VARIABILITY OF TREELINE MOUNTAIN BIRCH ESTABLISHMENT UNDER

HERBIVORY PRESSURE

A Thesis

by

TYNAN CASSADY GRANBERG

Submitted to the Office of Graduate Studies of Texas A&M University in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

August 2012

Major Subject: Geography

Variability of Treeline Mountain Birch Establishment under Herbivory Pressure

Copyright 2012 Tynan Cassady Granberg

VARIABILITY OF TREELINE MOUNTAIN BIRCH ESTABLISHMENT UNDER

HERBIVORY PRESSURE

A Thesis

by

TYNAN CASSADY GRANBERG

Submitted to the Office of Graduate Studies of Texas A&M University in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

Approved by:

Chair of Committee, David M. Cairns Committee Members, Charles W. Lafon Fred E. Smeins Head of Department, Vatche Tchakerian

August 2012

Major Subject: Geography

iii

ABSTRACT

Variability of Treeline Mountain Birch Establishment Under Herbivory Pressure.

(August 2012)

Tynan Cassady Granberg, B.A., Yale University

Chair of Advisory Committee: Dr. David Cairns

Alpine and arctic treelines have been viewed as sensitive indicators of global climate change. While many treelines have advanced under warmer climate regimes in recent decades, the response has not been uniform. Some of this variability may be attributable to the impacts of herbivores. This study investigates the interacting effects of herbivory, climate, and understory vegetation on mountain birch establishment at treeline in the Scandes Mountains of northern Sweden. An extensive dendrochronological database was created to determine periods of establishment, which were then regressed against reindeer (Rangifer tarandus, L.) population data and historical climate data. Vegetation classifications were also created and analyzed to determine if establishment patterns vary by understory vegetation type. I have tested the hypothesis that tree establishment varies within the treeline ecotone and that high reindeer stocking levels negatively impact establishment.

Weakly positive responses to herbivory were observed in patterns of tree establishment at treeline. This indicates that reindeer may modestly promote treeline advance at low densities, contradicting some previous research, but many of the results

iv

were not statistically significant. The climate variables found to have significant relationships with establishment were inconsistent across herding districts and aggregation levels. No connections between vegetation assemblages and establishment or between vegetation assemblages and reindeer use were observed.

v

ACKNOWLEDGEMENTS

First of all, I’d like to thank my advisor and committee chair, Dr. David Cairns, for giving me the opportunity to take the reins of this research project. His guidance and faith in my abilities were unwavering. Thanks also go to my other committee members,

Dr. Charles Lafon and Dr. Fred Smeins, for their unique insight and challenging questions, and to the Biogeography Research Cluster, an amazing group of grounded, dedicated scientists.

None of this would have been possible without help in the field, selflessly provided by Dr. Cairns, Dr. Lafon, Dr. Jon Moen of Umeå University, Adam Naito, and

Beeyoung Gun Lee. Special thanks go to Michelle Mouton and Brandon Wilcox, both of whom were stricken by mysterious, undergrad-targeting Nordic pathogens while conducting fieldwork in Sweden. Brandon—along with David Ethridge, Miguel Salazar,

Cody Keesee, Hilary Cooksley, Rachel Stuteville, and Laura Rudolf—also helped keep my sanity (and fingers) intact by completing vast amounts of microtome work, a key step in preparing our samples for dating. I’d like to thank Lin Bustamante of the Texas

A&M Veterinary School, as well, for the use of her microtome when our unit was under the weather.

Gratitude goes to the National Science Foundation and the Texas A&M

University Geography Department Graduate Enhance Fund for funding the project.

vi

Above all, I thank my mother for inspiring my love of learning, my father for instilling the work ethic necessary to attain that learning, and my teacher and good friend

Clair Thomas for teaching me just how much fun it could be. Thank you so much.

vii

TABLE OF CONTENTS

Page

ABSTRACT ...... iii

ACKNOWLEDGEMENTS ...... v

TABLE OF CONTENTS ...... vii

LIST OF FIGURES ...... ix

LIST OF TABLES ...... xi

1. INTRODUCTION ...... 1

1.1 Determinants of Treeline Position ...... 3 2.2 Establishment at Treeline ...... 8 2.3 Herbivore Influences on Treeline ...... 9

2. STUDY AREA ...... 13

3. METHODS ...... 17

3.1 Field Methods ...... 17 3.2 Laboratory Methods ...... 18 3.3 Age Structures ...... 19 3.4 Influence of Herbivory ...... 20 3.5 Influence of Climate ...... 21 3.6 Interacting Effects of Climate and Herbivory ...... 22 3.7 Influence of Vegetation ...... 23

4. RESULTS ...... 25

4.1 Age Structures ...... 25 4.2 Influence of Herbivory ...... 27 4.3 Influence of Climate ...... 27 4.4 Interacting Effects of Climate and Herbivory ...... 29 4.5 Influence of Vegetation ...... 29

5. DISCUSSION ...... 33

viii

Page

6. CONCLUSIONS ...... 38

REFERENCES ...... 39

APPENDIX A: FIGURES ...... 52

APPENDIX B: TABLES ...... 92

ix

LIST OF FIGURES

FIGURE Page

1 A mountain birch (Betula pubescens ssp. czerepanovii) seedling growing within a treeline study site, near Ritsem, Sweden...... 52

2 Geometrid moth larvae in the process of defoliating a mountain birch on the outskirts of Sarek National Park, Sweden...... 53

3 Conceptual model for differential impacts of herbivory pressure on different treeline types, presented in Cairns and Moen (2004)...... 54

4 Map of herding districts, weather stations, and study site locations ...... 55

5 Vegetation map for Norrbotten County study sites ...... 56

6 Vegetation map for Västerbotten County study sites ...... 57

7 An example of a diffuse treeline, near Nikkaluokta, Sweden ...... 58

8 Semi-domesticated reindeer (Rangifer tarandus), the primary herbivore at Swedish treelines ...... 59

9 A seasonal Sámi herding hut in Sarek National Park, Sweden ...... 60

10 A new reindeer calf is given the identifying earmark of its owner ...... 61

11 Historical reindeer stocking levels ...... 62

12 Reindeer stocking levels by year at study herding districts in Norrbotten County ...... 63

13 Reindeer stocking levels by year at study herding districts in Västerbotten County ...... 64

14 1 of 69 study sites established at treeline in Sweden ...... 65

15 District age class structures (annual data)...... 66

16 District age class structures (pentadal data)...... 67

x

17 District age class structures (decadal data)...... 68

18 Age class structures for all herding districts combined (annual, pentadal, and decadal data)...... 69

19 The impact of reindeer numbers on recent establishment...... 71

20 The relationship between stocking level and establishment was only significant in Sorkaitum ...... 72

21 Relationships between district stocking level and establishment were not significant in 4 of 5 districts...... 73

22 When grouped by pentad, the relationship between stocking level and establishment in Sorkaitum was no longer statistically significant ...... 75

23 Proportion of high v. low recruitment year under four different temperature/reindeer scenarios (All Districts, Laevas, and Girjas)...... 76

24 Proportion of high v. low recruitment year under four different temperature/reindeer scenarios (Sorkaitum, Vilhelmina Norra, and Vilhelmina Sodra) ...... 78

25 Pentadal data: Proportion of high v. low recruitment years under four combinations of high/low reindeer populations and high/low temperatures...... 80

26 Proportion of high v. low recruitment years under sixteen combinations of high/low/medium-high/medium-low (quartiles) reindeer populations and temperatures...... 82

27 Dendrogram of vegetation classifications produced via cluster analysis ... 84

28 Vegetation groups obtained using Indicator Species Analysis ...... 85

29 Distribution of vegetation classes at study sites in Norrbotten County ..... 87

30 Distribution of vegetation classes at study sites in Västerbotten County... 88

31 Age class structures by vegetation classification (annual data)...... 89

32 Age class structures by vegetation classification (pentadal data)...... 90

33 Age class structures by vegetation classification (decadal data)...... 91

xi

LIST OF TABLES

TABLE Page

1 Common vegetation types near treeline in Sweden ...... 92

2 Summary statistics for study sites, divided by herding district...... 93

3 Linear regression statistics for establishment as a function of mean monthly temperatures at different herding districts...... 94

4 Linear regression statistics of establishment as a function of the following winter's precipitation at different herding districts...... 100

5 Linear regression statistics of establishment as a function of degree days at different herding districts...... 100

6 Relative abundance of each plant species in the vegetation classifications 101

7 Relative frequency of each plant species in the vegetation classifications. 104

8 Indicator values of each plant species in the vegetation classifications. .... 107

9 Descriptions of vegetation classifications derived from cluster analysis. .. 110

10 Comparison of age class structures for herding districts within the 4 different vegetation classifications ...... 111

11 Comparison of age class structures for vegetation classifications within Sorkaitum ...... 111

12 Comparison of age class structures for vegetation classifications within Vilhelmina Norra ...... 112

1

1. INTRODUCTION

Vegetation distributions are closely tied to climate, soil, and disturbance regimes and the transition zones between vegetation types—known as ecotones—often mark critical thresholds of one or more environmental variables. One highly visible ecotone is treeline, where forests give way to tundra, grasslands, or other ecosystems. Alpine and arctic treelines are particularly compelling: they have been frequently viewed as indicators of global climate change due to the importance of temperature in determining their position (Kupfer & Cairns, 1996; Harsch et al., 2009). Some treelines have responded to climate warming with new tree establishment above the existing ecotone, but their responses have not been uniform. This variability is attributable to a number of factors, including past human land use, vegetation history, climate history, and herbivory, among others (Körner & Paulsen, 2004).

As the scope of investigation is narrowed from global to regional, landscape, and local scales—or from long to relatively short time frames—the predictive abilities of any one factor lessens accordingly. Site- and species-specific factors play a significant role in this variability (e.g., Mamet & Kershaw, 2011). Even at regional scales, treeline position does not advance in lockstep with climate change (Holtmeier et al., 2010).

In Fennoscandia, semi-domesticated reindeer have been a fixture of the landscape for thousands of years and their grazing may play an important role in influencing the

______This thesis follows the style of the Journal of Biogeography.

2 spatial and temporal heterogeneity of treeline responses to climatic conditions (Oksanen et al., 1995; Van Bogaert et al., 2011).

By improving our understanding of the impact of herbivory on treeline, we can improve our ability to use these sensitive ecotones as bellwethers of climate change and to predict future treeline positions. In northern Sweden, reindeer herding has been an economically and culturally important practice for the indigenous Samí people for centuries. Grazing by their animals at the transition between forest and tundra may influence the relative stability of the ecotone by shifting competitive balances, altering soil temperatures, and trampling and defoliating seedlings. The sustainability of the Samí herding lifestyle is dependent on the stability of the current landscape structure, as treeless tundra areas serve as critical, insect-free summer pasture for the herds. Treeline position also has important ramifications for carbon sequestration and similar climate change related phenomena.

This study investigates the interacting effects of herbivory, climate, and understory vegetation on mountain birch establishment at treeline in the Scandes

Mountains of northern Sweden. An extensive dendrochronological database was created to determine periods of establishment, which were then regressed against reindeer

(Rangifer tarandus, L.) population data and historical climate data. Vegetation classifications were also created and analyzed to determine if establishment patterns vary by understory vegetation type (i.e., herbaceous and low-growing shrub communities). I have tested the hypothesis that tree establishment varies within the treeline ecotone and that establishment is negatively influenced by high reindeer stocking levels.

3

1.1 Determinants of Treeline Position

Climate—particularly temperature—is considered the preeminent limiting factor for tree growth in arctic and subalpine conditions (Körner, 1998, 2003). At increasing elevation and latitude trees face concomitant climatic stresses. Low air and soil temperatures combine with high wind speeds, intense UV-exposure, short growing seasons and (at altitude) relatively low atmospheric CO2 levels to produce dieback and impaired tissue formation (Tranquillini & Benecke, 1979; Wardle, 1981; Hadley &

Smith, 1986; Cairns, 2001). At global scales, treelines match up well with isolines of temperature (Holtmeier & Broll, 2005). However, treeline and temperature become incongruent when viewed at regional and local scales and much research has gone into explaining this variability. While mean climatic conditions may fall within a tree’s tolerance range at a given location, the extremes or periodicity of these conditions may not be conducive to tree growth (Tranquillini & Benecke, 1979). All trees require a minimum period of time during the growing season in which they do not endure freezing temperatures. Mature trees can tolerate more extreme conditions than seedlings, so an episode of favorable climate can allow seedlings to establish and mature above the existing treeline and then persist when climate reverts to the mean (Holtmeier & Broll,

2005; Harsch & Bader, 2011). However, warm temperatures may be accompanied by drought stress, which limits growth (Lloyd & Fastie, 2002), and some have pointed out that the ratio of temperature to precipitation, rather than any absolute value, may be more important in many treeline systems (Ohse et al., 2012). Non-growing season climate variables such as snow cover are critical as well. Snow helps maintain higher winter soil temperatures, protects against desiccation, and provides moisture in the summer, but

4 excess snow cover can shorten the effective growing season. While early snowmelt has been shown to yield greater establishment and lower seedling mortality rates (Taylor,

1995; Kullman, 2002; Elliott & Kipfmueller, 2011; Barbeito et al., 2012), higher growth rates have been reported in years with relatively late snowmelt (Barbeito et al., 2012).

The climate history of a particular treeline is extremely important in explaining its location. A large body of research has focused on past fluctuations in treeline location as well as the current responses of treelines to climate change. Results have varied by study site and region, with some finding increased stand density (Payette & Filion, 1985), changes in growth form (Vallée & Payette, 2004), or species change (Landhäusser et al.,

2010) in lieu of treeline advance. Treeline has receded or remained unchanged in some locations (Butler et al., 1994; Payette, 2007). Still, many studies support the hypothesis of increasing treeline elevation with warming climate (Cullen et al., 2001; Leitner &

Gajewski, 2004; Dalen & Hofgaard, 2005; Jiménez-Moreno et al., 2008; Kullman &

Öberg, 2009).

The history of human land use is another significant factor in determining treeline position. The impact of this land use can alter the potential for trees to invade tundra areas far into the future (Staland et al., 2011). In the European Alps and elsewhere, logging for mining and other purposes has lowered the treeline well below the altitudinal limit prescribed by climate (Tranquillini & Benecke, 1979; Slatyer & Noble, 1992).

Some areas have been rapidly re-colonized by trees after human impacts have ceased and care must be taken not to attribute such advances to climate warming (Holtmeier & Broll,

2005; Camarero & Gutiérrez, 2007). For these reasons, some researchers have found it

5 useful to describe particular treeline environments as cultural—rather than natural— landscapes (Emanuelsson, 1987; Hofgaard, 1997).

Particularly in mountainous regions, treelines may form below climatic constraints due to topographic, geomorphic, or edaphic limitations. Microtopographic gradients can produce highly variable soil moisture and nutrient availability, even on relatively flat ground (Anschlag et al., 2008; Yang et al., 2011). Such gradients can be the deciding factor in tree establishment, especially in extreme treeline environments.

Nitrogen-acquisition has also been demonstrated to impact establishment and winter survival of mountain birch seedlings (Sveinbjörnsson et al., 1992; Weih & Karlsson,

1999). In the northern hemisphere, alpine treelines can generally be found higher on south-facing slopes due to increased direct solar radiation. Sheltered, leeward areas also tend to support trees at higher elevations than their windward neighbors (Resler et al.,

2005)—though the greater accumulation of snow and late melt-off in these protected areas may also negatively impact survival (Holtmeier, 2009).

Vegetation zones continue to adjust to geomorphic processes initiated by continental scale glaciation in the Pleistocene (Butler et al., 2009). These processes include debris flows, solifluction, and talus deposits that commonly suppress the ability of trees to establish at higher elevations (Butler & Walsh, 1994; Walsh & Butler, 1997;

Butler et al., 2009). Particularly in systems with steep slopes and extensive alpine areas, avalanches can be the preeminent control on treeline position (Butler & Walsh, 1990;

Walsh et al., 2004). Avalanches not only uproot trees, but also excavate soil that is then deposited downslope (Butler, 2001). Soil is commonly a limiting resource at high elevations, so avalanches combine with other edaphic processes such as frost heaving to

6 negatively impact tree establishment. Not all geomorphic processes inhibit treeline establishment, however, as the denudation process known as “turf exfoliation” has been shown to remove competing vegetation cover and reduce soil compaction, creating favorable microsites for seedling establishment (Pérez, 1992; Butler et al., 2004).

Harsch and Bader (2011) have attempted to resolve the debate over the dominant controls of treeline by categorizing treelines into four categories characterized by different spatial patterns: diffuse, abrupt, island, and krummholz. Each form, they argue, is governed by a different tree performance mechanism (growth limitation, dieback, or seedling mortality) and driven by stresses (e.g., desiccation, freezing damage, etc.) that are modified by neighbor interactions. Diffuse treelines are characterized by a gradual decrease in tree density and height along the ecotone caused by decreasing growing season temperatures with elevation. Trees in these ecotones are not strongly deformed, nor do they exhibit steep mortality gradients, indicating that dieback and seedling mortality are not significant in these systems. Instead, growth is the primary limitation.

Diffuse treelines have exhibited relatively direct responses to climate change (80% have advanced), while the remaining categories have displayed more complicated behaviors

(only 25% have advanced). This is because diffuse treelines are thought to be in equilibrium with growing season temperature.

Of the three remaining treeline categories, abrupt treelines are most frequently found at elevations lower than prescribed by climate. Trees at the limits of abrupt treelines remain tall and exhibit high fecundity, indicating that they are neither growth nor dieback limited. Instead, seedling mortality increases drastically beyond treeline because establishment requires facilitation from neighboring trees. In other words,

7 seedlings may establish within the microclimates produced by sheltering trees, but die rapidly when expanding beyond the forest limit, where positive feedbacks disappear.

Island treelines, as their name suggests, are patchy. They are maintained primarily by seedling mortality and dieback. Patches often form around minor topographic features such as boulders and expand in a leeward direction, with dieback occurring on the windward side (Alftine & Malanson, 2004). Reproduction in these treelines is often limited to layering (Bekker, 2005) and the degree of clumping has been used as a proxy for environmental severity (e.g., Bekker & Malanson, 2008). Island treelines may serve as transition states when the patches are outposts of an advancing treeline in the process of infilling, but patches may also remain stable when positive local feedbacks switch to negative feedbacks at a certain distance (Wilson & Agnew, 1992).

While island treelines may infill or advance when climate ameliorates, this is contingent on a number of variables beyond growing season temperature, such as snow cover and wind (Harsch & Bader, 2011).

Although not typically considered in treeline studies—because they do not fit the traditional 2 m height minimum—krummholz treelines are considered a separate category by Harsch and Bader (2011). Krummholz are often able to change their growth form when climatic conditions ameliorate (Smith et al., 2003). These treelines may grow either in patches or as more continuous belts and are strongly influenced by winter dieback—rather than summer growth limitation. Seedling mortality is extremely high.

8

1.2 Establishment at Treeline

By definition, an advancing treeline must establish new stems beyond the extent of mature trees at a rate exceeding mortality (Figure 1). With their large root systems and biomass, mature trees can tolerate more extreme and variable conditions than seedlings, which can be killed by a single severe storm during the growing season (Smith et al.,

2003). In fact, mortality of seedlings in their first year has been shown to exceed 90% in many treeline environments (Cui & Smith, 1991; Maher & Germino, 2006). Favorable temperatures and moisture regimes during early life stages are thought to be crucial for recruitment, promoting growth and shortening the time to maturity (Danby & Hik, 2007;

Mamet & Kershaw, 2011).

One determinant of treeline establishment may be the availability of viable seeds, as reproductive success from seed has been shown to diminish with increased elevation

(Sveinbjörnsson et al., 1996). If the treeline is elevational, rather than latitudinal, low elevation seed sources may exist nearby. However, polar treelines may exist far from consistent seed sources, instead producing viable seeds on their own only during periods of favorable climate (Sveinbjörnsson et al., 1996). Reproduction at treeline commonly occurs vegetatively, through root sprouting (in the case of Betula species) or adventitious roots (in the case of many coniferous species) (Stevens & Fox, 1991).

While some studies have successfully associated recent seedling establishment with climate variables (Wang et al., 2006; Lv & Zhang, 2011), others have observed a disconnect between treeline advance and mean climate conditions that may be due to a difference in mechanisms limiting seedling survival and tree growth (Harsch & Bader,

2011). While tree growth is closely linked to growing season temperature, seedlings are

9 less tightly coupled to atmospheric conditions (due to their short stature) and their survival may be mitigated by different interacting mechanisms. Van Bogaert (2011) found that cool summer temperatures suppressed radial growth in treeline birches but had no significant relationship with establishment. Seedlings are less tolerant of desiccation, drought stress, overheating, and photodamage than their elders (Hessl & Baker, 1997;

Danby & Hik, 2007) and their survival is more closely linked with winter and spring snow cover (Barbeito et al., 2012). Boulders can help ameliorate all of these conditions by sheltering seedlings from wind and excessive light and by enhancing snow deposition

(Resler et al., 2005; Batllori et al., 2009), thus promoting treeline advance.

Climate-establishment relationships may be further obfuscated by the presence of both clonal and seed reproduction at treeline. Treeline advance may be slower than expected in situations where seedling facilitation from neighbor interactions (shading, wind shielding, etc.) is particularly important or where clones predominate (Eränen &

Kozlov, 2008). In addition, seedlings are more likely to be significantly impacted by herbivory, due to their stature and the relative palatability of young leaves (Boertje,

1984).

1.3 Herbivore Influences on Treeline

This study focuses on the impact of herbivory on establishment at treeline. While not characteristic of all treeline systems, herbivores may significantly influence the stability of the ecotone where they occur through selective feeding or by triggering differential plant responses to damage (French et al., 1997; Cairns & Moen, 2004;

Herrero et al., 2011). Since treeline species exist at the edge of their range limits, even

10 relatively minor impacts from herbivores could have a significant influence on treeline dynamics.

In northern Sweden, mountain birch (Betula pubescens ssp. czerepanovii) are the preferred summer forage for semi-domesticated reindeer. Studies have shown that frequent grazing by sheep (Ovis aries) limits the establishment of new birch recruits at treeline in neighboring Norway (Speed et al., 2010, 2011), but reindeer are stocked at much lower densities and feed in a much less intensive manner than sheep. Still, reindeer are large ungulates and their influence on the position and structure of treeline has been previously reported (Oksanen et al., 1995; Oksanen & Virtanen, 1995; Helle, 2001).

Oksanen (1995) observed sudden, distinct treelines in areas of high reindeer presence and more gradual ecotones in areas with limited reindeer browsing. Van Bogaert et al. (2011) found that in a period of climate warming, disturbance by reindeer and the herders that accompany them suppressed both radial growth and recruitment near Lake Torneträsk.

Some studies of reindeer pastures have reported reductions in birch canopy area, biomass, height, and stem density, while grasses and sedges responded positively to reindeer grazing (Lempa et al., 2005). However, a 7-year exclosure study in northern Scandinavia found no change in birch cover when herbivores were removed from the system (Moen &

Oksanen, 1998). Another exclosure study, this time conducted over four years at sites along the length of the Scandes mountains, also found little to no effect of reindeer exclusion on birch forest vegetation communities (Eriksson et al., 2007).

Birch forests may also be defoliated and treelines depressed by outbreaks of herbivorous geometrid moths (Kallio & Lehtonen, 1973; Helle, 2001; Karlsson et al.,

2005) (Figure 2), which may promote establishment of aspen (Populus tremula) near

11 treeline (Van Bogaert et al., 2010). Reindeer and moths have been shown to have important interactive effects on the mortality of birch saplings (Lehtonen & Heikkinen,

1995), while moose (Alces alces) have been shown to combine with moth outbreaks to influence the competitive dynamics between mountain birch and aspen (Van Bogaert et al., 2009). Besides foliage consumption, herbivores may limit the advance of treeline by trampling seedlings and compacting the soil (Kozlowski, 1999; Herder & Niemelä,

2010). These negative impacts may be mitigated to some extent by positive effects, however. Studies have suggested that disturbance may promote treeline advance in the case of fire (Johnstone & Chapin, 2006) and thermokarst (Lloyd et al., 2003). Herbivory may play a similar role by reducing competition from tundra vegetation, increasing soil temperatures, and extending seed dispersal (Olofsson et al., 2001; Dullinger et al., 2004;

Van der Wal & Brooker, 2004). Animal activity may serve as the initial disturbance that triggers turf exfoliation at treeline (Pérez, 1992; Butler et al., 2009), a process that creates favorable microsites for seedling establishment. Significant correlations between increased reindeer populations, decreased lichen biomass, and increased tree biomass have been reported on the Finnmark plateau in Norway (Tommervik et al., 2004;

Tommervik et al., 2009). Cairns and Moen (2004) have presented a conceptual model that attempts to integrate the positive and negative effects of grazing based primarily on the relative palatability of the tundra vegetation (Figure 3). In this model, Mountain birch establishment at treeline is expected to follow a negative sigmoid curve under increasing herbivory pressure, with birch foliage preferred over field layer vegetation.

In this study, I investigate the effects of reindeer herbivory, tundra vegetation, and climate on the establishment of new birch stems within the treeline ecotone in northern

12

Sweden. I will follow the convention of defining treeline as existing from the forestline

(the uppermost limit of closed canopy, continuous forest) to the limit of tree-forming individuals (> 2m) (Moen et al., 2008). I hypothesize that tree establishment varies within the ecotone in northern Sweden—both spatially and temporally—and that this variation is driven by an inverse relationship between reindeer numbers and establishment.

13

2. STUDY AREA

The study area is located in the northern Scandes Mountains of Norrbotten and

Västerbotten Counties, near the 68th parallel in sub-arctic Sweden (Figure 4). The

Scandes are an ancient remnant of the Caledonian mountains. While relatively low in elevation (reaching ~ 2000 m), their position in the far north provides true alpine conditions. Mean annual temperature is approximately -1.0 º C. Mean annual precipitation varies greatly by elevation and proximity to Norway (and the Norwegian

Sea beyond), ranging from 500 to > 2000 mm annually. More than 750,000 hectares are protected as national parks and these areas have collectively been granted a United

Nations World Heritage designation. It is widely considered the largest remaining wilderness in Europe, though there is evidence of Sámi habitation since ~ 9,000 B.C.

(Ingman & Gyllensten, 2006).

Vegetation in the northern Scandes is a mosaic of heaths, meadows, bogs, shrublands, and deciduous forests (Figures 5 and 6). Several broadly scaled vegetation maps exist for the region (Andersson et al., 1984; Rafstedt, 1987). These are based on physiognomic and phytosociological factors and utilize infrared aerial imagery to classify broad swaths of the mountains (Eriksson et al., 2007). The most common vegetation types found near treeline are presented in Table 1 (adapted from Green, 2005; Eriksson et al., 2007)

At finer scales, a number of studies have resulted in vegetation classifications that subdivide one or more broad classes along environmental gradients (Jonasson, 1981;

Svensson & Rosswall, 1984), subdivide existing classes (Arnborg, 1990), or reclassify

14 small geographic regions (Sonesson & Kvillner, 1980; Krahulec et al., 1986). The classification presented in this study focuses on vegetation communities within the treeline ecotone. Vascular plants were sorted into classifications based solely on their phytosociology, then compared statistically to look for differences in reindeer use or birch establishment between classifications.

The mountain birch (Betula pubescens ssp. czerepanovii) is the exclusive treeline species in the northern Scandes, with an elevational limit of 600 to 800 m. Norway spruce (Picea abies) and Scots Pine (Pinus sylvestris) are common at lower elevations.

Birch trees are unique among northern hemisphere treeline genera in that they are deciduous. Monocormic (single-stem) individuals are typically found in areas of deep winter snow accumulation, while polycormic (multiple-stem) forms are more common on dry sites (Sonesson & Hoogesteger, 1983). Both growth forms become generally more stunted near the treeline, often forming krummholz patches beyond the ecotone.

Reproduction occurs by seed as well as through root-sprouting. Within the framework established by Harsch and Bader (2011), the treelines sampled in this study fall within the

‘diffuse’ category (Figure 7). Diffuse treelines have shown the strongest positive response to recent climate warming.

Mountain birch are browsed by semi-domesticated reindeer (Figure 8) raised by the indigenous Sámi people (Müller-Wille et al., 2006) (Figure 9). No wild reindeer have existed in Sweden since the late 18th-century (Moen & Danell, 2003). Reindeer are primarily harvested for their meat, though they used to be an important source of milk, hides, and transportation for the Sámi. The Sámi are organized along family lines into administrative unions called samebys, each with exclusive rights to summer and winter

15 grazing areas for their reindeer. In mountain samebys, reindeer are released in the high country during the summer and gathered up only twice: once to mark new calves early in the summer (Figure 10), and once in the fall to ship some animals to slaughter and the rest to the lowlands to last out the winter. For the remainder of the summer, the reindeer are free to roam hundreds of thousands of hectares of birch forest and tundra. Their movements are driven by conflicting motivations to maximize forage quality while avoiding insects (Hagemoen & Reimers, 2002; Mårell & Edenius, 2006). Insects are diminished—as is forage quality—at exposed, high elevation sites, while insects are maddeningly numerous in the more productive, low elevation, forested areas. Because of this, reindeer typically spend much of the long summer days above treeline, returning to the forests only in the evenings when insect activity is minimized (Skarin et al., 2010).

In many parts of Sweden, therefore, reindeer feeding and movements through the treeline ecotone follow diurnal patterns. Weather patterns (which influence insect levels) and tourists also influence reindeer movements, though habituation to hikers can reduce the impact of human disturbance (Skarin et al., 2010).

Reindeer populations in Sweden have fluctuated considerably since the first available census data in 1885. Numbers have ranged from a low of 162,000 animals in

1923 to a high of 300,000 in 1990 (Figure 11) (Swedish Board of Agriculture, unpublished). There is no trend in the data, as even the historically high stocking level of

1990 involved only 4,000 more reindeer than 1890. Other peaks in the data include the early 1930s and mid-1950s. The steady decline in numbers from 1885 to 1923 has been attributed to disease (Bernes, 1996), while disruptions from continent-wide hostilities in the early 1940s may have been responsible for the ebb during that era. Bernes (1996)

16 suggests that suppressed numbers in the 1960s and 1970s were caused by degraded winter grazing conditions, which may have been brought about by high grazing pressure on winter lichens in combination with acute ice formation on the ground. The primary limit on reindeer carrying capacity is the availability of lichens in winter (Klein, 1968).

Despite reduced Samí participation in reindeer husbandry (Beach, 1993), relatively high numbers of reindeer have been maintained in recent times due to the introduction of motor vehicles (helicopters and all-terrain vehicles), along with modern veterinary medicine and winter feed supplements (Kumpula, 2001; Lempa et al., 2005). In 2008, there were > 250,000 reindeer in Sweden.

Six herding districts were studied in the course of this research. Each is the province of a specific sameby and may be grazed at a maximum stocking level determined by the County Administration Board (Moen & Danell, 2003). These caps are based on a combination of vegetation inventories, local knowledge, and historical numbers. Laevas (9 study sites), Girjas (8 study sites), and Sorkaitum (21 study sites) herding districts experience moderate summer grazing densities of between 7 and 8 reindeer/km2, while Baste (4 sites) is typically grazed at a rate of 20 reindeer/km2. Baste is also the most alpine of the districts used in this study, with trees located only along its periphery. Vilhelmina norra (20 sites) and sodra (5 sites), located below the Arctic Circle in Västerbotten County, are grazed at an average density of 12 and 9 reindeer/km2 respectively (Swedish-Norwegian Border Grazing Commission, unpublished). Stocking levels in Norrbotten (Figure 12) and Västerbotten (Figure 13) Counties have followed national patterns. The great majority of Swedish reindeer husbandry takes place in

Norrbotten County.

17

3. METHODS

3.1. Field Methods

Sixty-nine field sites were established within the 6 different herding districts listed above (Figure 2). When selecting field sites, care was taken to avoid hillsides with calcareous bedrock, which commonly supports vastly different plant communities. Each field site consisted of a 10 x 50 m plot within the treeline ecotone. The narrow top edge was established near the upper limit of treeline (defined as the limit of individual trees reaching 2m in height) with the site extending downslope in the direction of the forestline

(the limit of closed canopy, continuous forest) (Figure 14). Sites were selected by navigating to random GPS coordinates on accessible hillsides and then moving up or downslope to the treeline ecotone, or by walking random distances along the ecotone.

Random distances were determined with a stopwatch. One crewmember held the running stopwatch and another announced when to stop it, at which point the digit in the 1/100th of a second column was recorded, and the watch was restarted. This was repeated three times, after which the recorded numbers were combined to form the three digits of the distance to be walked (limited to between 50 and 300 m).

A 1x1 m quadrat was placed randomly within each 10 x 50 m site to collect relative abundance data for understory vegetation, with all vascular plants identified to the species level. A rough map of each site was also sketched by hand to note microtopographic features (> ~ 3 m) and a ratio of concave to convex topography was estimated. Soil characteristics were not within the scope of this investigation, but since vegetative differences between sites may be linked to variation in soil temperature,

18 moisture, and nutrient relations (Weih, 1998), they may serve as indicators of underlying edaphic differences between sites. A general description was also recorded for each site, describing notable plants observed outside of the quadrat, presence of standing water or wet soil, significant amounts of exposed rock, etc.

Reindeer use of the site was estimated using a pellet group count (Wal et al.,

2001). This method was first used in the US as a census technique for deer (Bennett et al., 1940; Eberhardt & Van Etten, 1956). The amount of fecal material provides an effective proxy account of the amount of reindeer use, combining the number of reindeer and the amount of time spent on site. It does not, however, differentiate between short periods of intense herbivore use and more prolonged periods of light use. Reindeer

Pellets generally remain visible for 3-5 years after being deposited in these environments.

Within each site, basal cross-sections were collected for all live mountain birch seedlings and saplings < 5 cm in diameter. It is understood that both seed- and root- sprouted individuals occur at treeline in Sweden, but no attempt was made to differentiate between them for the purposes of this study—both were considered new establishment.

Henceforth, all small stems will be referred to as seedlings. Larger saplings and trees were counted and their diameters recorded, but samples were not collected for these individuals due to time limitations and the tendency for older birch trees to contain rotten pith. Dead trees were not measured or collected.

3.2. Laboratory Methods

In the lab, all birch cross-sections were dried and then sanded with progressively finer sand paper (80-400 grit). A phloroglucinol dye was then applied to improve ring

19 visibility (Patterson, 1959). Next, samples were placed in a 10% HCl solution, which reacted with the phloroglucinol to stain the lignin. After a rinse, cross-sections of relatively small samples (~ < 2 cm) were mounted in embedding medium and trimmed with a microtome to a thickness of 5 μ. These slices were then imaged with a ZeissTM digital microscope camera. Images were dated using dendrochronology software

(WinDENDROTM). The cellular structure of birch trees made automation within the software impractical, so all rings were tallied by the observer. Larger samples (~ 2-5 cm) were too wide to be mounted on slides using this method and were instead dated using a boom-mounted Fisher Scientific stereomicroscope at varying magnifications. The stereomicroscope was also used for small samples that were unable to be sliced cleanly with the microtome. All measurements were made by a single individual to maintain consistency. These methods are the most accurate currently available, especially considering the unreliability of age/diameter relationships for trees located at treeline.

Because of the mixed methods for determining tree ages—and the slight reduction in accuracy with the stereomicroscope vs. the slide-mounted technique (D.M. Cairns, unpublished)—the most meaningful results are likely to come when the data are aggregated by pentad or decade.

3.3. Age Structures

Age class structures were produced for each herding district and vegetation classification. A typical age class structure should consist of a large number of young individuals dwindling to relatively few older individuals along a negative exponential or power function curve (Daniels & Veblen, 2004). This is due to high seedling mortality

20 and factors such as crowding and competition as trees age. By fitting such a curve to the data and calculating the residuals, it is possible to see which age classes are over- or underrepresented in the data and thus which years had more or less establishment than expected (Szeicz & Macdonald, 1995; Mamet & Kershaw, 2011). In this manner, variation in establishment within the treeline ecotone was determined. Residuals were calculated for annual, pentadal, and decadal data.

3.4. Influence of Herbivory

The influence of reindeer herbivory on contemporary establishment (past 5 years) was tested using simple linear regression of establishment as a function of reindeer use at each site (estimated for the past 3 to 5 years using pellet group counts) (Fritts, 1976;

Daniels & Veblen, 2004). To lengthen the time scale, the spatial scale was broadened.

Sites within the same herding district were grouped and the number of birch stems in each age class regressed against district-wide stocking levels, for which historical data

(1931 to the present) was obtained from the Samí Parliament and Swedish Board of

Agriculture (Swedish Board of Agriculture, unpublished). Regressions were performed using both annual data and five-year averages of reindeer stocking and birch establishment. However, reindeer stocking was only recorded intermittently until 1961 in

Norrbotten and 1994 in Västerbotten, so 5-year averages of stocking were only calculated when there were counts in 3 of the 5 years in a pentad.

21

3.5. Influence of Climate

Regional climate data were obtained from the Swedish Meteorological and

Hydrological Institute and the Abisko Scientific Research Station. The nearest weather station for Laevas, Sorkaitum, and Girjas districts was Abisko (45-80 km), where mean daily temperature and precipitation data have been recorded since 1913. The

Västerbotten districts of Vilhelmina norra and Vilhelmina sodra were closest to the

Hemavan/Tärnaby weather station (65 km compared to 375 km to Abisko), for which only mean monthly temperature data (starting in 1966) were acquired. For analyses requiring precipitation or daily temperature measurements, Abisko data were used for all five districts.

To examine relationships between establishment and regional climate, the residuals for each herding districts were regressed against the following: Mean monthly temperature in the year of establishment, mean seasonal temperature in the year of establishment, mean monthly temperature in the year preceding establishment, mean monthly temperature in the following year, average mean monthly temperature over the five years following establishment, and precipitation total for the following winter

(October - April). Conditions in previous years allow evaluation of the importance of climate on seed production and root sprouting, which may only occur after a period of favorable climate. Looking at conditions in subsequent years addresses the concept of survivorship. Regressions were calculated for annual, pentadal, and decadal groupings.

Multiple linear regression models were also created for the previous year, current year, following year, and following five-year averages.

22

In addition to mean temperature and precipitation, regressions were calculated for the following: Annual degree days, growing season degree days (May to September), summer degree days (June to August), June degree days, July degree days, and August degree days. Degree days were compiled using a base temperature of 0 º C, the temperature at which mountain birch are able to start growing (Shutova et al., 2006).

Again, regressions were calculated for annual, pentadal, and decadal groupings.

3.6. Interacting Effects of Climate and Herbivory

One way to examine the interacting effects of climate and herbivory is to take a broader perspective and look at more generalized scenarios. Tree establishment residuals were divided along two axes (mean summer temperature and reindeer stocking levels), resulting in four possible scenarios (high temperature and high reindeer; low temperature and high reindeer; low temperature and low reindeer; high temperature and low reindeer.

The highest recruitment is expected to take place in years of low reindeer populations and high summer temperatures and lowest recruitment should take place when reindeer populations are high and summer temperatures are low.

For each year in which both monthly temperature data and annual reindeer stocking levels were available, a classification of "high" was assigned if it that variable was above the median for the period of record in that district and "low" if it was below. I then totaled the number of positive v. negative establishment residuals that fell under each of the four possible combinations. This was repeated using quartiles (high, low, medium-high, and medium-low scenarios).

23

As another test, multiple linear regression models were created using historical reindeer stocking data and the suite of climate variables listed above.

3.7. Influence of Vegetation

1 x 1 m quadrat vegetation data were analyzed using cluster analysis in PC-ORD

(McCune & Mefford, 1999) to identify plant community associations within the herbaceous plants and low-growing shrubs at treeline. A dendrogram was produced using Sørensen’s distance measure and Ward’s linkage method. The dendrogram was then pruned based on a combination of Indicator Species Analysis (ISA) (Dufrêne &

Legendre, 1997) and Multi-Response Permutation Procedure (MRPP). While originally designed as a method for assessing the value of species as indicators for environmental conditions (Niemi & McDonald, 2004), ISA can also provide an objective measure for selecting the most ecologically meaningful level at which to prune a dendrogram (cf.,

McCune & Grace, 2002; Aho et al., 2011; Lüth et al., 2011). In this method, an ISA is computed for each species at each level of grouping. Indicator values attest to how faithful (always found) a given species is to a group as well as how exclusive it is (i.e. never found in other groups). Low indicator values mean that the groups are too finely divided, while high internal heterogeneity is found when the grouping level is too coarse.

After averaging the p-values of all species at each grouping level, the grouping level with the lowest average p-value should be the most informative level. This can be verified by totaling the number of significant indicators within each grouping level—the level with the most indicators should also be meaningful. To further validate the meaningfulness of the grouping level, an MRPP was performed as a non-parametric method for testing the

24 null hypothesis of no difference between the groups (McCune & Grace, 2002). MRPP provides a T-statistic, measuring the separation between groups, a p-value indicating how likely it is that the difference was arrived at by chance, and an agreement statistic measuring the internal homogeneity of each group.

When this process was completed, Friedman tests (data were not normally distributed) were performed on the vegetation classifications to look for differences in reindeer use and/or birch establishment. For herding districts containing multiple vegetation classifications (minimum of 3 sites per classification), separate regression models of establishment as a function of the different climate variables were calculated.

Regressions of establishment against reindeer use were also calculated for each vegetation classification. Kolmogorov-Smirnov tests were performed to determine if there were significant differences between the age class structures of different herding districts within the same vegetation classification, or between different vegetation classifications in the same herding district.

25

4. RESULTS

4.1 Age Structures

Within the 69 field sites, 4906 birch cross-sections were collected and aged

(Table 2). Seedling densities within the 69 sites were highly variable, ranging from 0 to

> 430 seedlings (mean = 68, SD = 87). Pellet counts ranged from 2 to 117 (mean = 20,

SD = 18). Because of the possibility of missing the youngest birch seedlings even with thorough searches, the most recent period was removed from the age class structures before calculating residuals and performing any analyses. Besides their small size, newly established stems may have been missed if they were defoliated by herbivores before the arrival of the field crew and had not yet produced replacement leaves. I removed the most recent year of establishment (2009 for Laevas, Girjas, and Sorkaitum; 2010 for

Vilhelmina Norra and Sodra) from each annual establishment time series. For the 5- and

10-year binned time series, I removed the periods from 2005-2010 and 2000-2010 respectively. Indeed, only 1 seedling out of 4906 was determined to be 1-year of age.

The most recent period of establishment was also consistently identified as an outlier. No age structures were produced for Baste herding district because no seedlings were found at those sites. The oldest stem collected in the study began growing in 1888. Note that because trees greater than > 5 cm in diameter were not collected, the age class structures are not complete. They should still provide a good estimate of variation in recent establishment.

The age class structures are presented in Figures 15 through 18. Note that histogram columns falling below the fitted curve represent years with lower than expected establishment and columns rising above the curve represent years with high

26 establishment. Focusing on the pentadal data, Girjas herding district experienced reduced establishment in the most recent period (1995 to 2005), but greater than expected establishment from 1980 to 1995 and 1965 to 1975. The pentad of 1945 to 1950 showed positive establishment residuals surrounded by down years from 1950 to 1965 and 1930 to 1945. Vilhelmina sodra showed a similar pattern, aside from greater than expected establishment from 2000 to 2005 and below average years from 1960 to 1975. As in

Girjas, high establishment periods were seen from 1980 to 1995 and 1945 to 1950, while a period of low establishment was experienced form 1935 to 1945.

Focusing on the three herding districts with the largest sample sizes (Laevas,

Sorkaitum, and Vilhelmina norra), negative establishment residuals occurred from 2000 to 2005. From 1995 to 2000, establishment was roughly average in Sorkaitum and

Vilhelmina norra, but well below average in Laevas. Going back in time, Laevas experienced a prolonged period of high establishment from 1965 to 1995 and was only below average from 1950 to 1955. Sorkaitum had a shorter—but more pronounced— peak in establishment from 1980 to 1995 and was roughly average for the remainder of the recorded period. Vilhelmina norra also peaked from 1980 to 1995, but was preceded by an extended period of slightly depressed establishment from 1960 to 1980. Aside from slightly positive establishment residuals from 1950 to 1955, the remainder of the record for Vilhelmina norra was roughly average.

Overall, when examining the pentadal data, severely reduced establishment is apparent between 2000 and 2005 in every herding district other than Vilhelmina sodra and between 1995 and 2000 in every district besides Vilhelmina norra. The last quarter of the 20th century was generally a period of high birch establishment, with some

27 reduction in establishment between 1950 and 1970 and average establishment for the remainder of the record. Notably, all herding districts experienced extremely high establishment during the period of record high reindeer stocking levels around 1990.

4.2. Influence of Herbivory

Regression of recent establishment (last 5 years) as a function of reindeer pellet count showed a weak, but significant positive relationship across all herding districts (r2

= 0.142, p = 0.002) (Figure 19). Log10 transformation of the data to account for heteroscedasticity did not improve the results (r2 = 0.139, p = 0.003). The analysis was repeated after removing sites that experienced no establishment at all, but this reduced the significance and strength of the relationship (r2 = 0.076, p = 0.122). Sites without seedlings had a mean of 15 reindeer pellet groups per site, while those with seedlings had a mean of 23. However, standard deviations were high (13 for sites with establishment,

10 for sites without). Regressions of district-wide establishment residuals as a function of historical stocking levels only supported this relationship in Sorkaitum (r2 = 0.380, p

<= 0.001) (Figure 20). The remaining districts showed weak and insignificant positive relationships (Figure 21). When grouped into pentads, Sorkaitum showed a slightly stronger signal, but lost significance (r2 = 0.410, p = 0.087) and the remaining districts remained insignificant (Figure 22).

4.3. Influence of Climate

Generally weak, but significant correlations were found between establishment and a number of climate variables, but with little discernible pattern (Table 3). The

28 strongest relationships were found for decadal January mean temperature (r2= 0.51) in

Laevas and decadal mean May temperature (r2= 0.55) in Girjas. However, pentadal data showed no significant relationship with those months, but instead July (r2= 0.22), August

(r2= 0.24), and September (r2= 0.33) were significant in Laevas and July (r2= 0.21) in

Girjas. When all districts were combined, the only significant month in the pentadal data was September (r2= 0.21). Few variables showed consistent relationships across herding districts save January and March mean temperatures over the 5 years following establishment, and these were extremely weak outside of Sorkaitum (January r2= 0.21).

The following winter’s precipitation did prove to be weakly significant in Laevas (r2=

0.05), Sorkaitum (r2= 0.10), Vilhelmina norra (r2= 0.13), and when all districts were combined (r2= 0.12) (Table 4). July mean temperature over the 5 years following establishment was also weakly significant in Laevas (r2= 0.17), Girjas (r2= 0.05), and when all districts were combined (r2= 0.08). When multiple linear regression models were constructed, the only significant models proved to be current year’s July and

October temperature in Laevas (r2= 0.12) and the following 5 years’ January, March,

June, August, September, and October temperatures in Girjas (r2= 0.34). Seasonal temperature averages in the year of establishment proved insignificant once outliers were removed.

When regressed against different periods of degree days with pentadal data, establishment residuals showed significant relationships with annual degree days (r2 =

0.27) and growing season degree days (r2 = 0.29) in Laevas (Table 5). With all districts combined, again using pentadal data, only growing season degree days were significant

(r2 = 0.24) predictors of establishment.

29

4.4. Interacting Effects of Climate and Herbivory

When the data were not grouped into pentads, scenarios of high temperatures and low reindeer population had few years of high recruitment. Meanwhile, low temperature/high reindeer scenarios displayed some of the highest proportions of years with high recruitment (Figures 23 and 24). However, when aggregated by pentad, we see the highest recruitment in situations with low summer temperatures and low reindeer populations (Figure 25). The next highest proportion of high recruitment years comes in the scenario of low summer temperature and high reindeer population. When using quartiles instead of above/below-median scenarios, the positive impacts of high reindeer stocking levels are reinforced (Figure 26). The scenarios showing the highest proportion of positive establishment residuals are those with high reindeer levels and medium-low to high summer temperatures. Multiple linear regressions of establishment residuals as functions of reindeer stocking level and summer degree days showed no significant results.

4.5. Influence of Vegetation

Vegetation data collected in the 1 x 1 m quadrats were classified using cluster analysis (Sorensen distance measure and Ward’s linkage method) (Figure 27) and the sites were divided into 4 groups based on ISA (lowest p-value, 0.177, and 2nd highest number of significant indicators) (Tables 6 through 8; Figure 28). MRPP confirmed this as an informative level of grouping (p < 0.001, A= 0.212, T= -28.869). The resulting vegetation classifications were ecologically meaningful, closely aligning with the various heath and meadow classifications commonly found above the forest limit (Table 9;

30

Figures 29 and 30). Quantitative soil moisture measurements were not taken, but field observations allowed interpretation of the classifications along a moisture gradient.

Vegetation classification 2 (low herb meadow) was the wettest, followed by classification

3 (wet heath), classification 1 (fresh heath/grass heath), and classification 4 (dry heath).

Sites in classification 3 commonly included hummocks, allowing for considerable intra- site heterogeneity, but classification 2 was by far the most diverse.

Classification 1 was characterized by considerable Betula nana cover, which grew to a greater height than the B. nana found in classification 4. Other significant indicators of classification 1 included Vaccinium vitis-idaea and Festuca spp. Juniperus communis and Linnaea borealis were found frequently, but were not significant indicators. The highest numbers of Juncus and Luzula species were found at these sites. Classification 1 resembled the Fresh Heath vegetation type (although no Salix shrubs were found at these sites), with some Grass Heath characteristics.

Sites within classification 2 were moist, floristically diverse, and populated by mesic/hydrophytic plants such as Salix spp. and Trollius europaeus. In addition, Bartsia alpina, Bistorta vivipara, Hieracium alpinum, Pyrola spp., Solidago virgaurea,

Taraxacum spp., Thalictrum alpinum, Veronica alpina, and Viola biflora were also significant indicators. Other common plants included Geranium sylvaticum, Antennaria alpina, Hieracium spp., Parnassia palustris, Pedicularis spp., Ranunculus nivalis,

Saussurea alpina, and Alchemilla spp., though none of these were significant indicators.

Deschampsia cespitosa and Rubus chamaemorus were found in lesser numbers.

Classification 2 had the most mountain birch seedlings (both within the quadrats and site-wide), but this was not shown to be statistically significant (see below). These

31 sites match up well with the Low Herb Meadow vegetation type. They receive significant snow cover (Eriksson et al., 2007).

Classification 3 was identified by the greatest amount of Vaccinium myrtillis cover. Trientalis europaea proved to be another significant indicator. Hieracium alpinum, Koenigia islandica, Salix herbacea, Equisetum sylvaticum, and Salix spp. were present in lower numbers. Though not indicators, Deschampsia flexuosa and Rumex acetosa were found frequently. Species such as Rubus arcticus point to the hummocky nature of these sites. Classification 3 matches most closely with the Wet Heath vegetation type.

Sites grouped into classification 4 were dry and sparse, but contained significant cover from all three Vaccinium species (V. myrtillis, V. uliginosum, and V. vitis-idaea).

The greatest cover of Empetrum nigrum and Cornus suecica was found in these sites.

Arctostaphylos alpinum and Phyllodoce caerulea were also found. Classification 4 is similar to the Dry Heath vegetation type.

Friedman tests comparing reindeer pellet and recent birch establishment counts in each classification produced high p-values (lowest p = 0.083, remaining p > 0.317). This indicates that vegetation communities did not influence the amount of reindeer use or the amount of establishment. When the age class structures were subdivided by vegetation classification, classification 2 came closest to illustrating the idealized negative exponential curve and classification 1 showed the least recent establishment (Figure 31 through 33). However, two-sample Kolmogorov-Smirnov tests of the age distributions showed significant differences only with annual data (Tables 10 through 12). When grouped into pentads and decades, the noise in the data was reduced and there were no

32 significant differences between distributions. Lastly, north and south facing slopes did not exhibit significant differences in vegetation community, birch establishment, or reindeer use.

33

5. DISCUSSION

This study examined the roles of herbivory, climate, and understory vegetation on the variability of tree establishment within the treeline ecotone. The results show a weak relationship between herbivory pressure and establishment at treeline, but not in the manner that was expected. Because birch leaves are palatable to reindeer, it was thought that increasing reindeer numbers would result in reduced birch establishment. Instead, we see a positive relationship. Regressions for all herding districts and vegetation classifications demonstrated this positive relationship, though many of these were, admittedly, not statistically significant.

How does one reconcile this with past studies that have demonstrated suppressive effects of herbivory on treeline in Scandinavia (Speed et al., 2010, 2011; Van Bogaert et al., 2011)? I would suggest that mountain birch treelines, rather than following the type

II response to herbivory suggested by Cairns and Moen (2004), may exhibit a type III response. Such a response occurs when the negative impacts of arboreal foliage consumption are balanced by positive impacts, such as the consumption of upslope vegetation and the disturbance of upslope sites in preparation for tree establishment.

With increasing herbivory pressure, in this model, treeline migration potential follows a

Gaussian curve. Seedlings are released and high migration potential is found under moderate herbivory pressure. The impacts of browsing, trampling and—in other systems—seed predation ultimately limit treeline migration at higher herbivore densities.

While Speed (Speed et al., 2010, 2011) found that sheep may effectively limit the treeline, the “low” density treatment in that study (25 animals per km2) was significantly greater than the greatest density of animals observed in the present study (Baste herding

34 district, which had 20 per km2). Sheep also feed in a more intensive manner than reindeer. It should be noted that, while there were only 4 study sites in Baste, no birch seedlings were observed in that district. Taken together, it is reasonable to hypothesize that at the low herbivore densities seen in most of Sweden, herbivores may actually have a slightly positive impact on establishment at treeline. Only at higher densities does treeline migration potential begin to descend the right side of the Gaussian curve. This may especially be the case under warming, moistening conditions that might also promote denser tundra vegetation—the importance of herbivore disturbance to free space for seed implantation may be magnified under such a scenario. Further investigation is warranted to confirm that this is the case in mountain birch ecosystems.

Establishment responses to climate variables were generally poor. When significant correlations were found, they were inconsistent across herding districts and grouping levels (annual vs. pentadal vs. decadal aggregates). It is true that treelines frequently exhibit nonlinear responses to climate warming, as temperature and other climate variables interact to influence different life history stages (Holtmeier & Broll,

2005), or produce effects that lag behind the driving mechanism (Danby & Hik, 2007). It is not to be expected that all treelines be primarily limited by temperature. Still, diffuse treelines like the ones studied here have been identified as the most likely to benefit from increased temperatures (Harsch & Bader, 2011).

This is not the first study to report a lack of establishment response to climate in northern Sweden. In a less extensive study, Van Bogaert (2011) reported no correlation between establishment and summer mean temperature, winter mean temperature, summer precipitation, or winter precipitation. Significant connections between climate and

35 establishment have primarily involved coniferous species (Wang et al., 2006; Lv &

Zhang, 2011). However, despite finding strong correlations between establishment of older trees with climate, Mamet and Kershaw (2011) noted that this connection did not hold for recruitment in recent decades. Even with warming temperatures at their

Canadian treeline sites, many had no seedlings. Unfortunately for the present study, trees

> 5 cm in diameter were not sampled and thus the age structures presented do not stretch back far enough to test for a similar shift in establishment response. In the case of

Mamet and Kershaw (2011), they hypothesized that the relationship between climate and establishment had diverged due to a reduction in tree fecundity driven by moisture stress.

Warming has coincided with a higher frequency of dry summers in that region (New et al., 2000). A worsening correlation between temperature and radial growth—commonly attributed to moisture stress—has also been observed in northern hemisphere tree ring chronologies (Jacoby & D'Arrigo, 1995). Future studies should investigate whether these weakening relationships are driven by the same factors, or are isolated events.

In contrast to the drier conditions being experienced in much of North America,

Scandinavia has been getting wetter (Hurrell, 1995). Most global climate models (GCM) expect this trend to continue (http://climexp.knmi.nl). Further, poor correlations have previously been reported between precipitation and birch establishment (Van Bogaert et al., 2011) and growth (Young et al., 2011) in Sweden. Moisture stress is not likely to have spurred the disconnect between establishment and temperature found in this study.

One confounding factor in the research presented here is the possibility that trees spawned from clonal and seed reproduction may react differently to changes in climate.

It has been suggested that root-sprouted clones may be able to persist indefinitely at

36 treeline when climate precludes the establishment of seedlings (Kullman, 1989). Root- sprouted clones may be decoupled more from temperature, or coupled to different climate variables such as precipitation. It may also be the case that root-sprouts are actually more responsive to climate amelioration than seedlings. Since they are not reliant on seed availability and/or implantation, they may be able to take advantage of improved climatic conditions more readily. Clones are able to draw on a larger root base than individuals starting from seed, presumably making them more resistant to drought stress and more resilient when browsed by herbivores. They may also benefit from microclimates created by the mature tree from which they sprouted (Alftine & Malanson, 2004). On the other hand, vegetative reproduction requires greater investment from the parent plant (Bell &

Bliss, 1980). This study did not differentiate the reproductive methods responsible for individual trees, but many of the samples that were collected within the field sites came from clumps and are likely to have been clonal. While it is time-consuming and difficult to identify whether a given tree grew from seed or a root, future research should seek to differentiate between the two to help identify differential responses to climate, as well as herbivory. Treelines that advance via sexual reproduction are likely to have significantly different rates of spread than those that rely on vegetative reproduction for new stems.

A portion of the weak climate/establishment correlation reported here and elsewhere may be attributable to the coarse resolution of available climate data. Site conditions can vary widely within mountain ranges, making correlation with a single meteorological station difficult, even when it is relatively close. Topography, proximity to large impounded lakes, and other site-specific variables can significantly impact local microclimates. Such microclimates may be so decoupled from broader atmospheric

37 conditions that they exhibit completely opposite trends (e.g., cooling when the region at large is warming) (Dobrowski, 2011). Microclimates have been demonstrated to influence treeline position (Sundqvist et al., 2008), but data at this scale are scarce and lack temporal depth. Further development of topoclimatic models for northern Sweden

(Yang et al., 2011) may assist researchers in assessing the true conditions experienced at specific treeline locations. Furthermore, knowledge of local geometrid moth outbreaks, which defoliate large swaths of birch forest, may also help account for the apparent disconnect between establishment and climate (Karlsson et al., 2005).

This study did not find variation in the invasibility of different field layer vegetation assemblages. While tundra plants can affect establishment at a number of stages (implantation of the seed in the soil, germination of the seed, competition for light when the seedling is young) the species composition of the tundra was not supported as a significant factor. In the future, more intensive vegetation surveys should be completed

(including identification of lichens), as the methodology used here may not have sufficiently captured intra-site heterogeneity. Even if a connection had been made, more investigation would be prudent, as vegetation differences may commonly be the result of underlying edaphic variability (Weih, 1998) that was beyond the scope of this study.

38

6. CONCLUSIONS

To conclude, weakly positive responses to herbivory were observed in patterns of tree establishment at treeline, while the climate variables found to be significant were inconsistent across herding districts. No connection between vegetation assemblages and establishment was observed. Dendrochronology is a remarkable tool for peering into the past when direct observational data is impractical or unavailable. The strength of this research lies in the large dataset of seedling establishment dates that were compiled using dendrochronological techniques. Future research should incorporate this comprehensive treatment of seedling establishment with already broadly used techniques for dating mature trees. As more studies are completed, the complexity underlying treeline dynamics becomes increasingly apparent, with numerous interacting variables and time lags. Herbivory is one element of these systems that still requires further investigation.

39

REFERENCES

Aho, K., Weaver, T. & Regele, S. (2011) Identification and siting of native vegetation types on disturbed land: demonstration of statistical methods. Applied Vegetation Science, 14, 277-290.

Alftine, K.J. & Malanson, G.P. (2004) Directional positive feedback and pattern at an alpine tree line. Journal of Vegetation Science, 15, 3-12.

Andersson, L., Grundsten, C., Rafstedt, T. & von Südow, U. (1984) Vegetationskarta över de svenska fjällen. Naturvårdsverket, Stockholm, Sweden.

Anschlag, K., Broll, G. & Holtmeier, F.-K. (2008) Mountain birch seedlings in the treeline ecotone, subarctic Finland: variation in above- and below-ground growth depending on microtopography. Arctic, Antarctic, and Alpine Research, 40, 609- 616.

Arnborg, T. (1990) Forest types of northern Sweden. Plant Ecology, 90, 1-13.

Barbeito, I., Dawes, M., Rixen, C., Senn, J. & Bebi, P. (2012) Factors driving mortality and growth at treeline: a 30-year experiment of 92000 conifers. Ecology, 93, 389- 401.

Batllori, E., Camarero, J.J., Ninot, J.M. & Gutiérrez, E. (2009) Seedling recruitment, survival and facilitation in alpine Pinus uncinata tree line ecotones. Implications and potential responses to climate warming. Global Ecology and Biogeography, 18, 460-472.

Beach, H. (1993) A year in Lapland: guest of the reindeer herders. University of Washington Press, Seattle, WA.

Bekker, M.F. (2005) Positive feedback between tree establishment and patterns of subalpine forest advancement, Glacier National Park, Montana, USA. Arctic, Antarctic, and Alpine Research, 37, 97-107.

Bekker, M.F. & Malanson, G.P. (2008) Linear forest patterns in subalpine environments. Progress in Physical Geography, 32, 635.

40

Bell, K.L. & Bliss, L. (1980) Plant reproduction in a high arctic environment. Arctic and Alpine Research, 12, 1-10.

Bennett, L.J., English, P. & McCain, R. (1940) A study of deer populations by use of pellet-group counts. The Journal of Wildlife Management, 4, 398-403.

Bernes, C. (1996) The Nordic Arctic environment: unspoilt, exploited, polluted? Nordic Council of Ministers, Copenhagen, Denmark.

Boertje, R.D. (1984) Seasonal diets of the Denali caribou herd, Alaska. Arctic, 37, 161- 165.

Butler, D.R. (2001) Geomorphic process-disturbance corridors: a variation on a principle of landscape ecology. Progress in Physical Geography, 25, 237-238.

Butler, D.R. & Walsh, S.J. (1990) Lithologic, structural, and topographic influences on snow‐avalanche path location, eastern Glacier National Park, Montana. Annals of the Association of American Geographers, 80, 362-378.

Butler, D.R. & Walsh, S.J. (1994) Site characteristics of debris flows and their relationship to alpine treeline. Physical Geography, 15, 181-199.

Butler, D.R., Malanson, G.P. & Cairns, D.M. (1994) Stability of alpine treeline in Glacier National Park, Montana, U. S. A. Phytocoenologia, 22, 485-500.

Butler, D.R., Malanson, G.P. & Resler, L.M. (2004) Turf-banked terrace treads and risers, turf exfoliation and possible relationships with advancing treeline. Catena, 58, 259-274.

Butler, D.R., Malanson, G.P., Resler, L.M., Walsh, S.J., Wilkerson, F.D., Schmid, G.L. & Sawyer, C.F. (2009) Geomorphic patterns and processes at alpine treeline. Developments in Earth Surface Processes (ed. by G.P. Malanson, D.R. Butler, S.J. Walsh and D.B. Fagre), pp. 63-84. Elsevier, Atlanta, GA.

Cairns, D.M. (2001) Patterns of winter desiccation in krummholz forms of Abies lasiocarpa at treeline sites in Glacier National Park, Montana, USA. Geografiska Annaler: Series A, Physical Geography, 83, 157-168.

41

Cairns, D.M. & Moen, J. (2004) Herbivory influences tree lines. Journal of Ecology, 92, 1019-1024.

Camarero, J.J. & Gutiérrez, E. (2007) Response of Pinus uncinata recruitment to climate warming and changes in grazing pressure in an isolated population of the Iberian system (NE Spain). Arctic, Antarctic, and Alpine Research, 39, 210-217.

Cui, M. & Smith, W.K. (1991) Photosynthesis, water relations and mortality in Abies lasiocarpa seedlings during natural establishment. Tree physiology, 8, 37-46.

Cullen, L.E., Stewart, G.H., Duncan, R.P. & Palmer, J.G. (2001) Disturbance and climate warming influences on New Zealand Nothofagus tree-line population dynamics. Journal of Ecology, 89, 1061-1071.

Dalen, L. & Hofgaard, A. (2005) Differential regional treeline dynamics in the Scandes Mountains. Arctic, Antarctic, and Alpine Research, 37, 284-296.

Danby, R.K. & Hik, D.S. (2007) Responses of white spruce (Picea glauca) to experimental warming at a subarctic alpine treeline. Global Change Biology, 13, 437-451.

Daniels, L.D. & Veblen, T.T. (2004) Spatiotemporal influences of climate on altitudinal treeline in northern Patagonia. Ecology, 85, 1284-1296.

Dobrowski, S.Z. (2011) A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology, 17, 1022-1035.

Dufrêne, M. & Legendre, P. (1997) Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecological Monographs, 67, 345-366.

Dullinger, S., Dirnböck, T. & Grabherr, G. (2004) Modelling climate change driven treeline shifts: relative effects of temperature increase, dispersal and invasibility. Journal of Ecology, 92, 241-252.

Eberhardt, L. & Van Etten, R.C. (1956) Evaluation of the pellet group count as a deer census method. The Journal of Wildlife Management, 20, 70-74.

42

Elliott, G.P. & Kipfmueller, K.F. (2011) Multiscale influences of climate on upper treeline dynamics in the southern Rocky Mountains, USA: evidence of intraregional variability and bioclimatic thresholds in response to twentieth- century warming. Annals of the Association of American Geographers, 101, 1181-1203.

Emanuelsson, U. (1987) Human influence on vegetation in the Torneträsk area during the last three centuries. Ecological Bulletins, 38, 95-111.

Eränen, J.K. & Kozlov, M.V. (2008) Increasing intraspecific facilitation in exposed environments: consistent results from mountain birch populations in two subarctic stress gradients. Oikos, 117, 1569-1577.

Eriksson, O., Niva, M. & Caruso, A. (2007) Use and abuse of reindeer range. Acta Phytogeographica Suecica, 87, 1-6.

French, D., Miller, G. & Cummins, R. (1997) Recent development of high-altitude Pinus sylvestris scrub in the northern Cairngorm Mountains, Scotland. Biological Conservation, 79, 133-144.

Fritts, H.C. (1976) Tree rings and climate. Academic Press, London, England.

Green, J.A. (2005) An application of predictive vegetation mapping to mountain vegetation in Sweden. Masters Thesis, Texas A&M University, College Station.

Hadley, J.L. & Smith, W.K. (1986) Wind effects on needles of timberline conifers: seasonal influence on mortality. Ecology, 67, 12-19.

Hagemoen, R.I.M. & Reimers, E. (2002) Reindeer summer activity pattern in relation to weather and insect harassment. Journal of Animal Ecology, 71, 883-892.

Harsch, M.A. & Bader, M.Y. (2011) Treeline form–a potential key to understanding treeline dynamics. Global Ecology and Biogeography, 20, 582-596.

Harsch, M.A., Hulme, P.E., McGlone, M.S. & Duncan, R.P. (2009) Are treelines advancing? A global meta‐analysis of treeline response to climate warming. Ecology Letters, 12, 1040-1049.

43

Helle, T. (2001) Mountain birch forests and reindeer husbandry. Nordic Mountain Birch Ecosystems (ed. by F.E. Wielgolaski), pp. 279-291. Parthenon Publishing, New York, NY.

Herder, M. & Niemelä, P. (2010) Effects of reindeer on the re-establishment of Betula pubescens subsp. czerepanovii and Salix phylicifolia in a subarctic meadow. Rangifer, 23, 3-12.

Herrero, A., Zamora, R., Castro, J. & Hódar, J. (2011) Limits of pine forest distribution at the treeline: herbivory matters. Plant Ecology, 213, 1-11.

Hessl, A.E. & Baker, W.L. (1997) Spruce and fir regeneration and climate in the forest- tundra ecotone of Rocky Mountain National Park, Colorado, USA. Arctic and Alpine Research, 29, 173-183.

Hofgaard, A. (1997) Inter-relationships between treeline position, species diversity, land use and climate change in the central Scandes Mountains of Norway. Global Ecology and Biogeography Letters, 6, 419-429.

Holtmeier, F.K. (2009) Mountain timberlines: ecology, patchiness, and dynamics. Springer Verlag, Berlin, Germany.

Holtmeier, F.K. & Broll, G. (2005) Sensitivity and response of northern hemisphere altitudinal and polar treelines to environmental change at landscape and local scales. Global Ecology and Biogeography, 14, 395-410.

Holtmeier, F.K., Broll, G., Müterthies, A. & Anschlag, K. (2010) Regeneration of trees in the treeline ecotone: northern Finnish Lapland. Fennia-International Journal of Geography, 181, 103-128.

Hurrell, J.W. (1995) Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science, 269, 676-679.

Ingman, M. & Gyllensten, U. (2006) A recent genetic link between Sami and the Volga- Ural region of Russia. European Journal of Human Genetics, 15, 115-120.

Jacoby, G.C. & D'Arrigo, R.D. (1995) Tree ring width and density evidence of climatic and potential forest change in Alaska. Global Biogeochemical Cycles, 9, 227-234.

44

Jiménez-Moreno, G., Fawcett, P.J. & Scott Anderson, R. (2008) Millennial-and centennial-scale vegetation and climate changes during the late Pleistocene and Holocene from northern New Mexico (USA). Quaternary Science Reviews, 27, 1442-1452.

Johnstone, J.F. & Chapin, F.S. (2006) Effects of soil burn severity on post-fire tree recruitment in boreal forest. Ecosystems, 9, 14-31.

Jonasson, S. (1981) Plant communities and species distribution of low alpine Betula nana heaths in northernmost Sweden. Plant Ecology, 44, 51-64.

Kallio, P. & Lehtonen, J. (1973) Birch forest damage caused by Oporinia autumnata (Bkh.) in 1965-66 in Utsjoki, N Finland. Reports from the Kevo Subarctic Research Station, 10, 55-69.

Karlsson, P.S., Weih, M. & Borg, C. (2005) Mountain birch growth in relation to climate and herbivores. Plant Ecology, Herbivory, and Human Impact in Nordic Mountain Birch Forests (ed. by M.M. Caldwell, G. Heldmaier, R.B. Jackson, O.L. Lange, H.A. Mooney, E.D. Schulze, U. Sommer, F.E. Wielgolaski, P.S. Karlsson, S. Neuvonen and D. Thannheiser), pp. 71-86. Springer, Berlin, Germany.

Klein, D.R. (1968) The introduction, increase, and crash of reindeer on St. Matthew Island. The Journal of Wildlife Management, 32, 350-367.

Körner, C. (1998) A re-assessment of high elevation treeline positions and their explanation. Oecologia, 115, 445-459.

Körner, C. (2003) Alpine plant life: functional plant ecology of high mountain ecosystems. Springer, Berlin, Germany.

Körner, C. & Paulsen, J. (2004) A world-wide study of high altitude treeline temperatures. Journal of Biogeography, 31, 713-732.

Kozlowski, T. (1999) Soil compaction and growth of woody plants. Scandinavian Journal of Forest Research, 14, 596-619.

45

Krahulec, F., Rosén, E. & Maarel, E. (1986) Preliminary classification and ecology of dry grassland communities on Ölands Stora Alvar (Sweden). Nordic Journal of Botany, 6, 797-809.

Kullman, L. (1989) Recent retrogression of the forest-alpine tundra ecotone (Betula pubescens Ehrh. ssp. tortuosa (Ledeb.) Nyman) in the Scandes Mountains, Sweden. Journal of Biogeography, 16, 83-90.

Kullman, L. (2002) Rapid recent range‐margin rise of tree and shrub species in the Swedish Scandes. Journal of Ecology, 90, 68-77.

Kullman, L. & Öberg, L. (2009) Post-Little Ice Age tree line rise and climate warming in the Swedish Scandes: a landscape ecological perspective. Journal of Ecology, 97, 415-429.

Kumpula, J. (2001) Productivity of the semi-domesticated reindeer (Rangifer tarandus L.) stock and carrying capacity of pastures in Finland during 1960-1990's. University of Oulu Press, Oulu, Finland.

Kupfer, J.A. & Cairns, D.M. (1996) The suitability of montane ecotones as indicators of global climatic change. Progress in Physical Geography, 20, 253.

Landhäusser, S.M., Deshaies, D. & Lieffers, V.J. (2010) Disturbance facilitates rapid range expansion of aspen into higher elevations of the Rocky Mountains under a warming climate. Journal of Biogeography, 37, 68-76.

Lehtonen, J. & Heikkinen, R.K. (1995) On the recovery of mountain birch after Epirrita damage in Finnish Lapland, with a particular emphasis on reindeer grazing. Ecoscience, 2, 349-356.

Leitner, R. & Gajewski, K. (2004) Modern and Holocene stomate records of tree-line variations in northwestern Quebec. Canadian Journal of Botany, 82, 726-734.

Lempa, K., Neuvonen, S. & Tømmervik, H. (2005) Effects of reindeer grazing on pastures-a necessary basis for sustainable reindeer herding. Plant Ecology, Herbivory, and Human Impact in Nordic Mountain Birch Forests, pp. 157-164. Springer, Berlin, Germany.

46

Lloyd, A.H. & Fastie, C.L. (2002) Spatial and temporal variability in the growth and climate response of treeline trees in alaska. Climatic Change, 52, 481-509.

Lloyd, A.H., Yoshikawa, K., Fastie, C.L., Hinzman, L. & Fraver, M. (2003) Effects of permafrost degradation on woody vegetation at arctic treeline on the Seward Peninsula, Alaska. Permafrost and Periglacial Processes, 14, 93-101.

Lüth, C., Tasser, E., Niedrist, G., Dalla Via, J. & Tappeiner, U. (2011) Classification of the Sieversio montanae-Nardetum strictae in a cross-section of the eastern Alps. Plant Ecology, 212, 105-126.

Lv, L.X. & Zhang, Q.B. (2011) Asynchronous recruitment history of Abies spectabilis along an altitudinal gradient in the Mt. Everest region. Journal of Plant Ecology, 5, 147-156.

Maher, E.L. & Germino, M.J. (2006) Microsite differentiation among conifer species during seedling establishment at alpine treeline. Ecoscience, 13, 334-341.

Mamet, S.D. & Kershaw, G.P. (2011) Subarctic and alpine tree line dynamics during the last 400 years in north‐western and central Canada. Journal of Biogeography, 39, 855-868.

Mårell, A. & Edenius, L. (2006) Spatial heterogeneity and hierarchical feeding habitat selection by reindeer. Arctic, Antarctic, and Alpine Research, 38, 413-420.

McCune, B. & Mefford, M. (1999) PC-ORD. Multivariate analysis of ecological data. Version 5.0. MjM Software Design. Gleneden Beach, OR.

McCune, B. & Grace, J.B. (2002) Analysis of ecological communities. MjM Software Design, Gleneden Beach, OR.

Moen, J. & Oksanen, L. (1998) Long-term exclusion of folivorous mammals in two arctic-alpine plant communities: a test of the hypothesis of exploitation ecosystems. Oikos, 82, 333-346.

Moen, J. & Danell, Ö. (2003) Reindeer in the Swedish mountains: an assessment of grazing impacts. AMBIO: A Journal of the Human Environment, 32, 397-402.

47

Moen, J., Cairns, D.M. & Lafon, C.W. (2008) Factors structuring the treeline ecotone in Fennoscandia. Plant Ecology & Diversity, 1, 77-87.

Müller-Wille, L., Müller, F., Konstantinov, Y., Hukkinen, J., Gunslay, N., Forbes, B.C. & Bölter, M. (2006) Reindeer management in northernmost Europe. linking practical and scientific knowledge in social ecological systems. Springer, Berlin, Germany.

New, M., Hulme, M. & Jones, P. (2000) Representing twentieth-century space-time climate variability. Part II: Development of 1901-96 monthly grids of terrestrial surface climate. Journal of Climate, 13, 2217-2238.

Niemi, G.J. & McDonald, M.E. (2004) Application of ecological indicators. Annual Review of Ecology, Evolution, and Systematics, 35, 89-111.

Ohse, B., Jansen, F. & Wilmking, M. (2012) Do limiting factors at Alaskan treelines shift with climatic regimes? Environmental Research Letters, 7, 015505.

Oksanen, L. & Virtanen, R. (1995) Topographic, altitudinal and regional patterns in continental and suboceanic heath vegetation of northern Fennoscandia. Acta Botanica Fennica, 1-80.

Oksanen, L., Moen, J. & Helle, T. (1995) Timberline patterns in northemmost Fennoscandia. Acta Botanica Fennica, 153, 93-105.

Olofsson, J., Kitti, H., Rautiainen, P., Stark, S. & Oksanen, L. (2001) Effects of summer grazing by reindeer on composition of vegetation, productivity and nitrogen cycling. Ecography, 24, 13-24.

Patterson, A. (1959) Distinguishing annual rings in diffuse porous tree species. Journal of Forestry, 57, 126.

Payette, S. (2007) Contrasted dynamics of northern Labrador tree lines caused by climate change and migrational lag. Ecology, 88, 770-780.

Payette, S. & Filion, L. (1985) White spruce expansion at the tree line and recent climatic change. Canadian Journal of Forest Research, 15, 241-251.

48

Pérez, F. (1992) Processes of turf exfoliation (Rasenabschälung) in the high Venezuelan Andes. Zeitschrift fur Geomorphologie, 36, 81-106.

Rafstedt, T. (1987) Vegetationskarta över norra Härjedalens skogsland och lågfjäll Sånsfjället, Vemdalsfjällen, Klövsjöfjällen. Naturvårdsverket, Stockholm, Sweden.

Resler, L.M., Butler, D.R. & Malanson, G.P. (2005) Topographic shelter and conifer establishment and mortality in an alpine environment, Glacier National Park, Montana. Physical Geography, 26, 112-125.

Shutova, E., Wielgolaski, F., Karlsen, S., Makarova, O., Berlina, N., Filimonova, T., Haraldsson, E., Aspholm, P., Flø, L. & Høgda, K. (2006) Growing seasons of Nordic mountain birch in northernmost Europe as indicated by long-term field studies and analyses of satellite images. International Journal of Biometeorology, 51, 155-166.

Skarin, A., Danell, Ö., Bergström, R. & Moen, J. (2010) Reindeer movement patterns in alpine summer ranges. Polar Biology, 33, 1263-1275.

Slatyer, R. & Noble, I. (1992) Dynamics of montane treelines. Ecological Studies, 92, 346-359.

Smith, W.K., Germino, M.J., Hancock, T.E. & Johnson, D.M. (2003) Another perspective on altitudinal limits of alpine timberlines. Tree Physiology, 23, 1101- 1112.

Sonesson, M. & Kvillner, E. (1980) Plant communities of the Stordalen Mire: a comparison between numerical and non-numerical classification methods. Ecological Bulletins, 30, 113-125.

Sonesson, M. & Hoogesteger, J. (1983) Recent tree-line dynamics (Betula pubescens Ehrh. ssp. tortuosa (Ledeb.) Nyman) in northern Sweden. Nordicana, 47, 47-54.

Speed, J.D.M., Austrheim, G., Hester, A.J. & Mysterud, A. (2010) Experimental evidence for herbivore limitation of the treeline. Ecology, 91, 3414-3420.

49

Speed, J.D.M., Austrheim, G., Hester, A.J. & Mysterud, A. (2011) Growth limitation of mountain birch caused by sheep browsing at the altitudinal treeline. Forest Ecology and Management, 261, 1344-1352.

Staland, H., Salmonsson, J. & Hörnberg, G. (2011) A thousand years of human impact in the northern Scandinavian mountain range: long-lasting effects on forest lines and vegetation. The Holocene, 21, 379-391.

Stevens, G.C. & Fox, J.F. (1991) The causes of treeline. Annual Review of Ecology and Systematics, 22, 177-191.

Sundqvist, M.K., Björk, R.G. & Molau, U. (2008) Establishment of boreal forest species in alpine dwarf-shrub heath in subarctic Sweden. Plant Ecology & Diversity, 1, 67-75.

Sveinbjörnsson, B., Nordell, O. & Kauhanen, H. (1992) Nutrient relations of mountain birch growth at and below the elevational tree-line in Swedish Lapland. Functional Ecology, 6, 213-220.

Sveinbjörnsson, B., Kauhanen, H. & Nordell, O. (1996) Treeline ecology of mountain birch in the Torneträsk area. Ecological Bulletins, 45, 65-70.

Svensson, B.H. & Rosswall, T. (1984) In situ methane production from acid peat in plant communities with different moisture regimes in a subarctic mire. Oikos, 341-350.

Szeicz, J.M. & Macdonald, G.M. (1995) Recent white spruce dynamics at the subarctic alpine treeline of north-western Canada. Journal of Ecology, 83, 873-885.

Taylor, A.H. (1995) Forest expansion and climate change in the mountain hemlock (Tsuga mertensiana) zone, Lassen Volcanic National Park, California, USA. Arctic and Alpine Research, 27, 207-216.

Tommervik, H., Johansen, B., Tombre, I., Thannheiser, D., Hogda, K., Gaare, E. & Wielgolaski, F. (2004) Vegetation changes in the Nordic mountain birch forest: the influence of grazing and climate change. Arctic, Antarctic, and Alpine Research, 36, 323-332.

Tommervik, H., Johansen, B., Riseth, J., Karlsen, S., Solberg, B. & Hogda, K. (2009) Above ground biomass changes in the mountain birch forests and mountain

50

heaths of Finnmarksvidda, northern Norway, in the period 1957-2006. Forest Ecology and Management, 257, 244-257.

Tranquillini, W. & Benecke, U. (1979) Physiological ecology of the alpine timberline: tree existence at high altitudes with special reference to the European Alps. Springer-Verlag, Berlin, Germany.

Vallée, S. & Payette, S. (2004) Contrasted growth of black spruce (Picea mariana) forest trees at treeline associated with climate change over the last 400 years. Arctic, Antarctic, and Alpine Research, 36, 400-406.

Van Bogaert, R., Jonasson, C., De Dapper, M. & Callaghan, T.V. (2009) Competitive interaction between aspen and birch moderated by invertebrate and vertebrate herbivores and climate warming. Plant Ecology & Diversity, 2, 221-232.

Van Bogaert, R., Jonasson, C., De Dapper, M. & Callaghan, T.V. (2010) Range expansion of thermophilic aspen (Populus tremula L.) in the Swedish Subarctic. Arctic, Antarctic, and Alpine Research, 42, 362-375.

Van Bogaert, R., Haneca, K., Hoogesteger, J., Jonasson, C., De Dapper, M. & Callaghan, T.V. (2011) A century of tree line changes in sub-Arctic Sweden shows local and regional variability and only a minor influence of 20th century climate warming. Journal of Biogeography, 38, 907-921.

Van der Wal, R. & Brooker, R. (2004) Mosses mediate grazer impacts on grass abundance in arctic ecosystems. Functional Ecology, 18, 77-86.

Wal, R., Brooker, R., Cooper, E. & Langvatn, R. (2001) Differential effects of reindeer on high Arctic lichens. Journal of Vegetation Science, 12, 705-710.

Walsh, S.J. & Butler, D.R. (1997) Morphometric and multispectral image analysis of debris flows for natural hazard assessment. Geocarto International, 12, 59-70.

Walsh, S.J., Weiss, D.J., Butler, D.R. & Malanson, G.P. (2004) An assessment of snow avalanche paths and forest dynamics using Ikonos satellite data. Geocarto International, 19, 85-93.

51

Wang, T., Zhang, Q.B. & Ma, K. (2006) Treeline dynamics in relation to climatic variability in the central Tianshan Mountains, northwestern China. Global Ecology and Biogeography, 15, 406-415.

Wardle, P. (1981) Winter desiccation of conifer needles simulated by artificial freezing. Arctic and Alpine Research, 13, 419-423.

Weih, M. (1998) Seasonality of nutrient availability in soils of subarctic mountain birch woodlands, Swedish Lapland. Arctic and Alpine Research, 30, 19-25.

Weih, M. & Karlsson, P.S. (1999) The nitrogen economy of mountain birch seedlings: implications for winter survival. Journal of Ecology, 87, 211-219.

Wilson, J.B. & Agnew, A.D.Q. (1992) Positive-feedback switches in plant communities. Academic Press, London, England.

Yang, Z., Hanna, E. & Callaghan, T.V. (2011) Modelling surface air temperature variation over complex terrain around Abisko, Swedish Lapland: uncertainties of measurements and models at different scales. Geografiska Annaler: Series A, Physical Geography, 93, 89-112.

Young, A.B., Cairns, D.M., Lafon, C.W., Moen, J. & Martin, L.E. (2011) Dendroclimatic relationships and possible implications for mountain birch and Scots pine at treeline in northern Sweden through the 21st century. Canadian Journal of Forest Research, 41, 450-459.

APPENDIX A FIGURES

Figure 1. A mountain birch (Betula pubescens ssp. czerepanovii) seedling growing within a treeline study site, near Ritsem, Sweden. 52

Figure 2. Geometrid moth larvae in the process of defoliating a mountain birch on the outskirts of Sarek National Park, Sweden. 53

Figure 3. Conceptual model for differential impacts of herbivory pressure on different treeline types, presented in Cairns and Moen (2004). Type I responses occur where understory vegetation is more palatable than tree foliage and increased herbivory results in a competitive release for seedlings and root sprouts. Type II responses occur where tree foliage is preferred over understory vegetation, putting seedlings at a competitive disadvantage. Type III responses occur where negative impacts of arboreal foliage consumption are balanced by positive impacts, such as seedbed preparation through disturbance, at moderate densities. Mountain birch treelines have been thought to follow Type II responses.

54

55

Figure 4. Map of herding districts, weather stations, and study site locations.

Figure 5. Vegetation map for Norrbotten County study sites. Derived from data provided by the Swedish Surveying Agency (Latmäteriet). 56

57

Figure 6. Vegetation map for Västerbotten County study sites. Derived from data provided by the Swedish Surveying Agency (Latmäteriet).

Figure 7. An example of a diffuse treeline, near Nikkaluokta, Sweden. Trees gradually decrease in density and height along the ecotone.

58

Figure 8. Semi-domesticated reindeer (Rangifer tarandus), the primary herbivore at Swedish treelines.

59

Figure 9. A seasonal Sámi herding hut in Sarek National Park, Sweden. 60

61

Figure 10. A new reindeer calf is given the identifying earmark of its owner.

Swedish Reindeer Population by Year

Reindeer (in Thousands) 150 200 250 300 350

1900 1920 1940 1960 1980 2000 Year

Figure 11. Historical reindeer stocking levels. 62

Reindeer Population by Year: Study Herding Districts in Norrbotten County

Baste Laevas Girjas Sorkaitum 15000 10000 Reindeer 5000

1940 1960 1980 2000 Year

Figure 12. Reindeer stocking levels by year at study herding districts in Norrbotten County.

63

Reindeer Population by Year: Study Herding Districts in Vasterbotten County

Vilhelmina Norra Vilhelmina Sodra 15000 10000 Reindeer 5000

1940 1960 1980 2000 Year

Figure 13. Reindeer stocking levels by year at study herding districts in Västerbotten County. Note a large gap in accounting between

1965 and 1994. 64

Figure 14. 1 of 69 study sites established at treeline in Sweden. The top edge was placed near the upper limit of treeline, with the site extending downslope in the direction of the forestline. 65

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890

Girjas Laevas S orkaitum

N = 253 N = 1386 N = 1431 6

4

2

0 V ilhelmina norra V ilhelmina sodra

N = 1502 N = 183 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 6 Percent of Total

4

2

0

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Establishment Year

Figure 15. District age class structures (annual data).

66

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890

Girjas Laevas S orkaitum

N = 248 N = 1276 N = 1396 25

20

15

10

5

0 V ilhelmina norra V ilhelmina sodra

N = 1399 N = 170 25 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Percent of Total

20

15

10

5

0

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Establishment Year

Figure 16. District age class structures (pentadal data).

67

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880

Girjas Laevas S orkaitum 50 N = 228 N = 1138 N = 1292

40

30

20

10

0 V ilhelmina norra V ilhelmina sodra 50

N = 1146 N = 145 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880

Percent of Total 40

30

20

10

0 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880 Establishment Year

Figure 17. District age class structures (decadal data).

68

All Districts All Districts

N= 4659 N= 4393 4 20

3 15

2 10 Percent of Total Percent of Total

1 5

0 0

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Establishment Year Establishment Year

Figure 18. Age class structures for all herding districts combined (annual, pentadal, and decadal data).

69

All Districts

N= 3853 40

30

20 Percent of Total

10

0 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880 Establishment Year

Figure 18 continued. 70

Establishment (Last 5 Years) ~ Reindeer Activity Outliers Removed

12 Veg Class 1 Veg Class 2 Veg Class 3 10 Veg Class 4

8

6

Establishment Count 4

2

0

10 20 30 40 50 Pellets

Figure 19. The impact of reindeer numbers on recent establishment. Each symbol represents one plot.

71

Deviation from Predicted Establishment ~ District Stocking Level -- SORKAITUM

4

2

0 Establishment Residuals -2

-4

2000 4000 6000 8000 10000 SorkaitumReindeer

Figure 20. The relationship between stocking level and establishment was only significant in Sorkaitum (r2 = 0.380, p<= 0.001).

Each point represents one year of establishment/reindeer data. 72

Deviation from Predicted Establishment ~ Deviation from Predicted Establishment ~ District Stocking Level -- LAEVAS District Stocking Level -- GIRJAS

3 2

2 1

0 1

-1 0

Establishment Residuals -2 Establishment Residuals -1

-3

-2

2000 4000 6000 8000 10000 4000 6000 8000 10000 12000 14000 16000 Laevas Girjas Reindeer Reindeer

Figure 21. Relationships between district stocking level and establishment were not significant in 4 of 5 districts.

73

Deviation from Predicted Establishment ~ Deviation from Predicted Establishment ~ District Stocking Level -- VILHELMINA NORRA District Stocking Level -- VILHELMINA SODRA

3 4

2

2 1

0 0

Establishment Residuals -1 Establishment Residuals

-2 -2

4000 6000 8000 10000 12000 14000 4000 6000 8000 10000 12000

Vilhelmina.NReindeer Vilhelmina.SReindeer

Figure 21 continued. 74

Deviation from Predicted Establishment ~ District Stocking Level -- SORKAITUM

10

0

-10 Establishment Residuals

-20

4000 6000 8000

SorkaitumReindeer

Figure 22. When grouped by pentad, the relationship between stocking level and establishment in Sorkaitum was no longer 2 statistically significant (r = 0.410, p = 0.087). The other districts remained insignificant. 75

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: All Herding Districts Laevas Proportion of high vs. low recruitment Proportion of high vs. low recruitment -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 Summer Temp. High High Low Low

Reindeer Pop. Low High Low High Summer Temp. High High Low Low Reindeer Pop. Low High Low High

Figure 23. Proportion of high v. low recruitment years under four combinations of high/low reindeer populations and high/low temperatures. 76

Establishment Residuals Under Different Scenarios: Girjas Proportion of high vs. low recruitment -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

Summer Temp. High High Low Low Reindeer Pop. Low High Low High

Figure 23 continued. 77

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: Sorkaitum Vilhelmina Norra 0.5 0.0 -0.5 Proportion of high vs. low recruitment Proportion of high vs. low recruitment -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 -1.0

Summer Temp. High High Low Low Summer Temp. High High Low Low Reindeer Pop. Low High Low High Reindeer Pop. Low High Low High

Figure 24. Proportion of high v. low recruitment years under four combinations of high/low reindeer populations and high/low temperatures. 78

Establishment Residuals Under Different Scenarios: Vilhelmina Sodra Proportion of high vs. low recruitment -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4

Summer Temp. High High Low Low Reindeer Pop. Low High Low High

Figure 24 continued.

79

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: All Herding Districts Laevas 1.0 0.5 0.0 Proportion of high vs. low recruitment Proportion of high vs. low recruitment -0.5 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

Summer Temp. High High Low Low Summer Temp. High High Low Low Reindeer Pop. Low High Low High Reindeer Pop. Low High Low High

Figure 25. Pentadal data: Proportion of high v. low recruitment years under four combinations of high/low reindeer populations and high/low temperatures. Results for Vilhelmina Norra and Sodra are not displayed because the sample size was not sufficient. 80

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: Girjas Sorkaitum 1.0 0.5 0.5 0.0 0.0 -0.5 Proportion of high vs. low recruitment Proportion of high vs. low recruitment -0.5 -1.0

Summer Temp. High High Low Low Summer Temp. High High Low Low Reindeer Pop. Low High Low High Reindeer Pop. Low High Low High

Figure 25 continued. 8 1

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: All Herding Districts Laevas 0.5 0.0 -0.5 Proportion of high vs. low recruitment Proportion of high vs. low recruitment -1.0 -0.5 0.0 0.5 1.0

Summer Temp. H H H H M-HM-HM-HM-HM-LM-LM-LM-L L L L L Summer Temp. H H H H M-HM-HM-HM-HM-LM-LM-LM-L L L L L Reindeer Pop. L M-LM-H H L M-LM-H H L M-LM-H H L M-LM-H H Reindeer Pop. L M-LM-H H L M-LM-H H L M-LM-H H L M-LM-H H

Figure 26. Proportion of high v. low recruitment years under sixteen combinations of high/low/medium-high/medium-low (quartiles) reindeer populations and temperatures. Vilhelmina Norra and Sodra are not displayed because the sample size was insufficient. 82

Establishment Residuals Under Different Scenarios: Establishment Residuals Under Different Scenarios: Girjas Sorkaitum Proportion of high vs. low recruitment Proportion of high vs. low recruitment -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0

Summer Temp. H H H H M-HM-HM-HM-HM-LM-LM-LM-L L L L L Summer Temp. H H H H M-HM-HM-HM-HM-LM-LM-LM-L L L L L Reindeer Pop. L M-LM-H H L M-LM-H H L M-LM-H H L M-LM-H H Reindeer Pop. L M-LM-H H L M-LM-H H L M-LM-H H L M-LM-H H

Figure 26 continued. 83

Figure 27. Dendrogram of vegetation classifications produced via cluster analysis (Sorensen Distance Measure and Ward’s Linkage Method). Pruned to 4 Groups based on ISA and MRPP. 84

85

Figure 28. Vegetation groups obtained using Indicator Species Analysis (ISA) (cf., Dufrêne & Legendre, 1997). Species with indicator values > 25 % are listed under each group, at each grouping level, until their maximum indicator value (in bold) is reached.

86

Figure 28 continued.

Figure 29. Distribution of vegetation classes at study sites in Norrbotten County. 87

Figure 30. Distribution of vegetation classes at study sites in Västerbotten County. 88

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890

1 2

N = 1559 N = 846 6

5

4

3

2

1

0 3 4

6 N = 385 N = 1357 Percent of Total 5

4

3

2

1

0

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Establishment Year

Figure 31. Age class structures by vegetation classification (annual data).

89

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890

1 2 N = 1436 N = 792 20

15

10

5

0 3 4

N = 374 N = 1292

Percent of Total 20

15

10

5

0

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 Establishment Year

Figure 32. Age class structures by vegetation classification (pentadal data).

90

2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880

1 2

N = 1280 N = 638 50

40

30

20

10

0 3 4

N = 319 N = 1146 50 Percent of Total

40

30

20

10

0 2000 1990 1980 1970 1960 1950 1940 1930 1920 1910 1900 1890 1880 Establishment Year

Figure 33. Age class structures by vegetation classification (decadal data). 91

92

APPENDIX B

TABLES

Dry Heath: Dominated by low shrubs, particularly Vaccinium myrtillis, Empetrum nigrum, and low-growing Betula nana. Open plant cover. More species-rich on calcareous soils. Considerable lichen cover may be present on wind-exposed ridges.

Fresh Heath: Similar to Dry Heath, but with larger shrubs (Betula nana, Salix spp., Juniperus communis) afforded by greater winter snow cover.

Wet Heath: Heterogeneous heath with patchy tussocks of vegetation. May have patches of fresh heath interspersed with small mires.

Grass Heath: Sparse cover primarily consisting of Carex bigelowii., Juncus trifidus, Nardus stricta, Deschampsia flexuosa, and Festuca ovina. Similar to some meadows, but only well-irrigated immediately after snowmelt.

Low Herb Meadows: Dominated by grasses and herbs. Limited to areas with late snowmelt and considerable summer moisture. More species-rich on calcareous soils.

Willow Shrublands: Dense thickets of Salix spp. Primarily limited to consistently moist areas. Mountain birch (Betula pubescens ssp. czerepanovii) may establish if Salix cover is not too thick, or establish on the periphery.

Birch w/ Lichen: Birch forest with field layer consisting of heath-type plants. Lichens and dwarf shrubs (Juniperus communis, Empetrum nigrum) present. Found in relatively dry areas. Birches commonly polycormic.

Birch w/ Mosses: Most common birch forest. Understory consists of Vaccinium spp., Empetrum nigrum, Deschampsia flexuosa, mosses, and numerous herbs. Juniperus communis is locally common.

Table 1. Common vegetation types near treeline in Sweden (adapted from Green, 2005; Eriksson et al., 2007; personal observation, 2011)

Mean Reindeer Stocking Level Herding District # of Seedlings # of Sites (reindeer/km2)

Laevas 1369 9 7.11 Girjas 235 8 8.64 Sorkaitum 1410 21 7.49 Vilhelmina norra 1484 20 12.24 Vilhelmina sodra 164 5 9.15 Baste 0 4 20.62

Table 2. Summary statistics for study sites, divided by herding district.

93

Laevas

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.001 0.733 0.007 0.436 0.011 0.314 0.005 0.508 0.036 0.438 0.510 0.031 February 0.002 0.682 0.000 0.867 0.002 0.700 0.000 0.895 0.037 0.433 0.105 0.395 March 0.000 0.972 0.030 0.092 0.036 0.073 0.011 0.301 0.099 0.189 0.228 0.194 April 0.013 0.275 0.014 0.245 0.031 0.096 0.018 0.198 0.067 0.286 0.095 0.419 May 0.002 0.702 0.004 0.532 0.014 0.260 0.011 0.314 0.048 0.367 0.137 0.326 June 0.002 0.699 0.001 0.733 0.000 0.898 0.005 0.488 0.053 0.343 0.054 0.548 July 0.067 0.011 0.040 0.052 0.168 0.000 0.040 0.053 0.224 0.041 0.032 0.645 August 0.006 0.447 0.013 0.275 0.090 0.004 0.049 0.031 0.236 0.035 0.160 0.286 September 0.017 0.200 0.028 0.105 0.093 0.003 0.032 0.082 0.331 0.010 0.124 0.353 October 0.052 0.026 0.000 0.843 0.001 0.792 0.004 0.529 0.034 0.447 0.003 0.893 November 0.004 0.532 0.008 0.386 0.101 0.002 0.004 0.554 0.010 0.682 0.006 0.846 December 0.000 0.895 0.000 0.882 0.000 0.898 0.019 0.177 0.009 0.702 0.184 0.250

Table 3. Linear regression statistics for establishment as a function of mean monthly temperatures at different herding districts.

94

Girjas

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.017 0.211 0.002 0.692 0.084 0.005 0.000 0.890 0.002 0.873 0.174 0.264 February 0.006 0.440 0.001 0.799 0.008 0.396 0.005 0.510 0.009 0.703 0.040 0.605 March 0.023 0.140 0.022 0.151 0.104 0.002 0.000 0.869 0.005 0.765 0.053 0.552 April 0.002 0.635 0.000 0.873 0.023 0.149 0.013 0.270 0.000 0.940 0.217 0.207 May 0.001 0.739 0.005 0.510 0.004 0.574 0.012 0.295 0.008 0.721 0.551 0.022 June 0.015 0.238 0.008 0.375 0.056 0.024 0.001 0.752 0.000 0.995 0.002 0.917 July 0.016 0.214 0.024 0.134 0.055 0.026 0.055 0.022 0.209 0.049 0.271 0.151 August 0.000 0.916 0.001 0.788 0.010 0.337 0.007 0.413 0.123 0.140 0.412 0.063 September 0.005 0.512 0.001 0.807 0.001 0.717 0.005 0.496 0.134 0.124 0.015 0.755 October 0.001 0.826 0.001 0.813 0.023 0.147 0.006 0.474 0.010 0.689 0.164 0.280 November 0.005 0.501 0.011 0.304 0.029 0.107 0.019 0.188 0.017 0.592 0.017 0.738 December 0.010 0.341 0.009 0.365 0.058 0.021 0.015 0.238 0.009 0.705 0.091 0.431

Table 3 continued.

95

Sorkaitum

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.000 0.900 0.003 0.615 0.208 0.000 0.000 0.931 0.001 0.931 0.366 0.150 February 0.000 0.954 0.000 0.879 0.013 0.338 0.005 0.531 0.043 0.456 0.079 0.542 March 0.003 0.618 0.010 0.400 0.097 0.008 0.001 0.815 0.057 0.391 0.503 0.074 April 0.018 0.249 0.030 0.136 0.004 0.588 0.015 0.288 0.035 0.502 0.028 0.718 May 0.002 0.691 0.003 0.649 0.011 0.373 0.001 0.752 0.050 0.425 0.559 0.053 June 0.002 0.695 0.000 0.884 0.041 0.092 0.010 0.405 0.031 0.531 0.265 0.237 July 0.046 0.062 0.048 0.060 0.043 0.081 0.060 0.035 0.164 0.134 0.031 0.706 August 0.022 0.200 0.019 0.236 0.009 0.422 0.064 0.028 0.198 0.097 0.217 0.292 September 0.024 0.179 0.007 0.463 0.001 0.833 0.044 0.070 0.250 0.058 0.084 0.528 October 0.001 0.838 0.001 0.769 0.013 0.349 0.000 0.906 0.000 0.991 0.237 0.267 November 0.013 0.320 0.024 0.184 0.062 0.036 0.013 0.323 0.024 0.581 0.000 0.994 December 0.005 0.549 0.005 0.553 0.032 0.137 0.014 0.311 0.011 0.705 0.445 0.102

Table 3 continued.

96

Vilhelmina Norra

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.004 0.500 0.001 0.704 0.066 0.009 0.000 0.875 0.009 0.685 0.016 0.725 February 0.017 0.185 0.012 0.258 0.013 0.257 0.002 0.675 0.063 0.273 0.012 0.763 March 0.023 0.118 0.014 0.232 0.102 0.001 0.008 0.355 0.018 0.566 0.003 0.885 April 0.000 0.904 0.008 0.364 0.015 0.213 0.013 0.247 0.085 0.200 0.000 0.989 May 0.009 0.321 0.003 0.570 0.007 0.420 0.000 0.831 0.007 0.709 0.130 0.307 June 0.003 0.565 0.004 0.546 0.001 0.740 0.006 0.421 0.008 0.697 0.001 0.933 July 0.007 0.408 0.000 0.885 0.004 0.553 0.009 0.324 0.061 0.281 0.015 0.738 August 0.000 0.959 0.000 0.962 0.002 0.628 0.015 0.205 0.185 0.052 0.067 0.471 September 0.005 0.463 0.000 0.997 0.002 0.662 0.013 0.250 0.085 0.201 0.012 0.766 October 0.017 0.176 0.002 0.624 0.004 0.542 0.007 0.397 0.014 0.610 0.014 0.748 November 0.007 0.388 0.001 0.804 0.001 0.719 0.029 0.082 0.007 0.724 0.037 0.596 December 0.010 0.312 0.011 0.291 0.017 0.195 0.003 0.573 0.001 0.920 0.041 0.573

Table 3 continued.

97

Vilhelmina Sodra

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.000 0.870 0.005 0.495 0.008 0.401 0.000 0.959 0.002 0.847 0.030 0.656 February 0.001 0.752 0.004 0.531 0.006 0.456 0.010 0.342 0.005 0.784 0.018 0.730 March 0.008 0.378 0.001 0.726 0.073 0.009 0.058 0.019 0.032 0.464 0.001 0.935 April 0.019 0.176 0.011 0.304 0.034 0.079 0.017 0.207 0.019 0.571 0.001 0.952 May 0.020 0.165 0.014 0.254 0.003 0.610 0.000 0.958 0.090 0.213 0.264 0.157 June 0.013 0.263 0.001 0.714 0.000 0.995 0.016 0.215 0.029 0.489 0.002 0.917 July 0.003 0.602 0.010 0.339 0.013 0.280 0.003 0.579 0.002 0.848 0.014 0.763 August 0.001 0.820 0.005 0.503 0.003 0.613 0.041 0.050 0.041 0.403 0.189 0.242 September 0.000 0.909 0.031 0.089 0.014 0.266 0.013 0.271 0.099 0.189 0.000 0.977 October 0.002 0.671 0.000 0.910 0.006 0.455 0.000 0.942 0.106 0.175 0.005 0.850 November 0.003 0.627 0.000 0.995 0.011 0.333 0.006 0.458 0.006 0.751 0.099 0.409 December 0.007 0.432 0.007 0.427 0.002 0.686 0.003 0.583 0.040 0.413 0.097 0.416

Table 3 continued.

98

All Districts (~ Abisko Temps)

Contemporary Following Year Following 5 Years Previous Year 5 Year Aggregate 10 Year Aggregate 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value January 0.002 0.708 0.000 0.947 0.098 0.003 0.002 0.707 0.005 0.783 0.245 0.175 February 0.001 0.827 0.001 0.726 0.009 0.374 0.001 0.762 0.038 0.422 0.044 0.587 March 0.003 0.584 0.016 0.228 0.086 0.005 0.002 0.688 0.061 0.310 0.180 0.256 April 0.008 0.374 0.014 0.258 0.001 0.812 0.017 0.207 0.032 0.461 0.071 0.490 May 0.000 0.990 0.001 0.798 0.003 0.589 0.004 0.559 0.045 0.384 0.345 0.096 June 0.000 0.905 0.000 0.844 0.027 0.123 0.004 0.566 0.036 0.436 0.045 0.584 July 0.039 0.055 0.033 0.078 0.081 0.006 0.042 0.046 0.162 0.087 0.021 0.708 August 0.008 0.394 0.008 0.383 0.022 0.163 0.053 0.024 0.189 0.063 0.194 0.235 September 0.011 0.305 0.007 0.406 0.002 0.684 0.024 0.135 0.214 0.046 0.017 0.741 October 0.010 0.331 0.001 0.800 0.023 0.152 0.003 0.587 0.005 0.768 0.114 0.373 November 0.008 0.394 0.015 0.243 0.064 0.016 0.012 0.294 0.014 0.634 0.000 0.977 December 0.006 0.437 0.001 0.791 0.018 0.210 0.017 0.213 0.006 0.745 0.213 0.211

Table 3 continued.

99

R2 p-value Laevas 0.05 0.0322 Girjas 0.04 0.0660 Sorkaitum 0.10 0.0021 Vilhelmina norra 0.13 0.0005 Vilhelmina sodra 0.01 0.2603 All Districts 0.12 0.0012

Table 4. Linear regression statistics of establishment as a function of the following winter's precipitation at different herding districts.

Annual Growing Summer June July August 2 2 2 2 2 2 R p-value R p-value R p-value R p-value R p-value R p-value Laevas 0.275 0.021 0.294 0.016 0.188 0.064 0.040 0.412 0.159 0.091 0.201 0.054 Girjas 0.108 0.170 0.141 0.114 0.098 0.191 0.000 0.999 0.181 0.070 0.122 0.144 Sorkaitum 0.205 0.090 0.204 0.091 0.137 0.175 0.034 0.513 0.127 0.192 0.143 0.164 Vilhelmina norra 0.049 0.362 0.080 0.240 0.061 0.306 0.000 0.982 0.085 0.226 0.105 0.175 Vilhelmina sodra 0.000 0.988 0.006 0.755 0.020 0.566 0.003 0.826 0.025 0.514 0.015 0.620 All Districts 0.206 0.051 0.238 0.034 0.151 0.100 0.033 0.460 0.122 0.143 0.170 0.080

Table 5. Linear regression statistics of establishment as a function of degree days at different herding districts 100

101

Classification Species Max 1 2 3 4 Aconitum lycoctonum ssp. septentrionale 100 0 100 0 0 Alchemilla spp. 100 0 100 0 0 Antennaria dioica 100 100 0 0 0 Anthoxanthum alpinum 88 2 88 10 0 Arctostaphylos alpinum 53 53 2 0 46 Astragalus spp. 100 0 100 0 0 Bartsia alpina 99 0 99 0 1 Betula nana 50 50 12 11 28 Betula pubescens spp. czerapanovii 96 0 96 0 4 Bistorta vivipara 82 0 82 15 2 Calamagrostis spp. 52 33 52 6 9 Carex limosa 100 0 100 0 0 Carex spp. 44 22 44 9 25 Cerastium spp. 100 0 100 0 0 Cirsium spp. 100 0 100 0 0 Cornus suecica 100 0 0 0 100 Dactylorhiza viride 100 0 100 0 0 Deschampsia cespitosa 94 6 94 0 0 Deschampsia flexuosa 45 12 26 45 17 Diphasiastrum alpinum 57 7 57 35 1 Empetrum nigrum ssp. hermaphroditum 38 26 16 19 38 Epilobium anagallidifolium 100 0 100 0 0 Epilobium angustifolium 43 43 13 43 0 Epilobium spp. 100 0 100 0 0 Equisetum sylvaticum 68 0 68 0 32 Equisetum spp. 100 0 0 100 0 Euphrasia frigida 100 0 100 0 0 Euphrasia spp. 100 0 0 100 0 Festuca ovina 100 0 100 0 0 Festuca spp. 86 86 0 0 14 Geranium sylvaticum 99 0 99 1 0 Gnaphalium norvegicum 63 0 63 37 0 Gnaphalium supinum 100 0 100 0 0 Gymnocarpium dryopteris 49 0 49 49 2 Hieracium alpinum 52 0 52 46 2

Table 6. Relative abundance of species in the vegetation classifications. Dendrogram pruned at 4 groups. Relative abundance = % of perfect indication (average abundance of a given species in a given classification of sites over the average abundance of that species in all plots, expressed as a %).

102

Classification Species Max 1 2 3 4 Hieracium spp. 71 15 71 14 0 Hierochloė spp. 69 69 31 0 0 Juncus trifidus 100 0 100 0 0 Juncus spp. 71 71 0 29 0 Juniperus communis 46 37 46 5 12 Koenigia islandica 100 0 0 100 0 Linnaea borealis 95 95 0 4 2 Listera cordata 100 0 100 0 0 Loiseleuria procumbens 100 0 0 100 0 Luzula spicata 100 100 0 0 0 Luzula spp. 100 100 0 0 0 Lycopodium spp. 74 74 9 0 17 Melampyrum pratense 70 0 70 30 0 Melampyrum spp. 77 23 77 0 0 Melampyrum sylvaticum 63 0 63 22 15 Myosotis spp. 100 0 100 0 0 Oxyria digina 100 0 100 0 0 Parnassia palustris 100 0 100 0 0 Pedicularis lapponica 54 34 12 54 0 Pedicularis spp. 85 0 85 5 11 Phleum alpinum 100 0 100 0 0 Phyllodoce caerulea 61 0 20 19 61 Poa spp. 52 0 48 52 0 Potentilla erecta 100 0 100 0 0 Pyrola spp. 98 0 98 2 0 Ranunculus nivalis 100 0 100 0 0 Ranunculus spp. 100 0 100 0 0 Rubus arcticus 100 0 0 100 0 Rubus chamaemorus 78 78 0 0 22 Rumex acetosa 55 4 42 55 0 Salix herbacea 41 16 31 41 11 Salix spp. 70 1 70 22 7 Saussurea alpina 99 0 99 0 1 Sibbaldia procumbens 94 0 94 6 0 Silene dioica 100 0 100 0 0 Solidago spp. 79 0 19 79 2 Solidago virgaurea 53 8 53 38 1 Taraxacum spp. 100 0 100 0 0 Thalictrum alpinum 100 0 100 0 0

Table 6 continued.

103

Classification Species Max 1 2 3 4 Trientalis europaea 44 15 34 44 7 Trollius europaeus 100 0 100 0 0 Unknown forb 100 0 100 0 0 Unknown grass 100 0 100 0 0 Vaccinium myrtillis 36 8 23 36 32 Vaccinium uliginosum 54 13 32 1 54 Vaccinium vitis-idaea 38 36 20 7 38 Veronica alpina 100 0 100 0 0 Veronica spp. 100 0 100 0 0 Viola biflora 99 0 99 1 0

Table 6 continued.

104

Classification Species Max 1 2 3 4 Aconitum lycoctonum ssp. septentrionale 8 0 8 0 0 Alchemilla spp. 38 0 38 0 0 Antennaria dioica 6 6 0 0 0 Anthoxanthum alpinum 54 6 54 17 0 Arctostaphylos alpinum 21 19 8 0 21 Astragalus spp. 8 0 8 0 0 Bartsia alpina 31 0 31 0 4 Betula nana 69 69 46 42 64 Betula pubescens spp. czerapanovii 31 0 31 0 4 Bistorta vivipara 85 0 85 25 7 Calamagrostis spp. 69 50 69 25 11 Carex limosa 8 0 8 0 0 Carex spp. 54 13 54 33 29 Cerastium spp. 8 0 8 0 0 Cirsium spp. 8 0 8 0 0 Cornus suecica 36 0 0 0 36 Dactylorhiza viride 15 0 15 0 0 Deschampsia cespitosa 23 13 23 0 0 Deschampsia flexuosa 100 63 77 100 71 Diphasiastrum alpinum 33 6 23 33 4 Empetrum nigrum ssp. hermaphroditum 100 75 77 83 100 Epilobium anagallidifolium 8 0 8 0 0 Epilobium angustifolium 25 6 8 25 0 Epilobium spp. 8 0 8 0 0 Equisetum sylvaticum 8 0 8 0 4 Equisetum spp. 8 0 0 8 0 Euphrasia frigida 23 0 23 0 0 Euphrasia spp. 8 0 0 8 0 Festuca ovina 8 0 8 0 0 Festuca spp. 31 31 0 0 11 Geranium sylvaticum 77 0 77 8 0 Gnaphalium norvegicum 23 0 23 17 0 Gnaphalium supinum 15 0 15 0 0 Gymnocarpium dryopteris 8 0 8 8 4 Hieracium alpinum 54 0 54 50 7

Table 7. Relative frequency of each plant species in the vegetation classifications. Dendrogram pruned at 4 groups (lowest p-value, 2nd highest number of significant indicators in Indicator Species Analysis). Relative frequency = % of perfect indication (% of sites in a given classification where a given species is present).

105

Classification Species Max 1 2 3 4 Hieracium spp. 38 31 38 17 0 Hierochloė spp. 8 6 8 0 0 Juncus trifidus 8 0 8 0 0 Juncus spp. 19 19 0 8 0 Juniperus communis 31 31 23 8 11 Koenigia islandica 8 0 0 8 0 Linnaea borealis 63 63 0 8 11 Listera cordata 8 0 8 0 0 Loiseleuria procumbens 8 0 0 8 0 Luzula spicata 6 6 0 0 0 Luzula spp. 6 6 0 0 0 Lycopodium spp. 13 13 8 0 7 Melampyrum pratense 23 0 23 8 0 Melampyrum spp. 15 6 15 0 0 Melampyrum sylvaticum 31 0 31 8 4 Myosotis spp. 8 0 8 0 0 Oxyria digina 23 0 23 0 0 Parnassia palustris 31 0 31 0 0 Pedicularis lapponica 15 6 15 8 0 Pedicularis spp. 38 0 38 17 18 Phleum alpinum 8 0 8 0 0 Phyllodoce caerulea 43 0 23 42 43 Poa spp. 17 0 8 17 0 Potentilla erecta 8 0 8 0 0 Pyrola spp. 77 0 77 8 0 Ranunculus nivalis 54 0 54 0 0 Ranunculus spp. 8 0 8 0 0 Rubus arcticus 8 0 0 8 0 Rubus chamaemorus 6 6 0 0 4 Rumex acetosa 58 6 23 58 0 Salix herbacea 50 6 38 50 18 Salix spp. 92 6 92 42 25 Saussurea alpina 62 0 62 0 4 Sibbaldia procumbens 23 0 23 8 0 Silene dioica 8 0 8 0 0 Solidago spp. 25 0 15 25 7 Solidago virgaurea 77 31 77 67 7 Taraxacum spp. 38 0 38 0 0 Thalictrum alpinum 38 0 38 0 0

Table 7 continued.

106

Classification Species Max 1 2 3 4 Trientalis europaea 75 31 62 75 29 Trollius europaeus 62 0 62 0 0 Unknown forb 23 0 23 0 0 Unknown grass 15 0 15 0 0 Vaccinium myrtillis 100 44 85 100 93 Vaccinium uliginosum 71 31 54 17 71 Vaccinium vitis-idaea 100 88 54 33 100 Veronica alpina 38 0 38 0 0 Veronica spp. 23 0 23 0 0 Viola biflora 100 0 100 17 0

Table 7 continued.

107

Classification p- Species Max 1 2 3 4 value Aconitum lycoctonum ssp.septentrionale 8 0 8 0 0 0.000 Alchemilla spp. 38 0 38 0 0 0.605 Antennaria dioica 6 6 0 0 0 0.000 Anthoxanthum alpinum 47 0 47 2 0 0.512 Arctostaphylos alpinum 10 10 0 0 10 0.375 Astragalus spp. 8 0 8 0 0 0.003 Bartsia alpina 30 0 30 0 0 0.040 Betula nana 34 34 5 4 18 0.004 Betula pubescens spp. czerapanovii 30 0 30 0 0 0.000 Bistorta vivipara 70 0 70 4 0 0.007 Calamagrostis spp. 36 17 36 2 1 0.354 Carex limosa 8 0 8 0 0 0.099 Carex spp. 23 3 23 3 7 0.367 Cerastium spp. 8 0 8 0 0 0.025 Cirsium spp. 8 0 8 0 0 0.002 Cornus suecica 36 0 0 0 36 0.072 Dactylorhiza viride 15 0 15 0 0 0.015 Deschampsia cespitosa 22 1 22 0 0 0.000 Deschampsia flexuosa 45 7 20 45 12 0.245 Diphasiastrum alpinum 13 0 13 12 0 0.000 Empetrum nigrum ssp. hermaphroditum 38 20 12 16 38 0.349 Epilobium anagallidifolium 8 0 8 0 0 0.270 Epilobium angustifolium 11 3 1 11 0 0.364 Epilobium spp. 8 0 8 0 0 0.559 Equisetum sylvaticum 5 0 5 0 1 0.182 Equisetum spp. 8 0 0 8 0 0.008 Euphrasia frigida 23 0 23 0 0 0.166 Euphrasia spp. 8 0 0 8 0 0.364 Festuca ovina 8 0 8 0 0 0.009 Festuca spp. 27 27 0 0 1 0.000 Geranium sylvaticum 76 0 76 0 0 0.094 Gnaphalium norvegicum 15 0 15 6 0 0.059 Gnaphalium supinum 15 0 15 0 0 0.826 Gymnocarpium dryopteris 4 0 4 4 0 0.023 Hieracium alpinum 28 0 28 23 0 0.019

Table 8. Indicator values of each plant species in the vegetation classifications. Dendrogram pruned at 4 groups (lowest p-value, 2nd highest number of significant indicators in Indicator Species Analysis). Indicator values are based on combining the values for relative abundance and relative frequency.

108

Classification p- Species Max 1 2 3 4 value Hieracium spp. 27 5 27 2 0 0.720 Hierochloė spp. 4 4 2 0 0 0.364 Juncus trifidus 8 0 8 0 0 0.096 Juncus spp. 13 13 0 2 0 0.431 Juniperus communis 12 12 11 0 1 0.180 Koenigia islandica 8 0 0 8 0 0.000 Linnaea borealis 59 59 0 0 0 0.365 Listera cordata 8 0 8 0 0 0.173 Loiseleuria procumbens 8 0 0 8 0 0.588 Luzula spicata 6 6 0 0 0 0.595 Luzula spp. 6 6 0 0 0 0.316 Lycopodium spp. 9 9 1 0 1 0.043 Melampyrum pratense 16 0 16 3 0 0.134 Melampyrum spp. 12 1 12 0 0 0.031 Melampyrum sylvaticum 19 0 19 2 1 0.348 Myosotis spp. 8 0 8 0 0 0.009 Oxyria digina 23 0 23 0 0 0.001 Parnassia palustris 31 0 31 0 0 0.793 Pedicularis lapponica 4 2 2 4 0 0.005 Pedicularis spp. 33 0 33 1 2 0.348 Phleum alpinum 8 0 8 0 0 0.052 Phyllodoce caerulea 26 0 5 8 26 0.294 Poa spp. 9 0 4 9 0 0.354 Potentilla erecta 8 0 8 0 0 0.000 Pyrola spp. 75 0 75 0 0 0.000 Ranunculus nivalis 54 0 54 0 0 0.365 Ranunculus spp. 8 0 8 0 0 0.172 Rubus arcticus 8 0 0 8 0 0.667 Rubus chamaemorus 5 5 0 0 1 0.006 Rumex acetosa 32 0 10 32 0 0.136 Salix herbacea 20 1 12 20 2 0.000 Salix spp. 65 0 65 9 2 0.000 Saussurea alpina 61 0 61 0 0 0.372 Sibbaldia procumbens 22 0 22 1 0 0.375 Silene dioica 8 0 8 0 0 0.045 Solidago spp. 20 0 3 20 0 0.003 Solidago virgaurea 41 3 41 25 0 0.000 Taraxacum spp. 38 0 38 0 0 0.000 Thalictrum alpinum 38 0 38 0 0 0.025

Table 8 continued.

109

Classification p- Species Max 1 2 3 4 value Trientalis europaea 33 5 21 33 2 0.000 Trollius europaeus 62 0 62 0 0 0.011 Unknown forb 23 0 23 0 0 0.067 Unknown grass 15 0 15 0 0 0.000 Vaccinium myrtillis 36 4 20 36 30 0.003 Vaccinium uliginosum 39 4 17 0 39 0.001 Vaccinium vitis-idaea 38 31 11 2 38 0.000 Veronica alpina 38 0 38 0 0 0.007 Veronica spp. 23 0 23 0 0 0.000 Viola biflora 99 0 99 0 0 0.007

Table 8 continued.

110

______

Classification1: Greatest proportion of Betula nana; several species of grass. Vaccinium vitis-idaea, Linnaea borealis, and Festuca spp. common. Some Juniperus communis. Juncus and Luzula species in some locations. Similar to Fresh Heath/Grass Heath classifications.

Classification 2: Moist sites with numerous Trollius europaeus, Viola biflora, Salix spp., Veronica alpina, and Bistorta vivipara. Ranunculus nivalis, Geranium sylvaticum, Carex spp., and Parnassia palustris also found. The most diverse group. Similar to Low Herb Meadow classification.

Classification 3: Greatest proportion of Vaccinium myrtillis, but little or no other Vaccinium. Deschampsia flexuosa, Koenigia islandica, Rubus arcticus, and Salix spp. Vegetation typical of hummocky sites. Similar to Wet Heath classification.

Classification 4: Driest sites with greatest proportion of Empetrum nigrum, Vaccinium, and Cornus suecica. Some Arctostaphylos alpinum and Phyllodoce caerulea. Similar to Dry Heath classification. ______

Table 9. Descriptions of vegetation classifications derived from cluster analysis.

Veg Class Districts Compared Aggregation KS statistic p-value 1 Laevas - Girjas Annual 0.192 0.035 Pentadal 0.278 0.353 Decadal 0.306 0.638 2 Sorkaitum - Vilhelmina norra Annual 0.254 0.010 Pentadal 0.348 0.258 Decadal 0.339 0.660 3 Sorkaitum - Vilhelmina norra Annual 0.223 0.127 Pentadal 0.375 0.448 Decadal 0.500 0.552 4 Sorkaitum - Vilhelmina norra Annual 0.185 0.042 Pentadal 0.273 0.630 Decadal 0.309 0.821

Table 10. Comparison of age class structures for herding districts within the 4 different vegetation classifications (minimum of 3 sites in that classification).

Veg Classifications Compared Aggregation KS statistic p-value 2-3 Annual 0.292 0.016 Pentadal 0.339 0.501 Decadal 0.357 0.782 2-4 Annual 0.101 0.810 Pentadal 0.149 0.992 Decadal 0.286 0.909 3-4 Annual 0.224 0.131 Pentadal 0.284 0.734 Decadal 0.400 0.746

Table 11. Comparison of age class structures for vegetation classifications within Sorkaitum (minimum of 3 sites in that classification) 111

Veg Classifications Compared Aggregation KS statistic p-value 2-3 Annual 0.417 0.000 Pentadal 0.625 0.005 Decadal 0.625 0.093 2-4 Annual 0.273 0.001 Pentadal 0.313 0.263 Decadal 0.295 0.683 3-4 Annual 0.328 0.000 Pentadal 0.409 0.111 Decadal 0.455 0.332

Table 12. Comparison of age class structures for vegetation classifications within Vilhelmina Norra (minimum of 3 sites in that classification)

112