boots. 2009. Spatial– temporal analysis of species range expansion: the case of the mountain pine beetle, Dendroctonus ponderosae. Journal of biogeography 36:1446-1458. Rogala, J. K., M. Hebblewhite, J. Whittington, C. A. White, J. Coleshill, and M. Musiani. 2011. Human activity differentially redistributes large mammals in the Canadian Rockies National Parks. Ecology and Society 16. Rossi, G., G. Parolo, and T. Ulian. 2009. Human trampling as a threat factor for the conservation of peripheral plant populations. Plant Biosystems 143:104-113. Sandoz, F. A., S. Bindschedler, B. Dauphin, L. Farinelli, J. R. Grant, and V. Hervé. 2020. Biotic and abiotic factors shape arbuscular mycorrhizal fungal communities associated with the roots of the widespread fern Botrychium lunaria (Ophioglossaceae). Environmental Microbiology Reports 12:342-354. Scherrer, P., and C. M. Pickering. 2006. Recovery of Alpine Herbfield on a Closed Walking Track in the Kosciuszko Alpine Zone, Australia. Arctic, Antarctic, and Alpine Research 38:239-248. Sielaff, A. C., H. W. Polley, A. Fuentes-Ramirez, K. Hofmockel, and B. J. Wilsey. 2019. Mycorrhizal colonization and its relationship with plant performance differs between exotic and native grassland plant species. Biological Invasions 21:1981-1991. 112 Smith, G. R., A. C. Burger, B. W. Marlowe, E. P. Tristano, and J. E. Rettig. 2017. Effect of recreational trail traffic level on eastern red-backed salamander (Plethodon cinereus) relative abundance. Journal of North American Herpetology:1-4. Stohlgren, T. J., D. Binkley, G. W. Chong, M. A. Kalkhan, L. D. Schell, K. A. Bull, Y. Otsuki, G. Newman, M. Bashkin, and Y. Son. 1999. Exotic Plant Species Invade Hot Spots of Native Plant Diversity. Ecological Monographs 69:25-46. Sumanapala, and Wolf. 2019. Recreational Ecology: A Review of Research and Gap Analysis. Environments 6. Taylor, B. R., and S. Raney. 2013. Correlation Between ATV Tracks and Density of a Rare Plant (Drosera filiformis) in a Nova Scotia Bog. Rhodora 115:158-169. Taylor, K., T. Brummer, M. L. Taper, A. Wing, and L. J. Rew. 2012. Human‐mediated long‐distance dispersal: an empirical evaluation of seed dispersal by vehicles. Diversity and Distributions 18:942-951. Taylor, S. W., A. L. Carroll, R. I. Alfaro, and L. Safranyik. 2006. Forest, climate and mountain pine beetle outbreak dynamics in western Canada. The mountain pine beetle: A synthesis of biology, management, and impacts on lodgepole pine:67-94. Thompson, B. 2015. Recreational trails reduce the density of ground-dwelling birds in protected areas. Environmental Management 55:1181-1190. Thurston, E., and R. J. Reader. 2001. Impacts of experimentally applied mountain biking and hiking on vegetation and soil of a deciduous forest. Environmental Management 27:397-409. Törn, A., A. Tolvanen, Y. Norokorpi, R. Tervo, and P. Siikamäki. 2009. Comparing the impacts of hiking, skiing and horse riding on trail and vegetation in different types of forest. Journal of environmental management 90:1427-1434. Trip, N. v. V., and Y. F. Wiersma. 2015. A Comparison of All-Terrain Vehicle (ATV) Trail Impacts on Boreal Habitats Across Scales. Natural Areas Journal 35:266-278, 213. Tyser, R. W., and C. A. Worley. 1992. Alien Flora in Grasslands Adjacent to Road and Trail Corridors in Glacier National Park, Montana (U.S.A.). Conservation Biology 6:253- 262. Vanderhorst, J. P. 1997. Conservation Assessment of Sensitive Moonworts (Ophioglossaceae; Botrychium Subgenus Botrychium) on the Kootenai National Forest. Montana Natural Heritage Program. Vetaas, O. R., and J. A. Grytnes. 2002. Distribution of vascular plant species richness and endemic richness along the Himalayan elevation gradient in Nepal. Global Ecology and Biogeography 11:291-301. 113 von der Lippe, M., and I. Kowarik. 2007. Long-distance dispersal of plants by vehicles as a driver of plant invasions. Conserv Biol 21:986-996. Wagner, W., and F. S. Wagner. 1981. New species of moonworts, Botrychium subg. Botrychium (Ophioglossaceae), from North America. American Fern Journal 71:20-30. Wagner, W., and F. S. Wagner. 1983. Two moonworts of the Rocky Mountains; Botrychium hesperium and a new species formerly confused with it. American Fern Journal 73:53-62. Wagner, W., and F. S. Wagner. 1994. Another widely disjunct, rare and local North American moonwort (Ophioglossaceae: Botrychium subg. Botrychium). American Fern Journal 84:5-10. Watkins, R. Z., J. Chen, J. Pickens, and K. D. Brosofske. 2003. Effects of forest roads on understory plants in a managed hardwood landscape. Conservation Biology 17:411-419. Webb, R. H., H. C. Ragland, W. H. Godwin, and O. Jenkins. 1978. Environmental effects of soil property changes with off-road vehicle use. Environmental Management 2:219- 233. Wells, F. H., Lauenroth, William K., Bradford, John B. 2013. Recreational trails as corridors for alien plants in the Rocky Mountains, USA. Western North American Naturalist 72. Whittier, D.P. 1973. The effect of light and other factors on spore germination in Botrychium dissectum. Can. J. Bot. 51: 1791-1794. Whittington, J., C. C. St. Clair, and G. Mercer. 2005. Spatial responses of wolves to roads and trails in mountain valleys. Ecological Applications 15:543-553. Williams, J. N., C. Seo, J. Thorne, J. K. Nelson, S. Erwin, J. M. O’Brien, and M. W. Schwartz. 2009. Using species distribution models to predict new occurrences for rare plants. Diversity and Distributions 15:565-576. Williston, P. 2001. The Botrychiaceae of Alberta. Smithers, BC: Mnium Ecological Research. Williston, P. 2002. The botrychiaceae of Alberta : a survey of element occurrences of the genera Botrychium and Sceptridium in Alberta. Edmonton. Willoughby, M. G., Alexander, M. J., & Adams, B. W. (2008). Range plant community types and carrying capacity for the montane subregion. Alberta Sustainable Resource Development, and Agriculture and Agri-Food Canada. Edmonton, Alberta, Canada. Winther, J. L., and W. E. Friedman. 2007. Arbuscular mycorrhizal symbionts in Botrychium (Ophioglossaceae). American Journal of Botany 94:1248-1255. Wittemyer, G., P. Elsen, W. T. Bean, A. C. O. Burton, and J. S. Brashares. 2008. Accelerated human population growth at protected area edges. Science 321:123-126. 114 Wright, S. P. 1992. Adjusted p-values for simultaneous inference. Biometrics:1005-1013. Yoo, J., and N. Koper. 2017. Effects of shallow natural gas well structures and associated roads on grassland songbird reproductive success in Alberta, Canada. PloS one 12:e0174243. Zhou, T., X. Luo, Y. Hou, Y. Xiang, and S. Peng. 2020. Quantifying the effects of road width on roadside vegetation and soil conditions in forests. Landscape ecology 35:69-81. Zika, P. 1992. Draft management guide for rare Botrychium species (moonworts and grape ferns) for the Mount Hood National Forest. Unpublished report. Oregon Natural Heritage Program, Portland, OR. 115 APPENDIX 1: Chapter 2 Supplementary Materials Note: All my data has been archived in the Federated Research Data Repository, DOI: 10.20383/103.0665 Table A1.1. Final locations of trailside and off-trail transects surveyed in June to August of 2020 and 2021. Columns include the specific watershed each site was in, the date the transect was surveyed (D-M-Y), latitude and longitude of the transect site, elevation (meters above sea level), aspect (degrees), slope steepness (+ indicates East declination; - indicates West declination), soil compaction taken at the middle of the trail (kg/cm2), and the type of vegetation (Veg.) determined at each site (S = shrubland; G = grassland; B = broadleaf; C = coniferous; M = mixed). Site Watershed Date Lat. Long. Elev. Aspect Slope Comp. (m) (°) (°) (kg/cm2) Veg. t1 Carbondale 5-8-21 49.42 -114.55 1819 105 + 20 5 S t2 Carbondale 5-8-21 49.42 -114.55 1799 75 - 2 5 G t4 Carbondale 28-7-20 49.49 -114.41 1900 48 - 32 4 G t5 Carbondale 28-7-20 49.49 -114.42 1828 132 - 45 1.5 B t9 Carbondale 18-8-20 49.40 -114.38 1619 310 - 8 3.5 C t14 Carbondale 29-7-20 49.42 -114.43 1457 237 - 28 2.25 C t15 Carbondale 25-6-20 49.44 -114.39 1484 260 + 8 0.75 C t16 Carbondale 25-6-20 49.43 -114.38 1575 60 - 2 2.5 C t20 Carbondale 25-6-20 49.44 -114.39 1404 110 - 2 1.4 G t21 Carbondale 25-6-20 49.44 -114.41 1441 240 + 2 2.75 B t22 Carbondale 8-7-21 49.44 -114.5 1503 348 + 14 5 B t24 Carbondale 30-7-20 49.50 -114.5 1510 204 + 4 5 C t25 Carbondale 30-7-20 49.50 -114.51 1531 91 - 6 5 M t26 Carbondale 30-7-20 49.41 -114.49 1465 11 - 12 5 M t27 Carbondale 30-7-20 49.42 -114.48 1456 349 + 10 4.5 M t28 Carbondale 2-7-20 49.42 -114.45 1451 120 - 6 0.75 M t30 Carbondale 29-7-20 49.43 -114.45 1421 117 - 10 5 M t33 Carbondale 29-7-20 49.43 -114.44 1432 333 - 18 5 M t36 Carbondale 24-6-20 49.48 -114.45 1694 45 + 11 4.5 M t38 Carbondale 24-6-20 49.49 -114.46 1812 248 - 50 5 S t47 Drywood 21-7-20 49.23 -114.11 1907 122 - 40 5 C Creek t48 Drywood -114.08 1697 95 - 4 1.5 G Creek 21-7-20 49.24 t49 Drywood Creek 22-7-20 49.26 -114.05 1585 154 - 24 5 M t50 Drywood 22-7-20 49.26 -114.04 1539 357 - 10 5 G Creek t51 Drywood -114.1 1688 128 - 28 5 G Creek 25-8-20 49.27 116 t54 Drywood 25-8-20 49.27 -114.08 1612 110 - 32 5 M Creek t55 Drywood Creek 26-8-20 49.27 -114.08 1603 79 - 42 5 M t56 Drywood -114.08 1586 29 - 2 5 M Creek 26-8-20 49.28 t57 Drywood 29-6-21 49.28 -114.07 1510 234 - 8 5 G Creek t59 Drywood 23-7-20 49.27 -114.02 1509 10 - 12 5 B Creek t60 Drywood 23-7-20 49.27 -114.02 1510 344 - 12 5 B Creek t62 Drywood 23-7-20 49.27 -114.02 1506 33 - 20 5 B Creek t63 Drywood -114.1 1683 186 - 30 5 M Creek 25-8-20 49.27 t65 Drywood 22-7-20 49.26 -114.04 1540 74 - 16 5 G Creek t66 Drywood 22-7-20 49.27 -114.03 1523 56 - 22 5 G Creek t67 Drywood 22-7-20 49.27 -114.03 1515 62 - 10 5 S Creek t68 Drywood 25-8-20 49.27 -114.11 1707 200 - 36 5 S Creek t70 Drywood Creek 26-8-20 49.29 -114.07 1547 15 - 18 5 B t71 Middle Castle 3-7-20 49.36 -114.26 2193 140 - 10 5 G t72 Middle Castle 7-8-20 49.39 -114.34 1349 214 - 2 5 G t73 Middle -114.34 1350 240 - 4 5 S Castle 7-8-20 49.39 t74 Middle -114.26 1657 325 - 26 5 C Castle 6-8-20 49.37 t75 Middle 18-8-20 49.39 -114.38 1592 192 - 17 5 S Castle t76 Middle 18-8-20 49.39 -114.38 1608 154 - 23 5 C Castle t78 Middle 5-8-20 49.40 -114.31 1541 18 - 24 3.25 C Castle t79 Middle -114.38 1633 182 - 10 5 C Castle 18-8-20 49.40 t80 Middle Castle 20-8-20 49.41 -114.36 1558 176 - 6 5 C t81 Middle 19-8-20 49.41 -114.36 1530 48 - 20 5 C Castle t82 Middle 5-8-20 49.39 -114.29 1565 80 - 15 4.5 B Castle t83 Middle Castle 20-8-20 49.41 -114.36 1473 180 - 36 5 M t84 Middle Castle 2-7-20 49.41 -114.34 1347 120 - 14 0.8 B t85 Middle 2-7-20 49.43 -114.33 1338 130 - 20 1.5 B Castle 117 t87 Middle 6-8-20 49.38 -114.27 1479 121 - 16 5 M Castle t88 Middle 19-8-20 49.39 -114.37 1494 176 + 16 5 M Castle t89 Middle 5-8-20 49.40 -114.33 1378 280 - 22 5 M Castle t90 Middle Castle 20-8-20 49.40 -114.36 1496 164 - 34 5 M t91 Middle 25-6-20 49.40 -114.35 1368 130 - 12 3 M Castle t92 Middle Castle 2-7-20 49.43 -114.33 1344 163 - 8 1.75 C t93 Middle -114.33 1375 130 - 2 0 M Castle 2-7-20 49.44 t94 Middle Castle 6-8-20 49.38 -114.27 1538 50 - 11 5 S t95 Middle Castle 6-8-20 49.38 -114.28 1471 352 - 10 5 S t96 Middle Castle 7-8-20 49.38 -114.28 1483 258 - 2 5 M t99 Mill Creek 16-7-20 49.36 -114.18 1530 296 - 10 5 G t100 Mill Creek 15-7-20 49.36 -114.18 1491 270 - 8 2.5 G t101 Mill Creek 16-7-20 49.36 -114.19 1438 40 + 2 1.75 M t103 Mill Creek 16-7-20 49.36 -114.19 1462 172 - 14 3.25 S t104 Mill Creek 16-7-20 49.36 -114.2 1476 90 - 10 5 C t105 Mill Creek 15-7-20 49.32 -114.19 1511 338 - 12 5 C t106 Mill Creek 15-7-20 49.32 -114.19 1513 112 - 2 5 C t107 Mill Creek 14-7-20 49.34 -114.2 1505 17 - 20 1.8 C t108 Mill Creek 14-7-20 49.35 -114.2 1470 69 - 10 4 C t111 Mill Creek 6-8-20 49.37 -114.25 1687 58 - 12 0 G t112 Mill Creek 22-6-21 49.36 -114.23 1518 70 - 4 5 M t114 Mill Creek 22-6-21 49.36 -114.23 1514 125 + 3 3 M t115 Mill Creek 14-7-20 49.35 -114.2 1469 120 - 10 5 S t116 Mill Creek 14-7-20 49.35 -114.2 1477 275 - 12 2.5 C t118 Mill Creek 16-7-20 49.36 -114.17 1544 128 - 2 5 G t125 Upper Castle 21-6-21 49.36 -114.28 2137 169 + 8 5 G t125x Upper Castle 3-7-20 49.36 -114.29 1618 327 - 8 5 C t126 Upper Castle 13-7-21 49.23 -114.23 1544 275 + 16 5 C t128 Upper Castle 4-8-21 49.32 -114.33 1427 342 + 3 2 M t131 Upper Castle 26-6-20 49.37 -114.35 1401 240 - 27 1.3 S t134 Upper Castle 13-7-21 49.29 -114.28 1461 242 - 8 4 B t137 Upper Castle 13-7-21 49.26 -114.25 1518 280 - 4 5 M t139 Upper Castle 4-8-21 49.28 -114.33 1719 293 + 11 5 C t140 Upper Castle 4-8-21 49.30 -114.33 1586 242 + 26 5 M t142 Upper Castle 7-8-20 49.38 -114.29 1463 267 - 3 5 S 118 t145 Upper Castle 4-8-21 49.29 -114.33 1661 273 - 19 3.25 S t147 Upper Castle 3-7-20 49.36 -114.28 1764 125 - 52 5 C t149 Upper Castle 3-7-20 49.36 -114.31 1436 341 - 2 1.5 S t150 Upper Crowsnest 12-8-20 49.57 -114.63 1657 25 - 21 5 C t152 Upper -114.63 1664 320 - 2 5 C Crowsnest 12-8-20 49.58 t155 Upper Crowsnest 12-8-20 49.58 -114.64 1617 266 + 12 5 C t156 Upper 12-8-20 49.59 -114.66 1498 244 + 1 5 M Crowsnest t157 Upper Crowsnest 13-8-20 49.57 -114.57 1840 133 - 48 5 C t158 Upper -114.57 1847 150 - 51 5 S Crowsnest 13-8-20 49.57 t159 Upper Crowsnest 13-8-20 49.57 -114.57 1862 106 - 32 3 C t160 Upper Crowsnest 12-8-20 49.59 -114.65 1523 248 + 14 5 M t161 Upper -114.65 1541 230 + 6 4 M Crowsnest 12-8-20 49.59 t162 Upper Crowsnest 13-8-20 49.57 -114.57 1910 20 - 1 5 C t163 Upper -114.57 1914 141 - 26 5 S Crowsnest 13-8-20 49.57 t166 West Castle 9-7-20 49.37 -114.43 1794 163 - 12 5 G t167 West Castle 26-6-20 49.38 -114.35 1406 158 - 12 1.6 G t169 West Castle 8-7-20 49.29 -114.4 1521 334 - 24 4.25 C t170 West Castle 3-6-21 49.29 -114.4 1460 248 - 17 5 C t171 West Castle 9-7-20 49.37 -114.43 1775 107 - 20 5 C t172 West Castle 15-6-20 49.38 -114.36 1380 340 - 10 1.25 C t174 West Castle 19-8-20 49.38 -114.39 1544 112 + 3 5 S t175 West Castle 8-7-20 49.28 -114.4 1639 87 - 28 5 B t176 West Castle 8-7-20 49.27 -114.4 1686 67 - 38 3.5 M t177 West Castle 8-7-20 49.27 -114.41 1784 344 - 42 5 S t182 West Castle 10-7-20 49.34 -114.42 1405 241 - 12 5 M t183 West Castle 10-7-20 49.35 -114.41 1409 120 - 22 3.5 M t184 West Castle 9-7-20 49.37 -114.42 1751 138 + 10 5 C t185 West Castle 26-6-20 49.38 -114.35 1394 22 - 8 2.3 M t186 West Castle 19-8-20 49.38 -114.38 1417 112 - 4 3.75 M t189 West Castle 9-7-20 49.37 -114.42 1734 104 - 59 5 S t190 West Castle 9-7-20 49.37 -114.42 1686 115 - 48 5 S o3 Middle 29-7-21 49.38 -114.37 1492 200 + 10 NA G Castle o4 Middle 8-6-21 49.41 -114.35 1429 173 - 27 NA S Castle o10 Upper Castle 7-6-21 49.39 -114.33 1430 320 - 3 NA M 119 o13 Carbondale 28-6-21 49.42 -114.44 1374 276 - 13 NA C o18 Middle 30-6-21 49.40 -114.37 1513 82 - 2 NA C Castle o19 West Castle 3-6-21 49.29 -114.39 1649 250 - 20 NA C o21 Upper Castle 26-7-21 49.30 -114.33 1773 255 + 15 NA B o23 Mill Creek 24-6-21 49.35 -114.16 1492 224 + 1 NA C o25 Carbondale 3-8-21 49.41 -114.43 1703 170 + 19 NA C o31 West Castle 15-6-21 49.34 -114.43 1632 105 - 30 NA C o33 Carbondale 28-7-21 49.47 -114.5 1745 201 + 13 NA C o37 Upper Castle 12-7-21 49.31 -114.29 1591 200 + 11 NA S o42 Drywood 29-6-21 49.29 -114.06 1479 78 - 1 NA B Creek o45 Carbondale 17-6-21 49.54 -114.5 1834 222 + 32 NA S o54 Carbondale 23-6-21 49.42 -114.49 1612 102 + 2 NA S o57 West Castle 20-7-21 49.28 -114.38 1559 13 - 21 NA M o68 Carbondale 14-7-21 49.53 -114.55 1801 204 + 25 NA S o70 Drywood Creek 22-7-21 49.27 -114.1 2000 150 + 40 NA S o73 Upper 15-7-21 49.60 -114.66 1646 215 + 41 NA S Crowsnest o82 Carbondale 7-7-21 49.38 -114.53 1891 336 - 31 NA S o105 Carbondale 16-6-21 49.51 -114.51 1522 7 - 2 NA S o106 Upper Crowsnest 27-7-21 49.58 -114.56 1844 102 - 6 NA C o123 Drywood 21-7-21 49.24 -114.08 1720 160 + 17 NA B Creek o126 Carbondale 23-7-21 49.51 -114.55 1762 15 - 13 NA C 120 Table A1.2. Results of pairwise comparisons of estimated marginal mean maximum height of vegetation of each trail type compared between distances from trail. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom (df) are determined using Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean maximum height of vegetation between quadrats at different distances, without adjustments or with the Tukey adjustment for multiple tests. Trail use Distance emmean SE df lower upper Difference Difference type (m) (not (Tukey adjusted) adjusted) Control 0 70.5 7.46 411 55.8 85.1 a a 2 64.4 7.46 411 49.7 79 a a 5 67.9 7.46 411 53.3 82.6 a a 10 61.2 7.46 411 46.5 75.9 a a Footpath 0 68.7 7.49 411 53.9 83.4 a a 2 71.4 7.49 411 56.7 86.1 a a 5 68.1 7.49 411 53.4 82.8 a a 10 78.7 7.49 411 63.9 93.4 a a OHV 0 68 4.18 411 59.7 76.2 a a 2 78.9 4.18 411 70.7 87.2 b a 5 72.2 4.18 411 64 80.4 ab a 10 71.4 4.18 411 63.2 79.6 ab a Road 0 53.7 9.66 411 34.7 72.7 a a 2 60.5 9.66 411 41.6 79.5 a ab 5 63.1 9.66 411 44.1 82.1 ab ab 10 83.7 9.66 411 64.7 102.7 b b 121 Table A1.3. Results of pairwise comparisons of estimated marginal mean soil compaction values at each distance compared between trail types. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom (df) are determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean soil compaction values between quadrats without adjustments or with the Tukey adjustment for multiple tests. Distance Trail use emmean SE df lower upper Difference Difference type (not (Tukey adjusted) adjusted) 0m Control 0.837 0.225 315 0.3939 1.279 a a Footpath 1.581 0.229 329 1.1294 2.032 b ab OHV 1.795 0.124 303 1.5501 2.04 b b Road 2.668 0.293 319 2.0926 3.244 c c 2m Control 0.847 0.225 315 0.4043 1.29 a a Footpath 0.966 0.229 329 0.5144 1.417 a a OHV 1.121 0.124 303 0.8763 1.366 a a Road 2.16 0.293 319 1.584 2.736 b b 5m Control 1.035 0.225 315 0.5918 1.477 a a Footpath 0.965 0.229 329 0.5135 1.416 a a OHV 0.704 0.124 303 0.4586 0.949 a a Road 1.176 0.293 319 0.5997 1.751 a a 10m Control 0.92 0.225 315 0.4773 1.363 a a Footpath 0.776 0.229 329 0.3249 1.228 a a OHV 0.753 0.124 303 0.508 0.998 a a Road 0.676 0.293 319 0.0997 1.251 a a 122 Table A1.4. Results of pairwise comparisons of estimated marginal mean (emmean) soil compaction values for each trail type between distances from trail. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom (df) are determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean soil compaction values between quadrats at different distances, without adjustments or with the Tukey adjustment for multiple tests. Trail use Distance emmean SE df lower upper Difference Difference type (m) (not (Tukey adjusted) adjusted) Control 0 1.00 0.06 291 0.876 1.13 a a 2 1.01 0.06 291 0.879 1.13 a a 5 1.06 0.06 291 0.929 1.18 a a 10 1.02 0.06 291 0.888 1.14 a a Footpath 0 1.223 0.07 303 1.094 1.35 a a 2 1.044 0.07 303 0.915 1.17 b b 5 1.026 0.07 303 0.897 1.15 b b 10 0.960 0.07 303 0.831 1.09 b b OHV 0 1.27 0.04 280 1.200 1.34 a a 2 1.09 0.04 280 1.019 1.16 b b 5 0.96 0.04 280 0.889 1.03 c c 10 0.97 0.04 280 0.903 1.04 c c Road 0 1.44 0.08 294 1.277 1.61 a a 2 1.35 0.08 294 1.187 1.52 a a 5 1.09 0.08 294 0.928 1.26 b b 10 0.94 0.08 294 0.780 1.11 b b 123 Table A1.5. List of exotic species (n = 35) recorded during two survey seasons, from June to August of 2020 and 2021. Last four columns report the minimum, mean, median, and maximum elevation (meters above sea level) of each species surveyed. Scientific name Lifeform Lifecycle Min. Mean Med. Max. Elev Elev. Elev. Elev. (m) (m) (m) (m) Agrostis gigantea grass perennial 1378 1455 1456 1530 Agrostis stolonifera grass perennial 1432 1494 1491 1559 Alyssum alyssoides herb annual 1436 1436 1436 1436 Bromus inermis grass perennial 1350 1577 1572 1847 Capsella bursa-pastoris herb annual 1421 1421 1421 1421 Carum carvi herb biennial 1469 1469 1469 1469 Cerastium fontanum herb perennial 1457 1512 1513 1586 Cirsium arvense herb perennial 1368 1492 1467 1799 Dactylis glomerata grass perennial 1544 1544 1544 1544 Echium vulgare herb perennial 1812 1812 1812 1812 Festuca ovina grass perennial 1378 1421 1406 1479 Festuca trachyphylla grass perennial 1503 1503 1503 1503 Leucanthemum vulgare herb perennial 1349 1487 1477 1799 Linaria vulgaris herb perennial 1344 1485 1505 1617 Lolium pratense grass perennial 1421 1421 1421 1421 Matricaria discoidea herb annual 1510 1510 1510 1510 Medicago lupulina herb perennial 1349 1503 1510 1688 Melilotus alba herb annual 1683 1683 1683 1683 Melilotus officinalis herb annual 1585 1634 1634 1683 Phleum pratense grass perennial 1338 1524 1508 1900 Pilosella aurantiacum herb perennial 1694 1750 1745 1812 Plantago major herb perennial 1338 1490 1483 1657 Poa annua grass annual 1409 1537 1537 1664 Poa compressa grass perennial 1349 1529 1499 1900 Poa pratensis ssp. Pratensis grass perennial 1338 1499 1479 1828 Ranunculus acris herb perennial 1350 1520 1522 1794 Rumex acetosella herb perennial 1462 1462 1462 1462 Taraxacum officinale herb perennial 1338 1541 1510 1914 Thlaspi arvense herb annual 1421 1421 1421 1421 Tragopogon dubius herb perennial 1349 1444 1449 1506 Trifolium aureum herb perennial 1456 1459 1457 1465 Trifolium hybridum herb perennial 1338 1499 1494 1799 Trifolium pratense herb perennial 1349 1530 1478 1847 Trifolium repens herb perennial 1338 1500 1483 1799 Verbascum thapsus herb annual 1522 1608 1608 1694 124 Table A1.6. List of rare species (n = 15) recorded during two survey seasons, from June to August of 2020 and 2021. The last column indicates conservation status rank for Alberta (S1 = critically imperiled; S2 = imperiled). Scientific name Lifeform Lifecycle Provincial Status Adenocaulon bicolor herb perennial S2 Artemisia tridentata subshrub perennial S2 Carex geyeri sedge perennial S2 Draba reptans herb annual S2 Elymus scribneri grass perennial S2 Epilobium leptocarpum herb perennial S2 Festuca occidentalis grass perennial S2 Lupinus lepidus herb perennial S2 Melica spectabilis herb perennial S2 Melica subulata grass perennial S2 Microsteris gracilis herb annual S1 Paxistima myrsinites shrub perennial S2 Platanthera unalascensis herb perennial S2 Trisetum canescens grass perennial S2 Viola glabella herb perennial S2 125 Table A1.7. Results of pairwise comparisons of estimated marginal mean species richness at each distance for each vegetation type. Results are averaged over the most common trail type (OHV). Degrees of freedom (df) are determined using Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean species richness between quadrats at different distances, without adjustments or with the Tukey adjustment for multiple tests. Vegetation Distance mean SE df lower upper Diff. (not Diff. (m) adjusted) (Tukey adjusted) Grassland 0 11.3 1.28 289 8.77 13.8 a a 2 11.3 1.28 289 8.81 13.9 a a 5 13.5 1.28 289 10.96 16 a a 10 13.9 1.28 289 11.33 16.4 a a Shrubland 0 14.6 1.07 291 12.51 16.7 ab ab 2 16 1.07 291 13.87 18.1 a a 5 15.6 1.07 291 13.45 17.7 a a 10 12.8 1.07 291 10.64 14.9 b b Mixed 0 13.1 0.86 285 11.42 14.8 a a 2 14.5 0.86 285 12.84 16.2 a a 5 14.1 0.86 285 12.42 15.8 a a 10 12.9 0.86 285 11.18 14.6 a a Coniferous 0 12.7 0.94 291 10.82 14.5 a a 2 12.7 0.94 291 10.83 14.5 a a 5 11.2 0.94 291 9.37 13.1 ab ab 10 10.1 0.94 291 8.22 11.9 b b 126 Table A1.8. Results of pairwise comparisons of mean species richness (emmean) at each distance from trail for each trail type. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom (df) determined using Kenward- Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean species richness between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Trail use Distance emmean SE df lower upper Difference Difference type (m) (not (Tukey adjusted) adjusted) Control 0 12.16 1.17 288 9.87 14.5 a a 2 12.14 1.17 288 9.85 14.4 a a 5 12.31 1.17 288 10.01 14.6 a a 10 13.04 1.17 288 10.74 15.3 a a Footpath 0 16.43 1.16 293 14.16 18.7 a a 2 15.85 1.16 293 13.57 18.1 a a 5 13.75 1.16 293 11.47 16 ab ab 10 11.73 1.16 293 9.45 14 b b OHV 0 15.21 0.65 293 13.94 16.5 a a 2 13.99 0.65 293 12.72 15.3 ab a 5 13.7 0.65 293 12.43 15 b a 10 11.99 0.65 293 10.72 13.3 c b Road 0 7.89 1.51 287 4.92 10.9 a a 2 12.54 1.51 287 9.57 15.5 b b 5 14.62 1.51 287 11.64 17.6 b b 10 12.79 1.51 287 9.82 15.8 b b 127 Table A1.9. Results of estimated marginal means of linear trends comparing the slope of the relationship between species richness and distance from trail (trend) for the subset data of only footpaths and OHV trails within each vegetation type. Degrees of freedom (df) determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Bolded ‘p-value’ indicates significant differences between slopes. Vegetation type Trail type trend SE df lower upper p-value Grassland footpath -0.010 0.224 304 -0.45 0.431 0.647 OHV 0.118 0.167 304 -0.21 0.447 Shrubland footpath -0.895 0.205 304 -1.297 -0.493 0.008 OHV -0.232 0.139 304 -0.506 0.041 Mixed footpath -0.796 0.224 304 -1.237 -0.356 0.025 OHV -0.254 0.087 304 -0.426 -0.083 Coniferous footpath -0.289 0.205 304 -0.691 0.114 0.087 OHV -0.677 0.096 304 -0.867 -0.487 Table A1.10. Results of pairwise comparisons of mean community dissimilarity (emmean) within each vegetation type. Results are averaged over the most common trail type (OHV). Degrees of freedom (df) determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in final two columns denote significant differences in mean Bray-Curtis dissimilarity values between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Vegetation emmean SE df lower upper Differences Differences (not (Tukey adjusted) adjusted) Grassland 0.496 0.0367 132 0.423 0.569 a a Shrubland 0.664 0.0293 132 0.606 0.722 b b Mixed 0.629 0.0231 132 0.583 0.675 b b Coniferous 0.661 0.0258 132 0.61 0.712 b b 128 Table A1.11. Results of pairwise comparisons of mean Bray-Curtis dissimilarity values (emmean) between each trail type at each distance from trail. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom (df) determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in final two columns denote significant differences in mean Bray-Curtis dissimilarity values between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Distance Trail type emmean SE df lower upper Diff. (not Diff. compared adjusted) (Tukey to 10m adjusted) Control 0.536 0.0364 221 0.464 0.607 a a 0m Footpath 0.643 0.0368 228 0.571 0.716 bc ab OHV 0.724 0.0206 213 0.683 0.764 b b Road 0.771 0.0489 223 0.674 0.867 b b Control 0.548 0.0364 221 0.477 0.62 a a 2m Footpath 0.611 0.0368 228 0.538 0.683 ab a OHV 0.64 0.0206 213 0.599 0.68 b a Road 0.694 0.0489 223 0.597 0.79 b a Control 0.541 0.0364 221 0.469 0.613 a a 5m Footpath 0.546 0.0368 228 0.473 0.618 a a OHV 0.55 0.0206 213 0.509 0.59 a a Road 0.549 0.0489 223 0.452 0.645 a a Table A1.12. Results of pairwise comparisons of mean Bray-Curtis dissimilarity values (compared to the 10 m quadrat) for footpaths and OHV trails (emmean) for each vegetation type. Results are averaged over the levels of distance to trail. Degrees of freedom (df) determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in the final two columns denote significant differences in mean Bray-Curtis dissimilarity values between footpaths and OHV trails, without adjustments or with the Tukey adjustment for multiple tests. Vegetation Trail type emmean SE df lower upper Diff. (not Diff. adjusted) (Tukey adjusted) Grassland Footpath 0.38 0.062 94 0.26 0.506 a a OHV 0.55 0.049 94 0.45 0.644 b b Shrubland Footpath 0.78 0.057 94 0.663 0.888 a a OHV 0.65 0.038 94 0.574 0.726 a a Mixed Footpath 0.58 0.062 94 0.458 0.704 a a OHV 0.66 0.024 94 0.609 0.705 a a Coniferous Footpath 0.64 0.057 94 0.525 0.749 a a OHV 0.69 0.027 94 0.641 0.748 a a 129 Table A1.13. Results of pairwise comparisons of the mean probability of exotic species presence (emmean) within each vegetation type. Results are averaged over the most common trail type (OHV). Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Different lower-case letters in final two columns denote significant differences in mean probability of exotic species presence between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Vegetation emmean SE df lower upper Differences Differences type (not (Tukey adjusted) adjusted) Grassland 1.00 0.0019 Inf 0.924237 1 a a Shrubland 0.96 0.0522 Inf 0.617281 0.9973 ac ab Mixed 0.72 0.22 Inf 0.236775 0.9544 c bc Coniferous 0.02 0.03 Inf 0.000948 0.251 b c Table A1.14a. Results of pairwise comparisons of the mean probability of exotic species present (emmean) at each distance from trail for each trail type. Results are averaged over the most common vegetation type (mixed/broadleaf). Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Intervals are back- transformed from the logit scale. Different lower-case letters in pairwise significance column denote significant differences in the odds ratio of the probability of exotic species present between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Trail type Distance emmean SE df lower upper Diff. (not Diff. (m) adjusted) (Tukey adjusted) 0 0.32 0.34 Inf 0.023 0.907 a a Control 2 0.32 0.34 Inf 0.023 0.907 a a 5 0.17 0.23 Inf 0.009 0.829 a a 10 0.32 0.34 Inf 0.023 0.907 a a 0 1.00 0.00 Inf 0.854 0.999 a a Footpath 2 0.91 0.13 Inf 0.297 0.996 a ab 5 0.19 0.25 Inf 0.010 0.842 b b 10 0.03 0.05 Inf 0.001 0.464 b bc 0 1.00 0.00 Inf 0.992 1 a a OHV 2 0.97 0.03 Inf 0.795 0.997 bc bc 5 0.73 0.18 Inf 0.312 0.943 b b 10 0.65 0.20 Inf 0.245 0.917 b b 0 0.99 0.03 Inf 0.295 0.999 a a Road 2 1.00 0.01 Inf 0.608 1 ab a 5 0.93 0.16 Inf 0.107 0.999 a a 10 0.78 0.39 Inf 0.040 0.997 a a 130 Table A1.14b. Results of pairwise comparisons of the probability of exotic species present (emmean) for each trail type at each distance. Results are averaged over the most common vegetation type (mixed/broadleaf). Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Intervals are back-transformed from the logit scale. Different lower-case letters in pairwise significance column denote significant differences in the odds ratio of the probability of exotic species present between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Difference Difference Distance Trail type Prob. SE df lower upper (not (Tukey adjusted) adjusted) Control 0.32 0.33 Inf 0.023 0.907 a a 0m Footpath 1.00 0.0042 Inf 0.854 0.999 b ab OHV 1.00 0.0002 Inf 0.992 1 b b Road 0.99 0.03 Inf 0.295 0.999 ab ab Control 0.32 0.34 Inf 0.023 0.907 a a 2m Footpath 0.91 0.13 Inf 0.297 0.996 ab a OHV 0.97 0.03 Inf 0.795 0.997 b a Road 1.00 0.01 Inf 0.608 1 b a Control 0.17 0.23 Inf 0.009 0.829 a a 5m Footpath 0.19 0.25 Inf 0.010 0.842 a a OHV 0.73 0.18 Inf 0.312 0.943 a a Road 0.93 0.16 Inf 0.107 0.999 a a Control 0.32 0.34 Inf 0.023 0.907 a a 10m Footpath 0.03 0.05 Inf 0.001 0.464 ac a OHV 0.65 0.20 Inf 0.245 0.917 ab a Road 0.78 0.39 Inf 0.040 0.997 a a 131 Table A1.15. Results of estimated marginal means of linear trends comparing the slope of the relationship between the probability of at least one exotic species and distance from trail (trend) for the subset data of only footpaths and OHV trails within each vegetation type. Degrees of freedom (df) determined using the Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Bolded ‘p-value’ indicates significant differences between slopes. Vegetation Trail type trend SE df lower upper p-value Grassland footpath -0.000 1.204 401 -2.37 2.37 0.348 OHV -1.375 0.833 401 -3.01 0.263 Shrubland* footpath -4.759 1.453 401 -7.62 -1.902 0.013 OHV -1.261 0.303 401 -1.86 -0.665 Coniferous footpath -1.171 0.747 401 -2.64 0.298 0.033 OHV -3.586 0.946 401 -5.45 -1.726 *Shrubland includes shrubland, mixed, and broadleaf vegetation types due to low replication and to avoid unreasonable slope estimates. Table A1.16. Results of pairwise comparisons (emmeans) of the mean probability of at least one exotic species per transect compared between each vegetation type. Degrees of freedom (df) determined using Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Intervals are back- transformed from the logit scale. Different lower-case letters in pairwise significance column denote significant differences in the mean probability of exotic species present between transects, without adjustments or with the Tukey adjustment for multiple tests. Vegetation estimate SE df lower upper Diff. (not Diff. adjusted) (Tukey adjusted) Grassland 1 9.09E-13 Inf 2.22E-16 1 ab ab Shrubland 1 9.08E-13 Inf 2.22E-16 1 a a Mixed 1 9.08E-13 Inf 2.22E-16 1 ab ab Coniferous 1 9.08E-13 Inf 2.22E-16 1 b b 132 Table A1.17. Results of pairwise comparisons (emtrends) of the slope of the relationship between the probability of at least one exotic species per transect and elevation compared between each trail type. Results are averaged over the most common vegetation type (mixed/broadleaf). Degrees of freedom determined using Kenward-Roger method. Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Intervals are back-transformed from the logit scale. Different lower-case letters in pairwise significance column denote significant differences in the mean probability of exotic species present per transect, without adjustments or with the Tukey adjustment for multiple tests. Trail use emmean SE df lower upper Difference Difference type (not (Tukey adjusted) adjusted) Control 2.52 37.2 Inf -70.3 75.3 a a Footpath 10.74 39.8 Inf -67.3 88.8 ab ab OHV 5.36 37.2 Inf -67.5 78.2 b b Road 188.34 16361.2 Inf -31879 32255.7 ab ab Table A1.18. Results of pairwise comparisons of the probability of rare species present (emmean) compared between each trail type. Results are averaged over the most common vegetation type (mixed/broadleaf). Lower and upper confidence intervals for the estimates are included (confidence level = 0.95). Intervals are back-transformed from the logit scale. Different lower-case letters in pairwise significance column denote significant differences in the odds ratio of the probability of rare species present between quadrats, without adjustments or with the Tukey adjustment for multiple tests. Trail use emmean SE df lower upper Difference Difference type (not (Tukey adjusted) adjusted) control 0.4704 0.1148 Inf 0.2646 0.687 a a footpath 0.2869 0.1011 Inf 0.13251 0.515 ab a OHV 0.2375 0.0552 Inf 0.14627 0.361 b a road 0.0655 0.0648 Inf 0.00875 0.358 b a 133 APPENDIX 2: Chapter 3 Supplementary Materials Table A2.1. Results of the indicator species analysis for plots with or without Botrychium according to an indicator species analysis with 9,999 permutations. Values include the relative frequency and average relative abundance of species occurring in plots with Botrychium absent or present as well as the group each species has maximum indicator value for Botrychium (Occurrence- present or absent). Significant indicator species are shown in bold lettering. Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Abies balsamea 0.647 0.429 0.731 0.269 absent 0.183 Acer glabrum 0.588 0.143 0.908 0.092 absent 0.049 Achillea 0.824 0.857 0.464 0.536 present 0.659 millefolium Actaea rubra 0.353 0.286 0.531 0.469 absent 0.939 Agoseris 0.235 0.143 0.490 0.510 absent 1.000 aurantiaca Agoseris glauca 0.118 0.571 0.171 0.829 present 0.029 Agrostis scabra 0.118 0.143 0.407 0.593 present 0.844 Agrostis 0.059 0.000 1.000 0.000 absent 1.000 stolonifera Allium cernuum 0.529 0.429 0.602 0.398 absent 0.612 Alnus alnobetula 0.471 0.143 0.781 0.219 absent 0.243 Alnus incana ssp. 0.000 0.143 0.000 1.000 present 0.296 tenuifolia Alyssum 0.059 0.000 1.000 0.000 absent 1.000 alyssoides Amelanchier 0.765 0.571 0.629 0.371 absent 0.348 alnifolia Anaphalis 0.588 0.286 0.641 0.359 absent 0.370 margaritacea Anemone 0.412 0.714 0.403 0.597 present 0.223 multifida Anemone 0.000 0.143 0.000 1.000 present 0.287 parviflora Angelica arguta 0.118 0.000 1.000 0.000 absent 0.570 Angelica 0.412 0.571 0.368 0.632 present 0.281 dawsonii Antennaria 0.000 0.143 0.000 1.000 present 0.285 alpina Antennaria 0.176 0.429 0.490 0.510 present 0.548 howellii Antennaria media 0.000 0.143 0.000 1.000 present 0.289 134 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Antennaria 0.000 0.143 0.000 1.000 present 0.302 microphylla Antennaria 0.176 0.000 1.000 0.000 absent 0.529 parvifolia Antennaria 0.000 0.143 0.000 1.000 present 0.287 pulcherrima Antennaria 0.353 0.143 0.742 0.258 absent 0.428 racemosa Antennaria rosea 0.353 0.429 0.370 0.630 present 0.560 Antennaria 0.176 0.429 0.265 0.735 present 0.224 umbrinella Anticlea elegans 0.294 0.429 0.469 0.531 present 0.666 Anticlea 0.235 0.429 0.397 0.603 present 0.421 occidentalis Aphyllon uniflora 0.000 0.143 0.000 1.000 present 0.297 Apocynum 0.118 0.000 1.000 0.000 absent 0.567 androsaemifolium Aquilegia 0.176 0.286 0.325 0.676 present 0.367 flavescens Arabis nuttallii 0.000 0.286 0.000 1.000 present 0.074 Arctostaphylos 0.471 0.429 0.531 0.469 absent 0.861 uva-ursi Arnica cordifolia 0.824 0.571 0.613 0.387 absent 0.228 Arnica fulgens 0.000 0.143 0.000 1.000 present 0.288 Arnica latifolia 0.059 0.000 1.000 0.000 absent 1.000 Arnica ovata 0.000 0.286 0.000 1.000 present 0.076 Artemisia 0.059 0.000 1.000 0.000 absent 1.000 ludoviciana Artemisia 0.176 0.143 0.490 0.510 absent 1.000 michauxiana Artemisia 0.059 0.000 1.000 0.000 absent 1.000 tridentata Astragalus 0.059 0.143 0.292 0.708 present 1.000 alpinus Astragalus 0.000 0.143 0.000 1.000 present 0.287 bourgovii Astragalus 0.059 0.000 1.000 0.000 absent 1.000 canadensis Astragalus 0.000 0.286 0.000 1.000 present 0.079 vexilliflexus Balsamorhiza 0.059 0.000 1.000 0.000 absent 1.000 sagittata 135 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Berberis repens 0.647 0.429 0.638 0.362 absent 0.373 Betula 0.059 0.000 1.000 0.000 absent 1.000 glandulosa Boechera 0.118 0.429 0.292 0.708 present 0.126 lemmonii Boechera stricta 0.118 0.000 1.000 0.000 absent 0.564 Bromus carinatus 0.176 0.143 0.523 0.477 absent 0.884 Bromus ciliatus 0.059 0.143 0.215 0.785 present 0.789 Bromus inermis 0.294 0.286 0.507 0.493 absent 1.000 Bromus 0.059 0.143 0.198 0.802 present 0.784 pumpellianus Bromus vulgaris 0.118 0.000 1.000 0.000 absent 0.568 Calamagrostis 0.235 0.000 1.000 0.000 absent 0.283 canadensis Calamagrostis 0.000 0.143 0.000 1.000 present 0.295 purpurescens Calamagrostis 0.647 0.429 0.567 0.433 absent 0.609 rubescens Calamagrostis 0.059 0.000 1.000 0.000 absent 1.000 stricta Calochortus 0.235 0.143 0.712 0.288 absent 0.585 apiculatus Calypso bulbosa 0.000 0.143 0.000 1.000 present 0.293 Campanula 0.706 0.429 0.667 0.333 absent 0.231 rotundifolia Carex aurea 0.059 0.000 1.000 0.000 absent 1.000 Carex bebbii 0.059 0.000 1.000 0.000 absent 1.000 Carex capillaris 0.000 0.143 0.000 1.000 present 0.293 Carex 0.471 0.286 0.622 0.378 absent 0.571 concinnoides Carex deweyana 0.059 0.000 1.000 0.000 absent 1.000 Carex flava 0.059 0.000 1.000 0.000 absent 1.000 Carex geyeri 0.118 0.000 1.000 0.000 absent 0.560 Carex hoodii 0.235 0.286 0.579 0.421 absent 1.000 Carex microptera 0.118 0.143 0.452 0.548 present 1.000 Carex obtusata 0.059 0.143 0.171 0.829 present 0.289 Carex 0.000 0.286 0.000 1.000 present 0.075 phaeocephala Carex 0.000 0.143 0.000 1.000 present 0.297 praegracilis 136 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Carex raynoldsii 0.000 0.143 0.000 1.000 present 0.285 Carex rossii 0.588 0.429 0.664 0.336 absent 0.408 Carex siccata 0.059 0.000 1.000 0.000 absent 1.000 Castilleja hispida 0.176 0.286 0.475 0.525 present 0.919 Castilleja miniata 0.706 0.429 0.553 0.447 absent 0.528 Castilleja 0.059 0.286 0.121 0.879 present 0.078 occidentalis Ceanothus 0.118 0.000 1.000 0.000 absent 0.569 velutinus Cerastium 0.294 0.286 0.523 0.477 absent 0.891 arvense Cerastium 0.000 0.143 0.000 1.000 present 0.293 fontanum Cerastium nutans 0.059 0.286 0.171 0.829 present 0.133 Chamaenerion 0.882 0.714 0.536 0.464 absent 0.524 angustifolium Chamaenerion 0.059 0.143 0.171 0.829 present 0.281 latifolium Cherleria 0.118 0.286 0.331 0.669 present 0.552 obtusiloba Chimaphila 0.529 0.143 0.794 0.206 absent 0.167 umbellata Circaea alpina 0.059 0.000 1.000 0.000 absent 1.000 Cirsium arvense 0.118 0.286 0.198 0.802 present 0.378 Cirsium 0.000 0.143 0.000 1.000 present 0.296 flodmanii Cirsium 0.235 0.286 0.490 0.510 present 0.943 hookerianum Cirsium vulgare 0.059 0.143 0.292 0.708 present 1.000 Claytonia 0.059 0.000 1.000 0.000 absent 1.000 lanceolata Clematis 0.529 0.571 0.513 0.487 present 1.000 occidentalis Clintonia uniflora 0.647 0.286 0.680 0.320 absent 0.310 Coeloglossum 0.059 0.143 0.215 0.785 present 0.785 viride Collinsia 0.118 0.143 0.407 0.593 present 0.840 parviflora Collomia linearis 0.059 0.286 0.093 0.907 present 0.075 Comandra 0.176 0.143 0.523 0.477 absent 1.000 umbellata 137 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Corallorhiza 0.118 0.000 1.000 0.000 absent 0.573 maculata Corallorhiza 0.118 0.143 0.452 0.548 present 1.000 striata Corallorhiza 0.118 0.143 0.382 0.618 present 0.835 trifida Cornus 0.235 0.286 0.475 0.525 present 0.926 canadensis Cornus 0.059 0.143 0.452 0.548 present 1.000 stolonifera Crataegus 0.059 0.000 1.000 0.000 absent 1.000 chrysocarpa Cryptantha 0.059 0.000 1.000 0.000 absent 1.000 celosioides Cystopteris 0.294 0.429 0.354 0.646 present 0.439 fragilis Danthonia 0.059 0.000 1.000 0.000 absent 1.000 intermedia Danthonia 0.059 0.000 1.000 0.000 absent 1.000 spicata Dasiphora 0.471 0.714 0.368 0.632 present 0.213 fruticosa Doellingeria 0.529 0.143 0.742 0.258 absent 0.191 engelmannii Draba aurea 0.118 0.286 0.292 0.708 present 0.500 Draba 0.059 0.143 0.382 0.618 present 1.000 lonchocarpa Draba paysonii 0.000 0.143 0.000 1.000 present 0.295 Draba reptans 0.059 0.000 1.000 0.000 absent 1.000 Dracocephalum 0.000 0.143 0.000 1.000 present 0.287 parviflorum Dracocephalum 0.000 0.143 0.000 1.000 present 0.288 thymiflorum Dryas 0.059 0.143 0.121 0.879 present 0.297 drummundii Dryas hookeriana 0.000 0.143 0.000 1.000 present 0.294 Drymocallis 0.176 0.143 0.649 0.351 absent 0.655 arguta Drymocallis 0.000 0.143 0.000 1.000 present 0.285 glandulosa Echium vulgare 0.000 0.143 0.000 1.000 present 0.279 138 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Elaeagnus 0.059 0.143 0.407 0.593 present 1.000 commutata Elymus glaucus 0.471 0.000 1.000 0.000 absent 0.052 Elymus 0.000 0.143 0.000 1.000 present 0.291 lanceolatus Elymus scribneri 0.000 0.143 0.000 1.000 present 0.296 Elymus 0.000 0.143 0.000 1.000 present 0.287 trachycaulus ssp. trachycaulus Epilobium 0.059 0.143 0.236 0.764 present 0.782 anagallidifolium Epilobium 0.059 0.143 0.236 0.764 present 0.787 brachycarpum Equisetum 0.235 0.286 0.553 0.447 absent 1.000 arvense Equisetum 0.000 0.143 0.000 1.000 present 0.286 fluviatile Equisetum 0.000 0.286 0.000 1.000 present 0.081 hyemale Erigeron acris 0.000 0.143 0.000 1.000 present 0.302 Erigeron 0.059 0.000 1.000 0.000 absent 1.000 caespitosus Erigeron 0.176 0.429 0.265 0.735 present 0.215 compositus Erigeron 0.000 0.143 0.000 1.000 present 0.290 glabellus var. pubescens Erigeron 0.059 0.000 1.000 0.000 absent 1.000 peregrinus Erigeron 0.176 0.143 0.523 0.477 absent 1.000 speciosus Eriogonum 0.000 0.143 0.000 1.000 present 0.288 ovalifolium Eriogonum 0.235 0.000 1.000 0.000 absent 0.276 umbellatum Erythronium 0.235 0.571 0.292 0.708 present 0.137 grandiflorum Eurybia 0.765 0.571 0.559 0.441 absent 0.599 conspicua Eurybia sibirica 0.118 0.429 0.215 0.785 present 0.131 Festuca 0.176 0.286 0.407 0.593 present 0.603 campestris 139 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Festuca 0.235 0.143 0.602 0.398 absent 1.000 idahoensis Festuca 0.059 0.000 1.000 0.000 absent 1.000 occidentalis Festuca rubra 0.059 0.000 1.000 0.000 absent 1.000 Festuca 0.000 0.143 0.000 1.000 present 0.295 saximontana Festuca subulata 0.118 0.000 1.000 0.000 absent 0.558 Festuca 0.059 0.000 1.000 0.000 absent 1.000 trachyphylla Fragaria 0.941 1.000 0.497 0.503 present 0.955 virginiana Gaillardia 0.176 0.143 0.523 0.477 absent 1.000 aristata Galium boreale 0.765 0.714 0.498 0.502 absent 0.982 Galium triflorum 0.294 0.286 0.517 0.483 absent 1.000 Gentianella 0.235 0.000 1.000 0.000 absent 0.324 amarella Geranium 0.176 0.429 0.261 0.739 present 0.215 richardsonii Geranium 0.412 0.143 0.673 0.327 absent 0.441 viscosissimum Geum aleppicum 0.059 0.000 1.000 0.000 absent 1.000 Geum 0.059 0.143 0.292 0.708 present 1.000 macrophyllum Geum triflorum 0.059 0.143 0.215 0.785 present 0.787 Glycyrrhiza 0.059 0.000 1.000 0.000 absent 1.000 lepidota Goodyera 0.647 0.143 0.805 0.195 absent 0.069 oblongifolia Gymnocarpium 0.118 0.000 1.000 0.000 absent 0.574 dryopteris Hackelia 0.118 0.143 0.553 0.447 absent 1.000 micrantha Hedysarum 0.412 0.714 0.362 0.639 present 0.177 sulphurescens Heracleum 0.353 0.429 0.572 0.428 absent 0.991 maximum Heterotheca 0.118 0.000 1.000 0.000 absent 0.578 villosa Heuchera 0.353 0.286 0.536 0.464 absent 0.960 cylindrica 140 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Hieracium 0.471 0.143 0.819 0.181 absent 0.154 albiflorum Hieracium 0.176 0.000 1.000 0.000 absent 0.538 scouleri Hieracium triste 0.235 0.429 0.401 0.599 present 0.403 Hieracium 0.118 0.143 0.553 0.447 absent 1.000 umbellatum Juncus 0.059 0.000 1.000 0.000 absent 1.000 drummondii Juniperus 0.706 0.429 0.641 0.359 absent 0.286 communis Juniperus 0.176 0.143 0.622 0.378 absent 0.894 horizontalis Koeleria 0.294 0.143 0.641 0.359 absent 0.568 macrantha Lathyrus 0.471 0.571 0.426 0.574 present 0.560 ochroleucus Leucanthemum 0.118 0.286 0.236 0.764 present 0.332 vulgare Linaria vulgaris 0.118 0.143 0.354 0.646 present 0.847 Linnaea borealis 0.176 0.286 0.523 0.477 present 0.808 Linum lewisii 0.176 0.143 0.712 0.288 absent 0.662 Lithospermum 0.353 0.143 0.832 0.168 absent 0.295 ruderale Lomatium 0.176 0.143 0.553 0.447 absent 1.000 dissectum Lomatium 0.059 0.000 1.000 0.000 absent 1.000 triternatum Lonicera dioica 0.059 0.000 1.000 0.000 absent 1.000 Lonicera 0.412 0.429 0.490 0.510 present 1.000 involucrata Lonicera 0.588 0.000 1.000 0.000 absent 0.019 utahensis Lupinus sericeus 0.412 0.143 0.759 0.241 absent 0.299 Luzula parviflora 0.118 0.000 1.000 0.000 absent 0.562 Maianthemum 0.882 0.286 0.753 0.247 absent 0.019 racemosum Maianthemum 0.294 0.571 0.227 0.773 present 0.065 stellatum Medicago 0.000 0.286 0.000 1.000 present 0.075 lupulina 141 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Melica 0.059 0.143 0.292 0.708 present 1.000 spectabilis Melica subulata 0.353 0.000 1.000 0.000 absent 0.131 Menziesia 0.412 0.000 1.000 0.000 absent 0.117 ferruginea Micranthes 0.000 0.286 0.000 1.000 present 0.073 occidentalis Mitella breweri 0.059 0.000 1.000 0.000 absent 1.000 Mitella nuda 0.118 0.143 0.590 0.410 absent 1.000 Moehringia 0.118 0.143 0.452 0.548 present 1.000 lateriflora Monarda 0.294 0.143 0.658 0.342 absent 0.620 fistulosa Nassella viridula 0.000 0.143 0.000 1.000 present 0.296 Neottia cordata 0.118 0.143 0.553 0.447 absent 1.000 Orthilia secunda 0.529 0.286 0.682 0.318 absent 0.247 Osmorhiza 0.529 0.286 0.673 0.327 absent 0.299 berteroi Osmorhiza 0.059 0.143 0.292 0.708 present 1.000 depauperata Osmorhiza 0.059 0.429 0.121 0.879 present 0.063 occidentalis Oxyria digyna 0.000 0.143 0.000 1.000 present 0.284 Oxytropis 0.118 0.143 0.292 0.708 present 0.694 campestris var. spicata Oxytropis 0.059 0.143 0.121 0.879 present 0.292 splendens Packera cana 0.235 0.000 1.000 0.000 absent 0.327 Packera 0.471 0.571 0.440 0.560 present 0.561 pseudaurea Parnassia 0.059 0.000 1.000 0.000 absent 1.000 fimbriata Paxistima 0.059 0.000 1.000 0.000 absent 1.000 myrsinites Pedicularis 0.294 0.429 0.382 0.618 present 0.529 bracteosa Penstemon 0.118 0.286 0.340 0.660 present 0.511 albertinus Penstemon 0.588 0.429 0.507 0.493 absent 0.873 confertus 142 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Penstemon 0.000 0.143 0.000 1.000 present 0.289 ellipticus Penstemon lyallii 0.294 0.143 0.553 0.447 absent 0.806 Perideridia 0.059 0.000 1.000 0.000 absent 1.000 gairdneri Phacelia hastata 0.294 0.429 0.472 0.529 present 0.649 Phacelia sericea 0.118 0.286 0.248 0.752 present 0.422 Phleum pratense 0.294 0.571 0.340 0.660 present 0.204 Physaria 0.059 0.143 0.292 0.708 present 1.000 didymocarpa Picea glauca 0.706 0.714 0.500 0.500 present 0.999 Pilosella 0.059 0.000 1.000 0.000 absent 1.000 aurantiacum Pinus contorta 0.588 0.429 0.540 0.460 absent 0.726 Pinus flexilis 0.176 0.286 0.553 0.447 present 0.886 Platanthera 0.059 0.000 1.000 0.000 absent 1.000 dilatata Platanthera 0.000 0.143 0.000 1.000 present 0.292 huronensis Platanthera 0.059 0.000 1.000 0.000 absent 1.000 orbiculata Platanthera 0.176 0.000 1.000 0.000 absent 0.529 stricta Platanthera 0.059 0.000 1.000 0.000 absent 1.000 unalascensis Poa abbreviata 0.000 0.143 0.000 1.000 present 0.289 Poa alpina 0.059 0.286 0.171 0.829 present 0.187 Poa compressa 0.059 0.000 1.000 0.000 absent 1.000 Poa interior 0.059 0.000 1.000 0.000 absent 1.000 Poa palustris 0.118 0.143 0.407 0.593 present 0.844 Poa pratensis 0.235 0.286 0.390 0.610 present 0.702 ssp. Pratensis Poa secunda 0.000 0.286 0.000 1.000 present 0.084 Podagrostis 0.059 0.000 1.000 0.000 absent 1.000 humilis Polemonium 0.059 0.143 0.382 0.618 present 1.000 pulcherrimum Polygonum 0.118 0.143 0.452 0.548 present 1.000 douglasii Polystichum 0.059 0.143 0.292 0.708 present 1.000 lonchitis 143 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Populus 0.235 0.143 0.397 0.603 absent 1.000 balsamifera Populus 0.588 0.286 0.680 0.320 absent 0.326 tremuloides Potentilla 0.059 0.143 0.382 0.618 present 1.000 anserina Potentilla 0.000 0.143 0.000 1.000 present 0.285 glaucophylla Potentilla gracilis 0.235 0.429 0.362 0.639 present 0.411 Primula 0.059 0.143 0.121 0.879 present 0.286 conjugens Prosartes hookeri 0.059 0.000 1.000 0.000 absent 1.000 Prosartes 0.412 0.429 0.507 0.493 present 1.000 trachycarpa Prunella vulgaris 0.118 0.143 0.292 0.708 present 0.685 Prunus 0.118 0.143 0.673 0.327 absent 0.842 pensylvanica Prunus 0.118 0.000 1.000 0.000 absent 0.572 virginiana Pseudoroegneria 0.353 0.286 0.541 0.459 absent 0.910 spicata Pseudotsuga 0.235 0.000 1.000 0.000 absent 0.329 menziesii Pteridium 0.059 0.000 1.000 0.000 absent 1.000 aquilinum Pulsatilla 0.000 0.143 0.000 1.000 present 0.293 nuttalliana Pyrola asarifolia 0.294 0.143 0.641 0.359 absent 0.633 Pyrola 0.294 0.000 1.000 0.000 absent 0.271 chlorantha Pyrola picta 0.059 0.000 1.000 0.000 absent 1.000 Ranunculus acris 0.059 0.286 0.236 0.764 present 0.197 Ranunculus 0.000 0.143 0.000 1.000 present 0.296 cardiophyllus Ranunculus 0.000 0.143 0.000 1.000 present 0.293 eschscholtzii Ranunculus 0.059 0.143 0.215 0.785 present 0.790 uncinatus Rhinanthus minor 0.235 0.000 1.000 0.000 absent 0.283 Rhodiola 0.059 0.143 0.382 0.618 present 1.000 integrifolia 144 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Rhododendron 0.059 0.000 1.000 0.000 absent 1.000 albiflorum Ribes hirtellum 0.059 0.000 1.000 0.000 absent 1.000 Ribes inerme 0.000 0.143 0.000 1.000 present 0.288 Ribes lacustre 0.588 0.857 0.517 0.483 present 0.479 Ribes 0.294 0.143 0.523 0.477 absent 0.856 viscosissimum Rosa acicularis 0.353 0.429 0.467 0.533 present 0.970 Rosa arkansana 0.118 0.143 0.452 0.548 present 1.000 Rosa woodsii 0.353 0.429 0.465 0.535 present 0.928 Rubus idaeus 0.412 0.714 0.350 0.650 present 0.147 Rubus parviflorus 0.765 0.286 0.726 0.274 absent 0.066 Rumex acetosella 0.000 0.143 0.000 1.000 present 0.291 Sabulina rubella 0.000 0.286 0.000 1.000 present 0.074 Salix bebbiana 0.059 0.286 0.292 0.708 present 0.189 Salix discolor 0.353 0.000 1.000 0.000 absent 0.149 Salix 0.000 0.143 0.000 1.000 present 0.290 drummundiana Salix planifolia 0.059 0.000 1.000 0.000 absent 1.000 Salix scouleriana 0.412 0.286 0.622 0.378 absent 0.578 Salix vestita 0.000 0.143 0.000 1.000 present 0.293 Sambucus 0.235 0.000 1.000 0.000 absent 0.331 racemosa Sanicula 0.000 0.286 0.000 1.000 present 0.074 marilandica Saxifraga 0.059 0.286 0.236 0.764 present 0.192 bronchialis Saxifraga 0.000 0.143 0.000 1.000 present 0.296 mertensiana Sedum 0.294 0.714 0.322 0.678 present 0.088 lanceolatum Sedum 0.059 0.143 0.292 0.708 present 1.000 stenopetalum Selaginella densa 0.176 0.143 0.553 0.447 absent 1.000 Senecio fremontii 0.000 0.286 0.000 1.000 present 0.077 Senecio 0.000 0.143 0.000 1.000 present 0.297 hydrophiloides Senecio 0.059 0.143 0.292 0.708 present 1.000 triangularis 145 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Shepherdia 0.706 0.571 0.553 0.447 absent 0.630 canadensis Silene acaulis 0.000 0.143 0.000 1.000 present 0.294 Silene parryi 0.059 0.000 1.000 0.000 absent 1.000 Sisyrinchium 0.000 0.143 0.000 1.000 present 0.284 montanum Smelowskia 0.000 0.143 0.000 1.000 present 0.292 americana Solidago 0.059 0.000 1.000 0.000 absent 1.000 gigantea Solidago lepida 0.176 0.000 1.000 0.000 absent 0.531 var. salebrosa Solidago 0.118 0.000 1.000 0.000 absent 0.572 missouriensis Solidago 0.294 0.429 0.452 0.548 present 0.638 multiradiata Sorbus scopulina 0.471 0.286 0.755 0.245 absent 0.350 Sorbus sitchensis 0.059 0.000 1.000 0.000 absent 1.000 Spinulum 0.118 0.000 1.000 0.000 absent 0.785 annotinum ssp. Annotinum Spiraea lucida 0.882 0.714 0.517 0.483 absent 0.677 Stellaria crispa 0.059 0.000 1.000 0.000 absent 1.000 Stellaria longipes 0.059 0.143 0.292 0.708 present 1.000 Streptopus 0.235 0.000 1.000 0.000 absent 0.291 amplexifolius Symphoricarpos 0.765 0.571 0.628 0.372 absent 0.325 albus Symphyotrichum 0.588 0.714 0.468 0.532 present 0.616 laeve Symphyotrichum 0.000 0.143 0.000 1.000 present 0.299 puniceum Taraxacum 0.471 0.714 0.368 0.632 present 0.224 officinale Thalictrum 0.941 0.714 0.579 0.421 absent 0.280 occidentale Tiarella trifoliata 0.118 0.000 1.000 0.000 absent 0.567 var. trifoliata Toxicoscordion 0.059 0.143 0.215 0.785 present 0.784 venenosum Tragopogon 0.118 0.143 0.673 0.327 absent 0.848 dubius 146 Scientific Name Relative Relative Relative Relative Occurrence p- frequency frequency abundance abundance value (absent) (present) (absent) (present) Trifolium 0.118 0.143 0.215 0.785 present 0.680 hybridum Trifolium 0.118 0.286 0.171 0.829 present 0.150 pratense Trifolium repens 0.000 0.286 0.000 1.000 present 0.080 Trisetum 0.176 0.000 1.000 0.000 absent 0.528 canescens Trisetum 0.059 0.286 0.141 0.859 present 0.138 spicatum Trollius 0.059 0.000 1.000 0.000 absent 1.000 albiflorus Urtica dioica 0.118 0.000 1.000 0.000 absent 0.774 Vaccinium 0.235 0.143 0.742 0.258 absent 0.642 caespitosum Vaccinium 0.235 0.143 0.694 0.306 absent 0.719 membranaceum Vaccinium 0.353 0.000 1.000 0.000 absent 0.134 myrtillus Vaccinium 0.235 0.286 0.536 0.464 present 1.000 scoparium Valeriana 0.235 0.571 0.270 0.730 present 0.109 sitchensis Veratrum viride 0.647 0.143 0.788 0.213 absent 0.067 Verbascum 0.235 0.286 0.452 0.548 present 0.905 thapsus Veronica 0.000 0.143 0.000 1.000 present 0.289 serpyllifolia Veronica 0.000 0.143 0.000 1.000 present 0.284 wormskjoldii Veronica 0.000 0.143 0.000 1.000 present 0.287 wyomingensis Vibernum edule 0.059 0.000 1.000 0.000 absent 1.000 Vicia americana 0.529 0.429 0.563 0.437 absent 0.812 Viola adunca 0.412 0.571 0.500 0.500 present 0.709 Viola canadensis 0.294 0.143 0.641 0.359 absent 0.632 Viola glabella 0.059 0.143 0.121 0.879 present 0.291 Viola orbiculata 0.529 0.286 0.673 0.327 absent 0.283 Xerophyllum 0.059 0.000 1.000 0.000 absent 1.000 tenax Zizia aptera 0.000 0.143 0.000 1.000 present 0.292 147