Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Drought influences mixed-severity fire regimes across temporal and spatial scales in the Montane Cordillera… Chavardès, Raphaël Daniel 2019

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

Item Metadata

Download

Media
24-ubc_2019_september_chavardes_raphael.pdf [ 4.92MB ]
Metadata
JSON: 24-1.0380575.json
JSON-LD: 24-1.0380575-ld.json
RDF/XML (Pretty): 24-1.0380575-rdf.xml
RDF/JSON: 24-1.0380575-rdf.json
Turtle: 24-1.0380575-turtle.txt
N-Triples: 24-1.0380575-rdf-ntriples.txt
Original Record: 24-1.0380575-source.json
Full Text
24-1.0380575-fulltext.txt
Citation
24-1.0380575.ris

Full Text

 DROUGHT INFLUENCES MIXED-SEVERITY FIRE REGIMES ACROSS TEMPORAL AND SPATIAL SCALES IN THE MONTANE CORDILLERA OF CANADA by  Raphaël Daniel Chavardès B.Sc., The University of York, 2000 B.A., The University of British Columbia, 2011 M.Sc., The University of British Columbia, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2019  © Raphaël Daniel Chavardès, 2019 ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Drought influences mixed-severity fire regimes across temporal and spatial scales in the Montane Cordillera of Canada  submitted by Raphaël Chavardès in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Forestry  Examining Committee: Lori Daniels, Forestry Supervisor  Ze’ev Gedalof, Geography Supervisory Committee Member  Bianca Eskelson, Forestry Supervisory Committee Member Peter Marshall, Forestry University Examiner Brett Eaton, Geography University Examiner  Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member  iii  Abstract Understanding historical fire-drought associations, particularly in forests with mixed-severity fire regimes, is a research and fire management priority in western North America. My thesis investigates how drought variation across temporal and spatial scales drove such fire regimes in the Montane Cordillera of Canada. I developed three-interrelated studies written as independent chapters, all of which used crossdated fire-scars to represent historical fire years. The first two studies test fire-drought associations using monthly adaptations of the Drought Code (DC) from Canada’s Fire Weather Index System. First, I compared three monthly drought codes during the 20th and 21st centuries for montane forests of southeast British Columbia. Accuracy of monthly DC increased after accounting for overwinter drying, early fire season starts, and effective precipitation. June−August drought codes were significantly associated with historical fires. Variation in fire-season drought influenced fire severity, connecting modern fire-weather indices with historical mixed-severity fire regimes. Second, I investigated how historical drought variation drove mixed-severity fire regimes in the same location by developing a tree-ring proxy reconstruction of summer DC. Comparing summer DC against a local summer Palmer Drought Severity Index provided a nuanced understanding of inter-annual fire-drought associations and moisture content among forest fuels, namely in deep compact organics in the soil and large woody fuels, versus the duff layer. Fire years were associated with coinciding and previous year summer drought; but limited by coinciding and previous year summer wet conditions. Summer moisture conditions during fire years likely influenced ignitions and led to variable combustion of forest fuels. The final study encompassed broader spatial coverage by including 17 fire-history sites across the Montane Cordillera, and by testing historical associations between climate and fire based on years with evidence of fire at multiple sites, i.e., iv  fire synchrony. Fire synchrony was historically common, and associated with droughts at regional and subregional scales based on tree-ring proxy reconstructions of climate. My thesis provides information on drought as a driver of mixed-severity fire regimes across temporal and spatial scales. Ultimately, understanding how drought drove mixed-severity fire regimes across scales, helps fire managers anticipate how these fire regimes are shifting due to climate change. v  Lay Summary My research explores how drought influenced fires over time and space in western Canadian forests. I used weather and climate records, statistical models, tree-ring science, and fire-scar records to understand historical associations between droughts and fire. Using a suite of monthly drought indices, I found the onset, duration and degree of drought during the fire season influenced fire severity in montane forests of southeastern British Columbia. For these same forests, I used two different multi-century tree-ring reconstructions of summer drought, which represent moisture content in different forest fuels. I determined that the persistence and degree of drought influenced fire behaviour. Using networks of drought reconstructions and fire records from multiple sites across Central and Interior BC and western Alberta, I found that fires burning at multiple sites were common and occurred during droughts. Ultimately, my research helps fire managers foresee how future fires can be impacted by climate change.    vi  Preface Raphaël Chavardès was the main contributor to the identification, design, analyses and writing of this PhD thesis. The thesis was supported throughout by Drs. Lori Daniels, Ze’ev Gedalof, and Bianca Eskelson. Data were contributed by: Dr. Lori Daniels (fire-scar records in Chapters 2–4; tree cores in Chapter 3); Dr. Ze’ev Gedalof and Vesta Mather (scanned tree-core images in Chapter 3); Dr. Emma Watson (annual precipitation reconstructions in Chapter 4); Drs. Emily Heyerdahl and Jill Harvey, Rick Kubian, Gregory Greene, Hélène Marcoux, Alexandra Pogue, Eric Da Silva,  Jed Cochrane, John Nesbitt, Theresa Dinh, Wesley Brookes, Olivier Villemaire-Côté, and Raphaël Chavardès (fire-scar records in Chapter 4). Supporting contributions included those by Dr. Paul Pickell (developing the initial R script to calculate the Monthly Drought Code, which Raphaël Chavardès modified and further developed into the adjusted Monthly Drought Code), Dr. Jill Harvey (minor editorial advice in Chapter 4), Dr. Tongli Wang (answering questions regarding daily weather and monthly climate data), Dr. Estelle Arbellay (friendly reviews of Chapters 2 and 3), Kelsey-Copes Gerbitz (friendly review of Chapter 3), Alice Cecile (answering questions on dendrochronological standardization), Nicole Prehn (measuring Douglas-fir tree-ring widths), and Xianda Guo (supporting a preliminary study on climatic variation at two locations across the Canadian Rocky Mountains). Chapter 2 was published online by the International Journal of Wildland Fire on June 4th, 2019. Raphaël Chavardès generated the research questions and was the main contributor to the research design, analyses and writing of this publication. Dr. Lori Daniels provided the fire-scar data, advised on research approach and provided editorial advice. Dr. Bianca Eskelson helped develop the mixed-model statistical analyses and provided editorial advice. Dr. Paul Pickell co-developed the R script to calculate MDC and provided minor editorial advice. vii Table of Contents Abstract ......................................................................................................................................... iii Lay Summary ................................................................................................................................ v Preface ........................................................................................................................................... vi Table of Contents ........................................................................................................................ vii List of Tables ................................................................................................................................ xi List of Figures .............................................................................................................................. xii List of Abbreviations .................................................................................................................. xv Acknowledgements .................................................................................................................... xvi Chapter 1: Introduction ................................................................................................................1 1.1 Mega-fires Driven by Climate Change ........................................................................... 1 1.2 Forest Fire Ecology ......................................................................................................... 1 1.3 Mixed-severity Fire Regimes .......................................................................................... 2 1.4 Fire-drought Associations ............................................................................................... 3 1.5 Overarching Hypothesis and Research Questions .......................................................... 3 Chapter 2: Monthly adaptations of the Drought Code reveal nuanced fire-drought associations in montane forests with a mixed-severity fire regime ...........................................6 2.1 Introduction ..................................................................................................................... 6 2.2 Methods......................................................................................................................... 10 2.2.1 Study area.................................................................................................................. 10 2.2.2 Daily and monthly fire-weather data ........................................................................ 11 2.2.3 Computing monthly adaptations of the daily Drought Code, 1989–2013 ................ 12 2.2.4 Predictive model of the monthly mean Drought Code, 1989–2013 ......................... 12 viii 2.2.5 Comparing monthly adaptations of the daily Drought Code, 1989–2013 ................ 13 2.2.6 Comparing monthly adaptations of the daily Drought Code, 1901–2013 ................ 13 2.2.7 Fire-drought associations, 1901–2013 ...................................................................... 14 2.3 Results ........................................................................................................................... 15 2.3.1 Model predicting DCm .............................................................................................. 15 2.3.2 Improved predictions of DCm ................................................................................... 16 2.3.3 Differences in monthly adaptations of the daily Drought Code, 1901–2013 ........... 16 2.3.4 Fire-drought associations .......................................................................................... 18 2.4 Discussion ..................................................................................................................... 19 2.4.1 adjMDC improved drought estimation in montane forests of southeastern British Columbia ............................................................................................................................... 19 2.4.2 Implications of improved drought predictions for montane forests of southeastern British Columbia ................................................................................................................... 20 2.4.3 Drought variability influenced the historial mixed-severity fire regime .................. 22 2.5 Conclusion .................................................................................................................... 23 Chapter 3: Using complementary drought proxies improves interpretations of fire histories in montane forests of southeastern British Columbia ..............................................................33 3.1 Introduction ................................................................................................................... 33 3.2 Methods......................................................................................................................... 37 3.2.1 Study area.................................................................................................................. 37 3.2.2 Monthly drought indices ........................................................................................... 38 3.2.3 Developing the Douglas-fir latewood-width chronology into a detrended redisual chronology ............................................................................................................................ 39 ix 3.2.4 Climate-growth analyses ........................................................................................... 40 3.2.5 Reconstructions of summer drought ......................................................................... 40 3.2.5 Fire-drought associations .......................................................................................... 41 3.3 Results ........................................................................................................................... 43 3.3.1 Reconstruction of summer drought from Douglas-fir latewood widths ................... 43 3.3.2 Fire-drought associations .......................................................................................... 44 3.4 Discussion ..................................................................................................................... 45 3.4.1 Monthly DC and PDSI reveal subtle differences in drought influences on montane forests.. .................................................................................................................................. 45 3.4.2 Fire-drought associations highlight differences in moisture content over time…. ... 46 3.4.3 Fire severity indicators in montane forests with mixed-severity fire regimes .......... 47 3.5 Conclusion .................................................................................................................... 49 Chapter 4: Fire-climate associations in the Montane Cordillera of Canada: Climate drives fire synchrony in forests with mixed-severity fire regimes  .....................................................58 4.1 Introduction ................................................................................................................... 58 4.2 Methods......................................................................................................................... 61 4.2.1 Study area.................................................................................................................. 61 4.2.2 Historical fire records ............................................................................................... 62 4.2.3 Reconstructions of regional climate .......................................................................... 62 4.2.4 Regional fire-climate associations ............................................................................ 64 4.2.5 Subregional fire-climate associations ....................................................................... 65 4.3 Results ........................................................................................................................... 65 4.3.1 Historical fire records ............................................................................................... 65 x 4.3.2 Reconstructions of regional climate .......................................................................... 66 4.3.3 Regional fire-climate associations ............................................................................ 66 4.3.4 Subregional fire-climate associations ....................................................................... 67 4.4 Discussion ..................................................................................................................... 68 4.4.1 Fire synchrony mostly coincided with drought ........................................................ 68 4.4.2 Uncommon, synchronous widespread-fires were driven by pronounced drought ... 70 4.4.3 Changes during the 20th century ............................................................................... 72 4.5 Conclusion .................................................................................................................... 73 Chapter 5: Conclusion .................................................................................................................84 5.1 Thesis Contribution and Summary ............................................................................... 84 5.2 Management Implications ............................................................................................. 86 5.3 Directions for Future Research ..................................................................................... 87 5.3.1 Comparing the strength of fire-drought associations among fire regimes ................ 87 5.3.2 Anticipating drought influences on future fire activity in montane forests of southeastern British Columbia .............................................................................................. 88 5.3.3 Assessing changes in fire severity for forests of the Montane Cordillera ................ 89 5.3.4 Interregional fire-climate associations focusing on teleconnections ........................ 89 Bibliography .................................................................................................................................90 Appendix A .................................................................................................................................123 Appendix B .................................................................................................................................128 Appendix C .................................................................................................................................129 xi List of Tables Table 2.1 Statistics of the linear mixed-effects model to predict the monthly mean Drought Code from the adjusted Monthly Drought Code .................................................................................... 25 Table 2.2 Summary of daily and monthly precipitation at the Palliser fire-weather station during 25 fire seasons including 1989−2013 ........................................................................................... 26 Table 3.1 Detrended residual latewood-width chronology-based reconstruction of the summer Drought Code (DC) at the Palliser weather station including reconstruction, model and validation statistics ......................................................................................................................................... 51 Table 4.1 Characteristics of the 17 site-level fire records arranged from north to south in the Montane Cordillera of Canada ...................................................................................................... 75 xii List of Figures Figure 1.1 Study areas in southeastern British Columbia and the Montane Cordillera of Canada 5 Figure 2.1 Palliser fire-weather station and fire-scar sites within montane forests of southeastern British Columbia, Canada ............................................................................................................. 27 Figure 2.2 Monthly mean Drought code versus adjusted Monthly Drought Code including overwintering and an April 1 start of the fire season at the Palliser fire-weather station during 25 fire seasons including 1989–2013 ................................................................................................. 28 Figure 2.3 Boxplots of the monthly mean Drought Code (DCm), Monthly Drought Code (MDC), adjusted MDC, and predicted DCm at the Palliser fire-weather station during 25 fire seasons including 1989–2013 .................................................................................................................... 29 Figure 2.4 Months during each fire season with low, intermediate and high values of the Monthly Drought Code (MDC), adjusted MDC, and predicted monthly mean Drought Code from 1901–2013 for the Palliser fire-weather station ...................................................................................... 30 Figure 2.5 Fire-drought associations for the Monthly Drought Code (MDC), adjusted MDC, and predicted monthly mean Drought Code for the Palliser fire-weather station ............................... 31 Figure 2.6 Comparison of low, intermediate and high Monthly Drought Code (MDC), adjusted MDC, and predicted monthly mean Drought Code values for Palliser fire-weather station during 17 fire years indicated by fire scars .............................................................................................. 32 Figure 3.1 Location of the sampling sites for Douglas-fir increment cores, and fire-scar records, in montane forests of southeastern British Columbia, Canada. Also shown are the location for the summer Drought Code reconstruction at the Palliser fire-weather station, and the summer Palmer Drought Severity Index reconstruction at grid point 67 ............................................................... 52 xiii Figure 3.2 Bootstrapped correlation coefficients of the detrended residual latewood-width chronology against the monthly Drought Code, and monthly Palmer Drought Severity Index  at the Palliser fire-weather station .................................................................................................... 53 Figure 3.3 Association between reconstructions of the summer Drought Code and summer Palmer Drought Severity Index from 1686–2004 ......................................................................... 54 Figure 3.4 Superposed epoch analyses showing associations for the reconstructions of the summer Drought Code and the summer Palmer Drought severity Index during fire years and non-fire years ................................................................................................................................ 55 Figure 3.5 Bivariate event analyses showing associations for fire years and non-fire years during coinciding and preceding extremely dry or wet summers according to the reconstructions of the summer Drought Code and summer Palmer Drought severity Index ........................................... 56 Figure 3.6 Scatter plot for the reconstruction of the summer Drought Code versus reconstruction of summer Palmer Drought Severity Index during fire years and non-fire years ......................... 57 Figure 4.1 Centroid locations for the sites with fire records and locations of the climate reconstructions across the Montane Cordillera of Canada in western North America ................. 76 Figure 4.2 Fire records from 1735–2000 for the 17 sites within the Montane Cordillera ............ 77 Figure 4.3 Scatter plots for the reconstructions of regional climate for the Montane Cordillera . 78 Figure 4.4 Reconstructions of regional climate according to categories of synchrony in the Montane Cordillera ....................................................................................................................... 79 Figure 4.5 Superposed epoch analyses showing associations between reconstructions of regional climate during synchronous events from 1746–1945 ................................................................... 80 Figure 4.6 Reconstructions of mean annual precipitation z-scores interpolated across the Montane Cordillera during the 11 years with high fire synchrony ............................................... 81 xiv Figure 4.7 Reconstructions of mean summer Palmer Drought Severity Index values interpolated across the Montane Cordillera during the 11 years with high fire synchrony .............................. 82 Figure 4.8 Reconstructions of mean annual precipitation z-scores, and summer Palmer Drought Severity Index values interpolated across the Montane Cordillera from 1746–1945 during years with synchronous events ............................................................................................................... 83 xv List of Abbreviations adjMDC: adjusted MDC ARIMA: autoregressive integrated moving average asl: above sea level B: average bias BEA: bivariate event analysis DC: Drought Code DCm: monthly mean DC IDW: Inverse Distance Weighted MDC: Monthly Drought Code PC1: first principle component PC1PDSI: first principle component from summer PDSI reconstructions PC1PPT: first principle component from annual precipitation reconstructions PCA: principle components analysis PDSI: Palmer Drought Severity Index predDCm: predicted DCm RMSE: Root Mean Square Error SEA: superposed epoch analysis xvi Acknowledgements I acknowledge and thank my Supervisor Drs. Lori Daniels, my Supervisory Committee Members Drs. Ze’ev Gedalof and Bianca Eskelson, my internship Supervisor Dr. David Andison, and my mentor Dr. Paul Pickell for working with me on this PhD Research. I acknowledge and thank the people in the Faculty of Forestry at the University of British Columbia, including the Office of the Dean, Departments of Forest and Conservation Sciences, Forest Resources Management, and Wood Science, and the Tree-Ring Laboratory and Landscape Ecology Laboratory for providing and/or sharing resources and knowledge. I acknowledge and thank the people with fRI Research, Mistik Management, the University of Alberta, the Government of Saskatchewan, the Government of Alberta, the British Columbia Wildfire Service, the British Columbia Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Parks Canada, the Canadian Forest Service, Natural Resources Canada, and Environment Canada for their time, diligence, and efforts. I acknowledge and thank the Elders of the Musqueam First Nation for their generosity and warmth. I acknowledge and thank, the Natural Sciences and Engineering Research Council of Canada, fRI Research and the Healthy Landscapes Program, Mitacs, the Tree-Ring Laboratory at UBC, the University of British Columbia Faculty of Forestry (including all of the following donors: Asa Johal, James Gorman, Bruce Blackwell, Kerri Kirincic, Charlotte Jones, and Claire Vivier), and my family for providing funding that supported my research. Finally, I thank my communities, friends, and family for their consistent support and understanding over the years.  1 Chapter 1: Introduction 1.1 Mega-fires Driven by Climate Change Large forest fires driven by human-induced climate change have become increasingly difficult to manage (Bowman et al. 2009). Particularly large forest fires >10,000 ha, known as mega-fires, tend to include extreme fire behaviour, which can adversely impact socioeconomic factors (Stephens et al. 2014). For example, following persistent and extremely dry and warm weather that desiccated forest fuels, more than one million hectares burned in interior British Columbia, Canada, during each of the 2017 and 2018 fire seasons (Sankey 2019). These extreme fire-seasons in the region attributed to climate change (Kirchmeier-Young et al. 2019) resulted in high fire management costs, community evacuations, strained physical and mental health for many Canadians, damaged infrastructure, and disruptions to business and industry (Sankey 2019). Although these costs and effects are concerning, less well understood and possibly as important are the impacts that such fires have on ecological systems and the services that they provide. 1.2 Forest Fire Ecology Ecologically, forest fires burn the vegetation with a range of severities over time and space (Agee 1993). The range of fire severity is the result of interacting drivers that include but are not limited to weather, climate, and fuels (Pausas and Ribeiro 2013). Interactions among the drivers are generally complex, hence it is often preferable to focus research on few drivers to gain an understanding of how these influence temporal and spatial characteristics of fires within a determined timeframe and over a delineated area, i.e., the fire regime (Turner 2010; Falk et al. 2011). Fire regimes are usually described according to the frequency and severity of the 2  dominant fires (Stephens et al. 2013). Traditionally, research on fire regimes has focused on low- and high-severity fire regimes, at opposite sides of the fire-severity spectrum. Low-severity fire regimes have frequent surface fires causing minimal tree mortality and reducing surface fuels, hence the name stand-maintaining regimes (Halofsky et al. 2011). Conversely, high-severity fire regimes have infrequent crown fires causing high tree mortality and significantly reducing surface and canopy fuels, hence the name stand-replacing regimes (Halofsky et al. 2011).  1.3 Mixed-severity Fire Regimes Among fire regimes, those of mixed severity were most recently identified and are thus the least well understood (Perry et al. 2011). Mixed-severity fire regimes are defined by successive fires that burn at one location with a range of severities over time, and by fires that burn with a range of severities across space either simultaneously or through time (Halofsky et al. 2011). These regimes include combinations of low-severity surface fires and higher-severity crown fires (Daniels et al. 2017). Mixed-severity fire regimes have been described in many forests of western North America (e.g., Hessburg et al. 2019), eastern North America (e.g., Bergeron et al. 2004; Drobyshev et al. 2011, 2012; Xu et al. 2018), western South America (e.g., González et al. 2005, 2010; Veblen et al. 2009; Holz and Veblen 2011; Cóbar-Carranza et al. 2014), and northern Europe (e.g., Lampainen et al. 2004; Kuuluvainen 2009). However, research on their drivers remains limited relative to that of low- and high-severity fire regimes (Halofsky et al. 2011). Further research investigating drivers of mixed-severity fire regimes, for example by focusing on how climate, and more specifically drought, i.e., persistent warm and dry weather, influenced these fire regimes, can provide fire managers with baseline information that would help assess current and future characteristics of mixed-severity fire regimes. 3  1.4 Fire-drought Associations To test how drought influenced mixed-severity fire regimes, one can test associations between fire and drought. Fire-drought associations have been investigated in western North America, yet most have focused their efforts in the western United States (e.g., Westerling et al. 2006; Heyerdahl et al. 2008a; Littell et al. 2009; Trouet et al. 2010). To test fire-drought associations, distinct metrics and indices are commonly used (Littell et al. 2016). Most of these are related to temperature and precipitation, two climate variables which provide an indication of drought (Littell et al. 2016). Specifically, higher temperatures and lower to no precipitation increase the degree of drought (Heim 2002). Drought in turn decreases moisture content in forest fuels making them more susceptible to fire ignition, spread and combustion (Van Wagner 1987; Macias Fauria et al. 2011). Testing fire-drought associations using metrics and indices that were designed to account for changes in temperature, precipitation, and moisture content in forest soils and fuels across different temporal scales (e.g., monthly to interannual) and spatial scales (e.g., landscape to regional) can enhance the understanding of potential fire behaviour in mixed-severity fire regimes.  1.5 Overarching Hypothesis and Research Questions Our current understanding of how drought influences mixed-severity fire regimes over time and space remains insufficient. Improved understanding can provide fire managers with baseline information that can support assessments of current and future characteristics of fires that burn at a range of severities. For my thesis research, I conducted three interrelated studies written as distinct chapters, to investigate drought as a driver of mixed-severity fire regimes at different temporal and spatial scales in the Montane Cordillera of Canada (Figure 1.1.). In the 4  first study (Chapter 2), I use a suite of monthly drought indices and test fire-drought associations in montane forests of southeastern British Columbia to answer two questions: 1) How do monthly derivatives of the Drought Code differ over the fire? 2) How do fire-drought associations differ among these monthly derivatives? In the second study (Chapter 3), I test fire-drought associations in these same montane forests using two summer drought reconstructions, which represent moisture content in different the duff layer versus deep organics in the soil and large woody fuels. For this second study, I address two questions: 1) How related are summer Drought Code and summer Palmer Drought Severity Index reconstructions? 2) How do fire-drought associations differ between these reconstructions? The third study (Chapter 4) broadens its spatial scope by testing fire-drought associations in low- and mid-elevation forests across multiple locations in the Montane Cordillera. I use three summer and annual drought reconstructions to test associations between droughts and years with fires at multiple locations, i.e., fire synchrony. For this third study, I pose two questions: 1) How frequent was fire synchrony in the Montane Cordillera between 1746 and 1945? 2) What climate conditions were associated with such events at regional and subregional scales? Chapter 5 summarizes my main findings and explains the contributions of my thesis to research on drivers of mixed-severity fire regimes. I provide management implications and identify four directions for future research.  5   Figure 1.1 (A) Montane Cordillera of Canada in (B) western North America.    6  Chapter 2: Monthly adaptations of the Drought Code reveal nuanced fire-drought associations in montane forests with a mixed-severity fire regime  2.1  Introduction Globally, forests are facing increased pressures due to climate change, especially during drought (Choat et al. 2012; Sommerfeld et al. 2018). Droughts, caused by persistent warm and dry weather, directly affect tree health and survival by lowering available moisture (Van Mantgem et al. 2009; Allen et al. 2010, 2015), and indirectly by influencing climate-mediated disturbances such as fire (Daniels et al. 2011; Moritz et al. 2014). To date, few studies have compared drought conditions between recent and historical fires in western Canadian montane forests (elevation up to 1500 m above sea level (asl)) with mixed-severity fire regimes. At watershed to landscape scales, these montane forests include even-aged stands resulting from fairly high-severity fires that last burned one to several centuries ago, intermixed with uneven-aged, mixed-conifer, structurally diverse stands (Marcoux et al. 2015; Daniels et al. 2017). The latter stands commonly include trees with single and multiple fire scars, and historical  reconstructions using fire scars and forest demography reveal these stands burned at low- to moderate-severity once every 20–60 years, on average (Marcoux et al. 2013, 2015; Chavardès et al. 2016; Rogeau et al. 2016). Some montane forests in southeastern British Columbia have rich fire-scar records that extend throughout the 20th century (Cochrane 2007; Daniels et al. 2011). Testing how these fire-scar records are associated with drought over the instrumental record, can provide novel insights on mixed-severity fire regimes and influences of variable drought conditions within and among fire seasons. 7  In Canada, the daily Drought Code (DC) is a fire-weather index related to combustibility of slow-drying forest fuels, including deep compact organic matter and large-diameter woody fuels (Wotton 2009). The Monthly Drought Code (MDC) is a monthly adaptation of DC (Girardin and Wotton 2009). Both DC and MDC integrate cumulative effects of drought over the fire season (Girardin and Wotton 2009). A feature of DC is its capacity to account for fuel moisture depletion between fire seasons, called “overwintering”. In many high-latitude or high-altitude forests, where moisture is retained in frozen soils and insulated by a persistent snowpack, spring snowmelt saturates soils and overwintering effects are low, as is the degree of drought at the start of the fire season (Van Wagner 1987). In forests with dry climates, discontinuous or ephemeral snowpack, and moderately- to well-drained soils, fuels can dry over winter, increasing the degree of drought at the start of the fire season (Lawson and Armitage 2008). Under the latter conditions, a site-specific adjustment is applied to DC to account for overwintering (Van Wagner 1987); however, a comparable adjustment has not been developed for MDC owing to insufficient local soil properties information that can affect overwintering (Girardin and Wotton 2009).  To characterize 20th Century droughts during the fire season (May 1−October 31) across Canada, Girardin and Wotton (2009) developed two monthly adaptations of DC. The monthly mean DC (DCm) is the average of daily DC values calculated from noon temperature and precipitation observations. MDC is an adaptation of DCm calculated from the average of daily maximum temperatures and total precipitation over each month (Girardin and Wotton 2009). To calculate DCm and MDC across a broad spatial network from 1901–2002, Girardin and Wotton (2009) assumed overwintering effects were low and set initial values of DC and MDC to 15 on April 30 of each year (i.e., one day prior to the start of the fire season on May 1). This approach resulted in mean July MDC values from 46−300 across Canada (Girardin and Wotton 2009). In 8  contrast, DC calculated using daily fire-weather station records from southeastern British Columbia revealed higher mean July and August values. For example, fire-weather records from the Palliser (1989−2013), Johnson Lake (1986−2013), and Cranbrook (2001−2013) stations yielded July means of 338, 464, and 499 and August means of 472, 602, and 642, respectively (BC Wildfire Management Branch; unpublished data). I have sought to understand why the MDC values underestimated DCm during peak fire season in this region. I hypothesize that three factors explain the discrepancies between MDC and DCm. The first factor is overwintering and how it affects the initial DC value. Typically, the default DC value at the start of the fire season is set to 15 (Lawson and Armitage 2008). However, overwintering results in DC values >15 at the start of many fire seasons in montane forests of southeastern British Columbia (British Columbia Wildfire Management Branch; unpublished data) and should thus be accounted for when calculating DC at the monthly scale.  The second factor influencing DC and MDC values is the date assigned to the start of the fire season. The specific date varies among locations and years depending on snow melt or temperature. Daily DC calculations begin either the third day after snow has melted in regions with persistent snow cover or the third day that noon temperatures are ≥12°C in regions where snow cover is not consistent (Lawson and Armitage 2008). Typically, May 1 has been the start date when calculating MDC (Girardin and Wotton 2009; Girardin et al. 2009, 2013; Mori and Johnson 2013). However, the median start date of actual fire-weather records from 1981–2010 for the Canadian Southern Cordillera was April 12th (Wang et al. 2015). Moreover, recent evidence points to the fire season starting earlier in many parts of North America owing to climate change (Flannigan et al. 2003; McKenzie et al. 2004; Westerling et al. 2016; Wotton et 9  al. 2017). Therefore, initiating calculations of MDC on April 1 is justified to improve understanding of fire weather throughout the fire season now and in the future. The way in which precipitation is incorporated into DC and MDC is the third factor. With DC, daily precipitation must exceed 2.8 mm before it is considered to have a wetting effect on deep organic and coarse woody fuels (Girardin and Wotton 2009). For days with precipitation >2.8 mm, DC is calculated using daily effective precipitation that accounts for canopy and surface fuel interception as follows: daily effective precipitation = 0.83 x (daily total precipitation >2.8 mm) - 1.27 mm (Girardin and Wotton 2009). With MDC, monthly precipitation is multiplied by 0.83 to calculate monthly effective precipitation (Girardin and Wotton 2009). Thus, MDC overestimates effective precipitation by including days with precipitation ≤2.8 mm and not subtracting 1.27 mm from each day that precipitation exceeds 2.8 mm. More accurate calculation of MDC on a monthly time scale depends on the frequency of days per month when precipitation exceeds 2.8 mm. In summary, when the subtle effects of overwintering, start date, and effective precipitation are combined, MDC underestimates DC in montane forests of southeastern British Columbia. In the present study, my goal was to interpret drought conditions associated with historical fires in montane forests of southeastern British Columbia. Using fire-weather data from the Palliser station located in the montane forest, I computed and compared four monthly adaptations of DC (DCm, MDC, adjusted MDC, and predicted DCm) to address three questions. How does MDC differ from DCm over the fire season? Do overwintering and early start of the fire season adjustments improve MDC accuracy relative to DCm? Does predictive modelling to account for effective precipitation on a monthly scale further improve accuracy? To interpret historical fire-drought relations, I computed three monthly adaptations of the drought code 10  (MDC, adjusted MDC, and predicted DCm) from 1901–2013 and tested their associations with a crossdated fire-scar record compiled from 20 sites in montane forests surrounding the Palliser station (Cochrane 2007; Daniels et al. 2011). Combined, these analyses provided a nuanced understanding of fire-drought associations in a montane forest with a mixed-severity fire regime.  2.2 Methods 2.2.1  Study area The study area encompasses the Palliser fire-weather station (50°29’40”N, 115°39’59”W; 1100 m asl) and 20 fire-history research sites distributed along a 250 km north-south transect representing 255 400 hectares of mid-elevation montane forests (1100–1550 m asl) in southeastern British Columbia, Canada (Figure 2.1). These forests are dominated by interior Douglas-fir (Pseudotsuga menziesii var glauca (Beissn.) Franco), western larch (Larix occidentalis Nutt.), lodgepole pine (Pinus contorta Dougl. ex Loud.) and white spruce (Picea glauca (Moench) Voss). Climate is continental, and precipitation is influenced by the rainshadow of the Purcell Mountains located west of the study area (Pojar and Meidinger 1991). Climate normals for 1981–2010 at Palliser were calculated using ClimateNA 5.30 (Wang et al. 2016). At Palliser, mean annual temperature was 3.5°C, with means of -7.9°C for January and 15.4°C for July. Mean annual precipitation was 433 mm, with 293 mm (68%) during the fire season from April–October, and 113 mm as snow from November–March. Lightning-ignited forest fires are most common from July–August (Natural Resources Canada 2019).  11  2.2.2 Daily and monthly fire-weather data The British Columbia Wildfire Service operates circa 260 hourly fire-weather stations. The Palliser record is the longest and most continuous in montane forests of the study area. I acquired daily fire-weather records for Palliser from May 31, 1989, to September 30, 2013, including temperature (°C) and precipitation (mm, as water equivalent) measured at noon (British Columbia Wildfire Service, unpublished data). These records included the daily Drought Code (DC) values calculated by local authorities.  Acquiring monthly fire-weather data required multiple steps. I added 2.8°C to the daily noon temperature to generate maximum temperature (°C) (after Turner 1972; Van Wagner 1987; Girardin and Wotton 2009). Daily (April 1 to October 31) maximum temperature and DC for each month of the fire season were averaged. Daily precipitation for each month of the year was summed because monthly winter precipitation was also needed to initiate drought code calculations at the beginning of each fire season. I identified missing monthly maximum temperature and precipitation values, and monthly mean DC (DCm) values using the criteria for the Adjusted and Homogenized Canadian Climate Data (Mekis and Vincent 2011; Vincent et al. 2012). From 1989–2013, 21 (12%) monthly maximum temperatures, and 28 (16%) DCm values were missing from the beginning (April or May) or end (October) of the fire season. Seventy-six (26%) monthly precipitation values were missing from November–May, mostly during the months between fire seasons. To estimate missing values, I generated records of monthly maximum temperature from April–October, and monthly precipitation from January–December for 1989–2013 using ClimateNA (Wang et al. 2016). I used the ClimateNA-generated records to impute the monthly maximum temperature and precipitation values that were missing from the 1989–2013 Palliser monthly records. 12  2.2.3 Computing monthly adaptations of the daily Drought Code, 1989–2013 I computed two monthly adaptations of DC over each fire season from 1989–2013 using the Palliser station monthly maximum temperature and precipitation data as follows. Following Girardin and Wotton (2009), MDC was calculated for May−October assuming an initial value of 15 on April 30 in each year (Appendix A). Adjusted MDC (adjMDC) accounted for overwintering and an earlier start to the fire season. From an initial DC value of 195.7 measured on May 31, 1989, adjMDC was calculated for June−October of 1989. From 1990 to 2013, adjMDC was calculated from April−October with the April starting value accounting for overwinter drying between each fire season adapted to the monthly scale.   2.2.4  Predictive model of the monthly mean Drought Code, 1989–2013 I developed a linear mixed-effects model to predict DCm from adjMDC: 𝐘𝐘𝑖𝑖 = 𝐗𝐗𝑖𝑖𝛃𝛃 + 𝐙𝐙𝑖𝑖𝐛𝐛𝑖𝑖 + 𝛆𝛆𝑖𝑖  where i = month (April, May, …, October) or group of months (early fire season: April–May, mid–late fire season: June–October), Yi is a column vector of length ni of DCm observations for each group i, Xi and Zi are design matrices of dimension ni x 2 containing adjMDC values as predictor variables, β and 𝐛𝐛𝑖𝑖 are vectors of fixed- and random-effect coefficients, respectively, and 𝛆𝛆𝑖𝑖  ~ N𝑛𝑛𝑖𝑖(𝟎𝟎,𝛔𝛔𝜺𝜺𝑖𝑖𝟐𝟐 ) is a column vector of length ni of the errors for observations in i with covariance matrix 𝛔𝛔𝜺𝜺𝑖𝑖𝟐𝟐 . During model development, I tested whether it was justified to include (a) random effects (intercepts and slopes) for each month versus a random effect for each group of months, and (b) AR(1) correlation and/or unequal variance structure to account for autocorrelation and heteroscedasticity, respectively. Model selection was based on minimizing 13  Akaike’s information criterion, likelihood ratio tests and model simplicity (Bolker et al. 2009). To assess the fit of my model, I calculated for each month, and group of months, average bias (B) and root mean squared error (RMSE) using leave-one-out cross validation.  2.2.5  Comparing monthly adaptations of the daily Drought Code, 1989–2013 To assess the accuracy of MDC, adjMDC or predicted DCm (predDCm) relative to DCm from 1989–2013, I used boxplots with Mann-Whitney rank sum tests and calculated Pearson product moment correlations (Gorvine et al. 2018). To evaluate discrepancies among these drought codes, I used daily precipitation records for Palliser to calculate the percentage of days per month when precipitation was 0 mm, >0 but ≤2.8 mm, and >2.8 mm. Monthly precipitation records missing >5 days or >3 consecutive days were excluded from this analysis (Mekis and Vincent 2011). I calculated effective precipitation during a given month in two ways: (1) summing all daily precipitation values and multiplying their total by 0.83, i.e., monthly effective precipitation; and (2) summing only the daily precipitation values >2.8 mm, multiplying their total by 0.83, and then subtracting 1.27 mm per day that precipitation was >2.8 mm, i.e., daily effective precipitation summed over the month (Girardin and Wotton 2009). Overestimation of effective precipitation during a given month was the relative difference between monthly effective precipitation and daily effective precipitation summed over the month.  2.2.6 Comparing monthly adaptations of the daily Drought Code, 1901–2013 I used the ClimateNA-generated monthly data to calculate MDC from May−October, 1901−2013, using an initial value of 15 on April 30 of each year (Girardin and Wotton 2009). adjMDC was calculated continuously from April 1901 to October 2013 from an initial value of 14  15 on March 31, 1901 (Canadian Forestry Service 1984). I applied the linear mixed-effects model coefficients to predict DCm over the same period as adjMDC. To compare MDC, adjMDC, and predDCm from 1901–2013, I categorized their values as low (<200), intermediate (≥200 and <400), or high (≥400) according to thresholds corresponding to relative moisture content in compact organics and coarse woody fuels (after Bergeron et al. 2010). When drought codes are low, fuels are fairly wet and least likely to combust, whereas high values indicate dry combustible fuels, which is supported by de Groot et al.’s (2009) findings that more forest floor fuels are consumed by fire as drought codes increase. To assess the onset of drought during each fire season, I compared the first month in each year when values were ≥200 for each code. To assess the magnitude of fire season droughts, I compared the proportion of months categorized as low, intermediate, and high during all fire seasons from 1901–2013. For each pair of monthly adaptations of DC, I derived two-way contingency tables of observed values from which I calculated expected values and assessed differences using chi-square goodness-of-fit tests (Whitlock and Schluter 2015).  2.2.7 Fire-drought associations, 1901–2013 I combined crossdated fire-scar records from 20 montane forest sites in the study area (Cochrane 2007; Daniels et al. 2011) and identified 17 years from 1901–2013 in which ≥2 trees were scarred (hereafter, “fire years”). I tested fire-drought associations in two ways. First, I used Superposed Epoch Analysis in the Fire History Analysis and Exploration System (Brewer et al. 2015), to test for associations between fire occurrence and the following monthly adaptations of DC: MDC, adjMDC, and predDCm. I compared the mean value of each monthly adaptation during fire years relative to bootstrapped values from a Monte Carlo simulation of 1000 15  randomly selected years that provided 95.0, 99.0 and 99.9% confidence intervals (Sutherland et al. 2017). Second, to determine if MDC, adjMDC, and predDCm represented the magnitude of droughts during fire years similarly, I compared the proportion of months categorized as low, intermediate, and high during the 17 fire years. For each pair of monthly adaptations, I derived two-way contingency tables of observed values from which we calculated expected values and assessed differences using chi-square goodness-of-fit tests (Whitlock and Schluter 2015).  2.3 Results 2.3.1  Model predicting DCm The graph of DCm versus adjMDC revealed distinct slopes for two groups of months (group 1: April–May and group 2: June–October) (Figure 2.2). As a result, the final linear mixed effects model included random slopes by group of months representing the early fire season (April–May) and mid–late fire season (June–October). Among tested models, inclusion of an AR(1) correlation structure in the model resulted in the lowest AIC value. By definition of the fitted model, predictions by group of months were unbiased (Table 2.1). Across months, range of bias (B) was smallest over August and September (-2≤ B ≤4), largest over April and October (-22≤ B ≤30) and intermediate from May–July (-15≤ B ≤7). Root mean squared error (RMSE) across groups of months was greater over the early fire season (RMSE = 52) than mid–late fire season (RMSE = 43). Across months, RMSE was lowest from July–September (36≤ RMSE ≤42), greatest from April–June (51≤ RMSE ≤52) and intermediate over October (RMSE = 46).  16  2.3.2 Improved predictions of DCm Of the monthly adaptations of DC, MDC was least accurate and predDCm was the most accurate representation of DCm (Figure 2.3). Based on Mann-Whitney rank sum tests, medians of MDC were significantly lower than those for DCm (p <0.001); MDC ranges were also less than those for DCm. Over May, the correlation between MDC and DCm was weak and not significant (r2 = 0.29, p = 0.209), but correlations were strong and significant (r2 ≥0.78, p <0.001) from June−September. When comparing adjMDC with DCm, medians were similar early in the fire season (p = 0.841 and p = 0.091 in April and May, respectively), but adjMDC values were significantly lower than DCm for June−September (p ≤0.011 for all months). Ranges of MDC were also less than those of DCm later in the fire season. predDCm and DCm had similar medians and ranges for all months; medians did not differ significantly (p ≥0.351). Correlations of adjMDC and predDCm with DCm were strong and significant (r2 ≥0.87, p ≤0.027) for all months. At Palliser fire-weather station, monthly precipitation was greatest in June, increased from April–June and decreased from June–October (Table 2.2). Precipitation was >2.8 mm and sufficed to have a wetting effect on deep organics and coarse woody fuels thus altering daily DC on 9–25% of days per month. Monthly effective precipitation averaged 19–64 mm, whereas it averaged 9–47 mm when calculated from daily effective precipitation summed over the month. Overestimation of effective precipitation during a given month ranged from 36–117% and was greatest in October and least in June.  2.3.3 Differences in monthly adaptations of the daily Drought Code, 1901–2013 The degree, onset, and duration of droughts during the fire season differed among monthly adaptations of DC (Figure 2.4). MDC indicated the least amount of drought conducive 17  to fire. Over 113-years from 1901–2013, MDC was ≥200 in 4±1 months per year (mean ± s.d.; range 0–5). MDC first reached intermediate values in June in 3 years and July in 75 years. High values occurred in 42 years and persisted 2±1 months (range 1−3), for a total of 87 (13%) of 678 months assessed (i.e., May−October over 113 years). The adjMDC also was ≥200 in 4±1 months per year, with a wider range of 0−7 months. Compared to MDC, adjMDC reached intermediate values earlier: April or May in 11 years, June in 15 years, and July in 60 years. High values of adjMDC occurred in 47 years and persisted 2±1 months (range 1−4) for a total of 103 (13%) of the 791 months assessed (i.e., April−October over 113 years). Relative to MDC and adjMDC, predDCm indicated conditions more conducive to fire, with droughts that started earlier and were more pronounced. The predDCm was ≥200 in 5±1 months (range 3−7) per year. It reached intermediate values in April or May in 22 years, June in 61 years and July in 24 years. High predDCm values occurred in 95 years and persisted 3±1 months (range 1−6) for a total of 285 (36%) of the 791 months assessed.  Chi-square goodness-of-fit tests confirmed MDC and adjMDC had significantly fewer months with intermediate and high values and more months with low values than expected relative to predDCm from 1901–2013 (χ2 = 104.6, p <0.001 and χ2 = 115.8, p <0.001, respectively). MDC and adjMDC did not differ significantly across the proportion of months categorized as low, intermediate, and high from 1901–2013 (χ2 = 0.7, p = 0.703). 18  2.3.4 Fire-drought associations The monthly adaptations of DC revealed fire years were associated with a higher degree of drought conditions over multiple months of the fire season; however, of these associations, the strongest were for adjMDC and predDCm (Figure 2.5). Fire years were associated with high MDC values in mid–late fire season, with the strongest and most significant associations in July and August (p <0.01), followed by June (p <0.05). For adjMDC, associations were strongest from April–August (p <0.001), followed by September (p <0.05). Associations with predDCm were strongest in June and July (p <0.001), followed by April, May and August (p <0.01), and September (p <0.05). The degree, onset, and duration of droughts during the 17 fire years varied among MDC, adjMDC, and predDCm (Figure 2.6). In 16 fire years, MDC values were intermediate or high for 3−5 months; MDC remained low throughout the 1903 fire season. MDC values were low in May and reached intermediate values in June, July, or August during 2, 13, and 1 fire years, respectively. High MDC values occurred relatively late in the fire season in August, September, and October during 5, 11, and 8 fire years, respectively. Intermediate and high values of adjMDC persisted 3−7 months in all fire years, except in 1903 when adjMDC remained low. Compared to MDC, intermediate and high values of adjMDC were reached earlier during fire years. Values were intermediate in April and May in 3 and 4 fire years, respectively; high values were reached in July, August, and October in 2, 8 and 9 fire years, respectively. Compared to MDC and adjMDC, predDCm had the most intermediate values, starting as early as April (n = 5). Intermediate or high values persisted 4−7 months of each fire year. High values were reached as early as May and June in 1 and 3 fire years, respectively, and occurred in July, August, September, and October during 10, 16, 14, and 13 fire years, respectively. 19  Chi-square goodness-of-fit tests confirmed MDC and adjMDC had significantly fewer months with intermediate and high values and more months with low values than expected relative to predDCm during the 17 fire years (χ2 = 16.4 p <0.001 and χ2 = 14.7, p <0.001, respectively). MDC and adjMDC did not differ significantly across the proportion of months categorized as low, intermediate, and high during fire years (χ2 = 0.2, p = 0.898).  2.4 Discussion 2.4.1  adjMDC improved drought estimation in montane forests of southeastern British Columbia In the Canadian Fire Weather Index System, the daily Drought Code (DC) is used alongside other fire-weather indices to forecast fire danger, potential fire behaviour (Stocks et al. 1989; Taylor and Alexander 2006), and the effort required to extinguish a fire (Terrier et al. 2013). I aimed to determine whether the monthly mean Drought Code (DCm) was well represented by the Monthly Drought Code (MDC), an index commonly applied in boreal forests (Girardin et al. 2009; Bergeron et al. 2010; Drobyshev et al. 2013; Marchal et al. 2017), and adjMDC that accounted for overwinter drying and an earlier start to the fire season. For the Palliser station in montane forests of southeastern British Columbia, I found lower MDC values during April−May and lower MDC and adjMDC values during June−September relative to DCm. I found that MDC underestimated the degree of drought throughout the fire season in the dry montane forests of southeastern British Columbia. Therefore, I raise caution when interpreting MDC values in other montane forests where overwinter drying is common. Because adjMDC included the adjustment typically applied to daily DC calculations in dry western and some northern forests of Canada (Van Wagner 1987; Lawson and Armitage 2008), ranges of adjMDC 20  and DCm were similar during the early fire season. Nevertheless, adjMDC like MDC, underestimated DCm from June−September, when fires commonly burn in montane forests of British Columbia (Natural Resources Canada 2019). Assessment of the Palliser station daily precipitation records illustrates subtle effects of daily precipitation on the monthly adaptations of DC. Over 25 fire seasons from 1989−2013, precipitation exceeded 2.8 mm, effectively wetting deep organic fuels and coarse wood, only 9−25% of days per month. Monthly effective precipitation used to calculate MDC and adjMDC exceeded daily effective precipitation summed over the month used to calculate DCm by 36−117% per month, explaining the systematic underestimates of these two drought adaptations relative to DCm. April and May effective precipitation overestimates averaged 67 and 51%, respectively. Evidently, accounting for overwinter drying countered the effect of overestimating April and May effective precipitation because adjMDC was more accurate than MDC early in the fire season. June was wettest and most variable, with the lowest effective precipitation overestimate (36%). In July, August and September, overestimates ranged from 50−53%. These inaccuracies are cumulative; consequently, differences between MDC and adjMDC relative to DCm increased through the fire season.  2.4.2 Implications of improved drought predictions for montane forests of southeastern British Columbia Combined effects of the subtle factors overestimating monthly effective precipitation and underestimating DCm warranted the development of a predictive regression model for the Palliser station. The predDCm more accurately represented DCm than MDC or adjMDC and 21  improved understanding of the onset and duration of droughts during fire years from 1901−2013. The shift from low to intermediate drought codes provides critical information about increasing fire danger and risk of a sustained ignition. The predDCm revealed many years reaching intermediate values early in the fire season and improved detection of early-season drought making deep organic and coarse woody fuels more conducive to combustion. Given evidence of an earlier start to the fire season in recent decades (Jolly et al. 2015; Westerling et al. 2016; Wotton et al. 2017) with the trend expected to continue over the 21st century (Flannigan et al. 2013), my approach provides predictions of DCm that allow a more accurate understanding of how drought conditions influenced the onset of historical fire years in montane forests of southeastern British Columbia. In turn, these historical drought conditions can be compared to modern observations and future predictions to assess potential changes in the onset and degree of drought during each fire season over time. Droughts indicated by high predDCm values at Palliser were common and of adequate magnitude and duration to sustain fire, had an ignition occurred. Specifically, predDCm values for July and August from 1901–2013 averaged 345 and 452, respectively. These values are greater than mean July MDC values ≤300 reported across Canada from 1901–2002 (Girardin and Wotton 2009) and mean August MDC values <200 reported for the Montane Cordillera from 1901–2009 (Mori and Johnson 2013). These two previous studies suggested the likelihood of sustained fires remained intermediate or low even during peak fire season. In contrast, predDCm calculated at a finer spatial scale in southeastern British Columbia indicated more frequent and persistent droughts of greater magnitude during peak fire season that had greater likelihood of sustaining fires over the same period.  22  2.4.3 Drought variability influenced the historical mixed-severity fire regime Using crossdated fire-scar records in concert with the monthly adaptations of DC improved understanding of the range of variation in droughts that influenced the historical mixed-severity fire regime in the study area. Past fire-drought analyses assessed documentary records of area burned or number of large wildfires against MDC (Girardin et al. 2009; Bergeron et al. 2010; Girardin et al. 2013; Mori and Johnson 2013); however, this approach may be affected by documented fire suppression effects in British Columbia. Over the past 40 years, 92% of fires in British Columbia were suppressed before exceeding 4 ha in size (British Columbia Wildfire Management Branch 2012). As low-intensity fires are fairly easy to suppress (Arienti et al. 2006), fire-drought analyses based on modern area-burned records from British Columbia represent high-intensity fires that tend to burn under extreme fire-weather conditions (Johnson et al. 2001). Focusing on the subset of large intense fires that exceeded suppression capability biases interpretations of how drought variability influenced historical fire regimes. Using crossdated fire scars partly addressed this limitation. The fire-scar record does not overcome temporal effects of fire suppression. However, fire scars provide evidence of low- and moderate-severity surface fires, although scars can also form on the periphery of high-severity fires (Daniels et al. 2017). Therefore, fire-scar records are suitable for interpreting the range of drought conditions in which historical fires burned in montane forests with mixed-severity fire regimes. Combining fire-scar records with the monthly adaptations of DC provided novel insights into the duration and magnitude of droughts influencing historical fires. Among the monthly adaptations, adjMDC provided the strongest significant fire-drought associations, particularly during April, May and August. Fire-drought associations were also strong with predDCm, and 23  drought conditions were more accurately represented by predDCm than adjMDC over the fire season. I found both adjMDC and predDCm values were predominantly low during the early-fire season, indicating wet and/or cool conditions limited fuel combustibility. Six fire years, 1903, 1905, 1917, 1925, 1938, and 1971, had low and intermediate adjMDC and predDCm values from April–July. Fires burning under these conditions would have been low in severity; higher-severity fires would have been limited to later in the fire season when drought values increased. In contrast, springs with little precipitation and high temperatures, such as 1906, 1926, 1931, 1937, and 2001, resulted in intermediate to high adjMDC and predDCm values early in the fire season. As the fire season progressed and drought increased, more combustible fuels became available exacerbating the potential for high-severity fire. Corroborating these interpretations, montane forests of the study area include strong evidence of historical fires of a range of severities, including even-aged tree cohorts that established following moderate- to high-severity fires and uneven-aged forests with abundant fire-scarred trees recording lower-severity fires (Daniels et al. 2011; Marcoux et al. 2013, 2015). My findings provide an opportunity for future research that uses adjMDC and predDCm alongside fire-scar seasonality to test the hypothesis that early- versus late-season fires differed in severity, depending on the onset and duration of fire-season drought.   2.5 Conclusion Monthly adaptations of DC provided a nuanced understanding of the variability in drought within and among fire seasons in montane forests of southeastern British Columbia. Adjusting for overwintering and an early start to the fire season improved adjMDC relative to MDC estimates of DCm, particularly early in the fire season. Regression modelling based on 24  adjMDC accounted for differences between total and effective precipitation, yielding predDCm values that most accurately reproduced DCm. Hence, predDCm most accurately represented the degree, onset, and duration of droughts during fire seasons from 1901−2013. I found numerous and significant associations between fire-scar dates and adjMDC and predDCm for months throughout the fire season. Both adjMDC and predDCm showed variability in drought within and among fire seasons influencing historical fires in the study area.  Two monthly adaptations of DC, adjMDC and predDCm, provide fire scientists and managers with a practical connection between modern drought indices used to determine fire danger and historical droughts that were conducive to fires of a range of severities. Given growing evidence of mixed-severity fire regimes, I recommend continued investigation of these drought codes. Because precipitation varies along elevation gradients and among montane forests, comparative analyses among stations are warranted to determine the degree to which patterns observed at Palliser may be generalized and where they are not applicable. I acknowledge that calculating predDCm elsewhere requires daily fire-weather records, a substantive limitation due to spatial and temporal paucity of local records, as originally noted by Girardin and Wotton (2009). In contrast, adjMDC can be calculated for more locations and over broader areas using modelled monthly climate data. However, adjMDC remains less accurate than predDCm in the mid–late fire season owing to differences in how precipitation is integrated into these two monthly drought indices. I conclude that adjMDC and predDCm are practical indices of drought conditions conducive to fires of a range of severities and have strong potential to enhance understanding of historical fire-drought relations in forests where overwintering conditions dry fuels between fire seasons. 25  Table 2.1 Statistics of the linear mixed-effects model to predict the monthly mean Drought Code (DCm) from the adjusted Monthly Drought Code (adjMDC). Statistics included ni as the number of observations per month or group of months, B as average bias, and RMSE as root mean square error. Statistic  Group  Month Apr–May Jun–Oct Apr May Jun Jul Aug Sep Oct ni  25 118  5 20 25 25 25 25 18 B  1 0  -22 7 -15 -9 4 -2 30 RMSE  52 43  51 52 51 52 36 40 46                26  Table 2.2 Summary of daily and monthly precipitation at the Palliser fire-weather station during 25 fire seasons including 1989−2013. Monthly precipitation records missing >5 days or >3 consecutive days were excluded. Effective precipitation during a given month was calculated in two ways: (1) by summing all daily precipitation values and multiplying their sum by 0.83, i.e., monthly effective precipitation; and (2) by summing the daily precipitation values that exceeded 2.8 mm, multiplying their sum by 0.83, and then subtracting 1.27 mm per day that precipitation was >2.8 mm, i.e., daily effective precipitation summed over the month. The relative difference between (1) and (2) is the overestimate of effective precipitation that causes the Monthly Drought Code (MDC) and the adjusted MDC (adjMDC) to underestimate the monthly mean Drought Code (DCm). Numbers in parentheses indicate standard deviations.                       Month   Sample size (n)  Daily precipitation (% of days)  Effective precipitation (mm) Overestimate (%)     0mm >0 and ≤2.8mm >2.8mm   (1) (2) Apr  12  66 (11) 20 (11) 14 (9)  26 (16) 16 (14) 67 May  20  56 (8) 27 (8) 17 (6)  42 (16) 28 (15) 51 Jun  25  47 (12) 28 (10) 25 (10)  64 (32) 47 (29) 36 Jul  25  65 (13) 22 (9) 16 (8)  36 (17) 23 (13) 53 Aug  25  69 (14) 21 (9) 13 (9)  32 (23) 22 (19) 50 Sep  25  68 (15) 19 (10) 13 (8)  28 (17) 19 (13) 52 Oct   18   67 (11) 23 (8) 9 (5)   19 (8) 9 (5) 117               27    Figure 2.1 (a) Palliser fire-weather station and fire-scar sites within montane forests of (b) southeastern British Columbia, Canada.  28   Figure 2.2 Monthly mean Drought Code (DCm) versus adjusted Monthly Drought Code (adjMDC) including overwintering and an April 1 start of the fire season at the Palliser fire-weather station during 25 fire seasons including 1989–2013.  29   Figure 2.3 Boxplots of the monthly mean Drought Code (DCm), Monthly Drought Code (MDC), adjusted MDC (adjMDC), and predicted DCm (predDCm) at the Palliser fire-weather station during 25 fire seasons including 1989–2013. In each boxplot, the horizontal line represents the median; box boundaries are the 25th and 75th percentiles; bars are the 10th and 90th percentiles; and, dots are outliers.        30   Figure 2.4 Months during each fire season with low (<200; white areas), intermediate (≥200 and <400; light grey areas), and high (≥400; dark grey areas) values of the Monthly Drought Code (MDC), adjusted MDC (adjMDC), and predicted monthly mean Drought Code (predDCm) from 1901–2013 for the Palliser fire-weather station. Peaks at the top of each graphic represent the 17 fire years indicated by fire scars.          MonthMMDCadjMDCpredDCmJJASOMJJASOAMJJASOACalendar year1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 201031   Figure 2.5 Fire-drought associations for the Monthly Drought Code (MDC), adjusted MDC (adjMDC), and predicted monthly mean Drought Code (predDCm) for the Palliser fire-weather station. Black (white) bars represent mean values during fire (non-fire) years. Stars depict significant associations with fire years (1, 2, and 3 stars indicate the 95, 99, and 99.9 % confidence levels, respectively).  32   Figure 2.6 Comparison of low (<200), intermediate (≥200 and <400) and high (≥400) Monthly Drought Code (MDC), adjusted MDC (adjMDC), and predicted monthly mean Drought Code (predDCm) values for the Palliser fire-weather station during 17 fire years indicated by fire scars.      33  Chapter 3: Using complementary drought proxies improves interpretations of fire histories in montane forests of southeastern British Columbia  3.1  Introduction Drought-driven fire risk is expected to increase over the 21st century across many forests of western North America due to climate change (Stephens et al. 2013; Moritz et al. 2014). As a result of drought intensified by warmer temperatures and lower precipitation (Diffenbaugh et al. 2015), forest fuels have low moisture content making them more susceptible to ignite and burn at higher intensities and consequently with more severe impacts (Littell et al. 2009; Flannigan et al. 2013). Unfortunately, specific drought influences on fuels, including their ignition and combustion, remain insufficiently understood in montane forests of western North America characterized by mixed-severity fire regimes (Halofsky et al. 2011). Mixed-severity fire regimes are comparatively more complex than low- and high-severity fire regimes (Schoennagel et al. 2004; Halofsky et al. 2011). Low-severity fire regimes are dominated by frequent surface fires that cause minimal tree mortality but reduce surface fuels, whereas high-severity fire regimes are dominated by infrequent crown fires with high tree mortality that significantly reduce surface and crown fuels (Agee 1993). For mixed-severity fire regimes, fires burn with a range of severities across spatial and temporal scales (Schoennagel et al. 2004; Halofsky et al. 2011), resulting in forest stands that are compositionally and structurally diverse (Hessburg et al. 2016; Daniels et al. 2017). Fluctuation in fire severity is caused by interacting top-down drivers including weather and climate, which influence drought, and bottom-up drivers including topography, and vegetation composition and structure (Turner 2010; 34  Perry et al. 2011). An improved understanding of the interactions between drought and fires by investigating fire-drought associations (see Chapter 2; Chavardès et al. 2019) can assist fire managers in anticipating impacts of climate change in montane forests with mixed-severity fire regimes.  Research on contemporary relationships between fire occurrence and drought have used documentary fire records with instrumental drought records in the 20th and early 21st centuries in western North America (e.g., Westerling et al. 2006; Littell et al. 2016), including western Canada (e.g., Xiao and Zhuang 2007; Meyn et al. 2010a, 2010b). Yet, this period includes advances in fire suppression technologies (Morgan et al. 2008; Flannigan et al. 2009), which can impact fire-drought associations (Williams and Abatzoglou 2016). To address this limitation, multi-century proxy reconstructions of drought based on tree-ring chronologies have been developed (Cook et al. 2004; Girardin et al. 2004a). Such reconstructions are based on climate-growth analyses that test the relationship between specific monthly drought variables over the instrumental period against tree ring-, earlywood-, and/or latewood-width chronologies (Fritts 1976; Watson and Luckman 2002).  In western North America, interior Douglas-fir (Pseudotsuga menziesii var glauca (Beissn.) Franco) (hereafter “Douglas-fir”) radial growth is limited by low moisture availability according to climate-growth analyses (Chen et al. 2010). Within montane forests of western Canada, climate-growth analyses were conducted across multiple Douglas-fir ring-width chronologies (Griesbauer and Green 2010) and earlywood- and latewood-width chronologies (Watson and Luckman 2002). Specifically, Watson and Luckman (2002) found that intra-annual Douglas-fir growth was limited by low moisture availability, which is consistent with findings from the northwestern USA (Littell et al. 2008; Crawford et al. 2015). Seasonally, low moisture 35  availability during the months of June and July in the current growing season significantly decreased latewood widths (Watson and Luckman 2002; Crawford et al. 2015). Low moisture availability also influences fire activity during peak fire season in July and August within montane forests of western Canada (Natural Resources Canada 2019). Thus, conducting climate-growth analyses using a moisture-sensitive Douglas-fir latewood-width chronology is justified to better understand how drought historically varied over the fire season in these montane forests. Researchers have commonly used the Palmer Drought Severity Index (PDSI; Palmer 1965) in North America to conduct climate-growth analyses (e.g., Watson and Luckman 2002; Littell et al. 2008; Carnwath et al. 2012) and reconstruct historical droughts (e.g., Stahle et al. 2000; Cook et al. 2010). Based on PDSI records and moisture sensitive tree-ring chronologies, Cook et al. (2004) developed multi-century 2.5° x 2.5° gridded reconstructions of summer (June–August) PDSI over North America. These chronologies included those developed by Watson and Luckman (2004) for montane forests of western Canada. The multi-century depth and continental extent of the gridded reconstructions of summer PDSI made them suitable to test fire-drought associations at different spatial scales (e.g., Hessl et al. 2004; Trouet et al. 2006, 2010; Heyerdahl et al. 2008b), even though PDSI was originally developed as an agricultural index (Palmer 1965) and not as a fire-weather index. The Drought Code (DC) is a daily fire-weather index used in the Canadian Forest Fire Behaviour Prediction System to indicate the moisture content in deep compact organics in the soil and coarse woody fuels (Girardin and Wotton 2009; Podur and Wotton 2010; Chavardès et al. 2019). High DC values indicate that these fuels have low moisture content and are more likely to combust (de Groot et al. 2009). To investigate the associations between DC and historical fires in montane forests of southeastern British Columbia, Canada, Chavardès et al. (2019) developed 36  and compared several monthly adaptations of the DC (see Chapter 2). The predicted monthly mean Drought Code (hereafter, “monthly DC”) most accurately represented the onset, duration, and degree of drought during the fire season, and exhibited several strong and significant fire-drought associations in surrounding montane forests. A unique feature of the monthly DC is its capacity to adjust for soil moisture depletion between fire seasons, a condition common in many forests of western Canada (Lawson and Armitage 2008). Depletion of soil moisture can lead to sufficiently dry fuels that have the potential to burn even during the months of April and May in some forests (Chavardès et al. 2019). Given these particularities, the monthly DC is an ideal index to test climate-growth relationships using drought-sensitive Douglas-fir latewood-width chronologies. I hypothesize that its development into a reconstruction of summer DC (average monthly DC from June–August) predating instrumental records can facilitate comparisons to the nearest gridded reconstruction of summer PDSI.  Compared to monthly and summer DC, monthly PDSI and its summer reconstructions indicate moisture content in the surface and underlying soil layers (Hu and Willson 2000). Moisture is removed from these two layers through a temperature driven evapotranspiration algorithm whereas precipitation is added using a water balance method (Thornthwaite 1948; Palmer 1965). According to their description, the surface and underlying soil layers are best represented by the duff to uppermost soil mineral horizon (Miyanishi 2001). The duff layer is highly susceptible to ignition and combustion when dry (Wotton et al. 2005) thus monthly PDSI and its summer reconstructions are commonly used to test fire-drought associations in North America (Littell et al. 2016). Given the subtle difference in fuels represented by monthly and summer DC compared to monthly PDSI and its summer reconstructions, I hypothesize that 37  testing their associations with past fires has the potential to enhance understanding of fire behaviour in montane forests of southeastern British Columbia with mixed-severity fire regimes.    In this research, I test the hypothesis that using two proxy reconstructions of drought provides a more nuanced understanding of the type and degree of droughts associated with mixed-severity fire regimes. Specifically, I addressed three questions. (1) How do climate-growth relationships using monthly DC compare with monthly PDSI? (2) How does a reconstruction of summer DC based on monthly DC compare with the nearest gridded reconstruction of summer PDSI? (3) Do fire-drought associations differ among summer DC and summer PDSI reconstructions? To answer these questions, I developed a new summer drought reconstruction based on monthly DC using a moisture-sensitive Douglas-fir latewood-width chronology. I tested and compared fire-drought associations using this reconstruction and a reconstruction of summer PDSI representing an extensively used proxy-drought record in North America (e.g., Stahle et al. 2000; Cook et al. 2010). Practical outcomes from this research for fire managers include a nuanced understanding of fire ignition and fuel combustion in montane forests with historical mixed-severity fire regimes.  3.2 Methods 3.2.1 Study area This research focused on montane forests (circa 1100–1550 m above sea level (asl)) of southeastern British Columbia, Canada (Figure 3.1). These mid-elevation forests are dominated by conifers, including Douglas-fir, western larch (Larix occidentalis Nutt.), lodgepole pine (Pinus contorta Dougl. ex Loud.), white spruce (Picea glauca (Moench) Voss), and ponderosa pine. Such forests contain legacies of mixed-severity fire regimes (Cochrane 2007; Daniels et al. 38  2011; Marcoux et al. 2015), including fire scars and even-aged tree cohorts, providing evidence of low-, and moderate- to high-severity fires, respectively (Marcoux et al. 2013).  Climate in the study area is continental (Pojar and Meidinger 1991). Between 1981 and 2010, the Palliser fire-weather station (50°29’40”N, 115°39’59”W; 1100 m asl) recorded a mean annual temperature of 3.5°C, with mean monthly temperatures of -7.9°C for January and 15.4°C for July, and mean annual precipitation of 433 mm, with 293 mm (68%) from April–October, and 113 mm as snow from November–March (Chavardès et al. 2019). Despite most of the precipitation falling from April–October, peak activity of lightning-ignited forest fires was common from July–August (Natural Resources Canada 2019).  3.2.2 Monthly drought indices I calculated two monthly drought indices for the Palliser fire-weather station from 1901–2004 as follows. I compiled monthly maximum temperature (°C) and total monthly precipitation (mm; as water equivalent) from daily weather records and monthly climate data following procedures described in Chavardès et al. (2019). To generate monthly average temperature (°C), we used ClimateNA 5.30, a program that applies bilinear interpolation and elevation adjustments to downscale baseline gridded climate data to scale-free point data (Wang et al. 2016). For each fire season (April–October), I computed monthly DC and monthly PDSI. I calculated monthly DC from monthly precipitation and maximum temperature data following procedures described in Chavardès et al. (2019). Higher (lower) monthly DC values represent drier (wetter) moisture contents in deep compact organics in the soil and coarse woody fuels (Turner 1972). 39  I calculated monthly PDSI following Wells et al. (2004), using the monthly precipitation and average temperature data, average temperature normals calculated from 1901–2004, and available water-holding capacity of the soil of 480 mm corresponding to forested areas (Meyn et al. 2010b). Positive (negative) monthly PDSI values represent wetter (drier) moisture content conditions in surface and underlying soil layers (Palmer 1965), which comprise the duff to uppermost soil mineral horizon.  3.2.3 Developing the Douglas-fir latewood-width chronology into a detrended residual chronology I analyzed latewood widths of 147 Douglas-fir increment cores collected from six sites located in montane forests of southeastern British Columbia (Daniels et al. 2007). Cores were sanded and subsequently scanned at high resolution (1200 or 2400 dots per inch) to measure latewood widths using CooRecorder with a threshold of 70 to determine boundaries between earlywood and latewood (Larsson 2011). To develop a latewood-width chronology, I crossdated ring dates using the programs COFECHA (Holmes 1983) and CDendro (Larsson 2011). Only series with correlations ≥0.54 over the lifespan of the tree (n = 29) were combined into the latewood-width chronology to represent average latewood-width growth in the study area. To describe the latewood-width chronology I calculated statistics including the mean series intercorrelation (r), mean sensitivity (MS), and first order autocorrelation (AC1) (Cook and Kairiukstis 1990).  Using the Climatic Research Unit Standardization of the Tree-ring data program (Melvin and Briffa 2014), I applied signal-free detrending (Melvin and Briffa 2008) with a 100-year 50% frequency response to remove age- and stand-related growth trends (Cook and Peters 1981). We 40  computed the detrended chronology as a pre-whitened residual chronology to reduce temporal autocorrelation (Fritts 1976). I also stabilized variances and calculated the mean chronology as the weighted robust mean of the detrended chronology (Melvin and Briffa 2008). For the resulting detrended residual chronology, I calculated the expressed population signal (Wigley et al. 1984) from 1901–2004, the common period among the chronology and drought indices.  3.2.4 Climate-growth analyses To identify the strength of the climate-growth relationships between the detrended residual chronology, and the monthly drought indices, I conducted correlation function analyses (Fritts 1976) with the “treeclim” R package (Zang and Biondi 2015). I conducted these analyses over the 7-month window corresponding to the fire season (current year April–October) from 1901–2004. I calculated Pearson’s product moment correlation coefficients between the detrended residual chronology and each monthly drought index using 95% confidence intervals derived from 1,000 random bootstrapped samples (Zang and Biondi 2015).   3.2.5 Reconstructions of summer drought I used two reconstructions of summer drought: a reconstruction of summer (June–August) DC, which I developed in this study, and a published reconstruction of summer (June–August) PDSI for the nearest grid point from the Palliser fire-weather station (GP 67, 50°N, 115°W; Cook et al. 2004). I applied a linear model using the “lm” function in R (R Core Team 2018) to model summer DC as a function of the detrended residual chronology. To verify model assumptions were met with regards to normality, homoscedasticity, and serially independent data, I used 41  Shapiro-Wilk, Breusch-Pagan and Durbin-Watson tests, respectively, in addition to quantile-quantile and residual plots (Everitt and Skrondal 2010). The residual plot and Breusch-Pagan test revealed heteroscedastic error terms, which was accounted for by fitting a generalized least squares model (Fox and Weisberg 2019) with the error variance ln(𝛆𝛆2) = ln(𝐗𝐗). To assess model fit over 1901–2004, I calculated R2, reduction of error (RE) (Fritts 1976; Macias-Fauria et al. 2012), and root mean squared error (RMSE). To verify model consistency over time, I used the following split-sample periods: 1901–1952, 1953–2004, 1901–2003 every other year, and 1902–2004 every other year (after Littell et al. 2016). Validation statistics included R2 for the sub-sample models and RE for the sub-sample validations (after Littell et al. 2016). I tested the association between the reconstructions of summer DC and summer PDSI over their common period, hereafter the “period of analysis”. To test the association between both reconstructions, I calculated over the period of analysis their Pearson product moment correlation with associated p-value.   3.2.6 Fire-drought associations I followed three approaches to explore fire-drought associations during the period of analysis. Superposed epoch analysis was applied to determine whether fire years or non-fire years occurred after and during significantly drier or wetter than average summer conditions (Sutherland et al. 2015), bivariate event analysis was applied to determine whether fire years or non-fire years occurred after and during extremely dry or wet summer conditions (e.g., Harvey and Smith 2017), and logistic regression was applied to determine the summer drought thresholds that corresponded to presence or absence of fire years. For the historical fire records, I 42  used crossdated fire-scars collected from montane forests of southeastern British Columbia (Cochrane 2007; Daniels et al. 2011). I composited all site-level fire records, then filtered them to identify years in which ≥2 trees were scarred for a total of 68 fire-scar years (hereafter, “fire years”) over the period of analysis. I applied SEA in the Fire History Analysis and Exploration System (Brewer et al. 2015). To meet the assumptions of SEA, I used autoregressive integrated moving average (ARIMA) procedures to test for and remove autocorrelation (after Heyerdahl et al. 2008b) from the reconstruction of summer PDSI. ARIMA procedures showed that the reconstruction of summer PDSI was not white noise (p <0.001 up to six lags). I applied an AR(1) model to derive white noise residuals of the reconstruction of summer PDSI (p ≥0.547 up to 24 lags). I standardized the reconstruction of summer DC to z-scores (Salkind 2007) and calculated mean values for these z-scores and the reconstruction of summer PDSI residuals for fire years, non-fire years, and each of their preceding three years. To assess statistical significance of these years, I compared their mean values against bootstrapped values derived from a Monte Carlo simulation of randomly selected years that provided 95%, 99%, and 99.9% confidence intervals (Sutherland et al. 2015).  I applied BEA using the K1D software (Gavin 2010) to test whether extremely dry/wet summers (±1 standard deviation) were temporally synchronous, independent or asynchronous relative to fire years, non-fire years, and each of their preceding three years. Compared to SEA, BEA focuses on extreme rather than average conditions, and it presents the comparative advantage of avoiding problems due to serial autocorrelation because it does not rely on continuous drought records (Gartner et al. 2012). I conducted BEA over the period of analysis with forward selection such that fire years and non-fire years were preceded by extremely 43  dry/wet summers, and I generated 95% confidence intervals based on 1000 randomized Monte Carlo simulations (after Harvey and Smith 2017).  As a third approach, logistic regression was applied to predict the probability of a fire year based on the reconstructions of both summer DC and summer PDSI. An optimal pair of threshold values for summer DC and PDSI was derived from the cutoff probability at which the received operating characteristic (ROC) curve minimized the sum of false positives and false negatives (Toms and Villard 2015).  3.3 Results 3.3.1 Reconstruction of summer drought from Douglas-fir latewood widths The Douglas-fir latewood-width chronology was robustly crossdated with a high mean series intercorrelation (r = 0.62). Chronology statistics indicated low variation but high correlation between latewood-width measurements from year to year (mean sensitivity = 0.33 and first order autocorrelation (AC1) = 0.70, respectively), which were analogous to Watson and Luckman’s (2002) chronology statistics from Interior British Columbia. The detrended residual latewood-width chronology spanned from 1686–2004 and it achieved an expressed population signal >0.85 over 1901–2004, the common period among monthly drought indices. Increased drought during the fire season was significantly correlated with decreased Douglas-fir latewood width (Figure 3.2). Both monthly drought indices showed significant correlations with the chronology (p <0.05) during each month of the fire season; with negative correlations for monthly DC and positive correlations for monthly PDSI. Correlations with monthly DC were stronger from April–August, but weaker in September and October relative to monthly PDSI. 44  The model used to reconstruct summer DC met the assumptions of linear regression. The model showed moderate performance yielding an R2 of 0.46, RE of 0.45, and RMSE of 59.04 (Table 3.1). Validation statistics revealed similar values of R2 ranging from 0.41–0.49, and of RE ranging from 0.37–0.45. Over the common period of analysis spanning from 1686–2004, the summer DC reconstruction revealed a moderate negative Pearson product moment correlation with the summer PDSI reconstruction (r2 = -0.50; p <0.001) (Figure 3.3).  3.3.2 Fire-drought associations Summer drought was a driver of historical fires (Figure 3.4). From 1686–2004, the 68 fire years burned during summers with significantly drier conditions in the year of fire according to both drought reconstructions (p <0.001), and the year preceding fire according to the summer PDSI only (p <0.05). During the 251 non-fire years, summers were significantly wetter only during the coinciding year according to summer DC (p <0.05) and summer PDSI (p <0.01). As expected, fire years were synchronous with extremely dry summers and asynchronous with extremely wet summers (Figure 3.5). Fires were synchronous with extremely dry summers during the year of and prior year to fire according to summer DC (p <0.05), and year of and up to three years prior to fire according to summer PDSI (p <0.05). Corroborating these findings, fires were asynchronous with extremely wet summers during the year of and up to three years prior to fire according to summer DC and PDSI (p <0.05). Non-fire years remained independent (p ≥0.05) from extremely dry and wet summers. Based on the pair of identified thresholds, fire years were more likely to occur when summer DC ≥344 and summer PDSI ≤0.08 (Figure 3.6). Over 319 years, these conditions were met for 100 distinct years, 41% of which were fire years that were recorded at the greatest 45  percentage of sites (maximum 32% mean ±SD 9 ±6%). In contrast, when summer DC <344 and summer PDSI >0.08, only 5 of 109 years were fire years recorded at a lower percentage of sites (maximum 11% mean ±SD 7 ±4%) than under more pronounced and persistent drought conditions.  3.4 Discussion 3.4.1 Monthly DC and PDSI reveal subtle differences in drought influences on montane forests Subtle differences in climate-growth correlations across the fire season indicate that monthly DC and PDSI can be used together to better understand drought effects on historical fires in the montane forests of southeastern British Columbia. Compared to monthly PDSI, monthly DC was more strongly correlated with latewood-widths in the early months of the fire season. Monthly DC, developed as a fire-weather index, accounts for soil moisture depletion between fire seasons (Chavardès et al. 2019). Soil moisture can be partly retained between fire seasons by discontinuous frozen ground and insulating snowpack or it can be depleted by dry winters with early snowmelt (Van Wagner 1987; Lawson and Armitage 2008). The latter phenomenon is known as “overwintering” and is common in the study area. The capacity to account for overwintering likely explains the stronger climate-growth correlations from April–August for monthly DC than monthly PDSI. In contrast to monthly DC, monthly PDSI poorly accounts for snowfall (Riley et al. 2013), snowmelt, or frozen ground which can be problematic when using PDSI at high-latitudes and mountainous areas (Alley 1984). Nonetheless, during September and October, monthly PDSI displayed higher climate-growth correlations than monthly DC, which was consistent with correlations between monthly PDSI and soil moisture 46  that were strongest during late summer and autumn in the western USA (Dai et al. 2004). Combined, my findings provide support for the application of monthly DC along with monthly PDSI to improve interpretations of the effects of drought throughout the fire season in montane forests of southeastern British Columbia.  3.4.2 Fire-drought associations highlight differences in moisture content over time Fire-drought associations for summer DC and PDSI reflected the capacities of these two drought indicators to integrate moisture content over time in different forest fuels. Deep organics and coarse woody fuels have slow drying rates and hence long moisture content lag-times because of the low atmospheric influences on deep organics, including lack of influence by light precipitation events ≤2.8 mm (Van Wagner 1987; Girardin and Wotton 2009), and the small surface-area-to-volume ratio of coarse wood (Flannigan et al. 2016). In comparison, the duff layer responds more quickly to changes in moisture and is not influenced by overwintering effects (Wotton 2009). Differences in drying rates and overwintering effects meant that summer DC was significantly drier during fire years only. Summer DC during fire years and preceding years was also influenced by antecedent moisture conditions during and between prior fire seasons. Because average conditions were wetter two and three years prior to fire years, I found summer DC was drier, but not significantly so, during the preceding year. This difference in lagged response was reproduced when testing fire-drought associations during extremely dry summers. Summer PDSI showed synchrony during the year of and up to three years prior to fire, summer DC only revealed synchrony with extremely dry summers during the year of and one year prior to fire. Consequently, I conclude that for montane forests of southeastern British 47  Columbia, summer DC is a more suitable indicator of the degree and persistence of droughts across successive fire seasons than summer PDSI.  3.4.3 Fire severity indicators in montane forests with mixed-severity fire regimes In montane forests with mixed-severity fire regimes, summer DC and PDSI can be applied together with the fire scar record to make inferences on historical fire types and effects. Historically, fire years were most likely when moisture content was low in the duff layer and deep organic and coarse woody fuels. These conditions occurred during 100 distinct years between 1686 and 2004, implying high potential for sustained ignitions and fuel combustion. During 41 of these years, fire-scars were recorded at the greatest percentage of sites indicating low- to moderate-intensity fires that burned but did not necessarily kill trees. Because fire severity is related to the amount of combustible fuel (Keeley 2009), I would expect some fires burned at high severities given that most fuels had low moisture content. However, spatial variation in moisture gradients and the amount of combustible fuels likely influenced fire severity over space as found in other montane forests with mixed-severity fire regimes in western North America (Halofsky et al. 2011; Daniels et al. 2017). Despite mostly dry fuels, the spatial variation implied some fuels still had enough moisture or that the amounts of fuel were quite low in some stands. Alternatively, it is possible that these fires were of mixed severity, and that fire-scarred trees sampled were survivors in patches of low- to moderate-severity fires or at edges of high-severity fires.   Increased moisture content across fuels led to considerably fewer fire years relative to non-fire years and fewer sites recording fires. For 104 distinct years, when summer DC remained below 344 and summer PDSI was above 0.08 and ranged from ‘near normal’ to ‘extremely wet’ 48  (see Table 11 in Palmer 1965), no fire scars were detected across the study sites. Only five fire years were recorded under such mild drought conditions indicating forest fuels including the duff layer were less susceptible to sustain ignitions (Falk et al. 2007; Abatzoglou et al. 2016). When ignitions were sustained, higher moisture content across fuels likely limited the combustion of these fuels and the formation of fire scars at different sites. As a result, fire years were recorded at fewer sites than during years with more pronounced and persistent drought conditions coinciding with a fire year. Over the past 319 years, fire years were not common when summers coincided with low moisture content in deep organics and coarse woody fuels, but higher moisture content in the duff layer. Of the 55 years that coincided with summer DC ≥344, and summer PDSI ranging from ‘near normal’ to ‘very wet’ conditions, only 6 were fire years recorded at a maximum of 14% of sites (mean ±SD 6 ±4%). Daily and monthly DC values above 300 imply deep organics and coarse woody fuels are moderately to very dry and more likely to burn according to fire managers in Canada (Girardin and Wotton 2009). Combined, these conditions suggest deep organics and coarse wood had comparatively lower moisture content than the duff layer which may have been the result of two factors. First, it is possible that these six fire years followed a drought in the previous year and that overwintering conditions also led to moisture content remaining low in deep organics and coarse wood (Lawson and Armitage 2008). Second, given that moisture content in these fuels requires daily precipitation events that exceed 2.8 mm (Van Wagner 1987), it is also possible that these six years had relatively frequent light precipitation events over the summer that were <2.8 mm. If this were the case, then such precipitation events would have provided some moisture to the duff layer but not to deep organics and coarse woody fuels (Van Wagner 1987). Consistent with my interpretations, I found that all six years had 49  summer DC values that were >300 in the previous year, and had summer PDSI corresponding to ‘near normal’ to ‘incipient wet spell’ conditions during the fire year.   Sustained ignitions were more likely, but combustion of deep organics and coarse woody fuels less likely, when summer PDSI ranged from ‘near normal’ to ‘severe drought’, and summer DC was <344. Out of the 55 years, 16 were fire years recorded at a maximum of 20% of sites (mean ±SD 7 ±4%). Fires burning under these conditions emphasize the importance of a dry duff layer to sustain ignition and combustion (Wotton et al. 2005). However, because summer DC during the 16 fire years indicates that deep organics and coarse woody fuels were less combustible than under pronounced and persistent drought conditions, fires were more likely low- to moderate-severity fires. Similar findings were documented in other montane forests of western North America with evidence showing low- to moderate-severity fires burning during less pronounced and persistent droughts (Perry et al. 2011; Hessburg et al. 2016).  3.5 Conclusion Using the reconstructions of summer DC and PDSI in concert provided nuanced insights on the range of drought conditions driving mixed-severity fire regimes in montane forests of southeastern British Columbia. The development of a summer DC reconstruction in particular provided a practical connection between historical fires and a modern fire-weather index derived from empirical research on fire behaviour. Comparisons between summer DC and PDSI reconstructions revealed how moisture content among forest fuels, namely deep organics and coarse woody fuels versus the duff layer, can offer indications on the susceptibility to fire ignition and combustion. Distinct lag-times according to fire-drought associations reflected different drying rates, wetting susceptibility, and overwintering capacity among forest fuels 50  represented by each summer drought reconstruction. Variable fuel moisture conditions across fuels during fire years likely led to fires igniting and burning with a range of severities. Such information can be used by fire managers to understand how climate change could affect fire ignition and fuel combustion susceptibility in montane forests with historical mixed-severity fire regimes.  51  Table 3.1. Detrended residual latewood-width chronology-based reconstruction of the summer Drought Code (DC; dimensionless index) at the Palliser weather station including reconstruction, model and validation statistics.  Reconstruction coefficients  r (summer DC) Estimate Std. Error t-value P(>|t|) Intercept   603.1 30.2 19.940 <0.001 Detrended residual latewood-width chronology  -0.66 -258.7 27.8 -9.303 <0.001        Model diagnostics  Sub-sample model R2 Sub-sample validation RE R2 = 0.46  1901–1952 0.48 1901–1952 0.38 F-statistic = 86.54 on 1 and 102 df, p <0.001  1953–2004 0.41 1953–2004 0.37 RE = 0.44  1901–2003, every other year 0.49 1902–2004, every other year 0.38 RMSE = 59.04  1902–2004, every other year 0.41 1901–2003, every other year 0.45            52    Figure 3.1 (a) Location of the sampling sites for Douglas-fir increment cores (Daniels et al. 2007), and fire-scar records (Cochrane 2007; Daniels et al. 2011), (b) in montane forests of southeastern British Columbia (BC), Canada. (a) Also shown are the location for the summer Drought Code (DC) reconstruction at the Palliser fire-weather station, and the summer Palmer Drought Severity Index (PDSI) reconstruction (Cook et al. 2004) at grid point 67.  53   Figure 3.2 Bootstrapped correlation coefficients of the detrended residual latewood-width chronology against the monthly Drought Code (monthly DC; dimensionless), and monthly Palmer Drought Severity Index (monthly PDSI; dimensionless) at the Palliser fire-weather station. All correlation coefficients shown as grey bars were calculated for the period 1901–2004 and over April–October, the months corresponding to the fire season. Whiskers represent 95% confidence intervals.       54   Figure 3.3 Association between reconstructions of the summer Drought Code (DC) and summer Palmer Drought Severity Index (PDSI) from 1686–2004. The 68 fire years are shown as vertical grey bars.         55   Figure 3.4 Superposed epoch analyses showing associations for the reconstructions of the summer Drought Code (DC) and the summer Palmer Drought severity Index (PDSI) during fire years and non-fire years. Solid, long- and short-dashed lines represent confidence intervals of 99.9%, 99%, and 95%, respectively. Gray bars indicate statistically significant departures from the mean (p <0.05).       56    Figure 3.5 Bivariate event analyses showing associations for fire years and non-fire years during coinciding and preceding extremely dry or wet summers (±1SD) according to the reconstructions of the summer Drought Code (DC) and summer Palmer Drought severity Index (PDSI). Solid black lines represent reconstruction indices and dashed lines the 95% simulation envelopes, respectively. Reconstruction indices above the upper envelopes indicate synchrony with fire years or non-fire years. Reconstruction indices values falling between envelopes are statistically insignificant and indicate independence from fire years or non-fire years. Reconstruction indices below the lower envelopes indicate asynchrony with fire years or non-fire years. Portions highlighted in grey denote statistical significance (p <0.05).     57    Figure 3.6 Scatterplot for the reconstruction of the summer Drought Code (DC) versus reconstruction of summer Palmer Drought Severity Index (PDSI) during fire years (black circles) and non-fire years (grey points). The percentage of sites recording each fire year is represented by the size of each circle. Dashed lines correspond to summer DC = 344 and summer PDSI = 0.08.       58  Chapter 4: Fire-climate associations in the Canadian Montane Cordillera: Climate drives fire synchrony in forests with mixed-severity fire regimes  4.1  Introduction Anthropogenic climate change is leading to prolonged drought conducive to fire occurrence with high-severity impacts in forests of western North America (Flannigan et al. 2009; Westerling 2016; Wotton et al. 2017). Ultimately, this will result in more frequent fires placing human communities at risk (Moritz et al. 2014; Sankey 2019). For example, within the Montane Cordillera ecozone of Canada (hereafter, Montane Cordillera), many large (>200 ha) forest fires, including mega-fires (>10 000 ha), overwhelmed the deployment and control capabilities of fire suppression organizations during the 2017 and 2018 fire seasons (Natural Resources Canada 2019). Across the region, these fire seasons coincided with record breaking persistent warm and dry weather conditions corresponding to drought driven by anthropogenic climate change (Kirchmeier-Young et al. 2019). Such conditions over the fire season lower forest fuel moisture content, and thus facilitate the ignition, combustion and spread of fires (Gedalof 2011; Macias Fauria et al. 2011). The high number of widespread fires burning across a single fire season within the region has been described as historically unprecedented. Determining whether there is historical precedence for similar widespread-fire synchrony across the region is an important question that needs to be addressed to support fire managers in anticipating potential climate change impacts on fire regimes. Associations between modern fire records and climate (i.e., weather averaged over monthly to annual scales) have been investigated across multiple regions of western North 59  America, largely focusing on the mid-20th to early 21st centuries given the availability of modern fire records and instrumental climate and drought records (e.g., Morgan et al. 2008; Littell et al. 2009; Meyn et al. 2010a, 2010b; Westerling et al. 2016). However, this period overlaps with changes in fire management including improvements in fire suppression technology (Bowman et al. 2009; Flannigan et al. 2009), which can confound fire-climate associations (Williams and Abatzoglou 2016). To avoid the confounding effect of fire management, dendropyrochronologists test fire-climate associations over longer periods using crossdated fire-scar chronologies and multi-century proxies of summer temperature or drought or annual precipitation that extend prior to the 20th century (Swetnam and Anderson 2008; Littel et al. 2016; Williams and Abatzoglou 2016). Individual fire-scar chronologies and climate proxies provide site-level baseline information on the frequency of years with fire and the past conditions in which they burned. When sites are combined into regional networks, evidence of historical fire synchrony (Swetnam 1993; Falk et al. 2011) and interannual associations between synchronous or widespread fires and climate can be deduced (Heyerdahl et al. 2008a, 2008b; Trouet et al. 2010; Margolis and Swetnam 2013). Historical fire-climate associations can also be investigated at finer spatial scales through spatially explicit prediction models applied to networks of climate reconstructions (e.g., Heyerdahl et al. 2008b; Trouet et al. 2010). Thus, analyzing historical fire-climate associations across spatial scales can provide key insights on climate conditions under which synchronous fires burned, and the potential for future fire synchrony within the region. Many of the fire-scar chronologies developed for the Montane Cordillera and in the northwestern USA were collected in low- to mid-elevation coniferous forests with mixed-severity fire regimes (e.g., Hessburg et al. 2007, 2016; Perry et al. 2011; Heyerdahl et al. 2012; 60  Marcoux et al. 2013; Chavardès and Daniels 2016; Harvey and Smith 2017). Mixed-severity fire regimes are represented by fires that burn across space and time with a broad range of severities, from low-severity surface fires to high-severity crown fires (Perry et al. 2011; Daniels et al. 2017). The range of fire severities is reflected by the spatial and temporal variation in mortality effects on vegetation, and consequently forest stands across the landscape are compositionally and structurally diverse (Halofsky et al. 2011; Daniels et al. 2017). Although climate is a well-documented top-down driver of fires in this region (e.g., Heyerdahl et al. 2008b; Harvey and Smith 2017; Chavardès et al. 2018), understanding the more nuanced influences of climatic variation on the range of fire severities remains poorly understood. In this research, I tested the hypothesis that spatio-temporal variation in climate was an important driver of fire synchrony within and among regions but that extreme droughts that facilitated synchronous fires across sites were rare. Specifically, I address the following two questions: 1) How frequently did historical fires burn synchronously among sites within the Montane Cordillera? 2) What climate conditions were associated with various levels of fire synchrony at regional and subregional scales? To answer these questions, I used fire-scar records previously sampled at 17 sites across the Montane Cordillera in western Canada and quantified the occurrence and frequency with which fires synchronously burned at multiple sites and regions in a given year (i.e., fire synchrony). I reconstructed regional climatic proxies for summer temperature, annual precipitation, and summer drought and tested for associations between climate and years of fire synchrony at regional and subregional scales. I also applied a spatial model to networks of climate reconstructions to characterize fire-climate associations at a subregional scale. My findings reveal that fire synchrony driven by droughts was common but that widespread-fire synchrony was rare and driven by pronounced droughts. 61  4.2 Methods 4.2.1 Study area This research was conducted in the central and southern parts of the Montane Cordillera of Canada covering approximately 310,000 km2 and extending from the eastern slopes of the Coast and Cascade Mountains in British Columbia to the foothills of the Rocky Mountain in Alberta, and from the northern conterminous USA border to the northern extent of the Cariboo Mountains (Figure 4.1). Climate in the region is continental, with some maritime influences in the south modulated by rain-shadow effects of the Coast Mountains in the west, and the Columbia and Rocky Mountains in the east (Scudder and Smith 2011). Topographic influences on climate contribute to diverse forest ecosystems dominated by coniferous tree species (Scudder and Smith 2011). In the warmest and driest areas at low- to mid-elevations in the interior plateau and in the valleys within and between the eastern mountains, forests are predominantly composed of interior Douglas-fir (Pseudotsuga menziesii var glauca (Beissn.) Franco) and ponderosa pine (Pinus ponderosa Dougl. ex Laws.), with western larch (Larix occidentalis Nutt.) present in the valleys within and between the eastern mountains (Demarchi 2011). In mid-elevation forests of the eastern mountains and Rocky Mountain foothills, climate is cooler and wetter than in the interior plateau or low- to mid-elevation valleys in the eastern mountains (Environment Canada 2019). Coniferous species also present within these forests include western red cedar (Thuja plicata Donn ex D.Don) and western hemlock (Tsuga heterophylla (Raf.) Sarg.) in the Columbia Mountains, and lodgepole pine (P. contorta var. latifolia Douglas) and white spruce (Picea Glauca (Moench) Voss) in the Rocky Mountains and its foothills (Scudder and Smith 2011). 62  4.2.2 Historical fire records I acquired the crossdated fire-scar records from fire-history reconstructions conducted in the Montane Cordillera and composited fire-scar chronologies into 17 site-level fire records (Appendix B). The site-level fire records combined multiple plot-level fire records sampled by different researchers reconstructing historical fire occurrence (Table 4.1). In total, fire records from the 17 sites comprised 1,886 crossdated fire-scar samples. The recording period for each site was defined by the years in which ≥2 living, fire-scarred trees were present and had the potential to form a scar in the advent of fire. At the site level, the fire records were compiled in two different ways. (1) Fires recorded by ≥2 trees in a given year were considered a “fire year” (Heyerdahl et al. 2008a), and (2) fires recorded by ≥2 trees and ≥10% of recorder trees in a given year were a “widespread-fire year” (Trouet et al. 2010). For each year, the number of sites recording a fire year or widespread-fire year were summed, generating a regional fire record. Based on the number of sites recording a fire year, I designated each year in one of five categories of synchrony: 1) “no fire synchrony” for years with fire scars at one site only; 2) “low fire synchrony” for years with fire scars at two to three sites; 3) “moderate fire synchrony” for years with fire scars at four to five sites; 4) “high fire synchrony” for years with fire scars at six or more sites; and 5) “synchronous non-fire years” when no fires were recording at all 17 sites (adapted from Heyerdahl et al. 2008a).  4.2.3 Reconstructions of regional climate To represent climate throughout the Montane Cordillera, I used tree-ring proxy reconstructions of maximum temperature anomalies, precipitation, and the Palmer Drought Severity Index (PDSI), a drought index which combines the effects of temperature and 63  precipitation (Palmer 1965). I used the reconstruction of summer maximum temperature anomalies for Interior British Columbia (Wilson and Luckman 2003), reconstructions of annual  precipitation (previous July to current June; Watson and Luckman 2004) for 11 locations that provided the longest periods of record (i.e., Banff, Big Creek, Cranbrook, Jasper, Lillooet, Lytton, North Thompson, Oliver, Summerland, Waterton, and Williams Lake), and reconstructions of summer PDSI for 20 grid points (Cook et al. 2004) that covered the study region (grid points: 22–25, 29–32, 40–43, 52–55, and 65–68) (Appendix B).  I developed and compared regional climate reconstructions over the period of analysis from 1746–1945. This period corresponds to the common period between the regional fire record and climate reconstructions, all of which start prior to or in 1746. The period ends in 1945 because of subsequent documented influences that can confound fire-climate associations including fire exclusion policies imposed at the turn of the 20th Century, and the introduction of advanced and organized fire suppression after 1945 in western North America (Pyne 1982, 2007; Keane et al. 2002). A single regional climate reconstruction was needed for each of the three sets of climate reconstructions (i.e., summer maximum temperature anomalies, annual precipitation, and summer PDSI). The reconstruction of summer maximum temperature anomalies for Interior British Columbia was originally developed to represent the region and was used without modification. For the other two sets of climate reconstructions, I used Principle Components Analysis (PCA) (SAS Institute Incorporated 2017) to extract the first principle component (PC1) from the 11 reconstructions of annual precipitation (PC1PPT), and PC1 from the 20 reconstructions of summer PDSI (PC1PDSI). PC1PPT and PC1PDSI were considered representative of regional annual precipitation and summer PDSI, respectively. I tested for linear correlations among the three regional climate reconstructions using scatter plots and by calculating Pearson 64  product moment correlations between the reconstruction of summer maximum temperature anomalies, PC1PPT, and PC1PDSI.  4.2.4 Regional fire-climate associations Over the period of analysis, I used two approaches to test if increasing fire synchrony was associated with warm and dry regional climate. First, I compared values of each regional climate reconstruction across categories of synchrony using boxplots and analysis of variance of ranks followed by post-hoc Dunn’s tests (Gorvine et al. 2018). Second, I conducted additional analyses on three subsets of years with moderate to high degree of synchrony (hereafter, collectively referred to as “synchronous events”): (1) moderate–high fire synchrony (i.e., ≥4 sites recorded fire), (2) moderate–high widespread-fire synchrony (i.e., ≥4 sites recorded widespread fire), and 3) synchronous non-fire years during the period of analysis. To test the associations between synchronous events and the regional climate reconstructions, I used Superposed Epoch Analysis (SEA) in the Fire History Analysis and Exploration System (Brewer et al. 2015). To meet the assumptions of SEA, I used Autoregressive Integrated Moving Average (ARIMA) procedures to test for and remove autocorrelation from the reconstruction of summer maximum temperature anomalies, PC1PPT, and PC1PDSI. ARIMA procedures showed that the reconstruction of summer maximum temperature anomalies did not have statistically significant temporal autocorrelation (white noise tests p ≥0.188 up to 24 lags); however, PC1PPT and PC1PDSI were not white noise (p ≤0.016 up to six lags). I applied AR(1) models to derive white noise residuals of PC1PPT and PC1PDSI (p ≥0.57 up to 24 lags). For the reconstruction of summer maximum temperature anomalies, and PC1PPT and PC1PDSI residuals, I calculated mean values during the year coinciding with extreme synchronous events and the three preceding years and compared them to 65  bootstrapped values derived from a Monte Carlo simulation of randomly selected years that provided 95%, 99%, and 99.9% confidence intervals (Sutherland et al. 2015).  4.2.5 Subregional fire-climate associations Over the period of analysis, I used two approaches to visually explore climate conditions during years with high fire synchrony and synchronous events. For the first approach, to represent precipitation anomalies across space, I calculated z-scores (Salkind 2007) for each of the 11 reconstructions of annual precipitation. Using the multiple point reconstructions of annual precipitation z-scores (n = 11) and summer PDSI (n = 20) across the study area, I applied Inverse Distance Weighted (IDW) interpolation (Bivand et al. 2008) to develop continuous maps depicting climate conditions during each year with high fire synchrony. For the second approach, I calculated means for each of the 11 reconstructions of annual precipitation z-scores and 20 reconstructions of summer PDSI during synchronous events. I applied IDW interpolation to develop continuous maps depicting average climate conditions during synchronous events. For both approaches, to describe climate conditions and to colour-code the maps, annual precipitation z-scores, and summer PDSI values were assigned to one of 11 classes ranging from ‘extremely wet’ to ‘extreme drought’ (see Table 11 in Palmer 1965; Appendix C).  4.3 Results 4.3.1 Historical fire records The 17 site-level fire records revealed abundant fire activity between 1735 and 2000 (Figure 4.2; Table 4.1). From 1746–1945, fires scarred trees in 186 years; 90 of these showed widespread scarring at one or more sites. Of the 186 fire years, 48 were recorded at one site only. 66  Low, moderate, and high fire synchrony were recorded in 88, 39, and 11 years, respectively. Only 14 years had no fire scars at any of the sites.  4.3.2 Reconstructions of regional climate Based on PCA, variances explained for PC1PPT (33%) and PC1PDSI (73%) indicated the reconstructions of annual precipitation exhibited more spatial variability across the 11 locations than the reconstructions of summer PDSI across the 20 grid points. Diagnostic plots and Pearson product moment correlations revealed the strongest linear correlation was negative between the reconstruction of summer maximum temperature anomalies and PC1PDSI (r2 = -0.74, p <0.001) (Figure 4.3). In contrast, PC1PPT had a moderately strong positive linear correlation with PC1PDSI (r2 = 0.62, p <0.001), and a moderately weak negative linear correlation with the reconstruction of summer maximum temperature anomalies (r2 = -0.35, p <0.001).  4.3.3 Regional fire-climate associations Years when fires burned synchronously at multiple sites were generally associated with warmer summer maximum temperature anomalies and significantly associated with lower annual precipitation and summer PDSI, indicating relatively dry conditions (Figure 4.4). During the 50 years with moderate–high fire synchrony (fire interval mean and range: 4 yrs, 1–15 yrs), climate conditions were significantly drier and warmer in the year coinciding with fire (p <0.001) and/or previous year (p <0.05) (Figure 4.5a–c). Conditions were drier and warmer, on average, during the six years with moderate–high widespread-fire synchrony (fire interval mean and range: 26 yrs, 10–63 yrs); however, the associations were only statistically significant during the year coinciding with fire (p <0.01) (Figure 4.5d–f). In contrast to years with moderate–high fire 67  synchrony and widespread-fire synchrony, conditions during the 14 non-fire years were significantly wetter than average (p <0.01), and cooler and wetter in the preceding year (p <0.05) (Figure 4.5g–i).  4.3.4 Subregional fire-climate associations The 11 years with high fire synchrony coincided with a wide range of climatic conditions across the region. Annual precipitation z-scores spanned the full range of climatic conditions, from ‘extremely wet’ to ‘extreme drought’ (Figure 4.6). Compared to annual precipitation z-scores, summer PDSI spanned a narrower range of climatic conditions, from ‘slightly wet’ to ‘extreme drought’ (Figure 4.7). Distinct climatic patterns coincided with the three categories of synchronous events. During the 50 years with moderate–high fire synchrony, mean annual precipitation z-scores showed ‘mild drought’ at eastern sites and most northern sites, whereas ‘incipient drought’ to ‘near normal’ conditions were found at most western sites (Figure 4.8a; Appendix D). Mean summer PDSI values showed ‘near normal’ conditions at the northern sites along the continental divide and increased from ‘incipient drought’ to ‘moderate drought’ with decreasing latitude (Figure 4.8b). During the six years with moderate–high widespread-fire synchrony, mean annual precipitation z-scores corresponded to ‘severe drought’ and ‘extreme drought’ at eastern and northern sites, and ‘near normal’ to ‘moderate drought’ conditions at southwestern sites (Figure 4.8c). Mean summer PDSI values showed ‘incipient drought’ to ‘extreme drought’ conditions from the northwest to southeast (Figure 4.8d).  68  The 14 non-fire years mean annual precipitation z-scores and summer PDSI values showed mostly ‘incipient wet’ to ‘moderately wet’ conditions across the region with wetter conditions in the south and east (Figure 4.8e and 4.8f).  4.4 Discussion 4.4.1 Fire synchrony mostly coincided with drought In the Montane Cordillera of Canada, synchronous fires that scarred trees at ≥4 study sites (i.e., years of moderate-high fire synchrony) during the same year were common historically and generally facilitated by regionally warm and dry climate during the fire season and preceding year. My findings corroborate those found in several regions of western North America showing greater fire synchrony during droughts (e.g., Heyerdahl et al. 2008a, 2008b; Trouet et al. 2010; Margolis and Swetnam 2013); however, I also found that regional drought was not a necessary condition for fire synchrony. Several years with moderate–high fire synchrony occurred during regionally wetter-than-normal conditions (e.g., 1827, 1828, 1866, and 1915). As explained by Gedalof et al. (2005) and Macias Fauria et al. (2011), warm, dry, or windy weather for periods of only two to three weeks during the fire season can lower the moisture content of some forest fuels sufficiently to enable ignition and spread. Following short periods that dry fuels, low-intensity fires that scar but do not kill trees could have occurred in low- and mid-elevation forests in the region. These fires may have been constrained spatially due to adjacent forest fuels that maintained high moisture content and temporally due to the return of weather conditions less favourable for burning, such as precipitation, lower temperature, higher relative humidity, and lack of wind (Flannigan and Harrington 1988; Meyn et al. 2007). A likely 69  consequence of these spatial and temporal constraints would be fewer widespread fires, which is consistent with the fire-scars formed in 1827, 1828, 1866, and 1915.  On average, moderate–high fire synchrony coincided with significantly drier-than-average climate, particularly in the Columbia and Rocky Mountains that are located in the eastern part of the region. Specifically, most years with high fire synchrony revealed that fires in the Columbia and Rocky Mountains were more likely to burn during summer droughts that followed low antecedent annual precipitation. Such conditions occurred during 8 of the 11 years with high fire synchrony (1831, 1847, 1863, 1864, 1869, 1883, 1889, and 1896). My findings likely reflect the presence of mid-tropospheric anomalies, and possibly documented phases of Pacific oceanic-atmospheric teleconnections such as the El Niño-Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) that were conducive to fire synchrony. Mid-tropospheric anomalies in the form of upper-atmosphere blocking ridges were related to lower moisture, fuel drying, and increased fire activity over western Canada (Skinner et al. 1999), including the Canadian Rocky Mountains (Johnson and Wowchuck 1993). Over the Pacific Northwest, including some of the sites from the Interior Plateau of British Columbia, Kitzberger et al. (2007) and Heyerdahl et al. (2008b) found fire synchrony coincided with dry climate related to warm phases of ENSO and PDO. However, the findings from these studies contrasted the results of Macias Fauria and Johnson (2006, 2008) who reported cool phases of ENSO and PDO coincided with increased fire activity over most of the Montane Cordillera, and the findings of Harvey and Smith (2017) and Chavardès et al. (2018) who documented no significant relationships. Lack of congruence among these studies may be due to a strong expression of rain shadow effects on the Interior Plateau (Harvey and Smith 2017), and a climatic transition zone along the Columbia and Rocky Mountains (Macias Fauria and Johnson 2008). As a post-hoc analysis I used published 70  reconstructions of Pacific oceanic-atmospheric teleconnections including ENSO (Stahle et al. 1998, D’Arrigo et al. 2005; Braganza et al. 2009; Wilson et al. 2010), the Pacific North American circulation pattern (Trouet and Taylor 2010; Liu et al. 2017), PDO (Biondi et al. 2001; D’Arrigo et al. 2001; Gedalof and Smith 2001; D’Arrigo and Wilson 2006; MacDonald and Case 2005), and Pacific Decadal Variability (Black et al. 2016) that overlapped with the period of analysis, to test whether opposing phases were associated with categories of synchrony. I found none of the 11 reconstructions revealed significant associations between phases and categories of synchrony based on analysis of variance on ranks (p ≥0.381). A possible explanation for the lack of association with historical fire synchrony in the region is that these teleconnections had distinct subregional influences (Watson and Luckman 2005), that were temporally unstable (e.g., Knapp et al. 2002; Black et al. 2016).   4.4.2 Uncommon, synchronous widespread-fires were driven by pronounced drought Although widespread fires were common at the site level, synchronous widespread-fires that scarred trees at ≥4 sites were uncommon and exclusively coincided with pronounced drought at regional and subregional scales. In 1831, 1847, 1869, and 1889, pronounced drought during the summer preceded by low annual precipitation particularly in the eastern parts of the region coincided with widespread fires at sites in both the Interior Plateau and Rocky Mountains. Similarly, drought conditions were found to facilitate historical widespread fires in forests of the northwestern USA (Hessl et al. 2004; Gedalof et al. 2005; Heyerdahl et al. 2008a, 2008b). However, I found many individual years with regional drought conditions between 1746 and 1945 that lacked evidence of synchronous fires at our study sites. This finding indicates that regional drought does not guarantee fire occurrence and synchrony (Gedalof 2011). There are 71  several potential explanations for the lack of evidence of fire synchrony during these years. Despite drought pre-conditioning forest fuels for fire spread, there may have been few ignitions from lightning or people during those particular years. Fires burned but their intensity was insufficient to scar trees (Swetnam et al. 1999) or intense fires killed rather than scarred trees. Fires were recorded at only a few sites as a result of spatial variation in the degree of drought. Alternately, fires burned and scarred trees but the evidence was subsequently eliminated by successive fires, wood decomposition, and mechanical damage (Parson et al. 2007; Swetnam et al. 2011).  The high number of fire-scars formed during years with multiple widespread fires highlights the historic importance of surface fires even during regionally pronounced droughts affecting low- and mid-elevation coniferous forests of the Montane Cordillera. I surmise the historical extensiveness of surface fires may have been promoted by different factors. For example, in 1831 and 1847, pronounced summer drought conditions covered most of the region except in parts of the Interior Plateau that had near normal to incipient drought conditions. However, even under these fairly low drought conditions, sites in the Interior Plateau had abundant evidence of surface fires. In the low- to mid-elevation dry mixed-conifer forests of the Interior Plateau, fuels are more likely sufficiently desiccated by the regular periods of warm, dry weather therefore climate is less of a fire regime control than in more mesic forests (Heyerdahl et al. 2007, 2012; Harvey et al. 2017). As a result of surface fires capable of burning even under low drought conditions, fires can be fairly frequent, which is corroborated by the fire-scar records from sites in the Interior Plateau. Fairly frequent surface fires suggest that surface fuels and potential ladder fuels were more likely consumed thus further facilitating surface fires relative to crown fires (Agee 1993; Bond and Keeley 2005). Another possible factor is that some 72  of the surface fires may have burned over periods that did not coincide with peak fire season, such as during the early or late months of the fire season when drought conditions were less pronounced. Fires occurring during months with milder drought conditions likely consumed fewer fuels due to higher fuel moisture contents and as a result fires were more likely of lower severity (e.g., Chavardès et al. 2019).  4.4.3 Changes during the 20th century Decreased surface fire evidence across the Montane Cordillera after 1945 reflects findings from western North American forests, which relate the decrease to a period combining the imposition of fire exclusion policies at the turn of the 20th century, the modernization of fire suppression organizations (Keane et al. 2002; Donovan and Brown 2007; Pyne 1982, 2007), and less fire-conducive climate over several decades in the Northern Rocky Mountains of the USA (Higuera et al. 2015), and across the province of British Columbia (Meyn et al. 2010a). As a result of decreased fire activity, changes in fuel structure and availability were identified, including persistence of ladder fuels and increased canopy closure in some western USA forests (e.g., Taylor and Skinner 2003; Brown et al. 2004; Schoennagel et al. 2004; Higuera et al. 2015) and forests of the Montane Cordillera (e.g., Daniels et al. 2011; Marcoux et al. 2015; Chavardès and Daniels 2016; Harvey et al. 2017). My findings suggest that the resilience associated with frequent surface fires is eroding with the evidence of these fires from our records across the Montane Cordillera starting in the mid-20th century. As the climate of western North America became more conducive to fire after the mid-1980s via increased temperatures and prolonged fire seasons, fire activity also increased in western US forests (Westerling et al. 2006; Higuera et al. 2015; Westerling 2016). In Canada, 73  increased fire activity in forests of the Montane Cordillera was not detected until the early 21st Century (Filmon et al. 2003; Tzembelicos et al. 2018; Kirchmeier-Young et al. 2019). A possible explanation for the lagged response may stem from strong fire suppression policies maintained until the early 21st Century. For example, these policies were maintained until 2012 in British Columbia and it is only afterwards that more flexible management approaches were implemented such as permitting some fires to burn when these do not threaten communities and infrastructure (Government of British Columbia 2012). An added explanation is that lower temperatures at higher latitudes mitigated increases in lightning ignitions (Price and Rind 1994) and fire activity (Wang et al. 2014). However, as temperatures further increased into the 21st Century, and landscapes were subject to fewer surface fires, conditions likely shifted towards supporting large fires with high-severity effects that pose risks to human communities (Sankey 2019). Such fires in the region are likely to occur again given the availability of forest fuels, expansion of human communities, and climate-change predictions including longer fire seasons, increased cumulative fire severity ratings (Flannigan et al. 2013), and increased lightning ignitions (Price and Rind 1994; Krawchuk et al. 2009).  4.5 Conclusion I conclude that until 1945 years when fires scarred trees at ≥4 sites (i.e., moderate–high fire synchrony) were common within the Montane Cordillera, but that years when widespread fires scarred trees at ≥4 sites (i.e., moderate–high widespread-fire synchrony) and when none of the study sites included fire scars were fairly rare. Warm and dry climate particularly during the summer at regional and subregional scales facilitated moderate–high fire synchrony; although, these conditions needed to be more pronounced for moderate–high widespread-fire synchrony. 74  Decreased fire synchrony after 1945 coincided with advances in fire suppression and less fire conducive climate during several decades. The drop in fire synchrony across the region is consistent with increased ladder fuels and canopy cover found in many forests, suggesting that seasonal climate is becoming a less important driver of fire than fuel abundance and short term variations in weather. Regional predictions including longer fire-seasons, and increased fire-severity ratings suggest that these forests may include more frequent, high-severity fires that can synchronize as observed during the 2017 and 2018 fire seasons in the Montane Cordillera.                75  Table 4.1 Characteristics of the 17 site-level fire records arranged from north to south in the Montane Cordillera of Canada.                   Site Location No. of fire-scar samples Complete fire record (years) Period of analysis (1746–1945) First fire year Last fire year No. of fire years No. of widespread fire years 1 53°50'N 117°47'W 62 1640–2009 1756 1934 9 5 2 53°03'N 118°05'W 102 1604–2011 1772 1915 12 5 3 52°54'N 118°04'W 170 1735–1974 1846 1944 41 4 4 52°04'N 121°54'W 65 1558–2015 1747 1943 12 7 5 51°55'N 123°13'W 139 1513–2013 1747 1945 73 5 6 51°40'N 121°40'W 139 1491–1995 1756 1938 32 7 7 51°26'N 122°18'W 92 1520–2012 1748 1943 56 11 8 50°48'N 115°58'W 83 1693–2009 1848 1926 11 4 9 50°18'N 121°58'W 155 1511–1999 1754 1937 62 17 10 50°16'N 121°39'W 162 1608–1996 1749 1941 57 19 11 50°10'N 115°47'W 149 1431–2006 1757 1938 25 5 12 50°01'N 116°05'W 108 1353–2005 1748 1937 34 8 13 49°34'N 117°10'W 45 1665–2009 1834 1927 10 6 14 49°34'N 115°46'W 43 1522–2003 1751 1891 20 10 15 49°25'N 115°40'W 96 1510–2007 1751 1923 17 6 16 49°17'N 119°33'W 149 1611–2013 1759 1945 31 17 17 49°13'N 116°45'W 127 1407–2009 1768 1926 18 9                   Sources for each site-level fire record: 1: Amoroso et al. 2011; 2: Chavardès et al. 2018; 3: Dinh 2014; 4: Brookes 2019; 5: Harvey and Smith 2017; 6: Daniels and Watson 2003; 7: Harvey et al. 2017; 8: Kubian 2013; 9: Heyerdahl et al. 2008, 2012; 10: Heyerdahl et al. 2007; 11: Cochrane 2007; 12: Daniels et al. 2006, Gray and Daniels 2007; 13: Nesbitt 2010; 14: Nesbitt and Daniels 2009; 15: Marcoux et al. 2013, Villemaire-Côté 2014; 16: Pogue 2017; 17: Greene and Daniels 2017.   76   Figure 4.1 (A) Centroid locations for the sites with fire records and locations of the climate reconstructions across the Montane Cordillera of Canada in (B) western North America.          77   Figure 4.2 Fire records from 1735–2000 for the 17 sites within the Montane Cordillera. The top panel shows the fire record for each site organized by latitude from northernmost to southernmost locations, the temporal extent of each fire record from the starting year among fire records as a horizontal line, each fire year as a gray tick, and each widespread-fire year as a black tick. The bottom panel shows the number of sites with a given fire year as gray columns, and widespread-fire year as black columns. Years in which no sites recorded fire are synchronous non-fire years and shown as black columns extending downward for emphasis. The curve indicates the number of crossdated fire-scar samples over time.   No. of siteswith fire yearSiteNo. of samples06824101716151413121112345678910200010000NorthSouth1850 1950 20001750 1800 1900Calendar year78   Figure 4.3 Scatter plots for the reconstructions of regional climate for the Montane Cordillera. Reconstructions of regional climate corresponds to the reconstruction of summer maximum temperature anomalies (C°), the first principle component of the 11 reconstructions of annual precipitation (PC1PPT), and the first principle component of the 20 reconstructions of summer Palmer Drought Severity Index (PC1PDSI). Drier (wetter) conditions according to PC1PPT and PC1PDSI lie below (above) zero.    79   Figure 4.4 Reconstructions of regional climate according to categories of synchrony in the Montane Cordillera. Reconstructions of regional climate correspond to the reconstruction of summer maximum temperature anomalies (C°), the first principle component of the 11 reconstructions of annual precipitation (PC1PPT), and the first principle component of the 20 reconstructions of summer Palmer Drought Severity Index (PC1PDSI). Drier (wetter) conditions according to PC1PPT and PC1PDSI lie below (above) zero. Categories of synchrony correspond to: non-fire years = no fire years recorded at all 17 sites (n = 14); single site = fire year recorded at one site only (n = 48); low = fire year recorded at two to three sites (n = 88); moderate = fire year recorded at four to five sites (n = 39); and high = fire year recorded at six or more sites (n = 11) (adapted from Heyerdahl et al. 2008a). In each box plot, the black horizontal line represents the median, box boundaries are the 25th and 75th percentiles, and bars are the 10th and 90th percentiles. Different letters denote significant differences among median reconstructed regional climate values (α = 0.05).  80   Figure 4.5 Superposed epoch analyses showing associations between reconstructions of regional climate during synchronous events (moderate–high fire synchrony or widespread-fire synchrony, or synchronous non-fire years) from 1746–1945. Reconstructions of regional climate correspond to z-scores of the reconstruction of summer maximum temperature anomalies, white noise residuals of the first principle component for the 11 reconstructions of annual precipitation (PC1PPT residuals), and white noise residuals of the first principle component for the 20 reconstructions of summer Palmer Drought Severity Index (PC1PDSI residuals). Solid, long- and short-dashed lines represent confidence intervals of 99.9%, 99%, and 95%, respectively. Gray bars indicate statistically significant departures from the mean (p <0.05). 81   Figure 4.6 Reconstructions of mean annual precipitation z-scores interpolated across the Montane Cordillera during the 11 years with high fire synchrony. Darker reds (blues) indicate drier (wetter) conditions, whereas grey indicates near normal conditions (see Appendix D for more details). Small and large open circles indicate a fire and widespread fire was found at the site during the given year, respectively, whereas an ‘x’ indicates no fire scars were found at the site during the given year. 82   Figure 4.7 Reconstructions of mean summer Palmer Drought Severity Index (PDSI) values interpolated across the Montane Cordillera during the 11 years with high fire synchrony. Darker reds (blues) indicate drier (wetter) conditions, whereas grey indicates near normal conditions (see Appendix D for more details). Small and large open circles indicate a fire and widespread fire was found at the site during the given year, respectively, whereas an ‘x’ indicates no fire scars were found at the site during the given year. 83   Figure 4.8 Reconstructions of mean annual precipitation z-scores, and mean summer Palmer Drought Severity Index (PDSI) values interpolated across the Montane Cordillera from 1746–1945 during years with synchronous events (moderate–high fire synchrony or widespread-fire synchrony, or synchronous non-fire years). Darker reds (blues) indicate drier (wetter) conditions, whereas grey indicates near normal conditions (see Appendix D for more details). Open circles indicate the 17 sites. 84  Chapter 5: Conclusion 5.1 Thesis Contribution and Summary Understanding drought as a driver of mixed-severity fire regimes across temporal and spatial scales in the Montane Cordillera is important because some of the fire regimes in this region are expected to include increased frequency of large fires with a greater proportion of high-severity impacts (Daniels et al. 2011; 2017). In western North America, including the Montane Cordillera, warming temperatures were associated with longer fire seasons comprising more frequent and large fires (Westerling et al. 2006; Dennison et al. 2014; Kirchmeier-Young et al. 2019). When combined with 20th century increases in ladder fuels and canopy cover documented in some forests with mixed-severity fire regimes (e.g., Daniels et al. 2011; Marcoux et al. 2015; Chavardès and Daniels 2016; Harvey et al. 2017), the changes to the fire season can favour higher-severity fires in these forests. Moreover, further warming over the 21st Century is expected to increase fire-severity ratings (Flannigan et al. 2013) and the number of lightning ignitions (Price and Rind 1994; Krawchuk et al. 2009). The findings from this thesis contribute to the understanding of how drought influences mixed-severity fire regimes across temporal and spatial scales in the Montane Cordillera of Canada as follows. In Chapter 2, three monthly adaptations of the daily Drought Code (DC) of Canada’s Fire Weather Index System were compared and applied to interpret drought conditions associated with historical fires in montane forests of southeastern British Columbia within the Montane Cordillera. The three adaptations were compared with the monthly mean DC calculated from daily values for the Palliser fire-weather station. Two adaptations improved on the existing Monthly DC calculated from monthly climate data by (1) accounting for overwinter drying and an early start to the fire season, and (2) improving estimates of effective precipitation. Using a 85  crossdated fire-scar record from 20 sites in montane forests surrounding the Palliser station, revealed significant fire-drought associations from June−August with all adaptations, and significant associations in April and May with the two new adaptations. Of the 17 fire years from 1901−2013, six years had low initial drought conditions that increased late in the fire season, and five years had high drought conditions throughout the fire season. The findings from Chapter 2 highlight that variable drought conditions within and among fire seasons influenced fire severity.  In Chapter 3, drought influences on mixed-severity fire regimes in montane forests of southeastern British Columbia were further investigated. Initially, a Douglas-fir latewood-width chronology was developed and tested for associations with monthly drought records across the fire season. Associations were strong between records and latewood-widths, particularly from June–August. Based on the chronology, a summer Drought Code was reconstructed, and compared to a reconstruction of the Palmer Drought Severity Index. Comparing reconstructions provided a nuanced understanding of moisture content among forest fuels, namely in deep compact organics in the soil and coarse woody fuels versus the duff layer. Fire-drought associations were tested using the reconstructions and fire-scar records from montane forests of southeastern British Columbia. These tests revealed fire years were associated with previous and current year summer droughts; but limited by previous and current summer wet conditions. The findings from Chapter 3, highlight that variable fuel moisture conditions across fuels during fire years according to the reconstructions likely led to fires igniting and burning with a range of severities. Chapter 4 provided information on fires that occurred at multiple sites during historical fire seasons in the Montane Cordillera, and the climate conditions under which these fires burned. Local fire-scar chronologies were composited into 17 site-level records comprising 86  1,886 fire-scar samples. The composite fire record was tested against regional climate reconstructions to examine historical fire-climate associations for years with synchronous events, namely years when fires and widespread fires scarred trees at ≥4 sites and non-fire years, when none of the study sites included fire scars. Findings revealed that years with fires at ≥4 sites (i.e., moderate–high fire synchrony) occurred during 50 distinct years from 1746–1945. Years with widespread-fires at ≥4 sites (i.e., moderate–high widespread-fire synchrony) and no fires recorded at all sites were less common, occurring during 6 and 14 years, respectively. At regional and subregional scales, years with moderate–high fire synchrony coincided with warm and dry climate, with pronounced drought conditions occurring during years with moderate–high widespread-fire synchrony. From 1945 to 2000, decreased fire synchrony partly reflects improvements in fire suppression technology. Given anticipated longer fire-seasons and increased fire-severity ratings, more frequent, widespread, and potentially high-severity fires will have a greater chance of burning.  5.2 Management Implications The 2017 and 2018 fire seasons in the Montane Cordillera of western Canada challenged fire managers to accept that extreme back-to-back summer droughts with multiple large synchronous fires can overwhelm fire suppression capacities (Sankey 2019). To adapt to this new reality of climate change impacts on fire regimes within the region, fire managers should use novel information that can support them strategically. The drought indices and methods developed and applied in my research provide practical information on drought variability and fire-drought associations across temporal and spatial scales in the Montane Cordillera. Apart from gaining a historical perspective on the drought conditions during which fires burned at, fire 87  managers can compare this information to current drought conditions coinciding with fires and droughts anticipated due to climate change. Specifically, understanding how climate change could affect seasonal droughts can reveal critical information on the frequency of fire seasons with suitable and safe conditions for prescribed burns (Parks Canada and the Canadian Parks Council 2008; British Columbia Wildfire Service 2010). Likewise, detailed information about differences among drought indicators provides fire managers with long-term perspectives on fire behaviour such as the likelihood of sustained ignitions and fires burning across a range of severities. Such knowledge can ultimately guide sustainable fire management strategies that account for the historic range of variability in these forests (Keane et al. 2009). Finally, given the historical evidence that fires burned at multiple locations across the region during extreme droughts and climate change predictions indicating increased fire-severity ratings (Flannigan et al. 2013) and lightning ignitions (Price and Rind 1994), I suggest that fire managers should plan for future scenarios with higher frequency of fire seasons with extreme droughts and multiple large fires across the Montane Cordillera.  5.3 Directions for Future Research 5.3.1 Comparing the strength of fire-drought associations among fire regimes To test the hypothesis that the strength of fire-drought associations varies among fire regimes (Agee 1993; Schoennagel et al. 2004), I suggest calculating monthly DCs and testing their associations with historical and modern fire records at different locations across western Canada including similar montane forests, drier low-elevation forests, and cooler and more mesic subalpine forests and boreal forests. Compared to drier low-elevation forests, I expect stronger fire-drought associations in more mesic forests because they require prolonged periods of 88  drought that facilitate ignition and combustion (Macias Fauria et al. 2011). However, I also expect to find fewer fire years in these forests relative to drier forests. Fewer fire years will widen the confidence limits in statistical approaches such as superposed epoch analysis. To address this problem, a combined approach that also uses more flexible statistical models (e.g., generalized linear mixed models) may be more appropriate to test fire-drought associations. Ultimately, such research consolidates understanding of fire-drought associations across ecoclimatic regimes in western Canada.  5.3.2 Anticipating drought influences on future fire activity in montane forests of southeastern British Columbia The summer DC and Palmer Drought Severity Index (PDSI) thresholds can be applied to meteorological data and climate predictions to assess how climate change could impact fire activity for montane forests with mixed-severity fire regimes in southeastern British Columbia. Specifically, when summer DC ≥344 and summer PDSI ≤0.08, fire managers can expect high probability of fire at one to multiple sites in these forests. Past these thresholds, my findings suggest that the duff layer is dry and facilitates ignitions, and that fires are more likely to burn dry deep organics in the soil and coarse woody fuels. To assess the impacts of climate change on the mixed-severity fire regime in these montane forests, I suggest to compare drought conditions during historical and future years according to summer DC and PDSI and their respective thresholds.  89  5.3.3 Assessing changes in fire severity for forests of the Montane Cordillera The historical fire records including 1886 fire-scar samples from 17 sites provide evidence that low-severity surface fires were common in some low- and mid-elevation forests of the Montane Cordillera. By comparison, the many large fires in 2017 and 2018 included extensive high tree mortality indicting high-severity effects (Natural Resources Canada, unpublished data). I suggest to resurvey the sites from which the fire-scar records were collected to assess whether some of the 2017 and 2018 fires may have burned at high severity at these sites. If this is the case, then I may find some of the fire-scarred trees may have been consumed by these latter fires. Such a finding would provide evidence that some sites historically burned by fairly frequent low-severity surface fires are including more high-severity fires.  5.3.4 Interregional fire-climate associations focusing on teleconnections Although other research has detected influences of Pacific oceanic-atmospheric teleconnections on fires in western North America (Kitzberger et al. 2007; Trouet et al. 2010), including the Pacific Northwest (Hoffman et al. 2016), and Southern Interior British Columbia (Heyerdahl et al. 2008b), I did not find associations between such teleconnection reconstructions and the regional fire records from forests of the Montane Cordillera. Two potential limitations are that the teleconnections were unstable over time (e.g., Knapp et al. 2002; Black et al. 2016) and their reconstructions reproduce the instability (e.g., Kipfmueller et al. 2012), or the teleconnections had distinct interregional influences. I propose to test the latter hypothesis with fire-scar records from sites located in sufficiently distant study areas such that differences in fire-climate associations can be detected.  90  Bibliography Abatzoglou, JT, Kolden CA, Balch JK, Bradley BA (2016) Controls on interannual variability in lightning-caused fire activity in the western US. Environmental Research Letters 11 (4), 1–12, 10.1088/1748-9326/11/4/045005 Agee JK (1993) Fire Ecology of Pacific Northwest Forests. Island Press, Washington, DC, USA, pp. 1–505 Agriculture and Agri-Food Canada (2019) Drought Watch, Agroclimate Maps, Canadian Drought Monitor. Government of Canada, Agriculture and Agri-Food Canada, Science and Technology Branch. Alexander ME (1982) Calculating spring Drought Code starting values in the Prairie Provinces and Northwest Territories. Forest Management Note 12, Northern Forest Research Centre, Canadian Forest Service, Edmonton, AB, Canada, pp. 1–4 Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier M, Kitzberger T, Rigling A, Breshears DD, Hogg EH, Gonzalez P, Fensham R, Zhang Z, Castro J, Demidova N, Lim J, Allard G, Running SW, Semerci A, Cobb N (2010) A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecology and Management 259 (4), 660–684, 10.1016/j.foreco.2009.09.001 Allen CD, Breshears DD, McDowell NG (2015) On underestimation of global vulnerability to tree mortality and forest die‐off from hotter drought in the Anthropocene. Ecosphere 6 (8), 1–55, 10.1890/ES15-00203.1 Alley WM (1984) The Palmer Drought Severity Index: Limitations and assumptions. Journal of Climate and Applied Meteorology 23, 1100–1109, 10.1175/1520-0450(1984)023<1100:TPDSIL>2.0.CO;2 91  Amoroso MM, Daniels LD, Bataneih M, Andison DW (2011) Evidence of mixed-severity fires in the foothills of the Rocky Mountains of west-central Alberta, Canada. Forest Ecology and Management 262 (12), 2240–2249, 10.1016/J.FORECO.2011.08.016 Arienti MC, Cumming SG, Boutin S (2006) Empirical models of forest fire initial attack success probabilities: The effects of fuels, anthropogenic linear features, fire weather, and management. Canadian Journal of Forest Research 36 (12), 3155–3166, 101139/X06-188 Bergeron Y, Cyr D, Girardin MP, Carcaillet C (2010) Will climate change drive 21st century burn rates in Canadian boreal forest outside of its natural variability: Collating global climate model experiments with sedimentary charcoal data. International Journal of Wildland Fire 19 (8), 1127–1139, 101071/WF09092 Bergeron Y, Gauthier S, Flannigan M, Kafka V (2004) Fire regimes at the transition between mixedwood and coniferous boreal forest in northwestern Quebec. Ecological Society of America 85 (7), 1916–1932, 10.1890/02-0716 Biondi F, Gershunov A, Cayan DR (2001) North Pacific decadal climate variability since 1661. Journal of Climate 14 (1), 1–6, 10.1175/1520-0442(2001)014<0005::NPDCVS>2.0.CO;2 Bivand RS, Pebesma EJ, Gómez-Rubio V (2008) Applied spatial data analysis with R. Springer, New York, NY, USA, pp. 1–378 Black BA, Grissino-Mayer HD, Chavardès RD, Tucker CS, Thornton MP (2016) A 450-year reconstruction of Pacific Decadal Variability. Final Report of the 26th North American Dendroecological Fieldweek to the United States of America National Science Foundation, 48–53 92  Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens HH, White JS (2009) Generalized linear mixed models: A random practical guide for ecology and evolution. Trends in Ecology and Evolution 24 (3), 127–135, 101016/jtree200810008 Bond WJ, Keeley JE (2005) Fire as a global ‘herbivore’: The ecology and evolution of flammable ecosystems. Trends in Ecology and Evolution 20 (7), 387–394, 10.1016/j.tree.2005.04.025 Bowman DMJS, Balch JK, Artaxo P, Bond WJ, Carlson JM, Cochrane MA, D’Antonio CM, DeFries RS, Doyle JC, Harrison SP, Johnston FH, Keeley JE, Krawchuk MA, Kull CA, Marston JB, Moritz MA, Prentice IC, Roos CI, Scott AC, Swetnam TW, van der Werf GR, Pyne SJ (2009) Fire in the Earth System. Science 324 (5926), 481–484, 10.1126/science.1163886 Bradshaw LS, Deeming JE, Burgan RE, Cohen JD (1983) The 1978 National Fire-Danger Rating System: Technical documentation. General Technical Report INT-69, US Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station, Ogden, UT, USA, pp. 1–46 Braganza K, Gergis JL, Power SB, Risbey JS, Fowler AM (2009) A multiproxy index of the El Niño-Southern Oscillation, A.D. 1525-1982. Journal of Geophysical Research (Atmospheres) 114, 1–17, 10.1029/2008JD010896 Brewer PW, Velásquez ME, Sutherland EK, Falk DA (2015) Fire History Analysis and Exploration System (FHAES) version 2.0, http://www.fhaes.org. doi:10.5281/zenodo.593323 British Columbia Ministry of Environment (2019) BC Air Data Archive. BC, Canada British Columbia Ministry of Forests, Lands, Natural Resource Operations and Rural Development (2018) Impacts of 2017 fires on timber supply in the Cariboo region. Forest 93  Analysis and Inventory Branch, Office of the Chief Forester, British Columbia Ministry of Forests, Lands, Natural Resource Operations and Rural Development, BC, Canada, pp. 1–12 British Columbia Wildfire Management Branch (2012) Wildfire management branch strategic plan 2012–2017. Ministry of Forests, Lands and Natural Resource Operations, BC Wildfire Service, BC, Canada, pp. 1–24 British Columbia Wildfire Service (2010) Wildfire management strategy. Government of British Columbia, British Columbia Wildfire Service, BC, Canada, pp. 1–21 Brown RT, Agee JK, Franklin JF (2004) Forest restoration and fire: Principles in the context of place. Conservation Biology 81 (4), 903–912, 10.1111/j.1523-1739.2004.521_1.x Brookes WA (2019) Historical and contemporary disturbance regimes in central interior dry forests of British Columbia. MSc thesis, University of British Columbia, Vancouver, BC, Canada Canadian Forest Service (1984) Tables for the Canadian Forest Fire Weather Index System. Technical Report 25, Environment Canada, Canadian Forestry Service, Ottawa, ON, Canada, pp. 1–48 Carnwath GC, Peterson DW, Nelson CR (2012) Effect of crown class and habitat type on climate-growth relationships of ponderosa pine and Douglas-fir. Forest Ecology and Management 285, 44–52, 10.1016/j.foreco.2012.07.037 Chavardès RD, Daniels LD, Waeber PO, Innes JL, Nitschke CR (2013) Unstable climate–growth relations for white spruce in southwest Yukon, Canada. Climatic Change 116 (3–4), 593–611, 101007/s10584-012-0503-8 94  Chavardès RD, Daniels LD (2016) Altered mixed-severity fire regime has homogenized montane forests of Jasper National Park. International Journal of Wildland Fire 25 (4), 433–444, 10.1071/WF15048 Chavardès RD, Daniels LD, Gedalof Z, Andison DW (2018) Human influences superceded climate to disrupt the 20th century fire regime in Jasper National Park, Canada. Dendrochronologia 48, 10–19, 10.1016/j.dendro.2018.01.002 Chavardès RD, Daniels LD, Eskelson BNI, Pickell PD (2019) Monthly adaptations of the Drought Code reveal nuanced fire-drought associations in montane forests with a mixed-severity fire regime. International Journal of Wildland Fire, 10.1071/WF18119  Chen PY, Welsh C, Hamann A (2010) Geographic variation in growth response of Douglas-fir to interannual climate variability and projected climate change. Global Change Biology 16 (12), 3374–3385, 10.1111/j.1365-2486.2010.02166.x Choat B, Jansen S, Brodribb TJ, Cochard H, Delzon S, Bhaskar R, Bucci SJ, Feild TS, Gleason SM, Hacke UG, Jacobsen AL, Lens F, Maherali H, Martínez-Vilalta J, Mayr S, Mencuccini M, Mitchell PJ, Nardini A, Pittermann J, Pratt RB, Sperry JS, Westoby M, Wright IJ, Zanne AE (2012) Global convergence in the vulnerability of forests to drought. Nature 491, 752–755, 10.1038/nature11688 Cóbar-Carranza AJ, García RA, Pauchard A, Peña E (2014) Effects of Pinus contorta invasion on forest fuel properties and its potential implications on the fire regime of Araucaria araucana and Nothofagus antartica forests. Biological Invasions 16 (11), 2273–2291, 10.1007/s10530-014-0663-8 Cochrane JD (2007) Characteristics of historical forest fires in complex mixed-conifer forests of southeastern British Columbia. MSc thesis, University of British Columbia, Vancouver, BC, 95  Canada Cook ER, Peters K (1981) The smoothing spline: A new approach to standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bulletin 41, 45–53 Cook ER, Meko DM, Stahle DW, Cleaveland MK (1996) Tree-ring reconstructions of past drought across the conterminous United States: Tests of a regression method and calibration/verification results. In: Dean JS, Meko DM, Swetnam TW (Eds.) Tree Rings, Environment, and Humanity. University of Arizona, Tucson, AZ, USA, pp. 155–169 Cook ER, Meko DM, Stahle DW, Cleaveland MK (1999) Drought reconstructions for the continental United States. Journal of Climate 12, 1145–1162, 10.1175/1520-0442(1999)012<1145:DRFTCU>2.0.CO;2 Cook ER, Woodhouse CA, Eakin CM, Meko DM, Stahle DW (2004) Long-term aridity changes in the western United States. Science 306 (5698), 1015–1018, 10.1126/science.1102586 Cook ER, Seager R, Heim RR, Vose RS, Herweijer C, Woodhouse CA (2010) Megadroughts in North America: Placing IPCC projections of hydroclimatic change in a long-term paleoclimate context. Journal of Quaternary Science 25 (1), 48–61, 10.1002/jqs.1303 Crawford CJ, Griffin D, Kipfmueller KF (2015) Capturing season-specific precipitation signals in the northern Rocky Mountains, USA, using earlywood and latewood tree rings. Journal of Geophysical Research: Biogeosciences 120 (3), 428–440, 10.1002/2014JG002740 Dai A, Trenberth KE, Qian T (2004) A global dataset of Palmer Drought Severity Index for 1870–2002: Relationship with soil moisture and effects of surface warming. Journal of Hydrometeorology 5 (6), 1117–1130, 10.1175/JHM-386.1 96  Daniels LD, Watson E (2003) Climate-fire-vegetation interactions in the Cariboo Forests: A dendrochronological analysis. Report to Forest Innovation and Investment Forest Research Program, BC, Canada, pp 1–57 Daniels LD, Cochrane J, Gray RW (2006) Refining mixed-severity fire regimes in the Rocky Mountain Forest District. Report to Tembec Inc., Canadian Forest Products Ltd., and the Forest Investment Account of British Columbia, pp. 1–26 Daniels LD, Cochrane J, Gray RW (2007) Mixed-severity fire regimes: Regional analysis of the impacts of climate on fire frequency in the Rocky Mountain Forest District. Report to Tembec Inc., BC Division, Canadian Forest Products Ltd., Radium Hot Springs, and the Forest Investment Account of British Columbia, BC, Canada, pp 1–36 Daniels LD, Maertens TB, Stan AB, McCloskey SPJ, Cochrane JD, Gray RW (2011) Direct and indirect impacts of climate change on forests: Three case studies from British Columbia. Canadian Journal of Plant Pathology 33 (2), 108–116, 10.1080/07060661.2011.563906 Daniels LD, Yocom Kent LL, Sherriff RL, Heyerdahl EK (2017) Deciphering the complexity of historical fire regimes: Diversity among forests of Western North America. In: Amoroso MM, Daniels LS, Baker PJ, Camarero JJ (Eds.) Dendroecology: Tree-ring analyses applied to ecological studies, pp. 185–210, Springer International Publishing, Cham, Switzerland Daniels J (2017) Death toll in California wildfires grows to 42, PG&E sued by couple who lost home. CNBC, Eagleawood cliffs, NJ, USA Dansereau P, and Bergeron Y (1993) Fire history in the southern boreal forest of northwestern Quebec. Canadian Journal of Forest Research 23 (1), 25–32, 10.1139/x93-005 D'Arrigo R, Villalba R, Wiles G (2001) Tree-ring estimates of Pacific decadal climate variability. Climate Dynamics 18 (3–4), 219–224, 10.1007/s003820100 97  D'Arrigo R, Wilson R (2006) On the Asian Expression of the PDO. International Journal of Climatology 26 (12), 1607–1617, 10.1002/joc.1326 D'Arrigo R, Cook E, Wilson R, Allan R, Mann M (2005) On the variability of ENSO over the past six centuries. Geophysical Research Letters 32 (3), 1–4, 10.1029/2004GL022055 de Groot WJ, Pritchard J, Lynham TJ (2009) Forest floor fuel consumption and carbon emissions in Canadian boreal forest fires. Canadian Journal of Forest Research 39 (2), 367–382, 10.1139/X08-192 Demarchi DA (2011) An introduction to the ecoregions of British Columbia. Ecosystem Information Section, Ministry of Environment, Government of British Columbia, Victoria, BC, Canada, pp. 1–163 Diffenbaugh NS, Swain DL, Touma D (2015) Anthropogenic warming has increased drought risk in California. Proceedings of the National Academy of Sciences of the Unite States of America 112 (13), 3931–3936, 10.1073/pnas.1422385112 Dinh T (2014) Influence of humans and climatic variability on historic wildfire dynamics in Jasper National Park, Alberta, Canada. MSc thesis, University of Guelph, Guelph, ON, Canada Donovan GH, Brown TC (2007) Be careful what you wish for: The legacy of Smokey Bear. Frontiers in Ecology and the Environment 5 (2), 73–79, 10.1890/1540-9295(2007)5[73:BCWYWF]2.0.CO;2 Drobyshev I, Flannigan MD, Bergeron Y, Girardin MP, Suran B (2011) Variation in local weather explains differences in fire regimes within a Québec south-eastern boreal landscape. International Journal of Wildland Fire 19 (8), 1073–1082, 10.1071/WF09101 98  Drobyshev I, Goebel PC, Bergeron Y, Corace RG (2012) Detecting changes in climate forcing on the fire regime of a North American mixed-pine forest: A case study of Seney National Wildlife Refuge, Upper Michigan. Dendrochronologia 30 (2), 137–145, 10.1016/j.dendro.2011.07.002 Drobyshev I, Gewehr S, Berninger F, Bergeron Y (2013) Species specific growth response of black spruce and trembling aspen may enhance resilience of boreal forest to climate change. Journal of Ecology 101 (1), 231–242, 10.1111/1365-2745.12007 Environment Canada (2019) Canadian Climate Normals 1981–2010. National Climate Data and Information Archive, Ottawa, ON, Canada Falk DA, Miller C, McKenzie D, Black AE (2007) Cross-scale analysis of fire regimes. Ecosystems 10 (5), 809–823, 10.1007/s10021-007-9070-7 Falk DA, Heyerdahl EK, Brown PM, Farris C, Fulé PZ, McKenzie D, Swetnam TW, Taylor AH, Van Horne ML (2011) Multi-scale controls of historical forest-fire regimes: New insights from fire-scar networks. Frontiers in Ecology and the Environment 9 (8), 446–454, 101890/100052 Ferrell GT (1996) The influence of insect pests and pathogens on sierra forests. Sierra Nevada Ecosystem Project, vol. II, Assessments and scientific basis for management options. Centers for Water and Wildland Resources, University of California, Davis, CA, USA, pp. 1177–1192 Field RD, Spessa AC, Aziz NA, Camia A, Cantin A, Carr R, de Groot WJ, Dowdy AJ, Flannigan MD, Manomaiphiboon K, Pappenberger F, Tannipat V, Wang X (2015) Development of a global fire weather database. Natural Hazards and Earth System Sciences 15 (6), 1407–1423, 105194/nhess-15-1407-2015 99  Filmon G, Leitch D, Sproul J, Prescott L, Yau A, Armitage T, Zadravec D, Motter N, Maddess G, Mitchell D, Blackwell B, Gray R, Fikus J (2004) Firestorm 2003: Provincial review. Government of British Columbia, BC, Canada, pp. 1–100 Flannigan MD, Harrington JB (1988) A study of the relation of meteorological variables to monthly provincial area burned by wildfire in Canada (1953–80). Journal of Applied Meteorology 27, 441–452, 101175/1520-0450(1988)027<0441:ASOTRO>20CO;2 Flannigan MD, Stocks BJ, Weber MG (2003) Fire regimes and climatic change in Canadian forests. In: Veblen T, Baker W, Swetnam T (Eds.) Fire and climatic change in temperate ecosystems of the western Americas. Springer-Verlag, New York, NY, USA, pp. 97–119 Flannigan MD, Amiro BD, Logan KA, Stocks BJ, Wotton BM (2006) Forest fires and climate change in the 21st Century. Mitigration and Adaptation Strategies for Global Change 11 (4), 847–859, 10.1007/s11027-005-9020-7 Flannigan MD, Krawchuk MA, de Groot WJ, Wotton MB, Gowman LM (2009) Implications of changing climate for global wildland fire. International Journal of Wildland Fire 18 (5), 483–507, 10.1071/WF08187 Flannigan M, Cantin AS, de Groot WJ, Wotton M, Newbery A, Gowman LM (2013) Global wildland fire season severity in the 21st century. Forest Ecology and Management 294, 54–61, 10.1016/j.foreco.2012.10.022 Flannigan MD, Wotton BM, Marshall GA, de Groot WJ, Johnston J, Jurko N, Cantin AS (2016) Fuel moisture sensitivity to temperature and precipitation: Climate change implications. Climatic Change 134 (1–2), 59–71, 10.1007/s10584-015-1521-0 Fritts HC (1974) Relationships of ring-widths in arid-site conifers to variations in monthly temperature and precipitation. Ecological Monograph 44 (4), 411–440, 10.2307/1942448 100  Fritts HC (1976) Tree rings and climate. Academic Press, New York, New York, USA, pp. 1–567 Gartner MH, Veblen TT, Sherriff RL, Schoennagel TL (2012) Proximity to grasslands influences fire frequency and sensitivity to climate variability in ponderosa pine forests of the Colorado Front Range. International Journal of Wildland Fire 21 (5), 562–571, 10.1071/WF10103 Gavin DG (2010) K1D: Multivariate Ripley’s K-function for one dimensional data. University of Oregon, Department of Geography, pp. 1–8 Gedalof Z, Smith DJ (2001) Interdecadal climate variability and regime-scale shifts in Pacific North America. Geophysical Research Letters 28 (8), 1515–1518, 10.1029/2000GL011779 Gedalof Z, Peterson DL, Mantua NJ (2005) Atmospheric, climatic and ecological controls on extreme wildfire years in the northwestern United States. Ecological Applications 15 (1), 154–174, 10.1890/03-5116 Gedalof Z (2011) Climate and spatial patterns of wildfire in North America. In: McKenzie D, Miller C, Falk DA (Eds.) The Landscape Ecology of Fire. Springer, New York, NY, USA, pp. 89–115 Girardin MP, Tardif J, Flannigan MD, Bergeron Y (2004a), Multicentury reconstruction of the Canadian Drought Code from eastern Canada and its relationship with paleoclimatic indices. Climate Dynamics 23 (2), 99–115, 10.1007/s00382-004-0417-x Girardin MP, Tardif J, Flannigan MD, Wotton BM, Bergeron Y (2004b) Trends and periodicities in the Canadian Drought Code and their relationships with atmospheric circulation for the southern Canadian boreal forest. Canadian Journal of Forest Research 34 (1), 103–119, 10.1139/x03-195 101  Girardin MP, Wotton BM (2009) Summer moisture and wildfire risks across Canada. Journal of Applied Meteorology and Climatology 48, 517–533, 101175/2008JAMC19961 Girardin MP, Ali AA, Carcaillet C, Mudelsee M, Drobyshev I, Hély C, Bergeron Y (2009) Heterogeneous response of circumboreal wildfire risk to climate change since the early 1900s. Global Change Biology 15 (11), 2751–2769, 101111/j1365-2486200901869x Girardin MP (2010) Wildfire risk inferred from tree rings in the Central Laurentians of boreal Quebec, Canada. Dendrochronologia 28(3), 187–206, 10.1016/j.dendro.2009.05.006 Girardin MP, Ali AA, Carcaillet C, Gauthier S, Hély C, Le Goff H, Terrier A, Bergeron Y (2013) Fire in managed forests of eastern Canada: Risks and options. Forest Ecology and Management 294, 238–249, 101016/jforeco201207005 González ME, Veblen TT, Sibold JS (2005) Fire history or Araucaria-Nothofagus forests in Villarrica National Park, Chile. Journal of Biogeography 32 (7), 1187–1202, 10.1111/j.1365-2699.2005.01262.x González ME, Veblen TT, Sibold JS (2010) Influence of fire severity on stand development of Araucaria araucana-Nothofagus pumilio stands in the Andean cordillera of south-central Chile. Austral Ecology 35 (6), 597–615, 10.1111/j.1442-9993.2009.02064.x Goodrick SL (2002) Modification of the Fosberg Fire Weather Index to include drought. International Journal of Wildland Fire 11 (4), 205–211, 101071/WF02005 Gorvine B, Rosengren K, Stein L, Biolsi K (2018) Research methods: From theory to practice. Oxford University Press, New York, NY, USA, pp. 1–474 Gray RW, Daniels LD (2007) An investigation of fire history in the lower Gold/Joseph Creek watershed. Report to the City of Cranbrook, BC, Canada, pp. 1–19 102  Greene GA, Villemaire-Côté O, Daniels LD (2014) Climate change, wildfire, and landscape homogenization in western Canada. Report to the Rocky Mountain Trench Society. Greene GA, Daniels LD (2017) Spatial interpolation and mean fire interval analyses quantify historical mixed-severity fire regimes. International Journal of Wildland Fire 26 (2) 136–147, 10.1071/WF16084 Griesbauer HP, Green SD (2010) Regional and ecological patterns in interior Douglas-fir climate–growth relationships in British Columbia, Canada. Canadian Journal of Forest Research 40 (2), 308–321, 10.1139/X09-197 Halofsky JE, Donato DC, Hibbs DE, Campbell JL, Cannon MD, Fontaine JB, Thompson JR, Anthony RG, Bormann BT, Kayes LJ, Law BE, Peterson DL, Spies TA (2011) Mixed-severity fire regimes: lessons and hypotheses from the Klamath–Siskiyou ecoregion. Ecosphere 2 (40), 1–19, 10.1890/ES10-00184.1 Hansen AJ, Neilson RP, Dale VH, Flather CH, Iverson LR, Currie DJ, Shafer S, Cook R, Bartlein PJ (2001) Global change in forests: Responses of species, communities, and biomes: Interactions between climate change and land use are projected to cause large shifts in biodiversity. BioScience 51 (9), 765–779, 101641/0006-3568(2001)051[0765:GCIFRO]20CO;2 Harvey JE, Smith DJ (2017) Interannual climate variability drives regional fires in west central British Columbia, Canada. Journal of Geophysical Research: Biogeosciences 122 (7), 1759–1774, 10.1002/2016JG003661 Harvey JE, Smith DJ, Veblen TT (2017) Mixed-severity fire history at a forest–grassland ecotone in west central British Columbia, Canada. Ecological Applications 27 (6), 1746–1760, 10.1002/eap.1563 103  Heim RR Jr (2002) A review of twentieth-century drought indices used in the United States Bulletin of the American Meteorological Society 83, 1149–1165, 101175/1520-0477(2002)083<1149:AROTDI>23CO;2 Helsel P (2018) Some allowed back into devastated town of Paradise after deadly California fire. NBC News, Washington, DC, USA Hessburg PF, Salter RB, James KM (2007) Re-examining fire severity relations in pre-management era mixed conifer forests: Inferences from landscape patterns of forest structure. Landscape Ecology, 22, 5–24, 10.1007/s10980-007-9098-2 Hessburg PF, Spies TA, Perry DA, Skinner CN, Taylor AH, Brown PM, Stephens SL, Larson AJ, Churchill DJ, Povak NA, Singleton PH, McComb B, Zielinski WJ, Collins BM, Salter RB, Keane JJ, Franklin JF, Riegel G (2016) Forest Ecology and Management 366, 221–250, 10.1016/j.foreco.2016.01.034 Hessburg PF, Miller CL, Parks SA, Povak NA, Taylor AH, Higuera PE, Prichard SJ, North MP, Collins BM, Hurteau MD, Larson AJ, Allen CD, Stephens SL, Rivera-Huerta H, Stevens-Rumann CS, Daniels LD, Gedalof Z, Gray RW, Kane VR, Churchill DJ, Hagmann RK, Spies TA, Cansler CA, Belote RT, Veblen TT, Battaglia MA, Hoffman C, Skinner CN, Safford HD, Salter RB (2019) Climate, environment, and disturbance history govern resilience of western North American forests. Frontiers in Ecology and Evolution 7 (239), 1–27, 10.3389/fevo.2019.00239 Hessl AM, McKenzie D, Schellhaas R (2004) Drought and Pacific Decadal Oscillation linked to fire occurrence in the inland Pacific Northwest. Ecological Applications 14(2), 425–442, 10.1890/03-5019 104  Heyerdahl EK, Brubaker LB, Agee JK (2001) Spatial controls of historical fire regimes: A multiscale example from the interior west, USA. Ecology 82 (3), 660–678, 101890/0012-9658(2001)082[0660:SCOHFR]20CO;2 Heyerdahl EK, Brubaker LB, Agee JK (2002) Annual and decadal climate forcing of historical fire regimes in the interior Pacific Northwest, USA. The Holocene 12 (5), 597–604, 10.1191/0959683602hl570rp Heyerdahl EK, Morgan P, Riser II JP (2008a) Multi-season climate synchronized widespread historical fires in dry forests (1630–1900), Northern Rockies, USA. Ecology 89 (3), 705–716, 10.1890/06-2047.1 Heyerdahl EK, McKenzie D, Daniels LD, Hessl AE, Littell JS, Mantua NJ (2008b) Climate drivers of regionally synchronous fires in the inland Northwest (1651–1900). International Journal of Wildland Fire 17 (1), 40–49, 10.1071/WF07024 Heyerdahl EK, Lertzman K, Wong CM (2012) Mixed-severity fire regimes in dry forests of southern interior British Columbia, Canada. Canadian Journal of Forest Research 42 (1), 88–98, 10.1139/X11-160 Higuera PE, Abatzoglou JT, Littell JS, Morgan P (2015) The changing strength and nature of fire-climate relationships in the northern Rocky Mountains, U.S.A., 1902–2008. PLoS ONE 10 (6), 1–21, 10.1371/journal.pone.0127563  Hoffman KM, Gavin DG, Starzomski BM (2016) Seven hundred years of human-driven and climate-influenced fire activity in a British Columbia coastal temperate rainsforest. Royal Society 3 (10), 1–12, 10.1098/rsos.160608 Holmes RL (1983) Computer-assisted quality control in tree-ring dating and measurement. Tree-Ring Bulletin 43, 69–78 105  Holz A, Veblen TT (2011) The amplifying effects of humans on fire regimes in temperature rainforests in western Patagonia. Paleogeography, Paleoclimatology, Paleoecology 311 (1–2), 82–92, 10.1016/j.palaeo.2011.08.008 Hu QS, Willson GD (2000) Effects of temperature anomalies on the Palmer Drought Severity Index in the central United States. International Journal of Climatology 20, 1899–1911 Johnson EA, Wowchuck DR (1993) Wildfires in the southern Canadian Rocky Mountains and their relationship to mid-tropospheric anomalies. Canadian Journal of Forest Research 23 (6), 1213–1222, 10.1139/x93-153 Johnson EA, Gutsell SL (1994) Fire frequency models, methods and interpretations. Advances in Ecological Research 25, 239–287 Johnson EA, Miyanishi K, Weir JMH (1998) Wildfires in the western Canadian boreal forest: Landscape patterns and ecosystem management. Journal of Vegetation Science 9 (4), 603–610, 102307/3237276 Johnson EA, Miyanishi K, O’Brien N (1999) Long-term reconstruction of the fire season in the mixedwood boreal forest of western Canada. Canadian Journal of Botany 77 (8), 1185–1188, 10.1139/b99-110 Johnson EA, Miyanishi K, Bridge SRJ (2001) Wildfire regime in the boreal forest and the idea of suppression and fuel buildup. Conservation Biology 15 (6), 1554–1557, 101046/j1523-1739200101005x Jolly WM, Cochrane MA, Freeborn PH, Holden ZA, Brown TJ, Williamson GJ, Bowman DMJS (2015) Climate-induced variations in global wildfire danger from 1979 to 2013. Nature Communications 6 (7537), 1–11,10.1038/ncomms8537 106  Keane RE, Ryan KC, Veblen TT, Allen CD, Logan J, Hawkes B (2002) Cascading effects of fire exclusion in Rocky Mountain ecosystems: A literature review. In: Baron J (Ed.) Rocky Mountain Futures: An ecological perspective. Island Press, Washington, DC, USA, pp. 133–152 Keane RE, Hessburg PF, Landres PB, Swanson FJ (2009) The use of historical range of variability (HRV) in landscape management. Forest Ecology and Management 258 (7), 1025–1037, 10.1016/j.foreco.2009.05.035 Keeley JE (2009) Fire intensity, fire severity and burn severity: A brief review and suggested usage. International Journal of Wildfire 18 (1), 116–126, 101071/WF07049 Keetch JJ, Byram GM (1968) A drought index for forest fire control. Research Paper SE-38, US Department of Agriculture, Forest Service, Southeastern Forest Experiment Station, Asheville, NC, USA, pp. 1–35 Kipfmueller KF, Larson ER, St George S (2012) Does proxy uncertainty affect the relations inferred between the Pacific Decadal Oscillation and wildfire activity in the western United States? Geophysical Research Letters 39, 1–5, 10.1029/2011GL050645 Kirchmeier-Young MC, Gillett NP, Zwiers FW, Cannon AJ, Anslow FS (2019) Attribution of the influence of human-induced climate change on an extreme fire season. Earth’s Future 7 (1), 2–10, 10.1029/2018EF001050 Kitzberger T, Brown PM, Heyerdahl EK, Swetnam TW, Veblen TT (2007) Contingent Pacific-Atlantic Ocean influence on multi-century wildfire synchrony over western North America. Proceedings of the National Academy of Sciences of the USA 104 (2), 543–548, 10.1073/pnas.0606078104 107  Knapp PA, Grissino-Mayer HD, Soulé PT (2002) Climatic regionalization and the spatio-temporal occurrence of extreme single-year drought events (1500–1998) in the interior Pacific Northwest, USA. Quaternary Research 58 (3), 226–233, 10.1006/qres.2002.2376 Krawchuk MA, Cumming SG, Flannigan MD (2009) Predicted changes in fire weather suggest increases in lightning fire initiation and future area burned in the mixedwood boreal forest. Climatic Change 92 (1–2), 83–97, 10.1007/s10584-008-9460-7 Kubian R (2013) Characterizing the mixed-severity fire regime of the Kootenay Valley, Kootenay National Park. MSc thesis, University of Victoria, Victoria, BC, Canada Kuuluvainen T (2009) Forest management and biodiversity conservation based on natural ecosystem dynamics in Northern Europe: The complexity challenge. AMBIO: A Journal of the Human Environment 38 (6), 309–315, 10.1579/08-A-490.1  Lampainen J, Kuuluvainen T, Wallenius TH, Karjalainen L, Vanha-Majamaa I (2004) Long-term forest structure and regeneration after wildifre in Russian Karelia. Journal of Vegetation Science 15 (2), 245–256, 10.1111/j.1654-1103.2004.tb02259.x  Larsson L (2011) CDendro package version 7.4. Cybis Elektronik & Data AB Lawson BD, Armitage OB (2008) Weather guide for the Canadian Forest Fire Danger Rating System. Natural Resources Canada, Canadian Forest Service, Northern Forestry Centre, Edmonton, AB, Canada, pp. 1–87 Littell JS, Peterson DL, Tjoelker M (2008) Douglas-fir growth in mountain ecosystems: Water limits tree growth from stand to region. Ecological Monographs 78 (3), 349–368, 10.1890/07-0712.1 108  Littell JS, McKenzie D, Peterson DL, Westerling AL (2009) Climate and wildfire area burned in western U.S. ecoprovinces, 1916–2003. Ecological Applications, 19 (4), 1003–1021, 10.1890/07-1183.1 Littell JJ, Peterson DL, Riley KL, Liu Y, Luce CH (2016) A review of the relationships between drought and forest fire in the United States. Global Change Biology 22 (7), 1–17, 101111/gcb13275 Liu ZF, Tang YL, Jian ZM, Poulsen CJ, Welker JM, Bowen GJ (2017) Pacific North American circulation pattern links external forcing and North American hydroclimatic change over the past millennium. Proceedings of the National Academy of Sciences of the USA 114 (13), 3340–3345, 10.1073/pnas.1618201114 MacDonald GM, Case RA (2005) Variations in the Pacific Decadal Oscillation over the past millennium. Geophysical Research Letters 32 (8), 1–4, 10.1029/2005GL022478 Macias Fauria M, Johnson EA (2006) Large-scale climatic patterns control large lightning fire occurrence in Canada and Alaska forest regions. Journal of Geophysical Research, 111, 1–17, 10.10289/2006JG000181 Macias Fauria MM, Johnson EA (2008) Climate and wildfires in the North American boreal forest. Philosophical Transactions of the Royal Society Biological Sciences 363 (1501), 2315–2327, 101098/rstb20072202 Macias Fauria M, Michaletz ST, Johnson EA (2011) Predicting climate change effects on wildfires requires linking processes across scales. Wiley Interdisciplinary Reviews: Climate Change 2 (1), 99–112, 10.1002/wcc.92 109  Mantua NJ, Hare SR, Zhang Y, Wallace JM, Francis RC (1997) A Pacific interdecadal climate oscillation with impact on salmon production. Bulletin of the American Meteorological Society 78 (6), 1–11, 10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2 Marchal J, Cumming SG, McIntire JB (2017) Exploiting Poisson additivity to predict fire frequency from maps of fire weather and land cover in boreal forests of Québec, Canada. Ecography 40, 200–209, 101111/ecog01849 Marshall JD, Blair JM, Peters DPC, Okin G, Rango A, Williams M (2008) Predicting and understanding ecosystem responses to climate change at continental scales. Frontiers in Ecology 6 (5), 273−280, 10.1890/070165 Marcoux HM, Gergel SE, Daniels LD (2013) Mixed-severity fire regimes: how well are they represented by existing fire-regime classification systems? Canadian Journal of Forest Research 43 (7), 658–668, 10.1139/CJFR-2012-0449 Marcoux HM, Daniels LD, Gergel SE, Da Silva E, Gedalof ZM, Hessburg PF (2015) Differentiating mixed- and high-severity fire regimes in mixed-conifer forests of the Canadian Cordillera. Forest Ecology and Management 341, 45–58, 10.1016/J.FORECO.2014.12.027 Margolis EQ, Swetnam TW (2013) Historical fire-climate relationships of upper elevation fire regimes in the south-western United States. International Journal of Wildland Fire 22 (5), 588–598, 10.1071/WF12064 Mather V (2010) Climate relationships and sub-ring parameters of four conifer species in the Kootenay region of British Columbia, Canada. MSc thesis, University of Guelph, Guelph, ON, Canada 110  McAlpine RS, Stocks BJ, Van Wagner CE, Lawson BD, Alexander ME, Lynham TJ (1990) Forest fire behavior research in Canada. Proceedings of international conference on forest fire research, University of Coimbra, 1–12 McBride JM (1983) Analysis of tree rings and fire scars to establish fire history. Tree-Ring Bulletin 43, 51–67 McKee TB, Doesken NJ, Kleist J (1993) The relationship of drought frequency and duration to time scales. American Meteorological Society, Anaheim, CA, USA, pp. 1–6 McKenzie D, Gedalof Z, Peterson DL, Mote P (2004) Climatic change, wildfire, and conservation. Conservation Biology 18 (4), 890–902, 101111/j1523-1739200400492x McMahon T (2018) In wildfire-prone B.C. and California, urban sprawl and bad planning are fuelling future infernos. What can we do? The Globe and Mail, Toronto, ON, Canada Mekis É, Vincent LA (2011) An overview of the second generation adjusted daily precipitation dataset for trend analysis in Canada. Atmosphere-Ocean 49 (2), 163–177, 101080/070559002011583910 Melvin TM, Briffa KR (2008) A “signal-free” approach to dendroclimatic standardisation. Dendrochronologia 26 (2), 71–86, 10.1016/j.dendro.2007.12.001 Melvin TM, Briffa KR (2014) CRUST: Software for the implementation of Regional Chronology Standardisation: Part 1. Signal-Free RCS. Dendrochronologia 32 (1), 7–20, 10.1016/j.dendro.2013.06.002 Meyn A, Taylor SW, Flannigan MD, Thonicke K, Cramer W (2010a) Relationships between fire, climate oscillations, and drought in British Columbia, Canada, 1920–2000. Global Change Biology 16 (3), 977–989, 10.1111/j.1365-2486.2009.02061.x 111  Meyn A, Schmidtlein S, Taylor SW, Girardin MP, Thonicke K, Cramer W (2010b) Spatial variation of trends in wildfire and summer drought in British Columbia, Canada, 1920–2000. International Journal of Wildland Fire 19 (3), 272–283, 10.1071/WF09055 Minobe S (1999) Resonance in bidecadal and pentadecadal climate oscillations over the North Pacific: Role in climatic regime shifts. Geophysical Research Letters 26 (7), 855–585, 10.1029/1999GL900119 Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations and associated high-resolution grids. International Journal of Climatology 25 (6), 693–712, 101002/joc1181 Morgan P, Hardy CC, Swetnam TW, Rollins MG, Long DG (2001) Mapping fire regimes across time and space: Understanding coarse and fine-scale fire patterns. International Journal of Wildland Fire 10 (4), 329–342, 10.1071/WF01032 Morgan P, Heyerdahl EK, Gibson CE (2008) Multi-season climate synchronized forest fires throughout the 20th century, northern Rockies, USA. Ecology 89 (3), 717–728, 10.1890/06-2049.1 Mori AS, Johnson EA (2013) Assessing possible shifts in wildfire regimes under a changing climate in mountainous landscapes. Forest Ecology and Management 310, 875–886, 101016/jforeco201309036 Moritz MA, Batllori E, Bradstock RA, Gill AM, Handmer J, Hessburg PF, Leonard J, McCaffrey S, Odion DC, Schoennagel T, Syphard AD (2014) Learning to coexist with wildfire. Nature 515, 58–66, 10.1038/nature13946 112  Mote PW, Hamlet AF, Clark MP, Lettenmaier DP (2005) Declining mountain snowpack in western North America. Bulletin of the American Meteorological Society 86, 39−49, 10.1175/BAMS-86-1-39 Nash JE, Sutcliff JV (1970) Riverflow forecasting through conceptual models part I – A discussion of principles. Journal of Hydrology 10 (3), 282–290, 10.1016/0022-1694(70)90255-6 Nash CH, Johnson EA (1996) Synoptic climatology of lightning-caused forest fires in subalpine and boreal forests. Canadian Journal of Forest Research 26 (10), 1859–1874, 101139/x26-211 National Drought Mitigation Center (2019) U.S. Drought Monitor. University of Nebraska-Lincoln, Lincoln, NE, USA. Natural Resources Canada (2019a) Canadian Wildland Fire Information System, Interactive map. Government of Canada Natural Resources Canada (2019b) Canadian Wildland Fire Information System, CWFIS Datamart, National Fire Datadase fire point data. Government of Canada, http://cwfis.cfs.nrcan.gc.ca/datamart/metadata/nfdbpnt Nesbitt JH, Daniels LD (2009) Fire history in Cranbrook: Reconstructing fire frequency at McLeary Park & Rocky Mountain Airport. Report to RW Gray Consulting Ltd. and the City of Cranbrook, BC, Canada, pp. 1–14 Nesbitt JH (2010) Quantifying forest fire variability using tree rings Nelson, British Columbia 1700–present. MSc thesis, University of British Columbia, Vancouver, BC, Canada Palmer WC (1965) Meteorologic drought. Office of Climatology, United States Weather Bureau, Research Paper 45. (Washington, DC, USA) 113  Parks Canada and the Canadian Parks Council (2008) Principles and Guidelines for Ecological Restoration in Canada’s Protected Natural Areas. Parks Canada Agency, Gatineau, QC, Canada, 1–97 pp. Parsons RA, Heyerdahl EK, Keane RE, Dorner B, Fall J (2007) Assessing accurary of point fire intervals across landscapes with simulation modelling. Canadian Journal of Forest Research 37 (9), 1605–1614, 10.1139/X07-013 Pausas JG, Ribeiro E (2013) The global fire-productivity relationship. Global Ecology and Biogeography 22 (6), 728–736, 10.1111/geb.12043 Perry DA, Hessburg PF, Skinner CN, Spies TA, Stephens SL, Taylor AH, Franklin JF, McComb B, Riegel G (2011) The ecology of mixed severity fire regimes in Washington, Oregon, and Northern California. Forest Ecology and Management 262 (5), 703–717, 10.1016/J.FORECO.2011.05.004 Pogue AM (2017) Humans, climate and an ignitions-limited fire regime at Vaseux Lake. MSc thesis, University of British Columbia, Vancouver, BC, Canada Pojar J, Meidinger DV (1991) British Columbia: The environmental setting. In: Meidinger DV, Pojar J (Eds.) Ecosystems of British Columbia. British Columbia Ministry of Forests, Research Branch, Victoria, BC, Canada, pp. 339–366 Price C, Rind D (1994) The impact of a 2 X CO2 climate on lightning-caused fires. Journal of Climate 7, 1484–1494, 10.1175/1520-0442(1994)007<1484:TIOACC>2.0.CO;2 Pyne SJ (1982) Fire in America: A cultural history of wildland and rural fire. Princeton University Press, Princeton, NJ, USA, pp. 1–654 Pyne SJ (2007) Awful splendor: A fire history of Canada. University of British Columbia Press, Vancouver, BC, Canada, pp. 1–549 114  Riley KL, Abatzoglou JT, Grenfell IC, Klene AE, Heinsch FA (2013) The relationship of large fire occurrence with drought and fire danger indices in the western USA, 1984–2008: The role of temporal scale. International Journal of Wildland Fire 22 (7), 894–909, 10.1071/WF12149 Rogeau MP, Flannigan MD, Hawkes BC, Parisien MA, Arthur R (2016) Spatial and temporal variations of fire regimes in the Canadian Rocky Mountains and Foothills of southern Alberta. International Journal of Wildland Fire 25 (11), 1117–1130, 10.1071/WF15120 Salkind N (2007) Encyclopedia of measurement and statistics. Sage Publications, Thousand Oaks, CA, USA, 10.4135/9781412952644  Sankey S (2019) Blueprint for wildland fire science in Canada (2019–2029). Natural Resources Canada, Canadian Forest Service, Northern Forestry Centre, Edmonton, AB, Canada, pp. 1–60  SAS Institute Incorporated (2017) SAS 9.4 Documentation. SAS Institute Incorporated, Cary, NC, USA, http://support.sas.com/documentation/cdl_main/94/docindex.html Schoennagel T, Veblen TT, Romme WH (2004) The interaction of fire, fuels, and climate across Rocky Mountain forests. Bioscience 54 (7), 661–676, 10.1641/0006-3568(2004)054[0661:TIOFFA]2.0.CO;2. Schoennagel T, Veblen TT, Kulakowski D, Holz A (2007) Multidecadal climate variability and climate interactions affect subalpine fire occurrence, western Colorado (USA). Ecology 88 (11), 2891–2902, 101890/06-18601 Scudder GGE, Smith IM (2011) Introduction and summary of the Montane Cordillera Ecozone. In: Scudder GGE, Smith IM (Eds.) Assessment of species diversity in the Montane 115  Cordillera Ecozone, pp. 1–26, https://royalbcmuseum.bc.ca/assets/Montane-Cordillera-Ecozone.pdf Sherriff RL, Veblen TT (2006) Ecological effects of changes in fire regimes in Pinus ponderosa ecosystems in the Colorado Front Range. Journal of Vegetation Science 17 (6), 705–718, 10.1111/j.1654-1103.2006.tb02494.x Skinner WR, Stocks BJ, Martell DL, Bonsal B, Shabbar A (1999) The association between circulation anomalies in the mid-troposphere and area burned by wildland fire in Canada. Theoretical and Applied Climatology, 63 (1–2), 89–105, 10.1007/s007040050 Soil Landscapes of Canada Working Group (2010) Soil Landscapes of Canada version 32 Agriculture and Agri-Food Canada (digital map and database at 1:1 million scale) Sommerfeld A, Senf C, Buma B, D’Amato AW, Després T,  Díaz-Hormazábal I, Fraver S, Frelich LE, Gutiérrez ÁG, Hart SJ, Harvey BJ, He HS, Hlásny T, Holz A, Kitzberger T,  Kulakowski D, Lindenmayer D, Mori AS, Müller J, Paritsis J, Perry GLW, Stephens SL, Svoboda M, Turner MG, Veblen TT, Seidl R (2018) Patterns and drivers of recent disturbances across the temperate forest biome. Nature Communications 9 (4355), 1–9, 10.1038/s41467-018-06788-9 Stahle D, Cleaveland MK (1988) Texas drought history reconstructed and analyzed from 1698 to 1980. Journal of Climate 1, 59–74, 10.1175/1520-0442(1988)001<0059:TDHRAA>2.0.CO;2 Stahle DW, Cook ER, Cleaveland MK, Therrell MD, Meko DM, Grissino-Mayer HD, Watson E, Luckman BH (2000) Tree-ring data document 16th century megadrought over North America. Eos Transactions, American Geophysical Union 81 (12), 121–125, 10.1029/00EO00076 116  Stephens SL, Agee JK, Fulé PZ, North MP, Romme WH, Swetnam TW, Turner MG (2013) Managing forests and fire in changing climates. Science 342 (6154), 41–42, 10.1126/science.1240294 Stephens SL, Burrow N, Buyantuyev A, Gray RW, Keane RE, Kubian R, Liu S, Sijo F, Tolhurst KG, van Wagtendonk JW (2014) Temperature and boreal forest mega-fires: Characteristics and challenges. Frontiers in Ecology and the Environment 12 (2), 115–122, 10.1890/120332 Stocks BJ, Lynham TJ, Lawson BD, Alexander ME, Van Wagner RS, McAlpine RS, Dubé DE (1989) Canadian Forest Fire Danger Rating System: An Overview. The Forestry Chronicle 65 (4), 258–265, 105558/tfc65258-4 Sutherland EK, Brewer PW, Falk DA, Velásquez EM (2017) FHAES Fire History Analysis and Exploration System. Missoula Forestry Sciences Laboratory, Missoula, MT, USA Swetnam TW (1993) Fire history and climate change in giant sequoia groves. Science, 262 (5135), 885–889, 10.1126/science.262.5135.885 Swetnam TW, Baisan CH (1996) Historical fire regime patterns in the southwestern United States since AD 1700. In: Allen CD (Ed.) Fire effects in southwestern Forests: Proceedings of the 2nd La Mesa fire symposium, General Technical Report RM-GTR-286. USDA Forest Service, Rocky Mountain Research station, Fort Collins, CO, USA, pp. 11–32 Swetnam TW, Betancourt JL (1998) Mesoscale disturbance and ecological response to decadal climatic variability in the American southwest. Journal of Climate 11, 3128–3147, 10.1175/1520-0442(1998)011<3128:MDAERT>2.0.CO;2 Swetnam T, Allen C, Betancourt J (1999) Applied historical ecology: Using the past to manage for the future. Ecological Applications 9 (4), 1189–1206, 10.1890/1051-0761(1999)009[1189:AHEUTP]2.0.CO;2 117  Swetnam TW, Anderson SR (2008) Fire climatology in the western United States: Introduction to special issue. International Journal of Wildland Fire, 17 (1), 1–7, 10.1071/WF08016 Swetnam T, Falk DA, Hessl AE, Farris C (2011) Reconstructing landscape pattern of historical fires and fire regimes. In: McKenzie D, Miller C, Falk DA (Eds.) The Landscape Ecology of Fire. Springer, New York, NY, USA, pp. 165–192 Taylor AH, Skinner CN (2003) Spatial patterns and controls on historical fire regimes and forest structure in the Klamath Mountains. Ecological Applications 13 (3), 704–719, 10.1890/1051-0761(2003)013[0704:SPACOH]2.0.CO;2 Taylor SW, Alexander ME (2006) Science, technology, and human factors in fire danger rating: The Canadian experience. International Journal of Wildland Fire 15 (1), 121–135, 101071/WF05021 Terrier A, Girardin MP, Périé C, Legendre P, Bergeron Y (2013) Potential changes in forest composition could reduce impacts of climate change on boreal forests. Ecological Applications 23 (1), 21–35, 101890/12-04251 Toms JD, Villard MA (2015) Threshold detection: Matching statistical methodology to ecological questions and conservation planning objectives. Avian Conservation and Ecology 10 (1), 10.5751/ACE-00715-100102 TRIM (2016) Terrain Resource Information Management Program Version 5. British Columbia Government, Integrated Land Management Bureau, Crown Registry and Geographic Base Branch, BC, Canada Trouet V, Taylor AH (2010) Multi-century variability in the Pacific North American circulation pattern reconstructed from tree rings. Climate Dynamics 35 (6), 953–963, 10.1007/s00382-009-0605-9 118  Trouet V, Taylor AH, Wahl ER, Skinner CN, Stephens SL (2010) Fire-climate interactions in the American West since 1400 CE. Geophysical Research Letters 37 (4), 1–5, 10.1029/2009GL041695 Turner JA (1972) The Drought Code component of the Canadian Forest Fire Behavior System. Publication 1316. Department of the Environment, Canadian Forest Service, Ottawa, ON, Canada, pp. 1–14 Turner MG (2010) Disturbance and landscape dynamics in a changing world. Ecology 91 (10), 2833–2849, 101890/10-00971 Tzembelicos A, Erhardt A, de Faye B, Guy B, Kent C, Beldwell C, Wall D, Vermeulen D, Forsdick G, Farson H, Paulus I, Enns J, Stewart J, Fraser J, Bax K, Hague P, Pepin S, Siemens W (2018) Addressing the new normal: 21st century disaster management in British Columbia. British Columbia Government, pp. 1–148 Van der Kamp DW, Bürger G, Werner AT (2013) Evaluation of the monthly drought code as a metric for fire weather in a region of complex terrain, and uncertainties in future projections. The Pacific Climate Impacts Consortium, Victoria, BC, Canada, pp. 1–16 Van Mantgem PJ, Stephenson NL, Byrne JC, Daniels LD, Franklin JF, Fulé PZ, Harmon ME, Larson AJ, Smith JM, Taylor AH, Veblen TT (2009) Widespread increase of tree mortality rates in the western United States. Science 323 (5913), 521–524, 10.1126/science.1165000 Van Wagner CE (1974) Structure of the Canadian Forest Fire Weather Index. Publication Number 35. Canadian Forestry Service, Ottawa, ON, Canada, pp. 1–49 Van Wagner CE (1987) Development and structure of the Canadian Forest Fire Weather Index System. Forestry Technical Report 35. Canadian Forestry Service, Ottawa, ON, Canada, pp. 1–48 119  Veblen TT, Kitzberger T, Raffaele E, Mermoz M, González ME, Sibold JS, Holz A (2009) The historical range of variability of fires in the Andea-Patagonian Nothofaus forest region. International Journal of Wildland Fire 17 (6), 724–741, 10.1071/WF07152 Villemaire-Côté O (2014) Fire history near Cranbrook, British Columbia: Historical reconstruction using tree-ring evidence. Undergraduate thesis, University of British Columbia, Vancouver, BC, Canada Vincent LA, Wang XL, Milewska EJ, Wan H, Yang F, Swail V (2012) A second generation of homogenized Canadian monthly surface air temperature for climate trends analysis. Journal of Geophysical Research: Atmospheres 117 (D18110), 1–13, 101029/2012JD017859 Wang X, Parisien M, Flannigan MD, Parks SA, Anderson KR, Little JM, Taylor SW (2014) The potential and realized spread of wildfires across Canada. Global Change Biology 20 (8), 2518–2530, 10.1111/gcb.12590 Wang X, Thompson DK, Marshall GA, Tymstra C, Carr R, Flannigan MD (2015) Increasing frequency of extreme fire weather in Canada with climate change. Climatic Change 130 (4), 573–586, 101007/s10584-015-1375-5 Wang T, Hamann A, Spittlehouse D, Carroll C (2016) Locally downscaled and spatially customizable climate data for historical and future periods for North America. PLoS One 11 (6), 1–17, 101371/journalpone0156720 Watson E, Luckman BH (2001) Dendroclimatic reconstruction of precipitation for sites in the southern Canadian Rockies. The Holocene 11 (2), 203–213 10.1191/095968301672475828 Watson E, Luckman BH (2002) The dendroclimatic signal in Douglas-fir and ponderosa pine tree-ring chronologies from the southern Canadian Cordillera. Canadian Journal of Forest Research 32 (10), 1858–1874, doi:10.1139/x02-096 120  Watson E, Luckman BH (2004) Tree-ring based reconstructions of precipitation for the Southern Canadian Cordillera. Climatic Change 65 (1–2), 209–241, 10.1023/B:CLIM.0000037487.83308.02 Watson E, Luckman BH (2005) Spatial patterns of preinstrumental moisture variability in the Southern Canadian Cordillera. American Meteorological Society 18 (15), 2847–2863, 10.1175/JCL13416.1 Wells N, Goddard S, Hayes MJ (2004) A self-calibrating Palmer Drought Severity Index. Journal of Climate 17, 2335–2351, 10.1175/1520-0442(2004)017<2335:ASPDSI>2.0CO;2 Westerling AL, Swetnam TW (2003) Interannual to decadal drought and wildfire in the western United States. EOS, Transactions American Geophysical Union 84 (49), 545–560, 10.1029/2003EO490001 Westerling AL, Hidalgo HG, Cayan DR, Swetnam TW (2006) Warming and earlier spring increase western US Forest Wildfire Activity. Science 313 (5789), 940–943, 101126/science1128834 Westerling AL (2016) Increasing western US forest wildfire activity: sensitivity to changes in the timing of spring. Philosophical Transactions of the Royal Society B: Biological Sciences 371 (1696), 1–10, 10.1098/rstb.2015.0178 Whitlock MC, Schluter D (2015) The analysis of biological data. Roberts and Company Publishing, Greenwood Village, CO, USA, pp. 1–818 Wierzchowski J, Heathcott M, Flannigan MD (2002) Lightning and lightning fire, central cordillera, Canada. International Journal of Wildland Fire 11 (1), 41–51, 10.1071/WF01048. 121  Wigley TML, Briffa KR, Jones PD (1984) On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology 23, 201–213, 10.1175/1520-0450(1984)023<0201:OTAVOC>2.0.CO;2 Williams GW (2002) Aboriginal use of fire: Are there any “natural” plant communities? In: Kay C (Ed.) Wilderness and political ecology: Aboriginal influences and the original state of nature, University of Utah Press, Salt Lake City, UT, USA, pp. 179–214 Williams PA, Abatzoglou JT (2016) Recent advances and remaining uncertainties in resolving past and future climate effects on global fire activity. Current Climate Change Reports 2 (1), 1–14, 10.1007/s40641-016-0031-0 Wilson RJS, Luckman BH (2003) Dendroclimatic reconstruction of maximum summer temperatures from upper treeline sites in Interior British Columbia, Canada. The Holocene 13 (6), 851–861, 10.1191/0959683603hl663rp Wotton BM, Stocks BJ, Martell DL (2005) An index for tracking sheltered forest floor moisture within the Canadian Forest Fire Weather Index System. International of Wildland Fire 14 (2), 169–182, 10.1071/WF04038 Wotton BM (2009) Interpreting and using outputs from the Canadian Forest Fire Danger Rating System in research applications. Environmental and Ecological Statistics 16 (2), 107–131, 10.1007/s10651-007-0084-2 Wotton BM, Flannigan MD, Marshall GA (2017) Potential climate change impacts on fire intensity and key wildfire suppression thresholds in Canada. Environmental Research Letters 12 (9), 1–13, 101088/1748-9326/aa7e6e 122  Wright HE, Heinselman ML (2014) The ecological role of fire in natural conifer forests of western and northern North America – Introduction. Fire Ecology 10 (3), 4–13, 10.1007/BF03400628 Xiao JF, Zhuang QL (2007) Drought effects on large fire activity in Canadian and Alaskan forests. Environmental Research Letters 2 (4), 1–7, 10.1088/1748-9326/2/4/044003 Xu H, Tremblay F, Bergeron Y (2018) Importance of landscape features and fire refuges on genetic diversity of Thuya occidentalis L., in boreal fire dominated landscapes. Conservation Genetics 19 (5), 1231–1241, 10.1007/s10592-018-1091-6 Zang C, Biondi F (2015) Treeclim: An R package for the numerical calibration o proxy-climate relationships. Ecography 38 (4), 431–436, 10.1111/ecog.01335  123  Appendices Appendix A    A.1 Daily Drought Code The daily Drought Code (DC) is a dimensionless index which tracks, on a daily basis over the fire season, changes in moisture within the deep layer (~18 cm) of compact (~25 kg m-2 dry weight) organics in the soil (Van Wagner 1987) and large diameter woody fuels (Wotton 2009). The DC has a slow drying rate with a lagtime of 51 days (Field et al. 2015) where the lag represents the time required to lose roughly two-thirds of the free moisture content over 24 hours at 21.1 °C and relative humidity of 45 % (Van Wagner 1987). By default, the DC is given a value of 15 at the start of the fire season which is either the third day after snow has melted in regions normally covered by snow during the winter, or the third day that noon temperatures of 12 °C or higher have been recorded in regions where snow cover is not a significant feature (Lawson and Armitage 2008). The DC on any subsequent day of the fire season is calculated as: DC = 400 × ln(800 Q⁄ ), if Q >800 then Q = 800 where Q is the daily moisture equivalent (dimensionless) with maximum value 800 corresponding to saturation, and minimum value 0 corresponding to the driest condition (Van Wagner 1987). On each successive day of the fire season, DC is calculated by adding the effects of total daily precipitation (mm, as water equivalent) (Lawson and Armitage 2008) followed by daily potential evapotranspiration. The effect of total daily precipitation ppt is added to the previous day’s value of the DC (DCd-1) and the moisture equivalent after precipitation Qppt is calculated as: Qppt = 800 × e[−DCd−1/400] + 3.937 × PPTEFF, if ppt >2.8 mm or 124  Qppt = 800 × e[−DCd−1/400], if ppt ≤2.8 mm where PPTEFF = 0.833 × ppt − 1.27 is the effective precipitation (mm) after canopy and surface fuel interception (Girardin and Wotton 2009). The effect of potential evapotranspiration E (dimensionless) over a given day d of the fire season (Ed) is calculated as: Ed = 0.36(TMXd) + Lf, if Ed < 0 then Ed = 0, if TMXd < 0 °C then TMXd = 0 °C where TMXd is the daily maximum temperature (°C), and Lf is a standard day length adjustment factor which varies by month (April = 0.9, May = 3.8, June = 5.8, July = 6.4, August = 5.0, September = 2.4, October = 0.4, and November–March = -1.6) (Girardin and Wotton 2009). The present day’s value of the DC (DCd) is calculated as: DCd = 400 × ln�800 Qppt⁄ � + 0.5 × Ed By default, DC calculations are continued until snow covers the ground or noon temperatures drop below 12 °C for three consecutive days (Lawson and Armitage 2008).  A.2 Overwintered Drought Code In areas with moisture depletion in between fire seasons, the starting moisture equivalent value in the spring (Qs; i.e., Q on the first day of the fire season) is calculated as:  Qs = 𝑎𝑎 × Qf + 𝑏𝑏(3.94 × pptw) where Qf is the final moisture equivalent value in the fall (i.e., Q on the last day of the fire season), a is the carry-over fraction of prior fall moisture estimated as 0.5, 0.75 or 1.0, and b is the effectiveness of winter precipitation in recharging moisture reserves in spring estimated as 0.5, 0.75 or 0.9, and pptw is total winter precipitation (mm, as water equivalent) (Van Wagner 1987).  125  A.3 Monthly Drought Code The Monthly Drought Code (MDC) is a dimensionless index which tracks monthly changes in moisture in deep compact organics in the soil and large-diameter woody fuels (Girardin and Wotton 2009). Typically, the MDC is assigned a value of 15 on April 30 (Girardin and Wotton 2009) or March 31 (Bergeron et al. 2010), the day preceding the start of the fire season. The MDC in any given month of the fire season is calculated as: MDC = 400 × ln(800 Qm⁄ ), if Qm >800 then Qm = 800 where Qm is the moisture equivalent (dimensionless) during month m (Girardin and Wotton 2009). In their MDC calculations, Girardin and Wotton (2009) assumed that drying occurs in the first half of the month, total precipitation occurs in the middle of the month, and drying occurs again in the second half of the month. In a first step, potential evapotranspiration over a given month (Em) of the fire season is calculated as: Em = N[0.36(TMXm) + Lf], if Em <0 then Em = 0, if TMXm <0 °C then TMXm = 0 °C where TMXm is the monthly mean of daily maximum temperatures (°C), Lf is the standard day length adjustment factor applied to each month as in the daily DC, and N is the number of days in the given month (Girardin and Wotton 2009). Following drying in the first half of the month, MDC is calculated as: MDC1st half = MDCe + 0.25 × Em where MDCe is the MDC value at the end of the previous month. The effect of total monthly precipitation (PPTm; mm, as water equivalent) occurring mid-month affects the corresponding moisture equivalent (Qmr) as follows: Qmr = 800 × e[−(MDC1st half)/400] + 3.937 × PPTEFF 126  where PPTEFF = 0.833×PPTm is the monthly effective precipitation (Girardin and Wotton 2009). Following drying over the second half of the month, the MDC value of at the end of the month, MDC is calculated as: MDC2nd half = 400 × ln(800 Qmr⁄ ) + 0.25 × Em Averaging the MDC values at the end of the previous month and at the end of the given month yields the MDC value for the given month: MDCm = (MDCe + MDC2nd half) 2⁄  By default, MDC is calculations are discontinued at the end of each year on October 31 and does not account for potential overwinter drying of fuels (Girardin and Wotton 2009).  A.4 Adjusted Monthly Drought Code The adjusted Monthly Drought Code (adjMDC) accounts for overwinter drying of fuels and starts on April 1. With adjMDC, moisture equivalents at the end of the fire season on October 31 (QOct) and at the start of the fire season (QMar) are calculated as: QOct = 800 × e[−adjMDCOct/400]  And, QMar = 𝑎𝑎 × QOct + 𝑏𝑏(3.94 × PPTw), if QMar >800 then QMar = 800 where adjMDCOct is the adjMDC value on October 31, a is the carry-over fraction of prior fall moisture estimated as 0.5, 0.75 or 1.0, b is the effectiveness of winter precipitation in recharging moisture reserves in spring estimated as 0.5, 0.75 or 0.9, and PPTw is total winter precipitation (mm, as water equivalent) from November–March inclusive. I conducted a sensitivity analysis to select the most suitable combination a and b for our study area. I compared adjMDC, calculated with all possible combinations of a and b, against the overwintered DCm values for April or May 127  (depending on the start date for daily DC each year) from 1989−2013. Based on boxplots, Mann-Whitney rank sum tests, and Pearson product moment correlations adjMDC values were most suitable in April and May when a and b were both 0.75, corresponding to areas exposed to winter moisture depletion and sites that are moderately drained, respectively (Lawson and Armitage 2008). To begin each fire season, the adjMDC value on March 31 is calculated as:  adjMDCMar = 400 × ln(800 QMar⁄ )                  128  Appendix B     B.1 (A) Centroids for site-level fire records: 1: Amoroso et al. 2011; 2: Chavardès et al. 2018; 3: Dinh 2014; 4: Brookes 2019; 5: Harvey and Smith 2017; 6: Daniels and Watson 2003; 7: Harvey et al. 2017; 8: Kubian 2013; 9: Heyerdahl et al. 2008b, 2012; 10: Heyerdahl et al. 2007; 11:  Cochrane 2007; 12: Daniels et al. 2006, Gray and Daniels 2007; 13: Nesbitt 2010; 14: Nesbitt and Daniels 2009; 15: Marcoux et al. 2013, Villemaire-Côté 2014; 16: Pogue 2017; 17: Greene and Daniels 2017. (B) Approximate coverage of the reconstruction of maximum temperature anomalies for Interior British Columbia (dashed box). (C) Reconstructions of annual (previous July to current June) precipitation for Banff, Big Creek, Cranbrook, Jasper, Lillooet, Lytton, North Thompson, Oliver, Summerland, Waterton, and Williams Lake (Watson and Luckman 2004). (D) Reconstructions of summer Palmer Drought Severity Index (grid points: 22–25, 29–32, 40–43, 52–55, and 65–68) (Cook et al. 2004). 129  Appendix C     C.1 Thresholds and categories for the reconstructions of annual precipitation z-scores, and summer Palmer Drought Severity Index (PDSI; see Table 11 in Palmer 1965) values.     

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0380575/manifest

Comment

Related Items