An 11,000 Year Fire History of a Coastal Temperate Rainforest in Prince Rupert Harbour, British Columbia

AN 11,000 YEAR FIRE HISTORY OF A COASTAL TEMPERATE RAINFOREST IN PRINCE RUPERT HARBOUR, BRITISH COLUMBIA

Jonathan William Duelks

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF ARTS

THE FACULTY OF GRADUATE AND POSTDOCTURAL STUDIES (Anthropology)

THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)

April 2021

The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:

An 11,000 year fire history of a coastal temperate rainforest in Prince Rupert Harbour, British Columbia

submitted by Jonathan William Duelks in partial fulfillment of the requirements for the degree of Master of Arts in Anthropology

Examining Committee:

Andrew Martindale, Anthropology, UBC Supervisor Zhichun Jing, Anthropology, UBC Supervisory Committee Member

Camilla Speller, Anthropology, UBC Additional Examiner

iii

Abstract

This thesis utilizes high resolution charcoal analysis and charcoal morphology to reconstruct fire history for Prince Rupert Harbour (PRH), British Columbia between ~13,200 and 3500 cal BP.

Home to the North Coast Tsimshian, PRH is an ideal study location for evaluating demographic and related environmental patterns for its extensive and well-studied archaeological record yet it lacks local paleoenvironmental data. I use this fire history to: produce one line of localized palaeoecological data, assess variation in charcoal accumulation rates (CHAR) compared against regional and semi-local paleoclimate records, determine whether or not the presence of an anthropogenic signal exists in the charcoal record, and test the relationship between natural charcoal accumulation rates and demographic models for the PRH region between 6000 and 3500 cal BP. Results indicate that fire frequency and intensity were low with just fourteen potential fire events in the DIL CHAR record, 9 in Zone 2 (13,200-6000 cal BP) and 4 in Zone 1 (6000-3500 cal BP). Zone 2 peaks could not be confidently separated from the background signal and are interpreted with caution. Peaks charcoal morphology in Zone 1 in suggest local, low intensity fires.

The data suggests that no extreme fluctuation in climate as seen through fire regime occurred over the 11,000-year record, and while there is no correlation between demography and CHAR such that an anthropogenic driver could be posited, the only fire events confidently identified occurred after the earliest known villages formed. I suggest that additional fire histories capturing the past

3500 years when PRH saw significant demographic growth will be necessary and fruitful for understanding the human impact, if any, on the fire regime of the Prince Rupert Harbour region.

Lay Summary

This thesis utilizes high resolution charcoal analysis to reconstruct fire history for Prince Rupert

Harbour (PRH), British Columbia between ~13,200-3500 cal BP. The ~9000-year archaeological record as been studied extensively; however, we lack localized paleoenvironmental data to better understand human-environment interactions. This analysis finds that charcoal was present throughout the 11,000-year record in low quantities. Four fire events between 6000-3500 cal BP could be confidently distinguished from background noise. Low counts and few fire events indicate that PRH has not seen environmental change sever enough to alter the fire regime or rather, it has remained a cool and wet environment throughout the Holocene. No direct link could be made between the actions of humans and the occurrence of fire from 6000-3500 cal BP, but the presence of fires on the landscape as population began to increase at the end of the study period is intriguing and worthy of further study.

Preface

This thesis is ultimately based on data collected by the Tsimshian History and Archaeology research project (SSHRC Grant #410-2011-0814). The original data source, lake sediment cores, were collected by Bryn Letham during the 2012-2015 field seasons for his PhD, Long-Term

Human-Environment Interaction on Dynamic Coastal Landscapes: Examples From 15,000 Years of Shoreline and Settlement Change in the Prince Rupert Harbour Area. The data (diatoms) analyzed by Letham is distinctly different from the data (charcoal) analyzed here. The Sum

Probability Distributions of radiocarbon date densities used here as a proxy for demography were calculated by Thomas J. Brown and originally published in Proceedings of the National Academy of Sciences in November 2017. The authors on that paper are: Kevan Edinborough, Marko Porčić,

Andrew Martindale, Thomas Jay Brown, Kisha Supernant, and Kenneth M. Ames. Figure three,

‘Summed probability distribution of terrestrial radiocarbon dates from Prince Rupert Harbour’ in this thesis was created by Thomas J. Brown. None of the text of this thesis is taken directly from the above-mentioned publications. The charcoal analysis presented here, and the text of this thesis are original, unpublished, intellectual product of the author, Jonathan William Duelks.

Table of Contents

Abstract ...... iii Lay Summary ...... iv Preface...... v Table of Contents ...... vi List of Tables ...... viii List of Figures ...... ix Acknowledgements ...... x Chapter 1: Introduction ...... 1 Chapter 2: Modern and Paleoenvironmental Context ...... 2 Late Glacial-3 13,500 cal BP – 11,500 cal BP ...... 3 Early Post Glacial (EPG) 11,500 cal BP – 8,900 cal BP ...... 5 The Hypsithermal 8900 cal BP – 3800 cal BP ...... 6 Paleoclimate Summary ...... 8 Modern Ecology of the Prince Rupert Harbour Sub-Zone ...... 9 Chapter 3: Archaeology ...... 9 Radiocarbon Dates ...... 11 Settlement and Demography ...... 12 Artifacts/Technology ...... 16 Subsistence ...... 18 Archaeology Summary ...... 20 Chapter 4: Reconstructing Fire History ...... 21 Charcoal Morphology ...... 22 Identifying an Anthropogenic Signal ...... 25 Chapter 5: Data and Methods ...... 27 Core Samples...... 27 Coring and Site Selection ...... 29 Digby Island Lake and Philips Lagoon Cores ...... 30 Age-Depth Model ...... 30 Core Sampling and Processing ...... 31 Peak Analysis ...... 32 vii

Charcoal Morphology Identification ...... 33 Sum Probability Distributions of Radiocarbon Dates ...... 34 Statistical Methods ...... 34 Chapter 6: Charcoal Analysis and Results ...... 34 Sedimentation and Age Depth Model ...... 34 The Charcoal Record ...... 35 Peak Identification...... 36 Charcoal Morphology ...... 39 Statistical Results ...... 40 Chapter 7: Discussion ...... 42 Fire History ...... 43 Settlement Patterns, Demography, and Fire History at PRH ...... 46 Chapter 8: Conclusion...... 48 Bibliography ...... 51 Appendices ...... 60 Appendix A. Supplemental Tables...... 60 Appendix B. CHARanalysis Output ...... 62

viii

List of Tables

Table 1. Summary of environmental data relevant to the Prince Rupert Harbour study area ...... 5 Table 2 Descriptions of charcoal morphology based on Enache and Cumming (2006)...... 23 Table 3. Radiocarbon dates from the Digby Island Lake and Philips Lagoon Cores...... 30 Table 4. Frequency of charcoal peaks...... 36 Table 5. Charcoal morphotype for the Digby Island Lake core and mean morphotypes for Zones 1 & 2...... 40

List of Figures

Figure 1. Map of Prince Rupert Harbour and locations of core samples...... 4 Figure 2. Early Holocene Period sites and Early Period sites in Prince Rupert Harbour ...... 14 Figure 3. Summed probability distribution (SPD) of terrestrial radiocarbon dates from Prince Rupert Harbour ...... 15 Figure 4. Simplified charcoal morphotype diagram ...... 25 Figure 5. Charcoal accumulation rates from Digby Island Lake (~13,000-3500 cal BP) ...... 37 Figure 6. CHAR, background series, peaks, and peak threshold from the DIL core ...... 39 Figure 7. CHAR and SPD for Prince Rupert Harbour, 6000 cal BP to 3500 cal BP...... 42 Figure 8. Scatterplots of SPD and CHAR for Prince Rupert Harbour...... 42 Figure 9. SPD and unbinned CHAR from Digby Island Lake, 7000 cal BP – 3000 cal BP ...... 48

Acknowledgements

I would like to thank my advisor, Andrew Martindale for the opportunity to conduct this research as well as his input, guidance, and patience working with me. I would also like to thank Thomas

J. Brown for his friendship and his unending support and guidance on this project and my career as an archaeologist all together. I also thank my friend and mentor Kelly M. Derr for teaching me the ways of charcoal analysis and fire history reconstruction. I thank, my undergrad professors at

Portland State University: Virginia Butler, Kenneth M Ames, Cameron Smith, Shelby Anderson for your guidance, for the opportunities you gave me, and for inspiring my love of archaeology.

And finally, thank you to my family for your patience and support.

1. Introduction

Prince Rupert Harbour (PRH) is an ideal study location for evaluating demographic and related environmental patterns for its extensive archaeological record, and its long history of archaeological research (e.g., Ames 2005; MacDonald and Inglis 1981). Located on the Northwest

Coast of British Columbia, PRH is a part of the traditional territories of Coast Tsimshian people and has been their home for millennia. The cultural history of PRH is well documented by both the oral histories of the Tsimshian (the adawx) (Marsden 2002, Martindale 2006, Martindale and

Marsden 2003, Martindale et al 2017a), and the archaeological record (Ames 2005, Martindale et al 2017a,b) , which has been studied extensively for nearly a century (Ames and Martindale 2014), however, there is little localized and detailed paleoenvironmental data with which to frame questions about relationships between people and the environment.

Recent archaeological research has begun to address this issue in PRH. Letham et al.

(2016) collected a series of sediment cores from around PRH to reconstruct a 15,000-year history of relative sea level change via diatom analysis. Using data from these sediment cores, the Digby

Island Lake and Philips Lagoon cores, my research produced a ~9700-year (13,200 cal. BP-3500 cal. BP) fire history for the Prince Rupert Harbour area using charcoal as a fire proxy. These cores were selected for this study based on their availability, dated components, proximity to archaeological sites, and potential to capture environmental parameters for reconstructing local conditions. The cores are discussed in detail below.

The study area of Prince Rupert Harbour (PRH) is home to the Nine Tribes of the Coast

Tsimshian who today are based today in Lax Kw’alaams and Metlakatla (Ames and Martindale

2014:141). Located on the northern coast of British Columbia, PRH sits among the rocky coastal shores and islands of the Hecate lowlands between the mouths of the Nass river to the north and 2 the Skeena river to the south. The western shores of Digby Island are open to Chatham Sound in the west, with Metlakatla (Venn) Pass along the north coast, and the harbour to the east. Kaien

Island sits just east of Digby with Tuck inlet to the north and Fern Pass to the east. The Tsimshian

Peninsula forms the northern border of the harbour.

As this study asks whether there is a correlation between fire histories and demographic change in this region. The resultant charcoal accumulation rates (CHAR) will be compared against known settlement history for PRH and then statistically tested against pre-established models of demography that use Summed Probability Distributions (SPD) of radiocarbon dates as a proxy for demographic changes (Brown 2016, Martindale et al. 2017b). Fire history reconstructions for

Digby Island Lake and Philips Lagoon produce detailed and localized data that will contribute to the paleoecological, archaeological, and cultural history for Prince Rupert Harbour. As fire studies in temperate rainforests are relatively rare this research will produce the northernmost fire history for coastal British Columbia, an important expansion of this tool.

2. Modern and Paleoenvironmental Context

The modern and paleo-environmental data summarized here provides a coarse-grained summary of the ecological history of Prince Rupert Harbour and the northern Northwest Coast.

This ecological history is necessary for understanding and interpreting the fire history that will be produced here. Change or stasis in fire regimes must be understood in both natural and anthropogenic contexts. Climate and vegetation determine available fuel load and provide conditions that promote or prohibit fire occurrence, while human activities e.g., controlled burning or logging augment those conditions. Thus, historic environmental data on mean temperature, moisture level, and fuel availability through time provide a baseline for understanding a landscape’s potential for fire events. 3

There is a general lack of localized data on paleoenvironments and climate change for northern-coastal-mainland British Columbia, including the PRH region, or southeast Alaska.

Paleo-environments are better understood for Haida Gwaii, Vancouver Island, and southern coastal

B.C. Paleoenvironmental reconstructions for the PRH area are also limited: Banner et al. (1983) and Turunen and Turunen (2003) reconstruct bog development for an unnamed bog woodland on

Kaien Island at the base of Hayes Mountain and Diana Lake Bog respectively (Figure 1); Heusser in his landmark 1960 publication provides pollen-based paleoenvironmental reconstructions for the NWC from Southern Alaska to Northern California, three of the bogs sampled, Summit,

Rainbow Lake, and Prince Rupert (bog) are in the PRH region; and finally, McLaren (2008) provides a late-Pleistocene/Holocene reconstruction with a focus on the P/H transition for the nearby Dundas Island group, also part of Tsimshian territory. Myriad pollen profiles have been described for SE Alaska, Haida Gwaii, the central coast, and Vancouver Island, that contribute to the general understanding of paleoenvironments for the NWC and are typically referenced by more localized studies such as those described above, to better understand both local and regional variation. While these sources are vital for our understanding of vegetation changes and inferred climate data, these studies lack well dated transitions essential to answering environmental questions at archaeologically relevant time scales. Environmental data from the above-mentioned studies are summarized below (and see Table 1).

Late Glacial-3 13,500 cal BP – 11,500 cal BP The Late Glacial Period (LGP) as described by Heusser (1960) spans 16,500 to11,500 cal

BP and is divided into three subperiods, LG-1: 16,500-15,000 cal BP, LG-2: 15,000-13,500 cal

BP, and LG-3: 13,500-11,500 cal BP. Northern BC and Alaska are only represented in the LG-3 sample which coincides with the start of the DIL core. This lack of data may be explained, at least for PRH, by the fact that deglaciation occurred by 15,090-14,365 cal. BP (Letham et al. 2016). 4

Briefly, LG-1 on the southern NWC spans the last glacial maximum with pollen profiles indicating lodgepole pine parkland with fern, grass, and sedge communities reflective of modern northern tundra plant communities (Heusser 1960:181). LG-2, on the southern NWC was slightly warmer with lower humidity, an increase in arboreal pollen, and subsequent decrease in non-arboreal pollen. New tree species present in pollen profiles include Pseudotsuga menziesii (Douglas fir),

Tsuga heterophylla (western hemlock), Picea sitchensis (Sitka spruce) and Pinus monticola

(western white pine), while Pinus contorta (lodgepole pine) continue to thrive (Heusser 1960:181).

Heusser’s LG-3 marks a return to colder temperatures and vegetation community of Salix (willow),

Alnus (alder), and Pinus contorta in the area of PRH (Heusser 1960:182) while on Dundas Island a similar Salix, grass, sedge community is present at ~12,400 cal BP immediately followed by a pine parkland with herbaceous understory establishing between 12,400 and 12,250 cal BP

(Mclaren 2008:96-101). The LGP encompasses the Younger Dryas, an extreme cold period between 12,900 and 11,600 cal BP, related to a decline in northward circulation of warmer tropical waters, negative radiative forcing, and altered atmospheric circulation (Renssen et al. 2015).

Figure 1. Map of Prince Rupert Harbour and locations of core samples. 5

The effects of the YD are typically seen in pollen profiles for the NWC (Mathewes 1993). This is

generally true of the pollen assemblages for both PRH and Dundas as evidenced by the return to

cold in LG-3 and the similarities in pollen profiles to those of open steppe-like landscapes and/or

shrub tundra found in modern coastal Alaska (Heusser 1960, McLaren 2008).

Early Post Glacial (EPG) 11,500 cal BP – 8,900 cal BP Following the Younger Dryas, climate warmed on the NWC and elsewhere. In Northwest

BC, climate was cool and wet while still warmer than that of the LGP. Paludification, driven by

climate and localized topographic and hydrological characteristics whereby forested terrain

transitions to a mire system as a result of increased moisture and decreasing drainage, is underway

by the start of the EPG (Turunen and Turunen 2003).

Table 1. Summary of environmental data relevant to the Prince Rupert Harbour study area N. Amer. Time cal Environmental Climate Major Vegetation Temp (Vaiu BP Phase et al. 2006) 13,500 13,000 Colder than previous Late Glacial Open, steppe-like; salix-alnus-pinus 12,500 1.5ky Younger Dryas 13,500 - 11,500 contorta communities 3-4 degree 12,000 12,900 to 11,700 cal BP rapid increase 11,500 (Following the YD) 11,000 Post YD warming, a cool & Open Pinus contorta with herbaceous 10,500 Early Post wet climate understory transitioning to Tsuga 10,000 Glacial 11,500 heterophylla-Picea sitchensis-alnus- Cool and Dry 10k - 9k 9,500 - 8900 cal BP fern open community, 9,000 Wetter, cooler Paludification underway 1 degree slowed 8,500 Wetter & cooler, wetland Pinus contorta-Tsuga heterophylla- increase 8,000 expansions, 8.2k event Cupressaceae-Sphagnum bog 7,500 woodland Cooling Shrubby open forest: T. plicata-C. 7,000 Warm period (Initially nootkatensis-T. heterophylla rapid, then 6,500 Hypsithermal dominant, paludification increases gradual) 6,000 8900 - 3800 cal BP Gradual cooling towards Pinus contorta and thuja plicata 5,500 modern temperature, dominant with some alnus, ferns, and 5,000 mosses common, establishment of terrestrial productivity Warming 4,500 medern vegetation regime begins recovers ~6800 cal BP (Gradual) 4,000 establishing during this period, Timing is unclear 3,500 6

and Dundas Islands initiating between 12,000 and 10,000 cal BP (McLaren 2008:102-103). The earliest vegetation communities identified in the Early Post Glacial (EPG) were still dominated primarily by Pinus contorta and Alnus with fern and herbaceous understories that eventually gave way to an increase in Tsuga heterphylla, Alnus, Picea, Tsuga mertensiana (mountain hemlock), and Cupressaceae (cedar) while Pinus contorta declines in the latter part of the EPG (Heusser

1960, McLaren 2008:102-103, Turunen and Turunen 2003). A dry period occurs in the region between about 10,000 cal BP and 9000 cal BP associated with the insolation maximum, slowing the paludification process and is immediately followed (~9300-8300 cal BP) by an increase in precipitation, decrease in temperature, and the establishment of an Picea sitchensis-Tsuga heterophylla-alnus-fern open forest community (Turunen and Turunen 2003).

The Hypsithermal 8900 cal BP – 3800 cal BP The hypsithermal as defined by Heusser (1960) is a highly variable five-millennia long period that experiences oscillations in temperature and humidity throughout. With just one dated transition at ~6800 cal BP, it is difficult to the impact and severity of those oscillations, but the author describes the early period as a Picea stichensis (sitka spruce) dominant forest with Alnus

(Alder), Polypodiopsida (fern), and Lysichiton (skunk cabbage) wetlands (pre-6,800 cal BP) followed by a Tsuga heterophylla (western hemlock) dominant forest with Picea stichensis (sitka spruce), Cupressaceae (yellow cedar), Alnus (alder), and Tsuga mertensiana (doulgas fir) still present (Heusser 1960:184-186). Fortunately, data from Banner et al. (1983), McLaren (2008), and Turunen and Turunen (2003) provide some chronological resolution of vegetation succession during the Hypsithermal in the PRH region.

Between 8300 cal BP and 8000 cal BP increasingly cool and wet conditions intensify paludification triggering a shift from a forested environment to a less productive mire complex 7

(Turunen and Turunen 2003). All pollen profiles from the PRH region exhibit peaks in Lysichiton

(skunk cabbage) pollen and decreases in Pinus contorta pollen around this time indicating the proliferation of wetland conditions. This trend of increased rainfall and humidity coupled with a decrease in evaporation is seen throughout the NWC (Heusser 1960, Mathewes and Heusser 1981,

Turunen and Turunen 2003). This cool, moist climate allowed Cupressaceae to thrive and lead to the establishment of a Pinus contorta-Tsuga heterophylla-Cupressaceae-Sphagnum bog woodland with an understory of Ericaceae (heather) and Cyperaceae (sedge) species is established between

~8000-7500 cal BP (Turunen and Turunen 2003).

The 8.2k event, a severe, ~200-400 year long cold period occurs during the first portion of the Hypsithermal, possibly caused by the influx of freshwater as glacial Lake Agassiz drained into the Atlantic, is difficult to pick up in pollen profiles and is not addressed in studies referenced here.

The event is however, well documented in the GISP2 ice core which provides a record of post-

Pleistocene climate change via variation of O18/O16 ratios (Stuvier et al 1995) and is hypothesized to have had varying effects throughout the northern hemisphere (Hong et al. 2009). As with many widespread climatic processes, the 8.2k event did not necessarily play out as a single event, but rather would have been experienced with varying severity at different times and in different regions, potentially explains why it does not appear as a distinct event in the pollen records.

Following the 8.2k event, temperatures reach the warmest of the Holocene between ~7600 and 7000 cal BP (Heusser 1960), near modern, oceanic climate begins to set in ~6900 cal BP

(Turunen and Turunen 2003), and terrestrial productivity appears to have recovered by ~6800 cal

BP (Heusser 1960). Detailed discussions of pollen assemblage fluctuations are lacking and there are no major pollen zone changes within the remaining ~3000 years of the DIL core. Studies cited here all suggest a gradual continuation of cooling and increasing moisture throughout the 8 remainder of the Holocene with modern climates and forests established between 6000 cal BP and

2000 cal BP for the southern and northern NWC, respectively (Banner et al. 1983, Heusser 1960,

1985, McLaren 2008, Turunen and Turunen 2003).

Paleoclimate Summary As indicated above, the available data on paleoenvironments within the study area are few and all are temporally coarse. Here I briefly summarize the general trends. The Late Glacial (LG) period (13,500-11,500 cal BP) saw the lowest temperatures, coinciding with the Younger Dryas

(YD) (12,900-17,00 cal BP). Open, steppe-like landscapes gave way to pine parklands where alder, willow, and lodgepole pine were common with herbaceous understories consisting of grasses, sedge, and ferns (Heusser 1960, McLaren 2008). Following the YD, the Early Post Glacial (EPG)

(11,500-8900 cal BP) saw a 3-4 C° increase (Vaiu et al. 2006), with western hemlock-sitka spruce- alder open forest communities replacing those of the LG EPG (Heusser 1960, McLaren 2008: 102-

103, Turunen and Turunen 2003). Cedar appears towards the latter part of the EPG as lodgepole pine declines (Turunen and Turunen 2003). The EPG is characterized as cool and wet with a cool- dry spell between 10k-9k cal BP. Increased precipitation resulted in decreased terrestrial productivity (Banner et al. 1983, Turunen and Turunen 2003). Both temperature and humidity fluctuate throughout the Hypsithermal (8900 – 3800 cal BP) (Heusser1960), with poorly understood chronology. Cool, wet conditions hosting spruce and cedar dominated forests and expanding wetlands are present early on, with the 8.2k cold event (~8400-7800 cal BP) bringing on the brief but coldest temperatures since the YD (Hong et al. 2009). The warmest temperatures of the Holocene then occur between ~7600-7000 cal BP (Heusser 1960). As temperatures cool, a near modern oceanic climate began to set in by ~6800 cal BP (Turunen and Turunen 2003) and terrestrial productivity starts recovery (Heusser 1960). Around this time western redcedar dominates forest assemblages with western hemlock and lodgepole pine both common, gradual 9 cooling and increasing moisture leads to the eventual establishment of modern climate and forest somewhere between 6000 and 2000 cal BP (Banner et al. 1983, Heusser 1960, 1985, McLaren

2008, Turunen and Turunen 2003). The lack of temporal precision for significant events such as the establishment of modern climates is part of the impetus for this research.

Modern Ecology of the Prince Rupert Harbour Sub-Zone PRH falls within the Coastal Western Hemlock zone, Very Wet Hypermaritime subzone,

Central variant (CWHvh2). The CWHvh2 contains forest, woodland, and blanket bogs with low productivity forests at lower elevations, and productive forests at higher elevations (Banner et al.

2005:10). Typical tree species present include western redcedar (Thuja plicata), western hemlock

(Tsuga heterophylla), yellow cedar (Chamaecyparis nootkatensis), Sitka spruce (Picea sitchensis), mountain hemlock, amabilis fir (Abies amabilis), and red alder (Alnus rubra). Salal (Gaultheria shallon), blueberry, bunchberry, deer fern, skunk cabbage and a variety of mosses dominate the understory. Mean annual temperature in Prince Rupert Harbour is 6.7 oC, the warmest and coldest months mean temperatures are 13.11 oC and -0.2 oC respectively. Mean annual precipitation is

2,523 mm and there are ~233 days a year with greater than 0.2mm of rainfall (Banner et al.

2005:10-12). It is a predominantly cool and wet environment, fog is common and while the summers are warmer and drier than winter, warm dry periods are uncommon (Banner et al.

2005:10-12), although recent years have had drier conditions, possibly a result of climate change.

3. Archaeology

This section provides a brief overview of the history of Tsimshian people via archaeological data on settlement patterns and artifact and faunal assemblages from Prince Rupert

Harbour between ~9000 and ~3500 cal BP. Currently, there is little known archaeological data from the earliest parts of the study period (13,000 to 9000 ca BP) in PRH and data from 9000 to

6000 cal BP is limited. Brief descriptions of Late Pleistocene/Early Holocene assemblages from 10 similar regions of the NWC (e.g., Haida Gwaii & SE Alaska) are given to illustrate relevant histories, settlement, and subsistence strategies. No new analyses or data are presented here.

The archaeological history of Prince Rupert Harbour is typically described in three temporal phases: (1) Early Period - 6000-3500 cal BP, (2) Middle Period – 3500-1500 cal BP and,

(3) Late Period – 1500 cal BP to contact, defined by apparent changes in settlement patterns, logistical networks, resource use and intensification, site densities, and demography among other traits as reflected in the archaeological and oral records (Coupland 1996, MacDonald and Inglis

1981, Martindale and Marsden 2003, Martindale et al 2017b). The timing and characteristics of these periods, originally proposed by MacDonald (1969, see also MacDonald and Ingles 1981) have been refined as research in the PRH continued (Ames and Martindale 2014, Martindale et al.

2017a,b). The transition from Middle to Late Period is now understood to begin ~500 years later at

~1000 cal BP and there is a poorly understood early Holocene component, otherwise the generalized characteristics of the three periods have proven to hold true (Martindale and Marsden

2003; Martindale et al. 2017a). Several sites pre-dating the Early Period in PRH have been documented since MacDonald and Ingles (1981) developed the tripartite division of PRH history.

Ames (2005), Letham et al. (2015) Martindale et al. (2017b) briefly discuss those pre-6000 cal BP archaeological assemblages, however, pre-6000 cal BP data is sparse and the period difficult to define as it is equally likely that reported assemblage variation and composition reflects actual assemblage structure as it does sample size and related issues (Ames 2005). I refer to the pre-6000 cal BP period as the Early Holocene Period as the temporal span of this study (~13,200-3500 cal

BP) pre-dates the well-established ‘Early Period’ (6000-3500 cal BP) for PRH. This Early

Holocene Period in PRH, as defined by the presence of dated archaeological sites, would date from about 12,000-6000 cal BP. 11

Radiocarbon Dates The discussions of demographic trends in PRH that follow are based on the work of

Martindale et al. (2017b) and Edinborough et al. (2017) which produced Sum Probability

Distributions (SPD) of 14C dates producing models spanning ~6000 as a proxy for demography.

Based on the assumption that there is a monotonic relationship between the number of 14C dates and the magnitude of demographic trends (Rick 1987), SPDs of 14C dates are increasingly used as a proxy to reconstruct demographic trends (Shennan et al. 2013). In short, more people on the landscape will deposit more datable material, thus an increase in 14C dates indicates increasing populations and vice versa. SPDs use the full probability distribution of age estimates, making their composite results more accurate.

The PRH SPD (Figure 4) along with its data, methods, and interpretation were originally published in Martindale et al (2017b) and Edinborough et al. (2017). I refer the reader to these sources for a comprehensive discussion of the SPD, an abbreviated version of the data and methods follow. The Dundas Islands Project (Martindale et al. 2009) and the Prince Rupert Project

(Letham et al. 2016), both led by Martindale, employed percussion coring programs collecting 14C dates (n=291) from 37 village sites throughout the PRH region. These dates were combined with

356 dates from previous research in the area totaling dates n=647 (Martindale et al. 2017a). The

UCL method (Shennan et al. 2013) used to produce the SPD curve for PRH (with minor augmentation) accounts for: taphonomic processes and long-term populations trends via an exponential curve equation (Surrovell et al. 2009); researcher-sampling bias by binning dates in

100-year intervals, long-term population trends; and a 200-year moving average to smooth the effects of the calibration curve (Edinborough 2017:2, Martindale et al. 2017b).

Three variations of the PRH SPD were calculated, one on marine 14C dates, one on terrestrial 14C dates (Edinborough et al. 2017), and one combining marine and terrestrial 14C dates 12

(Martindale et al. 2017b). All three SPD’s illustrate the same general trends with minor deviations ensuring the validity of these variations as representative of general demographic trends. All dates calibrated using OxCal v.4.2.3 (Ramsey 2001, 2009) and IntCa13 atmospheric curve (Reimer et al. 2013) and a local ΔR of 273 ± 38 (Edinborough et al. 2016).

Settlement and Demography Early Holocene Period (13,200 – 6000 cal BP) Sites dating to the Early Holocene Period (EHP) of PRH provide limited evidence of settlement patterns (Martindale et al. 2017b). Five sites in Prince Rupert Harbour have components that date between 9000-6300 cal BP (Appendix A, Figure 2A). GbTo-185 is a ~9000-year-old lithic scatter/small camp (Letham 2017), GbTo-18 and 82, and GcTo-67 are small shell bearing sites, and GbTo-23 dates the earliest component of what would become a large village site, the latter four sites date to between about 6700-6300 cal BP (Martindale et al 2017a, Martindale et al.

2020). Considering data from northern coast Tsimshian territory the sample of sites with dates in the EHP climbs to 16 total – eight in the Dundas Island group and three from Stephen’s Island – with most representing basal deposits of sites that developed into villages in later periods. Most dated sites from the Early Holocene Period are in the Dundas Islands archipelago, five sites are located in PRH. These early dates are from components in shell bearing sites and were likely occupied by marine foragers as was typical of the time period for the northern NWC (Ames

2005:294). Many of these sites are associated with known Early and Middle Period village sites

(Letham et al. 2015:65). Two sites, GcTq-4 and GdTq-3 were small village occupations by 7000-

6500 cal BP (Letham et al. 2015; Martindale et al 2017a). Limited faunal data from these sites suggest that early occupants of PRH and surrounding environs relied on near- and off-shore resources at island localities with additional terrestrial resources utilized at mainland/PRH sites

(Brewster and Martindale 2011, Letham et al 2015:69-70). Archaeologically, NWC cultures are 13 characterized as ‘maritime oriented fisher-hunter-gatherers’, many sites like those in the PRH region that date to this time are on the coastal margins and outlying islands with faunal assemblages pointing to a reliance on local intertidal and off-shore marine resources (Ames 2005).

Early Period (6000 cal BP to 3500 cal BP) Settlement patterns for Early Period (EP) are somewhat clearer, with low density – as estimated by site frequency per km of coastline (Martindale et al, 2017b:147)- early settlements scattered around the Dundas Island Archipelago (DIA), Stephens Island, and in PRH, concentrated around Metlakatla Pass and in the interior near Kitselas Canyon (Letham et al. 2015, Martindale and Marsden 2003:20, Martindale et la. 2017b). The number of dated villages and large shell bearing sites in PRH doubles during the EP; GbTo-23, 31, 33, 34, 59, 66, GcTn-9, GcTo-6 & 27 are village sites and GbTn-19, GbTo-18 & 36 are large shell bearing sites (Figure 2b). Three village types are defined by Martindale et al (2017a) as seen in the archaeological record: small (<14 houses) straight rowed, large straight rowed, and large U-rowed. Small straight villages, the only type dated to the Early Holocene Period, first occur in the DIA prior to 6000 cal BP with their numbers increasing slightly throughout the harbour by ~5000 cal BP (Martindale et al. 2017a).

Small villages persist throughout the occupation of the harbour, both large types first appear by the end of the EP (Martindale et al. 2017a). Although volumetric samples for the early period are low and thus difficult to compare, the available data illustrates striking similarity to sites of this age in other parts of the northern NWC, at least for the first millennium of the EP (Ames 2005:294).

The post 5000 cal BP settlement pattern sees an increase in the density of small shell bearing camps and small villages occupying coastal landscapes throughout the PRH (Martindale et al. 2017b), and a slight decrease in the density of sites in the Dundas Island archipelago (DIA), and surrounding areas (Martindale and Marsden 2003:20, Martindale et al 2017b, Letham et al.

2015). Although exact timings are unclear, between 4500-3000 years ago village composition 14 changes in two ways. First, village composition shifts from uniformity of house size, all large or all small, with large and small houses co-occurring. Second, village size, measured by volume of shell and number of houses, increases (Martindale et al. 2017a:318).

Figure 2. Maps of (A) Early Holocene Period (9000-6000 cal BP) sites and (B) Early Period (6000-3500 cal BP) sites in Prince Rupert Harbour. By the end of the EP, circa 4000 cal BP there is further constriction of settlement with increased spatial clustering of sites on Dundas and even more so in the PRH (Martindale et al.

2017b). An increase in midden accumulation rates and an early rise in demography inferred from the SPD curve (Figure 3) coincide with that clustering of sites (Martindale et al. 2017b). The increased occurrence of village sites, the expansion of site size, like that seen at GbTo-31, and seasonality studies indicating year-round occupation of sites starting at the beginning of the millennium was coterminous with a shift towards semi- and full-sedentism seen throughout the

NWC (Ames 2005:298, Ames and Maschner 1999, Stewart and Stewart 1996).

During the EP in PRH, mobility appears tied to resource localities within localized catchment zones (Martindale and Marsden 2003:21). Ames (2005) suggests that settlement 15

Figure 3. Summed probability distribution of terrestrial radiocarbon dates from Prince Rupert Harbour, Dundas Islands, and Stevens Island. IntCal13 calibration curve used. 200-year bins. Gray shaded area is the SPD, red dashed line is the SPD smoothed 200-year moving average. patterns and artifact and faunal assemblages indicate that the shift towards logistical mobility may be underway by the end of the EP (2005:296). Small houses and villages are preset by ~6500 cal

BP in the broader Tsimshian region with larger house sizes and larger villages appearing between

~4500-3000 cal BP. (Martindale et al. 2017a). The semi- to full sedentism evident in the Middle

Period, may have its origins towards the EP (Ames 2005:296).

Middle Period 3500 – 1000 cal BP The span of the research presented here ends as the archaeology of the Tsimshian world in

PRH moves into the Middle Period (MP) where substantial changes in sedentism, social organization, subsistence, demography, and technology are clearly seen in the archaeological record. A dramatic increase in demography begins here and coincides with an increase in village sites both in number and in size (Martindale et al. 2017a,b). As village size increases so too does house size (Martindale et al. 2017a). As the number of occupied villages increase the density increases. Martindale et al. (2017b) show that when viewing the data in 100-year periods beginning 16 at 4000 cal BP villages begin to cluster in the Dundas Islands and even more-so in PRH, a pattern which continues until ~1200 cal BP. At its height between 2700 and 1200 cal BP approximately

50 sites are occupied, many contemporaneously, in PRH with ~50 sites occupied between 2700-

1200 cal BP (Letham et al. 2020, Martindale et al. 2017b). For a brief period between around 1000 cal BP there are no occupied sites in the region including the PRH, after which the harbour is quickly resettled with nearly all occupied sites clustered around the main entrances to the harbour

(Edinborough et al. 2017, Letham et al. 2020, Martindale et al. 2017b). Post-3500 cal BP artifact diversity increases dramatically as well (Ames 2005:302). Groundstone tools e.g., slate points, celts, mauls, and percussors are present in increasing quantities and the quantity and diversity of bone implements e.g., bipoints, harpoon valves, and bilaterally barbed harpoon points (Ames

2005:302). The MP also sees increased intensification of species and habitats as well as increased specialization including copper fabrication and more specialized wood working (Ames 2005:303).

Artifacts/Technology Artifact analysis, and more specifically artifact based cultural chronologies for PRH have proven difficult as there is little variation in form, and changes in assemblage structure are generally additive, new types are adopted while old types persist (Ames and Martindale 2014).

Technological trends appear cumulative, and with a few exceptions, most artifact classes that comprise the earlier assemblages persist in use through to contact with new styles adopted, while old are not retired. This trend was observed early on in PRH research (MacDonald and Inglis 1981) and persists through the most recent comprehensive analysis of artifacts from PRH (Ames 2005).

Early Holocene Period (13,200 – 6000 cal BP) Late Pleistocene/Early Holocene artifact assemblages from Prince Rupert Harbour are limited. Many early components have been dated via a coring program designed to sample basal/terminal dates of shell bearing sites (Lethem at al. 2019), providing invaluable data on 17 settlement patterns, midden accumulation rates, and site expansion via terraforming but preclude the recovery and analysis of artifacts. Letham et al. (2018) modeled relative sea level change and surveyed and tested paleoshorelines identifying three pre-6000 cal BP sites and several other potential sites. Of these sites one (GbTo-82) is a small shell midden located on a relic shoreline near Henry Point on the west side of Digby Island dating to 6727-6434 cal BP but no artifacts we recovered in the shovel tests. The other two sites, GcTo-67 and GbTo-185 contained small lithic artifacts. GcTo-67 located on the mainland north of Digby Island at the mouth of Scott Inlet contained a shell layer with a modified bird bone and quartz and quartzite flakes and debitage dated to 6643-6445 cal BP and a single cobble tool with usewear was found below a layer dating to 7196-6966 cal BP. GbTo-185 contained two hearths dating to 8348-8185 cal BP and 9302-9028 cal BP; artifacts associated with these hearths include hammerstones, a spokeshave, notched cobbles, and quartz and quartzite flakes (Letham et al. 2018).

The known sample of Late Pleistocene/early Holocene lithic technology on the NWC in general is also limited, but aggregate analysis of assemblages from SE Alaska to Northern

California illustrate broad patterns in the available data (Willis and Lauriers 2011:131-133). Late

Pleistocene-Early Holocene lithic technology from the greater NWC includes foliate/leaf bifaces, some stemmed points, unifacially modified flakes, scrapers, cobble choppers, and other expedient tools (Willis and Lauriers 2011). Microblades, absent from early sites in PRH, are a common hallmark at early Holocene sites elsewhere on the NWC (Fedje and Mackie 2005). The lack of microblades in PRH may be due to small sample size of sites. Wet sites from Haida Gwaii have produced bone implements including awls, barbed points, percussors, and awls as well as plant fiber-based cordage and basket remnants (Fedje and Mackie 2005, Fedje et al. 2001).

Early Period (6000 – 3500 cal BP) Bone awls, bone wood working tools, abraders, and spalls dominate PRH assemblages in the early period. An interesting attribute of Early Period (EP) assemblages is the presence of antler flakers considering the absence of flake stone technology; however, it is possible that flaked stone fragments and/or debitage was present but unrecovered due to dark, wet sediments and screening methods. Antler flakes do not appear in post 3500 assemblages (Ames 2005:302). Many styles common in later periods but unidentified in the early period include celts, mauls, net singers, complete ground slate points, and various harpoon types and components. At present, it is equally likely that the absence of tool types is a result of sampling as it is an accurate reflection of the technology of the times (Ames 2005:302).

Beginning somewhere around 3800 cal BP, artifact diversity in PRH increases exponentially, which is likely related to sampling as well as a shift towards increasing sedentism, intensification of resources and habitats, and increase in heavy woodworking (Ames 2005: 302-

303). It is in the succeeding middle period (3500-1500 cal BP) that the diversity of subsistence gear expands considerably along with; increases in population as evidenced by the SPD curve

(Figure 4), the proliferation of large and small villages, a second pulse in midden accumulation rates, increased regional interaction, and the development of the classic NWC pattern (Martindale et al. 2017b). While these developments fall outside of the temporal scope of the fire history presented here, when viewed as processes, these developments likely began in the early period.

Subsistence Our understanding of faunal assemblages for PRH during this study period is even more limited than that of the artifact assemblages due to sampling and collection issues and the relatively few studies focusing on faunal analysis (Ames 2005, but see Johnson 2019 and Patton et al. 2019).

Like the artifacts, there is relatively little change overall in the content of resource exploitation. 19

As previously discussed, faunal assemblages from the Dundas Islands Archipelago indicate extensive use of shellfish among other intertidal resources and offshore marine resources. Early deposits include shell lenses indicating that even the earliest inhabitants were harvesting shellfish

(Letham et al. 2015:70). Shellfish were harvested year-round throughout the occupation of Dundas

Islands, and seasonally during the spring and fall in PRH (Burchell et al 2013, Hallman et al. 2013).

Ethnographically, salmon is known as the most important resource in Tsimshian territory, with pink, coho, and sockeye ranking most important while chum and chinook salmon are used as well (Garfield 1966). Eulachon have been ranked second in importance as food, as well as a source of oil, important for preservation of perishable goods such as berries and for trade, herring was similarly used (Garfield 1966, Patton et al. 2019). The eulachon runs provided a well-timed addition to the diet as early spring runs coincided with potentially dwindling salmon stores. Cod and halibut were important resources available during winter months.

Salmon typically dominates faunal assemblages throughout the harbour (Coupland et al.

2009). Coupland et al. (2009) analyze faunal assemblages from five village sites and conclude that there is evidence for “extreme salmon specialization” between 2500-1000 cal BP. At McNichol

Creek,78.9% of total faunal remains were fish with salmon making up 97.6% of identified fish remains (Coupland et al. 2009:195). Similarly high ratios of fish and salmon were observed at the four other sites: Boardwalk, 95.1% of faunal remains were fish with salmon making up 95.1% of identified remains, Tremayne Bay -99.1% fish, 95.8% salmon, Philips Point - 99.2% fish, 94.4%

Salmon, and GbTo-77 - 95.4% fish, 90% salmon.

Eulachon and herring are also common, while spiny dogfish, ratfish, and flatfish are present in early period sites (Ames 2005). Brewster and Martindale (2011) find that eulachon dominate fish reamins at GcTq-13 (49%) and GcTq-6 (66%), two campsites in the Dundas Islands 20

Archipelago. Patton et al. (2019) revisit several sites in PRH employing fine mesh (6.3mm – 2mm) screens to investigate the history of eulachon fisheries. They find that smelts occur at 87% of sites investigated and make up between 0.25% and 37% based on NISP (Paton et al 2019:693).

Mammals played an important role in Tsimshian subsistence as well, although to a lesser extent. Canid are an important part of subsistence in the early period, tending to dominate Early

Period mammal assemblages, but their prevalence is decreasing by ~3500 cal BP at the same time as the presence of deer begin to increase through time. Present in all periods are deer, harbour seals, canids, and sea otters. Mountain goat remains appear more common in later period assemblages suggesting an increase in interaction between coast and interior (Ames 2005:270).

Archaeology Summary Archaeological data is minimal during the first ~7000 years of the fire history presented below, the remaining period, 6000 and 3500 cal BP coincides with the first few millennia of available archaeological data ending just as the archaeological record documents significant change in settlement, demography, and subsistence in PRH. Data from the Early Holocene Period

(EHP) is minimal compared to the following 6000 years. There are no sites that date between

13,000-9000 cal BP in PRH. In PRH, two small camp sites with small lithic assemblages have dates of ~9000 cal BP and 7000-6500 cal BP. These sites are located on paleoshorelines several meters above sites that to the mid-Holocene and later indicating long-term, persistent settlement of the harbour (Letham et al. 2020). The earliest villages are in the Dundas Islands, while several sites in PRH may have been established in some form prior to 6000 cal BP (Martindale et al.

2017a,b). Lithic assemblages from the EHP are few but generally fit with contemporaneous NWC assemblages reflecting generalized costal foragers (Willis and Des Lauriers 2011). Settlement data from the Early Period (EP) between 6000-3500 cal BP indicate scattered settlement around Dundas

Islands, Stephens Island, and PRH. Dated sites within PRH increase throughout the EP and a 21 pattern of increased site density that begins in the EP continues throughout the Middle Period (MP,

3500-1000 cal BP). Village and house size also increase throughout the MP (Martindale et al.

2017a). Mobility appears tied to resource specific sites throughout much of the EP (Martindale and

Marsden 2003) with the shift towards logistical mobility beginning towards the end of the EP

(Ames 2005). The MP, occurring outside the span of the fire history presented here, is a time of significant changes in Tsimshian history. Dramatic increases are seen in site numbers, size and density as well as in demography and diversity of artifacts (Ames 2005, Martindale et al 2017b).

4. Reconstructing Fire History

Simply put, fire history is reconstructed by counting charcoal from lake, bog, and sediment cores at equal and contiguous intervals (Whiltock and Laresn 2001). These sediments preserve data of past environmental conditions and can be used to reconstruct climate, vegetation, and fire histories (Whitlock and Larsen 2001). Charcoal is introduced into lake sediments in several ways.

Primary charcoal is deposited during or immediately after a fire while secondary charcoal is deposited in between fire events as a result of slope wash, sediment mixing, and transportation by streams leading into lakes (Whitlock and Larsen 2001:76).

While limnological records provide amazing time depth to study past environmental change, they can lack temporal and spatial precision. However, loose spatial and temporal precision can be refined with proper sampling, age-depth models, and an understanding of the properties of charcoal morphology and transport (Whitlock and Larsen 2001, Mooney and Tinner

2011). Pioneered by Iversen (1941), early fire histories used microscopic (<100µm) charcoal collected during pollen analysis, but more recent work focuses on macroscopic (>100µm) charcoal which is easier to process, quicker to quantify, and likely represents a more localized signal

(Whitlock and Larsen 2001). Recent research has since confirmed that macroscopic charcoal (>125 22

µm) is typically transported <10 km from a fire, thus it can reliably be interpreted as a local/extra- local signal (Mooney and Tinner 2011, Higuera et al. 2011).

While a simple counting method is valid and often still used, advanced statistical methods have been developed to allow for identification of background and fire event signals, and to compare and synthesize multiple datasets Higuera et al. 20010). Coupled with identification of charcoal morphology, these methods are well suited for studies answering questions about anthropogenic burning, fire intensity, and fuel (Derr 2012, Encahe and Cumming 2006).

Fire studies are primarily conducted in environments where wildfires are frequent in order to collect data on long-term fire dynamics that can be applied to current management strategies.

Thus, fire studies in high latitude, coastal temperate rainforests where naturally occurring fires are infrequent, e.g., the northwest coast of British Columbia, are rare and data on naturally and culturally induced fire dynamics are scant. It In the last few decades, fire ecologists and archaeologists have begun to focus on identifying the human influence on fire regimes and the use of fire to modify the landscape. On the NWC, where anthropogenic burning is well documented in the ethnographic record (Boyd 1999) new research has expanded the ecological boundaries of fire studies to include the coastal temperate rainforests of southern BC (Brown and Hebda 2002,

2003; Gavin et al. 2003, Hoffman et al. 2016a, 2016b, 2017)

Charcoal Morphology Charcoal morphology can be used in addition to raw counts to answer specific questions about fire type, fuel type, and charcoal transport (Derr 2012, Enache and Cummings 2006). Enache and Cummings (2006) define seven charcoal morphotypes occurring in two cores from interior

British Columbia (Table 2, Figure 4). These morphotypes can be useful throughout B.C. in areas where fuel sources are similar and have proven useful for coastal sites in cases where all seven morphotypes are present in a core sample (Derr 2012). Preliminary analysis of the Digby Island 23

Lake core has identified all seven morphotypes described by Enache and Cumming (2006), and so are suitable for this research.

Table 2 Descriptions of charcoal morphology based on Enache and Cumming (2006). Charcoal morphology Shape Morphotype Description

Irregular Type M With Structure, Highly porous, thin and prone to breakage

No Structure, Tiny amorphous sheets consisting of a black Irregular Type P powdery material Compact, structure, black, Similar to type C, but more Geometric Type S fragile due to structure

Compact, structure, partial black, Partially burnt, will often Geometric Type B have visible cell layers Compact, no structure, Compact, angular-irregular Geometric Type C geometric shape Elongated, with ramifications, Cylindrical fragments with Geometric Type D branchlike protrusions, Elongated, no ramifications, Cylindrical fragments no Geometric Type F ramifications

Enache and Cumming (2006) identify seven charcoal morphotypes in their study. These morphotypes are split into two general categories: Irregular (Types M and P) and Geometric

(Types S, C, B, D, and F), and further separated by shape, internal and external structure, and degree of burning (Table 2, Figure 4). Enache and Cumming describe Type “S” and “C” as the two most abundant charcoal morphotypes in their studies (2006, 2007). Type C charcoal is a compact, blocky-angular geometric fragment that does not exhibit any visible structure (Enache and Cumming 2006:285). While Type C charcoal has no visible structure, or wall porosity, it may appear fibrous which could be indicative of tree bark. Type S is compact fragment, angular and/or irregular in shape that is distinguished from Type C due to the presence of visible cell structure.

In addition to visible structure Type S may exhibit fibers and layers that can be indicative of wood fuel (Enache and Cumming 2006:285). It is also noted that when broken, Type C charcoal often 24 reveals visible internal structure. Thus, if Type C charcoal is broken during initial transport, prior to deposition in the lakebed or during the extraction process it would decrease and increase counts of Type C and S respectively. Type B charcoal is identical to Type S except that it is only partially burnt. This may be indicative of short lived or lower severity fires where fragments are deposited in the lake prior to full combustion. Type D are elongated and cylindrical fragments with ramifications (offshoots) present, while Type F are elongated fragments with no ramifications.

The two irregular morphotypes are M and P. Type M are quite fragile, with an irregular and highly porous structure resembling experimentally produced charcoals that originate from high heat and/or the burning of leaves and small branches. Type P fragments are irregular in shape with no structure and are described as a fragile “black powdery material” (Enache and Cumming

2006:285). The extremely fragile nature of Types M and P is helpful in interpreting not only the severity of fires, as combustion at higher temperatures produces fragile particles, but also as an indicator of primary deposition which may help distinguish between local, extra-local, and regional fires as there is less of a chance that these particles would survive long distance transport down in-feeding rivers or via slope wash (Enache and Cumming 2006:287-288).

Understanding the charcoal morphotypes present in a sample can help with interpreting fire events in number of ways. These morphological types can be assigned to generalized fuel type categories to define fuels consumed without the need for identification to species providing some indication of the nature of the fire. For instance, large, high intensity fires tend to produce predominantly large, wood charcoal fragments, whereas smaller understory fires less charcoal overall with smaller particles of charred grasses and sedges highly represented (Whitlock and

Larsen 2001). Thus, types D and F can be important for identifying low intensity fires and even anthropogenic signals in the charcoal record. Low intensity understory, grassland, or meadow fires 25 generally produce types D and F, as they do not burn hot enough to consume trees or create heat plumes that transport larger particles greater distances. So, when a sediment sample is dominated by these morphotypes, it can be interpreted resulting from a low intensity fire near the lake or bog sample (Enache and Cummings (2006). A regular presence of grass charcoal through time may be interpreted as maintenance fires that were burned somewhat regularly as part management practices (Derr 2012:197).

Figure 4. Simplified charcoal morphotype diagram developed by Enache and Cumming (2006: figure 2). Identifying an Anthropogenic Signal The use of fire by Indigenous peoples of North America to modify and manage landscapes has been well documented in the ethnographic record. On the Northwest Coast, prescribed burning was used to develop, expand, and maintain habitats for a variety of important root and berry species and to create grazing habitats for animals that may be hunted (Boyd 1999, Lepofsky et al. 2005,

Lepofsky and Lertzman 2008). Ethnographic records document knowledge and use of burning practices in the more recent past while the presence of plant species outside of their natural range 26 e.g., camas and Gary Oak parklands provide additional, indirect evidence for prescribed burning

(Lepofsky et al. 2005:219-220).

Although methods for reconstructing millennia-scale natural fire histories have been well developed over the past two decades (Gavin et al 2003. Higuera et al. 2007, 2009, 2010, Power et al. 2010), distinguishing between anthropogenic and natural fire events remains difficult. Both natural and human-set fires produce and deposit charcoal. The amount and morphology of charcoal produced are dependent on fire severity and fuel type, respectively. Both may be affected by post depositional processes. Fire severity is influenced by fuel load, fuel type, moisture levels, and weather. There is no singular signature left by either type of fire.

Many fire studies aim to understand local-to-regional scale histories in the context of modern forest management and while the identification of an anthropogenic signal is often referenced it is not often the main goal. In these studies, regional scale fire histories reconstruct trends in biomass burning or localized fire histories and assess those patterns in relation to paleoenvironmental reconstructions and indigenous occupation histories. Most rely on visual comparison of fire histories, pollen profiles, and past population estimates of varying empirical quality (Brown and Hebda 2002, Murphy 2012, Walsh et al. 2005, 2008). When fire research focuses on more recent history additional proxy evidence such as stand age and fire scarred tree rings can be combined with sedimentary charcoal and compared with known and dated former habitation sites to infer anthropogenic burning (Hoffman et al. 2016a, 2016b, 2017).

Derr (2012, 2016) offers a convincing method for identifying anthropogenic burning in deeper time. The analysis focuses on two bog cores from Valdez Island in the Strait of Georgia off

Vancouver Island. One core came from a bog next to DgRv-02, a Coast Salish village site with a

~4500-year occupation history – the “on-site” location – and the other “off-site” core was taken 27 from an unnamed bog located >6 km from the on-site core and not directly associated with known archaeological sites (Derr 2016). In addition to charcoal counts Derr (2016) uses charcoal morphology sensu Enache and Cumming (2006) to reconstruct fuel sources and fire severity. By identifying changes in fire severity and fuel type occurring in the on-site core, after the occupation of DgRv-02 and not seen in the off-site core data Derr (2016) offers compelling evidence of anthropogenic burning at a local scale.

The research presented below creates a localized fire history to compare against settlement patterns (Martindale et al. 2017b) and demographic reconstructions specific to PRH (Edinborough et al. 2017). It uses charcoal morphology to help understand fire type, and due to a lack of local data, interprets the local fire history in the context of regional paleoenvironmental reconstructions.

5. Data and Methods

Core Samples The DIL and PL cores were originally taken for relative sea level analysis (see Letham

2017). I selected them for charcoal analysis for two reasons: (1) their geographic location and relation to known archaeological sites and (2) for the time periods represented in each core.

Location was an important factor for several reasons. First, Digby Island Lake (DIL) core meets all criteria for fire study site selection as proposed by Whitlock and Larsen (2001:79-80).

Located near the center of the island DIL provides a long-term sedimentary record undisturbed by fluctuating sea levels after ~13,000 cal BP. Given DIL’s island locale, the lake’s watershed is small, and topography is low, meaning that long-distance secondary charcoal deposition via rivers and/or slope-wash is avoided (Whitlock and Larsen 2001). Philips Lagoon, however, does not meet the same site-selection criteria, but was chosen primarily for its proximity to known village sites as an effort to control for any anthropogenic signal. Coastal lagoons are cut-off from the sea by sandbars or reefs and often have freshwater inputs as well. Thus, there is potential that tidal action 28 may bring in pollen or charcoal from distant sources. While lagoons are not the most common sources for sediment based environmental reconstructions, studies have shown that they can reliably reflect local and extra-local conditions (Power et al. 2010).

Secondly, DIL was chosen as an ‘Off-site’ location (sensu Derr 2012), in that it is isolated in terms of its proximity to Tsimshian villages on Digby Island. The nearest known village site

(GbTo-6) is 2.2 km northeast of DIL while other villages on Digby Islands coast are ~ 2.5-3.5 km from the lake. Fire history reconstructions (Derr 2012, Hoffman et al. 2017) that have successfully identified an anthropogenic signal on the NWC do so by comparing records taken near (~few hundred meters) village locations with records from several kilometers away. Thus, DIL was chosen as an ‘off-site’ core under the assumption that CHAR would reflect ‘natural’ as opposed to anthropogenic signal should one exist at all. Likewise, the PL core was chosen to represent an ‘On- site’ location (sensu Derr 2012). Philips Lagoon is located between three known village sites,

GbTo-72 to the east and GbTo-57 and GbTo-59 to the west. Additionally, PL is within 2 km of seven other villages in Tsimshian territory (Figure 2).

Finally, the DIL and PL cores were chosen for the time period represented in both cores.

As discussed above, the DIL core represents ~9700 years (~13,206 cal BP to ~3947 cal BP) and the PL core represents ~13,000 years (~13,355 cal BP to present) of depositional history. The cores were chosen for analysis prior to receiving the additional limiting date of DIL and the two PL dates but under the assumption that both cores had a high likelihood of spanning the Holocene which would allow for fire-history reconstructions capturing the Pleistocene-Holocene transition, major climatic events (e.g., the Younger-Dryas and 8.2k events) and spanning the Tsimshian people’s archaeologically known occupation history of PRH (Letham personal communication).

Unfortunately, the DIL core did not capture the last ~3500 years of depositional history. 29

Coring and Site Selection The cores used in this study were collected during the 2012-2015 field seasons for Bryn

Letham’s PhD research which produced a localized post-glacial relative sea-level (RSL) curve for the Prince Rupert Harbour area. That research was a part of Andrew Martindale and Ken Ames’ larger Tsimshian history and archaeology research (SSHRC Grant #410-2011-0814). The RSL study area encompassed a ~18 km2 area centered around PRH. Lacustrine, geologic, and archaeological data points were taken from the south and southwest portion of the Tsimshian

Peninsula, Digby Island, and Kaien Island. Letham’s research utilized three data sources: (1) sediment cores collected from lakes, bogs, and intertidal zones, (2) relict marine sediments from exposed erosion faces, and (3) basal dates from archaeological sites (Letham 2017:40-47).

Macrofossils, sediment transitions, and limiting dates indicative of RSL changes were synthesized from these datasets to reconstruct sea level change across the Holocene (Letham 2016).

The Digby Island Lake (DIL) and Philips Lagoon (PL) cores from Letham’s (2017) work were chosen for their proximity to known archaeological sites and the time periods represented by each core. Site selection in a fire study would typically be targeted at specific locations and to answer specific questions. E.g., to identify anthropogenic burning, sites near archaeological sites and known use areas might be targeted for comparison to ‘offsite’ cores from areas with little known archaeology (Derr 2012). Site selection is typically based on watershed size -where larger watersheds lead to an overrepresentation of non-local charcoal and slope -as mass-wasting events following fires in areas with steep slope can redeposit charcoal from the same event multiple times over the course of several years thereby increasing that events’ signal (Whitlock and Larsen

2001:79-80)

Digby Island Lake and Philips Lagoon Cores The Philips Lagoon core recovered 180 cm of sediment with only two sediment zones originally described by Letham (2016); Zone 1: laminated gray sand, silt, and clay from 180-31 cm, varying amounts of shell are present between ~80-30 cm, Zone 2: shell rich brown sand and gravel in the top ~ 31cm of the core (Letham 2016:328). The PL core represents ~13,000 years

(~13,355 cal BP to present) of depositional history (Table 3).

The Digby Island Lake core recovered 260 cm of sediment and includes three sediment zones originally described by Letham (2016); sediment zone 1: dark brown peat/gyttja from 0 to

218 cm, Zone 2: medium brown silty mud from 218-235 cm, Zone 3: grey, clayey-sandy silt from

235 cm to 260 cm (Letham 2016:84). The DIL core represents ~9700 years (13,200 cal BP-3947 cal BP) of depositional history (Table 3).

Table 3. Radiocarbon dates from the Digby Island Lake and Philips Lagoon Cores. Median Calibrated Calibrated Depth Core Lab ID Age Cal Material Age From Age To (cm) BP DIL UOC-4167 4151 3809 3947 Twig 10 DIL D-AMS 008745 15,013 13,589 14,283 Twigs 211 PL UOC-4165 1938 1968 1942 Twig 10 PL UOC-4166 13,450 13,220 13,335 Organic macrofossil 74

Age-Depth Model Accurate dating and temporal control are crucial in any archaeological research, especially so when comparing proxy records to make inferences about human-environment interaction.

Ideally, core sediments would be annually laminated allowing for precise identification of annual fire patterns (Whitlock and Larsen 2001) however, neither the core exhibited clear laminae.

With two dates available for each core a simple age-depth model was possible. Using median dates, the difference between the top and bottom core 14C dates was calculated and the result divided by the length (cm) of the core between those limiting dates, providing an estimation 31 of n-years per cm3. Although a series of dates throughout the core is ideal for producing accurate age depth models there were no visible sedimentation changes between 14C dates for the DIL core so we trust that the model will be fairly accurate. While the PL core did contain sediment zone transitions between 14C dates, additional dates were not possible due to financial constraints, so I use the simple age depth model for PL as well.

Core Sampling and Processing Sampling cores and processing samples for analysis of macroscopic charcoal was adapted from Derr 2012 and Whitlock and Larsen 2001. The charcoal data used here was processed during the summer of 2017 and counted during the winter of the same year.

1. One cm3 samples of sediment were taken from the DIL and PL core at contiguous 1cm intervals as contiguous samples provide greater temporal resolution. 2. Individual samples were next placed 50 ml centrifuge tubes to soak in a mild deflocculant

solution (10% sodium hexametaphosphate (NaPO3)6 and distilled water) for 24-48 hours to loosen organic material without fragmenting charcoal. This helps prevent fragmenting of charcoal during processing which can result in false increases in charcoal counts. 3. Following the deflocculant soak, sediments were then sieved through nested screens of 500 µm and 125 µm size. Sample number, size, and depth were recorded, and the two size classes were transferred to separate vials.

4. Each sample was then soaked in a 9:1 H2O to sodium hypochlorite (NaClO, bleach) solution for 24 hours. This step serves to make charcoal identification easier by lightening organic materials and leaving charcoal unaltered. 5. Following the bleaching process, samples were rinsed with distilled water and placed back in clean vials until ready to count. 6. Each sample was then transferred to a gridded petri dish, charcoal fragments were counted and assigned a charcoal morphology (discussed below) under 40-100x magnification. 7. Counts of charcoal by morphotype were recorded on seven individual Staples Mechanical Tally Counters for each sample with tallies recorded in a spreadsheet.

Raw data was then converted to charcoal concentrations (n particles per cm-3) by dividing counts per sample by sample volume and charcoal accumulation rates (CHAR) were calculated by dividing charcoal concentrations by the deposition accumulation time per cm-1 (Whitlock and

Larsen 2001:85). The resulting CHAR values indicate particles/cm2/year deposited to account for variation in accumulation rates over time.

Peak Analysis Two methods were employed to identify peaks in the CHAR record, a simple count method, and a statistical decomposition of the charcoal timeseries. For the first, I identified peaks via raw counts of charcoal particles where N > 50 represents fire events (Derr 2012, Whitlock and

Larsen 2001). The decomposition method utilized CharAnalysis (Higuera 2009) to detrend the charcoal series into a slowly varying background component, the result of charcoal from regional

(distant) fires, changes in fuel source, secondary deposition, and post-depositional sediment mixing (Gavin et al. 2006:1725, Higuera et al. 2007), and a peak series, representing local fire events (Clark and Royall 1996, Long et al. 1998).

Identifying peaks using the raw count method is relatively simple. Samples with counts above 50 signify a fire event or events within the sample timespan, while counts below 50 represent the slowly varying background signal (Whitlock and Larsen 2001). Some studies conducted in coastal temperate rainforests of the NWC suggest reducing minimum counts in these hyper-wet maritime environments. Derr (2012), suggests samples with ≥20 fragments of charcoal are bracketed by samples with zero counts as representing fire events in the Gulf Islands Region.

Decomposition was performed in the CharAnalysis program (Higuera 2009). First,

CharAnalysis interpolates raw charcoal timeseries and sediment accumulation rates into equal intervals. The background signal was modeled with LOESS smoothing robust to outliers with a

900-year window, with the window width determined by the results of the sensitivity analysis, 33

Sensitivity analysis produces signal-to-noise (SNI) ratios providing a measure of the difference between fire and non-fire related peaks and KS-Goodness of fit test describes the relationship between the peaks component and the noise component, the sum of which indicated maximized results at 900 years (Higuera et al. 2009).

The peak series was defined as a residual by subtracting the background series from interpolated CHAR. Next, a threshold was specified to distinguish between fire-related peaks and non-fire related peaks- those that result from statistically insignificant variation in charcoal quantities (Gavin et al. 2006). I used a locally defined threshold, which assumes that variance around the background series changes through time (Higuera et al. 2008), with the threshold set at the 95th percentile and defining the noise component using a gaussian mixture model (Higuera et al. 2010, Higuera 2009). Minimum count was then set at 0.05, removing any peaks from the preceding 75 years with >5% chance of coming from the same Poisson distribution as count of the peak itself (Higuera 2009:9).

Charcoal Morphology Identification As discussed above charcoal morphology can provide information on fire type, fuel type, and charcoal transport. The seven charcoal morphotypes referenced here were defined in several studies from interior (Enache and Cumming 2006, 2007) and coastal British Columbia (Derr 2012)

(Figure 5). These morphotypes are useful throughout B.C. in areas where fuel sources are similar and when/where the seven morphotypes are present in a core (Derr 2012). All seven morphotypes were identified in the samples from PRH, thus they are used here. Morphotypes were identified during the initial counting step by visual comparison to published photographs and descriptions

(Enache and Cumming 20006, 2007), along with guidance and training from Dr. Kelly Derr.

Sum Probability Distributions of Radiocarbon Dates I use Sum Probability Distribution (SPD) of terrestrially derived 14C date densities from

Prince Rupert Harbour (discussed above) as a proxy for demography from 6,000-3,500 cal BP.

The SPD curve (Martindale et al. 2017b) was calculated in OxCal version 4.2.3 (Bronk Ramsey

2001, 2009) IntCa13 atmospheric curve (Reimer et al. 2013). I use the 100-year binned SPD densities from the 6,000-3,500 cal BP period (n=26) from the original authors database with permission (Edinborough et al. 2017: supplemental material).

Statistical Methods Following Williams et al. (2015) I tested statistical correlations between CHAR and the archaeological data (SPD) to identify correlations between fire history and demographic trends between 6000-3500 cal BP. First, CHAR values were binned in 100-year increments by taking mean CHAR at every 100-year increment so that the two datasets were temporally aligned.

Binning created a subset of CHAR data (n=26) encompassing the 6000-3500 cal BP period.

Next, a series of statistical correlations between CHAR and SPD are calculated to test for correlations between the datasets using PAST: Paleontological statistics Software package

(Hammer et al. 2001). Statistical correlations were tested between SPD and CHAR for the complete record, 6000-3500 cal BP period (SPD n=26, CHAR n=26), and then in overlapping

2000-year intervals (6000-4000 cal BP (n=20) and 5000-3500 cal BP (n=17)) (Williams et al.

2015:51). I use scatterplots and r2 values to look at the bivariate relationship between CHAR and

SPD data and Correlation Coefficients to test the strength of those relationships.

6. Charcoal Analysis and Results

Sedimentation and Age Depth Model The entirety of the DIL core has a total of three sediment zones, described above (Letham

2016:182). The terminal date (15,013-13,859 cal BP, median: 14,380 cal BP) comes from small 35 twig fragments taken from ~3cm below the transition to sediment zone 2 and the limiting date

(4151-3809 cal BP (87.6%) or 3803-3721 cal BP (7.8%), median: 3947 cal BP) for the core comes from a small twig fragment located 10 cm from the top of the core (Letham 2016:182). Although there are only two dates available there was no visible sedimentation change throughout the section of the core used thus the age depth-model indicates that 47 years are represented in each 1 cm3 sample suggesting an annual accumulation rate of 0.021 cm.

The Charcoal Record Charcoal load for the Digby Island Lake core is low, consistent with expectations of charcoal records from coastal temperate rainforests (Gavin et al. 2003). CHAR is discussed for the entire timespan represented by the Digby Island Lake core (13,206-3947 cal BP based on median dates) and for timeframe for which we have SPD based demographic models (6,000-3,500 cal BP).

The Philips Lagoon core did not produce enough charcoal to reconstruct fire history but is discussed briefly.

Sediments from the Philips Lagoon (PL) core did not yield enough charcoal to reconstruct fire history. Beginning at the top of the core (~13,000 cal BP) twenty-five consecutive 1cm3 samples produced only 4 pieces of charcoal total. Next, the sampling interval was expanded to 5 cm intervals to expedite processing time with the expectation that the sampling interval would return to consecutive 1cm3 samples if/when charcoal counts increased. No samples produced charcoal counts above 5 throughout the remainder of the core, a majority had none (n=134). That the PL core did not produce enough charcoal to reconstruct fire activity is not entirely unexpected.

PL is tidally influenced, given the 7.4m modern tidal range which mixes marine sediments and outflow sediments from the Skeena River with sediments from Philips Lagoons’ northern mainland inputs, even if charcoal had been present claims that they represent a local or even extra- local source would be tenuous at best. 36

The DIL core contained substantially more charcoal fragments than the PL core producing a ~9700-year charcoal record (13,206-3497 cal BP). The age depth-model indicates that each 1 cm3 sample represents ~47 years suggesting an annual accumulation rate of 0.021cm. Charcoal particles in the DIL core were exceptionally small, none were found in the 500µm sieves, all counts are from 125 µm. Mean charcoal counts for the DIL record is 9.84 (Median = 7). The mean count between 13,206-6000 cal BP is 11 (Median = 7), with a mean during the Early Period of Tsimshian history (6000-3500 cal BP) of 6.56 (Median = 5). With relatively low CHAR in the Digby Island

Lake core only three charcoal frequency classes (Table 4a) are defined. Table 4b splits the 1-49 frequency count category to better illustrate the distribution within that category. Just one sample had a charcoal count exceeding 50, the minimum count typically cited as evidence of a fire event

(Whitlock and Larsen 2001:85).

Table 4. Frequency of charcoal peaks. Charcoal Frequency of Charcoal Frequency of charcoal count charcoal Peaks ᵃ count Peaks ᵇ 0 26 0 26 1 - 49 180 1 – 25 166 >50 1 26-49 14 >50 1

Peak Identification Simple Count Method: CHAR is extremely low (n<4) or non-existent from the beginning of the record (~13,206 cal. BP) until an increase at ~11,600, just after the Younger Dryas (~12,900-

11,700 cal BP). The first of three periods in the Digby Island Lake record that record near significant increase in CHAR (n = 43) occurred between ~11,600-10,950 cal. BP. Between

~10,950-8780 cal. BP charcoal counts do not exceed 24, well below the threshold for discerning fire events from “background signals” in the charcoal record (Higuera et al. 2007).

There is a moderate increase in charcoal frequency beginning at ~8600 cal. BP (Figure 6) and this increase culminates in the records largest peak (n = 51, ~8318-8365 cal. BP), prior the 37 onset of the 8.2ka climatic cooling event (Thomas et al. 2007). The final, near significant event (n

= 47) occurs between 7519-7472 cal. BP, a period not associated with any major global climatic event. From 7472 cal BP until the end of the record, CHAR values are low and sporadic with a brief increase (n=34) at ~4600 cal BP.

With only one peak above 50 pieces per cm2, a conservative interpretation is that DIL has recorded only one fire event or period of increased fire. Located between 103-104 cm below the top of the core the peak occurred between ~8365-8318 cal BP (N=51). Low fire frequencies are not entirely surprising as time-since-last fire estimates for similar Cedar-hemlock, Hemlock-fir, and Sitka-spruce forests on Vancouver Island can exceed 12,000 years (Gavin et al. 2003:194).

Figure 5. Chart showing CHAR for Digby Island Lake. Red bars indicate peaks identified using the count method.

A less conservative view of the DIL charcoal record takes into account the fact that fires occurring in coastal temperate rain forests are likely to be of low-severity, producing and depositing less charcoal than fires that occur in drier, Douglas fir dominated forests (Gavin et al.

2003:187, Kelly et al. 2011). It is possible to interpret lower charcoal peaks in a “relative rather than absolute way,” such that when smaller peaks (N = 25-49) are bracketed by samples with zero 38 or very few counts they may reflect actual local fire events (Derr 2012:210). Thus, adjusting expectations of charcoal counts to better account for coastal temperate rainforest condition there have been approximately 3 fire events or periods with increased fire activity on Digby Island in the 9500-year period between 13,000-3500 cal. BP. These fires or periods of increased fire activity occurred between: 11,138-11,091 cal. BP (n=43, CHAR = 0.91), 8365-8318 cal. BP (n=51, CHAR

= 1.09), 7519-7472 cal. BP (n=47, CHAR = 1) (Appendix). An argument could be made for a final fire event between 4558-4605 cal BP (N = 34). However, the sample is not bracketed by samples with low counts but rather it falls in the middle of a brief period of increased CHAR relative to the preceding 2500 years. Excluding the potential peak at 4558-4605 cal BP, the three peaks present in the 9700-year DIL record indicate an mFRI (mean fire return interval) of 3200 years.

CharAnalysis: The DIL charcoal record has an acceptable global signal-to-noise index (SNI =

3.24) just above the suggested threshold cutoff of 3 (Kelly et al. 2011). Mean CHAR for the record is low, 0.21 pieces-cm-2yr-1, with similarly low mean background of 0.18 pieces-cm-2yr-1. Two temporal zones were designated for this analysis; Zone 1: spanning the Early Tsimshian Period

(6000-3500 cal BP) and Zone 2: Early Holocene Period (13,206-6000 cal BP). The local SNI for

Zone 1 is acceptable at 5.52, with mean CHAR of 0.14 pieces-cm-2yr-1 and mean background of

0.1143 pieces-cm-2yr-1. Zone 2 however, has a local SNI at 2.74, below the recommended threshold suggesting that peaks identified between 13,206-6000 cal BP may not be confidently separated from noise (Kelly et al. 2011), mean CHAR of 0.23 pieces-cm-2yr-1, and mean background of

0.2031 pieces-cm-2yr-1.

A total of 14 significant peaks were identified in the DIL record, 9 occur in Zone 2 and 4 in Zone 1 (Figure 7). The largest peaks occur in Zone 2 at 11,200 cal BP and between 8600-7400 cal BP. The mFRI for the entire record is 692 years. Zone 2 has an mFRI of 501 years. A single 39 mFRI could not be calculated for Zone 1 as the span between peaks are disparate enough that each interval became a distinct mFRI. Fire frequency is low throughout the record. Between 12,000-

7000 cal BP fire-frequency ranges from 1 to 1.5/1000yrs, the only exception being a slight increase to 2 fires per 1000yrs between 8-9000 cal BP. Fire frequency is 0-1/1000yrs between 7000-3500 cal BP except where two peaks occur between 4-5000 cal BP.

Figure 6. Digby Island Lake CHAR, background series, peaks, and peak threshold. The “+” symbol indicates peaks or fire events; gray dots are peaks that fail to pass the Poisson minimum-count criterion. Charcoal Morphology Charcoal morphologies were determined and assigned based on the morphotypes described above (Enache and Cumming 2006. The percentages of charcoal morphotypes for the DIL core are presented in Table 9. All seven of the morphotypes identified by Enache and Cumming (2006) were present in the Digby Island Lake core, thus their taxonomy is used here.

Morphotype C dominates the charcoal assemblage making up 46% of total charcoal.

Morphotype S is the next most common at 23% of the charcoal assemblage. Morphotypes C and

S represent different versions of the same potential fuel sources, e.g., wood and/or bark. These are the two most common morphotypes generally found in charcoal assemblages, Type C is often the only morphotype identified in assemblages with low charcoal concentrations (Enache and

Cumming 2006:285). 40

Types M and P represent leaf and shrub fuels and are present in the charcoal assemblage in lower quantities, 15% and 10% respectively. Type M has shown high statistical correlations with known fires that have burned within approximately 20 km of a study site (Enache and

Cumming 2006:287). Types B, D, and F occur infrequently in the assemblage making up 1%, 1%, and 4% respectively. Type F is thought to be produced by fragmentation of larger pieces of charcoal, although it may also represent grass and sedge fuels (Enache and Cummings 2006:290).

Type D may represent grasses, shrubs, or leaves.

Table 5. Count and percentage of charcoal by morphotype for the Digby Island Lake core and mean morphotypes and their percentages for Zones 1 & 2. Morphotype Type M Type P Type S Type C Type B Type D Type F Total Total 298 209 477 935 19 25 74 Percent 15 10 23 46 1 1 4 Zone 1 Mean 0.83 0.69 1.43 2.92 0.12 0.12 0.08 Percent 13 11 23 47 2 2 1 Zone 2 Mean 1.79 1.19 2.8 5.42 0.08 0.12 0.52 Percent 15 10 24 46 1 1 4

Following Derr (2012:221), I compared Morphotype frequencies between Zone 1 – 6000-

3500 cal BP and Zone 2 – 13,200-6000 cal BP to assess changes in fuel types between Zones

(Table 5). Proportionally, there is little difference in morphotype representation between the two periods. Type C dominates Zone 1 & 2 (47% and 46% respectively). Type S is the second most common morphotype in either zone, types M and P make up between 10% and 15%, while types

B, D, and F contribute marginally (Table 5).

Statistical Results Here, CHAR is compared to the SPD based demographic reconstruction (Edinborough et al. 2017) to assess the correlation if any between fire/environment and demography for the Prince 41

Rupert Harbour region between 6000-3500 cal BP. Visual assessment of the two datasets indicates little correlation beyond brief, concurrent increases between 5100-5000 cal BP, 3700-3600 cal BP, and potential negative correlations at 4700 cal BP, and at 4100 cal BP (Figure 7).

However, these apparent visual correlations prove untrue given the statistical analysis.

First, I assess the relationship between CHAR and the SPD for the complete overlap of the datasets

(6000-3500 cal BP). Then, following Williams et al. (2015) I test for statistical correlation between

CHAR and SPD in overlapping 2000-year intervals (51). Correlations were calculated using

Paleontological Statistics v.3.25 (Hammer et al. 2001).

The scatterplot of SPD (n=26) and CHAR (n=26) values show no linear relationship between SPD and CHAR from 6000-3500 cal BP (Figure 8a). The r2 value of 0.056 indicates a weak, if any relationship with only 5.6% of the variation in CHAR accounted for by the regression line. The Correlation Coefficient (0.236) verifies this weak relationship, CHAR and SPD are not correlated between 6000-3500 cal BP.

Figure 8b shows no relationship between SPD and CHAR for the 6000-4000 cal BP period.

The r2 value of 0.002 indicates no relationship, with only 1% of variation in CHAR accounted for by the regression line. The Correlation Coefficient (0.053) verifies a weak relationship, SPD and

CHAR for the 6000-4000 cal BP period are not correlated. Figure 8c shows a weak, negative linear relationship for the 5000-3500 cal BP period. The r2 value of 0.019 indicates a weak relationship.

The Correlation Coefficient (-0.140) confirms the weak correlation between SPD and CHAR for the 5000-3500 cal BP period. 42

Figure 7. CHAR (200-year bins) and SPD (200-year bins) for Prince Rupert Harbour, 6000 cal BP to 3500 cal BP.

Figure 8. Scatterplots of SPD and CHAR for Prince Rupert Harbour. (a) SPD and CHAR 6000 cal BP to 3500 cal BP, r2=0.056. (b) SPD and CHAR during the 6000 to 4000 BP period, r2=0.002. (c) SPD and CHAR during the 5000 cal BP to 3500 cal BP period for Prince Rupert Harbour. 7. Discussion

In this section I interpret the results of the fire history as they correspond to previously summarized paleoenvironmental records and answer the remaining research questions described above. (1) What does a fire history of DIL look like and what can charcoal morphology tell us about fire episodes, if any exist? (2) Is there any correlation between the human history -as described by the SPD and general archaeological trends- of PRH and the charcoal record? and (3)

Is there any evidence of an anthropogenic signal in the fire history for DIL?

Two methods were used to identify fire events in the DIL charcoal record, (1) the count method and (2) the CharAnalysis program. Although the two methods offer conflicting results in terms of timing and frequency of possible fire events both assessments provide analogous 43 interpretations of climate histories as seen through the lens of fire. Although fires have occurred, it seems unlikely that the Digby Island areas experienced climatic fluctuation severe enough to create prolonged and repeated dry seasons. While regional environmental data illustrate changing vegetation and temperature and moisture fluctuations, the area has likely been perpetually damp and not prone to much more than the occasional natural fire event between 13,200-3500 cal BP.

Fire History The CharAnalysis program identified nine significant peaks (Figure 6) in Zone 2 (13,200-

6000 cal BP). However, the SNI (2.74) is below the recommended threshold of three meaning that peaks cannot be confidently separated from the noise and thus should be interpreted with caution or not interpreted at all (Higuera et al. 2010, Kelly et al 2011).

Types S and C combined make up 69% of morphotypes in Zone 2, as expected in charcoal records with low counts (Enache and Cumming 2006). These types represent woody fuels likely produced by high intensity fires. With no confidently identified peaks in Zone 2 types S and C are interpreted as extra-local or regional inputs. The presence of M and P while low (26%) in Zone 2 is interesting. These types represent leaf and grass fuels, their fragile structure suggests they do not survive secondary transport well and are typically interpreted as evidence of local fire. Their presence could be the result of small fires undetected by peak analysis, post-depositional breakage including in the lab, or misidentification. At present, no interpretation beyond speculation is possible, I suggest additional research to assess whether similar counts are replicated.

Given the low SNI in Zone 2 and overall low CHAR throughout the DIL record it is safe to assume that if fires did occur on or near Digby Island at this time they were of low intensity.

The characteristics of low intensity fires add an additional factor undermining the validity of peaks detected in Zone 2. Low intensity fires produce small quantities of charcoal and smaller, more fragile pieces of charcoal such that a low severity fire signal is indistinguishable from the 44 background variation (Higuera et al. 2005). Peak detection is most reliable when fires are large and of high severity (Clark 1988b, Higuera et al. 2008, 2010), given the low SNI and low quantities of charcoal Zone 2 appears unsuitable for peak detection and we reject the peaks identified.

Although we reject the peaks identified in Zone 2, it is worth discussing the background series as a reflection of a regional signal. The background series from Zone 2 is relatively synchronous with early and mid- Holocene trends in biomass burning for NWNA described in

Marlon et al. (2013). Both the DIL record and NWNA datasets show high accumulation in the early Holocene (~12-10kya), followed by a sharp decline then, a steady rise (Marlon et al. 2013).

The increase in background series beginning at 11,650 cal BP coincide with the abrupt warming at the end of the YD (Vaiu et al. 2006) and is seen in both continental (Marlon et al. 2009) and regional NWNA (Marlon et al. 2013) paleofire records. Post-YD warming continues throughout the EPG (11,500 cal BP-8900 cal BP) while moisture and precipitation fluctuate (Turunen and

Turunen 2003, Vaiu et al. 2006). Biomass burning in NWNA continues to increase during the EPG

(Marlon et al. 2013) as does the background series from the DIL record.

The DIL record and NWNA biomass burning series diverge after ~8400 cal BP when the

DIL background series peaks and then declines steadily until the end of Zone 2 (6000 cal BP), whereas the NWNA series grows steadily until 5000 cal BP and levels off thereafter (Marlon et al

2013). The divergence of the DIL and regional signal may be reflective of conditions specific to

NWC environments. The onset of the Hypsithermal marks the beginning of wetland expansion and a further increase in moisture. A Pinus contorta-Tsuga heterophylla-Cupressaceae-Sphagnum bog woodland complex is established in PRH by ~7500 cal BP (Turunen and Turunen 2003). After a brief warm period between ~7500-7000 cal BP temperatures cool, moisture increases, and modern climates and forests begin to establish on the NWC (Banner et al. 1983, Heusser 1960, 45

1985, McLaren 2008, Turunen and Turunen 2003). This trend towards increased moisture, cooler temperatures, and wetland expansion is seen in pollen profiles throughout the northwestern

Canadian coast, thus it is plausible that the post-8400 cal BP decline in background signal in the

DIL core reflects semi-local trends in coastal forest conditions.

Zone 2 summary: Although charcoal is present throughout Zone 2 the peaks identified by

CharAnalysis cannot reliably be interpreted as representing local fire events. The background series follows regional and continental scale trends in biomass burning from ~13,200-8400 cal BP after which it declines in the DIL core while continuing to rise gradually in regional syntheses

(Marlon et al. 2009, 2013). That regional syntheses are combinations of dozens of datasets, variation in individual datasets should be expected and it is likely that the decline seen in the DIL core is reflective of more local conditions. Additionally, it is certainly plausible that early

Holocene fires did occur in the PRH region despite the rejection of peaks identified here. North of

PRH on Prince of Whales Island, AK charcoal and vitrified sands dating between ~10,000-10,500 cal BP were identified (Carlson 2007), and several fire events dating between ~10,500-8000 cal

BP have been identified on Hecate Island to the south (Hoffman et al. 2016). While this study has not confidently identified fire events in the DIL between ~13,200-6000 cal BP, we do not rule out the possibility of fire in the vicinity. However, further testing is needed to confirm these results.

Zone 1: The SNI for Zone 1 (~6000-3500 cal BP) is 5.52, well above the acceptable threshold. Four significant peaks were identified in Zone 1 (~6000-3500 cal BP), and with an acceptable SNI (5.52) I discuss the potential fire events below. The earliest peak at 5968-5921 cal

BP contained a raw count of 12, the accepted threshold to identify fire events for raw counts is 50 pieces, so this peak is viewed with skepticism. The remaining three, while lower than 50, are discussed as though they are empirically real given their identification by the CharAnalysis 46 program. Paleoenvironmental data from PRH and surrounding areas during this period is quite coarse-grained (Table 1), generally speaking, the entire 2.5k year period is described as gradually cooling, with modern climates and vegetation establishing between 6000-3500 cal BP.

Morphotypes S and C were again the most common (70%) here (Table 5). What is most interesting in Zone 1 is the higher counts of types M and P found in the samples representing the two of the peaks flagged by CHARanalysis (~4500 and ~4400 cal BP). Although total counts are relatively low (n=29 and 19 respectively) these are the only samples where types M and P are the most abundant morphotype. In the ~4500 cal BP peak type M is 53% and type P is 42% of the sample. In the ~4400 cal BP peak, typ M is 45% and type P is 34% of the sample. These highly fragile types are interpreted as having near shore origins (Enache and Cumming 2006). That these peaks were both flagged by CHARanalysis program and both have distinctly different morphotype ratios suggests that these are in fact real and local fire events. Although there is little that can be said beyond stating that fires happened at these times, this does confirm that this method can be successfully applied in coastal temperate rainforests.

Summary: Without finer grained paleoenvironmental data there is little to say about the effects of climate on fire between 6000-3500 cal BP in PRH. Cool and wet environments are not typically conducive to fire events, yet it is possible that a few dry seasons occurred over that 2500- year span. Coastal temperate rainforests produce ample amounts of fuel but without prolonged dry seasons they burn infrequently. Lightning is extremely rare in the PRH region and was likely rare in the past as well, but its occurrence is plausible. Taken together, fuel for wildfires would have been available between 6000-3500 cal BP and so it would only take a few drier than usual years and a natural or cultural ignition source to account for the fire events identified in Zone 1.

Settlement Patterns, Demography, and Fire History at PRH 47

I tested the relationship between CHAR and SPD based demographic trends to address two questions (1) is there an anthropogenic signal in the fire activity between 6000-3500 cal BP? and

(2) is there any shift in fire regimes that may reflect changes in climate that could have affected demography between 6000-3500 cal BP? No correlation was found for the three periods tested

(6000-3500 cal BP, 6000-4000 cal BP, 5000-3500 cal BP), it does not appear that fire and human activity affected one another at any scale detectable via these methods. There is however and interesting visual (not statistical) correlation between 2 fire events occurring between 4500-4400 cal BP and the brief rise in the SPD between 5000-4500 cal BP (Figure 7 & 9).

5000 cal BP marks the beginning of an increase in the density of small shell bearing sites and small villages (Martindale et al. 2017b). Eight sites are occupied within PRH at this time, although this number likely underrepresents sites from this time period due to sampling bias related to higher sea levels (Letham et al. 2020). If fires are anthropogenic in origin it is expected that the archaeological data should lead the fire record (Williams et al. 2015). This expectation is based on the idea that as population increases so too does the need for both space and resources. That need could promote the use of fire to either clear space for living/work area or to create space for and increase production of fire tolerant resources. Ethnographically fire was not known to be used by the coastal Tsimshian but its use has been documented further east in Gitxsan territory (Boyd 1999,

Johnson-Gottesfeld 1994).

While this apparent correlation between a population increase and subsequent fire events is intriguing, the less-than-robust data necessitates relatively large interpretive leaps needed to argue that the pattern is real. It is here, between 5000-3500 that the archaeological record begins to indicate a shift in settlement patterns concentrating in the harbour (Martindale et al. 2017b), but 48 it is not until 3500 that there is ample evidence to argue for an anthropogenic cause for these fire events. Additionally, although these peaks were flagged as real events by the CharAnalysis

Figure 9. SPD and unbinned CHAR. Summed probability distribution of terrestrial radiocarbon dates from Prince Rupert Harbour, Dundas Islands, and Stevens Island. IntCal13 calibration curve used. 200-year bins. Gray shaded area is the SPD, red dashed line is the SPD smoothed 200-year moving average. The dark grey line is CHAR.

program, their low raw counts necessitate some skepticism in their interpretation. I suggest that these results should be treated as an intriguing pilot study rather than definitive proof of an anthropogenic signal. Additional data is needed, and this signal should be investigated further as I will discuss below.

8. Conclusion

This study presents a 9700-year fire history extrapolated from the sedimentary charcoal record of Digby Island Lake, Prince Rupert Harbour. The DIL record provides northernmost fire history in Coastal British Columbia providing unique insight into CWHvh2 forest dynamics as related to fire. In addition, this high-resolution, localized fire history provides an additional line of data to better understand the ecological and cultural evolution of the Tsimshianic world of PRH. 49

This analysis did not confidently identify any local fire events in the DIL paleofire record between 13,200-6000 cal BP, suggesting that local conditions at this time were not conducive to natural fire events. Fire does not directly reflect climate/environment but does provide proxy evidence for changes in moisture and temperature. The absence of fire between 13,200-6000 cal

BP suggests that changes from warm-wet to cool-dry to warm-dry seen in local and regional pollen profiles were not severe enough to alter fire regimes. Four peaks were identified between 6000-

3500 cal BP, although the earliest peak is questionable due to low raw counts. Charcoal morphology suggests that the latter three peaks were the result of local fires. No statistical correlations suggesting a causal relationship between demography and fire were found in the DIL record but the 2 peaks between 4500-4400 cal BP occurring shortly after a spike in the SPD between 5000-4500 cal BP presents an interesting pattern worth exploring in future analyses.

There was a second reason for testing the relationship between SPD and CHAR. With the hypothesis that as populations increased along with village size and density, resource use and corresponding ecological footprint would also increase I posed the question: Does CHAR track demographic change thus providing a proxy narrative of the history of the Tsimshian people? The answer provided by the DIL data is no. However, the question could not be answered in full due to limitations of the data. The cores used in this analysis were taken for RSL reconstruction. The loose and waterlogged sediments that contain the 3500 cal BP to present sediments were discarded during extraction leaving behind the CHAR record for the last 5.5k years. Unfortunately, the time period lost encapsulates the most extraordinary changes in settlement pattern and increases in demography known for PRH. Thus, while the tentative answer to the question posed above is no, it does not preclude the possibility of revisiting the question in the future. 50

This study did not pose or answer questions regarding causal mechanisms in the development of cultural complexity. Rather, the intention was to add an additional line of paleoenvironmental evidence that can be used to better understand potential and suggested drivers of cultural change on the NWC and in PRH specifically. While climate/environment is not necessarily the prime explanation of the development of cultural complexity it has been suggested as playing a significant contributing role (Fladmark 1975) with varying degrees of emphasis

(Prentiss and Chatters 2003). Fire regimes are not a direct mirror of vegetation or climate changes however, outside of human impacts, the former is effectively controlled by the latter. Extreme changes in the fire regime that are not explainable by human agency are explained by climate.

Issues that arose during this analysis can help inform future paleoenvironmental work in

PRH. First, the attempt to recreate Derr’s (2012) analysis of anthropogenic burning by comparing

‘on-site’ and ‘off-site’ sediment cores was thwarted by the absence of charcoal in the Philips

Lagoon core. Future endeavors should establish a coring program/sampling strategy geared specifically towards answering these questions. A series of sediment cores should be taken to create a more robust composite fire history and to better answer questions about human- environment interactions and environmental history in general.

Bibliography

Ames, Kenneth M 2005 The North Coast Prehistory Project Excavations in Prince Rupert Harbour, British Columbia: The Artifacts. Portland State University.

Ames, Kenneth M, and Andrew Martindale 2014 Rope Bridges and Cables: A Synthesis of Prince Rupert Harbour Archaeology. Canadian Journal of Archaeology 38:140–178.

Banner, A., Pojar, J., Rouse, G. E. 1983 Postglacial Paleoecology and Successional Relationships of a Bog Woodland Near Prince Rupert, British Columbia. Canadian Journal of Forest Research 13(5):938–947.

Boyd, Robert T 1999 Indians, Fire, and the Land in the Pacific Northwest. 1st ed. Oregon State University, Corvallis.

Brewster, Natalie, and Andrew Martindale 2011 An Archaeological History of Holocene Fish Use in the Dundas Island Group, British Columbia. In The Archaeology of North Pacific Fisheries, edited by Madonna L. Moss and Aubrey Cannon, pp. 247–264. 1st ed. University of Alaska, Fairbanks.

Brown, K. J., and R. J. Hebda 2003 Coastal Rainforest Connections Disclosed Through a Late Quaternary Vegetation, Climate, and Fire History Investigation From the Mountain Hemlock Zone on Southern Vancouver Island, British Colombia, Canada. Review of Palaeobotany and Palynology 123(3–4):247–269. DOI:10.1016/S0034-6667(02)00195-1.

Brown, Kendrick J., and Richard J. Hebda 2002 Ancient Fires on Southern Vancouver Island, British Columbia, Canada: A Change in Causal Mechanisms at About 2,000 BP. Environmental Archaeology 7(March):1–12. DOI:10.1179/env.2002.7.1.1.

Brown, Thomas Jay 2016 Demography and the Evolution of Logistic Organization on the Northern Northwest Coast Between 11,000 and 5,000 cal BP. Portland State University.

Burchell, Meghan, Nadine Hallmann, Andrew Martindale, Aubrey Cannon, and Bernd R. Schöne 2013 Seasonality and Intensity of Shellfish Harvesting on the North Coast of British Columbia. Journal of Island and Coastal Archaeology 8(2):152–169. DOI:10.1080/15564894.2013.787566. 52

Carlson, Risa J. 2007 Current Models for the Human Colonization of the Americas: The Evidence From Southeast Alaska. Cambridge.

Coupland, Gary 1996 The Evolution of Multi-Family Households on the Northwest Coast of North America. In People Who Lived In Big Houses, Archaeological Perspectives On Large Domestic Structures, pp. 121–130.

Derr, Kelly M. 2012 Intensifying With Fire: Floral Resource Use and Landscape Management By the Precontact Coast Salish of Southwestern British Columbia. Washington State University.

2014 Anthropogenic Fire and Landscape Management on Valdes Island, Southwestern BC. Canadian Journal Of Archaeology 38:250–279.

Edinborough, Kevan, Marko Porcic, and Kenneth M Ames 2017 Radiocarbon Test For Demographic Events in Written and Oral History 114(47). DOI:10.1073/pnas.1713012114.

Enache, Mihaela D., and Brian F. Cumming 2007 Charcoal morphotypes in lake sediments from British Columbia (Canada): An assessment of their utility for the reconstruction of past fire and precipitation. Journal of Paleolimnology 38(3):347–363. DOI:10.1007/s10933-006-9084-8.

Enache, Mihaela D, and Brian F Cumming 2006 Tracking Recorded Fires Using Charcoal Morphology From the Sedimentary Sequence of Prosser Lake , British Columbia (Canada). Quaternary Research 65:282–292. DOI:10.1016/j.yqres.2005.09.003.

Murphy, Sinead M 2016 5,000-year Fire History in the Strait of Georgia Lowlands , British Columbia : Implications for Restoration and Management. Simon Fraser University

Fedje, Daryl W., and Quentin Mackie 2005 Overview of Cultural History. In Haida Gwaii: Human History and Environment From the Time of the Loon to the Time of the Iron People, pp. 154–162. UBC Press, Vancouver.

Fedje, Daryl, Quentin Mackie, Terri Lacourse, and Duncan McLaren 2011 Younger Dryas environments and archaeology on the Northwest Coast of North America. Quaternary International 242(2):452–462. DOI:10.1016/j.quaint.2011.03.042.

Fladmark, Knut 1975 A Paleoecological Model For Northwest Coast Prehistory. Ottowa.

Garfield, Viola E. 1966 The Tsimshian Indians and Their Arts. Seattle. University of Washington.

Gavin, Daniel G., Linda B. Brubaker, and Kenneth P. Lertzman 2003 Holocene fire history of a coastal temperate rain forest based on soil charcoal radiocarbon dates. Ecology 84(1):186–201. DOI:10.1890/0012- 9658(2003)084[0186:HFHOAC]2.0.CO;2.

Gavin, Daniel G., Douglas J. Hallett, Sheng Hu Feng, et al. 2007 Forest fire and climate change in western North America: Insights from sediment charcoal records. Frontiers in Ecology and the Environment 5(9):499–506. DOI:10.1890/060161.

Gavin, Daniel G, Linda B Brubaker, and Kenneth P Lertzman 2003 Holocene Fire History of a Coastal Temperate Rain Forest Based on Soil Charcoal Radiocarbon Dates Published by : Ecological Society of America. America 84(1):186–201.

Gottersfeld, Leslie M. 1994 Aboriginal Burning for Vegetation Management in Northwest British Columbia. Ecology 22(2):171–188.

Hallmann, N., M. Burchell, N. Brewster, A. Martindale, and B. R. Schöne 2013 Holocene climate and seasonality of shell collection at the Dundas Islands Group, northern British Columbia, Canada-A bivalve sclerochronological approach. Palaeogeography, Palaeoclimatology, Palaeoecology 373:163–172. DOI:10.1016/j.palaeo.2011.12.019.

Hammer, Ø., D. A. Harper, and P. D. Ryan 1999 PAST PAleontological STatistics Reference manual. Palaeontologia electronica 4(1):9.

Heusser, C J 1960 Late-Pleistocene Environments of North Pacific North America; an Elaboration of Late-Glacial and Postglacial Climatic, Physiographic, and Biotic Changes. American Geographical Society, Burlington.

Heusser, C J, and Linda E Heusser 1980 Quarternary Temperatures and Precipitation for the North-west Coast of North America. Nature 286(14):702–704.

Higuera, Philip 2009 CharAnalysis 0.9 : Diagnostic and analytical tools for sediment-charcoal analysis. Montana State University, Bozeman, MT(April):1–27.

Higuera, Philip E., Linda B. Brubaker, Patricia M. Anderson, Thomas A. Brown, Alison T. Kennedy, and Feng Sheng Hu 2008 Frequent fires in ancient shrub tundra: Implications of paleorecords for arctic environmental change. PLoS ONE 3(3):1–7. DOI:10.1371/journal.pone.0001744.

Higuera, Philip E., Linda B. Brubaker, Patricia M. Anderson, Feng Sheng Hu, and Thomas A. Brown 2009 Vegetation mediated the impacts of postglacial climate change on fire regimes in the south-central Brooks Range, Alaska. Ecological Monographs 79(2):201–219. DOI:10.1890/07-2019.1.

Higuera, Philip E., Daniel G. Gavin, Patrick J. Bartlein, and Douglas J. Hallett 2010 Peak detection in sediment-charcoal records: Impacts of alternative data analysis methods on fire-history interpretations. International Journal of Wildland Fire 19(8):996–1014. DOI:10.1071/WF09134.

Higuera, Philip E, Matthew E Peters, Linda B Brubaker, and Daniel G Gavin 2007 Understanding the Origin and Analysis of Sediment-Charcoal Records With a Simulation Model. Quaternary Science Reviews 26:1790–1809. DOI:10.1016/j.quascirev.2007.03.010.

Hoffman, K. M., D. G. Gavin, K. P. Lertzman, D. J. Smith, and B. M. Starzomski 2016 13,000 years of fire history derived from soil charcoal in a British Columbia coastal temperate rain forest. Ecosphere 7(7):1–13. DOI:10.1002/ecs2.1415.

Hoffman, Kira M., Daniel G. Gavin, and Brian M. Starzomski 2016 Seven hundred years of human-driven and climate-influenced fire activity in a British Columbia coastal temperate rainforest. Royal Society Open Science 3(10). DOI:10.1098/rsos.160608.

Hoffman, Kira M., Ken P. Lertzman, and Brian M. Starzomski 2017 Ecological legacies of anthropogenic burning in a British Columbia coastal temperate rain forest. Journal of Biogeography 44(12):2903–2915. DOI:10.1111/jbi.13096.

Hong, Y. T., B. Hong, Q. H. Lin, Y. Shibata, Y. X. Zhu, X. T. Leng, and Y. Wang 2009 Synchronous climate anomalies in the western North Pacific and North Atlantic regions during the last 14,000 years. Quaternary Science Reviews 28(9–10):840– 849. DOI:10.1016/j.quascirev.2008.11.011.

Johnson, Raini A 2019 Inter- and Intra- site Heterogeneity as Sources For Faunal Assemblage Variabilty: An Analysis of Fish Taxa From Northern Tsimshian Archaeological Sites. University of British Columbia.

Kelly, Ryan F., Philip E. Higuera, Carolyn M. Barrett, and Feng Sheng Hu 2011 A signal-to-noise index to quantify the potential for peak detection in sediment- charcoal records. Quaternary Research 75(1):11–17. DOI:10.1016/j.yqres.2010.07.011.

Lepofsky, Dana, and Ken Lertzman 2008 Documenting Ancient Plant Management in the Northwest of North America. Botany 86:129–145. DOI:10.1139/B07-094.

Lepofsky, Dana, Ken Lertzman, Douglas Hallett, and Rolf Mathewes 2005 Society for American Archaeology Climate Change and Culture Change on the Southern Coast of British Columbia 2400-1200: An Hypothesis. American Antiquity 70(2):267–293.

Letham, Bryn 2017 Long-Term Human-Environment Interaction on Dynamic Coastal Landscapes: Examples From 15,000 Years of Shoreline and Settlement Change in the Prince Rupert Harbour Area. University of British Columbia.

Letham, Bryn, Andrew Martindale, and Kenneth M. Ames 2020 Endowment, investment, and the transforming coast: Long-term human- environment interactions and territorial proprietorship in the Prince Rupert Harbour, Canada. Journal of Anthropological Archaeology 59(May):101179. DOI:10.1016/j.jaa.2020.101179.

Letham, Bryn, Andrew Martindale, Rebecca Macdonald, Eric Guiry, Jacob Jones, and Kenneth M. Ames 2016 Postglacial relative sea-level history of the Prince Rupert area, British Columbia, Canada. Quaternary Science Reviews 153:156–191. DOI:10.1016/j.quascirev.2016.10.004.

Letham, Bryn, Andrew Martindale, Kisha Supernant, Thomas J. Brown, Jerome S. Cybulski, and Kenneth M. Ames 2019 Assessing the Scale and Pace of Large Shell-Bearing Site Occupation in the Prince Rupert Harbour Area, British Columbia. Journal of Island and Coastal Archaeology 14(2):163–197. DOI:10.1080/15564894.2017.1387621.

Letham, Bryn, Andrew Martindale, Nicholas Waber, and Kenneth M. Ames 2018 Archaeological Survey of Dynamic Coastal Landscapes and Paleoshorelines: Locating Early Holocene Sites in the Prince Rupert Harbour Area, British 56

Columbia, Canada. Journal of Field Archaeology 43(3):181–199. DOI:10.1080/00934690.2018.1441575.

Long, Colin J., Cathy Whitlock, Patrick J. Bartlein, and Sarah H. Millspaugh 1998 A 9000-year fire history from the Oregon Coast Range, based on a high- resolution charcoal study. Canadian Journal of Forest Research 28(5):774–787. DOI:10.1139/x98-051.

MacDonald, George F., and Richard I. Inglis 1981 An Overview of the North Coast Prehistory Project ( 1966-1980 ). BC Studies Winter(48):37–63.

Marlon, J. R., Bartlein P. J., M. K. Walsh, 2009 Wildfire responses to abrupt climate change in North America. Proceedings of the National Academy of Sciences of the United States of America 106(8):2519–2524. DOI:10.1073/pnas.0808212106.

Marlon, Jennifer R., Patrick J. Bartlein, Anne Laure Daniau, Sandy P. Harrison, Shira Y. Maezumi, Mitchell J. Power, Willy Tinner, and Boris Vanniére 2013 Global biomass burning: A synthesis and review of Holocene paleofire records and their controls. Quaternary Science Reviews 65:5–25. DOI:10.1016/j.quascirev.2012.11.029.

Marlon, Jennifer, Patrick J. Bartlein, and Cathy Whitlock 2006 Fire-fuel-climate linkages in the northwestern USA during the Holocene. Holocene 16(8):1059–1071. DOI:10.1177/0959683606069396.

Marsden, Susan 2002 Adawx, Spanoxnox, and the Geopoltics of the Tsimshian. BC Studies(135).

Martindale, Andrew 2006 Tsimshian Houses and Households through the Contact Period. Household Archaeology on the Northwest Coast:140–158. DOI:10.2307/j.ctv8bt3gt.12.

Martindale, Andrew R C, and Susan Marsden 2003 Defining the Middle Period (3500 bp to 1500 bp) in Tsimshian History through a Comparison of Archaeological and Oral Records. BC Studies: The British Columbian Quarterly 0(138/9):13–50. DOI:10.14288/bcs.v0i138/9.1670.g1715.

Martindale, Andrew, Bryn Letham, Kisha Supernant, T. J. Brown, Kevan Edinborough, Jonathan Duelks, and Kenneth M. Ames 2017b Urbanism in Northern Tsimshian archaeology. Hunter Gatherer Research 3(1):133–163. DOI:10.3828/hgr.2017.8.

Martindale, Andrew, Susan Marsden, Katherine Patton, Angela Ruggles, Bryn Letham, Kisha Supernant, David Archer, Duncan McLaren, and Kenneth M. Ames 2017a The role of small villages in Northern Tsimshian territory from oral and archaeological records. Journal of Social Archaeology. Vol. 17.

Mathewes, Rolf W. 1993 Evidence for Younger Dryas-age cooling on the North Pacific coast of America. Quaternary Science Reviews 12(5):321–331. DOI:10.1016/0277-3791(93)90040- S.

Mathewes, Rolf W, and Linda E Heusser 1981 A 12,000 Year Palynological Record of Temperature and Precipitation Trends in Southwestern British Columbia. Canadian Journal of Botany 59.

Mclaren, Duncan 2008 Sea Level Change and Archaeological Site Locations on the Dundas Island Archipelago of North Coastal British Columbia. University of Victoria.

Mooney, Sd, and W Tinner 2011 The analysis of charcoal in peat and organic sediments. Mires and Peat 7:1–18.

Moss, Madonna L., Peteet, Dorothy M., Whitlock, Cathy 2007 Mid-Holocene Culture and Climate on the Northwest Coast of North America. In Climate Change and Cultural Dynamics : A Global Perspective on Mid-Holocene Transitions, edited by Daniel H. Anderson, David G., Maasch, Kirk, and Sandweiss, pp. 491–529.

McDonald, George F., and Richard I. Inglis 1980 An Overview of the North Coast Prehistory Project (1966-1980). BC Studies: The British Columbian Quarterly 0(48):37–63. DOI:10.14288/bcs.v0i48.1078.

Patton, A. Katherine, Andrew Martindale, Trevor J. Orchard, Sage Vanier, and Gary Coupland 2019 Finding eulachon: The use and cultural importance of Thaleichthys pacificus on the northern Northwest Coast of North America. Journal of Archaeological Science: Reports 23(December 2018):687–699. DOI:10.1016/j.jasrep.2018.11.033.

Power, M J, J R Marlon, P J Bartlein, and S P Harrison 2010 Fire history and the Global Charcoal Database : A New Tool for Hypothesis Testing and Data Exploration. Palaeogeography, Palaeoclimatology, Palaeoecology 291(1–2):52–59. DOI:10.1016/j.palaeo.2009.09.014.

Prentiss, William C., and James C. Chatters 2003 Cultural Diversification and Decimation in the Prehistoric Record. Current Anthropology 44(1):33–58. DOI:10.1086/344488.

Ramsey, Christopher Bronk 2001 Development of the Radiocarbon Calibration Program. Radiocarbon 43(2 PART I):355–363. DOI:10.1017/s0033822200038212.

Ramsey, Christopher Bronk 2009 Bayesian Analysis of Radiocarbon Dates. Radiocarbon 51(1):337–360. DOI:10.1017/s0033822200033865.

Reimer, Paula J, Edouard Bard, Alex Bayliss, J Warren Beck, Paul G Blackwell, Christopher Bronk, Ramsey Caitlin, E Buck Hai, and R Lawrence Edwards 2013 Intcal13 and Marine13 Radiocarbon Age Calibration curves 0 – 50,000 years cal bp. Radiocarbon 55(4).

Renssen, Hans, Aurelien Mairesse, Hugues Goosse, Pierre Mathiot, Oliver Heiri, Didier M. Roche, Kerim H. Nisancioglu, and Paul J. Valdes 2015 Multiple Causes of the Younger Dryas Cold eriPod. Nature Geoscience 8(12):946–949. DOI:10.1038/ngeo2557.

Rick, John W 1987 Society for American Archaeology Dates as Data: An Examination of the Peruvian Preceramic Radiocarbon Record. American Antiquity 52122126(1):55– 73.

Shennan, Stephen, Sean S. Downey, Adrian Timpson, Kevan Edinborough, Sue Colledge, Tim Kerig, Katie Manning, and Mark G. Thomas 2013 Regional Population Collapse Followed Initial Agriculture Booms in Mid- Holocene Europe. Nature Communications 4:1–8. DOI:10.1038/ncomms3486.

Stewart, Frances L, and Kathlyn M Stewart 1996 The Boardwalk and Grassy Bay Sites: Patterns of Seasonality and Subsistence on the Northern Northwest Cost of B.C. Canadian Journal of Archaeology 20:1996

Stuvier, Minze, Pieter M Grootes, and Thomas F Braziunas 1995 The GISP2 δ18O Climate Record of the Past 16,500 Years and the Role of the Sun, Ocean, and Volcanoes. Quaternary Research 44(3):341–354.

Surovell, Todd A., Judson Byrd Finley, Geoffrey M. Smith, P. Jeffrey Brantingham, and Robert Kelly 2009 Correcting temporal frequency distributions for taphonomic bias. Journal of Archaeological Science 36(8):1715–1724. DOI:10.1016/j.jas.2009.03.029.

Thomas, Elizabeth R., Eric W. Wolff, Robert Mulvaney, Jorgen P. Steffensen, Sigfus J. Johnsen, Carol Arrowsmith, James W.C. White, Bruce Vaughn, and Trevor Popp 2007 The 8.2 ka event from Greenland ice cores. Quaternary Science Reviews 26(1– 2):70–81. DOI:10.1016/j.quascirev.2006.07.017.

Turunen, Carrie, and Jukka Turunen 2003 Development History and Carbon Accumulation of a Slope Bog in Oceanic British Columbia, Canada. The Holocene 13(2):225–238. DOI:10.1191/0959683603hl609rp.

Viau, A. E., K. Gajewski, M. C. Sawada, and P. Fines 2006 Millennial-scale temperature variations in North America during the Holocene. Journal of Geophysical Research Atmospheres 111(9):1–12. DOI:10.1029/2005JD006031.

Walsh, Megan K., Cathy Whitlock, and Patrick J. Bartlein 2008 A 14,300-year-long record of fire-vegetation-climate linkages at Battle Ground Lake, southwestern Washington. Quaternary Research 70(2):251–264. DOI:10.1016/j.yqres.2008.05.002.

Walsh, Megan K, Jennifer R Marlon, et al. 2015 A Regional Perspective on Holocene Fire – Climate – Human Interactions in the Pacific Northwest of North America A Regional Perspective on Holocene Fire – Climate – Human Interactions in the Pacific Northwest of North America. Annals of the Association of American Geographers 105(6):1135–1157. DOI:10.1080/00045608.2015.1064457.

Whitlock, Cathy, and Chris Larsen 2001 Charcoal as a Fire Proxy. Tracking environmental change using lake sediments Vol 3 Terrestrial algal and siliceous indicators 3:75–97. DOI:10.1007/0-306- 47668-1_5.

Williams, Alan N, Scott D Mooney, Scott A Sisson, and Jennifer Marlon 2015 Exploring the relationship between Aboriginal population indices and fi re in Australia over the last 20,000 years. Palaeogeography, Palaeoclimatology, Palaeoecology 432:49–57. DOI:10.1016/j.palaeo.2015.04.030.

Willis, Samuel C., and Matthew R. Des Lauriers 2011 Early Technological Organization Along the Eastern Pacific Rim of the New World: A Co-Continental View:117–136. DOI:10.1007/978-1-4419-8219-3_5.

Appendices

Appendix A. Supplemental Tables

Number of sites and dates by region used to produce the SPD. Data Origin Number of Sites Number of 14C Dates PRH 23 163 Dundas 10 24 Stephens 2 11 Total 35 198

Sites with components dated to the Early Holocene Period (13,200 – 6000 cal BP ). Site Cal age BP Cal age BP Median Cal Location Number from to Age BP GcTr-8 Dundas 7412 7118 7257 GcTq-5 Dundas 9385 9010 9195 GdTq-1 Dundas 6550 6162 6338 GcTr-6 Dundas 11204 10885 11148 GcTq-2 Dundas 7415 7267 7314 GdTq-3 Dundas 7274 6970 7141 GcTq-4 Dundas 7250 6875 7069 GbTp-1 Dundas 9198 8800 9009 GbTo-23 Prince Rupert Harbour 6872 6446 6657 GbTo-185 Prince Rupert Harbour 9304 9028 9186 GbTo-18 Prince Rupert Harbour 6658 6003 6356 GbTo-82 Prince Rupert Harbour 6728 6463 6596 GcTo-67 Prince Rupert Harbour 6635 6445 6527 GaTp-10 Stephens 9629 9423 9509 GaTp-7a Stephens 7595 7494 7547 T416-1 Stephens 7348 7048 7209

Date ranges and charcoal counts for three potential fire events in the Digby Island Lake Core. Estimated Age Range Charcoal Count 7472 – 7519 cal. BP 47 8318 – 8365 cal. BP 51 11091 – 11138 cal. BP 43

Summary table of environmental, archaeological, and CHAR data. N. Amer. Temp Time cal Environmental Major Climate Environment/Climate (Vaiu et al. Fire Summary Culture Phase/Change BP Phase Event 2006) 13,500

13,000 Colder than previous 12,500 Late Glacial no fire activity 1.5 ky, Younger Dryas 3-4 degree rapid 12,000 12,900 to 11,700 increase No known sites date to cal BP ("broadly this period except a 11,500 resembles single date from GcTr-6 11,000 Post YD warming, a GISP2 data) ~11,000 cal BP 10,500 cool & wet climate Early Post Glacial 10,000 (EPG) 11,500 - low and sporadic 8900 cal BP Cool and Dry 10k - 9k 9,500 (high=21) 9,000 Wetter, cooler 1 degree slowed 8,500 Wetter & cooler, increase wetland expansions most Fire activity 8,000 8.2k cold event Pre-Early Period 9000 to peaks at 8500, 8300, 6000 cal BP 7,500 7500 cal BP 7,000 Warm period Cooling (Initially rapid, 6,500 Hypsithermal then gradual) low point, sporadic 6,000 8900 - 3800 cal BP Gradual cooling (high=12) 5,500 towards modern 5,000 temps, terrestrial Early Period 6000 to productivity recovers 3500 cal BP 4,500 Warming ~6800 cal BP low but spike 4,000 (Gradual) ~4500/4400 and a 3,500 small increase in last sample.

Appendix B. CHARanalysis Output

CHAR analysis.

Peak Sensitivity 63

Threshold determination details.