Holocene Carbon Dynamics in the Patterned Peatlands of the Hudson Bay Lowland, Canada: Reducing Landscape-Scale Uncertainty in a Changing Climate

Maara Susanna Packalen

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Geography University of Toronto

Holocene Carbon Dynamics in the Patterned Peatlands of the Hudson Bay Lowland, Canada: Reducing Landscape- Scale Uncertainty in a Changing Climate

Maara Susanna Packalen

Doctor of Philosophy

Department of Geography

University of Toronto

2015

Abstract

Northern peatlands have accumulated ~ 500 Pg of carbon (C) over millennia, and contributed to a net climate cooling. However, the fate of peatland C pools and related climate-system feedbacks remain uncertain under scenarios of a changing climate and enhanced anthropogenic pressure. Here, Holocene C dynamics in the Hudson Bay Lowland, Canada (HBL) are examined at the landscape scale with respect to glacial isostatic adjustment (GIA), climate, and ecohydrology. Results confirm that the timing of peat initiation in the HBL is tightly coupled with GIA, while contemporary climate explains up to half of the spatial distribution of the total C mass. Temporal patterns in C accumulation rates (CARs) are related to peatland age, ecohydrology, and possibly paleoclimate, whereby CARs are greatest for younger, minerotrophic peatlands. Rapid and widespread peatland expansion in the HBL has given rise to a globally significant C pool, in excess of 30 Pg C and two-thirds of which is of late Holocene age. Yet,

long-term decomposition of previously accrued peat has potentially resulted in some C losses, especially during the late Holocene when the landscape was occupied by an abundance of minerotrophic peatlands and climate was characterized by more precipitation and similar-to- colder temperatures than present. Model deconstruction of HBL C dynamics indicate that 85% of

C losses occurred during the late Holocene, while spatio-temporal scaling of modern methane

(CH4) emissions suggest a potential flux of 1 – 7 Pg CH4 to the late Holocene atmosphere, which provides evidence of a peatland contribution to the late Holocene CH4 rise recorded in ice cores.

Although HBL peatlands may continue to function as a net C sink, conservative climate scenarios predict warmer and wetter conditions in the next century – beyond the HBL’s range of past climate variability, yet within the peatland climate domain – with implications for primary production and decomposition. Further investigation into controls on spatial-temporal C dynamics may reduce uncertainty concerning the HBL’s potential to remain a net C sink under future climate and resource management scenarios, and contribute to our understanding of global peatland C-climate dynamics.

iii

Acknowledgements

Research Acknowledgements

I would like to express my deepest gratitude to my major supervisor Professor Sarah Finkelstein. You have been a tremendous mentor for me. I would like to thank you for guiding and encouraging my research, fostering idea development through countless discussions, and supporting unique opportunities for field work, collaboration, academic and professional enrichment, which greatly enhanced my graduate studies at the University of Toronto (U of T).

I would also like to thank my committee members at the U of T, Professors Nathan Basiliko (now at Laurentian University), Jing Chen, and William Gough, and Dr. James McLaughlin at the Ontario Ministry of Natural Resources and Forestry (MNRF). Each of you has contributed meaningful and unique perspectives at various stages of my doctoral research, and I am enormously grateful. Special thanks to Dr. McLaughlin for establishing a solid foundation to pursue peatland research in the Far North of Ontario, and specifically across the Hudson Bay Lowland. This logistically-challenging research would not have been possible, at the scale that it was conducted, without your tireless efforts to develop compelling research strategies and establish strong collaborations among government, academia, industry, and First Nation communities.

Special thanks are also extended to the MNRF, for supporting my request for an educational leave of absence from my permanent position with the Forest Research and Monitoring Section, which afforded dedicated time to pursue doctoral studies. And finally, to my external appraiser, Professor Merritt Turetsky (University of Guelph) – my sincerest appreciation is extended to you for reviewing my dissertation and providing extremely thoughtful and meaningful feedback. Your research contributions inspired me early in my career and primed me to undertake this amazing doctoral research journey; and your ongoing professional contributions and enthusiasm for research continue to inspire me today. Thank you.

Research funding and field support was provided by the MNRF’s Applied Research and Development Branch (now Science and Research Branch) and Far North Branch, under the auspices of projects CC-021 and FNIKM 028. Many thanks to MNRF field crews for peat coring and site surveys (2009 – 2011) and De Beers Canada for logistical support in the vicinity of the Victor Diamond Mine (2009 – 2012). Many thanks also to MNRF lab personnel for geochemical analyses. Additional support for field work and radiocarbon dating was provided by grants (327197-11 and 331284-11) from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Ontario Ministry of the Environment and Climate Change (MOECC) through the Climate Change and Multiple Stressor Research Program at Laurentian University. Many thanks for field and logistical support from the U of T (Department of Geography), Laurentian University (Living with Lakes Centre), Queen’s University (PEARL), Hearst Air Service, and Albert’s Fish Camp during a field campaign near Hawley Lake, ON (2011). Thanks also for field and logistical support from Western University, McGill University, Ministry of Environment and Climate Change (MOECC), and De Beers Canada, during a field campaign near the Victor Diamond Mine (2012). Extensive data syntheses were completed in support of this research. Thanks are extended to Dr. Arthur Dyke for providing access to the Canadian basal radiocarbon database; Professor Peter Kuhry for contributing raw peat core data; and Dr. Dan McKenney for access to gridded climate data.

Graduate stipend support was provided by an NSERC Alexander Graham Bell Canada Postgraduate Scholarship (CGSD2-426611-2012), Ontario Graduate Scholarship, several graduate student awards from the Department of Geography at the University of Toronto, and the Canadian Northern Studies Trust Scholarship from the Association of Canadian Universities for Northern Studies (2010 – 2014). Field research was further supported by graduate student research grants from the Society of Wetland Scientists (1) and Aboriginal Affairs and Northern Development Canada’s Northern Scientific Training Program grants (2).

I am extremely grateful for opportunities to enrich my graduate studies. I spent an invaluable three months as a visiting scientist at Columbia University, Lamont Doherty Earth Observatory (LDEO), under the auspices of the NSERC Michael Smith Foreign Study Supplement. This

special research opportunity supported extension of my doctoral research, under the guidance of Dr. Dorothy Peteet (co-affiliated with NASA-GISS), and included insightful and productive collaborations with Drs. Liz Corbett, Linda Heusser, and Jonathan Nichols. Thank you to each of you for welcoming me into your research group. Gratitude is also extended to Professor Zicheng Yu, at Lehigh University and Professors Tim Moore and Nigel Roulet at McGill University, for including me in stimulating peatland network meetings and research workshops. Graduate enrichment opportunities to study multivariate statistics at the University College London (2011) and to conduct field work in Sweden and across Finland (2012) were further supported by awards from the University of Toronto’s Centre for Global Change Science (2). Special thanks to Dr. Jukka Turunen from the Geological Survey of Finland, for mentorship and logistical support during my 2012 field campaign in Finland.

Personal Acknowledgements

My graduate journey would not have been as successful or fulfilling without the love and support of my parents, extended family and close friends. They have been my constant foundation, and I am forever indebted to them for their patience, advice, and interest in my work. Deepest thanks.

Many friends and colleagues at the U of T, within the peatland community, and from around the world have enriched my graduate experience beyond words. They are too numerous to name; however, I have deeply enjoyed the conversations, collaborations, and extra-curricular activities. I sincerely hope our paths will continue to cross, and wish each of them all the best.

Table of Contents

Abstract ...... ii

Acknowledgments...... iv

Table of Contents ...... vii

List of Tables ...... xi

List of Figures ...... xii

Chapter 1 Introduction ...... 1

1.1 Background ...... 1

1.1.1 Holocene carbon dynamics ...... 3

1.1.2 Peat development ...... 5

1.1.3 Circum-polar carbon accumulation...... 7

1.1.4 The Hudson Bay Lowland, Canada as a model ecosystem ...... 9

1.2 Research objectives ...... 11

1.2.1 Research questions ...... 12

1.2.2 Hypotheses ...... 13

1.3 General research approach ...... 13

1.3.1 Detailed peat records...... 14

1.3.2 Data syntheses ...... 14

1.4 Thesis structure and publication information ...... 15

1.4.1 Chapter 1 ...... 15

1.4.2 Chapter 2 ...... 15

1.4.3 Chapter 3 ...... 16

1.4.4 Chapter 4 ...... 16

vii

1.4.5 Chapter 5 ...... 16

Chapter 2 Carbon Storage and Potential Methane Production in the Hudson Bay Lowlands since Mid-Holocene Peat Initiation ...... 17

2.1 Abstract ...... 17

2.2 Introduction ...... 17

2.3 Methods...... 21

2.3.1 Data synthesis and new peat records ...... 21

2.3.2 Holocene peat initiation and carbon dynamics ...... 23

2.4 Results ...... 25

2.4.1 GIA and paleoclimate as controls of HBL peatland dynamics ...... 25

2.4.2 HBL carbon storage and potential Holocene methane emissions ...... 26

2.5 Discussion ...... 28

2.6 Conclusion ...... 36

2.7 Acknowledgements ...... 37

2.8 Author contributions ...... 38

2.9 Competing financial interests ...... 38

2.10 Tables ...... 39

2.11 Figures...... 49

Chapter 3 Quantifying Holocene Variability in Carbon Uptake and Release Since Peat Initiation in the Hudson Bay Lowlands, Canada ...... 52

3.1 Abstract ...... 52

3.2 Introduction ...... 53

3.3 Materials and Methods ...... 58

3.3.1 Study setting...... 58

3.3.2 Sample collection and data sources ...... 59

viii

3.3.3 Laboratory Analyses ...... 60

3.4 Results ...... 63

3.4.1 Carbon accumulation and peat decay rates ...... 63

3.4.2 Peatland carbon dynamics...... 65

3.5 Discussion ...... 67

3.5.1 Carbon accumulation patterns in the HBL ...... 67

3.5.2 Modeled peatland carbon dynamics in the HBL ...... 70

3.5.3 Net carbon balance and paleoclimate in the HBL ...... 72

3.6 Conclusion ...... 75

3.7 Acknowledgements ...... 76

3.8 Tables ...... 78

3.9 Figures...... 81

Chapter 4 Climate and Peat Type in Relation to the Spatial Distribution of the Peat Carbon Mass in the Hudson Bay Lowland, Canada ...... 86

4.1 Abstract: ...... 86

4.2 Introduction ...... 87

4.3 Study setting...... 92

4.4 Methods...... 93

4.4.1 Peat and bioclimatic data sources ...... 93

4.4.2 Peat physical properties ...... 94

4.5 Results ...... 95

4.5.1 Climate and carbon mass spatial relationships across the HBL ...... 95

4.5.2 Carbon mass variation among peat types ...... 98

4.6 Discussion ...... 99

4.7 Conclusions and future implications ...... 104

4.8 Acknowledgements ...... 106

4.9 Tables ...... 107

4.10 Figures ...... 109

Chapter 5 Conclusions and Future Research Directions ...... 115

5.1 Summary ...... 115

5.1.1 Peat initiation and carbon storage ...... 116

5.1.2 Holocene carbon dynamics ...... 117

5.1.3 Climatic controls of the distribution of the carbon mass ...... 118

5.2 Sources of Uncertainty and Future Research Directions ...... 119

5.2.1 HBL paleohydroclimate and carbon dynamics ...... 119

5.2.2 Carbon dynamics in permafrost peatlands and peatland pools ...... 120

5.2.3 Age-depth modeling and fen decay modeling ...... 120

5.2.4 Peat carbon dynamics in post-marine environments ...... 121

5.2.5 Peatland resilience and vulnerability to climatic change ...... 121

References ...... 122

List of Tables

Table 2.10-1 Location, age, and carbon mass of peat cores located in the Hudson Bay Lowlands, Canada (HBL), sorted by latitude...... 39

Table 3.8-1 14C-accelerator mass spectrometry (AMS) dating of peat macrofossil of known provenance for 10 new sites in the Hudson Bay Lowlands, Canada; sorted by increasing latitude...... 78

Table 3.8-2 Summary of deconstruction terms used to describe peatland carbon (C) dynamics in the Hudson Bay Lowlands, Canada (adapted from Yu, 2011)...... 80

Table 4.9-1 Peat physical properties of 42 sites in the Hudson Bay Lowland (HBL), Canada. . 107

Table 4.9-2 Peatland carbon mass distribution in the Hudson Bay Lowland, Canada...... 108

List of Figures

Figure 2.11-1 Land emergence and peatland expansion in the Hudson Bay Lowland, Canada. .. 49

Figure 2.11-2 Holocene peat initiation dynamics in the Hudson Bay Lowlands, Canada...... 50

Figure 3.9-1 Physical features and peat study locations in the Hudson Bay Lowlands, Canada. 81

Figure 3.9-2 Peat age, depth, and cumulative carbon (C) mass relationships and modeled exponential peat decay for the patterned peatlands of the Hudson Bay Lowlands, Canada...... 82

Figure 3.9-3 Holocene peat carbon (C) pools and modeled peat C terms (Yu, 2011) for the Hudson Bay Lowland, Canada, since peat initiation began ~ 8 ky BP...... 83

Figure 3.9-4 Holocene peatland area increase, carbon (C) accumulation, and net C balance, since peat inception for the Hudson Bay Lowlands, Canada...... 84

Figure 4.10-1 Physical features, study sites and carbon mass the Hudson Bay Lowland, Canada...... 109

Figure 4.10-2 Contemporary climate domain for peatlands of the Hudson Bay Lowland, Canada...... 110

Figure 4.10-3 Relationship between peat depth and carbon mass for well described peatlands in the Hudson Bay Lowland, Canada...... 111

Figure 4.10-4 Peat class distribution relative to peatland continentality in the Hudson Bay Lowland, Canada...... 112

Figure 4.10-5 Peat carbon mass relative to bioclimate in the Hudson Bay Lowland, Canada. .. 113

Figure 4.10-6 Total carbon mass (kg m-2) stored in bogs, fens and coastal peatland mesoforms, in the Hudson Bay Lowland, Canada...... 114

xii

Chapter 1 Introduction 1.1 Background

Peat-accumulating wetlands, termed peatlands, accrue carbon (C) rich terrestrial organic deposits over millennia, as a consequence of complex ecohydroclimatic interactions and under conditions where net primary productivity exceeds organic matter (OM) decomposition. Peatlands occupy

~ 3% of the global terrestrial extent, yet account for at least one third of the global soil organic C pool (Gorham, 1991; Hugelius et al., 2014). Although peatlands are globally distributed, northern high latitude peatlands represent ~ 90% of the global peatland distribution and store an estimated 250 – 500 Pg C (Loisel et al., 2014; Turunen et al., 2002;Yu, 2011). During the current interglacial, northern peatlands have sequestered atmospheric C dioxide (CO2) at a mean rate of

0.02 – 0.03 kg C m-2 y-1, equivalent to 25 to 50% of the atmospheric burden or at least 100 to 200 ppmv CO2 (Frolking and Roulet, 2007; Gorham, 1991). Hence, long-term C storage in peat deposits has resulted in a net climatic cooling over the Holocene (Frolking et al., 2011). Yet, complex interactions among internal peatland dynamics and external forcing suggest that the net

C balance of peatlands may vary over seasonal to millennial timescales (Yu, 2012) and may become increasingly vulnerable to loss under anticipated climate change scenarios.

Studies investigating the timing of peat initiation and lateral expansion reveal periods of significant methane (CH4) emissions, suggesting variability in the strength of peatlands as C sinks (MacDonald et al., 2006). Further, syntheses of global peat core records suggest that climate may be positively correlated with C accumulation at large spatial (degrees latitude) scales (Beilman et al., 2009) and long temporal (decadal to millennial) scales (Yu, 2012). The greatest rate of peat initiation in northern peatlands is reported to have occurred during the early

1 2

Holocene (11.5 – 7 ky), when climate was warmest and driest and favored rapid initiation and expansion of mostly minerotrophic peatlands (Jones and Yu, 2010; MacDonald et al., 2006).

However, since the early Holocene, peatlands appear to be on a trajectory of reduced C accumulation rates (Yu, 2011). The extent to which northern peatland initiation and expansion have influenced polar ice-core inferred Holocene atmospheric CH4 variability continues to fuel much debate (Reyes and Cooke, 2011), and uncertainty concerning the role of climate in peatland C dynamics persists. Hence, the mechanisms controlling northern peatland C dynamics and the role of northern peatlands in the Holocene C cycle are of great interest to the global C cycle research community.

Global circulation models predict increased hydroclimatic variability in northern high latitudes, whereby a warmer climate may enhance both vegetation productivity and OM decomposition depending in part upon concomitant alteration of the net moisture balance. Disproportionate and/or non-linear changes in the balance between production and decomposition, as a consequence of hydroclimatic variability and anthropogenic land conversion may have important implications for both the peatland C sink and C flux potentials. Thus, under future climate and land-use scenarios, the role and ecological functioning of northern peatlands could be significantly altered; yet, the magnitude of change and the mechanisms driving peatland C storage dynamics and climate-system feedbacks remain uncertain.

This dissertation research examines peatland C dynamics at the landscape scale and tracks C dynamics over the Holocene using complete peat profiles, for the purpose of better constraining the controls on the net C balance in northern peatlands. Here, the patterned peatlands of the HBL are considered as a model system to investigate controls on peat initiation, peatland development, and Holocene C dynamics vis-à-vis post glacial isostatic adjustment, paleoclimate,

3 and ecohydrology, as inferred by peat type. Additionally, the potential contribution of HBL peatlands to millennial-scale atmospheric CH4 trends is also investigated via model deconstruction of C sink-source dynamics and peatland development. The objectives of this research will be approached by coupling new and previously reported peat geochemical evidence with peat initiation and development chronologies, and model reconstructions of paleoclimate and paleotopography to track spatio-temporal changes in the peat C pool and elucidate possible controls on Holocene C dynamics in the HBL. Ultimately, this research is designed to improve our fundamental understanding of northern peatland C dynamics, and potentially provide a sound scientific rationale for Canada’s future resource management policies, in the context of a changing climate.

1.1.1 Holocene carbon dynamics

Although peatlands efficiently sequester CO2 from the atmosphere, some of the fixed CO2 may also be released as decomposition products in the form of CO2, CH4 and dissolved organic C

(DOC). Wetland ecosystems, including peatlands are important contemporary sources of CH4 to

-1 the atmosphere, releasing an estimated 20 – 45 Tg CH4 y (Mikaloff Fletcher et al., 2004). CH4 is an important greenhouse gas due to its high global warming potential (~ 24 x CO2), abundance

(third, after water vapor and CO2), chemical reactivity in the atmosphere, and sensitivity to changes in climate (Pickett-Heaps et al., 2011). Although highly variable, peatland C losses in the form of CH4appear to be greatest from minerotrophic peat forms and/or under ecohydrological conditions consistent with fen landforms, thermokarsts, and peatland flarks/pools (Pelletier et al., 2007; Bubier et al., 1993a; Olefeldt et al., 2013).

Polar ice-core records reveal that atmospheric CH4 concentrations rose twice during the

Holocene: first in the early Holocene during a period of continental deglaciation (11 – 8 ky BP)

and second in the late Holocene (4 ky BP – present), following a mid-Holocene atmospheric CH4 decline (Brook et al., 2000). Global syntheses of northern peatland initiation dates have correlated the rapid development of peatlands in the early Holocene to the rise of atmospheric

CH4 during the same period (MacDonald et al., 2006). Peak early Holocene atmospheric CH4 concentrations have been associated with the existence of predominantly warm, wet minerotrophic fens, recognized CH4 sources in the modern landscape. MacDonald et al. (2006) proposed that this was related to peak insolation, inferred from Milankovitch cycles, where warming effects were likely delayed in the northern hemisphere by ice albedo feedbacks during the retreat of continental ice sheet. By contrast, the mid-Holocene has been associated with a transition to mainly ombrotrophic Sphagnum bogs, resulting in reduced atmospheric CH4 contributions and moderate CO2 sequestration. However, atmospheric CO2 increased rather suddenly during the mid-to late Holocene and the oceanic buffering response (carbonate compensation pump) has been invoked as the most probable mechanism (Yu, 2011).

With declining insolation, below average apparent C accumulation in existing peatlands, and little evidence of new peat initiation, a connection between the late Holocene atmospheric CH4 rise and peatland initiation and expansion has not been well established. Consequently, many alternative hypotheses have been advanced to explain the late Holocene atmospheric CH4 rise, including: (1) early anthropogenic activity, namely rice cultivation and deforestation (Ruddiman et al., 2008); (2) mid-to-high latitude wetland expansion rather than initiation (Korhola et al.,

2010); and (3) tropical wetland contributions (Yu, 2011; Page et al., 2011). However, an apparent lack of evidence may stem from: (1) few peat initiation and expansion records from late emerging, and potentially globally significant peat complexes, such as the HBL (Yu, 2012); (2) a propensity to sample the deepest part of the peatland, which may result in an inability to effectively consider the role of peatland expansion in late Holocene C dynamics (Loisel et al.,

2014); and (3) the possibility of non-traditional peatland succession dynamics favouring a return to fen-like conditions due to changes in climate and/or other biophysical processes (van Bellen et al., 2013).

As a result, evidence in support of a clear connection between northern peatland initiation and expansion dynamics as a control on polar ice-core inferred Holocene atmospheric CH4 variability is lacking, and the source of the CH4 rise beginning during the mid-Holocene toward a pre- industrial maximum remains controversial (Ruddiman et al., 2011). Additional peatland initiation datasets from other time periods in the Holocene are needed, together with consideration of whole peat complexes, to understand how peatland development may account for variation in atmospheric CH4 concentrations, over centennial to millennial timescales.

1.1.2 Peat development

The peat profile is comprised of an acrotelm and a catotelm. The acrotelm by definition lies above the seasonal water table, typically within 0.5-m of the surface, and experiences the most variable seasonal water and C dynamics. The poorly- to moderately-decomposed organic matter that accumulates in the acrotelm is typically of low density resulting in large pore spaces and oxygenation. Consequently, relatively rapid decomposition occurs in the acrotelm compared with deeper portions of the peat profile. The portion of the peat profile that lies below the seasonal water table, termed catotelm, is characterized by slow decomposition and long-term peat accumulation. Density typically increases with depth along the peat profile, which reduced oxygen penetration and stimulates the development of anaerobic conditions, while decreasing hydraulic conductivity results in the decoupling of the accumulating peat column from the underlying mineral sediments. The boundary region between the aerobic and anaerobic states,

6 occurs at the mean summer water table minimum, and appears to be an extremely active zone for decomposition processes (Belyea and Baird, 2006).

While seasonally waterlogged conditions are frequently associated with peat accumulation, rates of accumulation vary as a function of a series of factors, including: vegetation assemblage, OM quality, biogeochemical interactions, bioclimate (e.g., temperature and moisture balance), topography, and disturbance (Belyea and Baird, 2006; Turetsky et al., 2005; Waddington et al.,

2015). Peatland hydrology is fundamental to comprehending peat accumulation dynamics and subsequent peatland development, C accumulation rates, and greenhouse gas exchanges

(Frolking and Roulet, 2007).Vegetation distribution and photosynthetic potential are closely related to the net water balance, which together influence peat accumulation through effective net primary production (Sonnentag et al., 2008), and contribute to the development of divergent peat classes (e.g., bogs and fens) that can be used as an indicator of ecohydrology. Further, ecohydrologically-driven vegetation distributions are coupled to biogeochemical processes, such as microbially-driven OM decomposition, through the influence of OM quality, cation exchange capacity, acidity, and nutrient cycling (Turetsky et al., 2008; Waddington et al., 2015). Coupled with temperature, decomposition processes produce DOC, CO2 and CH4 during terminal mineralization, all of which contribute to variability in spatio-temporal C dynamics (Turetsky,

2004). Topography is related to hydrological redistribution through surface and subsurface flow, resulting in feedbacks to vegetation assemblage, nutrient availability, and peatland development

(Glaser et al., 2004a; Glaser et al., 2004b). However, disturbances such as historical fire patterns and contemporary land management can result in sudden mass C losses (Turetsky et al., 2011).

Uncertainty exists regarding burn severity and fire regimes across the pan-boreal domain of northern peatlands, especially as it relates to understanding peatland vulnerability in terms of critical threshold conditions for changing net water balances.

1.1.3 Circum-polar carbon accumulation

Gorham (1991) estimated C accumulation for boreal and subarctic peatlands to be approximately

29 g C m-2 y-1. These estimates were based on mean peat depth, bulk density and peat accretion rate of 2.3 m, 112 g dm-3 and 0.5 mm y-1, respectively. However, peat decays gradually through time, as described by Clymo et al. (1998). In consideration of a constant decay process, the C accumulation rate reported in Gorham (1991) was adjusted to 23 g C m-2 y-1. Turunen et al.

(2002) presented a C accumulation rate 18.5 g C m-2 y-1 for over 1300 dated Finnish boreal and subarctic peatlands, with a mean peat depth of 1.52 m. Further analysis revealed that raised bogs accrued C at a greater rate than aapa-mires (also known as patterned peatlands), with reported C accumulation rates of 26.1 g C m-2 y-1 compared to 17.3 g C m-2 y-1, respectively. Similarly, bogs were reported to accrue C at a greater rate than fens (20.8 g C m-2 y-1 compared with 16.9 g C m-

2 y-1, respectively). The lower mean C accumulation rates reported in this study were attributed to at least three factors: (1) terrestrialized peatlands have traditionally been over-represented in the C accumulation literature relative to paludified systems; (2) mean peat depths in Finland were 1.5 m compared to 2.2 to 2.3 m reported for other northern peatlands; and (3) the mean bulk density used elsewhere may be too large, potentially resulting in an over-estimation of the total C pool.

Long-term mean C accumulation rates in the West Siberian peatlands, Russia ranged from 12.1 to 23.7g C m-2 y-1 (Turunen et al., 2001); while, C accumulation rates in this region during the last 2 ky ranged from 3.8 to 44.1 g C m-2 y-1 (Beilman et al., 2009). In the latter study, C accumulation was significantly correlated to modern mean annual air temperature, where maximum rates occurred between -1 and 0 °C. As well, fossil plant composition was significantly related to C accumulation, where C content was lowest in association with

Sphagnum remains (Beilman et al., 2009). C accumulation rates in Canadian peatlands are variable as well. Vitt et al. (2000) found that continental peatlands in western Canada accumulated C at a rate of 19.4 g C m-2 y-1 and began accumulating peat around 9 ky BP.

Western Canadian peatlands experienced maximum C accumulation rates during the mid-

Holocene, hindered in the early Holocene by enhanced drying due to maximum insolation and in the late-Holocene by permafrost development (Vitt et al., 2000). By comparison, peatlands in the

East James Bay Lowland, Quebec accumulated C at a mean rate of 16.2 to 18.5 g C m-2 y-1, while recent peat accumulated at a mean rate of 73.6 g C m-2 y-1 (Loisel and Garneau, 2010; van

Bellen et al., 2011a). Similar to the continental peatlands in western Canada, maximum C accumulation occurred during the mid-Holocene in the East James Bay Lowland, Quebec; however, in this case it was attributed to rapid peat initiation and lateral expansion beginning

~ 7.5 ky, as a consequence of delayed deglaciation. Further, reduced late-Holocene C accumulation rates were attributed to neo-glacial cooling and surface drying (van Bellen et al.,

2011b).

Today, there is a growing database of northern peatland C accumulation rates (Loisel et al.,

2014), with a noticeable gap in the HBL and Far East Russia (Yu, 2011). As many studies have quantified C accumulation using cores sampled from the deepest part of the peatland, uncertainty regarding the variability of C accumulation within a single peatland has arisen. As a result, this uncertainty in spatial variability reduces confidence in upscaled total peat C pool estimates.

Thus, quantifying the intra- and inter peatland C accumulation has been identified as a high priority in future C accumulation quantification (Korhola et al., 2010; van Bellen et al., 2011b).

1.1.4 The Hudson Bay Lowland, Canada as a model ecosystem

Development during the current interglacial of a nearly continuous peat cover in the Hudson Bay

Lowland, Canada (HBL) has yielded one of the world’s largest peatland complexes and a C reservoir of potential global significance. As such, the HBL may play an important role in moderating global climate and long-term atmospheric C dynamics. Yet, few data are available from this largely unexplored and undisturbed landscape to support either (1) quantification of the magnitude of the C reservoir or (2) an understanding the processes that control peatland development and long-term C dynamics.

The HBL is a poorly drained plain with peatlands covering of an area of 325, 000 km2 (Riley,

2011), second only in size to the Western Siberian Plain (Martini, 2006). This region was shaped by Pleistocene ice sheets, which resulted in the deposition of till and diamicton, eskers and moraines, and glacio-marine sediments, overtop of the underlying Paleozoic sedimentary bedrock (Martini, 2006). Climate in the HBL is a microthermal, strongly influenced by Arctic air masses and strong winds (Martini, 2006). Modern climate in the HBL is characterized by a mean annual temperature of -4 °C, with cold winters (January daily mean -15 to -18 °C) and warm summers (July daily mean 12 to 18 °C). Precipitation patterns are characterized by a moderately deep snow accumulation (203 to 241 cm) and moderate total precipitation (603 to 660 mm)

(Martini, 2006). Contemporary permafrost features in the HBL range from sporadic occurrences in the southern regions near James Bay, to discontinuous permafrost in the middle of the HBL and continuous permafrost in the region immediately south of Hudson Bay, which together potentially influence < 10% of the region.

Under the weight of the continental ice sheets, subsidence upwards of 200 m gave way to inundation by glacial seas upon glacial retreat (Peltier, 2004). During the early Holocene (< 7.5

10 ky) the HBL was largely covered by the Tyrrell Sea (Riley, 2011). The post-glacial HBL, influenced by isostatic rebound and glacial sea emergence, has an average slope of 0.5 m km-1 and contains expansive patterned peatlands with ridge-flark mesoforms, marsh wetlands and tidal flats alternating with beach ridges near the coast, and inland bog-fen peatland complexes, with bogs dominating the more inland reaches (Riley, 2011). Average peat depth ranges from less than 40 cm in coastal areas to greater than four meters at inland bogs (Riley, 2011).

Peatlands in the HBL likely began to initiate in the mid-Holocene as land emerged from the post- glacial Tyrrell Sea; however, the timing of peat initiation relative to land emergence remains unconfirmed. C accumulation rates are not well established for the patterned peatlands of the

HBL and few vertically dated peat records exists, thus the magnitude of the C store and

Holocene C dynamics remain unexplored. CH4 contributions from the modern HBL was recently estimated to be 2.3 Tg y-1(Pickett-Heaps et al., 2011), equivalent to one tenth of the global northern peatland annual contribution. Mesoform pattern development in the HBL of Sphagnum ridges, herbaceous lawns/flarks, and pools may provide a viable source of persistent CH4 emissions over seasonal to millennial timescales, through local redistribution of water and nutrients and associated ecological feedbacks.

The HBL is an important region for examining the role of peatland initiation and expansion on the late Holocene CH4 rise due to its delayed post-glacial emergence, yet it has hitherto not been considered in global-scale syntheses. Given the delayed peat initiation and expansion in the HBL relative to other northern peatlands and the size of the peat complex, investigation of HBL C dynamics in terms of potential global climate forcing is a high priority. Yet, the role of the HBL peat complex in long-term climate dynamics and the fate (and potential vulnerability) of its

11 peatland C pools and related climate system feedbacks remain uncertain under scenarios of a changing climate and enhanced anthropogenic pressure.

1.2 Research objectives

Uncertainty exists regarding the mechanisms driving northern peatland expansion and pattern development, the fate of peatland C pools following hydroclimatic change and long-term peatland-climate feedbacks. To date, an expanding peat inventory has yielding insight into rates of C accumulation in northern peatlands; however, few studies have deconstructed these rates into C flux terms (C uptake and C release) (Yu, 2011). Observed patterns in timing of peatland initiation frequency and possibly lateral expansion have been linked with global atmospheric

CO2 and CH4 fluctuations. An important next step in analyzing peatland C accumulation is to quantitatively evaluate the role of peatlands in the late-Holocene rise in CO2 and CH4. Further, a quantitative and mechanistic understanding of these fluxes over the Holocene may inform projections of the northern peatland C pool response to anticipated hydroclimatic variability and potential climate system feedbacks.

While climate has been invoked as the dominant control on northern peatland initiation (Jones and Yu, 2010; MacDonald et al., 2006), post-glacial isostatic adjustment may be an even more fundamental control of peatland initiation and expansion in the dynamic landscape of the HBL, where rates of isostatic rebound are among the highest globally. Working under the hypothesis that peatland initiation, expansion, and pattern development (as a proxy for ecohydrology) are mechanisms of persistent peatland-derived atmospheric CH4 sources, the overall objective of this dissertation research is to investigate landscape-scale controls on C dynamics in the HBL and the potential role of the HBL in contributing to the mid-to-late Holocene rise in atmospheric CH4. In

12 keeping with this objective, this dissertation research seeks to integrate long-term C accumulation, peatland development, and ecohydrology into a quantitative assessment of C storage and potential CH4 flux from a mid-to-late Holocene HBL landscape.

1.2.1 Research questions

The guiding question of this dissertation research considers whether (1) the timing of peatland initiation, (2) the rate of peatland lateral expansion, and (3) the paleoclimate interact at the landscape scale in the HBL to drive Holocene C dynamics and the observed rise in atmospheric

CH4 during the mid- to late Holocene. To address this overarching research question, each chapter provides evidence from a range of perspectives (spatio-temporal and local to landscape scales) with the objective to distinguish between allogenic and autogenic controls on peatland C dynamics in the HBL. Specific sets of research questions are detailed below.

1. How does the timing of peat initiation in the HBL vary in relation to glacial isostatic

adjustment-driven land emergence and paleoclimate; how much C is sequestered in the

HBL peat complex; what is the potential magnitude of peatland derived CH4 losses

during the period of record; and how do peatland initiation and expansion dynamics

relate to trends in atmospheric CH4 concentrations?

2. What is the apparent rate of C accumulation (CAR) for HBL peatlands; how does CAR

and its modeled C flux terms (C uptake and release) vary through the period of record;

and in relation to paleoclimate, land emergence, and peatland lateral expansion?

3. How does the total C mass vary within the HBL climate domain; and among major peat

types; and is there evidence of both allogenic and autogenic controls on the spatial

distribution of the peat C mass in the HBL?

1.2.2 Hypotheses

The hypotheses are: (1) peat initiation is tightly coupled with land emergence, and the principle mechanism of peat initiation is primary succession or paludification, driven by rapidly emerging post-glacial sediments; (2) the C mass in the HBL is of global significance, due in part to the size of the peat complex and the relative lack of disturbance since peat initiation began; (3) C will accumulate fastest during relatively warm periods due to increased primary productivity and potentially longer growing periods unless enhanced decomposition is also supported; (4) wet, cool periods will reduce C accumulation rates due to changes in vegetation assemblages and potentially greater C losses, including possibly CH4 in relation to C gains; (5) climate shall be related to the distribution of the C mass, such that warmer climates will support greater C masses; and (6) long-term C accumulation will occur faster in bog peatlands than in fen peatland ecosystems, due to variations in hydrothermal properties and decomposition processes, whereby vegetation assemblage will influence decomposition (e.g., vascular > Sphagnum) of the accumulated peat and C accumulation rates (e.g., Sphagnum > vascular).

1.3 General research approach

Detailed analyses of the new peat profiles were combined with previously reported peat records and publically available datasets to examine controls on Holocene C dynamics in the HBL.

Complete peat profiles were collected between 2009 and 2012 to support this dissertation research. Intensive study sites were established along a hydroclimatic gradient in each of the following regions: (1) sporadic permafrost zone in the southern HBL, near Kinoje Lake, Ontario

(52 °N, 82 °W); (2) discontinuous permafrost zone in the central HBL near Attawapiskat,

Ontario (53 °N, 84 °W); (3) continuous permafrost zone in the northern HBL near Hawley Lake,

Ontario and along the Hudson Bay coast (55.5 °N, 84.5 °W). At the intensive study sites, one to five transects measuring 500 m were established in both fens and bogs. One to five long cores

(defined as complete peat profiles to mineral contact) were collected every 100 m to capture the variability between the peat margins and centre to track peatland lateral expansion. Additional complete peat profiles were sampled across the HBL and along the former Tyrrell Sea margin

(western shore of the ~ 7-ky Hudson Bay coast), and includes sites located in the vicinity of: 54

°N, 88 °W; 51 °N, 87 °W; 51 °N, 86 °W; 51 °N, 80 °W.

1.3.1 Detailed peat records

Detailed peat records were developed using the newly sampled peat cores form the HBL. All new peat records presented in this dissertation were characterized for peat physico-chemical properties, including measurements of (1) bulk density; (2) C, nitrogen, calcium, potassium, phosphorus, and organic matter content; and (3) C radiometric dating of the peat-mineral contact

(termed 14C basal date). A selection of peat cores also includes vertical C radiometric dating

(14C), peat humification analysis, and plant macrofossil assessment. A literature review yielded additional detailed peat records from across the HBL that included at minimum basal-dated peat records, organic matter content, peat depth, and site location information.

1.3.2 Data syntheses

Publically available datasets were used to support landscape-scaling of controls on peatland development and long-term C dynamics. These include (1) a basal peat-age database, supported by a detailed literature review; (2) records of peat depths; (3) pollen-inferred paleoclimate that reported summer and winter temperature and annual precipitation anomalies relative to the contemporary climate normals; (4) gridded contemporary climate informed by climate station

15 outputs, interpolated continuously across the landscape; (5) spatially-explicit rates of glacial isostatic adjustment (one degree latitudinal-longitudinal grid) and paleotopography (0.5 ky elevation predictions); and (6) ice-core inferred atmospheric CO2 and CH4 concentration for the period of record examined in the dissertation.

1.4 Thesis structure and publication information

1.4.1 Chapter 1

Chapter 1 provides an introduction to the dissertation, including (1) a brief literature review in relation to existing knowledge gaps; (2) the context for the research objectives together with specific research questions and hypotheses; (3) a summary of the methodologies; and (4) detailed publication information.

1.4.2 Chapter 2

Chapter 2 provides the first estimate of the HBL C pool informed by detailed analyses of peat records sampled directly from the HBL. It also reports on the coupling between the timing of peat initiation and land emergence from the post-glacial Tyrrell Sea since peatland development began in the mid-Holocene, the relationship between peat initiation intensity and paleoclimate, and estimates potential CH4 efflux as peatlands expanded across the landscape, in the context of

Holocene-scale atmospheric trace gas dynamics. The manuscript was published in Nature

Communications (DOI: 10.1038/ncomms5078). The authors of the manuscript are Maara S.

Packalen (MSP), Sarah A. Finkelstein (SAF), and James W. McLaughlin (JWM). MSP designed and performed research, analyzed data, and wrote the paper; SAF and JWM designed and performed research. All authors commented on the manuscript at all stages. The manuscript was

16 subsequently selected and featured in the Permafrost Special Issue in the journal Nature (9 April

2015; www.nature.com/nature/focus/permafrost/).

1.4.3 Chapter 3

Chapter 3 quantifies C accumulation since mid-Holocene peat initiation in relation to paleoclimatic variation and peat type, and partitions potential C uptake and release as the HBL peat complex developed. The manuscript was published in The Holocene (DOI:

10.1177/0959683614540728). The authors of the manuscript are Maara S. Packalen (MSP) and

Sarah A. Finkelstein (SAF). MSP designed and performed research, analyzed data, and wrote the paper; SAF designed and performed research. Both authors commented on the manuscript at all stages.

1.4.4 Chapter 4

Chapter 4 examines the distribution of the HBL peat C mass in relation spatially explicate contemporary climate. It further reports on the peat physico-chemical parameters by major peat type (bog, fen and coastal peatlands). It is in revision in the Journal of Geophysical Research –

Biogeosciences (Manuscript #: 2015JG002938). The authors of the manuscript are Maara S.

1.4.5 Chapter 5

Chapter 5 provides an overall summary of the dissertation, outlines important sources of uncertainty, and identifies future research directions.

Chapter 2 Carbon Storage and Potential Methane Production in the Hudson Bay Lowlands since Mid-Holocene Peat Initiation 2.1 Abstract

Peatlands have influenced Holocene carbon (C) cycling by storing atmospheric C and releasing

methane (CH4). Yet, our understanding of contributions from the world’s second largest

peatland, the Hudson Bay Lowlands, Canada (HBL), to peat-climate-C dynamics is constrained

by the paucity of dated peat records and regional C data. Here, I examine HBL peatland

development in relation to Holocene C dynamics. I show that peat initiation in the HBL is tightly

coupled with glacial isostatic adjustment through most of the record, and occurred within

suitable climatic conditions for peatland development. HBL peatlands initiated most intensively

in the mid-Holocene, when glacial isostatic adjustment was most rapid and climate was cooler

and drier. As the peat mass developed, I estimate that the HBL potentially released 1 – 7 Tg CH4

y-1 during the late Holocene. My results indicate that the HBL currently stores a globally

significant C pool of ~ 30 Pg C and provide support for a peatland-derived CH4 contribution to

the late Holocene atmosphere.

2.2 Introduction

Peatlands are an important component of the global carbon (C) cycle, as long-term atmospheric

C dioxide (CO2) sinks and methane (CH4) sources. Holocene peat accumulation in northern high

latitudes occurs under cool, humid conditions. The storage of atmospheric C in peatlands, as

partially decomposed vegetation, has resulted in a net long-term climatic cooling. Quantifying

the influence of peatland development in the global C-climate system necessitates refined

17 18 estimates of peatland area extent and development in relation to paleoclimate, peat C accumulation and long-term potential atmospheric CO2 and CH4 dynamics (Gorham, 1991;

Schuldt et al., 2013). A key uncertainty stems from the paucity of detailed peatland age and C records from the world’s second largest continuous peatland, the Hudson Bay Lowlands (HBL),

Canada. HBL peatlands are assumed to be a globally significant C pool, and peatland development in this region is potentially driven by both glacial isostatic adjustment (GIA) and climate. However, these hypotheses remain untested at the regional scale and have never been supported by a larger dataset.

Prior to peatland initiation in the HBL, subsidence in excess of 200 m following the final collapse of the Laurentide ice sheet, resulted in inundation of the region by the post-glacial

Tyrrell Sea (Dyke et al., 2003) until ~ 8.5 ky. Since this time, land has been emerging from the

Hudson and James Bays at GIA rates that are among the fastest globally (Peltier, 2004; Tuittila et al., 2013;Webber et al., 1970). Today, a nearly continuous peat cover over low relief deposits of glacio-marine sediments (Martini, 2006) stretches from the shoreline of Hudson and James

Bays inland to the margin of the upland Precambrian Shield. Based upon available land classification data (Ontario Land Cover Database, 2000), peat accumulating features, dominated by bogs and fens, occupy up to 90% of the HBL’s 372,000-km2 landscape (Riley, 2011; Ontario

Land Cover Database, 2000; Martini, 2006) or ~ 10% of the northern peatland area (Tarnocai et al., 2009; Yu, 2012). Permafrost is estimated to occur in ~ 1% of the HBL, (Ontario Land Cover

Database, 2000; Riley, 2011), principally as continuous permafrost along the Hudson Bay coast

(Ontario Land Cover Database, 2000), and as sporadic to discontinuous permafrost in the southern and central HBL, respectively (Riley, 2011). The HBL’s climate is microthermal, influenced by Arctic air masses, strong winds, and ice cover on the Hudson Bay. Modern

19 average climate (1971 – 2000) is characterized by a mean annual temperature (MAT) of -2.5 ±

1.8 °C and a mean growing season temperature of 10.8 ± 1.0 °C over a period of 119 – 162 days

(McKenney et al., 2006). Total annual precipitation ranges from 430 – 740 mm with half to two- thirds of the precipitation occurring during the growing season (McKenney et al., 2006).

Since deglaciation, recent estimates suggest that circum-Arctic peatlands have accrued ~ 500 Pg

C (Tarnocai et al., 2009; Yu, 2011; Gorham, 1991), resulting in the sequestration of even more atmospheric C dioxide (CO2) than the 100 – 200 ppmv CO2 equivalent reported by Frolking and

Roulet (2007), which was based upon the earlier estimate of the northern peatland C sink of 250

– 450 Pg C. However, northern wetlands and especially minerotrophic peatlands (MacDonald et al., 2006; Yu et al., 2013), are one of the strongest natural sources of CH4, and release 21 – 43 Tg

-1 CH4 y to the atmosphere (Mikaloff Fletcher et al., 2004; Levine et al., 2011). Atmospheric CH4 concentrations are mostly source-driven (Zurcher et al., 2013; Levine et al., 2011), and peatland

C dynamics can account for part of this variation (Tarnocai et al., 2009; Jones and Yu, 2010;

MacDonald et al., 2006).

Polar ice-core records reveal that atmospheric CH4 concentrations rose twice during the

Holocene: first in the early Holocene during a period of continental deglaciation (11.6 – 8 ky

BP); and second, during the mid- to late Holocene (5 ky BP – present), following a brief decline in atmospheric CH4 (Brook et al., 2000). Although subject to large uncertainties, inter- hemispheric CH4 gradients suggest that northern terrestrial wet ecosystems, including peatlands can explain a portion of the variability in atmospheric CH4 during both the early and late

Holocene (Reyes and Cooke, 2011; MacDonald et al., 2006; Yu, 2011). Examination of the dynamic coupling between peatland development and Holocene atmospheric CH4 variability may provide additional insight into long-term peatland-climate-C dynamics.

During the early Holocene, high insolation and seasonality coupled with the availability of newly exposed surfaces following deglaciation, resulted in a tripling of peatland extent and precursor wetland conditions. Thus, post-glacial wet and reducing conditions together with rapid peatland expansion during the early Holocene are associated with rising atmospheric CH4 concentrations

(MacDonald et al., 2006; Yu, 2011). However, global syntheses of available northern peatland initiation dates have shown lower rates of peatland expansion after 9 ky BP, further implying reduced peatland-derived CH4 contributions to the mid- and late Holocene atmospheres

(MacDonald et al., 2006; Yu, 2011). However, given limited detailed evidence from spatially extensive and late emerging peatlands such as the HBL, quantification of the potential contribution of northern peatlands to the late Holocene CH4 rise has not been fully possible (Yu,

2012; Reyes and Cooke, 2011). While northern peatlands could contribute in part to the rise in

CH4 in the late Holocene, potential alternative terrestrial sources may include early anthropogenic activity, such as rice cultivation, biomass burning and deforestation (Ruddiman,

2003), lateral expansion of existing high latitude peatlands, permafrost dynamics (Tarnocai et al.,

2009), and/or contributions from tropical wet ecosystems (Yu, 2011; Zurcher et al., 2013).

Here, I report on the history of peat development in the HBL, using a synthesis of peat ages and detailed peat C records. With these new data, I examine the relationship between climatic and geophysical controls relative to peatland initiation and expansion in the HBL and evaluate the possible connection between HBL peatland development and atmospheric C dynamics, during the Holocene. This new evidence supports the hypothesis of potential peatland-derived CH4 emissions during the late Holocene, and fills a key gap in determining the potential contribution of northern peatlands to late Holocene atmospheric CH4 concentrations. While climate has been invoked as the dominant control on northern peatland initiation (Jones and Yu, 2010; MacDonald et al., 2006; Yu et al., 2013), GIA appears to be an even more fundamental control of peatland

21 initiation and expansion in the dynamic landscape of the HBL (Peltier, 2004; Glaser et al.,

2004a; Glaser et al., 2004b). I estimate that HBL peatlands have accumulated a globally significant peat C pool of ~ 30 Pg C over the Holocene. During the most intense period of peat initiation between 4 – 7 ky cal BP, thin peat profiles over marine sediments may limit the potential for CH4 emission, due to the possible influence of marine pore waters driving sulfate reduction (Liikanen et al., 2009; Koebsch et al., 2013) and the prevailing cold, dry climate inferred from a landscape scale pollen reconstruction (northern Quebec) (Viau and Gajewski,

2009). However, by 3 ky cal BP peatlands with a mean thickness of one meter occupied 80% of the HBL’s current extent, and were developing under climatic conditions similar to today (Viau

-1 and Gajewski, 2009), supporting a potential CH4 emission of 1 – 7 Tg CH4 y . Acknowledging the complexity of atmospheric CH4dynamics, a potential late Holocene CH4 emission of this magnitude from the HBL cannot alone explain the late Holocene rise in global atmospheric CH4 concentrations. Still, my new estimates provide evidence that peatland development in the HBL may constitute a potential CH4 contribution to late Holocene atmosphere that should be accounted for alongside other prospective anthropogenic and natural terrestrial CH4 sources in global C budgets.

2.3 Methods

2.3.1 Data synthesis and new peat records

I compiled 100 records of peat initiation (Table 2.10-1) from peat profiles located within the physiographic region of the HBL (Figure 2.11-1), spatially defined according to the North

American Environmental Atlas (2009). The records were compiled for the purpose of examining the temporal pattern of peatland initiation (Figure 2.11-2) in relation to summer and winter

22 insolation at 60 °N (Berger and Loutre, 1991), variability in ice-core inferred Holocene atmospheric CH4 (Brook et al., 2000) and CO2 (Monnin et al., 2004), post-glacial land emergence (Peltier, 2004), and pollen-reconstructed Holocene temperature and precipitation for the northern Quebec, Canada ecoregion (Viau and Gajewski, 2009). The northern Quebec ecoregion was selected for the inferred paleoclimate of the HBL due to its proximal location east of Hudson Bay, within a similar bio-climatic setting to the HBL, and having a similar paleoclimatic reconstruction to that produced from a pollen record of a lake core retrieved in the northeastern HBL (McAndrews and Campbell, 1993; McAndrews et al., 1982). Land emergence, area expansion, and the 5.5-ky and 3-ky shorelines were extracted from the 0.5-ky resolution paleotopography dataset (Peltier, 2002; Peltier, 2004). Peatland area extent and peat thickness were quantified to estimate the potential CH4 contribution from the HBL to the mid- and late

Holocene atmosphere, using previously reported snow-free CH4 emissions for the HBL (Pickett-

Heaps et al., 2011). The reported range used here was informed by both static chamber and airborne CH4 flux measurements.

A literature search and a radiocarbon database were used to prepare the synthesis of all available basal dated peat records from the HBL (Gorham et al., 2012; Gorham et al., 2007) (Table2.10-1 and original references listed therein), yielding 67 peat records meeting the following three criteria: location (geographic coordinates) within the physiographic region of the HBL; availability of 14C basal peat ages obtained from the peat-mineral interface; and basal peat material sampled from an ecosystem currently classified as a peatland. To supplement the series of 67 records, field sampling across the Ontario portion of the HBL in 2009 – 2011 resulted in my contribution of 33 new peat records. Basal dates for the 33 newly sampled peat cores were obtained using accelerator mass spectroscopy (AMS) 14C dating of plant remains of known

23 provenance and/or root-free bulk peat (n < 8) at either the Keck-CCAMS facility (Irvine, USA), the UGAMS facility (Athens, USA) or Beta Analytic, Inc. (Miami, USA). In the absence of time-weighted (vertically-dated) peat accumulation rates, paleo-peat thickness was estimated using linear accumulation following peat inception (n = 100); however, I acknowledge that the rate of peat accumulation may vary through time in the HBL.

2.3.2 Holocene peat initiation and carbon dynamics

All dates compiled from previous studies, together with the newly dated samples described above, were calibrated using IntCal09 calibration curve (Reimer et al., 2009) in the clam package for R (Blaauw, 2010). All calibrated most probable (median) ages are expressed as calendar years before present (1000 y cal BP = 1 ky cal BP), where BP is equal to AD 1950. While a histogram approach can be problematic when using calibrated radiocarbon dates spanning a probability distribution of > 200 years, I selected a bin size equal to the mean 2σ age uncertainty range of the calibrated radiocarbon dates considered in this study. Each date is represented by a single occurrence in the histogram, and is plotted at the location on the x-axis of the median calibrated initiation age (Gorham et al., 2012). Similar to other studies, I repeated the initiation age binning process using several different numerical approaches and bin sizes (100 – 500 y) and found that the major trends appeared to be insensitive to length of time used (Reyes and Cooke,

2011). Assessment of the lag between peat initiation and land emergence in the HBL was accomplished by spatially relating each peat age in my record with the corresponding time of land emergence. Using the geographic coordinates for each peat record, the corresponding time of land emergence for that location was extracted from the paleotopography dataset (Peltier,

2002; Peltier, 2004).

Quantification of the total C pool was based upon bulk density, loss-on-ignition (LOI) and C content measurements, using standard methods (Yu, 2012; Chambers et al., 2010/11), on 33 peat cores retrieved during the 2009 – 2011 field seasons, together with a re-interpretation of seven author-contributed raw datasets from previously published studies (Bunbury et al., 2012; Kuhry,

2008; Kettles et al., 2000; Kuhry, 1998; O'Reilly et al., 2014). Direct measurements of C content using homogenized subsamples spanning 4 – 10 cm of core depth were completed using an

Elementar Variomax CN analyzer (Elementar Analysensysteme GmbH Donaustraße 7 63452

Hanau Germany). The summed product of incremental bulk density (g cm-3) and peat depth (cm) resulted in the areal dry peat mass (g cm-2) term. Similar to other studies, areal dry peat mass for the peat profiles examined here was strongly related to basal peat age (Zoltai, 1991; Turunen et al., 2002). Here, I expressed areal dry peat mass as areal C mass (kg m-2), as it was similarly correlated with peat age. Linear regression analysis of C mass and basal peat age revealed a significant relationship (F = 41.52; R = 0.75, p < 0.001, n = 36) described by equation 2.3.2.1:

C mass (kg m-2) = 27.4 – [0.013 * basal peat age (y cal BP)]

(2.3.2.1)

Equation 2.3.2.1 was used to estimate areal C mass for the remaining 62 peat cores that lacked reported peat physical properties. The total C pool for the HBL was quantified by multiplying the mean C mass (n = 100) by the total peatland area. C pools for the inland and coastal areas were also calculated separately using mean C masses for these respective regions. The total C pool range was expressed in terms of the 2σ age uncertainty. Finally, the mean post-initiation C accumulation rate for each sample was calculated by dividing the total peat C pool by the age of the sample and expressed as g C m-2 y-1.

2.4 Results

2.4.1 GIA and paleoclimate as controls of HBL peatland dynamics

Today, land is emerging at a rate of 5 – 12 mm y-1 in the HBL (Peltier, 2004), with maximum isostatic uplift rates centered near the north-west and south-east reaches of the landscape (Figure

2.11-1). My synthesis of 33 new and 67 previously reported peat ages (Table 2.10-1) documents the spatio-temporal variability in peat initiation and expansion in the HBL, following retreat of the Laurentide ice sheet. The mean and standard deviation depth of peat is 203 ± 89 cm and the median basal age is 5 ky cal BP for this series of peat records, while the oldest basal date returned was 8.2 ky cal BP recovered from the margin of inundation by the post-glacial Tyrrell

Sea. It provides an oldest minimum peat age for the peatlands of the HBL.

Spatially associating basal peat age inventories with post-glacial land emergence histories reveals that peat initiation in the HBL was tightly coupled with GIA (Figure2.11-2). The theoretical rate of GIA decays exponentially with time, hence the fastest rates of isostatic uplift occur immediately following deglaciation (Peltier, 2004). Likewise, new peatland initiation in the HBL also appears to decline through time, with the greatest rates of peat initiation in the

HBL recorded during the period 4 – 7.5 ky cal BP, referred to here as the mid-Holocene (Figure

2.11-2). A small bias toward sampling older, deeper peatlands may contribute to the apparent reduction in the rate of peat initiation events later in the Holocene. However, I expect that a gridded sampling strategy across the entire HBL would reveal a comparable temporal pattern due to increasingly smaller additions of new land with time, as a result of long-term exponentially declining rates of GIA in the HBL. Thus a potential sampling bias toward deeper peatlands does not undermine my conclusions.

Available paleoclimate data (Viau and Gajewski, 2009; McAndrews et al., 1982) for the entire period of Holocene peatland development in the HBL indicate that climatic conditions were continuously suitable for peatland development (Yu, 2012). However, the coldest winters and summers together with the period receiving the least precipitation occurred between 4 – 7 ky cal

BP, synchronous with both the highest rates of peat initiation and emergence of new land available for paludification. Declining rates of peat initiation up to and during the period of

Neoglacial cooling (~ 2 ky cal BP), were associated with cooler temperatures and more precipitation based upon regional pollen-based paleoclimatic reconstructions (Figure 2.11-2c)

(Viau and Gajewski, 2009).Contemporary climate in the HBL likely varies spatially, due to the seasonal influence of the Hudson Bay. However, few paleoclimatic records are available for the

HBL, restricting examination of regional climate variation on peatland initiation and C dynamics during the Holocene.

2.4.2 HBL carbon storage and potential Holocene methane emissions

The physiographic region of the HBL is composed of two main ecoregions, which differ in terms of spatial extent, peat age and depth, and C storage capacity. The more recently emerged coastal

HBL ecoregion accounts for 62,370 km2, while the inland HBL and James Bay Lowland (JBL) ecoregions together account for 309,350 km2. Coastal HBL peatlands included in this study are occupied by thin peat deposits, with a median peat depth of 58 cm and a mean peat age of 1900 ±

1830 y cal BP (n = 13). By contrast, the inland HBL peatlands studied here have a median peat depth of 218 cm and a mean peat age of 5220 ± 1450 y cal BP (n = 86).

I estimated the total C pool for the HBL peatland complex and for each of the two principal ecoregions, based upon peat age and C content. My results reveal a mean areal C mass of 89 ±

27 kg C m-2 (n = 100) for the HBL that accumulated at a mean rate of 18.5 ± 5.7g C m-2 y-1 and has a mean C density of 0.05 ± 0.01 g cm-3. Multiplication of the mean total peat C mass by the total peatland surface area of 334,800 km2 results in an HBL C pool of approximately 30 Pg C

(29 – 31 Pg C). Permafrost occupies a relatively small fraction of the HBL, and although my record proportionally includes permafrost features, possible variation in C mass from permafrost- impacted peatlands could not be adequately assessed here due to sample limitations. However, I considered variation in areal C mass between the inland and coastal ecoregions. I found that the median C mass was significantly greater (U = 107, p < 0.001) inland (99 kg m-2) than near the coast (40 kg m-2).Taking this difference into consideration yields an area-weighted total C pool of ~ 27.6 Pg C for the inland HBL/JBL ecoregions and ~ 2.2 Pg C for the coastal ecoregion.

-2 -1 Modern seasonal CH4 fluxes range from 1 – 50 g CH4 m y for peatlands in the HBL (Bubier et al., 1993b; Klinger et al., 1994; Moore et al., 1994; Roulet et al., 1994), with scaled-up areal

-2 -1 emissions ranging from 5 – 20 g CH4 m y for the HBL region (Pickett-Heaps et al., 2011;

Christensen et al., 1996), for the HBL extent defined here. Although peatland CH4 fluxes are highly uncertain, I use the latter, more conservative regional CH4 flux range to estimate the potential CH4 contributions from the HBL to the mid- and late Holocene atmosphere. Given a

5.5-ky cal BP HBL landscape, I find that peat could cover 50% of the modern HBL extent, with

-1 a mean peat depth of 33 cm, and could potentially emit 0.7 – 3.7 Tg CH4 y . Climate at this time was colder and drier than today (Figure 2.11-2c) (Viau and Gajewski, 2009), and CH4 emissions may have stemmed from the terrestrial-aquatic interface. However, areas with thin peat in contact with marine sediments, as evidenced by the occurrence of salt marsh vegetation remains near the peat-mineral interface (O'Reilly et al., 2014), may conversely experience reduced potential CH4 flux, due in part to interactions with marine pore and surface waters (Koebsch et

28 al., 2013; Liikanen et al., 2009). By contrast, a 3-ky cal BP HBL peatland landscape could occupy approximately 80% of the modern peatland surface area, with a mean peat thickness of one meter. The accrual of a thick peat mass could limit the influence of the marine sediments upon late Holocene peatland dynamics. Accordingly, a 3-ky cal BP peatland extent, operating under climatic conditions similar today (as indicated by the regional paleoclimate record), could

-1 support a potential CH4 flux of 1.2 – 6.8 Tg CH4 y to the late Holocene atmosphere. Given a

-1 unit conversion factor 2.78 Tg CH4 ppb and an 8.4-y CH4 lifetime (Zurcher et al., 2013; Levine et al., 2011), the potential CH4 emissions from the HBL during the late Holocene could account for an increase of 4 – 21 ppb. I used a landscape scale estimate of snow-free CH4 flux for the geographic extent of the HBL, according to the spatial extent described here. Estimates from static chamber measurements are subject to large seasonal to inter-annual variability, and if considered alone may predict reduced potential CH4 emissions at the local scale (Klinger et al.,

1994; Roulet et al., 1994) or enhanced potential CH4 flux up to six times greater than these regional scale emission rates (Bubier et al., 1993a; Smith et al., 2004).

2.5 Discussion

My synthesis confirms that peat initiation in the HBL began several millennia later than many northern peatland regions, including the extensive peatlands of western Canada and Siberia. Two periods of intensified peat initiation are evident during the mid-Holocene. The first occurred between 5.6 – 7 ky cal BP, as new land emerged from the Tyrrell Sea approximately four times faster than the modern rate of isostatic uplift (Glaser et al., 2004b). Following a ~ 0.6 ky apparent gap in peat initiation, a second phase of intensified peat initiation occurred between 4.2

– 5 ky cal BP (Figure 2.11-2), when isostatic uplift was approximately three times faster than today. While the gap between the two most intense periods of initiation in the HBL may be due

29 to a lack of available peat records in this dataset, other studies have attributed similar peat initiation gaps to the occurrence of large scale drainage events. In the HBL, regional drainage events resulting from outbursts of dammed water features or uplift-mediated changes in stream flow paths have been documented (Sjörs, 1959; Glaser et al., 2004a).

Half of the peatlands examined here initiated synchronously with land emergence from the post- glacial Tyrrell Sea (Figure 2.11-2b). However, comparison of cumulative peat initiation and post-glacial land emergence curves reveal a divergence after 5.8 ky cal BP, suggesting a time lag between emergence of available land and establishment of peat-accumulating features on the parcels of newly emerged land. This apparent temporal lag in peat initiation relative to the emergence of the corresponding parcel of land appears to have continued until present. Syntheses of North American peat initiation records have reported lags between land emergence and peat initiation of up to 4 ky throughout the Holocene, which have been attributed in part to the slow migration of peatland plant propagules to the newly emerging surface (Gorham et al., 2012).

However in the HBL, peat-forming vegetation appears to have rapidly established on the newly emerging land between 5.6 –7 ky cal BP, thus the slow migration of peatland plant propagules does not adequately account for the time lag between land emergence and delayed peat initiation on emerged landscapes evident since 5 ky cal BP (Glaser et al., 2004b; Glaser et al., 2004a).

Temporal delays in peatland initiation may also be attributed in part to local inundation of the emerging land. The structure of peatland vegetation communities and water table position are tightly coupled, and together control vertical peat growth, C dynamics, and the subsequent redistribution of water within the peatland (Beilman et al., 2009; Belyea and Baird, 2006; Sjörs,

1959). The hydrogeologic setting in the HBL is influenced by regional (e.g., 102 km2) variability in GIA, and may account for the time lags in peat initiation in my record (Gorham et al., 2012;

Glaser et al., 2004a). Glaser et al. (2004a) hypothesize that faster rates of isostatic uplift in downstream reaches compared to corresponding headwaters can reduce slope, impede drainage, re-direct river channels, and contribute to anoxic conditions and peatland pattern development in peatlands of the southern HBL (Glaser et al., 2004a; Glaser et al., 2004b). My results suggest that similar processes occur across the HBL, whereby the main environmental factor driving the initiation and development of peatlands is the sustained occurrence of waterlogged soils (Glaser et al., 2004a; Glaser et al., 2004b; Gorham, 1991). In addition, pond development behind beach ridges during progressive shoreline retreat or winter ice scour in emerging high energy coastlines, and forest formation on beach ridge prior to peat initiation, as a consequence of regional isostatic uplift dynamics, are further evidence of the HBL’s dynamic landscape (Riley,

2011; Glaser et al., 2004b; Glaser et al., 2004a; Martini, 2006). Thus, the ecological response to hydrogeologic variability can delay paludification of available land and/or terrestrialization of lakes and ponds.

Previous global peatland syntheses indicate that the period of most intense northern peatland initiation has been driven by optimal climatic conditions, including high insolation and seasonality, during the early Holocene (Jones and Yu, 2010; MacDonald et al., 2006; Yu et al.,

2010). By contrast, comparisons with available paleoclimate reconstructions reveal an opposite trend in the HBL, such that the most intense period of peatland initiation in the HBL occurred under less favourable conditions. A regional (northern Quebec, Canada) pollen-derived temperature and precipitation reconstruction (Viau and Gajewski, 2009) indicates small climatic anomalies relative to modern climate in the HBL during the Holocene, with the warmest period occurring prior to land emergence from the Tyrrell Sea. Consequently, all peat initiation and expansion in the HBL progressed under conditions of decreasing summer insolation, increasing

31 winter insolation (Figure 2.11-2), and declining seasonality as inferred from summer-winter insolation differences. In contrast to the warm climate and high summer insolation that supported rapid peat initiation in other northern peatlands during the early Holocene (Yu, 2012;

MacDonald et al., 2006), maximum rates of peat initiation in the HBL occurred when the climate was coldest and driest (Figure 2.11-2). While it is possible that peat initiation and expansion may have occurred more rapidly in the HBL under optimal climatic conditions, as it did elsewhere in the early Holocene (MacDonald et al., 2006), the tight coupling between land emergence and peat initiation under cooler/drier mid-Holocene climatic conditions does not appear to have substantively limited the timing or rate of peat initiation. Since 3.5 ky cal BP, peat initiation in the HBL remained relatively low, consistent with declining rates of Neoglacial peat initiation for northern peatlands (Jones and Yu, 2010; MacDonald et al., 2006). However, declining rates of peat initiation in the HBL occurred under comparatively warmer and wetter conditions relative to the mid-Holocene HBL climate (Viau and Gajewski, 2009), and appear to lag land emergence.

Examination of North American pollen records of peatland indicators reveals a climatic optimum for peatland extent characterized by a MAT between -2 and 6 °C and mean total annual precipitation (MAP) between 630 and 1300 mm (Gajewski et al., 2001). Further evidence from the West Siberian Lowlands indicates that maximum peatland extent and C pools occur in association with a MAT of 0 °C, which may be a potential climatic optimum for maximizing the difference between net primary production and decomposition (Beilman et al., 2009; Sheng et al., 2004; Swanson et al., 2000). Modern climate across most of the HBL falls within these boundary conditions. Thus, in spite of some documented paleoclimatic variability in the HBL

(McAndrews et al., 1982; Bunbury et al., 2012; O'Reilly et al., 2014), peatland development has proceeded within climatic boundary conditions for northern peat initiation and expansion for

32 much of the peat record. Further, on-going rapid land emergence, which results in a lowering of the regional hydraulic gradient, reduced drainage and sustained waterlogged soil conditions, may mask the impact of climatic variation on peatland dynamics in the HBL. Hence, the weaker relationship between climate and peat initiation and expansion in the HBL may be accounted for by a limited ability to consider paleoclimatic variation spatially, and a lack of evidence in support of wide temporal variation in climate at the landscape scale in the HBL.

Anticipated high latitude warming may increase peatland primary productivity and C storage; however, enhanced hydroclimatic variability and intensified land use may augment peatland C decomposition and CO2/CH4 emissions, and result in a positive feedback to climatic warming.

Future climate scenarios for the HBL indicate a 2 – 5 °C temperature increase over the next century, with a concomitant increase in precipitation of 10 – 15% (McKenney et al., 2011). The implications of these projections may be that warmer, wetter conditions and a longer growing season in the HBL may support continued peatland development and sustained C sequestration via enhanced primary productivity (Charman et al., 2013). However, results from a multi-year investigation concerning water balance thresholds in the northern HBL provide confounding evidence, whereby the predicted precipitation increases over the next century in the HBL may not be sufficient to offset the growing season water deficit potential brought about by increasing temperatures and enhanced evaporation (Rouse, 1998). Therefore, disproportionate future temperature increases relative to concomitant precipitation increases may result in reduced peatland C storage capacity in the HBL, with C losses potentially exacerbated by increased fire risk associated with net drier conditions (Yu, 2012) or enhanced anthropogenic activity, such as infrastructure development (Kuhry, 2008). Hence, further examination of the spatio-temporal

33 relationship between paleoclimate and C dynamics is warranted, and may reveal a dynamic coupling between the two processes, as it has elsewhere (Charman et al., 2013).

Based upon new detailed records of peat age and areal C mass, I estimate that the HBL currently stores ~ 30 Pg C, equivalent to approximately 20% of the North American peatland C pool

(Gorham et al., 2012) or approximately 6% of the northern peatland C pool (Yu, 2012). While my synthesis includes C data for a range of forested and non-forested bog and fen features, I recognize that the C mass contribution from small bog/fen pools, permafrost features, and densely forested peatlands may be under- or over represented in my C pool estimate, and this requires additional investigation. Although refined estimates of land cover and C mass are needed in the HBL, my evidence provides the most comprehensive synoptic assessment of the modern HBL C pool and demonstrates that the HBL represents a globally significant C sink, underscoring its importance in regional and global C assessments. Still, three decades of peatland

CH4 emission measurements have shown that modern HBL peatlands can be persistent sources of CH4 to the atmosphere (Klinger et al., 1994; Moore et al., 1994; Pickett-Heaps et al., 2011;

Roulet et al., 1994), at a range of spatio-temporal scales.

Peatlands release CH4 to the atmosphere by ebullition, diffusion, and plant- mediated transport, however, potential flux estimates are subject to both spatial and temporal uncertainty, due in part to probable non-linear feedbacks among potential controls over peatland CH4 flux (Belyea and

-2 -1 Baird, 2006; Yu, 2011). Complex patterns and large ranges (at least 1 – 50 g CH4 m y ) in contemporary CH4 emissions in the HBL have been variously explained by a combination of moderate water table position, gradients in growing seasonal temperatures, and heterogeneities in water chemistry and vegetative cover (Bubier et al., 1993b; Klinger et al., 1994; Moore et al.,

1994; Roulet et al., 1994). Consequently, important sources of CH4 emissions in the HBL have

34 been reported from wet minerotrophic peatlands and those undergoing ombrotrophication, wetlands at the aquatic-terrestrial interface, and young coastal fens (Bubier, 1995; Bubier et al.,

1993b; Klinger et al., 1994; Moore et al., 1994; Whiting, 1994). However, recent evidence suggests that CH4 fluxes from shallow coastal systems may be suppressed by sulphur-rich marine sediments in brackish environments, depending in part upon the vegetation community present

(Koebsch et al., 2013; Liikanen et al., 2009). Thus, the potential for sulphur-mediated CH4 suppression is recognized here in relation to young or early emerging peatlands. Similarly, landscape-scale assessments using aircraft and surface measurements, together with a chemical transport model indicate that the inland HBL may release peatland-derived CH4 at twice the rate of the HBL’s coastal regions (Pickett-Heaps et al., 2011; Christensen et al., 1996). Accordingly, modern CH4 emissions from the coastal region of the HBL reportedly range between 5 and 10 g

-2 -1 CH4 m y , while the CH4 emission from the inland peatlands of the HBL range between 15 and

-2 -1 20 g CH4 m y . Owing to the unique geographic setting of this region in a zone of rapid isostatic uplift and the fact that emerging peatlands pass through a salt marsh or marine- influenced phase, this modern evidence supports my hypothesis that maximum CH4 flux from

HBL peatlands is related to the development of the peat mass rather than simply peat initiation

(Glaser et al., 2004a; O'Reilly et al., 2014).

Although estimates of CH4 flux from peatlands to the atmosphere can be highly dynamic, the possible connection between HBL peatland development and potential CH4 emissions can be evaluated using comparisons between HBL initiation histories and inferred potential CH4 fluxes

(Figure 2.11-2). Peatland development, rather than simply initiation, has been proposed to explain the relationship between CH4 flux and peat development in the early Holocene (Yu et al.,

2013; MacDonald et al., 2006; Smith et al., 2004; Jones and Yu, 2010). In the HBL, I find that

35 during the most intense period of peat initiation, ~ 5.5 ky cal BP, the landscape could potentially

-1 support CH4 emissions of 0.7 – 3.7 Tg CH4 y from minerotrophic peatlands and pre-cursor marsh conditions, the occurrence of which is evidenced by reconstruction of past environments following examination of testate amoebae, plant macrofossils and pollen (Bunbury et al., 2012;

O'Reilly et al., 2014; Glaser et al., 2004a; Glaser et al., 2004b). However, colder, drier conditions during the mid-Holocene together with a possible biogeochemical connection between shallow peat and the underlying marine sediments likely limited the CH4 flux to the lower end of the range.

Temporal lags in peat initiation relative to land availability observed after 5 ky cal BP, coupled with expansion and patterning of peatlands with C rich pools could further contribute to enhanced peat-derived CH4 from the HBL to the late Holocene atmosphere, owing to decomposition of both new and old peat (Korhola et al., 2010; Pelletier et al., 2007; Yu, 2011).

Therefore, following several millennia of peatland expansion, a 3-ky cal BP HBL landscape

-1 could have potentially contributed 1.2 – 6.8 Tg CH4 y to the late Holocene atmosphere, from the widespread minerotrophic to ombrotrophic peatlands (Glaser et al., 2004a; Klinger et al.,

1994; Riley, 2011) that were functioning under climatic condition similar to today. My late

Holocene estimates of CH4 emissions are comparable to those reported using an Earth system modeling approach (Kleinen et al., 2012; Schuldt et al., 2013; Zurcher et al., 2013). Using a

-1 modeling approach, the HBL was predicted to release ~ 5 Tg CH4 y , equivalent to ~ 10% of boreal CH4 emissions (Pickett-Heaps et al., 2011; Schuldt et al., 2013). Further, modeled CH4 emissions for the broader Hudson Bay region, which includes the HBL as defined here and the adjacent upland region on the Precambrian Shield, suggest a mean snow-free emission of 1.6 Tg

-1 CH4 y for the period 6 ky cal BP to pre-industrial levels (Kleinen et al., 2012; Schuldt et al.,

2013; Zurcher et al., 2013). Additionally, model evidence used to examine the mid-Holocene decline in CH4 emissions associated with the 8.2 ky cal BP event, which further confirmed a slight increase in CH4 emissions from the newly emerging HBL peatlands, at a time when reduced CH4 emissions from northern peatlands were predicted in response to prevailing cooler/drier climatic conditions (Zurcher et al., 2013). Considered alone, the potential CH4 fluxes from the HBL are small on a global scale. While acknowledging the contributions from other biospheric CH4 sources to late Holocene atmospheric CH4 dynamics, including tropical wet ecosystems and early anthropogenic activity, my estimates of potential CH4 flux do convey the role of the HBL in CH4 accounting. I suggest that a contribution from this large region, together with global peatlands, to atmospheric CH4 dynamics cannot be ignored.

2.6 Conclusion

In conclusion, my findings confirm that the most intense period of peat initiation in the HBL occurred during the mid-Holocene, when the climate was colder and received less precipitation than today (Figure 2.11-2c) (Viau and Gajewski, 2009) and insolation was decreasing relative to the early Holocene (Figure 2.11-2a) (Berger and Loutre, 1991). While climatic conditions in the

HBL remained suitable for peat initiation through the mid- to late Holocene, evidence from my study reveals a tight coupling between rapid land emergence from the Tyrrell Sea and peat initiation, as a consequence of GIA. The rate of isostatic uplift varies spatially across the HBL resulting in dynamic hydrogeologic conditions, and suggests that the mid-Holocene delay in peat initiation relative to apparent land emergence may be attributed to localized inundation of the land surface. Based upon new detailed records of peat C density, I estimate that the HBL currently stores ~ 30 Pg C, which is equivalent to ~ 20% of the North American (Gorham et al.,

2012) or ~ 6 % of the northern peatland C pool (Yu, 2012). The highest rate of peat initiation in

the HBL occurred prior to the late-Holocene rise in atmospheric CH4 inferred from ice-cores.

Using modern regional estimates of snow-free CH4 flux from the HBL, scaled to a mid- and late

Holocene HBL landscape, I further estimate a potential late Holocene CH4 flux of 1 – 7 Tg CH4

-1 y . This value represents a small but important potential CH4 emission, when compared to the 21

-1 – 43 Tg CH4 yr released by modern northern wetlands (Mikaloff Fletcher et al., 2004).

Increased drought frequency and severity are anticipated at many high latitude peatlands

(Gorham, 1991; Jones and Yu, 2010). However, future climate scenarios predict that the HBL may experience a warmer and wetter climate over the next century (McKenney et al., 2011), permitting the region to continue to function as a long-term C sink. Nevertheless, disproportionate hydroclimatic change and intensified anthropogenic pressure could decouple the peatland water balance in the HBL, reduce peatland C storage capacity, and result in a positive feedback to climatic warming. Improved quantification of modern peatland C stores and regional controls on Holocene peatland C dynamics in the HBL will reduce the uncertainty surrounding the sensitivity of northern peatland C pools under future climate scenarios.

2.7 Acknowledgements

Research funding and field support was provided by the Ontario Ministry of Natural Resources’

Applied Research and Development Branch and Far North Branch, under the auspices of projects

CC-021 and FNIKM 028 to JWM. Additional support for field work and radiocarbon dating was provided by grants (327197-11 and 331284-11) from the Natural Sciences and Engineering

Research Council of Canada (NSERC) and the Ontario Ministry of the Environment through the

Climate Change and Multiple Stressor Research Program at Laurentian University to SAF.

Graduate stipend support to MSP was provided in the form of an NSERC Alexander Graham

Bell Canada Postgraduate Scholarship (CGSD2-426611-2012), an Ontario Graduate Scholarship,

38 and an Association of Canadian Universities for Northern Studies, Canadian Northern Studies

Trust Scholarship. Field research grants from the Society of Wetland Scientists and Aboriginal

Affairs and Northern Development Canada’s Northern Scientific Training Program were also provided to MSP. We thank A. Dyke for providing access to the Canadian basal radiocarbon database and P. Kuhry for contributing raw peat core data. We also thank three anonymous reviewers for their thoughtful comments to previous draft manuscripts.

2.8 Author contributions

M.S.P. designed and performed research, analyzed data, and wrote the paper; S.A.F and J.W.M. designed and performed research. All authors commented on the manuscript at all stages.

2.9 Competing financial interests

The authors declare no competing financial interests.

2.10 Tables

Table 2.10-1 Location, age, and carbon (C)-mass of peat cores located in the Hudson Bay Lowlands, Canada (HBL), sorted by latitude. All conventional 14C dates were re-calibrated with the IntCal09 calibration curve (Reimer et al., 2009). New HBL basal peat dates (n=33) contributed by the authors, appear in bold. Note: *Basal peat depth not identified in reference. **C content determined via loss-on- ignition, at a level of 48% C. ***Areal C mass determined by direct measurement of complete peat profiles (n=38), including samples (N) 10, 12, 40-44, 52-62, 64-66, 68-71, 83-86, 89-92, 101, 102, 106, 128. If peat physical data were not available (n=62), total C mass was estimated using a regression model (see methods).

Basal Radiometric Conventional 14C Calibrated Age Carbon Lat Long Elevation Material 2σ Age N Site Name Depth Lab Age ± error (y (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Range (cm) Number BP) BP) m-2)*** basal 3010- 3 Squirrel R 50.07 -83.88 173 278 GSC-6321 woody 3010 ± 60 3210 68 (1) 3360 peat wood base 4420- 7 Harricana R 50.70 -79.33 59 276 Qu-498 4200 ± 120 4720 88 (2) of peat 5210 4320- 8 Birthday R 50.70 -79.33 59 218 Y-1164 basal peat 4110 ± 100 4640 87 (3) 4860 3980- 9 Birthday R 50.72 -79.33 50 200 GSC-1493 basal peat 3920 ± 130 4350 83 (4) 4810 UGAMS- wood 6760- Current 10 KR3A** 50.74 -84.60 123 271 6020 ± 30 6860 126 12713 fragments 6950 Study UGAMS- Sphagnum 2870- Current 12 PL101 50.94 -80.33 45 264 2840 ± 30 2950 109 12711 remains 3060 Study 6570- 13 Old Man bog 51.03 -84.53 97 420 XXX-06b basal peat 5980 ± 100 6830 115 (5) 7160 6500- 14 Oldman bog-8502 51.07 -84.50 99 445 Beta-42381 basal peat 5920 ± 90 6750 114 (6) 6960 6030- 15 Jaab L site 51.15 -83.05 108 353 WAT-2571 basal peat 5560 ± 110 6360 109 (7) 6640 Sesi M. SWF- 5760- 17 51.23 -83.03 102 149 Beta-54594 basal peat 5200 ± 60 5970 104 (6) 8506A 6180

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Age ± error (y BP) Range (cm) Number BP) m-2)*** 6300- 30 Albany Forks 51.42 -84.80 162 450 GSC-885 basal peat 5820 ± 150 6640 112 (10) 6980 800- 31 Coastal fen site 51.47 -80.62 9 81 WAT-2541 basal peat 1090 ± 70 1010 40 (7) 1180 4440- 32 Carling L site 51.50 -81.67 68 258 WAT-2543 basal peat 4110 ± 70 4640 87 (7) 4830 1730- 33 Interior fen site 51.51 -81.88 72 125 WAT-2542 basal peat 1960 ± 70 1910 52 (7) 2110 2370- 34 Kinosheo L, 300 51.55 -81.83 68 98 WAT-2549 basal peat 2580 ± 80 2630 61 (7) 2850 2790- 35 Kinosheo L, 100 51.55 -81.83 68 98 WAT-2554 basal peat 2870 ± 80 3010 66 (7) 3240 3490- 36 Kinosheo L, 200 51.55 -81.83 68 138 WAT-2556 basal peat 3450 ± 70 3720 75 (7) 3900 4100- 37 Kinosheo L, 300 51.55 -81.83 68 218 WAT-2557 basal peat 3910 ± 70 4340 83 (7) 4520 4440- 38 Kinosheo L, 390 51.55 -81.83 68 265 WAT-2572 basal peat 4110 ± 80 4640 87 (7) 4830 3220- 39 Kinosheo L, 490 51.55 -81.83 68 253 WAT-2567 basal peat 3160 ± 70 3390 71 (7) 3560 UGAMS- bryophyte 4520- Current 40 KJ101 51.55 -81.82 68 297 4100 ± 25 4610 130 12719 remains 4810 Study Kinosheo L wood near 4240- 41 51.55 -81.81 68 265 TO-4318 4000 ± 80 4480 85 (11) Bog** basal 4810

Table 2.10-1, continued

Basal Calibrated Age Carbon Lat Long Elevation Radiometric Material Conventional 14C 2σ Age N Site Name Depth (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Lab Number Dated Age ± error (y BP) Range (cm) BP) m-2)*** 6300- 30 Albany Forks 51.42 -84.80 162 450 GSC-885 basal peat 5820 ± 150 6640 112 (10) 6980 800- 31 Coastal fen site 51.47 -80.62 9 81 WAT-2541 basal peat 1090 ± 70 1010 40 (7) 1180 4440- 32 Carling L site 51.50 -81.67 68 258 WAT-2543 basal peat 4110 ± 70 4640 87 (7) 4830 1730- 33 Interior fen site 51.51 -81.88 72 125 WAT-2542 basal peat 1960 ± 70 1910 52 (7) 2110 2370- 34 Kinosheo L, 300 51.55 -81.83 68 98 WAT-2549 basal peat 2580 ± 80 2630 61 (7) 2850 2790- 35 Kinosheo L, 100 51.55 -81.83 68 98 WAT-2554 basal peat 2870 ± 80 3010 66 (7) 3240 3490- 36 Kinosheo L, 200 51.55 -81.83 68 138 WAT-2556 basal peat 3450 ± 70 3720 75 (7) 3900 4100- 37 Kinosheo L, 300 51.55 -81.83 68 218 WAT-2557 basal peat 3910 ± 70 4340 83 (7) 4520 4440- 38 Kinosheo L, 390 51.55 -81.83 68 265 WAT-2572 basal peat 4110 ± 80 4640 87 (7) 4830 3220- 39 Kinosheo L, 490 51.55 -81.83 68 253 WAT-2567 basal peat 3160 ± 70 3390 71 (7) 3560 UGAMS- bryophyte 4520- Current 40 KJ101 51.55 -81.82 68 297 4100 ± 25 4610 130 12719 remains 4810 Study Kinosheo L wood near 4240- 41 51.55 -81.81 68 265 TO-4318 4000 ± 80 4480 85 (11) Bog** basal 4810

Table 2.10-1, continued

Basal Radiometric Conventional 14C Calibrated Age Carbon Lat Long Elevation Material 2σ Age N Site Name Depth Lab Age ± error (y (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Range (cm) Number BP) BP) m-2)*** UGAMS- wood 3870- Current 42 KJ4-3 51.59 -81.78 66 131 3630 ± 25 3940 71 12717 fragments 4070 Study wood + UCIAMS- 4620- Current 43 KJ3-3 51.59 -81.79 65 176 herb 4170 ± 20 4720 80 97825 4830 Study stems wood + UCIAMS- 4840- Current 44 KJ2-3 51.59 -81.76 65 246 herb 4130 ± 25 4860 99 97824 4960 Study stems Sesi M SFen- 5950- 45 51.62 -82.28 69 166 Beta-53064 basal peat 5370 ± 80 6150 106 (6) 8505A 6300 BelecL bog- 4180- 46 51.62 -82.28 69 236 Beta-54598 basal peat 3960 ± 60 4420 84 (6) 8507A 4780 Belec L Int 4240- 47 51.62 -82.28 69 98 Beta-66733 basal peat 4010 ± 80 4500 85 (6) 9210A 4810 Belec L-fen- 4000- 48 51.62 -82.28 69 109 Beta-66735 basal peat 3840 ± 70 4250 82 (6) 9210B 4430 Wabassie bog- 4870- 49 51.72 -83.63 75 209 Beta-64925 basal peat 4500 ± 90 5140 93 (6) 9214 5450 3930- 50 Blackbear-8508A 51.72 -83.23 62 223 Beta-45905 basal peat 3730 ± 50 4080 80 (6) 4240 1090- 51 Albany R 52.25 -81.58 3 96 WAT-1504 basal peat 1350 ± 70 1270 44 (1) 1390 5920- Current 52 D001 52.67 -84.02 92 231 Beta-310877 bulk peat 5240 ± 30 5980 102 6180 Study 5750- Current 53 D002 52.71 -84.00 91 219 Beta-310878 bulk peat 5080 ± 30 5810 83 5910 Study

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Age ± error (y BP) Range (cm) Number BP) m-2)*** 5750- Current 54 D003 52.76 -84.14 92 223 Beta-310879 bulk peat 5100 ± 30 5810 80 5920 Study 6190- Current 55 D200 52.62 -84.12 118 240 Beta-310880 bulk peat 5420 ± 30 6240 84 6290 Study 4870- Current 56 D201 52.72 -84.25 104 201 Beta-310881 bulk peat 4400 ± 30 4960 106 5210 Study 5740- Current 57 D202 52.71 -84.32 105 215 Beta-310882 bulk peat 5060 ± 30 5820 99 5900 Study 2950- Current 58 D204 52.54 -84.58 143 129 Beta-310883 bulk peat 2900 ± 30 3740 64 3160 Study 7160- Current 59 D206 52.65 -84.58 142 257 Beta-310884 bulk peat 6300 ± 30 7220 105 7290 Study UGAMS- wood 6290- Current 60 VM4-5 52.70 -84.18 103 286 5530 ± 25 6320 108 11673 fragment 6400 Study conifer 6570- 61 VM4-3 52.71 -84.18 103 304 Beta-281004 5890 ± 40 6710 122 (12) fragments 6580 UGAMS- needles; 6300- Current 62 VM4-1 52.71 -84.19 105 311 5550 ± 25 6340 126 12718 wood 6400 Study UCIAMS- wood 6480- Current 63 MS15 R2 52.71 -84.17 102 254 5745 ± 20 5640 100 94010 fragments 6630 Study conifer 6320- 64 VM3-3 52.71 -84.17 102 242 Beta-281777 5640 ± 40 6420 104 (13) needles 6490 UGAMS- wood 6310- Current 65 VM3-2 52.71 -84.17 102 262 5620 ± 25 6400 127 12716 fragments 6450 Study

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Age ± error (y BP) Range (cm) Number BP) m-2)*** UGAMS- wood 6480- Current 66 VM3-5 52.71 -84.17 102 232 5750 ± 25 6550 139 11672 fragments 6640 Study UCIAMS- 5770- Current 67 TS01 52.72 -83.94 86 218 Seeds 5140 ± 20 5910 103 94011 5940 Study UGAMS- herbaceous 4870- Current 68 VM1-3 52.72 -83.94 86 210 4390 ± 20 4940 98 12715 remains 5040 Study UGAMS- Sphagnum 4980- Current 69 VM2-1 52.72 -83.94 86 211 4470 ± 25 5180 85 12710 remains 5290 Study UGAMS- 4450- Current 70 VM2-3 52.72 -83.94 86 207 Wood 4100 ± 30 4610 90 11426 4810 Study UGAMS- wood 5640- Current 71 VM2-5 52.72 -83.94 86 180 4980 ± 25 5700 84 11674 fragments 5850 Study 5590- 72 IB wetland 52.73 -77.77 88 188 Beta-254946 basal peat 4920 ± 40 5650 100 (14) 5730 5600- 73 OFL wetland 52.78 -77.51 102 207 Beta-251799 basal peat 4970 ± 40 5700 100 (14) 5880 Picea needles, 74 BF 52.73 -78.50 50 72 Beta-251816 110 ± 40 120 (-4)-270 29 (14) conifer bark 6280- 78 Attawapiskat 53.12 -85.42 117 198 GSC-31 basal peat 5670 ± 110 6470 110 (15) 6730 5490- 79 Attawapiskat R 53.13 -85.30 108 128 GrN-1925 basal peat 4940 ± 80 5690 100 (16) 5900 UGAMS- Sphagnum 8040- Current 80 ML201 54.24 -87.74 143 235 7350 ± 30 8160 108 12712 remains 8290 Study

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C Age 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated ± error (y BP) Range (cm) Number BP) m-2)*** 6000- 81 Hawley L 54.57 -84.67 137 292 GSC(24)7 basal peat 5580 ± 150 6380 109 (17) 6710 1700- 82 Palsa bog 54.57 -84.63 113 89 BGS-6 basal peat 1897 ± 63 1840 51 (18) 1990 UGAMS- Sphagnum 4420- Current 83 HL02 54.61 -84.60 85 230 4020 ± 25 4480 92 11422 remains 4570 Study UGAMS- Sphagnum 3290- Current 84 HL03 54.68 -84.60 99 269 2700 ± 25 3550 122 11423 remains 3880 Study UGAMS- wood 5610- Current 85 HL04A(West) 54.75 -84.53 89 93 4960 ± 30 5685 73 11424 fragments 5740 Study UGAMS- 4870- Current 86 HL04B(East) 54.75 -84.50 97 105 wood 4400 ± 25 4960 73 11251 5040 Study C Henrietta 800- 87 55.00 -82.33 2 122 GSC-231 basal peat 1210 ± 130 1130 42 (17) Maria 1360 Shagamu R 4240- 88 55.05 -86.75 126 -- BGS-702 basal peat 4020 ± 100 4510 85 (19) 027a* 4830 UGAMS- Sphagnum Current 89 PB3-5 55.23 -84.33 10 44 610 ± 20 600 550-650 22 11421 stems Study UGAMS- wood 1190- Current 90 PB-SF-5 55.23 -84.30 11 24 1340 ± 20 1280 22 11670 fragments 1300 Study UGAMS- Sphagnum Current 91 PB2-5 55.23 -84.33 10 45 F14C=1.01 ± 0.003 Modern -- 23 11420 stems Study UGAMS- Sphagnum Current 92 PB1-5 55.24 -84.32 9 42 610 ± 25 600 550-650 21 12714 stems Study

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Age ± error (y BP) Range (cm) Number BP) m-2)*** Shagamu R basal 6020- 93 55.35 -86.70 105 -- BGS-708 5600 ± 140 6400 109 (19) 208a* peat 6730 Shagamu R basal 2150- 94 55.60 -86.65 32 -- BGS-822 2350 ± 100 2420 58 (19) 62Ca* peat 2710 Shagamu R basal 1820- 95 55.65 -86.60 23 -- BGS-709 2060 ± 100 2040 54 (19) 62Ba* peat 2310 Shagamu R basal 96 55.70 -86.51 4 -- BGS-820 800 ± 100 740 560-930 37 (19) 62Aa* peat basal 6990- 97 Charlebois 56.67 -94.08 117 240 GSC-2760 6280 ± 80 7200 120 (20) peat 7420 basal 6940- 99 ThibaudeauStn 57.08 -94.16 122 155 GSC-5213 6240 ± 80 7150 119 (18) peat 7410 basal 3170- 100 Silcox 57.17 -94.24 142 73 GSC-5245 3120 ± 60 3340 70 (21) peat 3470 AECV- basal 6560- 101 Herchmer Bog** 57.38 -94.20 106 167 5970 ± 90 6810 148 (22) 1715C peat 7150 6210- 102 Herchmer Fen** 57.38 -94.20 106 203 AEVC-1714 bulk peat 5580 ± 80 6373 105 (23) 6550 Lost basal 4580- 103 57.57 -94.32 113 130 GSC-5221 4270 ± 70 4840 89 (21) Moose/O'Day peat 5040 McClintock AECV- 6410- 106 57.80 -94.21 79 163 wood 5810 ± 90 6610 143 (22) Bog** 1718C 6840 basal 3480- 108 Fletcher L 58.17 -93.83 49 324 GSC-3988 3400 ± 60 3650 74 (24) peat 3830

Table 2.10-1, continued

Basal Radiometric Calibrated Age Carbon Lat Long Elevation Material Conventional 14C 2σ Age N Site Name Depth Lab (median, y cal Mass (kg Footnotes (dd) (dd) (m asl) Dated Age ± error (y BP) Range (cm) Number BP) m-2)*** Shagamu R basal 5600- 123 55.05 -86.75 126 -- BGS-800 5050 ± 100 5800 102 (19) 027b* peat 5990 Shagamu R basal 5760- 124 55.35 -86.70 105 -- BGS-801 5270 ± 100 6060 105 (19) 208b* peat 6290 Shagamu R basal 1950- 125 55.60 -86.65 32 -- BGS-710 2250 ± 110 2250 56 (19) 62Cb* peat 2700 Shagamu R basal 1410- 126 55.65 -86.60 23 -- BGS-821 1725 ± 100 1640 48 (19) 62Bb* peat 1870 Shagamu R basal 127 55.70 -86.51 4 -- BGS-701 700 ± 100 660 520-900 36 (19) 62Ab* peat AECV- basal 4290- 128 McLintock Fen** 57.80 -94.20 81 148 4060 ± 100 4570 68 (23) 1717C peat 4840

Footnotes 1 Gorham, E., Lehman, C., Dyke, A., Clymo, D. & Janssens, J. Long-term carbon sequestration in North American peatlands. Quat. Sci. Rev.58, 77-82, doi:10.1016/j.quascirev.2012.09.018 (2012) 2 Dionne, J. C. Radiocarbon dates on peat and tree remains from the James Bay area, subarctic Quebec. Can. J. Forest Res.9, 125-129 (1979). 3 Stuiver, M., Deevey, E. S. & Rouse, I. Yale naturalradiocarbonmeasurements VIII. Radiocarbon 5, 312-341 (1963). 4 Skinner, R. Quaternary stratigraphy of the Moose River basin, Ontario. (Geologic Survey of Canada, Bulletin 225, 1973). 5 Hansen, B. C. Conifer stomate analysis as a paleoecological tool: An example from the Hudson Bay Lowlands. Can. J. Bot. 73, 244-252 (1995). 6 Glaser, P. H., Siegel, D. I., Reeve, A. S., Janssens, J. A. &Janecky, D. R. Tectonic drivers for vegetation patterning and landscape evolution in the Albany River region of the Hudson Bay Lowlands. J. Ecol.92, 1054-U1052, doi:10.1111/j.0022-0477.2004.00930.x (2004). 7 Klinger, L. F., Zimmerman, P. R., Greenberg, J. P., Heidt, L. E. & Guenther, A. B. Carbon trace gas fluxes along a successional gradient in the Hudson Bay Lowland. J. Geophys. Res.-Atmos.99, 1469-1494, doi:10.1029/93jd00312 (1994). 8 McNeely, R. Geological Survey of Canada Radiocarbon Dates XXXIII. (Geological Survey of Canada, Current Research 2001, 2002) . 9 Dionne, J. C. Formes et phenomenesperiglaciaires en Jamesie, Quebec subarctique. Geogr. Phys. Quat. 32,187 (1978). 10 Craig, B. G. Late-glacial and postglacial history of the Hudson Bay region. (Earth Science Symposium on Hudson Bay, Geological Survey of Canada, Paper 68- 53, 1969). 11 Kettles, I. M., Garneau, M. & Jette, H. Macrofossil, pollen, and geochemical records of peatlands in the Kinosheo Lake and Detour Lake areas, northern Ontario. (Geological Survey of Canada, Bulletin 545, 2000). 12 Bunbury, J., Finkelstein, S. A. & Bollmann, J. Holocene hydro-climatic change and effects on carbon accumulation inferred from a peat bog in the Attawapiskat River watershed, Hudson Bay Lowlands, Canada. Quat. Res.78, 275-284, doi:10.1016/j.yqres.2012.05.013 (2012).

13 O'Reilly, B. C., Finkelstein, S. & Bunbury, J. Pollen-derived paleovegetation reconstruction and long-term carbon accumulation at a fen site in the Attawapiskat River watershed, Hudson Bay Lowlands, Canada. Arctic, Antarctic and Alpine Research (2014). 14 Pendea, I.F., Costopoulos, A., Nielsen, C., & Chmura, G.L. A new shoreline displacement model for the last 7 ka from eastern James Bay, Canada. Quaternary Res. 73, 474-484 (2010). 15 Dyck, W. & Fyles, J. G. Geological Survey of Canada radiocarbon dates I and II (Geological Survey of Canada, Paper 63-21, 1963). 16 Terasmae, J. & Hughes, O.L. Glacial retreat in the North Bay Area, Ontario. Science 131, 1444-1446 (1960). 17 Blake, W., Dyck, W. & Fyles, J. G. Geological Survey of Canada Radiocarbon Dates IV. (Geological Survey of Canada, Paper 65-4, 1965). 18 Railton, J. B. & Sparling, I. H. Preliminary studies on the ecology of palsa mounds in northern Ontario. Can. J. Bot. 51, 1037-1044 (1973). 19 Protz, R., Ross, G.J., Martini, I. P. & Terasmae, J. Rate of podzolic soil formation near Hudson Bay, Ontario. Can. J. Soil Sci. 64, 31-49 (1984). 20 Klassen, R. W. Surficial geology of north-central Manitoba. (Geological Survey of Canada, Memoir 419, 1986). 21 Dredge, L. A. & Mott, R. J. Holocene pollen records and peatland development, northeastern Manitoba. Geogr. Phys. Quat.7, 7-19 (2003) . 22 Kuhry, P. Palsa and peat plateau development in the Hudson Bay Lowlands, Canada: timing, pathways and causes. Boreas 37, 316-327, doi:10.1111/j.1502- 3885.2007.00022.x (2008). 23 Kuhry, P. Late Holocene permafrost dynamics in two subarctic peatlands of the Hudson Bay Lowlands (Manitoba, Canada). Eurasian Soil Sci.31, 529-534 (1998). 24 Blake, W. Geological Survey of Canada radiocarbon dates XXV. (Geological Survey of Canada, Paper 85-7, 1986). 25 Reimer, P. J. et al.INTCAL09 and MARINE09 radiocarbon age calibration curves, 0-50,000 years cal BP. Radiocarbon 51, 1111-1150 (2009).

2.11 Figures

Figure 2.11-1 Land emergence and peatland expansion in the Hudson Bay Lowland, Canada. The physiographic region of the Hudson Bay Lowlands, Canada (HBL) includes the coastal Hudson Bay Lowland (Coastal) and the inland HBL and James Bay Lowland (Inland HBL/JBL) ecoregions (http://www.cec.org/naatlas/). Contours (grey gradient) represent modern isostatic uplift rates (mm y-1) (Peltier, 2004), while brown shading represents modern surface elevation (m, above sea level) generated using 1:250K Canadian digital elevation data (http://www.geobase.ca/geobase/en/data/cded/description.html). The 5.5-ky (blue dashed) and 3.0-ky (red dashed) paleo-coasts presented here were extracted from the paleotopography database (Peltier, 2002; 2004). Peatland study locations are colour-coded by basal 14C ages and grouped according to trends in peat initiation frequency since mid-Holocene peat initiation (Table 2.10-1).

Figure 2.11-2 Holocene peat initiation dynamics in the Hudson Bay Lowlands, Canada.

(a) Winter (blue dashed) and summer (red solid) insolation during last 8.5 ky at 60 °N (Berger and Loutre, 1991). (b) Frequency of post-glacial peatland initiation (red histogram) in the Hudson Bay Lowlands, Canada inferred from basal radiocarbon dates (n = 100, mean 2σ age range bins), cumulative peatland initiation (%; n = 100; black solid), and corresponding cumulative land emergence (%, n = 100; blue dashed). Here, the timing of land emergence, driven by the exponential decay in glacial isostatic adjustment, is spatially related to each peat record included in this study. Using the geographic coordinates for each peat record, the corresponding time of land emergence was extracted from a paleotopography dataset (Peltier, 2002; Peltier, 2004). The cumulative land emergence and peat initiation curves presented here are directly related and can be compared for the purpose of examining lags between the timing of land emergence and the timing of peat initiation at specific peatland sites (refer to Figure 2.11-1 for total surface area estimates through time using the paleo-coasts). (c) Reconstruction of January (black) and July (red) temperature and annual precipitation (blue bars) anomalies from the North American Pollen Database (northern Quebec pollen region) (Viau and Gajewski,

2009); and (d) GISP2 (Greenland) ice core record of atmospheric methane (CH4) concentration

(Brook et al., 2000) (red) and Antarctic ice core record of atmospheric carbon dioxide (CO2) concentrations (Monnin et al., 2004) (blue) during the Holocene.

Chapter 3 Quantifying Holocene Variability in Carbon Uptake and Release Since Peat Initiation in the Hudson Bay Lowlands, Canada 3.1 Abstract

Northern peatlands are a globally significant carbon (C) reservoir, yet also function as dynamic methane (CH4) sources to the atmosphere. The fate of peatland C stores and related climate system feedbacks remains uncertain under scenarios of a changing climate and enhanced anthropogenic pressure. Here, we present a synthesis of Holocene peatland C dynamics for the

Hudson Bay Lowlands, Canada (HBL) in relation to past atmospheric CH4 trends, glacial isostatic adjustment and paleoclimate. We report that peatland age and trophic status (e.g., fen- bog stage of peatland succession), together with paleoclimate, contribute to explaining some of the temporal variation in C accumulation rates (CAR) in the HBL. Our results show that younger, minerotrophic peatlands accumulate C faster, and although detailed paleoclimate data are not available, the results suggest the possibility of higher CAR in association with warmer

Holocene climates. Peat initiation rates and CAR were greatest during the mid-Holocene; however, our model results suggest that two-thirds of the HBL C pool is stored in peat of late

Holocene age, owing to long-term peatland expansion and development. Whereas the HBL has been a net C sink since mid-Holocene peat initiation, the HBL also appears to have been a modest C source, with 85% of the losses occurring during the late Holocene as a consequence of the gradual decay of previously accrued peat. Late Holocene peat decay, under wetter climatic conditions, and from a landscape occupied by an abundance of minerotrophic peatlands, indicates that the HBL may have been a natural terrestrial source of CH4 to the late Holocene atmosphere. While the peatlands of the HBL may continue to function as a globally significant C

52 53 store, ongoing C losses from the HBL may have important implications for the global C budget and climate system.

Key words: net carbon balance, methane, paleoclimate, glacial isostatic adjustment, mire, peat decay, carbon accumulation

3.2 Introduction

Northern peatlands are important repositories of atmospheric carbon (C), storing at least 500 Pg

C or about one third of the global C pool (Yu, 2011), within approximately 3% of the terrestrial surface area (Clymo et al., 1998). Although global syntheses confirm the importance of peatlands as net C sinks in the global C cycle and as cooling agents in the global climate system (Frolking et al., 2011; MacDonald et al., 2006; Yu, 2011), peatlands also function as important C sources, principally in the form of carbon dioxide (CO2), methane (CH4), and dissolved organic C (DOC).

Anticipated high-latitude changes in temperature and the net moisture balance may enhance both peatland net primary productivity (NPP) and decomposition. Recent evidence using peat records covering the last millennium suggests that NPP may be more important than peat decay in determining the rate of C sequestration (Yu, 2012; Charman et al., 2013). Further, peat initiation and long-term development appear to be sensitive to climate; however, the relationship between paleoclimate and temporal variability in the rate of peat C accumulation that is chronologically- controlled using multiple vertical peat dates (referred to here as C accumulation rate, or CAR), remains poorly constrained (Charman et al., 2013). As a result, improving our understanding of mechanisms that could augment peat C decomposition and CO2/CH4 emissions, which together could result in a positive feedback to climatic warming, continues to be a high priority (Jones and Yu, 2010). Assessments of long-term peatland-climate-C dynamics provide an important

54 perspective for informing our understanding of the future trajectory of peatland C accumulation, under conditions of a changing climate and enhanced anthropogenic pressure.

Long-term peatland development is influenced by the interaction of autogenic and allogenic processes and associated feedbacks. As a consequence of an imbalance between productivity and decay under generally waterlogged conditions, peat C accumulates over millennia. Examination of long-term peatland-climate-C cycle dynamics relies upon assessments of the synchronicity between regional to global syntheses of peat initiation and expansion, trophic status, and CAR in the context of inferred paleoclimate and long-term atmospheric trace gas variation. Results from these approaches provide evidence that the majority of northern peatland initiation occurred in the early Holocene, under conditions of maximum summer insolation and temperature seasonality (Yu et al., 2010), synchronous with a rapid rise in ice-core inferred, atmospheric CH4 concentrations (MacDonald et al., 2006). Further, the development of CH4 emitting pre-peatland marsh conditions coupled with rapid peatland initiation and expansion of minerotrophic systems

(both sources of CH4) has been linked to the early Holocene atmospheric CH4 rise (Jones and

Yu, 2010; Smith et al., 2004; Yu et al., 2013).

Late Holocene polar ice-core records reveal a second, more gradual rise of atmospheric CH4 concentrations, following a mid-Holocene reduction (Brook et al., 2000). However, more uncertainty exists regarding the contribution of northern peatlands to this rise in atmospheric

CH4 during the middle and late Holocene. Apparent declines in the rate of northern peatland expansion, coupled with potential ombrotrophication (fen to bog transition) of existing peatlands, may contribute to reductions in peatland-derived CH4 emissions (Yu, 2011). However, recent evidence from the extensive peatland region of northeastern Canada, document delayed ombrotrophication until the latter half of the late Holocene (Holmquist and MacDonald, 2014)

55 and a late Holocene, possibly climate-driven, return of minerotrophic conditions (van Bellen et al., 2013). Both scenarios would favor potential late Holocene, peatland-derived CH4 emissions.

However, late Holocene permafrost establishment (Kuhry, 1998; Kuhry, 2008; Lamarre et al.,

2012) may contribute to local reductions in potential CH4 emissions, which may be remobilized as established permafrost degrades under climate warming or disturbance scenarios (Kuhry et al.,

2010).

Alternative hypotheses assert that possible terrestrial CH4-sources during the mid- to late

Holocene CH4 rise may be related to early anthropogenic activity, such as rice cultivation and deforestation (Ruddiman, 2003); tropical wetland contributions (Brook et al., 2000; Yu et al.,

2010); and/or lateral expansion of existing high-latitude peatlands through the mid-to late

Holocene (Korhola et al., 2010). Until recently (Holmquist and MacDonald, 2014; Packalen et al., 2014), potential late Holocene CH4 emission estimates from northern peatlands have been limited by a lack of spatially explicit evidence of mid- to late Holocene peatland initiation and expansion dynamics from major peatland regions, such as the Hudson Bay Lowlands, Canada

(HBL). Chapter 2 details new evidence of late Holocene peatland initiation and expansion, so in consideration of this new evidence, and building upon the hypothesis of potential CH4 contributions related to peatland expansion (Korhola et al., 2010), together with the possibility of sustained minero- to weakly ombrotrophic conditions in important northern peatland regions, we suggest that further analyses may indicate some CH4 contributions from northern peatlands to the late Holocene CH4 rise (Behl, 2011; Packalen et al., 2014; Yu, 2012).

The timing of peat initiation in the HBL appears to be principally controlled by glacial isostatic adjustment (GIA) following retreat of the Laurentide ice sheet and land emergence from the post-glacial Tyrrell Sea, as presented in Chapter 2 (Glaser et al., 2004a; Packalen et al., 2014).

As well, evidence of peat initiation periodicity (Chapter 2) reveals that the most rapid and intense period of peat initiation in the HBL occurred during the mid-Holocene, in advance of the late

Holocene CH4 rise (Packalen et al., 2014). Although peatland succession trajectories in the HBL can be diverse (Klinger et al., 1994; Sjörs, 1959), newly initiated peatlands begin as nutrient- rich, minerotrophic systems that are dominated by herbaceous plant communities and more easily decomposable plant litter. Following millennia of succession, peatlands may transition to more nutrient-poor, ombrotrophic systems dominated by mossy vegetation producing litter that is more difficult to decompose. The trajectory between these two endpoints contributes to variation in long-term peatland C cycling.

While evidence suggests that climate is an important control on northern peatland initiation and

C accumulation dynamics, the link between regional climate in the HBL and variation in the rate of C accumulation is not well documented (Gorham et al., 2012). Delayed glacial retreat in northeastern Canada likely deferred the Holocene Thermal Maximum (HTM) until the mid-

Holocene, as orbitally-driven insolation and seasonal differences in temperature were declining

(Renssen et al., 2009). Millennial-scale climate variation in the vicinity of the HBL is approximated by a synthesis of pollen-inferred temperature and precipitation records obtained from northern Quebec, Canada (Viau and Gajewski, 2009). Although northern Quebec and the

HBL are not identical biophysical or physiographic environments, both share similar latitudes and are influenced by the Hudson Bay. These pollen-based temperature reconstructions indicate little temperature change since initiation of HBL peatlands began ~ 8 ky BP. Reconstructed

Holocene temperature fluctuations are on the order of ± 1 °C relative to modern climate (1961 –

1990 climate normal), with mostly cooler temperatures during the mid- and late Holocene, and evidence of similar summer temperatures and a trend toward warmer winters relative to today

57 between 4.5 and 2 ky BP. Reconstructed precipitation changes show a shift at about 3 ky from somewhat reduced precipitation in the earlier part of the record to increased precipitation in the late Holocene (Viau and Gajewski, 2009).

Here, I present variability in Holocene C balance in relation to the key potential drivers of that variability, including major temporal trends in paleoclimate, ice-core inferred trace gas variation, and frequency of regional peat initiation events in the HBL. Working under the assumption that both allogenic and autogenic processes account for variation in the rate of C sequestration in the

HBL, I quantify CAR and estimate long-term decay in the HBL, at 0.5-ky intervals to model changes in the net C balance (NCB) through time. Using my analysis of CAR in the HBL, together with the synthesis presented in Chapter 2 of all available peat initiation records for the

HBL (Packalen et al., 2014, and references therein), I model net C uptake and release terms at

0.5-ky intervals following the method of Yu (2011). I then compare the resulting patterns in C uptake and release in the HBL with mid- to late Holocene, pollen-inferred paleoclimate. I hypothesize that reduced precipitation associated with the HTM may enhance CAR, while wetter conditions associated with Neoglacial climate (post-3 ky BP) may suppress CAR. As a consequence of generally cooler, wetter conditions in the HBL during the mid-to late Holocene, I further suggest that such conditions may have supported long-term CH4 production, through sustained minerotrophic to weakly ombrotrophic peatland conditions, a growing peat mass as described in Chapter 2 (Packalen et al., 2014), and the potential for greater release of the products of anaerobic decomposition. Accordingly, I examine peatland development and C dynamics in the HBL in relation to mid-to late Holocene atmospheric CH4 trends, and evaluate the potential for HBL peatlands to function as a late-Holocene terrestrial source of CH4.

3.3 Materials and Methods

3.3.1 Study setting

The HBL is the second largest continuous peatland region globally (Riley, 2011), and is located between 50° and 60° N, and 76° and 100° W (Figure 3.9-1). The northern and eastern boundaries of the HBL stretch to the margins of Hudson and James Bays, while the southern and western margins of the HBL follow the contour of the Canadian Shield. For the purposes of this study, the area occupied by the HBL is defined according to the Hudson Plains Terrestrial Ecoregion in the North American Environmental Atlas (http://www.cec.org/naatlas/). Using this definition, the

HBL has a surface area of 372,000 km2, of which up to 90% is classified as peatland (Ontario

Land Cover Database, 2000). The HBL landscape was shaped by the Laurentide Ice Sheet (LIS).

Evidence of past glacial activity is present in the form of continuous and low-relief deposits of till and diamicton, eskers and moraines, and glacio-marine sediments that were deposited overtop of Paleozoic sedimentary bedrock (Martini, 2006). Deglaciation dynamics are described in

Chapter 2, such that high rates of GIA and subsequent flooding by the Tyrrell Sea, delayed land emergence and the onset of peat initiation until ~ 8 ky BP (Packalen et al., 2014). Further, the final collapse of the LIS in the vicinity of the HBL also delayed the onset of the HTM for northeastern Canada until the mid-Holocene (Renssen et al., 2009).

Climate in the HBL is a microthermal, strongly influenced by Arctic air masses and strong winds. Modern average climate (1971–2000) is characterized by a mean and standard deviation

(s.d.) annual temperature of -2.5 ± 1.8 °C and total annual precipitation ranging from 429 to 743 mm. The growing season in the HBL stretches over a period of 119 – 162 days, and is characterized by a mean ± s.d. growing season temperature of 10.8 ± 1.0 °C, and summer

59 precipitation, as rain, equivalent to half to two-thirds of the total annual precipitation (McKenney et al., 2006). Occasional sporadic to discontinuous permafrost features occur in the HBL peatlands, together with a narrow stretch of continuous permafrost limited to the northernmost reaches of the region, near the Hudson Bay coast (Riley, 2011).

3.3.2 Sample collection and data sources

During the 2009 – 2011 field seasons, complete peat profiles were collected from representative peatlands distributed across the HBL. To minimize peat compaction during sampling, surface peat (0–80 cm) was collected using a Wardenaar- or Jeglum-type box corer (10 x 10 x 80 cm), while the subsequent sampling to mineral contact was carried out using a Russian pattern side- cutting peat sampler (5 x 50 cm). Recovered core segments were wrapped in plastic and aluminum foil and stored in polyvinylchloride pipe at -10 °C until laboratory analysis.

For the purpose of examining C flux histories using the ‘Super Peatland’ approach described by

Yu (2011), CARs were quantified using measurements of bulk density, C content, and vertical radiocarbon dating collected from 17 complete peat profiles. Ten of these records are new contributions (14C- accelerator mass spectrometry (AMS)), while the remaining seven records were obtained from previously published literature and include both 14C-AMS and conventional dates (Bunbury et al., 2012; Kettles et al., 2000; Kuhry, 1998; Kuhry, 2008; O'Reilly et al.,

2014). Peatland area increase in the HBL was estimated in Chapter 2 using a previously published synthesis of peat initiation history for the HBL, which comprises 100 new and previously reported peat basal ages (Packalen et al., 2014, and references therein). Data obtained from previously published literature were selected on the basis of three criteria: (1) their

60 specified location within peatlands of the HBL ecoregion, (2) the availability of radiocarbon dates at the basal peat-mineral interface, and (3) the availability of total peat depth.

3.3.3 Laboratory Analyses

3.3.3.1 Peat physical properties

Bulk density, loss-on-ignition (LOI), and/or C content were obtained using standard methods

(Yu, 2012). Briefly, contiguous samples of known volume (5 – 10 cm3) were cut from fresh material at 2- to 4-cm intervals and oven-dried to constant mass. Direct measurement of C content was completed using an Elementar Variomax CN analyzer (Elementar Analysensysteme

GmbH Donaustraße 7 63452 Hanau Germany). Data from the seven previously reported peat records were re-interpreted using author-contributed raw datasets of bulk density, LOI and directly measured C content (Bunbury et al., 2012; O'Reilly et al., 2014) or C content at a rate of

50% LOI , where measured C content was not available (Kettles et al., 2000; Kuhry, 1998;

Kuhry, 2008).

Bulk density (g cm-3) was calculated by dividing the dry peat mass (g) by the fresh peat volume

(cm3), while C density (g cm-3) was calculated by multiplying the bulk density of each peat increment by the corresponding C concentration. Areal dry peat mass (g cm-2) was determined as a function of peat depth and cumulative dry mass was determined on an area basis as averages weighted by layer thickness. Here, I consider the rate of C accumulation using several conventions commonly reported in the literature, including: (1) the long-term apparent rate of C accumulation (LORCA), representing average C accumulation since peat inception; (2) CAR, which uses vertical chronological control to track temporal patterns in the rate of C accumulation; and (3) NCB, the peat decay adjusted product of peatland area and CAR

(discussed in detail below), which models the net C sequestration by peatlands through time and is potentially the most appropriate term to use when comparing historical peatland C dynamics and the global C cycle (Yu, 2012).

3.3.3.2 Radiocarbon dating

New radiocarbon dates were obtained using identified above-ground plant macrofossils sampled from a 1-cm peat segment from 10 cores (Table 3.8-1). 14C-AMS dating of these samples was completed at either the Keck-CCAMS facility (Irvine, USA), the UGAMS facility (Athens,

USA) or at Beta Analytic, Inc. (Miami, USA). All dates compiled from previous studies, which include 14C-AMS dated peat (Bunbury et al., 2012; O'Reilly et al., 2014) and both 14C-AMS and radiometric dates (Kettles et al., 2000; Kuhry, 1998; Kuhry, 2008), in addition to the newly dated samples, were calibrated using IntCal09 (Reimer et al., 2009) in the clam package for R

(Blaauw, 2010). All calibrated ages are expressed as calendar years before present (y cal BP), where present is AD 1950. For the 100 sites included in the peat initiation synthesis detailed in

Chapter 2 (Packalen et al., 2014) the timing of reported and recalibrated radiocarbon (14C, AMS and conventional) dates obtained from the peat- mineral interface is understood to correspond with primary peatland initiation in the HBL following LIS retreat and land emergence from the post-glacial Tyrrell Sea.

3.3.3.3 Peat decay

To estimate the true rate of peat C accumulation, the long-term rate of C decay in the peat profile must be quantified. Other studies where this model has been applied have concluded that the constant decay function provides the simplest best fit to the data (Yu, 2011). Accordingly, the

62 exponential bog decay model (equation 3.3.3.3.1) was applied to the subset of 17 HBL peat records (Clymo et al., 1998). This model states that:

M = (p/a) *(1-e-at), (3.3.3.3.1) where M is the cumulative peat C mass (g C cm-2); p is the annual peat addition to the catotelm and determines the slope of M versus peat age curve; a is the peat decay constant that determines the curvature of M versus peat age; and t is time. The subset of new and/or synthesized age-depth models was fitted using equation 3.3.3.3.1, such that depth was represented as cumulative peat C mass (g cm-2). Alternative and potentially more ecologically meaningful peat decay rules, such as linear and non-linear (quadratic) decay through time were also explored (Clymo et al., 1998).

Although they appeared to impact estimates of total C losses, they were not found to not improve curve fitting and were not used in subsequent analyses.

3.3.3.4 Modeling of carbon dynamics

Peatland C dynamics since mid-Holocene peat initiation for the HBL was quantified following the methods of Yu (2011), using mean CAR from the 17 peat profiles described above, binned at

0.5-ky intervals. Table 3.8-2 provides a summary of the terms used to describe peatland C dynamics in the HBL; however additional detail is presented in Yu (2011). Briefly, net C balance

(NCB) refers to the difference between net C uptake (NCU) at a given time interval and net C release (NCR) by the whole peatland, due to decay, prior to the given time interval. To quantify

NCB, mean time-weighted CAR (n = 17) were determined for 0.5-ky peat cohorts. The net C pool (NCP) was then quantified for each peat cohort by multiplying the mean CAR at each 0.5- ky bin by the corresponding total peatland area at that time. Peatland area at each time interval was estimated as a relative proportion of cumulative peat initiation, assuming a modern HBL

63 extent of 372,000 km2, with 90% occupied by peatlands. NCU for each peat cohort was calculated as the total amount of C at each time interval, accounting for previous peat losses due to decomposition using the exponential decay function calculated above (NCP*1/e-at). Finally,

NCR was quantified as the total amount of C released by the entire peatland, including previously accumulated peat, prior to the relative time period. Cumulative C pool, uptake, release, and balance were also calculated as cumulative sums of the respective net C terms to track the total C terms through time in the HBL.

3.4 Results

3.4.1 Carbon accumulation and peat decay rates

Vertical radiocarbon data for ten new peat profiles are presented in Table 3.8-1 and were combined with an additional seven vertically-dated peat profiles drawn from the published literature (Bunbury et al., 2012; Kettles et al., 2000; Kuhry, 2008; Kuhry, 1998; O'Reilly et al.,

2014). With the exception of two coastal peatland records from the northwest HBL, a region of permafrost influence, most records considered here are located in the inland peatland region and include bog and fen peatlands. Shallow coastal peatlands were not sufficiently well constrained in terms of their chronologies for inclusion in the present analysis. With that in mind, the total peat depth of the 17 vertically-dated peat records ranged from 131 to 311 cm, while basal peat ages ranged between 3940 and 6810 y cal BP.

C content per unit dry peat varied along the peat profile from 12 to 54%, with the lowest values being recorded near the mineral contact. The mean and standard error (s.e.) peat C content (n =

885 from 17 cores) was 48 ± 0.2% and the mean ± s.e. bulk density (n = 885 from 17 cores) was

97.4 ± 2.3 g dm-3. Using only the basal age of the peat deposit for the subset of peat cores used in

64 this study (n = 17), we found that median LORCA ranged between 11.2 and 28.3 g C m-2 y-1

(median: 18.3 g C m-2 y-1). Consideration of temporal variability in peat accumulation rates yielded CAR estimates that varied over two orders of magnitude among intervals, and ranged from 3.5 to 310 g C m-2 y-1. Incorporating vertical radiocarbon peat dates results in a Holocene

CAR of 25 ± 0.9 (s.e.) g C m-2 y-1 (median: 18.5 g C m-2 y-1). Upon closer inspection of the peat

C accumulation records, temporal variability in CAR appears to be related to periodic increases in age-depth model slopes through time, coupled with greater C content in fen peatland records, which has been observed elsewhere (Holmquist and MacDonald, 2014; Loisel et al., 2014).

Peat accumulation in the HBL varied among the 17 sites examined in this study, with concave, convex, and linear patterns of peat age versus peat accumulation observed (n = 80 from 17 cores). Considering all sites together, we found a linear relationship (Figure 3.9-2A) between peat depth and cumulative peat OM mass (R2 = 0.74, p < 0.0001, n = 80), as well as between peat depth and cumulative peat C mass (R2 = 0.93, p < 0.0001, n = 80). Fitting the C mass data with a conservative decay model (equation 3.3.3.1) that assumes constant decay through time, to the relationship between peat age and cumulative C mass for all HBL sites together (n = 80 from

17 cores) yielded a modeled peat C addition rate (PCAR) (Yu, 2011) of 21 g C m-2 y-1 and a decay constant, a = 0.0000562 y-1 for the HBL (Figure 3.9-2B). Application of alternative peat decay rules (Clymo et al., 1998), such as linear and non-linear (quadratic) decay through time increased the decay coefficients to 0.000067 y-1 and 0.000082 y-1, respectively; however, these were not found to be statistically significant in this study. Thus, for the purpose of this study, the simplest decay term was selected; however, further investigation is warranted into ecologically meaningful representations of decay in peatlands of differing trophic states, the results of which may better constrain total C mass estimates in the HBL.

3.4.2 Peatland carbon dynamics

Net C balance (Figure 3.9-3A) is a spatially constrained snapshot of net C sequestration by representative peatlands in the HBL using the most comprehensive regional record of peat initiation and expansion currently available. The sum of modeled NCB (CCB) must be equal to the sum of the measured NCP (CCP) (Yu, 2011); however, the trajectory in arriving at the total cumulative C term for each differs since peat initiation, owing to the accounting of delayed peat decay in the NCB term. According to the assumptions and limitations described above, the modeled potential NCU by HBL peatlands during the mid- to late Holocene ranged from 0.1 –

6.5 Pg C per 0.5-ky interval, with a cumulative C uptake (CCU) since peat initiation of 45 Pg C; assuming constant peat decay since peat inception and across the growing peat landscape revealed a potential loss (CCR) of 7.6 Pg C (Figure 3.9-3B). The difference between NCU and

NCR produces an NCB ranging from 0.1 – 5 Pg C, which summed results in a modeled total

HBL C pool of 37 Pg C. Consideration of alternative decay rules, such as non-linear peat decay, which may be more ecologically meaningful, resulted in the largest decay coefficient for the

HBL (a = 0.000082 y-1, presented above), and nearly doubles the total potential peat C loss to

~ 15 Pg C. Application of the non-linear decay coefficient to NCU reduces the modeled NCB for the HBL to ~ 30 Pg C, more consistent with the previous estimate reported in Chapter 2 using

LORCA and peatland expansion dynamics to estimate the magnitude of the HBL C pool

(Packalen et al., 2014). The outcome of curve fitting using non-constant decay terms yielded non-significant co-variates, thus such results are presented here for qualitative purposes alone to highlight the need for an improved understanding of peat decay dynamics in the HBL.

The peatlands of the HBL have been a persistent C sink since they began to develop during the mid-Holocene (Figure 3.9-4). During the past 7.5 ky of peat initiation and expansion in the HBL,

66 mean CAR (0.5-ky bins) ranged from 14 to 38 g C m-2 y-1 (Figure 3.9-4D). Late Holocene CAR in the HBL remained relatively stable, below the mean CAR (25 g C m-2 y-1). By contrast, mid-

Holocene CAR was highly variable in the HBL, and peaked (~ 35 g C m-2 y-1) twice during this period: first 6 – 6.5 ky BP, and again 4 – 4.5 ky BP, following a 1.5 ky period of below average peat CAR. The variation in CAR appears to track patterns in peatland initiation, such that maximum CAR occurred synchronously with peat initiation maxima (Figure 3.9-4E). CAR also peaks in the most recent past (0.5 ky cal BP), likely related to under-decomposed acrotelm peat.

Regional pollen-inferred climate records reveal relatively cold and dry conditions versus contemporary climate and a trend toward warmer and wetter conditions between 7.5 and 4.5 ky

BP, when peat CAR was greatest and most variable. Mid-Holocene CAR maxima also appear to be preceded by warmer summers and wetter conditions (Figure 3.9-4C), which may suggest that periodic enhanced peatland productivity may contribute to CAR variability. By contrast, relatively similar to slightly cooler conditions compared to modern climate coupled with anomalously wet conditions were associated with the lowest and least variable CAR, though some minor variation is noted.

We calculated a modeled peat C sequestration rate for the HBL of ~ 7 Tg C y-1 and a mean NCB of ~ 4.4 Tg y-1. Instantaneous C accumulation rates generated from the NCB term account for both C uptake and C release over the period of peatland development (Yu, 2012). In the HBL, the modeled instantaneous C accumulation rates ranged between 14 and 42 g C m-2 y-1, and were greatest during the mid-Holocene, coincident with the period of rapid land emergence from the

Tyrrell Sea and associated intense peat initiation. A delay in peat decay is apparent in the HBL, such that most of the modeled C was released during the late Holocene. Accordingly, our results indicate that ~ 6.4 Pg C was released between 0 – 4 ky BP, compared to a modeled mid-

Holocene (4 – 7.5 ky) release of ~ 1.2 Pg C, for a total of 7.6 Pg C since the time of peat initiation in the HBL.

3.5 Discussion

3.5.1 Carbon accumulation patterns in the HBL

Peat accumulation in the HBL has occurred on a topographically favorable landscape since the mid-Holocene (Chapter 2), against a backdrop of decreasing summer insolation and insolation seasonality, and within climatic boundary conditions suitable for peatland development (Viau and Gajewski, 2009; Beilman et al., 2009; Charman et al., 2013; Packalen et al., 2014). GIA- driven land emergence in the HBL, which decays exponentially through time, is the fundamental control on the timing of peat initiation in this region as presented in Chapter 2 (Packalen et al.,

2014). LORCA is a widely reported metric for comparing mean C accumulation among peatland regions (Tolonen and Turunen, 1996). The results presented in Chapter 2, estimated using a synthesis of 100 basal peat ages, reveal a mean ± s.d. LORCA equal 18.5 ± 5.7 g C m-2 y-1 for the HBL (Packalen et al., 2014). This rate is within the range of LORCA found for the subset of peat cores used in the present study (median 18.3 g C m-2 y-1; n = 17).

LORCA in the HBL is comparable to estimates reported for similarly aged peatlands, such as the

Eastmain region, Quebec, Canada, where inundation by the Tyrrell Sea delayed peat initiation until the mid-Holocene, and C accumulated at 16.2 to 18.5 g C m-2 y-1 (van Bellen et al., 2011a).

Further, the HBL LORCA is also similar to the rates reported for older peatland regions that began to initiate in the early Holocene, such as the West Siberian peatlands where the rates range from 12.1 to 23.7g C m-2 y-1 (Turunen et al., 2001), undrained Finnish peatlands with a rate of

18.5 g m-2 y-1 (Turunen et al., 2002), and the continental peatlands of western Canada with a rate of 19.4 g C m-2 y-1 (Vitt et al., 2000).

The rate of peat C accumulation following peat inception is related to both autogenic and allogenic factors, and likely varies through time. However, LORCA as it was originally defined

(Tolonen and Turunen, 1996) may not adequately account for peat mass losses through time due to fire and/or erosion, temporal gaps in C accumulation due to permafrost accretion or enhanced decomposition, and other disturbances (Clymo et al., 1998; Yu, 2012; Tarnocai et al., 2012).

Consequently, LORCA does not permit a complete assessment of temporal patterns in C accumulation in relation to changes in peat burial rates related to peat type and ombrotrophication, growing season length, and/or other potential environmental controls (Loisel et al., 2014).

As an alternative we also consider CAR, a metric that captures the apparent temporal variability in the rate of C accumulation, through vertical chronological control using multiple radiocarbon dates along the peat column. Using the subset of well-dated cores from the HBL (n = 17), our results reveal a Holocene CAR of ~ 25 g C m-2 y-1 (median: 18.5 g C m-2 y-1) for the HBL, which is similar to the recently reported northern CAR of 22.9 ± 2 (s.e.) g C m-2 y-1 (Loisel et al., 2014); and to values from nearby subarctic Quebec, Canada (~ 24 g C m-2 y-1) (Lamarre et al., 2012) and neighboring boreal shield peatland region of northwestern Ontario, Canada (~ 24 g C m-2 y-1)

(Holmquist and MacDonald, 2014).

Temporal variability in CAR may be related in part to the vegetation community composition and hence relative decomposability of the plant material through time. Previously, bog peat dominated by Sphagnum species, was shown to be less C rich than peat associated with more

69 minerotrophic conditions (Beilman et al., 2009). Further, periods of increased CAR identified for the northern peatland region (Loisel et al., 2014), were attributed to changes in plant composition and C density through time, rather than increasing peat bulk density due to compaction of older peat cohorts. In the HBL, plant macrofossil and pollen-based peat vegetation reconstructions provide evidence that minerotrophic conditions following peat inception can persist for millennia

(Bunbury et al., 2012; Holmquist and MacDonald, 2014; O'Reilly et al., 2014; Sjörs, 1959; van

Bellen et al., 2013). Thus, for the 17 peatlands examined here, C rich inputs from young, minerotrophic peatland vegetation, coupled with periods of rapid rates of peat accumulation, may contribute to explaining apparent CAR fluctuations in this record.

Peat initiation frequency in the HBL was most intense during the mid-Holocene (Chapter 2) followed by a reduction in the rate of peat initiation through the late Holocene (Packalen et al.,

2014). This peat initiation pattern of generally declining rates though time is associated with the

GIA-controlled reduction in new land available for occupation by new peatlands. Comparing these trends with HBL CAR reveals that the mid-Holocene periods of intensified peat initiation in the HBL were accompanied by elevated CAR of ~ 35 g C m-2 y-1, with maxima around 4 and

6 ky cal BP (Figure 3.9-4). As peat initiation rates declined through the late Holocene, partly in response to exponentially declining rates of glacial isostatic adjustment, late Holocene CAR was weakly variable and remained below the long-term mean CAR.

By the onset of the late Holocene (4 ky BP), ~ 75% of the HBL peatlands had initiated, and some of these peatlands were on a trajectory toward lower C density, Sphagnum-dominated, weakly ombrotrophic conditions. Previous evidence, inferred using testate amoebae and plant macrofossils, indicates that many modern bogs in the vicinity of the Hudson Bay transitioned from the fen stage in the last millennium or during latter half of the late Holocene (Holmquist

70 and MacDonald, 2014; Lamarre et al., 2012). Today, ~ 40% of the HBL remains classified as fen peatlands (Ontario Land Cover Database, second edition, 2000). Thus, lower rates of peat initiation in the HBL during the late Holocene, and the corresponding trend of lower CAR during this period may be more strongly influenced by peat succession dynamics occurring in existing

HBL peatlands, than by C inputs from newly emerging peatlands.

3.5.2 Modeled peatland carbon dynamics in the HBL

Using all available peatland records for the HBL with vertical chronological control, which included minerotrophic to ombrotrophic peatlands in both the presence and absence of permafrost, we assess the temporal variation in Holocene C uptake and release using the ‘Super

Peatland’ approach (Yu, 2011). As part of this process, peat C decay was estimated assuming constant decay and resulted in a slightly concave decay model of age versus depth. Though frequently considered, constant decay may be an oversimplified assumption, as individual peatlands or peat types may exhibit very different decay patterns or may not be well fitted to the decay model described by Clymo et al. (1998). In addition to this limitation, young, shallow coastal peatlands are not well represented in this study, owing to the challenges of dating the relatively young peat deposits. Thus we also consider alternative decay assumptions qualitatively, as model outputs yielded non-significant terms in the present study, and sample limitations precluded additional investigation.

Nevertheless, using a constant peat decay term (Clymo et al., 1998) summation of the NCB model output term, resulted in a total contemporary C mass equal to 37 Pg C (Figure 3.9-4).

Under conditions of constant peat decay, this modeled total C mass estimate is about 20% greater than a previous estimate of 30 Pg C presented in Chapter 2, and obtained using patterns of peat

71 initiation and expansion, land emergence, and measured C mass (Packalen et al., 2014). Still, peat decay is not well understood for HBL peatlands, especially regarding the relative role of fen versus bog peatlands. Consideration of alternative decay rules in the calculation of NCB, such as non-linear peat decay, nearly doubles the total C release term (NCR), and results in a reduced

HBL C mass estimate of 30 Pg C, consistent with the previous estimate presented in Chapter 2

(Packalen et al., 2014). However, we note here that the non-constant decay terms were not significant in this study, and thus are used only for qualitative purposes to demonstrate the effect of enhanced peat decay, at the landscape scale, on the total C pool in the HBL. Our results confirm that a better understanding of peat decay, especially concerning the influence of fen versus bog peat cover, is needed to fully explore Holocene C dynamics.

Although the absolute values for HBL C pools reported here are uncertain, temporal patterns in

Holocene C dynamics may be more meaningful and suggest that a total of 7.6 Pg C may been lost from the total peatland C mass since peat inception began in the HBL during the mid-

Holocene. Maximum modeled C sequestration and release was greatest during the late Holocene, as the peat complex expanded across the rapidly emerging low-relief landscape. Additional decomposition of C flux terms reveals a mid-Holocene (4 – 7.5 ky BP) net HBL C sequestration equal to 14 Pg C, which occurred at a mean rate of 4 Tg C y-1, and included a small C release of

1.2 Pg C. During the late Holocene, new peatland initiation coupled with rapid peat accretion in existing peatlands resulted in an additional C sequestration of 23 Pg C, at a mean rate of 6 Tg C y-1, and includes a larger potential C release of 6.4 Pg C. The larger late Holocene C release accounts for the gradual decay in the catotelm of previously deposited peat cohorts, which is expected (Yu, 2011); however, the cumulative effect of long-term decay is smaller in the HBL, due in part to the relatively young age of the HBL peat complex.

3.5.3 Net carbon balance and paleoclimate in the HBL

The timing of peat initiation events in the HBL is primarily driven by uplift; however, paleoclimate may have contributed to facilitating initiation events, and perhaps even more so, in influencing the rates of Holocene C accumulation and NCB in the millennia following initiation.

As a consequence of the delayed retreat of the LIS near the HBL, the timing of the HTM in northeastern Canada is thought to have been deferred until ~ 6 – 7 ky BP (Renssen et al., 2009).

The delay of the HTM in the HBL until the mid-Holocene corresponds to the earliest period of intensified peat initiation and a period of peak C accumulation in the HBL. Coupled climate model simulations and pollen reconstructions for northeast Canada suggest a possible warming of ~ 1 °C vs. pre-industrial temperatures, which may account for higher C sequestration rates via enhanced primary production during this time. However, regional paleoclimate reconstructions appear to confound this relationship.

Comparisons using apparent CAR among Canadian peatlands regions in the vicinity of the HBL reveal that maximum CAR occurred during the mid-Holocene, while the lowest CAR occurred in the late Holocene (van Bellen et al., 2011a; Vitt et al., 2000), and early Holocene for the continental peatlands of western Canada. Lower CAR in both the western Canadian peatlands and the Eastmain region, Quebec, during the late Holocene was attributed to a cooling climate and drier peatland conditions related to height-induced surface drying and/or permafrost accretion (van Bellen et al., 2011a; Vitt et al., 2000), while low early Holocene CAR was attributed to dry peatland conditions (Vitt et al., 2000). However, pollen-inferred climate reconstruction for northern Quebec, Canada (Viau and Gajewski, 2009) that are consistent with an independent paleoclimatic record from the HBL (McAndrews et al., 1982), imply lower than present temperature and precipitation during the mid-Holocene (until 4.5 ky BP). Thus, in the

HBL, the negative precipitation anomaly (i.e., reduced precipitation compared with the contemporary climate normal) corresponded to higher CAR during the mid-Holocene. Given the uncertainty in mid-Holocene climate in the HBL, the relationship between CAR and climate is interpreted cautiously here and warrants further investigation.

Regional paleoclimate reconstructions for the late Holocene near the HBL reveal climatic conditions similar to today, with the exception of the last 1.5 ky when increased precipitation and reduced temperatures associated with Neoglacial cooling are recorded. During this period, CAR appears to increase slightly as precipitation increases, and reaches a late Holocene peak CAR around 1.2 ky BP followed by a small decrease in CAR around 0.8 ky BP, around the Medieval

Climate Anomaly (MCA). Similarly, a decline in CAR during the MCA is also apparent in the northwestern Ontario peatlands, adjacent to the HBL (Holmquist and MacDonald, 2014); while a slight increase is noted at the time of the Little Ice Age (LIA). Evidence recorded in Quebec,

Canada peatlands to the east of Hudson Bay document a wet shift during the LIA and a return of minerotrophic conditions, which were attributed to a cooler, wetter climate (van Bellen et al.,

2013). However, our conclusions in this regard are highly limited by the resolution of the time series used for CAR in the present study (0.5-ky bins) compared with the century scale of the

MCA and LIA climatic events. This limitation highlights the need for further high resolution examinations of these time periods to better constrain the relationship between climate and CAR in the HBL.

Contemporary surface measurements have revealed that C losses from peatlands are tightly coupled with vegetation composition, trophic status and hydroclimatic conditions (Bubier, 1995;

Moore and Dalva, 1993) and can occur via aerobic decomposition in the acrotelm and by anaerobic decomposition in the catotelm (Clymo et al., 1998; Gorham, 1991). Further rapid

74 initiation and expansion of minerotrophic fens in the early Holocene have been proposed to have contributed to peak early Holocene CH4 concentrations (MacDonald et al., 2006; Smith et al.,

2004). Further, sources of early Holocene CH4 emissions have more recently been expanded to include pre-peatland wetlands such as marshes and wet fens as facilitators of biosphere CH4 transport to the atmosphere (Yu et al., 2013). In the HBL, surface and airborne measurements of contemporary snow-free CH4 flux, interpreted using a chemical transport model (Pickett-Heaps et al., 2011) suggest that the largest rates of net CH4 flux occur in association with the older, inland peatland region rather than the younger coastal region. Accordingly, Packalen et al.

(2014) recently hypothesized that the largest atmospheric contributions of peat-derived CH4 from the HBL may have occurred during the late Holocene, and suggested that a 3-ky BP peat mass

-1 could potentially release 1 – 7 Tg CH4 y to the late Holocene atmosphere (also presented in

Chapter 2).

Here, we examined patterns in Holocene NCB of HBL peatlands, for comparison with ice-core inferred atmospheric trace gas variation (Figure 3.9-4). Similar climate-CAR patterns are also apparent in northwestern Ontario peatlands, adjacent to the HBL (Holmquist and MacDonald,

2014). Assuming a constant decay term, our model estimations predict a late Holocene C release of 6.4 Pg C, equivalent to a mean rate of 1.6 Tg C y-1 (range: 1 – 2.5 Tg C y-1) from the more developed HBL peatland landscape. Alternative non-linear decay patterns, though not significant in this study, but may be more ecologically realistic, suggest even larger C losses from the HBL

C mass, such that total late Holocene C losses increase to ~ 10.2 Pg C, corresponding to a mean rate of ~ 2.6 Tg C y-1 (range: 1.5 – 4.1 Tg C y-1) during the late Holocene and from across the

HBL. Assuming a portion of these C losses occur as CH4, then this model evidence, irrespective of decay assumption is comparable to previous estimates of late Holocene CH4 production

75 potential from the HBL. Therefore, our analysis of modeled C dynamics provides an additional line of evidence of natural terrestrial C contributions from the HBL to the late Holocene atmosphere, particularly in the form of CH4 flux, as a consequence of the minerotrophic to weakly ombrotrophic peat patterning that characterizes the HBL landscape.

The trajectory of NCB in the HBL differs from other important peatland regions globally, such that C sequestration was greatest during the late Holocene, rather than at a minimum. While evidence of peat succession is present in the HBL, as peatlands shift from minerotrophic to ombrotrophic systems, our data show that the relationship between cumulative peat mass and peat depth in the HBL remains linear. Assuming that the relationship between peat mass and depth is related in part to trophic succession in the HBL, and given that there is little evidence to suggest that mass peat losses are occurring as a consequence of drying and oxidation of peat, the peatlands of the HBL may continue to accumulate peat (and sequester C) for some time to come.

Our findings convey the importance of the HBL in global C accounting, and suggest that significant C losses from the HBL may have important implications for the global C budget and climate system.

3.6 Conclusion

Here, we have examined trends in the timing and magnitude of variation among CAR and net C uptake, release, and balance since peat inception began in the HBL. Previous findings (Chapter

2) have confirmed the fundamental control of GIA over the timing of peat initiation in the HBL and that peatland development has occurred within the context of a weakly varying climate

(Packalen et al., 2014; Glaser et al., 2004a; Glaser et al., 2004b). Our findings do not reveal a tight coupling between climate and CAR in the HBL; however they do suggest a potential link

76 with climate that warrants further investigation. Further, we report that CAR may be more strongly related to peatland succession dynamics in the HBL. Accordingly, we find that CAR is related to the intensity of peat initiation, the latter of which determines the proportion of young peatlands with high values for CAR on the rapidly emerging HBL landscape. Our data show that while the HBL has been a persistent C sink for millennia, more than two-thirds of the total C mass accrued during the late Holocene, when CAR remained below the long-term mean. Most of the HBL peatlands initiated during the mid-Holocene, in advance of the late Holocene atmospheric CH4 rise. Moreover, our findings suggest that most of the potential C lost from the

HBL occurred during the late Holocene, likely owing to decay of previously deposited peat.

Modern evidence suggests the largest CH4 emissions in the HBL occur in association with older, inland patterned peatlands (Pickett-Heaps et al., 2011). Given that sustained minerotrophic to weakly ombrotrophic peat patterning typifies the HBL landscape, the persistent C release during the late Holocene may provide a first line of evidence of natural terrestrial C contributions from the HBL to the late Holocene atmosphere, particularly in the form of CH4.

3.7 Acknowledgements

Sincere thanks to Jim McLaughlin for research funding and field support, provided by the

Ontario Ministry of Natural Resources’ Applied Research and Development Branch and Far

North Branch, under the auspices of projects CC-021 and FNIKM 028. Additional support for field work and radiocarbon dating was provided by grants from the Natural Sciences and

Engineering Research Council of Canada (NSERC) and the Ontario Ministry of the Environment through the Climate Change and Multiple Stressor Research Program at Laurentian University to

SAF. Support in the form of an NSERC Alexander Graham Bell Canada Postgraduate

Scholarship and an Association of Canadian Universities for Northern Studies, Canadian

Northern Studies Trust Scholarship, and field research grants from the Society of Wetland

Scientists and Aboriginal Affairs and Northern Development Canada’s Northern Scientific

Training Program were provided to MSP. We also thank A. Dyke for providing access to the

Canadian basal radiocarbon database, P. Kuhry for contributing raw peat core data, and three reviewers (D. Beilman, Z. Yu, and one anonymous) for comments on an earlier draft of this manuscript.

3.8 Tables

Table 3.8-114C-accelerator mass spectrometry (AMS) dating of peat macrofossil of known provenance for 10 new sites in the Hudson Bay Lowlands, Canada; sorted by increasing latitude. Basal peat dates first reported in Packalen et al., 2014.

Peat Dated δ13C Median 2σ Age Lat Long Elevation Peatland Material 14C-AMS (y BP) Site Depth Lab Number Depth (‰ Age (y cal Range (y cal (d.d.) (d.d.) (m) Class. Dated (cm) (cm) PDB) BP)* BP)* Weakly wood KJ4-3 51.59 -81.78 66 131 UGAMS-11256 45-46 -26.4 130 ± 30 128 (-3) - 280 Ombrotrophic fragments wood UGAMS-11255 80-81 -25.7 1560 ± 20 1468 1400-1520 fragments wood UGAMS-12717 129-131 -27.4 3630 ± 25 3940 3870-4070 fragments Weakly Sphagnum KJ3-3 51.59 -81.79 65 176 UGAMS-11663 40-41 -27.0 F14C=1.024 ± 0.003 -5.50 (-4.4) - (-6.6) Ombrotrophic remains UCIAMS- wood + herb 175-176 4170 ± 20 4720 4620-4830 97825 stems KJ2-3 51.59 -81.76 65 Minerotrophic 246 UGAMS-11252 51-52 twig -23.8 1720 ± 20 1629 1560-1700 Sphagnum UGAMS-11253 130-131 -28.5 2960 ± 25 3137 3000-3240 stems UGAMS-11254 177-178 wood -28.8 3250 ± 25 3466 3400-3560

UCIAMS- wood + herb 245-246 4130 ± 25 4860 4840-4960 97824 stems Weakly VM4-5 52.70 -84.18 103 286 UGAMS-11668 53-54 wood -28.2 1390 ± 20 1302 1290-1340 Ombrotrophic wood UGAMS-11267 156-157 -26.7 3670 ± 25 4008 3910-4090 fragment wood UGAMS-11673 285-286 -26.9 5530 ± 25 6320 6290-6400 fragment Weakly VM4-1 52.71 -84.19 105 311 UGAMS-11265 46-47 wood -27.7 160 ± 20 186 (-2) - 280 Ombrotrophic UGAMS-11665 95-96 wood -26.8 1620 ± 20 1519 1420-1560

UGAMS-11666 143-144 wood -27.0 2470 ± 25 2571 2370-2710

UGAMS-11266 204-205 wood -28.5 3920 ± 25 4359 4260-4420

UGAMS-11667 253-254 wood -28.1 4860 ± 25 5600 5490-5650

needles; UGAMS-12718 310-311 -27.6 5550 ± 25 6340 6300-6400 wood

Table 3.8-1, continued

Peat Dated δ13C Lat Long Elevation Peatland Median Age 2σ Age Range Site Depth Lab Number Depth Material Dated (‰ 14 (d.d.) (d.d.) (m) Class. C-AMS (y BP) (y cal BP)* (y cal BP)* (cm) (cm) PDB) UGAMS- VM3-2 52.71 -84.17 102 Minerotrophic 262 87-88 wood -25.4 2530 ± 20 2625 2500-2740 11262 UGAMS- 122-123 wood -28.7 3640 ± 25 3951 3880-4080 11263 UGAMS- 167-168 wood -28.6 3790 ± 25 4174 4090-4240 11264 UGAMS- 206-207 wood fragments -27.9 5070 ± 25 5816 5750-5900 11664 UGAMS- 261-262 wood fragments -27.5 5620 ± 25 6400 6310-6450 12716 Weakly UGAMS- VM1-3 52.72 -83.94 86 210 72-73 wood -27.1 2060 ± 25 2028 1950-2110 Ombrotrophic 11257 UGAMS- herbaceous 209-210 -25.4 4390 ± 20 4940 4870-5040 12715 remains UGAMS- VM2-5 52.72 -83.94 86 Minerotrophic 180 41-42 twig -25.6 140 ± 20 142 (-3) - 280 11258 UGAMS- 94-95 wood -28.9 2150 ± 20 2143 2060-2300 11259 UGAMS- 127-128 wood -29.3 3890 ± 25 4336 4250-4410 11261 UGAMS- 179-180 wood fragments -29.0 4980 ± 25 5700 5640-5850 11674 Weakly UGAMS- HL02 54.61 -84.60 85 230 40-41 wood -26.4 1730 ± 25 1644 1570-1710 Ombrotrophic 11249 UGAMS- 78-79 wood fragment -26.3 2100 ± 20 2073 2000-2130 11675 UGAMS- 109-110 wood -29.4 2450 ± 25 2504 2360-2700 11250 UGAMS- 163-164 twigs -26.1 3150 ± 30 3379 3270-3450 11676 UGAMS- 229-230 Sphagnum remains -26.0 4020 ± 25 4480 4420-4570 11422 UGAMS- HL03 54.68 -84.60 99 Minerotrophic 269 70-71 Sphagnum remains -27.5 1880 ± 25 1832 1740-1880 12720 UGAMS- 127-128 Sphagnum remains -27.8 2560 ± 25 2723 2520-2750 12721 UGAMS- 190-191 Sphagnum remains -28.7 2920 ± 25 3067 2970-3200 12722 UGAMS- 268-269 Sphagnum remains -26.5 2700 ± 25 3550 3290-3880 11423 * calibrated with IntCal09 calibration curve (Reimer et al., 2009).

Table 3.8-2 Summary of deconstruction terms used to describe peatland carbon (C) dynamics in the Hudson Bay Lowlands, Canada (adapted from Yu, 2011).

Term* Explanation NCP Net C pool observed from present day peat cores and derived from the product of C accumulation and corresponding peatland extent, at 0.5 ky intervals. NCU Net C updake modeled for the entire peatland at 0.5 ky intervals, as the sum of observed NCP and predicted C released according to the exponential peat decay model (Eq. 3.3.3.3.1), for each peat cohort. NCR Net C release modeled for the entire peatland at 0.5 ky intervals, including previously accumulated peat, and derived from the exponential peat decay model (Eq. 3.3.3.3.1). NCB Net C balance modeled as the difference between NCU and NCR. CCP Cumulative C pool (sum of NCP) observed from present day peat cores and estimated for the entire peatland through time. CCU Cumulative C uptake (sum of NCU) modeled for the entire peatland through time. CCR Cumulative C released (sum of NCR) modeled for the entire peatland through time. *All terms reported in Pg C.

3.9 Figures

Figure 3.9-1 Physical features and peat study locations in the Hudson Bay Lowlands, Canada. The region includes the coastal Hudson Bay Lowland (HBL) and inland HBL and James Bay Lowland (JBL) ecoregions, where the boundary between the two regions is indicated by a thin black line (www.cec.org/naatlas/). Numbered contours (thick black lines) represent modern isostatic uplift rates (mm y-1) (Peltier, 2004), while grey shading represents modern surface elevation generated using 1:250K Canadian digital elevation data (www.geobase.ca). Available basal dated peat core locations for estimating peatland expansion (n = 100; black dots) (Packalen et al., 2014; and references therein) and vertically dated peat study locations for carbon analysis (n = 17, white triangles) are indicated.

Figure 3.9-2 Peat age, depth, and cumulative carbon (C) mass relationships and modeled exponential peat decay for the patterned peatlands of the Hudson Bay Lowlands, Canada. A. linear relationships among median peat age (n = 80 from 17 cores; dots) and 2σ age ranges (error bars) are presented for 17 vertically-dated peat cores sampled from the HBL. Peat ages were estimated following calibration of 14C ages using the IntCal09 curve (Reimer et al., 2009); and B. cumulative C mass was fit with an exponential peat decay model (Clymo et al., 1998). The best-fit peat C addition rate (PCAR) and long-term peat decay parameter (a) for the HBL are presented.

Figure 3.9-3 Holocene peat carbon (C) pools and modeled peat C terms (Yu, 2011) for the Hudson Bay Lowlands, Canada, since peat initiation began ~ 8 ky BP. A. peat age versus measured (n = 17 cores) net peat C pool (NCP) and modeled (n = 100 cores) net C uptake (NCU), release (NCR) and balance (NCB), where NCB = NCU – NCR.; and B. peat age versus cumulative measured peat C pool (CCP) and modeled cumulative peat C uptake (CCU) and release (CCR) following long-term constant peat decay, are presented at 0.5-ky intervals.

Figure 3.9-4 Holocene peatland area increase, carbon (C) accumulation, and net C balance, since peat inception for the Hudson Bay Lowland, Canada.

Data are presented in relation to regional paleoclimate, global insolation trends, and atmospheric methane (CH4) and carbon dioxide (CO2) variability. A. winter and summer insolation during last 7.5 ky at 60 °N (Berger and Loutre, 1991); B. GISP2 (Greenland) ice core record of atmospheric CH4 concentration (Brook et al., 2000) and Antarctic ice core record of atmospheric

CO2 concentrations (Monnin et al., 2004); C. reconstruction of winter and summer temperature and annual precipitation anomalies inferred using a synthesis of the North American Pollen Database for the northern Quebec, Canada pollen region (Viau and Gajewski, 2009); D. mean (± standard error) time-weighted C accumulation rates (CAR, n = 17) ; and modeled (Yu, 2011) net carbon uptake and balance, at 0.5-ky intervals (n = 100); and E. frequency of post-glacial peatland initiation in the HBL, inferred from basal radiocarbon dates (n = 100, 0.5-ky bins), cumulative peatland initiation (%; n = 100), and peatland surface area increase (%, n = 100) (Packalen et al., 2014).

Chapter 4 Climate and Peat Type in Relation to the Spatial Distribution of the Peat Carbon Mass in the Hudson Bay Lowland, Canada 4.1 Abstract:

Northern peatlands store more than 500 Gt of carbon (C); however, controls on the spatial distribution of the stored C (termed C mass, kg m-2) may differ regionally, owing to the complex interaction among climate, ecosystem processes, and geophysical controls. As a globally significant C pool, elucidation of controls on the spatial distribution of the peat C mass in the

Hudson Bay Lowland, Canada (HBL) is of particular importance. Although peat age in the HBL is closely related to timing of land emergence and peat depth, considerable variation in the total

C accumulated among sites of similar peat age suggests that other factors may explain trends in the distribution of the peat C mass. Here, we present detailed peat lithologies for bog, fen and coastal mesoforms, and consider climate as a control on the spatial distribution of the peat C mass (n = 364) across the HBL. We find that temperature, precipitation, and potential evapotranspiration each explain up to half of the variation in the peat C mass, such that regions in the HBL characterized by warmer and wetter conditions tend to support larger peat C masses.

Furthermore, we show that the widespread bog-fen patterning across the HBL is related to the observed spatial variability in C mass, suggesting that small scale topographic and ecohydrological controls are potentially important determinants of C mass accretion. Our findings support the hypothesis that both climate and ecohydrological factors are important drivers of peat C mass accretion, alongside geophysical controls on the timing of peat initiation in the HBL.

86 87

Keywords or Index terms: peatland, mean annual air temperature (MAAT), mean annual precipitation (MAP), potential evapotranspiration (PET), moisture balance, continentality

Key Points:

• Carbon-climate-peat linkages are presented for the Hudson Bay Lowland, Canada

• Climate may explain up to half of the spatial distribution of the peat C mass

• Wide C mass variation within local climate space suggests autogenic controls

4.2 Introduction

Northern peatlands occupying landscapes at latitudes north of 45°N, develop within a broad climatic domain, where mean annual air temperatures (MAAT) may range between -20 and

15°C, and mean annual precipitation (MAP) may range from < 100 to 3000 mm (Yu, 2012). Yet, northern high latitude peatlands occupy a unique biogeographic niche where net primary production (NPP) exceeds organic matter decomposition under sustained waterlogged conditions. Currently, northern peatlands are estimated to store 250 – 550 Pg carbon (C), most of which has accumulated during the present interglacial (Gorham, 1991; Loisel et al., 2014;

Turunen et al., 2002;Yu et al., 2010). Some of the main source of uncertainty in global peatland

C mass estimates stem from the need for region specific values of peat depth, dry peat mass, and

C density (Yu, 2012).

Northern high latitudes are expected to experience the most intense climatic change (McGuire et al., 2009), making cold climate peatlands especially vulnerable to changes in net C storage by potentially affecting rates of NPP and peat decomposition, enhancing disturbance, and/or unlocking large peat C pools from thawing permafrost. Further, climate-induced changes to C dynamics in northern peatlands may result in vastly different C flux scenarios depending upon

88 peat type (Loisel et al., 2014) and climatic conditions. While the role of climate and peat type as controls over the amount of C stored as peat remains poorly understood, contemporary evidence suggests that recent accelerated climatic change is impacting the net C stored in northern peatlands (Gorham et al., 2012). Under conditions of natural and anthropogenic climatic change,

NPP may be enhanced, suppressed, or offset by decomposition, altering the capacity of peatlands to function as long-term C reservoirs. Thus, alteration of peatland C dynamics may have important implications for the nature of future climate. As a result, anticipating the future trajectory of peatland C storage potential necessitates improved data on spatio-temporal controls on peatland C dynamics to support inclusion of these ecosystems in Earth and climate system models and C budget accounting.

In the present study, we quantify the roles of climate and peat type on the spatial distribution of the peat C mass in the Hudson Bay Lowland, Canada (HBL). Today, the HBL represents the second largest peatland region in the world. Within a nearly continuous peat cover patterned with bog- and fen-like features, the HBL maintains a globally significant C reservoir in excess of 30

Pg C, as quantified in Chapter 2 and 3 (Packalen and Finkelstein, 2014; Packalen et al., 2014) and that C mass has accumulated since mid-Holocene peat initiation. While climate has supported long-term C storage in the HBL, it remains unclear what factors control the size and spatial variation of the peat C mass in this region. Modern gridded climate data reveal a series of climate gradients, controlled by latitudinal position and proximity to the Hudson and James Bays where annual sea-ice formation is a major driver of regional climate (Gough et al., 2004). These gradients may explain patterns in the distribution of the total peat C mass across the HBL.

Although long-term paleoclimate records for the HBL are few, it is apparent from data presented in Chapters 2 and 3 that paleoclimate in the HBL exhibits relatively small departures from

89 contemporary climate normals over the past eight millennia (Packalen and Finkelstein, 2014;

Viau and Gajewski, 2009), and provide evidence that climate variability through space likely remained relatively consistent through time.

Glacial isostatic adjustment (GIA), especially during the mid-Holocene, is coupled to the rate and history of peat initiation in the HBL. Consequently, peat depth and C mass are related in part to the timing of land emergence. That being said, wide scatter within given time intervals suggest that other factors contribute to patterns in peatland development and C accretion in the HBL.

Moreover, peatlands initiated in many regions of the HBL under cooler- and wetter-than-present climatic conditions (Glaser et al., 2004a; Glaser et al., 2004b; Packalen et al., 2014).

Nonetheless, empirical and model evidence presented in Chapter 3 reveal that Holocene-scale variation in apparent C accumulation rates in the HBL appears to be related to peat type and to a lesser extent to climate (Packalen and Finkelstein, 2014). Accordingly, maximum C accumulation rates occurred during the mid-Holocene when young, early emerging fen peatlands dominated, while maximum total C mass development occurred during the late Holocene as the spatial extent of the accumulating peat mass expanded (Packalen and Finkelstein, 2014).

Notwithstanding the role of GIA, maximum and minimum temperature and precipitation have been previously reported to be important controls on the development of the circum-Arctic peatland C mass, together with bioclimatic parameters, such as potential evapotranspiration

(PET), net moisture balance (PET/MAP), growing degree days (GDD), and photosynthetically active radiation (PAR) (Charman et al., 2013; Eppinga et al., 2009; Holmquist and MacDonald,

2014; Loisel et al., 2012). MAAT, GDD and PAR have been invoked as controls on NPP for both vascular and non-vascular vegetation, whereby rates of peat accumulation rise in response to increasing MAAT, GDD, and PAR (Beilman et al., 2009; Charman et al., 2013). Further,

90 increasing PAR and growing season length, in the absence of moisture stress, lead to increased rates of C accumulation and a potential negative feedback to climatic warming (Loisel et al.,

2012). These trends are supported by Holocene-scale peat-paleoclimate records that reveal maximal peat C accretion during the warm early Holocene, when orbitally induced summer insolation peaked (Jones and Yu, 2010; MacDonald et al., 2006).

While climate may be an important control over the spatial variability in C mass at larger spatial scales (e.g., degree latitude), peat type as determined by local ecohydrological processes, trophic status, and vegetation type may also contribute to the spatial variation of the C mass at local to regional spatial scales. Further, non-linear responses by peatlands to autogenic controls, such as vegetation decomposition, and allogenic controls, such as internal (e.g., polar pattern,

ENSO)/external (e.g., PAR) forcing (Belyea and Baird, 2006) may disrupt the theoretical fen-to- bog peat succession trajectory. Still, maintenance of sustained waterlogged conditions is one of the principal mechanisms responsible for peat accumulation, and climate-driven changes may be reflected in peatland vegetation patterning and ecohydrology (Frolking et al., 2011; Jones and

Yu, 2010). In the HBL, precipitation, PET, and local drainage patterns are anticipated to contribute to the maintenance of differentially waterlogged conditions to support peatland development characterized by bog-fen patterning (Sjörs, 1959). However, changing trophic status and associated increases in bulk density with depth disconnect the surface peat over time from groundwater and nutrient supplies via progressively decreasing hydraulic conductivity. As a result, we expect that the younger, emerging landscapes near the coast will support more fen peatlands, while more inland reaches likely support a shift toward bog peatlands. As the landscape continues to emerge, effective moisture in near-surface peat, together with peat

91 temperature may become even more important predictors of peatland capacity to continue to accumulate peat and sequester C (Beilman et al., 2009; Loisel et al., 2012).

Given the HBL’s proximity to the Hudson and James Bays, the moisture balance (PET/MAP) and/or the relative continentality may be important explanatory factors for the distribution of the peat C mass in the HBL. PET-dominated regions (PET/MAP ratios approaching 1) are characterized by enhanced nutrient use efficiency among vegetation to support hummock-hollow patterning; however, drainage-dominated peatlands have moisture balances typically much less than one, and are associated with enhanced peat accumulation and ridge-flark (akin to bog- and fen-like features) pattern development (Eppinga et al., 2009). Greater temperature seasonality combined with PET-dominated climates reflect a drier, more continental climate and the implications of this on the total C mass remain uncertain under future climatic conditions.

Temperature seasonality in relation to PET/MAP can thus be used as an indicator of the relative continentality of climate, and previous results reveal faster Sphagnum growth rates in drier, more continental climates characterized by both greater seasonality and PET/MAP ratios (Loisel et al.,

2012). Although Sphagnum-dominated peat may be less C dense (Beilman et al., 2009; Loisel et al., 2014), potentially deeper peat may contribute to a larger total C mass that is more resistant to decomposition. Future C dynamics are likely to be more sensitive to sustained changes to the net moisture balance rather than short-term hydrologic variability (Dise, 2009), as rising temperatures potentially enhance evapotranspiration and support enhanced disturbance.

Consequently, temperature and continentality will be evaluated in the HBL for the whole peatland mass, and in relation to the spatial distribution of peat type across the landscape.

The objectives of this study are to elucidate important factors that explain the spatial distribution of the peat C mass in the HBL. I hypothesize that both climatic and ecohydrological controls, as

92 indicated by peat type, are important in explaining the spatial pattern in the distribution of the peat C mass in the HBL. I further hypothesize that increasing temperature, GDD and PAR would be positively correlated with the total peat C mass, while precipitation may be positively correlated with the peat C mass except in permafrost-impacted peatlands where both positive and negative relationships have been reported (Charman et al., 2013; Holmquist et al., 2014; Loisel et al., 2012). I expect a gradient in the relative continentality of the HBL from the coast inland, with implication on the distributions of the net moisture balance. Moreover, as the distribution of the peat C mass may also be related to the peat type, we distinguish among bog, fen and coastal peatland mesoforms in terms of both peat C characteristics and in relation to the major bioclimatic parameters presented here (Loisel et al., 2014). To this end, I hypothesize that fen peat will be shallower with greater C density than Sphagnum-dominated bog peat, owing in part to differences in the relative decomposability of the vegetation inputs to the accumulating peat mass.

4.3 Study setting

I define the HBL here using the inland and coastal HBL/James Bay Lowland (JBL) ecoregions described in the North American Environmental Atlas (http://www.cec.org/naatlas/).

Accordingly, the HBL stretches along the shoreline of Hudson and James Bays and inland toward the north-western margin of the Precambrian Shield (Figure 4.10-1). Under the influence of the Hudson and James Bays, the HBL has a microthermal climate, strongly influenced by

Arctic air masses and strong winds (Figure 4.10-2). During the early Holocene (until ~ 8.5 ky

BP), the HBL was largely covered by the Tyrrell Sea, providing a maximum age for HBL peatlands that is several millennia younger than the extensive peatlands of north-western North

America and Siberia. In the HBL, a rapidly rebounding landscape following the final collapse of

93 the Laurentide Ice Sheet (LIS) during the mid-Holocene, gave way to widespread peat initiation as documented in Chapter 2 (Packalen et al., 2014) over low-relief deposits of till and diamicton, eskers and moraines, and glacio-marine sediments overtop of Paleozoic sedimentary bedrock

(Martini, 2006). Approximately 8-ky hence, a nearly continuous peat cover, with a mean slope of

0.5 m km-1, has developed across the HBL’s ~ 372,000 km2 landscape (an updated spatial extent presented in Chapter 2) and accounts for nearly 10% of the arctic and subarctic peatland area

(Packalen et al., 2014; Tarnocai et al., 2009). Nearly 90% of the HBL is classified as peatland, characterized by alternating bog and fen features inland that becomes more fen-dominated in the coastal ecoregion. Further, the HBL includes the southernmost extent of non-alpine permafrost globally, which affects ~ 1% of the peatlands, principally as a narrow stretch of continuous permafrost along the Hudson Bay coast, and elsewhere as sporadic to discontinuous permafrost

(Riley, 2011). As well, coastal deltas, fen water tracks, networks of low order streams, small pools, and shallow lakes can be observed throughout the region, as was observed during several field campaigns and described in Chapters 2 and 3 (Glaser et al., 2004a; Glaser et al., 2004b;

Packalen and Finkelstein, 2014; Packalen et al., 2014; Sjörs, 1959). Consequently, little infrastructure development has occurred in the HBL, resulting in a large geographic region characterized by extreme remoteness, limited accessibility, and little evidence of anthropogenic disturbance.

4.4 Methods

4.4.1 Peat and bioclimatic data sources

Here, I use 42 detailed peat C records summarizing new and previously reported peat properties

(Table 4.9-1, and references therein), together with a larger synthesis of 364 peat depths drawn

94 from previously published literature (Riley, 2011), to scale C mass across the HBL (Figure 4.10-

1). Modern climate data (Figure 4.10-2) were used to examine potential climatic controls on the spatial distribution of total C pools across the HBL. Although subject to some uncertainty, these publicly available data are one of the most comprehensive spatial estimates currently available.

Briefly, these contemporary climate data are generated from the ANUSPLINE climate model, which interpolates verified North American meteorological station data using thin-plate smoothing splines to generate a continuous climate estimate (McKenney et al., 2006).

Bioclimatic variables considered in this analysis, as possible controls on the spatial distribution of the C mass in the HBL, include: MAAT, MAP, PET, and over the growing season, as well as, growing season length, timing, and GDD above 5 °C (GDD5). A strong correlation between

GDD and PAR (inferred from latitude and cloudiness) has previously been demonstrated

(Charman et al., 2013), owing to the shared growing season length among the two variables, thus only the former is presented here in relation to the spatial distribution of the HBL’s C mass.

Temperature seasonality (difference between summer and winter temperatures) in relation to the net moisture balance (PET/MAP) was used to evaluate the relative continentality of climate

(Loisel et al., 2012). Correlations among individual climatic variables and HBL peat C masses for all peatlands together (n = 364), and among major peat types, including bogs (n = 195), fens

(n = 121) and coastal/permafrost (n = 48) were evaluated using Pearson product moment correlation analysis, at a significance of p ≤ 0.05 (Table 4.9-2). Further, regression analysis of the most ecologically significant climate-peat C relationships is presented.

4.4.2 Peat physical properties

Quantification of the peatland C mass was based upon detailed analysis of 42 peat cores for bulk density, loss-on-ignition (LOI) and C content measurements (Table 4.9-1), using standard

95 methods described in Chapters 2 and 3 (Packalen and Finkelstein, 2014; Packalen et al., 2014).

Of these, 11 are re-interpreted from previously published studies (Bunbury et al., 2012;

Holmquist et al., 2014; Kettles et al., 2000; Kuhry, 1998; Kuhry; O'Reilly et al., 2014; van

Bellen et al., 2011a), while new peat physical characteristics for 31 previously described peat cores (Chapters 2 and 3) are presented (Packalen and Finkelstein, 2014; Packalen et al., 2014).

Local C mass (kg m-2) was calculated for all samples, by summing the product of incremental C density and increment length. Areal C mass was linearly related to peat depth (Figure 4.10-3), and this relationship was used to infer peat C mass for peatland sites with known peat depth (n =

322). The relationship (R2 = 0.81, p < 0.001) is described by equation 4.4.2.1:

C mass = 0.4 * peat depth + 17.1 (4.2.2.1)

Differences in peat physical features among peatland types were compared using a t-test or analysis of variance (ANOVA). For analyses that failed the normality test, ANOVA on ranks was completed, followed by multiple comparisons using Dunn’s method. Significance was determined at p ≤ 0.05.

4.5 Results

4.5.1 Climate and carbon mass spatial relationships across the HBL

4.5.1.1 Climate Space

The physiographic region of the HBL, as defined here, stretches across nearly 10 degrees of latitude, from ~ 50 – 60 °N. Modern average climate (1971 – 2000) is characterized by mean annual temperatures ranging between -7 and 0 °C (mean ± standard deviation: -2.5 ± 1.8 °C),

96 such that warmest conditions occur in the southern HBL, while the coldest conditions follow the

Hudson Bay coastline in a northwesterly direction (Figure 4.10-2a). The growing season for most vegetation types in the HBL begins in mid-May [earlier for Sphagnum spp. as growth may take place as soon as temperatures exceed freezing during the snow-free period (Charman et al.,

2013)], and extends for a period of 120 to 160 days with a mean growing season temperature of

10.8 ± 1.0 °C. Total annual precipitation ranges from 430 to 740 mm with half to two-thirds of the precipitation occurring during the growing season. The southern HBL receives the most precipitation and the northwestern HBL receives the least, with moderate amount of precipitation recorded along the Hudson Bay coast and the northeastern HBL (Figure 4.10-2a). Mean temperature seasonality is 49 ± 1.4 °C and mean PET/MAP ranges between 0.4 and 0.7.

Temperature seasonality and PET/MAP for the peatlands considered here are linearly related (R

= 0.93, p < 0.001). Accordingly, a greater moisture balance (lower PET/MAP ratio) was associated with lower temperature seasonality, reflecting a more maritime influence on the peatland ecosystems developing in the coastal regions of the HBL, compared to the more PET- dominated, more continental climate further inland (Figure 4.10-4).

4.5.1.2 Spatial distribution of peatlands within the HBL climate domain

The distribution of the peat C mass in the HBL for the 364 peatland sites considered here is presented in Figure 4.10-1, and our results reveal that the peat C mass ranges between 20 and

180 kg C m-2. Consideration of all peatlands together revealed significant positive correlations between the total C mass and total precipitation and mean temperature, albeit with wide variation

(Figure 4.10-5a, b; Table 4.9-2). GDD5, growing season length, and PET were also significantly and positively correlated with the distribution of the total peat C mass (Figure 4.10-5, Table 4.9-

2). A gradient of low to high C mass is apparent that increases with peat depth, from the coasts

97 of the Hudson and James Bay inland, reflecting the role of GIA in the timing of peat initiation.

However, no relationship between peat age (n=101) and either latitude (R2 = 0.02, p = 0.134) or longitude (R2 = 0.02, p = 0.214) is apparent, providing evidence for a lack of connection between

C mass and climate that is driven by the timing of peat initiation across the HBL (Packalen et al.,

2014). Further, substantial local variability is also apparent, confirming our hypothesis that other controls may contribute to explaining the distribution of the peat C mass in the HBL.

Both the length and the timing of the start of the growing season (GS) are similar for fens and bogs in the HBL, while the GS for the HBL’s coastal peatlands begins ~ 20 days later and last for 132 days (median), compared with 156 and 157 GS days (medians) for inland fens and bogs, respectively. The median temperature during the GS does not differ among fens and bogs (~ 13

°C) in the HBL; however, the coastal peatlands appear to be developing under cooler GS conditions (10 °C), with reduced seasonality. Moreover, the GGD5 for inland bogs (990) and fens (960) are nearly double the GDD5 for peatlands developing in the coastal (530) HBL.

From a surface wetness standpoint, our analysis of climatic patterns in the HBL reveals that bogs receive significantly (Kruskal-Wallis ANOVA on Ranks at p < 0.05) more GS precipitation

(median: 422 mm) than either fens (406 mm, Dunn’s method: Q = 2.9) or coastal peatlands (357 mm; Q = 10.2). Further, PET is lower for the coastal peatlands (211 mm), than for either bogs

(359 mm; Q = 10.4) or fens (349 mm; Q = 8.4), resulting in a greater net moisture balance

(PET/MAP) for coastal peatlands (0.42) than for bogs (0.55, Q = 8.5) and fens (0.55, Q = 7.7) in the HBL. Increased moisture availability near the coast may enhance permafrost accretion particularly when coupled with significantly lower winter precipitation (median: 62 mm), which provides less surface insulation from the significantly (p < 0.05) colder winter temperatures

(median: -22.4 °C, Q = 7.3 vs. fen and Q = 9.5 vs. bog) near the coast relative to the inland extent of the HBL (~ -19 °C).

4.5.2 Carbon mass variation among peat types

4.5.2.1 Peatland geography

The dataset includes a variety of peatland types and comparative statistics were used to evaluate differences among main peatland classifications used in this study. The mean peat depth of 207 ±

93 cm (n = 364) varied among peatland types across the HBL. Kruskal-Wallis ANOVA on ranks, followed by Dunn’s test revealed that the deepest peat was significantly associated with bog mesoforms (235 cm, n = 195, Q = 5 vs. fens and Q = 7.5 vs. coastal), followed by fen mesoforms (180 cm, n = 121, Q = 3.9 vs. coastal), and then coastal peatlands (100 cm, n = 48).

4.5.2.2 Peat carbon content among peat types

Mean (± standard deviation (s.d.)) peat characteristics for a selection of peat cores examined in detail here are presented in Table 4.9-1. C density for the subset of peatlands examined in this study was 48 (14) g dm-3, and was found to differ significantly among the major peatland classifications considered here. Kruskal-Wallis ANOVA on ranks indicated that bogs had the lowest median C density (45 g dm-3, n = 795 from 20 complete peat cores), while coastal peatlands had the greatest median C density (60 g dm-3, n = 38, from 7 complete peat cores). Fen features patterned with bogs in the interior reaches of the HBL were found to have an intermediate median C density of 49 g dm-3 (n = 603, from 15 complete peat cores). Mean total C mass among all peatland samples considered in the HBL was 92 ± 35 kg m-2, and did not differ significantly among bog (106 kg m-2, n = 20) and fen (98 kg m-2, n = 15) features (Figure 4.10-

6). However, coastal mesoforms had a significantly (t = 4.4 vs. bogs and t = 3.2 vs. fens, p <

0.05) lower median C mass (53 kg m-2, n = 7).

4.6 Discussion

As one of the largest continuous peatland regions globally (Chapter 2), the HBL currently stores a C pool of sufficient magnitude to potentially influence global climate via positive and negative feedbacks associated with changes in peat C dynamics (Packalen et al., 2014). As a result, understanding drivers of peat-C dynamics in the HBL is of high priority. While climate has been shown elsewhere to be a major factor in explaining the spatial patterning of the peat C mass across major northern peatland regions (Beilman et al., 2009; Charman et al., 2013), these relationships have not been well described for the HBL. Here, we consider the relationship between gridded modern bioclimatic parameters and the spatial distribution of the total C mass across the physiographic region of the HBL. Our results confirm that the total C mass is positively correlated with several climate parameters for the full extent of the HBL. Spatial climate patterns approximate a latitudinal gradient which skews toward the northwest around the

Hudson Bay. Under the influence of the thermal properties of the Hudson Bay, somewhat sharp transitions in annual and seasonal temperature trends are apparent just south of the Hudson Bay coast (Figure 4.10-2a). By contrast, precipitation patterns appear to be influenced by proximity to both the Hudson and James Bays in addition to elevation (Figure 4.10-2b). Accordingly, precipitation maxima occur in the southern HBL, and decreases north toward the higher elevation inland extents.

We find that the warmest annual temperatures and most precipitation in the HBL occur in the southern portions, where the peat C mass tends to be greatest (Figure 4.10-1). The southwestern

100 extent of the HBL is the most continental, experiencing the greatest seasonality and largest moisture losses by PET. Although the distribution of the peat C mass is related in part to the timing of peat initiation (Chapter 2), such that earlier initiation dates tend to be located inland and perpendicular to the coasts of the Hudson and James Bays (Packalen et al., 2014), similarly aged peatlands located across the latitudinal gradient in the HBL do not store equivalent C masses. Moreover, our analysis of the relationship between peat age versus latitude and longitude reveal that no correlation exists for peatlands in the HBL, suggesting that the relationships we report here between C mass and important climate variables is not controlled by the timing of peat initiation. Accordingly, our results confirm that similarly aged peat deposits in the southern HBL store more C than the more northern counterparts, providing additional evidence of a climatic control on the distribution of the peat C mass in the HBL.

Conversely, the coldest and driest climatic conditions occur toward the northwest HBL, where the continuous permafrost region lies. Within the coastal region of the HBL, a large range of peat

C masses are apparent (Figure 4.10-1), with some of the largest stores located in the most northwestern reaches of the HBL. Previous evidence indicates that permafrost developed in the

HBL during the late Holocene (Kuhry, 1998; Kuhry, 2008), which suggests the influence of permafrost in either (1) promoting the long-term storage of mid-Holocene age peat C (e.g., in the northwest HBL) or (2) preventing the accrual of newer peat C, such as in the northeast HBL, where land has only recently emerged from the Hudson Bay. Moreover, a study of late Holocene peat depths found that permafrost occurrence significantly reduced vertical peat growth during the late Holocene, and hence rates of C accumulation, during the past 2 ky (Holmquist et al.,

2014).

0 10

101

Several potential climatic controls of ecological importance to the accrual of the peat C mass were considered in this study. A longer growing season, with temperatures above 5 °C (GDD5) and a warmer mean GS temperature may influence both net primary production and/or decomposition, resulting in potential non-linear responses in peatland C dynamics. Given adequate moisture for peatland development, the total C mass increases with temperature,

GDD5, and growing season length in the HBL, and supports the hypothesis that the net C stored in peatlands is influenced by NPP and vertical peat growth. These results are consistent with peat-climate relationships reported for circum-Arctic peatlands. Beilman et al. (2009) found positive correlations between temperature and vertical peat growth for the last 2 ky in the West

Siberian Lowlands, and found that rates of C accumulation were significantly correlated with contemporary MAAT, where maximum rates occurred between -1 and 0 °C. Similarly,

Holmquist et al. (2014) reported positive correlations among vertical peat growth and MAAT,

GDD, precipitation for peatlands location on the Boreal Shield and Hudson Plains of northern

Ontario. Finally, Charman et al. (2013) found a positive correlation between the rate of C accumulation and GDD for the last 1 ky in circum-Arctic peatlands. However, warmer winters, with enhanced snow cover are thought to potentially enhance peat C losses due to respiration

(Jones and Yu, 2010), and this potential effect needs further consideration in the HBL.

Our analysis of the relationship among bioclimatic factors and the distribution of the peat C mass in the non-coastal extent of the HBL reveal significantly positive correlations among the distribution of the peat C mass in the HBL and each of the parameters considered. Although significant correlations among bioclimatic parameters and the distribution of the C mass in the

HBL are apparent, the spread of the data is large, suggesting an important contribution of local- scale ecohydrological processes in the accretion of peat C spatially, a result that was also found

1 10

102 in Chapter 3 to be important temporally in an analysis of Holocene trends (Packalen and

Finkelstein, 2014). As the peatlands of the HBL are patterned with ridge-flark features, our synthesis considers the distribution of C mass among bog- and fen-like features, as a proxy for ecohydrological variation (Figure 4.10-3). Our results reveal that C density was lowest in association with bog peatlands, which are characterized by poorly-decomposable bryophyte remains. These findings are consistent with previous investigations into the role of vegetation type on C mass development, which found that fossil plant composition was significantly related to the rate of C accumulation due to the low C content of Sphagnum remains (Beilman et al.,

2009; Loisel et al., 2014).

Mean total C mass for all HBL peatlands (91 ± 30 kg m-2) did not appear to vary significantly between inland patterned bog and fen features; however both stored significantly more C than coastal fen peatlands. Compared to circum-Arctic peatlands, the C content of HBL peatlands is comparable to those reported for Canadian peatlands, but lower than those reported for Finnish,

Swedish, and Siberian peatlands (Gorham et al., 2012). The bulk density of HBL bogs appear to be within the range of those reported for other northern peatlands. However, fen bulk densities in the HBL appear to be greater than Finnish fens (Turunen et al., 2002), and consistent with western Canadian shrub and open fens (Vitt et al., 2000). The fens considered in the present study include a range of peatland classifications, including open, shrub/treed, intermediate succession, and poor fens. This variability appears to be reflected in the mean and standard deviation of the fen peat characteristics. Considered together, the mean C and bulk density for

HBL peatlands appears to be consistent with the estimates originally reported by Gorham (1991) for North American peatlands and greater than that reported by Turunen et al. (2002) for Finnish peatlands.

2 10

103

Though bogs and fens are common mesoforms throughout the HBL, the bogs considered here tended to occur at higher elevations (median = 97 m) than fens (median = 79.5; p = 0.003). This finding is related in part to the proximity of the peatland mesoform to the coast, such that fens tended to occur ~ 10 km closer to the coast than did bogs (p = 0.01). Fen peatlands also tended to be located further north than bog peatlands (median latitude 52.7 °N versus 51.6 °N, respectively; p = 0.002). While peatlands tend to begin as fens and the younger peatlands are located in proximity to emerging shorelines, the occurrence of fen peatlands further north is not significantly related to peat age in the HBL. Rather the tendency of fens to occur further north than bogs may be partially explained by maritime nature of the climate related to the influence of the Hudson Bay, supported by the relationship between seasonality and PET/MAP. By contrast, bogs were associated with increased GGD5 (median = 962) likely due to the more southerly position in the HBL, which was significantly different from fens (median = 843). GDD5 was also significantly correlated with peat depth (R = 0.4, p < 0.001), supporting the hypothesis of increased NPP in association with warmer annual conditions.

Long-term records indicate that historical peat accumulation was maximal in circum-Arctic peatlands during the early Holocene thermal maximum, when plant production was high due to warm summers and respiration was low due to cold winters (Yu et al., 2010). Such conditions are consistent with more continental climates. In assessing the relative continentality of the HBL, we show that the greatest C mass occurred in the region of the HBL experiencing the largest seasonality. Moisture, as inferred from PET/MAP in the HBL ranged between 0.39 and 0.66, which is similar to the range observed at Degerö Stormyr, Sweden and lower than the more continental peatlands of Vasyugan Bog, Siberia (Eppinga et al., 2009). Previous studies have shown that as the PET/MAP ratio approaches 1, the C storage capacity of a peatland may

3 10

104 decline, and favour a nutrient accumulation-driven mechanism of peat patterning (Eppinga et al.,

2009). Under these conditions, peatland hummock development is favoured, as nutrients such as nitrogen and phosphorus accumulate and stimulate vegetation growth. Alternatively, when the

PET/MAP ratio is much less than 1, such as in the HBL, drainage-dominated peatland development favoring a peat accumulating mechanism results in patterning characterized by ridge-flark features and nutrients accumulating in sparsely vegetated hollows (Eppinga et al.,

2009). Wetter, warmer future climate conditions in the HBL may continue to favour reduced

PET/MAP ratios, supporting enhanced C storage capacity.

4.7 Conclusions and future implications

Although peat depth and age are closely related to timing of land emergence, wide scatter within given time intervals and across latitudes suggests other factors control peat development.

Climate is shown here to be an important factor in explaining landscape-scale spatial patterns in the distribution of the peat C mass in the HBL. The spatial patterns in the distribution of the peat

C mass are best explained by temperature, especially GS temperature and precipitation gradients.

Further, variables related to GS conditions (e.g., PAR, GDD5, growing season length and seasonality) significantly explain distribution of C mass in the HBL. Although all peat classes appear to be sensitive to temperature and precipitation trends in the HBL, fens may be more sensitive to growing season conditions, while bogs may be more sensitive to the net moisture balance. Climatic conditions also appear to explain major trends in the geographic distribution of the peat C mass within a given peat type (bog, fen, coastal) in the HBL. Yet, the widespread bog- fen patterning across HBL suggests that small scale topographic and ecohydrological controls are also critical determinants of C mass development, even when local climatic conditions remain

4 10

105 constant. Our results suggest that ecohydrological factors, together with climate are both important determinants of the distribution of the peat C mass in the HBL.

While conservative climate estimates suggest warmer and wetter conditions within the range of past climate variability, the simultaneous occurrence of warmer/wetter conditions is not documented in the HBL’s Holocene paleoclimatic record. Increased primary productivity is anticipated in the HBL as a consequence of a predicted warmer climate over the next century, lower PET/MAP ratios, and in association with longer growing seasons. However, warmer winters may enhance decomposition and net C losses, as it has elsewhere (Jones and Yu, 2010) and a net reduction in surface moisture conditions may further enhance mass C loss as a consequence of disturbance (e.g., fire). Nevertheless, potential climate feedbacks under future climate scenarios related to augmented CO2 and CH4 emissions, enhanced microbially-driven decomposition and thermokarst formation from thawing permafrost may confound these relationships in the HBL. Today, the largest peat C masses in the HBL appear to occur in association with more continental climates. Yet, analysis of contemporary patterns in PET/MAP reveal that the peatlands in the HBL occur in association with both maritime- and continental- type climates, suggesting that both net moisture balance and precipitation are important controls on peat accumulation. Although maintenance of moist surface conditions supports herbaceous productivity and the accretion of more C dense peat, Sphagnum productivity is greater under more continental climatic conditions. The result is a larger C mass, due in part to the accumulation of decay-resistant vegetation remains. If a more continental climate is coupled with a net moisture balance approaching one, then reduced C storage capacity is anticipated. By contrast, continued peat accumulation may be anticipated under warmer/wetter climatic conditions, if peatlands in the HBL respond in a manner similar to the boreal peatlands located to

5 10

106 the south. Consequently, consideration and elucidation of a moisture balance threshold for sustained C sequestration in the HBL is warranted, in anticipation of future climate scenarios.

4.8 Acknowledgements

Research funding and field support was provided by the Ontario Ministry of Natural Resources and Forestry’s Science and Research Branch and Far North Branch, under the auspices of projects CC-021 and FNIKM 028 to JWM. Additional support for field work and radiocarbon dating was provided by grants from the Natural Sciences and Engineering Research Council of

Canada (NSERC) and the Ontario Ministry of the Environment through the Climate Change and

Multiple Stressor Research Program at Laurentian University to SAF. Additional support in the form of an NSERC Alexander Graham Bell Canada Postgraduate Scholarship, Ontario Graduate

Scholarship, and an Association of Canadian Universities for Northern Studies, Canadian

Northern Studies Trust Scholarship, and field research grants from the Society of Wetland

Scientists and Aboriginal Affairs and Northern Development Canada’s Northern Scientific

Training Program were provided to MSP. We thank A. Dyke for providing access to the

Canadian basal radiocarbon database and P. Kuhry for contributing raw peat core data. Gridded climate data were provided by D. McKenney and P. Papadopol, Canadian Forest Service, Great

Lakes Forestry Centre, Sault Ste. Marie, Canada. We also thank J. Harden and an anonymous reviewer for thoughtful comments on an earlier draft manuscript. All data included in this study, unless publically available/otherwise stated, may be obtained from the corresponding author.

6 10

107

4.9 Tables Table 4.9-1 Peat physical properties of 42 sites in the Hudson Bay Lowland (HBL), Canada. Included here are new details for 31 previously reported cores (Packalen et al., 2014; Packalen and Finkelstein, 2014) and a summary of previously reported peat data, as referenced herein.

Elevation Basal Depth Mean Bulk Mean Mean Carbon Areal Carbon Site (m asl) (cm) Density (g cm-3) [Carbon] Density (g cm-3) Mass (kg m-2) Bog mesoforms (n=795) KR3A* 123 271 0.096 0.486 0.047 126 PL101 45 264 0.084 0.493 0.039 102 KJ101 68 297 0.087 0.499 0.044 130 Kinosheo L*1 68 265 0.074 0.475 0.035 93 KJ4-3 66 131 0.108 0.500 0.055 72 KJ3-3 65 176 0.093 0.490 0.046 81 D001 92 231 0.077 0.518 0.040 102 D003 92 223 0.104 0.531 0.055 80 D200 118 240 0.099 0.526 0.052 84 D206 142 257 0.073 0.490 0.036 105 VM4-5 103 286 0.076 0.497 0.038 109 VM4-32 103 304 0.087 0.462 0.040 123 VM4-1 105 311 0.081 0.503 0.041 126 VM1-3 86 210 0.096 0.485 0.046 97 ML201 143 235 0.096 0.481 0.045 106 HL02 85 230 0.079 0.506 0.040 92 JBL76 150 330 0.080 0.470 0.038 129 JBL46 108 176 0.086 0.480 0.041 72 LLC7 305 250 -- -- 0.043 106 Herchmer Palsa Bog*5 106 167 0.235 0.378 0.089 192 Mean (n=20) 109 243 0.095 0.488 0.045 106 s.d. 54 52 0.035 0.032 0.012 27 Fen mesoforms (n=603) KJ2-3 65 246 0.081 0.492 0.040 99 D002 91 219 0.099 0.527 0.052 83 D201 104 201 0.073 0.489 0.036 106 D202 105 215 0.077 0.471 0.036 99 D204 143 129 0.076 0.493 0.037 64 VM3-33 102 242 0.094 0.459 0.043 105 VM3-2 102 262 0.095 0.509 0.048 127 VM3-5 102 232 0.121 0.496 0.060 139 VM2-1 86 211 0.083 0.487 0.040 85 VM2-3 86 207 0.087 0.496 0.043 89 VM2-5 86 180 0.092 0.510 0.047 85 HL03 99 269 0.088 0.517 0.045 122

HL04A(West) 89 93 0.167 0.471 0.079 73 HL04B(East) 97 105 0.140 0.494 0.069 72 Herchmer Fen*4 106 203 0.120 0.429 0.051 119 Mean (n=15) 98 201 0.100 0.489 0.049 98 s.d. 17 54 0.027 0.024 0.012 22 Coastal mesoforms (n=38) JBL5 peat plateau6 143 141 0.135 0.465 0.063 78 McClintockFen*4 81 148 0.097 0.472 0.046 68 McClintock plateau*5 79 163 0.234 0.376 0.088 143 PB3-5 10 44 0.104 0.481 0.050 22 PB-SF-5 11 24 0.246 0.377 0.075 18 PB2-5 10 45 0.108 0.480 0.052 23 PB1-5 9 42 0.102 0.483 0.049 21 Mean (n=7) 49 87 0.147 0.448 0.060 53 s.d. 53 61 0.065 0.049 0.016 47 HBL peatlands Mean (n=42) 94 196 0.104 0.471 0.048 92 s.d. 48 77 0.042 0.076 0.014 35 Data obtained from previously published work: 1Kettles et al., 2000; 2Bunbury et al., 2012; 3O'Reilly et al., 2014; 4Kuhry et al., 5 6 7 1998; Kuhry, 2008; Holmquist et al., 2014; van Bellen et al., 2011a. *Carbon determined from LOI data at a level of 50% C.

7 10

108

Table 4.9-2 Peatland carbon mass distribution in the Hudson Bay Lowland, Canada. Correlation matrix (correlation coefficient, p-value and sample size) of the main annual (a) and growing season (b) climatic controls on the spatial variation of total carbon mass (kg m-2) in the Hudson Bay Lowlands, Canada (HBL). Data are presented for the complete peatland complex and by broad peatland classification, including bogs, fens, and the coastal peatland region. Bold font indicates significant correlation at p<0.05.

Temperature Potential Mean Annual Mean Annual Latitude Elevation Seasonality Evapo- Temperature Precipitation PET/MAP (°) (m, asl) (Annual transpiration (MAAT, °C) (MAP, mm) Range) (PET) All peatlands -0.428 0.459 0.440 0.486 0.437 0.471 0.312 1.30E-17 2.48E-20 1.06E-18 5.06E-23 2.24E-18 1.60E-21 1.12E-09 364 364 364 364 364 364 364 Bog peatlands -0.285 0.302 0.276 0.359 0.201 0.301 0.085 5.52E-05 1.78E-05 9.48E-05 2.60E-07 4.84E-03 1.96E-05 2.34E-01 195 195 195 195 195 195 195 Coastal peatlands 0.320 0.629 -0.127 -0.256 0.506 0.081 0.287 2.64E-02 1.65E-06 3.88E-01 7.92E-02 2.45E-04 5.84E-01 4.82E-02 48 48 48 48 48 48 48 Fen peatlands -0.366 0.366 0.370 0.468 0.311 0.395 0.180 3.69E-05 3.69E-05 3.02E-05 6.41E-08 5.14E-04 7.54E-06 4.87E-02 121 121 121 121 121 121 121

GS Mean GS Mean GS Length Timing of GS Temperature Precipitation GGD 5°C GS PET (days) GS initiation PET/GSP (GST, °C) (GSP, mm)

All peatlands 0.496 0.459 0.488 0.432 -0.469 0.470 0.394 5.61E-24 2.15E-20 3.33E-23 5.75E-18 2.83E-21 2.21E-21 6.11E-15 364 364 364 364 364 364 364 Bog peatlands 0.317 0.332 0.312 0.245 -0.270 0.298 0.184 6.31E-06 2.08E-06 9.04E-06 5.44E-04 1.37E-04 2.32E-05 9.97E-03 195 195 195 195 195 195 195 Coastal peatlands 0.357 -0.288 0.237 -0.057 -0.206 0.082 0.298 1.28E-02 4.71E-02 1.04E-01 7.01E-01 1.59E-01 5.57E-01 3.99E-02 48 48 48 48 48 48 48 Fen peatlands 0.406 0.427 0.409 0.361 -0.374 0.390 0.290 3.74E-06 1.03E-06 3.12E-06 4.75E-05 2.41E-05 9.62E-06 1.23E-03 121 121 121 121 121 121 121

8 10

109

4.10 Figures

Figure 4.10-1Physical features, study sites and carbon mass the Hudson Bay Lowland, Canada. The region includes the coastal Hudson Bay Lowland (HBL) and inland HBL and James Bay Lowland (JBL) ecoregions (www.cec.org/naatlas/). Contours represent modern isostatic uplift rates (Peltier, 2004), while shading represents modern surface elevation (0 – 400 m, above sea level) generated using 1:250K Canadian digital elevation data (www.geobase.ca). Peat depth locations are scaled according to the carbon (C)-mass at the site. Yellow triangles indicate site locations with detailed peat physical features.

9 10

110

Figure 4.10-2 Contemporary climate domain for peatlands of the Hudson Bay Lowland, Canada. Mean annual air temperature (MAAT) and mean annual precipitation (MAP) for the period 1971-2000 are presented.

0 11

111

Figure 4.10-3 Relationship between peat depth and carbon (C) mass for well described peatlands in the Hudson Bay Lowland, Canada.

1 11

112

Figure 4.10-4 Peat class distribution relative to peatland continentality in the Hudson Bay Lowland, Canada.

2 11

113

Figure 4.10-5 Peat carbon mass relative to bioclimate in the Hudson Bay Lowland, Canada. Data are presented for unique peatland mesoforms considered together as well as for bogs, fens, and coastal peatland classes. Carbon mass (n = 364) vs. mean annual air temperature (MAAT), mean annual precipitation (MAP), growing season (GS) temperature and precipitation, growing degree days above 5 °C (GDD5) and GS potential evapotranspiration (PET) are presented. All relationships are significant at p < 0.05 (see Table 4.9-2 for correlation coefficients).

3 11

114

Figure 4.10-6 Total carbon-mass (kg m-2) stored in bogs, fens and coastal peatland mesoforms, in the Hudson Bay Lowland, Canada. Median, 25th and 75th percentiles, and outliers of peat C mass are presented for all available peat cores (n = 42) with detailed peat C analyses. Differences among C mass stores in bog, fen, and coastal classes indicated by letters at a significance of p < 0.05.

4 11

Chapter 5 Conclusions and Future Research Directions 5.1 Summary

Based upon an understanding of controls on peatland trace gas (CO2 and CH4) flux combined with evidence of widespread peatland initiation in the early Holocene, compelling evidence was advanced that peatlands can contribute to variations in atmospheric CO2 and CH4 concentrations

(MacDonald et al., 2006). However, questions associated with the precise timing of peatland initiation versus the timing of atmospheric CO2 and CH4 changes have stimulated controversy concerning the connection between peatland ecosystem dynamics and the long-term trajectories of atmospheric trace gases (Reyes and Cooke, 2011). Moreover, a lack of evidence for a connection between peatland development and late Holocene trace gas dynamics, stemming from few records from late emerging peatlands, such as the HBL, and few data on the lateral expansion of peatlands through the late Holocene has perpetuated this debate. Climate-driven changes in peatland C dynamics are suggested and long-term, externally-forced climatic changes are potentially documented in the peat archive, such as cyclical climate variations associated with millennial- to centennial-scale solar forcing (Viau and Gajewski, 2009). Synchronous evidence of climate variation in multiple records (e.g., peatland, marine sediment, and ice core) increases confidence that perceived climate variations inferred from peat archives are indeed tracking allogenic processes (external changes).However strong mechanistic evidence of peatland induced climate perturbations is lacking, though it has been hypothesized that peatland expansion and pattern development may be viable mechanisms to substantiate the peatland- driven climate variability hypothesis (Korhola et al., 2010).

115 116

Peatlands may also record the effects of autogenic factors; that is, internal changes and feedbacks brought about during peatland development. Consequently, the peat archive can track changes in ecohydrology against the backdrop of relatively stable climate conditions (Charman et al., 2009).

However, peatland vegetation and hydrology may respond to seasonal temperature fluctuations through internal feedback mechanisms, and such internal feedbacks may contribute to longer term departures from stable climate conditions or non-linear responses to local ecohydrological variability. Alternatively, changes recorded in the peat archive may be the consequence of externally forced changes in stable climate conditions. Identifying drivers of C dynamics in the

HBL peat record necessitates distinguishing between autogenic factors related to peatland development, trends in orbital forcing, and regional climatic variation influenced by the HBL’s proximity to a large marine system with seasonal sea ice dynamics. Thus to contribute to the debate, disentangling the role of autogenic and allogenic factors in peatland pattern development is essential. Short-term empirical studies examining patterns in nutrients, hydrology and hydrochemistry can inform our understanding of short-term ecological processes (Eppinga et al.,

2009). However, predicting ecological responses to environmental change benefits from long- term systematic observation of ecohydroclimatic dynamics, which can be achieved through a combination of experimentation, mechanistic and ecological modeling, and examination of the paleoenvironmental record (Jackson et al., 2009).

5.1.1 Peat initiation and carbon storage

In Chapter 2, the first estimate of the total C sink of HBL peatlands together with a synoptic assessment of peat initiation frequency is presented. Based upon new detailed records of peat C density, the HBL is estimated to store ~ 30 Pg C (Packalen et al., 2014), equivalent to ~ 20% of the North American (Gorham et al., 2012) or ~ 6% of the northern peatland C pool (Yu, 2011).

6 11

117

Study results also confirm a tight coupling between peat initiation frequency and land emergence from the post glacial Tyrrell Sea, driven by the rate of glacial isostatic rebound. While climatic conditions in the HBL remained suitable for peat initiation through the period of record, the most intense period of peat initiation in the HBL occurred during the mid-Holocene, when the climate was colder and received less precipitation than today and insolation was decreasing relative to the early Holocene. Accordingly, the highest rate of peat initiation in the HBL occurred prior to the late-Holocene rise in atmospheric CH4concentration, providing additional evidence of a peat contribution to atmospheric trace gas dynamics. The late Holocene CH4 contribution from the

-1 HBL potentially ranged from 1 – 7 Tg CH4 y (Packalen et al., 2014), and represents a small but

-1 important potential CH4 emission, when compared to the 21 – 43 Tg CH4 yr released by modern northern wetlands (Mikaloff Fletcher et al., 2004). Improved quantification of modern peatland C stores and regional controls on Holocene peatland C dynamics in the HBL will reduce the uncertainty surrounding the sensitivity of northern peatland C pools under future climate scenarios.

5.1.2 Holocene carbon dynamics

Chapter 3 examines temporal C dynamics in the HBL at the landscape scale and partitions net C uptake, release, and balance since peat inception. Rather than climate variation, C accumulation in the HBL appears to be more strongly related to peatland succession dynamics (Packalen and

Finkelstein, 2014). Accordingly, CAR is most related to the intensity of peat initiation, the latter of which determines the proportion of young peatlands on the rapidly emerging HBL landscape that characteristically have high C accumulation rates. However, few temporally resolved paleoclimate records are available in the HBL, thus the potential link between C accumulation and climate warrants further investigation. Our data show that while the HBL has been a

7 11

118 persistent C sink for millennia, more than two-thirds of the total C mass accrued during the late

Holocene, when CAR remained below the long-term mean. Further, model evidence presented here suggest that most of the potential C lost from the HBL occurred during the late Holocene, likely owing to decay of previously deposited peat. Given that sustained minerotrophic to weakly ombrotrophic peat patterning typifies the HBL landscape, the persistent C release during the late

Holocene may provide an additional line of evidence of natural terrestrial C contributions from the HBL to the late Holocene atmosphere, particularly in the form of CH4 (Packalen et al., 2014;

Packalen and Finkelstein, 2014).

5.1.3 Climatic controls of the distribution of the carbon mass

Recognizing a limited temporal relationship between climate and peat C dynamics in the HBL

Chapter 4 (Packalen et al., in revision), examines spatially explicate climate factors in relation to the distribution of the total C mass. Although peat depth and age are closely related to timing of land emergence, wide scatter within given time intervals suggests other factors control peat development. Climate is shown here to be an important factor in explaining major spatial patterns in the distribution of the peat C mass in the HBL, which are best explained by temperature, especially growing season temperature, and then precipitation gradients. Climatic conditions also appear to explain major trends in the geographic distribution of the peat C mass within a given peat type (bog, fen, coastal) in the HBL. Yet, the widespread bog-fen patterning across HBL suggests that small scale topographic and ecohydrological controls are also critical determinants of C mass development, even when local climatic conditions remain constant.

8 11

119

5.2 Sources of Uncertainty and Future Research Directions

5.2.1 HBL paleohydroclimate and carbon dynamics

Today, the largest peat C masses in the HBL appear to occur in association with more continental climates. Yet analysis of contemporary patterns in PET/MAP reveal that the peatlands in the HBL occur in association with both maritime- and continental-type climates, suggesting that both net moisture balance and precipitation are important controls on peat accumulation. Although maintenance of moist surface conditions supports herbaceous productivity and the accretion of more C dense peat, Sphagnum productivity is greater under more continental climatic conditions. The result is a larger C mass, due in part to the accumulation of decay-resistant vegetation remains (Turetsky et al., 2008). Few paleoclimatic records are available for HBL peatlands to fully explore the relationship between long-term spatio-temporal climatic variation and peat C dynamics. Consequently, it is difficult to

9 11

120 disentangle the relative roles of climate and local autogenic processes in peat C dynamics in the

HBL.

5.2.2 Carbon dynamics in permafrost peatlands and peatland pools

The HBL is patterned with peat-pool complexes, some of which are or may have been influenced by permafrost. Currently, few data are available to accurately access the spatial extend of either small pools and permafrost features on the HBL landscape. Further, few records are available that detail the C dynamics of these features in the HBL. Consequently, the C store estimate presented here does not account for these uncertainties. Further, potential climate feedbacks under future climate scenarios related to augmented CO2 and CH4 emissions from pools and thermokarst ponds, and enhanced microbially-driven decomposition associated with pools and thawing permafrost features may confound our understanding of C dynamics in the HBL and its associated C sink potential.

5.2.3 Age-depth modeling and fen decay modeling

Temporally, C dynamics in the HBL were assessed here using a limited set of vertically dated peat records. Although the data presented here are in agreement with findings reported from circum-Arctic peatlands, additional well-dated peat records, together with higher resolution dating may help to disentangle the role of autogenic and allogenic controls on peat C dynamics.

Further, the decay models used here are conservative estimates of long-term peat decay, assume constant decay through time as a peatland proceeds through typical successional patterns.

Capturing more ecologically meaningful decay rates through time may improve temporal resolution of C sink-source dynamics in peatlands in the HBL and elsewhere.

0 12

121

5.2.4 Peat carbon dynamics in post-marine environments

Peat initiation in the HBL occurred over marine sediments, which may have limited early trace gas emissions (e.g., CH4), due to peat-sediment interactions with marine sulfur species. The estimates presented here make this assumption and thus provide an estimate for late Holocene peatlands that have developed sufficient peat depth to reasonably disconnect the potential sulphur-mediated suppression of peat CH4 emissions. However, few data are available to address this potential mechanism for either the HBL or for circum-Arctic peatlands that initiated in former marine environments. This mechanism may have important implications for both the assessment of long-term peat-climate-C dynamics, as well as, anticipated trace gas emissions from peatlands subject to the influence of sea level rise.

5.2.5 Peatland resilience and vulnerability to climatic change

Peatlands are anticipated to be subject to enhanced natural and anthropogenic disturbance as a consequence of a changing climate, intensified land-use and expanding resources management.

However, the vulnerability of peatlands to these changes remains unclear. While peatlands appear to control the local hydrology as it relates to supporting vegetation productivity, and historical peatland resilience in the context of changing climate is documented, more rapid, abrupt or widespread interference with peatland ecosystem functioning may compromise the future resilience of peatlands. The role of hydrology, fire, and depth of peat vulnerability all remain outstanding priorities for future research and approaches to ascertaining peatland vulnerability to climate and/or anthropogenic change may include identification of critical threshold conditions or probability of change under various climate and land-use scenarios.

1 12

References

Behl RJ. (2011) Glacial demise and methane's rise. Proceedings of the National Academy of Sciences of the United States of America 108: 5925-5926. Beilman DW, MacDonald GM, Smith LC, et al. (2009) Carbon accumulation in peatlands of West Siberia over the last 2000 years. Global Biogeochemical Cycles 23: 12. Belyea LR and Baird AJ. (2006) Beyond "The limits to peat bog growth'': Cross-scale feedback in peatland development. Ecological Monographs 76: 299-322. Berger A and Loutre MF. (1991) Insolation values for the climate of the last 10 million years. Quaternary Science Reviews 10: 297-317. Blaauw M. (2010) Methods and code for 'classical' age-modelling of radiocarbon sequences. Quaternary Geochronology 5: 512-518. Brook EJ, Harder S, Severinghaus J, et al. (2000) On the origin and timing of rapid changes in atmospheric methane during the last glacial period. Global Biogeochemical Cycles 14: 559-572. Bubier J, Costello A, Moore TR, et al. (1993a) Microtopography and methane flux in boreal peatlands, northern Ontario, Canada. Canadian Journal of Botany-Revue Canadienne De Botanique 71: 1056-1063. Bubier JL. (1995) The relationship of vegetation to methane emission and hydrochemical gradients in northern peatlands. Journal of Ecology 83: 403-420. Bubier JL, Moore TR and Roulet NT. (1993b) Methane emissions from wetlands in the midboreal region of northern Ontario, Canada. Ecology 74: 2240-2254. Bunbury J, Finkelstein SA and Bollmann J. (2012) Holocene hydro-climatic change and effects on carbon accumulation inferred from a peat bog in the Attawapiskat River watershed, Hudson Bay Lowlands, Canada. Quaternary Research 78: 275-284. Chambers FM, Beilman DW and Yu Z. (2010/11) Methods for determining peat humification and for quantifying peat bulk density, organic matter and carbon content for palaeostudeis of climate and peatland carbon dynamics. Mires and Peat 7: 1-10. Charman DJ, Barber KE, Blaauw M, et al. (2009) Climate drivers for peatland palaeoclimate records. Quaternary Science Reviews 28: 1811-1819. Charman DJ, Beilman DW, Blaauw M, et al. (2013) Climate-related changes in peatland carbon accumulation during the last millennium. Biogeosciences 10: 929-944. Christensen TR, Prentice IC, Kaplan J, et al. (1996) Methane flux from northern wetlands and tundra - An ecosystem source modelling approach. Tellus Series B-Chemical and Physical Meteorology 48: 652-661. Clymo RS, Turunen J and Tolonen K. (1998) Carbon accumulation in peatland. Oikos 81: 368- 388. Dise NB. (2009) Peatland Response to Global Change. Science 326: 810-811. Dyke AS, Moore A and Robertson L. (2003) Deglaciation of North America: Geologic Survey of Canada, Open File 1574. Eppinga MB, de Ruiter PC, Wassen MJ, et al. (2009) Nutrients and Hydrology Indicate the Driving Mechanisms of Peatland Surface Patterning. American Naturalist 173: 803-818. Frolking S and Roulet NT. (2007) Holocene radiative forcing impact of northern peatland carbon accumulation and methane emissions. Global Change Biology 13: 1079-1088.

122 123

Frolking S, Talbot J, Jones MC, et al. (2011) Peatlands in the Earth's 21st century climate system. Environmental Reviews 19: 371-396. Gajewski K, Viau A, Sawada M, et al. (2001) Sphagnum peatland distribution in North America and Eurasia during the past 21,000 years. Global Biogeochemical Cycles 15: 297-310. Glaser PH, Hansen BCS, Siegel DI, et al. (2004a) Rates, pathways and drivers for peatland development in the Hudson Bay Lowlands, northern Ontario, Canada. Journal of Ecology 92: 1036-1053. Glaser PH, Siegel DI, Reeve AS, et al. (2004b) Tectonic drivers for vegetation patterning and landscape evolution in the Albany River region of the Hudson Bay Lowlands. Journal of Ecology 92: 1054-U1052. Gorham E. (1991) Northern peatlands - Role in the carbon-cycle and probably responses to climatic warming. Ecological Applications 1: 182-195. Gorham E, Lehman C, Dyke A, et al. (2012) Long-term carbon sequestration in North American peatlands. Quaternary Science Reviews 58: 77-82. Gorham E, Lehman C, Dyke A, et al. (2007) Temporal and spatial aspects of peatland initiation following deglaciation in North America. Quaternary Science Reviews 26: 300-311. Gough WA, Cornwell AR and Tsuji LJS. (2004) Trends in Seasonal Sea Ice Duration in Southwestern Hudson Bay. Arctic 57: 299-305. Holmquist J, MacDonald G and Gallego-Sala A. (2014) Peatland Initiation, Carbon Accumulation, and 2 ka Depth in the James Bay Lowland and Adjacent Regions. Arctic, Antarctic, and Alpine Research 46: 19-39. Holmquist J and MacDonald GM. (2014) Peatland succession and long-term apparent carbon accumulation in central and northern Ontario, Canada. The Holocene 24: 1075-1089. Hugelius G, Strauss J, Zubrzycki S, et al. (2014) Estimated stocks of circumpolar permafrost carbon with quantified uncertainty ranges and identified data gaps. Biogeoscicences 11: 6573-6593. Jackson ST, Betancourt JL, Booth RK, et al. (2009) Ecology and the ratchet of events: Climate variability, niche dimensions, and species distributions. Proceedings of the National Academy of Sciences of the United States of America 106: 19685-19692. Jones MC and Yu ZC. (2010) Rapid deglacial and early Holocene expansion of peatlands in Alaska. Proceedings of the National Academy of Sciences of the United States of America 107: 7347-7352. Kettles IM, Garneau M and Jette H. (2000) Macrofossil, pollen, and geochemical records of peatlands in the Kinosheo Lake and Detour Lake areas, northern Ontario: Geological Survey of Canada, Bulletin 545. Kleinen T, Brovkin V and Schuldt RJ. (2012) A dynamic model of wetland extent and peat accumulation: results for the Holocene. Biogeosciences 9: 235-248. Klinger LF, Zimmerman PR, Greenberg JP, et al. (1994) Carbon trace gas fluxes along a successional gradient in the Hudson Bay Lowland. Journal of Geophysical Research- Atmospheres 99: 1469-1494. Koebsch F, Glatzel S and Jurasinski G. (2013) Vegetation controls methane emissions in a coastal brackish fen. Wetlands Ecology and Management 21: 323-337. Korhola A, Ruppel M, Seppa H, et al. (2010) The importance of northern peatland expansion to the late-Holocene rise of atmospheric methane. Quaternary Science Reviews 29: 611-617. Kuhry P. (1998) Late Holocene permafrost dynamics in two subarctic peatlands of the Hudson Bay Lowlands (Manitoba, Canada). Eurasian Soil Science 31: 529-534.

3 12

124

Kuhry P. (2008) Palsa and peat plateau development in the Hudson Bay Lowlands, Canada: timing, pathways and causes. Boreas 37: 316-327. Kuhry P, Dorrepaal E, Hugelius G, et al. (2010) Potential Remobilization of Belowground Permafrost Carbon under Future Global Warming. Permafrost and Periglacial Processes 21: 208-214. Lamarre A, Garneau M and Asnong H. (2012) Holocene paleohydrological reconstruction and carbon accumulation of a permafrost peatland using testate amoeba and macrofossil analyses, Kuujjuarapik, subarctic Quebec, Canada. Review of Palaeobotany and Palynology 186: 131-141. Levine JG, Wolff EW, Jones AE, et al. (2011) Reconciling the changes in atmospheric methane sources and sinks between the Last Glacial Maximum and the pre-industrial era. Geophysical Research Letters 38: 6. Liikanen A, Silvennoinen H, Karvo A, et al. (2009) Methane and nitrous oxide fluxes in two coastal wetlands in the northeastern Gulf of Bothnia, Baltic Sea. Boreal Environment Research 14: 351-368. Loisel, Gallego-Sala AV and Yu Z. (2012) Global-scale pattern of peatland Sphagnum growth driven by photosynthetically active radiation and growing season length. Biogeosciences 9: 2737-2746. Loisel, Yu Z, Beilman DW, et al. (2014) A database and synthesis of northern peatland soil properties and Holocene carbon and nitrogen accumulation. The Holocene. Loisel J and Garneau M. (2010) Late Holocene paleoecohydrology and carbon accumulation estimates from two boreal peat bogs in eastern Canada: Potential and limits of multi- proxy archives. Palaeogeography Palaeoclimatology Palaeoecology 291: 493-533. MacDonald GM, Beilman DW, Kremenetski KV, et al. (2006) Rapid early development of circumarctic peatlands and atmospheric CH4 and CO2 variations. Science 314: 285-288. Martini IP. (2006) Chapter 3 The cold-climate peatlands of the Hudson Bay Lowland, Canada: brief overview of recent work. In: I.P. Martini AMC and Chesworth W (eds) Developments in Earth Surface Processes. Elsevier, 53-84. McAndrews JH and Campbell ID. (1993) 6 ka mean July temperature in eastern Canada from Bartlein and Webb's (1985) pollen transfer functions: comments and illustrations. In: Telka A (ed) Proxy climate data and models of the six thousand years before present time interval: the Canadian perspective. The Royal Society of Canada, 22-25. McAndrews JH, Riley JL and Davis AM. (1982) Vegetation History of the Hudson Bay lowland: a postglacial pollen diagram from the Sutton Ridge. Le Naturaliste Canadien 109: 597- 608. McGuire AD, Anderson LG, Christensen TR, et al. (2009) Sensitivity of the carbon cycle in the Arctic to climate change. Ecological Monographs 79: 523-555. McKenney DW, Hutchinson MF, Papadopol P, et al. (2011) Customized spatial climate models for North America. Bulletin of the American Meteorological Society 92: 1611-1622. McKenney DW, Pedlar JH, Papadopol P, et al. (2006) The development of 1901-2000 historical monthly climate models for Canada and the United States. Agricultural and Forest Meteorology 138: 69-81. Mikaloff Fletcher SE, Tans PP, Bruhwiler LM, et al. (2004) CH4 sources estimated from atmospheric observations of CH4 and its C-13/C-12 isotopic ratios: 1. Inverse modeling of source processes. Global Biogeochemical Cycles 18: 17. Monnin E, Steig EJ, Siegenthaler U, et al. (2004) Evidence for substantial accumulation rate variability in Antarctica during the Holocene, through synchronization of CO2 in the

4 12

125

Taylor Dome, Dome C and DML ice cores. Earth and Planetary Science Letters 224: 45- 54. Moore TR and Dalva M. (1993) The influence of temperature and water-table position on carbon dioxide and methane emissions from laboratory columns of peatland soils. Journal of Soil Science 44: 651-664. Moore TR, Heyes A and Roulet NT. (1994) Methane emissions from wetlands, sourthern Hudson Bay Lowland. Journal of Geophysical Research-Atmospheres 99: 1455-1467. North American environmental atlas. Available at: http://www.cec.org/naatlas/ (2009). Ontario land cover database, second edition, Peterborough, Canada: The Queen's Printer for Ontario (2000). O'Reilly BC, Finkelstein SA and Bunbury J. (2014) Pollen-derived paleovegetation reconstruction and long-term carbon accumulation at a fen site in the Attawapiskat River watershed, Hudson Bay Lowlands, Canada. Arctic, Antarctic and Alpine Research 46: 6- 18. Olefeldt D, Turetsky MR, Crill PM, et al. (2013) Environmental and physical controls on northern terrestrial methane emissions across permafrost zones. Global Change Biology 19: 589-603. Packalen MS and Finkelstein SA. (2014) Quantifying Holocene variability in carbon uptake and release since peat initiation in the Hudson Bay Lowlands, Canada. The Holocene 24: 1063-1074. Packalen MS, Finkelstein SA and McLaughlin JW. (2014) Carbon storage and potential methane production in the Hudson Bay Lowlands since mid-Holocene peat initiation. Nat Commun 5. Page SE, Rieley JO and Banks CJ. (2011) Global and regional importance of the tropical peatland carbon pool. Global Change Biology 17: 798-818. Pelletier L, Moore TR, Roulet NT, et al. (2007) Methane fluxes from three peatlands in the La Grande Riviere watershed, James Bay lowland, Canada. Journal of Geophysical Research-Biogeosciences 112: 12. Peltier WR. (2002) Global glacial isostatic adjustment: palaeogeodetic and space-geodetic tests of the ICE-4G (VM2) model. Journal of Quaternary Science 17: 491-510. Peltier WR. (2004) Global glacial isostasy and the surface of the ice-age earth: The ICE-5G (VM2) model and GRACE. Annual Review of Earth and Planetary Sciences 32: 111-149. Pickett-Heaps CA, Jacob DJ, Wecht KJ, et al. (2011) Magnitude and seasonality of wetland methane emissions from the Hudson Bay Lowlands (Canada). Atmospheric Chemistry and Physics 11: 3773-3779. Reimer PJ, Baillie MGL, Bard E, et al. (2009) INTCAL09 and MARINE09 radiocarbon age calibration curves, 0-50,000 years cal BP. Radiocarbon 51: 1111-1150. Renssen H, Seppa H, Heiri O, et al. (2009) The spatial and temporal complexity of the Holocene thermal maximum. Nature Geoscience 2: 411-414. Reyes AV and Cooke CA. (2011) Northern peatland initiation lagged abrupt increases in deglacial atmospheric CH4. Proceedings of the National Academy of Sciences of the United States of America 108: 4748-4753. Riley JL. (2011) Wetlands of the Ontario Hudson Bay Lowland: A regional overview: Nature Conservancy of Canada. Roulet NT, Jano A, Kelly CA, et al. (1994) Role of the Hudson bay Lowland as a source of atmospheric methane. Journal of Geophysical Research-Atmospheres 99: 1439-1454.

5 12

126

Rouse WR. (1998) A water balance model for a subarctic sedge fen and its application to climatic change. Climatic Change 38: 207-234. Ruddiman WF. (2003) The anthropogenic greenhouse era began thousands of years ago. Climatic Change 61: 261-293. Ruddiman WF, Guo ZT, Zhou X, et al. (2008) Early rice farming and anomalous methane trends. Quaternary Science Reviews 27: 1291-1295. Ruddiman WF, Kutzbach JE and Vavrus SJ. (2011) Can natural or anthropogenic explanations of late-Holocene CO2 and CH4 increases be falsified? Holocene 21: 865-879. Schuldt RJ, Brovkin V, Kleinen T, et al. (2013) Modelling Holocene carbon accumulation and methane emissions of boreal wetlands - an Earth system model approach. Biogeosciences 10: 1659-1674. Sheng YW, Smith LC, MacDonald GM, et al. (2004) A high-resolution GIS-based inventory of the west Siberian peat carbon pool. Global Biogeochemical Cycles 18: 14. Sjörs H. (1959) Bogs and Fens in the Hudson Bay Lowlands. Arctic 12: 2-19. Smith LC, MacDonald GM, Velichko AA, et al. (2004) Siberian peatlands a net carbon sink and global methane source since the early Holocene. Science 303: 353-356. Sonnentag O, Chen JM, Roulet NT, et al. (2008) Spatially explicit simulation of peatland hydrology and carbon dioxide exchange: Influence of mesoscale topography. Journal of Geophysical Research-Biogeosciences 113: 16. Swanson DK, Lacelle B and Tarnocai C. (2000) Temperature and the boreal-subarctic maximum in soil organic carbon. Geographie Physique Et Quaternaire 54: 157-167. Tarnocai C, Canadell JG, Schuur EAG, et al. (2009) Soil organic carbon pools in the northern circumpolar permafrost region. Global Biogeochemical Cycles 23: 11. Tarnocai C, Kuhry P, Broll G, et al. (2012) Peatlands and their carbon dynamics: Comment on “Peatlands and their role in the global carbon cycle”. Eos, Transactions American Geophysical Union 93: 31-31. Tolonen K and Turunen J. (1996) Accumulation rates of carbon in mires in Finland and implications for climate change. The Holocene 6: 171-178. Tuittila ES, Juutinen S, Frolking S, et al. (2013) Wetland chronosequence as a model of peatland development: Vegetation succession, peat and carbon accumulation. Holocene 23: 25-35. Turetsky MR. (2004) Decomposition and organic matter quality in continental peatlands: The ghost of permafrost past. Ecosystems 7: 740-750. Turetsky MR, Crow SE, Evans RJ, et al. (2008) Trade-offs in resource allocation among moss species control decomposition in boreal peatlands. Journal of Ecology 96: 1297-1305. Turetsky MR, Donahue WF and Benscoter BW. (2011) Experimental drying intensifies burning and carbon losses in a northern peatland. Nature Communications 2: 5. Turetsky MR, Mack MC, Harden JW, et al. (2005) Spatial Patterning of soil carbon storage across boreal landscapes, New York: Springer. Turunen J, Tahvanainen T, Tolonen K, et al. (2001) Carbon accumulation in West Siberian mires, Russia. Global Biogeochemical Cycles 15: 285-296. Turunen J, Tomppo E, Tolonen K, et al. (2002) Estimating carbon accumulation rates of undrained mires in Finland - application to boreal and subarctic regions. Holocene 12: 69-80. van Bellen S, Dallaire PL, Garneau M, et al. (2011a) Quantifying spatial and temporal Holocene carbon accumulation in ombrotrophic peatlands of the Eastmain region, Quebec, Canada. Global Biogeochemical Cycles 25: 15.

6 12

127 van Bellen S, Garneau M, Ali AA, et al. (2013) Poor fen succession over ombrotrophic peat related to late Holocene increased surface wetness in subarctic Quebec, Canada. Journal of Quaternary Science 28: 748-760. van Bellen S, Garneau M and Booth RK. (2011b) Holocene carbon accumulation rates from three ombrotrophic peatlands in boreal Quebec, Canada: Impact of climate-driven ecohydrological change. Holocene 21: 1217-1231. Viau AE and Gajewski K. (2009) Reconstructing Millennial-Scale, Regional Paleoclimates of Boreal Canada during the Holocene. Journal of Climate 22: 316-330. Vitt DH, Halsey LA, Bauer IE, et al. (2000) Spatial and temporal trends in carbon storage of peatlands of continental western Canada through the Holocene. Canadian Journal of Earth Sciences 37: 683-693. Waddington JM, Morris PJ, Kettridge N, et al. (2015) Hydrological feedbacks in northern peatlands. Ecohydrology 8: 113-127. Webber PJ, Richardson JW and Andrews JT. (1970) Post-glacial uplift and substrate age at Cape Henrietta Maria, southeastern Hudson Bay, Canada. Canadian Journal of Earth Sciences 7: 317-325. Whiting GJ. (1994) CO2 exchange in the Hudson Bay lowlands: Community characteristics and multispectral reflectance properties. Journal of Geophysical Research: Atmospheres 99: 1519-1528. Yu ZC. (2011) Holocene carbon flux histories of the world's peatlands: Global carbon-cycle implications. Holocene 21: 761-774. Yu ZC. (2012) Northern peatland carbon stocks and dynamics: a review. Biogeosciences 9: 4071-4085. Yu ZC, Loisel J, Brosseau DP, et al. (2010) Global peatland dynamics since the Last Glacial Maximum. Geophysical Research Letters 37: L13402. Yu ZC, Loisel J, Turetsky MR, et al. (2013) Evidence for elevated emissions from high-latitude wetlands contributing to high atmospheric CH4 concentration in the early Holocene. Global Biogeochemical Cycles 27: 131-140. Zoltai SC. (1991) Estimating the age of peat samples from their weight: a study from west- central Canada. The Holocene 1: 68-73. Zurcher S, Spahni R, Joos F, et al. (2013) Impact of an abrupt cooling event on interglacial methane emissions in northern peatlands. Biogeosciences 10: 1963-1981.

7 12