Assessing the Factors That Alter Ecological Responses of Cold, Oligotrophic Lakes to Nutrient Subsidies

The University of Maine DigitalCommons@UMaine

Electronic Theses and Dissertations Fogler Library

Spring 5-8-2020

Benjamin Burpee University of Maine, [email protected]

Follow this and additional works at: https://digitalcommons.library.umaine.edu/etd

Part of the Fresh Water Studies Commons

Recommended Citation Burpee, Benjamin, "Assessing the Factors That Alter Ecological Responses of Cold, Oligotrophic Lakes to Nutrient Subsidies" (2020). Electronic Theses and Dissertations. 3253. https://digitalcommons.library.umaine.edu/etd/3253

This Open-Access Thesis is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Electronic Theses and Dissertations by an authorized administrator of DigitalCommons@UMaine. For more information, please contact [email protected]. ASSESSING THE FACTORS THAT ALTER ECOLOGICAL RESPONSES OF COLD, OLIGOTROPHIC

LAKES TO NUTRIENT SUBSIDIES

Benjamin Burpee

B.S., B.A. University of Maine, 2011

M.S. University of Maine, 2015

A DISSERTATION

Submitted in Partial Fulfillment of the

Requirements for the Degree of

Doctor of Philosophy

(in Ecology and Environmental Science)

The Graduate School

The University of Maine

May 2020

Advisory Committee:

Jasmine Saros, Professor of Paleolimnology and Lake Ecology, Advisor

Brain McGill, Professor of Biological Science

Amanda Klemmer, Assistant Research Professor of Food-web Ecology

Kevin Simon, Associate Professor of Environment

Mark Skidmore, Professor of Earth Sciences ASSESSING THE FACTORS THAT ALTER ECOLOGICAL RESPONSES OF COLD, OLIGOTROPHIC

LAKES TO NUTRIENT SUBSIDIES

By Benjamin T. Burpee

Dissertation Advisor: Dr. Jasmine Saros

An Abstract of the Dissertation Presented in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy (in Ecology and Environmental Science) May 2020

Arctic and alpine areas experience rapid environmental changes that are altering nutrient delivery to lakes. I review the current state of lake ecosystem subsidization research and summarize Arctic and alpine lake subsidies, their ecological importance, and ways they are changing. I identify current knowledge gaps in Arctic and alpine lake subsidization research and highlight the importance of Arctic and alpine lakes for ecosystem subsidy research.

Meltwater from the Greenland Ice Sheet (GrIS) exports sediment and nutrients to lakes, but the ecological effects of this subsidy remain unclear. To assess the effects, four glacially fed

(GF) lakes that receive GrIS meltwater were compared to snow and groundwater fed (SF) lakes.

Total phosphorus (TP) was six times higher in GF lakes compared to SF lakes, but biological and chemical assays called into question its bioavailability. GF lakes had higher algal biomass, which may be due to moderately higher DIN in GF lakes compared to SF lakes.

Increased nitrogen (N) deposition rates over the past century have affected North

American and European alpine lakes by shifting nutrient limitation status and increasing algal biomass. I evaluated predictors of alpine and subalpine lake sensitivity across North American and European mountain ranges. North American lakes were more sensitive to N deposition than European lakes. Local-scale analyses revealed distinction of lake sensitivity controls, such as land cover, bedrock geology, maximum lake depth, and elevation.

Altered Arctic lake thermal stratification may lead to hypolimnetic anoxia and sediment phosphorus release. To assess this, I investigated Greenland lakes that exhibit different stratification patterns. Among anoxia-susceptible lakes, morphometric and stratification strength metrics were not able to explain variation of anoxia. Cooler spring temperatures were associated with mid-summer lake anoxia. Increased TP concentration and algal biomass were associated with lake anoxia.

My dissertation research summarizes how environmental change is altering lake nutrient subsidization and cycling in remote Arctic and alpine areas. It highlights ecological responses to those alterations. With accelerating rates of warming and increasing N deposition rates in Arctic and alpine areas, this research provides timely insights of how lakes located in cold, remote areas will change in the coming century.

DEDICATION

I dedicate my dissertation to my amazing family: Johanna Cairns, Denise Burpee, Todd

Burpee, and Rory Burpee.

ii ACKNOWLEDGEMENTS

This dissertation would have been impossible to complete were it not for the help and support of many people. First and foremost, my advisor, Jasmine Saros, has always been extremely generous with her time, advice, and the opportunities she has offered me.

Importantly, she has always provided an outstanding and consistent example of what it means to be a scientist, a researcher, a teacher, and a mentor. I could not have asked for a better advisor, and I am thankful to have stumbled into her office to naively chat about microbial ecology all those years ago…

My committee members have demonstrated an amazing degree of patience and flexibility over the course of my degree completion. Their feedback and questions along the way was invaluable. Brian, Amanda, Kevin, and Mark: thank you.

The members of the Saros lab have changed since I joined in 2013, but it has always been a tight and vibrant community of scientists. They have provided assistance when needed, imparted valuable knowledge and skills, and inspired me with their research and discussions.

These folks include Rachel Fowler, Nora Theodore Hwang, Václava Hazuková, Robert

Northington, Joseph Mohan, Kathryn Warner, Carl Tugend, Amy Kireta, Heera Malik, Edna

Pedraza Garzon, Rob Brown, Kristin Strock, Kelsey Boeff, Simona Lukasik, and Matthew

Farragher.

Finally, I would like to thank my family—without their support, this endeavor would have been much harder, if not impossible. Mom, Dad, Rory, and Jo: I love you all and I am immensely grateful for you.

iii TABLE OF CONTENTS

DEDICATION...... ii

ACKNOWLEDGEMENTS...... iii

LIST OF TABLES...... viii

LIST OF FIGURES...... ix

LIST OF ABBREVIATIONS...... xiii

Chapter

1. INTRODUCTION………………………………………………………………………………………………….…………..1

2. CROSS-ECOSYSTEM NUTRIENT SUBSIDIES IN ARCTIC AND ALPINE LAKES:

IMPLICATIONS OF GLOBAL CHANGE FOR REMOTE LAKES………………...…..……….…….……....6

Abstract…………………………………………………....…………………………………….………………...6

Introduction……………………………………………………………………………………….....…………..6

Arctic lake subsidies...... 15

Cryosphere...... 15

Atmospheric deposition...... 23

Animal vectors...... 30

Alpine lake subsidies...... 34

Cryosphere...... 34

Atmospheric deposition...... 37

Animal vectors...... 41

Conclusions and synthesis...... 44

iv 3. ASSESSING ECOLOGICAL EFFECTS OF GLACIAL MELTWATER ON LAKES FED BY

THE GREENLAND ICE SHEET: THE ROLE OF NUTRIENT SUBSIDIES AND TURBIDITY...... 49

Abstract...... 49

Introduction…………………………………………………………………………………………….....……50

Methods...... 54

Study site...... 54

Water parameters...... 55

Sediment Psenner extractions...... 57

Extracellular enzyme activities...... 58

Diatom community analysis...... 59

Statistical analyses...... 59

Results...... 60

Water quality...... 60

Lake sediments...... 63

Biological metrics...... 67

Discussion...... 71

4. SCALE-DEPENDENT PREDICTORS OF LAKE SENSITIVITY TO ATMOSPHERIC

NUTRIENT DEPOSITION IN ALPINE REGIONS...... 79

Abstract...... 79

Introduction………………………………………………………………………………….....………………80

Methods...... 84

Results...... 88

v Northern hemisphere scale...... 88

Regional scale...... 95

Local scale...... 98

Southern Rocky Mountains (United States)...... 98

Central Rocky Mountains (United States)...... 100

Northern Rocky Mountains (United States & Canada)...... 102

Pacific Northwest Mountains (United States)...... 102

Canadian Rocky Mountains (Canada)...... 103

Sierra Nevada Mountains (United States)...... 105

Alps (Italy, Austria, and Germany)...... 105

Tatra Mountains (Czech Republic)...... 105

Bombervalds (Czech Republic, Germany)...... 106

Discussion...... 107

5. ARCTIC LAKE SEDIMENT PHOSPHORUS RELEASE DURING ANOXIA INCREASES

ALGAL BIOMASS...... 116

Abstract...... 116

Introduction...... 117

Methods...... 121

Study site...... 121

Quantifying susceptibility to lake sediment P release...... 125

Assessing lake anoxia and its predictor variables...... 126

Quantifying ecological responses to sediment P release...... 130

vi Statistical analyses...... 130

Results...... 131

Susceptibility of lake sediment P release...... 131

Lake thermal stratification, anoxia, and predictor variables...... 132

Lake water chemistry response to hypoxia and anoxia...... 139

Ecological response to sediment P release...... 143

Discussion...... 145

6. CONCLUSION...... 152

REFERENCES...... 158

BIOGRAPHY OF THE AUTHOR...... 195

vii LIST OF TABLES

Table 3.1. Lake characteristics and sampling information...... 55

Table 3.2. Microbial extracellular enzyme activities (EEAs) of glacially fed (GF) and

snow- and groundwater-fed (SF) lakes...... 68

Table 4.1. Alpine lake data coverage...... 86

Table 4.2. Generalized additive model (GAM) summaries for dissolved inorganic

nitrogen (DIN) and Chlorophyll a (Chl a)...... 93

Table 4.3. Summary of elevation (alt) effects in generalized additive models

(GAMs) for dissolved inorganic nitrogen (DIN) and Chlorophyll a (Chl a)...... 108

Table 5.1. Study lake characteristics...... 122

Table 5.2. Annual climate data and lake indices for lakes located in the warmer,

drier area...... 129

viii LIST OF FIGURES

Figure 2.1. Conceptual model of hierarchical drivers acting to change lake

ecosystem functioning...... 10

Figure 2.2. Conceptual model of permafrost degradation and variable responses to

this disturbance...... 19

Figure 2.3. Conceptual model comparing nitrogen (N) deposition, watershed

features, and epilimnion mixed layer depth between Arctic (a) and

alpine (b) lake ecosystems...... 24

Figure 2.4. A general comparison of lake nutrient subsidies between Arctic (a) and

alpine (b) lakes, and their ecological outcomes...... 45

Figure 3.1. Map of study area, showing the locations of lakes relative to the

Greenland Ice Sheet...... 54

Figure 3.2. Physical and chemical properties of glacially fed (GF) and snow and

groundwater fed (SF) lakes...... 61

Figure 3.3. Principal component analysis (PCA) plot of physical and chemical

properties of glacially fed (GF) and snow and groundwater fed (SF) lakes...... 62

Figure 3.4. Sequential sediment extractions for aluminum (Al; a – e), iron (Fe; f – j),

and phosphorus (P; k – o) from glacially fed (GF) and snow and

groundwater fed (SF) lake sediments...... 64

Figure 3.5. Ratios of glacially fed (GF) and snow and groundwater fed (SF) lake

sediment extractions...... 65

ix Figure 3.6. Principal component analysis (PCA) plot of sequential extractions from

glacially fed (GF) and snow and groundwater fed (SF) lake sediments...... 66

Figure 3.7. Chlorophyll a (Chl a) as a measure of algal biomass (a) and average

species richness, determined by rarefaction analysis (b)...... 69

Figure 3.8. Canonical correspondence analysis (CCA) plot of diatom species

distribution (right panel) across glacially fed (GF) and snow and

groundwater fed (SF) lakes (left panel), determined by nutrient and

turbidity variables (middle panel)...... 70

Figure 4.1. Map of study lake coverage (n = 339, data contributed by the Alpine

Lake Sensitivity Working Group) across North America and Europe...... 84

Figure 4.2. Global dissolved inorganic nitrogen (DIN) regression tree analysis...... 89

Figure 4.3. Global Chlorophyll a (Chl a) regression tree analysis...... 90

Figure 4.4. Ranking dissolved inorganic nitrogen (DIN) control variables by their

importance via random forest analysis...... 91

Figure 4.5. Ranking Chlorophyll a (Chl a) control variables by their importance via

random forest analysis...... 92

Figure 4.6. European dissolved inorganic nitrogen (DIN) regression tree analysis...... 94

Figure 4.7. North American dissolved inorganic nitrogen (DIN) regression tree analysis...... 96

Figure 4.8. European Chlorophyll a (Chl a) regression tree analysis...... 97

Figure 4.9. North American Chlorophyll a (Chl a) regression tree analysis...... 98

Figure 4.10. Southern Rockies dissolved inorganic nitrogen (DIN) regression tree

analysis...... 100

x Figure 4.11. Central Rockies dissolved inorganic nitrogen (DIN) regression tree

analysis...... 101

Figure 4.12. Pacific Northwest Mountains dissolved inorganic nitrogen (DIN)

regression tree analysis...... 102

Figure 4.13. Pacific Northwest Chlorophyll a (Chl a) regression tree analysis...... 103

Figure 4.14. Canadian Rocky Mountains dissolved inorganic nitrogen (DIN) regression

tree analysis...... 104

Figure 4.15. Canadian Rockies Chlorophyll a (Chl a) regression tree analysis...... 106

Figure 5.1. Map of ten study lakes in the Southwest Greenland Arctic...... 123

Figure 5.2. Sequential lake sediment extractions for aluminum (Al), iron (Fe), and

phosphorus (P) from lakes located in the warmer, drier area and those

located in the cooler, wetter areas...... 133

Figure 5.3. Ratios of lake sediment aluminum:iron (Al:Fe) and aluminum:phosphorus

(Al:P) from lakes located in the warmer, drier area and lakes located in

the cooler, wetter area...... 134

Figure 5.4. Comparison of lake sediment organic matter (OM) content of lakes

located in the warmer, drier area vs those in the cooler, wetter area...... 135

Figure 5.5. Comparison of temperature (solid line) and dissolved oxygen (DO; dotted

line) profiles in an anoxia-susceptible lake (SS1381) in July of 2015, when

it was anoxic, and 2016, when stratification was weak, and the lake

remained oxygenated...... 136

xi Figure 5.6. Comparison of hypolimnetic dissolved oxygen (DO) concentration

between lakes located in the warmer, drier area vs those located in the

cooler, wetter area...... 137

Figure 5.7. Individual lake hypolimnetic DO concentration in lakes located in the

warmer, drier area, averaged across sampling years (2013-2019)...... 138

Figure 5.8. Peak summer hypolimnetic dissolved oxygen (DO) concentration as a

function of mean spring temperature (March-May)...... 140

Figure 5.9. Lake hypolimnetic total phosphorus (TP) as a function of dissolved

oxygen (DO)...... 141

Figure 5.10. Lake hypolimnetic algal biomass (measured as Chlorophyll a, Chl a) as a

function of dissolved oxygen (DO)...... 142

Figure 5.11. Hypolimnetic nutrient limitation statuses (dissolved inorganic nitrogen

to total phosphorus, or DIN:TP) according to oxygen status...... 144

xii LIST OF ABBREVIATIONS

AIC Akaike information criterion

Al Aluminum

Al:Fe Al : FeNH4Cl+BD+NaOH ratio

Al:P AlNaOH : PNH4Cl+BD ratio

ANC Acid neutralizing capacity

ANOVA Analysis of variance

AP Phosphatase (AP)

As Surface area

BD Na2S2O4 buffered with NaHCO3

BF Brunt-Väisälä buoyancy frequency

BG β-1,4-glucosidase

C Carbon

Ca Calcium

CA:LA Catchment area : lake area ratio

CCA Constrained correspondence analysis

CEI Climate exposure index

Chl a Chlorophyll a

DIN Dissolved inorganic nitrogen

DO Dissolved oxygen

DOC Dissolved organic carbon

DOM Dissolved organic matter

xiii E Energy

EEA Extracellular enzyme activity

Fe Iron

GAM Generalized additive model

GCV Generalized cross-validation

GF Glacially-fed

GR Lake geometry ratio

GrIS Greenland Ice Sheet

K Potassium

LAP Leucine aminopeptidase

LC-PUFAs Long-chain polyunsaturated fatty acids

M Mass

MAM March, April, May

Mg Magnesium

N Nitrogen

NAG β-1,4-N-acetylglucosaminidase

NFUs Nephelometric turbidity units

+ NH4 Ammonium

- NO3 Nitrate

Nox Oxidized nitrogen

Nr Reactive nitrogen

Nred Reduced nitrogen

xiv OM Organic matter

P Phosphorus

PAR Photosynthetically active radiation

PCA Principal correspondence analysis

POM Particulate organic material

PUFAs Polyunsaturated fatty acids

Si Silica

SF Snow and groundwater-fed

SRP Soluble reactive phosphorus

SSI Schmidt’s stability index

TDN Total dissolved nitrogen

TN Total nitrogen

TP Total phosphorus

UVR Ultraviolet radiation

Zmax Maximum lake depth

CHAPTER 1

INTRODUCTION

Arctic and alpine lakes are sites of rapid environmental change. Alpine and Arctic climates are becoming warmer with increased radiative forcing, elongating ice- and snow-free growing seasons (Brown and Mote 2009; Bolch et al 2012; Haeberli 2013; Rabatel et al 2013).

Accelerated climate warming is drastically diminishing glaciers, permafrost, and perennial snow cover in mountain and high latitude regions across the world. Receding glaciers and enhanced permafrost thaw are likely to deliver increasing amounts of meltwater containing solutes, sediments, contaminants, and nutrients to connected lake and stream ecosystems (Hawkings

2015; 2016). Delivery of nutrients and materials across distinct habitats such as glaciers and lakes constitute a cross-boundary ecosystem subsidization (Polis et al 1997). Cross-ecosystem nutrient subsidies can have strong effects on lake systems by changing nutrient limitation

(Bergstrom and Jansson 2006; Elser et al 2009; Saros et al 2010; Brahney 2015), algal species community and diversity (Slemmons and Saros 2012; Slemmons et al 2015), and whole lake metabolic function (Cole et al 2000; Hanson et al 2003). In the case of glacier meltwater, subsidized aquatic ecosystems may experience high turbidity (Sommaruga 2015) and increased surface temperatures (Peter and Sommaruga 2017) and shifting nutrient limitation patterns

(Saros et al 2010).

In alpine regions, climate warming and glacial meltwater are not the only drivers of algal ecological changes and community reorganizations; rates of atmospheric nitrogen (N) deposition are increasing in alpine areas of North America (Neff et al 2008) and are another example of cross-boundary ecosystem subsidization (in this case, atmosphere-to-lake) driving

lake ecological change. Ecological responses to N deposition include changes in ecological communities and biodiversity (Wolfe et al 2001; 2003; Slemmons and Saros 2012) and increased primary productivity rates (Nydick 2004; Slemmons and Saros 2012). Nutrient subsidies to lakes often produce variable effects, however, since individual lakes have numerous intrinsic and extrinsic physical, chemical, and biological factors that interact with nutrient additions. This complexity results in different responses of lakes to similar amounts of nutrient addition (Magnuson et al 1990). Examples of these mediating factors include catchment area and vegetation cover, lake depth, initial nutrient concentrations, sediment aluminum and iron content, and starting microbial species composition. Thus, the responses of algal communities that are sensitive indicators of nutrient subsidy changes (Adrian et al 2009;

Williamson et al 2009; Baron et al 2011; Pardo et al 2011) differ among alpine lakes receiving the same amount of atmospheric nitrogen deposition (Spaulding et al 2015), and phytoplankton biomass responds differently among alpine lakes experiencing synchronous phosphorus (P) reduction (Anneville et al 2005). Being able to predict the sensitivity of a lake ecosystem to a particular nutrient subsidy, such as atmospheric N deposition, for example, has important management implications, as sensitive lake ecosystems can be targeted for monitoring or conservation efforts.

In addition to altering external inputs of nutrients, an effect of climate warming is the alteration of biogeochemical nutrient cycling within lake ecosystems. For instance, increased temperatures and radiative forcing results in earlier ice off and changes in thermal structure that produce more strongly stratified and shallower epilimnia (Saros et al 2003; Šmejkalová et al 2016). As strength and stability of stratification influences the extent of mixing between

hypolimnetic and epilimnetic water layers, it affects hypolimnetic oxygen concentration which can be depleted by heterotrophic processing of organic material (Jankowski et al 2006).

Sustained stratification prevents hypolimnetic oxygen replenishment following its consumption, which has implications for lake water chemistry. Under oxygenated conditions, lake sediment P is readily adsorbed to iron (Fe) and aluminum (Al) oxides which limit its availability to phytoplankton and other microorganisms. Depletion of oxygen in lake hypolimnia increases the environmental reduction potential and is associated with the release of P from lake sediments as it dissociates from Fe- and Al-oxides. Thus, higher P concentrations are typically observed in anoxic lake hypolimnia (Marsden 1989, Søndergaard et al 2003). As P is a common limiting nutrient, this has important implications for lake ecosystems. In temperate lakes, algal blooms are triggered when accumulation of P within the hypolimnion leads to a nutrient pulse in the lake upon mixing (Sommer 1985; French and Petticrew 2007; Wilhelm and Adrian 2008).

Despite the prediction of increased stratification intensity and stability under current trends of climate warming (Winder and Sommer 2012; Vincent et al 2013), this process is not well understood in remote Artic lake ecosystems.

The objectives of this dissertation research were to determine how environmental change is altering nutrient subsidization and nutrient cycling in remote Arctic and alpine lakes, and how lake ecosystems respond to those changes. Remote Arctic and alpine lakes are typically dilute, contain low nutrient concentrations, can have sparse catchment vegetation, and are biologically unproductive relative to temperate lakes. Because of this, they are particularly sensitive to even small nutrient subsidies or changes to in-lake nutrient cycling

(Adrian et al 2009). Algal communities found in Arctic and alpine lakes respond rapidly to

nutrient changes because of their high reproduction rates. Because relatively small nutrient additions can produce large ecological effects on lake algae communities, Arctic and alpine lakes are excellent study sites to evaluate the ecological consequences of changing nutrient subsidies and internal cycling. Because of their sensitivity and location in rapidly changing areas,

Arctic and alpine lake ecosystems serve as sentinel ecosystems that exhibit ecological responses which may be anticipated in warmer and more nutrient-rich temperate lakes

(Williamson et al 2009).

The following chapters together address my overall research objectives. Chapter 2 is a review and synthesis of lake subsidies in Arctic and alpine landscapes. I consider similarities and differences of ecosystem subsidy impacts between Arctic and alpine lakes and discuss how abrupt climatic changes in these remote areas will alter lake ecosystem subsidization. In

Chapter 3, I evaluate the physical and chemical characteristics of glacially fed lakes located in

West Greenland adjacent to the Greenland Ice Sheet. To understand the ecological effects that turbid and nutrient-rich meltwater has on lake ecological communities, I compare surface sediment diatom records of the glacially fed study lakes to nearby snow and groundwater fed reference lake ecosystems. This chapter is based on original data that I collected in the field. In

Chapter 4, to understand how cross-ecosystem subsidies of the same type and magnitude can produce different ecosystem responses, I assess the scale-dependent effects of lake and catchment characteristics that together control lake ecosystem sensitivity to N deposition across alpine regions of North America and Europe. This chapter is based on a database that I assembled from numerous previous studies by other investigators (as indicated in the chapter), and I selected and conducted all of the data analyses. In Chapter 5, to understand how climate

warming is impacting internal P cycling patterns in Arctic lakes (specifically, drivers of lake anoxia and subsequent lake sediment P release), I evaluate West Greenland lake sediment chemistry, climate records, and lake survey data collected from 2013-2019. To understand the ecological effects of lake sediment P release, I assess the relationship between anoxic lake conditions and hypolimnetic algal biomass. This chapter is based on original data that was collected by various members (including me) of the UMaine Diatom Ecology Lab for various purposes; I devised the questions and conducted the analyses for this chapter.

CHAPTER 2

CROSS-ECOSYSTEM NUTRIENT SUBSIDIES IN ARCTIC AND ALPINE LAKES: IMPLICATIONS OF

GLOBAL CHANGE FOR REMOTE LAKES

Abstract

Environmental change is continuing to affect the flow of nutrients, material and organisms across ecosystem boundaries. These cross-system flows are termed ecosystem subsidies. Here, we synthesize current knowledge of cross-ecosystem nutrient subsidies between remote lakes and their surrounding terrain, cryosphere, and atmosphere. Remote

Arctic and alpine lakes are ideal systems to study the effects of cross ecosystem subsidies because a.) they are positioned in locations experiencing rapid environmental changes, b.) they are ecologically sensitive to even small subsidy changes, c.) they have easily defined ecosystem boundaries, and d.) a variety of standard methods exist that allow for quantification of lake subsidies and their impacts on ecological communities and ecosystem functions. We highlight similarities and differences between Arctic and alpine systems and identify current knowledge gaps to be addressed with future work. It is important to understand the dynamics of nutrient and material flows between lakes and their environments in order to improve our ability to predict ecosystem responses to continued environmental change.

Introduction

Ecosystem subsidies, defined as cross-boundary material fluxes between ecosystems

(Polis et al 1997), are important because they enrich communities with nutrients, materials, and energy that might otherwise be scarce or unavailable. Commonly discussed nutrients in freshwater ecology research (as well as this review) include the elements nitrogen, phosphorus,

silica, iron, and carbon (N, P, Si, Fe, and C, respectively), their inorganic forms (such as dissolved inorganic N, or DIN) and organic forms (such as polyunsaturated fatty acids, or PUFAs) that are present in the environment as resources for organisms. In aquatic environments, nutrient subsidies can support algal communities dominated by few high-nutrient species (Slemmons and Saros 2012; Slemmons et al 2017), change ecosystem function and food web dynamics

(Nakano and Murakami 2001; Nydick et al 2004; Slemmons and Saros 2012; Piovia-Scott et al

2016), sustain species abundance, increase community turnover (Wolfe et al 2001; 2003;

Slemmons et al 2015), and influence biological richness of animals, plants, and microbiota

(Xiankai et al 2008; Bobbink et al 2010), but effects will vary depending on environmental nutrient limitation. In this review, we address the need for a synthesis of the growing body of research investigating subsidies to freshwater ecosystems. Due to their sensitivity and location within rapidly changing environments, we focus primarily on Arctic and alpine lake ecosystems.

We highlight the common themes, differences, and emerging challenges and opportunities for subsidy research within Arctic and alpine lakes and confirm these remote systems as important locales for continued subsidy research.

The study of cross-ecosystem subsidies emerged as ecologists assessed processes occurring over multiple spatial and temporal scales (Levin 1992). Growing from this recognition of complex and multiscale ecosystem processes, the two subdisciplines of landscape ecology and meta-ecosystems emerged. Landscape ecology recognized the importance of considering broad geographic scales and heterogeneous patterning of landscapes in shaping ecosystems and ecological communities (Turner 1989). Importantly, two of its main foci include “the interactions and exchanges across heterogeneous landscapes,” (172) and “the ways in which

fluxes are controlled within heterogeneous matrices” (Pickett 1995; 331). Within the schema of landscape ecology, ecosystem subsidies are exchanges between heterogeneous environments nested within a broader landscape. Further, the heterogeneity of a landscape (for instance, the variable vegetation cover of lake catchments moving up an elevation gradient) affects the quantity and quality of transported materials. Thus, ecosystem subsidies are centrally important in landscape ecology.

To explain how ecological processes operate over multiple spatial scales, Loreau et al

(2003) described the meta-ecosystem concept. A meta-ecosystem is a group of ecosystems that are connected through fluxes of material, species, and energy. This differs from the landscape ecology perspective in that connected ecosystems need not be spatially continuous or defined at a particular spatial scale (i.e. the landscape scale). Thus, ecosystem subsidies that originate remotely from the site where they are transported (such as agricultural fertilizers that are atmospherically deposited on distant alpine ecosystems) are included within this conceptual framework.

The singular topic of cross-ecosystem subsidies, which move across heterogeneous landscapes and enrich receiving ecosystems with nutrients or energy, was explicitly addressed by Polis et al (1997). In particular, Polis et al reviewed the effect of ecosystem subsidies on food web dynamics and described the connectedness of habitats via fluxes of material, nutrients, or organisms. Following Polis et al (1997), much of ecosystem subsidy research has remained focused on food web and community effects (e.g., Nakano et al 1999; Holt 2002; Huxel et al

2002; Sears et al 2004; Barrett et al 2005; Marczak et al 2007; Leroux and Loreau 2008;

Marcarelli et al 2011; Scharnweber et al 2014; Fellman et al 2015; Creed et al 2018).

At present, new challenges are evident in ecosystem subsidy research. For instance, over the past 20 years studies note that peak productivities and transfers between aquatic/terrestrial habitats can be asynchronous and seasonal (e.g., Nakano and Murakami

2001; Takimoto et al 2002). In other words, fluxes can vary temporally (Yang et al 2008). In the past decade research has demonstrated that subsidies of matter and organisms can be bidirectional, or reciprocal, across coupled ecosystems (Soininen et al 2015). Further, ecosystems are experiencing rapid climate change, a driver that exercises both direct and indirect effects upon the timing, direction, quality, and quantity of ecosystem subsidies. Leavitt et al (2009) demonstrated that in lakes experiencing climate warming, lake inputs can be divided into two categories, energy (E) and mass (M). While lakes will tend to respond coherently through time to changes in E inputs (for instance, increased PAR irradiance), lake responses to M inputs will likely be much more variable depending on a host of catchment and lake properties and processes. Cross-ecosystem subsidies can be categorized as M inputs. The topic of ecosystem subsidies is therefore complex: ecologists are not only concerned with structural ecosystem responses to unidirectional subsidies (e.g., Caraco and Cole 2004), they are concerned with effects of climate change upon the direction, timing, and magnitude of subsidies themselves. There is a hierarchy of factors and responses involved (Figure 2.1), and many are interactive, making it difficult to focus on any single pair of response/control variables.

10 ENVIRONMENTAL FORCERS

WARMIN G ( ) N, P DEPOSITION ( ) PRECIPITATION( )

20 LAKE SUBSIDY CHANGES BIDIRECTIONAL INPUTS

DOC Fe POM

DIN Si Turbidity

TP ANIMAL INPUTS

0 3 ECOLOGICAL (FUNC TIONAL) CHANGES

C CYCLING RATES SEDIMENTATION RATES

- PRIMARY PRODUCTION -- C, N, & P BURIAL

- RESPIRATION N & P CYCLING - METHANOGENESIS - LAKE SEDIMENT P RELEASE - NITRIFICATION AND DENITRIFICATION

Figure 2.1. Conceptual model of hierarchical drivers acting to change lake ecosystem functioning. Primary (1°) drivers are at the regional scale and include climate warming, atmospheric nutrient deposition, and changing precipitation patterns. These primary drivers alter secondary (2°) factors, which include changes in material transfer into and out of lakes, i.e. nutrient subsidies. Nutrient subsidies then have ecological impacts (tertiary or 3° responses) which change ecosystem functioning such as carbon and nutrient cycling rates. C = carbon, Fe = iron, N = nitrogen, P = phosphorus, Si = silica, DIN = dissolved inorganic N, DOC = dissolved organic C, POM = particulate organic matter, TP = total P.

In the face of this challenge, systems that exhibit naturally well-defined ecosystem boundaries, measurable fluxes of material or organisms, and measurable ecosystem responses to subsidies will lend themselves to ecosystem subsidy research. Lake ecosystems are therefore exemplary for such studies. While lakes are open systems, their physical boundaries are distinct, and they are relatively self-contained depending on water residency time. Following

Strayer et al (2003), we consider lake ecosystem boundaries in this review in terms of our investigative goals, relevant scale, and processes of interest. The ecosystem boundary of lakes in this review are therefore delineated by the basin and body of lake water, because we are focused on whole-lake ecological responses to ecosystem subsidies. However, we recognize lakes are part of their larger watershed (meta-ecosystem), and the lake subsidies we consider can be categorized as within-basin (e.g. benthic to pelagic transfers such as sediment P release), within-watershed (e.g. groundwater inputs of DOM), and extra-watershed transfers (e.g. atmospheric N deposition).

Fluxes into and out of lakes are measured or modeled with increasing ease and accuracy

(e.g., Coppin et al 2002; Qiang et al 2007; Nanus et al 2012). The ease with which ecological responses to cross-system transfers are registered and quantified in lakes also makes them convenient for subsidy research (Battarbee et al 2002; Leavitt et al 2002; Smol et al 2002;

Harper and McKay 2010). Because of their position at low points in the landscape, lakes behave as integrators and archives of the environmental changes that occur across landscapes and catchment areas (Williamson et al 2009). The sedimentation processes that occur within lake basins produce temporally continuous records of ecosystem change. Thus, the effects of subsidies, and the environmental factors that control those subsidies, can be recorded in lake

sediment records via quantification of algal pigments (Slemmons et al 2015), diatoms

(Slemmons et al 2017), and isotopes (Anderson et al 2018).

Finally, lakes are highly sensitive, behaving as sentinels of ecosystem change in response to environmental drivers (Williamson et al 2009). Mueller et al (2009) summarize why high

Arctic lakes located on Ellesmere Island, Canada serve as global sentinel systems of climate warming by drawing on four decades of data collection. Ellesmere Island lake ice-out phenology and mixing regime shifted abruptly in response to changes in air temperature when warming trends changed the lakes from permanently ice-covered systems to seasonally open water.

Seasonally open water exposed the lake surface to wind-driven mixing, thus changing the water column structure. Though biological metrics were not explicitly considered by Mueller et al

(2009), the ecological communities in Ellesmere Island lakes are dependent on water column mixing regime; thus, these lakes were ecologically sensitive to climate forcing and were likely vulnerable to cascading ecological regime shifts in response to the physical lake changes. As

Ellesmere Island lake changes occurred synchronously with ice shelf breakup and changes to permafrost, sea ice, and glaciers, the authors demonstrate that lakes serve as robust sentinels of environmental change beyond the regional scale. In summary, ecological sensitivity, physical boundaries, location within the landscape, and a plethora of standard evaluative methods mean lakes are ideal to study how rates of cross-ecosystem fluxes may shift due to anthropogenic pressures such as land use change, nutrient delivery through atmospheric deposition or non-point sources, and climate change.

Dilute, nutrient-poor and cold-water lakes with relatively simple ecological communities are particularly sensitive to cross-ecosystem subsidies. This is due to watershed characteristics

of high Arctic and alpine lakes that enhance transport of nutrients to lakes (e.g. Sickman et al

2002; Clow et al 2010; Nanus et al 2012) and because nutrient-poor lake ecosystems respond to lower thresholds of nutrient inputs (for example, alpine lakes in western US respond to lower amounts of atmospherically deposited N compared to other ecoregions Saros et al 2011; Baron et al 2011; Pardo et al 2011). Thus, high-elevation mountain and Arctic lakes are very responsive to changes in subsidy patterns (Adrian et al 2009; Mueller et al 2009). Because of

Arctic and alpine lake sensitivity, ecosystem responses observed first in these systems provides insight into changes ecologists might expect to eventually observe in lakes situated in warmer climates (i.e. lower latitude and lower elevation climates with higher annual mean air temperature compared to Arctic or high elevation environments) with more complex and diverse ecological communities (Williamson et al 2009). For these reasons, Arctic and alpine lakes are exceptional systems in which to conduct nutrient subsidy research.

Arctic and alpine lakes have served as systems to evaluate ecosystem subsidies for several decades. In the 1970s, nutrient fluxes were studied in Arctic ponds of Barrow, Alaska to account for ecosystem nutrient budgets (Hobbie 1980). This study provided early insights into the sources and fates of nutrients within the tundra landscape. Lake eutrophication in the high

Canadian Arctic resulting from N and P associated with sewage effluent was evaluated by

Schindler et al 1974. Ecological effects of enhanced N delivery to Lake Tahoe (located in the

Sierra Nevada Mountains of California) were evaluated by Goldman 1988 using a time series dating back to 1968. Eutrophication from land use changes and urbanization decreased lakewater clarity and increased lake primary productivity.

Arctic and alpine lakes are situated in areas that are experiencing accelerated rates of climate warming (Hassan et al 2005). The pan-Arctic region includes large areas of North

America, Europe, and Asia. Major alpine lake districts are distributed globally, including the

North American Rocky Mountains, European Alps, Himalayas and Tibetan Plateau in central and western Asia, and the South American Andes. Climate warming controls many environmental changes that are occurring in Arctic and alpine lake catchments, including glacier recession

(Meier et al 2003), catchment “greening” (Guay et al 2014), permafrost degradation (Jorgenson et al 2001; Jorgenson et al 2006), increased abundance of waterfowl (Fox et al 2005), altered quantity and timing of insect emergence (Greig et al 2012), and altered hydrological connectivity (Lane et al 2017). In turn, many of these environmental changes control the magnitude, frequency, timing, and quality of subsidies delivered to lakes from the atmosphere, cryosphere, and terrestrial catchment. Changes of cross-ecosystem subsidies are presently not well accounted for. Large scale environmental drivers such as climate warming will continue to impact ecosystem subsidies into the future, making this area of research timely and urgent.

Thus, alpine and Arctic areas are insightful landscapes in which numerous “natural experiments”, or comparative studies, can be conducted to understand how cross-ecosystem subsidies affect lakes. Already, researchers have compared effects of glacial meltwater (Saros et al 2010), permafrost slumping (Mesquita et al 2010), insect emergence (Dreyer et al 2012;

Bultman et al 2014), and dissolved organic carbon (DOC) impacts (Mariash et al 2018) on nearby paired Arctic and alpine lake ecosystems that are differentially influenced by environmental changes. Equally important are the paleoecological insights that lake sediment

records provide, whereby pre-Anthropocene algae assemblages or isotope chemistries can be compared to recent ones within the same lake ecosystem.

In this review we will synthesize such studies in order to outline the current state of ecosystem subsidy research in Arctic and alpine lakes. Previous studies have reviewed subsidy effects in remote lakes (Catalan et al 2006; Hobbs et al 2010; Catalan et al 2013). However, these papers focused on paleolimnological records (Hobbs et al 2010; Catalan et al 2013) or were location-specific (Catalan et al 2006). Our review surveys cross-ecosystem subsidies to remote lakes drawing on both contemporary ecological as well as paleolimnological research.

Because of its central importance to cross-ecosystem subsidies, we will highlight the current and future effects of abrupt climate change. Though we consider community and food web consequences of lake ecosystem subsidies, we emphasize biological and biogeochemical responses and trends in order to situate cross-ecosystem subsidy research as an important facet of Arctic and alpine carbon (C) and nutrient cycling and climate change feedback loops.

Conceptual models that illustrate current understanding of the drivers and effects of lake subsidies explore this paradigm. Finally, we will address certain challenges and open questions in lake ecosystem subsidy studies and call for research to address these knowledge gaps.

Arctic lake subsidies

Cryosphere

Across arctic and alpine landscapes alike, perhaps the most visually obvious change is the rapidly changing cryosphere involving the recession of glaciers, permafrost thawing, and diminishing snowpack. Glaciers concentrate and store atmospheric compounds, including anthropogenic pollutants. In this way, glaciers behave as “natural archives” because the

combined effects of emission trends and air circulation patterns is recorded within their ice

(Wang et al 2010). Thus, glaciers are reservoirs and secondary sources of anthropogenic pollutants (Schmid et al 2011), because these compounds are present in glacier meltwater and are delivered to downstream ecosystems. Climate warming increases glacial meltwater output and accelerates the release of accumulated pollutants (Bogdal et al 2010). While glacial pollutant storage is often associated with mountain glaciers that are in closer proximity to source emissions, anthropogenic organic pollutants have been detected in remote Arctic glacial ice and snowpack (Wania and Mackay 1993; Kawamura et al 1994; AMAP 2002). In addition to pollutants, glacier meltwater contains nutrient solutes as well as minerogenic particulates that are ecologically important. These nutrients and minerals may be sourced from the atmosphere and concentrated in glacier ice to be released upon melt (Daly and Wania 2005), similar to anthropogenic pollutants, or derived from bedrock material that gets weathered by biogeochemical and physical processes and washed out via subglacial flow (Williams et al 2007).

Ice sheets and glaciers store a significant amount of labile particulate and dissolved organic matter (DOM; Hood et al 2009; 2015), N (Saros et al 2010), P (Hodson et al 2004; Hawkings et al

2016; Burpee et al 2018), Si (Hawkings et al 2015; 2016), as well as many other micronutrients

(such as Fe; Bhatia et al 2013). Thus, Arctic glacier meltwaters represent an important nutrient subsidy for downstream freshwater ecosystems.

In southwest Greenland, Bhatia et al (2013) and Hawkings et al (2015) demonstrated significant amounts of P, N, Si, and Fe are contained in meltwater discharged from the

Greenland Ice Sheet. Ice sheet meltwater total P (TP) and DIN were important determinants of distinct algal communities and higher algal biomass in Greenland glacially-fed lakes compared

to nearby snow and groundwater-fed lakes (Burpee et al 2018). Similarly, a lake sediment record from northeast Greenland (Bunny Lake, fed by the Renland Ice Cap) show diatom community changes that likely coincided with glacier meltwater inputs starting a thousand years ago (Slemmons et al 2017). Glacially-fed lakes in Svalbard and northern Sweden also exhibited elevated TP concentrations (Hodson et al 2004). Together, these studies show the widespread potential for glacier meltwater being an important source of nutrient subsidies in the Arctic and highlight the potential importance of glacially-fed lakes along the ice sheet behaving as hotspots of C and nutrient cycling as a result of these glacial meltwater additions.

Fluxes of particulate and solute nutrients are greater in years of higher melt (Hawkings et al

2015); thus, future climate warming may increase nutrient delivery to freshwater systems.

These trends and interactions will likely have important implications for Arctic C and nutrient cycling (Lyons et al 2008).

While significant work has been completed on glacier nutrient subsidies in streams and rivers (e.g. Brittain et al 2001; Scott et al 2010; Fellman et al 2015), glacier fed Arctic lakes represent potential hotspots of nutrient cycling (McClain et al 2003) and should be more thoroughly investigated. Compared to alpine regions, little research has been conducted on

Arctic lakes to understand how glacial meltwater affects their ecology. Given that meltwater delivery to Arctic lakes will increase due to climate warming, as well as a projected increase in the abundance of glacially-fed Arctic lakes due to glacier recession (Anderson et al 2017), the ecological responses of these systems should be a research priority.

Permafrost is sensitive to changes in air and ground temperatures (Hinkel et al 2003;

Zhang et al 2005; Romanovsky et al 2007) as well as rainfall (Kokelj et al 2015). Similar to

glaciers, permafrost thaw in Arctic areas is therefore predicted to significantly increase in response to climate warming (Jorgenson et al 2001; Jorgenson et al 2006; Lawrence and Slater

2005; Anisimov and Reneva 2006). Permafrost thaw is an important source of lake nutrient subsidies. Thaw events can be episodic, representing pulse disturbances to lakes, or they can be gradual and persistent, representing press disturbances (Vonk et al 2015). Moreover, permafrost thaw contributes to lake nutrient subsidies directly and indirectly (Figure 2.2). The active layer of soil that sits above the permafrost controls the tundra landscape’s hydrology and biogeochemistry (Hinzman et al 1991; Waelbroeck et al 1997; Zhang et al 2005; Schuur et al

2008; Frey and McClelland 2009). As such, changes to the active layer depth affect groundwater and nutrient delivery to freshwater ecosystems (Hobbie et al 1999; Zhang et al 2005; White et al 2007). However, a deepening of the active layer may have variable results; DOM inputs to freshwaters may increase (e.g., Carey 2003; Frey and Smith 2005; Holmes et al 2008; Spencer et al 2015) or decrease due to mineral sorption (Kawahigashi et al 2004) or groundwater flow paths becoming deeper (Striegl et al 2005; Walvoord and Streigl 2007; Harms and Jones 2012;

Harms et al 2016). The quality of DOM transported to lakes may change to more biolabile pools

(Abbot et al 2014). Research in Alaska and the Canadian high Arctic suggested that DIN and TP flux to freshwaters will increase with continued warming (Hobbie et al 1999; McClelland et al

2007; Frey and McClelland 2009; Dugan et al 2012; Khosh et al 2017). Permafrost thaw that results in altered lake catchment biogeochemical and hydrological processes is an indirect control of nutrient subsidies. In contrast to trends in current literature suggesting Arctic lakes may represent hotspots of nutrient and C cycling, a recent study in Alaska revealed that the transfer and within-lake respiration of terrestrial organic C was minimal (Bogard et al 2019).

WARMING ( ) ATMOSPHERE

ACTIVE LAYER

O S

PERMAFROST

PULSE INDIRECT INPUT OR (MODULATED BY PRESS WATERSHED) DISTURBANCE

DIRECT INPUT MINERAL LAYER (SLUMPING EVENT) EXPOSED? YES NO GROUNDWATER

SORPTION DEEP FLOWPATHS? HIGH SEDIMENTATION RATES OF PARTICULATES + SOLUTES?

YES NO YES NO

DOM DOM DOM CLARITY IONIC CONC.

DIN + TP DOM LAKE TDN + TP

Figure 2.2. Conceptual model of permafrost degradation and variable responses to this disturbance. Indirect lake chemistry effects of permafrost degradation are mediated through the active layer of the lake watershed, whereas direct effects describe a direct transfer of material into a lake through permafrost degradation, i.e. slumping and thermokarst events. DIN = dissolved inorganic nitrogen, DOM = dissolved organic material, ionic conc. = ionic concentration, TDN = total dissolved nitrogen, TP = total phosphorus.

This study was located in an arid area of interior Alaska that may be broadly representative of arid Arctic landscapes. Instead of allochthonous DOC mineralization, periods of lake heterotrophy consisted of autotrophically produced organic C mineralization. Together, these studies suggest that permafrost thaw will have spatially variable lake subsidy outcomes depending on thaw patterns, active layer depth, soil type, biolability of C and nutrient pools, and local hydrology.

Other thermokarst processes, such as permafrost slumping, involve a direct physical transfer of particulate sediments and solutes from terrestrial environments into Arctic lakes.

While thermokarst events can form lakes that occupy tundra depressions where permafrost degraded, established lakes within the Arctic landscape (originated from other processes, such as glacial scouring) can be influenced by thermokarst processes within their watersheds. This is a common occurrence in the Canadian and Russian Arctic. For instance, one in ten lakes in certain areas of the Canadian uplands were affected by permafrost slumping (Kokelj et al 2009).

In this study, ionic concentrations were higher in lakes affected by permafrost slumping compared to undisturbed lakes. Higher ionic concentrations in affected lakes were due to the exposure of hitherto frozen watershed sediments to weathering processes, and subsequent transfer of soluble materials to lakes by surface runoff. Permafrost thermokarst events in northern Alaska increased DOC and DIN concentrations in feature outflows, with potential ecological consequences for downstream freshwaters (Abbot et al 2015). In contrast, sedimentation of organic material associated with slump sediments lowered lake nutrient concentrations (total dissolved N and TP) and increased water clarity, with more recently slump-affected basins exhibiting greater clarity (Thompson et al 2008; Kokelj et al 2009;

Thompson et al 2012; Thienpont et al 2013). Increased water clarity following slumping may have driven increased macrophyte production in affected lakes (Mesquita et al 2010) as well as diatom species turnover with increased prevalence and diversity of periphytic diatoms

(Thienpont et al 2013). In the Canadian uplands, lakes disturbed by permafrost slumping events exhibited doubled macroinvertebrate abundance (Moquin et al 2014). However, increasing water clarity may expose pelagic organisms to harmful UV radiation (Thompson et al 2008).

Accelerating rates of thermokarst activity in the Arctic in response to climate warming (Lantz and Kokelj 2008; Segal et al 2016), coupled with lake responses to such disturbance events described above, suggests that permafrost slumping will be an important modulator of lake nutrient subsidies.

Climate warming across the Northern Hemisphere pan-Arctic region is driving a general reduction in snow accumulation since 1979, though there is much regional variability associated with this trend (Liston and Hiemstra 2011). Snow accumulation is an important driver of ecological patterns, such as plant, lichen, and moss species distributions (Niittynen and Luoto

2018), plant phenology, peak plant greenness (Zeng et al 2013), and maximum plant greenness

(Pedersen et al 2018). As such, changing snow accumulation patterns are an important impact on lake catchment processes. Accumulated snow converts to meltwater during spring thaw

(Jones 1999), contributing nutrients, minerals, and particulates into lake ecosystems which may increase the relative abundance of littoral epilithic algae (McGowan et al 2018) or phytoplankton communities of Arctic lakes (Rosén et al 2009).

Shortened lake iced-over periods and earlier spring ice-out is a general trend across the

Pan-Arctic in response to climate warming (Šmejkalová et al 2016), and ice-out phenology has

important implications for Arctic lake stratification (whereby the top epilmnion and bottom hypolimnion physically separate due to temperature-based water density gradients). Increased intensity and stability of lake thermal stratification is an anticipated response to climate warming (Winder and Sommer 2012; Vincent et al 2013; Anderson et al 2017), though this may be modulated by changing ice-out phenology (Warner et al 2018). Increased stratification is important for within-lake benthic-to-pelagic subsidies, because hypolimnetic oxygen can become depleted (anoxic) when lake thermal stratification inhibits the diffusion of oxygen from the epilimnion (Jankowski et al 2006). Hypolimnetic anoxia increases the chemical reduction potential and favors the release of P from lake sediments that is otherwise associated with Fe and aluminum (Al) during oxygenated conditions (Marsden 1989; Søndergaard 2003). In low elevation temperate lakes, sediment P release is related to algae blooms (Sommer 1985; French and Petticrew 2007; Wilhem and Adrian 2008). Benthic to pelagic P subsidies may be an important Arctic lake response to climate warming, but there has been little research of anoxic sediment P release conducted in Arctic lakes. Determining the mechanisms that control Arctic lake sediment P release should therefore be a research priority, so that we may better predict

Arctic lake responses to increased climate warming and seasonal variability.

Reduced river ice jamming in the McKenzie River Delta is a cryosphere change driven by

Arctic warming trends that decreases the duration of hydrological connectivity between rivers and lakes at higher elevations within the Delta (5.2 m above sea level; Lesack and Marsh 2007).

In contrast, the duration of river-to-lake connection during peak water levels has increased in lower elevation areas (2.4 m above sea level) within the Delta, due to sea level rise. It is likely that these trends will have ecological effects because McKenzie River Delta lakes with frequent

river-to-lake connectivity were associated with enhanced colored DOM, P concentrations, and light attenuation rates but lower ammonium and non-colored DOM concentrations averaged over the open-water season (Spears and Lesack 2006). Patterns of lake nutrient and light availability were likely responsible for higher abundance and production rates of bacterioplankton within less-frequently flooded lakes. These studies demonstrate that even small environmental gradients (i.e. a change from 2.4 to 5.2 m elevation) can alter the directional trends in lake subsidies and the environmental factors that influence them.

Many factors determine the nature, direction, and magnitude of ecological impacts in aquatic systems from cryosphere-derived subsidies. For instance, glacier meltwater effects will depend on magnitude of meltwater delivery, glacier bedrock material, and quality of nutrients.

Permafrost effects will depend on soil quality, depth and biogeochemical activity of the active layer, water flow paths, and slumping events. Lastly, recent snow pack trends are highly variable throughout the Arctic, which will affect photosynthetically active radiation (PAR) and spring snowmelt nutrient additions. Parsing out this variability will be critical to understand why certain trends are occurring in some areas of the Arctic but not others, or why opposite responses follow seemingly identical disturbance events in different areas (e.g. DOC concentrations following permafrost degradation, or nutrient concentrations following slumping events).

Atmospheric deposition

N deposition is considered to be more important to lakes in mid-latitude areas that have experienced higher anthropogenic development, compared to high-latitude Arctic areas which are more remote and have lower rates of N deposition (Figure 2.3a; Galloway et al 2004;

VEGETATED CATCHMENT LOWER DEPOSITION RATES

DEVELOPED SOILS SHALLOW SLOPE

MIXED LAYER

SHALLOW BASIN

A.) ARCTIC LAKE

SPARSE

VEGETATION HIGHER UNDERDEVELOPED DEPOSITION

ROCKY RATES

SOILS OPE STEEP SL

MIXED

DEEP BASIN

B.) ALPINE LAKE

Figure 2.3. Conceptual model comparing nitrogen (N) deposition, watershed features, and epilimnion mixed layer depth between Arctic (a) and alpine (b) lake ecosystems. In comparison to alpine lakes, Arctic lakes may have relatively more vegetated watersheds with gradual slopes to mediate N transfer from the catchment into the lake, yet both are sensitive to small amounts of N deposition. Thus, despite the difference in N deposition rates between the two areas, N deposition is an important ecological impact for both Arctic and alpine lakes. In addition to modulating catchment-to-lake N transfer, developed soils of low latitude Arctic lakes leach dissolved organic carbon (DOC) into lakes, which an important control of mixed layer depth and stability. See text and Figure 4 for ecological implications of mixed layer depth and stability.

Hobbs et al 2010). For example, Hobbs et al (2010) suggested that changes observed in diatom beta diversity over the 20th century in the Greenland Arctic are driven by rapid climate warming, versus that observed in western US alpine lakes, driven by high rates of atmospheric

reactive N (Nr) deposition. However, Holmgren et al (2010) observed that recent changes in preserved Arctic lake sediment diatom assemblages on Svalbard are coincident with enhanced sediment N derived from anthropogenic sources, suggesting a causal link between lake ecological change and N deposition. Diatom sediment records demonstrated that anthropogenic N deposition in the Canadian Arctic, along with climate warming, had driven lake conditions and diatom communities to no-analogue states in two nearby lakes Wolfe et al

2006). Community changes include increased abundances of Aulacoseira distans, Brachysira vitrea, Fragilaria spp., Cyclotella rossii, and Discostella stelligera, which indicated nutrient enrichment. Similarly, Holmgren et al (2010) demonstrated that four Svalbard lakes exhibited diatom community changes over the past 30 years coherent with increased rates of N deposition. A study conducted in sub-Arctic Sweden demonstrated that lakes shifted from N to

P limitation as a result of N deposition (Bergström et al 2008). As the study lakes were remote, high latitude, dilute and unproductive, they serve as good analogues for Arctic systems. Climate warming cannot be ruled out as a synergistic driver of these ecological changes because of the coherence of warming, N deposition, and diatom responses (Wolfe et al 2006; 2013). Taken together, however, these studies suggest a causal link between sensitive Arctic lake ecosystem change and high latitude N deposition.

Though Greenland is remote, it is downwind of North American industrial emission sources and is thus vulnerable to modifications of the Northern Hemisphere atmosphere (Laj et

- al 1992). Several ice cores demonstrated recent increases in NO3 concentrations in Greenland

(Mayewski et al 1990; Laj et al 1992; Fischer et al 1998) Svalbard (Simões and Zagorodnov 2001;

Isaksson et al 2003), and the Canadian Arctic (Goto-Azuma and Koerner 2001; Yalcin and Wake

- 2001). Averaged across sites, these records showed a 30% increase of NO3 flux since preindustrial times (from 2.5 to 3.2 μg cm-2 yr-1; Burkhardt et al 2006), with more than a doubling in some locations such as Greenland (Laj et al 1992). N isotope analysis (δ15N)

- indicated that NO3 in Greenland ice is derived from increasing anthropogenic N emissions since

1850 (Hastings et al 2009). Matching these cryosphere records, the presence of anthropogenic

N deposition has coherently increased in lake sediment cores across high latitudes of the

Northern Hemisphere since the beginning of the 20th century (Holtgrieve et al 2011). An outstanding question, however, is whether N deposition rates in the Arctic are high enough to be ecologically important.

- -1 -1 N deposition in the Arctic is typically low (0.2 - 0.5 kg NO3 ha yr ; Wolfe et al 2006) but

Arctic freshwaters are sensitive to small inputs of nutrients. Though rates of N deposition have not been as high in Arctic areas compared to mid-latitude ones, there is a clearly increasing trend that could be ecologically important. Anderson et al (2018) used lake sediment cores to demonstrate that coastal Greenland lakes have received increased amounts of anthropogenically sourced Nr since 1860, despite relatively low deposition rates in this area

(0.4 kg ha-1 yr-1). This trend was not recorded in inland lakes possibly because lower deposition rates (0.2 kg ha-1 yr-1), and higher in-lake TN pools and microbial cycling rates could further confound the signal of external anthropogenic inputs. Arens et al (2008) demonstrated by nutrient addition experiments that small amounts of N deposition (0.5 g m-1 yr-1) have nonlinear

impacts on terrestrial Arctic ecosystem structure and function, increasing vegetation cover, photosynthesis, and CO2 exchange. Similarly, Gordon et al (2001) observed that their lowest N addition treatment (10 kg ha-1 yr-1) increased physiologically active bryophyte shoots in Arctic heath, indicating the minimum ecologically important rate of atmospheric N deposition is likely below this value. These results suggest even small increases in rates of atmospheric deposition will be ecologically important.

The vulnerability of Arctic lakes to environmental change is due to spatially variable acid neutralizing capacity (ANC; Betts-Piper et al 2004; Wolfe et al 2006; Lepori and Keck 2012), sparse catchment vegetation, short growing seasons, shallow active soil layers (Birks et al

2004), and low in-lake nutrient concentrations and productivity. Evidence for Arctic freshwater sensitivity comes from ecological and paleoecological studies. For instance, Benstead et al

(2005) enriched a small Arctic stream on Alaska’s north slope with NH4-N and soluble reactive P

(SRP) to 6.4 μM and 0.45 μM, respectively. This addition increased algal biomass (measured as

Chlorophyll a; Chl a), fungal biomass, rates of litter breakdown, and macroinvertebrate abundance and biomass (Benstead et al 2005). In a long-term, whole-lake manipulation experiment in Arctic northern Alaska, half of Lake N-2 was subsidized with 131 mmol N m-2 yr-1 and 10.4 mmol P m-2 yr-1, approximately five times the normal loading rates of nearby Toolik

Lake, while the other half was partitioned off as a control (O’Brien et al 2005). Ecological responses included increased phytoplankton biomass and productivity, lower water clarity, and eventual hypolimnetic anoxia coupled with sediment P release.

Wind can distribute fine particulate matter, including pollen, bacteria, diatoms, and dust, long distances to Arctic and alpine areas (Burckle et al 1988; Gayley et al 1989;

Donarummo et al 2003; Rousseau et al 2008; Harper and McKay 2010; Weide et al 2017). Dust production, emission, and deposition rates can be high in glacierized Arctic landscapes, such as

Canada, Alaska, Greenland, and Iceland (see Bullard 2013), comparable to rates of lower latitude regions such as New South Wales, Australia (McTainsh and Lynch 1996) or Nevada and

California, USA (Reheis and Kihl 1995). Glacier dust production occurs when glacial flour (fine mineral particulates resulting from subglacial bedrock erosion) is transported to floodplains by subglacial meltwater. Winds then deflate the glacier flour, and can transport it across tundra, lakes, rivers, and ocean. Key controls of dust production include subglacial substrate erodibility, drainage system size and structure, ice mass size, and meltwater runoff rates (Bullard 2013).169

The amount of high-latitude dust contributions to the earth’s dust budget is currently estimated to be ~5% and is expected to increase with climate warming (Bullard et al 2016). The magnitude of dust events in southwestern Greenland, for example, has increased since the

1990s (calculated using the Dust Storm Index; Bullard and Mockford 2018). Most of the dust produced in high latitude areas remains in these regions. In western Greenland for example, dust transport from the Ice Sheet margin was mostly confined to the local landscape, with a small percentage reaching the North Atlantic Ocean, and very little being transported back onto the ice sheet itself (Bullard and Mockford 2018). This was due to the consistent direction of the katabatic winds blowing west from the ice sheet (van den Broeke et al 1994), as well as a lack of thermal uplift in the Arctic that is required for long-distance transport of dust-containing stable air parcels (Arnalds et al 2016; Zwaaftink et al 2016). Conversely, dust plumes derived from glaciofluvial deposits along the Gulf of Alaska (GoA) coastal region can be deposited into the ocean several hundred kilometers beyond the continental shelf, serving as an Fe subsidy for Fe-

limited marine systems (Crusius et al 2011). It is unknown to what extent glaciofluvial-derived dust is deposited into lake ecosystems in this region, but glacially-sourced dust accumulates in nearby terrestrial systems (Muhs et al 2013). A sediment core from an interior western Alaska lake isolated from glacial processes suggested dust deposition is related to aridity, lack of catchment vegetation, and windiness as opposed to glacial processes (Dorfman et al 2015).

The ecological effects of dust deposition on Arctic lakes are not well understood (Bullard and Mockford 2018). Because dust production and deflation in Greenland occurs throughout the entire year including the winter months (Bullard and Mockford 2018), accumulated dust on and within lake ice and snow can cause a pulse disturbance to lake ecosystems during snowmelt and ice off periods in spring, in contrast to a more consistent press disturbance of dust addition to lakes throughout the summer months (Bullard 2017). Glacier flour in southwest Greenland is P and Fe-rich (Bhatia et al 2013; Hawkings et al 2016), properties which carry over into deflated dust from this area (Fowler et al 2018), and coastal Alaska (Crusius et al

2011; Muhs et al 2013). Fowler et al (2018) proposed DOM adsorption to Fe-rich dust as a mechanism for recent decadal DOC concentration declines observed in a number of Greenland lakes, though microcosm incubation experiments did not support this hypothesis. To the extent that the P content of glacially derived dust is bioavailable (Burpee et al 2018), dust addition to lakes could stimulate lake bacterial production and metabolism (Carlsson and Caron 2001; del

Giorgio and Newell 2012), or in-lake sediment P cycling.

More evidence is needed to assess the ecological effects of atmospheric N and P deposition in Arctic lakes. This mode of ecosystem subsidization has not been paid due attention in the Arctic, likely because of the relatively low absolute values of N deposition rates

compared to low elevation temperate regions. However, studies which attribute recent changes of algal communities in Arctic lake sediment records to climate warming alone (Smol et al 2005) may be missing the important contributions of anthropogenically sourced atmospheric deposition. High latitude dust production was only recently assessed as a significant contributor to the global dust budget (see Bullard 2013), positioning Arctic dust deposition as an intriguing topic of research for terrestrial and aquatic ecologists and biogeochemists. Further, climate

- warming will likely influence Nr and dust deposition alike. Greenland ice core NO3 , a proxy for atmospheric Nr deposition, was correlated with the North Atlantic Oscillation, suggesting greater deposition of anthropogenic Nr occurs when the NAO is in its positive phase (Burkhardt et al 2006). A climate change impact model on atmospheric N deposition suggested that increased precipitation will enhance N deposition rates in northern Europe (Hole and Enghardt

2008). Though changes in N deposition will be gradual in most areas, certain areas, including western Norway, may experience a 40% increase in N deposition by 2100 due to projected increases in precipitation. Warming could also release previously deposited and accumulated Nr from perennial snow packs and glaciers in the form of meltwater. Lastly, warming will likely increase dust production in high latitude areas because increased meltwater will deposit greater amounts of glacier flour into floodplain areas (Bullard 2013).

Animal vectors

Animals are important vectors for nutrient transfer between ecosystem types, and this mode of transfer between aquatic and terrestrial systems has received increased attention over the past decade. Unlike other transfers that have been discussed thus far, animals represent a bi-directional flow of energy and nutrients, whereby inputs of terrestrial C and nutrients to

lakes are cycled back to terrestrial environments by boundary-crossing organisms (Scharnweber et al 2014). The C that is delivered to terrestrial systems from emerging mosquitos, for instance, is a mix of terrestrial and aquatic material, and the terrestrial:aquatic C ratio depends on the magnitude of terrestrial inputs into the aquatic ecosystem, as well as light intensity (Kraus and

Vonesh 2012). Vander Zanden and Gratton (2011) highlighted that insect emergence and subsidies to terrestrial systems increased as a function of lake size, while rates of terrestrial inputs of particulate organic C to lakes decreased as a function of their size. Thus, reciprocal transfers between lakes and terrestrial habitats are modified by environmental variables. A detailed understanding of variables that influence reciprocal transfers between lakes and watersheds is an important research priority, because it will allow for better assessment of how reciprocal subsidies will be affected by different environmental changes.

In the Arctic, aquatic-terrestrial transfers often take place as abrupt seasonal pulses, and can be limited by distance from the lake, with an exponential decrease of insect infall with distance from lake (Gratton et al 2008; Gratton and Vander Zanden 2009). However, observed transfer distance was greater from lakes (150 m) than from streams (50 m; Petersen et al 1999;

Power et al 2004). For instance, midges (Diptera: Chironomidae) are an important nutrient subsidy from lakes to terrestrial systems. In Iceland, annual midge inputs from lakes were as high as 1200-2500 kg ha-1 yr-1, though these inputs decreased logarithmically from shore

(Gratton et al 2008). Peak rates of midge infall occured in August and can reach rates over 1500 kg ha-1 d-1 at a distance of 5 m from shore (median rate 290 kg ha-1 d-1; Gardarsson et al 2000).

Midge N content is 9.2%, and so this seasonal transfer was nutrient-rich, as high as 230 kg N ha-

1 yr-1 (Gratton et al 2008). δ13C and δ15N isotope analysis demonstrated this subsidy is

important for terrestrial arthropods (e.g. spiders), which consume midges directly or indirectly as a food source. Aquatic insects are a high-quality food source, in part because of their polyunsaturated fatty acid content. Martin-Creuzburg et al (2017) demonstrated that the terrestrial subsidies of polyunsaturated fatty acids from aquatic insect emergence was significant (mean rate was 150 mg m-2 yr-1 within 100 m of shore). Thus, in areas with high midge emergence rates, midges maintain higher abundances of terrestrial arthropods compared to areas with low midge emergence rates. In Iceland and northern Sweden, subsidized terrestrial arthropods include detritivore and herbivore species as well as predators

(Jonsson and Wardle 2009; Hoekman et al 2011; 2012; Dreyer et al 2012). An Icelandic study demonstrated that willow (Salix phylicifolia) located at high-midge emergence lakes contained

8-11% higher N content than those located at low-midge emergence lakes, and herbivorous caterpillars (Hydriomena furcata) were 72% heavier at the high-midge emergence sites

(Bultman et al 2014).

Arctic Chironomid midge emergence is synchronous, with the bulk of the community emerging from a single pond within 4 weeks, and a single species emerging within one week

(Butler 1980). In response to climate warming, Alaskan midge emergence has advanced by one week since the 1970s (Braegelman 2015). As midge adults only live for a few days, such shifts in aquatic insect phenology may have repercussions for Arctic predators, such as avian consumers. For example, in UK lakes in the South Pennines, climate warming shifted the dates of egg laying by an Arctic wader species (golden plover, Pluvialis apricaria) and tipulid (cranefly) emergence earlier, but at different rates (Pearce-Higgins et al 2005). Thus, climate predictions for 2070-2090 suggest that first-laying dates for golden plover will advance 25 days, but tipulid

emergence only 12 days. As emerging tipulids are an important prey resource for golden plover chicks, this phenological mismatch may reduce breeding success of this species at this location by 11%. Because Arctic lake midge emergence is synchronous, short in duration, and is an important resource for Arctic terrestrial and avian predators, phenology of Arctic lake midges and their predators require more investigation in order to determine if temporal mismatches will occur, and what ecological impact they will have.

With climate warming and other anthropogenic environmental changes, waterfowl habitat range and population sizes are increasing. Global goose population, for example, nearly doubled from 1996 (12.5 million; Madsen et al 1996) to 2006 (21.4 million; Wetlands

International 2006). Dense nesting colonies of waterfowl have the potential to transfer significant amounts of terrestrially derived nutrients (N and P) into Arctic lake ecosystems in the form of feces (Hahn et al 2007; Côté et al 2010). In Svalbard, goose guano has caused lake and pond eutrophication, increasing P concentrations four-fold since the 1960s (Hessen et al 2017).

A small-scale incubation experiment demonstrated that algal biomass increases from goose guano-derived nutrients (Tugend 2017). Sediment cores from Canadian Arctic ponds demonstrated strong relationships between algal biomass and chironomid heads in response to

N derived from seabird guano (Michelutti et al 2009; Griffiths et al 2010). Paleolimnological records demonstrated an increasing colony population enriches lake sediment δ15N, indicating a marine to lake linkage. Thus, coastal Arctic lake ecosystems may experience subsidies from different sources compared to inland lakes.

Particular areas of the Arctic, such as Iceland, have received attention regarding lake-to- land transfers of nutrients in the form of emerging aquatic insects, but there is a lack of

research in other Arctic areas where lake-to-land nutrient transfers would be seemingly important. For instance, some Greenland lakes experience high densities of emerging chironomid adults during peak summer, which have not been characterized in terms of nutrient quality and quantity. Further, with high rates of climate warming in western Greenland, outbreaks of moth larvae (Eurois occulta and Gynaephora groenlandica) have become more common (Lund et al 2017). These larvae commonly fall into lakes and streams (personal observation). Though larval outbreaks in western Greenland have been linked to terrestrial defoliation (Post and Pedersen 2008) and changes in C-burial rates (Lund et al 2017), the effect on aquatic ecosystems is unknown.

Alpine lake subsidies

Cryosphere

Alpine areas in North America have numerous glaciers and rock glaciers. 1500 glacier and 10000 rock glacier features have been identified in the western US alone (Fegel et al 2016).

Because of their abundance and location within the continental US and Europe, studies published to date that investigate the effects of nutrient-rich glacial meltwater on lake ecological structure and function have largely focused on alpine rather than Arctic lakes (e.g.,

Saros et al 2010; Slemmons and Saros 2012; Fegel et al 2016; Williams et al 2016). Climate warming has increased glacier melt to mountain surface waters in the Pacific Northwest region of the US over the past 70 years (Granshaw and Fountain 2006). Enhanced glacial melting has also increased glacial lake abundance and area across the Central and Patagonian Andes

(Wilson et al 2018). Glacier recession is widespread and accelerating throughout the Tibetan

Plateau (Yao et al 2007; Kehrwald et al 2008), and is cited as one of the factors leading to

increased lake levels across the region (Zhang et al 2011; Liu et al 2009; Zhang et al 2013). In addition to warming trends, climate-mediated dust deposition can alter glacier snow and ice albedo leading to greater glaciofluvial transport of glacier meltwater to high elevation lakes in southwestern Tibet (Conroy et al 2013). Glacier meltwater subsidizes systems with essential elements such as Ca, K, Mg, nutrients such as Si, Fe, N and DOM, and organisms such as microbes, though the characteristics of these subsidies differ among mountain ranges and between glacier types (rock glaciers vs glaciers; Birks et al 2004; Conroy et al 2013). In contrast to Arctic glacier meltwater, which typically has high P but low to medium N content, alpine

- glacier meltwater can have elevated NO3 , with subsidized lakes containing concentrations as

-1 -1 - high as 236 µg NO3 -N L (Saros et al 2010). Thus, NO3 -N enriched glacially-fed lakes in the

Beartooth Mountains of Wyoming and Montana exhibit lower diatom species richness (Saros et al 2010) and higher phytoplankton primary productivity rates and biomass compared to nearby snow and groundwater-fed lakes (Slemmons and Saros 2012). Although glacially-fed lakes in both Arctic and alpine areas are often turbid with suspended glacial flour, the target glacially- fed lakes from the Beartooth Mountains in these studies were clear due to either minimal subglacial weathering or upstream entrapment (Saros et al 2010). These studies are useful because the clarity of the glacially-fed lakes makes them similar to snow and groundwater-fed lakes. Thus, the ecological effects of these particular glacially-sourced N subsidies could provide insights into remote cold, oligotrophic lakes receiving other types of N subsidies (such as atmospheric deposition, or increased goose guano).

As in Arctic lakes (Burpee et al 2018), glacial transport of suspended mineral particulates that constitute milky-colored glacial flour can alter the transparency and light penetration of

alpine lakes, giving them a characteristic grey to turquoise hue (Sommaruga and Kandolf 2014).

Turbidity can reach high levels in alpine lakes (644 NTU recorded in the Swiss Alps; Bonalumi et al 2011; unit conversion by Sommaruga and Kandolf 2014). Sediment transport is increased during glacial recession as a function of meltwater flux and is therefore affected by climate warming (Leonard and Reasoner 1999; Vinebrooke et al 2010). Because of their small size (<32

μm; Brown et al 1996) and sharp edges, glacial flour particulates can have direct negative impacts on interception- or filter-feeding aquatic organisms (such as Daphnia, and heterotrophic nanoflagellates), that may ingest them (Koenings et al 1990; Sommaruga and

Kandolf 2014). Other physical consequences of suspended glacial flour in alpine lakes includes density underflows (also a function of cold meltwater temperature; Gilbert and Crookshanks

2009), a reduced euphotic zone and high UV light attenuation, both of which have consequences for ecosystem primary production rates (Sommaruga 2015), phytoplankton and zooplankton stoichiometry (Sterner et al 1997; Laspoumaderes et al 2013), and plankton survival (Sommaruga 2015). Further, surface temperature of glacially-fed alpine lakes can be surprisingly warm (late August mean surface temperature from 3 glacially fed lakes was 16.0 °C, calculated from Peter and Sommaruga 2017), similar to that of a nearby snow and groundwater-fed, clear lake (15.3 °C; Peter and Sommaruga 2017). Though mean lake water temperatures can be cooler in glacially-fed lakes compared to snow and groundwater-fed ones

(a 1.1 °C difference was demonstrated in Alaska, for example; Koenings et al 1990), the warming of the surface by attenuated solar radiation leads to discontinuous polymictic thermal stratification, a situation where thermal stratification of these lakes can undergo fast and

abrupt mixing throughout the summer, caused by cold night-time temperatures or precipitation events (Peter and Sommaruga 2017).

While Arctic glacier meltwater contains P that is associated with weathered minerogenic

- subglacial flour (Burpee et al 2018) the source of NO3 that is concentrated in alpine glacier

- meltwater is unclear (Williams et al 2016). Determining the extent to which meltwater NO3 is derived from anthropogenic sources, minerogenic material, or subglacial microbial processing of organic material may inform conservation or management decisions. Further, alpine glaciers are projected to disappear in the American Rocky Mountains by 2030 (Hall and Fagre 2003), and certain areas of the European Alps by 2050 (Paul et al 2004). It is currently an outstanding question what the ecological effects of glacier disappearance will be on glacially-fed lakes, and whether legacy effects of previously glaciated catchments will persist following deglaciation.

For instance, given the abundance of glaciers in the western United States (Fegel et al 2016), it will be important to determine how C and nutrient cycling may change as a lake goes from turbid and glacially-fed to clear and fed by groundwater sources.

Atmospheric deposition

Like Arctic lakes, alpine lakes are ecologically sensitive to atmospheric deposition due to their slow weathering bedrock and spatially variable ANC, absent or poorly developed catchment soils, sparse catchment vegetation cover, and low nutrient water chemistry (Psenner et al 1999; Nanus et al 2005; Clow et al 2010). In contrast to Arctic lakes, high elevation exposes alpine lakes to greater rates of atmospheric deposition (Williams and Tonnessen 2000). In mountain regions of the US, enhanced N deposition results from fossil fuel and agricultural emissions (Howarth et al 2002; Galloway et al 2004; Hundey et al 2014).

Atmospheric N deposition can change the nutrient limitation status of lake ecosystems.

Many North American alpine lakes were historically N-limited (Elser et al 1990; Nydick et al

2003; Fenn et al 2003; Saros et al 2010). However, in both North America and Europe, atmospheric N deposition has shifted lake nutrient limitation patterns from N-limitation to N and P co-limitation or P-limitation (Bergström and Jansson 2006, Elser et al 2009). Ecological responses to N deposition can be rapid, because primary producers with high reproductive rates, such as phytoplankton, are sensitive to nutrient changes (Baron et al 2011; Pardo et al

2011). Such responses include increased primary productivity (Goldman 1988; Jassby et al 1994;

Nydick 2004), increased phytoplankton biomass (Sickman et al 2003; Nydick 2004) decreased lake clarity (Goldman 1988; Jassby et al 1994), and directional algal community changes caused by increased dominance by opportunistic diatom species such as Fragilaria crotonensis,

Asterionella formosa, and Discostella stelligera (Baron et al 1986; Goldman 1988; Jassby et al

1994; Wolfe et al 2001; Wolfe et al 2003; Spaulding et al 2015; Williams et al 2016), that respond rapidly to nutrient enrichment (Saros et al 2005; Köster and Pienitz 2006; Saros et al

2010; 2012).

In addition to N deposition, alpine lakes are ecologically sensitive to P deposition. In mountain areas, N deposition is associated with both wet deposition (DIN associated with precipitation or fog droplets; Zannetti 1990; Burns 2003) and dry deposition (consisting of nitric acid and particulate ammonium and nitrate; Butler et al 2005). P deposition is associated with dry deposition (associated with particulate matter, or dust; Morales-Baquero et al 2006). Dust is sourced from anthropogenic activities and biomass burning (Mahowald et al 2008), and dust deposition can be related to climate. For instance, dust storms and strong winds have increased

in frequency in the Tibetan plateau over the past 44 years, observable via grain-size analysis in a high-elevation lake sediment record (Lake Sugan, 3,000 m above sea level; Qiang et al 2007). A longer lake sediment record spanning 1,600 years from Kusai Lake in the central Tibetan

Plateau suggests that dust deposition is positively related to summer temperature (Liu et al

2014). The link between dust and climate was also demonstrated in the southwest Tibetan

Plateau by Conroy et al (2006) by comparing western Tibet temperature (Bao et al 2003) and

Dasuopu Ice Core (Thompson et al 2000) dust deposition records over the past 1000 years. Dust deposition has increased 400% in some areas of the western US over the past twenty years, with mountainous areas being particularly impacted (Brahney et al 2013). Brahney et al (2015) compared paleoecological records from two Wind River Range lakes in Wyoming, North and

Lonesome, which were compared as dust-affected and control sites, respectively. The authors demonstrated a tripling in sediment P content, a tenfold increase in diatom production, increasing cyanobacteria, and diatom community changes in North Lake. These changes correspond to increasing dust flux starting around 1940. Brahney et al (2014) demonstrated that lakes of the Wind River Range were subsidized by dust deposition rates as high as 276 μg P m2 day-1. The source of this dust was from a local valley, and lakes closer to the source were more affected by P subsidies. Lake ecological effects included enhanced water and sediment P concentrations, greater phytoplankton and zooplankton biomass, and dominance of diatom communities by high-nutrient species, such as Asterionella formosa.

Dust may also originate from distant sources. In the Mediterranean region of Europe, for example, dust deposition was dominated by material from the Sahara-Sahel desert region in northern Africa, with maximum inputs occurring during spring and summer (Moulin et al 1997).

SRP contained in this dust and deposited in the SW Mediterranean stimulated bacterial abundance and decreased phytoplankton species diversity in an alpine lake in the Spanish

Sierra Nevada Mountains (Pulido-Villena et al 2008). Reche et al (2009) observed positive effects of dust inputs on bacterial abundance in two alpine lakes of the Sierra Nevada, Spain, though no effect on bacterial community composition was observed. In addition to P, Saharan dust can deliver chromophoric, aromatic, and fluorescent DOM to European alpine lakes

(Mladenov et al 2009). Saharan dust input could therefore be partially responsible for the relatively higher concentrations of DOC and CDOM in lakes of Sierra Nevada versus those of the

Pyrenees and Alps that do not receive this input (Mladenov et al 2009). Saharan dust also transfered viable bacteria (Gammaproteobacteria) that were deposited and grew in alpine lake water in the Austrian Alps (Peter et al 2014). Similarly, Reche et al (2009) observed growth of

Gammaproteobacteria in dust-inoculated water of the oligotrophic Quéntar Reservoir in Spain.

Together, these studies suggest that atmospheric dust deposition can indirectly affect lake microbial community assemblages by influencing environmental selection pressures (such as P availability or DOM quality) or microbial assemblages can be directly affected by the introduction of exogenous species.

Determining what factors determine lake ecological sensitivity will be a critical research priority that will be able to inform lake management schemes. For instance, critical loads of N deposition (thresholds below which effects of deposition cannot be detected; Burns et al 2008) are typically low (1.0-1.5 kg N ha-1 yr-1) for high elevation lake systems in the Sierras and Rocky

Mountains of western United States (Saros 2010; Baron et al 2011). However, nutrient deposition rates alone are not a good predictor of lake ecological sensitivity and response to

enrichment. For instance, elevation (positive association), talus cover (positive association), unvegetated catchment area (positive association), alluvium (negative association), and riparian areas (negative association) were important factors in determining surface water susceptibility

(measured from stream and lake outlets) to nutrient enrichment following atmospheric N deposition in Yosemite National Park, USA (Clow et al 2010). There are other factors, such as lake depth, clarity, and nutrient limitation that may be important in determining lake sensitivity. Analyses that include these factors across multiple spatial scales will be important in explaining the local and regional variation that we observe in lake responses to atmospheric deposition.

Animal vectors

Like Arctic lakes, high-elevation lakes are connected to their terrestrial environments via reciprocal flows of material. While inputs to alpine lake ecosystems are often passive hydrological or atmospheric processes determined by slope gradient and gravity, lake-to- terrestrial solid and liquid subsidies rely on animal vectors. For alpine lake ecosystems, the relative importance of aquatic- or terrestrial-derived subsidies to its neighboring environment can depend on lake elevation. For instance, based on research in the Sierra Nevada and

Klamath mountain ranges of the western United States, Piovia-Scott et al (2016) hypothesized that terrestrial C inputs were important controls for in-lake processes at lower elevations, and in contrast, in-lake processes were relatively more important for terrestrial consumers at higher elevation. Though lower-elevation lakes had higher absolute rates of lake-to-land C transfer than high-elevation lakes in the form of emergent insects and amphibians, the difference between lake C output vs. lake C input was smaller in low elevation lakes. This is due to the fact

that high elevation catchments deliver significantly less terrestrial C to lake ecosystems.

Further, because relatively little C was terrestrially derived in high elevation lakes, their ecosystems were dependent on inorganic C fixation, making them net C sinks. Lower elevation lake ecosystems, with greater amounts of terrestrial organic C, were instead CO2 sources.

Although on average lake-to-land animal C transfer decreased with lake elevation across the

Klamath mountain landscape, the density of terrestrial insect infall was higher in high-elevation lakes close to shore compared to lower elevation lakes, creating a hotspot of C and nutrient deposition (Piovia-Scott et al 2016).

Many species of trout have been introduced into naturally fishless alpine lakes for sport fishing (Nelson and Paetz 1992; Knapp 1996; Miró and Ventura 2013; Tiberti et al 2017). The presence of predatory fish can significantly reduce the quantity of lake-to-land animal C and nutrient transfer due to predation on insect and amphibian larvae (Piovia-Scott et al 2016). For instance, trout have reduced insect emergence in the Sierra Nevada (U.S.), such that fishless lakes can have 20 times more emergent insect biomass (Finlay and Vredendburg 2007).

Introduced trout have reduced amphibian abundance in mountain lakes by predation (Knapp et al 2005; Matthews et al 2001; Welsh et al 2006), as has the fungal pathogen Batrachochytrium dendrobatidis (Rachowicz et al 2006). Greig et al (2012) demonstrated that predatory fish reduce not only insect emergence directly by predation, but also reduce decomposition of terrestrial organic detritus indirectly, where increased phytoplankton blooms influenced by pelagic trophic cascades increase sedimentation rates and establish anoxic benthic conditions.

Thus, in addition to reducing lake-to-land subsidies, predatory fish may also modulate land-to- lake C transfer by reducing the amount of terrestrial C that is incorporated into aquatic

consumer pathways. Such food web interactions that decouple cross-system subsidies can have important effects on reciprocal communities; in Sierra Nevada (U.S.) lakes, introduced trout outcompete Rosey-finches (Leucosticte tephrocotis dawsoni) for mayfly larvae, so that Rosey- finches are 6 times more abundant at fishless lakes (Epanchin et al 2010). Introduced trout also outcompete and consume native frog species (Rana muscosa) such that these frog species are

10 times more abundant in fishless lakes of the Sierra Nevada (U.S.; Findlay and Vredenburg

2007).

Quality of a subsidy and its relative availability in the receiving environment is also an important factor when considering lake-to-land nutrient subsidies (Marczak et al 2007;

Marcarelli et al 2011; Fritz et al 2019). For instance, long-chain polyunsaturated fatty acids (LC-

PUFAs) are critical nutrients for organismal cell membranes, structure, development, function, and signaling (Brett and Muller-Navarra 1997; Arts and Kohler 2009; Parrish 2009). LC-PUFAs occur disproportionally in aquatic environments because they are synthesized by algae, but not by terrestrial plants (Brett and Muller-Navarra, 1997; Gladyshev et al 2013; Hixson et al 2015), and they bioaccumulate rapidly (twice the rate of bulk C) in higher trophic levels because of their nutritional status (Gladyshev et al 2011b). Thus, mobile or metamorphic organisms such as frogs, salamanders, aquatic insects, and birds (Gladyshev et al 2009; 2010; 2011a; 2011b;

Yermokhin et al 2018; Fritz et al 2019) represent an important terrestrial nutrient subsidy (LC-

PUFA) that is disproportionate to the mass bulk of the transferred material.

The ecological effects of climate warming on lake-to-land transfers in alpine areas requires more attention. For instance, it is unknown what effects climate warming will have on the timing, magnitude, species composition, and nutrient quality of lake-to-land transfers

across Northern hemisphere alpine areas. Further, the environmental factors that determine average lake respiration vs primary production along an elevation gradient (Piovia-Scott et al

2016) may become less distinct, so that high elevation lakes may become net heterotrophic, and lake-to-land nutrient contributions become relatively less important.

Conclusions and synthesis

The environmental forces acting upon Arctic and alpine lakes are similar: warming rates in the Arctic are greater than the Northern Hemisphere average (Serreze et al 2009), and many alpine regions exhibit accelerated warming as well (Pepin et al 2015). Atmospheric deposition is subsidizing both areas with N and P (Bergström and Jansson 2006; Elser et al 2009; Holtgrieve et al 2011; Brahney et al 2013; Bullard et al 2013), though deposition rates and sources of deposition differ between Arctic and alpine areas. Despite high spatial variability, annual precipitation is on average declining in Arctic areas (Liston and Hiemstra 2011), resulting in reduced annual snowpack and contributing to glacier shrinkage. Though these environmental factors are changing Arctic and alpine lakes nutrient subsidies (both in terms of material inputs and outputs), these changes are not entirely parallel (Figure 2.4). For instance, both Arctic and alpine landscapes contain an abundance of glacially-fed lakes, but nutrient subsidies delivered into lake ecosystems via glacier meltwater vary regionally depending on bedrock and atmospheric deposition patterns (e.g. Saros et al 2010; Burpee et al 2018).

Ecosystem subsidies across Arctic and alpine landscapes will lead to changes in lake ecosystem C and nutrient cycling depending on a suite of lake, landscape, and climatological variables. For instance, DOC groundwater exports are variable in Arctic landscapes depending on soil and groundwater flow path properties (Striegl et al 2005; Walvoord and Striegl 2007).

Figure 2.4. A general comparison of lake nutrient subsidies between Arctic (a) and alpine (b) lakes, and their ecological outcomes. Dissolved organic carbon (DOC) is an important factor for Arctic lakes with more vegetated watersheds. Greater DOC transfer to Arctic lakes may influence thermal stratification to create anoxic environments and may push ecosystems towards net heterotrophy. Conversely, the increase in nutrient additions to alpine lakes without the transfer of DOC from the catchment may push these lakes towards net autotrophy. N = nitrogen, P = phosphorus, DIN = dissolved inorganic N, hypo = hypolimnion, TP = total P.

Due to expected trajectories of climatological and environmental changes, Arctic lakes may become greater sources of CO2 and CH4 as respiration and methanogenesis increase from enhanced nutrient and C availability and stronger thermal stratification. Lower-latitude Arctic lakes generally have more vegetated watersheds and developed soils than alpine lakes, thus

DOC will likely be important in Arctic lakes compared to alpine ones. DOC is an important factor in determining thermal stratification of Arctic lakes to the extent that it contributes to PAR absorption (Saros et al 2016). Thus, lakes experiencing greater DOC subsidies from their surrounding catchments may exhibit enhanced thermal stratification (Rosén et al 2009; Figure

2.3). In turn, stronger thermal stratification could contribute to hypolimnetic anoxia resulting in greater in-lake P cycling as it is released from the sediments (Nürnberg 1984). Further, larger

DOC contributions could increase lake respiration, anoxic hypolimnetic methanogenesis, and denitrification rates (Kling et al 1991; Molot and Dillon 1993; Tranvik et al 2009; Northington and Saros 2016). In contrast, alpine lakes can be expected to increase rates of primary production and C fixation, because declining seasonal snowpack coupled with steep alpine catchments with sparse vegetation means that little DOC is transferred into alpine lake ecosystems (Piovia-Scott 2016). With relatively scarce organic C, increased nutrient loading from glaciers, snowmelt, and atmospheric deposition will likely increase rates of primary production in alpine lakes, establishing them as C sinks (Piovia-Scott 2016).

In the Arctic, increased waterfowl abundance enhances the transfer of terrestrial- and marine-derived nutrients to lakes and ponds (Hahn et al 2007; Michelutti et al 2009; Côté et al

2010; Griffiths et al 2010), with the potential to cause eutrophication of these remote ecosystems. Lake-to-land transfer of C and nutrients is dominated by insect emergence in both

alpine and Arctic lake ecosystems. Increased insect emergence that tracks warming and nutrient availability could lead to altered nutrient cycling in lake catchments. For instance, in lake catchments with high rates of emergent insect infall, nutrient content of plants and size and abundance of herbivorous caterpillars increased (Bultman et al 2014). Similarly, herbivorous, detritivorous, and predatory arthropod abundance increases (Dreyer et al 2012).

Climate warming and increased nutrient loading to Arctic and alpine lakes can be expected to change the timing, magnitude, and species composition of emergence (Greig et al 2012; Piovia-

Scott et al 2016). Phenological mismatches may occur in both Arctic and alpine landscapes between emergent insects and their terrestrial and avian predators but more Arctic- and alpine- specific research is needed to confirm and characterize these mismatches, as remote lakes are presently underrepresented in this relatively new research area.

Cross-ecosystem subsidies are mass (M) inputs, and as Leavitt et al (2009) describe, lake ecosystem responses to M inputs are variable. This variability comes from differences in lake catchments, ecological communities, physical structure, and water and sediment chemistry that interact with M inputs. Additionally, differences in lake energy budgets between Arctic and alpine areas will likely interact with subsidies in different ways. Many alpine lakes are located at low latitudes with greater annual solar irradiance than those at high latitudes. For instance, among 59 lakes surveyed across the Tibetan Plateau, latitude was positively related to ice cover duration, reflecting lower temperatures and decreased irradiance toward the north (Kropáček et al 2013). Thus, researchers can expect variation in responses between Arctic and alpine lakes to similar drivers, as well as variation of responses amongst lake groups in either region. From a lake management and conservation standpoint, an important outstanding research priority

involves accounting for this variation by assessing what factors determine lake sensitivity and responses to large-scale drivers, such as warming, permafrost thaw, or atmospheric deposition.

In closing, we have summarized cross-ecosystem subsidies to Arctic and alpine lake ecosystems from three major sources: the cryosphere, atmosphere, and animals. Though aquatic-terrestrial links were not addressed with their own explicit category, terrestrial-aquatic interactions were considered across all three types of nutrient subsidy classification. It is clear from this review that there are many questions to be addressed before we have an in-depth understanding of how nutrient subsidies are controlled by environmental factors, and how they in turn control the ecological behavior of cold, dilute, oligotrophic lakes. Research that assesses these factors will be crucial not only for Arctic and alpine ecologists and biogeochemists, but the broader ecological community. This is because ecosystem subsidies provide an important framework to understand and assess the types of environmental and ecosystem changes that will continue to occur in the 21st century. Our goal in this review has been not only to summarize recent ecosystem subsidy research in Arctic and alpine areas, but to promote remote lakes as exemplary systems with much to contribute to ecosystem subsidy research.

Remote lakes in rapidly changing areas have much insight to provide concerning the mechanisms, processes, and ecological impacts of cross-ecosystem nutrient subsidies that occur in other ecosystems, such as low-elevation temperate lakes and streams.

CHAPTER 3

ASSESSING ECOLOGICAL EFFECTS OF GLACIAL MELTWATER ON LAKES FED BY THE GREENLAND

ICE SHEET: THE ROLE OF NUTRIENT SUBSIDIES AND TURBIDITY

Abstract

Meltwater discharge from the Greenland Ice Sheet (GrIS) exports sediment, solutes, total phosphorus (TP), dissolved inorganic nitrogen (DIN), and other macro- and micronutrients to associated aquatic ecosystems. It remains unclear how this meltwater affects the ecology of glacially-fed (GF) lakes. We assessed a suite of physical, chemical, and biological features of four

GF lakes, and compared them to those of four nearby snow and groundwater fed (SF) lakes. We found that TP concentrations were six times higher in GF compared to SF lakes, but microbial extracellular enzyme activities and aluminum, iron, and phosphorus sediment fractions suggested that much of this TP in GF lakes is likely not biologically available. Turbidity was 15 times higher in GF lakes, and DIN was twice as high than in SF lakes, but these nitrogen differences were not significant. While diatom species richness did not significantly differ between lake types, GF lakes had higher water column chlorophyll a (Chl a). Diatom species distributions across all lakes were strongly associated with turbidity, TP and dissolved organic carbon (DOC). While cosmopolitan diatom taxa such as Discostella stelligera were found in both lake types, diatom communities differed across lake types. For instance, Fragilaria and

Psammothidium species dominated GF lakes, while Achnanthes species and Lindavia

49ormosa49 were dominant in SF lakes. In addition to turbidity, the moderate amounts of DIN in GF lakes may play an important role in shaping diatom communities. This is supported by the high abundance in GF lakes of taxa such as Fragilaria tenera and D. stelligera, which reflect N

enrichment in some lakes. Our results demonstrate how GrIS meltwaters alter the ecology of

Arctic lakes, and contribute to the growing body of literature that reveals spatial variability in the effects of glacial meltwaters on lake ecosystems.

Introduction

Glacial meltwater contains particulates, such as minerogenic glacial flour, and other solutes. Suspended glacier flour particles are as small as clay or fine silt. They cause high turbidity which give many glacially fed aquatic ecosystems a distinct milky brown or gray hue, and can limit light penetration into water, decreasing the amount of photosynthetically active radiation (PAR) and ultraviolet radiation (UVR) to which plankton are exposed (Sommaruga

2015). Glacial meltwater is also a source of nutrients such as phosphorus (P), derived from subglacial weathering of P-containing bedrock (Hodson 2004; Hawkings et al 2016), and nitrogen (N; Saros et al 2010), concentrated in glacier ice from the atmosphere (Daly and Wania

2005) or sourced from subglacial microbial weathering processes (Williams et al 2007).

Climate warming has increased glacier meltwater delivery into lakes and streams of alpine regions for the past 70 years (Granshaw and Fountain 2006). Glacially fed (GF) lakes form at the termini of glaciers and at the margins of ice caps and ice sheets as a function of meltwater production, ice dynamics, and environmental setting, such as topography and moraine location. GF lakes are predicted to increase in abundance and size around the world as a result of continued deglaciation (Carrivick and Tweed 2013). Because glacial meltwater contains high nutrient concentrations, GF lakes have the potential to be unique ecosystems and hot spots of nutrient cycling (McClain et al 2003).

Among the studies that have been published, the effects of nutrient-rich glacial meltwater on the ecological structure and function of receiving lakes have been largely investigated in GF alpine lakes (Saros et al 2010; Slemmons & Saros 2012; Fegel et al 2016;

Williams et al 2016). In the North American Rocky Mountains, nitrate-enriched glacial meltwater from small mountain glaciers has shifted lakes from N to P limitation (Saros et al

2010), increased lake primary production (Slemmons and Saros 2012), and decreased diatom species richness (Saros et al 2010; Slemmons and Saros 2012). There is currently a gap in the literature that evaluates how glacial meltwater affects Arctic GF lakes. Comparisons of lake features across GF and SF lake types in the Arctic will help to clarify meltwater-induced changes in lake features and subsequent community shifts.

The Arctic region of western Greenland presents a unique opportunity to investigate the ecological consequences of meltwater input into Arctic lakes because high amounts of glacial meltwater are exported from the Greenland Ice Sheet in this area (~158 km3 yr-1; Lewis and

Smith 2009), and particularly southwestern Greenland. GrIS meltwater contains nutrient concentrations that may be favorable for microbial growth in associated aquatic ecosystems.

For instance, high concentrations of the macronutrients nitrogen (N), phosphorus (P), and silica are exported in meltwaters from the Greenland Ice Sheet (GrIS) (Hawkings et al 2015; 2016).

Soluble reactive P (SRP) is a small portion of this P flux and reaches concentrations of 0.35 μM

(11 μg L-1). Most exported P associated with glacial sediments is in the particulate phase

(>97%). In total, GrIS P flux may contribute 15% of the bioavailable P input to Arctic oceans

(Hawkings et al 2016). Though GrIS meltwater from Leverett glacier in southwest Greenland contains low concentrations of DOC (41.4 and 17.7 μM, or 0.5 and 0.2 mg L-1 mean values for

2009 and 2010, respectively) it has a significant carbohydrate component and is highly bioavailable (Lawson et al 2013). Bhatia et al (2013) demonstrate that GrIS meltwater also contains high concentrations of dissolved and particulate iron (Fe), half of which is estimated to be bioavailable as a micronutrient. The large estimated annual fluxes of Fe in GrIS meltwater suggest it could be critical for supporting primary production in the North Atlantic Ocean at certain times during the year. The basal environments of continental ice sheets are biogeochemically reactive and weathering rates are high; their solute discharge rates are on the same scale as those of the world’s largest rivers (Wadham et al 2010).

The total nutrient load exported in meltwater increases with higher melt rates, suggesting nutrient export will increase in the future with climate warming (Hawkings et al

2016). Higher summer temperatures have increased the amount of meltwater runoff from the

GrIS since 1990 (Hanna et al 2008), and GrIS mass losses have increased by 100% since 1996

(van den Broeke et al 2009). Annual GrIS meltwater production is highest in the southwest of

Greenland (Lewis and Smith 2009). Thus, it is likely that highest nutrient export occurs in this region as well.

While nutrient export from the GrIS margin has been characterized in southwest

Greenland (Bhatia et al 2013; Lawson et al 2013; Hawkings et al 2015; 2016), the ecological effects of these subsidies in associated GF lake ecosystems remains unclear. GF lakes are prominent landscape features controlled by ice sheet dynamics. Russell Glacier, for instance, tightly controlled proglacial lake extent and physical structure by its recent advancement (from

1968 to 1999; Knight et al 2000). Proglacial lakes, rivers, and streams are more prevalent in surrounding areas with high meltwater output from the GrIS (Lewis and Smith 2009). Thus, GF

lakes may increase in number as rapid climate warming causes increased melt of the GrIS in the near future (Anderson et al 2017). Additionally, nutrient subsidies to lakes can increase rates of lake C cycling, for instance by increasing planktonic primary production and respiration rates

(Cole et al 2000; Pace and Cole 2000; Slemmons and Saros 2012). Lake bacterial cell-specific respiration rates of DOC decrease with P availability, and C-use efficiency increases (Smith and

Prairie 2004). With a highly labile DOC source (Lawson et al 2013), GF lakes have increasing potential to be biogeochemical hot spots for C cycling in the Greenland Arctic landscape

(McClain et al 2003). Determining how GrIS meltwater affects lake nutrient availability and subsequent ecological effects is important to understand the potentially important role Arctic

GF lakes play in elemental cycling.

Here we assess the ecological effects of solutes and nutrients from GrIS meltwater on

GF lakes along the ice sheet margin. Based on previous work on GrIS nutrient export (Bhatia et al 2013; Lawson et al 2013; Hawkings et al 2015; 2016), we hypothesized that GF lakes would have enhanced concentrations of biologically available P and dissolved inorganic N (DIN) compared to nearby snow and groundwater fed (SF) lakes due to meltwater inputs. We also hypothesized that similar to alpine lake ecosystems (e.g., Slemmons and Saros 2012), these nutrient subsidies would be biologically available, as indicated by geochemical analyses and microbial enzyme activities. In addition, we predicted that these subsidies, along with turbidity, would elicit different diatom assemblages between GF and SF lakes, with GF lakes having lower species richness. We address these questions by comparing water and sediment chemistry, microbial extracellular enzyme activities (EEAs), and diatom community structure between GF and SF lakes of similar size and proximity to the GrIS.

Figure 3.1. Map of study area, showing the locations of lakes relative to the Greenland Ice Sheet. Glacially fed (GF) lakes are indicated in light blue, and snow and groundwater fed (SF) lakes are dark blue.

Methods

Study site

The area around Kangerlussuaq in southwestern Greenland has warmed in recent decades, with 5.1 and 1.9 °C increases in spring and summer, respectively, over the period from

1981-2011 (Hannah et al 2012). Due to this rapid warming, and the fact that GrIS meltwater flux is highest in this region (Lewis and Smith 2009), lakes near the ice sheet margin in this area are ideal for addressing our questions. Bedrock in this area is composed of granitic gneiss (Nielson

2010). Target lakes were located along the GrIS and were sampled during the summer of 2015

(Figure 3.1). GF lakes included GL4, GL5, GL6, and GL7. Of those, GL4 and GL7 are headwater lakes, with both feeding into lakes GL5 and GL6. GF lake sizes ranged from 0.05 – 0.44 km2

(Table 3.1). SF lakes included SS901, SS903, SS906, and SS909. There is some surficial flow of water occasionally from Lake SS906 to Lake SS901. SF lake sizes ranged from 0.06 – 0.36 km2.

GL4, GL5, and GL6 were resampled in summer of 2016 for Chl a and surface sediments.

Table 3.1. Lake characteristics and sampling information. The only samples collected in 2016 were Chl a for the three indicated GF lakes and surface sediments from GL6. Surface sediments were used for both diatom counts and Psenner extractions.

All SF lakes included in this study stratify during summer, with oxygen profile data collected every summer since 2013 suggesting SF lake hypolimnia remain oxygenated (data not shown). Epilimnion depths in our target SF lakes excluding SS909 ranged from 5 m (SS906) to 11 m (SS903) in July of 2013 (Saros et al 2016). A weather station adjacent to SS903 indicated that average June temperatures at this location were 6.4 °C and 7.6 °C in 2013 and 2014, respectively, and July temperatures were 7.6 °C and 9.0 °C. We did not collect temperature and oxygen profile data to determine whether summer stratification and anoxia occurs in the GF target lakes. In turbid GF alpine lakes of the Austrian Alps, stratification occurs due to the absorbance of solar radiation by glacier flour particulates at the lake surface (Peter and

Sommaruga 2017). The surface temperatures of these stratified lakes can reach 16.65 °C. These stratified lakes often mix following precipitation events, due to the intrusion of cold water into the lake surface, and are classified as discontinuous cold polymictic. If southwest Greenland GF lakes stratify, the relatively small amount of summer precipitation in this region suggests that precipitation induced mixing events would not occur.

Water parameters

Chemical and physical water parameters were measured to determine differences between GF and SF lakes, and to assess their relationship to diatom community structure from

each lake type. Water temperature, conductivity, and pH were measured at the lake surface using a submersible HydroLab Datasonde 5A instrument. Turbidity was measured using a handheld turbidimeter (AquaFluor by Turner Designs). These parameters were measured from the deepest point of each lake, except for GL4, GL6, and GL7, which were measured near shore.

Water clarity was determined using a Secchi disk deployed from a raft and averaging two consecutive depth measurements. Secchi depth was not collected from GL7 due to logistical constraints. In SF lakes, surface water was collected using a van Dorn bottle deployed from the side of a rubber raft. In GF lakes, water was either collected with a van Dorn bottle deployed from a raft, or sampled near shore at 1 m depth in the case of GL4 and GL7.

Chl a was determined by filtering surface water through Whatman GF/F filters, which were stored frozen until analysis. Chl a was extracted from filters with 90% acetone, centrifuged, and quantified with a Varian Cary-50 Ultraviolet-Visible spectrophotometer

(Agilent Technologies; APHA 2000). Chl a was collected from target lakes in the summer of 2015 and again for GL4, GL5, and GL6 in summer of 2016. Values were averaged across years for those repeat samples. Total alkalinity was determined on unfiltered lake water by titration with

0.2N H2SO4 to pH 4.5 (APHA 2000). Total alkalinity was not collected from SF lake SS909 due to

+ - logistical constraints. Water used for dissolved nutrients (NH4 , NO3 , and SRP) and DOC was filtered through pre-rinsed Whatman GF/F filters and stored in acid-washed bottles. Dissolved and total nutrient samples (TN and TP) were analyzed on a Lachat QuickChem 8500 analyzer.

+ - NH4 , NO3 , and SRP were analyzed with the phenate, cadmium reduction, and ascorbic acid

+ - methods, respectively (APHA 2000). NH4 and NO3 were added together as dissolved inorganic

- nitrogen (DIN). Unfiltered TN and TP samples were measured as NO3 or SRP following persulfate digestion (APHA 2000). DOC was measured using a Shimadzu TOC analyzer.

Sediment Psenner extractions

Sediment Fe hydroxide (Fe(OH)3) and especially aluminum (Al) hydroxide (Al(OH)3) adsorb P in lake water and sediments, and can control P availability to microbes (Kopáček et al

2005; Norton et al 2011). Thus, as one measure of P bioavailability, we performed step-wise extractions of P, Fe and Al on the top 0-2 cm (0-3 cm for GL6) of lake sediments. Due to logistical constraints, GL4 and SS909 sediments were not collected for this analysis. This sequential analysis rapidly assesses sediment P, Fe, and Al concentrations. Ratios of these extractions can predict lake sediment P release in anoxic conditions (Kopáček et al 2005). Lake sediments were collected with a gravity corer from the deepest point we were able to locate in each lake. Sediments were extruded and stored frozen in Whirl-pak bags until analysis.

Following a modified method of Psenner et al (1984), four sediment extractions were performed: 1) 0.5 NH4Cl (in lieu of DI water, Tessier et al, 1979) extracts loose and readily- available P, Fe, and Al from sediment porewater; 2) 0.11 M Na2S2O4 buffered with 0.11 M

III NaHCO3 (BD) collects reducible species (particularly Fe oxyhydroxides) and associated P; 3) 0.1

M NaOH collects P associated with Al(OH)3 and organic material; and 4) 0.5 M HCl extracts P,

Fe, and Al associated with mineral material such as calcite and apatite (Kopáček et al 2005).

Following incubation on a shaker with each extract solution, samples were centrifuged for 15 minutes at 3000 rpm to separate the extract from the sediment. After the extract was collected, the centrifugation step was immediately repeated with fresh extract solution.

Supernatants from each extraction step were analyzed for P, Fe, and Al by inductively coupled plasma optical emission spectrometry (ICP-OES; Thermo Electron iCap 6300).

The potential for P release from lake sediments was evaluated using Al:FeNH4Cl+BD+NaOH

(Al:Fe) and AlNaOH:PNH4Cl+BD (Al:P) ratios. Briefly, if Al:Fe is below 3 and Al:P is below 25, then P is predicted to be released from lake sediments in anoxic conditions. If either of these ratios is greater than these thresholds, P release from lake sediments is likely minimal (Kopáček et al

2005).

The organic content of sediment (% organic matter, or %OM) was determined using loss on ignition (Heiri et al 2001). Briefly, wet sediment was weighed, dried at 100 °C for 16 hours, and reweighed to determine water content. The dried sediment was heated in a muffle furnace at 550 °C for 4 hours to combust all organic material. Finally, the combusted sediment was reweighed to determine %OM.

Extracellular enzyme activities

To assess microbial demand and acquisition efforts for C, N, and P, we measured microbial extracellular enzyme activities (EEAs) as described in Burpee et al (2016). Briefly, whole water samples were collected from lake epilimnia with a van Dorn bottle or by hand in acid washed bottles. Samples were stored unfiltered and frozen (-20 °C) until analysis. EEA samples were analyzed on a Thermo Electron Corporation Fluoroskan Ascent FL fluorescence spectrophotometer following the methods of Sinsabaugh and Foreman (2001), Findlay et al

(2003), and optimization procedures of German et al (2011). Since GF lakes had low EEAs, assays were run for 48 hours. The four enzymes selected for the assay together approximate microbial demand for macronutrients C, N, and P; these include β-1,4-glucosidase (BG), β-1,4-

N-acetylglucosaminidase (NAG), leucine aminopeptidase (LAP), and phosphatase (AP).

Respectively, these enzymes were measured using the fluorescent-labeled substrates 4-MUB-β-

D-glucoside, 4-MUB-N-acetyl-β-D-glucosaminide, L-Leucine-7-AMC, and 4-MUB-phosphate.

While EEAs of individual enzymes estimate total microbial demand for the corresponding nutrient, enzyme ratios BG:NAG+LAP and BG:AP estimate microbial investment in C acquisition efforts with respect to N and P (Sinsabaugh et al 2008; 2009).

Diatom community analysis

Diatom assemblages were enumerated from lake surface sediments. Single lake surface sediment samples (0-1 cm) were collected in 2015 from each lake with a gravity corer, except for GL5 and GL6 which were collected in 2016. SS909 sediments were not collected due to logistical constraints. Surface sediments for two glacial lakes, GL4 and GL7, were collected near shore from 1-m depth using a trowel, being careful to minimize disturbance. No obvious microbial mats were present in the littoral zone that could have altered diatom community structure collected in these areas. Sediments were treated with 10% HCl, followed by 30%

H2O2, which removed carbonate and organic material, respectively. Diatom slides were prepared following Battarbee (1986). Diatoms were counted with an Olympus BX51 microscope under 1000x magnification using oil immersion. A minimum of 300 diatom valves were counted and identified from each slide according to Krammer and Lange-Bertalot (1986) and Spaulding et al (2010).

Statistical analyses

Differences in physical and chemical factors between GF and SF lakes were evaluated using one-way analysis of variance (ANOVA). Because water chemistry and physical parameters

did not meet assumptions of normality, these were also evaluated using non-parametric

Kruskal-Wallis rank sum tests, which were in agreement with ANOVA tests at a significance level of p = 0.05. Thus, ANOVA results are reported. Lake sediment extraction data were log10 transformed to meet assumptions of normality. Principal component analyses (PCAs) were conducted to determine dominant environmental trends across lake types. Constrained correspondence analysis (CCA) was used to examine species distribution trends across sites and environmental variables. The analysis included the relative abundance values for the most dominant taxa (>5% relative abundance in a sample lake). TP, DIN, DOC, turbidity, and conductivity were log10 transformed and evaluated as explanatory environmental variables. To determine species richness across sample lakes, rarefaction analysis was conducted using a sample size of 300 individuals. All statistical analyses and plots were produced using R (version

3.3.2). ANOVAs were completed using the car package (version 2.1-4; Fox and Weisberg 2011), and rarefaction analysis, PCA, and CCA were conducted using the vegan package (version 2.4-2;

Oksanen et al 2017). Boxplots were made using ggplot2 (version 2.2.1; Wickham 2009).

Results

Water quality

Water chemistry was distinct between GF and SF lakes: TP concentrations were six times greater in GF lakes compared to SF ones (18 vs. 3 μg L-1, p < 0.01, n = 8; Figure 3.2a), though SRP

-1 -3 was below detection (<2 μg L PO4 -P) in all lakes. DOC concentrations were greater in SF lakes than GF ones (7 vs. <1 mg L-1, p = <0.01, n = 8; Figure 3.2g), as were TN concentrations (367 vs.

63 μg L-1, p < 0.01, n = 8; Figure 3.2c). DIN concentrations were higher in GF lakes, but not

-1 -1 - significantly so (9 μg L for SF lakes, 20 μg L for GF lakes, p = 0.10, n = 8; Figure 3.2e). NO3 or

+ NH4 , which make up the bulk of the DIN pool (Bergström 2010), were not significantly different between lake types when considered individually.

Figure 3.2. Physical and chemical properties of glacially fed (GF) and snow and groundwater fed (SF) lakes. Significant differences are indicated by ‘***’ (p<0.001), ‘**’ (p=0.001 – 0.01), or ‘*’ (p=0.01 – 0.05).

GF lakes along the GrIS were much more turbid than SF lakes (30.8 vs. 1.8 nephelometric turbidity units (NFUs), p < 0.01, n = 8; Figure 3.2b), and water clarity, measured by Secchi disk depth, was very low in GF compared to SF lakes (0.2 vs. 7.7 m, p < 0.01, n = 7;

Figure 3.2d). Conductivity (n = 8) and total alkalinity (n = 7) were lower in GF lakes (24 μS cm-1 and 4 mg L-1, respectively) compared to SF lakes (107 μS cm-1 and 46 mg L-1, p values = 0.03 and

0.02; Figure 3.2h). However, lake water pH was not significantly different between lake types

(6.5 for GF lakes, 5.9 for SF lakes, p = 0.18, n = 8). Average surface water temperature was similar across GF and SF lakes (8.4 and 10.6 °C, respectively, p = 0.39, n = 8; Figure 3.2f).

Figure 3.3. Principal component analysis (PCA) plot of physical and chemical properties of glacially fed (GF) and snow and groundwater fed (SF) lakes. GF lakes are labeled in red, SF lakes in black, and environmental variables in blue.

PCA axes 1 and 2 explained 78% of the variance in lake features (Figure 3.3). GF and SF lakes separated along the first axis, which closely tracked turbidity (PC1 score -0.94) and DOC concentration (PC1 score 0.96) and explained 55% of the variation. Conductivity also tracked the first axis (PC1 score 0.84) with higher values in SF lakes. Though DIN was not significantly different between GF and SF lakes, PCA suggested that higher concentrations are associated with GF lakes (PC1 score -0.58). The second axis explained 23% of measured water quality variation. Temperature and TP were oriented along the second axis (PC2 scores -0.85 and 0.61).

SF lakes exhibited the most separation along the second axis, compared to GF lakes.

Lake sediments

Differences in P chemistry between lake types were variable (n = 6): total extractable P

-1 (PTE) was similar between GF and SF lakes (16.4 and 13.7 μmol g dry sediment, p = 0.25; Figure

-1 3.4o), and the labile fraction of PTE (PNH4CL) was low across lake sediments (<0.1 vs. 0.1 mol g dry sediment, p = 0.01; Figure 3.4k). However, other fractions of sediment P were distinct

III between GF and SF lakes. Fe oxyhydride-associated P (PBD) was greater in SF compared to GF lakes (2.9 vs. 0.3 μmol g-1 dry sediment, p < 0.01; Figure 3.4l). P associated with amorphous

Al(OH)3 and organic material (PNaOH) was also greater in SF lakes compared with GF lakes (6.9 vs.

-1 2.0 μmol g dry sediment, p = 0.02; Figure 3.4m). In contrast, mineral-bound P (PHCl) was roughly four times higher in GF lakes than SF lakes (14.1 vs. 3.8 μmol g-1 dry sediment, p < 0.01;

Figure 3.4n).

Al and Fe sediment chemistry differed between lake types (n = 6). GF lake sediments

-1 contained more total extractable Al (AlTE) than SF lake sediments (127.4 vs. 31.3 μmol g dry sediment, p < 0.01; Figure 3.4e). Pore-water Al (AlNH4Cl) concentrations were both low (<1 μmol

-1 g dry sediment; Figure 3.4a) and did not differ between lake types (p = 0.60). AlBD was similarly low, and did not differ between lake types (1.1 and 1.3 μmol g-1 dry sediment in SF and GF lakes, respectively, p = 0.60; Figure 3.4b). In contrast, amorphous Al (AlNaOH) was nearly three times higher in GF lakes than SF lakes (43.8 and 15.2 μmol g-1 dry sediment, p = 0.01; Figure

3.4c). Mineral-associated Al (AlHCl) was five times higher in GF lakes compared to SF lakes (82.2 and 14.9 μmol g-1 dry sediment, respectively, p < 0.01; Figure 3.4d).

Figure 3.4. Sequential sediment extractions for aluminum (Al; a – e), iron (Fe; f – j), and phosphorus (P; k – o) from glacially fed (GF) and snow and groundwater fed (SF) lake sediments. Significant differences are indicated by ‘***’ (p<0.001), ‘**’ (p=0.001 – 0.01), or ‘*’ (p=0.01 – 0.05).

SF lake sediments contained more total extractable Fe (FeTE) than GF lake sediments

(245.4 vs. 112.8 μmol g-1 dry sediment, p = 0.03; Figure 3.4j). Both lake types had low pore-

-1 water Fe (FeNH4Cl; < 1 μmol g dry sediment, p = 0.01; Figure 3.4f). GF lake sediments had

III greater concentrations of Fe oxyhydrides (FeBD) compared to SF lake sediments (158.9 vs. 26.7

-1 μmol g dry sediment, p < 0.01; Figure 3.4g). FeNaOH concentrations similar between lake types

(p = 0.31; Figure 3.4h). Mineral-extractable Fe (FeHCl) was similar between lake sediments (70.6 and 77.2 μmol g-1 dry sediment for SF and GF lakes, respectively, p = 0.67; Figure 3.4i).

Figure 3.5. Ratios of glacially fed (GF) and snow and groundwater fed (SF) lake sediment extractions.

Phosphorus (P) release from sediments during anoxia is predicted if aluminum to iron (Al:FeNH4Cl+BD+NaOH) is

below 3 and if aluminum to phosphorus (AlNaOH:PNH4Cl+BD) is below 25.

Al:Fe and Al:P, indices that estimate the potential for P release from lake sediments under anoxic conditions (Kopáček et al 2005), were both higher in GF lakes compared to SF lakes (n = 6): 1.5 and 147.5 in GF lakes, respectively, and 0.1 and 5.4 in SF lakes (p values < 0.01;

Figure 3.5). Since P release is predicted if Al:Fe is below 3 and Al:P is below 25, these values indicate that P is more strongly bound to lake sediments in GF lakes, whereas P release from SF

lake sediments under anoxia is probable. Lake sediments from SF lakes contained more organic matter (OM) than GF lake sediments (12.0 vs. 1.7 % OM, p < 0.01).

Figure 3.6. Principal component analysis (PCA) plot of sequential extractions from glacially fed (GF) and snow and groundwater fed (SF) lake sediments. GF lakes are labeled in red, SF lakes in black, and extraction variables in blue. %OM = % organic matter.

PCA axes 1 and 2 explained 91% of the variation associated with lake sediment Al, Fe, and P speciation. The PCA demonstrated a separation of SF and GF lake sediments along the first axis, which tracked an organic (%OM, PC1 score 0.71) to mineral gradient (PHCl and AlHCl,

PC1 scores -0.71 and -0.70; Figure 3.6). Generally, Al and Fe also separated out across the first axis, with aluminum extractions associated with GF lakes and Fe extractions associated with SF lakes. Porewater Al (AlNH4Cl) and mineral Fe (FeHCl) were the exceptions to this pattern. FeHCl,

AlNH4Cl, and Al associated with reducible metal species (AlBD) tracked PCA axis 2 (PC2 scores -

0.65, 0.53, and 0.65, respectively) which separated both GF and SF lakes. While most P fractions were associated with SF lakes, PHCl was associated with GF lakes. The ratios Al:Fe and Al:P were strongly associated with GF lakes as well.

Biological metrics

AP activity, indicating microbial P demand, was similar between lake types (0.19 vs 0.17

μmol mL-1 h-1 in SF and GF lakes, respectively, p = 0.60, n = 8; Table 3.2). However, other enzyme activities were different between lake types. Both BG and NAG+LAP activities, indicating microbial C and N demand, respectively, were higher in SF lakes (0.10 and 0.23 μmol mL-1 h-1, respectively), and were low enough in GF lakes to reach limits of detection (both <0.01

μmol mL-1 h-1, p values = 0.03 and < 0.01, n = 8). BG:AP and BG:NAG+LAP ratios, indicative of microbial investment in C acquisition relative to P and N, respectively, were both higher in SF lakes (0.48 and 0.41) than GF lakes (< 0.01 and 0.05, p values < 0.01 and = 0.02, n = 8).

GF lakes contained higher water column Chl a concentrations (1.7 vs. 1.0 μg L-1 in SF lakes, p =

0.03, n = 8), possibly indicating greater planktonic algal biomass (Figure 3.7a). Though GF lakes had slightly lower diatom species richness, it was not significantly different between GF and SF lakes (22.1 and 33.0, respectively, p = 0.28, n = 7; Figure 3.7b).

Table 3.2. Microbial extracellular enzyme activities (EEAs) of glacially fed (GF) and snow- and groundwater-fed (SF) lakes. EEA units are μmol mL-1 h-1. Significant differences (p ≤ 0.05) between lake types are indicated by an asterisk (*).

Surface sediments (n = 7) demonstrated that both lake types were dominated by benthic diatom taxa, mostly belonging to the Achnanthidium or Psammothidium genera.

Abundant (>5% relative abundance) planktonic or tychoplanktonic taxa included Lindavia

68ormosa68, Discostella stelligera, Aulacoseira perglabra and Fragilaria vaucheriae, Fragilaria tenera, Staurosira construens, Staurosira construens var. venter, and Staurosirella pinnata.

Lindavia 68ormosa68 was only present in SS901 and SS906 which are hydrologically connected

(3.5 and 8.1 % relative abundance, respectively), while D. stelligera was present in all SF lakes

(11% mean relative abundance) and in GL5 and GL 6, also hydrologically connected (14% mean relative abundance). Achnanthes species were the most dominant in SS903 (all Achnanthes species representing 69% relative abundance). Achnanthidium minutissimum was also common in GF lakes (17% mean relative abundance). All Psammothidium species (Psammothidium chlidanos, Psammothidium scoticum, and Psammothidium subatomoides) were found solely in

GF lakes. For GF lakes, mean % relative abundances for P. chlidanos and P. scoticum were 10 and 23, respectively. Psammothidium subatomoides was only found in the littoral sediments of

GL4 (19% relative abundance). Fragilaria vaucheriae and F. tenera were only present in GF lakes

(7.0 and 5.4 mean % relative abundance, respectively). Staurisirella pinnata was observed mostly in SF lakes (9% mean relative abundance), though it was also present in GL5 and GL6 (4%

mean relative abundance). Staurosira construens was observed mostly in SF lakes (21% mean relative abundance), though it was also present in GL5 (0.8% relative abundance). Staurosira construens var. venter was only present in GL5 and GL6 (10.7 and 1.1% relative abundance, respectively).

Figure 3.7. Chlorophyll a (Chl a) as a measure of algal biomass (a) and average species richness, determined by rarefaction analysis (b). Significant differences are indicated by ‘***’ (p<0.001), ‘**’ (p=0.001 – 0.01), or ‘*’ (p=0.01 – 0.05).

CCA axes 1 and 2 explained 65% of the overall variation (and 69% of the constrained variation) of abundant diatom species distributions across lakes (Figure 3.8). When CCA axis 3 was included, the overall and constrained variation explained increased to 82% and 87%, respectively. Sites were distributed across a DOC (0.90 CCA1 biplot score) vs. TP and turbidity gradient (-0.89 and -0.94 CCA1 scores, respectively). Discostella stelligera was associated more with the second axis than the first (CCA1 score = 0.20 vs. CCA2 score = -0.80).

Figure 3.8. Canonical correspondence analysis (CCA) plot of diatom species distribution (right panel) across glacially fed (GF) and snow and groundwater fed (SF) lakes (left panel), determined by nutrient and turbidity variables (middle panel). ‘Turb’ = turbidity and ‘Cond’ = conductivity.

Discussion

We found that lakes fed by GrIS meltwater have distinct physical, chemical and biological characteristics. Compared to SF lakes, differences included higher concentrations of

TP and possibly DIN in GF lake water, though the difference in DIN was not significant. While the different sampling locations (pelagic versus littoral) in GF lakes may have affected our results, we note that the chemistry across GF samples was more similar than that in GF compared to SF lakes. These findings support our first hypothesis that GF lakes would have higher concentrations of phosphorus and DIN compared to SF lakes, though we could not assess P bioavailability from TP measurement alone. Conversely, SF lakes had higher concentrations of DOC and TN than GF lakes. DOC and TN tightly covary in SF lakes surrounding

Kangerlussuaq (Burpee et al 2016). Therefore the high TN concentrations in SF lakes are likely due to large DOC pools that contain organic N. We also determined that physical environments of GF lakes, measured by turbidity and water clarity, were different than those of SF lakes, while surface water temperatures were similar.

The enhanced P concentrations found in GF lakes in Greenland differ from patterns in

GF alpine lakes in North America, which have higher NO3-N than SF lakes but similarly low P

(Williams et al 2007; Saros et al 2010; Slemmons and Saros 2012). Bedrock underlying the GrIS at our study location consists of P-containing minerals, such as fluorapatite (Porder and

Ramachandran 2013; Hawkings et al 2016). Further, the subglacial environments of continental ice sheets such as the GrIS are biogeochemically reactive and have high weathering rates

(Wadham et al 2010; Hawkings et al 2016). High TP concentrations are present in other Arctic and alpine glacier meltwaters in Svalbard, Sweden, France, and Pakistan (Hodson et al 2004). In

contrast, low TP in North American GF alpine lakes could be attributable to different bedrock types and lower subglacial weathering rates. This region has slow weathering bedrock, for instance (Saros et al 2005).

Whether the source of nitrate in alpine glacier meltwater in the North American Rocky

Mountains is attributable to subglacial microbial and weathering processes or atmospheric deposition (i.e. the alpine distillery effect; Daly and Wania 2005) remains unclear (Williams et al

2007; Saros et al 2010). PCA associated DIN with GF lakes along the GrIS, and GF lakes had a higher mean concentration of DIN than SF lakes, though the difference was not statistically significant. Elevated DIN in GF lakes could be attributable to weathered minerogenic sources

(Holloway and Dahlgren, 2002) or to increases in anthropogenic N deposition in Greenland since the mid 19th century (Hastings et al 2009). Regardless of the sources, we found evidence that, similar to GF alpine lakes, glacial meltwaters to these Arctic lakes alter the nutrient chemistry compared to nearby SF lakes.

Though these GF lakes have higher concentrations of TP and higher algal biomass than

SF lakes, sediment chemistry analyses raise questions about the bioavailability of this P to microbial communities. While GF and SF lake sediments contained similar amounts of PTE, most

GF sediment P was in the mineral phase, whereas SF lakes had higher P content in more labile phases. Since the same suspended glacial flour that contributes to water turbidity settles to the bottom and constitutes GF lake sediments, it is likely that the P content of sediments and whole water samples are similar in GF lakes. Thus, the high TP measured in GF lake water is likely due primarily to mineral-associated P. The fact that SRP and DOC are both extremely low in GF meltwater minimizes these as potential sources contributing to TP. Our findings are consistent

with other work that reports glacial meltwaters in Svalbard, France, and Pakistan contain high

TP, but low ‘algal available’ P (PNaOH; Hodson et al 2004).

Higher aluminum content in GF lake sediments could further modify P availability to microbes. For instance, Al(OH)3 can adsorb to P, thus preventing its release from the sediment into the water column (Kopáček et al 2000; 2005). Norton et al (2011) demonstrated that over the course of 16.6 thousand years following deglaciation, Al species introduced into a northeast

American lake from developing catchment soils sequestered P from the water column. Diatom community changes tracked this declining P. Thus, high Al:Fe and Al:P ratios provide further evidence that GF lakes have less potential to release adsorbed P from glacial flour material

(Kopáček et al 2005). Lastly, our EEA results suggest high microbial demand for P relative to C and N, implying that P is biologically scarce in GF lakes, however these results should be interpreted with caution because the C and N enzyme activities were below detection. These low activities may reflect actual microbial demand, or possibly enzyme interference by adsorption to fine-grained mineral particles (Tietjen and Wetzel 2003). Together, these data suggest that while GF lakes have high TP concentrations, the bioavailablility of this pool may be low.

Some GF lakes in alpine regions are clear, not turbid; this is the case in many GF lakes of the central North American Rocky Mountains (Saros et al 2010; Slemmons and Saros 2012). In that area, high water clarity in alpine GF lakes is likely due to glacial flour settling and becoming entrained in lake inlet streams (Saros et al 2010), relatively reduced rates of weathering activity, and reduced amounts of glacier flour produced by comparatively smaller alpine glaciers. In other areas, turbidity is likely an important driver of lake ecology in GF lakes;

decreasing turbidity over a 400-year alpine lake sediment pigment record was one of the factors implicated in driving increased primary production (Vinebrooke et al 2010). In contrast,

Kammerlander et al (2016) demonstrated increased abundance and richness of algae and ciliates in an alpine GF turbid lake, compared to a clear lake. Our own evidence suggests that in addition to nutrient subsidies, lake turbidity is an important factor that corresponds to diatom species distribution across target lakes along the GrIS.

The similarity of surface temperatures between GF and SF lakes in Greenland is also consistent with patterns in some clear GF and SF alpine lakes (Slemmons and Saros 2012). We were unable to collect thermal profile data from the GF lakes studied here, hence it remains unclear whether vertical temperature gradients differ between GF and SF lakes in this region. In other regions such as the Austrian Alps, the turbidity of GF lakes alters thermal stratification patterns compared to those of SF lakes, with turbid GF alpine lakes being polymictic

(Sommaruga 2015; Peter and Sommaruga 2017). The surface temperatures of the target

Greenland GF lakes were lower than the alpine lakes observed by Peter and Sommaruga (2017), however, which can reach temperatures as high as 16.65 °C. The average surface temperature of the GF lakes included in this study (8.4 °C) suggests that stratification was not occurring at the time of observation.

Lakes surrounding Kangerlussuaq are mostly closed basins in which evaporative concentration controls lake water chemistry (Anderson et al 2001). Thus, higher conductivity in

SF lakes results from longer residence times and subsequently higher concentrations of ions due to evaporative concentration. GF lakes have shorter residence times than SF lakes owing to greater surface outflows.

Higher water column algal biomass in Greenland GF lakes is consistent with patterns in other clear and turbid alpine GF lakes (Slemmons and Saros 2012; Slemmons et al 2015;

Kammerlander et al 2016). This elevated planktonic algal biomass may be driven by P inputs, as even small amounts of labile P associated with mineral glacial flour can be biologically significant (Hodson et al 2004). Additionally, elevated DIN concentrations in GF and SF alpine lakes have a positive effect on algal biomass (Das et al 2005; Slemmons and Saros 2012;

Slemmons et al 2015), and could be responsible for the high water column Chl a in these GrIS- fed lakes. Lastly, higher Chl a concentrations that occur in some GF turbid lakes may be driven by altered light availability (Kammerlander et al 2016), which can lead to an increase in low- light adapted algae, such as cryptophytes (Gervais 1997), which were not counted in this study.

Furthermore, some algae can adapt to lower light intensities by increasing Chl a content per cell

(Jøsrgensen 1969). It must therefore be noted that though we use Chl a as a proxy for algal biomass, the proportional relationship between Chl a and biomass could differ between turbid

GF and clear SF lakes.

Algal biomass is often negatively correlated to species richness (Interlandi and Kilham

2001), but our hypothesis that GF lakes would have reduced species richness was not supported, despite this being observed in clear, N-subsidized alpine GF lakes (Saros et al 2010).

Species richness in Greenland GF lakes may be even higher than our reported value, which included littoral samples from GL4 and GL7. Some turbid GF lakes exhibit greater protistan

(ciliates and algae) species richness, likely owing to photoprotection and expanded resource niches (Kammerlander et al 2016). Thus, the reduced species richness expected of nutrient subsidized GF lake diatom communities (in accordance with resource competition theory;

Interlandi and Kilham 2001) may be offset by lake turbidity, which constrains light availability and reduces exposure to harmful light, thereby offsetting reductions in richness from nutrient enrichment.

Turbidity was the most important environmental variable that tracked diatom species distributions across the study lakes, closely followed by DOC and TP. Lakes that had similar landscape position and hydrological connectivity had more similar diatom communities. This was true for GL5 and GL6, and SS901 and SS906. Though GL4 and GL7 were not hydrologically connected, they had similar diatom communities that could reflect direct adjacency to the GrIS.

Benthic diatom taxa were abundant in both lake types, suggesting that benthic diatom abundance was not caused by differences in lake water chemistry. Benthic diatom dominance was unexpected in GF lakes because of high turbidity and low water clarity. The high proportion of GF lake benthic diatom abundance may be due to increased diatom productivity in the shallow littoral zones and relatively less productivity in the turbid water columns of GF lakes.

Benthic diatoms, including a high abundance of Psammothidium and Achnanthidium species, exist in supraglacial cryoconite holes on the surface of Antarctic and Arctic glaciers (Yallop and

Anesio 2010; Stanish et al 2013; Vinšová et al 2015). Thus, diatom assemblages of GF lakes along the GrIS could be dominated by supraglacial communities arriving via meltwater. The fact the surface sediments were collected from littoral zones for two out of the four GF lakes (GL4 and GL7) could also inflate the observed benthic diatom abundance. GL4 and GL7 clustered together in our CCA (Figure 3.8), suggesting similar environments from which the surface sediments were collected. Further, the sample size of GF lakes (n=4) is small in this study, and may not represent typical GrIS-fed systems. These considerations are important to consider

while interpreting our algal community data. Nevertheless, even GF lake surface sediments collected from greater depths (GL5 and GL6) contained high proportions of benthic diatom species such as A. minutissimum, S. construens var. venter, and F. vaucheriae.

Diatom communities were different between GF and SF lakes, and while some species were abundant in both lake types (D. stelligera and A. minutissimum), most were abundant in one or the other. The presence of F. tenera only in GF lakes is consistent with nutrient-enriched alpine

GF lakes (Slemmons et al 2015; Kammerlander et al 2016), and northeast Greenland Bunny

Lake (Slemmons et al 2016), and is associated with N enrichment (Das et al 2005; Sheibley et al

2014). In contrast, F. tenera was present in non-glacial central and northeastern Greenland coastal lakes, though lake nutrients were not evaluated in this study (Cremer and Wagner

2004). Paleolimnological evidence from North American alpine lakes suggests that, compared to a SF lake, the GF Jasper Lake had earlier dominance by key diatom species with moderate N requirements such as Asterionella 77ormosa and Fragilaria crotonensis (Slemmons et al 2015).

In East Greenland, the sedimentary diatom profile from Bunny Lake, fed by the Renland Ice Cap, revealed an increase in the relative abundances of taxa such as F. crotonensis, Fragilaria tenera, and Tabellaria flocculosa as glacial meltwater influx increased about 1,000 years ago (Slemmons et al. 2016). These diatom community shifts were attributed to changes in lake physical and nutrient characteristics induced by meltwater inputs, but contemporary data on these lake features were not available. Further, Discostella stelligera can be indicative of DIN enrichment

(Köster and Pienitz 2006; Perren et al 2017), and its high relative abundance in Greenland GF lakes is also consistent with that found in alpine GF lakes (Slemmons et al 2015). It is possible that D. stelligera is also in GL4 and GL7, but our sediment samples were from the littoral zones

of these lakes, and would underrepresent planktonic taxa if they are there. Together, these data suggest that the moderately elevated DIN in GF lakes is an important resource subsidy.

In conclusion, we show that GF lakes along the GrIS have their own distinct physical, chemical and biological features compared to SF lakes in the area. Though high turbidity, elevated TP, and moderate DIN appear to be important drivers of GF lake algal ecology, our work suggests that Al contributions to GrIS-associated lakes have important interactions with P and likely reduce its bioavailability in GF lakes. Our results contribute to the growing body of literature that reveals the spatial variability in the effects of glacial meltwaters on Arctic and alpine lakes worldwide. In turn, lakes play key roles in biogeochemical cycling in these regions.

Their growing numbers with glacial recession underscore their increasing importance in regional responses to global changes.

CHAPTER 4

SCALE-DEPENDENT PREDICTORS OF LAKE SENSITIVITY TO ATMOSPHERIC NUTRIENT

DEPOSITION IN ALPINE REGIONS

Abstract

Increased nitrogen (N) deposition rates over the past century have affected both North

American and European alpine lake ecosystems by shifting nutrient limitation status, increasing algal productivity and biomass, and decreasing water clarity. Ecological sensitivity of alpine lakes to N deposition varies, however, because chemical and biological responses are modulated by local watershed and lake properties. Thus, there is a need for an empirically based approach to predict mountain lake sensitivity to atmospheric N deposition for lake management and conservation. Here, we evaluate predictors of alpine and subalpine lake sensitivity across North American and European mountain ranges. Focal response variables of lakes included dissolved inorganic N (DIN) concentrations and algal biomass, measured as

Chlorophyll a (Chl a). Predictors of these responses were evaluated at three different spatial scales (hemispheric, regional, local) using three statistical methods in an exploratory fashion: regression tree, random forest analyses, and generalized additive model (GAM) analysis. N deposition rates (positive effect, +), % bare, unvegetated watershed (+), and bedrock geology (+ metamorphic, + sedimentary, but negative effects, -, for unconsolidated and igneous bedrock) were related to lake DIN at the Northern Hemisphere scale (R2 = 0.29, n = 165) and DIN (+), N deposition (+), and lake elevation (+) were related to lake algal biomass (R2 = 0.25, n = 118). At the regional (i.e. continental) scale, there was a distinction between North American and

European predictive factors of lake sensitivity. In North America, lake DIN was related to N

2 deposition rate (+), elevation (+), and maximum lake depth (Zmax, -; R = 0.31, n = 100) and lake algal biomass was related to N deposition rate (+), elevation (+), and the ratio of DIN to total phosphorus (DIN:TP, -; R2 = 0.34, n = 67). In Europe, DIN was related to % bare, unvegetated watershed (-; R2 = 0.11, n = 34) and algal biomass to DIN (-), TP (+), and their interaction (+; R2 =

0.85, n = 28). Local-scale (by mountain range) analyses revealed further distinction in predictor variables of lake sensitivity. In addition to N deposition rates, lake and watershed features such as land cover, bedrock geology, Zmax, and elevation were common modulators of lake DIN.

Localized algal biomass was consistently related to TP in Europe, while North American locales showed variable relationships to N or P. This study reveals scale-dependent ecosystem characteristics that render lakes vulnerable to atmospheric N deposition and provides important context to inform empirically based management strategies.

Introduction

In mountain regions, enhanced atmospheric N deposition results from fossil fuel combustion and agricultural practices (Howarth et al 2002; Galloway et al 2004; Hundey et al

2014). Alpine lakes are ecologically sensitive to atmospheric deposition due to their high elevation (Williams and Tonnessen 2000), slow weathering bedrock and low acid neutralizing capacity (ANC), absent or poorly developed watershed soils, sparse watershed vegetation cover, and low nutrient concentrations (Psenner et al 1999; Nanus et al 2005; Clow et al 2010).

Their high elevation exposes alpine lakes to higher rates of N deposition than their low- elevation counterparts (Williams and Tonnessen 2000). Enhanced atmospheric N deposition has shifted many lakes in North America and Europe from N-limitation to N and P co-limitation or P- limitation (Bergström and Jansson 2006, Elser et al 2009).

With short generation times, primary producers such as phytoplankton are sensitive to atmospheric nutrient deposition in oligotrophic alpine lakes (Baron et al 2011; Pardo et al

2011). Lake ecological effects include increased primary productivity (Goldman 1988; Jassby et al 1994; Nydick 2004), increased phytoplankton biomass (Sickman et al 2003; Nydick 2004), decreased lake clarity (Goldman 1988; Jassby et al 1994), and directional planktonic algal community changes caused by increased relative abundances of diatom taxa indicative of moderate nutrient enrichment such as Fragilaria crotonensis, Asterionella formosa, and

Discostella stelligera (Baron et al 1986; Goldman 1988; Jassby et al 1994; Wolfe et al 2001;

Wolfe et al 2003; Saros et al. 2005; Williams et al 2016). Together these ecological effects can alter the water quality of otherwise pristine alpine water resources.

Critical loads are a threshold value below which ecological effects of N deposition cannot be detected (Burns et al 2008). In alpine lakes, algal growth and diversity are highly sensitive to nutrient enrichment and respond to low DIN concentrations (3 - 43 µg L-1; Arnett et al 2012; Nanus et al 2012; Heard 2013), hence critical loads of N deposition that rely on algal responses can also be low for high elevation lake systems (1.0-1.5 kg N ha-1 yr-1 in the Sierras and Rocky Mountains of western United States; Saros 2011; Baron et al 2011). While critical loads are used as a management tool (especially in Canada and Europe; UNECE ICP 2004; Porter et al 2005; Ouimet et al 2006; Jeffries et al 2010), aquatic ecosystem response to N deposition rates is highly variable. Thus, N deposition rate alone is not always a good predictor of lake ecological sensitivity and response to enrichment.

Environmental variables and lake watershed features are important factors in determining lake responses to atmospheric N deposition. For instance, in a uniformly high N-

deposition area of the Rocky Mountains, lake phytoplankton biomass responded variably to nutrient addition bioassay experiments (Nydick et al 2003). Phytoplankton sensitivity to nutrient additions was modulated by certain lake characteristics, notably watershed vegetation, which was an important control of average lake nitrate and phosphate concentrations. Clow et al (2010) evaluated factors that contribute to lake sensitivity, specifically lake water nitrate concentration, to atmospheric N deposition, such as elevation, talus cover, unvegetated watershed area, alluvium, and riparian basin areas in central Sierra Nevada of California, US.

Acid neutralizing capacity (ANC) was an important determinant of lake sensitivity to

+ atmospheric deposition of SO2, NOx, and NH4 , which can cause acidification as well as eutrophication. Thus, lakes with low ANC were more vulnerable to N enrichment from atmospheric deposition. Nanus et al (2009) determined that lake ANC was predicted by watershed topography, bedrock, and vegetation type.

Though inclusion of lake watershed features improves the performance and predictive power of N deposition lake sensitivity models, variation that occurs at local scales (i.e., within mountain ranges, or across geographically close mountain ranges) can be important in explaining lake ecological responses to N deposition. Nanus et al (2012) modeled critical load values of lakes across the Rocky Mountains of the US, which included an estimated minimum

- -1 threshold of water NO3 -N concentration at which ecological responses would occur (7 µg L ), and watershed features such as slope, elevation, landcover type, soil type, and annual precipitation. Areas of critical load exceedance were then determined. Though sophisticated, this model predicted lakes in areas such as the Grand Teton Mountains, WY and Glacier

National Park, MT would be highly affected by N deposition. However, paleoecological research

from these areas indicates minimal recent changes in planktonic diatom communities.

Spaulding et al (2015) evaluated lake algal responses in the Grand Teton Mountains to atmospheric N deposition through time. Since their target lakes were shallow (<8.5 m) and clear, planktonic algal communities may have been constrained by ultraviolet light inhibition

(Vinebrooke and Leavitt 1999). Thus, sensitivity of planktonic taxa to N deposition through time was inconsistent among lakes. In Glacier National Park, lack of directional change in diatom communities to N deposition was attributed to lake P limitation (Saros et al 2011). Together, these studies suggest that local-scale factors can be important controls of lake sensitivity to N deposition. Since local-scale predictors of lake sensitivity likely differ among mountain ranges, they are important for water and wildlife managers to identify and include in their evaluation of target areas of concern.

Determining the factors that affect lake ecological sensitivity to atmospheric N deposition is a critical management priority with the potential to inform policy and management planning. There are locations that seem to deviate from predicted trends of lake ecological impact (e.g. Grand Teton Mountains and Glacier National Park), and factors that have not been assessed in previous lake sensitivity models, such as lake depth, clarity, and nutrient limitation, that may control lake sensitivity to atmospheric N deposition. To improve understanding of spatial variability of lake sensitivity to atmospheric N deposition, we evaluated lake and watershed factors across multiple spatial scales ranging from the Northern

Hemisphere (including North America and Europe) to local (single mountain ranges or mountain range groupings). Data used for this study were sourced from previously published lake surveys and research projects (Table 4.1). We used a combination of exploratory statistical approaches

to assess the relationships between deposition rates, watershed properties, physical and chemical lake characteristics and lake ecological sensitivity. Finally, we compared broad and local scale analyses to highlight the complexity and variation of lake sensitivity to N deposition.

Canadian Rockies (n = 109) Southern Rockies (n = 77) Northern Rockies (n = 21) Alps (n = 14) Pacific Northwest (n = 38) Bombervalds (n = 8) Central Rockies (n = 46) Tatras (n = 12) Sierras (n = 11)

Figure 4.1. Map of study lake coverage (n = 339, data contributed by the Alpine Lake Sensitivity Working Group) across North America and Europe. Colors refer to local mountain range groupings.

Methods

To evaluate lake sensitivity to atmospheric N deposition, we conducted an analysis of

339 alpine and subalpine lakes located in western North America and Europe (Figure 4.1). We used measurements of DIN (a metric of N enrichment) and Chl a concentration (an estimator for pelagic algal biomass) to evaluate lake sensitivity, since these variables represent N availability to lake algae and algal biomass, respectively. Predictor variables of lake sensitivity included a suite of physical, chemical, and biological variables listed in Table 4.1. These included

landcover type, slope steepness, and nutrient (DIN and TP) concentrations, all of which have been demonstrated to affect lake sensitivity to atmospheric N deposition in previous smaller- scale studies (Nydick et al 2003; Clow et al 2010). Additionally, we included lake depth, nutrient limitation status (DIN:TP; Bergström 2010), and bedrock geology, as we hypothesized these factors would be important at local scales (Saros et al 2010; Spaulding et al 2015). Dates of observations ranged from 1985 to 2017 and were limited to summer and autumn sampling to avoid spring snowmelt inputs and winter under ice conditions. Not all data variables were available for all sites (Table 4.1); thus, some local scale analyses included variables that were not considered in other locations. Many lakes had repeated observations from multiple years and seasons (1029 observations total, 128 lakes with multiple observations). Measurements were integrated if collected at different depths and averaged if sampled multiple times.

Since the availability of land cover data was variable across sites, vegetative covering categories (i.e. forests, meadow and shrub cover) and bare watershed categories (i.e. talus, scree, and bedrock) were combined, respectively, into two variables: % vegetation cover and % bare watershed. Bedrock geology data for US lake locations were obtained from the State

Geologic Map Compilation (SGMC; Horton et al 2017). Canadian bedrock geological data were from the Database of the Geologic Map of North America (Garrity and Soller 2009). European bedrock geology data were from the International Geological Map of Europe (IGME 5000; Asch

2005). Atmospheric deposition data for lake locations was obtained for the US from the

- + National Atmospheric Deposition Program (NO3 and NH4 from 1985-1999; total oxidized (Nox) and reduced N (Nred) from 2000-2017; NADP 2018). Canadian total annual N deposition data were obtained from the North American Climate Integration and Diagnostics model (NACID;

Table 4.1. Alpine lake data coverage. Country Region Number of Data Availability Lakes Slope Chl Data Source Landcover CA:LA Elevation Zmax TP DIN Steepness a US Southern Rocky 77 Baron J; Williams et al 2017 ± - - + ± ± + ± Mountains US Central Rocky 46 Brahney et al 2014; Ganz T; Mountains + ± ± + ± ± + ± Hundey et al 2014; 2016; Williams et al 2017 US Northern Rocky 21 Saros et al 2010; 2011; ± - - + + + ± ± Mountains Vinebrooke R; Williams et al 2017 US Pacific Northwest 38 Williams et al 2017 ± - - + + ± + + Mountains CA Canadian Rockies 109 Brahney J, Vinebrooke R - - ± + ± + ± +

US Sierra Nevada 11 Williams et al 2017 + - - + ± - + - Mountains AU, ITL, Alps 14 Koinig K, Sommaruga R + - + + + + + ± DE SK Tatras 12 Kopáček et al 2000 + - + + + + + +

CZ, DE Bombervalds 8 Kopáček et al 2004; 2006; Vrba et + - + + + + + + al 2003 “+” indicates data availability, “-” indicates data is unavailable, and “±” indicates partial data coverage. N = nitrogen, CA:LA = catchment area : lake area, Zmax = maximum lake depth, DIN = dissolved inorganic N, TP = total phosphorus, Chl a = chlorophyll a. Country abbreviations: AU = Austria, CA = Canada, CZ = Czech Republic, DE = Germany, ITL = Italy, SK = Slovakia, US = United States.

Hember 2018). As this model only had data to 2013, 2013 deposition data was used for lakes sampled after that year. European Nox and Nred deposition data were obtained from the

European Monitoring and Evaluation Programme (EMEP) model provided by the Norwegian

Meteorological Institute (MET Norway 2019).

To determine the spatial variability of lake sensitivity to atmospheric N deposition, data were evaluated at 3 different spatial scales: hemispheric, which included the whole dataset; regional, which split the data between North America and Europe; and local, which split the data according to areas of collection (see Figure 4.1). Lake sensitivity was assessed at each scale using three statistical techniques including regression tree analysis (De'ath and Fabricius 2000), random forest analysis (Breiman 2001), and generalized additive models (GAMs; Wood 2017).

These analyses were used in tandem due to the different strengths each offered. Regression tree analysis describes relationships between a dependent variable and its predictor variables but can be sensitive to small changes in the dataset. Random forest analysis is more robust than regression tree analysis but is not as descriptive in its summarization of relationships among variables. Based on these analyses, response variables can be selected and evaluated by

GAMs to determine the strength of a response to its predictor variables. Local scale data availability sometimes made some or all of these analyses inappropriate or impossible for one of the lake sensitivity metrics (lake DIN or Chl a).

To assess the relationships between response variables and their multiple predictor variables, regression tree analyses were performed using the R packages rpart (Therneau et al

2019) and rpart.plot (Milborrow 2019). To determine the relative importance of predictor variables in explaining response variable variation, random forest analysis was performed using

the R packages randomForest (Liaw and Wiener 2018) and caret (Kuhn et al 2019). Each random forest analysis was constructed using 1000 random trees with 3 random variables sampled at each split. To assess the ability of predictor variables to explain variation of lake sensitivity, GAM analyses were used to construct models of Northern Hemisphere lake responses (i.e. algal biomass or DIN concentration) to predictor variables identified in the regression tree and random forest analyses. GAMs of lake sensitivity were constructed from the predictors that regression tree and random forest analysis identified as important. These models were then evaluated for best performance, determined by the sample size (n), variance explained (R2), minimized generalized cross-validation (GCV) scores, and minimized Akaike information criterion (AIC) scores. GAM analyses were conducted using the mgcv R package

(Wood 2019). Generalized linear models were used in conjunction with the GAMs to confirm whether overall correlation between variable pairs was positive (+) or negative (-).

Results

Northern Hemisphere scale

Regression tree analysis demonstrated that across lakes, hemispheric lake DIN concentration was related to a combination of N deposition rates and lake watershed features

(mean = 132 µg DIN L-1, n = 250 observations; Figure 4.2). Average lake DIN was highest below

1156 m lake elevation (490 µg DIN L-1, 3% of study lakes). At or above this elevation, average lake DIN was 120 µg DIN L-1 (97% of study lakes). The amount of bare exposed watershed split this group further. Lakes with more than 11% bare watershed had mean DIN of 194 µg DIN L-1

(27%), which were further split into relatively low or high elevation (2029 m). Of those, low elevation lakes averaged 441 µg DIN L-1 (6%). High elevation lakes were further split by bedrock

geology. Lake DIN averaged 92 µg L-1 with igneous, sedimentary, or unconsolidated bedrock material (16%), or 237 µg DIN L-1 with metamorphic bedrock geology (5%). Lakes with less than

11% bare watershed had 91 µg DIN L-1. These lakes were split further into high or low N deposition groups (split by 5.2 kg N ha-1 yr-1). Low N deposition lakes averaged 46 µg DIN L-1

(32%), and high deposition lakes averaged 130 µg DIN L-1 (38%). High N deposition lakes were further modified by lake elevation, such that lower-elevation lakes had higher mean DIN.

Figure 4.2. Global dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value (top number) and relative sample size (bottom number, percentile value). alt = lake elevation, bare = % bare lake catchment, ndep = N deposition rate, geo = bedrock geology.

Mean Northern Hemisphere Chl a concentration was 2.1 µg L-1 (n = 218 observations).

Regression tree analysis indicated that N deposition was the primary control of lake algal

biomass (Figure 4.3). In lakes receiving N deposition rates above 7.9 kg N ha-1 yr-1, mean Chl a was 6.5 µg L-1 (11% of study lakes). These lakes were then split by 1050 m elevation, with lower and higher lakes averaging 2.7 µg L-1 (3%) and 8.1 µg L-1 (8%), respectively. In lakes in low N deposition areas (< 7.9 kg N ha-1 yr-1), mean Chl a was 1.6 µg L-1 (89%). DIN was the next important modulator of Chl a concentration in this group, with low-DIN lakes (<129 µg L-1) averaging 1 µg Chl a L-1 (71%), and lakes above 129 µg DIN L-1 averaging 3.9 µg Chl a L-1 (18%).

Figure 4.3. Global Chlorophyll a (Chl a) regression tree analysis. Each node displays average Chl a value and relative sample size. ndep = nitrogen (N) deposition rate, din = dissolved inorganic nitrogen (DIN), alt = lake elevation, tp = total phosphorus (TP), geo = bedrock geology.

Figure 4.4. Ranking dissolved inorganic nitrogen (DIN) control variables by their importance via random forest analysis. Analyses range from global (inclusive of all lakes in this study from North America and Europe) to local scale (single mountain range or mountain range grouping). IncNodePurity = increase in node purity, “alt” = lake elevation, “bare” = % bare lake catchment, “ndep” = N deposition rates, “geo” = bedrock geology, “zmax” = maximum lake depth, “ca.la” = lake catchment area:lake fetch area.

Figure 4.5. Ranking Chlorophyll a (Chl a) control variables by their importance via random forest analysis. Analyses range from global (inclusive of all lakes in this study from North America and Europe) to local scale (single mountain range or mountain range grouping). IncNodePurity = increase in node purity, tp = total phosphorus (TP), din = dissolved inorganic nitrogen (DIN), ndep = nitrogen deposition rates, alt = lake elevation, bare = % bare lake catchment, din.tp = DIN:TP ratio, zmax = maximum lake depth, geo = bedrock geology.

High DIN lakes (> 129 µg DIN L-1) were further modified by TP, with lakes above 4.4 µg TP L-1 having a mean value of 9.1 µg Chl a L-1 (4% of study lakes), and lakes below 4.4 µg TP L-1

averaging 2.5 µg Chl a L-1 (14% of study lakes). Lakes with less than 4.4 µg TP L-1 were further modified by DIN, such that above 415 µg DIN L-1 Chl a was low (0.86 µg L-1) but below it Chl a

-1 -1 was high (3 µg L ). The group of high Chl a lakes (3 µg L ) was divided by bedrock geology type

(metamorphic vs. igneous, sedimentary, or unconsolidated), with lower Chl a concentration in lakes with metamorphic bedrock (1.6 vs 4.2 µg L-1).

Table 4.2. Generalized additive model (GAM) summaries for dissolved inorganic nitrogen (DIN) and Chlorophyll a (Chl a). DIN Scale Area Predictors R2 n GCV AIC Global N. America & Europe N dep + % bare + geo 0.29 165 31109 2178 Regional N. America N dep + alt + Zmax 0.31 100 4287 1122 Regional Europe % bare 0.11 34 90411 486 Local Southern Rocky Mountains, US N dep + % bare + alt 0.08 48 9974 580 Local Central Rocky Mountains, US Zmax + geo 0.11 20 6893 234 Local Sierra Nevada Mountains, US N dep + geo 0.61 10 1436 102 Local Canadian Rocky Mountains, CA N dep + alt + geo 0.47 22 29469 290

Local Alps, DE, AU, I N dep * % bare 0.83 14 44763 190 Local Tatras, SK CA:LA + % bare 0.87 12 2466 129 Local Bombervalds, CZ alt + Zmax 0.41 8 61365 111 Chl a Scale Area Predictors R2 n GCV AIC Global N. America & Europe DIN + N dep + alt 0.25 118 13 639 Regional N. America N dep + alt + DIN:TP 0.34 67 10 347 Regional Europe TP*DIN 0.85 28 6 129 Local Southern Rocky Mountains, US N dep + % bare 0.60 5 13 26 Local Central Rocky Mountains, US DIN 0.95 11 3 44 Local Northern Rocky Mountains, US, CA TP 0.60 13 0 29 Local Pacific Northwest, US DIN 0.22 36 0.1 1 Local Canadian Rocky Mountains, CA % bare + TP 0.79 14 0.1 3 Local Alps, DE, AU, I TP * DIN 0.87 8 0.2 7 Local Tatras, SK DIN + TP 0.43 12 4 51 Local Bombervalds, CZ TP 0.82 8 20 48 Predictor abbreviations: N dep = nitrogen deposition, alt = elevation, Zmax = maximum lake depth, % bare = % bare catchment coverage, % vegetation = % vegetation catchment coverage, TP = total phosphorus. Abbreviations: N. America = North America, US = United States, CA = Canada, DE = Germany, AU = Austria, I = Italy, SK = Slovakia, CZ = Czech Republic.

Random forest analysis identified slightly different predictor variables for both Northern

Hemisphere DIN (Figure 4.4) and Chl a concentration (Figure 4.5). For Northern Hemisphere lake DIN concentration (n = 250 observations), % bare watershed and N deposition rate were the most important variables, with geology, elevation, and lake depth being the least important of variables considered. The most important variable for Northern Hemisphere Chl a concentration (n = 218 observations) was DIN, followed by TP and lake elevation. Bedrock geology and Zmax were among the least important variables for Chl a.

The best performing GAM for Northern Hemisphere DIN concentration included N deposition rates (+, p < 0.01), % bare watershed (+, p < 0.01), and bedrock geology type

(igneous, -, p = 0.01; metamorphic, +, p < 0.01; sedimentary, +, p < 0.01; and unconsolidated, -, p = 0.26; R2 = 0.29, n = 165 lakes; Table 4.2). The best performing Chl a model included DIN (+, p

= 0.03), N deposition rate (+, p < 0.01), and lake elevation (p = 0.09). This GAM explained 25% of the variance across 118 lakes (Table 4.2).

Figure 4.6. European dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. ndep = N deposition rate, zmax = maximum lake depth.

Regional scale

Regression tree analysis suggested differences in predictor variables of European versus

North American lake DIN concentration. European lake DIN concentration (mean DIN = 391 µg

L-1, n = 34 observations) was related to N deposition rates and lake depth (Figure 4.6). Lakes receiving high N deposition rates (above 8.1 kg N ha-1 yr-1) had higher DIN concentrations (653

µg L-1 on average, 32% of sample lakes). Lower N deposition European lakes had a lower mean

-1 DIN concentration (266 µg L , 68% of sample lakes) and were further split by a Zmax threshold of

10 m. Lakes deeper than 10 m contained higher DIN (358 µg L-1, 38% of study lakes), and shallower lakes contained lower DIN (145 µg L-1, 29% of study lakes). In contrast, North

American lakes had a variety of factors explaining lake DIN (n = 216 observations; Figure 4.7).

North American lakes were split along an N deposition rate threshold (5.2 kg N ha-1 yr-1), below which lakes had a lower average DIN concentration (49 µg L-1, 42%) and above which lakes were relatively DIN enriched (122 µg L-1, 58%). The group of lower N deposition lakes were further split by elevation: lakes receiving less than 5.2 kg N ha-1 yr-1 and located lower than 2200 m averaged 17 µg DIN L-1 (15%) and above 2200 m averaged 67 µg DIN L-1 (27%). The group of higher N deposition lakes (>5.2 kg N ha-1 yr-1) were split by elevation (2129 m). The lakes with highest concentration of DIN (263 µg L-1) were found between 1967 and 2129 m (6%). Above

2129 m elevation, mean lake DIN was 108 µg L-1 (46%). This group was split by elevation again

(3076 m), followed by bedrock geology type.

Random forest analysis predicted N deposition rates, % bare watershed coverage, maximum lake depth (Zmax), and elevation to be important predictor variables of Eurpoean lake

DIN (n = 34; Figure 4.4). However, the best performing GAM included % bare watershed (-, p =

0.02) as the sole DIN predictor variable (R2 = 0.11, n = 34; Table 4.2). Random forest analysis indicated lake elevation, N deposition rates, and Zmax were the most important predictor variables of lake DIN in North America (n = 216; Figure 4.6a); these variables were also the best

GAM predictor variables (lake elevation: +, p < 0.01; N deposition rate: +, p < 0.01; Zmax: -, p =

0.67) and explained 31% of DIN variation among North American lakes (n = 100; Table 4.2).

Figure 4.7. North American dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. ndep = N deposition rate, bare = % bare lake catchment, alt = lake elevation, geo = bedrock geology.

Regression trees indicated different predictor variables of Chl a between European lakes

(mean Chl a 3.7 µg L-1, n = 28 observations; Figure 4.8) and North American lakes (mean Chl a

1.9 µg L-1, n = 190 observations; Figure 4.9). TP emerged as an important predictor variable in

Europe, with 5.4 µg P L-1 as a threshold value. Below a TP concentration of 5.4 µg L-1

(representing 75% of study lakes), the model predicted a Chl a concentration of 1.8 µg L-1.

Above this concentration, Chl a was predicted at 9.4 µg L-1. North American controls of Chl a concentration were more complicated: lake elevation was most important, with elevation above 3094 m yielding a mean Chl a concentration of 5.2 µg L-1 (representing 19% of sample lakes). This group was further split by elevation (3394 m), with lakes below 3394 m averaging

8.7 µg Chl a L-1 (8%), and lakes above averaging 2.7 µg Chl a L-1 (11%). Lakes below 3094 m had a mean Chl a concentration of 1.2 µg L-1 (81% of sample lakes), and were split by DIN concentration. Above 148 µg DIN L-1, Chl a averaged 3.2 µg L-1 (9% of sample lakes) and below

148 µg DIN L-1, Chl a averages 0.9 µg L-1.

Figure 4.8. European Chlorophyll a (Chl a) regression tree analysis. Each node displays average Chl a value and relative sample size. alt = lake elevation, din = dissolved inorganic nitrogen (DIN), tp = total phosphorus (TP), ndep = nitrogen deposition rates.

Random forest analysis for European lakes suggested TP and DIN as the most important predictor variables for Chl a (Figure 4.5). The best performing GAM included TP (+, p = 0.23),

DIN (-, p = 0.17) and their interaction (+, p < 0.01) and explained 85% of the variation of lake Chl a concentration (n = 28; Table 4.2). Predictor variables identified for North American lake Chl a were DIN:TP, lake elevation, DIN, and N deposition rate (Figure 4.5). The best performing GAM for North American lake Chl a included N deposition rate (+, p < 0.01), elevation (+, p = 0.10), and DIN:TP (-, p = 0.87; R2 = 0.34, n = 67; Table 4.2).

Figure 4.9. North American Chlorophyll a (Chl a) regression tree analysis. Each node displays average Chl a value and relative sample size. alt = lake elevation, din = dissolved inorganic nitrogen (DIN), tp = total phosphorus (TP), ndep = nitrogen deposition rates.

Local scale

Southern Rocky Mountains (United States)

N deposition rate was identified as an important predictor variable in the regression tree of lake DIN concentration in the Southern Rocky Mountain range (n = 76; Figure 4.10).

Lakes that receive less than 4.7 kg N ha-1 yr-1 had a concentration of 64 µg DIN L-1 (29% of sample lakes), while lakes with greater deposition rates had an average concentration of 140 µg

DIN L-1 (71%). The subset of lakes that receive more than 4.7 kg N ha-1 yr-1 were split by bedrock geology, such that predominantly igneous and unconsolidated bedrock predicted lower DIN

(122 µg L-1 vs 176 µg L-1). Lakes with igneous and unconsolidated bedrock were further divided by the amount of N deposition they received, with low deposition lakes (< 5.6 kg N ha-1 yr-1) containing a mean of 84 µg DIN L-1 (12% of study lakes). Lakes receiving more than 5.6 kg N ha-1 yr-1 had a mean DIN concentration of 134 µg L-1. A subset of these lakes were modified by even higher N deposition rates (6.4 kg N ha-1 yr-1), which produced a mean DIN concentration of 183

µg L-1 (9% of study lakes). Lakes receiving between 5.6 and 6.4 kg N ha-1 yr-1 (26% of study lakes)

-1 had an average DIN concentration of 117 µg L and were divided by a threshold Zmax of 11 m, with 11 m or deeper lakes (17% of study lakes) containing less DIN than shallower lakes (9% of study lakes; 103 vs 143 µg L-1). Random forest analysis (n = 76) suggested N deposition rate as the most important indicator of lake DIN, followed by % bare watershed composition, elevation, and Zmax (Figure 4.4). The best performing GAM explained little variance of Southern

Rocky Mountain lake DIN and included N deposition rate (+, p = 0.09), % bare watershed (+, p =

0.12) and lake elevation (+, p = 0.13; R2 = 0.08, n = 48; Table 4.2).

Regression tree analysis only identified lake elevation (3480 m) as a predictor variable of

Chl a in the Southern Rockies (n = 30), with lakes above this threshold having a mean value of

1.9 µg Chl a L-1, and below 3480 m a mean value of 7.4 µg Chl a L-1. However, random forest identified % bare watershed and N deposition rate as being potentially important (Figure 4.5).

The best performing GAM model for Southern Rocky lake Chl a concentration included N

deposition rate (-, p = 0.21) and % bare watershed (+, p = 0.26) and explained 60% of the variation, though the sample size was small (n = 5; Table 4.2).

Figure 4.10. Southern Rockies dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. ndep = N deposition rate, geo = bedrock geology.

Central Rocky Mountains (United States)

Regression tree analysis demonstrated bare watershed was important in predicting DIN concentration of Central Rocky Mountain lakes (n = 43, Figure 4.11). Lakes with greater than

3.7% bare watershed (49% of study lakes) had a higher DIN concentration than lakes with less bare watershed (81 µg DIN L-1 vs 33 µg L-1). According to this model, the higher DIN lakes (>3.7% bare watershed) were further split into lakes with greater than or equal to 42% bare

100

watersheds, vs. lakes with bare watershed cover between 3.7 and 42% (37 µg DIN L-1 vs 130 µg

-1 DIN L , respectively). Random forest analysis, however, determined that Zmax was the most important predictor variable of lake DIN (Figure 4.10), and the best performing GAM used Zmax

(-, p = 0.93) and bedrock geology type (igneous: +, p = 0.67; metamorphic: -, p = 0.93; sedimentary: +, p = 0.15; unconsolidated: -, p = 0.96) as explanatory variables of DIN in Central

Rocky Mountain lakes (R2 = 0.11, n = 20; Table 4.2).

Figure 4.11. Central Rockies dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. geo = bedrock geology, ndep = nitrogen deposition rate.

There were not adequate data to yield a regression tree for Central Rocky Mountains

Chl a, nor random forest analysis. However, the best-performing GAM model revealed that DIN alone was the best predictor variable of lake Chl a (+, p < 0.01), though the data were limited

(R2 = 0.95, n = 11; Table 4.2).

101

Northern Rocky Mountains (United States & Canada)

There were not enough DIN data from the Northern Rockies to evaluate its controls.

Regression tree analysis did not suggest any control variables for Chl a, but random forest analysis determined that TP, DIN, % bare watershed, DIN:TP, and Zmax were most important variables for predicting lake Chl a concentration. The best performing GAM, however, included

TP as the only predictor variable (+, p < 0.01; R2 = 0.60, n = 13; Table 4.2).

Pacific Northwest Mountains (United States)

Regression tree analysis indicated that elevation was related to DIN concentration in lakes located in alpine areas of the Pacific Northwest US (Figure 4.12). Lakes above 1401 m had

9.9 µg DIN L-1 on average (82% of Pacific Northwest study lakes), and below this elevation, 41

-1 µg DIN L (41% of lakes). Random forest analysis suggested Zmax, followed by lake elevation, as the most important predictor variables for lake DIN (Figure 4.4). GAM analyses were not able to explain DIN concentration variation, however.

Figure 4.12. Pacific Northwest Mountains dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. alt = lake elevation.

Pacific Northwest lake Chl a was related to elevation and N deposition rate, according to regression tree analysis (n = 36; Figure 4.13). Lakes less than 1445 m in elevation (33% of study

102

lakes) contained a mean of 0.55 µg L-1 DIN. Above this elevation, lakes had less mean Chl a (0.34

µg L-1, 67% of study lakes). N deposition was a further modifier for this group of lakes, with lakes receiving more than 3.7 kg N ha-1 yr-1 having a mean Chl a concentration of 0.44 µg L-1

(31% of sample lakes). Lakes receiving less than 3.7 kg N ha-1 yr-1 had mean Chl a of 0.26 µg L-1

(36% of sample lakes). Random forest analysis suggested % bare watershed, N deposition rates, and lake DIN as the most important predictor variables of Chl a. The best performing GAM, however, included DIN as the sole predictor variable (+, p < 0.01; R2 = 0.22, n = 36; Table 4.2).

Figure 4.13. Pacific Northwest Chlorophyll a (Chl a) regression tree analysis. Each node displays average Chl a value and relative sample size. alt = lake elevation, ndep = nitrogen (N) deposition.

Canadian Rocky Mountains (Canada)

Regression tree analysis suggested that bedrock geology was an important predictor variable of lake DIN in the Canadian Rockies (n = 38; Figure 4.14). Lakes with metamorphic or sedimentary bedrock had a mean DIN concentration of 106 µg L-1 (63% of sample size). Lakes with igneous or unconsolidated bedrock material contained on average 300 µg DIN L-1 (37% of

103

sample size). There were not adequate data to perform random forest analysis for Canadian

Rocky lake DIN. The best performing GAM included N deposition rate, lake elevation, and bedrock geology for explanatory variables (R2 = 0.47, n = 22; Table 4.2).

Figure 4.14. Canadian Rocky Mountains dissolved inorganic nitrogen (DIN) regression tree analysis. Each node displays average DIN value and relative sample size. alt = lake elevation.

Chl a was related to DIN concentration according to the regression tree model, with TP further modulating Chl a (Figure 4.15). For lakes that had DIN below the threshold concentration of 148 µg L-1 (81% of sample lakes), the average Chl a concentration was 1 µg L-1.

This group was further divided by TP concentration; lakes below 9.5 µg TP L-1 contained 0.85 µg

Chl a L-1, and lakes above this threshold contained 1.5 µg Chl a L-1 on average. Lakes that had

DIN concentration greater than 148 µg L-1 contained on average 3 µg Chl a L-1. Random forest analysis suggested that % bare watershed was the most important predictor of lake Chl a, followed by TP, lake nutrient status, and Zmax (Figure 4.5). GAMs were in agreement with random forest analysis, as the best performing model included both % bare watershed (-, p <

0.01) and TP concentration (+, p = 0.02) as predictor variables (R2 = 0.79, n = 14).

104

Sierra Nevada Mountains (United States)

There were not adequate data for regression tree analysis of Sierra Nevada lake DIN.

However, random forest analysis suggested that Zmax, lake elevation, and N deposition rate were related to DIN concentration (Figure 4.4). The best performing GAM included N deposition

(-, p < 0.01) and bedrock geology (igneous: +, p < 0.01; metamorphic: -, p = 0.02), though data were limited (R2 = 0.61, n =10; Table 4.2).

Alps (Italy, Austria, and Germany)

Regression tree analysis was not appropriate given the amount of data available for lake

DIN in the Alps. Random forest analysis identified lake elevation as the most important predictor variable for DIN, followed by % bare watershed (Figure 4.4). The best performing

GAM included N deposition rate (+, p < 0.01), % bare watershed (+, p < 0.01) and their interaction (-, p < 0.01) as predictor variables (R2 = 0.83, n = 14; Table 4.2).

Random forest analysis identified TP as the most important predictor variable of lake Chl a in the Alps, followed by elevation and N deposition rate (Figure 4.5). The best performing

GAM included TP (+, p < 0.01), DIN (+, p < 0.01), and their interaction (-, p = 0.01) as predictors of Chl a, though data were limited (R2 = 0.87, n = 8; Table 4.2).

Tatra Mountains (Slovakia)

There were not adequate data to conduct regression tree analysis for lake DIN in the

Tatra Mountains. Random forest analysis suggested that % bare watershed was the most important predictor variable of lake DIN, followed by Zmax and the ratio of watershed area to

105

lake area (CA:LA; Figure 4.4). The best performing GAM used % bare watershed and CA:LA as predictor variables (R2 = 0.87, n = 12; Table 4.2).

Figure 4.15. Canadian Rockies Chlorophyll a (Chl a) regression tree analysis. Each node displays average Chl a value and relative sample size. din = dissolved inorganic nitrogen (DIN), tp = total

phosphorus (TP), zmax = maximum lake depth (Zmax).

As with lake DIN, there were not enough data to build a regression tree model for Tatra

Mountain lake Chl a. Random forest analysis identified DIN:TP, lake DIN, and TP to be the best three predictor variables of Chl a (Figure 4.5). The best performing GAM used DIN (-, p = 0.41) and TP (+, p = 0.16) as predictor variables (R2 = 0.43, n = 12; Table 4.2).

Bombervalds (Czech Republic, Germany)

Data were scarce in the Bombervald Mountains, as there were only 8 observations.

Random forest analysis determined lake DIN was related to CA:LA and lake elevation, followed

106

by Zmax, and to a lesser extent % bare watershed (Fig. 4). The best performing GAM included

2 lake elevation (+, p = 0.09) and Zmax (+, p = 0.15) as predictor variables (R = 0.41; Table 4.2).

TP and lake elevation were identified as the two primary predictor variables of lake Chl a

(Fig. 5). The best performing GAM used TP as the only predictor variable for Chl a (+, p < 0.01;

R2 = 0.82, n = 8; Table 4.2).

Discussion

At all scales, lake algal biomass was tightly coupled with lake nutrient status, but there were regional differences of nutrient importance. Our findings suggest North American lakes could currently be more ecologically sensitive to atmospheric N deposition than European lakes. For instance, North American algal biomass was related to N-associated factors (N deposition, DIN:TP, and lake elevation—see below for connection between elevation and lake

N). The relationship of N-related factors to the North American lake algal biomass suggests that these high-elevation lakes are generally N-limited. In contrast, European lake algal biomass was related to TP, DIN, and DIN-TP interaction, but TP had the greatest effect. Caution is recommended when interpreting positive relationships between TP and algal biomass, however, as these factors covary since algal biomass contributes to the TP pool (Stow and Cha

2013). Still, P control of algal biomass is likely reflective of the fact that European lakes are now

P limited after prolonged N enrichment (Kopáček et al 2005; Bergström 2006; Elser et al 2009).

Supporting this, mean European lake DIN (across locations) was 391 ug N L-1, compared to 91

µg L-1 across North American lakes. Thus, P deposition may be relatively more important than N deposition in N-enriched alpine lake ecosystems. Kopáček et al (2011) demonstrate that CA:LA modifies the importance of P deposition in alpine lakes of the Tatra Mountains, such that

107

atmospheric P contributions are more important in small watershed lakes with large surface areas. P limitation was demonstrated in lakes of the French Alps, located in areas receiving high rates of N deposition (Jacquemin et al 2018). Future projections for this area indicate that N deposition will remain unchanged or slightly decrease (Galloway et al 2004), but dust- associated P deposition sourced from North Africa will likely increase (Moulin & Chiapello

2004). In the Pyrenees mountains, declines in lake DIN that were concurrent with increases in atmospheric N deposition were related to atmospheric P deposition (Camarero and Catalan

2012). The authors provide evidence that following historical P limitation caused by N deposition, increasing P deposition rates are forcing Pyrenean lake ecosystems back to N limitation. Together, these studies suggest that European alpine lakes located in areas of historically high N deposition may be more sensitive to future inputs of P rather than N.

Table 4.3. Summary of elevation (alt) effects in generalized additive models (GAMs) for dissolved inorganic nitrogen (DIN) and Chlorophyll a (Chl a). DIN Scale Area Predictors R2 n Elevation effect (μg DIN L-1 m-1) -2 Regional N. America N dep + alt + Zmax 0.31 100 2.3 x 10 Local Canadian Rocky N dep + alt + geo 0.47 22 -0.40 Mountains, CA Local Bombervalds, CZ alt + Zmax 0.41 8 2.3 Chl a Scale Area Predictors R2 n Elevation effect (μg Chl a L-1 m-1) Global N. America & Europe DIN + N dep + alt 0.25 118 7.2 x 10-4 Regional N. America N dep + alt + DIN:TP 0.34 67 8.5 x 10-4

Locally, European mountain range (Alps, Tatras, and Bombervalds) algal biomass was consistently related to TP, which agreed with regional-scale European findings and suggest that these local-scale areas would be relatively insensitive to N deposition. Local lake algal biomass predictor variables in North America were more diverse. The Southern and Central Rocky

108

Mountains appeared to be primarily N-limited. These findings were surprising, as the Southern

Rocky Mountains were grouped almost entirely within Colorado. In the 1980s, N was a common limiting nutrient among Colorado alpine lakes (Morris and Lewis 1988), but the Colorado Front

Range has been a site of historically high and increasing N deposition rates (Lewis and Grant

1980; Lewis et al 1984; Baron et al 2000; Burns 2003), registering lake ecosystem responses that date back to the 1950s (Wolfe et al 2001; 2003). Our findings that Southern Rocky

Mountain lakes are still sensitive to N deposition contrasts somewhat with those of Williams et al (1996), who demonstrated N saturation within watersheds of the Colorado Front Range.

Further, Gardner et al (2008) determined that lake phytoplankton in the Colorado Front Range are now P-limited and do not respond to N enrichment. Lakes within the Colorado Front Range located east and west of the Continental Divide are differentially affected by N deposition, however (Morris and Lewis 1988; Baron et al 2000). For instance, N deposition rates are 3-5 kg ha-1 yr-1 on the East side, but 1-2 kg ha-1 yr-1 on the West side (Baron et al 2000). Thus, lakes west of the Continental Divide may exhibit higher sensitivity to N deposition, but our analyses were not fine enough to register this difference. Central Rocky Mountain algal biomass sensitivity to N is consistent with research in the Uinta Mountain Range that demonstrated 5 out of 6 lakes exhibited N or N and P co-limitation (according to DIN:TP; Hundey et al 2014).

Spaulding et al (2015) found directional community changes in diatom sediment cores that were coincident with anthropogenic N deposition in shallow Teton Mountain lakes. However, contemporary water chemistry from these lakes suggested mostly P or N and P co-limitation

(based on TN:TP).

109

In contrast, the Northern and Canadian Rocky Mountain ranges across the Northern

US/Southern Canada were P limited. This scheme matches interpretations of lake sediment diatom records from the Northern Rocky Mountains showing little to no response of algae communities to relatively high rates of N deposition in the area of Glacier National Park, MT

(Saros et al 2011). Increased phytoplankton and zooplankton biomass in response to P deposition in the Canadian Rockies is consistent with P limitation (Brahney et al 2014; 2015).

Nitrogen enrichment bioassays in Canadian Rocky Mountain lakes demonstrated general P limitation and insensitivity to N enrichment, suggesting these lakes will not be sensitive to N deposition (Murphy et al 2010). In the Eastern Front Range Mountain Lakes of the Canadian

Rockies, Cook et al (2020) determined a prevalence of P or N and P co-limitation. Thus, our findings of general N-insensitivity for Northern and Canadian Rocky Mountain ranges are in agreement with other research.

DIN concentration was dependent on environmental factors such as N deposition rates and watershed features, and predictor variables of lake DIN were more variable amongst spatial scales than Chl a. For instance, at the regional scale, N deposition rate was identified as an important predictor variable of North American lake DIN. N deposition gradients were reduced at local scales, thus analysis of Central Rocky Mountain lake DIN concentration was instead related to factors that modulate N deposition effects such as lake bathymetry, bedrock geology, and elevation. In contrast, N deposition was not related to European lake DIN at regional and most local scales, except for the Italian, German, and Austrian Alps. European lake

DIN was instead mostly related to the amount of bare, unvegetated watershed cover at the regional scale, and at local scales in the Alps and Tatras. This finding is comparable to those

110

from Clow et al (2010), who determined that bare watershed was positively correlated to surface water nitrate concentration in Yosemite National Park in the Californian Sierra Nevada

(Spearman rank correlation = 0.44, p < 0.01), due to facilitation of atmospheric N delivery to surface waters through their watersheds. It is possible that consistently high N deposition rates have reduced N deposition gradients across the European study areas (Kopáček et al 2005;

Bergström 2006; Elser et al 2009), thus regional and local scale European analyses instead identified factors that modulate N input into alpine lakes via their watersheds.

Lake elevation was an important variable for lake DIN concentration in many of the analyses at all scales (Table 4.3). While we expected elevation to be a negative predictor for lake DIN and Chl a, there was variability in the direction of the response (some of the smaller branches of the Northern Hemsiphere DIN regression tree, for example, show higher DIN within a node split to be associated with higher elevation). Though elevation itself is not a control of lake DIN or Chl a concentration, it is likely capturing other explanatory variable gradients. These could include N deposition rates, soil development, land coverage, precipitation patterns, and hydrological inputs, as have been considered in other research (Goldman 1961; Nanus et al

2009; Clow et al 2010; Nanus et al 2012). For example, Piovia-Scott et al (2016) consider how reciprocal land-lake cross-ecosystem carbon and nutrient transfers change with elevation in the

Californian Sierra Nevada and Klamath mountains. They observe that lake nitrate concentration increases with elevation, while epilimnion temperature, watershed vegetation cover, and DOC decrease. The increase of nitrate in higher elevation lakes may be attributable to differences in soil nutrient retention. Soil nutrient retention is affected by litter accumulation, snowpack dynamics, and soil root development (Miller et al 2005; Johnson et al 2009). Thus, in higher

111

elevation lakes that have less watershed vegetation cover, lakes receive more inorganic N because of decreased soil retention, and in lower elevation lakes soil retention modulates the amount of inorganic N that lakes receive from the watershed. A pattern of increasing DIN and

DIN:TP with elevation was also shown in the Californian Sierra Nevada mountain range by Sadro et al (2012). Though our analyses did not identify this trend in the Sierras, we demonstrated

DIN is positively related to lake elevation in the Bombervalds. At this location Chl a is related to

TP, suggesting that N subsidization from their watersheds and airsheds makes Bombervald lakes P-limited. Sadro et al (2012) did not find a significant relationship between lake elevation and algal biomass, and Piovia-Scott et al (2016) only considered bacterial abundance, which was negatively associated with elevation. In our study, we observed increased biomass with lake elevation though the effect sizes were small. The instances of algal biomass increase along elevation could be due to increased inorganic nutrient availability along elevation but is likely local differences in algal productivity and biomass between mountain ranges. For instance,

Figure 4.8 presents the North American Chl a regression tree analysis, where the first node splits the dataset into lower elevation lakes (< 3094 m) with a mean Chl a concentration of 1.2

μg L-1, and higher elevation lakes (≥ 3094 m) with a mean Chl a concentration of 5.2 μg L-1. The lakes which are located at this higher elevation are in the Southern Rocky Mountains (n = 75),

Central Rocky Mountains (n = 26), and to a lesser extent, the Sierra Nevada Mountains (n = 5).

These three groups represent the southernmost locations considered in the analysis. Further, they represent relatively highly impacted areas by atmospheric N deposition (see discussion of

Colorado Front Range, above). Thus, regional scale analysis of algal biomass presents elevation as a significant predictor variable, when in fact it is likely spatially covariable.

112

Our study is an important precursor to empirically based management strategies and approaches. Our findings provide a detailed evaluation of lake sensitivity to atmospheric N deposition that complements the critical load framework already in place and commonly practiced in North America and Europe (UNECE ICP 2004; Porter et al 2005; Ouimet et al 2006;

Jeffries et al 2010). While critical loads depend on a threshold rate of N deposition to determine ecological vulnerability of watersheds, our approach provides a method and strong justification for evaluating additional factors that can influence lake ecosystem sensitivity, especially at local scales where N deposition gradients may become less variable. In addition to identifying sensitive ecosystems for management and conservation purposes, the approach in this study is capable of identifying sentinel lakes that will likely manifest ecological change in response to atmospheric deposition (Williamson et al 2009). Thus, sensitive lake ecosystems could be targeted and prioritized for monitoring in areas of concern.

Trends that were identified at different scales have important management implications and will be useful for a variety of policy goals. For instance, we determined that at hemispheric and regional scales, N deposition is an important predictor variable of lake DIN and algal biomass. In other words, our large-scale analyses identified large-scale processes that influence broad areas. Thus, environmental policy being implemented at international and national scales should focus on reductions of N deposition by regulating agricultural, industrial, and fossil fuel emissions. At more local scales, for instance at the national, state, or park level, factors such as vegetation cover, elevation, bedrock geology, lake depth, and nutrient status will be important in deciding the lakes that should be targeted for monitoring due to increased ecological sensitivity.

113

Challenges to this project included variable data coverage, such that local areas had observations of different variables. Thus, some local analyses were able to include variables that were not present in others. This can make comparison between localities with mismatched variables difficult. Further, N deposition and land cover data were collected from different sources with different scales of resolution. With regards to N deposition data, we feel this limitation should not significantly affect the quality of our data or their interpretation, as our smallest local scale approach still covered relatively large geographical areas, i.e. entire mountain ranges. Further, the extension of atmospheric N deposition data to timeframes outside of the dataset’s range (for instance, extending the Canadian N deposition dataset forward in time, and the European dataset backward) should also not adversely affect the accuracy of our models, as N deposition rates did not change drastically over short-term time frames in these areas.

In summary, we demonstrate that predictive factors of alpine and subalpine lake sensitivity to atmospheric N deposition will be variable and distinct between localities and scales. We provide evidence that large scale processes, such as N deposition, are important at hemispheric and national scales, and that local lake and watershed characteristics determine lake sensitivity at smaller scales. Very often, ecological sensitivity is contingent upon multiple environmental factors. Thus, our empirically based approach that is capable of evaluating multiple environmental factors at multiple scales is proposed as a robust complementary method to critical load-informed management styles, which rely on a threshold value of atmospheric deposition to identify vulnerable ecosystems. This research serves to increase predictive capabilities of wilderness and freshwater resource management agencies and

114

simultaneously provide a flexible method to evaluate controls of ecological sensitivity that can meet diverse policy and regulation goals.

115

CHAPTER 5

ARCTIC LAKE SEDIMENT PHOSPHORUS RELEASE DURING ANOXIA INCREASES ALGAL BIOMASS

Abstract

Vertical temperature gradients in Arctic lakes are tightly linked to climate, and altered thermal stratification patterns may lead to increased occurrence and persistence of anoxic deep waters. While the effects of warming on external phosphorus (P) loading to Arctic surface waters have been investigated, internal P release from lake sediments during anoxic events has received less attention. Here, we use a combination of lake survey data from multiple years

(2013-2019) in combination with stepwise extraction analysis of lake sediment chemistry to assess: 1., Are Arctic lakes susceptible to internal sediment P release during hypoxic and anoxic events? 2., Are there specific lake or climate variables that predict the development of anoxic conditions? and 3., What are the ecological implications of sediment P release? To address these questions, we compared a group of lakes located in a dry, warm area of the Greenland

Arctic to a group of lakes located in a cooler, wetter, and windier area 40 km away. Summer stratification is stronger in lakes from the warmer, drier area compared to those in the cooler, wetter area. Lakes from both groups exhibited similar sediment chemistries, with ratios of iron

(Fe), aluminum (Al), and P suggesting sediments from both lake groups are conducive to P release during anoxia (mean Al:Fe = 0.36 and 0.09, and mean Al:P = 9.5 and 4.9 for lakes in the warmer, drier area and those in the cooler, wetter area, respectively). Deep water

(hypolimnetic) anoxia was only observed in lakes located in the warmer, drier area. The occurrence of anoxia was variable between years and occurred more frequently in certain individual lakes of the warmer, drier area. Individual lake morphometric features and thermal

116

stratification strength and stability metrics were not predictive of hypolimnetic dissolved oxygen (DO) concentration. Mid-summer hypolimnetic DO concentration was positively controlled by average springtime temperature (R2 = 0.19, p = 0.01) amongst lakes in the warmer, drier area, suggesting a climatic link to anoxic events. Across both regions, TP increased as hypolimnetic DO concentration declined (R2 = 0.23, p < 0.01), with hypolimnion DO status (oxygenated, hypoxic, or anoxic) being a significant predictor of TP concentration (p <

0.01). Lake hypolimnetic nutrient limitation status was not affected by DO, but algal biomass was higher at lower DO (R2 = 0.23, p < 0.01). Together, our results suggest that a lake benthic- to-pelagic subsidy, in the form of sediment P release during anoxia, is related to climate. This study highlights another dimension of Arctic lake sensitivity to warming, providing evidence that internal Arctic lake P cycling will be responsive to rapid climate change.

Introduction

Arctic lakes are located in areas experiencing rapid climate warming (Screen and

Simmonds 2010; Jeong et al 2014). Physical alterations of Arctic lakes in response to climate warming include shortened iced-over periods, with earlier spring ice-out (Šmejkalová et al

2016), and increased water temperature (Schindler et al 1990; McDonald et al 1996; Drake et al

2018). In northern temperate lakes, climate warming induces earlier onset and increased stability of lake thermal stratification (King et al 1997; Gerten and Adrian 2001). Lake thermal stratification refers to the vertical separation of water according to temperature-dependent density, with warmer, less dense water in the top epilimnion, denser, colder water in the bottom hypolimnion, and a zone of temperature transition (the metalimnion) between these two layers. Increased intensity and stability of lake thermal stratification was demonstrated in

117

Northern temperate lakes in response to climate change (Kraemer et al 2015; Richardson et al

2017) and is an anticipated response of Arctic lakes to climate warming (Winder and Sommer

2012; Vincent et al 2013; Anderson et al 2017).

Lake thermal stratification strongly affects lake chemistry and ecology, including benthic-pelagic coupling. It influences the extent of mixing between nutrient-rich hypolimnetic water and nutrient-depleted epilimnetic water (Livingstone 2003; O’Reilly et al 2003).

Therefore, peaks in algal biomass often correspond to lake turnover periods. A key limnological feature affected by lake thermal stratification is hypolimnetic oxygen concentration (Jankowski et al 2006). Hypolimnetic oxygen is consumed by heterotrophic processing of settling organic material and scavenging by reduced species such as methane that are derived from lake sediment (Watson et al 2016). Sustained stratification prevents hypolimnetic oxygen replenishment following its consumption, which has implications for lake water chemistry.

Under oxygenated conditions, lake sediment iron (Fe) and aluminum (Al) oxides exert absorptive control of phosphorus (P), a commonly limiting nutrient for lake phytoplankton.

Depletion of oxygen in lake hypolimnia increases the environmental reduction potential and is associated with the release of P from lake sediments as it dissociates from Fe- and Al-oxides.

This process alters benthic-pelagic coupling, whereby the benthic environment provides a P nutrient subsidy for the pelagic ecosystem (Baustian et al 2014). Thus, higher P concentrations are typically observed in anoxic lake hypolimnia (Marsden 1989, Søndergaard et al 2003), though this process is also controlled by a variety of other lake-specific factors (Hupfer and

Lewandowski 2008). One such factor includes the relative amount and speciation of Fe and Al,

118

which can create redox-insensitive associations with P, thereby controlling its availability even during anoxia (Kopaček et al 2005).

The mechanistic control of lake benthic-to-pelagic P subsidies (via sediment P release) is important in temperate lakes experiencing eutrophication and warming (Cooke et al 1993;

2005; Gibbons 2015). Accumulation of P within the hypolimnion leads to a nutrient pulse in the lake upon mixing. Thus, temperate lake algal blooms are triggered by P-enriched water following anoxic events during the breakdown of thermal stratification (Sommer 1985; French and Petticrew 2007; Wilhelm and Adrian 2008), though the frequency of these events is determined by the lake mixing regime (Ptacnik et al 2003; Huisman et al 2004). Though sediment P release and its ecological consequences are well understood for temperate lake systems, less is understood about these processes in remote Arctic lakes despite the prediction of increased stratification intensity and stability under current trends of climate warming

(Winder and Sommer 2012; Vincent et al 2013).

Effects of accelerated climate warming on Arctic lakes are numerous and include changes in P availability due to shifts in Arctic lake connectivity and cross-ecosystem transfers of organic material and nutrients (Hobbie et al 1999). Permafrost degradation has been identified in Arctic areas as a potential cause of enhanced external P loading from the watershed to surface waters (Hobbie et al 1999; Frey and McClelland 2009). P-rich dust from subglacial outwash that accumulates and deflates in glacial floodplains has also been considered an important aeolian input into Arctic lake systems in Canada, Alaska, Iceland, and

West Greenland (Bullard et al 2013). Lakes that are hydrologically connected to Arctic glacial meltwater outflows in West Greenland contain greater concentrations of lake water total P

119

(TP), though the bioavailability of this P pool is uncertain due to its mineral association

(Hawkings et al 2016; Burpee et al 2018). External inputs of P to lake ecosystems have important ecological effects: P-rich dust additions to lakes could stimulate bacterial production and lake metabolic rates (Carlsson and Caron 2001; del Giorgio and Newell 2012). In Alaska, sustained P loading to an experimental lake increased primary production rates and phytoplankton biomass, thus suggesting these effects will occur with permafrost-to-lake P transfer (Hobbie et al 1999). TP-rich glacially fed lakes in West Greenland contained greater algal biomass, moderately lower diatom biodiversity, and diatom assemblages (inclusive of the nutrient-rich species Fragilaria tenera; Das et al 2005; Sheibley et al 2014) that were different from nearby non-glacially fed lakes (Burpee et al 2018). Together, these studies demonstrate that climate warming is affecting P inputs into Arctic lake systems from the landscape, atmosphere, and cryosphere with important ecological impacts. Yet, the potential for Arctic lake benthic-to-pelagic P subsides has largely been ignored even though broad-scale Arctic environmental trends are conducive to positively influence internal P release from lake sediments.

Here, we investigate Arctic lake sediment P release under anoxic conditions during summer thermal stratification. Our research questions are 1.) Are Arctic lakes susceptible to internal sediment P release during hypoxic and anoxic events? 2.) Are there specific lake or climate variables that predict the development of anoxic conditions? and 3.) What are the ecological implications of benthic-to-pelagic P subsidies? To address these questions, we compare two groups of Arctic lakes: a group that exhibits relatively stronger stratification located in a warmer, drier area of the Greenland Arctic, and a group that exhibits relatively

120

weaker stratification located in a cooler, wetter, and windier area of Greenland. To measure the susceptibility of lake sediment P release, we employ sediment chemical extraction techniques and lake survey data from 2013-2019. To evaluate the ecological effects of sediment

P release, we assess water chemistry and measurements of lake algal biomass from the sampling period. Our research demonstrates how climate warming interacts with lake thermal dynamics to affect nutrient availability and lake biology. We consider the future trajectories of

Arctic lake sediment P release under current trends of rapid environmental warming, thus increasing our understanding of how remote Arctic lakes will respond to future climate change.

Methods

Study site

Kangerlussuaq and its surrounding area in West Greenland has a low Arctic continental climate with a mean summer temp of 10 °C, but it has experienced recent trends of warming, increasing springtime temperatures by 5.1 °C and summer temperatures by 1.9 °C from 1981 to

2011 (Hanna et al 2012). Annually, Greenland air temperature has increased by 3 °C and melting degree days have increased by 100% comparing the periods 2007-2012 to 1979-2000

(Mayewski et al 2014). The region surrounding Kangerlussuaq is rich in lakes (>20,000;

Anderson et al 2001) and therefore represents an excellent opportunity to assess the effects that rapid climate warming will have on sensitive Arctic lake ecosystems.

121

Surface area Zmax Mean Hypo DO Hypo TP Area Mean BF (s-1) Lake Lat Long Al:Fe Al:P (km2) (m) depth (m) (mg L-1) (μg L-1) SS2 66.99 -50.96 0.368 12 6 0.3 11.5 7.0 6.3 SS8 67.01 -51.08 0.146 10 4 0.6 5.6 2.0 14.5 SS85 66.98 -51.06 0.246 11 4 0.6 18.4 4.5 4.9 Drier, warmer 32 x 10-4 SS1341 66.99 -51.14 0.070 14 5 0.3 6.6 5.7 5.0 SS1381 67.02 -51.12 0.215 19 6 0.3 13.2 3.5 4.9 SS1590 67.01 -50.98 0.243 18 5 0.1 1.9 3.1 6.7 SS86 66.96 -49.80 0.042 12 6 0.1 4.4 11.2 4.9 SS901 67.13 -50.24 0.106 15 8 0.1 4.3 10.1 4.9 Colder, wetter 12 x 10-4 SS903 67.13 -50.17 0.354 29 11 0.1 3.4 10.4 3.2 SS906 67.12 -50.25 0.085 18 7 0.1 7.5 9.2 5.6 Table 5.1. Study lake characteristics. BF = buoyancy frequency; Zmax = maximum depth; Al:Fe = lake sediment aluminum to iron ratio; Al:P = lake sediment aluminum to phosphorus ratio; ‘Hypo DO’ refers to hypolimnetic dissolved oxygen concentration; ‘Hypo TP’ refers to hypolimnetic total phosphorus concentration. BF is the July 2013 mean among lakes within each category (data from Saros et al 2016); a higher BF indicates stronger thermal stratification. DO and TP concentrations are the mean values from 2013-2019 for each lake.

122

Figure 5.1. Map of ten study lakes in the Southwest Greenland Arctic. Lakes located in the warm, dry area are indicated with pink markers. Lakes located in the colder, wetter area are located proximal to the Greenland Ice Sheet and are indicated with blue pins.

123

Ten study lakes (Table 5.1) were divided into two comparison groups consisting of those located in a warmer, drier area and those located in a wetter, cooler and windier area 40 km away (Figure 5.1). Lakes in both regions are low-nutrient, oligotrophic systems, which is typical for the area (Anderson et al 2001; Perren et al 2009) and representative of Arctic lakes generally. All lakes included in this study stratify during summer except for SS86 (included within the cooler, wetter area lake group), which is located on within a nunataq surrounded by the Greenland Ice Sheet. Lakes in the warmer, drier area stratify more strongly than those in the cooler, wetter area. For instance, mean Brunt-Väisälä buoyancy frequency (BF), a measure of stratification strength, was 32 x 10-4 s-1 among sample lakes located in the warmer, drier area in July 2013 (SD = 14 x 10-4 s-1), compared to 12 x 10-4 s-1 (SD = 9 x 10-4 s-1) among sample lakes located in the cooler, wetter area (calculated from Saros et al 2016, in addition to SS86). The euphotic zone (lower limit defined by the depth at which PAR equals 1% of surface irradiance) exceeds average lake depth in the study lakes (Malik et al 2017). This suggests that light in the hypolimnetic environment of the study lakes is sufficient to support phytoplankton growth.

Most lakes in the area have long residence times (Anderson and Stedmon 2007; Leng and

Anderson 2003) and minimal surface water connectivity (Bosson et al 2013). The cooler, wetter area is located near the Greenland Ice Sheet, though the study lakes at this location are not hydrologically connected to it. Easterly katabatic winds coming off the Ice Sheet are frequent and strong in this area (Bullard and Austin 2011). Summer temperatures here are colder than the warmer and drier area, which is located west of the Kangerlussuaq International Airport.

For instance, a weather station located in the cooler, wetter area indicated that average June temperatures at this location were 6.4 °C and 7.6 °C in 2013 and 2014, respectively. In contrast,

124

average June temperatures at the Kangerlussuaq International Airport were 10.9 °C and 11.7 °C in 2013 and 2014, respectively. Lakes were sampled in June-July from 2013-2019, as well as in

August in 2014. May-June is often when the lake ice comes off starting with lakes in the warmer, drier area and followed by the lakes located closer to the Ice Sheet. Lakes stratify shortly after ice-off (Brodersen and Anderson 2000). The end of June through July is considered peak summer season, when lakes are usually all stratified.

Quantifying susceptibility to lake sediment P release

Aluminum (Al) and to a lesser extent, Iron (Fe) hydroxides (Al(OH)3 and Fe(OH)3, respectively) can adsorb P in lake sediments and control its availability to bacterioplankton and phytoplankton (Kopáček et al 2005; Norton et al 2011). We therefore determined the Al, Fe, and P to better understand the potential for sediment P release in our study lakes. We performed a stepwise extraction (known as Psenner extraction; Psenner et al 1984) to measure concentrations of Al, Fe, and P in the top 2 cm of lake sediments. Lake sediments were collected with a gravity corer from the side of a rubber raft in 2015 from the deepest points of lake basins. Sediments were extruded on site and stored frozen in Whirl-pak bags until analysis.

A modified Psenner et al (1984) method was used to perform 4 extractions: First, 0.5 M NH4Cl

(instead of DI water; Tessier et al 1979) extracts readily available P, Fe, and Al from sediment porewater; second, 0.11 M Na2S2O4 buffered with 0.11 M NaHCO3 (BD) extracts reducible Al and Fe species (particularly FeIII oxyhydroxides) and associated P; third, 0.1 M NaOH collects P associated with organic material and Al(OH)3; fourth, 0.5 M HCl extracts Fe, Al, and P associated with mineral material (e.g. apatite and calcite; Kopáček et al 2005). Samples were incubated on a shaker with each extraction solution, then centrifuged for fifteen minutes at 3000 rpm to

125

separate the sediment and extract. The supernatant was collected, then centrifugation was immediately repeated with fresh extract solution. Extracts were analyzed for P, Fe, and Al by inductively coupled plasma optical emission spectrometry (ICP- OES; Thermo Electron iCap

6300).

To assess the potential for lake sediment P release, we determined Al:Fe and Al:P ratios.

Al:Fe is determined as AlNH4Cl+BD+ NaOH:BDNH4Cl+BD+ NaOH, and Al:P as AlNaOH:PNH4Cl+BD. P is predicted to disassociate from lake sediments during anoxic conditions when Al:Fe is below 3 and Al:P below 25 (Kopáček et al 2005).

Sediment organic matter content (%OM) was determined by loss on ignition (Heiri et al

2001). Wet sediment was weighed, dried at 100 °C for at least 16 hours, and reweighed to determine water content. The organic matter associated with the dried sediment was combusted at 550 °C for 4 hours in a muffle furnace, and the combusted sediment was reweighed to determine %OM.

Assessing lake anoxia and its predictor variables

To evaluate lake thermal stratification and hypolimnetic anoxia, lakes were sampled annually (sometimes twice per year) following ice-off and during peak summer season from the side of a rubber raft. Water column temperature and dissolved oxygen (DO) were measured every meter using a Hydrolab® DataSonde 5a instrument (OTT Hydromet, Loveland, Colorado) from 2013-2015, and a YSI EXO2 multiparameter sonde (YSI Incorporated, Yellow Springs, Ohio) from 2016-2019. DO thresholds were used to categorize hypolimnion status as oxygenated, hypoxic, or anoxic for comparison across lakes. The threshold of DO below which anoxic biogeochemical processes occur is difficult to determine for any given lake as it is dependent on

126

other factors such as organic carbon inputs and microorganism community and processes

(Nealson and Saffarini 1994). However, a general guideline for anoxia that we employ here is <2 mg DO L-1 (Nürnberg 1995). Hypoxia is also difficult to determine as it is system-dependent, but here we use 2-4 mg DO L-1, which a conservative threshold compared to those used by other water management agencies and organizations (Nürnberg et al 2002). Thus, hypolimnia with

DO concentrations >4 mg DO L-1 were categorized as oxygenated.

Lake stratification is an important predictor of lake hypolimnetic anoxia (Jankowski et al

2006). Lake morphometry (depth, surface area, and volume) influences lake thermal stratification (Gorham and Boyce 1989) and how lake thermal structure responds to climate warming (Kraemer et al 2015). Therefore, we assessed the relationship of lake morphometry, stratification strength and stability, and climate effects to hypolimnetic DO concentration using linear regression analysis. These analyses only included lakes located in the warmer, drier area because they were susceptible to hypolimnetic anoxia (anoxia did not occur in lakes located in the cooler, wetter area). To assess the relationship between lake morphometry and anoxia, three morphometry-dependent indices were considered: lake geometry ratio (GR), climate exposure index (CEI), and Schmidt’s stability index (SSI). Briefly, the GR index is a ratio of lake

0.25 surface area (As) to maximum depth (Zmax), expressed as As :Zmax, and is a relative measure of lake stratification susceptibility (Gorham & Boyce 1989). CEI is the ratio of lake As to mean depth (As:Zavg) and estimates lake exposure to climate (Magnuson et al 1990). SSI calculates the stability of a thermally stratified water column, or the resistance to mechanical mixing due to the potential energy contained within the stratified water and is calculated from temperature and lake volumetric profiles (Schmidt 1928; Hutchinson 1957; Idso 1973). To assess the

127

relationship between individual lake stratification intensity and anoxia, we evaluated the relationship between BF and lake hypolimnetic DO concentration. BF is calculated from lake temperature profiles and is a measure of stratification strength determined by vertical density gradients (Read et al 2011). BF was determined for each 1 m depth down a lake’s thermal profile and averaged across the water column. The R package ‘rLakeAnalyzer’ was used to calculate SSI and BF (Winslow 2019). Lake surface area, mean depths, and volumes were calculated from bathymetric maps, created as described in Saros et al (2016).

To account for annual variability of lake anoxia, the relationship between spring and summer average air temperature (Table 5.2) and hypolimnetic DO was assessed among lakes in the warmer, drier area. Because the majority of our survey data were collected in June and early July, the effect of July mean temperature was not included as a potential predictor of hypolimnetic DO. August (i.e., late summer) lake data that were collected in 2014 were excluded from statistical analyses in order to reduce seasonal variability between mid and late summer. Regression analysis that considered only peak summer data included data that were collected between June 25 – July 31. To assess the relationship between lake anoxia and climatic variables, average temperature data for March-June were obtained from the Danish

Meteorological Institute (DMI) weather station at the Kangerlussuaq airport. Annual spring mean temperature was calculated by averaging March, April, and May (MAM) temperatures, which represents the timeframe leading up to lake ice-out.

128

Lake SS2 SS8 SS85 SS1341 SS1381 SS1590 GR (m-0.5) 2.1 2.0 2.1 1.2 1.1 1.2 CEI (m) 68478 40252 74729 18598 37153 50882 2013 Anoxia Yes Yes Yes Yes Yes Yes SSI (J m-2) 3 17 14 11 26.1 5 BF (s-1) 3.0 x 10-5 3.8 x 10-4 3.9 x 10-4 1.6 x 10-4 1.9 x 10-4 8.0 x 10-5 MAM -3.8 °C June 10.9 °C 2014 Anoxia No Yes No - No No SSI (J m-2) 34 19 4 - 54 35 BF (s-1) 6.0 x 10-4 5.8 x 10-4 6.3 x 10-4 - 4.7 x 10-4 4.7 x 10-4 MAM -8.1 °C June 11.7 °C 2015 Anoxia No Yes - No Yes Yes SSI (J m-2) 38 57 29 37 120 54 BF (s-1) 5.6 x 10-4 5.6 x 10-3 1.4 x 10-3 6.8 x 10-4 8.4 x 10-4 6.1 x 10-4 MAM -9.6 °C June 9.5 °C 2016 Anoxia No No No No No No SSI (J m-2) 41 16 7 20 67 46 BF (s-1) 6.9 x 10-4 6.8 x 10-4 5.6 x 10-4 4.2 x 10-4 5.7 x 10-4 5.4 x 10-4 MAM -0.5 °C June 10.4 °C 2017 Anoxia No Yes No - Yes Yes SSI (J m-2) 41 34 19 - 72 36 BF (s-1) 5.4 x 10-4 1.1 x 10-3 6.8 x 10-4 - 6.1 x 10-4 3.9 x 10-4 MAM -7.6 °C June 9.9 °C 2018 Anoxia No Yes No - No Yes SSI (J m-2) 28 13 14 - 32 28 BF (s-1) 4.5 x 10-4 5.0 x 10-4 4.3 x 10-4 - 2.8 x 10-4 3.2 x 10-4 MAM -7.4 °C June 9.0 °C 2019 Anoxia No Yes No - No Yes SSI (J m-2) 48 18 12 - 55 51 BF (s-1) 6.9 x 10-4 6.8 x 10-4 6.7 x 10-4 - 5.1 x 10-4 5.1 x 10-4 MAM -2.3 °C June 11.3 °C

Table 5.2. Annual climate data and lake indices for lakes located in the warmer, drier area. GR = lake geometry ratio; CEI = climate exposure index; “Yes” or “No” indicates whether anoxia was observed during the sampling season of the indicated year; SSI = Schmidt’s stability index; BF = buoyancy frequency; MAM = March, April, and May’s average temperature; June = June’s average temperature. If lakes had more than one SSI or BF value per year, values for that year were averaged.

129

Quantifying ecological response to sediment P release

Lake water chemistry and biological metrics were assessed to determine how lake hypolimnetic anoxia affects internal lake P loading and algal biomass. During the lake sampling period, water samples from lake metalimnia and hypolimnia were collected using a van Dorn sampling bottle to quantify water nutrient chemistry and pelagic algal biomass, measured as

Chlorophyll a (Chl a). Chl a was measured by filtering lake water through Whatman GF/F filters.

Filters were stored frozen until analysis within three weeks of collection. Filtered Chl a was extracted using 90% acetone, followed by centrifugation and quantification on a Varian Cary-50

Ultraviolet-Visible spectrophotometer (Agilent Technologies; APHA 2000). P and nitrogen (N) nutrient samples were analyzed on a Lachat QuickChem 8500 analyzer. Total phosphorus (TP) was measured from whole water samples stored at 4 °C in acid washed bottles until analysis. TP was digested using the persulfate method and analyzed as soluble reactive P (SRP) using the

+ − ascorbic acid method (APHA 2000). Dissolved nitrogen (NH4 and NO3 ) sample water was filtered through Whatman GF/F filters and stored in acid washed bottles at 4 °C until

+ − analysis. NH4 and NO3 were analyzed with the phenate and cadmium reduction methods,

+ − respectively (APHA 2000). NH4 and NO3 were then added together as the sum of dissolved inorganic nitrogen (DIN). The ratio of DIN:TP was determined in order to measure lake nutrient limitation status; this ratio indicates a shift from lake N to P limitation as it increases from 1.5 to

3 (Bergström et al 2010).

Statistical analyses

To evaluate differences in lake water and sediment chemistry between groups (the warmer, drier vs cooler, wetter areas), Welch’s two-sided t-tests (confidence interval = 0.95)

130

were used. The effect of oxygen status on response variables was assessed using one-way analysis of variance (ANOVA). Differences between variables within each oxygen status group were compared using Tukey multiple comparison of means. To evaluate the relationship between climate and lake anoxia, MAM and June temperatures were modeled as predictors of hypolimnion DO using ordinary least squares regression. Ordinary least squares regression models were also used to assess the relationship between anoxia and lake morphometry and stratification stability (to determine lake-specific predictors of anoxia), TP and DO (to determine the effect of anoxia on sediment P release), and algal biomass and DO (to determine lake ecological responses to anoxia). In assessing the relationship between TP and DO, three outlier

TP values from lakes SS903, SS906, and SS2 were identified by the influence they were exerting upon the model and removed. As TP measurement is a sensitive assay, these outliers were likely caused by measurement error. The removal of these three data points did not change the direction or significance of the relationship, but improved data normality and the amount of variation that was able to be explained. Lake sediment bar plots were constructed using

Microsoft Excel, while all other figures and statistical analyses were performed in base R

(version 3.5.1).

Results

Susceptibility of lake sediment P release

Lake surface sediment chemistry analyses indicated that total extractable Al and P content (AlTE and PTE, respectively) were similar between the two study areas (Figure 5.2). Total extractable Fe (FeTE) content of lake sediments in the cooler, wetter area was higher than that of those in the warmer, drier area (214 vs 75 μmol g−1 dry sediment, p = 0.03) as lakes in the

131

cooler, wetter area contained more Fe-oxides and minerogenic Fe than lakes in the warmer, drier area. Individual lake analysis demonstrated relative consistency within regions, though in warmer, drier area SS1590 had high Fe (due to Fe-oxides, organic Fe, and minerogenic Fe) and P

(due to metal oxide-associated and organic P).

Mean lake sediment Al:Fe was greater in the warmer, drier area vs the cooler, wetter area (0.36 vs. 0.09, p = 0.01) as was Al:P, though the difference was not significant (9.5 vs 4.9, p

= 0.12). However, both regions satisfied criteria for potential anoxic P release from sediments, because Al:Fe and Al:P were below 3 and 25, respectively (Kopáček et al 2005; Figure 5.3). Thus, both groups were identified as being susceptible to sediment P release if hypolimnetic anoxia were to occur. Lake sediments in the warmer, drier area contained more organic matter than those in the cooler, wetter area (20% vs 13%, p < 0.01; Figure 5.4).

Lake thermal stratification, anoxia, and predictor variables

No instances of hypoxia or anoxia were observed in lake hypolimnia in the cooler, wetter area during our sampling period. Among lakes that stratified in the warmer, drier area, there was variability in the occurrence of hypolimnetic anoxia across years. For instance, lake

SS1381’s hypolimnion was anoxic in July of 2015 when it was strongly stratified (i.e. shallow epilimnion and steep metalimnetic temperature gradient), but not in July of 2016, when stratification was weaker (i.e. deeper epilimnion and more gradual metalimnetic temperature gradient; Figure 5.5). Out of a total of 51 observations across 6 lakes from 2013-2019 in the warmer, drier area, there were 5 instances of hypolimnetic hypoxia and 22 instances of anoxia.

Lake hypolimnia in the cooler, wetter area were more oxygenated than lake hypolimnia in the warmer, drier area during summer (10 mg DO L-1 vs 4 mg DO L-1, p < 0.01; Figure 5.6).

132

0-2 cm sediment Al 0-2 cm sediment Fe 0-2 cm sediment P

Warmer, drier area

Colder, wetter area

Warmer, drier area drier Warmer,

wetter area wetter Colder, Colder, μmol g-1 dry sediment μmol g-1 dry sediment μmol g-1 dry sediment

Figure 5.2. Sequential lake sediment extractions for aluminum (Al), iron (Fe), and phosphorus (P) from lakes located in the warmer, drier area and those located in the cooler, wetter areas. Lake sediment extractions are averaged by region (top row) and are listed by individual lake (bottom two rows).

133

Al:Fe Al:P 4.0 30.0

3.5 25.0 3.0 20.0 2.5

2.0 15.0

1.5 10.0 1.0 5.0 0.5

0.0 0.0 Warm, dry Cool, wet Warm, dry Cool, wet

Figure 5.3. Ratios of lake sediment aluminum:iron (Al:Fe) and aluminum:phosphorus (Al:P) from lakes located in the warmer, drier area and lakes located in the cooler, wetter area. These ratios are indicative of P release potential of lake sediments under anoxic conditions if Al:Fe is below 3 and Al:P is below 25 (dashed lines).

134

Figure 5.4. Comparison of lake sediment organic matter (OM) content of lakes located in the warmer, drier area vs those in the cooler, wetter area.

Lake survey data indicated individual lakes were more susceptible to summer anoxia than others in the warmer, drier area, notably SS8, SS1381, and SS1590 (Figure 5.7). Variation of hypolimnetic DO among lakes located in the warmer, drier area could not be explained by lake morphometric or stratification stability indices such as GR (R2 = 0.03, p = 0.26), CEI (R2 =

0.02, p = 0.28), SSI (R2 = 0.00, p = 0.86) or BF (R2 = 0.01, p = 0.45, regressions not shown; see

Table 5.2 for lake index data). Thus, susceptibility to anoxia was not related to thermal stratification stability or to lake-specific factors that influence stratification like lake surface area, Zmax, mean depth, and volumetric profile.

135

Figure 5.5. Comparison of temperature (solid line) and dissolved oxygen (DO; dotted line) profiles in an anoxia- susceptible lake (SS1381) in July of 2015, when it was anoxic, and 2016, when stratification was weak, and the lake remained oxygenated. Hypolimnetic total phosphorus (TP) concentration was 8.5 μg L-1 in 2015 and 4.9 μg L-1 in 2016. Hypolimnetic chlorophyll a (Chl a) was 4.7 μg L-1 in 2015 and 2.6 μg L-1 in 2016.

136

Figure 5.6. Comparison of hypolimnetic dissolved oxygen (DO) concentration between lakes located in the warmer, drier area vs those located in the cooler, wetter area.

137

Figure 5.7. Individual lake hypolimnetic DO concentration in lakes located in the warmer, drier area, averaged across sampling years (2013-2019). Horizontal dashed line indicates hypoxia threshold of 4 mg DO L-1. Dotted line indicates anoxia threshold of 2 mg DO L-1.

138

Hypolimnetic DO concentration was positively related to average spring temperature but the relationship was not significant (R2 = 0.08, p = 0.06; not shown). The relationship was stronger when peak summer data were exclusively considered (R2 = 0.19, p = 0.01; Figure 5.8).

Thus, in years with warmer spring temperatures, summer hypolimnetic DO concentration was greater. Lake hypolimnetic DO concentration (whether June-July or July only) had no relationship to average June temperature (e.g. the effect of June temperature on June-July hypolimnetic DO concentration was not significant: R2 = 0.00, p = 0.89; not shown).

Lake water chemistry response to hypoxia and anoxia

TP was negatively correlated to DO concentration across both regions (R2 = 0.23, p <

0.01; Figure 5.9A). Mean TP in oxygenated lake hypolimnia was 6.0 µg L-1, 6.0 µg L-1 in hypoxic hypolimnia, and 11.0 µg L-1 in anoxic hypolimnia. Lake oxygen status was a significant predictor of TP (p < 0.01; Figure 5.9B), and the difference between anoxic and oxygenated hypolimnia TP was significant (p < 0.01), but differences between hypoxic-anoxic and oxygenated-hypoxic hypolimnetic TP were not (p = 0.20 and 1.00, respectively). This relationship between TP and oxygen status remained consistent even when lakes located in the warmer, drier area were exclusively considered, suggesting that spatial variability between study areas was not controlling lake TP concentration. Thus, lake oxygen status in the warmer, drier area was a significant predictor of TP (p < 0.01) and oxygenated, hypoxic, and anoxic hypolimnetic TP concentration was 5.6, 6.0, and 11.0 µg L-1, respectively. Similar to the analysis that included both areas, difference between oxygenated-anoxic hypolimnetic TP was significant amongst lakes in the warmer, drier area (p < 0.01), but differences between hypoxic-anoxic and oxygenated-hypoxic hypolimnia were not (p = 0.14 and 0.99, respectively).

139

Figure 5.8. Peak summer hypolimnetic dissolved oxygen (DO) concentration as a function of mean spring temperature (March-May). This analysis only included lakes located in the warmer, drier study area, which were susceptible to hypolimnetic anoxia.

140

A B

Figure 5.9. A) Lake hypolimnetic total phosphorus (TP) as a function of dissolved oxygen (DO). Vertical dashed line indicates hypoxia threshold of 4 mg DO L- 1. Vertical dotted line indicates anoxia threshold of 2 mg DO L-1; B) TP concentration of oxic, hypoxic, and anoxic hypolimnia.

141

A B

Figure 5.10. A) Lake hypolimnetic algal biomass (measured as Chlorophyll a, Chl a) as a function of dissolved oxygen (DO). Vertical dashed line indicates hypoxia threshold of 4 mg DO L-1. Vertical dotted line indicates anoxia threshold of 2 mg DO L-1; B) Algal biomass of oxic, hypoxic, and anoxic hypolimnia.

142

Ecological response to sediment P release

Algal biomass was higher in anoxic hypolimnia (R2 = 0.23, p < 0.01; Figure 5.10A). This relationship was consistent within warmer, drier area lakes (R2 = 0.15, p < 0.01) and cooler, wetter area lakes (R2 = 0.30, p < 0.01), suggesting the relationship was not controlled by regional differences in algal productivity. Hypolimnetic oxygen status was a significant predictor of algal biomass (p < 0.01; Figure 5.10B). Mean Chl a concentration of anoxic hypolimnia was

3.7 µg L-1, 3.1 µg L-1 in hypoxic epilimnia, and 2.1 µg L-1 in oxygenated hypolimnia. Differences in algal biomass were significant between anoxic and oxygenated hypolimnia (algal biomass was higher in anoxic hypolimnia; p < 0.01), but not between hypoxic-anoxic and oxygenated-hypoxic hypolimnia (p = 0.66 and 0.35, respectively). Hypolimnetic DO concentration did not affect metalimnetic algal biomass (R2 = 0.03, p = 0.12; not shown). Metalimnetic algal biomass was not significantly higher in anoxic lakes vs oxygenated ones.

Hypolimnetic nutrient limitation was not related to DO concentration, nor was it different amongst oxygen statuses, though there was an insignificant increase in P limitation between oxygenated and hypoxic hypolimnia (DIN:TP = 4.2 vs 11.1, p = 0.16; Figure 5.11a).

Mean anoxic hypolimnetic DIN:TP was 4.7. Mean DIN:TP ratios for all three oxygen statuses were above 3, indicating general P limitation (Bergström et al 2010). The pattern of moderately intensified metalimnetic P limitation tracked DIN availability (Figure 5.11b): oxygenated lakes had the lowest hypolimnetic DIN concentration (12 µg L-1) followed by anoxic lakes (33 µg L-1) and hypoxic lakes had the highest (70 µg L-1). The difference between hypoxic and oxygenated was significant (p < 0.01). The differences between oxygenated and anoxic or hypoxic and anoxic lake DIN were not significant (p = 0.06 and 0.09, respectively).

143

A B

Figure 5.11. A) Hypolimnetic nutrient limitation statuses (dissolved inorganic nitrogen to total phosphorus, or DIN:TP) according to oxygen status; B) Hypolimnetic DIN concentration according to oxygen status.

144

Discussion

Our findings demonstrate that the Arctic lakes in this study are susceptible to benthic- to-pelagic subsidies via sediment P release during anoxia. Importantly, lake hypolimnetic DO concentration was related to mean spring air temperature, suggesting a climate influence of lake hypoxia and anoxia. This is likely due to deeper mixed layers of lake surface water (i.e. weaker lake stratification) resulting from early lake ice-out in years with warm spring temperatures (Warner et al 2018). TP concentration and algal biomass increased in lake hypolimnia when DO concentration was low. Previous research has focused on changing external inputs of P into Arctic surface water ecosystems due to climate warming (Hobbie et al

1999; Frey and McClelland 2009). Here, we show that internal Arctic lake sediment P release plays an important role in P availability, is linked to lake thermal structure, and is an important part of climate-driven lake ecosystem effects.

Arctic lakes are generally low-nutrient systems, and P limitation is a common feature

(Schindler 1974; Rigler 1978; Gregory-Eaves et al 2000; Levine and Whalen 2001; Granéli et al

2004; Brutemark 2006). Lakes in the area of this study can be N or N and P co-limited (Hogan et al 2014; Whiteford et al 2016; Malik 2017). However, P-limitation is common in summer

(Brutemark et al 2006; Whiteford et al 2016; Malik et al 2017), with an intensification of P limitation in early summer (June) vs mid-summer (July; Burpee et al 2016). This is consistent with our findings of general P limitation in lake hypolimnia. Thus, P released from lake sediments is an important subsidy for Arctic lake pelagic environments that exhibit P limitation

(Schindler 1974; Rigler 1978; Gregory-Eaves et al 2000; Levine and Whalen 2001; Granéli et al

2004). Because P is a limiting nutrient, algae respond to it rapidly, as our data indicated by

145

higher Chl a concentration during anoxia. These results are similar to those found in eutrophic temperate lakes, where Chl a concentration increases coherently with sediment P release (e.g.

French and Petticrew 2007). The fact that Chl a concentration increased in the hypolimnia, and to a lesser extent the metalimnia of our study lakes suggest that the concentration of lake algal biomass could be deepened in the water column under anoxic conditions, due to greater nutrient availability in hypolimnia. In deep, oligotrophic alpine lakes, lake chlorophyll maxima move to deeper water with higher nutrient availability (Saros et al 2005). Movement of lake chlorophyll maxima has implications for zooplankton migration (Williamson et al. 1996; Winder et al. 2003) and water transparency (Morris et al 1995).

Lake sediment Al and Fe content and speciation indicated P release would occur from lake sediments during anoxia. Lake sediment chemistry was relatively similar between the two lake groups, save for Fe. The Fe content of lakes in the cooler, wetter area was higher than that of lakes in the warmer, drier area. This likely reflects greater hydrological conductivity of lakes in the cooler, wetter area compared to those in the warmer, drier area, as Fe is readily mobilized through the watershed as it complexes with DOC (Norton et al 2011). Similarly, the relatively high Fe content of SS1590, located in the warmer, drier area was likely due to hydrological connectivity, as it has a small inlet on its eastern side. The fact that lake sediment chemistry and P-release susceptibility are similar between the lake groups suggests that factors controlling the development of hypoxia and anoxia, not sediment chemistry alone, are the mechanisms controlling lake sediment P release.

Climatic controls of Arctic lake thermal stratification will be highly spatially variable.

Steep climatic gradients between inland and coastal tundra have been demonstrated for our

146

study area (Mernild et al 2015; Curtis et al 2018) as well as the pan-Arctic (Raynolds et al 2008).

This study took place across a climate gradient from the Greenland Ice Sheet margin to inland tundra. As the two target research areas in this study were only 40 km apart, we were able to demonstrate spatial variability of climate influence on Arctic lake thermal regimes. Across lakes in the warmer, drier area, the climate gradient disappeared, and variation of hypoxia and anoxia occurrence could no longer be accounted for. For instance, anoxia frequently occurred in lakes SS8 and SS1590, but only once in lake SS2 in 2013. There were no clear relationships between hypolimnetic DO concentration and lake GR, CEI, or SSI, indicating that lake bathymetry and morphometry does not determine anoxia in the warmer, drier area. For instance, CEI is a ratio of lake surface area to mean depth, and relevant to thermal stratification because surface area is an important determinant of stratification and thermocline depth

(Gorham and Boyce 1989). Lake SS1341 was rarely anoxic and had the lowest CEI amongst lakes in the warmer, drier area. Conversely, Lakes SS85 and SS2 rarely were anoxic and had the highest CEI values from the lakes located in the warmer, drier area. Thus, while climate differences between the two study areas likely influence differences in lake anoxia, localized controls of anoxia amongst lakes in the warmer, drier area remain unclear.

We found that Arctic lake hypolimnetic DO concentration was related to climate, because greater concentrations of DO in hypolimnia occurred when spring temperature was warmer. This relationship was especially strong for peak summer hypolimnetic DO, probably indicative of the time required for anoxia to occur following stratification in May and June. Over the time period investigated, warmer spring temperature resulted in weaker lake summer stratification. In their analysis involving three of the lakes in this study (SS2, SS85, and SS1590),

147

Warner et al (2018) found deeper mixing layers in years with early ice-out. The authors attributed this to differences in the length of the spring turnover period and the amount of solar insolation in early versus late ice-out years. In this case, when ice out is early (mid May in this area), this occurs before peak annual solar insolation, and air temperatures can be quite variable between mid May to mid June; these conditions may lead to a longer turnover period and slower onset of stratification (Olsen et al. 2012). When ice out is later (mid to late June in this area), this occurs at peak insolation and air temperatures are generally and consistently higher, leading to rapid onset of thermal stratification and reduced turnover. Lakes stratify shortly after ice-out with a short spring turnover period, which is important in determining the time, depth, and stability of stratification (Prowse et al 2006). Later ice-out occurs closer to the annual solar insolation peak, therefore lake surface water layers rapidly warm and set up a shallow epilimnion relative to earlier ice-out years (Warner et al 2018).

Lake ice-out trends respond directly and coherently to climate (Magnuson et al 2000) and across the Arctic there is a broad trend of earlier lake ice-out (Šmejkalová et al 2016). In

West Greenland, lake ice-out became 6 days earlier during 1973-2015 (Saros et al 2019). That trend, however, becomes more variable after 2010 due to changes in seasonality, suggesting the possibility of increased variability in summer hypolimnetic anoxia frequency. In areas of the

Arctic where lake stratification increases in duration and stability, sediment P release is likely to become more frequent.

Predictors of lake stratification stability include a suite of physical factors including lake surface area (Fee et al 1996), fetch (Gorham and Boyce 1989), depth, volume (Kraemer et al

2015), and light attenuation (Hocking and Straškraba 2013). Water transparency is an important

148

determinant of light attenuation and therefore lake thermal stratification (Fee et al 1996) and it is controlled by algal chlorophyll and DOC (Morris et al 1995). Permafrost can be a large reservoir of terrestrial DOC with the potential to alter lake thermal structure, but Arctic surface water DOC trends are spatially variable. In some Arctic areas permafrost DOC delivery to surface waters is increasing (west Siberia; Frey and Smith 2005) but in other areas it is declining

(Carey 2003; Kawahigashi et al 2004; Striegl et al 2005; McClelland et al 2007). Lake DOC is on average declining in West Greenland lakes, but rapid warming raises the question of how

Greenland lake thermal structure will change in the near future despite declining DOC. For instance, Saros et al (2016) determined that Greenland lake epilimnion depth is related to PAR attenuation (DOC quantity and quality are among the predictors of PAR attenuation) and epilimnion temperature. Saros et al (2016) demonstrated that Greenland lakes were more transparent (i.e. exhibited less PAR attenuation) in 2014 than in 2013, but epilimnion temperatures were warmer by 3 °C, resulting in equal or shallower epilimnion depths in 2014 vs

2013. Shallower epilimnion depths with warmer temperatures can result from stronger stratification and stability (King et al 1999; Warner et al 2018) and could increase the possibility of hypolimnetic anoxia.

This study takes place in an Arctic region that is experiencing overall drying (Mernild et al 2014) and decreased hydrological connectivity (Anderson et al 2017) though there is a positive trend in winter precipitation since 1994 (Saros et al 2019). Despite external P loads being important determinants for lake sediment P retention capacity (Hupfer et al 2008) and the amount of sediment P able to be released, it is expected that external P inputs into lakes in the Kangerlussuaq area are minimal save for spring snowmelt (McGowan et al 2017) and

149

aeolian transported glacial dust (Hogan et al 2014; Bullard et al 2018). The fact that Arctic lakes are susceptible to sediment P release under anoxic conditions is therefore an important finding, because increased internal sediment P release could offset decreased hydrological connectivity that some Arctic lakes, like those in Greenland, may face under climate warming and changing precipitation patterns (Anderson et al 2017).

The results of this study serve as an impetus for further investigation. For instance, we determined the correlation between anoxia and higher TP, but not causation. Thus, further experimentation is needed to confirm anoxia is causing P loading from lake sediments, or to determine if bacterial productivity in lake sediments during peak summer is causing anoxia and elevated TP is correlated to their biomass or metabolic activities (Hupfer et al 2008). Another alternative mechanism of P release could involve decomposition of organic P during and following sedimentation of organic material settling out of the water column (Hupfer et al

1995). Lastly, in low-productivity Arctic lakes, bulk algal biomass is an important metric of ecological response because it is the base of aquatic food webs and a point of entry for inorganic C into a lake’s C cycle. However, the presence of certain phytoplankton taxa and community turnover can indicate environmental changes such as nutrient enrichment (Saros et al. 2005; Williams et al 2016) or changes in lake thermal stratification (Saros et al 2016). Thus, future research should assess how internal Arctic lake P loading will affect algal community composition as well as abundance.

Reports of anoxia in Arctic lake ecosystems are not uncommon, but the concern with enhanced climate warming is that anoxia will increase in frequency and duration (Winder and

Sommer 2012; Vincent et al 2013). We found that Arctic lakes in the West Greenland Arctic are

150

susceptible to sediment P release during anoxia, and that TP and algal biomass increase when

DO concentration is low. Further, we determined that spring air temperature plays an important role in determining Arctic lake summer hypolimnetic DO concentrations. To the extent that Arctic lake stratification increases in response to current trends of warming, our findings suggest internal P cycles in Arctic lakes are likely to respond. This study highlights another dimension of Arctic lake sensitivity to climate and furthers understanding of Arctic lake biogeochemical and ecological responses to rapid environmental warming.

151

CHAPTER 6

CONCLUSION

The objectives of this dissertation were to determine how environmental change is altering nutrient subsidization and nutrient cycling in remote Arctic and alpine lakes, and how lake ecosystems respond to those changes. This work demonstrated that nutrient subsidies are an important dimension of rapid environmental change that is occurring in remote Arctic and alpine areas. With ongoing abrupt climate warming in Arctic and alpine regions, together with increased N deposition rates, this work provides a survey of how remote lake ecosystems often located in pristine areas are responding to changing nutrient subsidization and cycling patterns.

Chapter 2 reviewed cross-ecosystem subsidies in Arctic and alpine lakes. In this review, I introduced the concept of ecosystem subsidies to lakes from the cryosphere, atmosphere, terrestrial environment, and animals. I summarized and compared Arctic and alpine lake subsidies and their drivers through a series of conceptual diagrams. The literature review highlighted current research trends and identified knowledge gaps. A notable knowledge gap that I pointed out involves assessing the factors that determine lake ecological sensitivity and responses to large-scale nutrient subsidization. This knowledge gap is directly related to my dissertation research, and I addressed this topic in Chapter 4. I also made the argument that

Arctic and alpine lake ecosystems are ideal systems to study the effects of cross-ecosystem subsidization because they are ecologically sensitive, they have easily defined ecosystem boundaries, and many standard methods exist to quantify the flux of subsidies and their ecological effects. This review therefore provides valuable context for the motivation of the following dissertation chapter research topics.

152

In Chapter 3, I assessed cryosphere-to-lake subsidization in Arctic lakes. I found that total phosphorus (TP) was 6 times higher in glacially fed lakes adjacent to the Greenland Ice

Sheet in West Greenland compared to nearby snow and groundwater fed lakes. Additionally, dissolved inorganic nitrogen (DIN) was twice as high in the glacially fed lakes compared to snow and groundwater fed lakes, though there was not a significant difference. Stepwise surface sediment extraction and extracellular enzyme activities called into question the bioavailability of the glacial meltwater-derived TP. For instance, extracellular enzyme activities suggested that lake microbes were limited by phosphorus (P) and sediment extractions determined that the majority of P associated with glacial flour suspended in glacial meltwater was minerogenic, therefore recalcitrant. Differences in diatom community composition and higher algal biomass in Greenland glacially fed lakes compared to snow and groundwater fed lakes may therefore be attributable to moderately higher DIN concentrations in glacially fed lakes. Supporting this conclusion was the high abundance of two nitrogen (N) associated diatom taxa, Fragilaria tenera and Discostella stelligera, in glacially fed lakes. This study highlighted that the quality and not just the quantity of cross-ecosystem subsidies is ecologically important. This study contributed to the growing body of literature that reveals spatial variability of glacial meltwater effects on lake ecosystems, which is a prevalent subsidy type across both Arctic and alpine landscapes.

In Chapter 4, I evaluated the predictors of lake ecosystem sensitivity to nutrient subsidies, specifically atmospheric N deposition. Using lake DIN and algal biomass as focal variables to assess sensitivity, I found that at large scales (i.e. the Northern hemisphere) lake

DIN and algal biomass were related to large-scale processes such as N deposition rates.

153

However, gradients in N deposition disappeared at smaller, localized scales (i.e. individual mountain ranges) and other lake and watershed features such as land cover, bedrock geology, and lake depth became important predictor variables. Thus, I determined that scale is important in determining the predictors of lake ecological sensitivity. Sensitivity to N deposition differed between North America and European alpine lakes: North American lake algal biomass was positively related to N-associated factors such as N deposition rate and N availability

(though there was local variation), whereas European lakes were consistently related to P- associated factors such as TP. This distinction is likely reflective of prolonged N enrichment of

European alpine lakes due to high N deposition rates (Kopáček et al 2005; Bergström 2006;

Elser et al 2009). In sum, this study demonstrates that lake environmental factors modulate ecological response to nutrient subsidies. Therefore, lake ecological sensitivity to a nutrient subsidy such as atmospheric N deposition can be spatially variable. This study serves to complement current critical-load based lake management approaches and increase predictive capabilities of agencies that wish to identify lakes which are ecologically vulnerable to nutrient inputs.

Chapter 5 demonstrated that internal Arctic lake nutrient cycling is another dimension of remote lake sensitivity to abrupt climate change. Climate warming can increase stability and duration of lake thermal stratification. Lake hypolimnion dissolved oxygen (DO) depletion is an important consequence of increased stratification strength because it limits mixing of surface and deep-water layers. Hypolimnetic anoxia promotes dissociation of P from sediment aluminum (Al)- and iron (Fe)-oxides. We demonstrated that Arctic lake sediments in West

Greenland are susceptible to P release during anoxia based on sediment extraction chemistry.

154

We also demonstrated that anoxia readily occurs in a group of lakes located in a warm, dry area that exhibit relatively stable stratification compared to lakes located in a cooler, wetter area proximal to the ice sheet. Mean spring temperature was a positive predictor variable of mid- summer hypolimnion DO concentration, suggesting that cooler springs were associated with summer anoxia. DO concentration was a negative predictor of TP, confirming that Arctic lake anoxia promotes sediment P release. DO was also a negative predictor of hypolimnetic algal biomass, suggesting that Arctic lake sediment P release can have important ecological impacts.

While other studies have focused on external inputs of P subsidies to Arctic lake surface waters

(Hobbie et al 1999; Frey and McClelland 2009; Bullard et al 2013; Hawkings et al 2016; Burpee et al 2018) this study highlights the ecological importance of climate-mediated internal P cycling in Arctic lake ecosystems.

My dissertation research assesses ways in which environmental change affects lake nutrient subsidies, cycling, and ecological responses. While several research questions were addressed in this document, new questions are raised for future research. Chapter 2 in particular identified several outstanding research priorities regarding Arctic and alpine lake nutrient subsidies, including the broad assessment of factors that determine lake sensitivity and responses to large-scale drivers, such as warming, permafrost thaw, or atmospheric deposition.

This question was pertinent for the research topics of Chapters 4 and 5. While I addressed this research priority explicitly in Chapter 4, my analysis was limited to the consideration of N deposition, but P deposition rates are increasing in alpine areas and have already had ecological effects on alpine lakes in the North America including increased phytoplankton and zooplankton biomass (Brahney et al 2014). My findings that European lakes are currently more

155

sensitive to P than N inputs also highlight the importance of a comprehensive analysis that assesses the factors determining lake sensitivity to P deposition. Further, Chapter 4’s analysis used data collected within the last 35 years. Paleoecological data from lake sediment cores would allow for the assessment of past lake sensitivity (for instance, by measurements of diatom community turnover rates) in lakes that are no longer sensitive to N deposition due to environmental N saturation or relief from biological N limitation. An analysis of lake factors that influenced past ecological sensitivity and responses to N deposition would increase predictive power of current lake sensitivity in areas that are still sensitive to N deposition.

Lake sensitivity to warming represents another question requiring further attention;

Chapter 5 identified that cooler spring temperatures were associated with summer Arctic lake anoxia, but I was not able to explain anoxia frequency of individual lakes based on the metrics that I considered. Thus, it will be important to identify lake-specific factors that determine susceptibility to anoxia to allow better predictability and understanding of climate warming impacts on lake P cycling across the Arctic. The relationship between cooler spring temperatures and summer Arctic lake anoxia also suggests that changing seasonality, and not annual average or summer climate warming, is more important for Arctic lake stratification

(Warner et al 2018) and P cycling dynamics. This relationship could be explored in more detail, for instance by evaluating the influence of ice-out dates on summer lake anoxia. Long-term ecological effects of lake sediment P release in Arctic lakes that consistently experience anoxia

(e.g. SS8 and SS1590 in Chapter 5) could also be evaluated with paleoecological analysis to better understand ecological trajectories of Arctic lakes that experience more frequent anoxia due to changing climate and seasonality.

156

Arctic and alpine lakes are situated in areas experiencing rapid environmental change including climate warming and atmospheric nutrient deposition. In response, nutrient cycling and delivery to lakes from other ecosystems are changing. Together, my dissertation chapters provide a survey of such changes, highlight the importance of Arctic and alpine lakes to ecosystem subsidy research, and address gaps of knowledge surrounding Arctic and alpine lake biogeochemical and ecological responses to climate warming and altered nutrient patterns. I demonstrate that nutrient subsidy quality is important in Arctic lakes receiving glacial meltwater, that scale-dependent lake and catchment features play an important role in determining ecological sensitivity to atmospheric N deposition across the Northern

Hemisphere, and that internal nutrient cycling dynamics of cold, oligotrophic Arctic lakes are responsive to climate with ecological implications. All of these insights add to our general understanding of how lakes located in cold, remote areas will change over the coming century with current projections of climate warming and atmospheric deposition patterns.

157

REFERENCES

Abbott BW, Jones JB, Godsey SE, Larouche JR, Bowden WB. Patterns and persistence of hydrologic carbon and nutrient export from collapsing upland permafrost. Biogeosciences. 2015; 12(12): 3725-40.

Abbott BW, Larouche JR, Jones Jr JB, Bowden WB, Balser AW. Elevated dissolved organic carbon biodegradability from thawing and collapsing permafrost. Journal of Geophysical Research: Biogeosciences. 2014; 119(10):2049-63.

Adrian R, O'Reilly CM, Zagarese H, Baines SB, Hessen DO, Keller W, Livingstone DM, Sommaruga R, Straile D, Van Donk E, Weyhenmeyer GA. Lakes as sentinels of climate change. Limnology and oceanography. 2009; 54(6 part 2): 2283-97.

AMAP Assessment 2002: Persistent organic pollutants, Arctic Monitoring and Assessment Programme (AMAP), Oslo, Norway, 2002.

American Public Health Association (APHA): Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association, Water Environment Federation, Washington, D.C, 2000.

Anderson, NJ and Stedmon, CA. The effect of evapoconcentration on dissolved organic carbon concentration and quality in lakes of SW Greenland. Freshwater Biology. 2007: 52(2): 280-289.

Anderson NJ, Curtis CJ, Whiteford EJ, Jones VJ, McGowan S, Simpson GL, Kaiser J. Regional variability in the atmospheric nitrogen deposition signal and its transfer to the sediment record in Greenland lakes. Limnology and Oceanography. 2018; 63(5): 2250-65.

Anderson NJ, Harriman R, Ryves DB and Patrick ST. Dominant factors controlling variability in the ionic composition of West Greenland lakes. Arctic, Antarctic, and Alpine Research. 2001; 33(4): 418-425.

Anderson NJ, Saros JE, Bullard JE, Cahoon SM, McGowan S, Bagshaw EA, Barry CD, Bindler R, Burpee BT, Carrivick JL, Fowler RA. The Arctic in the twenty-first century: Changing biogeochemical linkages across a paraglacial landscape of Greenland. BioScience. 2017; 67(2): 118-33.

Anisimov O, Reneva S. Permafrost and changing climate: the Russian perspective. AMBIO: A Journal of the Human Environment. 2006; 35(4): 169-76.

Arens SJ, Sullivan PF, Welker JM. Nonlinear responses to nitrogen and strong interactions with nitrogen and phosphorus additions drastically alter the structure and function of a high arctic ecosystem. Journal of Geophysical Research: Biogeosciences. 2008; 113(G3), doi:10.1029/2007JG000508.

158

Arnalds O, Dagsson-Waldhauserova P, Olafsson H. The Icelandic volcanic aeolian environment: Processes and impacts—A review. Aeolian Research. 2016; 20: 176-95.

Arnett HA, Saros JE and Mast MA. A caveat regarding diatom-inferred nitrogen concentrations in oligotrophic lakes. Journal of Paleolimnology. 2012; 47(2): 277-291.

Arts MT, Kohler CC. Health and condition in fish: the influence of lipids on membrane competency and immune response. In Lipids in Aquatic Ecosystems 2009 (pp. 237-256). Springer, New York, NY.

Bao Y, Bräuning A, Yafeng S. Late Holocene temperature fluctuations on the Tibetan Plateau, Quaternary Science Reviews. 2003; 22(21-22): 2335-2344.

Baron JS, Driscoll CT, Stoddard JL, Richer EE. Empirical critical loads of atmospheric nitrogen deposition for nutrient enrichment and acidification of sensitive US lakes. Bioscience. 2011; 61(8): 602-13.

Baron J, Norton SA, Beeson DR, Herrmann R. Sediment diatom and metal stratigraphy from Rocky Mountain lakes with special reference to atmospheric deposition. Canadian Journal of Fisheries and Aquatic Sciences. 1986; 43(7): 1350-62.

Baron JS, Rueth HM, Wolfe AP, Nydick KR, Allstott EJ, Minear JT, Moraska B. Ecosystem responses to nitrogen deposition in the Colorado Front Range. Ecosystems. 2000; 3: 352–368.

Barrett K, Anderson WB, Wait DA, Grismer LL, Polis GA, Rose MD. Marine subsidies alter the diet and abundance of insular and coastal lizard populations. Oikos. 2005; 109(1): 145-53.

Battarbee RW. Diatom Analysis. In Berglund, B. E. (ed.), Handbook of Holocene Palaeoecology and Palaeohydrology 1986 Wiley: 527-570.

Battarbee RW, Jones VJ, Flower RJ, Cameron NG, Bennion H, Carvalho L, Juggins S. Diatoms, in Tracking Environmental Change Using Lake Sediments, eds. J. P. Smol, H. J. Birks, W. M. Last, Springer, Dordrecht, Netherlands, 2002, 155-202.

Baustian MM, Hansen GJA, de Kluijver A, Robinson K, Henry EN, Knoll LB and Carey CC. Linking the bottom to the top in aquatic ecosystems: mechanisms and stressors of benthic-pelagic coupling. In Eco-DAS X Symposium Proceedings 2014 (Vol. 2014, pp. 25-47).

Benstead JP, Deegan LA, Peterson BJ, Huryn AD, Bowden WB, Suberkropp K, Buzby KM, Green AC, Vacca JA. Responses of a beaded Arctic stream to short-term N and P fertilisation. Freshwater Biology. 2005; 50(2): 277-90.

159

Bergström AK. The use of TN:TP and DIN:TP ratios as indicators for phytoplankton nutrient limitation in oligotrophic lakes affected by N deposition. Aquatic Sciences. 2010; 72: 277-281.

Bergström AK, Jansson M. Atmospheric nitrogen deposition has caused nitrogen enrichment and eutrophication of lakes in the northern hemisphere. Global Change Biology. 2006; 12(4): 635-43.

Bergström AK, Jonsson A, Jansson M. Phytoplankton responses to nitrogen and phosphorus enrichment in unproductive Swedish lakes along a gradient of atmospheric nitrogen deposition. Aquatic Biology. 2008; 4(1): 55-64.

Betts-Piper AM, Zeeb BA, Smol JP. Distribution and autecology of chrysophyte cysts from high Arctic Svalbard lakes: preliminary evidence of recent environmental change. Journal of Paleolimnology. 2004; 31(4): 467-81.

Bhatia MP, Kujawinski EB, Das SB, Breier CF, Henderson PB, Charette MA. Greenland meltwater as a significant and potentially bioavailable source of iron to the ocean. Nature Geoscience. 2013; 6(4): 274.

Birks HJ, Jones VJ, Rose NL. Recent environmental change and atmospheric contamination on Svalbard as recorded in lake sediments–an introduction. Journal of Paleolimnology. 2004; 31(4): 403-10.

Bobbink R, Hicks K, Galloway J, Spranger T, Alkemade R, Ashmore M, Bustamante M, Cinderby S, Davidson E, Dentener F, Emmett B. Global assessment of nitrogen deposition effects on terrestrial plant diversity: a synthesis. Ecological applications. 2010; 20(1): 30-59.

Bogard MJ, Kuhn CD, Johnston SE, Striegl RG, Holtgrieve GW, Dornblaser MM, Spencer RG, Wickland KP, Butman DE. Negligible cycling of terrestrial carbon in many lakes of the arid circumpolar landscape. Nature Geoscience. 2019; 12(3): 180.

Bogdal C, Nikolic D, Lüthi MP, Schenker U, Scheringer M, Hungerbühler K. Release of legacy pollutants from melting glaciers: model evidence and conceptual understanding, Environmental Science & Technology, 2010; 44(11): 4063-4069.

Bonalumi M, Anselmetti FS, Kaegi R, Wüest A. Particle dynamics in high-Alpine proglacial reservoirs modified by pumped-storage operation. Water Resources Research. 2011; 47(9): DOI: 10.1029/2010WR010262.

Bosson E, Lindborg T, Berglund S, Gustafsson LG, Selroos JO, Laudon H, Claesson LL and Destouni G. Water balance and its intra-annual variability in a permafrost catchment: hydrological interactions between catchment, lake and talik. Hydrology and Earth System Sciences Discussions. 2013; 10(7): 9271-9308.

160

Braegelman SD. Seasonality of some Arctic Alaskan chironomids (doctoral dissertation), North Dakota State University, 2015.

Brahney J, Ballantyne AP, Kociolek P, Spaulding S, Otu M, Porwoll T, Neff JC. Dust mediated transfer of phosphorus to alpine lake ecosystems of the Wind River Range, Wyoming, USA. Biogeochemistry. 2014; 120(1-3): 259-78.

Brahney J, Ballantyne AP, Sievers C, Neff JC. Increasing Ca2+ deposition in the western US: The role of mineral aerosols. Aeolian Research. 2013; 10: 77-87.

Brahney J, Mahowald N, Ward DS, Ballantyne AP, Neff JC. Is atmospheric phosphorus pollution altering global alpine Lake stoichiometry? Global Biogeochemical Cycles. 2015; 29(9): 1369-83.

Breiman L. Random forests. Machine learning. 2001; 45(1): 5-32.

Brett MT, Müller-Navarra DC. The role of highly unsaturated fatty acids in aquatic foodweb processes. Freshwater Biology. 1997; 38(3): 483-99.

Brittain JE, Milner AM. Ecology of glacier-fed rivers: current status and concepts. Freshwater Biology. 2001; 46(12): 1571-8.

Brodersen KP and Anderson NJ. Subfossil insect remains (Chironomidae) and lake-water temperature inference in the Sisimiut–Kangerlussuaq region, southern West Greenland. Weather. 2000; 32(17): 36.

Brown GH, Tranter M, Sharp MJ. Experimental investigations of the weathering of suspended sediment by alpine glacial meltwater. Hydrological Processes. 1996; 10(4): 579-97.

Brutemark A, Rengefors K and Anderson NJ. An experimental investigation of phytoplankton nutrient limitation in two contrasting low arctic lakes. Polar Biology. 2006: 29(6): 487-494.

Bullard JE. Contemporary glacigenic inputs to the dust cycle. Earth Surface Processes and Landforms. 2013; 38(1): 71-89.

Bullard JE. The distribution and biogeochemical importance of high-latitude dust in the Arctic and Southern Ocean-Antarctic regions. Journal of Geophysical Research: Atmospheres. 2017; 122(5): 3098-103.

Bullard JE and Austin MJ. Dust generation on a proglacial floodplain, West Greenland. Aeolian Research. 2011; 3(1): 43-54.

Bullard JE, Baddock M, Bradwell T, Crusius J, Darlington E, Gaiero D, Gasso S, Gisladottir G, Hodgkins R, McCulloch R, McKenna-Neuman C. High-latitude dust in the Earth system. Reviews of Geophysics. 2016; 54(2): 447-85.

161

Bullard JE, Mockford T. Seasonal and decadal variability of dust observations in the Kangerlussuaq area, west Greenland. Arctic, Antarctic, and Alpine Research. 2018; 50(1): DOI: 10.1080/15230430.2017.1415854.

Bultman H, Hoekman D, Dreyer J, Gratton C. Terrestrial deposition of aquatic insects increases plant quality for insect herbivores and herbivore density. Ecological entomology. 2014; 39(4): 419-26.

Burckle LH, Gayley RI, Ram M, Petit JR. Diatoms in Antarctic ice cores: some implications for the glacial history of Antarctica, Geology. 1988; 16(4): 326-329.

Burkhart JF, Bales RC, McConnell JR, Hutterli MA. Influence of North Atlantic Oscillation on anthropogenic transport recorded in northwest Greenland ice cores. Journal of Geophysical Research: Atmospheres. 2006; 111(D22): DOI: 10.1029/2005JD006771.

Burns, D.A., 2003. Atmospheric nitrogen deposition in the Rocky Mountains of Colorado and southern Wyoming-A review and new analysis of past study results. Atmos. Environ. 2003; 37: 921–932.

Burns DA, Blett T, Haeuber R, Pardo LH. Critical loads as a policy tool for protecting ecosystems from the effects of air pollutants. Frontiers in Ecology and the Environment. 2008; 6(3): 156-9.

Burpee BT, Anderson D, Saros JE. Assessing ecological effects of glacial meltwater on lakes fed by the Greenland Ice Sheet: The role of nutrient subsidies and turbidity. Arctic, Antarctic, and Alpine Research. 2018; 50(1): DOI: 10.1080/15230430.2017.1420953.

Burpee B, Saros JE, Northington RM and Simon K S. Microbial nutrient limitation in Arctic lakes in a permafrost landscape of southwest Greenland. Biogeosciences. 2016; 13(2): 365-374.

Butler MG. Emergence phenologies of some arctic Alaskan Chironomidae, in Chironomidae: Ecology, Systematics, Cytology and Physiology, ed. D. A. Murray, Pergamon Press, Oxford, UK, 1980, 307-314.

Butler TJ, Likens GE, Vermeylen FM, Stunder BJ. The impact of changing nitrogen oxide emissions on wet and dry nitrogen deposition in the northeastern USA, Atmospheric Environment. 2005; 39(27): 4851-4862.

Camarero L and Catalan J. Atmospheric phosphorus deposition may cause lakes to revert from phosphorus limitation back to nitrogen limitation. Nature Communications. 2012; 3: 1118.

Caraco N, Cole J. When terrestrial organic matter is sent down the river: the importance of allochthonous carbon inputs to the metabolism of lakes and rivers. Food webs at the landscape level. 2004: 301-16.

162

Carey SK. Dissolved organic carbon fluxes in a discontinuous permafrost subarctic alpine catchment. Permafrost and Periglacial Processes. 2003; 14(2): 161-71.

Carlsson P, Caron DA. Seasonal variation of phosphorus limitation of bacterial growth in a small lake. Limnology and Oceanography. 2001; 46(1): 108-20.

Carrivick JL, and Tweed FS. Proglacial lakes: character, behavior and geological importance. Quaternary Science Reviews. 2013; 78: 34–52.

Catalan J, Pla-Rabés S, Wolfe AP, Smol JP, Rühland KM, Anderson NJ, Kopáček J, Stuchlík E, Schmidt R, Koinig KA, Camarero L. Global change revealed by palaeolimnological records from remote lakes: a review, Journal of Paleolimnology. 2013; 49(3): 513-535.

Clow DW, Nanus L, Huggett B. Use of regression-based models to map sensitivity of aquatic resources to atmospheric deposition in Yosemite National Park, USA. Water Resources Research. 2010; 46(9): DOI: 10.1029/2009WR008316.

Cole JJ, Pace ML, Carpenter SR and Kitchell JF. Persistence of net heterotrophy in lakes during nutrient addition and food web manipulations. Limnology and Oceanography. 2000; 45(8): 1718-1730.

Commission for the Geological Map of the World, Asch K and Bellenberg S.. The 1: 5 million international geological map of Europe and adjacent areas (IGME 5000). Bundesanstalt für Geowissenschaften und Rohstoffe 2005.

Conroy JL, Overpeck JT, Cole JE, Liu KB, Wang L, Ducea MN. Dust and temperature influences on glaciofluvial sediment deposition in southwestern Tibet during the last millennium, Global and Planetary Change. 2013; 107: 132-144.

Cook J, Stuparyk BR, Johnsen MA and Vinebrooke RD. Concordance of chemically inferred and assayed nutrient limitation of phytoplankton along a depth gradient of alpine lakes in the Canadian Rockies. Aquatic Sciences. 2020; 82(1): 17.

Cooke GD, EB Welch SA & PR Newroth. Restoration and Management of Lakes and Reservoirs, 2nd ed. 1993 Lewis Publishers, Boca Raton.

Cooke GD, EB Welch, SA Peterson & SA Nichols. Restoration and Management of Lakes and Reservoirs, 3rd ed. 2005 Taylor and Francis, Boca Raton.

Coppin P, Lambin E, Jonckheere I, Muys B. Digital change detection methods in natural ecosystem monitoring: a review, in Analysis of Multi-Temporal Remote Sensing Images, eds. L. Bruzzone and P. Smits, Proceedings of the First International Workshop on Multitemp 2001, University of Trento, Italy, 2002, 3-36.

163

Côté G, Pienitz R, Velle G, Wang X. Impact of geese on the limnology of lakes and ponds from Bylot Island (Nunavut, Canada). International Review of Hydrobiology. 2010; 95(2): 105-29.

Creed IF, Bergström AK, Trick CG, Grimm NB, Hessen DO, Karlsson J, Kidd KA, Kritzberg E, McKnight DM, Freeman EC, Senar OE. Global change-driven effects on dissolved organic matter composition: Implications for food webs of northern lakes. Global change biology. 2018; 24(8): 3692-714.

Cremer H and Wagner B. Planktonic diatom communities in High Arctic lakes (Store Koldewey, Northeast Greenland). Canadian Journal of Botany. 2004; 82(12): 1744-1757.

Crusius J, Schroth AW, Gasso S, Moy CM, Levy RC, Gatica M. Glacial flour dust storms in the Gulf of Alaska: Hydrologic and meteorological controls and their importance as a source of bioavailable iron, Geophysical Research Letters. 2011; 38(6): L06602, DOI: 10.1029/2010GL046573.

Curtis CJ, Kaiser J, Marca A, Anderson NJ, Simpson G, Jones V and Whiteford E. Spatial variations in snowpack chemistry, isotopic composition of NO3− and nitrogen deposition from the ice sheet margin to the coast of western Greenland. Biogeosciences. 2018; 15(2): 529-550.

Daly GL, Wania F. Organic contaminants in mountains. Environmental science & technology. 2005; 39(2): 385-98.

Das B, Vinebrooke RD, Sanchez-Azofeifa A, Rivard B and Wolfe AP. Inferring sedimentary chlorophyll concentrations with reflectance spectroscopy: a novel approach to reconstructing historical changes in the trophic status of mountain lakes. Canadian Journal of Fisheries and Aquatic Sciences. 2005; 62(5): 1067-1078.

De'ath G and Fabricius KE. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology. 2000; 81(11): 3178-3192. del Giorgio PA, Newell RE. Phosphorus and DOC availability influence the partitioning between bacterioplankton production and respiration in tidal marsh ecosystems. Environmental microbiology. 2012; 14(5): 1296-307.

Donarummo Jr J, Ram M, Stoermer EF. Possible deposit of soil dust from the 1930's US dust bowl identified in Greenland ice, Geophysical Research Letters. 2003; 30(6): 1269, DOI: 10.1029/2002GL016641.

Dorfman JM, Stoner JS, Finkenbinder MS, Abbott MB, Xuan C, St-Onge G. A 37,000-year environmental magnetic record of aeolian dust deposition from Burial Lake, Arctic Alaska, Quaternary Science Reviews. 2015; 128: 81-97.

164

Drake TW, Holmes RM, Zhulidov AV, Gurtovaya T, Raymond PA, McClelland JW and Spencer RG. Multidecadal climate-induced changes in Arctic tundra lake geochemistry and geomorphology. Limnology and Oceanography. 2019; 64(S1): S179-S191.

Dreyer J, Hoekman D, Gratton C. Lake-derived midges increase abundance of shoreline terrestrial arthropods via multiple trophic pathways. Oikos. 2012; 121(2): 252-8.

Dugan HA, Lamoureux SF, Lewis T, Lafrenière MJ. The impact of permafrost disturbances and sediment loading on the limnological characteristics of two high Arctic lakes. Permafrost and Periglacial Processes. 2012; 23(2): 119-26.

Elser JJ, Andersen T, Baron JS, Bergström AK, Jansson M, Kyle M, Nydick KR, Steger L, Hessen DO. Shifts in lake N:P stoichiometry and nutrient limitation driven by atmospheric nitrogen deposition. Science. 2009; 326(5954): 835-7.

Elser JJ, Marzolf ER, Goldman CR. Phosphorus and nitrogen limitation of phytoplankton growth in the freshwaters of North America: a review and critique of experimental enrichments. Canadian Journal of fisheries and aquatic sciences. 1990; 47(7): 1468-77.

Epanchin PN, Knapp RA, Lawler SP. Nonnative trout impact an alpine-nesting bird by altering aquatic-insect subsidies. Ecology. 2010; 91(8): 2406-15.

Fee EJ, Hecky RE, Kasian SEM and Cruikshank DR. Effects of lake size, water clarity, and climatic variability on mixing depths in Canadian Shield lakes. Limnology and Oceanography. 1996; 41(5): 912-920.

Fegel TS, Baron JS, Fountain AG, Johnson GF, Hall EK. The differing biogeochemical and microbial signatures of glaciers and rock glaciers. Journal of Geophysical Research: Biogeosciences. 2016; 121(3): 919-32.

Fellman JB, Hood E, Raymond PA, Hudson J, Bozeman M, Arimitsu M. Evidence for the assimilation of ancient glacier organic carbon in a proglacial stream food web. Limnology and Oceanography. 2015; 60(4): 1118-28.

Fenn ME, Baron JS, Allen EB, Rueth HM, Nydick KR, Geiser L, Bowman WD, Sickman JO, Meixner T, Johnson DW, Neitlich P. Ecological effects of nitrogen deposition in the western United States. BioScience. 2003; 53(4): 404-20.

Findlay SE, Sinsabaugh RL, Sobczak WV and Hoostal M. Metabolic and structural response of hyporheic microbial communities to variations in supply of dissolved organic matter. Limnology and Oceanography. 2003; 48(4): 1608-1617.

Finlay JC, Vredenburg VT. Introduced trout sever trophic connections in watersheds: consequences for a declining amphibian. Ecology. 2007; 88(9): 2187-98.

165

Fischer H, Wagenbach D, Kipfstuhl J. Sulfate and nitrate firn concentrations on the Greenland ice sheet: 1. Large-scale geographical deposition changes. Journal of Geophysical Research: Atmospheres. 1998; 103(D17): 21927-34.

Fowler RA, Saros JE, Osburn CL. Shifting DOC concentration and quality in the freshwater lakes of the Kangerlussuaq region: An experimental assessment of possible mechanisms. Arctic, Antarctic, and Alpine Research. 2018; 50(1), DOI: 10.1080/15230430.2018.1436815.

Fox AD, Madsen J, Boyd H, Kuijken E, Norriss DW, Tombre IM, Stroud DA. Effects of agricultural change on abundance, fitness components and distribution of two arctic-nesting goose populations. Global Change Biology. 2005; 11(6): 881-93.

Fox, J. and Weisberg, S. An {R} Companion to Applied Regression. 2nd ed. 2011, Thousand Oaks, CA: Sage.

French TD and Petticrew EL. Chlorophyll a seasonality in four shallow eutrophic lakes (northern British Columbia, Canada) and the critical roles of internal phosphorus loading and temperature. Hydrobiologia. 2007; 575(1): 285-299.

Frey KE, McClelland JW. Impacts of permafrost degradation on arctic river biogeochemistry. Hydrological Processes: An International Journal. 2009; 23(1): 169-82.

Frey KE, Smith LC. Amplified carbon release from vast West Siberian peatlands by 2100. Geophysical Research Letters. 2005; 32(9), DOI: 10.1029/2004GL022025.

Fritz KA, Whiles MR, Trushenski JT. Subsidies of long-chain polyunsaturated fatty acids from aquatic to terrestrial environments via amphibian emergence. Freshwater Biology. 2019; 64(5): 832-42.

Galloway JN, Dentener FJ, Capone DG, Boyer EW, Howarth RW, Seitzinger SP, Asner GP, Cleveland CC, Green PA, Holland EA, Karl DM. Nitrogen cycles: past, present, and future. Biogeochemistry. 2004; 70(2): 153-226.

Gardarsson A, Einarsson A, Jonsson E, Gislason GM, Olafsson JS, Hrafnsdottir T, Ingvason HR. Population indices of Chironomidae at Myvatn over 20 years, 1977–1996. Myvatn, Iceland: Myvatn Research Station. 2000.

Gardner EM, McKnight DM, Lewis Jr WM and Miller MP. Effects of nutrient enrichment on phytoplankton in an alpine lake, Colorado, USA. Arctic, Antarctic, and Alpine Research. 2008; 40(1): 55-64.

166

Garrity CP and Soller DR. 2009, Database of the Geologic Map of North America; adapted from the map by J.C. Reed, Jr. and others (2005): U.S. Geological Survey Data Series 424 [https://pubs.usgs.gov/ds/424/].

Gayley RI, Ram M, Stoermer EF. Seasonal variations in diatom abundance and provenance in Greenland ice, Journal of Glaciology. 1989; 35(120): 290-292.

German DP, Weintraub MN, Grandy AS, Lauber CL, Rinkes ZL, and Allison SD. Optimization of hydrolytic and oxidative enzyme methods for ecosystem studies. Soil Biology and Biochemistry. 2011; 43(7): 1387-1397.

Gerten D and Adrian R. Differences in the persistency of the North Atlantic Oscillation signal among lakes. Limnology and Oceanography. 2001; 46(2): 448-455.

Gervais F. Light-dependent growth, dark survival, and glucose uptake by cryptophytes isolated from a freshwater chemocline. Journal of Phycology. 1997; 33(1): 18-25.

Gibbons KJ. Effect of temperature on phosphorus release from anoxic western Lake Erie sediments (Doctoral dissertation, University of Toledo) 2015.

Gilbert R, Crookshanks S. Sediment waves in a modern high-energy glacilacustrine environment. Sedimentology. 2009; 56(3): 645-59.

Gladyshev MI, Arts MT, Sushchik NI. Preliminary estimates of the export of omega-3 highly unsaturated fatty acids (EPA+ DHA) from aquatic to terrestrial ecosystems. In Lipids in Aquatic Ecosystems 2009 (pp. 179-210). Springer, New York, NY.

Gladyshev MI, Kharitonov AY, Popova ON, Sushchik NN, Makhutova ON, Kalacheva GS. Quantitative estimation of dragonfly role in transfer of essential polyunsaturated fatty acids from aquatic to terrestrial ecosystems. In Doklady Biochemistry and Biophysics 2011 Jun 1 (Vol. 438, No. 1, pp. 141-143). MAIK Nauka/Interperiodica.

Gladyshev MI, Krylov AV, Sushchik NN, Malin MI, Makhutova ON, Chalova IV, Kalacheva GS. Transfer of essential polyunsaturated fatty acids from an aquatic to terrestrial ecosystem through the fish-bird trophic pair. In Doklady Biological Sciences 2010 Apr 1 (Vol. 431, No. 1, p. 121). Springer Science & Business Media.

Gladyshev MI, Sushchik NN, Anishchenko OV, Makhutova ON, Kolmakov VI, Kalachova GS, Kolmakova AA, Dubovskaya OP. Efficiency of transfer of essential polyunsaturated fatty acids versus organic carbon from producers to consumers in a eutrophic reservoir. Oecologia. 2011; 165(2): 521-31.

Gladyshev MI, Sushchik NN, Makhutova ON. Production of EPA and DHA in aquatic ecosystems and their transfer to the land. Prostaglandins & Other Lipid Mediators. 2013; 107: 117-26.

167

Goldman CR. The contribution of alder trees (Alnus tenuifolia) to the primary productivity of Castle Lake, California. Ecology. 1961; 42(2): 282–288, doi:10. 2307/1932080.

Goldman CR. Primary productivity, nutrients, and transparency during the early onset of eutrophication in ultra-oligotrophic Lake Tahoe, California-Nevada. Limnology and Oceanography. 1988; 33(6): 1321-33.

Gordon CW, Wynn JM, Woodin SJ. Impacts of increased nitrogen supply on high Arctic heath: the importance of bryophytes and phosphorus availability. New Phytologist. 2001; 149(3): 461- 71.

Gorham E and Boyce FM. Influence of lake surface area and depth upon thermal stratification and the depth of the summer thermocline. Journal of Great Lakes Research. 1989; 15(2): 233- 245.

Goto-Azuma K, Koerner RM. Ice core studies of anthropogenic sulfate and nitrate trends in the Arctic. Journal of Geophysical Research: Atmospheres. 2001; 106(D5): 4959-69.

Granéli W, Bertilsson S and Philibert A. Phosphorus limitation of bacterial growth in high Arctic lakes and ponds. Aquatic Sciences. 2004; 66(4): 430-439.

Granshaw FD, Fountain AG. Glacier change (1958–1998) in the north Cascades national park complex, Washington, USA. Journal of Glaciology. 2006; 52(177): 251-6.

Gratton C, Donaldson J, Vander Zanden MJ. Ecosystem linkages between lakes and the surrounding terrestrial landscape in northeast Iceland. Ecosystems. 2008; 11(5): 764-74.

Gratton C, Zanden MJ. Flux of aquatic insect productivity to land: comparison of lentic and lotic ecosystems. Ecology. 2009; 90(10): 2689-99.

Gregory-Eaves, I., Smol, J.P., Finney, B.P., Lean, D.R. and Edwards, M.E., 1999. Characteristics and variation in lakes along a north-south transect in Alaska. Archiv für Hydrobiologie, pp.193- 223.

Greig HS, Kratina P, Thompson PL, Palen WJ, Richardson JS, Shurin JB. Warming, eutrophication, and predator loss amplify subsidies between aquatic and terrestrial ecosystems. Global Change Biology. 2012; 18(2): 504-14.

Griffiths K, Michelutti N, Blais JM, Kimpe LE, Smol JP. Comparing nitrogen isotopic signals between bulk sediments and invertebrate remains in High Arctic seabird-influenced ponds. Journal of Paleolimnology. 2010; 44(2): 405-12.

168

Guay KC, Beck PS, Berner LT, Goetz SJ, Baccini A, Buermann W. Vegetation productivity patterns at high northern latitudes: a multi-sensor satellite data assessment. Global Change Biology. 2014; 20(10): 3147-58.

Hahn S, Bauer S, Klaassen M. Estimating the contribution of carnivorous waterbirds to nutrient loading in freshwater habitats. Freshwater Biology. 2007; 52(12): 2421-33.

Hall MH, Fagre DB. Modeled climate-induced glacier change in Glacier National Park, 1850– 2100. BioScience. 2003; 53(2): 131-40.

Hanna E, Huybrechts P, Steffen K, Cappelen J, Huff R, Shuman C, Irvine-Fynn T, Wise S and Griffiths M. Increased runoff from melt from the Greenland Ice Sheet: a response to global warming. Journal of Climate. 2008; 21(2): 331-341.

Hanna E, Mernild SH, Cappelen J and Steffen K. Recent warming in Greenland in a long-term instrumental (1881–2012) climatic context: I. Evaluation of surface air temperature records. Environmental Research Letters. 2012; 7(4): 045404.

Harms TK, Edmonds JW, Genet H, Creed IF, Aldred D, Balser A, Jones JB. Catchment influence on nitrate and dissolved organic matter in Alaskan streams across a latitudinal gradient. Journal of Geophysical Research: Biogeosciences. 2016; 121(2): 350-69.

Harms TK, Jones Jr JB. Thaw depth determines reaction and transport of inorganic nitrogen in valley bottom permafrost soils: Nitrogen cycling in permafrost soils. Global Change Biology. 2012; 18(9): 2958-68.

Harper MA and McKay RM. Diatoms as markers of atmospheric transport, in The Diatoms: Applications for the Environmental and Earth Sciences, 2nd ed., eds. J. M. Smol and E. F. Stoermer, Cambridge Univerisity Press, Cambridge, UK, 2010, 552-559.

Hassan R, Scholes R, Ash N. Ecosystems and human well-being: current state and trends. Volume 1. Washington, DC: Island Press; 2005.

Hastings MG, Jarvis JC, Steig EJ. Anthropogenic impacts on nitrogen isotopes of ice-core nitrate. Science. 2009; 324(5932): 1288.

Hawkings JR, Wadham JL, Tranter M, Lawson E, Sole A, Cowton T, Tedstone AJ, Bartholomew I, Nienow P, Chandler D, Telling J. The effect of warming climate on nutrient and solute export from the Greenland Ice Sheet. Geochem. Perspect. Lett. 2015; 1: 94-104.

Hawkings J, Wadham J, Tranter M, Telling J, Bagshaw E, Beaton A, Simmons SL, Chandler D, Tedstone A, Nienow P. The Greenland Ice Sheet as a hot spot of phosphorus weathering and export in the Arctic. Global Biogeochemical Cycles. 2016; 30(2): 191-210.

169

Heard AM. Global change and mountain lakes: establishing nutrient criteria and critical loads for Sierra Nevada lakes (Doctoral dissertation, UC Riverside), 2013.

Heiri O, Lotter AF, and Lemcke G. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. Journal of Paleolimnology. 2001; 25(1): 101-110.

Hember RA. Spatially and temporally continuous estimates of annual total nitrogen deposition over North America, 1860–2013. Data in brief. 2018; 17: 134-140.

Hessen DO, Tombre IM, van Geest G, Alfsnes K. Global change and ecosystem connectivity: How geese link fields of central Europe to eutrophication of Arctic freshwaters. Ambio. 2017; 46(1): 40-7.

Hinkel KM, Nelson FE. Spatial and temporal patterns of active layer thickness at Circumpolar Active Layer Monitoring (CALM) sites in northern Alaska, 1995–2000. Journal of Geophysical Research: Atmospheres. 2003; 108(D2), DOI: 10.1029/2001JD000927.

Hinzman LD, Kane DL, Gieck RE, Everett KR. Hydrologic and thermal properties of the active layer in the Alaskan Arctic. Cold Regions Science and Technology. 1991; 19(2): 95-110.

Hixson SM, Sharma B, Kainz MJ, Wacker A, Arts MT. Production, distribution, and abundance of long-chain omega-3 polyunsaturated fatty acids: a fundamental dichotomy between freshwater and terrestrial ecosystems. Environmental Reviews. 2015; 23(4): 414-24.

Hobbie JE. Limnology of Tundra Ponds, Barrow, Alaska, Dowden, Hutchinson & Ross, Stroudsburg, PA, USA, 1980.

Hobbie JE, Peterson BJ, Bettez N, Deegan L, O'Brien WJ, Kling GW, Kipphut GW, Bowden WB, Hershey AE. Impact of global change on the biogeochemistry and ecology of an Arctic freshwater system. Polar Research. 1999; 18(2): 207-14.

Hobbs WO, Telford RJ, Birks HJ, Saros JE, Hazewinkel RR, Perren BB, Saulnier-Talbot É, Wolfe AP. Quantifying recent ecological changes in remote lakes of North America and Greenland using sediment diatom assemblages. PloS one. 2010; 5(4), DOI: 10.1371/journal.pone.0010026.

Hocking GC and Straškraba M. The effect of light extinction on thermal stratification in reservoirs and lakes. International Review of Hydrobiology. 1999; 84(6): 535-556.

Hodson A, Mumford P, Lister D. Suspended sediment and phosphorus in proglacial rivers: bioavailability and potential impacts upon the P status of ice-marginal receiving waters. Hydrological processes. 2004; 18(13): 2409-22.

170

Hoekman D, Bartrons M, Gratton C. Ecosystem linkages revealed by experimental lake-derived isotope signal in heathland food webs. Oecologia. 2012; 170(3): 735-43.

Hoekman D, Dreyer J, Jackson RD, Townsend PA, Gratton C. Lake to land subsidies: experimental addition of aquatic insects increases terrestrial arthropod densities. Ecology. 2011; 92(11): 2063-72.

Hogan EJ, McGowan S and Anderson NJ. Nutrient limitation of periphyton growth in arctic lakes in south-west Greenland. Polar Biology. 2014; 37(9): 1331-1342.

Hole L, Engardt M. Climate change impact on atmospheric nitrogen deposition in northwestern Europe: a model study. AMBIO: A Journal of the Human Environment. 2008; 37(1): 9-18.

Holloway JM and Dahlgren RA. Nitrogen in rock: occurrences and biogeochemical implications. Global Biogeochemical Cycles. 2002; 16(4): 1118.

Holmes RM, McClelland JW, Raymond PA, Frazer BB, Peterson BJ, Stieglitz M. Lability of DOC transported by Alaskan rivers to the Arctic Ocean. Geophysical Research Letters. 2008; 35(3), DOI: 10.1029/2007GL032837.

Holmgren SU, Bigler C, Ingólfsson O, Wolfe AP. The Holocene–Anthropocene transition in lakes of western Spitsbergen, Svalbard (Norwegian High Arctic): climate change and nitrogen deposition. Journal of Paleolimnology. 2010; 43(2): 393-412.

Holt RD. Food webs in space: on the interplay of dynamic instability and spatial processes. Ecological Research. 2002; 17(2): 261-73.

Holtgrieve GW, Schindler DE, Hobbs WO, Leavitt PR, Ward EJ, Bunting L, Chen G, Finney BP, Gregory-Eaves I, Holmgren S, Lisac MJ. A coherent signature of anthropogenic nitrogen deposition to remote watersheds of the northern hemisphere. Science. 2011; 334(6062): 1545- 8.

Hood E, Battin TJ, Fellman J, O'neel S, Spencer RG. Storage and release of organic carbon from glaciers and ice sheets. Nature geoscience. 2015; 8(2): 91.

Hood E, Fellman J, Spencer RG, Hernes PJ, Edwards R, D’Amore D, Scott D. Glaciers as a source of ancient and labile organic matter to the marine environment. Nature. 2009; 462(7276): 1044.

Horton JD. The State Geologic Map Compilation (SGMC) Geodatabase of the Conterminous United States, 2017.

Howarth RW, Boyer EW, Pabich WJ, Galloway JN. Nitrogen use in the United States from 1961– 2000 and potential future trends. AMBIO: A Journal of the Human Environment. 2002; 31(2): 88-97.

171

Huisman J, Sharples J, Stroom JM, Visser PM, Kardinaal WEA, Verspagen JM and Sommeijer B. Changes in turbulent mixing shift competition for light between phytoplankton species. Ecology. 2004; 85(11): 2960-2970.

Hundey EJ, Moser KA, Longstaffe FJ, Michelutti N, Hladyniuk R. Recent changes in production in oligotrophic Uinta Mountain lakes, Utah, identified using paleolimnology. Limnology and Oceanography. 2014; 59(6): 1987-2001.

Hupfer M, Gächter R and Giovanoli R. Transformation of phosphorus species in settling seston and during early sediment diagenesis. Aquatic Sciences. 1995; 57(4): 305-324.

Hupfer M and Lewandowski J. Oxygen controls the phosphorus release from lake sediments–a long-lasting paradigm in limnology. International Review of Hydrobiology. 2008; 93(4-5): 415- 432.

Hutchinson GE. A Treatise on Limnology, 1. Geography, Physics and Chemistry. J. Wiley & sons, N.Y., 1957, 1015 pp.

Huxel GR, McCann K, Polis GA. Effects of partitioning allochthonous and autochthonous resources on food web stability. Ecological Research. 2002; 17(4): 419-32.

Idso, S.B., 1973. On the concept of lake stability 1. Limnology and oceanography. 1973; 18(4): 681-683.

Interlandi SJ and Kilham SS. Limiting resources and the regulation of diversity in phytoplankton communities. Ecology. 2001; 82(5): 1270-1282.

Isaksson E, Hermanson M, Hicks S, Igarashi M, Kamiyama K, Moore J, Motoyama H, Muir D, Pohjola V, Vaikmäe R, van de Wal RS. Ice cores from Svalbard––useful archives of past climate and pollution history. Physics and Chemistry of the Earth, Parts A/B/C. 2003; 28(28-32): 1217- 28.

Jacquemin C, Bertrand C, Oursel B, Thorel M, Franquet E and Cavalli L. Growth rate of alpine phytoplankton assemblages from contrasting watersheds and N-deposition regimes exposed to nitrogen and phosphorus enrichments. Freshwater Biology. 2018; 63(10): 1326-1339.

Jankowski T, Livingstone DM, Bührer H, Forster R and Niederhauser P. Consequences of the 2003 European heat wave for lake temperature profiles, thermal stability, and hypolimnetic oxygen depletion: Implications for a warmer world. Limnology and Oceanography. 2006; 51(2): 815-819.

172

Jassby AD, Reuter JE, Axler RP, Goldman CR, Hackley SH. Atmospheric deposition of nitrogen and phosphorus in the annual nutrient load of Lake Tahoe (California-Nevada). Water Resources Research. 1994; 30(7): 2207-16.

Jeffries DS, Semkin RG, Gibson JJ and Isaac WONG. Recently surveyed lakes in northern Manitoba and Saskatchewan, Canada: characteristics and critical loads of acidity. Journal of Limnology. 2010; 45-55.

Jeong JH, Kug JS, Linderholm HW, Chen D, Kim BM and Jun SY. Intensified Arctic warming under greenhouse warming by vegetation–atmosphere–sea ice interaction. Environmental Research Letters. 2014; 9(9): 094007.

Jones HG. The ecology of snow-covered systems: a brief overview of nutrient cycling and life in the cold. Hydrological processes. 1999; 13(14-15): 2135-47.

Jonsson M, Wardle DA. The influence of freshwater-lake subsidies on invertebrates occupying terrestrial vegetation. Acta Oecologica. 2009; 35(5): 698-704.

Jøsrgensen EG. The adaptation of plankton algae IV. Light adaptation in different algal species. Physiologia plantarum. 1969; 22(6): 1307-1315.

Johnson DW, Miller WW, Susfalk RB, Murphy JD, Dahlgren RA, and Glass DW. Biogeochemical cycling in forest soils ofthe eastern Sierra Nevada Mountains, U.S.A. For. Ecol. Manage. 2009; 258(10): 2249–2260, doi:10.1016/ j.foreco.2009.01.018.

Jorgenson MT, Racine CH, Walters JC, Osterkamp TE. Permafrost degradation and ecological changes associated with a warming climate in central Alaska. Climatic change. 2001; 48(4): 551- 79.

Jorgenson MT, Shur YL, Pullman ER. Abrupt increase in permafrost degradation in Arctic Alaska. Geophysical Research Letters. 2006; 33(2), DOI: 10.1029/2005GL024960.

Kammerlander B, Koinig KA, Rott E, Sommaruga R, Tartarotti B, Trattner F and Sonntag B. Ciliate community structure and interactions within the planktonic food web in two alpine lakes of contrasting transparency. Freshwater Biology. 2016; 61(11): 1950-1965.

Kawahigashi M, Kaiser K, Kalbitz K, Rodionov A, Guggenberger G. Dissolved organic matter in small streams along a gradient from discontinuous to continuous permafrost. Global Change Biology. 2004; 10(9): 1576-86.

Kawamura K, Suzuki I, Fuji Y, Watanabe O. Ice core record of polycyclic aromatic hydrocarbons over the past 400 years, Naturwissenschaften, 1994; 81(11): 502-505.

173

Kehrwald NM, Thompson LG, Tandong Y, Mosley-Thompson E, Schotterer U, Alfimov V, Beer J, Eikenberg J, Davis ME. Mass loss on Himalayan glacier endangers water resources, Geophysical Research Letters. 2008; 35(22): L22503, DOI: 10.1029/2008GL035556.

Khosh MS, McClelland JW, Jacobson AD, Douglas TA, Barker AJ, Lehn GO. Seasonality of dissolved nitrogen from spring melt to fall freezeup in Alaskan Arctic tundra and mountain streams. Journal of Geophysical Research: Biogeosciences. 2017; 122(7): 1718-37.

King JR, BJ Shuter, and AP Zimmerman. The response of the thermal stratification of south bay (Lake Huron) to climatic variability. Can. J. Fish. Aquat. Sci. 1997; 54: 1873–1882.

King JR, Shuter BJ and Zimmerman AP. Signals of climate trends and extreme events in the thermal stratification pattern of multibasin Lake Opeongo, Ontario. Canadian Journal of Fisheries and Aquatic Sciences. 1999; 56(5): 847-852.

Kling GW, Kipphut GW, Miller MC. Arctic lakes and streams as gas conduits to the atmosphere: implications for tundra carbon budgets. Science. 1991; 251(4991): 298-301.

Knapp RA. Non-native trout in natural lakes of the Sierra Nevada: an analysis of their distribution and impacts on native aquatic biota. In Sierra Nevada ecosystem project: final report to Congress 1996 (Vol. 3, pp. 363-407).

Knapp RA. Effects of nonnative fish and habitat characteristics on lentic herpetofauna in Yosemite National Park, USA. Biological Conservation. 2005; 121(2): 265-79.

Knight PG, Waller RI, Patterson CJ, Jones AP and Robinson ZP. Glacier advance, ice-marginal lakes and routing of meltwater and sediment: Russell Glacier, Greenland. Journal of Glaciology. 2000; 46(154): 423-426.

Koenings JP, Burkett RD, Edmundson JM. The exclusion of limnetic Cladocera from turbid glacier-meltwater lakes. Ecology. 1990; 71(1): 57-67.

Kokelj SV, Tunnicliffe J, Lacelle D, Lantz TC, Chin KS, Fraser R. Increased precipitation drives mega slump development and destabilization of ice-rich permafrost terrain, northwestern Canada. Global and Planetary Change. 2015; 129: 56-68.

Kokelj SV, Zajdlik B, Thompson MS. The impacts of thawing permafrost on the chemistry of lakes across the subarctic boreal-tundra transition, Mackenzie Delta region, Canada. Permafrost and Periglacial Processes. 2009; 20(2): 185-99.

Kopáček J, Hejzlar J, Borovec J, Porcal P and Kotorova I. Phosphorus availability in an acidified watershed-lake ecosystem. Limnology and Oceanography. 2000; 45(1): 212-225.

174

Kopàček J, Hejzlar J, Vrba J and Stuchlík E. Phosphorus loading of mountain lakes: terrestrial export and atmospheric deposition. Limnology and Oceanography. 2011; 56(4): 1343-1354.

Kopáček J, Borovec J, Hejzlar J, Ulrich KU, Norton SA and Amirbahman A. Aluminum control of phosphorus sorption by lake sediments. Environmental Science & Technology. 2005; 39(22): 8784-8789.

Köster D, Pienitz R. Seasonal diatom variability and paleolimnological inferences–a case study. Journal of Paleolimnology. 2006; 35(2): 395-416.

Kraemer BM, Anneville O, Chandra S, Dix M, Kuusisto E, Livingstone DM, Rimmer A, Schladow SG, Silow E, Sitoki LM and Tamatamah R. Morphometry and average temperature affect lake stratification responses to climate change. Geophysical Research Letters. 2015; 42(12): 4981- 4988.

Krammer, K., and Lange-Bertalot, H., 1986-1991: Bacillariophyceae In Ettl H., Gartner, G., Gerloff, J., Heynig, H., and Mollenhauer D. (eds.), Sußwasserflora von Mitteleuropa, V. 2 (1-4). Stuttgart/Jena. Gustav Fischer Verlag. (Freshwater flora of central Europe.)

Kraus JM, Vonesh JR. Fluxes of terrestrial and aquatic carbon by emergent mosquitoes: a test of controls and implications for cross-ecosystem linkages. Oecologia. 2012; 170(4): 1111-22.

Kropáček J, Maussion F, Chen F, Hoerz S, Hochschild V. Analysis of ice phenology of lakes on the Tibetan Plateau from MODIS data, The Cryosphere. 2013; 7: 287-301.

Kuhn et al 2019. r package ‘caret.’

Laj P, Palais JM, Sigurdsson H. Changing sources of impurities to the Greenland ice sheet over the last 250 years. Atmospheric Environment. Part A. General Topics. 1992; 26(14): 2627-40.

Lane SN, Bakker M, Gabbud C, Micheletti N, Saugy JN. Sediment export, transient landscape response and catchment-scale connectivity following rapid climate warming and Alpine glacier recession. Geomorphology. 2017; 277: 210-27.

Lantz TC, Kokelj SV. Increasing rates of retrogressive thaw slump activity in the Mackenzie Delta region, NWT, Canada. Geophysical Research Letters. 2008; 35(6), DOI: 10.1029/2007GL032433.

Laspoumaderes C, Modenutti B, Souza MS, Bastidas Navarro M, Cuassolo F, Balseiro E. Glacier melting and stoichiometric implications for lake community structure: zooplankton species distributions across a natural light gradient. Global change biology. 2013; 19(1): 316-26.

Lawrence DM, Slater AG. A projection of severe near-surface permafrost degradation during the 21st century. Geophysical Research Letters. 2005; 32(24), DOI: 10.1029/2005GL025080.

175

Lawson EC, Wadham JL, Tranter M, Stibal M, Lis GP, Butler CE, Laybourn-Parry J, Nienow P, Chandler D and Dewsbury P. Greenland Ice Sheet exports labile organic carbon to the Arctic oceans. Biogeosciences. 2014; 11(14): 4015-4028.

Leavitt PR, Fritz SC, Anderson NJ, Baker PA, Blenckner T, Bunting L, Catalan J, Conley DJ, Hobbs WO, Jeppesen E, Korhola A. Paleolimnological evidence of the effects on lakes of energy and mass transfer from climate and humans. Limnology and Oceanography. 2009; 54(6part2): 2330- 48.

Leavitt PR, Hodgson DA. Sedimentary pigments, in Tracking Environmental Change Using Lake Sediments, eds. J. P. Smol, H. J. Birks, W. M. Last, Springer, Dordrecht, Netherlands, 2002, 295- 325.

Leng MJ and Anderson NJ. Isotopic variation in modern lake waters from western Greenland. The Holocene. 2003; 13(4): 605-611.

Leonard EM, Reasoner MA. A continuous Holocene glacial record inferred from proglacial lake sediments in Banff National Park, Alberta, Canada. Quaternary Research. 1999; 51(1): 1-13.

Lepori F, Keck F. Effects of atmospheric nitrogen deposition on remote freshwater ecosystems. Ambio. 2012; 41(3): 235-46.

Leroux SJ, Loreau M. Subsidy hypothesis and strength of trophic cascades across ecosystems. Ecology letters. 2008; 11(11): 1147-56.

Lesack LF, Marsh P. Lengthening plus shortening of river-to-lake connection times in the Mackenzie River Delta respectively via two global change mechanisms along the arctic coast, Geophysical Research Letters. 2007; 34(23): L23404, DOI: 10.1029/2007GL031656.

Levin SA. The problem of pattern and scale in ecology: the Robert H. MacArthur award lecture. Ecology. 1992; 73(6): 1943-67.

Levine MA and Whalen SC. Nutrient limitation of phytoplankton production in Alaskan Arctic foothill lakes. Hydrobiologia. 2001; 455(1-3); 189-201.

Lewis WM and Grant MC. Acid precipitation in the western United States. Science. 1980; 207(4427): 176-177.

Lewis Jr WM, Grant MC and Saunders III JF. Chemical patterns of bulk atmospheric deposition in the state of Colorado. Water Resources Research. 1984; 20(11); 1691-1704.

Lewis SM and Smith LC. Hydrologic drainage of the Greenland Ice Sheet. Hydrological Processes. 2009; 23(14): 2004-2011.

176

Liaw and Wiener 2018. Package ‘randomForest.’

Liston GE, Hiemstra CA. The changing cryosphere: Pan-Arctic snow trends (1979–2009). Journal of Climate. 2011; 24(21): 5691-712.

Liu J, Wang S, Yu S, Yang D, Zhang L. Climate warming and growth of high-elevation inland lakes on the Tibetan Plateau, Global and Planetary Change. 2009; 67(3-4): 209-217.

Liu X, Yu Z, Dong H, Chen HF. A less or more dusty future in the Northern Qinghai-Tibetan Plateau? Scientific Reports. 2014; 4(6672): DOI: 10.1038/srep06672.

Livingstone DM. Impact of secular climate change on the thermal structure of a large temperate central European lake. Climatic change. 2003; 57(1-2): 205-225.

Loreau M, Mouquet N, Holt RD. Meta-ecosystems: a theoretical framework for a spatial ecosystem ecology. Ecology Letters. 2003; 6(8): 673-9.

Lund M, Raundrup K, Westergaard-Nielsen A, López-Blanco E, Nymand J, Aastrup P. Larval outbreaks in West Greenland: instant and subsequent effects on tundra ecosystem productivity and CO2 exchange. Ambio. 2017; 46(1): 26-38.

Lyons W, Finlay J. Biogeochemical processes in high-latitude lakes and rivers. Oxford University Press, Oxford; 2008.

Madsen J, Reed A, Andreev A. Status and trends of geese (Anser sp., Branta sp.) in the world: a review, updating and evaluation. Gibier Faune Sauvage. 1996; 13(2): 337-53.

Magnuson JJ, Benson BJ and Kratz TK. Temporal coherence in the limnology of a suite of lakes in Wisconsin, USA. Freshwater Biology. 1990; 23(1): 145-159.

Magnuson JJ, DM Robertson, BJ Benson, RH Wynne, DM Livingstone, T Arai, RA Assel, RG Barry, V Card, E Kuusisto, NG Granin, TD Prowse, KM Stewart, and VS Vuglinksi. Historical trends in lake and river ice cover in the northern hemisphere. Science. 2000; 289: 1743–46.

Mahowald N, Jickells TD, Baker AR, Artaxo P, Benitez-Nelson CR, Bergametti G, Bond TC, Chen Y, Cohen DD, Herut B, Kubilay N. Global distribution of atmospheric phosphorus sources, concentrations and deposition rates, and anthropogenic impacts. Global biogeochemical cycles. 2008; 22(4), DOI: 10.1029/2008GB003240.

Malik HI, Northington RM and Saros JE. Nutrient limitation status of Arctic lakes affects the responses of Cyclotella sensu lato diatom species to light: implications for distribution patterns. Polar Biology. 2017; 40(12): 2445-2456.

177

Marcarelli AM, Baxter CV, Mineau MM, Hall Jr RO. Quantity and quality: unifying food web and ecosystem perspectives on the role of resource subsidies in freshwaters. Ecology. 2011; 92(6): 1215-25.

Marczak LB, Thompson RM, Richardson JS. Meta-analysis: trophic level, habitat, and productivity shape the food web effects of resource subsidies. Ecology. 2007; 88(1): 140-8.

Mariash HL, Cazzanelli M, Rautio M, Hamerlik L, Wooller MJ, Christoffersen KS. Changes in food web dynamics of low Arctic ponds with varying content of dissolved organic carbon. Arctic, Antarctic, and Alpine Research. 2018; 50(1), DOI: 10.1080/15230430.2017.1414472.

Marsden MW. Lake restoration by reducing external phosphorus loading: the influence of sediment phosphorus release. Freshwater biology. 1989; 21(2): 139-162.

Martin-Creuzburg D, Kowarik C, Straile D. Cross-ecosystem fluxes: Export of polyunsaturated fatty acids from aquatic to terrestrial ecosystems via emerging insects. Science of the Total Environment. 2017; 577: 174-82.

Matthews KR, Pope KL, Preisler HK, Knapp RA. Effects of nonnative trout on Pacific treefrogs (Hyla regilla) in the Sierra Nevada. Copeia. 2001; 2001(4): 1130-7.

Mayewski PA, Lyons WB, Spencer MJ, Twickler MS, Buck CF, Whitlow S. An ice-core record of atmospheric response to anthropogenic sulphate and nitrate. Nature. 1990; 346(6284): 554.

Mayewski PA, Sneed SB, Birkel SD, Kurbatov AV and Maasch KA. Holocene warming marked by abrupt onset of longer summers and reduced storm frequency around Greenland. Journal of Quaternary Science. 2014; 29(1): 99-104.

McClain ME, Boyer EW, Dent CL, Gergel SE, Grimm NB, Groffman PM, Hart SC, Harvey JW, Johnston CA, Mayorga E, McDowell WH. Biogeochemical hot spots and hot moments at the interface of terrestrial and aquatic ecosystems. Ecosystems. 2003; 6(4): 301-12.

McClelland JW, Stieglitz M, Pan F, Holmes RM, Peterson BJ. Recent changes in nitrate and dissolved organic carbon export from the upper Kuparuk River, North Slope, Alaska. Journal of Geophysical Research: Biogeosciences. 2007; 112(G4), DOI: 10.1029/2006JG000371.

McDonald M, Hershey AE & Miller MC. Global warming impacts on lake trout in Arctic lakes. Limnol. Oceanogr. 1996; 41: 1102–1108.

McGowan S, Gunn HV, Whiteford EJ, Anderson NJ, Jones VJ, Law AC. Functional attributes of epilithic diatoms for palaeoenvironmental interpretations in South-West Greenland lakes. Journal of paleolimnology. 2018; 60(2): 273-98.

178

McTainsh GH, Lynch AW. Quantitative estimates of the effect of climate change on dust storm activity in Australia during the Last Glacial Maximum. Geomorphology. 1996; 17(1-3): 263-71.

Meier MF, Dyurgerov MB, McCabe GJ. The health of glaciers: Recent changes in glacier regime. Climatic change. 2003; 59(1-2): 123-35.

Mernild SH, Hanna E, McConnell JR, Sigl M, Beckerman AP, Yde JC, Cappelen J, Malmros JK and Steffen K. Greenland precipitation trends in a long-term instrumental climate context (1890– 2012): Evaluation of coastal and ice core records. International Journal of Climatology. 2015; 35(2): 303-320.

Mernild SH, Hanna E, Yde JC, Cappelen J and Malmros JK. Coastal Greenland air temperature extremes and trends 1890–2010: annual and monthly analysis. International Journal of Climatology. 2014; 34(5): 1472-1487.

Mesquita PS, Wrona FJ, Prowse TD. Effects of retrogressive permafrost thaw slumping on sediment chemistry and submerged macrophytes in Arctic tundra lakes. Freshwater Biology. 2010; 55(11): 2347-58.

Michelutti N, Liu H, Smol JP, Kimpe LE, Keatley BE, Mallory M, Macdonald RW, Douglas MS, Blais JM. Accelerated delivery of polychlorinated biphenyls (PCBs) in recent sediments near a large seabird colony in Arctic Canada. Environmental pollution. 2009; 157(10): 2769-75.

Milborrow, S. and Milborrow, M.S., 2019. Package ‘rpart. plot’.

Miller WW, Johnson DW, Denton C, Verburg PSJ, Dana GL, and Walker RF. Inconspicuous nutrient laden surface runoff from mature forest Sierran watersheds. Water Air Soil Pollut. 2005; 163(1–4): 3–17, doi:10.1007/ s11270-005-7473-7.

Miró A, Ventura M. Historical use, fishing management and lake characteristics explain the presence of non-native trout in Pyrenean lakes: Implications for conservation, Biological Conservation. 2013; 167: 17-24.

Mladenov N, Lípez-Ramos J, McKnight DM, Rechea I. Alpine lake optical properties as sentinels of dust deposition and global change. Limnology and oceanography. 2009; 54(6part2): 2386- 400.

Mladenov N, Pulido-Villena E, Morales-Baquero R, Ortega-Retuerta E, Sommaruga R, Reche I. Spatiotemporal drivers of dissolved organic matter in high alpine lakes: Role of Saharan dust inputs and bacterial activity. Journal of Geophysical Research: Biogeosciences. 2008; 113(G2): DOI: 10.1029/2008JG000699.

Molot LA, Dillon PJ. Nitrogen mass balances and denitrification rates in central Ontario lakes. Biogeochemistry. 1993; 20(3): 195-212.

179

Moquin PA, Mesquita PS, Wrona FJ, Prowse TD. Responses of benthic invertebrate communities to shoreline retrogressive thaw slumps in Arctic upland lakes. Freshwater Science. 2014; 33(4): 1108-18.

Morales-Baquero R, Pulido-Villena E, Reche I. Atmospheric inputs of phosphorus and nitrogen to the southwest Mediterranean region: Biogeochemical responses of high mountain lakes. Limnology and Oceanography. 2006; 51(2): 830-7.

Morris DP and Lewis Jr WM. 1988. Phytoplankton nutrient limitation in Colorado mountain lakes. Freshwater Biology. 1988; 20(3): 315-327.

Morris DP, Zagarese H, Williamson CE, Balseiro EG, Hargreaves BR, Modenuti B, and Quiemalin˜os C. The attenuation of solar UV radiation in lakes and the role of dissolved organic carbon. Limnology and Oceanography. 1995; 40: 1381–1391.

Moulin C and Chiapello I. Evidence of the control of summer atmospheric transport of African dust over the Atlantic by Sahel sources from TOMS satellites (1979–2000). Geophysical research letters. 2004; 31(2).

Moulin C, Lambert CE, Dulac F, Dayan U. Control of atmospheric export of dust from North Africa by the North Atlantic Oscillation. Nature. 1997; 387(6634): 691.

Mueller DR, Van Hove P, Antoniades D, Jeffries MO, Vincent WF. High Arctic lakes as sentinel ecosystems: Cascading regime shifts in climate, ice cover, and mixing. Limnology and Oceanography. 2009; 54(6 part 2): 2371-85.

Muhs DR, Budahn JR, McGeehin JP, Bettis III EA, Skipp G, Paces JB, Wheeler EA. Loess origin, transport, and deposition over the past 10,000 years, Wrangell-St. Elias National Park, Alaska, Aeolian Research. 2013; 11, 85-99.

Murphy CA, Thompson PL and Vinebrooke RD. Assessing the sensitivity of alpine lakes and ponds to nitrogen deposition in the Canadian Rocky Mountains. Hydrobiologia. 2010; 648(1): 83-90.

Nakano S, Murakami M. Reciprocal subsidies: dynamic interdependence between terrestrial and aquatic food webs. Proceedings of the National Academy of Sciences. 2001; 98(1): 166-170.

Nakano S, Miyasaka H, Kuhara N. Terrestrial–aquatic linkages: riparian arthropod inputs alter trophic cascades in a stream food web. Ecology. 1999; 80(7): 2435-41.

Nanus L, Campbell DH, Williams MW. Sensitivity of alpine and subalpine lakes to acidification from atmospheric deposition in Grand Teton National Park and Yellowstone National Park, Wyoming. Denver, Colorado: US Geological Survey; 2005.

180

Nanus L, Clow DW, Saros JE, Stephens VC and Campbell DH. Mapping critical loads of nitrogen deposition for aquatic ecosystems in the Rocky Mountains, USA. Environmental Pollution. 2012; 166: 125-135.

Nanus L, Williams MW, Campbell DH, Tonnessen KA, Blett T and Clow DW. Assessment of lake sensitivity to acidic deposition in national parks of the Rocky Mountains. Ecological Applications. 2009; 19(4): 961-973.

National Atmospheric Deposition Program (NRSP-3). 2020. NADP Program Office, Wisconsin State Laboratory of Hygiene, 465 Henry Mall, Madison, WI 53706.

Nealson KH and Saffarini D. Iron and manganese in anaerobic respiration: environmental significance, physiology, and regulation. Ann. Rev. Microbiol. 1994; 48: 311–343.

Nelson JS, Paetz MJ. The fishes of Alberta. University of Alberta; 1992.

Nielsen AB. Present conditions in Greenland and the Kangerlussuaq area. Posiva Oy. Working Report no. 2010-07. 2010.

Niittynen P, Luoto M. The importance of snow in species distribution models of arctic vegetation. Ecography. 2018; 41(6): 1024-37.

Northington RM, Saros JE. Factors controlling methane in arctic lakes of southwest Greenland. PloS one. 2016; 11(7), DOI: 10.1371/journal.pone.0159642.

Norton SA, Perry RH, Saros JE, Jacobson GL, Fernandez IJ, Kopáček J, Wilson TA and SanClements MD. The controls on phosphorus availability in a Boreal lake ecosystem since deglaciation. Journal of Paleolimnology. 2011; 46(1): 107-122.

Norwegian Meteorological Institute (MET Norway 2019). https://www.emep.int/mscw/mscw_moddata.html.

Nürnberg GK. The prediction of internal phosphorus load in lakes with anoxic hypolimnia. Limnology and oceanography. 1984; 29(1): 111-24.

Nürnberg GK. Quantifying anoxia in lakes. Limnology and oceanography. 1995; 40(6): 1100- 1111.

Nürnberg GK. Quantification of oxygen depletion in lakes and reservoirs with the hypoxic factor. Lake and Reservoir Management. 2002; 18(4): 299-306.

181

Nydick KR, Lafrancois BM, Baron JS, Johnson BM. Lake-specific responses to elevated atmospheric nitrogen deposition in the Colorado Rocky Mountains, USA. Hydrobiologia. 2003; 510(1-3): 103-14.

Nydick KR, Lafrancois BM, Baron JS, Johnson BM. Nitrogen regulation of algal biomass, productivity, and composition in shallow mountain lakes, Snowy Range, Wyoming, USA. Canadian Journal of Fisheries and Aquatic Sciences. 2004; 61(7): 1256-68.

O'Brien WJ, Barfield M, Bettez N, Hershey AE, Hobbie JE, Kipphut G, Kling G, Miller MC. Long- term response and recovery to nutrient addition of a partitioned arctic lake. Freshwater Biology. 2005; 50(5): 731-41.

Oerlemans J. Extracting a climate signal from 169 glacier records. Science. 2005; 308(5722): 675-677.

Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O'Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, and Wagner H. vegan: Community Ecology Package. 2017. R package version 2.4-2.

O’Reilly CM, Alin SR, Plisnier PD, Cohen AS & McKee BA. Climate change decreases aquatic ecosystem productivity of Lake Tanganyika, Africa. Nature. 2003; 424: 766–768.

Ouimet R, Arp PA, Watmough SA, Aherne J and DeMerchant I. Determination and mapping critical loads of acidity and exceedances for upland forest soils in Eastern Canada. Water, Air, and Soil Pollution. 2006; 172(1-4): 57-66.

Pace ML and Cole JJ. Effects of whole-lake manipulations of nutrient loading and food web structure on planktonic respiration. Canadian Journal of Fisheries and Aquatic Sciences. 2000; 57(2): 487-496.

Pardo LH, Fenn ME, Goodale CL, Geiser LH, Driscoll CT, Allen EB, Baron JS, Bobbink R, Bowman WD, Clark CM, Emmett B. Effects of nitrogen deposition and empirical nitrogen critical loads for ecoregions of the United States. Ecological Applications. 2011; 21(8): 3049-82.

Parrish CC. Essential fatty acids in aquatic food webs. In Lipids in Aquatic Ecosystems 2009 (pp. 309-326). Springer, New York, NY.

Paul F, Kääb A, Maisch M, Kellenberger T, Haeberli W. Rapid disintegration of Alpine glaciers observed with satellite data. Geophysical research letters. 2004; 31(21), DOI: 10.1029/2004GL020816.

Pearce-Higgins JW, Yalden DW, Whittingham MJ. Warmer springs advance the breeding phenology of golden plovers Pluvialis apricaria and their prey (Tipulidae), Oecologia, 2005; 143(3): 470-476.

182

Pedersen SH, Liston GE, Tamstorf MP, Abermann J, Lund M, Schmidt NM. Quantifying snow controls on vegetation greenness. Ecosphere. 2018; 9(6), DOI: 10.1002/ecs2.2309.

Pepin N, Bradley RS, Diaz HF, Baraër M, Caceres EB, Forsythe N, Fowler H, Greenwood G, Hashmi MZ, Liu XD, Miller JR. Elevation-dependent warming in mountain regions of the world. Nature Climate Change. 2015; 5(5): 424.

Perren BB, Axford Y and Kaufman DS. Alder, nitrogen, and lake ecology: terrestrial-aquatic linkages in the postglacial history of Lone Spruce Pond, southwestern Alaska. PloS ONE. 2017; 12(1): 0169106.

Perren BB, Douglas MS and Anderson NJ. Diatoms reveal complex spatial and temporal patterns of recent limnological change in West Greenland. Journal of Paleolimnology. 2009; 42(2): 233- 247.

Peter H, Hörtnagl P, Reche I, Sommaruga R. Bacterial diversity and composition during rain events with and without Saharan dust influence reaching a high mountain lake in the Alps. Environmental microbiology reports. 2014; 6(6): 618-24.

Peter H, Sommaruga R. Alpine glacier-fed turbid lakes are discontinuous cold polymictic rather than dimictic. Inland Waters. 2017; 7(1): 45-54.

Petersen I, Winterbottom JH, Orton S, Friberg N, Hildrew AG, Spiers DC, Gurney WS. Emergence and lateral dispersal of adult Plecoptera and Trichoptera from Broadstone Stream, UK. Freshwater Biology. 1999; 42(3): 401-16.

Pickett ST, Cadenasso ML. Landscape ecology: spatial heterogeneity in ecological systems. Science. 1995; 269(5222): 331-4.

Piovia-Scott J, Sadro S, Knapp RA, Sickman J, Pope KL, Chandra S. Variation in reciprocal subsidies between lakes and land: perspectives from the mountains of California. Canadian Journal of Fisheries and Aquatic Sciences. 2016; 73(11): 1691-701.

Polis GA, Anderson WB, Holt RD. Toward an integration of landscape and food web ecology: the dynamics of spatially subsidized food webs. Annual review of ecology and systematics. 1997; 28(1): 289-316.

Porder S and Ramachandran S. The phosphorus concentration of common rocks—a potential driver of ecosystem P status. Plant and soil. 2013; 367(1-2): 41-55.

Porter E, Blett T, Potter DU and Huber C. Protecting resources on federal lands: implications of critical loads for atmospheric deposition of nitrogen and sulfur. BioScience. 2005; 55(7): 603- 612.

183

Post E, Pedersen C. Opposing plant community responses to warming with and without herbivores. Proceedings of the National Academy of Sciences. 2008; 105(34): 12353-8.

Power ME, Rainey WE, Parker MS, Sabo JL, Smyth A, Khandwala S, Finlay JC, McNeely FC, Marsee K, Anderson C. River-to-watershed subsidies in an old-growth conifer forest. Food webs at the landscape level. 2004: 217-40.

Prowse TD, Wrona FJ, Reist JD, Gibson JJ, Hobbie JE, Lévesque LM and Vincent WF. Climate change effects on hydroecology of Arctic freshwater ecosystems. AMBIO: A Journal of the Human Environment. 2006; 35(7): 347-359.

Psenner R. Living in a dusty world: airborne dust as a key factor for alpine lakes. Water, Air, and Soil Pollution. 1999; 112(3-4): 217-27.

Psenner R, Pucsko R and Sager M. Die Fraktionierung organischer und anorganischer Phosphorverbindungen von Sedimenten — Versuch einer Definition ökologisch wichtiger Frakionen. Archive fur Hydrobiologie Supplement. 1984; (70): 111–155.

Ptacnik R, Diehl S and Berger S. Performance of sinking and nonsinking phytoplankton taxa in a gradient of mixing depths. Limnology and Oceanography. 2003; 48(5): 1903-1912.

Pulido-Villena E, Reche I, Morales-Baquero R. Evidence of an atmospheric forcing on bacterioplankton and phytoplankton dynamics in a high mountain lake. Aquatic sciences. 2008; 70(1): 1-9.

Qiang M, Chen F, Zhang J, Zu R, Jin M, Zhou A, Xiao S. Grain size in sediments from Lake Sugan: a possible linkage to dust storm events at the northern margin of the Qinghai–Tibetan Plateau, Environmental Geology. 2007; 51(7): 1229-1238.

Rachowicz LJ, Knapp RA, Morgan JA, Stice MJ, Vredenburg VT, Parker JM, Briggs CJ. Emerging infectious disease as a proximate cause of amphibian mass mortality. Ecology. 2006; 87(7): 1671-83.

Raynolds MK, Walker DA, Munger CA, Vonlanthen CM and Kade AN. A map analysis of patterned-ground along a North American Arctic Transect. Journal of Geophysical Research: Biogeosciences. 2008; 113(G3).

Read JS, Hamilton DP, Jones ID, Muraoka K, Winslow LA, Kroiss R, Wu CH and Gaiser E. Derivation of lake mixing and stratification indices from high-resolution lake buoy data. Environmental Modelling & Software. 2011; 26(11): 1325-1336.

184

Reche I, Ortega-Retuerta E, Romera O, Villena EP, Baquero RM, Casamayor EO. Effect of Saharan dust inputs on bacterial activity and community composition in Mediterranean lakes and reservoirs. Limnology and Oceanography. 2009; 54(3): 869-79.

Reheis MC, Kihl R. Dust deposition in southern Nevada and California, 1984–1989: Relations to climate, source area, and source lithology. Journal of Geophysical Research: Atmospheres. 1995; 100(D5): 8893-918.

Richardson D, Melles S, Pilla R, Hetherington A, Knoll L, Williamson C, Kraemer B, Jackson J, Long E, Moore K and Rudstam, L. Transparency, geomorphology and mixing regime explain variability in trends in lake temperature and stratification across Northeastern North America (1975–2014). Water. 2017; 9(6): 442.

Rigler FH. Limnology in the high Arctic: a case study of Char Lake: With 7 figures and 1 table in the text. Internationale Vereinigung für theoretische und angewandte Limnologie: Verhandlungen. 1978; 20(1): 127-140.

Romanovsky VE, Sazonova TS, Balobaev VT, Shender NI, Sergueev DO. Past and recent changes in air and permafrost temperatures in eastern Siberia. Global and Planetary Change. 2007; 56(3- 4): 399-413.

Rosén P, Cunningham L, Vonk J, Karlssona J. Effects of climate on organic carbon and the ratio of planktonic to benthic primary producers in a subarctic lake during the past 45 years, Limnology and Oceanography. 2009, 54(5), 1723-1732.

Rousseau DD, Schevin P, Ferrier J, Jolly D, Andreasen T, Ascanius SE, Hendriksen SE, Poulsen U. Long-distance pollen transport from North America to Greenland in spring, Journal of Geophysical Research: Biogeosciences. 2008; 113(G2): G02013, doi:10.1029/2007JG000456.

Sadro S, Nelson CE, and Melack JM. The influence oflandscape position and catchment characteristics on aquatic biogeochemistry in high-elevation lake-chains. Ecosystems. 2012; 15(3): 363–386. doi:10.1007/s10021-011-9515-x.

Saros J, Anderson NJ, Juggins S, McGowan S, Yde J, Telling J, Bullard J, Yallop M, Heathcote A, Burpee B and Fowler R. Arctic climate shifts drive rapid ecosystem responses across the West Greenland landscape. Environmental Research Letters. 2019.

Saros JE, Clow DW, Blett T and Wolfe AP. Critical nitrogen deposition loads in high-elevation lakes of the western US inferred from paleolimnological records. Water, Air, & Soil Pollution. 2011; 216(1-4): 193-202.

Saros JE, Interlandi SJ, Doyle S, Michel TJ and Williamson CE. Are the deep chlorophyll maxima in alpine lakes primarily induced by nutrient availability, not UV avoidance? Arctic, Antarctic, and Alpine Research. 2005; 37(4): 557-563.

185

Saros JE, Michel TJ, Interlandi SJ, Wolfe AP. Resource requirements of Asterionella formosa and Fragilaria crotonensis in oligotrophic alpine lakes: implications for recent phytoplankton community reorganizations. Canadian Journal of Fisheries and Aquatic Sciences. 2005; 62(7): 1681-9.

Saros JE, Northington RM, Anderson DS and Anderson NJ. A whole-lake experiment confirms a small centric diatom species as an indicator of changing lake thermal structure. Limnology and Oceanography Letters. 2016; 1(1): 27-35.

Saros JE, Northington RM, Osburn CL, Burpee BT, John Anderson N. Thermal stratification in small arctic lakes of southwest Greenland affected by water transparency and epilimnetic temperatures. Limnology and Oceanography. 2016; 61(4): 1530-42.

Saros JE, Rose KC, Clow DW, Stephens VC, Nurse AB, Arnett HA, Stone JR, Williamson CE, Wolfe AP. Melting alpine glaciers enrich high-elevation lakes with reactive nitrogen. Environmental science & technology. 2010; 44(13): 4891-6.

Saros JE, Stone JR, Pederson GT, Slemmons KE, Spanbauer T, Schliep A, Cahl D, Williamson CE, Engstrom DR. Climate-induced changes in lake ecosystem structure inferred from coupled neo- and paleoecological approaches. Ecology. 2012; 93(10): 2155-64.

Scharnweber K, Vanni MJ, Hilt S, Syväranta J, Mehner T. Boomerang ecosystem fluxes: organic carbon inputs from land to lakes are returned to terrestrial food webs via aquatic insects. Oikos. 2014; 123(12): 1439-48.

Schindler DW, Beaty KG, Fee EJ, Cruikshank DR, DeBruyn ER, Findlay DL, Linsey GA, Shearer JA, Stainton MP and Turner MA. Effects of climatic warming on lakes of the central boreal forest. Science. 1990; 250(4983): 967-970.

Schindler DW, Kalff J, Welch HE, Brunskill GJ, Kling H, Kritsch N. Eutrophication in the High Arctic—Meretta Lake, Cornwallis Island (75 N Lat.), Journal of the Fisheries Board of Canada. 1974; 31(5): 647-662.

Schindler DW, Welch HE, Kalff J, Brunskill GJ and Kritsch N. Physical and chemical limnology of Char Lake, Cornwallis Island (75 N lat.). Journal of the Fisheries Board of Canada. 1974; 31(5): 585-607.

Schmid P, Bogdal C, Blüthgen N, Anselmetti FS, Zwyssig A, Hungerbühler K. The missing piece: sediment records in remote mountain lakes confirm glaciers being secondary sources of persistent organic pollutants, Environmental Science & Technology. 2011; 45(1): 203-208.

Schmidt W. Über Temperatur und Stabilitätsverhältnisse von Seen. Geogr. Ann. 1928; 10: 145- 177.

186

Schuur EA, Bockheim J, Canadell JG, Euskirchen E, Field CB, Goryachkin SV, Hagemann S, Kuhry P, Lafleur PM, Lee H, Mazhitova G. Vulnerability of permafrost carbon to climate change: Implications for the global carbon cycle. BioScience. 2008; 58(8): 701-14.

Scott D, Hood E, Nassry M. In-stream uptake and retention of C, N and P in a supraglacial stream. Annals of Glaciology. 2010; 51(56): 80-6.

Screen JA and Simmonds I. The central role of diminishing sea ice in recent Arctic temperature amplification. Nature. 2010; 464(7293): 1334.

Sears AL, Holt RD, Polis GA. Feast and famine in food webs: the effects of pulsed productivity. Food webs at the landscape level. University of Chicago Press, Chicago, Illinois, USA. 2004: 359- 86.

Segal RA, Lantz TC, Kokelj SV. Acceleration of thaw slump activity in glaciated landscapes of the Western Canadian Arctic. Environmental Research Letters. 2016; 11(3): DOI: 10.1088/1748- 9326/11/3/034025.

Serreze MC, Barrett AP, Stroeve JC, Kindig DN, Holland MM. The emergence of surface-based Arctic amplification. The Cryosphere. 2009; 3(1): 11-9.

Sheibley RW, Enache M, Swarzenski PW, Moran PW and Foreman J.R. Nitrogen deposition effects on diatom communities in lakes from three national parks in Washington State. Water, Air, & Soil Pollution. 2014; 225(2): 1857.

Sickman JO, Melack JM, Clow DW. Evidence for nutrient enrichment of high-elevation lakes in the Sierra Nevada, California. Limnology and Oceanography. 2003; 48(5): 1885-92.

Sickman JO, Melack JM, Stoddard JL. Regional analysis of inorganic nitrogen yield and retention in high-elevation ecosystems of the Sierra Nevada and Rocky Mountains, in The Nitrogen Cycle at Regional to Global Scales, eds. E. W. Boyer and R. W. Howarth, Springer, Dordrecht, Netherlands, 2002, 341-374.

Simões JC, Zagorodnov VS. The record of anthropogenic pollution in snow and ice in Svalbard, Norway. Atmospheric Environment. 2001; 35(2): 403-13.

Sinsabaugh RL and Foreman CM. Activity profiles of bacterioplankton in a eutrophic river. Freshwater Biology. 2001; (46): 1239-1249.

Sinsabaugh RL, Lauber CL, Weintraub MN, Ahmed B, Allison SD, Crenshaw C, Contosta AR, Cusack D, Frey S, Gallo ME, Gartner TB, Hobbie SE, Holland K, Keeler BL, Powers JS, Stursova M, Takacs-Vesbach C, Waldrop MP, Wallenstein MD, Zak DR, and Zeglin LH. Stoichiometry of soil enzyme activity at global scale, Ecology Letters. 2008; 11(11): 1252–1264.

187

Sinsabaugh RL, Hill BH and Shah JJF. Ecoenzymatic stoichiometry of microbial organic nutrient acquisition in soil and sediment. Nature. 2009; 462(7274): 795-798.

Slemmons KE, Saros JE. Implications of nitrogen-rich glacial meltwater for phytoplankton diversity and productivity in alpine lakes. Limnology and Oceanography. 2012; 57(6): 1651-63.

Slemmons KE, Medford A, Hall BL, Stone JR, McGowan S, Lowell T, Kelly M, Saros JE. Changes in glacial meltwater alter algal communities in lakes of Scoresby Sund, Renland, East Greenland throughout the Holocene: Abrupt reorganizations began 1000 years before present. The Holocene. 2017; 27(7): 929-40.

Slemmons KE, Saros JE, Stone JR, McGowan S, Hess CT, Cahl D. Effects of glacier meltwater on the algal sedimentary record of an alpine lake in the central US Rocky Mountains throughout the late Holocene. Journal of paleolimnology. 2015; 53(4): 385-99.

Šmejkalová T, Edwards ME and Dash J. Arctic lakes show strong decadal trend in earlier spring ice-out. Scientific reports. 2016; 6: 38449.

Smith EM and Prairie YT. Bacterial metabolism and growth efficiency in lakes: the importance of phosphorus availability. Limnology and Oceanography. 2004; 49(1): 137-147.

Smol JP, Birks HJ, Last WM. Using biology to study long-term environmental change, in Tracking Environmental Change Using Lake Sediments, eds. J. P. Smol, H. J. Birks, W. M. Last, Springer, Dordrecht, Netherlands. 2002; 1-3.

Smol JP, Wolfe AP, Birks HJ, Douglas MS, Jones VJ, Korhola A, Pienitz R, Rühland K, Sorvari S, Antoniades D, Brooks SJ. Climate-driven regime shifts in the biological communities of arctic lakes. Proceedings of the National Academy of Sciences. 2005; 102(12): 4397-402.

Soininen J, Bartels PI, Heino J, Luoto M, Hillebrand H. Toward more integrated ecosystem research in aquatic and terrestrial environments. BioScience. 2015; 65(2): 174-82.

Sommaruga R. When glaciers and ice sheets melt: consequences for planktonic organisms. Journal of plankton research. 2015; 37(3): 509-18.

Sommaruga R, Kandolf G. Negative consequences of glacial turbidity for the survival of freshwater planktonic heterotrophic flagellates. Scientific reports. 2014; 4: 4113.

Sommer U. Seasonal succession of phytoplankton in Lake Constance. BioScience. 1985; 5: 351– 357.

Søndergaard M, Jensen JP & Jeppesen E. Role of sediment and internal loading of phosphorus in shallow lakes. Hydrobiologia. 2003; 506: 135–145.

188

Spaulding, S. A., Lubinski, D. J. and Potapova, M., 2010: Diatoms of the United States. http://westerndiatoms.colorado.edu Accessed on 11 March, 2017.

Spaulding SA, Otu MK, Wolfe AP, Baron JS. Paleolimnological records of nitrogen deposition in shallow, high-elevation lakes of Grand Teton National Park, Wyoming, USA. Arctic, Antarctic, and Alpine Research. 2015; 47(4): 703-17.

Spears BM, Lesack LF. Bacterioplankton production, abundance, and nutrient limitation among lakes of the Mackenzie Delta (western Canadian arctic), Canadian Journal of Fisheries and Aquatic Sciences. 2006; 63(4): 845-857.

Spencer RG, Mann PJ, Dittmar T, Eglinton TI, McIntyre C, Holmes RM, Zimov N, Stubbins A. Detecting the signature of permafrost thaw in Arctic rivers. Geophysical Research Letters. 2015; 42(8): 2830-5.

Stanish LF, Bagshaw EA, McKnight DM, Fountain AG and Tranter M. Environmental factors influencing diatom communities in Antarctic cryoconite holes. Environmental Research Letters. 2013; 8(4): 045006.

Sterner RW, Elser JJ, Fee EJ, Guildford SJ, Chrzanowski TH. The light: nutrient ratio in lakes: the balance of energy and materials affects ecosystem structure and process. The American Naturalist. 1997; 150(6): 663-84.

Stow CA and Cha Y. Are chlorophyll a–total phosphorus correlations useful for inference and prediction?. Environmental science & technology. 2013; 47(8): 3768-3773.

Strayer DL, Power ME, Fagan WF, Pickett ST, Belnap J. A classification of ecological boundaries, BioScience. 2003; 53(8): 723-729.

Striegl RG, Aiken GR, Dornblaser MM, Raymond PA, Wickland KP. A decrease in discharge- normalized DOC export by the Yukon River during summer through autumn. Geophysical Research Letters. 2005; 32(21), DOI: 10.1029/2005GL024413.

Takimoto G, Iwata T, Murakami M. Seasonal subsidy stabilizes food web dynamics: balance in a heterogeneous landscape. Ecological Research. 2002; 17(4): 433-9.

Tessier A, Campbell P and Bisson M. Sequential extraction procedure for the speciation of trace metals. Analytical Chemistry. 1979; (51): 844-851.

Therneau T. User written splitting functions for RPART. 2019. Mayo Clinic. Disponível.

189

Thienpont JR, Ruehland KM, Pisaric MF, Kokelj SV, Kimpe LE, Blais JM, Smol JP. Biological responses to permafrost thaw slumping in Canadian Arctic lakes. Freshwater Biology. 2013; 58(2): 337-53.

Thompson LG, Yao T, Mosley-Thompson E, Davis ME, Henderson KA, Lin PN. A high-resolution millennial record of the South Asian monsoon from Himalayan ice cores, Science. 2000; 289(5486): 1916-1919.

Thompson MS, Kokelj SV, Prowse TD, Wrona FJ. The impact of sediments derived from thawing permafrost on tundra lake water chemistry: an experimental approach. In Proceedings of the 9th International Conference on Permafrost, edited by: Kane, DL and Hinkel, KM 2008 (Vol. 2, pp. 1763-1768).

Thompson MS, Wrona FJ, Prowse TD. Shifts in plankton, nutrient and light relationships in small tundra lakes caused by localized permafrost thaw. Arctic. 2012: 367-76.

Tiberti R, Nelli L, Brighenti S, Iacobuzio R, Rolla M. Spatial distribution of introduced brook trout Salvelinus fontinalis (Salmonidae) within alpine lakes: evidences from a fish eradication campaign, The European Zoological Journal. 2017; 84(1): 73-88

Tietjen T and Wetzel RG. Extracellular enzyme-clay mineral complexes: enzyme adsorption, alteration of enzyme activity, and protection from photodegradation. Aquatic Ecology. 2003; 37(4): 331-339.

Tranvik LJ, Downing JA, Cotner JB, Loiselle SA, Striegl RG, Ballatore TJ, Dillon P, Finlay K, Fortino K, Knoll LB, Kortelainen PL. Lakes and reservoirs as regulators of carbon cycling and climate. Limnology and oceanography. 2009; 54(6part2): 2298-314.

Tugend, C.A., III, and Saros, Jasmine E. Effects of Geese and Caribou on Nutrient and Carbon Cycling in Southwest Greenland. Thesis for the completion of a Master of Science degree, University of Maine. 2017.

Turner MG. Landscape ecology: the effect of pattern on process. Annual review of ecology and systematics. 1989; 20(1): 171-97.

UNECE ICP, 2004. Manual on Methodologies and Criteria for Modelling and Mapping Critical Loads & Levels and Air Pollution Effects, Risks and Trends. In: Convention on Long-range Transboundary Air Pollution. United Nations Economic Commission for Europe, International Cooperative Programme. http://www.icpmapping.org. van den Broeke M, Bamber J, Ettema J, Rignot E, Schrama E, van de Berg WJ, van Meijgaard E, Velicogna I and Wouters B. Partitioning recent Greenland mass loss. Science. 2009; 326(5955): 984-986.

190

van den Broeke MR, Duynkerke PG, Oerlemans J. The observed katabatic flow at the edge of the Greenland ice sheet during GIMEX-91. Global and Planetary Change. 1994; 9(1-2): 3-15.

Vander Zanden MJ, Gratton C. Blowin’in the wind: reciprocal airborne carbon fluxes between lakes and land. Canadian Journal of Fisheries and Aquatic Sciences. 2011; 68(1): 170-82.

Vincent WF, Laurion I, Pienitz R and Walter Anthony KM. Climate impacts on Arctic lake ecosystems. Climatic Change and Global Warming of Inland Waters: Impacts and Mitigation for Ecosystems and Societies. 2013; 27-42.

Vinebrooke RD and Leavitt PR. Differential responses of littoral communities to ultraviolet radiation in an alpine lake. Ecology. 1999; 80(1): 223-237.

Vinebrooke RD, Thompson PL, Hobbs W, Luckman BH, Graham MD, Wolfe AP. Glacially mediated impacts of climate warming on alpine lakes of the Canadian Rocky Mountains. Internationale Vereinigung für theoretische und angewandte Limnologie: Verhandlungen. 2010; 30(9): 1449-52.

Vinšová P, Pinseel E, Kohler TJ, Van de Vijver B, Žárský JD, Kavan J and Kopalová, K. Diatoms in cryoconite holes and adjacent proglacial freshwater sediments, Nordenskiöld glacier (Spitsbergen, High Arctic). Czech Polar Reports. 2015; 5: 112-133.

Vonk JE, Tank SE, Bowden WB, Laurion I, Vincent WF, Alekseychik P, Amyot M, Billet MF, Canario J, Cory RM, Deshpande BN. Reviews and syntheses: Effects of permafrost thaw on Arctic aquatic ecosystems. Biogeosciences. 2015; 12(23): 7129-67.

Wadham JL, Tranter M, Skidmore M, Hodson AJ, Priscu J, Lyons WB, Sharp M, Wynn P and Jackson M. Biogeochemical weathering under ice: size matters. Global Biogeochemical Cycles. 2010; 24(3): GB3025.

Waelbroeck C, Monfray P, Oechel WC, Hastings S, Vourlitis G. The impact of permafrost thawing on the carbon dynamics of tundra. Geophysical Research Letters. 1997; 24(3): 229-32.

Walvoord MA, Striegl RG. Increased groundwater to stream discharge from permafrost thawing in the Yukon River basin: Potential impacts on lateral export of carbon and nitrogen. Geophysical Research Letters. 2007; 34(12), DOI: 10.1029/2007GL030216.

Wang X, Gong P, Zhang Q, Yao T. Impact of climate fluctuations on deposition of DDT and hexachlorocyclohexane in mountain glaciers: Evidence from ice core records, Environmental Pollution. 2010; 158(2): 375-380.

Wania F, Mackay D. Global fractionation and cold condensation of low volatility organochlorine compounds in polar regions, Ambio. 1993; 22(1): 10-18.

191

Warner K, Fowler R, Northington R, Malik H, McCue J and Saros J. How Does Changing Ice-Out Affect Arctic versus Boreal Lakes? A Comparison Using Two Years with Ice-Out that Differed by More Than Three Weeks. Water. 2018; 10(1): 78.

Watson SB, Miller C, Arhonditsis G, Boyer GL, Carmichael W, Charlton MN, Confesor R, Depew DC, Höök TO, Ludsin SA and Matisoff G. The re-eutrophication of Lake Erie: Harmful algal blooms and hypoxia. Harmful algae. 2016; 56: 44-66.

Weide DM, Fritz SC, Brinson BE, Thompson LG, Billups WE. Freshwater diatoms in the Sajama, Quelccaya, and Coropuna glaciers of the South American Andes, Diatom Research. 2017; 32(2): 153-162.

Welsh Jr HH, Pope KL, Boiano D. Sub-alpine amphibian distributions related to species palatability to non-native salmonids in the Klamath mountains of northern California. Diversity and Distributions. 2006; 12(3): 298-309.

Wetlands International. Waterbird population estimates, 4th ed. Wageningen, The Netherlands: Wetlands International. 2006.

White D, Hinzman L, Alessa L, Cassano J, Chambers M, Falkner K, Francis J, Gutowski WJ, Holland M, Holmes RM, Huntington H. The arctic freshwater system: Changes and impacts. Journal of geophysical research: Biogeosciences. 2007; 112(G4), DOI: 10.1029/2006JG000353.

Whiteford EJ, McGowan S, Barry CD and Anderson NJ. Seasonal and regional controls of phytoplankton production along a climate gradient in South-West Greenland during ice-cover and ice-free conditions. Arctic, Antarctic, and Alpine Research. 2016; 48(1): 139-159.

Wickham H. 2009: ggplot2: Elegant Graphics for Data Analysis. New York: Springer-Verlag. Wilhelm S and Adrian R. Impact of summer warming on the thermal characteristics of a polymictic lake and consequences for oxygen, nutrients and phytoplankton. Freshwater Biology. 2008; 53(2): 226-237.

Williams MW, Baron JS, Caine N, Sommerfeld R and Sanford R. Nitrogen saturation in the Rocky Mountains. Environmental Science & Technology. 1996; 30(2): 640-646.

Williams JJ, Beutel M, Nurse A, Moore B, Hampton SE and Saros JE. Phytoplankton responses to nitrogen enrichment in Pacific Northwest, USA Mountain Lakes. Hydrobiologia. 2016; 776(1): 261-276.

Williams MW, Knauf M, Cory R, Caine N, Liu F. Nitrate content and potential microbial signature of rock glacier outflow, Colorado Front Range. Earth Surface Processes and Landforms: The Journal of the British Geomorphological Research Group. 2007; 32(7): 1032-47.

192

Williams JJ, Nurse A, Saros JE, Riedel J, Beutel M. Effects of glaciers on nutrient concentrations and phytoplankton in lakes within the Northern Cascades Mountains (USA). Biogeochemistry. 2016; 131(3): 373-85.

Williams MW, Tonnessen KA. Critical loads for inorganic nitrogen deposition in the Colorado Front Range, USA. Ecological Applications. 2000; 10(6): 1648-65.

Williamson CE, Sanders RW, Moeller RE and Stutzman PL. Utilization of subsurface food resources for zooplankton reproduction: Implications for diel vertical migration theory. Limnology and Oceanography. 1996; 41(2): 224-233.

Williamson CE, Saros JE, Vincent WF, Smol JP. Lakes and reservoirs as sentinels, integrators, and regulators of climate change. Limnology and Oceanography. 2009; 54(6 part 2): 2273-82.

Wilson R, Glasser NF, Reynolds JM, Harrison S, Anacona PI, Schaefer M, Shannon S. Glacial lakes of the Central and Patagonian Andes, Global and Planetary Change. 2018; 162: 275-291.

Winder M and Sommer U. Phytoplankton response to a changing climate. Hydrobiologia. 2012; 698(1): 5-16.

Winder M, Boersma M and Spaak P. On the cost of vertical migration: are feeding conditions really worse at greater depths? Freshwater Biology. 2003; 48(3): 383-393.

Winslow L, Read J, Woolway R, Brentrup J, Leach T, Zwart J, Albers S and Collinge D. 2019. Package ‘rLakeAnalyzer’.

Wolfe AP, Baron JS, Cornett RJ. Anthropogenic nitrogen deposition induces rapid ecological changes in alpine lakes of the Colorado Front Range (USA). Journal of Paleolimnology. 2001; 25(1): 1-7.

Wolfe AP, Cooke CA, Hobbs WO. Are current rates of atmospheric nitrogen deposition influencing lakes in the eastern Canadian Arctic? Arctic, Antarctic, and Alpine Research. 2006; 38(3): 465-76.

Wolfe AP, Hobbs WO, Birks HH, Briner JP, Holmgren SU, Ingólfsson Ó, Kaushal SS, Miller GH, Pagani M, Saros JE, Vinebrooke RD. Stratigraphic expressions of the Holocene–Anthropocene transition revealed in sediments from remote lakes. Earth-Science Reviews. 2013; 116: 17-34.

Wolfe AP, Van Gorp AC, Baron JS. Recent ecological and biogeochemical changes in alpine lakes of Rocky Mountain National Park (Colorado, USA): a response to anthropogenic nitrogen deposition. Geobiology. 2003; 1(2): 153-68.

Wood SN. 2017. Generalized additive models: an introduction with R. Chapman and Hall/CRC.

193

Wood SN. 2019. R Package ‘mgcv.’

Xiankai L, Jiangming M, Shaofeng D. Effects of nitrogen deposition on forest biodiversity. Acta Ecologica Sinica. 2008; 28(11): 5532-48.

Yalcin K, Wake CP. Anthropogenic signals recorded in an ice core from Eclipse Icefield, Yukon Territory, Canada. Geophysical Research Letters. 2001; 28(23): 4487-90.

Yallop ML and Anesio AM. Benthic diatom flora in supraglacial habitats: a generic-level comparison. Annals of Glaciology. 2010; 51(56): 15-22.

Yang LH, Bastow JL, Spence KO, Wright AN. What can we learn from resource pulses? Ecology. 2008; 89(3): 621-34.

Yao T, Pu J, Lu A, Wang Y, Yu W. Recent glacial retreat and its impact on hydrological processes on the Tibetan Plateau, China, and surrounding regions, Arctic, Antarctic, and Alpine Research. 2007; 39(4): 642-650.

Yermokhin MV, Sushchik NN, Tabachishin VG, Kalacheva GS, Kolmakova AA, Gladyshev MI. Amphibia as a Vector of Transfer of Long-Chain Polyunsaturated Omega-3 Fatty Acids from Aquatic to Terrestrial Ecosystems. In Doklady Biochemistry and Biophysics 2018 (Vol. 481, No. 1, pp. 195-197). Pleiades Publishing.

Zannetti P. Dry and Wet Deposition, in Air Pollution Modeling: Theories, Computational Methods, and Available Software, ed. P. Zannetti, Springer, Boston, USA, 1990, 249-262.

Zeng H, Jia G, Forbes BC. Shifts in Arctic phenology in response to climate and anthropogenic factors as detected from multiple satellite time series. Environmental Research Letters. 2013; 8(3), DOI: 10.1088/1748-9326/8/3/035036.

Zhang G, Xie H, Kang S, Yi D, Ackley SF. Monitoring lake level changes on the Tibetan Plateau using ICESat altimetry data (2003–2009), Remote Sensing of Environment. 2011; 115(7): 1733- 1742.

Zhang G, Yao T, Xie H, Kang S, Lei Y. Increased mass over the Tibetan Plateau: From lakes or glaciers? Geophysical Research Letters. 2013; 40(10): 2125-2130.

Zhang Y, Chen W, Smith SL, Riseborough DW, Cihlar J. Soil temperature in Canada during the twentieth century: Complex responses to atmospheric climate change. Journal of Geophysical Research: Atmospheres. 2005; 110(D3), DOI: 10.1029/2004JD004910.

Zwaaftink CG, Grythe H, Skov H, Stohl A. Substantial contribution of northern high-latitude sources to mineral dust in the Arctic. Journal of Geophysical Research: Atmospheres. 2016; 121(22): 13-678.

194

Biography of the Author

Ben grew up in Maine playing in vernal pools by his house. He currently lives in

Massachusetts with his spouse, Johanna Cairns, and two Maine coon cats, Peety and Charlie. He is a candidate for the Doctor of Philosophy degree in Ecology and Environmental Science at the

University of Maine in May 2020.

195