CLIMATE AND BIOLOGICAL RESILIENCE AT THE NORTHERN LIMIT OF BRITISH COLUMBIA’S INLAND TEMPERATE RAINFOREST by Nathan Malcomb B.A., The College of Wooster, 2007 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN NATURAL RESOURCES AND ENVIRONMENTAL STUDIES UNIVERSITY OF NORTHERN BRITISH COLUMBIA August 2025 © Nathan Malcomb, 2025 Abstract Climate-mediated shifts in forest productivity hold uncertain impacts for temperate rainforest ecosystems. While considerable focus has been devoted to detecting, monitoring, quantifying, and modeling ecological change, variable tree responses across biogeographic gradients complicate ongoing research. In British Columbia, arid montane forests are dynamically reorganizing to meet evolving environmental conditions, as evidenced by largescale disturbance processes such as mega-fires and insect outbreaks. Less is known about the more subtle climate responses impacting the Inland Temperate Rainforest. As these forests are among the most productive in the world, shifts in forest growth may have lasting implications for global climate and ecological, economic, and cultural systems. This thesis documents the influence of the recent climate variability on an ITR ecosystem in the northern Rocky Mountains of British Columbia. Findings complement the existing body of limited research in this biologically rich but relatively understudied region by producing multi-species tree-ring chronologies from high- and low-elevation sites. Using multiple analytical techniques, results from tree-ring, biomass, and dendroclimatic analyses highlight the role of biogeography in mediating climate sensitivity and tree growth across species and elevation gradients. Stand-level biomass estimates reveal the significant carbon storage potential of lower-elevation forests, rivaling productive temperate rainforests in the coastal Pacific Northwest. Dendroclimatic analysis highlights the role of temperature and snow in limiting tree growth, with the highest productivity periods occurring during years with slightly aboveaverage temperatures, below-average snowpack, and average precipitation. As expected at I this northern latitude site, low growth occurs during cold years with heavy snowpack. Still, trees also display negative growth responses to above-average temperatures and drought, particularly at low elevations. Despite near-normal precipitation in the recent decade, thermal stress during periodic "heat domes" is becoming an overarching driver of reduced biomass accumulation in old-growth western red cedar, an iconic keystone species and dominant carbon pool in the ITR. Intervals of reduced growth throughout the 121-year study period have been followed by notable plasticity across cedar and other co-dominant species, showing potential for increased productivity under projected climate scenarios. The long-term trajectory of the ITR will hinge on species-level adaptations to nonanalog warming conditions projected for the next century and conservation measures that protect the structural and compositional resiliency of these globally significant ecosystems. Remarkable adaptability is evident in study species which thrive from the Northern Rockies (Engelmann spruce and subalpine fir) to the Sierra Madre of central Mexico (Douglas-fir) and along the Pacific Northwest coast to northern California (western hemlock, western red cedar, Douglas-fir). Dedicated research and conservation efforts across these ranges are essential for enabling the forests of the ITR to adapt to changing environmental conditions. This thesis underscores the ITR’s critical role as a globally significant carbon sink and biodiversity reservoir. Building on existing conservation efforts, it calls for the creation of the Great Caribou Rainforest Initiative. Modeled after the Great Bear Rainforest Framework, this initiative aligns scientific evidence with cultural and economic values to create best-practice climate change mitigation strategies safeguarding the ITR’s adaptive capacity, carbon storage, and biodiversity. II Table of Contents Abstract ...................................................................................................................................... I List of Tables .........................................................................................................................VII List of Figures .......................................................................................................................XII Acknowledgments................................................................................................................ XIV Dedication.............................................................................................................................. XV Chapter 1: Introduction, Organization, Imperative, Literature Review, and Hypotheses .... 1 1.1 Introduction and Organization .................................................................................... 1 1.2 The Imperative .............................................................................................................. 2 1.3 Climate Change and Forest Response in Western North America .......................... 3 1.4 Biogeographic Drivers on Snow, Temperature and Tree Growth ........................... 5 1.5 Research Gaps, Study Rationale, and Theoretical Framework ............................... 8 1.6 Research Questions ..................................................................................................... 10 1.7 Conclusions .................................................................................................................. 12 Chapter 2: Field Methods, Growth Trend, and Productivity Analysis ................................ 13 2.1 Introduction ................................................................................................................. 13 2.2 Plot Selection and Study Area ................................................................................... 13 2.3 Biometric Inventory .................................................................................................... 16 2.4. Dendrochronological Processing .............................................................................. 17 2.5 Growth Trend and Data Visualization Analyses ..................................................... 21 III 2.5.1 Long-term Growth Trends ..................................................................................... 21 2.5.2 Long-Term Growth Patterns and Breakpoint Detection ........................................ 24 2.5.3 Spectral and Wavelet Analysis .............................................................................. 30 2.5.4 Species and Stand-Level Biomass Calculations .................................................... 31 2.5.5 Scaled Biomass Timeseries.................................................................................... 34 2.5.6 Biomass Trends Across Exceptional Growth Periods ........................................... 36 2.6 Biomass Volatility Index ............................................................................................ 40 2.7.1 Introduction ............................................................................................................ 43 2.7.2 Species and elevation-specific variability .............................................................. 43 2.7.3 Biomass dynamics and structural contributions .................................................... 44 2.8 Conclusions and Future Directions ........................................................................... 45 Chapter 3: Climate Sensitivity of High and Low-elevation Study Sites............................... 47 3.1 Introduction ................................................................................................................. 47 3.2 ClimateNA Data and Validation................................................................................ 48 3.3 Site Climate Conditions: Contemporary and Paleoclimate Perspectives .............. 51 3.3.1 Temperature ........................................................................................................... 51 3.3.2 Precipitation ........................................................................................................... 52 3.3.3 Summer Heat Moisture Index (SMH).................................................................... 53 3.3.4 Historical Context and Holocene Legacy .............................................................. 53 3.4 Climate Trend Analysis .............................................................................................. 54 3.4.1 Temperature Trends ............................................................................................... 55 3.4.2 Precipitation Trends ............................................................................................... 55 IV 3.4.3 Snowfall Trends ..................................................................................................... 56 3.5 Random Forest Climate Response Analysis ............................................................. 60 3.6 Principal Component Analysis .................................................................................. 67 3.7 Linear Mixed Effects Models ..................................................................................... 68 3.7.1 Temperature and Energy Balance .......................................................................... 71 3.7.2 Winter Temperature and Continentality ................................................................ 73 3.7.3 Degree Days ........................................................................................................... 74 3.7.4 Moisture and Snowpack Dynamics ....................................................................... 74 3.8 Influence of Pacific Decadal Oscillation ................................................................... 76 3.9 Climate Period Analysis: Impacts on Growth and Productivity ............................ 82 3.9.1 T-test comparisons of climate conditions .............................................................. 82 3.9.2 Random Forest classification of high vs. low growth periods ............................... 83 3.9.3 Elevation-dependent divergence in response ......................................................... 83 3.9.4 Species-specific climate responses ........................................................................ 84 3.10 Climate Impact Index (CII) ..................................................................................... 88 3.11 Discussion Climate Sensitivity and Space-for-Time Substitution Framework ... 90 3.11.1 Species and Elevation-Specific Climate Sensitivity ............................................ 90 3.11.2 Evaluating the Space-for-Time Substitution Framework .................................... 94 Chapter 4: Forecasting Forest Futures and Carbon Resilience in the ITR ........................ 97 4.1 Synthesis of Climate Sensitivity and Carbon Dynamics in the ITR....................... 97 4.2 Regional Analogs....................................................................................................... 100 V 4.3 Monitoring Adaptation and Future Forest Evolution ........................................... 102 4.4 The Great Caribou Rainforest Initiative ................................................................ 103 References Cited ................................................................................................................... 106 Appendix A. Dendrochronological Distribution and Normality Assessments ...... 122 Appendix B. Spectral and Wavelet Analysis of Chronology Periodicity ..................... 124 Appendix C. Species- and Stand-Level Biomass Dynamics .................................... 128 Appendix D. Climate Summary Tables by Elevation and Period ...................................... 131 Appendix E. Key to Random Forest Climate Variables ......................................... 134 VI List of Tables Table 1. Descriptive Dendrochronology Statistics. Presents key metrics for evaluating treering data variability, signal strength, and responsiveness to environmental changes, including Standard Deviation (STD), Mean Series Correlation (MSC), First Order Autocorrelation (AR1), Expressed Population Signal (EPS), Signal-to-Noise Ratio (SNR), and Mean Sensitivity (MS). ..................................................................................................................... 20 Table 2. Long-term linear growth trends of dimensionless tree-ring indices between 19012021. Slope directions are reported for the purpose of interpretation, although no trends significantly differ from zero at p= 0.05. ................................................................................ 22 Table 3. Linear growth trend analysis on the raw ring-width series of the Cedar High and Cedar Low Chronologies. The values in parentheses represent the numbers of trends that differ significantly from zero at p= 0.05. ................................................................................ 22 Table 4. Summary of identified high and low growth periods across the seven Longworth tree-ring chronologies. Periods exceeding a +0.5 threshold are classified as high growth, while those below -0.5 are marked as low growth based on significant deviations observed in four or more chronologies. ...................................................................................................... 26 Table 5. Frequency values for species presence and absence data during high and low growth periods. *High-elevation values have been scaled to reflect one fewer sampled species. ...... 27 Table 6. Breakpoint analysis results for all seven Longworth chronologies. Years listed with negative signs denote breakpoints occurring during periods of declining growth. ................ 28 Table 7. Forest biometrics of tree populations in the low and high-elevation plots of the Longworth study area. Diameter at breast height (DBH) and height (HT) reflect the mean measurements for all tallied trees within species grouping across subplot and macro plot designations. Basal area (BA) and trees-per-area (TPA) values are presented as area-scaled averages, combining subplot and macro plot data. ................................................................. 33 Table 8. Basal area (m²/ha), trees per hectare (TPH), and biomass (Mg C ha⁻¹) estimates for the high and low-elevation sites in the Longworth study area................................................ 33 Table 9. Species-level biomass estimates for high and low-elevation plots at the Longworth Study area................................................................................................................................ 34 Table 10. Low and high-elevation biomass accumulation percent gains and declines for the high and low growth periods. Percent gains and losses are calculated from modeled period biomass accumulation (Table 11), and the empirical 2021 biomass baseline estimates are presented in Tables 8 and 9. .................................................................................................... 40 Table 11. Relative species-, stand-, and landscape-level biomass (Mg C ha⁻¹) for the seven Longworth chronologies during the two historic high-growth periods, two low-growth VII periods, and the recent interval. Biomass values represent the sum of annual increments, scaled for unequal period lengths. .......................................................................................... 40 Table 12.Summary Table of growth trends and biomass for the seven Longworth Chronologies. .......................................................................................................................... 46 Table 13. Linear regression results between the Barkerville instrumental records and Barkerville Climate NA data (top) for the Barkerville Town Site. MAT = Mean annual temperature, MAP = Mean annual precipitation, and PAS = Precipitation as snow. ............. 50 Table 14. Linear regression results between the Barkerville instrumental records and Longworth Climate NA high-elevation data. MAT = Mean annual temperature, MAP = Mean annual precipitation, and PAS = Precipitation as snow. ............................................... 50 Table 15. Climate NA 1901-2021 normals for the Longworth high and low-elevation study sites. All climate measures exhibit significant elevation-related differences, confirmed by ttests (p < 1.50 × 10⁻¹³). ........................................................................................................... 57 Table 16. Lag distributions of aggregated species-level predictor variables across low and high-elevation chronologies. ................................................................................................... 63 Table 17. Percent distributions of variable lags for the Longworth high-elevation chronologies. ........................................................................................................................... 64 Table 18. Percent distributions of variable lags for the Longworth low-elevation chronologies. ........................................................................................................................... 64 Table 19. Percent distribution of species-specific variable groupings irrespective of lags for the Longworth sites. ................................................................................................................ 64 Table 20. Variables selected by RF modeling for low-elevation species at the Longworth Site. Variables highlighted in green represent lag 0, variables highlighted in blue represent lag 1, and variables highlighted in red represent lag 2. Please see Tables 33-35 for a key to the abbreviated variable names. .............................................................................................. 65 Table 21. Variables selected by RF modeling for the high-elevation species at the Longworth Site. Variables highlighted in green represent lag 0, variables highlighted in blue represent lag 1, and variables highlighted in red represent lag 2. Please see Tables 33-35 for a key to the abbreviated variable names. .............................................................................................. 66 Table 22. Table of variables included in PCA. PCA was conducted separately for each elevation gradient using Climate NA metrics specific to each site. Values in bold were selected by RF but will be integrated mixed effects modeling as stand-alone variables to evaluate the influences of specific effects. Please see Tables 34-36 for a key describing climate variable abbreviations. ............................................................................................... 68 VIII Table 23. Random effects, AIC, and R² values from LME models for the seven Longworth chronologies. Values show standard deviations for plot, tree, and residual variance, along with marginal and conditional R²............................................................................................ 71 Table 24. LME results for the seven Longworth Chronologies. Significant positive correlations are highlighted in Blue, and significant negative correlations are highlighted in Red. Please refer to Tables 34-36 for a key to the variable abbreviations.............................. 76 Table 25. Parametric linear regression and non-parametric Spearman rank correlations between PDO and the seven Longworth chronologies. Correlations in bold are significant at p<0.035. .................................................................................................................................. 80 Table 26. Climate WNA anomalies for historic high, low, and recent periods. The long-term average between 1901-2021 is included for reference. The long-term average between 19012021 is included for reference. Values significantly different from the long-term average are shown in bold. The Climate Impact Index is calculated as a composite index of Mean Annual Temperature (MAT) and Precipitation as Snow (PAS). ............................................ 86 Table 27. Results of random forest analysis focused on predicting the climatic drivers of high and low-growth periods identified in this study. While none of the correlations were significant, the model displayed moderate overall accuracy, exhibiting 100% accuracy in predicting high-growth periods and 60% in predicting low-growth periods. ......................... 86 Table 28. High elevation Climate NA averages for high and low growth periods. Values in bold are significantly different from the period mean as per the results of a t-test p<0.05. ... 87 Table 29. Low elevation Climate NA averages for high and low growth periods. Values in bold are significantly different from the period mean as per the results of a t-test p<0.05. ... 88 Table 30. This table presents the results of the initial normality test results, Box-Cox transformation lambda values, and standard deviations after transformation, which are used to assess and correct for normality and variance in the annual ring width. .......................... 122 Table 31. Biomass coefficients were used to calculate biomass at the Longworth site. Values are sourced directly from Ung et al., 2008............................................................................ 128 Table 32. Percent biomass gains and declines for the seven Longworth chronologies during the same intervals shown in Table 12. Percent change is calculated from trough to peak (or vice versa), and BVI reflects combined climate sensitivity and proportional variability. .... 129 Table 33. Net biomass changes (Δ Mg C ha⁻¹) from the 2021 baseline for categorized moisture tolerance and elevational species groupings during high and low growth periods.129 Table 34. Percent change from the 2021 baseline for categorized moisture tolerance and elevational species groupings during high and low growth periods. .................................... 130 IX Table 35. Table of high elevation Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), and Precipitation as Snow (PAS) 1901-2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. .......................................... 131 Table 36. Table of low elevation Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), and Precipitation as Snow (PAS) between 1901-2021.Values denoted in bold are significantly different from the long-term mean at p < 0.05................................... 131 Table 37. Table of low-elevation Maximum Temperatures (TMax) between 1901-2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. .... 132 Table 38. Table of low-elevation Minimum Temperatures (TMin) between 1901 and 2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. .... 132 Table 39. Table of high-elevation Maximum Temperatures (TMax) between 19012021.Values denoted in bold are significantly different from the long-term mean at p < 0.05. ............................................................................................................................................... 132 Table 40. Table of high-elevation Minimum Temperatures (TMin) between and 9012021.Values denoted in bold are significantly different from the long-term mean at p < 0.05. ............................................................................................................................................... 133 Table 41. Key to temperature variables selected by RF modeling across the seven Longworth chronologies. ...................................................................................................... 134 Table 42. Key to the drought and humidity variables selected by RF modeling across the seven Longworth chronologies. ............................................................................................ 135 Table 43. Key to the snow, precipitation, and frost variables selected by RF modeling across the seven Longworth chronologies. ...................................................................................... 136 Table 44. Correlation results for high elevation chronologies during high growth periods and Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. ..... 137 Table 45. Correlation results for high elevation chronologies during low growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. ................................................................................................................. 138 Table 46. Correlation results for low elevation chronologies during high growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. ................................................................................................................. 139 Table 47. Correlation results for low elevation chronologies during low growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation X (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. ................................................................................................................. 140 XI List of Figures Figure 1. Map of the Study area in Longworth, BC ............................................................... 15 Figure 2. Layout of nested subplots used for biometric sampling at Longworth, BC. The outer radius in red is 17.84 m and the inner radius is 7.98 m. ................................................ 16 Figure 3. Long-term growth trend analysis for the four low-elevation Longworth chronologies between 1901-2021. The long-term trendline is colored in gray, and the LOESS smoothing curve is in bold. LW Fir stands for Douglas-fir Low............................................ 23 Figure 4 Long-term growth trend analysis for the three high-elevation Longworth Chronologies between 1901-2021. The long-term trendline is colored gray, and the LOESS smoothing curve is bold. ......................................................................................................... 23 Figure 5. Annual ring-width series for the seven Longworth chronologies paired with 10-year moving averages of high and low-elevation chronologies. .................................................... 26 Figure 6. High and low-elevation z-scores for the 1901-2021 study period. High growth periods common to four or more chronologies are highlighted in blue, whereas low periods common to four or more chronologies are highlighted in red. Dashed lines denote the zscores of 0.75 and -0.75, respectively. .................................................................................... 28 Figure 7. Standardized chronologies show an exponential smoothing curve in blue, annual standardized chronologies in gray, and statistical breakpoints denoted by dashed vertical lines. ........................................................................................................................................ 29 Figure 8. Scaled biomass series for the seven Longworth Chronologies representing relative annual biomass increments (Mg C ha⁻¹) through the 1901-2021 study period. ..................... 36 Figure 9. Scatter plot of regression between Longworth Climate NA and the long-term Barkerville instrumental dataset. ............................................................................................ 51 Figure 10. High-elevation Climate NA Mean Annual Temperature, Mean Annual Precipitation, and Precipitation as Snow 1901-2021 for the Longworth Site......................... 58 Figure 11. Low-elevation Climate NA Mean Annual Temperature, Mean Annual Precipitation, and Precipitation as Snow 1901-2021 for the Longworth Site......................... 59 Figure 12. High and low-elevation Climate NA maximum and minimum temperatures between 1901-2021 for the Longworth Site. .......................................................................... 60 Figure 13. Smoothed 10-year moving z-score averages of high and low-elevation chronologies paired with PDO. On the PDO curve, areas shaded in red represent negative phase PDO, whereas areas shaded in blue represent positive phases. .................................... 81 Figure 14. Violin plots for the seven Longworth chronologies represent the distribution of standardized tree-ring growth indices for different species grouped by elevation. The width XII of each violin indicates the density of data points at different values, with bulges suggesting a more frequent distribution of growth indices and tapers indicating less frequent values. The central dot in each violin marks the median of the distribution. ........................................... 123 Figure 15. Spectral decomposition of the seven Longworth chronologies. This plot shows the relative power of periodicities between 1-40 years. ....................................................... 125 Figure 16. Morlet wavelet analysis of the four Longworth low-elevation tree-ring chronologies. The color-coded power spectral density illustrates a power gradient across various spectral bands. Areas of significance at the 99% confidence level are outlined in black. The Cone of Influence (COI) is delineated by a black crosshatch pattern. ................ 126 Figure 17. Morlet wavelet analysis of the three Longworth high-elevation tree-ring chronologies. The color-coded power spectral density illustrates a gradient of power across various spectral bands. Areas of significance at the 99% confidence level are outlined in black. The Cone of Influence (COI) is delineated by a black crosshatch pattern. ................ 127 Figure 18. Field sampling at the Low Elevation Site............................................................ 141 Figure 19. Low Elevation site in winter................................................................................ 142 Figure 20. Photograph of upper elevation site, showcasing smaller diameter, more densely spaced trees. .......................................................................................................................... 142 Figure 21. Image of canopy decline and dieback directly upslope from the Low Elevation site. ........................................................................................................................................ 143 Figure 22. Photograph taken just below the High Elevation site showcasing rot-driven blowdown damage. ............................................................................................................... 144 Figure 23. Photograph taken just below the High Elevation site showcasing rot-driven blowdown damage. ............................................................................................................... 145 XIII Acknowledgments This work would not have been possible without the support, guidance, and encouragement of an extended web of individuals who contributed to each idea, tree core, yoga pose, R session, pedal stroke, revision, and shared meal along the way. Primarily, thank you to my advisor, Dr. Che Elkin, for your steady guidance, insights, and availability. I am grateful for your belief in me, your technical support, our days in the field, and your leadership, exemplified by your daily bike commute up the hill. To my committee members, Dr. Joseph Shea and Dr. Hardy Griesbauer: thank you for your expertise and feedback, which challenged me to refine this work into its final form. I am incredibly grateful to my colleagues in the Mixedwood Ecology Lab for welcoming me with open arms and sharing trail runs, rides, and Kokanee dinners. And to Dan, my UNBC brother from Minnesota, although we did not make it to the finish line together in person, your enduring smile, belief in possibility, and encouragement remained steadfast. To my early academic mentors, Dr. Greg Wiles and Dr. Neil Pederson, thank you for igniting the spark and inspiring my journey into climate science. Dr. Lauren Oakes, thank you for helping me rediscover hope and resilience amidst ecological change. Your reunification of heart in science is deeply imprinted on this work. A heartfelt thank you to my dear friend and R tutor, Tom Day, for your patience and steadfast resolve in helping me navigate the steep coding learning curve, even on the high seas of Alaska. You were my syntax lighthouse. To Dr. Sarah Klein, Dr. Kathryn Fentress, Clare Jones, Emily Milner, Daniel Leonard, and Eli Buren: your counsel and care brought balance through thick and thin. To my business partner, Nick, your no-nonsense mantra, "get this thing done, we have work to do", kept me moving as we built carbon projects across the continent. To my yoga teachers, Alisha, Liza, Annie, and Jen Lo, thank you for holding space to renew. To my friends, thank you for your patience during my long disappearance into the woods and Suzzallo Library. Your generosity, cold dips, and humor kept me going with a full belly and a smile on my face. To my family: Greg Clarke, for offering me a place to stay in Canada; Peggy Clarke and Kevin Malcomb, for gifting me the computer upon which countless hours of analysis and writing occurred. To my brother, Jake, thank you for our weekly conversations about forests, carbon, and statistics. Your curiosity and insights have been a constant source of inspiration. To my rediscovered sister, Molly Hatcher, our connection this spring brought inspiration and momentum on the home stretch. Finally, to my partners past and present, thank you for your wisdom, loving presence, and patience that gave me the courage to take this leap. You taught me what it means to embody purpose, receive love, and reflect it back through the unique gifts we offer the world. This work is a testament to the collective effort behind meaningful research. Thank you for your kindness, wisdom, and belief in me. XIV Dedication To those who forged the path before me, carving trails to new ideas and ways of knowing; To those who believed in me and stood by me when the doctors told me to stay home and take art classes at community college; And to the cedars of what will become the Great Caribou Rainforest May you continue to thrive, inspire, and lead us toward a more beautiful world than we ever knew was possible. XV Chapter 1: Introduction, Organization, Imperative, Literature Review, and Hypotheses 1.1 Introduction and Organization This work investigates the climatic response of an Inland Temperate Rainforest ecosystem on the western slope of the Canadian Rocky Mountains in Longworth, British Columbia (BC). Combining forest biometric and dendroclimatological techniques, this research addresses gaps in understanding how individual tree climate sensitivity influences forest growth and stand-level dynamics. The findings from this study contribute to an essential understanding of forest evolution under changing climate regimes and will inspire and support future studies and conservation efforts across broader geographic gradients in both interior and coastal rainforest ecosystems of BC. There are four core chapters in this thesis. Chapter 1 serves as an introduction and literature review on recent drought in western North America, followed by a discussion of previous dendrochronological studies documenting forest responses to climate variability. The chapter concludes with research questions and hypotheses to be evaluated in Chapters 2 and 3. Chapter 2 presents the site selection rationale and methodology, forest biometric sampling, and tree-ring collection techniques, followed by a detailed review of chronology development, growth, and productivity trend analysis. Chapter 3 documents original dendroclimatic research detailing methodology, analysis, and results highlighting forest climate sensitivity at upper and lower-elevation study sites. Finally, Chapter 4 synthesizes the findings from previous chapters to address the influence of recent climate variability on forest productivity in Longworth, BC. This thesis concludes with a data-driven call for strengthened conservation measures in the Inland Temperate Rainforest (ITR), allowing time and space for critical adaptation in the face of future climate change. 1 1.2 The Imperative Forested ecosystems are significant terrestrial carbon sinks, sequestering up to 30% of anthropogenic greenhouse gas emissions (Pan et al. 2011; Mo et al. 2023). Recognized for their dual role as carbon sinks and sources, considerable attention has focused on forest carbon stability amidst climate change and extreme events (IPCC 2021). In western North America, old-growth temperate rainforests of coastal and interior BC are among the most carbon-dense ecosystems on Earth, playing a vital role in sequestering atmospheric CO₂ and mitigating climate change (Keith et al. 2009; Anderegg et al. 2019; Case et al. 2021; DellaSala et al. 2022). While much research has been focused on arid montane and boreal forests, less attention has focused on understanding the subtler impacts of recent climate variability on the temperate rainforest ecosystems (Halofsky et al. 2018a; Hammond et al. 2022; Kannenberg and Maxwell 2022). These rainforests are important stores of carbon and biodiversity, with old-growth stands valued for their high-volume, long-lived trees and organic-rich soils (Luyssaert et al. 2008; DellaSala et al. 2022)(Luyssaert et al., 2008; DellaSala et al., 2022b). The structural complexity enhances of these forest systems, enhances ecosystem stability during climatic shifts, promoting resilience against disturbance (DellaSala et al. 2021). However, British Colombia’s Inland Temperate Rainforest remains largely underprotected and underappreciated their non-timber goods and services. Since the late 20th century, 286,000 ha of old-growth rainforests in the ITR have been harvested, leaving only 4.6% of the original 1.3-million-hectare extent (DellaSala et al., 2021). Logging in the ITR critically impacts its carbon storage potential and undermines its function as a climateresilient carbon sink (Frey et al. 2016; DellaSala et al. 2021; Birdsey et al. 2023). 2 In contrast, British Colombia’s coastal rainforests benefit from more robust protections through initiatives like the Great Bear Rainforest Act, which offers a model for climate mitigation and ecosystem protection (Griess et al. 2019). With the ITR facing fewer protections and continued development pressures, research directed towards understanding its climate response and carbon cycling is imperative for informing evidence-based, bestpractice conservation efforts and ensuring the region’s climatic resiliency. 1.3 Climate Change and Forest Response in Western North America Short- and long-term climate variability significantly influences tree growth and forest productivity through shifts in growing season quality and duration. Across Canada, temperatures have risen by 1.7°C since 1948, which is twice the global average (Case et al. 2021; Hammond et al. 2022). Projections estimate a further 1.8 to 6.0°C increase by the century’s end, depending on future greenhouse gas emissions (Bush and Lemmen 2019). In BC, the mean temperature rose by 1.9°C between 1948 and 2016, with the most dramatic warming in winter months (3.7°C). This warming trend is more pronounced at higher latitudes, where northern and interior BC are heating faster than the coastal regions (White et al. 2016). These temperature increases are accompanied by shifting precipitation patterns. While BC’s temperate rainforests receive substantial precipitation, future rainfall increases of 4 to 14% may not be sufficient to offset rising evaporative demands and seasonal moisture redistribution that favors winter months (White et al., 2016; Bush and Lemmen, 2019). Winter warming and increased precipitation are expected to decrease snowpack in the Canadian Rockies by as much as 72% by the end of the 21st century (Cho et al. 2021). 3 Reduced snowpack, earlier snowmelt, and more frequent rain-on-snow events will significantly alter hydroclimatic conditions, particularly in ecosystems reliant on snowpack to sustain soil moisture into the growing season (Casirati et al. 2023). Overlaying long-term climate trends, there has been a rise in extreme weather events across western North America (Swain et al. 2020). Between 2013 and 2021, a large area of warm Pacific Ocean water, commonly referred to as “The Blob”, amplified by high-pressure ridging, brought sustained 2 to 5°C above-average temperatures to BC and adjacent US states (OWSC 2015; Szeto et al. 2016; Tseng et al. 2017; Mass et al. 2024). This rise in temperature was accompanied by a 40 to 80% reduction in snowpack and a 75% decrease in summer precipitation across the Pacific Northwest (PNW) (Coulthard et al. 2016; Marlier et al. 2017; Reyes and Kramer 2023). Though “The Blob” dissipated in 2019, evidence suggests that marine heatwaves and “heat dome” events may become a persistent feature of Pacific Northwest climate, accelerating the arrival of climate extremes that were not projected until later in the 21st century (Frölicher and Laufkötter 2018; Heeter et al. 2023). The most visible consequence of these recent extremes has been an unprecedented surge in regional wildfires. Between 2013 and 2017, over 8.8 million hectares of forest burned across the Pacific Northwest, with fires often overlapping areas experiencing severe insect outbreaks (Young et al. 2017; Halofsky et al. 2018b). While these disturbances are well-documented, the subtler impacts of recent drought and heat events on forest productivity, particularly in temperate rainforest ecosystems, remain under-researched. Dendroclimatological studies across western North America have illuminated the nonlinear and often divergent responses of forest species to warming even prior to recent heat waves (Gedalof and Smith 2001; D’Arrigo et al. 2008; Anchukaitis et al. 2013; Salzer et al. 4 2014; Wilson et al. 2016). For example, at coastal Pacific Northwest sites, Gedalof and Smith (2001) found that Tsuga mertensiana (mountain hemlock, T. mertensiana) and Abies lasiocarpa (subalpine fir, A. lasiocarpa) responded positively to deeper snowpacks at moisture-limited sites. At mesic sites, however, snowpack delayed budburst and reduced growth. Similarly, Salzer et al. (2014) observed that Pinus longaeva (bristlecone pine, P. longaeva) at high elevations shifted from temperature-limited growth to drought sensitivity over recent decades. Trees growing just 60 to 80 meters higher in altitude showed the opposite trend, exhibiting enhanced growth in response to warming. In BC’s interior, similar studies have focused on Pseudotsuga menziesii var. glauca (Douglas-fir, P. menziesii), western red cedar, and other hybrid spruce species across elevation and moisture gradients (Miyamoto et al. 2010; Griesbauer and Scott Green 2010; Konchalski 2015; Wiley et al. 2018). These studies reveal varied responses to snowpack and seasonal drought. Growth–snowpack associations appear to vary by elevation and timing. While early or persistent snow cover may truncate the growing season or interfere with cold hardening, mid- and late-winter snowpack provides critical protection and improves moisture availability for early season growth (Laroque and Smith 1999; Wiley et al. 2018). 1.4 Biogeographic Drivers on Snow, Temperature and Tree Growth At mesic sites in coastal BC and Alaska (AK), recent research has emphasized the importance of winter snowpack in protecting the shallow root systems of yellow cedar and more recently, western hemlock, from freeze-thaw injury (Beier et al. 2008; Jarvis et al. 2013; McGrath et al. 2016; Comeau and Daniels 2022). Decline in yellow cedar has been most evident in low elevation stands where diminishing snowpacks fail to insulate fine roots 5 from frequent freeze-thaw cycles caused by alternating coastal maritime and Arctic air masses (Beier et al., 2008). These events can trigger dehardening and early sap flow, leaving trees vulnerable to subsequent cold damage (Hennon et al. 2006). Tree-ring chronologies developed in these stands help clarify these dynamics, revealing high interannual variability and divergent growth responses. Some trees show sharply reduced growth associated with winter injury, while others nearby exhibit growth release, likely due to reduced competition as neighboring trees decline (Beier et al. 2008; Wiles et al. 2012). These patterns suggest a complex interplay between physiological vulnerability and microclimatic variation. Generally, low-elevation yellow cedar and western hemlock exhibit a positive growth association with winter snowpack, supporting the notion that deep snow protects roots and helps maintain dormancy until spring conditions become favorable for growth (Beier et al., 2008; Wiles et al., 2012). Numerous low-elevation chronologies from coastal Alaska show a transition over the past century. Before 1950, tree growth was positively associated with warm growing seasons. Since that time, many sites have exhibited increasingly negative responses to summer temperatures. In contrast, upper elevation populations of the same species continue to display positive and even amplified growth responses to recent warming and reduced snowpack (Beier et al., 2008; Jarvis et al., 2013; McGrath et al., 2016). These divergent trends underscore the complex role of snowpack and elevation in shaping forest climate sensitivity. Similar studies in BC’s interior and coastal zones have investigated the role of snowpack and seasonal moisture variability across species and elevation gradients. Research on Douglas-Fir, hybrid spruce, and western red cedar has revealed diverse responses to fall and winter snowpack (Miyamoto et al., 2010; Griesbauer and Green, 2010; Konchalski, 6 2015; Wiley et al., 2018). On some sites, early snow cover appears to truncate the growing season or disrupt cold-hardening processes, leading to negative growth associations. Conversely, mid- and late-winter snowpack appears to provide critical protection for roots and contributes to moisture availability during the early growing season (Laroque and Smith, 1999; Wiley et al., 2018). Temperature effects are nuanced and vary by site. While warmer spring temperatures are generally associated with increased growth, hotter summers often correlate with reduced productivity, particularly at lower elevations or on moisture-limited sites (Stan and Daniels 2014; Wiley et al. 2018). In the Robson Valley, Konchalski (2015) found no consistent temperature-growth relationships in western red cedar based on north- versus south-facing aspects. However, significant elevation-related differences were noted: low-elevation stands showed strong negative growth responses to summer heat, whereas high-elevation stands exhibited minimal temperature sensitivity. In a nearby region of interior BC, Wiley et al. (2018) documented negative growth responses to summer heat in hybrid spruce but not in the more drought-tolerant Douglas-Fir. Intriguingly, soil moisture availability alone did not account for these differences. The authors suggest that other biogeographic factors, such as aspect and snowmelt timing, may mediate thermal stress, especially in mixed species stands. Together, these findings suggest that interior populations of species like western red cedar may experience temperature stress earlier in the growing season than their coastal counterparts, reflecting differences in elevation, snowpack retention, and the seasonal timing of drought. Collectively, these findings underline the importance of resolving forest-climate interactions at fine spatial scales, where species’ physiology, microsite conditions, and 7 topographic variation mediate climate sensitivity. Divergent growth responses observed across elevation bands and among co-occurring species indicate that both growth enhancement and decline may occur concurrently within the same forest stand. This spatial heterogeneity highlights the need for hierarchical, multi-species approaches to climateresponse analysis that can detect threshold behaviors, non-linear dynamics, and lagged effects, factors essential for anticipating future forest trajectories under continued climate forcing. 1.5 Research Gaps, Study Rationale, and Theoretical Framework Geographic variation across a species’ distribution provides a unique opportunity to study growth responses to climate across different physiological thresholds and niche requirements. In the context of climate change, understanding how trees have responded to past temperature and hydroclimatic variability can help refine models forecasting future forest response. This approach aligns with the principle of uniformitarianism—that the present is the key to the past, and vice versa (Wilmking et al. 2017). However, when projected future climate scenarios diverge from historical conditions, this assumption becomes problematic. To address this, the space-for-time substitution model provides an important lens for evaluating climate sensitivity across existing gradients in temperature, snowpack, and moisture availability. By comparing species growing in cooler or more moisture-limited regions with those in warmer or more favorable ones, this approach can help forecast how forests might respond under future climate conditions (Littell et al. 2008; Klesse et al. 2020). 8 For example, Klesse et al. (2020) evaluated Douglas-Fir growth across its latitudinal range from Mexico to British Columbia using a hierarchical mixed-effects model. The findings showed distinct gradients in productivity and climate sensitivity depending on geographic context. Rather than simply comparing raw growth rates, authors focused on growth–climate relationships revealing that species response to climate is not uniform but varies across landscapes. This approach offers a more robust framework for anticipating how warming and changing moisture regimes may affect future productivity and forest resilience. Despite the wealth of knowledge generated by previous studies and emerging examinations of the space-for-time substitution model, research gaps remain. Many existing studies focus on a single species or site, use inconsistent methodologies, or rely on outdated chronologies that fail to capture contemporary responses to rapid climate change. Particularly in British Columbia's Inland Temperate Rainforest, relatively few studies have sought to understand forest sensitivity to recent climate anomalies. While the biodiversity of the ITR has been increasingly documented, only two inventory-based biomass surveys (Matsuzaki et al. 2013; DellaSala et al. 2022) and one unpublished dendroclimatology thesis (Konchalski, 2015) have focused on the region. This limited body of work has offered foundational insights into biomass accumulation, climate sensitivity, and disturbance dynamics. However, given the accelerating pace of climate change and the ongoing industrial pressure placed on these forests, there is an urgent need for more comprehensive analyses. Specifically, a multispecies approach is needed that captures both species and stand-level implications for carbon storage and productivity. 9 This thesis seeks to address these gaps by examining the growth and climate response of seven dominant tree species at upper and lower elevation sites in the ITR. Using tree-ring and forest biometric data, the study investigates species-level responses to recent warming events, including the 2013–2021 “Blob” marine heatwave and explores how elevation, moisture availability, and temperature regime shape forest sensitivity. The study adopts a space-for-time substitution framework to infer how trees growing in cooler or higher elevation environments may respond under future climate conditions. By evaluating these patterns within a hierarchical structure, from tree to stand to landscape, this work contributes critical insight into how forests in the ITR may adapt in the face of global change. As the northernmost extent of the Inland Temperate Rainforest, the Longworth study area also captures the upper latitudinal range limits for western red cedar, western hemlock, and Douglas-fir. This adds a valuable geographic and ecological dimension to the space-fortime substitution model, complementing the elevational and old-growth dimensions already embedded in the study design. 1.6 Research Questions This study investigates the intricate relationships between biogeography, climate variability, carbon cycling, and long-term forest evolution through a focused set of three core research questions. To explore these dynamics, we selected a research site near the northernmost extent of the Inland Temperate Rainforest, where several keystone tree species approach the limits of their latitudinal range. This positioning adds a valuable geographic and ecological dimension to the space-for-time substitution framework, complementing the elevational and old-growth gradients already embedded in the study design. Together, these 10 spatial and ecological contrasts support a multi-scalar lens for assessing forest resilience, informing the design and interpretation of the study’s core questions. 1. How have trees in the ITR responded to both historical and recent climate variability? By analyzing tree-ring records over the instrumental climate period, this question explores long-term growth responses to climatic fluctuations and identifies whether the warm-dry conditions of the 2013–2021 drought period produced unique or amplified growth responses in the context of climate variability in the 20th century. It also considers whether these impacts are detectable across species and elevations. 2. How do species and elevation influence climate sensitivity? This question evaluates whether species growing at different elevations show distinct responses to temperature, snowpack, and seasonal moisture variability. By comparing growth-climate relationships, it seeks to identify which species or site conditions confer greater resilience or greater vulnerability under changing climate regimes. 3. Can space-for-time substitution effectively anticipate future forest response? This question examines the utility of using spatial gradients in elevation and species distributions as a proxy for temporal change. It assesses whether present-day variability in growth responses can help predict how forests may respond to future warming and altered hydroclimate regimes in the ITR and similar ecosystems. 11 1.7 Conclusions Chapter 1 introduced the ecological and climatological significance of British Columbia’s Inland Temperate Rainforest and outlined key research gaps related to forest carbon dynamics and climate sensitivity. A literature review on recent warming, snowpack decline, and species-specific growth responses highlighted the need for multi-scalar, elevation-based approaches. The chapter concluded by presenting three guiding research questions centered on growth variability, climate sensitivity, and the utility of space-for-time substitution. These questions frame the subsequent chapters, which integrate biometric and dendroclimatic data to investigate forest resilience across species and elevation gradients in Longworth, BC. 12 Chapter 2: Field Methods, Growth Trend, and Productivity Analysis 2.1 Introduction Chapter 2 presents an integrated analysis of forest biometrics, species growth trends, and above-ground biomass across two elevation gradients in the Inland Temperate Rainforest. The objectives are threefold: (1) to describe the field and dendrochronological methods used to quantify stand structure and develop seven species-level ring-width chronologies; (2) to examine tree growth patterns using linear and non-linear techniques; and (3) to estimate biomass trends using scaled chronologies as a proxy for net primary productivity (NPP). While the spatial scope of this work is limited, emphasis is placed on replicable methods and robust data processing approaches applicable to other temperate rainforest ecosystems in British Columbia and the Pacific Northwest. This chapter focuses on temporal variability in growth and biomass; interpretation of environmental drivers is reserved for Chapters 3 and 4. The results provide an empirical foundation for subsequent analysis of climate sensitivity and carbon storage potential in the region. 2.2 Plot Selection and Study Area Study sites were selected to capture elevation-mediated variation in species-specific climate responses within the Interior Cedar–Hemlock (ICH) zone of the Northern Rocky Mountains. Candidate stands were identified through GIS-based analysis of the Biogeoclimatic Ecosystem Classification (BEC) database focusing on the Robson Valley, an area at the northernmost extent of the Inland Temperate Rainforest. To enhance the climate sensitivity of forests surveyed in our analysis, we screened for south facing slopes and 13 pronounced elevation gradients to maximize thermal and moisture variability between upper and low elevation stands. The focus on old-growth served three purposes: (1) to reduce growth signals originating from natural and human disturbance; (2) to climate response, and productivity in forests of high conservation value; and (3) to sample intact, mixed-species communities reflecting long-term growth dynamics. Field reconnaissance of remote-sensed candidate sites confirmed the Longworth Lookout Trail area near Longworth, BC, as an optimal location for addressing the study’s core research (Figure 1). The site features well-developed old-growth structure and mixedspecies composition across both elevation bands. Notably, Longworth captures several major canopy species near their northernmost range limits, providing a natural setting for evaluating spatial gradients as proxies for future climate scenarios. Dominant species include western redcedar (Thuja plicata), western hemlock (Tsuga heterophylla), Engelmann spruce (Picea engelmannii), Douglas-fir (Pseudotsuga menziesii var. glauca), and subalpine fir (Abies lasiocarpa). At each elevation, circular nested subplots were arranged in a four-point cloverleaf pattern to capture a spatially representative footprint of forest structure. Each subplot included a smaller 7.98 m radius plot to inventory all trees ≥10 cm DBH and a larger 17.84 m radius to include trees ≥70 cm DBH (Figure 2). This design optimized the sampling of widely spaced, large-diameter trees typical of old-growth stands while accurately representing forest density and species composition. 14 Figure 1. Map of the Study area in Longworth, BC 15 2.3 Biometric Inventory Within each subplot, all canopy and co-dominate trees within size class thresholds were targeted for biometric (ie. Species ID, DBH, and Length) and dendrochronological sampling. Here canopy-dominant trees were considered as those with crowns extending above the general canopy and co-dominant trees were considered as those with crowns forming part of the main canopy layer receiving light primarily from above. Two increment cores were extracted at perpendicular angles at DBH to account for irregular stem geometry. Where in-plot tally of species, such as Douglas-fir and subalpine fir, were insufficient for cross-dating, additional canopy-dominant trees were sampled nearby, taking care to match plot conditions as closely as possible. Figure 2. Layout of nested subplots used for biometric sampling at Longworth, BC. The outer radius in red is 17.84 m and the inner radius is 7.98 m. 16 2.4. Dendrochronological Processing Increment cores were mounted and sanded using progressively finer sandpaper to enhance ring boundary visualization. Annual ring widths were dated visually by counting from bark to pith, with cross-dating conducted using marker years that were verified at both tree and species levels following standard dendrochronological procedures (Maxwell and Larsson 2021). Cores with missing or false rings, especially common in western redcedar, were compared with the opposite core from the same tree and to well-dated reference samples to assist with dating challenges. Core images were scanned at high resolution and measured in CooRecorder (v.9.6; Cybis Elektronik 2020). Preliminary chronologies were constructed using 8–10 wellpreserved cores per species, then iteratively developed core by core by incorporating more difficult to date samples. CDendro (v.9.6) was used for early cross-dating, with final quality control and chronology validation performed in R using the dplR package (Bunn 2008, 2010). Each tree-ring series was pre-whitened, detrended and auto correlation was removed using autoregressive modeling, prior to standardization and comparison against species-level master chronologies. Chronological coherence and potential dating errors were assessed using 50-year moving segments and inter-series correlation statistics. Ring-width series were detrended using a conservative cubic smoothing spline with a 0.50 frequency response (Cook et al. 1990 p. 19), selected to balance biological persistence with interannual variability. Due to extensive heartwood rot and the large size of many canopy-dominant trees, modern age-dependent standardization methods such as Regional Curve Standardization (RCS) were not feasible. Instead, a spline approach was applied across 17 all species, with dimensionless growth indices calculated by dividing observed values by expected values to produce standardized series. This yielded well-dated, annually resolved growth indices suitable for characterizing long-term radial growth and establishing specieslevel productivity baselines. These standardized series served as the foundation for constructing site-level chronologies and evaluating inter-tree growth coherence across elevation and species. Following standardization, tree-ring indices were averaged using Tukey’s biweight robust mean to develop species chronologies (Cook et al. 1990; Bunn 2008). Descriptive statistics were then applied to assess signal strength, variability, and biological persistence (Fritts et al. 1990) ( Table 1). Mean series correlation (MSC) was used to quantify coherence between individual series and the overall species signal. MSC values were higher at upperelevation sites, ranging from 0.61 (Subalpine Fir High) to 0.68 (Cedar High), compared to 0.46–0.67 at lower elevations. This suggests slightly greater growth synchrony in upper elevation stands, likely reflecting stronger climatic constraints such as temperature and snowpack duration. Autocorrelation metrics aided in detecting the persistence of growth trends. Firstorder autocorrelation (AR1) values ranged from 0.62 to 0.81, indicating that prior-year conditions consistently influenced annual growth across species and elevations (Table 1). Expressed Population Signal (EPS) values exceeded the standard dendrochronological threshold of 0.85 for all chronologies (Wigley et al. 1984), highlighting strong common signals throughout the dataset. Signal-to-noise ratio (SNR) values revealed distinct contrasts across elevation and species. Both cedar chronologies exhibited high SNR values (≥45), indicating strong 18 coherence in growth responses amongst chronologies. In contrast, Douglas-fir Low and Hemlock Low showed lower SNR values (<12), pointing to greater intra-species variability potentially linked to differences in microsite conditions, physiology, or disturbance history. These findings parallel MSC patterns, reinforcing the interpretation that high-elevation stands are shaped by more uniformly limiting environmental conditions (Table 1). Mean sensitivity (MS) values, used to evaluate interannual growth variability, were moderate and relatively uniform across species (0.20–0.24; Table 1). This indicates broadly similar growth responses across the dataset, with minor contrasts suggesting potential species-level differences in phenology and resource allocation. Recent critiques caution that mean sensitivity is dependent on the underlying variance structure of ring-width series, which varies by species, age, and site conditions (Bunn et al. 2013; Buras 2017). As such, while MS remains a useful complementary indicator, it should be interpreted in conjunction with the other statistics explored here. To further assess growth distributions and identify departures from normality, violin plots were generated for each standardized chronology (Figure A1). These plots visualize both the density and spread of ring-width indices, revealing distinct distributional patterns across elevations and species. Most low-elevation chronologies exhibited left-skewed shapes, indicative of frequent low-growth years and heightened sensitivity to environmental stress. High-elevation cedar and spruce series, by contrast, displayed more symmetrical distributions, suggesting greater uniformity in climate response. Subalpine Fir High, notably distinct from the other chronologies, exhibited a right-skewed distribution marked by frequent high-growth years, potentially reflecting more favorable responses to environmental conditions. 19 Informed by growth distribution patterns visualized in violin plots, Box-Cox transformations were applied to each chronology to address non-normality and reduce heteroscedasticity prior to time-series decomposition and statistical modeling. Using the boxcox function in the MASS package in R, optimal lambda (λ) values were applied to normalize variance and stabilize residuals. Transformation parameters and posttransformation standard deviations are reported in Table A1. This preprocessing step enhanced cross-species comparability and minimized the risk of propagating distributional artifacts into subsequent time-series analyses, including breakpoint detection, biomass modeling, and climate analyses. Table 1. Descriptive Dendrochronology Statistics. Presents key metrics for evaluating treering data variability, signal strength, and responsiveness to environmental changes, including Standard Deviation (STD), Mean Series Correlation (MSC), First Order Autocorrelation (AR1), Expressed Population Signal (EPS), Signal-to-Noise Ratio (SNR), and Mean Sensitivity (MS). Chronology n STD MSC AR1 EPS SNR MS Cedar Low 23 0.27 0.67 0.75 0.99 44.7 0.24 Spruce Low 23 0.16 0.66 0.81 0.93 13.98 0.2 Hemlock Low 20 0.19 0.48 0.65 0.92 11.32 0.23 Douglas-fir Low 20 0.16 0.46 0.68 0.91 10.65 0.24 Cedar High 29 0.20 0.68 0.74 0.98 45.46 0.2 Spruce High 24 0.17 0.65 0.72 0.96 24.87 0.2 Subalpine Fir High 23 0.22 0.61 0.62 0.97 36.09 0.22 20 2.5 Growth Trend and Data Visualization Analyses 2.5.1 Long-term Growth Trends To evaluate long-term trends in forest growth, we applied linear regression lines to standardized ring-width chronologies for seven dominant species across high- and lowelevation sites. Most chronologies exhibited subtle slope direction, with no trends statistically different from zero (Table 2; Figure 3). Slight positive trends were observed in Douglas-fir Low and Subalpine Fir High, while Cedar High and Spruce Low showed minor negative slopes. These weak signals may reflect both true biological variability and statistical dampening introduced by the standardization process (Peters et al. 2015). In particular, the use of spline detrending, while effective for isolating interannual variability, can reduce lowfrequency variance and compress long-term growth trends near the beginning and end of the series, potentially obscuring subtle increases or declines in recent growth. To explore the effects of standardization and inter-series averaging, we examined raw ring-width data for individual trees within the Cedar High and Cedar Low chronologies. This analysis revealed contrasting patterns: most Cedar High series exhibited declining growth, while Cedar Low trees were more evenly split between increasing and decreasing trends (Table 3). These findings suggest that while species-level chronologies provide useful generalizations, they may mask important individual-level dynamics. Because tree growth is inherently nonlinear, we applied LOESS (Locally Weighted Scatterplot Smoothing) to assess changes in growth trajectory over time. This method revealed consistent downward trends in five of the seven chronologies, including those with flat or positive linear slopes (Table 2, Figures 3 and 4). Only Douglas-fir Low maintained a positive slope under both linear and LOESS methods. These results highlight the utility of 21 LOESS in capturing subtle growth decelerations potentially missed by linear models, particularly in older trees or forests experiencing emerging stress. Together, these analyses indicate that although net growth trends remain statistically flat across the study period, several species and elevations show signs of long-term change. Cedar High and Spruce Low may be experiencing gradual growth declines, potentially linked to age, structure, emerging stressors, or artifacts of standardization. Further analysis in Chapter 3 explores the role of climate drivers in shaping these observed patterns. Table 2. Long-term linear growth trends of dimensionless tree-ring indices between 19012021. Slope directions are reported for the purpose of interpretation, although no trends significantly differ from zero at p= 0.05. Linear Trend LOESS Slope Chronology Intercept Direction P Value Residual SE DF Direction Subalpine Fir 0.88 Increase 0.93 0.22 119 Decrease Cedar High 1.08 Decrease 0.93 0.20 119 Decrease Spruce High 0.86 Increase 0.89 0.17 119 Decrease Cedar Low 0.87 Increase 0.93 0.27 119 Decrease Hemlock Low 0.80 Increase 0.86 0.19 119 Decrease Spruce Low 1.19 Decrease 0.82 0.17 119 Decrease Douglas-fir Low 0.885 Increase 0.903 0.16 119 Increase Table 3. Linear growth trend analysis on the raw ring-width series of the Cedar High and Cedar Low Chronologies. The values in parentheses represent the numbers of trends that differ significantly from zero at p= 0.05. % Positive % Negative n n Species Slope Growth Growth Positive Negative Cedar High -4.69E-05 30% 70% 9 (7) 20 (18) Cedar Low 5.78E-05 52% 48% 12 (7) 11 (6) 22 Figure 3. Long-term growth trend analysis for the four low-elevation Longworth chronologies between 1901-2021. The long-term trendline is colored in gray, and the LOESS smoothing curve is in bold. LW Fir stands for Douglas-fir Low. Figure 4 Long-term growth trend analysis for the three high-elevation Longworth Chronologies between 1901-2021. The long-term trendline is colored gray, and the LOESS smoothing curve is bold. 23 2.5.2 Long-Term Growth Patterns and Breakpoint Detection To explore multi-year fluctuations in forest growth, we calculated 10-year moving averages for each of the seven species-level chronologies. This approach smooths interannual variability and highlights sustained periods of above- or below-average growth across the 1901–2021 period. The resulting curves illustrate broad periods of sustained synchronous high or low growth, particularly during the early 1940s, mid-1960s to early 1970s, early 2000s, and the most recent decade (Figure 5). These intervals serve as visual confirmation of shared dynamics among species and elevation bands. Smoothed elevation series revealed similar overall variability between low- and highelevation sites, though multi-year lags were more evident in the first half of the 20th century. Low-elevation chronologies often exhibited growth anomalies one to two years earlier than their high-elevation counterparts (Figure 5). After 1960, this temporal offset weakened, with growth patterns becoming increasingly synchronized, likely reflecting stronger regional climate forcing that overrode the buffering effects of elevation. Complementing this visual analysis, z-scores were calculated for each chronology (Fritts 1976). Years with z-scores exceeding ±0.5 were considered significant positive or negative anomalies. Years in which four or more chronologies exhibited synchronous highor low-growth anomalies were classified as shared events (Figure 6; Table 4). Notable multiyear low-growth periods between 1925–1933, 1936–1942, 1959–1965, and 1991–2001 were consistently identified across z-score, breakpoint, and moving average approaches, while high-growth intervals in 1944–1952, 1982-1986 and 2004–2012 were also detected across all methods (Figures 5–7; Tables 4, 6). Additional shorter pulses, such as 1912–1917 and 1977– 24 1983, appeared in both the smoothed and z-score series and were corroborated by breakpoint records. Breakpoint analysis was conducted using the "breakpoints" function from the strucchange package in R (Zeileis et al. 2003). To focus on the most meaningful shifts, we limited detection to five potential breakpoints per chronology over the 1901–2021 period. Prior to analysis, an exponential smoothing factor of 0.2 was applied to reduce short-term noise while preserving medium- to long-term variability. Breakpoints were interpreted not as isolated, discrete events, but as approximate temporal markers of broader structural change. Compared to moving averages and z-score anomalies, breakpoint analysis highlighted additional divergence between species and elevation bands. While z-score and smoothed series emphasized shared multi-year intervals of high or low growth, breakpoint results revealed finer-scale and species-specific synchronous and divergent shifts that were sometimes lagged over the course of several years (Table 6; Figure 7). For instance, in 1918, both Cedar High and Subalpine Fir High exhibited positive breakpoints, while Cedar Low showed a negative breakpoint in 1920. Similar elevation-based divergences occurred in the late 1930s to early 1940s, mid- to late 1950s, and around 2000–2001. These contrasts suggest elevation- and site-specific responses to environmental conditions or disturbance legacies that may not significantly alter average growth but still shift the underlying trajectory. Periods of asynchronous response, especially between Cedar High and Low, further underscore the role of local factors such as phenology and disturbance history. Documented Hemlock Looper outbreaks in the 1920s, 1950s, and early 1990s (Konchalski 2015) may have contributed to these divergent or delayed responses, both within individual species chronologies and across elevation bands. 25 Figure 5. Annual ring-width series for the seven Longworth chronologies paired with 10-year moving averages of high and low-elevation chronologies. Table 4. Summary of identified high and low growth periods across the seven Longworth tree-ring chronologies. Periods exceeding a +0.5 threshold are classified as high growth, while those below -0.5 are marked as low growth based on significant deviations observed in four or more chronologies. Period Range Species High 1912-1917 Hemlock Low, Cedar Low, Cedar High, Spruce High High 1944-1952 Douglas-fir Low, Spruce Low, Cedar Low, Cedar High, Spruce High, Subalpine Fir High High 1982-1986 Hemlock Low, Spruce Low, Spruce High, Subalpine Fir High High 2004-2012 Douglas-fir Low, Spruce Low, Hemlock, Cedar Low, Cedar High, Spruce High, Subalpine Fir High Low 1901-1904 Douglas-fir Low, Spruce Low, Hemlock Low, Cedar Low, Spruce High, Subalpine Fir High Low 1925-1933 Douglas-fir, Spruce Low, Hemlock Low, Cedar Low, Cedar High, Spruce High, Subalpine Fir High Low 1936-1942 Cedar Low, Cedar High, Spruce High, Subalpine Fir High Low 1959-1965 Hemlock Low, Spruce Low, Spruce High, Subalpine Fir Low 1991-2001 Douglas-fir, Spruce Low, Hemlock Low, Cedar Low, Cedar High, Spruce High, Subalpine Fir High Low 26 Table 5. Frequency values for species presence and absence data during high and low growth periods. *High-elevation values have been scaled to reflect one fewer sampled species. High Low Species Total Growth Growth Cedar Low 7 3 4 Spruce Low 7 3 4 Hemlock Low 7 3 4 Douglas-fir Low 5 2 3 Cedar High 6 3 3 Subalpine Fir High 8 3 5 Spruce High 9 4 5 Low Elevation 26 11 15 High Elevation* 30.67 13.33 17.33 27 Figure 6. High and low-elevation z-scores for the 1901-2021 study period. High growth periods common to four or more chronologies are highlighted in blue, whereas low periods common to four or more chronologies are highlighted in red. Dashed lines denote the zscores of 0.75 and -0.75, respectively. Table 6. Breakpoint analysis results for all seven Longworth chronologies. Years listed with negative signs denote breakpoints occurring during periods of declining growth. Spruce High Subalpine Fir Cedar High Cedar Low Spruce Low 1927 1918 1918 -1920 -1936 1965 1983 -1945 1952 -1940 1938 1954 1983 -2001 - -1970 -1958 1956 -1973 -2001 - - 1989 1976 -1983 1991 - - - - 2000 -2001 - - - 28 Douglas Hemlock -fir Figure 7. Standardized chronologies show an exponential smoothing curve in blue, annual standardized chronologies in gray, and statistical breakpoints denoted by dashed vertical lines. 29 2.5.3 Spectral and Wavelet Analysis Spectral and wavelet decomposition were used to examine dominant temporal patterns in tree-ring growth. These analyses help to characterize whether growth variability occurs primarily on short, intermediate, or long-term timescales, and whether those patterns remain stable or shift across the 1901–2021 period. Spectral power analysis was conducted using the spectrum() function from the R package forecast, which decomposes time series into constituent frequencies (Hyndman and Khandakar 2008). This analysis revealed consistent peaks in the 2–4-year range across all species and elevations, reflecting the influence of short-term, multi-year variability. Additional decadal-scale peaks in the 11–12 and 16–20-year ranges were more prominent in high-elevation chronologies, especially Subalpine Fir High and Cedar High. The cooccurrence of multiple spectral peaks suggests that tree growth patterns are shaped by both interannual and decadal-scale variability. The more pronounced low-frequency signals in high-elevation chronologies likely reflect increased sensitivity to the persistent influence of large-scale climate oscillations that modulate growing season temperature, snowpack duration and moisture availability, highlighting the role of enduring environmental constraints in shaping growth responses at upper elevations. To examine how these frequency patterns evolved through time, we conducted wavelet analysis using the Morlet wavelet transform as implemented in the R package dplR (Torrence and Compo 1998; Bunn et al. 2013). Wavelet power spectra highlighted strong 2– 6 year bands in many species between the 1940s and 1990s, while longer, more stable decadal patterns were again most prominent in high-elevation species. Toward the end of the 30 time series, wavelet power declined in most chronologies, suggesting reduced periodic signal strength in recent decades. Full spectral and wavelet visualizations are presented in Appendix C (Figures C1– C3). These results are summarized in the Chapter 2 discussion and are explored further in Chapter 3 alongside climate response modeling. 2.5.4 Species and Stand-Level Biomass Calculations Forest inventory data collected from field plots in 2021 were input into speciesspecific allometric equations to estimate tree, species, and stand-level biomass. These estimates provided a static baseline from which to model annual biomass accumulation over the 1901–2021 period and served as inputs for the productivity analysis detailed later in this chapter (Table 7). Biomass coefficients were sourced from Boudewyn et al. (2007) and Ung et al. (2008), who developed empirical volume-to-biomass equations based on extensive inventories across Canadian forests (Table C1) (Boudewyn et al. 2007; Ung et al. 2008). Each tree’s gross biomass was calculated by summing modeled estimates of stem, bark, and branch components using the following equations: Biomassstem=exp(astem+bstem×ln(DBH)×ln(Height) (Eq 1) Biomassbark= exp(abark+bbark×ln(DBH)×ln(Height) (Eq 2) Biomassbranches = exp(abranches +bbranches ×ln(DBH) ×ln(Height) (Eq 3) Biomassgross= Biomassstem + Biomassbark + Biomassbranches (Eq 4) 31 where: a and b are species-specific coefficients DBH is in centimeters (cm), and Height is in meters (m) Gross biomass was aggregated across species and scaled from plot-level to perhectare estimates for both high- and low-elevation sites. While some uncertainty exists due to hollow boles in large-diameter western redcedar biomass deductions were not applied in this study due to the complexity of estimating rot distributions. Observations hollow boles in fallen trees and the high variability in internal rot patterns in this study stress the importance of further research using full cross-sections to develop rot deductions for cedar, as proposed by Matsuzaki et al. (2013). Although the high-elevation site exhibited greater tree density and total basal area, the low-elevation plots contained more than three times the total biomass, approximately 825 Mg C ha⁻¹ compared to 260 Mg C ha⁻¹ at the upper elevation site (Table 8). This disparity reflects contrasting forest structures: high-elevation stands were denser and dominated by smaller-diameter trees, while low-elevation plots were characterized by more widely spaced, large-diameter western redcedar and Engelmann spruce. The lower stem density, reduced competition, and more favorable site conditions at low elevations supported greater individual tree biomass and higher overall carbon accumulation per hectare. At the species level, cedar and spruce contributed the majority of biomass across both elevation bands. In high-elevation plots, biomass was more evenly distributed among subplot and macro plot trees. In contrast, low-elevation plots showed a strong concentration of biomass in individuals exceeding 70 cm DBH. Hemlock, Douglas-fir, and subalpine fir 32 contributed modestly to total biomass due to their smaller average sizes and lower overall abundance (Tables 8–9). Table 7. Forest biometrics of tree populations in the low and high-elevation plots of the Longworth study area. Diameter at breast height (DBH) and height (HT) reflect the mean measurements for all tallied trees within species grouping across subplot and macro plot designations. Basal area (BA) and trees-per-area (TPA) values are presented as area-scaled averages, combining subplot and macro plot data. Species Elev. DBH (cm) HT (m) BA (ha) TPA (ha) Subalpine Fir High 41 19 8 58 Hemlock High 59 26 6 17 Spruce High 58 26 51 158 Cedar High 66 21 156 367 Cedar Low 62 25 105 267 Spruce Low 66 38 27 75 Hemlock Low 34 19 15 125 Douglas-fir Low 91 38 11 17 Subalpine Fir Low 36 23 3 25 Table 8. Basal area (m²/ha), trees per hectare (TPH), and biomass (Mg C ha⁻¹) estimates for the high and low-elevation sites in the Longworth study area. Total BA BA TPH BM (Mg C 2/ Elevation (m ha) (ha) (ha) ha⁻¹) High 26 221 600 260 Low 19 161 508 825 33 Table 9. Species-level biomass estimates for high and low-elevation plots at the Longworth Study area. Species BM (Mg C ha⁻¹) % BM Macro % of Total BM Cedar High 122 22 47 Spruce High 60 18 23 Subalpine Fir 77 0 30 Spruce Low 268 82 33 Douglas-fir 26 100 3 Cedar Low 442 78 54 Hemlock Low 89 85 11 2.5.5 Scaled Biomass Timeseries To estimate variability in stand and species-level net primary production through time, we generated annually resolved biomass time series by scaling standardized tree-ring chronologies with empirical biomass measurements (Table 9). This method offers a streamlined alternative to traditional basal area index (BAI) or raw ring-width approaches, enabling efficient reconstruction of above-ground biomass allocation over time. Standardized (STD) chronologies, being dimensionless, mitigate biases related to bole geometry, internal decay, and age-related growth trends, particularly useful for irregularly shaped old growth cedar. Still, as previously cautioned, conservative detrending may obscure long-term trends which introduces a limitation of this approach (Galván et al. 2014; Peters et al. 2015). To generate biomass timeseries, we multiplied each annual value of the species-level chronology by its proportional contribution to total stand biomass, as derived from 2021 field measurements (Figure 14; Table 11). This produced a scaled, annually resolved biomass 34 index that integrates relative species biomass contributions with interannual growth dynamics. Mathematically, this is expressed as: ScaledWeightedBiomasssspecies,year=stdspecies,year×TotalBiomass (Eq 5) /Biomassspecies where: • ScaledWeightedBiomasssspecies, year denotes the scaled biomass for a given species during a specific period • stdspeciesyear is the standardized growth index for the species in that year • Biomassspecies,2021, is the measured biomass of the species in the 2021 baseline • TotalBiomass2021 is the sum of measured biomasses for all species in the 2021 baseline. 35 Figure 8. Scaled biomass series for the seven Longworth Chronologies representing relative annual biomass increments (Mg C ha⁻¹) through the 1901-2021 study period. 2.5.6 Biomass Trends Across Exceptional Growth Periods To evaluate how species and stand-level dynamics respond to diverse climate conditions, we conducted a landscape-scale analysis across the full study period. By identifying exceptional high- and low-growth intervals, this approach provides a framework for interpreting historical variability and projecting potential biomass trajectories under future climate change scenarios. We first generated a master chronology by averaging the seven standardized, dimensionless ring-width chronologies and used the pracma package in R to identify the two most distinct intervals of elevated or suppressed productivity. This allowed us to isolate prominent periods of high and low growth while avoiding biases inherent in a composite biomass series, which would disproportionately reflect high-biomass species. Although this 36 composite chronology masks some species- and site-level biomass dynamics, it provides a standardized lens for interpreting long-term forest-wide growth trends and enables detection of both convergent and divergent species responses. The identified periods were then used as focal intervals for analyzing species- and elevation-specific biomass trends. At the landscape scale, biomass series retained the multimodal variability observed in earlier z-score and moving average analyses, skillfully capturing exceptional growth periods. We identified two high-growth intervals (1936–1946, 1994–2002) and two low-growth intervals (1950–1958, 1984–1994), which were compared against the recent period (2013– 2021) encompassing the Northeast Pacific marine heatwave (“The Blob”) and the 2021 Pacific Northwest Heat Dome anomalies (Figure 8; Table 11). During the high-growth periods, biomass accumulation notably exceeded the 2021 baseline (1,085 Mg C ha⁻¹), reaching 1,593 Mg C ha⁻¹ in 1936–1946 and 1,929 Mg C ha⁻¹ in 1994–2002. Conversely, low-growth periods demonstrated narrower biomass accumulation, ranging from 845 to 865 Mg C ha⁻¹, reflecting broader forest-wide declines in productivity. The recent period, with total accumulation of 855 Mg C ha⁻¹, ranked comparably to the historic low periods, underscoring growth reductions associated with recent climate extremes. Elevation-level analysis highlighted both divergent and convergent shifts in productivity throughout the study interval (Tables 10-11). During the earlier high-growth period (1936–1946), high-elevation species drove biomass accumulation (+95%), exceeding gains at low elevations (+32%) (Table 10). In contrast, during the more recent high-growth interval (1994–2002), low-elevation species dominated biomass gains (+92%), while highelevation species showed more modest increases (+32%). Low-growth intervals revealed suppressed biomass accumulation across all elevations, with reductions ranging from −17% 37 to −26%. These patterns, which were slightly more pronounced at low elevations, highlight greater environmental conditions across the forest system and the influence of the previously documented Hemlock Looper outbreak in the early 1990s. Notably, the asymmetry between gains during high-growth periods and losses during low-growth intervals reflects a pattern of forest resilience and net-positive biomass variability at the Longworth site. Stratification of landscape-level results reveals underlying biogeographic and speciesspecific drivers of biomass dynamics (Tables 11, C2, and C3). During 1936 to 1946, upperelevation species including Cedar High (+103%), Spruce High (+88%), and Subalpine Fir (+88%) exhibited robust growth increases, whereas lower-elevation species displayed more modest responses, ranging from Cedar Low (+38%) to Douglas-fir Low (+2%) (Table C2). Conversely, the 1994 to 2002 interval featured pronounced responses from low-elevation species, notably Hemlock Low (+290%) and Douglas-fir Low (+115%), reflecting rapid post-disturbance recovery following the early 1990s Hemlock Looper outbreak (Table C2; Figure 8). Species responses during low-growth intervals emphasized variability in relation to both environmental and disturbance-related factors. For example, Douglas-fir Low and Hemlock Low demonstrated substantial biomass reductions (−56% and −71%, respectively) during 1984–1994 (Table 12). In contrast, Cedar High and Cedar Low exhibited more moderate declines (−23% and −7%), highlighting the greater of this species (Tables C2 and C3). Spruce High and Subalpine Fir displayed consistent resilience across multiple lowgrowth intervals, suggesting greater stress tolerance at upper elevations or lesser susceptibility to defoliation by Hemlock Looper (Table C2). 38 To evaluate how species-level traits mediate sensitivity to climate variability, we grouped species by moisture tolerance based on assumptions derived from biogeographic distributions and previous studies (Wiley et al. 2018; Klesse et al. 2020). These classifications are generalizations and observed growth responses may also reflect sitespecific which interact with genetic, phenologic, and morphological traits to shape species productivity (Tables 11 and C3). Presumed mesic species (Cedar High, Cedar Low, Hemlock Low) exhibited substantial biomass gains during high-growth periods, with increases of 113 Mg C ha⁻¹ from 1936 to 1946 and 201 Mg C ha⁻¹ from 1994 to 2002, and moderate declines of 41 to 56 Mg C ha⁻¹ during low-growth intervals. Corresponding proportional changes, calculated as percent difference from the 2021 baseline, ranged from 26 to 45 percent during favorable intervals and 9 to 13 percent during adverse ones, indicating high dendroclimatic plasticity (Table C4). Drought-tolerant species such as Subalpine Fir and both Spruce chronologies (High and Low) showed more modest total changes in biomass, with gains ranging from 43 to 61 Mg C ha⁻¹. However, they exhibited relatively larger proportional swings, with percentage increases between 32 and 102% and declines between 25 and 48% (Table C4). In some cases, this was due to smaller baseline biomass values, which can amplify relative changes even when absolute changes are modest. However, species like Douglas-fir showed sharper variability largely due to disturbance history, particularly Hemlock Looper outbreaks that caused abrupt declines and subsequent recovery. These patterns highlight that both species traits and disturbance regimes interact to shape the magnitude and consistency of biomass response across gradients. 39 Table 10. Low and high-elevation biomass accumulation percent gains and declines for the high and low growth periods. Percent gains and losses are calculated from modeled period biomass accumulation (Table 11), and the empirical 2021 biomass baseline estimates are presented in Tables 8 and 9. Elevation 1936-1946 1994-2002 1950-1958 1984-1994 2013-2021 High Elevation 95% 32% -23% -17% -21% Low Elevation 32% 92% -22% -21% -26% Table 11. Relative species-, stand-, and landscape-level biomass (Mg C ha⁻¹) for the seven Longworth chronologies during the two historic high-growth periods, two low-growth periods, and the recent interval. Biomass values represent the sum of annual increments, scaled for unequal period lengths. Species 1936-1946 (High) 1994-2002 (High) 1950-1958 (Low) 1984-1994 2013-2021 (Low) Recent 2021 Baseline Cedar High 248 180 103 95 82 122 Spruce High 113 76 42 58 56 60 Subalpine Fir 145 88 54 63 69 77 Spruce Low 316 453 198 202 223 268 Douglas-fir 27 56 22 12 23 26 Cedar Low 611 731 361 410 327 442 Hemlock 133 346 65 25 77 89 Low Elev. 1087 1585 646 649 650 825 High Elev. 507 344 199 216 206 260 All Species Total 1593 1929 845 865 855 1085 2.6 Biomass Volatility Index In this thesis, we developed a Biomass Volatility Index (BVI) using biomass coefficients and growth rate percentages derived from our period analysis. This index helps to better define the broader implications of climate variability on species-level growth by 40 combining response rates during historical extremes with modeled species-specific biomass volumes for those periods. Because modeled biomass estimates are proportional to species volumes measured in the 2021 baseline, this index offers a metric to gauge possible resilience and vulnerabilities in the forest carbon cycle pool based on compositional volume and sensitivity. While this index was calculated using previously identified extreme periods to showcase the upper bounds of volatility, BVI could also be calculated over extended time periods to assess longer-term trends. The Biomass Volatility Index is calculated by multiplying the variance of weighted biomass values by the mean percent change in biomass, as expressed by: (Eq 6) BVI=𝝈2 x pi where: 𝝈2 represents the variance of biomass values over selected high and low periods, indicating the dispersion of biomass data points from their mean. pi is the mean of the percent changes for the selected periods, representing the average rate of biomass change (please see data found in Table 14). The sign of the mean is important in determining the sign of the BVI. A normalization step was applied to facilitate comparisons across species. Each BVI value was divided by the highest observed BVI score and scaled to 100. This allows for an intuitive, relative index of volatility that can be used to evaluate species' roles within the overall carbon cycle. The sign of the BVI (positive or negative) is also retained, providing an indication of whether a species' growth patterns suggest potential for biomass gain (sink) or loss (source) 41 under climate variability. Depending on the research context, BVI could be adapted to reflect absolute volatility (e.g., using absolute values of percent change) or separated into indices for positive and negative periods. In this study, we retained the signed mean to reflect the balance between gains and declines in species-level productivity. Interestingly, no discernible elevation gradient trends were observed in BVI scores (Table C1). Instead, values were primarily shaped by species-specific proportional biomass contributions and the magnitude of growth rate variability across periods. For example, Cedar Low exhibited the highest BVI (25.52), driven by its dominant contribution to total site biomass and its pronounced sensitivity to both high and low-growth periods. In contrast, species like Subalpine Fir High, Douglas-fir Low, and Spruce High represent smaller shares of total biomass and express smaller BVI values (0.57, 0.01, and 0.34 respectively) (Table 1). Other species including Hemlock Low and Cedar High exhibited moderate BVI scores, reflecting both substantial biomass contributions and measurable growth variability. Notably, Hemlock Low’s moderate BVI score (11.0) resulted from particularly sharp growth reductions during low periods, including the early 1990s Hemlock Looper outbreak, as well as high gains in subsequent years (Table C1; Figure 8). While BVI focuses on biomass volatility, its interpretation can be enriched by crossreferencing with signal-to-noise ratio (SNR) statistics (Table 1). Species with both high BVI and high SNR such as cedar may be considered key climate-sensitive contributors to forest carbon dynamics due to their high responsiveness and dominance in stand composition. Conversely, species with low BVI and low SNR, like Douglas-fir, may play stabilizing roles in the ecosystem, contributing less to dynamic variability but supporting long-term carbon balance. 42 2.7 Growth and Biomass Discussion 2.7.1 Introduction The results presented in this chapter reveal complex, multi-level patterns of forest growth and productivity across species, elevations, and time. By examining trends from individual trees to species-level chronologies and stand-level biomass estimates, we gain a deeper understanding of the drivers of net primary productivity (NPP) in the Inland Temperate Rainforest. Our integrative approach, which combines tree-ring analysis, forest inventory data, and visualization tools, offers a holistic interpretation of how species composition, elevation, and forest structure shape above-ground biomass dynamics in one of the most carbon-rich ecosystems in the world. 2.7.2 Species and elevation-specific variability Descriptive dendrochronological statistics generated from the tree-ring chronologies (Table 1) provide essential insights into data quality, coherence among trees, and variability in species- and elevation-specific growth patterns (Fritts et al. 1990; Cook and Pederson 2011). This analysis addresses our first research question regarding the nature of speciesand elevation-specific variability in growth and climate sensitivity. Mean series correlations (MSC) align closely with prior regional studies, indicating relatively high coherence in species-level chronologies, particularly at upper-elevation sites (Konchalski 2015; Stan and Daniels 2022). Signal-to-noise ratios (SNR) (Table 1) and growth-distribution violin plots (Figure A1) further highlight stable high-elevation growth patterns contrasted with skewed, stress-responsive dynamics at lower elevations. 43 Biological persistence, assessed via autocorrelation, wavelet and spectral analyses (Table 1; Appendix Figures B1-B3), underscores biological and environmental continuity in annual to decadal lagged tree growth patterns. While both chronologies show multi-year periodicities likely related to biological persistence and higher frequency modes of variability including El Nino, higher elevations exhibit longer cycles (approximately 18 to 20 years) associated with Pacific Decadal Oscillation (Gedalof et al. 2002; Littell et al. 2016) 2.7.3 Biomass dynamics and structural contributions Integrating dendrochronological and forest biometric data establishes a robust baseline for biomass allocation and climate-linked fluctuations, directly addressing our second research question on how growth variability translates into biomass accumulation. Structural differences significantly influence biomass dynamics. Low-elevation forests dominated by large-diameter Cedar and Spruce contain a majority of total landscape level biomass despite the higher overall density at upper elevations (Tables 8 and 9; Figures G1G4). These results align with findings from other carbon-rich ecosystems in the Pacific Northwest and underscore the regional significance of low-elevation forests as substantial carbon reservoirs, where the structural dominance of large-diameter, old-growth trees plays a key role in shaping long-term biomass distribution and carbon dynamics (Smithwick et al. 2002; Gray 2003; Matsuzaki et al. 2013; Krankina et al. 2014; DellaSala et al. 2022). Observed differences in biomass response and recovery following disturbance events underscore the importance of compositional and structural factors in mediating landscapescale resilience. Productivity analyses identified exceptional growth intervals following Hemlock Looper outbreaks, particularly in low-elevation Cedar and Spruce stands, where recovery was rapid and often followed by pulses in biomass accumulation (Table 10; 44 Appendix Table C2). These patterns highlight biogeographic divergence between sites, with mesic low-elevation stands exhibiting greater plasticity and more pronounced recovery, while higher-elevation sites, despite receiving greater overall precipitation, showed more muted responses. This contrast may reflect interactions between biological factors such as higher stem density and greater competition, along with microsite influences including increased solar exposure and potentially better-drained soils. Alternatively, it is also possible that Hemlock Looper outbreaks are more constrained at higher elevations due to cooler temperatures and less stressful climate conditions, which may enhance tree resistance or reduce insect outbreaks. Together, these findings illustrate how forest composition, structural development, disturbance history, and microclimatic variation collectively interact to shape long term biomass accumulation and trajectory 2.8 Conclusions and Future Directions This chapter demonstrated that forest productivity in the Inland Temperate Rainforest is shaped by species composition, structural development, and site-specific conditions across elevation gradients. High-biomass species such as cedar and spruce played a dominant role in stand-level carbon dynamics, with low-elevation forests showing greater plasticity in growth responses and rapid recovery following disturbance. At the same time, high-elevation species often exhibited more stable growth trajectories, suggesting a degree of buffering under recent climate stress. By linking dendrochronological data with biomass modeling and introducing the Biomass Volatility Index, this work provides a strong foundation for assessing long-term forest response. While elevation explained some variability, recent climate extremes revealed the overriding influence of local adaptation and structure, calling into question the predictive strength of space-for-time models alone. Future work should integrate microsite hydrology, 45 genetic variability, and improved climate-growth modeling to refine projections. The next chapter builds on these insights by examining climatic drivers of growth, followed by a synthesis of how climate sensitivity translates into changes in productivity and carbon storage. Table 12.Summary Table of growth trends and biomass for the seven Longworth Chronologies. Growth % of Total Spectral Species Trend Biomass Peaks BVI Biomass Impact Cedar High Slight Negative 11.20% 4, 7, 18-20 years Spruce High Marginal Positive 5.50% 16-32 years (1940sStable contributions, lower weight 1980s) Very Low compositionally Subalpine Fir Marginal Positive 7.10% 4-6 years Hemlock Low Marginal Positive High Low Major high-elevation contributor, high volatility Contributions are limited due to tree size and shorter lifespan 8.20% 3, 10-16 years (1980s90s) Moderate Sensitive to environmental stress/insect outbreak Cedar Low Marginal Positive 40.70% 4, 7, 11-12, >20 years Very High Major biomass contributor and significant volatility Spruce Low Slight Negative 24.70% 1-5, 11-12 years High Major biomass contributor and high volatility Douglas-fir Marginal Positive 2.40% Less pronounced Low Stable contributor but low weight compositionally Notes for Table Use: • Growth Trend: Indicates the overall trend from linear analysis (positive, negative, stable). • % of Total Landscape Biomass shows the percentage contribution of each species to the total landscape-level biomass of 1085 Mg C ha-1. • Spectral Peaks Year Ranges: Lists the year ranges identified for significant spectral peaks for each species, providing a concise reference to periodic influences on growth. • BVI (Biomass Volatility Index): Measures the variability and sensitivity of biomass contributions (low, moderate, high, very high). • Biomass Impact: Describes the impact or contribution of the species to the total forest biomass, especially during noted growth periods. 46 Chapter 3: Climate Sensitivity of High and Low-elevation Study Sites 3.1 Introduction This chapter applies multiple statistical techniques to assess the climate sensitivity of high- and low-elevation tree-ring series. The objective is to evaluate how recent and historical climate variability has influenced forest growth dynamics. Building on the biomass and growth analyses from Chapter 2, we expand our focus to examine how biogeographic context shapes species- and elevation-specific responses. Here the hierarchical structure of the study, from individual trees to species, stands, and landscapes, supports a critical evaluation of the climate-centered assumptions behind the space-for-time substitution framework (Klesse et al. 2020). By analyzing plots along an elevation gradient with similar species composition but contrasting thermal and snowpack regimes, the study creates a realworld analog for forecasting the effects of warming on cooler, snow-dominated forests. Understanding these contrasts helps identify ecological and climatic factors that modulate tree growth across spatiotemporal scales. The chapter begins by validating modeled climate data from ClimateNA against local instrumental records. It then characterizes temperature, precipitation, and snowpack differences between elevation bands in the context of regional climate trends and recent extremes in the Longworth region. Random Forest models are used to identify key climatic predictors of growth, which are then transformed via Principal Component Analysis (PCA). These components inform species- and tree-level Linear Mixed Effects Models (LME) that quantify growth response across elevation gradients. This integrative modeling framework help elucidate links between climate variability and tree growth in the Interior Temperate 47 Rainforest and sets the stage for Chapter 4’s exploration of how climate sensitivity affects long-term productivity and carbon storage. 3.2 ClimateNA Data and Validation Due to the limited spatial coverage of meteorological stations and long-term climate records near the Longworth site, we used ClimateNA v7.31 to generate site-specific climate variables for dendroclimatic analysis. ClimateNA is a statistical downscaling and interpolation tool that generates high-resolution climate data (~1 km) by combining historical weather station records with terrain-adjusted estimates derived from elevation, slope, and aspect (Wang et al. 2016). For this study, we extracted climate variables for each plot by inputting the coordinates and elevations of our high- and low-elevation sites, enabling consistent comparison of modeled climate conditions across species and elevations. Published quality assessments have highlighted the model’s strong predictive capacity for temperature and moderate skill for precipitation and snow, especially in topographically complex regions like British Columbia (Krebs et al. 2018; Ye et al. 2022). In a comparative study of 11 field stations, Krebs et al. (2018) found strong agreement between observed and modeled temperature values (r² ≥ 0.9), with weaker correlations for monthly precipitation. These findings are consistent with observed biases in modeled hydroclimatic variables due to sparse ground-station coverage, sub-grid microclimates, and vegetation-mediated snowpack retention (De Frenne et al. 2013; Pepin et al. 2022). To assess model fidelity in our study area, we conducted a correlation analysis comparing modeled ClimateNA data for Longworth with the nearby Barkerville, BC, instrumental record. Barkerville, located approximately 93 km south of Longworth in a 48 climatically similar subzone of the Inland Temperate Rainforest, contains the most complete and long-standing meteorological records in the region (Figure 1). First, we validated ClimateNA temperature data by regressing modeled values against the Barkerville instrumental series for the same location. Results indicated a high degree of correlation between the two sites (r² = 0.86, RMSE = 1.84°C for temperature), while modeled precipitation and snow showed moderate predictive power (r² = 0.81 and 0.77, respectively; Table 13). We then compared modeled data from Longworth’s high-elevation site to the Barkerville instrumental dataset, which showed an even stronger correlation for mean annual temperature (r² = 0.98) but with a slightly higher RMSE (2.28°C; Table 14). Precipitation and snowfall correlations were lower (r² = 0.48 and 0.69, respectively), consistent with previous findings and reflective of regional-scale variability in hydroclimate/ measurement challenges at metrological sites. The modeled 1.6°C difference in mean annual temperature across the 442 m elevation gradient at Longworth corresponds to a lapse rate of approximately 3.6°C/km, within the lower bounds of the observed range for temperate mountain forests (3.5–6.5°C/km) (Körner et al. 2016; Pepin et al. 2022). Following this quality control exercise, we chose to proceed with the modeled Longworth ClimateNA data for several key reasons. First, strong correlations with the Barkerville instrumental record indicated sufficient reliability for use in growth–climate analysis. Second, the tool allowed extraction of precise elevation-specific data tailored to each plot location. Lastly, using the Barkerville station alone would have masked hydroclimatic differences between elevation bands, particularly those related to snowpack that are central to understanding species- and elevation-specific climate sensitivity. Given 49 these advantages, and its widespread use forest ecology and dendroclimatology, ClimateNA offered a robust framework for climate-growth modeling in this study (Wang et al. 2016). Table 13. Linear regression results between the Barkerville instrumental records and Barkerville Climate NA data (top) for the Barkerville Town Site. MAT = Mean annual temperature, MAP = Mean annual precipitation, and PAS = Precipitation as snow. Climate Variable 1901-2021 Normals RMSE R2 P Value MAT 1.65 °C 1.843 (°C) 0.86 1.29E-05 MAP 987.16 mm 26.7 (mm) 0.81 0.000177 PAS 472.38 (mm) 16.68 (mm) 0.77 1.76E-05 Table 14. Linear regression results between the Barkerville instrumental records and Longworth Climate NA high-elevation data. MAT = Mean annual temperature, MAP = Mean annual precipitation, and PAS = Precipitation as snow. Climate Variable 1901-2021 Normals RMSE R2 P Value MAT 1.22 °C 2.28 (°C) 0.98 < 2.2e-16 MAP 1281.37 mm 33.5 (mm) 0.48 < 2.2e-16 PAS 626.31 cm 33.02 (cm) 0.69 < 2.2e-16 50 Figure 9. Scatter plot of regression between Longworth Climate NA and the long-term Barkerville instrumental dataset. 3.3 Site Climate Conditions: Contemporary and Paleoclimate Perspectives Climate differences between the high- and low-elevation plots at Longworth highlight elevation influences variables critical to tree growth and forest structure. Although only 2.4 km apart, the two sites exhibit significant gradients in temperature, precipitation, and growing season moisture that shape species responses through biogeographic mediation of hydroclimate and thermal balances. These environmental contrasts, coupled with the longevity of trees at both sites, provide a valuable opportunity to examine how historic and ongoing climate variability shape forest development and sensitivity across elevation bands. 3.3.1 Temperature Modeled ClimateNA mean annual temperature (MAT) estimates for the upper and lower elevation Longworth sites show a clear thermal gradient, with the upper plot averaging 1.8 °C and the lower site averaging 3.4 °C (Table 15). This 1.6 °C difference is comparable to or greater than the total regional warming observed since the Little Ice Age (1–3 °C) (Shi et 51 al. 2022; Heeter et al. 2023). Seasonal differences are even more pronounced in winter, with the Mean Winter Temperature (MWT) at the low-elevation site recorded at -6.6 °C, compared to -10.1 °C at high elevation. These geographically mediated discrepancies influence phenological and physiological processes including dormancy, bud flush, exposure to heat stress, and freeze thaw cycles (Körner et al. 2016; Liang and Camarero 2018; Huang et al. 2020). During the shoulder seasons, even minor temperature differences can determine whether air temperatures rise above freezing, whether precipitation falls as snow or rain, and whether trees initiate or suspend photosynthetic activity. 3.3.2 Precipitation At the Longworth study area, the upper elevation plot receives significantly more precipitation than the lower site, with mean annual totals of 1,298 mm and 992 mm, respectively (Table 15). This pattern extends into both growing and cool seasons, with the upper site receiving more total precipitation (480 mm vs. 398 mm), more snowfall (641 mm vs. 397 mm), and a lower proportion of summer rainfall (37% vs. 40%), reflecting greater snowpack accumulation, delayed spring melt, and a more rain-dominated regime at lower elevations (Table 15). These hydroclimatic differences influence not only the onset and duration of the growing season, but also the availability of soil moisture during critical phenological windows, with higher growing season precipitation and likely lower evaporative demand at the upper site enhancing late-season moisture availability and buffering against summer drought stress. 52 3.3.3 Summer Heat Moisture Index (SMH) The Summer Heat Moisture Index (SMH), which integrates growing season temperature and precipitation into a measure of drought stress, is significantly lower at the high-elevation site (26.2) than the low-elevation site (37.1) (Table 15). This lower SMH reflects both cooler temperatures and higher moisture availability at higher elevation. Although lower SMH values typically suggest reduced photosynthetic capacity during early summer, they may also signal conditions favorable for growth later in the season, particularly in cool-adapted or shade-tolerant species. At lower elevations, the higher SMH reflects warmer and drier summer conditions, which may impose physiological limits on growth or select for species with drought avoidance or tolerance strategies. 3.3.4 Historical Context and Holocene Legacy These present-day climatic contrasts become especially intriguing when interpreted through the perspective of long-term forest evolution in the late Holocene. Some of the largest western redcedar and Douglas-fir trees sampled at the Longworth site likely germinated during the Roman or Medieval Warm Periods, while cores from subalpine fir, Engelmann spruce, and western hemlock date back to the Little Ice Age. This uneven age structure illustrates the capacity of ITR species to establish and persist under a wide range of climatic regimes. Supporting this interpretation, paleoecological records show that Cedar– Hemlock forests were established in the ITR between 2000 and 6000 years ago, following a shift to wetter conditions after the Holocene Thermal Maximum (Gavin et al. 2003, 2021; Thompson et al. 2022). During this period, gene flow between coastal and interior populations suggests active post-glacial migration through narrow mountain corridors linking the Pacific Coast and northern Rockies (Ruffley et al. 2022). These findings imply that 53 present-day species composition are shaped by both long-distance dispersal and millennia of climate-driven niche expansion. The 120-year climate window captured in this study spans a warming trend of up to 3 °C above Little Ice Age norms and approximately 1 °C above Medieval Warm Period conditions (Sanborn et al. 2006; Shi et al. 2022; Heeter et al. 2023). Yet despite the legacy of adaptability in stands of millennial aged trees, the magnitude and pace of warming observed over the past century may now exceed any interval experienced since at least the Roman Warm Period, with recent decades warming faster than any comparable period in the last two millennia (Heeter et al. 2023). 3.4 Climate Trend Analysis Building upon the site-specific climate conditions outlined in Section 3.3, this section explores long-term trends in temperature, precipitation, and snowfall using modeled data from ClimateNA (Wang et al., 2016). These trends provide temporal context for understanding forest sensitivity to recent climatic change, especially considering the longerterm variability observed in the Holocene record (Gavin et al. 2003, 2021; Sanborn et al., 2006; Thompson et al. 2022). While previous sections emphasized geographically mediated climate differences across elevation bands, the focus here explores temporal change. By examining how temperature and moisture variables have shifted over the last 120 years, we can better evaluate the novelty and magnitude of recent climate change. To evaluate long-term variability, we partitioned modeled climate data for both elevation bands into four multi-decadal intervals, 1901–1940, 1941–1976, 1977–2021, and 2013–2021, that correspond with major Pacific Decadal Oscillation (PDO) regime shifts and 54 broadly align with the exceptional growth periods identified through z-score and breakpoint analyses in Chapter 2 (Tables D1- D6). These intervals also encompass key recent events such as “The Blob” and the 2021 heat dome. We evaluated trends in mean annual temperature (MAT), mean annual precipitation (MAP), and precipitation as snow (PAS) against the long-term mean to characterize shifts in climatic regimes and provide temporal resolution for the climate-growth response modeling that follows. 3.4.1 Temperature Trends Our ClimateNA series reveal a statistically significant increase in MAT across the 1901–2021 interval. At both elevation bands, warming has been nonlinear, with inflection points corresponding to major PDO regime shifts in 1940 and 1976. Since 1901, MAT has increased by approximately 1.4 °C. The warmest decadal interval on record is 2013–2021, which is up to 0.9 °C warmer than the long-term mean and over 1.2 °C warmer than the 1941–1976 cold phase PDO interval (Tables D1–D2). Seasonal breakdowns show the most pronounced warming during the winter months, consistent with broader trends observed across western North America (White et al. 2016; Bush and Lemmen 2019). Minimum winter temperatures in Longworth site have risen by more than 3 °C, increasing freeze-thaw frequency and potentially altering dormancy timing, snowpack duration, and insect outbreak dynamics (Tables D4 and D6; Figure 12). 3.4.2 Precipitation Trends Mean annual precipitation (MAP) has increased slightly at both elevation bands but fluctuations were highly variable over time. The 1941–1976 period for example is marked by anomalously high moisture balances, with values more than 5% above the long-term mean 55 (Tables D1–D2; Figures 10–11). A second high-precipitation interval occurred in the late 1990s through 2012. In contrast, the 1920s, 1980s, and 1990s were marked dry intervals, which may have contributed to the low-growth events and the hemlock looper outbreaks described in Chapter 2 (Figures 10–11). Despite these decadal fluctuations, no sustained trends in MAP were observed over the 120-year interval. This suggests that while the ITR remains among the wettest inland forest ecosystems in the world, future moisture availability may be increasingly governed by changes snowpack dynamics and intra-annual variability rather than changes in annual precipitation totals alone. 3.4.3 Snowfall Trends In contrast to the modest MAP changes, modeled snowfall (PAS) shows a clear and statistically significant decline. At both elevation bands, PAS was highest during the 1940– 1976 interval, with values 11–14% above the long-term mean and declined steadily thereafter (Tables D1–D2; Figures 10 and 11). The most recent decade (2013–2021) shows the lowest PAS values on record at both elevations, approximately 10–13% below average, corresponding to significant winter warming and shifts in the form and timing of cold-season precipitation (Table D1 and D2). These declines could hold important implications for the ITR where snowpack governs the onset and duration of the growing season, provides late-season moisture, and shapes disturbance regimes (e.g., insect survival and soil frost dynamics). The observed PAS trends are consistent with broader findings across western North America, supporting the hypothesis that snow loss, rather than total moisture decline, may be a key determinate of future forest evolution (White et al. 2016; Sanmiguel-Vallelado et al. 2021). 56 Table 15. Climate NA 1901-2021 normals for the Longworth high and low-elevation study sites. All climate measures exhibit significant elevation-related differences, confirmed by ttests (p < 1.50 × 10⁻¹³). MAT MWT MAP MSP PAS % Precip % Precip 1 2 3 4 5 6 Site Lat/Long Elev (m) Slope (℃) (℃) (mm) (mm) (mm) SMH Summer7 Snow8 Low 53.925, 121.451 726 27° 3.4 -6.6 992 398 397 37.1 40 40 High 53.953, 121.479 1168 31° 1.8 -10.1 1298 480 641 26.2 37 49 1 MAT ℃ = Mean Annual Temperature, 2MWT ℃= Mean Winter Temperature, 3MAP (mm)= Mean annual precipitation,4MSP (mm)= Mean Summer Precipitation, 5PAS (mm), 6 SMH = Summer Heat Moisture Index, 7% Precip Summer = Percent of precipitation falling during the summer, 8% Precip Snow = Percent of precipitation falling as snow. 57 Figure 10. High-elevation Climate NA Mean Annual Temperature, Mean Annual Precipitation, and Precipitation as Snow 1901-2021 for the Longworth Site. 58 Figure 11. Low-elevation Climate NA Mean Annual Temperature, Mean Annual Precipitation, and Precipitation as Snow 1901-2021 for the Longworth Site. 59 Figure 12. High and low-elevation Climate NA maximum and minimum temperatures between 1901-2021 for the Longworth Site. 3.5 Random Forest Climate Response Analysis As a precursor to our climate-growth modeling, Random Forest (RF) analysis was used to identify the most influential climate predictors across species and elevational gradients prior to LME runs. RF is particularly well-suited for use as a screening tool due to 60 its flexibility in modeling complex, nonlinear relationships; its tolerance for multicollinearity; and its capacity to handle a large number of predictor variables, including temporally lagged climate effects, without the overfitting risks associated with traditional regression approaches (Breiman 2001; Cutler et al. 2007). In this analysis, RF also functions as a standalone tool for detecting dendroclimatic responses, including multi-year memory effects and non-additive interactions between climate variables that influence growth. We constructed and evaluated RF models using several complementary packages in R. Random Forest modeling was performed using randomForestSRC, employing a minimum depth variable importance framework to assess the relative influence of climate predictors (Ishwaran et al. 2010). Each model consisted of 1,000 RF decision trees, and outof-bag (OOB) error estimates were used for internal model validation. Final variable importance rankings were based on mean minimum depth, a robust and interpretable measure of predictor stability across all trees in the ensemble (Ishwaran et al., 2010). Additional models were trained using the randomForest and ranger packages to assess consistency in variable importance outputs across RF implementations (Wright and Ziegler 2017). Predictor rankings from these models were based on the percentage increase in mean squared error (%IncMSE) and impurity-based importance, respectively. Results were visualized and compared to the minimum depth rankings to evaluate stability and interpretability. RF results revealed elevation- and species-specific differences in the temporal distribution and type of climate predictors most strongly associated with tree growth at the Longworth site. At high elevations, the majority of importance variables occurred with a oneor two-year lag (43% and 40%, respectively), indicating that prior-year climate conditions 61 were frequently retained in the model structure (Table 16). This pattern differed at low elevations, where lag 0 and lag 1 variables were most prominent (23% and 45%, respectively), suggesting stronger associations with current or immediately preceding conditions (Table 16). Species-level distributions further illustrate these trends. Subalpine Fir and Spruce High showed a predominance of lagged variables, while Cedar High exhibited a more even distribution across lag intervals (Table 17). At lower elevations, Cedar and Douglas-fir Low were primarily associated with lag 0 and lag 1 predictors, whereas Spruce Low and Hemlock Low had more balanced distributions across the three lag periods (Table 18). To explore the most frequently associated climate variables influencing tree growth, RFselected variables were grouped into four primary categories: temperature, drought, precipitation, and snow (Table 19). In general, high-elevation species showed higher frequencies of temperature-related variables, while low-elevation species, particularly Cedar Low and Hemlock Low, had a greater proportion of snow and drought predictors (Table 19). Species-specific variable selections further illustrate these patterns. Cedar High was associated with several cool-season climate variables, including Tmin_at, PAS10, and Eref07 (Table 21; Appendix Tables E1, E3). Subalpine Fir High showed high frequencies of spring climate variables such as DD5_sp and RH_sp (Table 21; Appendix Tables E1–E3), while Spruce High was most frequently linked with warm-season variables including DD5_sm, CMD, and DD18_sm (Table 21; Appendix Tables E1–E3). At low elevations, Cedar Low had a relatively high number of snow and drought-related variables selected by the model, including PAS_at, CMI07, and Eref06 (Table 20; Appendix Tables E2 and E3). Douglas-fir Low was most frequently associated with growing season 62 temperature and drought indicators, including DD5_01, Tmax_wt, CMD07, and CMI11 (Table 20; Appendix Tables E1–E3). Spruce Low and Hemlock Low both showed frequent associations with a mix of snow, drought, and humidity-related predictors such as PAS10, CMD_sp, RH_at, and CMI06 (Table 20; Appendix Tables E2 and E3). Snow-related variables were more frequently selected for low-elevation species than for those at higher elevations. Spruce Low, Cedar Low, and Hemlock Low all had several snow predictors in the upper tier of importance rankings, including PAS, NFFD10, and bFFP (Table 19; Table 20; Appendix Table E3). In contrast, Spruce High had relatively fewer snow variables in model results (Table 21; Appendix Table E3). These patterns correspond with elevation-dependent differences in snowpack persistence and variability. The prominence of lagged predictors across many chronologies is consistent with the autocorrelation and spectral results presented earlier, which indicated persistent multi-year influences in growth records (Table 1; Figures B1–B3). RF model outputs reinforce these findings by identifying climate variables from previous years as important predictors across species and elevations. These results demonstrate the utility of Random Forest methods in capturing complex climate–growth relationships and in identifying the timing and types of variables most frequently associated with tree-ring variability at the Longworth site. Table 16. Lag distributions of aggregated species-level predictor variables across low and high-elevation chronologies. Stand Lag 0 Lag 1 Lag 2 High elevation 17% 43% 40% Low elevation 23% 45% 32% 63 Table 17. Percent distributions of variable lags for the Longworth high-elevation chronologies. High elevation Lag0 Lag1 Lag2 Cedar High 30% 40% 30% Subalpine Fir 10% 60% 30% Spruce High 21% 50% 29% Table 18. Percent distributions of variable lags for the Longworth low-elevation chronologies. Low elevation Lag0 Lag1 Lag2 Cedar Low 20% 50% 30% Douglas-fir 17% 58% 25% Spruce Low 33% 33% 33% Hemlock Low 20% 40% 40% Table 19. Percent distribution of species-specific variable groupings irrespective of lags for the Longworth sites. Species Temperature (%) Snow (%) Cedar High 50% 15% 15% 20% Subalpine Fir 50% 10% 35% 5% Spruce High 45% 5% 20% 5% Cedar Low 25% 15% 35% 25% Douglas-fir Low 40% 10% 30% 20% Spruce Low 35% 20% 25% 20% Hemlock Low 25% 20% 30% 25% 64 Drought (%) Precipitation (%) Table 20. Variables selected by RF modeling for low-elevation species at the Longworth Site. Variables highlighted in green represent lag 0, variables highlighted in blue represent lag 1, and variables highlighted in red represent lag 2. Please see Tables 33-35 for a key to the abbreviated variable names. Rank Cedar Low Douglas-fir Low Spruce Low Hemlock Low 1 Eref06 PAS_at PPT_wt DD5_01 2 NFFD10 NFFD01 PAS11 NFFD04 3 Tmin10 DD5_01 CMD06 CMD_at 4 PAS_wt RH CMI06 DD5_sp 5 PAS10 CMI07 CMI11 RH_at 6 eFFP PPT07 PAS05 DD18_05 7 DD18_05 TD MCMT CMD 8 Tmax_wt CMD07 DD5_sm CMI_wt 9 Tmin_sm CMD05 CMD_sm CMD_sp 10 RH02 PPT_sm Eref06 CMI06 11 PPT_wt PAS11 CMI_sm PPT07 12 PAS RH_at Tave_sm FFP 13 TD PAS01 RH02 TD 14 Tave10 PPT_sm DD5_sp CMI_sp 15 bFFP PPT_wt TD Tmin_sm 16 Eref DD18_05 PAS01 SHM 17 Eref_sp MCMT DD_18_sm MCMT 18 Tmax10 PAS CMD_sp PPT_wt 19 CMI_wt PPT01 Eref07 TD 20 DD_0_wt DD5_07 CMD CMD07 65 Table 21. Variables selected by RF modeling for the high-elevation species at the Longworth Site. Variables highlighted in green represent lag 0, variables highlighted in blue represent lag 1, and variables highlighted in red represent lag 2. Please see Tables 33-35 for a key to the abbreviated variable names. Rank Cedar High Subalpine Fir High Spruce High 1 Tmin_at RH_wt DD1040 2 Tmax07 Eref CMD 3 PPT_wt DD1040 DD5_sm 4 PAS10 bFFP MWMT 5 Eref07 Tmin_at DD_18_sm 6 PAS_at DD_0_at Tave_sm 7 PPT01 PPT_sm DD18_sm 8 Tave_at PAS CMD_sm 9 MWMT DD5_01 bFFP 10 CMI01 PPT_wt DD5 11 Tmin_wt RH SHM 12 MCMT Tmax08 DD18 13 PAS_sp PAS_wt RH_sp 14 PPT_sp bFFP PPT_sp 15 DD_18_at NFFD01 RH 16 TD RH_at PAS 17 DD_0_sp CMD_sm Tmax_sm 18 Tave10 Tmin10 Tmax_sm 19 CMD_sp NFFD_at bFFP 20 DD_18_wt NFFD10 DD_0 66 3.6 Principal Component Analysis Principal Component Analysis (PCA) is a dimensionality reduction technique used to simplify complex datasets. In this study, PCA was employed to reduce variable redundancy, improve interpretability, and limit the likelihood of multicollinearity or pseudo-replication in subsequent mixed-effects models. Variables selected through Random Forest analysis were grouped into six categories: temperature, degree days, continentality, frost, snow, and drought, and were entered into group-specific PCA runs (Table 22). Each run generated uncorrelated principal components that retained the main variability of the original dataset. This transformation enabled us to reduce the dimensionality of closely related climate variables without losing ecologically meaningful predictors. To account for delayed responses between climate and tree-ring series, each group was analyzed using 1-year (t–1) and 2-year (t–2) lagged datasets, consistent with patterns identified in the RF analysis. PCA was conducted using the princomp function in base R after normalizing and centering all variables to ensure comparability. We retained only the first principal component for each group, which explained between 45 and 74 percent of the total variance, depending on variable group and elevation gradient (Table 22). Although some excluded variance may be ecologically relevant, limiting each group to PC1 helped prevent overfitting from collinear or marginal predictors. PCA was performed separately for high- and low-elevation sites using site-specific ClimateNA inputs. Variables strongly linked to key ecological processes, such as CMD and PAS, and identified as important in RF, were retained as standalone predictors in the final LME models to allow for direct hypothesis testing. For a key to climate variable abbreviations, see Appendix Tables E1–E3. 67 Table 22. Table of variables included in PCA. PCA was conducted separately for each elevation gradient using Climate NA metrics specific to each site. Values in bold were selected by RF but will be integrated mixed effects modeling as stand-alone variables to evaluate the influences of specific effects. Please see Tables 34-36 for a key describing climate variable abbreviations. Groupings R² highR² Low elevation Elevation Variables Temperature 0.48 0.48 Tave 05-08, Tave SP-SM, tmax_wt, tmin_wt, tmax_sm, tmin_sm Degree Days 0.48 0.52 DD 5-8, DD1040 18, 1040, DD sp, sm, at, DD 05,07, SP-SM Continentality 0.63 0.64 TD, MCMT Frost 0.46 0.45 NFFD10, FFP, bFFP, NFFD_sm, NFFD_at, NFFD_wt Snow 0.71 0.72 PAS, PAS WT, SP 0.73 Eref, Eref_sm, CMD_sm, Eref06, 07, CMD06, 07, CMI06,07, CMI_sm, RH06, RH_sm Drought 0.74 3.7 Linear Mixed Effects Models Forest growth response to climate variability was assessed at multiple spatiotemporal scales using linear mixed-effects models (LME). LMEs are well-suited to tree-ring datasets, which exhibit strong spatial and temporal autocorrelation at the tree, species, stand, and regional levels. These models allow for hierarchical variance partitioning, distinguishing fixed effects (e.g., elevation, species) from random effects representing variation among individual trees within plots (Zuur et al. 2009; Harrison et al. 2018). In each model, specieslevel annual growth series served as response variables, with climate predictors and species included as fixed effects. We tested interactions between climate and species, while random effects were used to account for the nested structure and residual variance at the tree and plot levels (Zuur et al. 2009; Galván et al. 2014). 68 To address known temporal autocorrelation in tree-ring data, we applied an autoregressive correlation structure (corAR1), which assumes correlation decreases with increasing time lag (Schielzeth and Forstmeier 2009; Galván et al. 2014; Harrison et al. 2018). This structure improved the estimation of climate effects while reducing the likelihood of Type I and II errors. Model development followed a top-down selection strategy aimed at balancing model fit and parsimony (Zuur et al. 2009). Two sequential LME runs were conducted for each species at both elevation bands. The first used all candidate predictors and interactions derived from prior RF and PCA screening. The second retained only those predictors found to be statistically significant (p < 0.05). All models were fitted using the nlme package in R. Plot-level variance was low across all species, suggesting that local site differences had relatively minimal influence on growth (Table 23). In contrast, tree-level standard deviations were relatively high, particularly for Douglas-fir and hemlock, indicating strong variation among individual trees (Table 23). This variability likely reflects differences in microsite conditions, competitive status, and physiological response, as well as disturbance legacies such as Hemlock Looper outbreaks. Residual variance remained high across all models, suggesting that a large share of growth variability was not captured by the modeled fixed or random effects (Table 23). Contributing factors that may explain these results include non-linear responses, biological/environmental persistence, and disturbance. Further although PCA helped reduce redundancy among predictors, it may have also removed subtle or interacting climate signals present in the original RF-selected variables that could be both ecologically and temporally meaningful. Random Forest results further support this interpretation, indicating that lagged 69 and non-linear climate responses during both the growing and cool seasons may influence growth in ways that linear models can only partially detect. To more deeply evaluate model performance and partition explained variance, we examined both marginal and conditional R² values. Marginal R² values were generally low, reflecting the limited explanatory power of fixed effects alone. In contrast, conditional R² values were substantially higher, suggesting that random effects, particularly at the tree level, captured a large portion of the variance (Table 23). While model variance remained relatively high, the emergence of interpretable and partially synchronous climate-growth relationships across species-specific models lends support to the ecological relevance of the retained predictors and suggests that the LME framework captured meaningful aspects of species-level climate sensitivity. Shared sensitivities to temperature, drought, continentality, and snow were observed across many species and elevation bands, while degree days, frost, and lagged variables contributed to more localized or species-specific responses (Table 24). The following sections explore these climate predictors in greater detail, highlighting their varying influence across species, elevations, and temporal scales. 70 Table 23. Random effects, AIC, and R² values from LME models for the seven Longworth chronologies. Values show standard deviations for plot, tree, and residual variance, along with marginal and conditional R². Standard Cedar Hemlock Douglas- Spruce Spruce Cedar Subalpine Deviation Low Low Fir Low Low High High Fir High Plot-level 2.14E-01 1.34E-04 1.12E-04 2.31E-01 2.78E-04 1.16E-04 2.91E-05 Tree-level 2.04E-01 4.49E-01 4.19E-01 1.98E-03 3.86E-01 2.80E-01 3.70E-01 Residual 2.14E-01 5.28E-01 3.94E-01 6.73E-01 4.21E-01 3.17E-01 5.07E-01 Model AIC -6,013.65 1,400.74 618.10 1,342.73 163.14 -1,859.55 1,123.07 Marginal R2 0.04 0.02 0.03 0.04 0.03 0.01 0.03 Conditional R2 0.5 0.43 0.54 0.14 0.47 0.44 0.38 3.7.1 Temperature and Energy Balance LME models consistently selected temperature as a dominant driver of growth variability across species and elevations. As identified in the RF analysis, temperature effects were strongest during the growing season (May to August), with positive correlations observed for Douglas-fir, Spruce High, and Cedar High (Table 24). These findings align with the hypothesis that growing season temperature is climatically limiting at upper elevations and latitudes in the ITR. Interestingly, Douglas-fir was the only species to show a statistically significant positive response to maximum summer temperature, consistent with RF results and previous studies that highlighted its ability to span warm, dry environments in western North America (Littell et al. 2008; Griesbauer and Green 2010; Klesse et al. 2020). Negative growth responses to annual, seasonal, and extreme (maximum or minimum) temperature variables suggest that tree growth is constrained by both upper and lower thermal limits, shaped by species- and site-specific physiological thresholds across elevation gradients. Among the seven chronologies, subalpine fir was the only species to exhibit negative correlations with mean annual temperature, suggesting thermal stress (Table 24). 71 Additionally, several lower elevation species showed negative correlations with minimum summer temperatures, again pointing to the influence of latitude at the northern extent of the ITR, regardless of elevation or species. Interestingly, the absence of significant correlations with minimum temperatures at high elevations may be related to the moderating influence of snowpack during the shoulder seasons. It is also possible that by the time snowmelt initiates the transition out of dormancy at upper elevations, temperatures are already within an optimal range for growth. Lagged temperature effects were most pronounced in cedar and spruce across elevation gradients, aligning with Random Forest results that identified prior summer and fall conditions as influential predictors. In the LME models, high-elevation cedar and both spruce chronologies showed positive correlations with growing season temperatures from the previous year (lag 1), consistent with multi-annual physiological effects (Table 24). RF analysis identified similar lag structures, highlighting the capacity of these species to integrate past climate signals into current-year growth. Interestingly, Douglas-fir and low-elevation cedar showed mixed or even inverse responses across current and lagged temperature variables (Table 24). LME models revealed that while current-year temperatures promoted growth, elevated temperatures two years prior were negatively correlated. These findings complement RF’s identification of lagged responses, where physiological carryover effects, resource trade-offs, prolonged snowpack or heat stress recovery could drive physiological processes over the course of multiple growing seasons. This complexity was also reflected in Douglas-fir’s positive correlation with degree days lag 2, suggesting that accumulated heat in prior years may influence delayed growth investment in roots or foliage infrastructure. 72 3.7.2 Winter Temperature and Continentality LME models identified a split response to winter temperature across the elevational gradient, consistent with RF findings. Low-elevation chronologies (Douglas-fir, hemlock low, and cedar low) were positively associated with warmer winter conditions, particularly the mean coldest month temperature (MCMT), while high-elevation species (spruce high, subalpine fir high, and cedar high) showed negative correlations (Table 24). This contrast may reflect elevation-dependent differences in exposure to cold-related stress. At low elevations, positive responses to MCMT suggest that colder winters may limit growth, possibly through increased risk of cold injury during radiative cooling or inversion events. At high elevations, where average MCMT is only slightly lower, negative correlations may reflect constraints related to chilling exposure, dormancy maintenance, soil temperature, or snowpack duration (Table D4 and D6). Continentality (TD), defined as the difference between MWMT and MCMT, was positively associated with growth in all chronologies except for spruce low (Table 24). RF models also emphasized the importance of TD over multi-annual timescales. Together, these results suggest that broad seasonal temperature gradients may influence key phenological rhythms, including budburst and dormancy. However, TD is often strongly correlated with other temperature variables, particularly at high elevations, and may reflect a composite signal of both thermal opportunity and physiological stress. This underscores the importance of interpreting PCA-derived and standalone variables in tandem and recognizing that related predictors may capture overlapping and or distinct ecological processes. 73 3.7.3 Degree Days Degree day variables captured biologically relevant heat accumulation and were among the most consistently selected predictors in RF models. LME results showed that degree days were positively associated with growth in low-elevation cedar but negatively associated in high-elevation spruce and cedar, suggesting elevation-dependent thermal thresholds (Table 24). RF results showed similar patterns and helped clarify the timing of these effects, including differences between mid- and late-summer degree days. Lagged degree-day variables mirrored this trend, with positive correlations in low-elevation cedar and negative associations in high-elevation spruce and cedar. These findings suggest that heat accumulation in prior years may support growth in some species and settings but contribute to cumulative stress in others, depending on elevation and site conditions. 3.7.4 Moisture and Snowpack Dynamics Snowfall-related variables were frequently selected as important in RF models, suggesting that snow plays a notable role in shaping growth dynamics across species and elevations. Because LME results include information on directionality, they offer additional insight into whether snow tends to act as a limiting or facilitating factor. In many cases, snowfall metrics showed negative correlations with growth, particularly in the current growth year. However, there were exceptions, such as high-elevation cedar and low-elevation spruce, which exhibited weak or positive associations in lagged years. While LME models appeared to capture some of these inhibitory effects directly, RF may have helped illustrate potential threshold responses or interactions with temperature and drought (Table 24). The frost eigenvector, which represented variables related to frost-free periods and the number of frost-free days, showed species- and elevation-specific patterns that aligned with 74 RF-derived insights (Table 24). In both methods, frost emerged as a key modulator of growing season length and quality. Negative correlations at low elevations (Douglas-fir, hemlock, and low-elevation spruce) and in subalpine fir suggest that late frosts or early dormancy limit growth, while positive correlations for high-elevation cedar and spruce may reflect the benefits of extended frost-free periods or the role of chilling in bud development (Table 24). RF models emphasized climatic moisture deficit (CMD) and other water balance metrics as important growth drivers, particularly at low elevations. LME models largely supported this pattern, identifying drought-related variables as significant predictors across elevation gradients. Interestingly, positive correlations with the drought eigenvector were observed even in some high-elevation species, such as spruce high and subalpine fir. This may reflect indirect benefits of mild drought conditions that coincide with warmer, drier periods favorable for photosynthesis. In contrast, Douglas-fir showed negative correlations with drought variables, suggesting that its growth is more strongly constrained by growing season moisture and more tightly coupled to local hydrologic conditions, potentially influenced by the offsite samples collected from the steeper, rockier slopes above the lowelevation study area (Table 24). 75 Table 24. LME results for the seven Longworth Chronologies. Significant positive correlations are highlighted in Blue, and significant negative correlations are highlighted in Red. Please refer to Tables 34-36 for a key to the variable abbreviations. Cedar Hemlock Douglas-fir Spruce Spruce Cedar Subalpine Fir Variable Low Low Low Low High High High drought temp dd cont snow tmax_wt tmin_wt tmax_sm tmin_sm mcmt temp_lag1 temp_lag2 dd_lag1 dd_lag2 cont_lag1 cont_lag2 frost_lag1 frost_lag2 snow_lag1 snow_lag2 3.8 Influence of Pacific Decadal Oscillation The Pacific Decadal Oscillation (PDO) is a dominant mode of sea surface temperature (SST) variability in the North Pacific that influences winter temperature, precipitation, and snowpack across British Columbia (Mantua and Hare 2002; Mote 2006; Newman et al. 76 2016). Originally described as the leading empirical orthogonal function of sea surface temperature anomalies between 20°N and 70°N, the PDO has often been viewed as a lowfrequency analog to El Niño Southern Oscillation (ENSO). Recent studies suggest that the PDO is not a discrete or independent climate mode, but rather an integrated pattern shaped by interactions among several large-scale systems, including ENSO, the North Pacific Gyre Oscillation, and Arctic circulation patterns (Newman et al. 2016). Positive-phase PDOs are characterized by warmer sea surface temperatures and elevated sea level pressure off the Pacific coast. These conditions strengthen the Aleutian Low, enhancing the advection of warm, moist air into interior British Columbia and reducing winter snowpack (Bitz and Battisti 1999; Whitfield et al. 2010). In contrast, negative-phase PDOs are associated with cooler SSTs, a weakened Aleutian Low, and enhanced cold-season precipitation and snowpack in the Canadian Rockies and Coast Mountains (Mote 2006; McCabe et al. 2008; Whitfield et al. 2010). Two full PDO cycles are evident in 20th-century instrumental records. Warm phases occurred from 1925 to 1946 and from 1977 to 1998, while cool phases spanned 1890 to 1924 and 1947 to 1976 (Gedalof et al. 2002; MacDonald and Case 2005). Since 1998, PDO phases have become less defined, with greater interannual variability, possibly due to intensified ENSO activity and Arctic Sea ice loss (Fang et al. 2018; Lorenzo et al. 2023; Jahfer et al. 2023). Spectral and wavelet analyses of Longworth chronologies revealed periodicities consistent with PDO (20 to 60 years), with persistent power at decadal scales (Figures 7 to 9). However, parametric linear regressions between annual PDO values and growth were generally non-significant across all chronologies with only three series, Cedar High, Spruce High, and Douglas-fir Low, showing significant Spearman correlations (Table 25). This 77 pattern is consistent with other dendrochronological studies that report intermittent or weak PDO and growth correlations, likely due to the seasonal aggregation and nonlinear nature of the PDO signal (Gedalof et al. 2002; St. George 2014). Given these mixed results, PDO was not included as a standalone predictor in our LME models. Instead, we focused on direct climate variables identified through RF screening and PCA aggregation, such as temperature, drought, snowpack, and continentality that are strongly influenced by PDO. Species- and elevation-specific differences support this interpretation. Cedar High and Spruce High, which had the strongest PDO correlations, also showed sensitivity to winter temperature and snowpack in both RF and LME models (Table 40). Douglas-fir Low’s positive correlation with PDO may reflect a sensitivity to warm-phase conditions, which are often associated with elevated temperatures and reduced moisture availability. This aligns with LME results identifying drought and lagged heat stress as important drivers of growth. Visual interpretation of 10-year moving averages of PDO and growth chronologies revealed lagged and concurrent synchronicities (Figure 13). All chronologies entered the 20th century emerging from a cool-phase PDO (1855 to 1896) and a prolonged growth suppression. A mild positive phase from 1896 to 1909 coincided with recovery in cedar and spruce, followed by a strong warm phase in the 1920s that overlapped with a known Hemlock Looper outbreak. Growth declined sharply during this interval, especially in cedar. Recovery continued through the 1930s and 1940s during a sustained positive-phase PDO, until the 1947 regime shift to a cool phase brought colder winters and deeper snowpack. This transition coincided with one of the most severe growth suppressions in the record, as confirmed by breakpoint, z-score, and biomass analyses (Figures 6, 7 and 13). 78 The 1976 regime shift to a warm-phase PDO temporarily boosted productivity across all species except Cedar High (Figure 13). The 1980s and 1990s were marked by alternating PDO phases and generally moderate tree-ring responses. A brief return to negative PDO conditions in the early 1990s coincided with minor declines, followed by more significant reductions during the 1992–1994 Hemlock Looper outbreak. All species were impacted, with Hemlock and Douglas-fir experiencing the most pronounced downturns (Figure 11). Recovery began in the early 2000s, particularly at low elevations, where modeled biomass accumulation reached its highest levels on record. Since 1998, the PDO has remained predominantly in a negative phase, punctuated by intermittent warm anomalies. Tree productivity increased modestly through 2010 but declined sharply between 2012 and 2021, coinciding with the lowest net primary productivity values modeled in this study (Table 11). This downturn aligns with a suite of climatic stressors, including severe drought, anomalously warm winters, snowpack collapse, and the 2021 heat dome (Tables 26; Figures 10 and 11) (Peterson et al. 2016; Cornwall 2019; Heeter et al. 2023). 79 Table 25. Parametric linear regression and non-parametric Spearman rank correlations between PDO and the seven Longworth chronologies. Correlations in bold are significant at p<0.035. Chronology Linear Regression R² Spearman Correlation Subalpine Fir 0.0378 0.064 Cedar High 0.0965 -0.34 Spruce High 0.1374 -0.346 Spruce Low 0.0406 0.16 Hemlock Low 0.0161 -0.106 Cedar Low 0.0262 -0.184 Douglas-fir Low 0.0044 0.204 80 Figure 13. Smoothed 10-year moving z-score averages of high and low-elevation chronologies paired with PDO. On the PDO curve, areas shaded in red represent negative phase PDO, whereas areas shaded in blue represent positive phases. 81 3.9 Climate Period Analysis: Impacts on Growth and Productivity To assess the influence of climatic variability on biomass accumulation and tree growth, we conducted a targeted climate period analysis based on intervals identified in Chapter 2 (Table 4; Figure 6). To better define growing conditions for these periods, we calculated anomalies for mean annual temperature (MAT), mean annual precipitation (MAP), precipitation as snow (PAS), and the annual Pacific Decadal Oscillation (PDO) using ClimateNA for each elevation gradient (Tables D1–D6; Table 26). Building on period and breakpoint findings from Chapter 2, this analysis defines the climatic backdrop of exceptional growth intervals to assess whether the growing environment and climate sensitivities identified in the RF and LME models contributed to observed variation in growth and biomass accumulation. 3.9.1 T-test comparisons of climate conditions T-tests comparing climate variables across growth periods and against long-term means reveal patterns of climate-mediated influence on tree growth (Table 26). Statistically significant anomalies included elevated MAT during the high-growth intervals of 1984–1994 and 2013–2021, and elevated PAS during the low-growth interval of 1950–1958. The 1950s also exhibited the coldest MAT in the dataset along with above-average snowpack, coinciding with a strongly negative PDO phase (Figure 13). In contrast, warm intervals such as 1936–1946 and 2004–2012 featured above-average MAT and near- to below-average PAS (Table 26). These conditions corresponded with increased tree growth and modeled biomass accumulation (Tables 10 and 11; Figure 8). MAP remained close to average during most intervals supporting earlier findings that precipitation alone may not be a primary constraint on growth at the Longworth site (Tables D1 and D2; Figures 10 and 11). 82 3.9.2 Random Forest classification of high vs. low growth periods To further evaluate the climate drivers of exceptional growth periods, we performed a Random Forest classification using averaged MAT, MAP, PAS, and PDO values from each interval. The model was trained to predict high (1) or low (0) growth outcomes, using an 80/20 data split and out-of-bag validation. The model accurately identified all high-growth periods but correctly classified only 60% of low-growth periods (Table 27). This discrepancy may reflect the more complex nature of low-growth conditions, where both warm-dry and cold-snowy extremes have the potential to limit growth depending on species and sitespecific thresholds. Despite this limitation, RF classification identified MAT and PAS as the strongest predictors, consistent with results from full-period RF and LME models (Tables 20, 21, and 24). MAP and PDO contributed to classification accuracy but were less influential. These findings support the possibility that tree growth at Longworth is shaped by the upper and lower bounds of temperature and snowpack (Körner et al. 2016; Klesse et al. 2020; Gao et al. 2022), with high productivity generally occurring during warm years with moderate snowpack and growth declines arising under both cold-snowy and hot-dry conditions. 3.9.3 Elevation-dependent divergence in response Across the full set of climate intervals, several patterns highlight elevation-dependent buffering and divergence in climate sensitivity. As observed in the RF, LME, and period analysis results, low-elevation chronologies typically exhibited stronger and more variable correlations with hydroclimatic variables such as MAP and PAS, while high-elevation species showed more consistent associations with MAT and snowpack. For instance, during the warm and relatively dry period of 2004–2012, low-elevation species such as Cedar Low 83 exhibited stronger correlations with temperature and snowpack variables, while Cedar High showed more buffered responses (Table F1; Discussed in more detail below). This contrast, consistent with RF and LME findings, highlights elevation-related differences in snowpack exposure and climate sensitivity. Building on this pattern, earlier intervals also revealed elevation-driven divergence in response direction. During the 1925–1933 low-growth period, Cedar High showed a weak positive correlation with MAT while Cedar Low showed a strong negative correlation, highlighting contrasting sensitivities across the elevation gradient (Tables F1, F4). A similar split was observed in Spruce, where low-elevation growth was more negatively impacted by temperature, while high-elevation counterparts responded more positively or neutrally. These findings suggest that elevation may mediate not only the strength but also the direction of climate sensitivity, potentially due to differences in snowpack duration, thermal accumulation, or stress thresholds. 3.9.4 Species-specific climate responses To further evaluate climate sensitivities across species, we calculated Pearson correlation coefficients between standardized ring-width chronologies and climate variables (MAT, MAP, PAS, PDO) during selected high- and low-growth intervals. Although shortwindow correlations reduce statistical power and increase the risk of Type I and II errors, they allowed us to isolate responses during presumed peak sensitivity periods. This approach also helped to circumvent heteroscedasticity and nonlinear dynamics that may obscure interpretations of statistical relationships in full-period models (Zuur et al. 2009). Correlation results generally aligned with trends identified in the RF and LME analyses but offered more detailed insight into species-level thresholds. Supporting previous 84 analyses, MAT tended to show positive correlations during high-growth intervals and negative correlations during low-growth intervals (Tables F1–F4), particularly in highelevation species such as Cedar High and Spruce High. Responses in Subalpine Fir High and many low-elevation species were more variable, with some associations shifting depending on growth interval. Notably, during the warm, low-snow interval of 1982–1986, spruce and cedar at both elevations showed negative correlations with MAT, while Hemlock Low, Douglas-fir Low, and Subalpine Fir High showed positive associations. During the same interval, Spruce Low and Cedar Low also showed elevated sensitivity to PAS, which may indicate that even moderate snowpack can provide a buffering effect. In contrast, more drought-tolerant species such as Douglas-fir and Subalpine Fir appeared less influenced by snowpack reductions. These results are broadly consistent with LME findings that highlight the role of snowpack as a potential modulator of early-season growth for moisture-sensitive species (Tables 24, D1–D2). Similarly, increased MAP sensitivity in some species during warm intervals may reflect seasonal water deficits despite average total precipitation. One of the clearest expressions of this pattern occurred during 2004–2012, when Cedar Low exhibited strong negative correlations with PAS (r = –0.92) and MAP (r = –0.36), alongside strong positive correlations with MAT (r = 0.83) and PDO (r = 0.67) (Table F3). While these patterns suggest that warmer temperatures may have promoted growth, concurrent snowpack loss may have increased exposure to freeze–thaw events or reduced soil moisture availability. In comparison, Cedar High, which experienced greater snow during this period, showed weaker climate correlations and more stable growth (Tables D1, D2, F1, 85 and F2; Figure 5), indicating that elevation-related differences in site conditions may influence the magnitude of climate sensitivity. Table 26. Climate WNA anomalies for historic high, low, and recent periods. The long-term average between 1901-2021 is included for reference. The long-term average between 19012021 is included for reference. Values significantly different from the long-term average are shown in bold. The Climate Impact Index is calculated as a composite index of Mean Annual Temperature (MAT) and Precipitation as Snow (PAS). Variable 1936-1946 (High) 1994-2002 1950-1958 1984-1994 (High) (Low) (Low) 2009-2021 (Recent) Long-term Average Annual PDO 0.52 0.094 -0.753 0.556 0.117 0.045 MAT (°C) 1.59 1.67 0.52 1.86 1.92 1.22 MAP (mm) 1214.27 1319 1299.22 1234.73 1278 1281.37 PAS (mm) 556.09 597.33 708 569.36 571.31 626.31 CII 86 73 1 96 96 37 Table 27. Results of random forest analysis focused on predicting the climatic drivers of high and low-growth periods identified in this study. While none of the correlations were significant, the model displayed moderate overall accuracy, exhibiting 100% accuracy in predicting high-growth periods and 60% in predicting low-growth periods. Variable Importance Score R-Value P-Value PAS High 0.033 0.866 MAT Highest 0.019 0.9226 MAP Moderate 0.045 0.8199 PDO Moderate 0.127 0.5197 86 Table 28. High elevation Climate NA averages for high and low growth periods. Values in bold are significantly different from the period mean as per the results of a t-test p<0.05. Range Group MAT (°C) MAP (mm) PAS (mm) 1912-1917 High 0.47 1173 585 0.02 1936-1946 High 1.61 1214 556 0.52 1944-1952 High 0.87 1214 649 -0.66 1982-1986 High 1.02 1317 626 0.86 1994-2002 High 1.68 1319 597 0.09 2004-2012 High 1.88 1334 585 -0.43 1901-1904 Low 0.95 1261 643 0.14 1925-1933 Low 1.13 1142 536 0.22 1936-1942 Low 1.68 1259 553 0.93 1950-1958 Low 0.53 1299 708 -0.75 1959-1965 Low 1.02 1515 749 -0.53 1984-1994 Low 1.83 1235 569 0.56 1991-2001 Low 1.86 1266 584 0.23 2013-2021 Low 2.49 1286 543 0.86 1901-2021 Long-term 1.21 1282 628 0.04 87 PDO Table 29. Low elevation Climate NA averages for high and low growth periods. Values in bold are significantly different from the period mean as per the results of a t-test p<0.05. Range Period MAT (°C) MAP (mm) PAS (mm) PDO 1912-1917 High 2.65 861 326 0.02 1936-1946 High 3.78 892 302 0.52 1944-1952 High 3.02 888 372 -0.66 1982-1986 High 3.2 980 330 0.86 1994-2002 High 3.86 971 300 0.09 2004-2012 High 4.09 984 298 -0.43 1901-1904 Low 3.13 925.5 356 0.14 1925-1933 Low 3.32 839 291 0.22 1936-1942 Low 3.86 923 293 0.93 1950-1958 Low 2.72 952 402 -0.75 1959-1965 Low 3.21 1125 404 -0.53 1984-1994 Low 4.04 906 296 0.56 1991-2001 Low 4.05 932 293 0.23 2013-2021 Low 4.68 943 278 0.86 1901-2021 Long-term 3.4 942 338 0.04 3.10 Climate Impact Index (CII) To complement the species- and elevation-specific modeling results, we developed a preliminary Climate Impact Index (CII) to characterize climate conditions associated with enhanced tree growth across the Longworth site. Drawing from the Random Forest and regression analyses, the index is designed to reflect the combined influence of warm mean annual temperature (MAT) and low snowfall (PAS), which emerged as the most consistently influential predictors across all chronologies. Mean annual precipitation (MAP) was excluded due to its inconsistent relationship with growth. 88 Prior to index calculation, MAT values were min-max normalized so that higher values corresponded to warmer conditions. PAS values were normalized in reverse by subtracting normalized values from one, such that lower snowpack contributed positively to the index. Both variables were scaled based on the 10th and 90th percentiles of their distributions to reduce the influence of outliers. The CII was then computed as the average of the scaled MAT and PAS values, assuming equal weighting, and rescaled to a 0 to 100 range for interpretability. CII values varied across periods, ranging from 1 to 96 (Table 26). Lower CII values were generally associated with high-growth intervals, suggesting that moderate warmth combined with reduced snowfall may favor biomass accumulation. The 1994–2002 period, which corresponded with the highest modeled productivity across the study, also had the lowest CII value (Table 26). This pattern may indicate that climate conditions during this time were favorable for growth at a landscape scale. In contrast, the cold and snowy 1950– 1958 period, which recorded the lowest productivity, also had one of the lowest CII values. At the opposite end of the spectrum, the most recent interval (2013–2021) exhibited elevated CII values but relatively low modeled productivity (Table 26). This may reflect conditions where warming and declining snowpack begin to exceed optimal thresholds for some species. These results are broadly consistent with patterns identified in the Random Forest and LME models. Although the current version of the CII is simplified and based on annual means, it offers a useful starting point for tracking climate conditions associated with enhanced or diminished growth. Future refinements could incorporate seasonal or monthly climate variables, as well as short-term extremes known to influence physiological stress 89 responses (Klesse et al. 2020; Kharouba and Williams 2024). Additional predictors such as drought and degree-day metrics, which were important in several individual species models, could improve assessments of forest response under future climate scenarios. 3.11 Discussion Climate Sensitivity and Space-for-Time Substitution Framework This chapter explores the climate sensitivity of seven Longworth tree-ring chronologies as a window into how forest stands and individual species in the Inland Temperate Rainforest respond to shifting environmental conditions. Anchored in the spacefor-time substitution framework, our approach draws on present-day biogeographic contrasts to anticipate potential future forest trajectories under continued climate change. Within the study region, recent trends reveal substantial shifts in temperature and precipitation patterns, including a 1.4 °C increase in mean annual temperature, a decline in snowfall, and a rise in extreme weather events such as droughts and heatwaves (Tables D1–D6; OWSC, 2015; Bush and Lemmen, 2019; Heeter et al., 2023). These changes highlight the importance of understanding how forest growth, productivity, and carbon dynamics may evolve under intensifying climate pressures. In response, Chapter 3 addresses core research questions focused on species- and stand-level climate responses across elevation bands, while evaluating the extent to which space-for-time modeling can offer meaningful insight into the future dynamics of high-latitude and high-elevation forests. 3.11.1 Species and Elevation-Specific Climate Sensitivity Dendroclimatic analyses revealed pronounced species- and elevation-specific patterns of climate sensitivity, aligning with prior hypotheses and growth trends discussed in Chapter 2. At upper elevations, species such as Spruce High and Cedar High exhibited strong positive 90 responses to summer temperatures and negative correlations with both persistent snowpack and late-season growing degree day accumulation (Tables 20, 24). These results indicate that while early snowmelt and mid-season warmth are likely beneficial at these sites, prolonged snow cover may shorten the growing season, and excess late-summer heat can reduce productivity. Snowpack influences the timing of dormancy break and growth cessation, acting as a control on both the onset and duration of the growing period (SanmiguelVallelado et al., 2021). Interactions between snowpack and growing season temperatures highlight the importance of both thermal and moisture conditions in shaping growth, where early melt supports productivity, but lingering snow or excessive late-season heat can shorten the growing season, limiting radial growth. Low-elevation species demonstrated more complex and multidimensional climate sensitivities, with a broader set of hydroclimatic variables influencing growth responses. Several chronologies, particularly Douglas-fir Low and Cedar Low, showed strong negative associations with snowpack and cool growing season temperatures in the LME and period analyses (Tables 24 and F4), suggesting that, as in the upper plots, delayed melt may constrain the duration of the growing season. However, unlike their high-elevation counterparts, these low-elevation species also exhibited heightened sensitivity to drought and evaporative demand. Random Forest analyses identified spring and early summer drought indices, including CMD, Eref, and CMI, as important drivers of growth variability in Douglas-fir Low and Cedar Low (Tables 20, 21: see Table E2 for variable key). These effects were most pronounced at lag 0 and lag 1, reflecting both immediate and carryover impacts of moisture stress (Tables 16 and 17). LME models further supported these findings, with negative growth correlations linked to summer CMD and positive associations 91 with relative humidity and spring precipitation in Douglas-fir Low, Cedar Low, and Hemlock Low (Table 24). Collectively, these results suggest that low-elevation trees operate within a narrower seasonal envelope where both thermal and hydrological extremes interact to limit growth. While reduced snowpack may extend the growing season, the gains are often offset by increased exposure to freeze–thaw variability and growing season drought, which together reduce water availability, limiting growth. Yet the apparent benefit of snowpack to regional hydroclimate forcing and tree growth may become more significant in the context of future climate change, due increased exposure to more mild winter temperatures, cool season rain, drought stress and heatwaves, especially in the context of observed trends in declining PAS and increasing evapotranspiration (Heeter et al. 2023; Reyes and Kramer 2023). Notably, the persistence and stability of growth at upper elevations, observed in chapter 2, may be closely associated with prolonged snowpack and cooler MAT (Tables 1 and 5; Figure 5), suggesting these factors mediate the biological memory and adaptive capacity of these forests to multi-year climatic fluctuations. RF models confirmed broader and more lagged climatic sensitivity at high elevations, with predictor selections distributed across multiple prior years (Tables 17 and 21). Conversely, lower-elevation species exhibited stronger associations with current-year climatic conditions, suggesting reduced buffering capacity and more immediate responsiveness in warmer environments (Tables 18 and 24). Interestingly, subalpine fir emerged as an intermediary species, displaying sensitivity patterns that spanned both elevation zones. Although it occupies upper-elevation sites at Longworth, its climate responses more closely resembled those of lower-elevation species, characterized by weaker 92 associations with temperature and stronger sensitivity to hydroclimatic variables such as precipitation and snowpack (Tables 19, 21, and 24). This pattern may reflect its position near the lower margin of its ecological range in the study area and suggests that subalpine fir could serve as an early indicator of shifting climate thresholds in upper-montane forests (Peterson et al. 2002). Our period analysis helps place results from RF and LME analyses within the context of real-world exceptional growth intervals. Both high- and low-growth periods reflected distinct combinations of temperature, snowpack, and moisture availability, reinforcing modeled sensitivities to these variables. Productive intervals such as 1936–1946 and 1994– 2002 were characterized by elevated mean annual temperatures and moderate snowpack (Tables 26, and 28), aligning with the positive influence of summer warmth observed in Cedar High and Spruce High. In contrast, low-growth periods such as 1950–1958 and 2013– 2021 revealed divergent climatic constraints: the former was cold and snow-rich, while the latter was hot and dry, with drought signals especially pronounced in Douglas-fir Low and Cedar Low, which showed strong negative correlations with PAS and MAP and heightened sensitivity to evaporative demand (Table F3). A Random Forest classification trained on climate variables during these intervals identified MAT and PAS as the strongest predictors of growth, supporting the interpretation that forest productivity at Longworth is constrained by both warm and cool season thresholds (Table 27). Moreover, insights from spectral, wavelet, and autocorrelation analyses suggest that historically stable climate–growth relationships may be shifting. The increasingly unstable behavior of the Pacific Decadal Oscillation (PDO) in recent decades has resulted in shorter regime phases and more frequent extreme events such as the 2021 Heat Dome. While 93 earlier positive PDO phases were associated with elevated growth and biomass, this relationship appears to be uncoupling in recent years (Figure 13; Tables D1–D6). Between 2013 and 2021, warm Pacific anomalies altered PDO periodicity and were expressed as nonstationary, annular oscillations, diverging from earlier decadal-scale cycles (Zhang et al., 2023). Although warm intervals have historically enhanced productivity, the declining growth trends during this period may signal that the frequency or intensity of hightemperature events is beginning to exceed physiological adaptation thresholds (Table 15; Figure 14). Supported by autocorrelation statistic and spectral decomposition results, findings from this study indicate that long-lived trees at our Longworth study site integrate multiannual to decadal-scale climate signals into their growth trajectories, and that greater intervals between PDO regime shifts have likely allowed species to acclimate prevailing environmental conditions. As regime shifts accelerate and extreme events intensify, these adaptive windows may narrow, increasing the vulnerability of both high- and low-elevation stands to future climate volatility. 3.11.2 Evaluating the Space-for-Time Substitution Framework Building on conclusions from chapter 2, our results support the partial utility of the space-for-time substitution model in identifying potential future responses of upper latitude and high-elevation forests to climate change. Elevation-driven contrasts in climate sensitivity, especially those observed in Spruce and Cedar across sites, demonstrate how current low-elevation dynamics may offer a glimpse into the future conditions that upperelevation species could face. Species such as Spruce High and Cedar High, which currently exhibit balanced lagged responses and sensitivity to snowpack and temperature, may 94 increasingly resemble their lower-elevation counterparts in responsiveness under future climate change scenarios. Nevertheless, our findings also point to important limitations in the assumptions of the model. In particular, the strong site-specific and species-level differences in climate associations, including the negative responses to late-summer heat at high elevations, highlight the role of local adaptation and phenological constraints in modulating forest growth. Degree-day results from the LME models indicate that even at high elevations, current warming trends may already be inducing thermal stress in sensitive species (Table 24). While future adaptive capacity may modulate response, these findings caution against simple extrapolation of growth trends across elevation gradients without accounting for local site dynamics and species-specific physiological thresholds (Körner et al., 2016; Klesse et al., 2020). Moreover, insights from period, spectral and wavelet analyses suggest that historically stable climate-growth relationships may be shifting. (Figure 19; Tables D1–D6). The 2013–2021 interval represents a distinct phase during which Pacific Ocean warming resulted in non-stationary, annular oscillations as detailed in 3.11.1 (Zhang et al., 2023). Although warm periods have historically coincided with increased growth, the declining trends in tree growth and biomass accumulation during this interval suggest that the frequency and intensity of recent climate extremes may be surpassing physiological adaptation thresholds (Table 15; Figure 14). Supported by our autocorrelation and spectral decomposition analyses, these findings imply that long-lived trees at the Longworth site integrate multi-annual to decadal climate variability into their phenotypic and physiological 95 trajectories, and that longer intervals between regime shifts may provide critical windows for adaptation to prevailing environmental conditions (Figures 7–9). These observations challenge previously observed forest-ocean teleconnection patterns, providing an early indication that the emergence of non-analog conditions and unprecedented fluctuations between extremes will likely challenge insights from the growth dynamics observed in this and similar studies. Better understanding the impacts of ongoing climate change on forest evolution will thus require more extensive monitoring across diverse biogeographic environments and novel models addressing dynamic interactions between natural and anthropogenic forcing on regional climate processes. These themes are expanded in Chapter 4, which directly examines how climate sensitivity translates into spatial and temporal patterns of forest productivity and aboveground carbon dynamics across the Longworth landscape (DellaSala et al. 2022). 96 Chapter 4: Forecasting Forest Futures and Carbon Resilience in the ITR The final chapter of this thesis draws together findings from growth, biomass, and climate sensitivity analyses to explore how forest productivity and carbon storage may shift in British Columbia’s Interior Temperate Rainforest under continued climate change. Bridging results presented in Chapters 2 and 3, it highlights key challenges and insights, examines patterns across elevation and species, and considers the implications for conservation and long-term resilience. The chapter closes with reflections on future research needs, stewardship strategies, and policy directions aimed at sustaining carbon-rich ecosystems in an era of rapid environmental change. 4.1 Synthesis of Climate Sensitivity and Carbon Dynamics in the ITR This study contributes to a growing body of research highlighting the global significance of the Interior Temperate Rainforest (ITR) as a critical carbon sink (DellaSala et al., 2022). Collectively, forests worldwide and highly productive ecosystems like the ITR are increasingly recognized for their outsized role in regulating the global carbon cycle (Pan et al. 2011, 2023; Mo et al. 2023). However, the carbon storage potential of these systems, which underpins IPCC scenarios and nature-based mitigation frameworks, cannot be assumed to remain time stable. Ultimately, the ability of forest ecosystems to function as reliable long-term carbon sinks depends not only on the climate sensitivity of individual species, but also on complex interactions between site conditions, disturbance regimes, phenological thresholds, and human management (Moore and Schindler 2022). Drawing on the results presented in earlier chapters, this synthesis links climate sensitivity at the tree, species, and stand levels to net primary productivity and long-term 97 biomass accumulation. Across species and elevations, dominant trees at the northern limit of the ITR exhibited broadly synchronous responses to variation in temperature, snowpack, and precipitation, though the magnitude and timing of these responses often diverged, with some periods marked by asynchronous growth patterns. Differences in trends were likely shaped by physiological thresholds, phenological triggers, and lagged climate effects, though the underlying mechanisms were not directly tested. Observed patterns may reflect processes such as shifts in the timing of cold hardening, varying sensitivity to drought and winter warming, and species-specific responses to disturbances like western hemlock looper outbreaks. As explored throughout this thesis, a central interpretive framework for understanding these patterns is the space-for-time substitution model. By leveraging elevation and latitude as proxies for future climate conditions, this approach allowed for evaluation of how upperelevation or northern sites might respond to warming by comparing them to contemporary conditions at warmer, lower-elevation and/or lower latitude sites (White et al., 2016; Klesse et al., 2020; Bush and Lemmen, 2019). At Longworth, this framework revealed that many upper-elevation chronologies already show signs of convergence with lower-elevation growth patterns, a shift that coincides with a rise in mean annual temperature at the high-elevation site to just above 1 °C since 1960. This warming appears to be reducing snowpack persistence, altering phenological cues, and weakening the historical buffering capacity of upper montane forests. As a result, high-elevation stands may increasingly resemble low-elevation forests in climate sensitivity. These observations signal that the predictive skill of the space-for-time model in the future might be tempered by non-linear climate–growth relationships, microsite 98 variability/local adaptation, and the potential emergence of non-analog climate conditions in the future (Kharouba and Williams 2024). Still notable patterns emerge in this study that allow us to make projections around productivity and carbon storage in the face of future forest evolution. Within the Longworth site, species such as Engelmann spruce, Douglas-fir, and subalpine fir demonstrated relatively stable or improving growth under recent warm intervals, suggesting that these species may possess greater near-term resilience to warming and drought, particularly in stands where snowpack limitations have historically constrained productivity. In contrast, western redcedar, while historically a significant contributor to biomass accumulation, appears increasingly vulnerable to hydroclimatic stress and disturbance interactions. Observed declines in crown health and rot-related bole failure in redcedar may signal that populations at lower elevations are already nearing or exceeding their climatic thresholds (Figures G4–G6). Nevertheless, the notable recovery and adaptability observed in lowelevation species such as cedar and hemlock following hemlock looper disturbances, along with their high productivity at lower elevation and latitude sites across the Pacific Northwest, highlight their potential to adapt to changing environmental conditions. While this study initially hypothesized that projected declines in productivity at lower elevations due to drought stress might be offset by improved growing conditions in upperelevation forests, results from Longworth suggest that this compensatory dynamic is currently limited. Although upper-elevation stands did not exhibit significant declines in biomass accumulation over the past century, they also did not show meaningful gains that would indicate a strong positive response to recent warming. Instead, long-term productivity trends at remained largely stable over the study window, even as mean annual temperatures 99 increased, snowpack decreased, and precipitation trended slightly upward. These findings suggest that upper-elevation forests may be constrained by local adaptations that have not yet fully capitalized on expanded thermal windows. Although low-elevation stands showed greater volatility, their productivity gains during favorable periods significantly outweighed low-growth intervals, suggesting notable resilience. These findings offer valuable insights for site-level carbon storage projections, highlighting opportunities to sustain and even enhance biomass accumulation in lowelevation forests, particularly those dominated by old-growth cedar. While upper-elevation stands have yet to demonstrate the anticipated rebounds in productivity, their relative stability underscores their complementary role in regional carbon dynamics. As warming continues and hydrologic regimes shift, preserving the structural and functional integrity of existing low-elevation old-growth forests will be essential for maintaining regional carbon balance (Sillett et al., 2025). Recognizing the adaptive capacity of these ecosystems, alongside their legacy carbon pools, can inform proactive management and monitoring strategies to maximize future carbon storage potential. 4.2 Regional Analogs Building on the elevation-based framework at Longworth, the broader latitudinal application of the space-for-time model offers additional insight into how climate sensitivity varies across species and regions. Several keystone species in the Interior Temperate Rainforest, including Douglas-fir, western redcedar, western hemlock, Engelmann spruce, and subalpine fir, occur in both coastal and interior environments. Examining growth patterns 100 across these gradients helps illuminate species-specific adaptations, vulnerabilities, and likely trajectories under continued climate change. In coastal and lower-elevation forests, prolonged exposure to warmer, wetter conditions has shaped distinct patterns of growth, productivity, and structure, offering useful analogs for interior forest trajectories under future climate scenarios. While earlier research has highlighted the resilience of Engelmann spruce and subalpine fir across diverse thermal environments, continued investigation is needed to assess how these species respond under contemporary climate extremes (Peterson et al. 2002; Wilson et al. 2014; Wiley et al. 2018) . As conditions diverge from historical baselines, especially considering recent droughts and heat events, many existing chronologies may only partially reflect the range of stress responses now unfolding, underscoring the need for updated datasets and long-term monitoring. As discussed in Chapter 1, a prominent example of ongoing forest change, which began even before recent periods of climatic variability, is the centuries-long decline in yellow cedar observed across low-elevation coastal forests in British Columbia and Alaska (Beier et al. 2008; Buma et al. 2017; Krapek et al. 2017). This decline highlights how populations established during the Little Ice Age are now trending toward disequilibrium with current environmental conditions. These patterns largely mediated by snowpack, which buffers trees from freeze–thaw damage during warm winter events. Recent studies also suggest that snowpack provides critical winter protection for co-occurring western hemlock, underscoring the importance of cold-season conditions in governing long-term growth responses (Jarvis et al. 2013; Comeau and Daniels 2022). 101 Notably, although the warming conditions that contributed to yellow cedar decline in coastal forests may still be decades from fully manifesting at higher elevations in the interior, coastal systems offer valuable analogs for how snow-dependent species might respond to sustained losses in winter insulation and altered thermal regimes. At Longworth, similar stress mechanisms may already be emerging, particularly at lower elevations manifesting in canopy decline and rot-driven blowdown (Figure G3, G4), while upper-elevation stands are beginning to display growth responses suggestive of growing season heat stress. These earlystage symptoms mirror coastal trends and underscore the value of continued monitoring of forest responses at Longworth and across the ITR. 4.3 Monitoring Adaptation and Future Forest Evolution Emerging shifts in forest responses across the Pacific Northwest underscore the need for robust tools to evaluate the vulnerability and resilience of high-carbon, high conservationvalue stands. The Biomass Volatility Index (BVI), introduced in Chapter 2, quantifies interannual variability in biomass accumulation and offers a practical measure of speciesand site-level sensitivity to climate stress. When paired with long-term field inventories and remote-sensing data, the BVI could help identify early warning signals and support forecasts of future forest function across climatic gradients and time. The Longworth site, located at the northern margin of the Interior Temperate Rainforest, serves as a useful case study for assessing long-term forest responses to climate variability and for evaluating the stability of carbon sinks across the region. Many old-growth trees at this site and in similar forests throughout the Pacific Northwest likely established during periods spanning the Medieval Warm Period to the Little Ice Age, surviving through centuries of environmental change. Their persistence reflects a combination of species-level 102 adaptations and the buffering influence of intact forest structure, which helps maintain microclimatic stability and reduce exposure to environmental extremes. Long-term resilience appears to depend on extended intervals without major disturbance, along with structural and compositional diversity that supports recovery from short-term stress events such as drought or insect outbreaks. As warming accelerates and climate conditions move beyond historical bounds, the future of these forests will hinge not only on persistence but on adaptation. Forest productivity will increasingly depend on the interaction between species traits and the integrity of surrounding ecological conditions. Conservation strategies that protect these conditions, particularly spatial continuity, structural complexity, and sufficient recovery windows will be essential to sustaining the ITR’s role as a deep reservoir of carbon and biodiversity. 4.4 The Great Caribou Rainforest Initiative This study highlights the importance of integrating long-term ecological monitoring with conservation planning to support the resilience of British Columbia’s Interior Temperate Rainforest (ITR). As climate stressors intensify and old-growth systems face increasing pressure from cumulative impacts and land use change, the need for coordinated protection strategies becomes more urgent. In this context, the Great Caribou Rainforest Initiative is proposed as a regional framework for ecosystem-based conservation in the Robson and Cariboo Mountains, modeled in part after the Great Bear Rainforest Agreement (Price et al. 2009; Low and Shaw 2011). 103 The GBR offers a precedent for multi-stakeholder agreements that incorporate ecological science, Indigenous governance, and economic planning into land management decisions. Drawing from this model, the Great Caribou Rainforest Initiative would aim to identify and protect high conservation value areas (HCVAs), restore degraded sites to support long-term carbon storage, and promote forest practices compatible with biodiversity conservation and climate adaptation (Palm et al. 2020; Areendran et al. 2020; DellaSala et al. 2021) (Palm et al., 2020; Areendran et al., 2020; DellaSala et al., 2021). Conservation financing mechanisms, such as carbon offsets, could help support these objectives by creating economic incentives for landholders and tenure holders to pursue longterm conservation outcomes. Similar to the GBR, additional funding streams could include support for community-based stewardship, Indigenous-led tourism and education programs, and sustainable forestry initiatives. While such mechanisms require careful design and oversight to ensure ecological integrity and equitable benefits, they offer a potential pathway to align conservation goals with local livelihoods, while increasing research interests and steward ship of ITR carbon stocks. Recent efforts by Conservation North, the Valhalla Wilderness Society, and others have already laid the groundwork for expanded protections in the ITR. Some progress has been made at the provincial level, including the deferral of harvest in selected old-growth areas (BC Office of Legislative Counsel 2020), yet large portions of ecologically significant forest remain unprotected. The Great Caribou Rainforest Initiative could serve to coordinate existing efforts, provide a structure for data-driven decision-making, and support the implementation of long-term monitoring and adaptive management programs. The findings of this thesis underscore the ecological value of old-growth forests in the 104 ITR and the importance of protecting the conditions that support their long-term function. As forests in this region face increasingly novel climate conditions, conservation strategies that integrate scientific research, regional coordination, and Indigenous leadership will be essential to maintaining their role in carbon storage and biodiversity conservation. 105 References Cited Anchukaitis, K.J., D’Arrigo, R.D., Andreu-Hayles, L., Frank, D., Verstege, A., Curtis, A., Buckley, B.M., Jacoby, G.C., and Cook, E.R. 2013. Tree-Ring-Reconstructed Summer Temperatures from Northwestern North America during the Last Nine Centuries*. J. Clim. 26(10): 3001–3012. doi:10.1175/JCLI-D-11-00139.1. Anderegg, W.R.L., Anderegg, L.D.L., Kerr, K.L., and Trugman, A.T. 2019. Widespread drought-induced tree mortality at dry range edges indicates that climate stress exceeds species’ compensating mechanisms. Glob. Change Biol. 25(11): 3793–3802. doi:10.1111/gcb.14771. Areendran, G., Sahana, M., Raj, K., Kumar, R., Sivadas, A., Kumar, A., Deb, S., and Gupta, V.D. 2020. A systematic review on high conservation value assessment (HCVs): Challenges and framework for future research on conservation strategy. Sci. Total Environ. 709: 135425. doi:10.1016/j.scitotenv.2019.135425. BC Office of Legislative Counsel. 2020. Forest and Range Practices Act SPECIAL TREE PROTECTION REGULATION B.C. Reg. 229/2020. Ministry of Attorney General, Victoria, BC. Beier, C.M., Sink, S.E., Hennon, P.E., D’Amore, D.V., and Juday, G.P. 2008. Twentiethcentury warming and the dendroclimatology of declining yellow-cedar forests in southeastern Alaska. Can. J. For. Res. 38(6): 1319–1334. doi:10.1139/X07-240. Birdsey, R.A., DellaSala, D.A., Walker, W.S., Gorelik, S.R., Rose, G., and Ramírez, C.E. 2023. Assessing carbon stocks and accumulation potential of mature forests and larger trees in U.S. federal lands. Front. For. Glob. Change 5. Frontiers. doi:10.3389/ffgc.2022.1074508. Bitz, C.M., and Battisti, D.S. 1999. Interannual to Decadal Variability in Climate and the Glacier Mass Balance in Washington, Western Canada, and Alaska. J. Clim. 12(11): 3181–3196. doi:10.1175/1520-0442(1999)012<3181:ITDVIC>2.0.CO;2. Boudewyn, Song, Magnussen, and Gillis. 2007. Model-based, volume-to-biomass conversion for forested and vegetated land in Canada - NRCan Open S&T Repository. Available from https://ostrnrcan-dostrncan.canada.ca/entities/publication/29ed30f7-ba82-47508d5c-e043b4cab898 [accessed 30 March 2024]. 106 Breiman, L. 2001. Random Forests. Mach. Learn. 45(1): 5–32. doi:10.1023/A:1010933404324. Buma, B., Hennon, P.E., Harrington, C.A., Popkin, J.R., Krapek, J., Lamb, M.S., Oakes, L.E., Saunders, S., and Zeglen, S. 2017. Emerging climate-driven disturbance processes: widespread mortality associated with snow-to-rain transitions across 10° of latitude and half the range of a climate-threatened conifer. Glob. Change Biol. 23(7): 2903–2914. doi:10.1111/gcb.13555. Bunn, A.G. 2008. A dendrochronology program library in R (dplR). Dendrochronologia 26(2): 115–124. doi:10.1016/j.dendro.2008.01.002. Bunn, A.G. 2010. Statistical and visual crossdating in R using the dplR library. Dendrochronologia 28(4): 251–258. doi:10.1016/j.dendro.2009.12.001. Bunn, A.G., Jansma, E., Korpela, M., Westfall, R.D., and Baldwin, J. 2013. Using simulations and data to evaluate mean sensitivity (ζ) as a useful statistic in dendrochronology. Dendrochronologia 313 250-254 31(4): 250–254. doi:10.1016/j.dendro.2013.01.004. Buras, A. 2017. A comment on the expressed population signal. Dendrochronologia 44: 130– 132. doi:10.1016/j.dendro.2017.03.005. Bush and Lemmen. 2019. Canada’s Changing Climate Report. Government of Canada. Available from https://changingclimate.ca/CCCR2019/. Case, M.J., Johnson, B.G., Bartowitz, K.J., and Hudiburg, T.W. 2021. Forests of the future: Climate change impacts and implications for carbon storage in the Pacific Northwest, USA. For. Ecol. Manag. 482: 118886. doi:10.1016/j.foreco.2020.118886. Casirati, S., Conklin, M.H., and Safeeq, M. 2023. Influence of snowpack on forest water stress in the Sierra Nevada. Front. For. Glob. Change 6. Frontiers. doi:10.3389/ffgc.2023.1181819. Cho, E., McCrary, R.R., and Jacobs, J.M. 2021. Future Changes in Snowpack, Snowmelt, and Runoff Potential Extremes Over North America. Geophys. Res. Lett. 48(22): e2021GL094985. doi:10.1029/2021GL094985. 107 Comeau, V.M., and Daniels, L.D. 2022. Multiple divergent patterns in yellow-cedar growth driven by anthropogenic climate change. Clim. Change 170(3): 22. doi:10.1007/s10584-021-03264-0. Cook, E., Briffa, K., Shiyatov, S., Mazepa, V., and Jones, P.D. 1990. Data Analysis. In Methods of Dendrochronology: Applications in the Environmental Sciences. Edited by E.R. Cook and L.A. Kairiukstis. Springer Netherlands, Dordrecht. pp. 97–162. doi:10.1007/978-94-015-7879-0_3. Cook, E.R., and Pederson, N. 2011. Uncertainty, Emergence, and Statistics in Dendrochronology. In Dendroclimatology: Progress and Prospects. Edited by M.K. Hughes, T.W. Swetnam, and H.F. Diaz. Springer Netherlands, Dordrecht. pp. 77– 112. doi:10.1007/978-1-4020-5725-0_4. Cornwall, W. 2019. A new ‘Blob’ menaces Pacific ecosystems. Science 365(6459): 1233– 1233. doi:10.1126/science.365.6459.1233. Coulthard, B., Smith, D.J., and Meko, D.M. 2016. Is worst-case scenario streamflow drought underestimated in British Columbia? A multi-century perspective for the south coast, derived from tree-rings. J. Hydrol. 534: 205–218. doi:10.1016/j.jhydrol.2015.12.030. Cutler, D.R., Edwards, T.C., Beard, K.H., Cutler, A., Hess, K.T., Gibson, J., and Lawler, J.J. 2007. Random Forests for Classification in Ecology. Ecology 88(11): 2783–2792. doi:10.1890/07-0539.1. D’Arrigo, R., Wilson, R., Liepert, B., and Cherubini, P. 2008. On the ‘Divergence Problem’ in Northern Forests: A review of the tree-ring evidence and possible causes. Glob. Planet. Change 60(3): 289–305. doi:10.1016/j.gloplacha.2007.03.004. De Frenne, P., Rodríguez-Sánchez, F., Coomes, D.A., Baeten, L., Verstraeten, G., Vellend, M., Bernhardt-Römermann, M., Brown, C.D., Brunet, J., Cornelis, J., Decocq, G.M., Dierschke, H., Eriksson, O., Gilliam, F.S., Hédl, R., Heinken, T., Hermy, M., Hommel, P., Jenkins, M.A., Kelly, D.L., Kirby, K.J., Mitchell, F.J.G., Naaf, T., Newman, M., Peterken, G., Petřík, P., Schultz, J., Sonnier, G., Van Calster, H., Waller, D.M., Walther, G.-R., White, P.S., Woods, K.D., Wulf, M., Graae, B.J., and Verheyen, K. 2013. Microclimate moderates plant responses to macroclimate warming. Proc. Natl. Acad. Sci. 110(46): 18561–18565. Proceedings of the National Academy of Sciences. doi:10.1073/pnas.1311190110. 108 DellaSala, D.A., Keith, H., Sheehan, T., Strittholt, J., Mackey, B., Connolly, M., Werner, J.R., and Fredeen, A.L. 2022. Estimating carbon stocks and stock changes in Interior Wetbelt forests of British Columbia, Canada. Ecosphere 13(4): e4020. doi:10.1002/ecs2.4020. DellaSala, D.A., Strittholt, J.R., Degagne, R., Mackey, B., Werner, J.R., Connolly, M., Coxson, D., Couturier, A., and Keith, H. 2021. Red-Listed Ecosystem Status of Interior Wetbelt and Inland Temperate Rainforest of British Columbia, Canada. Land 10(8): 775. Multidisciplinary Digital Publishing Institute. doi:10.3390/land10080775. Fang, K., Cook, E., Guo, Z., Chen, D., Ou, T., and Zhao, Y. 2018. Synchronous multidecadal climate variability of the whole Pacific areas revealed in tree rings since 1567. Environ. Res. Lett. 13(2): 024016. IOP Publishing. doi:10.1088/17489326/aa9f74. Frey, S.J.K., Hadley, A.S., Johnson, S.L., Schulze, M., Jones, J.A., and Betts, M.G. 2016. Spatial models reveal the microclimatic buffering capacity of old-growth forests. Sci. Adv. 2(4): e1501392. American Association for the Advancement of Science. doi:10.1126/sciadv.1501392. Fritts, H.C., Guiot, J., Gordon, G.A., and Schweingruber, F. 1990. Methods of Calibration, Verification, and Reconstruction. In Methods of Dendrochronology: Applications in the Environmental Sciences. Edited by E.R. Cook and L.A. Kairiukstis. Springer Netherlands, Dordrecht. pp. 163–217. doi:10.1007/978-94-015-7879-0_4. Frölicher, T.L., and Laufkötter, C. 2018. Emerging risks from marine heat waves. Nat. Commun. 9(1): 1–4. doi:10.1038/s41467-018-03163-6. Galván, J.D., Camarero, J.J., and Gutiérrez, E. 2014. Seeing the trees for the forest: drivers of individual growth responses to climate in Pinus uncinata mountain forests. J. Ecol. 102(5): 1244–1257. doi:10.1111/1365-2745.12268. Gao, S., Liang, E., Liu, R., Babst, F., Camarero, J.J., Fu, Y.H., Piao, S., Rossi, S., Shen, M., Wang, T., and Peñuelas, J. 2022. An earlier start of the thermal growing season enhances tree growth in cold humid areas but not in dry areas. Nat. Ecol. Evol. 6(4): 397–404. Nature Publishing Group. doi:10.1038/s41559-022-01668-4. 109 Gavin, D.G., Brubaker, L.B., and Lertzman, K.P. 2003. Holocene Fire History of a Coastal Temperate Rain Forest Based on Soil Charcoal Radiocarbon Dates. Ecology 84(1): 186–201. Gavin, D.G., White, A., Sanborn, P.T., and Hebda, R.J. 2021. Deglacial landforms and Holocene vegetation trajectories in the northern interior cedar-hemlock forests of British Columbia. In Untangling the Quaternary Period—A Legacy of Stephen C. Porter. Edited by R.B. Waitt, G.D. Thackray, and A.R. Gillespie. Geological Society of America. p. 0. doi:10.1130/2020.2548(05). Gedalof, Z., Mantua, N.J., and Peterson, D.L. 2002. A multi-century perspective of variability in the Pacific Decadal Oscillation: new insights from tree rings and coral. Geophys. Res. Lett. 29(24): 57-1-57–4. doi:10.1029/2002GL015824. Gedalof, Z., and Smith, D.J. 2001. Dendroclimatic response of mountain hemlock (Tsuga mertensiana) in Pacific North America. Can. J. For. Res. 31(2): 322–332. NRC Research Press. doi:10.1139/x00-169. Gray, A. 2003. Monitoring stand structure in mature coastal Douglas-fir forests: effect of plot size. For. Ecol. Manag. 175(1): 1–16. doi:10.1016/S0378-1127(02)00078-6. Griesbauer, H.P., and Scott Green, D. 2010. Assessing the climatic sensitivity of Douglas-fir at its northern range margins in British Columbia, Canada. Trees 24(2): 375–389. doi:10.1007/s00468-009-0407-z. Griess, V.C., Man, C.D., Leclerc, M.-E., Tansey, J., and Bull, G.Q. 2019. Carbon stocks and timber harvest. Alternative policy approaches for the Great Bear rainforest and their consequences. For. Policy Econ. 103: 147–156. doi:10.1016/j.forpol.2017.06.001. Halofsky, J.E., Andrews-Key, S.A., Edwards, J.E., Johnston, M.H., Nelson, H.W., Peterson, D.L., Schmitt, K.M., Swanston, C.W., and Williamson, T.B. 2018a. Adapting forest management to climate change: The state of science and applications in Canada and the United States. For. Ecol. Manag. 421: 84–97. doi:10.1016/j.foreco.2018.02.037. Halofsky, J.S., Donato, D.C., Franklin, J.F., Halofsky, J.E., Peterson, D.L., and Harvey, B.J. 2018b. The nature of the beast: examining climate adaptation options in forests with stand-replacing fire regimes. Ecosphere 9(3): e02140. doi:10.1002/ecs2.2140. 110 Hammond, W.M., Williams, A.P., Abatzoglou, J.T., Adams, H.D., Klein, T., López, R., Sáenz-Romero, C., Hartmann, H., Breshears, D.D., and Allen, C.D. 2022. Global field observations of tree die-off reveal hotter-drought fingerprint for Earth’s forests. Nat. Commun. 13(1): 1761. Nature Publishing Group. doi:10.1038/s41467-02229289-2. Harrison, X.A., Donaldson, L., Correa-Cano, M.E., Evans, J., Fisher, D.N., Goodwin, C.E.D., Robinson, B.S., Hodgson, D.J., and Inger, R. 2018. A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ 6. doi:10.7717/peerj.4794. Heeter, K.J., Harley, G.L., Abatzoglou, J.T., Anchukaitis, K.J., Cook, E.R., Coulthard, B.L., Dye, L.A., and Homfeld, I.K. 2023. Unprecedented 21st century heat across the Pacific Northwest of North America. Npj Clim. Atmospheric Sci. 6(1): 1–9. Nature Publishing Group. doi:10.1038/s41612-023-00340-3. Hennon, P., D’Amore, D., Wittwer, D., Johnson, A., Schaberg, P., Hawley, G., Beier, C., Sink, S., and Juday, G. 2006. Climate warming, reduced snow, and freezing injury could explain the demise of yellow-cedar in southeast Alaska, USA. Available from https://www.nrs.fs.fed.us/pubs/39685 [accessed 9 January 2020]. Huang, J.-G., Ma, Q., Rossi, S., Biondi, F., Deslauriers, A., Fonti, P., Liang, E., Mäkinen, H., Oberhuber, W., Rathgeber, C.B.K., Tognetti, R., Treml, V., Yang, B., Zhang, J.-L., Antonucci, S., Bergeron, Y., Camarero, J.J., Campelo, F., Čufar, K., Cuny, H.E., De Luis, M., Giovannelli, A., Gričar, J., Gruber, A., Gryc, V., Güney, A., Guo, X., Huang, W., Jyske, T., Kašpar, J., King, G., Krause, C., Lemay, A., Liu, F., Lombardi, F., Martinez del Castillo, E., Morin, H., Nabais, C., Nöjd, P., Peters, R.L., Prislan, P., Saracino, A., Swidrak, I., Vavrčík, H., Vieira, J., Yu, B., Zhang, S., Zeng, Q., Zhang, Y., and Ziaco, E. 2020. Photoperiod and temperature as dominant environmental drivers triggering secondary growth resumption in Northern Hemisphere conifers. Proc. Natl. Acad. Sci. 117(34): 20645–20652. Proceedings of the National Academy of Sciences. doi:10.1073/pnas.2007058117. Hyndman, R.J., and Khandakar, Y. 2008. Automatic Time Series Forecasting: The forecast Package for R. J. Stat. Softw. 27: 1–22. doi:10.18637/jss.v027.i03. IPCC. 2021. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. Available from https://www.ipcc.ch/report/ar6/wg1/. 111 Ishwaran, H., Kogalur, U.B., Gorodeski, E.Z., Minn, A.J., and Lauer, M.S. 2010. HighDimensional Variable Selection for Survival Data. J. Am. Stat. Assoc. 105(489): 205–217. Taylor & Francis. doi:10.1198/jasa.2009.tm08622. Jahfer, S., Ha, K.-J., Franzke, C.L.E., Lee, S.-S., Rodgers, K.B., and Lee, D.-E. 2023. Decoupling of Arctic variability from the North Pacific in a warmer climate. Npj Clim. Atmospheric Sci. 6(1): 1–9. Nature Publishing Group. doi:10.1038/s41612023-00480-6. Jarvis, S.K., Wiles, G.C., Appleton, S.N., D’Arrigo, R.D., and Lawson, D.E. 2013. A Warming-Induced Biome Shift Detected in Tree Growth of Mountain Hemlock [Tsuga mertensiana (Bong.) Carrière] Along the Gulf of Alaska. Arct. Antarct. Alp. Res. 45(2): 211–218. doi:10.1657/1938-4246-45.2.211. Kannenberg, S.A., and Maxwell, J.T. 2022. Disentangling the drivers of non-stationarity in tree growth. Tree Physiol. 42(6): 1128–1130. doi:10.1093/treephys/tpac031. Keith, H., Mackey, B.G., and Lindenmayer, D.B. 2009. Re-evaluation of forest biomass carbon stocks and lessons from the world’s most carbon-dense forests. Proc. Natl. Acad. Sci. 106(28): 11635–11640. Proceedings of the National Academy of Sciences. doi:10.1073/pnas.0901970106. Kharouba, H.M., and Williams, J.L. 2024. Forecasting species’ responses to climate change using space-for-time substitution. Trends Ecol. Evol. doi:10.1016/j.tree.2024.03.009. Klesse, S., DeRose, R.J., Babst, F., Black, B.A., Anderegg, L.D.L., Axelson, J., Ettinger, A., Griesbauer, H., Guiterman, C.H., Harley, G., Harvey, J.E., Lo, Y.-H., Lynch, A.M., O’Connor, C., Restaino, C., Sauchyn, D., Shaw, J.D., Smith, D.J., Wood, L., Villanueva-Díaz, J., and Evans, M.E.K. 2020. Continental-scale tree-ring-based projection of Douglas-fir growth: Testing the limits of space-for-time substitution. Glob. Change Biol. 26(9): 5146–5163. doi:10.1111/gcb.15170. Konchalski, C.S. 2015. Sensitivity of western redcedar to climate and western hemlock looper in British Columbia’s inland temperate rainforest. University of Northern British Columbia. doi:https://doi.org/10.24124/2015/bpgub1025. 112 Körner, C., Basler, D., Hoch, G., Kollas, C., Lenz, A., Randin, C.F., Vitasse, Y., and Zimmermann, N.E. 2016. Where, why and how? Explaining the low-temperature range limits of temperate tree species. J. Ecol. 104(4): 1076–1088. doi:10.1111/13652745.12574. Krankina, O.N., DellaSala, D.A., Leonard, J., and Yatskov, M. 2014. High-Biomass Forests of the Pacific Northwest: Who Manages Them and How Much is Protected? Environ. Manage. 54(1): 112–121. doi:10.1007/s00267-014-0283-1. Krapek, J., Hennon, P.E., D’Amore, D.V., and Buma, B. 2017. Despite available habitat at range edge, yellow-cedar migration is punctuated with a past pulse tied to colder conditions. Divers. Distrib. 23(12): 1381–1392. doi:10.1111/ddi.12630. Krebs, C.J., Henry, J.D., Kenney, A.J., and Hofer, E.J. 2018. Empirical test of the ClimateWNA model for local accuracy in the Kluane Lake area of the southern Yukon, Canada. Clim. Res. 75(2): 111–115. doi:10.3354/cr01511. Laroque, C.P., and Smith, D.J. 1999. Tree-ring analysis of yellow-cedar (Chamaecyparis nootkatensis) on Vancouver Island, British Columbia. Can. J. For. Res. 29(1): 115– 123. doi:10.1139/x98-185. Liang, E., and Camarero, J.J. 2018. Threshold-dependent and non-linear associations between temperature and tree growth at and below the alpine treeline. Trees 32(2): 661–662. doi:10.1007/s00468-017-1651-2. Littell, J.S., Pederson, G.T., Gray, S.T., Tjoelker, M., Hamlet, A.F., and Woodhouse, C.A. 2016. Reconstructions of Columbia River Streamflow from Tree-Ring Chronologies in the Pacific Northwest, USA. JAWRA J. Am. Water Resour. Assoc. 52(5): 1121– 1141. doi:10.1111/1752-1688.12442. Littell, J.S., Peterson, D.L., and Tjoelker, M. 2008. Douglas-Fir Growth in Mountain Ecosystems: Water Limits Tree Growth from Stand to Region. Ecol. Monogr. 78(3): 349–368. 113 Lorenzo, E.D., Xu, T., Zhao, Y., Newman, M., Capotondi, A., Stevenson, S., Amaya, D.J., Anderson, B.T., Ding, R., Furtado, J.C., Joh, Y., Liguori, G., Lou, J., Miller, A.J., Navarra, G., Schneider, N., Vimont, D.J., Wu, S., and Zhang, H. 2023. Modes and Mechanisms of Pacific Decadal-Scale Variability. Annu. Rev. Mar. Sci. 15(Volume 15, 2023): 249–275. Annual Reviews. doi:10.1146/annurev-marine-040422-084555. Low, M., and Shaw, K. 2011. Indigenous Rights and Environmental Governance: Lessons from the Great Bear Rainforest. BC Stud. Br. Columbian Q. (172): 9–33. doi:10.14288/bcs.v0i172.2247. Luyssaert, S., Schulze, E.-D., Börner, A., Knohl, A., Hessenmöller, D., Law, B.E., Ciais, P., and Grace, J. 2008. Old-growth forests as global carbon sinks. Nature 455(7210): 213–215. Nature Publishing Group. doi:10.1038/nature07276. MacDonald, G.M., and Case, R.A. 2005. Variations in the Pacific Decadal Oscillation over the past millennium. Geophys. Res. Lett. 32(8). doi:10.1029/2005GL022478. Mantua, N.J., and Hare, S.R. 2002. The Pacific Decadal Oscillation. J. Oceanogr. 58(1): 35– 44. doi:10.1023/A:1015820616384. Marlier, M.E., Xiao, M., Engel, R., Livneh, B., Abatzoglou, J.T., and Lettenmaier, D.P. 2017. The 2015 drought in Washington State: a harbinger of things to come? Environ. Res. Lett. 12(11): 114008. doi:10.1088/1748-9326/aa8fde. Mass, C., Ovens, D., Christy, J., and Conrick, R. 2024. The Pacific Northwest Heat Wave of 25–30 June 2021: Synoptic/Mesoscale Conditions and Climate Perspective. Weather Forecast. 39(2): 275–291. American Meteorological Society. doi:10.1175/WAF-D23-0154.1. Matsuzaki, E., Sanborn, P., Fredeen, A.L., Shaw, C.H., and Hawkins, C. 2013. Carbon stocks in managed and unmanaged old-growth western redcedar and western hemlock stands of Canada’s inland temperate rainforests. For. Ecol. Manag. 297: 108–119. doi:10.1016/j.foreco.2012.11.042. Maxwell, R.S., and Larsson, L.-A. 2021. Measuring tree-ring widths using the CooRecorder software application. Dendrochronologia 67: 125841. doi:10.1016/j.dendro.2021.125841. 114 McCabe, G.J., Betancourt, J.L., Gray, S.T., Palecki, M.A., and Hidalgo, H.G. 2008. Associations of multi-decadal sea-surface temperature variability with US drought. Quat. Int. 188(1): 31–40. doi:10.1016/j.quaint.2007.07.001. McGrath, S., Luna, E., Wiesenberg, N., and Wiles, G. 2016. EXAMINING THE EFFECTS OF A CHANGING CLIMATE ON ALASKA YELLOW CEDAR: ANALYSES OF RING WIDTH AND BLUE INTENSITY TREE RING CHRONOLOGIES. doi:10.1130/abs/2016AM-280922. Miyamoto, Y., Griesbauer, H.P., and Scott Green, D. 2010. Growth responses of three coexisting conifer species to climate across wide geographic and climate ranges in Yukon and British Columbia. For. Ecol. Manag. 259(3): 514–523. doi:10.1016/j.foreco.2009.11.008. Mo, L., Zohner, C.M., Reich, P.B., Liang, J., de Miguel, S., Nabuurs, G.-J., Renner, S.S., van den Hoogen, J., Araza, A., Herold, M., Mirzagholi, L., Ma, H., Averill, C., Phillips, O.L., Gamarra, J.G.P., Hordijk, I., Routh, D., Abegg, M., Adou Yao, Y.C., Alberti, G., Almeyda Zambrano, A.M., Alvarado, B.V., Alvarez-Dávila, E., Alvarez-Loayza, P., Alves, L.F., Amaral, I., Ammer, C., Antón-Fernández, C., Araujo-Murakami, A., Arroyo, L., Avitabile, V., Aymard, G.A., Baker, T.R., Bałazy, R., Banki, O., Barroso, J.G., Bastian, M.L., Bastin, J.-F., Birigazzi, L., Birnbaum, P., Bitariho, R., Boeckx, P., Bongers, F., Bouriaud, O., Brancalion, P.H.S., Brandl, S., Brearley, F.Q., Brienen, R., Broadbent, E.N., Bruelheide, H., Bussotti, F., Cazzolla Gatti, R., César, R.G., Cesljar, G., Chazdon, R.L., Chen, H.Y.H., Chisholm, C., Cho, H., Cienciala, E., Clark, C., Clark, D., Colletta, G.D., Coomes, D.A., Cornejo Valverde, F., CorralRivas, J.J., Crim, P.M., Cumming, J.R., Dayanandan, S., de Gasper, A.L., Decuyper, M., Derroire, G., DeVries, B., Djordjevic, I., Dolezal, J., Dourdain, A., Engone Obiang, N.L., Enquist, B.J., Eyre, T.J., Fandohan, A.B., Fayle, T.M., Feldpausch, T.R., Ferreira, L.V., Finér, L., Fischer, M., Fletcher, C., Frizzera, L., Gianelle, D., Glick, H.B., Harris, D.J., Hector, A., Hemp, A., Hengeveld, G., Hérault, B., Herbohn, J.L., Hillers, A., Honorio Coronado, E.N., Hui, C., Ibanez, T., Imai, N., Jagodziński, A.M., Jaroszewicz, B., Johannsen, V.K., Joly, C.A., Jucker, T., Jung, I., Karminov, V., Kartawinata, K., Kearsley, E., Kenfack, D., Kennard, D.K., Kepfer-Rojas, S., Keppel, G., Khan, M.L., Killeen, T.J., Kim, H.S., Kitayama, K., Köhl, M., Korjus, H., Kraxner, F., Kucher, D., Laarmann, D., Lang, M., Lu, H., Lukina, N.V., Maitner, B.S., Malhi, Y., Marcon, E., Marimon, B.S., Marimon-Junior, B.H., Marshall, A.R., Martin, E.H., Meave, J.A., Melo-Cruz, O., Mendoza, C., Mendoza-Polo, I., Miscicki, S., Merow, C., Monteagudo Mendoza, A., Moreno, V.S., Mukul, S.A., Mundhenk, P., Nava-Miranda, M.G., Neill, D., Neldner, V.J., Nevenic, R.V., Ngugi, M.R., Niklaus, P.A., Oleksyn, J., Ontikov, P., Ortiz-Malavasi, E., Pan, Y., Paquette, A., ParadaGutierrez, A., Parfenova, E.I., Park, M., Parren, M., Parthasarathy, N., Peri, P.L., Pfautsch, S., Picard, N., Piedade, M.T.F., Piotto, D., Pitman, N.C.A., Poulsen, A.D., Poulsen, J.R., Pretzsch, H., Ramirez Arevalo, F., Restrepo-Correa, Z., Rodeghiero, 115 M., Rolim, S.G., Roopsind, A., Rovero, F., Rutishauser, E., Saikia, P., Salas-Eljatib, C., Saner, P., Schall, P., Schelhaas, M.-J., Schepaschenko, D., Scherer-Lorenzen, M., Schmid, B., Schöngart, J., Searle, E.B., Seben, V., Serra-Diaz, J.M., Sheil, D., Shvidenko, A.Z., Silva-Espejo, J.E., Silveira, M., Singh, J., Sist, P., Slik, F., Sonké, B., Souza, A.F., Stereńczak, K.J., Svenning, J.-C., Svoboda, M., Swanepoel, B., Targhetta, N., Tchebakova, N., ter Steege, H., Thomas, R., Tikhonova, E., Umunay, P.M., Usoltsev, V.A., Valencia, R., Valladares, F., van der Plas, F., Van Do, T., van Nuland, M.E., Vasquez, R.M., Verbeeck, H., Viana, H., Vibrans, A.C., Vieira, S., von Gadow, K., Wang, H.-F., Watson, J.V., Werner, G.D.A., Wiser, S.K., Wittmann, F., Woell, H., Wortel, V., Zagt, R., Zawiła-Niedźwiecki, T., Zhang, C., Zhao, X., Zhou, M., Zhu, Z.-X., Zo-Bi, I.C., Gann, G.D., and Crowther, T.W. 2023. Integrated global assessment of the natural forest carbon potential. Nature 624(7990): 92–101. Nature Publishing Group. doi:10.1038/s41586-023-06723-z. Moore, J.W., and Schindler, D.E. 2022. Getting ahead of climate change for ecological adaptation and resilience. Science 376(6600): 1421–1426. American Association for the Advancement of Science. doi:10.1126/science.abo3608. Mote, P.W. 2006. Climate-Driven Variability and Trends in Mountain Snowpack in Western North America*. J. Clim. 19(23): 6209–6220. doi:10.1175/JCLI3971.1. Newman, M., Alexander, M.A., Ault, T.R., Cobb, K.M., Deser, C., Di Lorenzo, E., Mantua, N.J., Miller, A.J., Minobe, S., Nakamura, H., Schneider, N., Vimont, D.J., Phillips, A.S., Scott, J.D., and Smith, C.A. 2016. The Pacific Decadal Oscillation, Revisited. J. Clim. 29(12): 4399–4427. doi:10.1175/JCLI-D-15-0508.1. OWSC. 2015. OWSC, 2015. Available from http://www.climate.washington.edu/events/2015winter/ [accessed 27 September 2018]. Palm, E.C., Fluker, S., Nesbitt, H.K., Jacob, A.L., and Hebblewhite, M. 2020. The long road to protecting critical habitat for species at risk: The case of southern mountain woodland caribou. Conserv. Sci. Pract. 2(7): e219. doi:10.1111/csp2.219. Pan, C., Li, C., An, A., Deng, G., Lin, J.K., He, J., Li, J.F., Zhu, X., Zhou, G., Shrestha, A.K., Kozak, R., and Wang, G. 2023. Canada’s Green Gold: Unveiling Challenges, Opportunities, and Pathways for Sustainable Forestry Offsets. Forests 14(11): 2206. Multidisciplinary Digital Publishing Institute. doi:10.3390/f14112206. 116 Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L., Shvidenko, A., Lewis, S.L., Canadell, J.G., Ciais, P., Jackson, R.B., Pacala, S.W., McGuire, A.D., Piao, S., Rautiainen, A., Sitch, S., and Hayes, D. 2011. A Large and Persistent Carbon Sink in the World’s Forests. Science 333(6045): 988–993. American Association for the Advancement of Science. doi:10.1126/science.1201609. Pepin, N.C., Arnone, E., Gobiet, A., Haslinger, K., Kotlarski, S., Notarnicola, C., Palazzi, E., Seibert, P., Serafin, S., Schöner, W., Terzago, S., Thornton, J.M., Vuille, M., and Adler, C. 2022. Climate Changes and Their Elevational Patterns in the Mountains of the World. Rev. Geophys. 60(1): e2020RG000730. doi:10.1029/2020RG000730. Peters, R.L., Groenendijk, P., Vlam, M., and Zuidema, P.A. 2015. Detecting long-term growth trends using tree rings: a critical evaluation of methods. Glob. Change Biol. 21(5): 2040–2054. doi:10.1111/gcb.12826. Peterson, D.W., Peterson, D.L., and Ettl, G.J. 2002. Growth responses of subalpine fir to climatic variability in the Pacific Northwest. Can. J. For. Res. 32(9): 1503–1517. NRC Research Press. doi:10.1139/x02-072. Peterson, W., Bond, N., and Robert, M. 2016. The Blob is gone but has morphed into a strongly positive PDO/SST pattern. PICES Press Sidney 24(2): 46-47,50. Price, K., Roburn, A., and MacKinnon, A. 2009. Ecosystem-based management in the Great Bear Rainforest. For. Ecol. Manag. 258(4): 495–503. doi:10.1016/j.foreco.2008.10.010. Reyes, L., and Kramer, M.G. 2023. High-elevation snowpack loss during the 2021 Pacific Northwest heat dome amplified by successive spring heatwaves. Npj Clim. Atmospheric Sci. 6(1): 1–12. Nature Publishing Group. doi:10.1038/s41612-02300521-0. Ruffley, M., Smith, M.L., Espíndola, A., Turck, D.F., Mitchell, N., Carstens, B., Sullivan, J., and Tank, D.C. 2022. Genomic evidence of an ancient inland temperate rainforest in the Pacific Northwest of North America. Mol. Ecol. 31(10): 2985–3001. doi:10.1111/mec.16431. 117 Salzer, M.W., Larson, E.R., Bunn, A.G., and Hughes, M.K. 2014. Changing climate response in near-treeline bristlecone pine with elevation and aspect. Environ. Res. Lett. 9(11): 114007. doi:10.1088/1748-9326/9/11/114007. Sanborn, P., Geertsema, M., Jull, A.J.T., and Hawkes, B. 2006. Soil and sedimentary charcoal evidence for Holocene forest fires in an inland temperate rainforest, eastcentral British Columbia, Canada. The Holocene 16(3): 415–427. SAGE Publications Ltd. doi:10.1191/0959683606hl937rp. Sanmiguel-Vallelado, A., Camarero, J.J., Morán-Tejeda, E., Gazol, A., Colangelo, M., Alonso-González, E., and López-Moreno, J.I. 2021. Snow dynamics influence tree growth by controlling soil temperature in mountain pine forests. Agric. For. Meteorol. 296: 108205. doi:10.1016/j.agrformet.2020.108205. Schielzeth, H., and Forstmeier, W. 2009. Conclusions beyond support: overconfident estimates in mixed models. Behav. Ecol. 20(2): 416–420. doi:10.1093/beheco/arn145. Shi, F., Sun, C., Guion, A., Yin, Q., Zhao, S., Liu, T., and Guo, Z. 2022. Roman Warm Period and Late Antique Little Ice Age in an Earth System Model Large Ensemble. J. Geophys. Res. Atmospheres 127(16): e2021JD035832. doi:10.1029/2021JD035832. Sillett, S.C., Kramer, R.D., Fuentes, B., Carroll, A.L., and Antoine, M.E. 2025. Contributions of exceptional trees to improved management of coast redwood forests in California. For. Ecol. Manag. 594: 122943. doi:10.1016/j.foreco.2025.122943. Smithwick, E.A.H., Harmon, M.E., Remillard, S.M., Acker, S.A., and Franklin, J.F. 2002. Potential Upper Bounds of Carbon Stores in Forests of the Pacific Northwest. Ecol. Appl. 12(5): 1303–1317. doi:10.1890/10510761(2002)012[1303:PUBOCS]2.0.CO;2. St. George, S. 2014. An overview of tree-ring width records across the Northern Hemisphere. Quat. Sci. Rev. 95: 132–150. doi:10.1016/j.quascirev.2014.04.029. Stan, A.B., and Daniels, L.D. 2014. Growth releases across a natural canopy gap-forest gradient in old-growth forests. For. Ecol. Manag. 313: 98–103. doi:10.1016/j.foreco.2013.11.004. 118 Swain, D.L., Singh, D., Touma, D., and Diffenbaugh, N.S. 2020. Attributing Extreme Events to Climate Change: A New Frontier in a Warming World. One Earth 2(6): 522–527. doi:10.1016/j.oneear.2020.05.011. Szeto, K., Zhang, X., White, R.E., and Brimelow, J. 2016. The 2015 Extreme Drought in Western Canada. Bull. Am. Meteorol. Soc. 97(12): S42–S46. doi:10.1175/BAMS-D16-0147.1. Thompson, A.J., Zhu, J., Poulsen, C.J., Tierney, J.E., and Skinner, C.B. 2022. Northern Hemisphere vegetation change drives a Holocene thermal maximum. Sci. Adv. 8(15): eabj6535. American Association for the Advancement of Science. doi:10.1126/sciadv.abj6535. Torrence, C., and Compo, G.P. 1998. A Practical Guide to Wavelet Analysis. Bull. Am. Meteorol. Soc. 79(1): 18. Tseng, Y.-H., Ding, R., and Huang, X. 2017. The warm Blob in the northeast Pacific—the bridge leading to the 2015/16 El Niño. Environ. Res. Lett. 12(5): 054019. doi:10.1088/1748-9326/aa67c3. Ung, C.-H., Bernier, P., and Guo, X.-J. 2008. Canadian national biomass equations: new parameter estimates that include British Columbia data. Can. J. For. Res. 38(5): 1123–1132. NRC Research Press. doi:10.1139/X07-224. Wang, T., Hamann, A., Spittlehouse, D., and Carroll, C. 2016. Locally Downscaled and Spatially Customizable Climate Data for Historical and Future Periods for North America. PLOS ONE 11(6): e0156720. Public Library of Science. doi:10.1371/journal.pone.0156720. White, Wolf, Anslow, and Werner. 2016. Indicators of climate change for British Columbia. British Columbia Ministry of the Enivronment. Available from https://www2.gov.bc.ca/gov/content/environment/research-monitoringreporting/reporting/environmental-reporting-bc/previous-reports-indicators [accessed 21 May 2024]. Whitfield, P.H., Moore, R.D. (Dan), Fleming, S.W., and Zawadzki, A. 2010. Pacific Decadal Oscillation and the Hydroclimatology of Western Canada—Review and Prospects. 119 Can. Water Resour. J. Rev. Can. Ressour. Hydr. 35(1): 1–28. Taylor & Francis. doi:10.4296/cwrj3501001. Wigley, T.M.L., Briffa, K.R., and Jones, P.D. 1984. On the Average Value of Correlated Time Series, with Applications in Dendroclimatology and Hydrometeorology. J. Appl. Meteorol. Climatol. 23(2): 201–213. American Meteorological Society. doi:10.1175/1520-0450(1984)023<0201:OTAVOC>2.0.CO;2. Wiles, G.C., Mennett, C.R., Jarvis, S.K., D’Arrigo, R.D., Wiesenberg, N., and Lawson, Daniel.E. 2012. Tree-ring investigations into changing climatic responses of yellowcedar, Glacier Bay, Alaska. Can. J. For. Res. 42(4): 814–819. doi:10.1139/x2012-028. Wiley, E., Rogers, B.J., Griesbauer, H.P., and Landhäusser, S.M. 2018. Spruce shows greater sensitivity to recent warming than Douglas-fir in central British Columbia. Ecosphere 9(5): e02221. doi:10.1002/ecs2.2221. Wilmking, M., Scharnweber, T., van der Maaten-Theunissen, M., and van der Maaten, E. 2017. Reconciling the community with a concept—The uniformitarian principle in the dendro-sciences. Dendrochronologia 44: 211–214. doi:10.1016/j.dendro.2017.06.005. Wilson, R., Anchukaitis, K., Briffa, K.R., Büntgen, U., Cook, E., D’Arrigo, R., Davi, N., Esper, J., Frank, D., Gunnarson, B., Hegerl, G., Helama, S., Klesse, S., Krusic, P.J., Linderholm, H.W., Myglan, V., Osborn, T.J., Rydval, M., Schneider, L., Schurer, A., Wiles, G., Zhang, P., and Zorita, E. 2016. Last millennium northern hemisphere summer temperatures from tree rings: Part I: The long term context. Quat. Sci. Rev. 134: 1–18. doi:10.1016/j.quascirev.2015.12.005. Wilson, R., Rao, R., Rydval, M., Wood, C., Larsson, L.-Å., and Luckman, B.H. 2014. Blue Intensity for dendroclimatology: The BC blues: A case study from British Columbia, Canada. The Holocene 24(11): 1428–1438. SAGE Publications Ltd. doi:10.1177/0959683614544051. Wright, M.N., and Ziegler, A. 2017. ranger: A Fast Implementation of Random Forests for High Dimensional Data in C++ and R. J. Stat. Softw. 77: 1–17. doi:10.18637/jss.v077.i01. 120 Ye, Z., O’Neill, G.A., and Wang, T. 2022. Climate data for field trials: onsite micro stations versus ClimateNA. Can. J. For. Res. 52(7): 1028–1041. NRC Research Press. doi:10.1139/cjfr-2022-0051. Young, D.J.N., Stevens, J.T., Earles, J.M., Moore, J., Ellis, A., Jirka, A.L., and Latimer, A.M. 2017. Long-term climate and competition explain forest mortality patterns under extreme drought. Ecol. Lett. 20(1): 78–86. doi:10.1111/ele.12711. Zuur, A., Ieno, E.N., Walker, N., Saveliev, A.A., and Smith, G.M. 2009. Mixed Effects Models and Extensions in Ecology with R. Springer-Verlag, New York. Available from https://www.springer.com/gp/book/9780387874579 [accessed 24 September 2019]. 121 Appendix A. Dendrochronological Distribution and Normality Assessments This appendix provides diagnostic summaries of the statistical characteristics of each tree-ring chronology. Table A1 presents results from normality tests, Box-Cox transformations, and post-transformation standard deviations used to ensure suitability for parametric analysis. Figure A1 displays violin plots for each chronology, illustrating the distribution and central tendency of standardized ring-width indices across species and elevation bands. Table 30. This table presents the results of the initial normality test results, Box-Cox transformation lambda values, and standard deviations after transformation, which are used to assess and correct for normality and variance in the annual ring width. Chronology Initial Normality p-value Lambda for Box-Cox Std Dev (Transformed) Douglas-fir 0.04 1.29 0.16 Spruce Low 0.10 0.72 0.17 Hemlock Low <0.001 1.79 0.17 Cedar Low 0.31 0.96 0.27 Cedar High 0.33 0.87 0.20 Spruce High 0.59 1.30 0.17 Subalpine Fir 0.00 1.84 0.20 122 Figure 14. Violin plots for the seven Longworth chronologies represent the distribution of standardized tree-ring growth indices for different species grouped by elevation. The width of each violin indicates the density of data points at different values, with bulges suggesting a more frequent distribution of growth indices and tapers indicating less frequent values. The central dot in each violin marks the median of the distribution. 123 Appendix B. Spectral and Wavelet Analysis of Chronology Periodicity This appendix presents spectral and time-frequency analyses used to assess periodic patterns in tree growth across the Longworth chronologies. Figure B1 shows the results of spectral decomposition, highlighting the relative power of growth cycles between 1–40 years. Figures B2 and B3 display Morlet wavelet analyses for low- and high-elevation chronologies, respectively, revealing how the strength and timing of growth periodicities vary over the 20th and early 21st centuries. These analyses help contextualize decadal-scale variability in relation to climate oscillations such as the PDO. 124 Figure 15. Spectral decomposition of the seven Longworth chronologies. This plot shows the relative power of periodicities between 1-40 years. 125 Figure 16. Morlet wavelet analysis of the four Longworth low-elevation tree-ring chronologies. The color-coded power spectral density illustrates a power gradient across various spectral bands. Areas of significance at the 99% confidence level are outlined in black. The Cone of Influence (COI) is delineated by a black crosshatch pattern. 126 Figure 17. Morlet wavelet analysis of the three Longworth high-elevation tree-ring chronologies. The color-coded power spectral density illustrates a gradient of power across various spectral bands. Areas of significance at the 99% confidence level are outlined in black. The Cone of Influence (COI) is delineated by a black crosshatch pattern. 127 Appendix C. Species- and Stand-Level Biomass Dynamics This appendix provides supplementary biomass calculations used to assess speciesand stand-level productivity at the Longworth site. Table C1 presents species-specific biomass coefficients sourced from Ung et al. (2008), applied to diameter and height measurements. Tables C2 through C4 summarize percent change and absolute shifts in modeled biomass across high- and low-growth periods, disaggregated by species, moisture tolerance, and elevation groupings. These data support the interpretation of climate-driven productivity changes and the development of the Biomass Volatility Index (BVI). Table 31. Biomass coefficients were used to calculate biomass at the Longworth site. Values are sourced directly from Ung et al., 2008. Species DBH Coefficient DBH SE HT Coefficient HT SE Cedar 1.31 0.06 1.53 0.06 Spruce 1.33 0.09 1.68 0.10 Subalpine Fir 1.65 0.05 1.71 0.06 Douglas-fir 1.53 0.06 1.36 0.07 Hemlock 1.93 0.05 1.11 0.04 128 Table 32. Percent biomass gains and declines for the seven Longworth chronologies during the same intervals shown in Table 12. Percent change is calculated from trough to peak (or vice versa), and BVI reflects combined climate sensitivity and proportional variability. 1936-1946 1994-2002 1950-1958 1984-1994 2013-2021 Species (High) (High) (Low) (Low) (Recent) BVI Cedar High 103% 47% -16% -23% -33% 5.47 Spruce High 88% 26% -31% -4% -8% 0.34 Subalpine Fir 88% 15% -29% -18% -11% 0.57 Spruce Low 18% 69% -26% -25% -17% 3.48 Douglas-fir 2% 115% -17% -56% -11% 0.01 Cedar Low 38% 65% -18% -7% -26% 25.52 Hemlock Low 50% 290% -27% -71% -13% 11 Table 33. Net biomass changes (Δ Mg C ha⁻¹) from the 2021 baseline for categorized moisture tolerance and elevational species groupings during high and low growth periods. Δ Mesic Δ High Δ Low Δ Drought-Tolerant (Mg C (Mg C elevation elevation Period ha⁻¹) ha⁻¹) (Mg C ha⁻¹) (Mg C ha⁻¹) 1936-1946 (High) 43 113 82 66 1994-2002 (High) 61 201 28 190 1950-1958 (Low) -29 -41 -20 -45 1984-1994 (Low) -24 -41 -14 -44 2013-2021 (Recent) -15 -56 -17 -44 129 Table 34. Percent change from the 2021 baseline for categorized moisture tolerance and elevational species groupings during high and low growth periods. DroughthighlowTolerant Mesic elevation elevation Period Δ% Δ% Δ% Δ% 1936-1946 (High) 72 26 32 8 1994-2002 (High) 102 45 11 23 1950-1958 (Low) -48 -9 -8 -5 1984-1994 (Low) -40 -9 -5 -5 2013-2021 (Recent) -13 -7 -5 -25 130 Appendix D. Climate Summary Tables by Elevation and Period This appendix presents summary statistics for temperature and precipitation variables across the Longworth study site. Tables D1 and D2 report long-term trends in mean annual temperature (MAT), mean annual precipitation (MAP), and precipitation as snow (PAS) for high- and low-elevation sites. Tables D3 through D6 provide seasonal minimum and maximum temperatures over selected multi-decadal periods. Statistically significant departures from the 1901–2021 mean are indicated in bold (p < 0.05), offering context for climate anomalies linked to observed forest growth trends and the timing of Pacific Decadal Oscillation (PDO) phase shifts. Table 35. Table of high elevation Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), and Precipitation as Snow (PAS) 1901-2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period MAT (°C) MAP (mm) PAS (mm) 1901-1940 0.84 1218.60 617.83 1940-1976 0.91 1344.00 696.44 1977-2021 1.80 1287.07 577.76 2013-2021 2.09 1278.67 561.11 1901-2021 1.22 1281.37 626.31 Table 36. Table of low elevation Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), and Precipitation as Snow (PAS) between 1901-2021.Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period MAT (°C) MAP (mm) PAS (mm) 1901-1940 3.04 362.80 339.13 1940-1976 3.10 395.08 383.92 1977-2021 3.99 404.98 299.00 2013-2021 4.29 373.44 293.00 1901-2021 3.41 388.09 337.53 131 Table 37. Table of low-elevation Maximum Temperatures (TMax) between 1901-2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period Winter Tmax (°C) Spring Tmax (°C) Summer Tmax (°C) Autumn Tmax (°C) 1901-1940 -3.99 10.29 21.56 9.44 1941-1976 -3.8 9.9 21 8.9 1977-2021 -2.63 10.76 21.36 8.75 2013-2021 -2.46 11.19 22.03 8.9 Table 38. Table of low-elevation Minimum Temperatures (TMin) between 1901 and 2021. Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period Winter Tmin (°C) Spring Tmin (°C) Summer Tmin (°C) Autumn Tmin (°C) 1901-1940 -13.53 -3.55 5.73 -1.63 1941-1976 -13.01 -3.34 6.42 -1.17 1977-2021 -10.77 -2.03 7.07 -0.64 2013-2021 -10.43 -2.11 7.64 -0.42 Table 39. Table of high-elevation Maximum Temperatures (TMax) between 19012021.Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period Winter Tmax (°C) Spring Tmax (°C) Summer Tmax (°C) Autumn Tmax (°C) 1901-1940 -5.01 5.95 16.99 6.21 1941-1976 -4.83 5.56 16.44 5.66 1977-2021 -3.65 6.43 16.79 5.51 2013-2021 -3.48 6.85 17.46 5.66 132 Table 40. Table of high-elevation Minimum Temperatures (TMin) between and 9012021.Values denoted in bold are significantly different from the long-term mean at p < 0.05. Period Winter Tmin (°C) Spring Tmin (°C) Summer Tmin (°C) Autumn Tmin (°C) 1901-1940 -13.69 -5.07 4.31 -2.90 1941-1976 -13.17 -4.86 5 -2.43 1977-2021 -10.92 -3.53 5.64 -1.90 2013-2021 -10.57 -3.6 6.21 -1.66 133 Appendix E. Key to Random Forest Climate Variables This appendix defines the climate variables selected by Random Forest (RF) modeling across the seven Longworth chronologies. Tables E1 through E3 organize these variables into temperature (Table E1), drought and humidity (Table E2), and snow, precipitation, and frost (Table E3) categories. Variables include annual and seasonal metrics of thermal accumulation, moisture balance, and snowpack, all derived from ClimateNA outputs. These definitions provide context for interpreting variable importance and specieslevel responses in RF and LME analyses throughout Chapter 3. Table 41. Key to temperature variables selected by RF modeling across the seven Longworth chronologies. Temperature Variables TD Temperature Difference °C between MWMT and MCMT (continentality) Tmax_wt Winter Mean Max Temperature (°C) MCMT Mean Coldest Month Temperature (°C) DD_18_sm Degree-Days above 18°C in summer DD18 Degree-Days above 18°C (cooling degree-days) DD_18_wt Degree-Days below 18°C in winter (heating degree-days) DD_18_at Degree-Days below 18°C in autumn (heating degree-days) DD5 Degree-Days above 5°C (growing degree-days) DD5_sp Spring Degree-Days above 5°C DD_0 Degree-Days below 0°C (chilling degree-days) DD_0_sp Degree-Days below 0°C in spring (chilling degree-days) DD0_at Autumn Degree-Days below 0°C 134 Table 42. Key to the drought and humidity variables selected by RF modeling across the seven Longworth chronologies. Drought and Humidity Variables CMD Hargreaves Climatic Moisture Deficit (mm) CMD_sm Hargreaves Climatic Moisture Deficit for Summer (mm) CMD_at Hargreaves Climatic Moisture Deficit for Autumn (mm) CMD_wt Hargreaves Climatic Moisture Deficit for Winter (mm) RH Mean Annual Relative Humidity (%) RH_wt Winter Relative Humidity (%) RH_at Autumn Relative Humidity (%) RH_sp Spring Relative Humidity (%) CMI Climate Moisture Index (mm) CMI_wt Winter Climate Moisture Index (mm) CMI_sp Spring Climate Moisture Index (mm) CMI06 Climate Moisture Index (mm) June CMI11 Climate Moisture Index (mm)November Eref Hargreaves reference evaporation (mm) Eref06 Hargreaves reference evaporation (mm) June Eref07 Hargreaves reference evaporation (mm) July Eref_sp Hargreaves reference evaporation (mm) Spring SHM Summer heat-moisture index 135 Table 43. Key to the snow, precipitation, and frost variables selected by RF modeling across the seven Longworth chronologies. Snow, Precip, and Frost Variables PAS_at Precipitation as Snow in Autumn (mm) PAS10 Precipitation as Snow in October (mm) PAS General precipitation as snow (mm) PAS_wt Winter Precipitation as Snow (mm) PAS_sp Spring Precipitation as Snow (mm) PPT01 January Precipitation (mm) PPT_wt Winter Precipitation (mm) PPT_sp Spring Precipitation (mm) PPT_sm Summer Precipitation (mm) FFP Frost-free period bFFP The day of the year on which FFP begins NFFD_at Autumn number of frost-free days NFFD_at Autumn number of frost-free days NFFD10 The number of frost-free days in October 136 Appendix F. Climate Period Correlation Tables This appendix contains species- and elevation-specific correlation tables used to evaluate tree-ring responses to climate variables (MAT, MAP, PAS, PDO) during defined high- and low-growth intervals. These tables support the climate period analysis discussed in Chapter 3 and offer detailed insight into species thresholds during key intervals. Table 44. Correlation results for high elevation chronologies during high growth periods and Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. Period Species MAT MAP PAS PDO 1912-1917 Cedar High 0.17 0.23 0.41 0.93 1944-1952 Cedar High 0.74 0.42 -0.11 0.69 1982-1986 Cedar High -0.87 0.45 0.73 -0.89 2004-2012 Cedar High -0.38 -0.12 0.1 -0.39 1936-1946 Cedar High 0.71 0.1 -0.4 0 1994-2002 Cedar High 0.52 -0.25 -0.34 -0.65 1912-1917 Spruce High 0.65 -0.22 -0.59 0.5 1944-1952 Spruce High 0.01 -0.3 -0.48 0.47 1982-1986 Spruce High -0.49 0.01 0.68 -0.76 2004-2012 Spruce High 0.16 0.05 -0.28 -0.42 1936-1946 Spruce High 0.24 -0.4 -0.47 -0.24 1994-2002 Spruce High 0.47 -0.31 -0.32 -0.55 1912-1917 Subalpine Fir 0.39 -0.06 -0.25 0.6 1944-1952 Subalpine Fir 0.55 0.23 -0.65 0.62 1982-1986 Subalpine Fir 0.19 -0.32 -0.05 -0.65 2004-2012 Subalpine Fir 0.01 0.34 -0.28 -0.26 1936-1946 Subalpine Fir 0.33 -0.49 -0.66 -0.06 1994-2002 Subalpine Fir -0.07 -0.16 -0.35 -0.52 137 Table 45. Correlation results for high elevation chronologies during low growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. Period Species MAT MAP PAS PDO 1901-1904 Cedar High -0.56 0.49 0.6 -0.14 1925-1933 Cedar High 0.35 -0.02 0.31 0.13 1950-1958 Cedar High -0.07 0.02 -0.33 0.02 1959-1965 Cedar High -0.3 0.66 0.03 -0.4 1984-1994 Cedar High -0.78 0.52 0.59 -0.49 1991-2001 Cedar High 0.25 -0.03 0.38 -0.14 2013-2021 Cedar High 0.33 0.19 -0.29 0.04 1901-1904 Spruce High -0.29 -0.33 0.43 -0.1 1925-1933 Spruce High 0.12 -0.08 0.05 0.19 1950-1958 Spruce High -0.43 0.27 -0.12 0.24 1959-1965 Spruce High -0.08 0.33 -0.1 -0.28 1984-1994 Spruce High -0.39 0.18 0.27 -0.14 1991-2001 Spruce High 0.22 -0.07 -0.01 0.1 2013-2021 Spruce High 0.36 0.29 -0.28 0.13 1901-1904 Subalpine Fir -0.13 0.23 0.02 -0.31 1925-1933 Subalpine Fir 0.32 -0.05 0.08 0.09 1950-1958 Subalpine Fir -0.35 0.16 -0.3 -0.06 1959-1965 Subalpine Fir -0.29 0.15 -0.12 -0.31 1984-1994 Subalpine Fir -0.24 0.13 0.3 -0.28 1991-2001 Subalpine Fir 0.28 -0.04 -0.01 0.02 2013-2021 Subalpine Fir 0.28 0.11 -0.16 0.01 138 Table 46. Correlation results for low elevation chronologies during high growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. Period Species MAT MAP PAS PDO 1912-1917 Cedar Low -0.07 0.13 0.5 0.66 1944-1952 Cedar Low 0.76 -0.13 -0.34 0.81 1982-1986 Cedar Low -0.89 0.77 0.76 -0.97 2004-2012 Cedar Low 0.83 -0.36 -0.92 0.67 1936-1946 Cedar Low 0.78 0.22 -0.43 0.3 1994-2002 Cedar Low 0.57 -0.49 -0.27 -0.65 1912-1917 Spruce Low 0.26 0.15 0.41 0.23 1944-1952 Spruce Low -0.12 -0.19 -0.18 -0.19 1982-1986 Spruce Low 0.01 -0.34 0.3 0.05 2004-2012 Spruce Low -0.18 0.09 0.22 -0.11 1936-1946 Spruce Low -0.01 0.03 -0.26 0.06 1994-2002 Spruce Low 0.02 -0.1 -0.09 -0.29 1912-1917 Hemlock -0.36 0.29 0.29 -0.01 1944-1952 Hemlock 0.43 0.34 0.25 0.46 1982-1986 Hemlock 0.49 0.02 0.53 0.44 2004-2012 Hemlock -0.1 0.09 -0.36 0.18 1936-1946 Hemlock 0.36 0.36 -0.38 0.23 1994-2002 Hemlock 0.55 -0.01 -0.23 -0.45 1912-1917 Douglas-fir -0.06 0.39 -0.06 0.18 1944-1952 Douglas-fir 0.03 -0.03 -0.41 0.16 1982-1986 Douglas-fir 0.55 -0.12 -0.78 0.5 2004-2012 Douglas-fir 0.26 0.14 -0.42 0.11 1936-1946 Douglas-fir 0.41 -0.18 -0.67 0.22 1994-2002 Douglas-fir -0.03 0 -0.2 0.01 139 Table 47. Correlation results for low elevation chronologies during low growth periods, showing correlations between Mean Annual Temperature (MAT), Mean Annual Precipitation (MAP), Precipitation as Snow (PAS), Pacific Decadal Oscillation (PDO), and standardized ring width indices. Period Species MAT MAP PAS PDO 1901-1904 Cedar Low -0.56 0.54 0.7 -0.14 1925-1933 Cedar Low -0.32 0.67 0.45 -0.53 1950-1958 Cedar Low -0.12 -0.17 -0.09 -0.04 1959-1965 Cedar Low -0.36 0.13 0.25 0.2 1984-1994 Cedar Low -0.12 0.52 0.23 -0.56 1991-2001 Cedar Low -0.07 -0.09 -0.03 -0.25 2013-2021 Cedar Low 0.33 0.17 -0.24 0.2 1901-1904 Spruce Low -0.74 0.36 0.56 -0.24 1925-1933 Spruce Low -0.26 0.18 -0.23 -0.45 1950-1958 Spruce Low -0.69 0.05 0.17 -0.11 1959-1965 Spruce Low 0.14 -0.38 -0.13 0.06 1984-1994 Spruce Low -0.19 0.26 0.17 -0.5 1991-2001 Spruce Low 0.23 -0.08 0.09 0.07 2013-2021 Spruce Low 0.4 0.11 -0.29 -0.14 1901-1904 Hemlock -0.45 0.72 0.7 -0.25 1925-1933 Hemlock -0.26 0.16 0.09 -0.33 1950-1958 Hemlock -0.55 -0.17 -0.04 -0.16 1959-1965 Hemlock -0.64 0.44 0.01 -0.22 1984-1994 Hemlock -0.17 0.15 -0.22 -0.46 1991-2001 Hemlock 0.29 -0.06 0.17 0.1 2013-2021 Hemlock 0.27 0.06 -0.13 -0.07 1901-1904 Douglas-fir -0.84 -0.04 0.39 0.08 1925-1933 Douglas-fir -0.6 0.38 0.15 -0.45 1950-1958 Douglas-fir -0.38 -0.14 -0.45 -0.16 1959-1965 Douglas-fir -0.12 -0.29 0.1 0.05 1984-1994 Douglas-fir -0.27 0.5 -0.13 -0.49 1991-2001 Douglas-fir 0.34 -0.24 0.21 0.13 2013-2021 Douglas-fir 0.41 0.05 -0.19 -0.22 140 Appendix G. Longworth Photos This appendix presents representative photographs of high- and low-elevation forest stands in the Longworth study area. Additional images document structural damage and tree defects, including canopy dieback and internal stem decay. These features may reflect longterm physiological stress and emerging responses to changing climate conditions, particularly at lower elevations. Figure 18. Field sampling at the Low Elevation Site. 141 Figure 19. Low Elevation site in winter. Figure 20. Photograph of upper elevation site, showcasing smaller diameter, more densely spaced trees. 142 Figure 21. Image of canopy decline and dieback directly upslope from the Low Elevation site. 143 Figure 22. Photograph taken just below the High Elevation site showcasing rot-driven blowdown damage. 144 Figure 23. Photograph taken just below the High Elevation site showcasing rot-driven blowdown damage. 145