Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Plant succession in the High Arctic : patterns & mechanisms O'Kane, Katriina 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_september_okane_katriina.pdf [ 34.6MB ]
Metadata
JSON: 24-1.0365968.json
JSON-LD: 24-1.0365968-ld.json
RDF/XML (Pretty): 24-1.0365968-rdf.xml
RDF/JSON: 24-1.0365968-rdf.json
Turtle: 24-1.0365968-turtle.txt
N-Triples: 24-1.0365968-rdf-ntriples.txt
Original Record: 24-1.0365968-source.json
Full Text
24-1.0365968-fulltext.txt
Citation
24-1.0365968.ris

Full Text

PLANT SUCCESSION IN THE HIGH ARCTIC:  PATTERNS & MECHANISMS by KATRIINA O’KANE Hons. B.Sc., Queen’s University, 2013 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Geography) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2018 © Katriina O’Kane, 2018 The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis entitled: Plant Succession in the High Arctic: Patterns and Mechanisms submitted by Katriina O’Kane in partial fulfillment of the requirements for the degree of Mater of Science in Geography Examining Committee: Greg Henry, Geography Supervisor  Brian Klinkenberg, Geography Supervisory Committee Member  !iiABSTRACT Succession is defined as species change over time, and investigations into its nature over the past century have shown that it is a highly variable process, dependent on local environmental conditions and species pools. The High Arctic is a landscape currently experiencing rapid change, and the response of ecosystems to certain changes can be better predicted by understanding succession. However, little research has been conducted about succession in High Arctic environments. Consequently, in this thesis we investigate the patterns of and mechanisms behind plant succession at and around Alexandra Fiord, Ellesmere Island, in the Canadian High Arctic. In Chapter 2 we resurvey three glacial forelands originally surveyed 21 years ago to investigate patterns of primary succession. We find that species advance predictably in a directional manner towards the retreating glacial margin, however the rates and patterns of change are unique to each species, and species do not behave as well define communities. However, distinctly different species dominated on successional areas compared to older mature areas. In Chapter 3 we survey a topographically heterogeneous foreland to investigate the mechanisms driving primary succession. We find that micro-environmental influences played the most important role. Variation in substrate grain size explained the largest amount of variation in vegetation patterns. Other important influences included facilitation from moss and other vascular plant species, time since deglaciation, intrinsic life history traits, and distance to a seed source. In Chapter 4 we visit a site of secondary succession that has been recovering for 31 years. We find that this site was far more advanced in its recovery than the glacial forelands, but it had not yet reached a community composition similar to that of the surrounding mature tundra. Many successional species were the same at both the site of secondary succession and the site of primary succession. This work has provided new insights into the patterns and mechanisms of succession in the High Arctic, emphasizing the importance of understanding succession at a local scale, and providing testable hypothesis for further research. 
!iiiLAY SUMMARY In this thesis we investigate how High Arctic ecosystems recover after disturbances, focusing on plants. Studying vegetation recovery on land released from a retreating glacier, we found that the patterns of what plants grow where were unique to each species, however the patterns of all species changed relative the amount of time the land has been recovering. Different species were more abundant on younger compared to older land. We also found that substrate type (for example large rocks versus fine sediment) and the presence of moss influenced why plants grow where they do. Studying the recovery of vegetation at a site that was subject to less severe human disturbance, we found that vegetation at this site recovered faster than on land near the glacier. The results highlight the variable responses of plants to disturbances, and emphasize the importance of understanding ecosystem recovery at a small local scale. !ivPREFACE This thesis is my original, unpublished work. Supervision and guidance for this research was provided by Dr. Greg Henry (University of British Columbia, Geography). The 1995 study described and discussed in Chapter 2 was designed and conducted by Glenda A. Jones and Dr. Greg Henry. The study plots and 1988 vegetation survey described in Chapter 4 were set up and conducted by Dr. Josef Svoboda and Dr. Greg Henry. I designed and conducted the 2016 survey in Chapter 2, the study in Chapter 3, and the 2016 survey in Chapter 4, with valued input from Dr. Greg Henry. I analyzed all data presented in this thesis, with valued help from Dr. Guillaume Blanchet in Chapter 3. Parts of chapters 2 and 3 were presented at the Arctic Change Conference (Quebec, QC, December 2017), the Canadian Society of Ecology and Evolution Meeting (Victoria, BC, May 2017), and at two ArcticNet Annual Scientific Meetings (Winnipeg, MB, December 2016; and, Vancouver, BC, December 2015). !vTABLE OF CONTENTSAbstract iii ...................................................................................................................................Lay Summary iv ..........................................................................................................................Preface v .......................................................................................................................................Table of Contents vi ....................................................................................................................List of Tables ix ...........................................................................................................................List of Figures x ...........................................................................................................................List of Abbreviations xiii ............................................................................................................Acknowledgements xiv ...............................................................................................................Dedication xv ...............................................................................................................................Introduction 1 ..............................................................................................................................Chapter 1 - Literature Review 4 ................................................................................................1.1 Introduction 5 ......................................................................................................................1.1.1 Changing climate in the Arctic 5 .................................................................................1.1.2 Economic Development in the Canadian Arctic 6 ......................................................1.2 Succession 8 ........................................................................................................................1.2.1 Succession Theory and History 8 ................................................................................1.2.2 The Arctic Environment  11 ........................................................................................1.2.3 Primary Succession in the Arctic and sub-Arctic 13 ...................................................1.2.4 Secondary Succession in the Arctic and sub-Arctic 18 ...............................................1.2.5 Spatial Analysis in Arctic and sub-Arctic Succession 20 ............................................1.2.6 Ecological Debates Concerning Succession 21 ..........................................................1.3 Alexandra Fiord 24 .............................................................................................................1.3.1 Primary Succession on Twin Glacier Foreland 24 ......................................................1.3.2 Secondary Succession at Alexandra Fiord  25 ............................................................1.4 Conclusions 27 ....................................................................................................................Chapter 2 - Patterns at Three Glacial Forelands: Repeat Survey after 21 years 28 ............2.1 Summary 29 ........................................................................................................................2.2 Introduction 30 ....................................................................................................................2.3 Methods 34 .........................................................................................................................!vi2.3.1 Study Sites 34 ..............................................................................................................2.3.2 Methods 39 ..................................................................................................................2.4 Results 47 ............................................................................................................................2.4.1 Twin Glacier 47 ...........................................................................................................2.4.2 Beistad Fiord 53 ..........................................................................................................2.4.3 Sverdrup Pass 55 .........................................................................................................2.5 Discussion 61 ......................................................................................................................2.5.1 Successional patterns 61 .............................................................................................2.5.2 Species patterns 63 ......................................................................................................2.5.3 Chronosequence 68 .....................................................................................................2.5.4 Future Research 69 ......................................................................................................2.6 Conclusions 70 ....................................................................................................................Chapter 3 - Mechanisms at Twin Glacier Foreland: Effects of Topographic Heterogeneity on Succession 71 ..........................................................................................................................3.1 Summary 72 ........................................................................................................................3.2 Introduction 73 ....................................................................................................................3.3 Methods 76 .........................................................................................................................3.3.1 Study Site 76 ...............................................................................................................3.3.2 Vegetation Survey 77 ..................................................................................................3.3.3 Environmental Variables 80 ........................................................................................3.3.4 Vegetation Analysis  84 ...............................................................................................3.4 Results 89 ............................................................................................................................3.4.1 Environmental Variables 89 ........................................................................................3.4.2 Vegetation Maps 93 .....................................................................................................3.4.3 Hierarchical Modelling of Species Communities 101 ................................................3.5 Discussion 109 ....................................................................................................................3.5.1 Mechanisms influencing succession 109 ....................................................................3.5.2 Limitation of the study 112 .........................................................................................3.5.3 Future Research 113 ....................................................................................................3.6 Conclusions 115 ..................................................................................................................Chapter 4 - Secondary Succession at Alexandra Fiord 116 ....................................................!vii4.1 Introduction 117 ..................................................................................................................4.2 Methods 119 ........................................................................................................................4.2.1 Study Site 119 .............................................................................................................4.2.2 Vegetation and Soil Survey 119 ..................................................................................4.2.3 Data analysis 123 ........................................................................................................4.3 Results 124 ..........................................................................................................................4.3.1 Disturbed Farm Area vs. Mature Tundra 124 ..............................................................4.3.2 Secondary Succession vs. Primary Succession 128 ....................................................4.4 Discussion 130 ....................................................................................................................4.5 Conclusion 132 ...................................................................................................................Conclusion 133 ............................................................................................................................Bibliography 136 .........................................................................................................................Appendices 150 ............................................................................................................................Appendix A: Relocating Chapter 2 Study Area 151 .................................................................Appendix B: Rare Species 156 .................................................................................................Appendix C: Historical Data from the Farm 161.....................................................................!viiiLIST OF TABLES Table 2.1. PERMANOVA results, significance of the change in species composition.................50 Table 3.1. Description of foreland sections...................................................................................77  Table 3.2. Results of HMSC variance partitioning for the six most common species................103 Table 4.1. PERMANOVA results, significance of the difference in vegetation between the disturbed farm area and the surrounding undisturbed tundra......................................................124 Table 4.2. Soil properties in disturbed farm plots and in the surrounding mature tundra...........128 !ixLIST OF FIGURES Chapter 2Figure 2.1. Map showing surveyed sites.......................................................................................34 Figure 2.2. Photographs of Twin Glacier and the adjacent survey area........................................35 Figure 2.3. Aerial photograph of Twin Glacier, showing the survey area, glacial retreat, and lichen trimline................................................................................................................................36 Figure 2.4. Photographs of Beistad Glacier and the adjacent survey area....................................37 Figure 2.5. Photographs of Teardrop Glacier, Sverdrup Pass and the adjacent survey area.........38 Figure 2.6. Aerial view of Twin Glacier survey area overlain with grid of zones surveyed and historical positions of the glacial margin ......................................................................................39 Figure 2.7. Survey design at Twin Glacier....................................................................................40 Figure 2.8. Transect layout on the forelands of Beistad Glacier and Teardrop Glacier, Sverdrup Pass................................................................................................................................................42 Figure 2.9. NMDS visualizations representing plant cover across belts in 1995 and in 2016.....45 Figure 2.10. Total percent cover and total species richness across the foreland in both 1995 and 2016................................................................................................................................................48 Figure 2.11. Shannon’s Diversity in both 1995 and 2016, and Jaccard’s Similarity between 1995 an 2016...........................................................................................................................................49 Figure 2.12. Statistically significant indicator species in 1995 and in 2016.................................51 !xFigure 2.13. Relative frequency of vascular plant species across Twin Glacier foreland in 2016................................................................................................................................................54 Figure 2.14. Summary of vegetation presence/absence at Beistad Fiord.....................................56 Figure 2.15. Summary of vegetation presence/absence at Teardrop Glacier, Sverdrup Pass...........................................................................................................................................57-59 Chapter 3Figure 3.1. Photograph of the topographically heterogeneous foreland at Twin Glacier..............76 Figure 3.2. Study area and sections, locations of 331 survey points, 56 temperature sensors, and 34 locations of soils soil samples and nutrient measurements.......................................................78 Figure 3.3. Image of the sampling grid surveyed at each point....................................................79 Figure 3.4. Series of maps showing presence and abundance of substrate classes and relict vegetation.......................................................................................................................................90 Figure 3.5. Series of maps showing visualizations of distance to the glacier, slope angle, slope aspect, and solar radiation..............................................................................................................91 Figure 3.6. Series of maps showing available soil nutrients, including nitrate, ammonium, phosphorous, as well as organic matter, total carbon, and soil moisture.......................................92 Figure 3.7. Interpolated predictions from temperature, soil moisture, and soil organic matter models............................................................................................................................................94 Figure 3.8. Shannon’s diversity and species richness across the foreland....................................96 !xiFigure 3.9. Presence and abundance of vascular plant species and moss across the foreland...................................................................................................................................97-100 Figure 3.10. The results of the HMSC variance partitioning......................................................102 Figure 3.11. Species-to-species association matrix showing the residual correlation at the scale of individual points......................................................................................................................105 Figure 3.12. Species-to-species association matrix showing residual correlation at the scale of sections.........................................................................................................................................106 Figure 3.13. Species-to-species association matrix showing residual correlation attributed to spatial autocorrelation..................................................................................................................107 Chapter 4Figure 4.1. Photographs of the ’Green Igloos’ experimental farm in 1982 and the revegetation experiment established thereafter in 1985...................................................................................120 Figure 4.2. Aerial photograph of the Alexandra Fiord glacial valley in 2016 showing the location of the experimental farming site, and two on-the-ground photographs of the study site............121 Figure 4.3. The arrangement of the plots and soil sample locations surveyed in 2016..............122 Figure 4.4. The average standardized percent cover vegetation, substrate, standing dead, and litter on Disturbed Farm and Mature Tundra plots, and on Belts 80 and 90 at Twin Glacier Foreland.......................................................................................................................................125 Figure 4.5. The average standardized percent cover of vegetation on Disturbed Farm and Mature Tundra plots, and on Belts 80 and 90 at Twin Glacier Foreland..................................................126 
!xiiLIST OF ABBREVIATIONS AIC - Akaike’s information criterion Ca - Calcium  cm -  Centimetre DEM - Digital elevation model  GIS - Geographic information system GPS - Global positioning system ha - Hectare HMSC - Hierarchical Modelling of Species Communities IDW - Inverse distance weighted InSp - Indicator species  K - Potassium  km - Kilometre LIA - Little ice age LOESS - Local regression m - Metre Mg - Magnesium  NH4-N - Ammonium  NMDS - Non-metric multidimensional scaling NO3-N - Nitrate  OM - Organic matter PERMANOVA - Permutational multivariate analysis of variance spp. - Species TWI - Topographic wetness index 
!xiiiACKNOWLEDGEMENTS I am extremely grateful to have had to opportunity to work on this project. For this opportunity I thank Greg Henry, who over 30 years has cultivated the conditions for a very interesting study about succession at a very beautiful location. I also thank Greg for his supervision, advise, positive encouragement, and support throughout the project. Thank you to Guillaume Blanchet, who introduced me to some very interesting statistical methods for my data analysis in Chapter 3, and who took the time to explain HMSC and answer many of my questions. Thank you also to Esther Frei, Brian Klinkenberg, and Jennifer Williams for their valuable feedback and advise throughout the project. A big thank you to my three outstanding field assistants: to Silly Seal, whose big heart and team spirit made camp feel like home, and whose friendship provided much needed support throughout the process; to Cassandra Elphinstone, whose constant stream of ideas have positively contributed to many aspects of the project, and whose company in philosophical conversations have advanced my ideas about the nature of science; and to Katie MacIntosh, whose humour and puns provided endless laughs and kept everyone’s spirits high. Thank you to Geoffrey Boulangé for his support through two cross-continental years and two satellite telephone summers, and for his patience, care, ideas and encouragement. And thank you to many other friends and family members for the support and joy they have provided.  Thank you to the citizens of Canada and of Quebec, whose tax dollars have made this project possible financially. Thank you to the W. Garfield Weston Foundation and to the Koerner Foundation for their financial support which has allowed me to dedicate more time to this project. Thank you to the dedicated staff at the Polar Continental Shelf Project who took care of our safety and well being in the field, and to the Royal Canadian Mounted Police for their logistical support. And finally, thank you to Inuit of Nunavut, and specifically to the HTO offices in Resolute and Grise Fiord who gave us permission to conduct this research on their beautiful land. 
!xivDEDICATION To the Arctic wind. !xvINTRODUCTION “It is this process of recovery of ecosystems after disturbance that provides the clean air and water and fertile soils that humans and all organisms need to survive.”       - Walker & del Moral, 2003 !1Succession is how an ecosystem rebuilds itself. It is how barren land becomes rich with life: with plants and with animals. Models of succession can be used to predict how landscapes will change with a warming climate. They can be used in revegetation initiatives, to develop successful and ecologically appropriate strategies. It is both a predictive and an organizational tool - albeit one that is riddled with contingencies. Despite a century of study, few generalizations have arisen about the patterns and mechanisms of succession. Observational and experimental evidence point to the site specific nature of succession. The patterns of species change over time depend on the intrinsic biotic characteristics of the plants in the area, as well as on extrinsic environmental characteristic that play out on various spatial scales. At larger scales, differences in geology, topography, and climate influence which species are adapted to live in the area, while at smaller scales differences in for example moisture or exposure define which species have an advantage. Because patterns of succession are contingent on so many different factors, a local understanding is necessary.  The High Arctic environment is experiencing rapid environmental changes as a result of amplified global warming. Yet very few studies have been conducted about succession in the High Arctic (reviewed in Chapters 1 and 2). To help researcher understand and predict future changes to High Arctic ecosystems, a better understanding of plant succession is useful. In the thesis we investigate succession at and around Alexandra Fiord, on Ellesmere Island, Nunavut. We set out to better understand the patterns and mechanisms of both primary and secondary succession. We use the term patterns of succession to describe the “what” of succession, i.e. the sequence in and rate at which the abundances and identities of  different plant species change across a disturbed landscape over time. We use the term mechanism of succession to describe the “why” of succession, i.e. why does vegetation change over time, what biotic and abiotic processes act to produce changes in the patterns observed during succession.  !2In Chapter 1 we review literature pertinent to this project. We examine the historical theoretical and experimental literature that has attempted to describe succession, and focus on work that has been done in Arctic and sub-Arctic environments. In Chapter 2 we revisit three High Arctic glacial forelands that were first surveyed 21 years ago. We investigate the patterns of succession, and test the validity of using the chronosequence approach applied in these surveys. In Chapter 3 we survey a topographically heterogeneous glacial foreland to investigate the mechanisms of succession. In Chapter 4 we survey a site of secondary succession, and compare the patterns to those of the surrounding mature tundra, as well as to the patterns of primary succession observed in Chapter 2. !3CHAPTER 1 - LITERATURE REVIEW Literature Review !41.1 Introduction This literature review summarizes research concerning various aspects of plant succession in Arctic environments. We begin by introducing two important contemporary changes influencing the Arctic environment (climate change and economic development), and by describing how knowledge of successional patterns and mechanisms can help us manage these changes. In the second section we review literature relating to succession. We focus first on historical theoretical literature, before turning our  attention to describing the Arctic environment and to summarizing studies about both primary and secondary succession in Arctic and sub-Arctic environments. We spend some time describing two studies that investigating the spatial dynamics of vegetation across successional landscapes. We then summarize two recent debates (niche versus neutral models, and the use of chronosequences) which can influence the design of succession studies. In the the final section, we summarize research that has been conducted about succession around our  study area at Alexandra Fiord, Ellesmere Island. 1.1.1 Changing climate in the ArcticSuccessional processes are responsible for having re-vegetating the Arctic since the end of the last ice age, and will continue to play an important role in the coming decades as the physical landscape is altered due to the changing climate. Land ice in the form of glaciers and ice caps cover over 400,000 km2 of the Arctic worldwide (Sharp et al., 2015), and the dynamics of this ice has an impact on tundra vegetation. When glaciers expand, plant communities along their termini regress, and when glaciers retreat, the emerged forelands experience succession.   The dynamics of glaciers are governed by their mass balance, which in turn depends on long-term trends in temperature and precipitation (i.e. climate). Gardner et al. (2011) link recent mass loss of glaciers in the Canadian Arctic Archipelago directly to warmer surface air temperatures in !5the summer. However, memory held in the flow of the glacier creates a time lag where the terminus position may not reflect the current mass balance. Dowdeswell (1995) suggests that this timescale for adjustment to changing mass balance is in the order of 100 years for many Arctic glaciers. This is further complicated by the effect of glacier size on terminus response, where larger glaciers have longer lag times.   Glaciers have undergone many cycles of expansion and retreat since the end of the last ice age due to the changing climate, and are today almost ubiquitously retreating. There is consensus about a warm period in the early to mid Holocene in the Arctic, although its timing is inconsistent and regionally variable. The warm period is believed to have been driven by solar insolation, but affected locally by ice sheets (Gajewski & Atkinson, 2003). Following the warm period temperatures have shown a cooling trend, although with high interdecadal variability, and punctuated by a Medieval Warm Period around 1000 A.D. (Dhal-Jenson et al., 1998). The cooling trend culminated in the Little Ice Age, which lasted until the mid 19th century. Since then, temperatures have been increasing, and much of the warming since has been attributed to the anthropogenic release of greenhouse gases (Overpeck et al., 1997; Crowley, 2000; Mann et al., 1998).    This recent warming has led to glacial retreat and thinning (Gardner et al., 2011), thawing of permafrost (Overpeck et al., 1997), declining annual extent of snow cover (Post et al., 2009), and expansion of species distributions (Post et al., 2009). Vegetation and ecosystem response to each of these consequences will be governed by successional processes, and models of succession can aid in predicting future tundra ecosystem dynamics. 1.1.2 Economic Development in the Canadian Arctic Anthropogenic influences in the Arctic extend beyond the recent impacts on climate, and include direct effects on the environment due to mineral resource exploration and extraction, and the !6development of infrastructure. The Canadian Northern Economic Development Agency’s (2013) Strategic Framework for 2013 to 2018 emphasizes the importance of resource development in stimulating economic development in the North. It states that, in 2013, there were 160 active exploration projects underway in the three northern territories, and 28 of those were in advanced environmental assessment and permitting phases. That is a large number considering there were only eight operational mines in the three territories in 2015 (Department of Energy, Mines and Resources, Government of Yukon, 2015; NWT & Nunavut Chamber of Mines, 2015). Both active and exploration phases of the mining process cause disturbances, and lead to primary or secondary succession depending on the severity of the disturbance. However, revegetation treatments can speed up succession. Designing ecologically appropriate revegetation treatments depends on a local understanding of successional patterns and mechanisms. !71.2 Succession 1.2.1 Succession Theory and History The most simple and general definition of succession is species change over time (Walker & del Moral, 2003). The evolution of succession theory over the past 100 years calls for such a general definition, as succession has proved to be such a complicated and unpredictable process that it cannot be easily conceptualized. None-the-less, many authors have made important contributions to succession theory, and each contribution has helped develop a more wholesome understanding. Frederic Clements (1916) presented the first comprehensive theory of plant succession. He viewed succession as a series of predictable seres (vegetation units) that converge towards a climax formation defined by the climactic conditions of the region. He compared the climax formation to an organism, which not only arises, grows, matures, and dies, but also reproduces itself, repeating with fidelity its stages of development. Each successive vegetational sere modified the environment, facilitating the development of the next sere. His model of succession was so romantically orderly that ecologists studying unruly ecosystems clung to it for the first half of the 20th century. Despite early warnings from Gleason (1917) that individualistic and stochastic processes played important roles, it wasn't until much later that experimental and observational evidence finally forced ecologists to abandon Clements' theory. Since the mid-1970s, there has been a shift towards reductionist and mechanistic approaches, emphasizing the importance of site-specific differences (Glenn-Lewin et al., 1992). There has also been a move away from the climax concept (Glenn-Lewin et al., 1992). !8One of the most important papers that helped redefine succession after this shift was by Connell and Slayter (1977). They suggested two alternative modes of interaction between species, other than facilitation as suggested by Clements. While facilitation assumes that earlier plants have a net positive effect on later ones, tolerance assumes a net zero effect, and inhibition assumes a net negative effect. Facilitation occurs when, for example, early colonists build up the soil nutrient pool or retain moisture, helping later colonists grow and develop. Tolerance occurs when early colonizers neither increase nor reduce the rates of recruitment and growth of later colonists, which simply arrive later or grow slower. Inhibition results when early colonists out-compete later colonists for resources, and are only replaced after they are damaged or killed. In a re-evaluation of their paper ten years later (Connell et al., 1987), they redefined the three modes as representing extremes of a continuum of effects that earlier species can have on later ones. They also stated that it is possible to have more than one mode operating within the same successional sequence (e.g. a shrub could both add nutrients to the soil, and shade out other species). However, they emphasize that at each stage there is one dominant net effect, which could still be defined within the bounds of their model. This new way of thinking presented by Connell and Slayter was important, as it emphasized the possibility of both multiple mechanisms and multiple pathways, providing researcher with new testable hypotheses about species interactions. Connell and Slayter (1977) also touched on the concept of stability, which relates to the Clementsian climax, another topic of controversy in succession theory. Clements originally defined climax as the point when “the occupation and reaction of a dominant are such as to exclude the invasion of another dominant” (Clements, 1916, p.105). Clements (1916) considered this point as being stable and thus unchanging, unless there was a large variation of climate or the development of a new flora. Contemporary theoretical ecologists are more hesitant to declare the existence of a climax, or even confirm the stability of a seemingly dominant vegetation community. Regular disturbances that are stochastic in nature can redirect successional trajectories, and result in a landscape mosaic of patches that are each at a different stage of successional development (Pickett & White, 1985, quoted in Walker & del Moral, 2003; Moreau et al., 2008). Raup (1981), who conducted his research in the Arctic, suggested that succession is !9simply a process of continual recovery from disturbance, rather than a progress towards a climax (quoted in Walker & del Moral, 2003). If, however, a community can persist without major interruption, then it reaches a stage where on a small scale individuals die and are replaced. Connell and Slayter's hypothesis about stability suggested that if individuals are replaced with other individuals of the same species, the community is considered stable. If they are replaced by different species, the community is unstable (Connell & Slayter, 1977). Despite the debate in theoretical ecology about climax and stability, the climax concept is still widely used in fields such as forestry and wildlife management (Walker & del Moral, 2003). These applications are convenient, but may misrepresent the true dynamics of ecosystems. Other important additions to successional theory include Egler’s initial floristic composition concept and Tilman’s (1985) resource-ratio hypothesis. Studying secondary succession in abandoned fields, Egler (1954) concluded all species involved in succession were capable of establishing at the beginning, and that dominance was a function of life history traits such as growth rates and height at maturity. This view emphasizes the importance of intrinsic species traits during succession, an idea which has since been widely adopted. Tilman's (1985) resource-ratio hypothesis explains succession as adaptations of different species to different limiting resources (most importantly soil resources and light). As resources change over time, so do the species that are most successful. These additions, together with many others, have helped develop a more rounded understanding of succession. So, where does that leave succession theory today? After 100 years, we seem further away today from achieving the goal of a model for predicting species change over time. After abandoning Clements' theory, ecologists have acknowledged that there are multiple modes, such as competition and facilitation, that can lead succession through different pathways of species replacement. Individual species life history traits, and abiotic factors such as resource availability, exert significant influences on species composition. The most important conclusion that can be made about our current understanding of succession is that it is highly variable, and thus, site specific. While no overarching theory of succession may ever be achieved, !10understanding the relative importance of different mechanisms at different sites can help develop local predictive models of succession. 1.2.2 The Arctic Environment  The first step to understanding local mechanisms in succession is understanding the local environment. The Arctic is a region of harsh environmental conditions, and thus low biological diversity. Only species that have evolved unique adaptations and are highly stress tolerant have been able to develop viable populations. Various environmental factors contribute to making the Arctic an extreme habitat, mostly relating to the low amounts of energy available. The soil is poor in nutrients and organic matter, because low temperatures limit bacterial activity and decomposition rates (Bliss, 1962). Light intensity is reduced, although much of the Arctic receives sunlight for 24 hours a day during the growing season. Low temperatures limit plant growth by slowing metabolic processes (Bliss, 1962). However, a gradient exists, where temperatures near the ground can be higher by a several degrees compared to air temperatures. Other consequences of low temperatures include the potential for brief periods of freezing temperatures during the summer, and shorter growing seasons, lasting from May or early June to August. There is also very little summer precipitation in parts of the Arctic, thus plants are often reliant on the moisture provided by snowmelt. As each of these environmental factors push the limits of the life they support, micro-environments become increasingly important (Billings & Mooney, 1968). Micro-topographic differences of even a few centimetres can make a significant difference in temperature, wind, or moisture. Despite these unfavourable environmental conditions, a collection of plants have evolved strategies to persist in the Arctic. The most obvious adaptation is their reduced height to take advantage of the higher temperatures near the ground surface. Many plants will often form mats no higher than six to eight centimetres (Billings & Mooney, 1968). Such mats create warmer !11micro-environments that permit faster metabolism (Bliss, 1962). For example, Wilson (1957) measured the temperature in a Saxifraga oppositifolia mat near Resolute in the summer. While air temperature was only 2.5oC, the temperature just beneath the surface of the Saxifraga mat was 12.5oC. This effect is further magnified in plants that have developed hollow stems, where temperatures within the cavity may be as much as 20oC above external air temperatures (Billings & Mooney, 1968). These plants are then able to carry out internal photosynthesis with these high temperatures, using recycled carbon dioxide (Billings & Mooney, 1968).  Even when not taking advantage of warmer micro-environments, Arctic plants have a metabolic system that can capture, store, and utilize energy at lower temperatures (Billings & Mooney, 1968). They are also extremely stress tolerant, being able to cope with both frost and desiccation throughout the growing season (Bliss, 1962). Another major adaptation of Arctic plants is their rapid growth after the melting of snow-cover in the spring. Most plants are perennials, and store energy in their roots, rhizomes, and stems throughout the winter. At the breaking of dormancy, that energy is used within a few weeks until the shoots are 75-90% complete (Billings & Mooney, 1968). Almost all perennials also have pre-formed flower buds, allowing for rapid flowering (Billings & Mooney, 1968). These adaptations imply that growth is dependent not only on current growing season temperatures and conditions, but also on conditions in the previous year, when energy stores were being built and flower buds were being formed. In addition to growth adaptations, Arctic plants have also developed several reproductive strategies that enable them to persist in their harsh environment. Many plants rely on asexual reproduction, through rhizomes or bulbils, which shortens the steps necessary and requires less energy (Billings & Mooney, 1968). Nonetheless, sexual reproduction is still prolific. Following rapid flower growth, many plants are able to self-pollinate in the absence of abundant pollinators (Bliss, 1962). Whether or not seeds ripen and germinate thereafter depends upon if the weather remains favourable. Plants are adapted to take advantage of good years, being able to produce considerable amounts of seeds in one year, even if that is followed by many years when few, if any, seeds are produced (Bliss, 1962). Seed dormancy is also common, and dormant seeds can !12stay viable for many years though bad conditions, to finally germinate in favourable years (Billings & Mooney, 1968).   Despite these adaptations, the Arctic remains a marginal environment where plants are challenged by their abiotic environment. Theories of succession should reflect these abiotic factors. Svoboda and Henry (1987) proposed a model where succession is defined as a function of biological driving forces and environmental resistance. Biological driving forces are the intrinsic characteristics of invading and establishing species, for example genetics, germination capacity, or ability to survive adverse conditions. Environmental resistance, on the other hand, represents the sum of the extrinsic hindering abiotic and biotic factors, such as low temperatures, drought, or predation. They show how this model can be used to describe three types of succession. Directional-replacement occurs in low resistance environments, where succession proceeds similar to lower latitudes. Directional-nonreplacement occurs in high resistance environments, when invading species succeed in slowly expanding, without replacing earlier species. Nondirectional-nonreplacement occurs in the most extreme environments, where very few species succeed at establishing, and continuously fluctuate in their cover, while other species invade but fail to establish permanently.  The unique environment and plants of the Arctic highlight the need for a local model of succession. Past research about succession in Arctic and sub-Arctic environments can provide us with insights into patterns and mechanisms that operate on a regional scale. 1.2.3 Primary Succession in the Arctic and sub-Arctic The first successional studies in a Northern environment were carried out at Glacier Bay, Alaska, where a young William S. Cooper (1923) began surveying vegetation and setting up permanent quadrats in 1916. The opportunity presented by glacial foreland chronosequences to study species change over known time intervals was just beginning to be realized. Since then, much !13has been learned. Most of the research has concentrated on glacial forelands, since they are many in cold polar regions. However, research has also been conducted on beaches following isostatic rebound (Bliss & Gold, 1994), recently emerged volcanic islands (Magnusson et al., 2014), and floodplains (Walker et al., 1986). First, a definition of primary and secondary succession is useful. Primary succession is a process of species change over time that occurs on barren surfaces where severe disturbances have removed almost all traces of biological activity. There is no biological legacy, no surviving plants, animals, soil microbes, seeds, or organic soil. This contrasts with secondary succession, where the disturbance has been less severe, and where there are surviving biological agents and soil that speed up and direct succession. Rather than two separate modes, primary and secondary succession should be seen as points along a continuum (Walker & del Moral, 2003). Glacier Bay has continued to be an important site for research about primary succession, and its legacy of research has helped reach important conclusions. Chapin et al. (1994) conducted a study that explained the interaction of multiple mechanisms behind succession at Glacier Bay. They initially set out to test the hypothesis that the major effect of  nitrogen fixing colonizers (including Dryas drummondii and Alnus sinuata) was to facilitate establishment of late-successional dominants, but found that this facilitating effect was of minor importance. Instead they concluded that the life history traits of plants determined the pattern of succession. Early successional species (Chamerion latifolium and Dryas drummondii) had smaller seeds, a younger age of first reproduction, shorter life-spans, a shorter height at maturity, and were shade intolerant. Mid-successional species (Alnus sinuata) had less success in getting their seeds to areas newly released by the glacier, but grew taller and eventually shaded out early species. Late successional species (Picea sitchensis) also arrived in negligible numbers during the pioneer stage, but grew slowly, lived longer, and eventually out-competed Alnus sinuata with their height at maturity. Walker et al. (1986) found similar results from their research on an Alaskan floodplain, where seed dispersal, growth rates, and height at maturity played key roles in determining the successional sequence. Chapin et al. (1994) further concluded at Glacier Bay !14that simultaneously facilitative and inhibitory effects of plants on each other also direct succession. Fixation of nitrogen and addition of organic matter from the Dryas drummondii and Alnus sinuata stages facilitate the growth of Picea sitchensis by providing essential nutrients. In fact, Lawrence et al. (1967) estimated that the presence of Dryas drummondii sped up the rate of succession to a fully developed forest by 20 to 30 years at Glacier Bay. Other facilitative effects of vegetation include increasing soil temperatures and decreasing wind speeds (Cutler, 2011). However, all plants also compete with each other for soil resources and light. Both Dryas drummondii and Alnus sinuata had a net inhibitory effect on initial establishment (nearly 100%) of Alnus sinuata and Picea sitchensis, respectively. This supports Egler's (1954) initial floristic composition theory, that new species must disperse to sites before initial colonizers modify the environment for establishment. While the results from this study give much insight into succession at Glacier Bay, they have also been criticized for their application of a potentially erroneous chronosequences (Walker et al., 2010; Fastie, 1995). Seed dispersal and seedling success have also been discussed as important factors in Arctic succession. Elven and Ryvarden (1975) studied a glacier foreland in Norway, and found that 87% of diaspores came from a distance of less than five meters. They also found that smooth and heavy diaspores (lacking any apparent morphological adaptation to dispersal) had dispersed several hundred meters. This agrees with results from Schwienbacher et al. (2012) who found that seed mass did not correspond to position on the foreland. However, it contradicts others who have found that pioneer species do have diaspores well adapted to wind dispersal (Stöcklin & Bäumler, 1996; Chapin et al., 1994; Walker et al., 1986). Others studying glacial forelands have equally reported that colonization is constrained by seed availability (Erschbamer et al., 2001; Fastie, 1995; Jones, 1997). Germination was found by Schwienbacker et al. (2012) to respond to temperature, where pioneer species performed better at colder temperatures than late succession species. Seedling mortality is equally limiting, and is due to either drought (Bell & Bliss, 1980) or frost (Marcante et al., 2012). Bell and Bliss (1980) showed that low germination (between 0 and 40% for different species) and high annual seedling mortality (about 50%) played a major role in restricting plant cover development in High Arctic polar deserts and semi-deserts. They !15found large differences in success between two years of different conditions, which led them to suggest that occasional good years are what drive development, which lags in the mostly bad years. They further argued that reproduction by seed is a rare event. In addition to life history traits, species interactions, and seedling success, local abiotic differences play important roles in determining successional development. Bliss and Gold (1994) studied a costal topographic gradient at Truelove Lowland, Devon Island, that represented stages of primary succession due to isostatic rebound. They concluded that availability of fresh water and the length of time that sites remained wet played important roles in the build up of soil and the kinds of species occupying a site. Cutler (2011) investigated vegetation-environment interactions on a chronosequence of lava flows in Iceland, and stressed the importance of differences in micro-topography. Vegetation on hillocks persisted as a thick, well-developed moss layer, while vascular species became dominant in hollows. Several authors have stressed the importance of ‘safe sites’ in seedling establishment (Harper et al., 1965). ‘Safe sites’ have beed described as concave surfaces, proximity to large rocks, coarse substrate, or vicinity of other plants (Jumpponen et al., 1999; Erschbamer et al., 2001). Matthews (1992) has spent his career researching forelands primarily in Norway, and wrote a book summarizing his knowledge about glacial foreland succession. He summarized the factors affecting primary succession on a glacial foreland in a conceptual equation: v =  f (t, cl, o, r, p …) Where vegetation (v) is a function of primarily terrain age (t), and then of other environmental factors such as climate (cl), organisms (o), relief (r) and parent material (p). Climate defines the temperature, moisture, snow distribution, and topography. Organisms such as herbivores can limit successional development, while others such as soil microbes fixing nutrients can aid development. Relief in terms of altitude, aspect and micro-topography determine shading, protection and resource access. And parent material includes the texture and lithology of the substrate, which impacts soil chemistry, water availability, and frost action.  !16In addition, Matthews (1992) summarized successional trends of plants on glacial forelands. Vegetation cover increases with terrain age, however patchiness develops due to spatial heterogeneity of the substrate and plant-plant interactions. Species richness also increases with terrain age, at least on younger ground. He notes that several investigations have detected a peak in richness. For example Dolezal et al. (2008) studied primary succession on a glacier foreland in the Russian Far East, and found that richness peaked at the 80 year old moraine, after which alder began to dominate in dense stands. Finally, Matthews (1992) notes that most studies recognize stages in succession, where pioneer groups differ substantially from later stages, with the intermediate stages being less well defined. He acknowledges that causes for these patterns are poorly understood, but suggests that initial composition may depend primarily on immigration and establishment, while subsequent changes involve competitive balance, differential arrival, longevity and growth rates, and environmental change. Together, biotic and abiotic factors influence succession, which can either converge along one pathways as it develops, or diverge to multiple pathways. A pathway describes the unique order of development of species and of other abiotic factors that takes place post-disturbance. Most studies of pathways during primary succession in the Arctic have found evidence of divergence to multiple pathways as opposed to convergence. Often succession proceeds though different stages of development due local topographic, substrate, or stochastic factors. Both Bliss and Gold (1994) and Cutler (2011) noted divergent trajectories due to their described differences in abiotic site conditions. Dolezal et al. (2008) found that the trajectory diverged after 25 years depending on the substrate texture and topographic position. Moraine crests with fine-grained substrate developed species poor communities dominated by alder and grasses, moraine flanks with coarse grained substrate developed species rich communities dominated by legumes and forbs, and depressions were dominated by willow and sedge communities. Robbins and Matthews (2010) surveyed 39 glacier forelands in south-central Norway, and found that rates of succession and mature communities changed with altitude. Sub-alpine (below 1000 m) vegetation transitioned from pioneer plants to birch woodland in about 70 years, intermediate altitudes developed to dwarf-shurb and snow-bed vegetation within 250 years, and high alpine !17(above 1600 m) vegetation persisted as pioneer herbs indefinitely. Fastie (1995) documented three different pathways of vegetation change at Glacier Bay, resulting in communities that were dominated by either Picea sitchensis, Alnus sinuata or Poplus trichocarpa. He attributed these differences largely to differential seed rain, with sites being at varying distances from particular parent species at time of deglaciation.  This literature on primary succession in Arctic and sub-Arctic environments highlights the importance of individual species characteristics and local abiotic and biotic factors in driving and determining successional patterns. Life history characteristics such as height at maturity and reproductive strategies, and environmental factors such as temperature, moisture availability, micro-topography, and parent material all have important influences on the diverging pathways of Arctic succession. The literature strongly confirms the fact that succession is highly variable across space and time. The need for local knowledge of successional patterns and mechanisms is apparent. 1.2.4 Secondary Succession in the Arctic and sub-ArcticWhile studies of primary succession dominate research on succession in the Arctic, studies of secondary succession are important when questions concerning restoration are involved. The largest factor in determining patterns of secondary succession is the extent and severity of the disturbance - how much of the plant cover is removed and over how large an area. If some plant cover remains, it is usually restricted to more resistant species, such as Salix arctica, Dryas integrifolia, or Saxifrage oppositifolia in the High Arctic (Babb & Bliss, 1974). Salix arctica was shown by Babb and Bliss (1974) to be the most resistant, requiring only the roots and several buds to remain intact for the individual plant to continue growing. These remaining plants direct succession thereafter. If no plant cover remains, other forms of biological legacy may still help speed up succession, such as rich soils or a seed bank. In such cases, rapid invaders in High Arctic secondary succession include graminoids, Bistorta vivipara, Saxifraga spp., Draba spp., !18and Papaver spp. (Babb & Bliss, 1974; Cannone et al., 2010). Besides plant cover, moisture is another factor that has been shown to affect secondary succession. Studies support the hypothesis that moisture availability aids in speeding up the process of succession (Cargill & Chapin, 1987; Forbes, 2001). Warmer summer temperatures may also reduce the time needed for recovery (Cannone et al., 2010). Many studies on secondary succession have focused on the effects of reclamation treatments, however results vary. Hagen et al. (2014) compared seeded and unseeded sites over 21 years in an alpine environment in Norway. They found that total vegetation was higher on seeded sites, but that unseeded sites had more native vegetation and higher species richness. Gretarsdottir et al. (2004) studied long term (20-45 years) effects of reclamation treatments in Iceland, including fertilization and seeding. They found that plant cover was higher on areas with reclamation treatment, and species were more diverse, and proposed that the seeded species - which had almost disappeared in later stages -  had facilitated the colonization of native plants. Cargill and Chapin (1987) suggested that whether or not to seed depends on the moisture regime of the site. Sowing exotic grass species in mesic disturbed sites where succession already progresses rapidly is likely to inhibit establishment of native species. However, sowing in xeric disturbed sites could facilitate establishment of native species, by increasing nutrient availability (Cargill & Chapin,1987). This variability in response again points to the importance of local environment in determining successional outcomes. These studies have advanced the understanding of secondary succession in the Arctic and sub-Arctic, by determining which species are the most resistant and which are rapid invaders, by showing the importance of moisture and of temperature, and by uncovering the variable success of reclamation treatments. !191.2.5 Spatial Analysis in Arctic and sub-Arctic SuccessionDue to the time consuming nature of vegetation sampling, research on succession is often limited to small areas spatially, which are expected to represent the larger patterns occurring. However, due to the heterogeneous and divergent nature of succession described in the previous sections, such a small sample may be ill-equipped to reveal larger patterns. For this reason, studies that map out species distributions over an entire landscape (for example, an entire glacial foreland) are useful. Matthews (1976) was the first to start mapping species over an entire foreland, producing a series of species abundance maps as part of his PhD. Before the days of GPS, he used aerial photographs to locate 638 sites across the entire Storbreen glacier foreland, at which he sampled local shoot frequency of each vascular plant in 16 m2 quadrats. He then used a computer to turn these into maps. He analyzed the maps to describe three distinct vegetation “cores”. Pioneer species arose on ground up to 50 years. Heath species were most abundant on older ground, and at a altitudes lower than 1200 m for north-facing slopes, or 1300 m for south-facing slopes. Snow-bed species occurred at higher altitude sites that had been deglaciated for a relatively long period of time, but were quite variable in their distribution depending on the species. The maps display the patchiness of species distribution. Thirty years later, technology had evolved and researchers were able to take advantage of satellite imagery and GPS to study species distribution. An excellent paper was written by Moreau et al. (2005) that described their application of remote sensing, GIS, and field sampling to create a vegetation map and perform a spatial analysis of the Midre Lovénbreen glacier foreland in Svalbard. Aerial photographs and satellite images were used to determine glacial retreat and identify runoff features. A DEM of the foreland was generated using 40,000 GPS points, which was further used to model slope, radiation, aspect, temperature, and winds. Field sampling was conducted to represent six micro-topographic relief forms identified from the GIS !20database. This produced a map, which again shows the patchiness of species distributions. Nonetheless, it still shows three concentric rings relating to three temporal periods. Abiotic variables deemed most important include runoff, terrain age, and moisture. Despite the importance of capturing the variability within heterogeneous successional landscapes, relatively few studies have surveyed at such a large scale. Modern research about succession could take advantage of satellite imagery, remote sensing, GPS and GIS technologies to better characterize forelands and arrive at more representative results. 1.2.6 Ecological Debates Concerning SuccessionHere we will focus on two debates in ecology: niche versus neutral models, and the use of chronosequences.  Niche versus Neutral Models A recent debate in ecology has centred around Hubbell’s (2001) Unified Neutral Theory of Biodiversity and Biogeography. This theory asserts that species are equivalent to each other in all important ecological respects, including probability of giving birth, dying, migrating, and speciating. The theory thus relies on stochastic processes such as ecological drift, random migration and random speciation to determine species distributions. Niche theory, on the other hand, has a long history originating from Charles Darwin. It asserts that species respond to environmental factors, for example resources, predators, or competitors. In doing so, the species carves out a “niche” of optimal existence.  Much of the recent discussion about niche versus neutral models has focused on the integration of the two seemingly opposing ideas. Chase (2014) emphasizes the importance of scale in determining whether niche or neutral processes dominate. He argues that at larger landscape !21scales, the importance of deterministic factors like environmental conditions and interspecific interactions dominate, implying niche processes. At smaller scales, there are fewer individuals and less habitat heterogeneity, and the importance of stochastic events increases, implying neutral processes. Gravel et al. (2006) suggests that niche and neutral theories are simply the extremes of a continuum, from competitive to stochastic exclusion. In an equilibrium community, they would expect to see both processes having affected species distributions to various degrees.  The Use of Chronosequences Another debate that has been highlighted recently is whether or not chronosequences are effective tools in studies of succession. A chronosequence is a set of sites formed on the same substrate, that differ only in time since they were formed (Walker et al., 2010). Vegetative succession is a process that can last decades to millennia, and so the use of this space-for-time substitution allows ecologists to uncover trends over time periods longer then their lifetimes. However, chronosequences rely on one critical assumption: all sites start with identical environmental conditions. As previously highlighted, micro-environmental factors are key in succession. Other factors could effect chronosequence interpretation, such as differences in substrate material, climatic history, past disturbances, or unequal input of propagules (Finegan, 1984).   These uncertainties have led Johnson and Miyanishi (2008) to argue for the discontinued use of unvalidated chronosequences in the study succession. In their paper, they provide evidence from several non-chronosequence methods that invalidate the chronosequences previously inferred. They use classic examples, including Cowles (1899) study of sand dune succession, Cooper’s (1923, 1931, 1939) study of succession at Glacier Bay, and others. They suggest that any future use of chronosequences must be validated using data from other methods, such as long-term plots, palynology, stand reconstruction, or remotely sensed imagery. !22Walker et al. (2010) replied two years later with a paper that argued chronosequences were “important and often necessary” in studying succession, but acknowledged there were certain instances where their use was inappropriate. They develop a framework that ranks different trajectories of successional development according to how appropriate chronosequences are for their study. Their paper outlines several other important considerations, such as disturbance frequency, time scale, purpose of study, and biodiversity of the site. It further recommends guidelines to be used for developing appropriate chronosequence studies. This research shows that chronosequences can still be valuable, and provides direction on about how best to use them. In this section we have described successional theory and what is known about succession in Arctic and sub-Arctic environments. One of the key conclusions is the variability of succession over space and time, and thus the need for local knowledge about the patterns and mechanisms of succession. In the last section we will focus on past research about succession conducted at our study site at Alexandra Fiord, Ellesmere Island, Nunavut 
!231.3 Alexandra Fiord Alexandra Fiord is a unique site in the Canadian High Arctic, in that it is one of the few places that has been studied for a relatively long time (30 years). Amongst many other projects, this research includes studies about succession, namely two studies conducted on the foreland of Twin Glacier, and one on a disturbed site further down in the valley. These studies provide valuable information about local patterns and mechanisms. 1.3.1 Primary Succession on Twin Glacier ForelandTwin Glacier foreland is one of the few forelands in the Canadian High Arctic where succession has been studied. In addition, the retreat of Twin Glacier has also been marked every year since 1980 (except from 1986-1991). Two studies have been conducted on plant succession on the foreland: by Jones (1997; Jones & Henry, 2003), and by Young (2002). Jones (1997) conducted her Master’s research on primary plant succession in the High Arctic. This consisted primarily of a detailed vegetation survey on a section of the foreland of Twin Glacier. She also collected information about environmental factors, including substrate, litter, soil characteristics, snow depth, and terrain age. She further collected soil samples and used seed traps to determine seed bank and seed rain characteristics. Lastly, she conducted a brief survey of other forelands in the region to better understand how succession patterns at Twin Glacier foreland compared regionally. The main finding from her study was that succession proceeded through a directional-replacement pathway, as described by Svoboda and Henry (1987). It passed through four main stages of dominance in 44+ years: mosses —> graminoid-forb —> deciduous shrub-moss —> evergreen dwarf-shrub-moss. Vegetation was most strongly correlated with terrain age. There was a significant positive correlation between above-ground vegetation cover and the germinable seed bank, which suggests that species colonization is largely constrained by !24the availability of seeds. Brief surveys of other glacier forelands revealed that directional-nonreplacement and nondirectional-nonreplacement pathways may also operate in the region. Young (2002) studied primary succession on a different section of the Twin Glacier foreland. He first used various dating techniques to extend the annual chronology for the foreland. He then sampled vegetation cover and environmental factors. His findings showed some stark differences from Jones’, and highlight the heterogeneity of response within a single foreland. Unlike the results reported by Jones (1997), Young (2002) found that terrain age played a secondary role, and species distribution was more a function of moisture availability and substrate size. He also found a general absence of many species commonly found on Alexandra Fiord lowland, and attributed this scarcity to the fact that his study site was located on a small plateau, and thus geographically isolated. Young (2002) did not find evidence of directional-replacement, and concluded that there was no clear successional pattern. These studies point to the variability of successional processes over this topographically heterogeneous foreland. To account for these differences, further studies should survey a broader area. 1.3.2 Secondary Succession at Alexandra Fiord In 1982, ‘Green Igloos’ Experimental Farm was established at Alexandra Fiord to test the possibility of growing food crop in the High Arctic. In developing the farm, approximately 0.3 ha of ground was stripped of natural vegetation and the soil layer disturbed. In 1985, the farm was dismantled and a revegetation experiment was initiated with two treatments: fertilization and surface roughness. Fifteen years later, Chan (2001) studied the residual effects of these treatments on the growth and reproductive output of Papaver spp. By measuring total nitrogen content in the soil, total nitrogen content in the vegetative tissue, and in the aboveground biomass, Chan (2001) found no significant differences between fertilized and unfertilized plots. !25However, she found that surface roughness played a role, where disturbance provided safe sites for establishment. While these results are interesting, a more comprehensive vegetative survey would be help uncover characteristics of the secondary succession process occurring here. !261.4 Conclusions In this literature review we have provided an overview of plant succession in the Arctic. Climate change and development currently occurring in the Arctic make it crucial to understand succession, which is a complex process that is dependent on local environmental factors and individuals species characteristics. Studies that incorporate broad spatial patterns of successional development are most representative. Alexandra Fiord provides a rare chance to use historical research on succession to help develop a deeper understanding of successional patterns and mechanisms operating in High Arctic environments.  !27CHAPTER 2 - PATTERNS AT THREE GLACIAL FORELANDS: REPEAT SURVEY AFTER 21 YEARS Patterns at Three Glacial Forelands:  Repeat Survey after 21 years !282.1 Summary In this chapter we investigated the patterns of plant colonization and community development on three High Arctic glacial forelands, by repeating surveys conducted 21 years ago and testing the validity of the chronosequence approach used in these surveys. In the summer of 2016, we re-visited three glacial forelands eastern coast of Ellesmere Island originally surveyed in 1995. Our primary study area was at Twin Glacier, Alexandra Fiord (78°53’ N; 75°45’ W). We surveyed vascular plant species using percent cover estimates and presence/absence counts. We used indicator species analysis to create a visualization of changes at our primary study area, and a PERMANOVA to test the significance of these changes. Our results show that at Twin Glacier foreland, total percent cover and species richness have increased since 1995. Luzula confusa and Papaver spp. were the most abundant early successional species. Other successional species included Saxifraga cernua, Draba species, Minuartia rubella, and other forbs and graminoids. Salix arctica was abundant on older successional terrain. The species composition observed beyond the Little Ice Age glacial maximum was distinctly different from the composition  of successional communities. The two other forelands surveyed showed both similarities and differences to our primary study area. We conclude that the species observed on all three forelands have advanced in a predictable manner towards the retreating glacial margin, confirming the validity of the chronosequence approach used in these surveys. Species acted individualistically rather than as communities, advancing in a directional manner. Local species diversity played an important role in determining patterns of community development. Patterns of individual species showed agreement and disagreement with other studies, suggesting that both time since deglaciation and local environmental factors play important roles in defining successional patterns. 
!292.2 Introduction In the Arctic, a warming climate is altering landscapes by melting glaciers (Gardner et al., 2011), thawing permafrost (Vincent et al., 2017), and causing shifts in species ranges (Post et al., 2009). It is also rendering the North more accessible to development. Direct effects on the environment come from mineral resource exploration and extraction, increased tourism, and the construction of infrastructure. To model and predict how vegetation will respond to a changing Arctic, and to attempt to manage tundra ecosystems, a thorough understanding of plant succession is needed. The most general definition of succession is species change over time (Walker & del Moral, 2003). The evolution of succession theory over the past 100 years calls for such a broad definition, as succession has proved to be a complicated process with seemingly infinite contingencies, and as a result cannot be easily conceptualized. After a romantic half-a-century of subscribing to the Clementsian deterministic model of succession, experimental and observation evidence eventually forced ecologists to shift towards more reductionist and mechanistic approaches, emphasizing the individualistic nature of species and the importance of site-specific differences (Glenn-Lewin, et al., 1992). Ideas emerged highlighting more diverse mechanisms such as facilitation, tolerance, and inhibition (Connell & Slayter, 1977). Ideas surrounding multiple pathways suggested they could either converge or diverge or something in between (Walker & del Moral, 2003), or that pathways could be determined by the composition of the first species to become established (Egler, 1954). Other ideas about the importance of species life history traits (Chapin, et al., 1994; Egler, 1954) and about abiotic factors such as resource availabilities (Tilman, 1985) continued to render succession theory more realistic, albeit more complex and site specific. In the Arctic, vegetation is subjected to extreme environmental conditions. Low available energy creates conditions of low levels of soil nutrients and organic matter, short cool growing seasons, and slows plant metabolic processes (Bliss, 1962). In addition, much of the Arctic is a desert, !30receiving little summer precipitation and often relying on moisture from snowmelt. These conditions create an environment where abiotic factors can have an important negative effect on plant growth and community development, therefore also hindering succession. Svoboda and Henry (1987) proposed a model of succession in the Arctic that takes these effects of “environmental resistance” into consideration. They suggested that succession in marginal Arctic environments is a function of biological driving forces (intrinsic characteristics of colonizing species such as genetics, dispersal and germination capacity, or ability to survive adverse conditions) and environmental resistance (extrinsic hindering abiotic and biotic factors, such as low temperatures, drought, or predation). They proposed that this could result in three types of succession. Directional-replacement occurs in low resistance environments, where succession proceeds similar to lower latitudes. Directional-nonreplacement occurs in high resistance environments, when invading species succeed in slowly expanding, without replacing earlier species. Nondirectional-nonreplacement occurs in the most extreme environments, where very few species succeed at establishing and continuously fluctuate in their abundance, while other species establish periodically but fail to survive permanently. The site specific nature of successional patterns and processes necessitates a local understanding. Recent research on glacial forelands in Svalbard provides some new information about High Arctic plant primary succession. Hodkinson et al. (2003) investigated two contrasting chronosequences on the nutrient poor foreland of an upland glacier compared to nutrient enriched islands released by a tidewater glacier. Moreau conducted vegetation surveys in front of retreating glaciers and compared the patterns to landscape properties (Moreau et al., 2005; Moreau et al., 2008). Nakatsubo et al. (2010) investigated the drivers and process of Salix polaris (Wahlenb.) development on glacial forelands. Prach and Rachlewicz (2012) studied the patterns of plant succession on five glacial forelands. Těšitel et al. (2014) studied the population structure of Saxifraga oppositifolia and Braya purpurascens on a glacial foreland. All of the mentioned studies showed evidence of directional succession driven by time since deglaciation. One study found evidence of patterns of non-replacement in plant community development (Prach & Rachlewicz, 2012), and the other studies investigating several species reported certain early !31colonizers decreasing in abundance in later stages of succession. Several studies also emphasized the role of frequent disturbances (erosion by proglacial streams, frost heave, slumping) in creating heterogeneous forelands with horizontal discontinuities in stages of succession (Hodkinson et al., 2003; Moreau et al., 2005; Moreau et al., 2008). In the Canadian High Arctic, there have been three sites where primary plant succession on glacial forelands has recently been studied. Forelands in a valley near Oobloyah Bay, Ellesmere Island (about 100 kilometres NE of Eureka) have been surveyed since 2002 (Okitsu et al., 2004; Mori et al., 2006; Mori et al., 2008; Mori et al., 2013). Here again the researchers found evidence of directional non-replacement succession, where colonization was aided by favourable micro-sites. Breen and Lévesque (2006) surveyed Teardrop Glacier foreland in Sverdrup Pass, Ellesmere Island, focusing on the importance of biological soil crusts in facilitating the establishment and growth of plants during early stages of succession. The third site is at Alexandra Fiord on the central eastern coast of Ellesmere Island, and the surrounding region. Here Jones and Henry (2003) carried out an intensive vegetation survey on the foreland of Twin Glacier in 1995, along with rapid surveys of four other glacial forelands. At Twin Glacier, they found evidence of directional-replacement succession driven by time since deglaciation. They suggest that succession proceeded though four main stages, namely: (1) pioneer mosses, (2) graminoid-forb, (3) deciduous shrub-moss, and (4) evergreen dwarf-shrub-moss. However, the surveys at the other sites showed different patterns, both directional and non-directional, with and without replacement of species.   In this chapter we returned to Alexandra Fiord to resurvey the vegetation at three of the forelands originally surveyed in 1995. To our knowledge, only one other study of this kind exists in the High Arctic. In 2006, Moreau resurveyed vegetation along transects at two glacier forelands first surveyed in 1975 (Moreau et al., 2009). They observed directional succession, which they described as progressing through seven stages that remained comparable at the two points in !32time. However, a drying of the environment and disturbances most often caused by proglacial water erosion created trajectories that were less predictable.  Such repeat studies are especially important when investigating glacial foreland succession. Almost all studies exploring primary succession on glacial forelands assume the foreland represents a chronosequence, i.e. a set of sites formed on the same substrate that differ only in time since they were formed (Walker et al., 2010). As plant succession in the High Arctic is a process that lasts hundreds or thousands of years (Svoboda & Henry, 1987), the use of this space-for-time substitutions enables ecologists to uncover trends over time periods longer than their lifetimes. However, chronosequences themselves rely on one critical assumption: all sites start with identical environmental conditions. In practice, other factors can influence chronosequence interpretation, such as differences in substrate material, climatic history, disturbances, or unequal input of propagules (Finegan, 1984). These uncertainties have led to criticism of the use of chronosequences (Johnson & Miyanishi, 2008; Walker et al., 2010). However, repeat studies of vegetation patterns along chronosequences is one way of validating their underlying assumptions. Comparing the predicted future to the actual future gives researchers new insights and helps resolve uncertainties surrounding chronosequence interpretation. In this chapter we revisited three High Arctic chronosequences 21 years after they were first surveyed. We investigated the patterns of plant colonization and community development, and tested the validity of the chronosequence approach used in these surveys. 
!332.3 Methods 2.3.1 Study Sites Twin Glacier, Alexandra Fiord, Ellesmere Island (78°50’ N; 75°47’ W) The primary study site was located on the foreland of the western lobe of Twin Glacier, Alexandra Fiord, Ellesmere Island in the Canadian High Arctic (Figure 2.1). The two tongues of Twin Glacier stretch down from the Prince of Wales Ice Cap on Johan Peninsula, and form the southern end of a deglaciated valley. The valley (8 km2) is bound by steep cliffs to the east and west, and the coastline of Alexandra Fiord forms the northern edge. The topographic position of the valley results in enhanced moisture and warmth, and has resulted in high plant diversity (92 vascular plant species), cover, and production compared to the surrounding polar desert environment (Freedman et al., 1994). For this reason, the valley is considered a polar oasis. !34Figure 2.1. Map showing the locations of the three forelands surveyed.The rocky gneiss-granite foreland of the western lobe of Twin Glacier is characterized by highly variable topography and two glacial streams (Figure 2.2). The sampling area was on a very gentle east facing slope, located at the base of small hill on top of which lies the current glacial margin (Figure 2.3). A unique aspect of the foreland is the abundance well-preserved pre-LIA plant communities and relatively rich organic soil that have been released from the glacier. According to radio carbon dating, the paleo-communities within our survey area were entombed between the year 1410 and 1690 (Bergsma et al., 1984) and perhaps as early as 1120 (Jones, 1997). The paleo-soil generally supports higher plant cover and a slightly greater species diversity than the glaciofluvial sediment (Jones & Henry, 2003). !35Figure 2.2. (a) Aerial photograph of the western lobe (right) and southern lobe (left) of Twin Glacier in July 2015. The lichen trimline can be seen near the glacier, where recently the recently deglaciated area is characterized by lighter coloured lichen-free rocks, and the area past the LIA glacial maxima is characterized by darker lichen-covered rocks. (b) Ground view of the study area on Twin Glacier foreland.  a) b)Beistad Glacier, Beistad Fiord, Ellesmere Island (79°02’ N; 78°55’ W) This granite-gneiss glacier foreland is at the head of Beistad Fiord (Figure 2.1), and surrounded by steep slopes (Figure 2.4). Several meltwater streams flow across the flat foreland, surrounded by moist graminoid-dominated depressions, and intermixed with elevated gravel and boulder rich ridges. Although paleo-vegetation and soil exists, it is rare. Teardrop Glacier, Sverdrup Pass, Ellesmere Island (79°07’ N; 79°44’ W) This foreland surrounds a glacier commonly referred to as Teardrop Glacier (Figure 2.1). It is in the relatively lush valley of Sverdrup Pass, also considered a polar oasis (Breen & Lévesque, 2006). The total area of the oasis is much greater than that of the Alexandra Fiord lowland (ca. 800 km2), and it hosts a greater number of herbivores, including a healthy population of muskoxen (Henry et al., 1986; Raillard & Svoboda, 2000). In comparison it has a higher species !36Figure 2.3. Aerial photograph of Twin Glacier’s western lobe in July 2015. The survey area is indicated by the yellow shape. Approximate glacial retreat between 1995 and 2015 is shown by the white patterning. The dashed white line shows the lichen trim line, indicating the maximum extent of the glacier during the Little Ice Age.!37Figure 2.4. (a) Aerial photograph of Beistad Glacier at the head of Beistad Fiord, taken in July 2016.  (b) Ground view panorama in front of Beistad Glacier. a) b)richness, and individual plants were larger and lusher than at Alexandra Fiord. The foreland itself is gently sloped to the north, and very moist with several small streams (Figure 2.5). Similar to the Twin Glacier foreland, this foreland has patches of paleo-vegetation and soil. !38Figure 2.5. (a) Aerial photograph of Teardrop Glacier, Sverdrup Pass taken in July 2016. The boulder line can be seen, delineating the recent LIA glacial maximum. (b) Ground view panorama of the transects in front of Teardrop Glacier. a) b)2.3.2 Methods Vegetation Surveys – Twin Glacier The survey design used a chronosequence approach where distance from the glacier represents increasing time since deglaciation (Jones 1997). In 1995 the sampling area extended 300 metres from the glacial margin, to just beyond the Little Ice Age glacial maximum. In 2016 we sampled the same area, however, it is now about 115 m from the current glacial margin (Figure 2.6). The !39Figure 2.6. Aerial view of the study area in July 2016. The white grid shows the actual layout and size of the 144 zones surveyed, as referenced by GPS coordinates. Blue lines show the historical position of the glacial margin, as staked by Josef Svoboda (1980-1985) and Greg Henry (1992-1995) and their teams. area was stratified into thirty 10 m long belts that run parallel to the glacier, further subdivided into six zones (Figure 2.6; Figure 2.7). The number of zones decreased with distance from the glacier because of the limits imposed by a river canyon, and two zones were missing in the first belt because of slope topography, resulting in a grid consisting of 144 zones. Transects in the study area were marked with metal stakes and tags in 1995, which were nearly all relocated in 2015 allowing us to recreate the sampling area with high fidelity. Appendix A provides information on relocating the transects and study area for future research. In July 2016 we conducted two parallel vegetation surveys representing two scales of observation. At a finer scale, each of the 144 zones was surveyed using six randomly located 50 cm x 50 cm quadrats, that were further subdivided into 25 squares (Figure 2.7, Small Quadrats). Within each quadrat, percent cover of vascular plant species was estimated visually. All visual !40Figure 2.7. Survey design. The study area extends 300 meters from the 1995 glacial margin, and is divided into 30 ten metre long belts that run parallel to the glacier. The belts are subdivided into zones, for a total of 144 zones. Within each zone, two levels of surveying were undertaken in 2016. At a coarse scale, each zone was divided into four Large Quadrants. At a detailed scale, six 50 cm x 50 cm Small Quadrats were chosen randomly within each zone.estimates of plant cover were done by one observer to reduce bias (Kercher et al., 2003). At a coarser scale, each zone was divided into four equal area quadrants (Figure 2.7, Large Quadrants). Within each of these quadrants, presence/absence of vascular plant species was noted after a survey on foot that lasted on average 15 minutes.  Our survey mimics and expands upon the one conducted during the growing seasons of 1994 and 1995. The original survey also used the same 50 cm x 50 cm quadrats randomly placed within each zone to visually estimate the cover of vascular plant and moss species. However, a higher density of quadrats was used, resulting in a total of 1411 surveyed quadrats in the original study compared to a total of 864 surveyed quadrats in 2016. No coarse scale survey with the larger quadrants was conducted in 1995. For further details about the original survey see Jones (1997) and Jones & Henry (2003). Vegetation Surveys – Beistad & Teardrop Glaciers Presence/absence of vascular plants was recorded along transects leading away from the glacial margin at both sites on July 27, 2016. At Beistad Glacier, three 70-m transects were sampled along slightly elevated gravel ridges (Figure 2.8). At all three transects, presence/absence of plant species was noted in 50 cm x 50 cm quadrats placed every three metres. At transects 1 and 3, presence/absence of plant species was also noted in 2 m x 2 m quadrats placed every eight metres. At Teardrop Glacier in Sverdrup Pass, three 215-m transects were surveyed. Presence/absence of plant species was noted at all three transects in 50 cm x 50 cm quadrats placed every three metres for the first 70 m, every five metres from 70 m - 125 m, and every 10 metres thereafter. At transects 1 and 3, presence/absence of plant species was also noted in 2 m x 2 m quadrats placed every 10 m for the entire length of the transect. Appendix A provides information on relocating the transects surveyed in 2016 at both Beistad and Teardrop Glaciers. In 1995 the survey methods differed slightly (Jones & Henry, 2003). At Beistad Fiord, transects originally also ran on slightly elevated portions of the foreland. Presence/absence of vascular plant and moss species was recorded in 50 cm x 50 cm quadrats every three meters along two 57-!41m long transects. At Sverdrup pass, presence/absence of vascular plant and moss species was surveyed in 50 cm x 50 cm quadrats along three 145 -m transects. Quadrats were placed every three metres for the first 30 m, every five metres for the next 25 m, every 10 metres for the next 30 m and every 20 metres for the last 60 m.  Data Analysis –  comparison of 1995 and 2016 Our main interest was to compare patterns of species presence and abundance on the forelands between 1995 and 2016, and to do so we could only use data that were comparable between the !42Figure 2.8. Transect layout on the forelands of Beistad Glacier and Teardrop Glacier, Sverdrup Pass. two years. At Twin Glacier, this consisted of the percent cover survey of vascular plants in the 50 cm x 50 cm quadrats, while at the two other forelands it was the presence/absence survey along the transects also in the 50 cm x 50 cm quadrats. Data analysis was done using the R programming environment version 3.4.0 (R Core Team, 2017). At Twin Glacier, total percent cover of all vascular plant species was averaged across belts, while species richness was summed across belts. A Shannon-Weaver diversity index was calculated for each belt using the package ‘vegan’ (Oksanen et al., 2017) as: H’ = - ∑(pi) * ln(pi)            (2.1) where pi is the proportional abundance of species i. Temporal change in overall community composition was quantified by calculating the Jaccard’s Similarity Index using the ‘betapart’ package (Baselga et al., 2017) as: J(A,B) = |A ∩ B| / |A ∪ B|        (2.2) where A is the community composition of a zone in 1995, and B is the community composition of the same zone in 2016. The results were plotted with a LOESS curve. To formally test whether species composition on the recently deglaciated portion of the study area had changed between 1995 and 2016 we used Permutational Multivariate Analysis of Variance (PERMANOVA; Anderson, 2001). Plant data were split into recently deglaciation (0 - 250 m) and post-LIA glacial maximum (250 - 300 m), and a PERMANOVA was run using the “adonis” function from the ‘vegan’ package in R (Oksanen et al., 2017), with Bray-Curtis distances and 9999 permutations. To assess successional patterns at Twin Glacier foreland in a more detailed manner, we were interested in analyzing the patterns of important species. For this we used the indicator species !43analysis developed by Dufrêne and Legendre (1997), available through the R package ‘indicspecies’ (De Caceres & Legendre, 2009). However, this analysis requires sites to be grouped.  To determine appropriate groupings that were consistent in the two years, we used hierarchical clustering to cluster the 30 belts into 10 groups. The clustering was based on a matrix of Bray-Curtis dissimilarities calculated from species averages across the belts. To visualize the differences between belts, we used Non-Metric Multidimensional Scaling (NMDS), and coloured the belts according to the 10 clusters (Figure 2.9). Acknowledging the spatial sequence of the belts, the resulting two figures were visually assessed and compared to determine the following four groups: Group 1 = 0 - 70 m, Group 2 = 70 - 180 m, Group 3 = 180 - 250 m, Group 4 = 250 - 300 m. The grey polygons in Figure 2.9 show these grouping in the NMDS space. While the groups sometimes overlap in the NMDS space, following the spatial sequence of the belts and creating groups that were consistent in both years was more important than complete fidelity to the NMDS spatial arrangement of the belts. The ‘vegan’ package was used to help compute dissimilarities and to perform the NMDS (Oksanen et al., 2017).  We determined the indicator species (InSp) for these four groups of sites based on percent cover of species. The analysis is based on the frequency of species within a group, and the concentration of abundance of a species within the group (De Caceres & Legendre, 2009). An IdSp is defined as the most characteristic species of each group, found mostly in a single group or combination of groups, and present in the majority of the quadrats belonging to that group (Dufrêne & Legendre, 1997). We used 999 permutations to aid in determining the significance of the indicator values calculated for each species, and only species with significant (p > 0.05) indicator values were used to produce the final visualizations. Significant InSp for the two years were graphed in Photoshop CC 2017 to show the groups they represent. Width and shading of bars was used to represent relative cover of each InSp. !44!45Figure 2.9. NMDS visualizations, calculated from average percent cover of plant species in small quadrats in (a) 1995 and (b) 2016. Belts are coloured according to the group assigned by hierarchical cluster analysis. Grey polygons show the groups to which the belts were assigned (Group 1 = 0 - 70 m, Group 2 = 70 - 180 m, Group 3 = 180 - 250 m, Group 4 = 250 - 300 m).At Beistad and Teardrop Glaciers, presence/absence of plant species was visualized along an axis representing distance from the glacial margin for data from both 1995 and 2016. Data Analysis – 2016 additional In addition to the comparative analysis between the two years, new data collected in 2016 provided further information about the patterns of less common species. Because of the very low density of many species on High Arctic glacial forelands, their patterns only become apparent in surveys of larger areas. At Twin Glacier, we surveyed species presence/absence in Large Quadrants (Figure 2.7). Using this presence/absence data, we calculated a relative frequency of each species in each belt, and visualized this along an axis representing distance from the 1995 glacier front. At Beistad and Teardrop Glaciers, presence/absence of plant species in 2 m x 2 m quadrats was visualized along an axis representing distance from the glacier front.  !462.4 Results 2.4.1 Twin GlacierCommunity Patterns The total percent cover of vascular plant species has increased overall since 1995, especially on the most recently deglaciated portion of the foreland (Figure 2.10a). The dominant change was in the first 100 m, where total percent cover increased on average by 2.4% between 1995 and 2016. Previous monitoring of the glacial retreat showed that the Belt 100 in our study area was deglaciated in 1980 (Figure 2.6). Beyond the LIA glacial maximum (i.e. 250 - 300 m), percent cover has remained relatively similar. There was the sharp decrease in percent cover observed in both years at Belts 230 and 240. This occurs right before the LIA glacial maximum. Species richness in 1995 showed an increasing trend moving away from the glacial margin (Figure 2.10b), beginning with 1 species in Belt 0, and increasing to a maximum of 14 species by Belt 190.  Twenty-one years later in 2016, species richness has increased to 6 species in Belt 0. From there species richness increased to maximum of 16 species in Belt 130, after which it decreased along the rest of the recently deglaciated area, returning to a minimum of 3 species in Belt 230. Similar to the decrease in percent cover at Belts 230 and 240, there was also a decrease in species richness. Plant diversity was greater in 2016 than in 1995 up until Belt 180 (Figure 2.11a). Then diversity in 2016 declined and was lower than in 1995 until Belt 240, after which diversity was more or less the same in both years. The similarity in community composition between the two years increased on average until Belt 180, after which it decreased (Figure 2.11b). !47!48Figure 2.10. (a) Total percent cover and (b) total species richness across the foreland in both 1995 (dashed line) and 2016 (solid line). The vertical dotted line indicates the location of the LIA glacial maximum, a time discontinuity in the chronosequence. a) b)!49Figure 2.11. (a) Shannon’s Diversity in both 1995 (dashed line) and 2016 (solid line). (b) Jaccard’s Similarity between zones in year 1995 compared to 2016. a) b)The PERMANOVA (Table 2.1) indicats differences in community composition between years was more pronounced ion the recently deglaciated foreland (0 - 250 m) compared to the area beyond the LIA glacial maximum (250 - 300 m). For the area beyond the LIA glacial maximum, there was not a significant difference between years (p > 0.05). For the recently deglaciated area, the difference in composition between years was significant (p < 0.05). Indicator Species In 1995, Luzula confusa was the most abundant species across the foreland (on average 1.23 %), followed by Salix arctica (0.50 %) and then Papaver spp.  (0.16 %) (Figure 2.12, 1995). For 1Group 1, no species proved to be an InSp. For Groups 2 to 4, many InSp represent several groups, however certain species were more abundant in Group 2 (Luzula confusa, Poa arctica, and Draba lactea), while others were more abundant in Group 3 (Papaver spp., Draba nivalis, and Chamerion latifolium). Only Saxifraga cernua was loyal to Group 2, and Draba oblongata was loyal to Group 3. Group 4, which lies beyond the LIA glacial maximum, had a distinctly different composition. Several species (Dryas integrifolia, Cassiope tetragona, Carex nardina,  While arctic poppies at Alexandra Fiord have previously been referred to as Papaver radicatum, here we follow 1updates from the online Flora of the Canadian Arctic Archipelago that suggest Papaver radicatum is restricted to Iceland and Scandinavia (Aiken et al., 2007).  According to this classification, species found on Ellesmere Island include Papaver lapponicum, Papaver dahlianum, and Papaver cornwallisense (Aiken et al., 2007).!50DF MS F model pRecently deglaciated: 0 - 250 mYear 1 0.87 5.24 0.0004Error 48 0.17Post-LIA glacial maximum: 250 - 300 mYear 1 0.30 2.14 0.062Error 8 0.14Table 2.1. PERMANOVA results testing the significance of the change in species composition on the recently deglaciation foreland (0 - 250 m) and the area beyond the LIA glacial maximum (250 - 300 m).  !51Figure 2.12. Statistically significant indicator species (InSp) in 1995 (left) and 2016 (right) calculated from percent cover of vascular plant species in small quadrats. Coloured boxes show only the groups for which the species was an InSp, despite the fact that the species often existed in groups where it was not an InSp. Width of the bar represents the relative abundance of the species compared to the most abundant InSp (Luzula confusa in 1995, and Salix arctica in 2016), considering only the cover in the groups for which it was an InSp. As such, the width of the bar represents the relative cover of the species across the foreland, where wide bars indicate more abundant cover compared to thin bars. Shading of the bar represents the relative abundance of the species across the groups it represents, with dark shading indicating higher cover. The colours of the bars represent different life-forms, where yellow = graminoids, green = forbs, and red = shrubs. The horizontal dotted line indicates the location of the LIA glacial maximum, a time discontinuity in the chronosequence.Saxifraga oppositifolia, Cardamine bellidifolia, Festuca brachyphylla, and Saxifraga tricuspidata) represented only Group 4, and suggest a very different community exists beyond this time discontinuity on the landscape. Salix arctica was most abundant in Group 4, however it had begun to encroach into Group 3 as well.  In 2016, many of the indicator species were the same as in 1995, including the three most abundant InSp across the foreland (Figure 5, 2016). However, by 2016 Salix arctica had replaced Luzula confusa as the most abundant species, although the latter was still the second most abundant species. This represents an overall increase in the abundance of Salix arctica across the surveyed area over the 21 years, which now covered on average 0.94 %. Papaver spp. had also increased in abundance, now averaging 0.29%. Meanwhile, Luzula confusa decreased in abundance, now averaging 0.94 %.  All InSp on the recently deglaciated portion of the foreland (0 - 250 m) that were present in both years had moved closer to the glacial margin, both in the groups they represent and in their abundance. This includes Luzula confusa, Papaver spp., Saxifraga cernua, Draba lactea, Draba oblongata, and Salix arctica.  In the cluster of species that represented Group 4, many species were present in both years (Dryas integrifolia, Carex nardina, Festuca brachyphylla and Saxifraga tricuspidata). In 2016, Cassiope tetragona represented both Group 1 and Group 4, however it was much more abundant in Group 4. This resemblance in species composition between the two years, and the fact that these species had not advanced onto the recently deglaciated portion of the foreland in a significant way, suggest that the communities in Groups 1-3 are still far from being mature. There were a few discrepancies in the indicator species of Group 4 between the two years, however these discrepancies represent species that were found in very low abundance and cover (Saxifraga oppositifolia 1995 = 0.06 % mean cover in Group 4, Cardamine bellidifolia 1995 = 0.04 %, Silene acaulis 2016 = 0.01 %, Saxifraga nivalis 2016 = 0.02 %). Therefore, these differences are likely a result of species being missed by the 90 quadrats sampled in this area.  !52Large Quadrants This large scale survey emphasized the patterns of less common plants on the foreland (Figure 2.13). As in the survey of smaller quadrats, certain species were more common in younger belts on the foreland, including Saxifraga cernua, Draba lactea, Cerastium alpinum, Sagina intermedia, and Saxifraga cespitosa. Other species were more common on older successional belts, including Draba oblongata and Erysimum pallasii. Still other species did not show as much variation in abundance, however they were clearly more common on the recently deglaciated portion of the foreland than past the LIA glacial maximum, including Draba nivalis, Minuartia rubella, Melandrium spp., and Stellaria longipes.  The patterns of species that dominated after the LIA glacial maximum showed a different pattern when viewed at this scale. Species characteristic of a more mature community had indeed begun to advance onto the recently deglaciated portion of the foreland. Species including Cassiope tetragona, Silene acaulis, Saxifraga tricuspidata, Festuca brachyphylla, Saxifraga nivalis, Dryopteris fragrans, Vaccinium uliginosum, Cardamine bellidifolia, Dryas integrifolia, and Carex nardina were all more common after the LIA glacial maximum, but were also present with less frequency on the recently deglaciated area. Similar to the pattern in the 2016 InSp analysis, Cassiope tetragona had a higher frequency on younger belts, followed by a decrease in frequency, and then an increase again after Belt 150. 2.4.2 Beistad FiordThis foreland had consistently lower species diversity compared to the other forelands surveyed in this study. However, it is likely that the quadrat size chosen in 1995 was too small to capture the patterns of plant diversity across the foreland, as species diversity of vascular plants was significantly higher in the 2 m x 2 m quadrats (17 species), compared to the 50 cm x 50 cm quadrats (9 species in 2016).  !53!54Figure 2.13. Relative frequency of vascular plant species across Twin Glacier foreland in 2016, based on the total number of large quadrants containing the species within a belt (refer to Figure 2.7 for terminology). The colours of the shapes represent different life-forms, where yellow = graminoids, green = forbs, and red = shrubs. The vertical dotted line indicates the location of the LIA glacial maximum, a time discontinuity in the chronosequence.Most species present in 1995 have advanced spatially towards the glacier by 2016 (Figure 2.14). Most of the species present in both years are recorded having their first appearance closer to the glacier margin in 1995 compared to 2016, including Draba subcapitata, Salix arctica, Dryas integrifolia, and Poa abbreviata. The only exception was the most abundant species, Saxifraga oppositifolia, which was recorded first appearing three metres earlier in 2016 compared to 1995. Two species observed in 1995 were not observed in 2016, including Draba corymbosa and Poa arctica. There is a chance that this is an error due to misidentification, as both species could resemble other species that were observed on the foreland, and surveys in 2016 occurred after the flowering of Draba, making identification more difficult. In 2016 there was an abundance of Festuca brachyphylla observed across the foreland, which was not observed in 1995, suggesting this species arrived post-1995. Draba nivalis was also observed in numerous quadrats in 2016, whereas it was not observed in 1995, suggesting it might also be a new arrival. However as is the case with all Draba species, there is a chance that misidentification played a role in this difference of observations. 2.4.3 Sverdrup PassThe foreland along the margin of Teardrop Glacier was remarkably moist and rich in species. The most significant difference between the two years can be seen in the distance of species to the glacial margin. In 2016, most species were first observed much further from the glacial margin than in 1995 (Figure 2.15). Considering the accelerated retreat rates of Teardrop glacier in the past decades (La Farge et al., 2013), this could suggest that species have not had time to disperse into the newly exposed area. In addition, certain individual species showed changes. Puccinellia andersonii, Braya purpurascens and Cerastium alpinum were observed in several quadrats in 1995 but not in 2016, while Pedicularis capitata, Saxifraga rivularis, and Trisetum spicatum were observed only in 2016. There was some confusion about Draba species between the two years, where Draba nivalis was observed in 2016 but not in 1995, and Draba corymbosa !55!56Figure 2.14. Summary of vegetation presence/absence at Beistad Fiord, surveyed in 50 cm x 50 cm quadrats in (a) 1995 and (b) 2016, and (c) in 2 m x 2 m quadrats in 2016. The solid dark-grey vertical line represents the LIA glacial maximum, and the dotted dark-grey vertical line in (b) and (c) represent the location of the glacier front in 1995. The colours of the shapes represent different life-forms, where yellow = graminoids, green = forbs, and red = shrubs. b) a) c)!57(a)!58(b)!59Figure 2.15. Summary of vegetation presence/absence at Teardrop Glacier, Sverdrup Pass in 50 cm x 50 cm quadrats in (a) 1995 and (b) 2016, and (c) in 2 m x 2 m quadrats in 2016. For full description see caption for Figure 2.14.(c)and Draba subcapitata were observed in 1995 but not in 2016. This confusion may be due to misidentification rather than actual change in species, as surveying in 2016 occurred after the flowering period of Draba. Certain species were observed more frequently in 2016, including Poa abbreviata, Chamerion latifolium, Saxifraga tricuspidata, Dryas integrifolia and Bistorta vivipara. Other species were observed more frequently in 1995, including Minuartia rubella, Saxifraga nivalis, and Cardamine bellidifolia. These differences are likely a result of both a change in species composition on a foreland undergoing succession, as well as differences due to chance because of the different locations of the transects and placement of quadrats between the two years.!602.5 Discussion 2.5.1 Successional patterns The most evident pattern that arose across all three forelands, in both years, and at both large and small surveying scales is that species are advancing along the successional gradient individualistically rather than as well defined communities. This is in agreement with many other studies which describe individualistic responses of plant species to changing environmental conditions (Glenn-Lewin, 1980; Williams et al., 2004; Feagin et al., 2005; Erschbamer, 2007; Chapin & Shaver, 1985). However, many studies characterizing vegetation patterns during primary succession in the High Arctic define stages of succession by using ordination and classification techniques to split species into communities along the gradient of terrain age (Bliss & Gold, 1994; Jones & Henry, 2003; Moreau et al., 2005; Moreau et al., 2009). Organizing a large number species into smaller number of communities is a convenient way of communicating the patterns of species change. However, it may be counterproductive in that it encourages community-oriented conclusions rather than focusing on species specific patterns and drivers of succession.   In describing the patterns of species as individualistic, we are not implying that species are necessarily independent. Callaway (1997) has argued that the focus on individualistic species responses to abiotic drivers goes contrary to strong evidence of positive species interactions. In this study, we indeed observe individualistic arrangement along the terrain-age gradient. However, Callaway argues that the evidence for positive interactions between species suggests that their distributions are not truly independent. Furthermore, he describes how the importance of these positive interactions becomes even more significant as abiotic conditions deteriorate. Indeed, others have also point to the importance of facilitation and positive interactions in marginal Arctic environments (Olofson, 2004; Carlsson & Callaghan 1991; Callaghan & Emanuelsson, 1985). In describing species patterns as individualistic, we simply imply that there !61are no clearly defined ‘stages of succession’. Instead species arrange themselves individualistically in micro-environments that satisfy their needs. The individualistic patterns may or may not be aided by growing in association with or proceeding other species. Another pattern that arises, and is especially evident in the comparison between the two years, is that succession proceeds in a directional manner. Species that were present in 1995 are also present in 2016, and have all advanced in both their presence and abundance towards the glacial margin. This is in agreement with other High Arctic primary succession studies (Bliss & Gold, 1994; Mori et al., 2008; Okitsu et al., 2004; Jones & Henry, 2003; Moreau et al., 2009; Prach & Rachlewicz, 2012). Despite papers which suggest otherwise (Mori et al., 2008; Okitsu et al., 2004), directional succession is not uncommon in High Arctic environments where relatively lush vegetation is found (Svoboda & Henry, 1987). Both the forelands at Twin Glacier and Teardrop Glacier are adjacent to polar oases, rich with plant species and influenced by ameliorated abiotic temperature and moisture conditions.  Comparing the three different forelands, another pattern that arises is the importance of local species diversity in determining species patterns on a successional landscape. The foreland at Beistad Glacier was by far the least diverse and was dominated by the very sparse presence of Saxifraga oppositifolia. This foreland was surrounded by steep hills, and the area of land between the glacier and the ocean was very small, inhibiting the establishment of a significant vegetation community to act as a seed source. The foreland at Twin Glacier was dominated by Luzula confusa, Papaver spp., and Salix arctica, but was host to a relatively diverse selection of other species. This foreland is at the end of a valley described as a polar oasis (Freedman et al., 1994) and therefore is influenced by a much more diverse vegetation community and supply of seeds. At Teardrop Glacier, both Luzula confusa and Salix arctica are important, but so are Saxifraga oppositifolia and Stellaria longipes, and the diversity of successional species is even greater than at Twin Glacier. This foreland is surrounded one of the largest and most diverse polar oases in the region. Evidence that successional patterns are driven by location (and as an extension, seed source) has been shown in other studies in the Arctic (Fastie, 1995; Cannone et !62al., 2010). Further, others have touched on the importance of site specific factors in influencing species (Boulanger-Lapointe et al., 2014) and community (Elmendorf et al., 2012a) responses to ecosystem drivers. Finally, an interesting pattern seen at Twin Glacier was the drop in both percent cover and species richness right before the LIA glacial maximum. This is striking because the pattern occurs in both years in the same belts (Belts 230 and 240). We hypothesize that this pattern is related to geomorphological action at the LIA glacial terminus. Bergsma et al. (1984) studied a 1959 aerial photograph of Twin Glacier and noted that the terminus had only retreated ~45 m from its LIA glacial maximum, compared to ~135 m in the 22 proceeding years. We interpret this as evidence that the glacier likely remained at its maximum position for several decades, if not centuries, thus allowing it time to modify the abiotic environment and conditions at this position. However, when we investigated graphs of the substrate surveys performed in both 1995 and 2016, there were no anomalous patterns in substrate size at these two belts.  2.5.2 Species patternsAs we have argued for a greater focus on describing individual species patterns, in this section we direct our effort to such a discussion. We focus specifically on the description of eight species that were important at the forelands investigated. Luzula confusa This species was by far the most important species on Twin Glacier foreland, dominating as an early successional vascular plant species in 1995, and retaining its dominance in 2016. It was not present at Beistad Glacier; however, it was important on Teardrop Glacier foreland. This is in agreement with Breen and Levesque (2006), who also described Luzula confusa as an important early- to mid-successional species on Teardrop Glacier foreland. However, Mori et al. (2008) !63note the highest presence of Luzula confusa on the oldest moraine at their study site in Oobloyah Valley, and did not find it on the youngest moraine where Chamerion latifolium dominated instead.  In 1994 and 1995, seed traps were set and soil seed bank samples were collected across the study area at Twin Glacier foreland (Jones 1997). These showed that the dominance of Luzula confusa on the foreland was mirrored by its dominance in the germinable seed bank (78.7% of emergent seedlings), and by its near exclusivity in the fall-winter seed rain (98.4%). Abundant seed production and particularly high germinability have not been found in other studies (Addison & Bliss, 1984; Cooper et al., 2004; Alsos et al., 2013; Müller et al., 2011), suggesting that the study area on Twin Glacier foreland was particularly well suited for the reproduction of Luzula confusa. While Luzula confusa is able to persist in a wide range of environments (Aiken et al., 2007), it prefers sites of intermediate moisture and requires higher than normal soil moisture for seedling establishment (Addison & Bliss, 1984). This preference for moisture might be a reason Luzula confusa is found in higher abundance on younger terrain, which often benefits from moisture from the melting glacier. Papaver spp. Appearing slightly after Luzula confusa, Papaver spp.  were the second most abundant early 1successional species on Twin Glacier foreland, increasing in abundance by 2016. It was present but rare at Beistad Glacier, but was present as an important early successional species at Teardrop Glacier. This agrees with the results from Breen and Levesque (2006) who similarly found Papaver spp. important early - to mid- successional species, but disagrees with Mori et al. (2008) who found it in highest frequency on the oldest moraine.  That Papaver spp. are important successional species agrees with their description as one of the hardiest vascular plants in the High Arctic, preferring open substrates with mineral soil and low  See footnote on page 50 regarding the nomenclature of Papaver. 1!64organic content (Aiken et al., 2007; Svalbard Flora, 2018). While their percent germination is often low (Svalbard Flora, 2018; Bell & Bliss, 1980; Klady et al., 2011; Bliss, 1958; Alsos et al., 2013; Müller et al., 2011), individual plants can produce more than 400 seeds in a favourable year (Svalbard Flora, 2018), giving them an advantage as an r-selected species. Saxifraga cernua Saxifraga cernua was a less abundant, but clearly an early- and mid-successional species on Twin Glacier foreland, disappearing on older terrain. It was not present at Beistad Glacier, but was a persistent early-to-mid successional species on Teardrop Glacier. The results from Breen and Levesque (2006) agree, however other studies are less conclusive. Mori et al. (2008) and Okitsu et al. (2004) found Saxifraga cernua on intermediate moraines in Oobloyah Valley, and Hodkinson et al. (2003) found it arriving in the oldest stage (> 150 years) at their study sites in Svalbard.  This species is very successful at reproducing vegetatively through bulbils (Svalbard Flora, 2018) and often shows high percent germination (90 % + in optimal conditions) from these bulbils (Bell & Bliss, 1980; Alsos et al., 2013; Müller et al., 2011). It prefers cool and moist habitats, such as snow-beds (NA Flora, 2018; Aiken et al., 2007; Svalbard Flora, 2018; Bell & Bliss, 1980). Its strategy of prolific reproduction and successful germination, along with its preference for moist cool habitats, is favourable for success as an early successional species. Saxifraga oppositifolia At our study area on Twin Glacier foreland, Saxifraga oppositifolia had a low overall abundance and does not show a particularly interesting pattern with distance from glacier. However, at Beistad Glacier it was the earliest arriving vascular plant species, and persisted along the entire chronosequence. At Teardrop Glacier, it also arrived early and persisted along the rest of the foreland. In other studies, Saxifraga oppositifolia has a similar pattern of arriving early and !65gradually growing and increasing in size over time (Breen & Levesque, 2006; Hodkinson et al., 2003; Moreau et al., 2008; Mori et al., 2008; Těšitel et al., 2014).  Saxifraga oppositifolia is the northernmost flowering plant species in the world, and one of the most common and widespread species in harsh polar desert environments (Svalbard Flora, 2018; Aiken et al., 2007). It is one of the first species to flower in the spring (Aiken et al., 2007), making it more likely for plants to produce ripe seeds even in less favourable years. It is not as common in environments where there is competition from other species (Aiken et al., 2007). Salix arctica By 2016, Salix arctica had become the most important species in our study area on Twin Glacier foreland, having increased in percent cover since 1995. In our study area the species appeared to arrive later on older terrain. However, on other sections of the foreland seedlings of Salix arctica were found very close to the glacial margin (Chapter 3). At Beistad Glacier this species was present but not very abundant. In 2016 it was observed in mid-stages and persisted into later stages, but in 1995 only appeared once in mid-stages. At Teardrop Glacier, Salix arctica was one of the earliest arriving vascular plant species, and persisted along the entire chronosequence. Other studies also point to its early arrival and gradual increase in cover over time (Breen & Levesque, 2006; Okitsu et al., 2004; Mori et al., 2008; Cannone et al., 2010). Salix arctica is an early flowerer, prolific seed producer, and able persist in a wide range of habitats (NA Flora, 2018; Aiken et al., 2007). It is able to live anywhere from waterlogged sedge meadows to xeric wind and snow blasted ridges. Its seeds are well adapted to wind-dispersal facilitating its early arrival to successional landscapes, while its slow growing prostate growth form means that it will gradually increase in cover over time. !66Cassiope tetragona  Cassiope tetragona was for the most part a mature species, but only present on Twin Glacier foreland and not at Beistad or Teardrop Glaciers. The highest percent cover of this species was found in the mature post-LIA belts. However, in 2016 young plants were also found with higher frequency in the younger belts, and did not show a directional pattern. Instead they were more common on younger belts, then decreased in frequency until about 150 m, after which they increased again. Other High Arctic studies mostly point to Cassiope tetragona as a later successional and mature species (Breen & Levesque, 2006; Okitsu et al., 2004; Hodkinson et al., 2003; Cannone et al., 2010), however Mori et al. (2008) also found this species on mid-successional moraines, gradually increasing in frequency over time. This species is strictly adapted to insect pollination (Svalbard Flora, 2018), and not as successful at germinating (Alsos et al., 2013; Bliss 1958; Cooper et al., 2004). It is slow growing, and its more erect growth form prefers areas that have sufficient protective snow cover in the winter, but soils that are well drained in the summer (Svalbard Flora, 2018).  Dryas integrifolia At all three investigated forelands, Dryas integrifolia was a late-successional or mature species. Other High Arctic studies agree that Dryas integrifolia is a mid- to late-successional species (Breen & Levesque, 2006; Okitsu et al., 2004; Mori et al., 2008), however it is suggested to be a pioneer species in calcareous gravelly and rocky sites (Au & Tardif, 2007; Aiken et al., 2007). It is a widespread mat forming species that prefers drier areas. While its seeds are well adapted to wind dispersal, germination is only average (Alsos et al., 2013; Müller et al., 2011; Klady et al., 2011; Bliss 1958; Cooper et al., 2004).  Carex nardina Carex nardina was clearly a mature species at Twin Glacier foreland. It appeared almost exclusively on post-LIA belts, where it was one of the most abundant graminoids. It was not !67present at Beistad Glacier, however appeared very infrequently in mid-successional stages at Teardrop Glacier. Other studies similarly noted this species as a later successional species (Okitsu et al., 2004; Mori et al., 2008; Moreau et al., 2008). It is characteristic of exposed dry areas with calcareous soils of low organic content (NA Flora, 2018; Aiken et al., 2007; Svalbard Flora, 2018). It reproduces only by seed. Few studies have investigated its germination success, with only Alsos et al. (2013) noting a 0% germination under optimal conditions.  2.5.3 ChronosequenceProblems in the interpretation of results from chronosequences have arisen when the key assumption has not been addressed (Johnson & Miyanishi, 2008). This assumption asserts that time is the only factor differentiating sites along the gradient, and therefore all sites have the same environmental history. One method of validating chronosequences is using long-term studies (Johnson & Miyanishi, 2008). If the results agree with what was expected from the original chronosequence study, then we can conclude that time is indeed the main factor driving change. The patterns observed in 2016 are for the most part what was expected. While the patterns of individual species proved to be more important than community patterns, all the species did indeed advance in a predictable manner toward that glacial margin. The patterns of some species at Twin Glacier (such as Cassiope tetragona and Chamerion latifolium) were not linear in their increases in abundance, suggesting that other environmental factors are also contributing to creating heterogeneous species distributions. The distribution of micro-sites could be one important factor, especially in the High Arctic (Cooper et al., 2004; Mori et al., 2013). However for the most part, we can confirm the interpretations and validity of using chronosequences on these three small, relatively environmentally homogeneous sections of three High Arctic glacial foreland. !682.5.4 Future ResearchRepeat surveys are valuable tools in ecology to characterize species change over time in light of climate change (Danby et al., 2011; Daniëls et al., 2011; Myers-Smith et al., 2011; Elmendorf et al., 2012b; Savage & Vellend, 2015), to investigate successional patterns (Moreau et al. 2009; Jírová et al., 2012; Buma et al., 2017), and to validate the use of other tools such as chronosequences (Johnson & Miyanishi, 2008). To maximize the validity of results from repeat surveys, careful consideration of survey methods is important (Moreau et al., 2009). The surveys on the foreland of Twin Glacier employed percent cover estimates, a method known to be observer-biased (Morrison, 2016). While we controlled for these errors as much as possible by having only one observer conduct the 2016 survey, there was no way of validating the cover estimates made by the observer in 1995 to the estimates made by the observer in 2016. To avoid these uncertainties in the future, those interested in repeat surveys would be better off using methods that are less observer-biased, such as presence/absence surveys (Morrison, 2016; Vittoz et al., 2010). Three possibilities for Arctic plant surveys are the methods employed in Matthews (1978), Moreau et al. (2009), and by the vegetation survey method we present in Chapter 3. These methods employ a sampling strategy based on a grid of presence/absence counts at various points across the foreland, giving a relative frequency of plant species at each point. As we have shown, species develop individualistically across a successional landscape, and to better understand patterns and drivers of succession, future research should focus on understanding the patterns and drivers of individual species. For example, it would be interesting to identify species that are early arrivers, and if these early arrivers are consistent regionally or across the Arctic. If so, what traits make them adept at growing in recently deglaciated areas? If not, why is there no regional consistency? Do the different early arrivers still have common traits? Research around plant traits has been popular in the past years (Kattge et al., 2011), but no research has been done about traits in High Arctic successional environments. 
!692.6 Conclusions In this chapter we observed that species advance along the successional gradient individualistically rather than as well defined communities. We have observed that succession proceed directionally, and that local species diversity is important in determining species patterns on a successional landscape. We have discussed the patterns of eight important species at our forelands, including: Luzula confusa, Papaver spp., and Saxifraga cernua which were important early- to mid-successional species; Saxifraga oppositifolia and Salix arctica which arrived early but grew slowly in size, becoming important on older terrain, and which thrived to differing degrees on the three forelands; and Cassiope tetragona, Dryas integrifolia, and Carex nardina which were late-successional to mature species. For the most part, the patterns observed in 2016 agreed with expectations following the 1995 survey, therefore confirming the chronosequence approach applied in these studies. !70CHAPTER 3 - MECHANISMS AT TWIN GLACIER FORELAND: EFFECTS OF TOPOGRAPHIC HETEROGENEITY ON SUCCESSION Mechanisms at Twin Glacier Foreland: Effects of Topographic Heterogeneity on Succession !713.1 Summary In this chapter we investigate the mechanisms influencing patterns of plant presence and abundance observed across a topographically heterogeneous High Arctic glacial foreland. We conducted a vegetation survey on the foreland of the western lobe of Twin Glacier, Alexandra Fiord, Ellesmere Island in the Canadian High Arctic (78°50’ N; 75°47’ W). The foreland is characterized by high and flat plateaus, steep slopes, river valleys, and gentle sloping lowlands. To describe the environmental variability, we also collected soil samples, soil nutrient measurements, recorded temperatures, and captured aerial photographs which helped create a DEM. Results from these surveys were mapped, and Hierarchical Modelling of Species Communities was used to help uncover mechanisms. We found that micro-environmental influences played the most important role, where variation in substrate type explained the largest amount of variation in vegetation patterns. Other influences deemed important included facilitation from moss and other vascular plant species, time since disturbance, intrinsic life history traits of plant species, and distance to seed source. While these conclusions provide interesting insights, the number of sites surveyed for both vegetation and soil properties were likely to few to uncover other important mechanisms structuring vegetation patterns across the foreland. Further research could improve conclusions by experimentally testing the hypothesized mechanisms proposed here. !723.2 Introduction Extrinsic and intrinsic biotic factors have been hypothesized to influence patterns of species composition. The first model of succession proposed by Frederic Clements (1916) focused heavily on the role of positive biotic interactions, suggesting that early successional species modified the environment in such a way as to make it favourable to species which dominated later. Since then, other modes of plant interaction during succession have also been acknowledged, including inhibitory competitive effects and neutral modes of mutual tolerance (Connell & Slayter, 1977). Intrinsic life history traits of plants have also proved to be important drivers of successional patterns (McCook, 1994; Chapin et al., 1994; Prach et al., 1997). Certain traits tend to favour early establishment, such as smaller seeds, long distance dispersal, rapid growth, and shorter life-spans, while other traits tend to favour later dominance, such as longevity and resistance to damage (Suter & Edwards, 2013; Řehounková & Prach, 2010; Prach et al., 1997; McCook, 1994; Chapin et al., 1994).  In addition to biotic drivers of community development, landscape and environmental suitability are important in determining successional trajectories. This is especially true in Arctic environments, where environmental resistance creates conditions that are marginal for life (Svoboda & Henry, 1987). Robbins and Mathews (2010) observed the effect of altitude on rates and pathways of succession in Norway. Dolezal (2008) and Mori et al. (2006) described the patterns of plant community development on different parts of Arctic glacial moraines. In Svalbard, Moreau et al. (2008) shed light on the influence of paraglacial runoff dynamics in creating heterogeneous forelands with horizontal discontinuities in successional patterns. At Glacier Bay, Fastie (1995) used dendrochronological evidence to show the influence of distance to seed source in creating three distinct pathways of succession. Jumpponen et al. (1999) and Mori et al. (2013) describe the importance of 'safe sites' in providing favourable micro-environments that facilitate early plant colonization, and suggest that the location of these sites can play a deterministic role in defining the distribution of plant species. And finally, Tilman’s !73(1985) resource-ratio hypothesis suggests that patterns of species abundance during succession is a function of their adaptation to different resources, especially to light and soil nutrients. Each species is a superior competitor for a certain amount of a limiting resource, and when the relative availability of that resource changes, so does the abundance of different species. While most researchers have focused their attention on uncovering deterministic mechanisms driving species distribution and community composition during succession, not all ecologists subscribe to their views. Lawton (1999) controversially argued for the abandonment of community ecology, stating “there are painfully few fuzzy generalizations, let alone rules or laws, and the necessary contingent theory looks unworkably complicated.” In this view, the search for mechanisms driving community composition during succession is futile, because the site specific differences are too numerous and therefore no generalizations can be made. Instead Lawton suggests observing patterns at either smaller or larger scales than the community. Taking an equally critical stance, Hubbell’s (2001) Unified Neutral Theory suggests that species are equivalent to each other in all important ecological respects, including probability of reproducing, dying, and migrating. As a result, species distribution is controlled not by deterministic niche preferences, but by stochastic processes such as ecological drift and random migration. Despite the controversy these positions created, they have contributed to an acknowledgment of the importance of stochastic processes. Modern community ecologists are finding a middle ground, acknowledging that both stochastic and deterministic processes shape the presence and abundance of species (Ovaskainen et al., 2017; Gravel et al., 2006; Leibold & McPeek, 2006). In the High Arctic, few studies have investigated primary succession on glacial forelands. However, most of them have discussed their interpretation of some of the driving mechanisms behind the patterns of plant species. Almost all studies have pointed to the importance of time since deglaciation in being the dominant driving mechanism (e.g. Jones & Henry, 2003; Okitso et al., 2004; Moreau et al., 2009; Hodkinson et al., 2003; Prach & Rachlewiz, 2012). However, other abiotic drivers have also been discussed, including paraglacial fluvial dynamics (Moreau et !74al., 2008), the presence of ‘safe sites’ (Cooper et al., 2004; Mori et al., 2013), nutrient inputs (Hodkinson et al., 2003) and the presence of organic rich paleo-soils (Jones & Henry, 2003). Biotic facilitation has also been observed, especially from biological soil crusts which often appear much earlier than vascular plants (Breen & Lévesque, 2006; Hodkinson et al., 2003). Negative competitive effects, while observed in less extreme Arctic environments (Chapin et al., 1994; Evlyn & Ryvarden, 1975), has not been observed in the High Arctic in successional environments. However, the role of stochastic processes in shaping species distributions has been noted, in particular random seed dispersal and disturbance events (Hodkinson et al., 2003; Moreau et al., 2008; Mori et al., 2008).   In this chapter we attempt to uncover the mechanisms driving patterns of plant presence and abundance across the topographically heterogeneous successional landscape of Twin Glacier foreland. We expand on the vegetation survey in Chapter 2 with another vegetation survey covering a much larger area of the foreland, and simultaneously collect information allowing us to create GIS layers describing various abiotic components of the landscape. We use hierarchical modelling of species communities to shed light on both biotic and abiotic processes occurring at various spatial scales. We aim to deepen our understanding of why we observe certain patterns of species presence and abundance. !753.3 Methods 3.3.1 Study Site The foreland under investigation borders the western lobe of Twin Glacier, Alexandra Fiord, Ellesmere Island in the Canadian High Arctic (78°50’ N; 75°47’ W). The glacier lies at the head of an 8 km2 valley, which is a known polar oasis (Freedman et al., 1994).  The valley boasts high plant cover and diversity (96 species of vascular plants), however its small area reduces the regular presence of larger herbivores such as muskoxen and caribou.  The rocky gneiss-granite glacial foreland is characterized by highly variable topography and two glacial streams (Figure 3.1; Figure 2.2). High and flat plateaus, steep slopes, river valleys, and !76Figure 3.1. Topographically heterogeneous foreland of Twin Glacier, characterized by plateaus, valleys, steep slopes, glacial streams, and gently sloping lowlands.gently sloping lowlands all make up part of the terrain. The foreland also has an abundance of well-preserved pre-Little Ice Age (LIA) plant communities and organically rich soil (Bergsma et al., 1984), which tends to support a higher cover and diversity of plants (Jones & Henry, 2003). 3.3.2 Vegetation Survey In late June and early July 2016, a total of 331 points were randomly surveyed by four observers in stratified sections (Figure 3.2c). The sections were chosen to represent different terrain types, including relatively flat and stable terrain, steep slopes, and valleys with glacial streams (Table 3.1). Three-hundred-and-one points were sampled in sections representing the recently deglaciated (post-LIA) foreland. These sections were further subdivided into sub-sections by lines running parallel to the glacier to get a more even distribution of points along the !77Section DescriptionF1a Narrow flat plateau near the SE margin of the glacierF1b Medium grade slope, SE facingF1c Very gentle slope, nearly flat, low-lying landF2 Small flat plateau, surrounded by three steep slopes and the glacierF3 Large flat plateau, surrounded by thee steep slopes and the NW margin of the glacierS1 Steep slope, medium grade in certain areas, N facingS2 Steep slope, S facingS3 Steep slope, NW facingS4 Steep slope, medium grade in certain areas, SE facingV1 Large flat valley, low-lying, with braided glacial streamV2 Narrow gently sloped valley, with glacial stream running underneath rocksctrl1 Beyond the LIA glacial maximum, on flat terrain beyond F1cctrl2 Beyond the LIA glacial maximum, on flat terrain in the river valley beyond V1ctrl3 Beyond the LIA glacial maximum,  on flat terrain beneath a steep slope separating it from F2ctrl4 Beyond the LIA glacial maximum, on flat terrain beyond F3Table 3.1. Foreland sections and a description of their characteristics (see Figure 3.2b for map).!78Figure  3.2.  (a)  Aerial  photograph  of  the  study area. (b) The study area was divided into sections based  on  topography,  where  yellow  =  relatively flat and stable terrain, purple = steep slopes, blue = valleys with glacial streams, and brown = control points (see Table 3.1 for full  descriptions).  (c) A total of 331 points were sampled, randomly chosen within sub-sections of the foreland. In addition, (d) temperature was recorded at 56 points, and (e) soil samples and nutrient measurements were collected at 34 points.d)a) b)c)e)chronosequence. An additional 30 points were sampled beyond the LIA glacial maximum to act as control points. The LIA maxima of Twin Glacier was interpreted as the extent of the lichen-kill zone, where rocks on the recently deglaciated foreland were largely devoid of darker colour crustose lichens characteristic of older rocks (Lévesque & Svoboda, 1999). QGIS (QGIS Development Team, 2018) was used to randomly assign points within polygons representing each of the sub-sections. Each point was at the centre of four 1 m x 1 m quadrats, each of which was further subdivided into a grid of 25 smaller squares (Figure 3.3). Within each of these smaller squares, the presence/absence of vascular plant species and moss was recorded. This gave us a frequency count for each plant species and for moss on a scale of 0 to 100 at each point. This method is a modified version of methods used by Matthews (1978) and Moreau et al. (2009), and provides a rapid assessment in High Arctic environments where vegetation is found in low density. Unlike !79Figure 3.3. Each of the 331 sampled points was at the centre of a grid of four 1 m x 1 m quadrats. Each quadrat  was  further  subdivided  into  25  smaller  squares,  inside  of  which  the  presence/absence  of vascular plant species, moss, dead vegetation, and substrates types was noted. This gave each attribute a value from 0 to 100 at each point.subjective percent cover estimation methods, this method relies on simple counts, making the results much more comparable between different observers. 3.3.3 Environmental Variables Substrate & Dead Vegetation Substrate and dead vegetation was characterized at each of the 331 points using the same grid of four 1 m x 1 m quadrats used in the vegetation survey, again as presence/absence counts in the 100 smaller squares (Figure 3.3). Substrate classes included boulders (> 30 cm in length), large rocks (10 - 30 cm), small rocks (0.5 - 10 cm), light coloured sandy fine sediment (< 0.5 cm), and dark colour organically rich fine sediment (< 0.5 cm). Dead vegetation included a combined class for standing dead (dead material attached to living vegetation as well as dead material rooted in the ground) and litter, as well as a class for relict paleo-vegetation as described by Bergsma et al. (1984). A spreadsheet containing records of these variables at each of the 331 points was imported into R and mapped with the help of the ‘sp’ package (Pebesma & Bivand, 2005; Bivand et al., 2013). Topographic Variables Aerial photographs were taken from a helicopter on a sunny and calm July 27, 2016. A Canon EOS 7D with a 15 mm lens in manual mode was used with the following settings: shutter speed = 1/800 seconds, aperture = f/7.1, ISO = 400, white balance = standard daylight setting. The focus was manually set to infinite, and images were captured in RAW. The first flight over the foreland was at an average elevation of 507 masl, and the second flight at an average elevation of 944 masl. Flight tracks were simultaneously captured with a Garmin GPSMAP® 64 device.  Locations were interpolated for each photograph from the flight tracks, and then used with the photographs to create a digital elevation model (DEM) using Agisoft PhotoScan (AgiSoft !80PhotoScan Professional, 2017), creating a raster with a 1 m x 1 m resolution. The coordinate system used for the DEM and all other mapping was NAD83 / UTM zone 17N. The DEM was used to calculate slope angle, slope aspect, and plan and profile curvatures in R using the packages ‘raster’ (Hijmans, 2016) and ‘sp’ (Pebesma & Bivand, 2005; Bivand et al., 2013). Further, it was used to calculate topographic wetness index (TWI) using the package ‘dynatopmodel’ (Metcalfe et al., 2017) and solar radiation using the package ‘insol’ (Corripio, 2014). Each of these calculations created a GIS raster layer that also had a 1 m x 1 m resolution, except for the TWI which due to slower computation times was at a resolution of 4 m x 4 m. These raster layers were mapped in R with the help of the ‘raster’ package (Hijmans, 2016). Distance to Glacier Distance to the glacier was used as a proxy for terrain age. The margin of the glacier was walked and recorded using a Garmin GPSMAP® 64 device during the summer of 2016. Distance was calculated from this vector line to each sampling point and raster pixel using the “gDistance” function in the ‘rgeos’ package in R (Bivand & Rundel, 2017). The distance raster was mapped in R with the help of the ‘raster’ package (Hijmans, 2016). Temperature HOBO Pendant temperature sensors were deployed from June 14 to July 30, 2016 at 52 locations across the foreland and control areas, mostly along transects leading away from the glacier (Figure 3.2d). Nine transects ran along flatter stable terrain, four ran along slopes, and two ran in the river valleys. The HOBO Pendants were placed on tent pegs inside of white plastic cups that acted as radiation shields, at a height of 8.5 cm from the ground.  The recorded temperatures at each location were averaged over the time deployed. The averages were used to model temperature across the foreland, where the predictor variables tested were distance to glacier, elevation, slope angle, slope aspect, and solar radiation. Variable selection !81sought to minimize the Akaike’s information criterion (AIC) and maximize the adjusted R squared. The linear model used for temperature took the form of: Temperature2 ~ log(Distance to Glacier) + Solar Radiation + Elevation   (3.1) An inverse distance weighted (IDW) interpolation of the residuals was performed using the ‘gstat’ package (Pebesma, 2004; Gräler et al., 2016) with the maximum distance set to 200 m. This improved the root mean squared error and the mean absolute error of the model after a leave-one-out cross validation (1.01 and 0.84 respectively for the model including IDW, versus 1.09 and 0.88 respectively without IDW). The model including the interpolation of the residuals was used to interpolate temperature across a 1 m x 1 m raster covering the same area as the DEM, thus making it another GIS layer. This raster layer was mapped in R with the help of the ‘raster’ package (Hijmans, 2016). Soil properties Soil samples were collected on July 7, 2016 at 34 locations across the foreland (Figure 3.2e). A small shovel was used to sample soil from the top 5 cm into a 185 cm3 can. Two samples were collected at all but one location, the first representing light coloured sandy soil, and the other dark coloured organically rich soil. Samples were weighed for their wet weight, air dried for a month in a heated cabin, and weighed again to determine their percent water content (soil moisture). Samples were then sent to the Analytical Chemistry Laboratory of the BC Ministry of Environment where percent organic matter (OM) content was determined using loss on ignition, along with total carbon and total nitrogen content. Soil moisture, OM, and total carbon values at each of the 34 locations were mapped with the help of the ‘sp’ package (Pebesma & Bivand, 2005; Bivand et al., 2013). To measure soil nutrients, four ion exchange membrane pairs (PRS®-probes, Western AG, Saskatoon) were placed at 33 location across the foreland on June 27, 2016. The probes were placed around the locations were soil samples were collected (Figure 3.2e). At each location, two !82probe pairs were placed in dark coloured organically soil, and two pairs were placed in light coloured sandy soil. The probes were retrieved on July 31, 2016, rinsed with deionized water, and sent to WesternAG, Saskatoon for analysis of soil nutrient availability. Ammonium, nitrate, and phosphorous availability values at each of the 34 locations were mapped with the help of the ‘sp’ package (Pebesma & Bivand, 2005; Bivand et al., 2013). Models for soil moisture and organic matter were developed. Predictor variables tested include distance to glacier (Dist), elevation, slope angle (Slope), slope aspect (Aspect), plan curvature, profile curvature, TWI, x and y geographic coordinates, a categorical variable classifying the sample locations according to the section of the foreland they belonged to (Locat), a binary variable classifying the sample as light coloured sandy soil or not (Sandy), and a binary variable classifying the sample as dark coloured organically rich soil or not. Variable selection sought to minimize the AIC and maximize the adjusted R squared, while also seeking to create a simple model defined by fewer predictor variables. The final linear model for soil moisture took the form of: Log(% Water) ~  log(Dist) + Sandy + y + Locat      (3.2) The final linear model for soil OM content took the form of:  Log(% OM) ~ log(Dist) + TWI + cos(Aspect) + log(Slope) + Sandy + x + Locat (3.3) Because both models include the variable Sandy, the models cannot be used to interpolate directly to the DEM raster, for which no information is available on soil type. However, the 331 points have an abundance count for each soil type. Therefore, both models were used to interpolate values directly to the 331 points. Estimates at these points were used to interpolated soil moisture and organic matter to a 1 m x 1 m raster across the foreland. These raster layers were then mapped in R with the help of the ‘raster’ package (Hijmans, 2016). !833.3.4 Vegetation Analysis Species Mapping Maps of the presence and abundance of individuals species and moss across the foreland were created. In addition, species richness and diversity was calculated and mapped. Richness was calculated as the total number of vascular plant species present at a point. Diversity was calculated at each point using the Shannon-Weaver diversity index in the ‘vegan’ package (Oksanen et al., 2017). Diversity is defined the same way as in Chapter 2 (equation 2.1). All vegetation maps were created in R with the help of the ‘sp’ package (Pebesma & Bivand, 2005; Bivand et al., 2013). Hierarchical Modelling of Species Communities Mechanisms driving patterns of species presence and abundance across the foreland were investigated using a hierarchical generalized linear mixed modelling framework developed for modelling spatially structured community data. This framework builds on previous work in species distribution modelling. Species distribution models aim to relate observations of species presence and abundance to the environmental characteristics of the observation locations (Elith & Leathwick, 2009). Their purpose is to provide a description of mechanisms influencing species distributions, and/or to predict species distributions across space or time (Elith & Leathwick, 2009). Here we are interested in the descriptive applications of modelling, which can be used to make data-driven hypotheses about the mechanisms driving succession.  However, standard species distribution models are effective at modelling only one species at a time. Thus, to apply them to community data three methods were proposed in a review by Ferrier & Guisan (2006). They include: (i) ‘assemble first, predict later’, where species data are first summarized by indices or ordination to community-level entities (such as species richness), and !84then these entities are modelled as a function of environmental variables; (ii) ‘predict first, assemble later’, where species are first modelled individually as a function of environmental variables, and then the individual models are summed to produce community-level conclusions; and (iii) ‘assemble and predict together’, in which all species are modelled as a function of environmental variables in a single integrated model. The third option originally referred more to constrained ordination and classification techniques, but additional methodological progress has led to joint species distribution models (Ovaskainen et al. 2017). Joint species distribution models assess the influences of environmental variables both at the species and community levels simultaneously, while also accounting for species co-occurrences (Ovaskainen et al. 2016b; Warton et al., 2015).  A recent addition to joint species distribution models has been the inclusion of latent variables to model species co-occurrences at different spatial scales (Ovaskainen et al. 2016a, Ovaskainen et al. 2016b). Statistical species co-occurrence patterns are created either by biotic interactions or by a shared response to an unmeasured environmental variable (Ovaskainen et al. 2016b), therefore investigating co-occurrences at different spatial scales can help uncover the reasons behind co-occurrence. Including spatially structured latent variables can also help account for spatial autocorrelation (Ovaskainen et al. 2016b), which was observed for most species that we surveyed. Inclusion of these spatially structured latent variables led Ovaskainen et al. (2017) to propose a framework based on hierarchical modelling, which they call Hierarchical Modelling of Species Communities (HMSC). The HMSC framework, as described by Ovaskainen et al. (2017), models the occurrence or abundance of species j at site i using a generalized linear model, such that yij ~ D ( Lij, σ2j )          (3.4) where D is the statistical distribution, Lij is the linear predictor, and σ2j is the variance associated with species j. In our study we modelled the presence and abundance of all 29 observed vascular !85plant species at 331 points across the foreland. We used a logit binomial model, where each small square in the grid was considered a trial, resulting in 100 trials for each sampling location. The linear predictor Lij was modelled with a set of fixed effects (F) that account for the different environmental variables, and a set of random effects (R) that include one spatially autocorrelated and two non-spatially autocorrelated random effects, such that  Lij = LijF + LijR .          (3.5) The fixed component is a function of the environmental variables, such that LijF = ∑ xikβjk           (3.6) where xik is the environmental variable k recorded at site i, and the βjk is the response of species j to the environmental variable k. In our model we include the following environmental variables: distance to glacier; relict paleo-vegetation; the five substrate classes (boulders, large rocks, small rocks, light coloured sandy fine sediment, and dark coloured organically rich fine sediment); moss; predictions from the models for solar radiation, temperature, soil moisture, and soil organic matter; and a slope-aspect interaction represented by 9 dummy variables (N, NE, E, SE, S, SW, W, NE, and flat). Because we suspected that some vascular plant species might have a unimodal rather than linear response to some of these environmental variables, we first scaled all the variables using the equation  Xscaled  =  ∑ (x - x̄)/σx         (3.7)                      and then we squared the scaled variables. All variables were squared, except the predictions from the four models (solar radiation, temperature, soil moisture, and soil organic matter), which showed a fair number of extreme values and were therefore sensitive to being squared. The slope-aspect interaction was also not squared. This fixed component models the overall !86prevalence of each species (the intercept), as well as the species environmental niches (the unique regression parameters for each environmental variable). To connect individual models to a community structure, we combined species-specific models into a single hierarchical model as described by Ovaskainen & Soininen (2011). Including this hierarchical structure improves the predictions for rare species, as the model learns to recognize the general response of species to a given environmental variable, and applies this response to species with very limited data, where such a response may not have been detected in a single-species model. This is done by organizing the regression coefficients βjk into a n X q matrix, where n is the number of species and q is the number of environmental variables. We then assume that the responses of species to environmental variables adhere to a multivariate normal distribution, such that βj. ~ N ( µ, V ) .          (3.8) The period in βj. singles out a row in the matrix, where βj. is a vector of q regression coefficients describing how species j responds to the environmental variables. The expected response of a typical species is described by the vector µ, and the variation around this expectation is captured in the variance-covariance matrix V. The second component of the model LijR  deals with random effects, which is where we investigate spatial structure. This second component models left over variation in species occurrences and co-occurrences that has not already been attributed to the response of species to environmental variables (Ovaskainen et al. 2017). In our model, we include three sets of latent variables, such that LijR  = ɛijP + ɛS(i)jS + ɛijA + ϵij         (3.9) !87where the random effects ɛijP and ɛS(i)jS model variation at the point and section levels respectively (Figures 3.2c and 3.2b), and ɛijA models variation attributed to spatial autocorrelation in the three dimensional space of latitude, longitude, and elevation. S(i) indicates the section in which point i is found. These random effects are modelled as ɛiP. ~ N (0, ΩP) ,          (3.10) ɛiS. ~ N (0, ΩS) , and         (3.11) ɛiA .~ N (0, ΩA)          (3.12) where ΩP, ΩS, and ΩA are residual species-to-species variance-covariance matrices. The off-diagonal elements of each matrix measure the covariance between two species at each spatial scale, assessing whether the two species occur more or less often then expected by chance, after the environmental variables have been accounted for. This HMSC method was implemented in R using the ‘HMSC’ package (Blanchet et al., 2018). To compare the relative importance of different predictors, explained variance was partitioned into variance attributed to environmental variables (fixed effects) and latent variables (random effects), and results were graphed. To measure the strength of species-to-species associations at different spatial scales, correlation matrices were calculated from the residual species-to-species variance-covariances matrices with the help of the function “corRandomEff” (‘HMSC’; Blanchet et al., 2018). Results were graphed with the help of the R package ‘corrplot’ (Wei & Simko, 2017). !883.4 Results 3.4.1 Environmental VariablesSubstrate & Dead Vegetation Boulders, large rocks, small rocks, and light coloured sandy fine sediment appear ubiquitously across the foreland, with boulders in higher abundance on slopes, and sandy fine sediment appearing more often near the glacier and on flatter terrain (Figure 3.4). Dark coloured organically rich fine sediment was less common, and often found in higher abundance on flat terrain, showing high abundance near the glacier in section F3 (for section locations and descriptions, see Figure 3.2b and Table 3.1). Relict paleo-vegetation was also found in high abundance near the glacier in section F3, and was less common on slopes and in river valleys.  Topographic Variables & Distance to Glacier Figure 3.5 presents the maps of distance to the glacier, slope angle, slope aspect, and solar radiation, providing context for the interpretation of species presences and abundances. Soil Properties (Figure 3.6) Soil moisture was highest near the NW margin of the glacier in section F3, and in valley V1 (Figure 3.6f). Total organic matter, total carbon, and total nitrogen were almost perfectly linearly related. All three were highest in section F3 (Figure 3.6d & 3.6e). Nitrate and ammonium availability were highly irregular. The highest nitrate availability was at a point in section F3 about 250 m from the glacier in organically rich soil, and two other points with higher availability were found about 50 m from the glacier, in section S2 in organically rich soil, and in section F3 in sandy soil (Figure 3.6a). The two points with highest ammonium availability were both found in section F3 about 10 m away from the glacier, in sandy soil (Figure 3.6b). The highest phosphorous availability was found in section F3 10 m from the glacier in sandy soil, !89!90d)  Fine Sediment - Sandya)  Boulders b)  Large Rocks c)  Small Rockse)  Fine Sediment - Organic f)  Relict VegetationFigure 3.4.   (a-e)  Presence and abundance of substrate classes and (f)  relict  paleo-vegetation as surveyed across the foreland. Coloured dots show a presence of the substrate class or relict  paleo-vegetation,  while faint  crosses show an absence.  Legends indicate the number of squares (out of 100) in which the substrate class or relict paleo-vegetation was found. !91Figure 3.5.Visualizations of (a) distance to the glacier, (b) slope angle, (c) slope aspect, and (d) solar radiation .c)  Slope Aspecta)  Distance b)  Slope Angled)  Solar Radiation!92d)  Organic Mattera)  Nitrate b)  Ammonium c)  Phosphorouse)  Total Carbon f)  Soil MoistureFigure 3.6. Available soil nutrients including (a) nitrate, (b) ammonium, and (c) phosphorous as determined by the ion exchange membranes. Nutrients availabilities are measured in supply rates (micro grams/10cm2 over 26 days). Percent (d) organic matter, (e) total carbon, and (f) soil moisture, as measured from the collected soil samples. while the second highest was found in section F2 in organically rich soil (Figure 3.6c). There were no general trends in any of the nutrient availabilities, as values seemed to be low overall, and quite variable. Soil and Temperature Models (Figure 3.7) The temperature model was significant (p < 0.01) and had an adjusted R squared of 0.51. The coldest predicted locations were near the glacier, with the head of valley V1 being particularly frigid. North facing slopes were also predicted as being colder. Warmer areas were predicted to be in section F1c (for section locations and descriptions, see Figure 3.2b and Table 3.1).  The soil moisture model was significant (p < 0.01) and the adjusted R squared was 0.74. However, the residuals for the model showed heteroscedasticity, begging caution when interpreting model significance. Soil moisture was predicted to be the highest in section F3, and on slope S1. The driest predicted areas were in section F1. The OM model was significant (p < 0.01) and the adjusted R squared was 0.66. The areas with the highest predicted OM were in the control sections ctrl1 and ctrl2. On the foreland, OM was predicted to be slightly higher on section F3, and along the margins of valley V1. 3.4.2 Vegetation MapsSpecies Richness & Diversity Species richness and diversity tended to co-occur (Figure 3.8). Highest richness and diversity occurred in sections F1 and F3 on the foreland, generally increasing with distance from the glacier up to a certain point. Control points were less rich and diverse than points on the foreland. On section F3, high species richness and diversity were also found very close to the glacier, in the same area where higher soil moisture and nutrients were also noted. This particular !93!94Figure  3.7.  Interpolated  predictions from (a) temperature, (b) soil moisture, and  (c)  soil  organic  matter  models. Errors  are  shown  at  each  sampling point. Errors for the temperature model are from leave-one-out cross validation.c)  Organic Mattera)  Temperature b)  Soil Moisturearea boasting both favourable soil parameters in conjunction with diverse vegetation was not investigated in depth by our study, but could be interesting to re-visit.  Vascular Plant Species and Moss The three most abundant species were Papaver spp., Luzula confusa, and Salix arctica. Papaver spp. (Figure 3.9i) were most abundant on section F3, although they were also common but less abundant on F1 and F2 and in valley V1 and slope S1 and S4. Luzula confusa (Figure 3.9m) was by far most frequently observed on section F1, slope S1, and valley V1, all of which are connected to each other and at lower elevations. It was much less frequent on higher plateaus, including F1a and F2, and while it was more frequently observed on plateau F3, it was still lower in abundance than on F1. Salix arctica (Figure 3.9c) was most abundant on older parts of sections F1c, S1 and V1. It is also present in smaller abundance in section F3, where it is found remarkably close to the glacial front. It is interesting to note that the locations were it is found in close proximity to the glacier correspond to the area of abnormally favourable soil parameters and diverse vegetation. Salix arctica was almost completely absent from plateau F2, on slopes S2, S3, and S4, and in valley V2. The presence of Salix arctica tended to be clustered together. Luzula confusa and Papaver spp. were more common on the recently deglaciated portion of the foreland, while Salix arctica was found both on the foreland and at control points. Some species were more frequent at control points beyond the LIA glacial maximum of Twin Glacier. These species include Carex nardina (Figure 3.9l), Cassiope tetragona (Figure 3.9a), and Dryas integrifolia (Figure 3.9b). Out of these three species, Cassiope tetragona appeared to be the most adventurous in making its way onto the recently deglaciated foreland, and was found in low abundance and frequency on sections F1, S1, V1, F2 and F3.  Different Draba species were also frequently observed across the foreland, however, they had different patterns. Draba lactea (Figure 3.9e) was found less frequently, and mostly on section F1c. Draba nivalis (Figure 3.9f), on the other hand, was found across almost all sections, and found in highest abundances at points on section F3. Unidentified Draba species (Figure 3.9g) !95!96Figure 3.8. (a) Shannon’s diversity index and (b) species richness (in number of species) at the 331 points across the foreland. Coloured dots show a presence of plant species, while faint crosses show an absence.a)  Shannon’s Diversity b) Species Richness!97c)  Salix arcticaa)  Cassiope tetragona b) Dryas integrifoliad)  Moss Figure 3.9 a-d. Presence and abundance of (a-c) different shrub species and (d) moss across the foreland. Coloured dots show a presence of the species or moss, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species or moss was found.!98g)  Draba spp.e)  Draba lactea f) Draba nivalisFigure 3.9 e-g. Presence  and  abundance  of  Draba species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends  indicate  the  number  of squares  (out  of  100)  in  which  the species was found.!99j)  Saxifraga cernuah)  Minuartia rubella i) Papaver spp.k)  Saxifraga oppositifoliaFigure 3.9 h-k. Presence and abundance of different forb species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species was found.!100Figure 3.9 l-n. Presence  and  abundance  of  different graminoid  species  across  the  foreland. Coloured  dots  show  a  presence  of  the species,  while  faint  crosses  show  an absence. Legends indicate the number of squares (out of 100) in which the species was found.n)  Poa arctical)  Carex nardina m) Luzula confusawere also broadly scattered across all sections, again most abundant at points on section F3. All Draba species were rarely found at control points. Three other forbs and one other graminoid were also amongst the more common plants across the study area. Minuartia rubella (Figure 3.9h) was found infrequently across sections F1, F2, F3, S4, and at control points. Poa arctica (Figure 3.9n) was found in similar areas as Luzula confusa, however less frequently. Saxifraga cernua (Figure 3.9j) was most common on section F1, but also found on slope S1, valley V1, and plateau F3. It was not observed at control points. Saxifraga oppositifolia (Figure 3.9k) was found in higher abundance and frequency at control points, however it was also present regularly in sections F1, S1, V1, F2 and F3.  Maps of the less abundant vascular plant species are presented in Appendix B. Moss (Figure 3.9d) was present almost ubiquitously across the foreland, except for on very steep slopes and at a few points adjacent to the glacier. In sections F1 and V1, its abundance appears to increase with distance from the glacier, while in section F3 it is frequently found in high abundance near the glacier.  3.4.3 Hierarchical Modelling of Species CommunitiesVariance Partitioning The influence of different environmental variables and random effects on species presence and abundance is described in Figure 3.10. Because models of more common species have more data points to determine the relationships, the results of the variance partitioning for the six most common species are further presented in Table 3.2.  !101!102Figure  3.10.  The results of the HMSC variance partitioning. Variation in species occurrences is partitioned into responses to environmental variables and random effects. The bars showing species-specific responses are ordered according to decreasing prevalence of species from left to right. The number of points at which species are observed is recorded in the numbers next to the species names.  The legend records the average responses across all species to environmental variables (and to their scaled squares) and to random effects.The most important environmental variable influencing species occurrences was substrate type. This was a grouped class describing the response of species to the five different substrate classes. Substrate type described 27.8% of the total explained variation when considering all species, and 19.6% considering the six most common species. Other important environmental variables included distance to the glacier (15.1% all species; 16.5% six most common species) and the presence of moss (8.6%; 13.1%). Predictions from all four models (including solar radiation, temperature, soil moisture, and soil organic matter) described very little of the variation (a total of 7.5% for all species; 5.4% for the six most common species). The slope-aspect interaction was not very important (5.1%; 3.0%). Presence of relict vegetation was less important for more common species (3.5% six most common species), however it was more important for describing the variation in rare species (describing 9.1% of the variation considering all species). !103Papaver spp.Luzula confusaSalix arcticaPoa arcticaDraba nivalisSaxifraga oppositifolia TOTALDistance 0.9% 11.8% 19.1% 3.9% 5.2% 10.4% 8.6%Distance2 21.7 8.2 1.3 2.1 10.7 3.6 7.9Slope*Aspect 3.4 2.0 2.7 2.8 3.2 3.9 3.0Relict 1.6 1.1 0.8 2.3 4.2 1.1 1.9Relict2 1.6 4.3 0.6 1.0 1.0 0.8 1.6Substrate 4.8 10.0 25.5 10.3 13.1 12.4 12.7Substrate2 4.2 3.5 5.9 6.2 6.7 14.7 6.9Moss 10.6 12.5 9.3 13.1 8.0 5.3 9.8Moss2 7.0 0.6 0.6 2.1 5.3 4.0 3.3Temperature 3.5 0.6 0.9 0.9 1.3 1.1 1.4Solar Radiation 0.5 0.4 0.5 0.7 1.3 1.0 0.7Soil Moisture 4.5 2.8 0.5 0.6 1.1 1.4 1.8Soil OM 0.5 0.4 1.5 0.6 1.4 4.6 1.5Latent, point 33.8 15.6 28.3 23.5 35.0 28.4 27.4Latent, section 0.5 22.9 0.7 4.6 0.2 3.2 5.4Latent, auto 0.7 3.2 1.6 25.4 2.3 4.2 6.2Table 3.2. Results of HMSC variance partitioning for the six most common species. Explained variance is partitioned into responses to environmental variables (and to their scaled squares) and to random effects. The column total describes the average results including only these six species.Observing the differences between scaled environmental variables and their squares gives us clues about the type of influence the variable has on species occurrences. For example, Salix arctica was better described by distance than by distance squared, which implies that it has a linear response to this variable. As distance from the glacier increases, so does the presence and abundance of Salix arctica (Figure 3.9c). Other species dominated by this linear response include Draba lactea, Minuartia rubella, Cassiope tetragona, Dryas integrifolia, Carex nardina, Luzula nivalis, and Potentilla hyparctica. On the other hand, certain species such as Papaver spp. were better described by distance squared, which implies that this species has a unimodal response to this variable. As distance from the glacier increases, the presence and abundance of Papaver spp. at first increase, and then decrease towards further distances (Figure 3.9i). This unimodal response to distance dominates for a few other species as well, including Draba nivalis, unidentified Draba species, Saxifraga cernua,  Saxifraga nivalis, and Saxifraga cespitosa.  The spatially structured random effects played a significant role in describing the variation in species occurrences (26.9% all species; 39.1% six most common species). By far the most important random effect was at the scale of individual points (19.3%; 27.5%).  Species associations Species-to-species associations calculated in the HMSC framework measure the residual correlations between species pairs after having accounted for the variation caused by environmental variables. These residual associations can be explained either by biotic interactions between species (e.g. competition for negative associations, or facilitation for positive associations), or by a shared response to a missing environmental variable (Ovaskainen et al. 2017). Here we present the results of species associations at three spatial scales. At the scale of individual points, species association were predominantly positive (Figure 3.11). This suggests that species were either simultaneously present or absent from individual points. This response could be due to positive biotic interactions, but because overall cover of species is !104so low, it is likely that this pattern also indicates a shared response to missing environmental variables that play out at this fine scale. At the scale of sections (Figure 3.2b), species association showed a clear structure, where they were sorted into three groups (Figure 3.12). The first group includes Saxifraga cernua, Draba lactea, Poa arctica, Melandrium apetalum, Salix arctica, Papaver spp., Cerastium alpinum, Stellaria longipes, Erysimum pallasii, and Luzula confusa. These species tended to occur in the !105Figure  3.11.  Species-to-species  association  matrix  showing  residual  correlation  at  the  scale  of individual points. Species are sorted to better illustrate the relationship among pairs of species. The strengths of positive (blue) and negative (red) associations are shown.same sections, but not in the sections where species from the second group were found. The second group of species includes Saxifraga oppositifolia, Cassiope tetragona, Dryas integrifolia, Carex nardina, Silene acaulis, Saxifraga tricuspidata, Minuartia rubella, Cardamine bellidifolia, Saxifraga cespitosa, unidentified Draba species, Vaccinium uliginosum, Chamerion latifolium, Festuca brachyphylla, Potentilla hyparctica, and Saxifraga nivalis. These species tended to occur in the same sections, but not in the sections where species from the first group were found. The first group is made of up species more common on younger terrain, while the second group !106Figure 3.12. Species-to-species association matrix showing residual correlation at the scale of sections. Species are sorted to better illustrate the relationship among pairs of species. The strengths of positive (blue) and negative (red) associations are shown.of species consists of species found in more mature communities. The third group of species occurred independently from other species, and included Draba nivalis, Carex aquatilis, Arctagrostis latifolia, and Luzula nivalis. Draba nivalis was found quite ubiquitously throughout the foreland, while the other three species had very low occurrences. Species associations attributed to spatial autocorrelation showed weaker patterns, however species were also sorted into three groups (Figure 3.13). Species in the first group tend to appear close to each other, but not close to species in the second group. This first group includes Draba lactea, Cerastium alpinum, unidentified Draba species, Saxifraga cernua, Saxifraga !107Figure 3.13.  Species-to-species  association matrix showing residual  correlation attributed to spatial autocorrelation.  Species  are  sorted  to  better  illustrate  the  relationship  among pairs  of  species.  The strengths of positive (blue) and negative (red) associations are shown.oppositifolia, Salix arctica, and Carex nardina. The second group includes Poa arctica, Luzula confusa, Chamerion latifolium, Saxifraga cespitosa, Saxifraga tricuspidata, Festuca brachyphylla, Draba nivalis, Vaccinium uliginosum, and Saxifraga nivalis. Species in this group tended to occur in points close to each other, but not close to species in the first group. The third group of species (the remaining species) tended to occur independently of other species. 
!1083.5 Discussion This study aimed to shed light on the different factors influencing the presence and abundance of plant species across an environmentally heterogeneous High Arctic glacial foreland. We first discuss the mechanisms observed influencing vegetation patterns, before turning to a discussion of limitations of the study and suggestions for future research. 3.5.1 Mechanisms influencing succession Micro-environmental influences The two most important variables describing the variation in presence and abundance of plant species were substrate type and the random effects associated with point-scale variation. These variables point to the importance of local micro-environmental influences. Variation in the presence and abundance of boulders, large rocks, small rocks, sandy fine sediment, and dark fine sediment was useful in explaining the variation in presence and abundance of all plant species, emphasizing the importance of micro-topography and soil type in influencing patterns of plant growth. Meanwhile, the patterns of residual variation at the scale of individual points was more important than at the broader section scale or than patterns related to autocorrelation. The species-to-species correlations at this scale were predominantly positive, which suggests that if a species occurs at a point, it is likely that other species will also occur at that point. This implies that certain points are simply more favourable for plant growth than other points. As suggested earlier, it is likely that this pattern of co-occurrence is due at least in part to missing micro-environmental variables at this finer scale. These results agree with others who have pointed to the influence of micro-topography on patterns of vegetation development during primary succession (Burga et al., 2010; Cutler, 2010; !109Cannone et al., 2004; Whittaker, 1989). In addition, the distribution of 'safe sites' has been discussed by many as an important determinant of vegetation patterns (Harper et al., 1965; Cooper et al., 2004; Jumpponen et al., 1999; Mori et al., 2013; Marteinsdóttir et al. 2010). Safe sites have been described as concave surfaces, proximity to large rocks, coarse substrate, or vicinity of other vegetation (Jumpponen et al., 1999; Erschbamer et al., 2001). Other environmental variables that could play out at this small scale include differences in soil moisture and nutrient availability. Facilitation The presence and abundance of moss also had a strong influence on vegetation occurrence patterns. Covariance between moss and different vascular species revealed that this influence was mostly positive, with the exception of three species that had small negative covariances (Dryas integrifolia, Carex nardina, and Vaccinium uliginosum, all of which are found mostly at control points). This strong positive influence of moss on species occurrences suggests that moss could be facilitating the presence of vascular plant species. This is in agreement with a High Arctic study by Gornall et al. (2011), which found that shrub and forb biomass were higher when there was a shallow moss layer than when they were in bare soil. They suggest this was due to the moisture retaining effects of moss, as they found that soil with a shallow moss layer had higher soil moisture than bare soil. This positive effect was not found for graminoid species however, and if the moss layer was too deep all plants responded with lower biomass. However, we observed the moss layer across the recently deglaciated (post-LIA) foreland to be less developed and thus shallow. In addition, facilitation might also be occurring between different species of vascular plants. Positive species associations at the scale of individual points might be due to a shared response to environmental variables as discussed above, but it could also be due to positive biotic interactions (Ovaskainen et al., 2017). Facilitative interactions between vascular plant species have long been presented as an important mechanism influencing succession (Clements, 1916; Connell & Slayter, 1977;  Glenn-Lewin, 1992; Chapin et al., 1994; Pulsford et al., 2016). !110Time since disturbance Distance from the glacier, while not the dominant influence, still had an important effect on vegetation patterns (Figure 3.10). The response was not the same for all species, with certain species showing a linear response of increasing abundance with distance from the glacier, while others showed a unimodal response. Species associations at the scale of sections also points the importance of distance from the glacier in structuring plant communities. Certain earlier successional species were indeed more commonly found in the same sections, but not in the sections containing more mature species, which themselves were found together in other sections.  Time since disturbance is the most frequently cited mechanism for structuring species occurrences in studies of succession in the High Arctic (Matthews, 1992; Jones & Henry, 2003; Moreau et al., 2009; Hodkinson et al., 2003; Prach & Rachlewiz, 2012; Okitso et al., 2004). While it is important here, it does not appear to be the dominant factor explaining variation in presence and abundance of species. Life history traits Individual species graphs, as well as the section-scale species association matrix, suggest that species are either more common on the recently disturbed foreland or in older control sections. For example, Papaver spp., Luzula confusa, and Draba nivalis are all found broadly across the recently deglaciated (post-LIA) foreland, but are very rare beyond the LIA maximum at control points. Other species including Cassiope tetragona, Carex nardina, and Dryas integrifolia are found almost exclusively at control points. This suggests certain species could be disturbance specialist while others are slower growing and dominant only in older mature stages. Intrinsic life history traits of plants are commonly recognized as mechanisms determining species occurrences (Svoboda & Henry, 1987; McCook, 1994; Chapin et al., 1994; Prach et al., 1997) and are discussed in more detail in Chapter 2. !111Distance to seeds The clustering of species observed in certain species maps, as well as the fact that most species occurrences showed autocorrelation, suggests that individuals of a species are often found close to each other. This is demonstrated particularly well by Salix arctica, which appears to cluster together in two main areas on the foreland. While there are many reasons clustering of individuals might occur, such as in response to a biotic or abiotic variable, a likely explanation could also be that seeds do not disperse far from their origin. This would agree with results from Elven and Ryvarden (1975) who, studying a glacier foreland in Norway, found that 87% of diaspores came from a distance of less than five metres. Others have also pointed to the importance of seed availability and distance to seed source in determining the patterns of species presence and abundance on glacial forelands (Fastie, 1995; Erschbamer et al., 2001; Jones, 1997). In addition, studies that demonstrate nucleation also provide evidence of the influence of distance to seed source (Cutler et al., 2008; Yarranton & Morrison, 1974; Blundon et al., 1993). Nucleation occurs when an individual plant succeeds at establishing itself in a safe micro-site on a recently disturbed landscape, and thereafter aids in the further development vegetation by either acting as a source of seeds, or by improving the micro-climatic conditions so as to allow for the establishment of other plants (Cutler et al., 2008).  3.5.2 Limitation of the studyWhile the results point to many interesting conclusions, these conclusions likely fall short of the full story. Sampling effort was probably too small to capture many of the important environmental influences characterizing species occurrence patterns. Qualitative observations made by team members did not always align with the results.  Very steep slopes were observed to be dominated by boulders and lack vegetation, except for the occasional Papaver spp. However, while the number of points sampled on steep slopes was !112proportional to the area they covered, the total number sampled on steep slopes was very small. As a result, the influence of slope angle was likely underrepresented in the results. Similarly, the number of soil samples collected was likely not enough to make accurate conclusions about the influences of soil moisture, OM, and nutrients. Although the soil moisture and OM models were significant and had acceptable adjusted R squared values (0.74 and 0.66 respectively), they likely failed to capture the micro-environmental variability of these factors. We hypothesize that soil moisture in particular likely had a more important role in shaping species occurrences than observed in our results. The soil moisture and OM models could have benefited from the collection of additional soil samples, or from measurements made at each point surveyed for vegetation. However, this is time consuming. Soil nutrients likely also played a role, but were not addressed here. Although models for ammonium, nitrate, and phosphorus were made from the points sampled for nutrients, these models were so poor that they were not included in the results, nor were they used to address variation in species occurrences.  3.5.3 Future Research This study uses correlations to describe the importance of different mechanisms influencing succession. While correlative vegetation models based on observational studies are useful for generating testable hypotheses, they are unable to infer causation. Therefore, an obvious next step would be to experimentally test the potential mechanisms described in this study.  Another interesting pattern that begs further investigation is the area on section F3 that boasts unusually high soil nutrients and species richness. This area was also noted qualitatively by observers as being particularly productive and diverse considering its proximity to the glacier. Too few soil samples were collected to reliably assert claims of higher nutrients in this area, and even if higher nutrient availabilities were the case, it is not understood why. Understanding why !113this area is a “hotspot” could help uncover mechanisms driving and limiting species advances across successional landscapes.  And finally, a further investigation into patterns of seed dispersal across topographically heterogeneous glacial forelands could potentially lead to interesting insights. A spatial survey of the occurrences of individuals could help visualize whether migration pathways exists, and why species do or do not end up growing in certain areas. 
!1143.6 Conclusions In this chapter we have attempted to describe the mechanisms driving patterns of plant presence and abundance across the topographically heterogeneous successional landscape of Twin Glacier foreland. That is, we have tried to explain why we observe certain vegetation patterns. Vegetation maps and Hierarchical Modelling of Species Communities gave us some insights. We found that micro-environmental influences played the most important role. Variation in micro-topography and soil type explained a large amount of variation in vegetation patterns, as did the random effects representing point-scale spatial structure. Facilitation was also hypothesized to be important, as the presence and abundance of moss related positively to vascular plant occurrences. In addition, point-scale species associations were predominantly positive, which could be due in part to positive species interactions. Time since disturbance also proved to be important, however it was not the dominant factor. Species responded differently to this variable, with some species increasing linearly as time progressed, while other species showed unimodal responses. Intrinsic species life history traits likely also play a role in determining patterns, as species occurrences were divided between presence on recently deglaciated sections of the foreland or presence in mature sections beyond the LIA glacial maximum. Lastly, distance to seed source might also play a role in creating clustered species distributions across the foreland. While our data point to these mechanisms as being important, limitations in the number of sites surveyed led us to conclude that there were likely other mechanisms also structuring species occurrences that we were not able to capture. Future research could provide new insights by experimentally testing the hypothesized mechanisms proposed here.  
!115CHAPTER 4 - SECONDARY SUCCESSION AT ALEXANDRA FIORD Secondary Succession at Alexandra Fiord !1164.1 Introduction The first two chapters discussed patterns (Chapter 2) and mechanisms (Chapter 3) of primary succession at Twin Glacier foreland, Alexandra Fiord. In this final chapter we move down from the foreland and into the valley to examine secondary succession. Secondary succession differs from primary succession in that the disturbed area has some amount of biological legacy from the pre-disturbance biotic communities.  Previous research of secondary succession in the Arctic has described some common patterns and mechanisms. The largest factor determining pathways during secondary succession is the extent and severity of the disturbance - how much of the plant cover is removed and over how large an area. If some plant cover remains, it is usually restricted to more resistant species, such as Salix arctica, Dryas integrifolia, or Saxifraga oppositifolia in the High Arctic (Babb & Bliss, 1974). Salix arctica was shown by Babb and Bliss (1974) to be the most resistant, requiring only the roots and several buds to remain intact for the individual plant to continue growing. These remaining plants steer succession thereafter. If no plant cover remains, other forms of biological legacy may still help speed up succession, such as rich soils, soil microfauna, and a seed bank. In such cases, rapid invaders in High Arctic secondary succession include graminoids, Bistorta vivipara,  Saxifraga  spp., Draba  spp., and Papaver  spp.  (Babb & Bliss, 1974; Cannone et al. 2010). Besides plant cover, moisture is another factor that has been shown to affect secondary succession. Studies support the hypothesis that moisture availability aids in speeding up the process of succession (Cargill & Chapin, 1987; Forbes, 2001). Warmer summer temperatures may also reduce the time needed for recovery (Cannone et al., 2010). In this chapter we re-survey a site of secondary succession in the Alexandra Fiord glacial valley. We compare our results with similar surveys of the surrounding mature tundra and with the results from the vegetation survey on the Twin Glacier foreland presented in Chapter 2. We ask two questions: do differences persist in vegetation patterns and soil properties between the !117disturbed farm area and the surrounding mature tundra after 31 years; and, how do the patterns of secondary succession at Alexandra Fiord differ from the patterns of primary succession? !1184.2 Methods 4.2.1 Study SiteIn 1982 at Alexandra Fiord, the ‘Green Igloos’ Experimental Farm was established to test the feasibility of growing food crop in the High Arctic (Figure 4.1a, 4.1b, 4.1c). In developing the farm, the ground was stripped of natural vegetation and the soil layer was disturbed. In 1985, the farm was dismantled and a revegetation experiment was initiated with two different treatments: fertilization and surface roughness (Figure 4.1d). Two surveys have been conducted since then. In 1988 density of plant species was surveyed in plots with different treatments. In 2000, the growth and reproductive output of Papaver spp. was observed (Chan, 2001). The original setup of the experiment and a summary of results from the 1988 survey can be found in Appendix C.  This disturbed farm area was situated on the flat, slightly elevated bank of a small braided river flowing from Twin Glacier, through the valley, and into Alexandra Fiord (Figure 4.2). The 8 km2 glacial valley of Alexandra Fiord is considered a polar oasis relative to the surrounding uplands, with high plant diversity (92 vascular plant species), cover, and production (Freedman et al., 1994). It was deglaciated about 8 - 9 ka BP, and the soils have a relatively high concentration of organic matter despite being young and poorly developed (Freedman et al., 1994). 4.2.2 Vegetation and Soil SurveySurvey plots were established in 2016 as seen in Figure 4.3a. This included 17 plots in the disturbed farm area, and three plots in the surrounding undisturbed tundra. Within each plot, three 50 cm x 50 cm quadrats were surveyed (Figure 4.3b). Percent cover of vascular plant species was estimated visually, along with percent cover of moss, standing dead vegetation (both !119!120Figure 4.1. (a-c) ’Green Igloos’ experimental farm in 1982, and (d) the revegetation experiment established thereafter in 1985. Photographs credit of Greg Henry. a) b)  c) d)!121Figure 4.2. (a) Arial photograph of the Alexandra Fiord glacial valley in July 2016, with the location of the experimental farming site indicated by the yellow star. (b, c) Photographs of the study site in August 2016. The revegetation study sign post can be seen lying on the ground in photograph b. a) c) b)dead material attached to living vegetation as well as dead material rooted in the ground), litter, and substrate in five classes (fine sediment (<0.5 cm), pebbles (0.5-3 cm), small rocks (3-10 cm), large rocks (10-30 cm) and boulders (>30 cm)). In addition, a presence/absence survey of vascular plant species was conducted of the entire plot in a survey lasting approximately 10 minutes.  Two soil samples were collected from each of the 17 plots in the disturbed farm area on July 8, 2016. On the same date two samples were also collected from four locations in the surrounding mature tundra, shown in Figure 4.3a. The top 6 cm of soil was sampled using a soil corer 6 cm in diameter. Samples were weighed for their wet weight, air dried for a month in a heated cabin, and weighed again to determine their percent water content. Samples were then sent to the Analytical Chemistry Laboratory of the BC Ministry of Environment where total carbon and total nitrogen were measured, and percent organic matter (OM) content was determined using loss on ignition. In the same plots and mature tundra locations that soil samples were collected !122 a) b)Figure 4.3. (a) The arrangement of the plots (yellow squares) and soil sample locations (green dots) surveyed in 2016. This included 17 plots in the disturbed farm area (1-17), and three plots in the surrounding undisturbed tundra (C1-C3). The adjacent river is highlighted in blue. (b) Three 50 cm x 50 cm quadrats were surveyed in each plot.from, soil nutrient availability was measured using two ion exchange membranes (PRS® probes, Western Ag, Saskatoon) that were inserted and remained in the soil from July 7, 2016 to August 2, 2016. The probes were rinsed with deionized water and sent to WesternAG, Saskatoon for analysis of soil nutrient availability.  4.2.3 Data analysisVegetation, litter, substrate cover as well as soil parameters were averaged across: (1) the 17 plots in the disturbed farm area, and (2) the three plots in the surrounding undisturbed tundra. The averages of percent cover were standardized so that their total added up to 100%. Vegetation, litter, and substrate cover from Belts 80 and 90 at Twin Glacier (Chapter 2) were also averaged and standardized. These two belts represented areas that are estimated to have been deglaciated around 1982-1988, based on posts marking glacial retreat (Figure 2.6). These averages were visualized with the help of Tableau (Tableau Desktop, 2017). In addition, visualizations were also made showing only vegetation cover (vascular plants and moss). Means and 95% confidence intervals for all soil variables were calculated. A Permutation Multivariate Analysis of Variance (PERMANOVA; Anderson, 2001) was used to formally test whether cover in the disturbed farm area differed from that in the surrounding undisturbed tundra. The PERMANOVA was run using the “andois” function from the ‘vegan’ package (Oksanen et al., 2017) in R (R Core Team, 2017), with Bray-Curtis distances and 9999 permutations. In the first test we investigated the significance of differences in cover of vascular plant species, moss, substrate, litter and standing dead. In the second test we investigated the significance of difference in cover of only vascular plant species and moss. 
!1234.3 Results 4.3.1 Disturbed Farm Area vs. Mature TundraThe PERMANOVA indicates a statistically significant (p < 0.05) difference in cover between plots in the disturbed farm area compared to the surrounding mature tundra (Table 4.1). Differences were significant in both tests, suggesting that both the total cover and plant community composition differ between disturbed and mature plots. This implies that the disturbed farm area has not yet reached a maturity after 31 years of recovery.  To describe the differences between the disturbed farm plots and the mature tundra plots in more detail, we turn to the visualization in Figures 4.4 and 4.5. Figure 4.4 shows that the majority of cover (52.5%) in the disturbed farm plots was standing dead vegetation, which was much less important in mature tundra plots (8.3%). While we did not identify the species of the standing dead vegetation as part of the survey, we observed that most of it was graminoids. The most abundant cover in the mature tundra plots was litter (28.8%), which we observed as being mostly dead leaves of Salix arctica. Litter was also an important cover at the disturbed farm plots, with !124Table 4.1. PERMANOVA testing the significance of the difference in vegetation, litter, and substrate (Substrate and Plants), and in vegetation only (Plants only), between the disturbed farm area and the surrounding undisturbed tundra.DF MS F model pSubstrate and PlantsDisturbed 1 1.77 33.63 0.0001Error 58 0.05Plants onlyDisturbed 1 0.65 5.85 0.0003Error 58 0.11!125Figure 4.4. The average standardized percent cover of vegetation (red = shrubs, yellow = graminoids, green = forbs, blue = moss), substrate (grey), standing dead (beige), litter (beige) and relict vegetation (beige, only present at Twin Glacier) on Disturbed Farm and Mature Tundra plots, and on Belts 80 and 90 at Twin Glacier foreland (see Chapter 2 for description of Belts). !126Figure 4.5. The average standardized percent cover of vegetation (red = shrubs, yellow = graminoids, green = forbs, blue = moss) on Disturbed Farm and Mature Tundra plots, and on Belts 80 and 90 at Twin Glacier foreland (see Chapter 2 for description of Belts). the third highest cover (12.4%). Salix arctica was the most abundant species in both areas, although it has a higher cover the mature tundra plots (28.8%) compared to the disturbed farm plots (14.4%). Moss was the second most abundant vegetation cover type, again being more important in mature tundra plots (17.1%) compared to disturbed farm plots (5.0%). Bare ground in the form of fine sediment was important in mature tundra plots (14.2%), but was not as important in disturbed farm plots (4.7%). While Salix arctica and moss were the most important vegetation types in both areas, the species of secondary importance show differences in the vegetation communities (Figure 4.5). Graminoids were more important in disturbed farm plots (Luzula confusa = 3.94%; Poa arctica = 3.19%; Festuca brachyphylla = 1.72%), while shrubs were more important in mature tundra plots (Dryas integrifolia = 4.37%; Cassiope tetragona = 2.25%). The shrubs found in mature tundra were also present in the disturbed farm plots, but were much less abundant (Dryas integrifolia = 0.57%; Cassiope tetragona = 0.15%). Several other forbs and graminoids made up less that 1% of the cover at both sites. Total species richness of all the disturbed farm plots observed in the presence/absence survey was 26. When a species accumulation curve was plotted, it levelled off to this number. In contrast, the total species richness of all the mature tundra plots was 25. However only three plots were surveyed, and the species accumulation curve showed that this was not enough to achieve a true representation of the richness. Another way of comparing species richness is to compare average richness per plot. The average species richness was not different between the sites, with considerable overlap between confidence intervals: mean richness ± C.I. was 13.4 ± 4.7 in the disturbed site and 14.7 ± 7.0 in the undisturbed tundra. Soil properties There were some difference in soil properties between the disturbed farm plots and the surrounding mature tundra (Table 4.2). Soil conditions were generally more favourable in the mature tundra, having higher soil moisture, nitrogen, carbon, organic matter, and more available !127ammonium, calcium, and magnesium. The disturbed farm plots had more available nitrate and potassium. However, all values showed very high variability from sample to sample, and almost all means from the two areas were within a 95% confidence intervals of each other. The only exception is the value for ammonium in disturbed farm plots, which was consistently lower than in mature tundra sites.  4.3.2 Secondary Succession vs. Primary SuccessionThe differences between the disturbed farm area and Twin Glacier foreland were remarkable. Despite both sites recovering for the same amount of time, there is much more vegetation, or remnants of vegetation, at the site of secondary succession on the farm. At Twin Glacier foreland, 65.9% of the cover is bare substrate, compared to only 4.7% at the farm. While standing dead vegetation made up 52.5% of the cover at the farm, it made up only 6.8% at Twin Glacier foreland. The vegetation with the highest cover at Twin Glacier foreland was moss !128Farm TundraSoil Moisture (%) 23.39 ± 8.84 31.13 ± 27.74Total Nitrogen (%) 0.23 ± 0.16 0.31 ± 0.52Total Carbon (%) 3.51 ± 2.24 4.38 ± 6.22Organic Matter (%) 7.55 ± 4.36 9.94 ± 13.76NO3-N 20 ± 16 15 ± 13NH4-N 3 ± 8 26 ± 40Ca 300 ± 206 371 ± 217Mg 175 ± 115 217 ± 117K 110 ± 118 78 ± 53Table 4.2. The mean and 95% confidence interval of measured soil properties both from disturbed farm plots (Farm) and from the surrounding mature tundra (Tundra). Nutrients are in supply rates (micro grams/10cm2 over 26 days).(2.09%), and while moss was hardly the most important vegetation type at the farm, it still had higher cover there (5.01%). The most important vascular plant species at Twin Glacier foreland was Luzula confusa (1.05%). At the farm Luzula confusa was again less important relative to other species, but still had higher cover (3.94%) than at the foreland. At the farm, the most important vascular plant species was Salix arctica (14.41%), which was the third most abundant vascular species at the foreland, but had a much lower cover (0.45%). Papaver spp. were the only vegetation type to be more abundant on the foreland (0.46%) compared to the farm (0.01%). !1294.4 Discussion After 31 years, the ‘Green Igloos’ disturbed farm area had still not returned to a similar cover and community composition as the surrounding mature tundra. Considering the slow rates of plant growth and reproduction in the Arctic, this is not surprising. Cannone et al. (2010) studied secondary succession on permafrost landslide disturbances on Ellesmere Island and report that 50 years were required for the vegetation to achieve similar composition as communities in undisturbed conditions. At the time of our study, standing dead vegetation made up the largest portion of cover at the disturbed farm sites. Much of this standing dead was observed to originate from graminoids (mostly Luzula confusa). Certain species of graminoids are often noted as being important during early stages of secondary succession in Arctic environments, although the specific species seems to vary according to location (Babb & Bliss, 1974; Cannone et al., 2010; Desforges, 2000; Kemper & Macdonald, 2009; Forbes et al., 2001). However the dominant live plant species at the disturbed farm plots was the deciduous shrub Salix arctica. In disturbances such as ours where all plant cover is removed, shrubs usually appear only in later stages of secondary succession (Cannone et al., 2010; Desforges, 2000; Babb & Bliss, 1974). These two pieces of evidence suggest that a shift in vegetation might be underway. The large amount of standing dead vegetation could suggest that early-arriving graminoids are dying and giving way to slower growing shrubs. Salix arctica is expected to continue increasing in cover while graminoids are expected to continue decreasing, as seen by their respective abundances in mature tundra plots. While the site of secondary succession at the farm was still far from reaching maturity, it was much more developed than the site of primary succession at Twin Glacier foreland. Vegetation are the foreland remained extremely sparse. The biological legacy of rich soils and the proximity of propagules likely aided in the recovery of vegetation at the site of secondary succession. However, despite their differences in total cover, there were some similarities in the vegetation !130communities. Graminoids were important constituents of the vegetation communities at both locations, with Luzula confusa being the dominant graminoid at both locations. Salix arctica was also important at both locations, although was much less abundant on the foreland. Shrubs typical of mature communities in the High Arctic (Dryas integrifolia and Cassiope tetragona) were rare both at the farm and on the foreland. !1314.5 Conclusion The site of secondary succession at the ‘Green Igloos’ farm is still far from reaching similar cover and community composition as the surrounding mature tundra, however it is much more advanced than terrain of similar age at the Twin Glacier foreland. The farm site seems to have progressed passed an early stage of graminoid dominance - seen by an abundance of graminoid dominated standing dead vegetation -  and is now progressing towards a more mature community with a higher abundance of shrub species, notable Salix arctica. The availability of ammonium is still much less in disturbed farm plots compared to the surrounding mature tundra. However, the farm site is much closer to a mature community than the Twin Glacier foreland site, which is still dominated by bare ground.  !132CONCLUSION !133This thesis has investigated the patterns of and mechanisms driving both primary and secondary succession at and around Alexandra Fiord, Ellesmere Island, in the Canadian High Arctic. The literature review described how succession has proved to be a complex process dependant on local environmental characteristics and individual species traits. This research has further confirmed the variability and site specific nature of succession. We therefore beg for a local understanding of succession if theory is used in applications such as prediction of future ecosystems or restoration of disturbed landscapes. In Chapter 2 we resurveyed three glacial forelands originally surveyed 21 years ago to better understand the patterns of primary succession. We observed species advance along successional gradients individualistically rather than as well defined communities. Their presences and abundances advanced predictably in a directional manner, confirming the use of the chronosequence approach in these studies. Local species diversity was important in determining species patterns, yet certain species had comparable patterns across all three forelands. The patterns of eight important species were discussed, including: Luzula confusa, Papaver spp., and Saxifraga cernua which were important early- to mid-successional species; Saxifraga oppositifolia and Salix arctica which arrived early but grew slowly in size, becoming important on older terrain; and Cassiope tetragona, Dryas integrifolia, and Carex nardina which were late-successional to mature species.  In Chapter 3 we surveyed a topographically heterogeneous glacial foreland to better understand mechanisms behind primary succession. We found that micro-environmental influences played the most important role, where variation in micro-topography and soil type explained a large amount of variation in vegetation patterns, as did the random effect representing fine-scale spatial structure. Facilitation was also hypothesized to be important, as demonstrated by the important positive relationship between moss and vascular plant occurrences, as well as the positive fine-scale species-to-species associations. Time since deglaciation also proved important, however was not the dominant factor. Species responded differently to time, with the occurrences of some species increasing linearly as time progressed, while other species showed a !134unimodal response. Intrinsic species life history traits likely also played a role in determining patterns, as did distance to a seed source. While these finding provide interesting insights, limitations in the number of sites surveyed for both vegetation and environmental variables led us to conclude that there were likely other mechanisms also structuring vegetation patterns that were not captured by our study.  In Chapter 4 we surveyed a site of secondary succession that had been recovering for 31 years. We found that this site seemed to have progressed past an earlier stage of graminoid dominance, and was now moving towards a later stage where the deciduous shrub Salix arctica was becoming more important. This site was far more advanced in its recovery than the site of primary succession near the foreland, however its community composition was still not comparable to that of the surrounding mature tundra. Despite the different rates of recovery, the sites of both primary and secondary succession shared similarities in the identities of certain key successional species, including Luzula confusa and Salix arctica.  This thesis has also pointed to possible future directions for research. The individualistic nature of species described in Chapter 2 led us to conclude that a description of patterns during succession is better done by describing the evolution of individual species rather than by describing communities or groups of species. Chapter 3 used correlative methods to make data-driven hypotheses about mechanisms driving succession, however these hypotheses could be tested experimentally. Considering the rarity of studies concerning succession in the Canadian High Arctic, any further research will aid in developing a better understanding of this highly variable process. The High Arctic environment is currently experiencing rapid change. Understanding how vegetation recovers from disturbances can help researcher better predict future vegetation patterns. With the goal of better understanding succession, this work has provided a detailed description of the patterns and mechanisms influencing this complex process in the Canadian High Arctic. 
!135BIBLIOGRAPHY Addison, P. A., & Bliss, L. C. (1984). Adaptations of Luzula confusa to the polar semi-desert environment. Arctic, 37, 121–132.  AgiSoft PhotoScan Professional (2017). Software, version 1.3.5. Retrieved on October 15, 2017, from: http://www.agisoft.com/downloads/installer/ Aiken, S.G., Dallwitz, M.J., Consaul, L.L., McJannet, C.L., Boles, R.L., Argus, G.W., … Harris, J.G. (2007). Flora of the Canadian Arctic Archipelago: Descriptions, illustrations, identification, and information retrieval. NRC Research Press, National Research Council of Canada, Ottawa. Retrieved on January 25, 2018, from: http://nature.ca/aaflora/data  Alsos, I. G., Müller, E., & Eidesen, P. B. (2013). Germinating seeds or bulbils in 87 of 113 tested Arctic species indicate potential for ex situ seed bank storage. Polar Biology, 36, 819–830. Anderson, M. J. (2001). A new method for non-parametric multivariate analysis of variance. Austral Ecology, 26, 32–46.  Au, R., & Tardif, J. C. (2007). Allometric relationships and dendroecology of the dwarf shrub Dryas integrifolia  near Churchill, subarctic Manitoba. Canadian Journal of Botany,  85,  585–597.Babb, T. A., & Bliss, L. C. (1974). Effects of Physical Disturbance on Arctic Vegetation in the Queen Elizabeth Islands. Journal of Applied Ecology, 11, 549–562.  Bell, K. L., & Bliss, L. C. (1980). Plant Reproduction in a High Arctic Environment. Arctic and Alpine Research, 12(1), 1–10. Bergsma, B. M., Svoboda, J., & Freedman, B. (1984). Entombed plant communities released by a retreating glacier at central Ellesmere Island, Canada. Arctic, 37, 49–52. Billings, W. D., & Mooney, H. A. (1968). The ecology of Arctic and Alpine Plants. Biological Review, 43, 481–529. Bivand, R. S., Pebesma, E. J., Gomez-Rubio, V. (2013). Applied spatial data analysis with R, Second edition. Springer, New York, NY. !136Bivand, R. & Rundel, C. (2017). rgeos: Interface to Geometry Engine - Open Source (GEOS). R package, version 0.3-26. Retrieved on November 1, 2017, from: https://CRAN.R-project.org/package=rgeos  Blanchet, F. G., Tikhonov, G., & Norberg, A. (2018). HMSC: Hierarchical Modelling of Species Communities. R package, version 2.1-1. Retrieved on March 20, 2018, from: https://github.com/guiblanchet/HMSC  Bliss, L. C. (1958). Seed germination in arctic and alpine species. Arctic, 11, 180–188. Bliss, L. C. (1962). Adaptations of arctic and alpine plants to environmental conditions. Arctic, 15, 117–144. Bliss, L. C., & Gold, W. G. (1994). The patterning of plant communities and edaphic factors along a High Arctic coastline: Implications for succession. Canadian Journal of Botany, 72, 1095–1107.  Blundon, D. J., MacIsaac, D. A., & Dale, M. R. T. (1993). Nucleation during primary succession in the Canadian Rockies. Canadian Journal of Botany, 71, 1093–1096.  Boulanger-Lapointe, N., Lévesque, E., Boudreau, S., Henry, G. H. R., & Schmidt, N. M. (2014). Population structure and dynamics of Arctic willow (Salix arctica) in the High Arctic. Journal of Biogeography, 41, 1967–1978.  Breen, K., & Lévesque, E. (2006). Proglacial succession of biological soil crusts and vascular plants: Biotic interactions in the High Arctic. Canadian Journal of Botany, 84, 1714–1731.  Buma, B., Bisbing, S., Krapek, J., & Wright, G. (2017). A foundation of ecology rediscovered: 100   years of succession on the William S. Cooper plots in Glacier Bay, Alaska. Ecology, 98, 1513–1523.Burga, C. A., Krüsi, B., Egli, M., Wernli, M., Elsener, S., Ziefle, M., … Mavris, C. (2010). Plant succession  and  soil  development  on  the  foreland  of  the  Morteratsch  glacier  (Pontresina, Switzerland): Straight forward or chaotic? Flora: Morphology, Distribution, Functional Ecology of Plants, 205, 561–576. Callaghan, T. V., & Emanuelsson, U. (1985). Population structure and processes of tundra plants and  vegetation.  In  J.  White  (Ed.),  The  population  structure  of  vegetation  (pp.  399–439). Dordrecht: Springer Science + Business Media.  !137Callaway,  R.  M.  (1997).  Positive  interactions  in  plant  communities  and  the  individualistic-continuum concept. Oecologia, 112, 143–149. Canadian Northern Economic Development Agency (2013). Building a Strong North Together: Strategic  Framework  2013-2018.  Retrieved  on  November  20,  2015  from:  http://www.cannor.gc.ca/DAM/DAM-CANNOR-CANNOR/STAGING/texte-text/framework_strat-plan_1387761468037_eng.pdf  Cannone, N., Guglielmin, M., & Gerdol, R. (2004). Relationships between vegetation patterns and periglacial landforms in northwestern Svalbard. Polar Biology, 27, 562–571.  Cannone, N., Diolaiuti, G., Guglielmin, M., & Smiraglia, C. (2008). Accelerating climate change impacts on alpine glacier gorefield ecosystems in the European Alps. Ecological Applications, 18, 637–648.  Cannone, N., Lewkowicz, A. G., & Guglielmin, M. (2010). Vegetation colonization of permafrost-related landslides, Ellesmere Island, Canadian High Arctic. Journal of Geophysical Research: Biogeosciences, 115, 1–10.  Cargill, S. M., & Chapin, F. S. (1987). Application of successional theory to tundra restoration: a review. Arctic and Alpine Research, 19, 366–372. Carlsson, B. A., & Callaghan, T. V. (1991). Positive plant interactions in tundra vegetation and the importance of shelter. Journal of Ecology, 79, 973–983.  Chan, C. (2001). Residual Effects of Fertilization in Experimental Plots in the High Arctic. Directed Study: University of British Columbia. Chapin, F. S., & Shaver, G. R. (1985). Individualistic growth response of tundra plant species to environmental manipulations in the field. Ecology, 66, 564–576.  Chapin, F. S., Walker, L. R., Fastie, C. L., & Sharman, L. C. (1994). Mechanism of primary succession following deglaciation at Glacier Bay, Alaska. Ecological Monographs, 64, 149–175. Chase, J. M. (2014). Spatial scale resolves the niche versus neutral theory debate. Journal of Vegetation Science, 25, 319–322. Clements, F. E. (1916). Plant succession: an analysis of the development of vegetation (No. 242). Carnegie Institution of Washington. !138Connell, J. H., Noble, I. R., & Slatyer, R. O. (1987). On the Mechanisms Producing Successional Change. Oikos, 50, 136–137. Connell, J. H., & Slatyer, R. O. (1977). Mechanisms of succession in natural communities and their role in community stability and organisation. The American Naturalist, 111, 1119–1144.  Cooper, W. S. (1923). The Recent Ecological History of Glacier Bay, Alaska: The Interglacial Forests of Glacier Bay. Ecology, 4, 93–128. Cooper, W. S. (1931). A Third Expedition to Glacier Bay , Alaska. Ecology, 12, 61–95. R Cooper, W. S. (1939). A Fourth Expedition to Glacier Bay, Alaska. Ecology, 20, 130–155. Cooper, E. J., Alsos, I. G., Hagen, D., Smith, F. M., Coulson, S. J., & Hodkinson, I. D. (2004). Plant recruitment in the High Arctic: Seed bank and seedling emergence on Svalbard. Journal of Vegetation Science, 15, 115–124. Cowles, H. C. (1899). The Ecological Relations of the Vegetation on the Sand Dunes of Lake Michigan. Doctoral Thesis: The University of Chicago. Corripio, J. G. (2014). insol: Solar Radiation. R package, version 1.1.1. Retrieved on October 20, 2017, from: https://CRAN.R-project.org/package=insol Crowley, T. J. (2000). Causes of Climate Change Over the Past 1000 Years. Science, 289, 270–277.  Cutler, N. (2010). Long-term primary succession: A comparison of non-spatial and spatially explicit inferential techniques. Plant Ecology, 208, 123–136. Cutler, N. (2011). Vegetation-environment interactions in a sub-arctic primary succession. Polar Biology, 34, 693–706. Dahl-Jensen, D., Mosegaard, K., Gundestrup, N., Clow, G. D., Johnsen, S. J., Hansen, A. W., & Balling, N. (1998). Past Temperatures Directly from the Greenland Ice Sheet. Science, 282, 268–271.  Danby, R. K., Koh, S., Hik, D. S., & Price, L. W. (2011). Four decades of plant community change in the alpine tundra of southwest Yukon, Canada. Ambio, 40, 660–671.  Daniëls,  F.  J.  a,  De Molenaar,  J.  G.,  Chytrý,  M.,  & Tichý,  L.  (2011).  Vegetation change in southeast Greenland? Tasiilaq revisited after 40 years. Applied Vegetation Science, 14, 230–241. !139De Cáceres, M., & Legendre, P. (2009). Associations between species and groups of sites: indices and statistical inferences. Ecology, 90, 3566-3574.  Desforges, M. (2000). Plant Succession after Active Layer Detachment Slides in High Arctic Tundra, Fosheim Peninsula, Ellesmere Island, Canada (Master’s Thesis). University of British Columbia. Department of Energy Mines and Resources, Government of Yukon (2015). Yukon Mining and Exploration Projects 2015. Retrieved on November 20, 2015, from: http://www.emr.gov.yk.ca/mining/pdf/yukon-mining-exploration-projects-8.5x11-oct2015.pdf  Dolezal, J., Homma, K., Takahashi, K., Vyatkina, M. P., Yakubov, V., Vetrova, V. P., & Hara, T. (2008). Primary succession following deglaciation at Koryto Glacier Valley, Kamchatka. Arctic, Antarctic, and Alpine Research and Alpine Research, 40, 309–322. Dowdeswell, J. A. (1995). Glaciers in the High Arctic and Recent Environmental Change. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 352, 321–334.  Dufrêne, M., & Legendre, P. (1997). Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs, 67, 345-366Egler, F. E. (1954). Vegetation science concepts I. Initial floristic composition, a factor in old-field vegetation development. Vegetatio Acta Geobotanica, 4, 412–417.  Elith, J., & Leathwick, J. R. (2009). Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annual Review of Ecology, Evolution, and Systematics, 40, 677–697. Elmendorf, S. C., Henry, G. H. R., Hollister, R. D., Björk, R. G., Bjorkman, A. D., Callaghan, T. V., … Wookey, P. A. (2012a). Global assessment of experimental climate warming on tundra vegetation: Heterogeneity over space and time. Ecology Letters, 15, 164–175. Elmendorf,  S.  C.,  Henry,  G.  H.  R.,  Hollister,  R.  D.,  Björk,  R.  G.,  Boulanger-Lapointe,  N., Cooper, E. J., … Wipf, S. (2012b). Plot-scale evidence of tundra vegetation change and links to recent summer warming. Nature Climate Change, 2, 453–457.Elven, R., & Ryvarden, L. (1975). Dispersal and primary establishment of vegetation. In F. E. Wielgolaski  (Ed.),  Fennoscandian  tundra  ecosystems  (pp.  82–85).  Springer-Verlag  Berlin Heidelberg.!140Erschbamer, B. (2007). Winners and losers of climate change in a central alpine glacier foreland. Arctic, Antarctic, and Alpine Research, 39, 237–244.Erschbamer, B., Kneringer, E., & Schlag, R. N. (2001). Seed rain, soil seed bank, seedling recruitment, and survival of seedlings on a glacier foreland in the Central Alps. Flora, 304–312. Fastie,  C.  L.  (1995).  Causes  and  ecosystem consequences  of  multiple  pathways  of  primary succession at Glacier Bay, Alaska. Ecology, 76, 1899–1916.  Feagin, R. A., Wu, X. B., Smeins, F. E., Whisenant, S. G., & Grant, W. E. (2005). Individual versus  community  level  processes  and  pattern  formation  in  a  model  of  sand  dune  plant succession. Ecological Modelling, 183, 435–449.  Ferrier, S., & Guisan, A. (2006). Spatial modelling of biodiversity at the community level. Journal of Applied Ecology, 43, 393–404. Finegan, B. (1984). Forest Succession. Nature, 312, 109–114.  Forbes, B. C., Ebersole, J. J., & Strandberg, B. (2001). Anthropogenic disturbance and patch dynamics in circumpolar ecosystems. Conservation Biology, 15, 954–969.  Freedman, B., Svoboda, J. & Henry, G.H.R. (1994) Alexandra Fjord – an ecological oasis in the polar desert. In J. Svoboda and B. Freedman  (Eds.), Ecology of a polar oasis, Alexandra Fjord, Ellesmere Island, Canada (pp. 1–9). Captus University Publications, Toronto.  Gajewski, K. (2012). The Evolution of Polar Desert and Tundra Ecosystems. In H. French & O. Slaymaker (Eds.), Changing Cold Environments (pp. 48–65). Chichester, UK: John Wiley & Sons, Ltd. Gajewski, K., & Atkinson, D. A. (2003). Climatic change in northern Canada. Environmental Reviews, 11, 69–102. Gardner, A. S., Moholdt, G., Wouters, B., Wolken, G. J., Burgess, D. O., Sharp, M. J., … Labine, C. (2011). Sharply increased mass loss from glaciers and ice caps in the Canadian Arctic Archipelago. Nature, 473, 357–360.  Gleason, H. A. (1917). The Structure and Development of the Plant Association. Bulletin of the Torrey Botanical Club, 44, 463–481. Glenn-Lewin,  D.  C.  (1980).  The  individualistic  nature  of  plant  community  development. Vegetatio, 43, 141–146.!141Glenn-Lewin, D. C., Peet, R. K., & Veblen, T. T. (1992). Plant succession: theory and prediction. London, UK: Chapman & Hall. Gravel, D., Canham, C. D., Beaudet, M., & Messier, C. (2006). Reconciling niche and neutrality: the continuum hypothesis. Ecology Letters, 9, 399–409. Gräler, B., Pebesma, E., & Heuvelink, G. (2016). Spatio-Temporal Interpolation using gstat. The R Journal 8, 204-218.Gretarsdottir, J., Aradottir, A. L., Vandvik, V., Heegaard, E., & Birks, H. J. B. (2004). Long-term effects of reclamation treatments on plant succession in Iceland. Restoration Ecology, 12, 268–278. Hagen, D., Hansen, T. I., Graae, B. J., & Rydgren, K. (2014). To seed or not to seed in alpine restoration: Introduced grass species outcompete rather than facilitate native species. Ecological Engineering, 64, 255–261. Harper, J. L., Williams, J. T., & Sagar, G. R. (1965). The behaviour of seeds in soil. I. The heterogeneity of soil surfaces and its role in determining the establishment of plants from seed. Journal of Ecology, 53, 273–286. Henry, G., Freedman, B., & Svoboda, J. (1986). Survey of vegetated areas and muskox populations in east-central Ellesmere Island. Arctic, 39, 78–81. Hijmans, R. J. (2016). raster: Geographic Data Analysis and Modeling. R package, version 2.5-8. Retrieved on April 5, 2017, from: https://CRAN.R-project.org/package=raster  Hodkinson, I. D., Coulson, S. J., & Webb, N. R. (2003). Community assembly along proglacial chronosequences in the High Arctic: Vegetation and soil development in north-west Svalbard. Journal of Ecology, 91, 651–663.  Hubbell, S. P. (2001). The unified neutral theory of biodiversity and biogeography. Princeton: Princeton University Press. Hudson, J. M. G., & Henry, G. H. R. (2009). Increased plant biomass in a high arctic heath community from 1981 to 2008. Ecology, 90, 2657–2663. Jírová, A., Klaudisová, A., & Prach, K. (2012). Spontaneous restoration of target vegetation in old-fields  in  a  central  European  landscape:  A repeated  analysis  after  three  decades.  Applied Vegetation Science, 15, 245–252.  !142Johnson, E. A., & Miyanishi, K. (2008). Testing the assumptions of chronosequences in succession. Ecology Letters, 11, 419–431. Jones, G. A. (1997). Primary plant succession on the Twin Glacier foreland, Alexandra Fjord, Ellesmere Island, Canadian High Arctic (Master’s Thesis). University of British Columbia.  Jones, G. A., & Henry, G. H. R. (2003). Primary plant succession on recently deglaciated terrain in the Canadian High Arctic. Journal of Biogeography, 30, 277–296. Jumpponen, A., Väre, H., Mattson, K. G., Ohtonen, R., & Trappe, J. M. (1999). Characterization of “safe sites” for pioneers in primary succession on recently deglaciated terrain. Journal of Ecology, 87, 98–105. Kattge, J., Díaz, S., Lavorel, S., Prentice, I. C., Leadley, P., Bönisch, G., … Wirth, C. (2011). TRY - a global database of plant traits. Global Change Biology, 17, 2905–2935. Kaufmann, R. (2002). Glacier foreland colonisation: Distinguishing between short-term and long-term effects of climate change. Oecologia, 130, 470–475. Kemper, J. T., & Macdonald, S. E. (2009). Effects of Contemporary Winter Seismic Exploration on Low Arctic Plant Communities and Permafrost. Arctic, Antarctic, and Alpine Research, 41, 228–237. Kercher, S. M., Frieswyk, C. B., & Zedler, J. B. (2003). Effects of sampling teams and estimation methods on the assessment of plant cover. Journal of Vegetation Science, 14, 899-906.  Klady,  R.  A.,  Henry,  G.  H.  R.,  &  Lemay,  V.  (2011).  Changes  in  high  arctic  tundra  plant reproduction in response to long-term experimental warming. Global Change Biology, 17, 1611–1624.Lawrence, D., Schoenike, R., Quispel, A., & Bond, G. (1967). The role of Dryas drummondii in vegetation development following ice recession at Glacier Bay, Alaska, with special reference to its nitrogen fixation by root nodules. The Journal of Ecology, 55, 793–813. Lawton, J. H. (1999). Are there general laws in ecology? Oikos, 84, 177-192. Leibold, M. A., & McPeek, M. A. (2006). Coexistence of the niche and neutral perspectives in community ecology. Ecology, 87, 1399–1410. Levesque, E., & Svoboda, J. (1999). Vegetation re-establishment in polar lichen-kill’’ landscapes: a case study of the Little Ice Age impact. Polar Research, 18, 221–228.!143Magnússon, B., Magnússon, S. H., Ólafsson, E., & Sigurdsson, B. D. (2014). Plant colonization, succession and ecosystem development on Surtsey with reference to neighbouring islands. Biogeosciences, 11, 5521–5537. Mann, M. E., Bradley, R. S., & Hughes, M. K. (1998). Global-scale temperature patterns and climate forcing over the past six centuries. Nature, 392, 779–787. Marcante, S., Sierra-Almeida, A., Spindelböck, J. P., Erschbamer, B., & Neuner, G. (2012). Frost as a limiting factor for recruitment and establishment of early development stages in an alpine glacier foreland? Journal of Vegetation Science, 23, 858–868.  Marteinsdóttir, B., Svavarsdóttir, K., & Thórhallsdóttir, T. E. (2010). Development of vegetation patterns in early primary succession. Journal of Vegetation Science, 21, 531–540. Matthews, J. A. (1976). A phytogeography of a gletschervorfeld: Storbreen-i-Leirdalen, Jotunheimen, Norway. Ph.D. Thesis: University of London, King’s College. Matthews, J. A. (1978). Plant colonisation patterns on a gletschervorfeld, southern Norway: A meso-scale geographical approach to vegetation change and phytometric dating. Boreas, 7, 155–178. Matthews, J. A. (1992). The Ecology of Recently-Deglaciated Terrain: A Geoecological Approach to Glacier Forelands and Primary Succession. Cambridge, UK: Cambridge University Press. McCook, L. J. (1994). Understanding Ecological Community Succession: Causal Models and Theories, a Review. Vegetatio, 110, 115–147.Metcalfe, P., Beven, K., & Freer, J. (2017). dynatopmodel: Implementation of the Dynamic TOPMODEL Hydrological Model. R package, version 1.2.1. Retrieved on October 15, 2016 from: https://CRAN.R-project.org/package=dynatopmodel   Moreau, M., Laffly, D., Joly, D., & Brossard, T. (2005). Analysis of plant colonization on an arctic moraine since the end of the Little Ice Age using remotely sensed data and a Bayesian approach. Remote Sensing of Environment, 99, 244–253.  Moreau, M., Mercier, D., Laffly, D., & Roussel, E. (2008). Impacts of recent paraglacial dynamics on plant colonization: A case study on Midtre Lovénbreen foreland, Spitsbergen (79°N). Geomorphology, 95, 48–60.  !144Moreau, M., Laffly, D., & Brossard, T. (2009). Recent spatial development of Svalbard strandflat vegetation over a period of 31 years. Polar Research, 28, 364–375. Mori, A., Osono, T., Iwasaki, S., Uchida, M., & Kanda, H. (2006). Initial recruitment and establishment of vascular plants in relation to topographical variation in microsite conditions on a recently-deglaciated moraine on Ellesmere Island, High Arctic Canada. Polar Bioscience, 19, 85–95.  Mori, A. S., Osono, T., Uchida, M., & Kanda, H. (2008). Changes in the structure and heterogeneity of vegetation and microsite environments with the chronosequence of primary succession on a glacier foreland in Ellesmere Island, high arctic Canada. Ecological Research, 23, 363–370.  Mori, A. S., Uchida, M., & Kanda, H. (2013). Non-stochastic colonization by pioneer plants after deglaciation in a polar oasis of the Canadian High Arctic. Polar Science, 7, 278–287.  Müller,  E.,  Cooper,  E.  J.,  &  Alsos,  I.  G.  (2011).  Germinability  of  arctic  plants  is  high  in perceived optimal conditions but low in the field. Botany, 89, 337–348.  Morrison,  L.  W.  (2016).  Observer  error  in  vegetation  surveys:  A review.  Journal  of  Plant Ecology, 9, 367–379. Myers-Smith,  I.  H.,  Hik,  D. S.,  Kennedy, C.,  Cooley, D.,  Johnstone, J.  F.,  Kenney, A. J.,  & Krebs,  C.  J.  (2011).  Expansion  of  canopy-forming  willows  over  the  twentieth  century  on Herschel Island, Yukon Territory, Canada. Ambio, 40, 610–623.Nakatsubo, T., Fujiyoshi, M., Yoshitake, S., Koizumi, H., & Uchida, M. (2010). Colonization of the polar willow Salix polaris on the early stage of succession after glacier retreat in the High Arctic, Ny-Ålesund, Svalbard. Polar Research, 29, 385–390.  NWT & Nunavut Chamber of Mines. (2015). Mines Actively Producing in the NWT and Nunavut. Retrieved November 16, 2015, from: http://www.miningnorth.com/mines  Okitsu, S., Sawaguchi, S., Hasegawa, H., & Kanda, H. (2004). Vegetation development on the glacier moraines in Oobloyah Valley, Ellesmere Island, High Arctic Canada. Polar Bioscience, 17, 83–94.  Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., … Wagner, H. (2017). vegan: Community ecology package. R package, version 2.4-3. Retrieved on September 23, 2016, from: https://CRAN.R-project.org/package=vegan!145Olofsson, J. (2004). Positive and negative plant-plant interactions in two contrasting arctic-alpine plant communities. BioOne, 36, 464–467.Ovaskainen, O., & Soininen, J. (2011). Making more out of sparse data  : hierarchical modelling of species communities. Ecology, 92, 289–295. Ovaskainen, O., Abrego, N., Halme, P., & Dunson, D. (2016a). Using latent variable models to identify large networks of species-to-species associations at different spatial scales. Methods in Ecology and Evolution, 7, 549–555. Ovaskainen, O.,  Roy, D. B.,  Fox, R.,  & Anderson, B. J.  (2016b).  Uncovering hidden spatial structure  in  species  communities  with  spatially  explicit  joint  species  distribution  models. Methods in Ecology and Evolution, 7, 428–436. Ovaskainen, O., Tikhonov, G., Norberg, A., Guillaume Blanchet, F., Duan, L., Dunson, D., … Abrego, N. (2017). How to make more out of community data? A conceptual framework and its implementation as models and software. Ecology Letters, 20, 561–576.Overpeck, J., Hughen, K., Hardy, D., Bradley, R., Case, R., Douglas, M., … Zielinski, G. (1997). Arctic Environmental Change of the Last Four Centuries. Science, 278, 1251–1256. Pebesma, E. J. (2004). Multivariable geostatistics in S: the gstat package. Computers & Geosciences, 30, 683-691.  Pebesma, E. J., Bivand, R. S. (2005). Classes and methods for spatial data in R. R News, 5. Retrieved on September 25, 2017, from: https://cran.r-project.org/doc/Rnews/ Peet, R. K., & Christensen, N. L. (1980). Succession: A population process. Vegetatio, 43, 131–140.  Pickett, S. T., & White, P. S. (1985). The ecology of natural disturbance and patch dynamics. Orlando, Florida: Academic Press.  Post, E., Forchhammer, M. C., Bret-Harte, M. S., Callaghan, T. V., Christensen, T. R., Elberling, B., … Aastrup, P. (2009). Ecological dynamics across the arctic associated with recent climate change. Science, 325, 1355–1358. Prach, K., Šmilauer, P., & Pyšek, P. (1997). Changes in species traits during succession  : a search for pattern. Oikos, 79, 201–205.Prach, K., & Rachlewicz, G. (2012). Succession of vascular plants in front of retreating glaciers in central Spitsbergen. Polish Polar Research, 33, 319–328.  !146Pulsford, S. A., Lindenmayer, D. B., & Driscoll, D. A. (2016). A succession of theories: Purging redundancy from disturbance theory. Biological Reviews, 91, 148–167. QGIS Development Team (2018). QGIS Geographic Information System. Open Source Geospatial Foundation Project. Retrieved on September 20, 2015, from: http://qgis.osgeo.org R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Retrieved on January 5, 2016, from: https://www.R-project.org/ Raillard, M., & Svoboda, J. (2000) High grazing impact, selectivity, and local density of muskoxen in central Ellesmere Island, Canadian High Arctic, Arctic, Antarctic, and Alpine Research, 32, 278-285. Raup, H. M. (1981) Physical disturbance in the life of plants. In M. H. Nitechi (Ed.), Biotic Crises in Ecological and Evolutionary Time (pp. 39-52). New York: Academic Press. Řehounková, K., & Prach, K. (2010). Life-history traits and habitat preferences of colonizing plant  species  in  long-term spontaneous succession in  abandoned gravel-sand pits.  Basic  and Applied Ecology, 11, 45–53. Robbins, J. A., & Matthews, J. A. (2010). Regional Variation in Successional Trajectories and Rates of Vegetation Change on Glacier Forelands in South-Central Norway. Arctic, Antarctic, and Alpine Research, 42, 351–361.Savage, J.,  & Vellend, M. (2015).  Elevational shifts,  biotic homogenization and time lags in vegetation change during 40 years of climate warming. Ecography, 38, 546–555. Schwienbacher, E., Navarro-Cano, J. A., Neuner, G., & Erschbamer, B. (2012). Correspondence of seed traits with niche position in glacier foreland succession. Plant Ecology, 213(3), 371–382. Sharp, M., Wolken, G., Burgess, D., Cogley, J. G., Copeland, L., Thomson, L., … Pelto, M. (2015). Glaciers and ice caps outside Greenland. In J. Blunden & D. S. Arndt (Eds.), State of the Climate in 2014, Special Supplement to the Bulletin of the American Meteorological Society, 96, 135–137. Stöcklin, J., & Bäumler, E. (1996). Seed rain, seedling establishment and clonal growth strategies on a Glacier Foreland. Journal of Vegetation Science, 7, 45–56. !147Suter, M., & Edwards, P. J. (2013). Convergent succession of plant communities is linked to species’ functional traits. Perspectives in Plant Ecology, Evolution and Systematics, 15(4), 217–225.Svoboda, J., & Henry, G. H. R. (1987). Succession in marginal arctic environments. Arctic and Alpine Research, 19, 373–384.  Tableau Desktop (2017). Tableau Desktop: Professional Edition. Tableau Software, Inc. Retrieved on September 16, 2017, from: https://www.tableau.com/products/desktop   Těšitel, J., Těšitelova, T., Bernardová, A., Drdová, E. J., Lučanová, M., & Klimešová, J. (2014). Demographic population structure and fungal associations of plants colonizing High Arctic glacier forelands, Petuniabukta , Svalbard. Polar Research, 33, 20797.  Tilman, D. (1985). The resource-ratio hypothesis of plant succession. The American Naturalist, 125, 827–852.  Vincent, W. F., Lemay, M., & Allard, M. (2017) Arctic permafrost landscapes in transition: towards an integrated Earth system approach. Arctic Science, 3, 39-64.  Vittoz, P., Bayfield, N., Brooker, R., Elston, D. A., Duff, E. I., Theurillat, J. P., & Guisan, A. (2010).  Reproducibility  of  species  lists,  visual  cover  estimates  and  frequency  methods  for recording high-mountain vegetation. Journal of Vegetation Science, 21, 1035–1047.Walker, L. R., & del Moral, R. (2003). Primary succession and ecosystem rehabilitation. Cambridge, UK: Cambridge University Press.  Walker, L. R., Zasada, J. C., & Chapin III, F. S. (1986). The role of life history processes in primary succession on an Alaskan floodplain. Ecology, 67, 1243–1253.  Walker, L. R., Wardle, D. A., Bardgett, R. D., & Clarkson, B. D. (2010). The use of chronosequences in studies of ecological succession and soil development. Journal of Ecology, 98, 725–736.  Warton, D. I., Blanchet, F. G., O’Hara, R. B., Ovaskainen, O., Taskinen, S., Walker, S. C., & Hui, F. K. C. (2015). So Many Variables: Joint Modeling in Community Ecology. Trends in Ecology and Evolution, 30, 766–779.  Wei, T. & Simko, V. (2017). R package “corrplot”: Visualization of a Correlation Matrix. R package, version 0.84. Retrieved on March 23, 2018, from: https://guithub.com/taiyun/corrplot !148Whittaker, R. J. (1989). The Vegetation of the Storbreen Gletschervorfeld, Jotunheimen, Norway. III . Vegetation- Environment Relationships. Journal of Biogeography, 16, 413–433. Williams, J. W., Shuman, B. N., Webb, T., Bartlein, P. J., & Leduc, P. L. (2004). Late-Quaternary vegetation dynamics in north America: Scaling from taxa to biomes. Ecological Monographs, 74, 309–334. Wilson, J. W. (1957). Observations on the Temperatures of Arctic Plants and Their Environment. Journal of Ecology, 45, 499–531. Wilson, J. B., Gitay, H., Roxburgh, S. H., King, W. M., Tangney, R. S., & Roxburgh, H. (1992). Egler’s concept of “Initial Floristic Composition” in succession: Ecologists citing it don’t agree what it means. Oikos, 64, 591–593. Yarranton, G. A., & Morrison, R. G. (1974). Spatial Dynamics of a Primary Succession : Nucleation. Journal of Ecology, 62, 417–428. Young, A. T. (2002). Primary Succession at a High Arctic Glacial Foreland. Directed Study: University of British Columbia. !149APPENDICES !150Appendix A: Relocating Chapter 2 Study Area The following Figures and Tables provide detailed information about relocating and reconstructing the study areas in Chapter 2.  !151(a)                    (b)Figure A.1. The two types of markers left on the foreland. The first consists of a square metal marker (a), and the other consists of a bent over metal marking flag (b). Flagging tape was sometimes tied onto markers. All markers were marked with a metal tag, following the labelling scheme described in Figure A.2.!152Figure A.2. Layout of the study area. Thirty belts run parallel to the glacial margin, and each belt is divided into six zones. Black labels within the zones are the names used in 2016 during the survey. Blue belt names are the names given in 1995 to the belts by Jones (1997). Metal stakes were re-tagged and left on the foreland at all locations represented by an orange dot. These tags were labelled with a combination of the transect name (T5 - TB) and a distance measure given by Jones that represents the distance from the 1992 glacial margin (0 m - 260 m), both labelled in purple below. ID X Y          ID X Y1959Rock  -75.7732533145  78.8462916045 Transect B  TB-00   -75.7823781071  78.8456976073 TB-10   -75.7819537672  78.8457202697  TB-20  -75.7815343863  78.8457486156  TB-30  -75.7811360491  78.8457726365  TB-40   -75.7807156018  78.8458059065 TB-50   -75.7804133954  78.8458343586 TB-60   -75.780004627  78.8458587194 TB-70   -75.7795898527  78.8458923414  TB-80   -75.7791777294  78.8459294132 TB-90   -75.7787426815  78.8459666263 TB-100   -75.7782922765  78.8460069102 TB-110   -75.7779034173  78.8460525924 TB-120   -75.7775351257  78.846102925 TB-130   -75.7771457895  78.8461504914 TB-140   -75.7767605572  78.8461941729 TB-150   -75.776401837  78.8462457701 TB-160   -75.7760223082  78.8462996554 TB-170   -75.7756117521  78.8463451988 TB-180   -75.7751932752  78.8463864581 TB-190   -75.7748424558  78.8464322055 TB-200   -75.7744584572  78.8464819139 TB-210   -75.7740782267  78.846527319 TB-220   -75.7737032364  78.8465742944 TB-230   -75.7733325827  78.8466252301 TB-240   -75.7729269677  78.8466735372 TB-250   -75.7725508562  78.8467239378 TB-260   -75.7721849254  78.8467640865 TB-270   -75.7717708735 78.8468217919 Transect A  TA-00   -75.7823366389  78.8455818064 TA-10   -75.7818788481  78.8455990977 TA-20   -75.7815166753  78.8456183221 TA-30   -75.7810333745  78.84565755 TA-40   -75.7806094923  78.8456708852 TA-50   -75.7801605498  78.8457056011 TA-60   -75.7797819585  78.8457400514 TA-70   -75.7793616892  78.8457795224 TA-80   -75.7789432816  78.8458100394 TA-90   -75.778619594 78.8458454454  TA-100   -75.7781150447  78.8458915611 TA-110   -75.7777319226 78.8459345796  TA-120   -75.7773186962  78.8459751117 TA-130   -75.776906608  78.8460125845 TA-140   -75.7764652883  78.8460505007 TA-150   -75.7760425434  78.8460899366  TA-160   -75.7756469584  78.8461255217  TA-170   -75.7752115294  78.8461673433  TA-180   -75.7748090798  78.8462032309  TA-190   -75.7743850991  78.8462463221  TA-200  -75.7740030231  78.8462886599  TA-210   -75.7735830379  78.8463315409  TA-220   -75.7732211129  78.8463734065  TA-230   -75.7728067754  78.8464291056  TA-240   -75.7724233357  78.8464717629  TA-250   -75.77200758  78.8465112921  TA-260   -75.7716141563  78.8465503412  TA-270   -75.7712004594  78.8465917074 Transect 1  T1-1993  -75.7830881441  78.8453741076  T1-1994  -75.7835430245  78.8453728716  T1-1995  -75.7839640663  78.8453773034  T1-00   -75.7826407273  78.8453975116  T1-10   -75.7821863479  78.8454074449  T1-20   -75.7817372856  78.8454242653  T1-30   -75.7812548254  78.8454405508  T1-40   -75.7808935294  78.8454645809  T1-50   -75.7803748765  78.8454716731  T1-60   -75.7799699633  78.8454888759  T1-70   -75.7795227665  78.8455190142  T1-80   -75.779087716  78.8455495503  T1-90   -75.7786213271  78.8455865608  T1-100   -75.7781906282  78.8456091028  T1-110   -75.7777486562  78.8456342017  T1-120   -75.7773075466  78.8456605697  T1-130   -75.7768733025  78.8456906057  T1-140   -75.7764174911  78.8457236273  T1-150   -75.7760282652  78.8457618712  T1-160   -75.7755844575  78.8458072111  T1-170   -75.7752103365  78.8458470893  T1-180   -75.7746504746 78.8459041126  T1-190   -75.7742830238  78.8459451988  T1-200   -75.773721081  78.8459873538  T1-210   -75.7733002477  78.8460276884  T1-220   -75.7731231523  78.8460387342  T1-230   -75.7727057429  78.8460707087  T1-240   -75.7723085071  78.8460937252  T1-250   -75.7718605305  78.8461033688  T1-260   -75.7714545042  78.8461142769  T1-270   -75.770990194  78.8461261491
!153Table A.1. GPS coordinates corresponding the transect-distance labeling scheme described in Figure A.2. “1959_Rock” is the location of two large erratic boulders. ID X Y          ID X YTransect 2  T2-1993  -75.7835188374  78.845174467 T2-1994  -75.7838578758  78.8451514675 T2-1995  -75.7841749507  78.8451705549 T2-00   -75.7827506414  78.8452177979 T2-10   -75.7823543853  78.8452382362 T2-20   -75.7818931688  78.8452496555 T2-30   -75.7814628419 78.8452611215 T2-40   -75.7810075305  78.8452729121 T2-50   -75.7805370926  78.8452780098 T2-60   -75.7801038805  78.8453014396 T2-70   -75.7796791631  78.8453082618 T2-80   -75.7791867579  78.8453268778 T2-90   -75.7787696467  78.8453416157 T2-100   -75.7783159823  78.8453560854 T2-110   -75.7778892634  78.8453613224 T2-120   -75.7774064188  78.8453860381 T2-130   -75.7770048459  78.845410382 T2-140   -75.776535517  78.8454387313 T2-150   -75.7761241549  78.8454631596 T2-160   -75.7756559695  78.8454952627 T2-170   -75.7752361872  78.8455158641 T2-180   -75.7747709158  78.8455404241 T2-190   -75.7743489935  78.8455585365 T2-200   -75.7739095987  78.8455822278 T2-210   -75.7734720765  78.8456097751 T2-220   -75.7730287134  78.8456340733 T2-230   -75.7727072428  78.8457065432 T2-240  -75.7724049307  78.8457711479  T2-250   -75.7720972543 78.8458370638 T2-260   -75.77172899  78.8458941543 T2-270   -75.7713575152  78.8459489495 Transect 3  T3-00   -75.7829670474  78.8450208684 T3-20   -75.7820429887  78.8450343747 T3-40   -75.781114646  78.8450313788 T3-60   -75.7802007849  78.8450640216 T3-80   -75.7793186053  78.8450794804 T3-100   -75.7784171847  78.8451113436 T3-120   -75.7775407921  78.845157911 T3-140  -75.7766564985  78.8452058143  T3-160   -75.7757764921  78.8452659436 T3-180   -75.7748807426  78.8453179548 T3-200  -75.7739585415  78.8453622956 Transect 4  T4-1993  -75.783791531  78.8449528275  T4-1994  -75.7841687453  78.8449648921  T4-1995   -75.7844210057  78.8449789482  T4-00   -75.783247781  78.844919918  T4-20   -75.7823252882  78.8448560987  T4-40   -75.7813955128  78.844864645  T4-60   -75.7805271802  78.8448484814  T4-80   -75.7795740378  78.8448404054  T4-90   -75.7791509455  78.8448423298  T4-100   -75.7787725479  78.8448455704  T4-120   -75.7777889692  78.8448763161  T4-140   -75.7768942748  78.8448909854  T4-160   -75.7760035967  78.844941457  Transect 5  T5-1993  -75.7841670351  78.8447366265  T5-1994  -75.7846049109  78.8447568023  T5-1995  -75.7849315233  78.8447770327  T5-00   -75.7835902317  78.8447149709  T5-20   -75.7827324836  78.8446800872  T5-40   -75.7816994922  78.8446676007  T5-60   -75.7808754879  78.844649342  T5-80   -75.7799697623  78.8446248164  T5-100   -75.7791001688  78.8446090584  T5-120   -75.7781628108  78.8446313185
!154ID X Y  Beistad Glacier T1-end -78.9257968518 79.0400015655T1-start -78.9294469085 79.0396606515T2-start -78.929682648 79.0397652077T3-start -78.9306700929 79.0403346908T3-end -78.9266224385 79.0405973629T2-end -78.9257429562 79.0401891575Sverdrup Pass – Tear Drop Glacier T1-0m -79.7447383469 79.132343447T1-70m -79.7438024615 79.1328963809T1-125m -79.7435843291 79.1333760618T1-210m -79.7429635323 79.1343152193T2-0m -79.7433216127 79.1322483177T2-95m -79.742525872 79.1331831 T2-210m -79.7413812196 79.1345798527T3-0m -79.7420126654 79.1321783098T3-69m -79.7415852618 79.132779992T3-125m -79.7415248433 79.1332791716T3-215m -79.7406357054 79.1340840071!155Table A.2. GPS coordinates of the beginning and end points of the transects at Beistad Glacier and at Teardrop Glacier, Sverdrup Pass.Appendix B: Rare Species Figue B.1 presents maps showing the presence and abundance of less common species across Twin Glacier foreland. !156!157c)  Cerastium alpinuma)  Vaccinium uliginosum b) Cardamine bellidifoliad)  Chamerion latifoliumFigure B.1 a-d. Presence and abundance of vascular plant species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species was found.!158g)  Saxifraga cespitosae)  Erysimum pallasii f) Melandrium apetalumh)  Saxifraga nivalisFigure B.1 e-h. Presence and abundance of vascular plant species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species was found.!159k)  Stellaria longipesi)  Saxifraga tricuspidata j) Silene acaulisl)  Potentilla hyparcticaFigure B.1 i-l. Presence and abundance of vascular plant species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species was found.
!160o)  Festuca brachyphyllam)  Arctagrostis latifolia n) Carex aquatilisp)  Luzula nivalisFigure B.1 m-p. Presence and abundance of vascular plant species across the foreland. Coloured dots show a presence of the species, while faint crosses show an absence. Legends indicate the number of squares (out of 100) in which the species was found.Appendix C: Historical Data from the Farm  !161Figure C.1. Original arrangement of plots and treatments applied for the revegetation experiment set up in 1985.!162Table C.1. Diversity and evenness data from the plots in 1988.  For Treatment: NFS= Non-fertilized smooth; NFC = Non-fertilized coarse; FS = fertilized smooth; FC = fertilized coarse.  S = Richness = number of non-zero elements in row.  E = Evenness  = H / ln (Richness).  H = Shannon`s diversity index = - sum (Pi*ln(Pi)).  D = Simpson`s diversity index for infinite population = 1 - sum (Pi*Pi).  where Pi = importance probability in element i (element i relativized by row total)Plots Treatment Mean Stand.Dev. Sum Minimum Maximum S E H D’1 NFS 1.100 2.647 26.400 0.000 12.200 10 0.715 1.647 0.72702 NFC 1.175 3.020 28.200 0.000 12.200 8 0.675 1.404 0.69453 FS 0.742 1.518 17.800 0.000 6.000 9 0.812 1.785 0.79114 FC 0.467 1.056 11.200 0.000 4.600 10 0.752 1.732 0.75385 NFS 0.575 1.143 13.800 0.000 4.600 11 0.798 1.914 0.80076 NFC 0.592 1.264 14.200 0.000 5.600 10 0.783 1.803 0.77607 FS 0.625 1.077 15.000 0.000 3.600 10 0.873 2.010 0.83988 FC 0.525 1.435 12.600 0.000 6.800 10 0.652 1.500 0.66019 NFS 0.775 1.269 18.600 0.000 3.600 11 0.849 2.037 0.851210 FS 0.342 0.781 8.200 0.000 3.000 7 0.814 1.584 0.749611 NFC 0.850 1.946 20.400 0.000 7.400 7 0.805 1.566 0.749112 FC 0.675 1.559 16.200 0.000 7.200 10 0.752 1.732 0.745313 NFC 0.508 1.602 12.200 0.000 5.800 5 0.595 0.958 0.561714 NFS 0.558 1.387 13.400 0.000 5.800 10 0.695 1.601 0.712015 FC 0.750 2.267 18.000 0.000 10.800 9 0.593 1.304 0.593616 FS 0.342 0.689 8.200 0.000 2.800 8 0.861 1.791 0.7960Tundra N/A 1.192 1.989 28.600 0.000 7.900 20 0.747 2.238 0.8471Averages 0.694 1.568 16.647 0.000 6.465 9.7 0.751 1.683 0.7440

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0365968/manifest

Comment

Related Items