Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Relationships between climate and annual radial growth in three coniferous species in interior British… Lo, Yueh-Hsin 2009

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

Item Metadata

Download

Media
24-ubc_2009_spring_lo_yuehhsin.pdf [ 17.62MB ]
Metadata
JSON: 24-1.0067154.json
JSON-LD: 24-1.0067154-ld.json
RDF/XML (Pretty): 24-1.0067154-rdf.xml
RDF/JSON: 24-1.0067154-rdf.json
Turtle: 24-1.0067154-turtle.txt
N-Triples: 24-1.0067154-rdf-ntriples.txt
Original Record: 24-1.0067154-source.json
Full Text
24-1.0067154-fulltext.txt
Citation
24-1.0067154.ris

Full Text

RELATIONSHIPS BETWEEN CLIMATE AND ANNUAL RADIAL GROWTH IN THREE CONIFEROUS SPECIES IN INTERIOR BRITISH COLUMBIA, CANADA  by Yueh-Hsin Lo  B.Sc. National Taiwan University, 1997 M.Sc. National Taiwan University, 1999  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2009  © Yueh-Hsin Lo, 2009  ABSTRACT The relationship between climate and tree ring chronologies has been considered mainly in relation to temporal variations in the climate regime, and has mainly focused on trees on moisture-stressed sites and/or at the edge of their range. In this thesis, I examined whether temporal patterns in tree ring chronologies within and between three western North American tree species, all growing on zonal (mesic) sites, were consistent across an elevational gradient, and the degree to which ring width variations reflected estimates of past variations in temperature and precipitation. Increment cores were taken from hybrid spruce (Picea glauca x engelmannii), lodgepole pine (Pinus contorta) and Douglas-fir (Pseudotsuga menziesii) along an elevational gradient spanning three biogeoclimatic (BEC) zones (Engelmann Spruce - Subalpine Fir, Montane Spruce, and Interior Douglas-fir zones) in southern interior British Columbia, Canada. Ring width chronologies were prepared and then compared with estimates of past climatic regimes using the climate extrapolation model MTCLIM to create elevational sequences of past temperature and precipitation conditions based on extrapolation from low elevation climate stations. These extrapolations were tested against data from high elevation climate stations in the same general area. Results from correlation, regression and principal component analysis suggest that tree growth/climate relationships over time in these three species were consistent, but that the strength of the response within a species differed by BEC zone. Results indicated that the response to climate was species-dependent, with the strongest response in Douglas-fir and the weakest in hybrid spruce. Finally, a simple computer model was constructed and used to simulate the net primary production for each tree species in relation to recorded climate. Results from this simple modelling exercise explained only a small part of the inter-annual variability of tree-ring growth, but indicated the possibility of successfully simulating the influence of climate on tree growth, provided the submodel is successfully linked to a more complex modelling framework. Results from this thesis will be used to modify an existing ecosystem-level model (FORECAST) to provide it with the capability to simulate climate change effects on tree growth.  ii  TABLE OF CONTENTS  ABSTRACT………………………………………………………………….. ................................................ii TABLE OF CONTENTS.........................................................................................................................iii LIST OF TABLES ..................................................................................................................................vi LIST OF FIGURES ..............................................................................................................................viii LIST OF SYMBOLS AND ABREVIATIONS ..........................................................................................xi ACKNOWLEDGEMENTS………………………………………………………………………………….. ..xv DEDICATION .......................................................................................................................................xvi CO-AUTHORSHIP STATEMENT .......................................................................................................xvii 1. INTRODUCTION .............................................................................................................................................1 1.1. W HAT IS CLIMATE CHANGE AND WHAT EVIDENCE DO WE HAVE FOR IT? ....................................................3 1.1.1. Global climate change ....................................................................................................................6 1.1.2. Climate history of British Columbia............................................................................................. 11 1.1.3. Impacts of climate change on forest ecosystems.....................................................................12 1.2. THE PHYSIOLOGICAL CONCEPTS UNDERLYING TREE RING FORMATION...................................................15 1.2.1. Global research in dendroclimatology ........................................................................................17 1.2.2 Dendroclimatology research in British Columbia .......................................................................19 1.3. RELATIONSHIP BETWEEN CLIMATE AND TREE GROWTH: MODELLING APPROACHES................................20 1.3.1. Climate models ..............................................................................................................................20 1.3.2. Biological models...........................................................................................................................22 1.4. REFERENCES ...........................................................................................................................................24 2. VALIDATION OF THE MOUNTAIN MICROCLIMATE SIMULATION MODEL (MTCLIM) IN INTERIOR BRITISH COLUMBIA, CANADA ................................................................................................33 2.1. INTRODUCTION .........................................................................................................................................33 2.2. MODEL DESCRIPTION ...............................................................................................................................35 2.3. METHODOLOGY TO TEST THE PERFORMANCE OF MTCLIM IN THE RESEARCH AREA .............................36 2.4. RESULTS ...................................................................................................................................................41 2.5. DISCUSSION .............................................................................................................................................61 2.6. REFERENCES ...........................................................................................................................................64 3. RELATIONSHIPS BETWEEN CLIMATE AND TREE RADIAL GROWTH IN INTERIOR BRITISH COLUMBIA, CANADA .....................................................................................................................................67 3.1. INTRODUCTION .........................................................................................................................................67 3.2. METHODS .................................................................................................................................................68 3.2.1. Sampling sites................................................................................................................................68 3.2.2. Sample collection and process....................................................................................................70 3.2.3. Sampling regime............................................................................................................................71 3.2.4. Core preparation and creation of tree-ring chronologies .........................................................72 3.2.5. Climate data ...................................................................................................................................73 3.2.6. Statistical analysis .........................................................................................................................73 3.3. RESULTS ...................................................................................................................................................75 3.3.1. Tree ring general information.......................................................................................................75 3.3.2. Growth-Climate relationships.......................................................................................................80 3.4. DISCUSSION .............................................................................................................................................93 3.4.1. Sample selection rules .................................................................................................................93 iii  3.4.2. X-ray densitometer ........................................................................................................................93 3.4.3. Tree ring analysis results..............................................................................................................95 3.5. CONCLUSIONS ........................................................................................................................................104 3.6. REFERENCES .........................................................................................................................................108 4. DEVELOPMENT AND PRELIMINARY TESTING OF A TREE PRODUCTIVITY -CLIMATE MODEL WITH DENDROCLIMATOLOGICAL DATA ................................................................................................113 4.1. INTRODUCTION .......................................................................................................................................113 4.2. MODELS OF TREE GROWTH AND CLIMATE ..............................................................................................116 4.2.1. FORECAST.................................................................................................................................. 116 4.2.2. ForWaDy (Forest Water Dynamics) model.............................................................................. 119 4.3. DEVELOPMENT OF A STELLA TREE PRODUCTIVITY – CLIMATE MODEL ................................................120 4.3.1. Tree productivity-climate model description ............................................................................121 4.3.2. Simulation results ........................................................................................................................125 4.4. DISCUSSION AND CONCLUSIONS ............................................................................................................140 4.5. REFERENCES .........................................................................................................................................143 5. CONCLUDING REMARKS .......................................................................................................................146 5.1. MAIN FINDINGS OF THIS WORK ...............................................................................................................146 5.2. STRENGTHS AND WEAKNESSES OF THIS RESEARCH .............................................................................152 5.3. CONCLUSIONS ........................................................................................................................................154 5.4. FUTURE RESEARCH ................................................................................................................................155 5.5. REFERENCES .........................................................................................................................................158 APPENDIX A. SUPPLEMENTARY INFORMATION FOR THE VALIDATION OF MTCLIM ...............161 A.1. GEOGRAPHIC WEATHER STATIONS INFORMATION IN THE PRELIMINARY STUDY ....................................161 A.2. CALCULATION OF LAPSE RATES AND PRECIPITATION ISOHYETS FOR THE EXPERIMENTAL SITES ..........162 A.3. VALIDATION RESULTS OF THE PRELIMINARY STUDY...............................................................................164 A.4. REGRESSION RESULTS OF MONTHLY PRECIPITATION DATA IN THREE SITES..........................................166 APPENDIX B. ADDITIONAL STATISTICAL OUTPUT FOR RELATIONS TREE RING WIDTH – CLIMATE VARIABLES...................................................................................................................................168 B.1. PRINCIPAL COMPONENT ANALYSIS ........................................................................................................168 B.1.1. Lodgepole pine............................................................................................................................168 B.1.2. Hybrid spruce ..............................................................................................................................170 B.1.3. Douglas-fir....................................................................................................................................171 B.2. CORRELATIONS ......................................................................................................................................172 B.2.1. Lodgepole pine............................................................................................................................172 B.2.2. Hybrid spruce ..............................................................................................................................176 B.2.3. Douglas-fir....................................................................................................................................179 B.3. REGRESSIONS .......................................................................................................................................182 B.3.1. Lodgepole pine............................................................................................................................182 B.3.2. Hybrid spruce ..............................................................................................................................189 B.3.2. Douglas-fir....................................................................................................................................191 APPENDIX C. REVIEW OF MODELS LINKING TREE GROWTH AND CLIMATE AND ADDITIONAL GRAPHICAL OUTPUT OF THE MODEL TEST.........................................................................................198 C.1. STAND LEVEL CLIMATE CHANGE MODELS: A REVIEW .............................................................................198 C.1.1. PnET.............................................................................................................................................198 C.1.2. Forest-BGC and Tree-BGC .......................................................................................................199 C.1.3. BIOMASS.....................................................................................................................................201 C.1.4. LINKAGES ...................................................................................................................................202 C.1.5. G’DAY ...........................................................................................................................................203 C.1.6. 3-PG (Physiological Principles in Predicting Growth) ...........................................................204 C.1.7. CENTURY, TREEDYN3, and TRIPLEX ..................................................................................206 iv  C.1.8. Canadian models: BEPS, EASS, ECOSYS and CN-CLASS...............................................209 C.1.9. References ..................................................................................................................................213 C.2. TREE PRODUCTIVITY -CLIMATE MODEL .................................................................................................216 C.3. ADDITIONAL GRAPHICAL OUTPUT FOR THE TEST OF THE TREE PRODUCTIVITY-CLIMATE MODEL ..........217 C.3.1. Graphical comparisons of seasonal limiting factors changing for different combinations of temperature and precipitation growth scenarios................................................................................217 C.3.2. Graphical comparisons of NPP predictions and tree-ring chronologies. ............................219 C.3.3. Results of regressions of NPP predictions vs. tree-ring chronologies ................................222  v  LIST OF TABLES Table 1. 1. Typology of climate extremes. .................................................................................7 Table 2. 1. Geographic information for the weather stations used to test MTCLIM...............38 Table 2. 2. Monthly temperature and precipitation summary of the testing sites. ...................40 Table 2. 3. Descriptive statistics of model performance for the simulation of average monthly maximum temperatures for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White ..............................................................................41 Table 2. 4. Descriptive statistics of model performance for the simulation of average monthly minimum temperature for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White.........................................................................................42 Table 2. 5. Descriptive statistics of model performance for the simulation of monthly total precipitation in Hedley NP Mine, Vernon Silver Star Lodge and Big White ...........42 Table 3. 1. Plot location and sample information summary.....................................................71 Table 3. 2. Site elevation and descriptive statistics of the three tree species ...........................75 Table 3. 3. Pearson’s correlation coefficients (r) between different plots................................76 Table 3. 4. Significant correlations between tree-ring residual chronologies and climate variables for lodgepole pine......................................................................................86 Table 3. 5. Significant correlations between tree-ring residual chronologies and climate variables for hybrid spruce........................................................................................87 Table 3. 6. Significant correlations between tree-ring residual chronologies and climate variables for Douglas-fir. ..........................................................................................88 Table 3. 7. Significant regressions (p < 0.05) between tree-ring residual index (TRI) and selected climate variables, for different tree species at different BEC zones.. .........91 Table 3. 8. Models for tree-ring residual index (TRI) selected by stepwise regression for different climate variables, for different species, BEC zones and transects. ............92 Table 3. 9. Reported relationships between ring growth and climate variables for spruce and lodgepole pine ..................................................................................................100 Table 3. 10. Reported relationships between Douglas-fir ring growth and climate variables ..................................................................................................................101 Table 4. 1. Descriptive statistics of projected June and July potential NPP for high and low elevation climate data inputs............................................................................128 Table 4. 2. Descriptive statistics of lodgepole pine (Pl), hybrid spruce (Sx) and Douglas-fir (Fd) in different BEC zones (ESSF, MS and IDF). .............................128 Table 4. 3. Linear regression results between simulation outputs vs. tree ring chronologies for Douglas-fir in the MS zone. ..............................................................................135 Table 4. 4. Linear regression results between simulation outputs vs. tree ring chronologies for Douglas-fir in the IDF zone. .............................................................................136 Table 4. 5. Linear regression results between previous year simulation outputs vs. tree ring chronologies for hybrid spruce in both ESSF and MS zones..........................139 Table A. 1. Geographic information for the weather stations used to test MTCLIM in the preliminary study. ...................................................................................................161 Table A. 2. Descriptive statistics of model performance for the simulation of average monthly maximum temperatures for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White ............................................................................164 vi  Table A. 3 Descriptive statistics of model performance for the simulation of average monthly minimum temperature in Hedley NP Mine, Vernon Silver Star Lodge and Big White. . ......................................................................................................165 Table A. 4. Descriptive statistics of model performance for the simulation of monthly total precipitation in Hedley NP Mine, Vernon Silver Star Lodge and Big White. 165 Table C. 1. Comparison of different ecosystem processes, climate input included and main features in several stand-level models. .......................................................... 211 Table C. 2. Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the ESSF zone................................................222 Table C. 3 Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the MS zone.........................................................................223 Table C. 4. Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the IDF zone. .................................................224 Table C. 5. Linear regression results between simulation outputs vs. tree ring chronologies for hybrid spruce in the ESSF zone...................................................225 Table C. 6. Linear regression results between simulation outputs vs. tree ring chronologies for hybrid spruce in the MS zone. .....................................................226  vii  LIST OF FIGURES Figure 1. 1. Example of ecosystem variation inside of a biological zone .................................2 Figure 1. 2. Ice core evidence of atmospheric concentrations of carbon dioxide over the last 10,000 years and since 1750..............................................................................7 Figure 1. 3. Change in climate averages compared to the period 1961-1990............................8 Figure 1. 4. Surface temperature changes continentally and globally from observations and model simulations .............................................................................................9 Figure 1. 5. Northern Hemisphere temperature variation from 1000 to 2000 AD relative to 1961 to 1990 ......................................................................................................10 Figure 1. 6. The climate history of the Earth ...........................................................................10 Figure 1. 7. Change of temperature and precipitation over the last century in British Columbia................................................................................................................11 Figure 1. 8. Predicted climate change using CGCM1 and CGCM2........................................21 Figure 2. 1. Flow chart showing the steps followed by MTCLIM to estimate daily microclimate data in mountainous terrain..............................................................36 Figure 2. 2. Location of the weather stations used to test MTCLIM in the Okanagan Valley area (Interior B.C.)......................................................................................39 Figure 2. 3. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed and simulated data at Hedley Mine ..................44 Figure 2. 4. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed and simulated data at Hedley Mine ..................45 Figure 2. 5. Monthly relative-frequency distribution histogram of daily precipitation comparing observed and simulated data at Hedley Mine ......................................46 Figure 2. 6. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed and simulated data at Silver Star ......................47 Figure 2. 7. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed and simulated data at Silver Star ......................48 Figure 2. 8. Monthly relative-frequency distribution histogram of daily precipitation comparing observed and simulated data at Silver Star ..........................................49 Figure 2. 9. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed and simulated data at Big White.......................50 Figure 2. 10. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed and simulated data at Big White.......................51 Figure 2. 11. Monthly relative-frequency distribution histogram of daily precipitation comparing observed and simulated data at Big White...........................................52 Figure 2. 12. Monthly regressions of predicted on observed daily maximum temperature at Hedley Mine.......................................................................................................53 Figure 2. 13. Monthly regressions of predicted on observed daily minimum temperature at Hedley Mine.......................................................................................................54 Figure 2. 14. Monthly regressions of predicted on observed monthly total precipitation at Hedley Mine...........................................................................................................55 Figure 2. 15. Monthly regressions of predicted on observed daily maximum temperature at Silver Star...........................................................................................................56 Figure 2. 16. Monthly regressions of predicted on observed daily minimum temperature at Silver Star...........................................................................................................57 viii  Figure 2. 17. Monthly regressions of predicted on observed monthly total precipitation at Silver Star...............................................................................................................58 Figure 2. 18. Monthly regressions of predicted (y axis) on observed (x axis) daily maximum temperature and daily minimum temperature at Big White .................59 Figure 2. 19. Monthly regressions of predicted on observed monthly total precipitation at Big White ...............................................................................................................60 Figure 3. 1.The location of the study site in Tree Farm License 49 ........................................69 Figure 3. 2. Climatic diagrams of Westwold (1971 – 2002)....................................................70 Figure 3. 3. Lodgepole pine growth indexed residual chronologies for the ESSF, MS and IDF zones. ..............................................................................................................77 Figure 3. 4. Spruce growth indexed residual chronologies for the ESSF and MS zones. .......78 Figure 3. 5. Douglas-fir growth indexed residual chronologies for the MS and IDF zones....79 Figure 3. 6. Plots of regressions of residual indexed chronologies within species between different BEC zones. ..............................................................................................81 Figure 3. 7. Plots of regressions of residual indexed chronologies within BEC zones between different species. ......................................................................................82 Figure 3. 8. Pearson’s Correlation Coefficients between the residual indexed chronology of lodgepole pine and climate variables from 1922 to 1997..................................83 Figure 3. 9. Pearson’s Correlation Coefficients between the residual indexed chronology of hybrid spruce and climate variables from 1922 to 1997. ..................................84 Figure 3. 10. Pearson’s Correlation Coefficients between the residual indexed chronology of Douglas-fir and climate variables from 1922 to 1997.......................................85 Figure 3. 11. Relative position of x-ray photo and line plot ....................................................94 Figure 3. 12. Eight increment core conditions which affect the quality of X-ray densitometer output................................................................................................95 Figure 3. 13. The relationship between climate variables and tree ring formation during the growing season...............................................................................................105 Figure 3. 14. The relationship between climate variables and tree ring formation prior to the growing season...............................................................................................106 Figure 4. 1. Flow chart of climate change impacts on forest ecosystem production.............115 Figure 4. 2. Files and program structure of FORECAST.......................................................117 Figure 4. 3. Flow chart of ForWaDy compartments and processes (Seely et al., 1997)........120 Figure 4. 4. Flow chart of tree productivity-climate model...................................................123 Figure 4. 5. Relationship between growth rate multiplier and air temperature and precipitation for three hypothetical species of each climate variables. ............124 Figure 4. 6. Relationship between frost days of the month and frost index for the hypothetical species. ............................................................................................125 Figure 4. 7. Seasonal limiting factors changing for different combinations of temperature and precipitation growth adaptation scenarios at high and low elevation for three consecutive years. .......................................................................................126 Figure 4. 8. Average and standard deviation of projected potential NPP in June and July and tree ring index. ..............................................................................................129 Figure 4. 9. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at the MS zone. ......................................130 Figure 4. 10. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at the MS zone. ...............................131 ix  Figure 4. 11. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology in the IDF zone. ..............................132 Figure 4. 12. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at IDF zone. ....................................133 Figure 4. 13. Regressions between tree ring indexes from the chronology of Douglas-fir at IDF zone and potential early growing season NPP calculated with climate from a high elevation site.....................................................................................137 Figure 4. 14. Regressions between tree ring indexes from the chronology of Douglas-fir at IDF zone and potential early growing season NPP calculated with climate from a low elevation site......................................................................................138 Figure A. 1. Location of the weather stations used in the preliminary study to test MTCLIM in the Okanagan Valley area ...............................................................162 Figure A. 2. Relationship between altitude and average monthly maximum temperature in the Okanagan Valley.............................................................................................163 Figure A. 3. Relationship between altitude and average monthly minimum temperature in the Okanagan Valley.............................................................................................163 Figure A. 4. Relationship between altitude and annual total precipitation in Okanagan Valley ...................................................................................................................164 Figure A. 5. Regression between observed and predicted values of monthly precipitation in Hedley Mine. ...................................................................................................166 Figure C. 1. Tree productivity-climate model structure written in STELLA. ...................... 216 Figure C. 2. Seasonal limiting factors changing for different combinations of temperature and precipitation growth scenario at high elevation (climate for ESSF zone) for three consecutive years.................................................................................. 217 Figure C. 3. Seasonal limiting factors changing for different combinations of temperature and precipitation growth scenario at low elevation (climate for IDF zone) for three consecutive years. ...................................................................................... 218 Figure C. 4. Simulation results of Jun and July NPP vs. tree ring index data forlodgepole pine chronology at ESSF zone............................................................................ 219 Figure C. 5. Simulation results of Jun and July NPP vs. tree ring index data for lodgepole pine chronology at MS zone. .............................................................................. 219 Figure C. 6. Simulation results of Jun and July NPP vs. tree ring index data for lodgepole pine chronology at IDF zone............................................................................... 220 Figure C. 7. Simulation results of Jun and July NPP vs. tree ring index data for hybrid spruce chronology at ESSF zone. ....................................................................... 220 Figure C. 8. Simulation results of Jun and July NPP vs. tree ring index data for hybrid spruce chronology at MS zone............................................................................ 221  x  LIST OF SYMBOLS AND ABREVIATIONS AGCM Annual accum. GDD above 5 °C Annual accum. GDD above 10 °C Annual CMI Annual P Annual PET August Temperature BEC C-CIARN CGCMs CMI CMI January ESSF GCM GDD at 5 from May to August GDD at 5 from June to August GDD at 10 from May to August GDD at 10 from June to August GHGs GPP IDF January Temperature LAI MAE m a.s.l. ME MER MS NPP SDO SDM OAGCMs OGCMs P PAnnual PAnnual-1  Atmospheric General Circulation Models. Accumulated degree-days above 5 °C of the current year Accumulated degree-days above 10 °C of the current year Accumulated climate moisture index of the current year Accumulated annual precipitation of the current year Accumulated potential evapotranspiration of the current year Monthly mean temperature of August Biogeoclimatic zone of British Columbia Canadian Climate Impacts and Adaptation Research Network Coupled General Circulation Model Climate moisture index Accumulated climate moisture index for January. (Idem for the other months) Engelmann spruce – sphagnum biogeoclimatic zone General Circulation Model. Accumulated degree-days above 5 °C from May to August of the current year Accumulated degree-days above 5 °C from June to August of the current year Accumulated degree-days above 10 °C from May to August of the current year Accumulated degree-days above 10 °C from June to August of the current year Greenhouse gas Gross primary production Interior Douglas-fir biogeoclimatic zone January monthly mean temperature Leaf area index Mean absolute error Meters above sea level Modelling efficiency Mean error (average bias) Mountain spruce biogeoclimatic zone Net primary production Standard deviation of the observations Standard deviation of the MTCLIM predictions Oceanic-Atmospheric General Circulation Models. Oceanic General Circulation Models. Monthly precipitation Accumulated annual precipitation of the current year Accumulated annual precipitation of the previous year  xi  PJan PJan -1 PDec-1 to PFeb PDec -1 to PFeb < 300 mm PGrowing season POct-1 to PMar PSpring PSpring-1 P from May to August < 150 mm  P from May to August < 200 mm  P from May to August < 250 mm  P from May to August < 300 mm  P from June to August < 100 mm  P from June to August < 200 mm  P from June to August < 250 mm  P from June to August < 300 mm  PET PETAnnual t PET January PET Jan  Accumulated precipitation of current January. (Idem for other months) Accumulated precipitation of previous January. (Idem for other months) Accumulated precipitation from previous December to current February (winter) Accumulated precipitation from previous December to current February (winter), accounting for this value only if it is under 300 mm Accumulated precipitation of the current growing season. Accumulated precipitation from previous October to current March Accumulated current spring precipitation. Idem for other seasons. Accumulated previous spring precipitation. Idem for other seasons. Accumulated precipitation from current May to August accounting for this value only if it is less than 150 mm Accumulated precipitation from current May to August accounting for this value only if it is less than 200 mm Accumulated precipitation from current May to August accounting for this value only if it is less than 250 mm Accumulated precipitation from current May to August accounting for this value only if it is e under 300 mm Accumulated precipitation from current June to August accounting for this value only if it is less than 100 mm Accumulated precipitation from current June to August accounting for this value only if it is less than 200 mm Accumulated precipitation from current June to August accounting for this value only if it is less than 250 mm Accumulated precipitation from current June to August accounting for this value only if it is less than 300 mm Potential evapotranspiration Potential evapotranspiration of the current year Accumulated potential evapotranspiration for current January (Idem for the other months) Accumulated potential evapotranspiration for current January (Idem for other months) xii  PETJan -1 PP12 to P2 PP12 to P2 < 150 mm  PP12 to P2 < 200 mm  PP12 to P2 < 250 mm  PP12 to P2 < 300 mm  Precipitation from May to August Precipitation from June to August Precipitation from PP6 to PP8 Previous CMI Previous Annual PET U W m2 T TGrowing season TJan TJan -1 TM tM Tm tm TMAX TMIN TOct -1 to TApr TSpring TSpring-1 TWinter max TDI TFL  Accumulated potential evapotranspiration of previous January (Idem for other months) Accumulated precipitation from previous December to current February Accumulated precipitation from previous December to current February (winter) with this value under 150 mm Accumulated precipitation from previous December to current February (winter) with this value under 200 mm Accumulated precipitation from previous December to current February (winter) with this value under 250 mm Accumulated precipitation from previous December to current February (winter) with this value under 300 mm Accumulated precipitation from current May to August Accumulated precipitation from current June to August of the current year Precipitation from previous June to previous August Accumulated climate moisture index of the previous year Accumulated annual potential evapotranspiration of the previous year Theil’s inequality coefficient Watts per square metre Annual mean temperature of the current year Average temperature of the current growing season. Current January monthly mean temperature (Idem for other months) Previous January monthly mean temperature (Idem for other months) Annual absolute maximum temperature Average daily maximum temperature Absolute minimum temperature Average daily minimum temperature Daily maximum temperature Daily minimum temperature Average temperature from previous October to current April Average temperature of current spring (Idem for other seasons) Average temperature of previous spring (Idem for other seasons) Maximum temperature of the winter Transpiration Deficit Index Tree Farm License xiii  TRI UBC UNBC VPD  Tree-ring Residual Index chronology University of British Columbia University of Northern British Columbia Vapour pressure deficit  xiv  ACKNOWLEDGEMENTS  I would like to sincerely thank Dr. J.P. Hamish Kimmins for his time, direction and motivation for helping me to gain a passion in forest ecology. Dr. Clive Welham, Dr. Andrew Black, Dr. Robert Guy and Dr. Brad Seely provided extensive project advice and critical review of the manuscripts, without their help thesis would not have been possible! Even though Dr. Peter Joliffe was not able to be my committee member at the last moment, I still would like to thank him for previous advice and direction in my research. Extreme thanks go out to Dr. Lori Daniels for her in-depth help and advice on the training of dendrochronology and project design. I also gratefully thank Dr. Shawn Mansfield for permission to use his X-ray densitometer. Thanks to Dr. Lloyd for his kindness, training in the field and advice regarding research. Finally, thanks go to the Riverside Forest Products Ltd (now Tolko Industries Ltd) for its help with site selection and mapping. For help with fieldwork, thanks to Brock, Kristina, Julie, and Demon! Thanks to Kyu-Young and Ian for their help with lab work. Tremendous thanks go out to my parents for their encouragement and love. Thanks to my roommate, Szu-Chi, for her unlimited support and help in both life and financial! And finally, but most importantly, to my friend and ‘supervisor’, Dr. Juan Blanco, for his endless pushing, help, editing and encouragement with the thesis, and lots of support!  This research was supported by an International Student Scholarship from the University of British Columbia, and Graduate Research Assistantships from Dr. Clive Welham, Dr. Brad Seely, Dr Hamish Kimmins (UBC) and Dr. Dan Moore (UBC / C-CIARN).  xv  DEDICATION  gÉ Åç ytÅ|Äç  xvi  CO-AUTHORSHIP STATEMENT Chapter Two: Validation of the Mountain Microclimate Simulation Model (MTCLIM) in Interior British Columbia, Canada.  Yueh-Hsin Lo executed the research, carried out the statistic analysis and wrote the manuscript. Dr. Juan Blanco and Dr. Clive Welham helped to interpret the statistical analysis and provided edits for the manuscript. Dr. Hamish (J.P.) Kimmins and Dr. Brad Seely helped to identify the main guidelines of this research and provided edits for the manuscript.  Chapter Three: Relationships between Climate and Tree Radial Growth in Interior British Columbia, Canada.  Yueh-Hsin Lo executed the research, carried out the field work and sample analysis and wrote the manuscript. Dr. Juan Blanco, Dr. Brad Seely and Dr. Clive Welham helped to interpret the statistical analysis and provide edits for the manuscript. Dr. Hamish (J.P.) Kimmins helped to identify the main guidelines of this research and provide edits for the manuscript.  Chapter Four: Development and Preliminary Testing of a Tree Productivity Model with Dendroclimatological Data.  Yueh-Hsin Lo executed the research, carried out statistical analysis and wrote the manuscript. Dr. Brad Seely helped to develop the model. Dr. Juan Blanco and Dr. Brad Seely helped to interpret the statistical analysis and provide edits for the manuscript. Dr. Hamish Kimmins and Dr. Clive Welham helped to identify the main guidelines of this research and provide edits for the manuscript.  xvii  1. Introduction Almost daily, the media report on public concerns about global warming: the melting of the Arctic ice sheet and glaciers, abnormally high temperatures, and severe wind, snow and rain storms (BBC, 2008; CBC, 2008; Climate Change News Digits, 2008). It is now generally accepted that human-accelerated global climate change is occurring and that it constitutes one of the major environmental risks that humans and other species face (King, 2005; IPCC, 2007). Because most instrument-derived weather data are recent, dating only from the early 1900’s, discussion on the climate change issue at longer time scales must rely on natural records, such as Arctic ice cores, pollen or lake sediment profiles, and tree ring data (Alfsen, 2001; IPCC, 2001; Mayewski and White, 2002; Gajewski and Atkinson, 2003). Dendroclimatology is the science that studies the relationship between tree ring data and climate variables. This branch of ecology was developed in Arizona (U.S.A.) starting in 1901 when the father of dendrochronology, Andrew E. Douglas, was on a trip through the forest of northern Arizona (Fritts, 1976). He grew up in New England and knew that the trees growing there were influenced mainly by shading and competition. When he saw the tree rings from the stumps in Arizona forest, he noticed patterns that were quite different from those observed in New England tree rings. He concluded that the patterns reflected annual variations in soil moisture, and that this offered a possibility of using tree ring patterns as a tool to trace past climates, assuming that the growth / environment relationship was the same through time. Since then, dendroclimatology has developed greatly (Beniston, 2002; Bräker, 2002). However, based on the principle of limiting factors (e.g. growth cannot proceed faster than is allowed by the most limiting factor, Begon et al., 2006), most of the research has been done either at the tree line or in arid sites. According to Krajina (1969), this is a biased sampling strategy if we want to understand the growth / climate relationship at the ecosystem level. Forty years ago, Dr. V.J. Krajina developed an ecosystem classification for British Columbia (Krajina, 1969). He identified climatic units that set the overall vegetation potential (biogeoclimatic zones and subzones), and quantified how this climatically-determined potential was modified by local gradients of soil moisture and nutrients associated with local topographic gradients and soil variations (Figure 1.1).  1  Figure 1. 1. Example of ecosystem variation inside of a biological zone (Adapted from Krajina, 1969). Tree symbols and their size reflect their growth class (example of Lodgepole pine (Pinus contorta) in the Engelmann Spruce-Subalpine Fir (ESSF) and Interior Douglas-fir (IDF) zones). The vertical axis is the soil moisture gradient (0 – very xeric to 8 – hydric). The horizontal axis is the soil nutrient regime (A – oligotrophic to F – hypereutrophic). Black represents shade tolerant while white represents shade intolerant. Circled Roman numbers indicate growth classes of trees while the small numbers (uncircled) represent the specific biogeocoenosis. The contour curves serves as the projection of biogeocoenoses and growth class. Detail descriptions are in Krajina (1969).  As concern over climate change increases, the need for predictive tools to estimate climate change impacts is becoming increasingly urgent. One major concern is species distribution change or extinction due to shifts in climate. Many studies have used a bioclimatic envelope modeling approach to assess climate change impacts (Pearson and Dawson, 2003; Beaumont et al., 2005; Hamann and Wang, 2005). This approach uses spatial tree distribution-climatic variable relationships to generate new maps of species distributions under various predicted future climate conditions. Geographical information systems (GIS) have been used to map species ranges or ecosystem characteristics with respect to their climatic envelope (Hamann and Wang, 2005). However, these models have lacked of some important ecological factors (e.g. biotic interactions, evolutionary change, species dispersal, natural disturbance factors and soil-climate relationships) which renders their predictions questionable. In this chapter, I review the literature on climate change, how key atmospheric factors (i.e.  2  temperature, precipitation, CO2) affect plant growth, the dendroclimatology research which has been done globally and in B.C., the current climate and ecosystem modelling approach to the climate change issue, and how we link forest models with field data. Chapter two introduces the climate extrapolation model, MTCLIM (Thornton et al., 1997) that was used to generate the historical climate record to be compared to tree ring width series, and validated its predictions for my study area. In Chapter three, I describe my tree ring research at Tree Farm License 49 (TFL49), near Kelowna, British Columbia (Riverside Forest Products Ltd., 2001). Because the goal of this research was to seek a climate/growth relationship at different elevations, I used ring width patterns of three tree species along an elevational (i.e., a climate) gradient of three biogeoclimatic (BEC) zones (i.e. ESSF, MS and IDF). Correlation analysis, principal component analysis and regression analysis were used to test for a relationship between tree ring and climate variables. In Chapter four I review stand level growth models, describe the major models (i.e. FORECAST and ForWaDy) which will be linked together for simulating possible climate change impacts, and a computer model developed to simulate the influence of climate on tree productivity. The last chapter presents conclusions and suggestions about future research and how the results of the thesis work could be used to simulate climate change effects in the FORECAST-Climate model (the FORECAST model with climate change simulation ability).  1.1. What is climate change and what evidence do we have for it? The Intergovernmental Panel on Climate Change (IPCC) has defined climate change as “a statistically significant variation in either the mean state of the climate or in its variability, persisting for an extended period (typically decades or longer)” (IPCC, 2001a). In 2007, it reached the conclusion that the climate is changing rapidly and the frequency of abnormal climatic phenomena is increasing. In the 2007 report (IPCC 2007) it pointed out that “…the warmth of the last half century is unusual in at least the previous 1,300 years…” Although few scientists believe that the warming trend is part of the natural history of the earth (Loehle, 2004), most scientists and much of society believe that human activities are causing or are accelerating the speed of global warming (IPCC, 1995, 2001a, b). The major climate change factor is the change in the concentration of carbon dioxide and other anthropogenically produced greenhouse gases (GHGs) (e.g., methane and nitrous oxide) in the 3  atmosphere, which increase retention of energy emitted by the Earth, changing temperature means and variation, and changing precipitation temporal patterns and amount. The temporal pattern of global temperature change has been reflected in the pattern of atmospheric CO2 concentration change (IPCC, 2001a). Although the causal relationship between atmospheric temperature and CO2 is not clear, they are strongly correlated with each other. Since the 19th century, large quantities of carbon dioxide have been emitted to the atmosphere due to fossil fuel use and extensive land use change. Based on model simulations, Houghton et al. (1996) predicted that the average annual rate of increase of atmospheric CO2 will continue to be 1.8 ppm yr-1 leading to a doubling of CO2 concentration by the end of 21st century compared to before the start of the industrial revolution (cited in Norby et al., 2001). From ice cores, tree ring chronologies, pollen evidence and lake sediments, scientists have discovered that the latter half of the 20th century was the warmest period of the past 10,000 years (Gajewski and Atkinson, 2003). Compared with other climate change indicators mentioned above, tree rings provide what is probably the highest resolution, continuous, long-term climatic proxy data (Trotter et al., 2002). Many studies have demonstrated that tree-ring characteristics are correlated with environmental factors, especially temperature and precipitation (Dang and Lieffers, 1989; MacDonald et al., 1998; Dobry and Klinka, 1998; Miina, 2000). By studying rings of annual tree stem growth and the causes of their variation from year to year, scientists can reconstruct past climates and develop hypotheses regarding the potential effects of climate change (Fritts, 1976; Wilson and Luckman, 2003; Schongart et al., 2006). Bigelow et al. (2003) concluded that climate change has the potential to dramatically alter the vegetation distribution in arctic ecosystems. From the fossil record, we also know that animal populations have migrated in the past in response to climatic fluctuations (Toweill, 1998). However, human populations now occupy large portions of the earth and block traditional species migration routes. Under such conditions, we may cause the fragmentation of animal habitats and cause the extinction of some species in the future by preventing their adaptation to changing climates by moving with their habitats (Roberts, 1988). Climate change does not only relate to the risk of species extinction, but it also has impacts on human society (i.e. health, energy use, economics and insurance), water resources, forest ecosystems, agricultural systems, and coastal ecosystems (Bjørke and Seki, 2001; IPCC, 2001b). Because precipitation patterns can change, therefore, floods and droughts will occur more 4  frequently and more severely and affect forest and agricultural systems (Taylor and Taylor, 1997; Merritt et al., 2006). For coastal organisms, climate change may cause habitat loss or salt stress as a result of sea level change (Roberts, 1988). As mentioned above, forest species may change their distribution under global warming (Kohlmaier et al, 1995; Shafer et al., 2001). Those species that cannot migrate fast enough or cannot adapt to the future climate may be locally eliminated or even become extinct (Roberts, 1988). There are many different scenarios used in climate change modelling to predict future climates (IPCC, 2001a). Scenarios that assume only moderate warming may increase forest productivity. On the other hand, under scenarios that assume greater temperature increases, climate change may increase drought stress and decrease forest productivity (Bachelet et al., 2001). The temporal variation in the difference between increased biomass due to increased photosynthetic rate and biomass loss due to disturbances, stresses and respiration results in variation in carbon budgets over time (Bachelet et al., 2001; McKinnon et al., 2003). Climate change also affects disturbance regimes (Bachelet et al., 2001). Fire is a major disturbance in temperate and northern forest ecosystems (Long et al., 1998; Hély et al., 2001). There are three key factors, which determine the frequency, type and severity of fire: weather, plant community structure and fuel mass (Keane et al., 1995; Hély et al., 2001). Based on the charcoal and fossil pollen in lake sediments, Clark et al. (1996) and Long et al. (1998) found that over the long term, the vegetation structure is highly correlated with the fire regimes. Normal fire regimes, however, could be changed due to climate change because temperature and precipitation strongly influence fire frequency (Long et al., 1998). During long periods of high temperatures and drought, the interval between fires will be shorter than during cool, wet periods. After a fire, plants species composition and community structure will change. Hence, fire and vegetation influence each other. The IPCC reports points out that according to current General Circulation Models (GCMs), many parts of North America will become warmer and wetter (IPCC, 2001a, b). However, higher amounts of precipitation may not necessarily prevent droughts because precipitation may not be evenly distributed through the year. There may still be periods of drought, which may be associated with fire and accompanying change of vegetation. Other important disturbances are insects and diseases. In the Canadian boreal forest, the wood loss due to insects is 1.3-2.0 times as much as the annual average fire loss (Volney and Fleming, 2000). Studies show that climate change will directly (warmer winter) or indirectly (change of 5  plant distribution and defence ability) influence the outbreak of insects (Ayres and Lombardero, 2000; Volney and Fleming, 2000). From the research of Steve Taylor and Bob Erickson (Canadian Forest Services, personal communication) we know that mountain pine beetle outbreaks are related to climate and forest structure. Under global warming, the winter temperature may not be sufficiently low and sustained long enough to kill the larva of the beetle. Meanwhile, the warmer summer also will increase the ability of the adults to migrate and may result in an additional generations per year. As a consequence, if there is enough food for insects, the area of forest damage will increase as the temperature increases. This effect of climate change can already be seen in the unprecedented outbreak of mountain pine beetle in British Columbia, which is expected to keep extending under the predicted future climate (Pacific Forest Center, Canadian Forest Services, 2003; Eng et al., 2005; Kurz et al., 2008).  1.1.1. Global climate change Figures 1.2 to 1.5 show historical records of changing atmospheric carbon dioxide concentrations, global surface temperatures, global sea level and Northern Hemisphere snow cover (IPCC, 2007). Based on the temperature variation over the past one and half centuries (Figure 1.3), the global mean temperature has increased 0.8-1.0°C (IPCC, 2007). However, the rate of change over the last century was very rapid compared to the rates before the 20th century (Figures 1.2 and 1.5). On the other hand, if we consider a longer time period (i.e. 230 million years ago when fossil fuels formed, Cannell (1995), Figure 1.6) the temperature variation over the last century is still within the historical record range. That is why some researchers still hold to the idea that current global warming is a normal phenomenon in earth climate history. However, in the last IPCC report (2007), it was concluded that current global warming, especially the frequency and magnitude of recently abnormal climate events are increasing and they may cause many environmental issues and important amounts of financial losses. Therefore, it is an important issue that we should be concerned about. The discussion above has focused on the trend of the means of key climatic variables. However, the variation around the mean also has influences on society and ecosystems. In fact, these variations may play an even more important role in the climate change issue than simply the overall temporal trends of means. For example, the change of minimum temperature affects the survival rate of herbivores and pathogens (Ayres and Lombardero, 2000) and it also affects 6  the level of frost damage to plants. The severity and frequency of extreme weather events (e.g. hurricanes, droughts, ice storms, etc.) also have great impacts on both human society and ecosystems, and cause extended financial losses. Droughts and summer thunderstorms are major initiators of forest fires and will alter the components and structure of forest ecosystem. Table 1.1 illustrates the typology of climate extremes (IPCC, 2001b). Table 1. 1. Typology of climate extremes (IPCC, 2001b). Type Simple extremes  Description Individual local weather variables exceeding critical level on a continuous scale  Examples of events Heavy rainfall, high/low temperature, high wind speed  Typical method of characterization Frequency/return period, sequence and/or duration of variable exceeding a critical level  Complex extremes  Severe weather associated with particular climate phenomena, often requiring a critical combination of variables A plausible future climatic state with potentially extreme large-scale or global outcomes  Tropical cyclones, droughts, ice storms, ENSO-related events  Frequency return period, magnitude, duration of variable(s) exceeding a critical level, severity of impacts  Unique or singular phenomena  Collapse of major ice Probability of occurrence and sheets, cessation of magnitude of impact thermohaline circulation, major circulation changes * Stakeholders also can be engaged to define extreme circumstances via thresholds that mark a critical level of impact for the purposes of risk assessment. Such critical levels often are locally specific, so they may differ between regions.  Figure 1. 2. Ice core evidence of atmospheric concentrations of carbon dioxide over the last 10,000 years (large panel) and since 1750 (inset panel). The Y-axis on the right hand is the corresponding radiative forcings1. Different colors represent different studies and red lines represent atmospheric samples. (IPCC, 2007). 1  Radiative forcing is defined as the change between the incoming radiation energy and the outgoing radiation energy in a given climate system. 7  Figure 1. 3. Change compared to the period 1961-1990 in (a) global average surface temperature; (b) global average sea level rise from tide gauge (blue) and satellite (red) data; and (c) Northern Hemisphere snow cover for March-April. Each dot is the yearly values. The black line represents the decadal averaged values. The shaded areas are the estimated uncertainty intervals (IPCC, 2007).  8  Figure 1. 4. Surface temperature changes continentally and globally from observed (line) and model simulation (shade). The base line is the average surface temperature of period 1901-1950. Black lines are the decadal average from 1906-2005. Lines are dashed where spatial coverage is less than 50%. The blue shaded area shows the results of 19 simulations from 5 climate models under 90% confidence interval using only the natural forcings due to solar activity and volcanoes. The red shaded area shows the results of 58 simulations from 14 climate models under 90% confidence interval using both natural and anthropogenic forcings (IPCC, 2007).  9  Figure 1. 5. Northern Hemisphere temperature variation from 1000 to 2000AD relative to 1961 to 1990 (IPCC, 2001a).  Figure 1. 6. The climate history of the Earth. The curves are the representations of postulated departures from present global means, dashed portions are estimated from other sources of data. The time unit is a million year. Mean global temperature and precipitation are based on the record from air bubbles in the long ice cores and are roughly estimated in a million year scale (Alfsen, 2001).  10  1.1.2. Climate history of British Columbia Over the past 50 years, the regional pattern of climate in Canada has shown a warming trend in south-western Canada while in north-eastern B.C. it became cooler (C-CIARN and UNBC, 2003). Environment Canada (1997) reported that the warming in Canada amounted to approximately 1 ºC in the last century, with the B.C. Ministry of Water, Land and Air Protection (2002) summarised past climate change in B.C. over the last 100 years as: • “Average annual temperature warmed by 0.6ºC on the coast, 1.1 ºC in the interior, and 1.7 ºC in northern B.C.” (Fig. 1.7) • “Night-time temperatures increased across most of B.C. in spring and summer.” • “Precipitation increased in southern B.C. by 2 to 4 percent per decade.”(Fig. 1.7) • “Sea surface temperatures increased by 0.9 ºC to 1.8 ºC along the B.C. coast.” • “Two large B.C. glaciers retreated by more than a kilometre each.”  Figure 1. 7. Change of temperature and precipitation over the last century in British Columbia. (Ministry of Water, Land and Air Protection, 2002). © Province of British Columbia. All rights reserved. Reprinted with permission of the Province of British Columbia.  Figure 1.7 shows that the ranges of temperature and precipitation change are different from place to place, but the general trends are that since the last century, south and western British Columbia have become warmer and wetter. However, not only the annual temperature and precipitation have changed, but also seasonal temperature, maximum and minimum temperature and snowpack (Ministry of Water, Land and Air Protection, 2002). For example, spring in most of British Columbia is now warmer compared to the mid-20th century. Annual daytime maximum temperatures increased by 0.9 ºC in the Southern Interior, and annual night-time 11  minimum temperatures increased by 0.9 ºC and by 1.3 ºC to 1.7 ºC per century on the coast and parts of the interior, respectively. According to the IPCC (2001b), precipitation in North America has very likely increased by 0.5 to 1.0 percent per decade over the 20th century, while according to Ministry of Water, Land and Air Protection historical data, precipitation increase in B.C. has exceeded these values (Figure 1.7). This means we may have more water resources under future climate. However, depending on when and how much rain falls during the year, it may have positive influences (e.g. increased falling rain during growing season) or negative influences (e.g. rain and snow storms) on human society, forest and agricultural systems (Cohen et al., 2004). Regrettably, research for specific changes in species in British Columbia is minimal (Kimmel, 2009), but other climate-related changes observed in B.C. include snow pack which has decreased over the past 50 years, with significant regional variation. Spring runoff/snowmelt are occurring earlier, water temperatures are increasing, and the fire season is longer with more fire activity in the boreal forest (Gayton, 2008; Wilson and Hebda, 2008).  1.1.3. Impacts of climate change on forest ecosystems All three major components of climate change (i.e. temperature, precipitation and CO2) affect forest ecosystems in different ways. I will describe the influences of each component in the following sections.  a. Temperature The temperature component of climate change includes changes in mean annual air temperature, seasonal variation, maximum and minimum values, and the variability from year to year in each variable. There is evidence, for example, that over the past 50 years minimum night time temperatures have been increasing, and at a rate that exceeds the increase in daily maximum temperature (IPCC, 2001 a, b), thereby reducing the diurnal temperature range (Walther et al. 2002). Temperature is one of the key factors determining plant growth and distribution (Walther et al., 2002; Peñuelas and Boada, 2003). Since the 1960’s, spring activities of plants such as bud formation, leaf emergence and flowering have occurred progressively earlier (Menzel and Fabian, 1999). In the North Hemisphere, spring has started about 1.5 days earlier per decade (Schwartz et 12  al., 2005). In Europe, the length of growing season has increased at about 3.6 days per decade over the past 50 years. Fall leaf colour changes have become progressively delayed by 0.3 to 1.6 days per decade (Walther et al., 2002). A more recent report suggests that the mean advance of spring/summer has been 2.5 days per decade, and leaf colouring and fall have been delayed by 1.0 day per decade (Menzel et al., 2005). Because early blooming and herbaceous plants have greater responses to winter warming than late blooming and woody plants (Walther et al., 2002), global warming will change the competition pattern between different species. Even though higher temperature may be good for plants because it increases the photosynthetic rate and causes a longer growing season, not all biological processes are favoured by high temperature. Flower blossoming and seed germination for some species need a period of time below a certain minimum temperature to stimulate the reactions (Kimmins, 2004), which means, warming may cause reproduction failure and decrease the number of offspring. In some areas, high temperatures cause water stress or increase the metabolic rate. Therefore, they would either decrease the growth rate or consume extra energy. Research has shown that climate change not only influences plant growth and their competitive ability, but also plant distribution. In boreal forests, the upper and lower temperature limits depend on the chilling requirement and cold damage, respectively (Shafer et al., 2001). Some evidence shows that global warming leads to plant population shifts poleward and upward in elevation (Walther et al., 2002; Peñuelas and Boada, 2003), but there is also evidence to the contrary (MacDonald et al., 1998). Recent research has shown additional proofs of migration of tree species in the U.S.A. (Woodall et al., 2009) and increased rates of mortality in the Pacific Northwest (Van Matgem et al., 2009). Plant-growth modellers tend to use reported temperature (or precipitation) thresholds to predict future species distribution (discussed further below). However, the boundaries we set today reflect those species’ realized niches (the area that the species occupy now) instead of their fundamental niche (the area where the species have the potential to occupy) (Loehel and LeBlanc, 1996). If we want to build a model to simulate ecosystem response to climate change, this may cause biased model simulations (Loehle, 1996). Because climate change also affects the variation of temperature, extreme temperatures and rapid changes of temperature can damage plants (e.g. frost damage, cell damage) (Saxe et al., 2001). When temperature combines with other factors, like precipitation, it causes different kinds of damage (e.g. wilt, snow break). These either decrease the competitive ability or increase 13  mortality rate. Climate change could also impact disturbance regimes. For example, global warming may increase fire frequency and severity (Running, 2006; Westerling et al., 2008), which in turn can induce changes in stand structure, species composition, nutrient availability, pH and soil flora and fauna (Landhäusser and Wein, 1993; Kimmins, 2004). The frequency and severity of climate change also changes the duration of the different stages of stand dynamics. For example, more frequent but less severe fire (due to lack of fuel) keeps the stand in a later (e.g. understory re-initiation, Oliver and Larson, 1996) stage while severe fire tends to keep the stand in the stand initiation and stem exclusion stages.  b. Precipitation Changes in precipitation regimes (quantity, intensity and timing of precipitation, and the different forms: snow, rain, fog, dew, hail, etc.) are also an important components of climate change. Climate change can affect the scale and frequency of precipitation, which impacts water resources and ecosystem water availability (IPCC, 2001b; Ministry of Water, Land and Air Protection, 2002; Cohen et al., 2004). In any ecosystem, water is an important component. It is “the material that makes life possible” (Kimmins, 2004). Water in the soil (soil moisture) affects photosynthesis and decomposition rates (Saxe et al., 2001). Water availability and temperature are important determinants of vegetation type (Begon et al., 2006). Glaciers, streams and rivers can change terrain features and create different ecosystem environments. Also, precipitation can be seen as a disturbance regime such as drought or floods. It also has links with other factors that cause disturbance. For example, a dry environment has higher potential for fire and insect outbreak than a humid environment that is more susceptible to pathogens (Long et al., 1998; Ayres and Lombardero, 2000; Volney and Fleming, 2000). At the individual tree level, soil moisture is a key factor that affects growth and competition. It affects photosynthesis rates directly through photochemical processes or indirectly by affecting stomatal closure and therefore changing the amount of carbon dioxide that diffuses into the leaf (Fritts, 1976). It also affects soil nutrient availability via decomposition. Typically, it has been assumed that a plant’s water use efficiency will increase as CO2 concentration increases. Recent results, however, suggest that this assumption might be wrong because other factors such as 14  vapour pressure deficit (VPD) and leaf temperature also affect water use (Reichstein et al., 2002). Like temperature, precipitation affects the ecosystem in a variety of ways. Oversimplification of these interactions can lead to erroneous conclusions about climate change effects  c. Carbon dioxide (CO2) Atmospheric CO2 concentration is another major factor affecting forest ecosystems. At the tree level, CO2 affects plants directly through its positive effects on photosynthesis, respiration and transpiration (Ehleringer and Field, 1993). However, the effects of elevated CO2 concentrations at the stand level are mixed (McMurtrie, 1991; Norby et al., 2001; Sallas et al., 2003). In the short term and for seedlings, CO2 does increase growth, but researchers have also found that CO2 changes the carbon allocation patterns (i.e. produce more root biomass than leaf biomass) leading to the conclusion that increased CO2 represents a negative feedback to growth due to less leaf area (Sallas et al., 2003). Data on the response of mature trees and the effects of long-term exposure are scarce (McMurtrie, 1991). McMurtrie (1991) pointed out that any CO2 fertilization effect would be significant only if it was matched by an increase in nutrient supply. At the ecosystem level, Norby et al. (2001) noted that long-term ecosystem productivity responses to elevated CO2 may ultimately depend on N availability to plants and on the ability of plants to use N. They used a meta-analysis to assess the effects of elevated CO2 concentration on litterfall mass loss and found that increased CO2 concentrations in the atmosphere did not have a significant effect on mass loss during decomposition. In other words, the nutrient supply (here meaning N supply) may not match the speed of change in CO2 supply and the photosynthetic rate could be limited due to lack of N. Hence, at the stand level, CO2 response in a particular forest can only be assessed after understanding the nutrient supply conditions and the main factors which control the growth of that particular stand (Kirschbaum et al., 1998).  1.2. The physiological concepts underlying tree ring formation In woody vascular plants, variation in cell wall thickness and ring density can be used as an index related to the factors that affect tree ring formation (Fritts, 1976). Two major processes control the size of growing tissue and the speed of growth. One is nutrient and water movement: the speeds of which are affected by temperature, soil moisture, air humidity, stomatal aperture, 15  light intensity and CO2 concentration. The other is the synthesis and assimilation of organic matter, which are accomplished by photosynthesis and respiration. The factors directly limiting photosynthesis are wavelength and light spectral composition and intensity, CO2 concentration, temperature, and water availability (Kramer and Kozlowski, 1979; Larcher, 1995). However, these are the factors at leaf level; at tree level, leaf age, stomatal conductance, acclimation to sun or shade conditions and nutritional status are additional factors that affect the rate of photosynthesis (Landsberg and Gower, 1997). At the species level, different groups of species (C3, C4, and CAM) have different photosynthetic efficiencies and rates. As for respiration, its rate is influenced by both environmental factors (i.e. temperature, water deficit, oxygen), and physiological factors (i.e. age of both tissue and tree and the time of the year) (Fritts, 1976; Landsberg and Gower, 1997). Photosynthesis and respiration rates increase as temperature increases, but the preferred or optimal range is different for each process. A temperature between 0 °C to 15 °C causes photosynthesis rates to increase rapidly, but above that range, the rate remains the same or even decreases. On the other hand, respiration rate increases slowly at low temperatures. However, in the 10 °C to 30 °C temperature range, respiration rate increases considerably, about 2 to 2.5 times for each 10 °C throughout the temperature range. Above the optimum temperature for maximum photosynthesis rate, respiration exceeds photosynthesis which causes the net assimilation rate to decrease (Fritts, 1976) Plant age and physiological activity also influence photosynthesis and respiration rates (Salisbury and Ross, 1992). For example, respiration rates in younger tissues are higher than in older tissues. Also, respiration during the growing season is higher than during the dormancy period. On the other hand, the photosynthetic rate increases as leaves or needles grow, but it declines after plants reach mature and/or over-mature stage (Fritts, 1976). At the tree level, photosynthetic efficiency sometimes declines with tree age. This phenomenon occurs because respiration consumption of C exceeds photosynthetic production (Salisbury and Ross, 1992; Landsberg and Gower, 1997). From what is said above, it is clear that environmental factors influence tree growth and therefore ring formation. Dendrochronology is the science which studies these relationships, and it is based on seven basic principles: 1) the uniformitarian principle, 2) the limiting factor principle, 3) the ecological amplitude principle, 4) the site selection principle, 5) the aggregate 16  tree growth principle, 6) the crossdating principle, and 7) the replication principle. A detailed description of each one can be found in Fritts’ (1976) book. By using these principles, we can date each tree ring, compare the pattern between samples (i.e. crossdating), identify an overall pattern which most of the samples fall into it, build a master chronology which represents the growth response to environmental factors in the sampled area and then use this master chronology to explore the relationship between tree growth and the environmental factors we are interested in. By assuming that the climate/tree-ring relationship is relatively constant over time, we can then reconstruct past climates. However, this approach has limitations. Several studies have found that the relationship between ring width and temperature has changed markedly over the last 40 years in some boreal forests (Wilmking et al., 2004; Rocha et al., 2006).There might have been bias when using a fixed relationship to reconstruct palaeoclimatic data.  1.2.1. Global research in dendroclimatology Dendrochronology has been widely used around the world to provide evidence of climate change (Cook et al., 2003; Sarris et al., 2007; Hegerl et al., 2007), and as a tool to explore ecosystem response to climate (Yu et al., 2007). When compared to ice core and river sediment studies or other proxy records, tree-ring is an easy-to-access tool to quantify the evidence of climate change and has the highest resolution among others (Bräker, 2002; Gajewski and Atkinson, 2003). In Europe, Lindholm and Eronen (2002) used Scots pine (Pinus sylvestris L.) ring width to reconstruct mid-summer temperatures in northern Fennoscandia dating back to 50 AD. McCarroll et al. (2003) also did a pilot study in northern Finland by using multi-proxy data of Scots pine (i.e. earlywood, latewood and annual ring width; earlywood, latewood and maximum density; stable carbon isotope ratio; height increment; needle production; pollen deposition and cambium dynamics) from three sites along a latitudinal gradient to build a record of past climates. Skomarkova et al. (2006) used radial growth, wood density and carbon isotope ratios in tree rings of European beech (Fagus sylvatica L.) to explore the inter-annual and seasonal variability of growth in Germany and Italy. They found the response period of ring width and density is different, but it is mostly limited by soil moisture during the growing season. They also found that the climate signal is over-ridden by effects of stand density and crown structure. Using dendroclimatology, Sarris et al. (2007) found that rainfall declined rapidly after the late 1970’s 17  for the entire Mediterranean basin. Sidorova et al. (2007) reported that the radial growth of larch in North Central Siberia was influenced by July temperatures. Webber et al. (2007) calculated response functions between tree ring width and soil moisture in the dry Valais valley (Switzerland) by using principal component analysis and they found the species response to climate differently. They also found there is a sub-regional difference in species response pattern. Savva et al. (2006) examined Norway spruce along an altitudinal gradient in the Tatra Mountains, Poland. They found a consistent decline in radial growth as altitude increased. In Asia, Liang et al. (2001) used dendroclimatology to evaluate climate-growth relationships of Meyer spruce (Picea meyeri) in northern China. Cook et al. (2003) reconstructed the past climates in the Himalayas, and found that over the past 400 years the winter temperature (October to February) has increased. But they also found that there has been a temperature decline in spring and summer (February to June) since 1960. Yu et al. (2007) studied tree line Betula ermanii forests in Changbai Mountain, Northeast China. They used correlation, response function coefficients and regression analysis and found soil moisture deficit was the limiting factor for growth in these forests. In North America, Sauchyn and Beaudoin (1998) used tree ring and other proxy records (i.e. weather records, river sediments) to reconstruct recent environmental change in the south-western Canadian plains and found that during the last millennium there has been a trend of increased drought severity. MacDonald et al. (1998) reported that the growth of black and white spruce at the central Canadian northern tree line reflected temperature variation. Hogg et al. (2002) used light coloured aspen rings and a climate moisture index to reconstruct past defoliation histories and to demonstrate how climate change had affected insect outbreaks in north-western Alberta. Peterson and others conducted a series of studies in the North Cascades National Park, Washington (Holman and Peterson, 2006, Case and Peterson, 2005, 2007). They examined the effects of climatic variability during the 20th century on the growth of lodgepole pine (Pinus contorta) and Douglas-fir (Pesudotsuga menziesii) along an elevational, gradient. They used multivariate analysis and correlation analysis to identify the relationship between climate and growth, and factorial analysis to separate their samples into mid-elevation and high-elevation chronologies. They found the chronology at different elevations responded differently to climate factors. The high-elevation group responded positively to annual temperature while the mid-elevation group responded negatively to growing season maximum 18  temperature but positively to growing season precipitation. In South America, Lara et al. (2001) used radial growth to reconstruct past precipitation in the central Andes of Chile. In Africa, Schongart et al. (2006) used regression to establish a relationship between annual ring width indices and climate variables (i.e. local precipitation, sea surface temperature and El Niño- Southern Oscillation (ENSO)) in central-western part of Benin and north-eastern Ivory Coast; they found a relationship between local precipitation and tree growth, but not with ENSO. Pearson and Searson (2002) summarized the history of dendrochronology development in Australia and made recommendations for future research. Even though in some regions Dendrochronology research has only just started (Pearson and Searson, 2002; Schongart et al., 2006), it is becoming more and more important in ecosystem research and climate change research (Rocha et al., 2006; IPCC, 2007)  1.2.2 Dendroclimatology research in British Columbia A number of studies have explored tree ring response to temperature and precipitation in British Columbia. Zhang et al. (1999) used different conifer species to identify regional climatic anomalies and insect outbreaks (i.e. spruce beetle) over the past four centuries in central British Columbia. Zhang and Hebda (2004) used Douglas-fir to explore the radial growth of trees in mountainous areas; they found that growth response was subject to conditions associated with changes in elevation. For example, growing season precipitation influenced growth for both high and low elevation Douglas-fir, but the response to temperature was different at different elevations. Watson and Luckman (2001) summarized the dendroclimatology research in southern Canadian Rockies and used multiple regression models to reconstruct annual precipitation (previous August to current July or previous July to current June) for the Banff, Jasper and Cranbrook areas. Luckman and Wilson (2005) also reconstructed past summer temperatures in the Canadian Rockies. Bower et al. (2005) used ring width, density and mass components in conjunction with regression analysis of growing season soil moisture deficit to build a drought response coefficient for Douglas-fir in costal B.C. They only found a significant drought response on very dry sites; at moister sites, the coefficient was very close to zero. All things considered, these are only evidence of climate change and its relationship with tree-ring growth. They are just the foundations of the work that deals with the study of forest management under a future changing environment. To actually predict the future and 19  management the forest, we need model simulation to help us forecast the future and make plans.  1.3. Relationship between climate and tree growth: modelling approaches Due to the complexities involved in the assessment of climate change and its consequences, there is an increasing need for credible models with which to forecast the effects of climate change. Simple models are easy to create and run, but unless designed with sufficient complexity and incorporating the key determinants, they are unlikely to be useful for application in climate change research. There are two kinds of models used to assess climate change. The first are the climate models, which generate future climate data (i.e., predictions) and can be scaled down from global to regional or local scale, and then used to provide guidance on general policy and management initiatives. The second kind are biological models which allow us to input future climate conditions from the first type of models, and then make predictions about what might happen to forests and other ecosystems in the future. I describe these two kinds of models in the following sections.  1.3.1. Climate models By using global models, we can simulate and study the climate system (IPCC, 2001a). In the 1920’s, scientists advanced the idea of simulating atmospheric motion, but because of the complex calculations, the application of these ideas only became possible in the 1950’s with the development and availability of computers. After the 1960’s, General Circulation Models (GCMs) became more common but were restricted to atmospheric circulation. By the 1970’s, atmospheric global circulation models (or AGCMs) had become the central tool for simulating climate. Meanwhile, ocean modellers started to build the oceanic general circulation models (OGCMs) for understanding of oceanic general circulation. Also, in this period, scientists became interested in long-term climate predictions based on a combination of AGCMs and OGCMs. By the mid-1980s, combined models (OAGCM) had been developed and became the new standard for the future climate modeling (American Institute of Physics, 2008). 20  The latest type of climate models in Canada is the second generation of coupled general circulation models (CGCM2) (Flato et al., 2000; Flato and Boer, 2000). The Canadian CGCM2 is an OAGCM model with a resolution of 3.7° x 3.7° for the earth surface grid and 1.8° x 1.8° for the ocean (Environment Canada, 2008). Simulation results from a run using the IS92a emission scenario (i.e. increase of CO2 at a rate of 1% per year (compounded) till year 2100) (IPCC, 2001a) are shown in Figure 1.8 (Environment Canada, 2008). These results predict that future average annual temperature in BC may increase by 1ºC to 4ºC.  Figure 1. 8. Predicted results using CGCM1 and CGCM2. Left: Global annual average surface temperature change, relative to the 1900-1929 average as produced by CGCM1 and CGCM2 for various forcing scenarios. Right: Annual mean surface air temperature change, 1971-1990 to 2041-2060 as projected by CGCM1 (upper panel) and CGCM2 (lower panel). (Environment Canada, 2008).  A number of local climate models have been developed. These include PRISM (Parameter-elevation Regressions on Independent Slopes Model; Daly and Neilson, 1992) and MTCLIM (MounTain microCLIMate simulation model, Running et al. 1987), which are widely used in North America and overseas (Thornton et al., 2000; Li et al., 2001; Daly et al., 2002; Almeida and Landsberg, 2003; Hunter and Meentemeyer, 2005; Cienciala and Tatarinov, 2006). 21  ANUSPLIN (Australian National University SPLINe routine, Hutchinson and Bischof, 1983) was originally developed in Australia but now is also used in North America (Custer et al., 1996; McKeeny et al., 2006). All of these models can be used at the landscape scale and can generate local climate data from GCM outputs2. 1.3.2. Biological models Many models have been developed to assess the impacts of climate change on forest ecosystems. Loehle and LeBlanc (1996) reviewed models which focus on the abundance and distribution of species and classified them into three groups: (a) ecological response surface models, (b) biogeographic correlation models (also referred to as bioclimatic envelope models), and (c) forest stand simulation models. The first two were developed under the assumptions that climate ultimately restricts species distributions and abundances, and the relationships between climate and species distributions and abundances in the past (or now) will remain the same in the future (Pearson and Dawson, 2003; Hamann and Wang, 2005). The third was developed on the basis of knowledge of plant physiology and ecology. Many scientists have favoured bioclimatic models because: First, the cost of field data collection for calibration of other types of models of the effects of climate change can be prohibitive. Second, they are easy to run because they usually do not have many variables. Third, for many species we lack the knowledge of the ecology and biology needed to apply physiological models. Fourth, many datasets consist of presence-only data so that bioclimatic models are often the most accessible. Consequently, bioclimatic models have become a widely used tool for assessing the potential responses of species ranges to climate change (Beaumont et al., 2005). However, many researchers have criticized this approach because bioclimatic models don’t represent biotic interactions, evolutionary change (adaptability) and species-dispersal strategies and limitations (Pearson and Dawson, 2003). These critics have frequently turned to forest-stand simulation models such as JABOWA (Botkin et al., 1972; Botkin, 1993) and FORET (Shugart, 1984). These models integrate species-specific information regarding species characteristics and their interactions with environmental factors, something that cannot be addressed in bioclimatic models. Other models have assessed the impact of climate change from an energy-flow perspective. 2  For a discussion of these landscape scale climate models and their pros and cons, see Chapter 2. 22  These models can calculate how much primary production and biomass will be produced under various climate-change scenarios. Such models include Forest-BGC (Korol et al., 1995; Running and Coughlan, 1988), BIOMASS (McMurtrie et al., 1990; McMurtrie and Landsberg, 1992), and LINKAGES (Pastor and Post, 1985). Detailed model descriptions and comparisons are in Chapter four and Appendix C. In this study, the objectives were: 1) to examine and quantify the relationship between key climate variables and tree ring width chronologies along a major elevation gradient to determine whether the responses of individual species varied among different elevations. I used the tree ring width of lodgepole pine (Pinus controrta Dougl. ex Loud var. latifolia), Douglas-fir (Pesudotsuga menziesii (Mirb.) Franco var. glauca) and hybrid white spruce (Picea glauca x engelmannii), growing in three biogeoclimatic (BEC) zones along an elevational gradient: Engelmann Spruce-Subalpine Fir (ESSF), Montane Spruce (MS), and Interior Douglas-fir (IDF) biogeoclimatic zones (altitudinal variation), and examined the relationships between growth indices and climate variables. 2) To develop and test a tree productivity – climate model. The knowledge gained here will be used to modify the ecosystem-level stand growth model FORECAST (Kimmins et al., 1986; Kimmins et al., 1999), in order to simulate possible climate change impacts on stand-level ecosystem processes and forest growth. Rocha et al. (2006) pointed out that only a few studies have attempted to bridge between ecological modeling and dendrochronological approaches. This tree productivity – climate model will be an example of such bridging and the modified FORECAST will be a useful tool for the forest manager and policy maker to assess the possible consequences of climate change under different management strategies.  23  1.4. References Alfsen, K.H. 2001. Climate change and sustainability in Europe. CICERO Policy Note 2001: 03. American Institute of Physics. 2008. A History of Atmospheric General Circulation Models. Available at: http://www.aip.org/history/sloan/gcm/intro.html. Retrieved Dec 15th, 2008. Ayres, M.P. and M.J. Lombardero. 2000. Assessing the consequences of global change for forest disturbance from herbivores and pathogens. The Science of the Total Environment. 262: 263-286. Bachelet, D., R.P. Neilson, J.M. Lenihan and R.J. Drapek. 2001. Climate change effects on vegetation distribution and carbon budget in the United States. Ecosystems. 4: 164-185. Barry, R.G. and R.J. Chorley. 1998. Atmosphere, Weather and climate 7th Ed. Routledge. New York. NY. U.S.A. Pp. 409. B.B.C. 2008. BBC News / In Depth / Climate Change. Available at: http://news.bbc.co.uk/2/hi/in_depth/sci_tech/2000/climate_change/default.stm. Retrieved July 20, 2008. Beaumont, L.J., L. Hughes and M. Poulsen. 2005. Predicting species distributions: use of climatic parameters in BIOCLIM and its impact on predictions of species’ current and future distributions. Ecological Modelling. 186: 250-269. Begon, M., C.R. Townsend and J.L. Harper. 2006. Ecology, from individuals to ecosystems 4th Ed. Blackweel Publishing. Oxford, U.K. Pp.738. Beniston, M. 2002. Climate modeling at various spatial and temporal scales: where can dendrochronology help? Dendrochronologia. 20: 117-131. Bigelow N.H., L.B. Brubaker, M.E. Edwards, S.P. Harrison, I.C. Prentice, P.M. Anderson, A.A. Andreev, P.J. Bartlein, T.R. Christensen, W. Cramer, J.O. Kaplan, A.V. Lozhkin, N.V. Matveyeva, D.F. Murray, A.D. McGuire, V.Y. Razzhivin, J.C. Ritchie, B. Smith, D.A. Walker, K. Gajewski, V. Wolf, B.H. Holmqvist, Y. Igarashi, K. Kremenetskii, A. Paus, M.F.J. Pisaric and V.S. Volkova. 2003. Climate change and Arctic ecosystems: 1. Vegetation changes north of 55 degrees N between the last glacial maximum, mid-Holocene, and present. Journal of Geophysical Research-Atmospheres. 108 (D19): Alt. No. 8170. Bjørke, S.Å. and M. Seki. 2001. Vital climate graphics: the impacts of climate change. United Nations Environment Programme (UNEP); Arendal, Norway : GRID-Arendal published. Available at: http://www.grida.no/climate/vital/index.htm Retrieved March 26, 2004. Botkin, D. B. 1993. Forest Dynamics: An Ecological Model. Oxford University Press, New York. Pp. 309. Botkin, D.B., J.G. Janak and J.R. Wallis. 1972. Some ecological consequences of a computer model of forest growth. Journal of Ecology. 60: 849-872. Bower, A.D., W.T. Adams, D. Birkes and D. Nalle. 2005. Response of annual growth ring components to soil moisture deficit in young, plantation-grown Douglas-fir in coastal British Columbia. Canadian Journal of Forest Research. 35: 2491-2499. Bräker, O.U. 2002. Measuring and data processing in tree-ring research - a methodological introduction. Dendrochronologia. 20: 203-216.  24  Canadian Climate Impacts and Adaptation Research Network (C-CIARN) Forest Sector and the University of Northern British Columbia (UNBC). 2003. Climate Change in the Western and Northern Forests of Canada: Impacts and Adaptations. Available at: http://c-ciarn-bc.ires.ubc.ca/resources/pgreport.pdf Retrieved Dec 4, 2004. Cannell, M. 1995. Forests and the Global Carbon Cycle in the Past, Present and Future. European Forest Institute Research Report 2. Joensuu, Finland. Pp. 66. Case, M.J. and D.L. Peterson. 2005. Fine-scale variability in growth-climate relationship of Douglas-fir, North Cascade Range, Washington. Canadian Journal of Forest Research. 35: 2743-2755. Case, M.J. and D.L. Peterson. 2007. Growth-climate relations of lodgepole pine in the North Cascades National Park, Washington. Northwest Science. 81: 62-75. C.B.C 2008. CBC. Ca News – Climate Change. Available at: http://www.cbc.ca/news/background/climatechange/ Retrieved July 20, 2008. Cienciala, E and F.A. Tatarinov. 2006. Application of BIOME-BGC model to managed forests 2. Comparison with long-term observations of stand production for major tree species. Forest Ecology and Management. 237: 252-266. Clark, J.S., P.D. Royall and C. Ghumbley. 1996. The role of fire during climate change in an eastern deciduous forest at Devil’s Bathtub, New York. Ecology. 77: 2148-2166. Climate Change News Digest. 2008. Available at: http://www.climatechangenews.org/ Retrieved July 20, 2008. Cohen, S., D. Neilsen and R. Welbourn. 2004. Expanding the Dialogue on Climate Change & Water Management in the Okanagan Basin, British Columbia. Final Report. Environment Canada. Victoria, British Columbia. Canada. Pp. 257. Cook, E.R., P.J. Krusic and P.D. Jones. 2003. Dendroclimatic signals in long tree-ring chronologies from the Himalayas of Nepal. International Journal of Climatology. 23: 707-732. Creed, I.F., L.E. Band, N.W. Foster, I.K. Morrison, J.A. Nicolson, R.S. Semkin and D.S. Jeffries. 1996. Regulation of nitrate-N release from temperate forests: A test of the N flushing hypothesis. Water Resources Research. 32: 3337-3354. Custer, S.G., P. Farnes, J.P. Wilson and R.D. Snyder. 1996. A comparison of hand and spline-drawn precipitation maps for mountainous Montana. Water Resources Bulletin. 32: 393-405. Daly, C. and R.O. Neilson. 1992. A digital topographic approach to modeling the distribution of precipitation in mountainous terrain. In: Interdisciplinary Approaches in Hydrology and Hydrogeology. American Institute of Hydrology. p.447-454. Daly, C., W.P. Gibson, G.H. Taylor, G.L. Johnson and P. Pasteris. 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research. 22: 99-113. Dang, Q. L. and V. J. Lieffers. 1989. Climate and annual ring growth of black spruce in some Albertal peatlands. Canadian Journal of Botany. 67: 1885-1889. Dobry, J and K. Klinka. 1998. Reconstructing temperature from tree rings of Pacific silver fir in coastal British Columbia. Northwest Science. 72: 81-87. Dodson, R. and D. Marks. 1997. Daily air temperature interpolated at high spatial resolution over a large mountainous region. Climate Research. 8: 1-20. Ehleringer, J.R. and C.B. Field. 1993. Scaling Physiological Processes: Leaf to Global. Academic Press. London. U.K. Pp. 388.  25  Eng, M., A. Fall, J. Hughes, T. Shore, B. Riel, P. Hall and A. Walton. 2005. Provincial level projection of the current mountain pine beetle outbreak: An overview of the model (BCMPB v2) and results of year 2 of the project. Mountain Pine Beetle Initiative Working Paper 2005-20. Canadian Forest Service and the B.C. Forest Service, Victoria, B.C. Canada. Pp. 54. Environment Canada. 1997. The Canada Country Study: climate impacts and adaptation. National Summary for Policy Makers. Ottawa. ON. Canada. Pp. 24. Environment Canada. 2008. The Second Generation Coupled Global Climate Model (CGCM2). Available at: http://www.cccma.ec.gc.ca/models/cgcm2.shtml. Retrieved December 15, 2008. Flato, G.M. and G.J. Boer. 2001 Warming asymmetry in climate change simulations. Geophysical Research Letters. 28: 195-198. Flato, G.M., G.J. Boer, W.G. Lee, N.A. McFarlane, D. Ramsden, M.C. Reader and A.J. Weaver. 2000. The Canadian Centre for Climate Modelling and Analysis global coupled model and its climate. Climate Dynamics. 16: 451-467. Fritts, H.C. 1976. Tree Ring and Climate. Academic Press, London, U.K., Pp. 567. Fritts, H.C. and T.W. Swetnam. 1989. Dendroecology: a tool for evaluating variations in past and present forest environments. Advances in Ecological Research. 19: 111-189. Gajewski K. and D.A. Atkinson. 2003. Climatic change in northern Canada. Environmental Reviews. 11: 69-102. Hamann, A. and T.L. Wang. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology. 128: 211-221. Hegerl, G.C., T.J. Crowley, M. Allen, W.T. Hyde, H.N. Pollack, J. Smerdon and E. Zorita. 2007. Detection of human influence on a new, validated 1500-year temperature reconstruction. Journal of Climate. 20: 650-666. Hély, C., M. Flannigan, Y. Bergeron and D. McRae. 2001. Role of vegetation and weather on fire behavior in the Canadian mixedwood boreal forest using two fire behavior prediction systems. Canadian Journal of Forest Research. 31: 430-441. Hogg, E.H., J.P. Brandt and B. Kochtubajda. 2002. Growth and dieback of aspen forests in northwestern Alberta, Canada, in relation to climate and insects. Canadian Journal of Forest Research. 32: 823-832. Holman, M.L. and D.L. Peterson. 2006. Spatial and temporal variability in forest growth in the Olympic Mountains, Washington: sensitivity to climatic variability. Canadian Journal of Forest Research. 36: 92-104. Hunter, R.D. and R.K. Meentemeyer. 2005. Climatologically aided mapping of daily precipitation and temperature. Journal of Applied Meteorology. 44: 1501-1510. Hutchinson, M.F. and R.J. Bischof. 1983. A new method for estimating the spatial distribution of mean seasonal and annual rainfall applied to the Hunter Valley, New South Wales. Australian Meteorological Magazine. 31: 179-184. Intergovernmental Panel on Climate Change (IPCC). 1995. IPCC Second Assessment Synthesis of Scientific-Technical Information Relevant to Interpreting Article 2 of the UNFCCC. Available at: http://www.ipcc.ch/pub/sa(E).pdf Retrieved March 26, 2004. Intergovernmental Panel on Climate Change (IPCC). 2001a. Climate Change 2001: The Scientific Basis. Cambridge University Press. Pp. 881. Available at: http://www.grida.no/climate/ipcc_tar/wg1/index.htm Retrieved Dec 4, 2004.  26  Intergovernmental Panel on Climate Change (IPCC). 2001b. Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge University Press. Pp. 1032. Available at: http://www.grida.no/climate/ipcc_tar/wg2/index.htm Retrieved December 4, 2004. Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007: The Physical Science Basis. Summary for Policymakers. Available at: http://hosted.ap.org/specials/interactives/_documents/climate_report.pdf Retrieved February 1, 2009. Keane, R.E., K. Ryan and S.W. Running. 1995. Simulation the effects of fire and climate change on northern Rocky Mountain landscapes using the ecological process model FIRE-BGC. General Technical Report RM. 262: 34-38. Kimmins, J.P. 2004. Forest Ecology. A Foundation for Sustainable Management and Environmental Ethics in Forestry. 3rd ed. Prentice Hall, New Jersey. NJ. U.S.A. Pp. 720. Kimmins, J.P., D. Mailly and B. Seely. 1999. Modelling forest ecosystem net primary production: the hybrid simulation approach used in FORECAST. Ecological Modelling. 122: 195-224. Kimmins, J.P., K.A. Scoullar, R.E. Bigley, W. Kurz, P.G. Comeau and L. Chatarpaul. 1986. Yield prediction models: the need for a hybrid ecosystem-level approach incorporating canopy function and architecture. In: T., Fujimory and D. Whitehead. (Eds.). Crown and Canopy Structure in Relation to Productivity. FFPRI, Ibaraki, Japan, Pp. 26-48. King, D. 2005. Climate change: the science and the policy. Journal of Applied Ecology. 42: 779-783. Kirschbaum, M.U.F., B.E. Medlyn, D.A. King, S. Pongracic, D. Murty, H. Keith, P.K. Khanna, P. Snowdon and R.J. Raison. 1998. Modelling forest-growth response to increase CO2 concentration in relation to various factors affecting nutrient supply. Global Change Biology. 4: 23-41. Kohlmaier, G.H., Ch. Hager, A. Nadler, G. Wurth and M.K.B. Ludeke. 1995. Global carbon synamics of higher latitude forests during an anticipated climate change: ecophysiological versus biome-migration view. Water, Air and Soil Pollution. 82: 455-464. Korol, R.L., S.W. Running and K.S. Milner. 1995. Incorporating inter-tree competition into an ecosystem model. Canadian Journal of Forest Research. 25: 413-424. Kramer, P.J. and T.T. Kozlowski. 1979. Physiology of trees, 2nd ed. McGraw-Hill, New York. NY. U.S.A. Pp. 642. Krajina, V.J. 1969. Ecology of forest trees in British Columbia. Ecology of Western North America. 2: 1-146. Kurz W.A., C.C. Dymond, G. Stinson, G.J. Rampley, E.T. Neilson, A.L. Carroll, T. Ebata and L. Safranyik 2008. Mountain pine beetle and forest carbon feedback to climate change. Nature. 452: 987-990. Landsberg, J.J. and R.H. Waring. 1997. A generalized model of forest productivity using simplified concepts of radiation-use efficiency, carbon balance and partitioning. Forest Ecology and Management. 95: 209-228. Landsberg, J.J. and S.T. Gower. 1997. Applications of Physiological Ecology to Forest Management. Academic Press, NY. U.S.A. Pp. 354. Landhäusser, S.M. and R.W. Wein. 1993. Postfire vegetation recovery and tree establishment at the Arctic treeline: climate-change-vegetation-response hypotheses. Journal of Ecology. 81: 665-672. 27  Lara, A., J.C. Aravena, R. Villalba, A. Wolodarsky-Franke, B. Luckman, and R. Wilson. 2001. Dendroclimatology of high-elevation Nothofagus pumilio forests at their northern distribution limit in the central Andes of Chile. Canadian Journal of Forest Research. 31: 925-936. Larcher, W. 1995. Physiological Plant Ecology 3rd. Springer-Verlag Berlin Heidelberg, Germany. Pp. 506. Li, H.-T., W.-Q. Shen and W.-G. Sang. 2001. Research situation and application of MTCLIM model. Journal of Mountain Science. 19: 533-540. Liang, E., X. Shao, Y. Hu and J. Lin. 2001. Dendroclimatic evaluation of climate-growth relationships of Meyer spruce (Picea meyeri) on a sandy substrate in semi-arid grassland, north China. Trees. 15: 230-235. Lindholm, M. and M. Eronen. 2000. A reconstruction of mid-summer temperatures from ring widths of Scots pine since AD 50 in northern Fennoscandia. Geogr. Ann. 82A: 527-535. Loehel, C. 1996. Forest response to climate change. Do simulations predict unrealistic dieback? Journal of Forestry. 94: 13-15. Loehel, C. 2004. Climate change: detection and attribution of trends from long-term geologic data. Ecological Modelling. 171: 433-450. Loehel, C. and D. LeBlanc. 1996. Model-based assessments of climate change effects on forests: a critical review. Ecological Modelling 90: 1-31. Long, C.J., C. Whitlock, P.J. Bartlein and S.H. Millspaugh. 1998. A 9000-year fire history from the Oregon coast range, based on a high-resolution charcoal study. Canadian Journal of Forest Research. 28: 774-787. Luckman, B.H., K.R. Briffa, P.D. Jones and F.H. Schweingruber. 1997. Tree-ring based reconstruction of summer temperatures at the Columbia Icefield, Alberta, Canada, AD 1073-1983. Holocene. 7: 375-389. Luckman, B.H. and R.J.S. Wilson. 2005. Summer temperatures in the Canadian Rockies during the last millennium: a revised record. Climate Dynamics. 24: 131-144. MacDonald, G.M., J.M. Szeicz, J. Claricoates and K.A. Dale. 1998. Response of central Canadian treeline to recent climatic changes. Annals of the association of American Geographers. 88: 183-208. Mayewski, P.A. and F. White. 2002. The Ice Chronicles. The Quest to Understand Global Climate Change. University of New Hampshire. University Press, NE. U.S.A. Pp. 268. McCarroll, D., R. Jalkanen, S. Hicks, M. Tuovinen, M. Gagen, F. Pawellek, D. Eckstein, U. Schmitt, J. Autio and O. Heikkinen. 2003. Multiproxy dendroclimatology: a pilot study in northern Finland. Holocene. 13: 829-838. McKeeny, D.W., J.H. Pedlar, P. Papadopol, and M.F. Hutchinson. 2006. The development of 1901-2000 historical monthly climate models for Canada and the United States. Agricultural and Forest Meteorology. 138: 69-81. McKinnon, G.A., S.L. Webber, and N.A. MacKendrick. 2003. Climate Change in the Western and Northern Forests of Canada: Impacts and Adaptations. A report on the Workshop held February 17-19, 2003. in Prince George, British Columbia. Natural Resource of Canada. McMurtrie, R.E. 1991. Relationship of forest productivity to nutrient and carbon supply - a modeling analysis. Tree Physiology. 9: 87-99.  28  McMurtrie, R.E. and J.J. Landsberg. 1992. Using a simulation model to evaluate the effects of water and nutrients on the growth and carbon partitioning of Pinus radiata. Forest Ecology and Management. 52: 243-260. McMurtrie, R.E., D.A. Rook and F.M. Kelliher. 1990. Modelling the yield of Pinus radiata on a site limited by water and nitrogen. Forest Ecology and Management. 30: 381-413. Menzel, A. and P. Fabian, 1999. Growing season extended in Europe. Nature. 397: 659. Menzel, A., T.H. Sparks, N. Estrella, E. Koch, A. Aasa, R. Ahas, K. Kübler, P. Bissolli, O. Braslavska, A. Briede , F.M. Chmielewski, Z. Crepinsek, Y. Curnel, Å. Dahl, C. Defila, A. Donnelly, Y. Filella. K. Jatczak, F. MåGe, A. Mestre, Ø. Nordli, J. Peñuelas, P. Pirinen, V. RemišOvá, H. Scheifinger, M. Striz, A. Susnik, A.J.H. V. Vliet, F-E. Wielgolaski, S. Zach and A. Zust. 2006. European phenological response to climate change matches the warming pattern. Global Change Biology. 12: 1969-1976. Merritt, W.S., Y. Alila, M. Barton, B. Taylor, S. Cohen and D. Neilsen. 2006. Hydrologic response to scenarios of climate change in sub watersheds of the Okanagan basin, British Columbia. Journal of Hydrology 326: 79-108. Miina, J. 2000. Dependence of tree-ring, earlywood and latewood indices of Scots pine and Norway spruce on climatic factors in eastern Finland. Ecological Modelling. 132: 259-273. Ministry of Water, Land and Air Protection. 2002. Indicators of climate change for British Columbia, 2002. Available at: http://wlapwww.gov.bc.ca/air/climate/indicat/pdf/indcc.pdf Retrieved December 4, 2004. Norby, R.J., M.F. Cotrufo, P. Ineson, E.G. O’Neill and J.G. Canadell. 2001. Elevated CO2, litter chemistry and decomposition: a synthesis. Oecologia. 127: 153-165. Oliver, C.D., B.C. Larson. 1996. Forest Stand Dynamics, Update ed. Wiley, New York, Pp. 520. Pacific Forest Center, Canadian Forest Services. 2003. Mountain Pine Beetle. Available at: http://www.pfc.forestry.ca/entomology/mpb/index_e.html Retrieved November 15, 2003. Pastor, J. and W.M. Post. 1985. Development of a Linked Forest Productivity-Soil Process Model. U.S. Dept. of Energy, ORNL/TM-9519. Pp. 167. Pearson, S.G. and M.J. Searson. 2002. High-resolution data from Australian tree. Australian Journal of Botany. 50: 431-439. Pearson, R.G. and T.P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: are climate envelope models useful? Global Ecology and Biogeography. 12: 361-371. Peñuelas, J. and M. Boada. 2003. A global change-induced biome shift in the Montseny mountain (NE Spain). Global Change Biology. 9: 131-140. Reichstein, M., J.D. Tenhunen, O. Roupsard, J.M. Ourcival, S. Rambal, F. Miglietta, A. Peressotti, M. Pecchiari, G. Tirone and R. Valentini. 2002. Severe drought effects on ecosystem CO2 and H2O fluxes at three Mediterranean evergreen sites: revision of current hypotheses? Global Change Biology. 8: 999-1017. Riverside Forest Products Ltd. 2001. Riverside’s Tree Farm Licence 49 Ecological Forest Stewardship Project. Available at: http://www.riverside.bc.ca/woodlands/tfl49-index.htm Retrieved November 15, 2003. Roberts, L. 1988. Is there life after climate change? Science. 242: 1010-1012.  29  Rocha, A.V., M.L. Goulden, A.L. Dunn and S.C. Wofsy. 2006. On linking interannual tree ring variability with observations of whole-forest CO2 flux. Global Change Biology. 12: 1378-1389. Running, S. 2006. Climate change: is global warming causing more, larger wildfires? Science. 313: 927-928. Running, S.W. and J.C. Coughlan. 1988. A General model of Forest ecosystem process for regional applications. I. Hydrologic balance, canopy gas exchange and primary production processes. Ecological Modelling. 42: 125-154. Running, S.W., R.R. Nemani and R.D. Hungerford. 1987. Extrapolation of synoptic meteorological data in mountainous terrain and its use for simulating forest evapotranspiration and photosynthesis. Canadian Journal of Forest Research. 17: 472-483. Salisbury, F.B. and C.W. Ross. 1992 Plant physiology 4th ed. Wadsworth Publishing Co. Belmont, CA. U.S.A. Pp.682. Sallas, L., E.M. Luomala, J. Utriainen, P. Kainulainen and J. Holopainen. 2003. Contrasting effects of elevated carbon dioxide concentration and temperature on Rubisco activity, chlorophyll fluorescence, needle ultrastructure and secondary metabolites in conifer seedlings. Tree Physiology. 23: 97-108. Sarris, D., D. Christodoulakis and C. Korner. 2007. Recent decline in precipitation and tree growth in the eastern Mediterranean. Global Change Biology. 13: 1187-1200. Sauchyn, D. and A.B. Beaudoin. 1998. Recent environmental change in the southwestern Canadian plains. Canadian Geographer (42), 4; CBCA Reference p.337. Savva, Y., J. Oleksyn, P.B. Reich, M.G. Tjoelker, E.A. Vaganov and J. Modrzynski. 2006. Interannual growth response of Norway spruce to climate along an altitudinal gradient in the Tatra Mountains, Poland. Trees-Structure and Function. 20: 735-746. Saxe, H., M.G.R. Cannnel, Ø. Johnsen, M.G. Ryan and G. Vourlitis. 2001. Tree and forest functioning in response to global warming. New Phytologist. 149: 369-400. Schongart, J., B. Orthmann, K.J. Hennenberg, S. Porembski and M. Worbes. 2006. Climate-growth relationships of tropical tree species in West Africa and their potential for climate reconstruction. Global Change Biology. 12: 1139-1150. Schwartz, M.D., R. Ahas and A. Aasa. 2005. Onset of spring starting earlier across the Northern Hemisphere. Global Change Biology. 12: 343–351. Shafer, S.L., P.J. Bartlein and R.T. Thompson. 2001. Potential changes in the distributions of western North America tree and shrub taxa under future climate scenarios. Ecosystems. 4: 200-215. Shugart, H.H. 1984. A Theory of Forest Dynamics the ecological implications of forest succession models. Springer-Verlag, New York. NY. U.S.A. Pp. 278. Sidorova, O.V., E.A.Vaganov, M.M. Naurzbaev, V.V. Shishov, and M.K. Hughes. 2007. Regional features of the radial growth of larch in North Central Siberia according to millennial Tree-Ring chronologies. Russian Journal of Ecology. 38: 90-93. Skomarkova, M.V., E.A. Vaganov, M. Mund, A. Knohl, P. Linke, A. Boerner and E.D. Schulze. 2006. Inter-annual and seasonal variability of radial growth, wood density and carbon isotope ratios in tree rings of beech (Fagus sylvatica) growing in Germany and Italy. Trees. 20: 571-586.  30  Taylor E. and B. Taylor. 1997. Responding to Global Climate Change in British Columbia and Yukon. Volume I of the Canada Country Study: Climate Impacts and Adaptation. Ministry of Environment, Lands and Parks. Ottawa. ON. Canada. Pp.363. Thornton, P.E, H. Hasenauer and M.A. White. 2000. Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: an application over complex terrain in Austria. Agriculture and Forest Meteorology. 104: 255-271. Thornton, P.E., S.W. Running and M.A. White. 1997. Generating surfaces of daily meteorological variables over large regions of complex terrain. Journal of Hydrology. 190: 214-251. Toweill, D.E. 1998. Climate change and wildlife: what can we expect? p. 27-32 in F.H. Wagner and J. Baron (Eds.) Proceedings of the rocky mountain/great basin regional climate-change workshop. U.S. National Assessment of The Consequences of Climate Change. February 16-18, 1998. Salt Lake City, Utath. Pp.166. Trotter, R.T., N.S. Cobb and T.G. Whitham. 2002. Herbivory, plant resistance, and climate in the tree ring record: Interactions distort climatic reconstructions. Proceedings of The National Academy of Sciences of The United States of America. 99: 10197-10202. Van Mantgen, P.J., N.L. Stephenson, J.C. Byrne, L.D. Daniels, J.F. Franklin, P.Z. Fulé, M.E. Harmon, A.J. Larson, J.M. Smith, A.H. Taylor, T.T. Veblen. 2009. Widespread increase of tree mortality rates in the western United States. Science. 323: 521-524. Van Vliet, A.J.H., R.S. de Groot, Y. Bellens, P. Braun, R. Bruegger, E. Bruns, J. Clevers, C. Estreguil, M, Flechsig, F.O. Jeanneret, M. Maggi, P. Martens, B. Menne, A. Menzel and T. Sparks. 2003. The European Phenology Network. International Journal of Biometeorology. 47: 202-212. Volney, W.J.A. and R.A. Fleming. 2000. Climate change and impacts of boreal forest insects. Agriculture. Ecosystems and Environment. 82: 283-294. Walther, G-R, E. Post, P. Convey, A. Menzal, C. Parmesan, T.J. C. Beebee, J-M Fromentin, O. Hoegh-Guldberg and F. Bairlein. 2002. Ecological responses to recent climate change. Nature. 416: 389-395. Watson, E. and B.H. Luckman. 2001. Dendroclimatic reconstruction of precipitation for sites in the southern Canadian Rockies. Holocene. 11: 203-213. Weber, P., H.Bugmann, and A.Rigling. 2007. Radial growth responses to drought of Pinus sylvestris and Quercus pubescens in an inner-Alpine dry valley. Journal of Vegetation Science. 18: 777-792. Westerling, A.L., H.G. Hidalgo, D.R. Cayan and T.W. Swetnam. 2008. Warming and Earlier Spring Increase Western U.S. Forest Wildfire Activity. Science. 313: 940-943. Wilmking, M., G.P. Juday, V.A . Barber and H.S.J. Zaldw. 2004. Recent climate warming forces contrasting growth responses of white spruce at treeline in Alaska through temperature thresholds. Global Change Biology. 10: 1–13. Wilson, R.J.S. and B.H. Luckman. 2003. Dendroclimatic reconstruction of maximum summer temperatures from upper treeline sites in Interior British Columbia, Canada. Holocene. 13: 851-861. Woodall, C.W., C.M. Oswalt, J.A. Westfall, C.H. Perry, M.D. Nelson, A.O. Finley. 2009. An indicator of tree migration in forests of the eastern United States. Forest Ecology and Management. In press.  31  Yu, D.P., G.G. Wang, L.M. Dai and Q.L. Wang. 2007. Dendroclimatic analysis of Betula ermanii forests at their upper limit of distribution in Changbai Mountain, Northeast China. Forest Ecology and Management. 240: 105-113. Zhang, Q.B., R.I. Alfaro and R.J. Hebda. 1999. Dendroecological studies of tree growth, climate and spruce beetle outbreaks in Central British Columbia, Canada. Forest Ecology and Management. 121: 215-225. Zhang Q.B. and R.J. Hebda. 2004. Variation in radial growth patterns of Pseudotsuga menziesii on the central coast of British Columbia, Canada. Canadian Journal of Forest Research. 34: 1946-1954.  32  2. Validation of the Mountain Microclimate Simulation Model (MTCLIM) in Interior British Columbia, Canada3 2.1. Introduction As hydrological and ecological research on the potential impacts of climate change on ecosystems and processes continues, there is an increasing demand for reliable meteorological data (Thornton et al., 1997; Thornton and Running, 1999; Price et al., 2001; Hamann and Wang, 2004; McKeeny et al., 2006). However, due to the cost of building and maintaining long-term weather monitoring installations (Glassy and Running, 1994) we rarely have sufficient directly measured weather data for particular locations. As a consequence, when climate data are needed as inputs to run ecosystem models, it is often necessary to use data extrapolated from the nearest weather station to the site of interest (Thornton et al., 1997; McKeeny et al., 2006). In general, we know that as elevation increases, temperature decreases; the widely-used temperature/elevation lapse rate is 6.4 °C per 1000 m (Dodson and Marks, 1997; Barry and Chorley, 1998; Lookinbill and Urban, 2003), and precipitation generally reflects an equivalent lapse rate, with more rain and snow at higher than at lower elevations. However, topography, slope, aspect, land cover, etc., can have an influence on these simple relationships (Creed et al., 1996; Barry and Chorley, 1998). For example, the variation of temperature at a site near a lake will be less than for a site near a desert because lakes have a buffering effect on temperature due to their large heat storage capacity (Dodson and Marks, 1997). Statistical methods have been developed to extrapolate climate variables (e.g. temperature and precipitation) from local weather stations to sites of interest. These include inverse-distance methods, optimal interpolation procedures such as kriging or smoothing splines (Custer et al., 1996; Thornton et al., 1997; Hamann and Wang, 2004). The advantages of these methods are that they are easy to calculate, if the area has similar topography and a number of weather stations that can be cross-referenced, results are quite reliable, and calculations do not require much input information (Thornton et al., 1997; McKeeny et al., 2006). One drawback is that these methods require assumptions about the distribution of the climate variables of interest and most of the 3  A version of this chapter will be submitted for publication as Lo Y.-H., Blanco J.A., Seely B., Welham C., Kimmins J.P. “Validation of the Mountain Microclimate Simulation Model (MTCLIM) in Interior British Columbia, Canada”. 33  time these calculations are dependent on the weight function used for extrapolation. However, in many cases topography is complex and there are few weather stations to provide input data. Hamann and Wang (2005) pointed out that an unbalanced distribution of sample stations can be problematic. Also, without considering local factors such as slope, aspect, elevation and land cover, these calculations also fail to generate reliable climate data for those areas where the topography is complicated (Custer et al., 1996; Almeida and Landsberg, 2003). In these cases, extrapolation from local climate stations is best achieved using regional climate models (RCMs). Several climate models have been developed and compared for this purpose. Two of these models are PRISM (Parameter-elevation Regressions on Independent Slopes Model, Daly and Neilson, 1992) and MTCLIM (MounTain microCLIMate simulation model; Running et al. 1987) both of which are widely used in North America and overseas (Glassy and Running, 1994; Thornton et al., 1997; Thornton et al., 2000; Li et al., 2001; Daly et al., 2002; Almeida and Landsberg, 2003; Hunter and Meentemeyer, 2005; Cienciala and Tatarinov, 2006). Another model is ANUSPLIN (Australian National University SPLINe routine, Hutchinson and Bischof, 1983), which was originally developed in Australia but is now also applied in North America (Custer et al., 1996; McKeeny et al., 2006). All these models can be used at the landscape scale to generate climate data. However, for stand level or daily time step simulations, MTCLIM is a better choice among the three because it has been designed for specific site simulation at daily time steps. There is also one local model that is used in British Columbia: MMFCLiM (Benton, 1997). It is a GIS-based monthly climate model and is used in the McGregor Model Forest in Prince George (British Columbia, Canada). Because MTCLIM is widely used and its temporal and spatial scales fit my own research requirements, I chose this model as my climate generation tool in this research. MTCLIM has been applied successfully in several countries (Li et al., 2001; Almeida and Landsberg, 2003, and references above), but has never been tested in the semi-arid conditions of the Okanagan Valley (interior B.C.). Hence, before using MTCLIM as the climate generator for my dendroclimatology work, I first carried out a validation exercise to verify its appropriateness for this region. It is important to note, however, that all models are simplified, incomplete representations of reality (Kimmins, 2004) and to validate a model is not necessarily to prove that it is ‘correct’ (Oreskes et al., 1994; Sterman, 2002; Oreskes, 2003). In many cases, model validation is an exercise to show that predictions are close enough to independent empirical data 34  to make them useful for specific and practical applications, and that decisions based on model output are defensible (Popper, 1963; Soares et al., 1995; Rykiel, 1996). The appropriateness of different methods for model validation has engendered considerable discussion (Gardner and Urban, 2003) but there is general agreement that predictions should be tested against independent data (Aber, 1997). In a strict sense, a given model evaluation or validation is acceptable only for the conditions under which it was conducted. Nevertheless, the greater the number of cases of good agreement between observed and predicted values, the greater the confidence in the model (see Oreskes et al. (1994) and Rykiel (1996) for further discussion).  2.2. Model description MTCLIM (Figure 2.1, Running et al., 1987) is a weather data generating model which uses daily data from a reference weather station. Input variables include daily maximum and minimum temperature and precipitation data, and geographic information for the reference and target location (i.e. elevation, slope, aspect, latitude and albedo) to extrapolate temperature and moisture regimes on the basis of lapse rate and precipitation isohyets, respectively (Thornton et al., 1997).4 Based on the input data the model produces daily maximum and minimum temperature, daily mean temperature, daily precipitation, vapour pressure deficit, daily solar radiation and the length of day (Coughlan and Running, 1997; Kimball et al., 1997; Thornton and Running, 1999; Chiesi et al., 2002).  4  One important requirement of this model is that the input weather data can not have missing values. 35  Figure 2. 1. Flow chart showing the steps followed by MTCLIM to estimate daily microclimate data in mountainous terrain. The inputs required by the model are listed in site factor and base station categories. (Adapted from Running et al., 1987)  2.3. Methodology to test the performance of MTCLIM in the research area I tested the predictions from MTCLIM using three pairs of weather stations in the Kamloops forest region (Lloyd et al., 1990). These stations were selected because they are close to my tree ring research experimental site (see Chapter Four) and each one contains a representation of both high and low elevation climates. The low elevation site (the reference site) was an input into MTCLIM and the high elevation site (the target site) was used to test the model simulation ability. These pairs were Hedley vs. Hedley NP Mine, Vernon vs. Vernon Silver Star Lodge, and McCulloch vs. Big White. The locations (Figure 2.2) and simulation years for these weather stations are listed in Table 2.1. All stations are located in the southern very dry zone of the Kamloops forest region (Lloyd et al., 1990). First, because of the intolerance of MTCLIM to 36  missing data within each year, I examined the recorded weather data of the lower elevation weather stations (i.e. Hedley, Vernon and McCulloch) and eliminated the years which had too many missing values and were unable to fill the gap by estimation. Then I used MTCLIM to generate daily maximum, minimum and mean temperatures, daily precipitation, vapour pressure deficit, solar radiation and day length for each of the corresponding high elevation site in each pair. I used the following lapse rates for temperature: 6 °C /1000 m for maximum temperature and 4 °C /1000 m minimum temperature; annual precipitation isohyets of 78.3 cm, 69.4 cm and 88.5 cm were used in Hedley Mine, Silver Star and Big White, respectively (Dodson and Marks, 1997; Barry and Chorley, 1998; Lookinbill and Urban, 2003). The methodology to calculate lapse rate is described in Appendix A. Because precipitation is a discontinuous event or series of events and therefore daily predictions are difficult to compare to recorded events, I summed the daily precipitation into monthly values. The high elevation recorded data were then compared with the high elevation predictions (extrapolated by MTCLIM from the low elevation data) using frequency histograms and regression analysis. I also calculated several goodness-of-fit indices that compare observed and predicted values to evaluate how well the model performed. These indexes were: Pearson’s correlation coefficient (r), coefficient of determination (r2, i.e. the amount of variability in observed values accounted for by the linear model), mean error (MER), mean absolute error (MAE), modeling efficiency (ME) and Theil’s inequality coefficient (U). Mean error (MER) was calculated to identify the overall directionality of bias:  MER =  1 n ∑ ( Pi − Oi ) n i =1  where n is the number of pairs, Pi is the ith predicted value and Oi is the ith observed value. Positive and negative MER values indicate over- and under-prediction, respectively. Mean absolute error (MAE) was computed to determine overall magnitude of error:  MAE =  1 n ∑ ( Pi − Oi ) n i =1  37  Modelling efficiency was calculated as defined by Vanclay and Skovsgaard (1997) as: n  ∑D  2 i  i =1  ME = 1 −  n  ∑ (O  i  − P i )2  i =1  where Di = Oi – Pi. This statistic provides a simple index of performance on a relative scale, where  ME = 1 indicates a perfect fit, ME = 0 indicates that the model is no better than a simple average of the prediction values, while negative values indicate poor model performance. Theil’s inequality coefficient (Theil, 1966) is another index to exam the model performance.  n  ∑D  2 i  U=  i=q n  ∑O  2 i  i =1  U can assume values of 0 and greater. If U = 0 then the model produces perfect predictions. If U = 1 the model produces predictions of system behaviour that are not better than a no-change prediction. If U > 1, then the predictive power of the model is worse than the no-change prediction.  Table 2. 1. Geographic information for the weather stations used to test MTCLIM5. Station  Latitude  Longitude Elevation (m) Record year  Simulation years 1909-2001 (59 years)  Hedley  N 49.357° W120.077°  517  1904 - 2002  Hedley NP Mine  N 49.369° W 120.022°  1,707  1904 - 2002  Vernon  N 50.233° W 119.283°  556  1920 - 2000  Vernon Silver Star Lodge N 50.358° W 119.056°  1,572  1971 - 2002  Mcculloch  N 49.800° W 119.200°  1250  1936 - 1993  Big White  N 49.733° W 118.933°  1,841  1981 - 1999  5  1972 – 2000 (27 years) 1971 – 1993 (20 years)  Because of the intolerance of MTCLIM, the actual simulation years are less than the difference between starting and ending simulation years. 38  Figure 2. 2. Location of the weather stations used to test MTCLIM in the Okanagan Valley area (Interior B.C.). Westwold weather station was not part of this test but it was used in Chapter three. (Google Inc., 2009)  39  Table 2. 2. Monthly temperature and precipitation summary of the testing sites. (Environment Canada, 2008). Site  Hedley (1971-2000) Hedley Mine3 1909-2001 (59 years) Vernon (1971-2000) Silver Star3 1972 – 2000 (27 years) Mcculloch (1971-2000) Big White3 1971 – 1993 (20 years)  Variable  Jan  Feb  Mar  Apr  May  Jun  Jul  Aug  Sep  Oct  Nov  Dec  Year  Daily Average (°C)  -4  -0.5  4.4  8.9  13.2  16.8  19.9  19.7  14.7  8.1  1.4  -3.5  8.3  Precipitation (mm)  34.3  20.1  19.6  25.5  40.8  44.5  38.1  36.7  26.1  22.2  33.1  35.9  376.8  Daily Average (°C)  -7.42  -5.41  -2.92  2.87  5.48  8.81  12.60  12.34  8.99  3.37  -3.00  -6.12  2.47  Precipitation (mm)  51.09  43.31  38.55  40.64  63.27  69.23  49.16  46.28  33.67  36.41  47.97  55.42  575.01  Daily Average (°C)  -4.2  -1.2  3.8  8.5  13  17  19.7  19.6  14.3  7.9  1.2  -3.2  8.1  Precipitation (mm)  32.8  26.2  26.9  27.7  40.1  42.4  37.5  33.8  32.9  26.6  40.4  42.7  409.9  Daily Average (°C)  -6.14  -5.25  -2.71  1.41  4.59  9.76  14.04  13.72  10.24  1.94  -3.14  -6.96  2.62  Precipitation (mm)  112.11  95.67  87.97  46.83  12.75  49.40  40.60  36.41  41.47  75.16  101.05  134.25  833.67  Daily Average (°C)  -7.7  -5.3  -1.9  2.5  7.1  10.5  13.1  13  8.7  3.5  -3.2  -7.4  2.74  Precipitation (mm)  72.8  66.2  54.2  48.7  64.4  78.2  54.7  53.7  45.8  40.4  62.1  86.6  727.8  Daily Average (°C)  -7.01  -5.345  -3.02  -0.36  -  -  -  -  -  -  -5.09  -8.45  -  Precipitation (mm)  141.55  97.43  81.30  41.86  -  -  -  -  -  -  105.75  149.90  -  40  2.4. Results Monthly temperature and precipitation summary of all sites is shown in Table 2.2. Tables 2.3 to 2.5 present indexes of model performance of comparing recorded and simulated monthly maximum temperature, minimum temperature and monthly total precipitation for the three target sites (Hedley NP Mine, Vernon Silver Star Lodge and Big White). For predictions of daily maximum temperature, MTCLIM performance showed seasonal variation. In winter MTCLIM monthly estimates were lower (between 0.01 to 5.27 °C) than the locally recorded data. From spring to fall the model’s simulated values for maximum temperature were higher (between 0 to 4.62 °C) than the recorded data. As for daily minimum temperature predictions, MTCLIM predicted values higher than observed data. Predicted monthly precipitation values were significantly lower than recorded values at all three sites for all months except May and July precipitation in Silver Star. Table 2. 3. Descriptive statistics of model performance for the simulation of average monthly maximum temperatures for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White (n = 12).  Simulation years  Hedley NP Mine  Vernon Silver Star Lodge  Big White  1909-2001 (59 years)  1972 – 2000 (27 years)  1971 – 1993 (20 years)  record Annual T (Average ±SD)  simulation  record  simulation  7.73 ± 9.45 6.92 ± 11.42 5.24 ± 9.92 5.98 ± 10.95  record  simulation  6  5.57 ± 10.24  -  r  0.99  0.99  0.98  2  0.97  0.99  0.95  MER (°C)  0.50  0.20  -0.16  MAE (°C)  2.43  1.52  1.61  ME  0.87  0.95  0.94  Theil’s U  0.27  0.18  0.21  r  Pearson’s correlation coefficients between the observed data and the simulated data for maximum and minimum temperatures at all three sites had high values (all r values are larger than 0.90). For precipitation r values were also very high for the Hedley data pair (r = 0.83) but not for McCulloch and Vernon data pairs (r = 0.31 and 0.41, respectively). The coefficients of determination (r2) which determines the proportion of variance explained were also quite high for both maximum and minimum temperature simulations (i.e. all r2 > 0.9), with slightly lower values in Big White. For precipitation predictions, All the  6  No years with full 12 months records were available at Big White. 41  variances from three weather stations explained by model simulation results were moderate to low (i.e. r2 = 0.69, 0.10 and 0.28 for Hedley Mine, Silver Star and Big White, respectively). Table 2. 4. Descriptive statistics of model performance for the simulation of average monthly minimum temperature for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White. (n = 12).  Simulation years  Hedley NP Mine  Vernon Silver Star Lodge  Big White  1909-2001 (59 years)  1972 – 2000 (27 years)  1971 – 1993 (20 years)  record Annual T (Average ±SD)  simulation  record  simulation  -2.50 ± 7.89 -2.53 ± 8.05 -1.89 ± 8.31 -0.94 ± 7.91  record  simulation  4  -6.00 ± 7.99  -  r  0.99  0.99  0.95  2  r  0.98  0.98  0.91  MER (°C)  -0.12  -0.23  3.15  MAE (°C)  0.99  0.80  3.15  ME  0.97  0.97  0.73  Theil’s U  0.16  0.16  0.53  Table 2. 5. Descriptive statistics of model performance for the simulation of monthly total precipitation in Hedley NP Mine, Vernon Silver Star Lodge and Big White (n = 12).  Simulation years Annual precipitation (Average ±SD)  Hedley NP Mine  Vernon Silver Star Lodge  Big White  1909-2001 (59 years)  1972 – 2000 (27 years)  1971 – 1993 (20 years)  record 521.43 ± 134.70  record 704.52 ± 225.70  simulation 713.83 ± 183.11  simulation 789.37 ± 152.35  record --4  simulation 1237.88 ± 212.51  r  0.83  0.31  0.53  2  0.69  0.10  0.28  MER (mm)  -11.25  7.14  9.26  MAE (mm)  11.91  22.24  24.57  ME  0.90  0.86  0.92  Theil’s U  0.27  0.35  0.27  r  Under different model performance indices, MTCLIM simulation bias (MER) for maximum temperature ranged from -0.16 to 0.50 °C. For minimum temperatures, the model simulation bias was from -0.12 to 3.15 °C. As for precipitation, the monthly biases were from -11.25 to 9.26 mm. These are about 1 to 2 percent biases compared with annual total precipitation. The mean absolute errors (MAE) showed that the data generated by the model had higher variation for predicted maximum temperate than minimum temperature with the 42  exception of minimum temperature at Big White. The MAE for precipitation ranged from 11.91 to 24.57 mm. Both model efficiency and Theil’s U index showed that model performed well and better than the extrapolation of a simple average for temperature. Similarly, in the precipitation part, both indices showed that the model performed well even though there were some biases. Figures 2.3 to 2.5 compare monthly relative frequency histograms of observed and predicted climate variables (i.e. daily maximum, minimum temperature and monthly precipitation) at Hedley Mine. The distributions of observed and simulated data were similar for both daily maximum and minimum temperature, but the means were not. During cold months (i.e. November to February), the model predicted results were lower than observed data while in warm months (i.e. April to August), the model’s predictions were opposite compared with cold months results. Depending on the time of the year, the shift directions were similar but the relative differences were not. Precipitation data were not predicted as well as that for temperature. It also showed that there might be a simulation bias in MTCLIM or an effect of the small sample size. The same differences appeared in Vernon Silver Star lodge and Big White results (Figures 2.6 to 2.11), and there, the difference between observed and simulated precipitation distributions was even greater. A regression analysis was also done to determine the monthly relationships between observed and simulated data. For Hedley Mine (Figures 2.12 and 2.13) and Vernon Silver Star lodge (Figures 2.15 and 2.16), both daily maximum temperature and minimum temperature showed significant positive linear relationships between the simulated and observed data. However, for daily maximum temperature, the highest r2 occurred in summer while for daily minimum temperature the highest r2 occurred in winter. As for precipitation (Figures 2.14 and 2.17), the highest r2 occurred in summer at Hedley but no pattern was found in Silver Star. The regression results in Figures 2.18 and 2.19 showed that MTCLIM was successful at simulating temperature at Big White with high r2 but not for precipitation. It did well at simulating results in some months (e.g. March and December) while at other months it simulated poorly (e.g. February and April). However, the relationships between the predicted and observed data were better described by a linear regression after pooling monthly precipitation data together, as showed below (figures in Appendix A):  Hedley Mine:  Predicted P = 13.93 + 1.08 × Observed P  r2 = 0.50, n =303  Silver Star:  Predicted P = 31.93 + 0.44 × Observed P  r2 = 0.30, n = 166  Big White:  Predicted P = 70.06 + 0.44 × Observed P  r2 = 0.23, n = 52  43  HEDLEY MINE TMAX  Figure 2. 3. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed (grey areas) and simulated (black line) data at Hedley Mine for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  44  HEDLEY MINE TMIN  Figure 2. 4. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed (grey areas) and simulated (black line) data at Hedley Mine for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  45  HEDLEY MINE P  Figure 2. 5. Monthly relative-frequency distribution histogram of daily precipitation comparing observed (grey area) and simulated (black line) data at Hedley Mine for the years given in Table 2.1. Solid line indicates the average value of observed records. Broken line indicates the average value of MTCLIM projections. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in mm) of observed records. SDM stands for value of standard deviation (in mm) of MTCLIM projections.  46  SILVER STAR TMAX  Figure 2. 6. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed (grey areas) and simulated (black line) data at Silver Star for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  47  SILVER STAR TMIN  Figure 2. 7. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed (grey areas) and simulated (black line) data at Silver Star for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  48  SILVER STAR P  Figure 2. 8. Monthly relative-frequency distribution histogram of daily precipitation comparing observed (grey area) and simulated (black line) data at Silver Star for the years given in Table 2.1. Solid line indicates the average value of observed records. Broken line indicates the average value of MTCLIM projections. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in mm) of observed records. SDM stands for value of standard deviation (in mm) of MTCLIM projections.  49  BIG WHITE TMAX  Figure 2. 9. Monthly relative-frequency distribution histograms of daily maximum temperature comparing observed (grey areas) and simulated (black line) data at Big White for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  50  BIG WHITE TMIN  Figure 2. 10. Monthly relative-frequency distribution histograms of daily minimum temperature comparing observed (grey areas) and simulated (black line) data at Big White for the years given in Table 2.1. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in °C) of observed records. SDM stands for value of standard deviation (in °C) of MTCLIM projections.  51  BIG WHITE P  Figure 2. 11.. Monthly relative-frequency distribution histogram of daily precipitation comparing observed (grey area) and simulated (black line) data at Big White for the years given in Table 2.1. Solid line indicates the average value of observed records. Broken line indicates the average value of MTCLIM projections. Solid vertical lines indicate the average value of observed records. Broken vertical lines indicate the average value of MTCLIM projections. SDo stands for value of standard deviation (in mm) of observed records. SDM stands for value of standard deviation (in mm) of MTCLIM projections.  52  Figure 2. 12. Monthly regressions of predicted (y axis) on observed (x axis) daily maximum temperature at Hedley Mine recorded between 1909 and 2001. Also shown are the regression line and its equation. All regressions were significant at P < 0.05.  53  Figure 2. 13. Monthly regressions of predicted (y axis) on observed (x axis) daily minimum temperature at Hedley Mine recorded between 1909 and 2001. Also shown are the regression line and its equation. All regressions were significant at P < 0.05.  54  Figure 2. 14. Monthly regressions of predicted (y axis) on observed (x axis) monthly total precipitation at Hedley Mine recorded between 1909 and 2001. Also shown are the regression line and its equation. A star ‘*’ before the equation stands for significance at level P < 0.05.  55  Figure 2. 15. Monthly regressions of predicted (y axis) on observed (x axis) daily maximum temperature at Silver Star recorded between 1972 and 2000. Also shown are the regression line and its equation. All regressions were significant at P < 0.05.  56  Figure 2. 16. Monthly regressions of predicted (y axis) on observed (x axis) daily minimum temperature at Silver Star recorded between 1972 and 2000. Also shown are the regression line and its equation. All regressions were significant at P < 0.05.  57  Figure 2. 17. Monthly regressions of predicted (y axis) on observed (x axis) monthly total precipitation at Silver Star recorded between 1972 and 2000. Also shown are the regression line and its equation. A star ‘*’ before the equation stands for significance at level P < 0.05.  58  Figure 2. 18. Monthly regressions of predicted (y axis) on observed (x axis) daily maximum temperature (left) and daily minimum temperature (right) at Big White recorded between 1971 and 1993. Also shown are the regression line and its equation. All regressions were significant at P < 0.05.  59  Figure 2. 19. Monthly regressions of predicted (y axis) on observed (x axis) monthly total precipitation at Big White recorded between 1971 and 1993. Also shown are the regression line and its equation. A star ‘*’ before the equation stands for significance at level P < 0.05.  60  2.5. Discussion The capability of MTCLIM to simulate temperature in central Okanagan Valley was generally good. The model performance at a yearly time scale showed the correlation coefficients were all above 0.95. Compared to MTCLIM used in other areas (Almeida and Landsberg, 2003), the r2 values presented here are higher. This fact indicates that a high proportion of the variance in the simulation results is explained by observed data and it also indicates that MTCLIM performs well in the Okanagan Valley region. The biases between the simulated and observed data were from -0.16 to 0.50 °C for daily maximum temperature and –0.12 to 3.15 °C for daily minimum temperature. These biases are relatively small compared with other studies (Benton, 1997; Hamann and Wang, 2005; McKeeny et al., 2006). The range of values to compare results from a preliminary study (Appendix A) also indicated that as the distance between the base and the target stations increased so did the bias, though the range was still within the acceptable limits (see Dodson and Marks, 1997; McKeeny et al., 2006). Two other indices of the model performance also indicated that the model’s predictions were good. Most of the model efficiency (ME) values were above 0.8 and most of the Theil’s U indices were close to zero, with the exception of minimum temperature at Big White. This demonstrated that MTCLIM’s predictions were better when weather stations were closer geographically and when longer data series were used. The shapes of the distributions were similar with similar standard deviations but with slightly different means. Among the three experimental sites, both Hedley Mine and Vernon Silver Star Lodge had similar monthly patterns. I also found that as the number of simulated years increased, the shapes of the distributions of observed and simulated data became less biased. This is a common occurrence for climate-generating models (McKeeny et al., 2006) because the more information used in the simulation, the better the model is parameterized thereby generating more accurate predictions. MTCLIM performed reasonably well at predicting climate on a monthly time scale. Monthly data are important in broad scale ecological studies like this one, because tree growth responds to monthly temperature and precipitation during average weather conditions (see Chapter Three for detailed discussion on this topic). However, some daily extreme weather events like frost or heat waves can affect tree growth. The distribution of predicted daily values for each month was very similar to the climate records, although some systematic bias was evident. In winter (October to 61  February), the model’s predicted values were lower than the recorded values for daily maximum temperature. This was also found for minimum temperature but the difference was not as large. On the other hand, from March to August, the model’s simulated values were higher than recorded values for both maximum and minimum temperatures. One possible reason is that these weather stations are close to the Okanagan Lake (Figure 2.1), and lakes usually have a buffering effect on temperature due to their capacity to store latent heat, the lakes release heat in the winter making air warmer and they store heat in the summer reducing air temperature (Barry and Chorley, 1998; Kimmins, 2004; McKeeny et al., 2006). MTCLIM did not predict precipitation as reliably as temperature. This is a concern since precipitation is a critical component of all climate models. Rainfall patterns are more difficult to simulate because they vary more on spatial and temporal scales than temperature and they are not as strongly related to altitude as is temperature (Chiesi et al., 2002). However, in general, in the Okanagan Valley the bottom of the valley is drier than the mountain tops (Lloyd et al., 1990). Precipitation can also be markedly affected by air masses and other atmosphere circulation cycles (e.g. storms, El Niño and La Niña events, etc.). Even though the monthly simulation biases of the model were about one to two percent (-11.25 to 9.26 mm) of the annual average, the biases were within acceptable range as compared to the PRISM model and the model of Rehfeldt et al. (1999, 2001), whose deviations of predicted mean annual precipitation were typically higher, around 100 mm and 400 mm, respectively (Hamann and Wang, 2005). Although the r2 values were lower than desirable, they were acceptable (except for spring values, in the pairs without long data record) and were higher than found in another MTCLIM study (Chiesi et al., 2002). Meanwhile, just as for temperature, as the sample size (number of years simulated) increased, the simulation ability also increased (Chiesi et al., 2002). However, as pointed out by Cienciala and Tatarinov (2006), even with careful selection of the input parameters (i.e., temperature and precipitation), the bias in precipitation part is still large (Chiesi et al., 2002). These results are consistent with previous research (Thornton et al., 1997). One important distinction though is that I carried out the analysis for a longer time scale using only a small number of stations, whereas typically a larger number of stations are used on relatively short time scales (a year or less). The three pairs of stations represented three different geographic conditions. The Hedley Mine pair is the most ideal pair which has long-term climate data and the two weather stations 62  are close together. It is also located in the driest part of the Okanagan Valley, which has an increasing rainfall gradient form south to north (Table 2.2; Lloyd et al., 1990). The Big White pair represents a relatively short climate record and the two weather stations are far apart. Although this situation is not desirable, it is common in most of studies that the weather station is not close to the experimental site. In the case of Silver Star, the pair has the shortest climate data record of my study, but the two stations are close to each other. The above results showed that in the Okanagan region MTCLIM predicted temperature much better than precipitation. This is not surprising because in dry regions, temperature but especially precipitation are more variable than in wet regions (Dodson and Marks, 1997; Barry and Chorley, 1998; Kimball et al., 1997). However, the good performance of MTCLIM in temperature prediction is also positive for predicting precipitation because in dry regions it is important to accurately estimate minimum temperatures to calculate dew point (used in the precipitation algorithm) and precipitation events are more critical than in humid areas (Kimball et al., 1997). Finally, the bias in precipitation is still in the acceptable range. Besides, Glassy and Running (1994) pointed out that in ecological model simulation, the patterns of the climate data are more important than the exact values of the variables. In the present study, MTCLIM produced a consistent pattern between the observed and simulated data. All things considered, my results showed that if we have an adequate sample size and proper input parameters, especially the lapse rate and precipitation isohyet, MTCLIM can provide quite reliable weather data for its use in mountainous terrain. For this reason, I conclude that it was suitable for my dendroclimatological research to use MTCLIM as a climate generator for the sites where I collected the tree cores (see Chapter Three). I also consider MTCLIM as a suitable weather model for similar studies to be carried out in this area.  63  2.6. References Aber, J.D. 1997. Why don't we believe the models? Bulletin of the Ecological Society of America. 78: 232-233. Almeida A.C. and J.J. Landsberg. 2003. Evaluating methods of estimating global radiation and vapor pressure deficit using a dense network of automatic weather stations in coastal Brazil. Agricultural and Forest Meteorology. 118: 237-250. Barry, R.G. and R.J. Chorley. 1998. Atmosphere, Weather and Climate. 7th Ed. Routledge. New York. NY. U.S.A. Pp. 409. Benton, R.A. 1997. MMFCLiM: A GIS Based Monthly Climate Model for the McGregor Model Forest: Theory & Model Description v0.9. Report prepared for McGregor Model Forest Association. Prince George, BC, Canada. Pp56. Chiesi, M., F. Maselli, M. Bindi, L. Fibbi, L. Bonora, A. Raschi, R. Tognetti, J. Cermak and N. Nadezhdina. 2002. Calibration and application of FOREST-BGC in a Mediterranean area by the use if conventional and remote sensing data. Ecological Modelling. 154: 251-262. Cienciala, E and F.A. Tatarinov. 2006. Application of BIOME-BGC model to managed forests 2. Comparison with long-term observations of stand production for major tree species. Forest Ecology and Management. 237: 252-266. Coughlan, J.C. and S.W. Running. 1997. Regional ecosystem simulation: A general model for simulating snow accumulation and melt in mountainous terrain. Landscape Ecology. 12: 119-136. Creed, I.F., L.E. Band, N.W. Foster, I.K. Morrison, J.A. Nicolson, R.S. Semkin and D.S. Jeffries. 1996. Regulation of nitrate-N release from temperate forests: A test of the N flushing hypothesis. Water Resources Research. 32: 3337-3354. Custer, S.G., P. Farnes, J.P. Wilson and R.D. Snyder. 1996. A comparison of hand and spline-drawn precipitation maps for mountainous Montana. Water Resources Bulletin. 32: 393-405. Daly, C., W.P., Gibson, G.H. Taylor, G.L. Johnson, and P. Pasteris. 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research. 22: 99-113. Daly, C. and R.O. Neilson. 1992. A digital topographic approach to modeling the distribution of precipitation in mountainous terrain. In: Interdisciplinary Approaches in Hydrology and Hydrogeology. American Institute of Hydrology. p. 447-454. Dodson, R. and D. Marks. 1997. Daily air temperature interpolated at high spatial resolution over a large mountainous region. Climate Research. 8: 1-20. Environment Canada. 2008 National Climate Data and Information Archive. Available at: http://climate.weatheroffice.ec.gc.ca/Welcome_e.html Retrieved January 6, 2009. Gardner, R.H. and D.L. Urban. 2003. Model validation and testing: past lessons, present concerns, future prospects. In: Models in ecosystem science. C.D. Canham, J.J. Cole, and W.K. Lauenroth. Princeton University Press, Princeton, NJ. U.S.A. p. 184-203. Glassy, J.M. and S.W. Running. 1994. Validating diurnal climatology logic of the MT-CLIM model across a climatic gradient in Oregon. Ecological Applications. 4: 248-257. Google Inc. 2009. Google Maps. Available at: http://maps.google.ca. Retrieved January 6, 2009.  64  Hamann, A. and T.L. Wang. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology. 128: 211-221. Hunter, R.D. and R.K. Meentemeyer. 2005. Climatologically aided mapping of daily precipitation and temperature. Journal of Applied Meteorology. 44: 1501-1510. Hutchinson, M.F. and R.J. Bischof. 1983. A new method for estimating the spatial distribution of mean seasonal and annual rainfall applied to the Hunter Valley, New South Wales. Australian Meteorological Magazine. 31: 179-184. Kimball, J.S., S.W. Running and R. Nemani. 1997. An improved method for estimating surface humidity from daily minimum temperature. Agricultural and Forest Meteorology. 85: 87-98. Kimmins, J. P. 2004. Forestry Ecology: A Foundation for Sustainable Management and Land ethics.3rd ed. Printice Hall, Upper Saddle River, NJ. U.S.A. Li, H.-T., W.-Q. Shen and W.-G. Sang. 2001. Research situation and application of MTCLIM model. Journal of Mountain Science. 19: 533-540. Lloyd, D., K. Angove, G. Hope and C. Thompson. 1990. A Guide to Site Identification and Interpretation for the Kamloops Forest Region. British Columbia Ministry of Forests. Victoria. BC. Canada. Pp 407. Lookingbill, T.R. and D.L. Urban. 2003. Spatial estimation of air temperature differences for landscape-scale studies in montane environments. Agricultural and Forest Meteorology. 114: 141-151. McKeeny, D.W., J.H. Pedlar, P. Papadopol and M.F. Hutchinson. 2006. The development of 1901-2000 historical monthly climate models for Canada and the United States. Agricultural and Forest Meteorology. 138: 69-81. Oreskes, N. 2003. The role of Quantitative models in science. In: C.D. Canham, J.J. Cole and W.K. Lauenroth. (Eds). Models in ecosystem science. Princeton University Press, Princeton, NJ. U.S.A. p. 13-31. Oreskes, N., K. Shrader-Frechette and K. Belitz. 1994. Verification, validation and confirmation of numerical models in the Earth Sciences. Science. 263: 641-646. Popper, K.R. 1963. Conjetures and refutations. Routledge & Kegan Paul. London, UK. Price, D.T., N.E. Zimmermann, P.J. Van der Meer, M.J. Lexer, P. Leadley, I.T.M. Jorritsma, J. Schaber, D.F. Clark, P. Lasch, S. McNulty, J. Wu and E. Smith. 2001. Regeneration in gap models: Priority issues for studying forest responses to climate change. Climate Change. 51: 475-508. Rehfeldt, G.E., W.R. Wykoff and C.C. Ying. 2001. Physiologic plasticity, evolution, and impacts of a changing climate on Pinus contorta. Climatic Change. 50: 355-376. Rehfeldt, G.E., C.C. Ying, D.L. Spittlehouse and D.A. Hamilton. 1999. Genetic responses to climate in Pinus contorta: niche breadth, climate change, and reforestation. Ecological Monographs. 69: 375-407. Running, S.W., R.R. Nemani and R.D. Hungerford. 1987. Extrapolation of synoptic meteorological data in mountainous terrain and its use for simulating forest evapotranspiration and photosynthesis. Canadian Journal of Forest Research. 17: 472-483. Rykiel, E.J. 1996. Testing ecological models: the meaning of validation. Ecological Modelling. 90: 229-244.  65  Soares, P., M. Tomé, J.P. Skovsgaard and J.K. Vanclay. 1995. Evaluating a growth model for forest management using continuous forest inventory data. Forest Ecology and Management. 71: 251-265. Sterman, J. 2002. All models are wrong: reflections on becoming a systems scientist. Systems Dynamics Reviews. 18: 501-531. Theil, H. 1966. Applied econometric forecasting. North-Holland, Amsterdam, The Netherlands Pp.474. Thornton, P.E, H. Hasenauer and M.A. White. 2000. Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: an application over complex terrain in Austria. Agriculture and Forest Meteorology. 104: 255-271. Thornton, P.E. 2000. General notes for MTCLIM version 4.3. Numerical Terradynamic Simulation Group. School of Forestry, University of Montana. MN. U.S.A. Thornton, P.E., S.W. Running and M.A. White. 1997. Generating surface of daily meteorological variables over large regions of complex terrain. Journal of Hydrology. 190: 214-251. Thornton, P.E. and S.W. Running. 1999. An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity and precipitation. Agricultural and Forest Meteorology. 93: 211-228. Vanclay, J.K. and J.P. Skovsgaard. 1997. Evaluating forest growth models. Ecological Modelling. 98: 1-12.  66  3. Relationships between Climate and Tree Radial Growth in interior British Columbia, Canada7 3.1. Introduction Originally developed as a tool for archaeologists in the early 20th century (Fritts, 1976), dendroclimatology has developed greatly in the last few years (Beniston, 2002; Bräker, 2002) and has been used widely in reconstructing historical climate patterns (Alfsen, 2001; IPCC, 2001; Gajewski and Atkinson, 2003; Cook et al., 2003; Hegerl et al., 2007; Sarris et al., 2007). The basic approach is to construct a tree ring master chronology, which represents the average growth response in the sampled area. The master chronology is then matched to the climatic factor(s) of interest. Dendrochronology has also been used as a tool to explore the ecosystem response to climate change (Yu et al., 2007; Chhin et al., 2008). For example, a number of studies have examined tree ring response to temperature and precipitation. Zhang et al., (1999) used different conifer species to identify regional climatic anomalies and insect outbreaks (e.g. spruce beetle) for the past four centuries in central British Columbia. Zhang and Hebda (2004) used Douglas-fir to explore the radial growth of trees in mountainous areas. They found that growing-season precipitation influenced growth for both high- and low-elevation Douglas-fir, but temperature played different roles at different elevations. Watson and Luckman (2001a,b) summarized the dendroclimatology research in the south Canadian Rockies and used multiple regression models to reconstruct annual precipitation (previous August to current July or previous July to current June) for Banff and Jasper (Alberta) and Cranbrook (B.C.). Luckman and Wilson (2003, 2005) also reconstructed past summer temperatures in the Canadian Rockies. Bower et al. (2005) used ring width, density and mass components in conjunction with regression analysis of growing season soil moisture deficit to build a drought response coefficient for Douglas-fir, although the effect of drought was limited to very dry sites. Peterson and others conducted a series of studies in the North Cascades National Park, Washington (Holman and Peterson, 2006; Case and Peterson, 2005, 2007). They examined the 7  A version of this chapter will be submitted for publication as Lo, Y.-H., Blanco, J.A. Seely, B., Welham, C., Kimmins, J.P. “Relationships between Climate and Tree Radial Growth in interior British Columbia, Canada”. 67  effects of climatic variability during the 20th century on the growth of lodgepole pine (Pinus contorta) and Douglas-fir (Pesudotsuga menziesii) along an elevational gradient. Multivariate analysis and correlation analysis were used to identify the relationship between climate and growth, and a factorial analysis to separate their samples into mid-elevation and high-elevation chronologies; the chronologies at each elevation responded to different climate factors. The high-elevation plots responded positively to annual temperature while the mid-elevation plots responded negatively to growing season maximum temperature but positively to growing season precipitation. Chhin et al. (2008) used lodgepole pine in Alberta to examine the relationships between climate and growth across elevational sequences of ecoregions and for different diameter size classes. They found annual growth of lodgepole pine was generally sensitive to heat and moisture stress in late summer of the previous year, the degree of winter harshness, and the timing of the start of the growing season. However, a criticism of traditional dendrochronological studies is that, most of the research has been done either at tree line or on arid sites, because they are the locations where climate clearly acts as the growth limiting factor (Begon et al., 2006). This could be a biased sampling strategy if we want to understand the growth / climate relationship at the population and ecosystem levels (Krajina, 1969). Here, I use the classic dendroclimatology approach using three tree species (Douglas-fir, lodgepole pine, and hybrid white spruce) in southern interior British Columbia. The principle objective is to examine and quantify the relationship between key climate variables and tree ring width chronologies along a major elevation gradient to determine whether the responses of individual species varied among different elevations.  3.2. Methods 3.2.1. Sampling sites Study sites were located on Tolko Industries Ltd’s Tree Farm License 49 (TFL 49), near Kelowna, B.C. (previously Riverside Forest Products Ltd., 2001) (see Figure 4.1). Mean July temperature for this region was 19.1°C for 1971 – 2002, and mean December temperature –2.9°C. Average annual temperature is 6.2°C and the mean annual total precipitation 391.8 mm. A detailed historical climate summary (1971 – 2002) is shown in Figure 3.2 (Environment Canada, 2003). 68  The study area contains five forested biogeoclimatic (BEC) zones: the Ponderosa Pine (PP), Interior Douglas-fir (IDF), Montane Spruce (MS), Engelmann Spruce-Subalpine Fir (ESSF) and Interior Cedar Hemlock (ICH) zones (Pojar et al., 1987; Lloyd et al., 1990). My study transect lines passed through three of these zones (i.e. ESSF, MS and IDF) representing a gradient of elevation-induced climates from a low elevation sub-continental warm dry summer and cool winter (IDF), to high elevation sub-continental subalpine with cold snowy winters and cool summers (ESSF). The MS zone represents the local mid-elevation environments with an intermediate climate (Lloyd et al., 1990). Based on the GIS database from TFL 49, there are thirteen species of trees in the study area, the three most abundant being lodgepole pine (Pinus contorta var. latifolia), Douglas-fir (Pseudotsuga menziesii var. glauca) and hybrid white spruce (Picea glauca x engelmannii) (Nitschke, 2006). I selected these three species as the focus for my study.  Figure 3. 1.The location of the study site in TFL 49. The three yellow lines are the study transects (The data from the dotted line were lost stolen and could not be replaced). The red triangle on the TFL map shows the location of the weather station at Westwold (50°28' N, 119°45' W, 609 m a.s.l.). (Google Inc., 2009; Riverside Forest Products Ltd., 2001).  69  Figure 3. 2. Climatic diagrams of Westwold (1971 – 2002). Dashed line: mean monthly precipitation including snow; Solid line: mean monthly temperature; y: number of years considered; T: mean annual temperature (°C); P: mean annual amount of precipitation (mm); TM : absolute maximum temperature (°C); tM : mean daily maximum temperature (°C); tm: mean daily minimum temperature (°C); Tm : absolute minimum temperature (°C). Oblique striped area shows months with an absolute minimum temperature below 0 °C.  3.2.2. Sample collection and process The original sampling design had three topographic transect lines across the TFL 49 region. However, at the end of the first summer (2003) the field data from the first transect were stolen and the increment cores which I had obtained were not of adequate quality. For these reasons, this transect was abandoned. In the second summer (2004), the two remaining transect lines crossing three BEC zones (ESSF, MS and IDF) were sampled (Figure 3.1). The original sampling design was to have 40 trees per plot per species, but due to the uneven distribution of the three tree species and limited road access, the final sample size for each plot was sometimes lower (Table 3.1). In total, I obtained 80 cores of lodgepole pine in the ESSF zone, 40 cores in the MS zone and 40 cores in the IDF zone. For spruce, I obtained 65 cores in the ESSF and 40 in the MS zone. I obtained 41 cores of Douglas-fir in the MS zone and 66 in the IDF zone. Access to elevation sequences in the area was limited to those roads that ascend the steep slopes from the Okanagan Valley floor and bypass cliffs and the barriers to field sampling. 70  Sampling was limited to stands that could be reached and sampled within a work day from an access road. Sample plots were deliberately located on zonal or mesic sites (sensu Krajina 1969) in which moisture experiences of trees are dominated by local precipitation and climate-induced water balance rather than factors such as thin, coarse textured soils, slope, or very deep coarse textured soils (azonal sites). This strategy was adapted because I wanted to characterize the response of the majority of the tree populations within the BEC zones, and not just trees on the extreme dry end of the soil moisture gradient.  Table 3. 1. Plot location and sample information summary BEC zones  Longitude Latitude  Elevation (m)  ESSFxc2 W119°57' N50°25'  1598 – 1796 m  Transect 1 MSdm2 W119°57' N50°24'  1455 – 1550 m  IDFdk1 W119°56' N50°22'  1089 – 1168 m  ESSFxc2 W119°39' N50°16'  1473 – 1542 m  IDFdk1 W119°50' N50°20'  1244 – 1356 m  Transect 2  Species Lodgepole pine Spruce Lodgepole pine Spruce Douglas-fir Lodgepole pine Douglas-fir Lodgepole pine Spruce Lodgepole pine Douglas-fir  DBH (cm) 34.7±4.6 43.9±7.4 36.1±5.9 43.3±8.5 47.6±9.9 53.9±9.2 37.3±4.8 47.4±7.7 37.4±6.3 53.0±10.5  Height (m) 24.1±2.9 28.1±4.7 25.9±5.9 30.5±3.7 27.5±4.9 30.4±6.1 26.4±4.1 30.7±5.7 24.7±2.7 27.7±7.6  Sample size 40 35 40 40 41 40 40 30 40 26  3.2.3. Sampling regime TFL 49 has been severely attacked by mountain pine beetle (Dendroctonus ponderosae) with widespread mortality in lodgepole pine. I therefore sampled mixed stands rather than pure stands since the former had more surviving pine than the latter, and also by having other two species as reference I sought to separate climate-caused growth variations versus insect-caused growth variations (Zhang et al., 1999). I selected five sampling sites (see Table 3.1) based on the TFL49 BEC zone maps of species distribution, availability of stands of the target species and accessibility of the stands. At each site, I identified healthy (i.e. no scars or large numbers of insect emergence holes on the stem) canopy dominant trees with DBH > 20 cm. The selection criteria were chosen to avoid trees whose growth may have been influenced strongly by competition for light and/or nutrients; the objective was to sample trees for which a major factor affecting its later growth (post-stem-exclusion stage; Oliver and Larson, 1990) has been climate. One core was taken per tree at ~ 1.3 m above the forest floor using an increment borer. Given the 71  overlap of some BEC zones in the study area, I also examined the understory vegetation composition to ensure a mesic site condition and that the sample sites came from sites in the target BEC subzone (Lloyd et al., 1990). Therefore, while taking the cores, site location (i.e. latitude, longitude, elevation, aspect and slope) and for each tree DBH, tree height, ground vegetation and its percentage coverage were recorded.  3.2.4. Core preparation and creation of tree-ring chronologies Cores were air dried, soaked in an acetone solution for 24 hours to remove the extractives, and then air dried again. I used a saw with parallel circular blades to prepare a flat strip approximately 1.5 mm thick for analysis in an X-ray densitometer (Model QTRS-01X, QMS, Quintek Measurement Sys. Inc., Knoxville, Tennessee, U.S.A.). This instrument measures the width of early wood and late wood, and the annual tree ring and overall annual ring density. The annual ring width was used for cross dating using the COFECHA software (Holmes, 1983; Grissino-Mayer, 2001). This program uses a segmented time series correlation technique to assess the quality of cross-dating and to determine if there are false rings or dating errors in the cores (Grissino-Mayer, 2001). Cores of poor quality (e.g., fragmented, rotted, not cross-datable) were excluded from further analysis. I used ARSTAN (developed by Dr. Edward R. Cook, Tree-Ring Laboratory, Lamont-Doherty Earth Observatory, Columbia University; Cook, 1985; Cook and Holmes, 1986; Grissino-Mayer et al. 1996) to detrend the biological effects of age and competition and to get a residual chronology of each species for each BEC zone (Cook and Holmes, 1986). ). A double detrending method was chosen. First, a negative exponential curve was used to detrend the measured tree-ring sequences to remove the biological growth trend related to the tree’s age. The subsequent tree-ring index sequences were then detrended a second time with a cubic spline of 50% frequency-response cutoff at 32 years to remove a low-frequency trend related to stand dynamics. The ring-width indices chronologies following the double detrending were averaged together by year across different samples for each site with a robust mean calculation to further remove the random signals related to local disturbances (Cook, 1985). The result is a residual tree-ring chronology representing the common signal for the site from the different samples. These residual chronologies were then used to analyze for relationships between ring width and selected climate variables.  72  3.2.5. Climate data The climate data for 1921 – 2002 were derived from the Westwold weather station (Figure 3.1). Based on site information, lapse rate and annual precipitation isohyet regression functions, I generated climate data for each BEC zone study site for which I lacked field measured climatic data by using the regional climate generation model MTCLIM (Thornton et al., 1997; see Chapter 2). This model was specifically designed to generate mountain weather data from available (generally valley bottom) weather stations. Despite having field-measured data from 1921 to 2002, the restriction of MTCLIM input requirements limited me to generating climate data for my study sites from 1922 to 1997. Using this slightly restricted set of generated climate data I explored the relationship between maximum, minimum and daily temperatures, precipitation, vapour pressure deficit, daily incident solar radiation and day length data and the tree ring data8.  3.2.6. Statistical analysis I ran correlations, principal component and simple and multiple regression analyses in this examination of growth-climate relationships. In the first set of tests, I used monthly mean temperature and monthly total precipitation from the previous April to current October (i.e. 19 months; a total of 38 variables) (see also Watson and Luckman, 2001; Wilson and Luckman, 2003). I then used principal component analysis (PCA) to explore the interrelationship between variables and to reduce the number of variables by identifying the most important sources of variance within a data set - the major factors that explain how climate has affected tree growth at the study sites (Legates, 1991; Wilson and Luckman, 2003). I then repeated the analysis using DendroClim 2002 (developed at the Department of Geography, University of Nevada, Reno; Biondi and Waikul 2004). DendroClim 2002 is a statistical tool commonly used in dendroclimatological studies that uses bootstrapping techniques (Halfon, 1989) to establish the confidence interval and estimate the significance of both correlation and response function coefficients. By inputting tree ring index and climatic variables (i.e. monthly values of air temperature and precipitation or others), the program outputs both graphic and text results that define the relationship between climatic variables and ring 8  “Growth” means dbh ring growth in this thesis. 73  index. In a second set of statistical analysis I chose additional climatic variables and plotted the residual tree-ring index against them to explore whether there were any additional linear or curvilinear relationships. The variables were August mean temperature, January mean temperature, annual mean temperature, annual maximum and minimum daily temperature, annual total precipitation, annual growing degree days above 5 °C and 10 °C, growing season (i.e. May to August when it is the frost free period in Westwold) degree days and growing season precipitation. Two types of growing season were defined for this analysis, a long season (May to August) and a short season (June to August). Two additional composite variables were also included in the analysis, winter precipitation (previous December to current February) and the sum of two successive growing seasons precipitation (i.e. previous May to August precipitation plus current May to August precipitation). These accumulated precipitations were calculated with all the available data and also using thresholds of precipitation (150, 200, 250 and 300 mm; Seely and Welham, 2008). I also calculated monthly and annual potential evapotranspiration (PET) using the Jensen-Haise method (Bonnan 1989; Jensen et al. 1990) and PET of the previous year. Monthly climate moisture index (CMI) was calculated as monthly precipitation minus PET (Hogg 1994, 1997). CMI of the previous year was also included in the analyses. Finally, Pacific decadal oscillation monthly values were also included (Mantua et al. 1997, Zhang et al. 1997). The final analysis carried out was a multiple regression. Using the variables that were significantly correlated (p < 0.05) with the residual tree-ring index chronology, an extra forward stepwise regression was carried out with a level of significance of 0.10 specified for a variable to enter into the model. To reduce problems of co-linearity among variables included in the stepwise regression analysis, only the single variable from the same family of variables with the highest correlation coefficient was included. For example, if both annual degree-days above 5°C and 10°C for the same period were significantly correlated to the residual tree-ring index chronology, only one of them was included in the stepwise analysis, because both variables were highly correlated with each other (Quinn and Keough, 2002). Statistical analyses were carried out with SAS 8.02 (SAS Institute, NC, USA).  74  3.3. Results 3.3.1. Tree ring general information Summary statistics of the chronologies for all three species across three BEC zones are listed in Table 3.2. The number of cores analyzed ranged from 17 to 54 cores per chronology, which is 53 to 94 percent of the original sample size9. Mean inter-correlation (i.e. the average correlation between all series) ranged from 0.40 to 0.63, with an overall mean (±standard error) for all plots of 0.51 ± 0.08. Mean sensitivities (i.e. a measure of the annual variability in tree rings) for each plot ranged from 0.17 to 0.23, with an overall mean (±standard error) of 0.19 ± 0.03. The shortest master chronologies of all plots were 103 and 104 years (e.g. lodgepole pine at both ESSF and MS zones, respectively), while the longest core was Douglas-fir at IDF zone, with a length of 215 years. For all chronologies, the percentage correlation between current year and previous year (autoregressive model of order 1) was from 70% to 95% (i.e. correlation between the ring width recorded for one year and the one recorded two, three or four years before, respectively). Table 3. 2. Site elevation and descriptive statistics of the dendrochronology data of the three tree species. BEC Zone  ESSF  Species Lodgepole pine Spruce  Elevation(m)  1473-1796  Lodgepole pine MS  Spruce  1455-1550  Douglas-fir IDF  Lodgepole pine Douglas-fir  1089-1356  % of Auto-regressive model Number of Original Mean Master Series Mean cores sample Range intercorrelation sensitivity order 1 order 2 order 3 order 4 analyzed trees  37 54 17 23 33 19 52  53 58 32 28 35 35 59  0.40 0.51 0.54 0.49 0.63 0.41 0.57  0.23 0.17 0.19 0.15 0.19 0.20 0.21  1900-2003 1867-2003 1889-2003 1884-2003 1861-2003 1900-2003 1788-2003  91% 4% 78% 13% 71% 29% 91% 9% 94% 6% 79% 21% 87% 6%  7%  4% 2%  8%  Residual growth indexed chronologies of all species across three BEC zones are shown in Figures 3.3 to 3.5. For each species, the five-year moving average (thick line) shows similar patterns across the BEC zone; while between different species, the patterns have slight differences. However, after 1950, all seven time series show a strong similarity and have two big dips around 1960 and 1970. For each species across different elevations, the patterns of the residual indexed chronologies are similar; especially after 1950, when the patterns are almost overlaid on each other, showing synchronous periods of growth above and below the average. 9  Cores of poor quality (e.g., fragmented, rotted, not cross-datable) were excluded from further dendrochronological analysis. 75  In order to analyze how similar the sites are in terms of tree growth patterns, I also calculated Pearson’s correlation coefficients (Table 3.3) between different plots and plotted the residual indexed chronologies of each site against each other (Figure 3.6).  Table 3. 3. Pearson’s correlation coefficients (r) between different plots. The thick line separates the correlation coefficient of different species. (Pl: lodgepole pine, Sx: hybrid spruce, Fd: Douglas-fir).  ESSF Pl MS Pl IDF Pl  ESSF Pl MS Pl 1.00 0.56 1.00 0.33 0.41  IDF Pl ESSF Sx MS Sx MS Fd IDF Fd  1.00  ESSF Sx MS Sx  0.24 0.08  0.45 0.34  0.31 0.23  1.00 0.59  1.00  MS Fd IDF Fd  0.39 0.27  0.68 0.46  0.26 0.33  0.38 0.23  0.21 0.26  1.00 0.54  1.00  The correlations between populations of the same species in different BEC zones were slightly higher for spruce and Douglas-fir (i.e. 0.59 and 0.54, respectively) than those of lodgepole pine (0.56, 0.33 and 0.41). However, comparing the correlations between different species in the same BEC zone, the correlation coefficients were relatively smaller (i.e. 0.24, 0.34, 0.21 and 0.33) than individual-species correlation coefficients, except for lodgepole pine and Douglas-fir in the MS zone (i.e. 0.68). Consistent results are shown in Figures 3.6 and 3.7. In Figure 3.6, all five comparisons had slopes around 0.37 to 0.68 with the coefficient of determination (r2) varying from 0.11 to 0.35. Even though the plots are scattered, it was still possible to see a positive linear correlation within species between different BEC zones. In Figure 3.7, the linear correlation of the comparisons was not as strong as those in Figure 3.6 (e.g. most slope ranges were from 0.16 to 0.29) except lodgepole pine versus Douglas-fir in the MS zone where the slope was 0.71 with r2 = 0.11. The data in Figure 3.7 are also more scattered than for comparisons within species, (Figure 3.6) except the one just mentioned.  76  Figure 3. 3. Lodgepole pine growth indexed residual chronologies for the ESSF, MS and IDF zones. The thin lines represent the tree ring index chronology, the thick lines represent the five-year moving average trend, and the horizontal dotted lines represent the average tree growth for the period.  77  Figure 3. 4. Spruce growth indexed residual chronologies for the ESSF and MS zones. The thin lines represent the tree ring index chronology, the thick lines represent the five-year moving average trend and the horizontal dotted lines represent the average tree growth for the period.  78  Figure 3. 5. Douglas-fir growth indexed residual chronologies for the MS and IDF zones. The thin lines represent the tree ring index chronology, the thick lines represent the five-year moving average trend and the horizontal dotted lines represent the average tree growth for the period. Data prior to 1860 for the IDF are not shown.  79  3.3.2. Growth-Climate relationships a. Correlation coefficients and principle component analysis (PCA)  Using SAS 8.02 and DendroClim 2002, I calculated the Pearson’s correlation coefficient between the residual indexed chronology for each species in each BEC zone and the climate variables from 1922 to 1997. Due to limitations of MTCLIM (see Thornton, 2000), a longer time series of climatic data could not be used for statistical purposes. The climate variables used in this analysis are mean monthly temperature from the previous April to current October and the total monthly precipitation from previous April to current October. The results are presented in Figures 3.8 to 3.10. All correlation coefficients were medium to low, ranging from -0.3 to 0.4. For temperature, lodgepole pine ring growth in the ESSF and MS zones had positive correlations with previous April and May monthly mean temperature, winter monthly mean temperature (i.e. previous October to current March) and current August and October monthly mean temperature. They had negative correlations with previous growing season monthly mean temperature (previous June to September). For precipitation, lodgepole pine in these two BEC zones had positive correlations with previous August, September and current June and July monthly total precipitation. They had negative correlations with previous April to June monthly total precipitation and winter monthly total precipitation (i.e. previous October to current February). The patterns for lodgepole pine were quite similar in these two zones. On the other hand, lodgepole pine growth in the IDF zone showed a different pattern from those in the ESSF and MS zones. It had a positive correlation with most monthly mean temperatures, except with the previous August, September, November, December and current September. It also showed a positive correlation with most monthly total precipitation values, except for the previous summer months (i.e. previous April to July).  80  Figure 3. 6. Plots of regressions of residual indexed chronologies within species between different BEC zones. The black line is the linear correlation and its mathematical function is listed in the right bottom corner with the associated coefficient of determination (r2).  81  Figure 3. 7. Plots of regressions of residual indexed chronologies within BEC zones between different species. The black line is the linear correlation and its mathematical function is listed in the right bottom corner with the associated coefficient of determination (r2). 82  ESSF Pl vs Monthly temperature  ESSF Pl vs Monthly precipitation  0.40  0.40  0.30  0.30  0.20  0.20  0.10  0.10  0.00 -0.10  0.00 A  M  J  J  A  S  O  N  D  J  F  M  A M  J  J  A  S  O  -0.10  -0.20  A M  J  J  Previous year  -0.40  -0.30  Current year  -0.40  0.40  0.30  0.30  0.20  0.20  0.10  0.10  D  J  F  M  A M  J  J  A  S  O  S  O  S  O  Current year  0.00  0.00 A M  J  J  A  S  O  N  D  J  F  M A  M  J  J  A  S  O  -0.10  A M  J  J  A  S  O  N  D  J  F  M A  M  J  J  A  -0.20  -0.20 -0.30  Previous year  -0.40  -0.30  Current year  Previous year  Current year  -0.40  IDF Pl vs Monthly temperature  IDF Pl vs Monthly precipitation  0.40  0.40  0.30  0.30  0.20  0.20  0.10  0.10  0.00  0.00 A  M  J  J  A  S  O  N  D  J  F  M  A  M  J  J  A  S  O  -0.10  A M  J  J  A  S  O  N  D  J  F  M A  M  J  J  A  -0.20  -0.20 -0.40  N  MS Pl vs Monthly precipitation  0.40  -0.30  O  Previous year  MS Pl vs Monthly temperature  -0.10  S  -0.20  -0.30  -0.10  A  Previous year  -0.30  Current year  -0.40  Previous year  Current year  Figure 3. 8. Pearson’s Correlation Coefficients between the residual indexed chronology of lodgepole pine and climate variables (left panels temperature and right panels precipitation) from 1922 to 1997. Yellow bars mean those values are statistically significant. The X axis on the left and right are months from previous April to current October. The Y axis is the Pearson’s correlation coefficient.  83  ESSF Sx vs Monthly precipitation  ESSF Sx vs Monthly temperature 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  0.40  Previous year  Current year  0.30 0.20 0.10 0.00  A M  J  J  A  S  O  N  D  J  F  M A  M  J  J  A  S  O  -0.10  A M  J  J  -0.30  N  D  J  F  M A  M  J  J  A  S  O  Previous year  S  O  Current year  MSSx vs Monthly precipitation 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  0.30 0.20 0.10 0.00 A M  J  J  A  S  O  N  D  J  F  M A  M  J  J  A  S  O  -0.20 -0.40  O  -0.40  0.40  -0.30  S  -0.20  MSSx vs Monthly temperature  -0.10  A  Previous year  Current year  A M  J  J  A  S  Previous year  O  N  D  J  F  M A  M  J  J  A  Current year  Figure 3. 9. Pearson’s Correlation Coefficients between the residual indexed chronology of hybrid spruce and climate variables (left panels temperature and right panels precipitation) from 1922 to 1997. Yellow bars mean those values are statistically significant. The X axis on the left and right are months from previous April to current October. The Y axis is the Pearson’s correlation coefficient.  The patterns for spruce were similar to those of lodgepole pine for precipitation but less so for temperature (Figure 3.9). For both zones, spruce showed a positive correlation with previous growing season precipitation (i.e. previous April to September), but a negative correlation with monthly mean temperature for the same period. As before, the correlation coefficients for the other climate variables examined were all very small, except for current June monthly temperature and precipitation in the ESSF zone. It had a positive correlation with current June monthly mean temperature and a negative correlation with current June monthly total precipitation.  84  MS Fd vs Monthly temperature 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  A M  J  J  A  S  O  N  D  J  F  M  A  Previous year  M  MS Fd vs Monthly precipitation  J  J  A  S  0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  O  Current year  A M  J  J  A M  J  J  A  S  Previous year  O  N  D  J  F  M A  S  O  N  D  J  F  Previous year  IDF Fd vs Monthly temperature 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  A  M A  M  J  J  A  S  O  S  O  Current year  IDF Fd vs Monthly precipitation  M  J  J  A  S  0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40  O  Current year  A M  J  J  A  S  Previous year  O  N  D  J  F  M A  M  J  J  A  Current year  Figure 3. 10. Pearson’s Correlation Coefficients between the residual indexed chronology of Douglas-fir and climate variables (left panels temperature and right panels precipitation) from 1922 to 1997. Yellow bars mean those values are statistically significant. The X axis on the left and right are months from previous April to current October. The Y axis is the Pearson’s correlation coefficient.  Douglas-fir in both MS and IDF zones had similar patterns of relationships for both temperature and precipitation variables (Figure 3.10). Douglas-fir showed positive correlations with previous April and May temperatures and also winter temperature (i.e. previous October to current March). This is similar to the relationship for lodgepole pine in the ESSF and MS zones. Douglas-fir in these two zones also had positive correlations with previous August and September precipitation and current growing season precipitation (i.e. current March to August). In the MS zone, Douglas-fir showed a negative correlation with winter precipitation (i.e. previous October to current February) while in the IDF zone there was almost no correlation between the chronology indices and winter precipitation variables.  85  Table 3. 4. Significant correlations between tree-ring residual chronologies and climate variables for lodgepole pine. See Appendix B for extra statistical output on non-significant correlations. BEC zone Variable  Correlation  Signif  ESSF  PP12 to P2 < 30 mm PET May PDO January PDO April  -0.3632 -0.3951 0.3208 0.2261  0.0032 0.0006 0.0047 0.0496  MS  T Tm PET February PET March Annual PET PDO January PDO February PDO April PDO June CMI February PP12 to P2 PP12 to P2 < 300 mm  0.2942 0.3102 0.3425 0.2463 0.2348 0.2517 0.2228 0.2562 0.2398 -0.2729 -0.2551 -0.3689  0.0128 0.0085 0.0035 0.0384 0.0487 0.0283 0.0530 0.0255 0.0369 0.0213 0.0387 0.0027  IDF  Tm P from May to Aug < 150 mm PET February PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PP12 to P2 CMI February  0.4147 0.3207 0.3169 -0.3517 -0.2650 -0.2619 -0.2619 -0.3453  0.0003 0.0530 0.0071 0.0051 0.0329 0.0336 0.0336 0.0032  Plot Correlation (limits -1 to +1) -1 0  1  In considering the secondary climate variables, I found that lodgepole pine tree-ring residual chronology in the ESSF zone was negatively correlated with previous winter precipitation previous May PET, and January and April PDO (Table 3.4), while in the MS zone, lodgepole pine was still negatively correlated with previous winter precipitation, but also negatively correlated with February CMI. However, it was positively correlated with February, March and annual PET, annual mean temperature, annual minimum temperature and January, February, April and June PDO. As in the IDF zone, lodgepole pine had the same negatively correlated response with previous winter precipitation in the ESSF and IDF zones and negatively correlated response to February CMI in the MS zone. It was also positively correlated with February PET and annual minimum temperature as MS zone. The only difference between the IDF zone and the others is its positively correlated response to current long growing season precipitation under the threshold of 150 mm. Spruce tree-ring residual chronology in the ESSF zone showed significant positive correlation 86  with June PET but a negative correlation with CMI for the same month. In the MS zone, the spruce tree-ring residual chronology showed a positive correlation with previous year’s short growing season precipitation (previous June to August) (Table 3.5). Table 3. 5. Significant correlations between tree-ring residual chronologies and climate variables for hybrid spruce. See Appendix B for extra statistical output on non-significant correlations. BEC zone Variable Correlation Signif Plot Correlation (limits -1 to +1) -1 0 1 PET June 0.2637 0.0263 ESSF CMI June -0.2751 0.0202 MS  Precipitation from PP6 to PP8  0.2757  0.0250  For Douglas-fir (Table 3.6), tree-ring residual chronologies in both MS and IDF zones were positively correlated with January temperature and PET. Douglas-fir radial growth in the MS zone was also positively correlated with annual minimum temperature and January, February, April and June PDO, but negatively correlated with previous winter precipitation. While in the IDF zone, it was positively correlated with annual and growing season precipitation, November PET, May, Jun and July’s CMI. Finally, it was negatively correlated with May and July’s PET, annual and growing season GDD > 5°C.  I did principal components analysis using SAS 8.02 to identify the major determinants of the chronology indices. Unfortunately, none of the PCA analyses was able to pick up less than 13 principal components (see Appendix B). Using the DendroClim2002 program, however, I found that lodgepole pine ring growth in the ESSF zone was positively correlated with previous April monthly mean temperature (r = 0.20) and current January and July precipitation (r = 0.27, 0.27 respectively); it was negatively correlated with current January temperature (r = -0.14). In the MS zone, lodgepole pine was positively correlated with current January and July precipitation (r = 0.29, 0.25 respectively) and was negatively correlated with previous June precipitation (r = -0.23). In the IDF zone, lodgepole pine was positively correlated with previous April monthly mean temperature (r = 0.18), current July precipitation (r = 0.26) and was negatively correlated with previous December monthly mean temperature (r = -0.23).  87  Table 3. 6. Significant correlations between tree-ring residual chronologies and climate variables for Douglas-fir. See Appendix B for extra statistical output on non-significant correlations. BEC zone  Variable  Correlation  Signif  MS  January Temperature Tm PET January PDO January PDO February PDO April PDO June PP12 to P2 < 300 mm  0.3046 0.2431 0.3046 0.3346 0.2875 0.2490 0.2399 -0.2965  0.0098 0.0411 0.0098 0.0043 0.0150 0.0363 0.0439 0.0174  IDF  January Temperature Annual P Precipitation May to August Precipitation June to August P from May to Aug < 150 mm P from May to Aug < 200 mm P from May to Aug < 250 mm P from May to Aug < 300 mm P from Jun to Aug < 150 mm P from Jun to Aug < 200 mm P from Jun to Aug < 250 mm P from Jun to Aug < 300 mm PET January PET November CMI May CMI June CMI July PET May PET July GDD at 5 for May to August Annual accum. GDD >above 5°C  0.2390 0.2861 0.3863 0.3116 0.3437 0.4385 0.3869 0.3863 0.3145 0.3235 0.3613 0.3116 0.2390 0.2410 0.3605 0.3245 0.2784 -0.2413 -0.2627 -0.2396 -0.2560  0.0447 0.0156 0.0009 0.0082 0.0373 0.0009 0.0013 0.0009 0.0183 0.0086 0.0021 0.0082 0.0447 0.0429 0.0020 0.0058 0.0187 0.0427 0.0269 0.0441 0.0312  Plot Correlation (limits -1 to +1) -1 0  1  Significant response functions (p<0.05) for lodgepole pine in the ESSF and MS zones were:  ESSF:  TRI = 0.22 × PJan + 0.19 × P Jul  MS:  TRI = 0.29 × PJan + 0.21 × PJul  Spruce tree-ring residual chronologies in the ESSF and MS zones were positively correlated (P < 0.05) with current July precipitation (r = 0.25 and r = 0.18, respectively). However, no significant response functions were found for spruce. As for Douglas-fir in the MS zone, ring width residual index chronology was positively correlated with previous April (r = 0.20), current May and June monthly mean temperature (r = 0.20, 0.20 respectively) and also current July precipitation (r = 0.38). It was negatively correlated 88  with previous December temperature (r = -0.22) and previous June precipitation (r = -0.21). In the IDF zone, Doulas-fir was positively correlated with current May and June monthly mean temperature (r = 0.19, 0.15 respectively) and with previous October precipitation (r = 0.25). The response functions for Douglas-fir are:  MS:  TRI = -0.22 × PJun -1 + 0.24 × PJul  IDF:  TRI = 0.18 × POct -1  b. Regression analysis Because all the correlation coefficients for the three tree species across the BEC zones were small, I used not only the individual monthly variables but also explored other climate variables and used regression analysis to seek any other relationships between ring width residual indexed chronology and these climatic variables. Results are summarized in Table 3.7. No strongly linear relationships between the residual chronology and climatic variables were found, but some weak significant regression trends could be identified. All three species showed quite similar patterns across different BEC zones, but with slight differences. Lodgepole pine had a relatively high negative regression with previous winter precipitation across three zones, while in MS and ESSF there was a negative linear relationship with previous August temperature and a positive linear relationship with previous December temperature. For lodgepole pine in the MS and IDF zones, there were positive linear relationships with annual minimum temperature and February temperature, and PET, but a negative linear relationship with December CMI. For spruce, the responses in the ESSF zone were stronger than those in the MS zone (e.g. the 2  r values for the ESSF were bigger than those for the MS zone, Appendix B) with a similar trend across different climatic variables. In both zones, tree radial growth had negative linear relationships with previous July temperature and positive relationships with previous July and August precipitation. Douglas-fir showed positive relationships with previous April, November and current January mean temperature, and also January PET in both MS and IDF zones. For the MS zone, the residual chronology showed a positive relationship with previous May and December mean  89  temperature and annual minimum temperature, and a negative relationship with winter precipitation. For the IDF zone of tree-ring residual chronology showed significant relationships with additional climate variables (e.g. GDD from May to Aug, annual precipitation, some PET and CMI, see Appendix B) For significant climate variables, I then did a multivariate regression analysis (Table 3.8). For lodgepole pine in the ESSF zone, most of the variables picked for the best model were those from the previous year (i.e. previous winter precipitation, previous April and November temperature) plus current May PET and January PDO. These variables explained 39% of the variance in the tree-ring index. In the MS zone, most of the variables were affected by the previous year’s climate: previous year precipitation together with previous May temperature and annual PET. In the IDF zone, however, the variables picked were annual minimum temperature and current February precipitation. For spruce in the ESSF zone, all the variables picked were temperature-related, while in the MS zone, all the variables picked were precipitation-related. For Douglas-fir, most of the variables in the MS zone came from the previous year and were more related to temperature, while in the IDF zone; selected variables included previous July temperature, current May precipitation, current July PET and annual total precipitation.  90  Table 3. 7. Significant regressions (p < 0.05) between tree-ring residual index (TRI) and selected climate variables, for different tree species at different BEC zones. Regression graphs and p values can be found in Appendix B. ESSF Variable LODGEPOLE PINE T Previous April T Previous August T Previous November T Previous December T Current May Accum. PP12 to P2 < 300 mm PET May PDO January PDO April  HYBRID SPRUCE T Previous June T Previous July T Previous August T Previous July T Previous August P Current June PET June CMI June DOUGLAS FIR  MS  IDF  r2  Variable  r2  Variable  r2  0.11 0.07 0.09 0.08 0.16 0.13 0.16 0.10 0.05  T Previous May T Previous August T Previous December T Current February T Current March Annual T mean Min of Tmin P Previous August P Previous October PP12 to P2 Accum. PP12 to P2 < 300 mm PET February PET March CMI February Annual PET PDO January PDO April PDO June  0.08 0.11 0.09 0.12 0.06 0.09 0.10 0.15 0.06 0.07 0.14 0.12 0.06 0.07 0.06 0.06 0.07 0.06  T Previous June T Current February Min of Tmin P Current February P from May to Aug < 150 mm Accum PP12 to P2 Accum PP12 to P2 < 200 mm Accum PP12 to P2 < 250 mm Accum PP12 to P2 < 300 mm PET February CIM February CIM March  0.07 0.10 0.17 0.08 0.10 0.07 0.12 0.07 0.07 0.10 0.12 0.07  0.07 0.14 0.14 0.09 0.11 0.06 0.07 0.08  T Previous July P Previous July P Previous August Accum Jun. P to August P  0.06 0.06 0.06 0.08  T Previous April T Previous May T Previous November T Previous December T Current January Min of Tmin Accum PP12 to P2 < 300 mm PET January PDO January PDO February PDO April PDO June  0.10 0.14 0.07 0.11 0.06 0.06 0.09 0.09 0.11 0.08 0.06 0.06  T Previous April T Previous July T Previous November T Current May T Current July P Current May P Current June P Current July GDD above 5°C May to Aug. GDD above 10°C May to Aug. P May to August Accum P May to Aug < 150 mm Accum P May to Aug < 200 mm Accum P May to Aug < 250 mm Accum P May to Aug < 300 mm P June to August Accum P June to Aug < 150 mm Accum P June to Aug < 200 mm Accum P June to Aug < 250 mm Accum P June to Aug < 300 mm Annual P PET January PET May PET July PET November CMI May CMI June CMI July  0.07 0.07 0.06 0.06 0.07 0.12 0.10 0.07 0.06 0.07 0.15 0.12 0.19 0.15 0.15 0.10 0.10 0.10 0.13 0.10 0.08 0.06 0.06 0.07 0.06 0.13 0.11 0.08  91  Table 3. 8. Models for tree-ring residual index (TRI) selected by stepwise regression for different climate variables, for different species, BEC zones and transects. Cp: Mallow’s Cp coefficient.  BEC  Model  r2  Cp  0.39  4.79  0.53  5.42  0.27  1.00  zone Lodgepole pine ESSF  TRI = 1.577 – 0.041 PETMay - 0.0071 (PDec -1 to PFeb < 300 mm) + 0.032 PDOJan + 0.014 TNov -1  MS  TRI = 0.821 + 0.028 PDOApril + 0.013 PAug -1 – 0.012 POct -1 + 0.022 TMay -1 – 0.006 (PDec -1 to PFeb ) + 0.019 PETFeb  IDF  TRI = 1.352 + 0.009 TMinimum – 0.027 PFeb  Hybrid spruce ESSF  TRI = 1.386 – 0.018 TAug -1 – 0.020 TJul -1 + 0.016 TJun  0.32  6.35  MS  TRI = 0.918 + 0.006 PJul -1 + 0.007 PAug -1  0.12  2.65  0.41  4.09  0.43  -1.20  Douglas fir MS  TRI = 0.967 + 0.028 TMay -1 + 0.010 TJan + 0.019 TNov -1 - 0.006(PDec -1 to PFeb < 300 mm)  IDF  TRI = 1.905 + 0.032 PMay – 0.030TJul -1 – 0.041 PETJul + 0.003 PAnnual  92  3.4. Discussion 3.4.1. Sample selection rules The common approach in dendroclimatology is to sample trees from sites experiencing extreme climatic conditions (i.e. cold, dry), or where soil conditions render soil moisture relatively unavailable. Tree ring growth on such sites is very sensitive to climatic variation and provides data that are suitable for reconstruction of past climates (Fritts, 1976; Hughes, 2002; Briffa et al., 2004; Kienast et al., 2007). On the other hand, Krajina (1969) pointed out that, even within the same biogeoclimatic zone, trees grow differently according to site-related variations in soil nutrient and moisture content. In my research, the goal was to find tree-ring / climate relationships that could be used in either predictive growth and yield models or ecosystem management models account for changing climate conditions. Consequently, I selected sample trees from stands on zonal (mesic) site that would represent the growth performance of the majority of the population of trees in that specific biogeoclimatic zone, and to derive functional relationships between growth and climate from that zone. The risk with this approach is that it may generate a weaker response to climate than sampling according to traditional methods. Indeed, the correlations reported in this study were usually in the middle-lower range as compared to studies in the Pacific Northwest that used the traditional approach (Zhang and Hebda, 2004; Holman and Peterson, 2006, Case and Peterson, 2005, 2007).  3.4.2. X-ray densitometer There is increasing use of X-ray densitometers in dendrochronology research (Bower et al., 2005; Skomarkova et al., 2006). The advantage of this instrument is that it is convenient, and it can measure both density and ring width at the same time. However, there are also some issues with this instrument, as summarized in Tree-ring Bulletin in the 1970’s (Parker and Meleskie, 1970; Polge, 1970; Schweingruber et al., 1978; Figure 3.11) The wood tissue quality, the orientation of the ring, if the ring is parallel to the x-ray beam, or the standard threshold value are some of the variables that affect the outputs of X-ray densitometers (Schweingruber et al., 1978). In addition, it has been reported that chronologies of wood density from early and late wood can provide different and sometimes contradictory relationships with climate variables (Hughes, 2002). Even following carefully recommended procedures to minimize the errors, these still occur from time to time. During my research, I ran the x-ray densitometer not only to get output 93  automatically, but I also manually adjusted the output graph, which defines the start and end point of each ring. However, due to problems with the tree cores (e.g. fragile cores, unclear cutting surface, non-parallel rings, low density wood, Figure 3.12), the only information I could use in my analysis was the ring width.  Figure 3. 11. Relative position of x-ray photo and line plot. (Schweingruber et al., 1978).  94  Figure 3. 12. Eight increment core conditions which affect the quality of X-ray densitometer output. a. misaligned xylem cells (early wood), b. cracks, c. misaligned xylem cells (late wood), d. rotten wood, e. foreign matter, f. misaligned rings (early wood + late wood), g. foreign matter, h. limb distortion. (Parker and Meleskie, 1970).  3.4.3. Tree ring analysis results a. Tree-ring chronologies general trends The results of my data are consistent with a similar study carried out with lodgepole pine and Douglas-fir in Washington state (Holman and Peterson, 2006; Case and Peterson, 2005, 2007). Mean inter-correlation values for Douglas-fir and hybrid spruce were in the same range as that of other studies carried out in the region with these same species (Case and Peterson, 2005; Zhang and Hebda, 2004). Values were moderately high, indicating a generally similar pattern for tree-ring width in the trees used to create each chronology. However, inter-correlation values for lodgepole pine in the ESSF and IDF zones were lower than for the other two species (Table 3.2), an indication that radial growth of pines varies more between different individual trees than is the case for the other species in these two zones. This low inter-correlation for pine, however, may also be partially due to the growth performance in unfavourable conditions, an interpretation which is supported by the higher inter-correlation in the MS zone. The mean sensitivity range for all the species was from 0.15 to 0.23. These values are slightly lower than reported by Watson and Luckman (2001a, b), but are within the typical range for Pacific Northwest (Case and Peterson, 2005, 2007; Zhang and Hebda 2004). Most of the tree-ring chronologies detected in my samples had a high first order autoregressive component, which means that previous year radial condition/performance affected current year radial growth  95  (Fritts, 1976). Values of correlation between chronologies and their first order autoregressive component were similar to those of other previously reported studies for these species (Watson and Luckman, 2001a; Case and Peterson 2005, 2007; Zhang and Hebda, 2004) Compared to the original sample size, only about half of the lodgepole pine cores were included in the data analysis. This reflects the poor quality of cores of this species, which in turn reflects the wood quality of the trees cored, even though they were healthy trees. Also, the quality of X-ray densitometer output for this species was often inadequate for further analysis. I sought samples which had the highest correlation with other cores within the plot and also those had the highest response to climate variables, and as a consequence rejected data from many cores which either had low inter-correlation values with other cores or had low sensitivities in the tree-ring chronologies. This procedure is not unusual but the methods used to eliminate cores are rarely reported in detail. In a similar study carried out in Washington State (Case and Peterson, 2005, 2007), only 18% of the original lodgepole cores taken were analyzed, compared with 45% of the original sample size for Douglas-fir. As a consequence, the sample size used to create lodgepole pine chronologies was at the low end of the range of sample sizes recommended by Fritts (1976), and results from these chronologies should therefore be interpreted with caution. The fluctuations in the tree-ring index series showed similar patterns to the variation of the Pacific Decadal Oscillation (PDO), especially with a reduced tree ring growth around the 1970s that matches the last cold period of the PDO (Mantua et al., 1997; Zhang et al., 1997). Finally, the residual chronologies were not the same across the three species although within each species across the elevational transects of BEC zones the patterns matched well. As Table 3.3 shows, there are higher similarities between chronologies for a given species across different BEC zones than for different species within the same BEC zone. For lodgepole pine, the greater the distance between pine plots, the weaker the correlation coefficient. Above suggests that the growth response to climate is different between species, but within a species the results show a similar general response at locations with different climatic conditions modulated by site specific conditions. The high correlation between lodgepole pine and Douglas-fir in the MS zone is notable, which could be an indication that Douglas-fir and lodgepole pine are climate sensitive species, exhibiting a similar response when they grow in a favourable environment, whereas spruce is not as sensitive in the site I sampled (Frits, 1976; Luckman et al., 1997; Zhang et al., 96  1999; Watson and Luckman, 2001a, b; Wilson and Luckman, 2003; Luckman and Wilson, 2005).  b. Correlations between tree-ring chronologies and climate variables  The correlation coefficient values ranged from -0.37 to 0.37, which is typical for this type of study (Bower et al., 2005; Luckman et al., 1997; Luckman and Wilson, 2005; Wilson and Luckman, 2003; MacDonald et al., 1998; Watson and Luckman, 2001; Holman and Peterson, 2006; Case and Peterson, 2005, 2007). However, the response of each species varied between different BEC zones.  b.1. Lodgepole pine For lodgepole pine, the correlation patterns from previous April to current October across three BEC zones were similar (i.e. positive and negative correlation patterns). There was a positive correlation with previous April and May monthly mean temperatures, winter monthly mean temperature (i.e. previous October to current March) and current August and October monthly mean temperatures, but a negative correlation with previous June to September monthly mean temperature (or August and September monthly mean temperature in the IDF zone). These results suggest that lodgepole pine growth is limited by low temperatures in winter (frost damage or snow break) and also by high temperatures in summer (drought effect) (Case and Peterson, 2007). They also show that the length of the previous year’s growing season (defined by the spring temperatures of April and May) influences current growth. This could be because a longer growing season allows lodgepole pine to produce more biomass and then allocate more to buds when there is no nutrient and water limitation (Litton et al., 2007). The correlation pattern for precipitation is not as consistent as that for temperature. Lodgepole pine in the ESSF and MS zones showed positive correlations with previous August, September and current June and July monthly total precipitation, which supports the idea that growing season soil moisture is a major factor affecting growth of this species. Lodgepole pine showed negative correlation with previous April to June monthly total precipitation and winter monthly total precipitation (i.e. previous October to current February). These negative relationships could result from low temperatures in the early period of the growing season, precipitation in the previous growing season, and also a negative growth effect from snow - the most common form of winter 97  precipitation. On the other hand, the lodgepole pine growing in the IDF zone showed slightly different patterns from those of the ESSF and MS zones. It showed positive correlation with most monthly mean temperatures except previous August, September, November, December and current September. It also showed positive correlation with most monthly total precipitation except previous summer (i.e. previous April to July). This may be an indication that in this lower zone when there is no water stress (low temperature and more precipitation) from the previous growing season, lodgepole pine will increase the radial growth of the current year. As for the winter snow, it could be a reservoir for next growing season water supply. It was noticed by Krajina (1969) and Lloyd et al. (1990) that lodgepole pine can be very sensitive to summer drought and water stress. Similar tree-ring/climate relationship patterns have been reported by Case and Peterson (2007). Significant correlations between ring residual index and climate (yellow bars in Figures 3.8 to 3.10) were different among the three BEC zones. It seems that the significant monthly temperatures are delayed by one month as the elevation decreases (e.g. positive correlation with previous June monthly mean temperature in the IDF zone; while in the MS and ESSF zones, the significantly positive correlation came from previous May and April monthly mean temperature, respectively). From these results I conclude that irrespective of which zone it is growing in, lodgepole pine optimum growth is at higher temperatures early in the growing season, and there is a longer growth period if the water supply is sufficient (i.e. positive correlation with monthly mean temperature from both previous and current April, May and October). Winter snowfall in the IDF zone is good for water supply in summer, whereas snowfall in the MS zone could cause damage to branches or buds and reduce the growth in the following year (Levitt, 1980; Havranek and Tranquillini, 1995).  b.2. Hybrid spruce Spruce showed a positive correlation with previous growing season precipitation (i.e. previous April to September), but a negative correlation with monthly mean temperature for the same period for both ESSF and MS zones. This is likely an indication that water stress during the growing season is the most important limiting factor for spruce. Compared with lodgepole pine and Douglas-fir, which had more significant correlations in temperature variables than 98  precipitation variables, spruce seemed to response equally to temperature and precipitation variables. The correlation coefficients for other temperature and precipitation variables (i.e., insignificant ones) were all very small, except for the positive correlation with current June monthly mean temperature in the ESSF zone. This could be an indication that spruce in high elevation sites is little affected by interannual climate fluctuations, and that specific tree- or stand-level agents (such as competition history, gap formation, microtopography, nutrient availability, carbon allocation shifts etc.) could be more influential in determining tree ring growth. In both the ESSF and MS zones, the previous growing season water deficit appeared to be the main factor causing a ring growth decrease (i.e. negatively correlated with previous July and August mean temperature and positively correlated with previous July and August total precipitation). In contrast to the hypothesis that spruce is more sensitive to temperature than soil moisture (MacDonald et al., 1998), in this study spruce was more sensitive to soil moisture than temperature (Zhang et al., 1999; Savva et al., 2006). Table 3.8 summarizes reported relationships between ring growth and climate variables for spruce and lodgepole pine. My results are generally in agreement with these values. Differences most probably reflect the site-specificity of tree ring response as reported by Zhang and Hebda (2004), Green and Miyamoto (2005), Savva et al. (2006), Pichler and Oberhuber (2007) and Su et al. (2007).  99  Table 3. 9. Reported relationships between ring growth and climate variables (i.e. temperature and precipitation) for spruce and lodgepole pine. Altitude in m.a.s.l.; T stands for temperature; P for precipitation; suffix “-1” indicates data from previous year; “variance” indicates the percentage of variance explained by models including the variables cited; and “-” indicates data not provided.  Reference  Region  Altitude  Positive relations  Negative relations  Variance  SPRUCE Kienast et al. (1987)  Colorado – USA  3400  Wilson and Luckman (2003)  Interior BC  -  Zhang et al. (1999)  Interior BC  SBS zone  Green and Miyamoto (2006)  Interior BC  -  This work  Interior BC  MS - 1500  This work  Interior BC  ESSF - 1700  PJan, PFeb, PMar, PApr, PMay  TDec, TJan, TFeb  -  TJun, TJul, TSep, TDec -1  TAug-1  -  TJul, TAug-1 TApr  0.30 -  PJul-1. PAug-1  TJul-1  0.12  TJun, PJul-1. PAug-1  TJul-1, TAug-1, PJun  0.32  LODGEPOLE PINE Chhin et al. (2008)  Alberta (Boreal forest)  878  TOct -1 to TApr, PMay-1 to PSep-1  TJul-1. TAug-1, TSep-1  0.33  Chhin et al. (2008)  Alberta (Foothills)  1262  TOct-1, TNov-1, TFeb, TMar, TApr  TJul-1. TAug-1, POct-1 to PMar  0.32  Chhin et al. (2008)  Alberta (Cypress Hills)  1382  TNov-1, PNov-1, PFeb  TAug-1, TSep-1, POct-1 to PMar  0.20  Chhin et al. (2008)  Alberta (Rocky Mnt.)  1653  TOct-1, TNov-1, TFeb, TMar, TApr  TAug-1, TSep-1, POct-1 to PMar  0.23  Oleksyn and Fritts (1991)  California – USA  Graumlich (1991)  Sierra Nevada – USA  3200  PJul-1, PAug-1PJun, PJul, PAug, , PSummer, PWinter  TApr, TNov-1  0.27  Case and Peterson (2007)  Washington – USA  <1000  PGrowing season  TGrowing season  0.41  Case and Peterson (2007)  Washington – USA  >1000  TSummer-1  Snow depth  0.55  Nakawatase and Peterson (2006)  Washington – USA  >1000  Green and Miyamoto (2006)  Interior BC  -  Anonymous (1998)  Interior BC  IDF  This work  Interior BC  IDF - 1300  TJul-1, TOct-1, TApr, PAug-1, PNov-1, PMay  This work  Interior BC  MS - 1500  TMay-1, TDec-1, TMar, TAug  This work  Interior BC  ESSF - 1700  -  PWinter, PSpring, TSpring  0.30  TSummer TOct-1, TNov-1, TDec-1 PAug-1  TApr-1, TJun  100  -  TJun, TAug-1  0.50 0.27  PAug-1, PDec-1  0.48 0.36  Table 3. 10. Reported relationships between Douglas-fir ring growth and climate variables (i.e. temperature and precipitation). Altitude in m.a.s.l.; T stands for temperature; P for precipitation; suffix “-1” indicates data from previous year; “variance” indicates the percentage of variance explained by models including the variables cited; and “-“ indicates data not provided.  Reference  Region  Altitude  Positive relations TSep, TMay, PJun-1, PSep-1, POct-1, PFeb, PMar, PApr, PMay, PJun  Negative relations  variance  DOUGLAS-FIR Fritts et al. (1971)  Colorado – USA  -  Kienast et al. (1987)  Colorado – USA  1900  Gonzalez-Elizondo et al. (2005)  Central Mexico  2600 – 3100  Biondi (1997)  Central West USA  Zhang et al. (2000)  -  POct-1, PNov-1, PMar, PApr, PMay  TJan, TFeb, TMar, TJun  PAnnual, PWinter, PSpring  TM, TWinter max  -  PApr, PMay, PJun  TJul  Vancouver Island  -  TAug, TSep-1, TNov-1, PApr, PMay, PJun, PJul, PAug-1  Zhang and Hebda (2004)  Coastal BC  -  TMar, PMay, PJun, PJul  Nakawatase and Peterson (2006)  Washington – USA  <500  Nakawatase and Peterson (2006)  Washington – USA  500 – 1000  Case and Peterson (2005)  Washington – USA  <1000  PGrowing season  Case and Peterson (2005)  Washington – USA  >1000  T, TGrowing season  Green and Griesbauer (2007)  Interior BC  -  Zhang et al. (1999)  Interior BC  Anonymous (1998)  0.84 -  TAug-1  -  PAnnual -1, T-1  -  PSummer  TGrowing season  0.41 0.55  TWinter  TGrowing season, PJul-1  -  SBS zone  Tnovember-1, Pnovember-1, Pmay  TJul-1, TJun-1, TJul-1  0.45  Interior BC  IDF zone  PAug-1  TJun  0.50  This work  Interior BC  IDF - 1300  TApr-1, TNov-1, TJan, PMay, PJun, PJul  TJul-1, TMay, TJul  0.433  This work  Interior BC  MS - 1500  TApr-1, TMay-1, TWinter-1  TJul-1, PMay  0.409  101  b.3. Douglas-fir Douglas-fir in the MS and IDF zones showed similar patterns for both temperature and precipitation variables. There was a positive correlation with previous April and May temperature and also with winter temperature (i.e. previous October to current March). This was the same response as lodgepole pine in the ESSF and MS zones, and it may be a result of more vigorous growth with extended growing seasons in previous years combined with a warm late winter – early spring in the current year following a warmer fall and winter. For precipitation, Douglas-fir showed a positive correlation with previous August and September precipitation and current growing season precipitation (i.e. current March to August). This suggests that, in the lower elevation zone, water stress during the growing season could also be a limiting factor for tree growth during dry years, which is in agreement with previous work (Zhang et al., 2000; Case and Peterson, 2005). In the MS zone, Douglas-fir had a negative correlation with winter precipitation (i.e. previous October to current February), while in the IDF zone there was almost no correlation between the chronology indices and winter precipitation variables. This may be because in the MS most winter precipitation falls as snow and snow could be limiting growth by branch breakage or by reducing the length of the growing season. Also, later snowmelt has implications for soil and nutrient cycling processes that affect tree growth. The Douglas-fir chronologies for the MS and IDF zones were more sensitive to temperature than precipitation: the longer the growing period, the better the growth (i.e. positively correlated with previous April, May, and previous October to current January mean temperature). But the summer drought effect also influences tree growth, with this variable being more important for the IDF zone than the MS zone. The reason why Douglas-fir shows consistent relationships between climate and ring growth in both IDF and MS zones could be that it is a more climate sensitive species in comparison with lodgepole pine (Fritts, 1976; Watson and Luckman, 2001; Zhang and Hebda, 2004). As a consequence, I found a similarity in the response of Douglas-fir chronologies to climate variables across its elevation range. Like spruce and lodgepole pine, I also compared my results with other summarized literature that reported relationships between ring growth and climate variables (Table 3.9). My results are somewhat different from these reported relationships, suggesting that they may be site-specific (Zhang and Hebda, 2004).  102  c. PCA and DendroClim 2002  I did not find any major climate component that explains the relationship between tree ring and climate when I used SAS to run principal component analysis. Therefore, I used it as an alternative the widely used dendroclimatology program DendroClim2002, which employs the bootstrap technique to calculate the relationship between tree ring index and climate variables and picks up the ones which are significant (P<0.05). DendroClim2002 and PCA results showed that for lodgepole pine in ESSF and MS zones July precipitation has a positive influence on ring growth because it reduces water stress in the middle of the growing season. I did not find exactly this relationship in the literature, but Graumlich (1991) and Case and Peterson (2007) both reported a positive relationship between tree-ring and growing season/summer precipitation. The positive correlations with January precipitation (snow) are contradictory to the correlation analysis results. But other studies (Table 3.8) have reported similar but not identical relationships (Oleksyn and Fritts, 1991; Graumlich 1991); one study reported a different relationship (Chhin et al., 2008). Therefore, I can only conclude that the relationship between winter precipitation and radial growth is not clear in my study. Based on the methods I used in data analysis, winter precipitation could have either a positively influence on radial growth (snowpack becomes soil water during the subsequent growing season) or a negative influence (e.g. breaking of branches due to snow load). For Douglas-fir, the response functions in the MS and IDF zones agreed with some of my correlation results, but they were also contradictory with other correlation results. Based on DendroClim2002 results, it seems clear that Douglas-fir was sensitive to water stress in an extended growing season that ranges from June to October, and precipitation during this period enhanced tree-ring growth. As for lodgepole pine in the IDF zone and also for spruce in either ESSF or MS zones, DendroClim2002 did not find any significant variables from which to form a response function. It suggested that these two species in these zones may not respond strongly enough to the climate variables to establish a response function.  d. Simple and multivariate regressions  Because none of the relationships I found was strong, I also used regression analysis to  103  explore other climate variables to see if I missed any information in my data. I did not find any strong linear relationships between tree ring indices and climate variables. The lack of significant regression relationships may reflect the weakness of any individual relationships that exist, but comparing across different zones within each species, the regressions show the same pattern across different BEC zones, and are also consistent with what I found in the correlation analysis. Moderate to low values of regression coefficient (r2) are common in dendroclimatological studies like this one (Yu et al., 2007), and low percentage of variance explained by climate variables is common in the literature. My values are in the same range as other previously reported dendroclimatological studies (Briffa et al. 1990, Graumlich 1991, Li et al. 2006, Macías et al. 2006, Girardin et al. 2008).  3.5. Conclusions All three species had some significant correlations with climate variables in the previous year. Previous growing season water stress affects current year radial growth, but the sensitivity of ring growth and the critical month vary from species to species and from zone to zone. Current growing season temperature and water stress also affect tree growth, but the effect was not as strong as these conditions in the previous year. In addition to growing season climate, winter precipitation also acts as an important determinant in tree growth in the following growing season. Depending upon the species and the variables considered, it either acts as the water resource in the following year (e.g. lodgepole pine in ESSF and MS zone in PCA analysis) or as a snow damage factor (lodgepole pine in ESSF and MS zone in correlation and multi-regression analysis). This seasonal change in the effects of climatic factors was reflected in Fritt’s conceptual models (Figures 3.13 and 3.14). Long term climatic changes such as PDO and maybe ENSO had influences on tree growth in this area, with higher growth rate when the North Pacific Ocean was warmer and atmospheric air pressure was higher.  104  Figure 3. 13. The relationship between climate variables (i.e. temperature and precipitation) and tree ring formation during the growing season. (Fritts et al., 1971).  Both conceptual models represent low precipitation and high temperature conditions, and both predict narrow rings as a consequence of these combined conditions. Deterministic relationships vary between these two models. This emphasizes that in these kinds of tree ring formation models, the relationship between tree ring growth and climate is very complex, and different combinations of mechanisms can result in the same patterns of variation in tree ring. But we have to keep in mind that any approach using dendrochronology only describes the relative responsiveness of tree-ring series of that specific species and for that specific site, and it only measures tree ring response at the height at which the core was taken. The absolute radial growth of trees will differ among sites, in large part depending upon environment, and climatic variation may cause changes in tree stem taper, allocation to below ground, and to branch and foliage varies, and allocation to branch and foliage replacement. Breast height ring growth is thus a very incomplete index of whole tree response to climate change. There are other alternative tree growth measurements, such as height, basal area or wood density that could provide additional insight in these relationships (Nigh et al., 2004; Biondi and Qeadan, 2008). While one might expect that high-elevation trees have lower growth rates than lower-elevation trees at the same age, the reasons for this difference may vary, and will probably vary between ecological sites  105  types within any zone.  Figure 3. 14. The relationship between climate variables (i.e. temperature and precipitation) and tree ring formation prior to the growing season. (Fritts et al., 1971).  Despite our recognition that temperature and precipitation affect tree ring growth, the detailed mechanisms involved are not always clear (Rocha et al., 2006) and these relative importances of mechanisms may vary from site to site. Also, while it is known that temperature and precipitation alter tree growth and thereby affect harvestable timber production (Ogaya and Peñuelas, 2007), there is much less literature on how they affect the mechanisms of carbon allocation for individual trees or the overall carbon allocation within forest ecosystem (Raich et al., 2006; Litton et al, 2007). The only well documented review concluded that carbon allocation is related to soil nutrient and moisture conditions (Litton et al, 2007). Perhaps the main take-home-message from my results and their comparison with the literature is that tree growth response to variations in climate variables is complex, will differ between different species that have different climate-related adaptations and growth strategies, and between different BEC zones and site types within each zone. If this conclusion is correct, it poses difficulties for the prediction of forest landscape-scale response to climate change, and raises questions about the predictions of simple models such as biological climatic envelope models. However, I have found some relationship between tree growth and climate, most of which related to growing  106  season water stress and winter precipitation conditions. I also found that lodgepole pine and Douglas-fir were more sensitive to climate compared than hybrid spruce. This information can be used in other area of North America as a reference for future research.  107  3.6. References Alfsen, K.H. 2001. Climate change and sustainability in Europe. CICERO Policy Note. 2001: 03. Anonymous. 1998. Climate-radial growth relationships in some major tree species of British Columbia. Scientia Silvica, extension series, 13. Begon, M., C.R. Townsend, and J.L. Harper. 2006. Ecology, from individuals to ecosystems 4th Ed. Blackweel Publishing. Oxford, U.K. Pp.738. Beniston, M. 2002. Climate modeling at various spatial and temporal scales: where can dendrochronology help? Dendrochronologia. 20: 117-131. Biondi, F. 1 997. Evolutionary and moving response functions in dendroclimatology. Dendrochronologia. 15: 139-150 Biondi, F. and K. Waikul. 2004. DENDROCLIM2002: A C++ program for statistical calibration of climate signals in tree-ring chronologies. Computer and Geosciences. 30: 303-311. Biondi, F. and F. Qeadan. 2008. A theory-driven approach to tree-ring standardization: defining the biological trend from expected basal area increament. Tree-Ring Research. 64: 81-96. Bonnan, G.B. 1989. A computer model of the solar radiation, soil moisture, and soil thermal regimes in boreal forests. Ecological Modelling. 45: 275-306. Bower, A.D., W.T. Adams, D. Birkes and D. Nalle. 2005. Response of annual growth ring components to soil moisture deficit in young, plantation-grown Douglas-fir in coastal British Columbia. Canadian Journal of Forest Research. 35: 2491-2499. Bräker, O.U. 2002. Measuring and data processing in tree-ring research - a methodological introduction Dendrochronologia. 20: 203-216. Briffa, K.R., T.S. Bartholin, D. Eckstein, P.D. Jones , W. Karlén , F.H. Schweingruber and P. Zetterber. 1990. A 1,400-year tree-ring record of summer temperatures in Fennoscandia. Nature. 346: 434-439. Briffa, K.R., T.J. Osborna and F.H. Schweingruber. 2004. Large-scale temperature inferences from tree rings: a review. Global and Planetary Change. 40: 11-26. Case, M.J. and D.L. Peterson. 2005. Fine-scale variability in growth-climate relationship of Douglas-fir, North Cascade Range, Washington. Canadian Journal of Forest Research. 35: 2743-2755. Case, M.J. and D.L. Peterson. 2007. Growth-climate relations of lodgepole pine in the North Cascades National Park, Washington. Northwest Science. 81: 62-75. Chhin S., E.H.T. Hogg, V.J. Lieffers and S. Huang. 2008. Potential effects of climate change on the growth of lodgepole pine across diameter size classes and ecological regions. Forest Ecology and Management. 256: 1692-1703. Cook, E.R. 1985. A time series analysis approach to tree-ring standardization. PhD dissertation, University of Arizona, Tucson. AR. U.S.A. Cook, E.R. and R. Holmes. 1986. Guide for computer program ARSTAN. Laboratory of Tree-Ring Research. University of Arizona, Tucson, AR. U.S.A. Cook, E.R., P.J. Krusic and P.D. Jones. 2003. Dendroclimatic signals in long tree-ring chronologies from the Himalayas of Nepal. International Journal of Climatology. 23: 707-732.  108  Environment Canada. 2003 National Climate Data and Information Archive. Available at: http://climate.weatheroffice.ec.gc.ca/Welcome_e.html Retrieved February 15, 2005. Fritts, H.C. 1976. Tree Rings and Climate. Academic Press, London, Pp.567. Fritts, H.C., T.J. Blasing, B.P. Hayden and J.E. Kutzbach. 1971. Multivariate Techniques for Specifying Tree-Growth and Climate Relationships and for Reconstructing Anomalies in Paleoclimate. Journal of Applied Meteorology. 10: 845-864. Gajewski, K and D Atkinson. 2003. Climate change in the Canadian Arctic. Environmental Reviews. 11: 69-102. Girardin, M.P., F. Raulier, P.Y. Bernier and J.C. Tardiff. 2008. Response of tree growth to a changing climate in boreal centra Canada: A comparison of empirical, process-bases, and hybrid modelling approaches. Ecological Modelling. 213: 2909-228. González-Elizondo M., E. Jurado, J. Návar, M.S. González-Elizondo, J. Villanueva, O. Aguirre and J. Jiménez. 2005. Tree-rings and climate relationships for Douglas-fir chronologies from the Sierra Madre Occidental, Mexico: A 1681-2001 rain reconstruction. Forest Ecology and Management. 213: 39-53. Graumlich, L.J. 1991. Subalpine tree growth, climate, and increasing CO2: an assessment of recent growth trends. Ecology. 72: 1-11. Green S. and H. Griesbauer. 2007. Predicting the responses of interior Douglas-fir to climate change in British Columbia. Report B.C. Forest Science Program #Y071270. Prince George, BC. Canada. Pp. 5. Green S. and Y. Miyamoto. 2006. Characterizing the growth responses of three co-occurring northern conifer tree species to climate variation across a range of conditions. A summary report for a study funded by the B.C. Forest Science Program FSP project #Y061107. Prince George, BC. Canada. Pp. 6. Grissino-Mayer, H.D. 2001. Evaluating crossdating accuracy: a manual and turorial for the computer program COFECHA. Tree-ring Research. 57: 205-221. Grissino-Mayer, H.D., R.L. Holmes and H.C. Fritts. (Eds). 1996. The International Tree-Ring Data Bank program library, user’s manual, version 2. International Tree-Ring Data Bank, Tucson, AR. U.S.A. Google Inc. 2009. Google Maps. Available at: http://maps.google.ca. Retrieved January 6, 2009. Halfon, E. 1989. Probabilistic validation of computing simulations using the bootstrap. Ecological Modelling. 46: 213-219. Havranek, W., W. Tranquillini. 1995. Physiological processes during winter dormancy and their ecological significance. In: Smith, W.K, T.M. Hinckley (Eds.). Ecophysiology of coniferous forests. Academic Press, San Diego, USA. p. 95-124. Hegerl, G.C., T.J. Crowley, M. Allen, W.T. Hyde, H.N. Pollack, J. Smerdon and E. Zorita. 2007. Detection of human influence on a new, validated 1500-year temperature reconstruction. Journal of Climate. 20: 650-666. Hogg, E.H. 1994. Climate and the southern limit of the western Canadian boreal forest. Canadian Journal of Forest Research. 24: 1835-1845. Hogg, E.H. 1997. Temporal scaling of moisture and the forest-grassland boundary in western Canada. Agricultural and Forest Meteorology. 84: 115-122. Holman, M.L. and D.L. Peterson. 2006. Spatial and temporal variability in forest growth in the Olympic Mountains, Washington: sensitivity to climatic variability. Canadian Journal of Forest Research. 36: 92-104.  109  Holmes, R.L. 1983. Computer-Assisted quality control in tree ring dating and measurement. Tree-Ring Bulletin. 43: 69-78. Hughes, M.K. 2002. Dendrochronology in climatology – the state of the art. Dendrochronologia. 20: 95-116. Hunt, E.R. Jr., F.C. Martin and S.W. Running. 1991. Simulating the effects of climatic variation on stem carbon accumulation of a ponderosa pine stand: comparison with annual growth increment data. Tree Physiology. 9: 161-171. Intergovernmental Panel on Climate Change (IPCC). 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge University Press. Pp.1032. Jones, F.W. and M.L. Parker. 1970. G.S.C. Tree-ring scanning densitometer and data acquisition system. Tree-ting Bulletin. 30: 23-31. Jensen M.E., R.D. Burman and R.G. Allen. (Eds.) 1990. Evapotranspiration and irrigation water requirements. Manuals and Reports on Engineering Practice 70. American Society of Civil Engineers. New York. NY. U.S.A. Pp.332. Kienast F., F.H. Schweingruber, O.U. Bräker and E. Schär. 1987. Tree-ring studies on conifers along ecological gradients and the potential of single-year analyses. Canadian Journal of Forest Research. 17: 683-696. Kienast, F., O.Wildi and S. Ghosh. 2007. On selected issues and challenges in dendroclimatology. A Changing World. Challenges for Landscape Research. p. 113-132. Krajina, V.J. 1969. Ecology of forest trees in British Columbia. Ecology of Western North America. 2: 1-146. Legates, D.R. 1991. The effect of domain shape on principal components analysis. International Journal of Climatology. 11: 135-146. Levitt, J. 1980. Responses of plant to environmental stresses. Academic Press, London, UK. Pp. 297. Li, C., M.D. Flannigan and I.G.W. Corns. 2000. Influence of potential climate change on forest landscape dynamics of west-central Alberta. Canadian Journal of Forest Research. 30: 1905-1912. Litton, C.M., J.W. Raich and M.G. Ryan. 2007. Carbon allocation in forest ecosystems. Global Change Biology. 13: 2089-2109. Lloyd, D., K. Angove, G. Hope and C. Thompson. 1990. A Guide to Site Identification and Interpretation for the Kamloops Forest Region. British Columbia Ministry of Forests. Victoria. BC. Canada. Pp. 407. Luckman, B.H. and R.J.S. Wilson. 2005. Summer temperatures in the Canadian Rockies during the last millennium: a revised record. Climate Dynamics. 24: 131-144. Luckman, B.H., K.R. Briffa, P.D. Jones and F.H. Schweingruber. 1997. Tree-ring based reconstruction of summer temperatures at the Columbia Icefield, Alberta, Canada, AD 1073-1983. Holocene. 7: 375-389. MacDonald, G.M., J.M. Szeicz, J. Claricoates and K.A. Dale. 1998. Response of central Canadian treeline to recent climatic changes. Annals of the association of American Geographers. 88: 183-208. Macías, M., L. Andreu, O. Bosch, J.J. Camarero and E. Gutiérrez. 2006. Increasing aridity is enhancing silver fir (Abies alba mill.) water tress in its south-western distribution limit. Climatic change. 79: 289-313.  110  Mantua, N.J., S.R. Hare, Y. Zhang, J.M. Wallace, R.C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society. 78: 1069-1079. Nakawatase J.M. and D.L. Peterson. 2006. Spatial variability in forest growth – climate relationships in the Olympic Mountains, Washington. Canadian Journal of Forest Research. 36: 77-91. Nigh, G.D., C.C. Ying and H. Qian. 2004. Climate and productivity of major conifer species in the Interior of British Columbia, Canada. Forest Science. 50: 659-671. Nitschke, C.R. 2006. Integrating climate change into forest planning: a spatial and temporal analysis of landscape vulnerability. PhD thesis. University of British Columbia. Vancouver, BC. Canada. Oleksyn J. and H.C. Fritts. 1991. Influence of climatic factors upon tree rings of Larix decidua and L. decidua x L. kaempfer from Pulawy, Poland. Trees-estructure and function. 5: 75-82. Ogaya, R. and J. Peñuelas. 2007. Tree growth, mortality, and above-ground biomass accumulation in a holm oak forest under a five-year experimental field drought. Plant Ecology. 189: 291-299. Parker, M.L. and K.R. Meleskie. 1970. Preparation of X-ray Negatives of tree-ring specimens for Dendrochronological Analysis. Tree-ting Bulletin. 30: 11- 22. Pichler, P. and W. Oberhuber. 2007. Radial growth response of coniferous forest trees in an inner Alpine environment to heat-wave in 2003. Forest Ecology and Management. 242: 688-699. Pojar J., K. Klinka and D.V. Meidinger. 1987. Biogeoclimatic Ecosystem Classfication in British Columbia. Forest Ecology and Management. 22: 119-154. Polge, H. 1970. The use of X-ray densitometric methods in Dendrochronology. Tree-ring Bulletin. 30: 1-10. Quinn, G.P. and M.J. Keough. 2002. Experimental design and data analysis for biologist. Cambridge University Press. Cambridge. U.K. Pp. 537. Raich, J.W., A.E. Russell, K. Kitayama, W.J. Parton and P.M. Vitousek. 2006. Temperature Influences Carbon Accumulation in Moist Tropical Forests. Ecology. 87: 76-87. Riverside Forest Products Ltd. 2001. Riverside’s Tree Farm Licence 49 Ecological Forest Stewardship Project. Available at: http://www.riverside.bc.ca/woodlands/tfl49-index.htm Retrieved on January 31, 2003. Rocha, A.V., M.L. Goulden, A.L. Dunn and S.C. Wofsy. 2006. On linking interannual tree ring variability with observations of whole-forest CO2 flux. Global Change Biology. 12: 1378-1389. Sarris, D., D. Christodoulakis and C. Korner. 2007. Recent decline in precipitation and tree growth in the eastern Mediterranean. Global Change Biology. 13: 1187-1200. Savva, Y., J. Oleksyn, P.B. Reich, M.G. Tjoelker, E.A. Vaganov and J. Modrzynski. 2006. Interannual growth response of Norway spruce to climate along an altitudinal gradient in the Tara Mountains, Poland. Trees. 20: 735-746. Schweingruber, F.H., H.C. Fritts, O.U. Bräker, L.G. Drew and E. Schar. 1978. The X-ray technique as applied to Dendroclimatology. Tree-ting Bulletin. 38: 61-91. Seely, B. and C. Welham. 2008. Factorial analysis of soil cover, competing vegetation, aspect, and climate effects on tree water stress in Oil Sands reclamation using the ForWaDy model. Unpublished report. FORRX Consulting Inc. Belcarra, B.C. Canada. Pp. 22.  111  Skomarkova, M.V., E.A. Vaganov, M. Mund, A. Knohl, P. Linke, A. Boerner and E.D. Schulze. 2006. Inter-annual and seasonal variability of radial growth, wood density and carbon isotope ratios in tree rings of beech (Fagus sylvatica) growing in Germany and Italy. Trees. 20: 571-586. Su, H.X., W.G. Sang, Y.X. Wang and K.P. Ma. 2007. Simulating Picea schrenkiana forest productivity under climatic changes and atmospheric CO2 increase in Tianshan Mountains, Xinjiang Autonomous Region, China. Forest Ecology and Management. 246: 273-284. Thornton, P.E. 2000. General notes for MTCLIM version 4.3. Numerical Terradynamic Simulation Group. School of Forestry, University of Montana. MN. U.S.A. Thornton, P.E., S. W. Running and M.A. White. 1997. Generating surfaces of daily meteorological variables over large regions of complex terrain. Journal of Hydrology. 190: 214-251. Watson, E. and B.H. Luckman. 2001a. Dendroclimatic reconstruction of precipitation for sites in the southern Canadian Rockies. Holocene. 11: 203-213. Watson, E and B.H. Luckman. 2001b. The development of a moisture-stressed tree-ring chronology network for the southern Canadian cordillera. Tree-ring Research. 57: 149-168. Wilson, R.J.S. and B.H. Luckman. 2003. Dendroclimatic reconstruction of maximum summer temperatures from upper treeline sites in Interior British Columbia, Canada. Holocene: 13: 851-861. Yu, D.P., G.G. Wang, L.M. Dai and Q.L. Wang. 2007. Dendroclimatic analysis of Betula ermanii forests at their upper limit of distribution in Changbai Mountain, Northeast China. Forest Ecology and Management. 240: 105-113. Zhang, Y., J.M. Wallace, D.S. Battisti. 1997. ENSO-like interdecadal variability: 1900-93. Journal of Climate 10: 1004-1020. Zhang Q.B., R.I. Alfaro and R.J. Hebda. 1999. Dendroecological studies of tree growth, climate and spruce beetle outbreaks in Central British Columbia, Canada. Forest Ecology and Management. 121: 215-225. Zhang, Q.B. and R.J. Hebda. 2004. Variation in radial growth patterns of Pseudotsuga menziesii on the central coast of British Columbia, Canada. Canadian Journal of Forest Research. 34: 1946-1954. Zhang Q.-B., R.J. Hebda, Q.-J. Zhang and R.I. Alfaro. 2000. Modeling tree-ring growth responses to climatic variables using artificial neural networks. Forest Science. 46: 229-239.  112  4. Development and Preliminary Testing of a Tree Productivity -Climate Model with Dendroclimatological Data10  4.1. Introduction There are many models already developed for assessing the impacts of climate change on forest ecosystems in different aspects (Loehel and LeBlanc, 1996). Some of them focus on species distributions and abundances and are categorized as ecological response surface models or biogeographic correlation models. These models are built based on the assumptions that climate ultimately restricts species distributions and abundances, and that the relationships between climate and species distributions and abundances in the past (or now) will remain the same in the future (Pearson and Dawson, 2003; Hamann and Wang, 2005). However, they lack any representation of biotic interactions, species evolutionary change (adaptability) and species dispersal strategies and limitations (Pearson and Dawson, 2003), not to mention they also lack to consider the abiotic factors such as hydrological and nutrient cycles in a forest stand. Therefore other researchers have looked for another approach: forest stand simulation models, which integrate species-specific information regarding species characteristics and their interactions with environmental factors. Examples of this type of model include JABOWA (Botkin et al., 1972; Botkin, 1993) and FORET (Shugart, 1984). These models deal with changing of specie composition within a stand and also the biomass changes under different climate conditions. Following the ideas from stand simulation models, one group of models assess the impacts of climate change from an energy flow perspective. These models calculate how much primary production and biomass will be produced under climate change scenarios. Such models include Forest-BGC (Korol et al., 1995; Running and Coughlan, 1988), BIOMASS (McMurtrie et al., 1990, McMurtrie and Landsberg, 1992), and LINKAGES (Pastor and Post, 1985), among others. Each of these models has its advantages and shortcomings, but are all reasonably good when being used to explore the impacts of climate in terms of ecophysiological responses that can be  10  A version of this chapter will be submitted for publication as Lo, Y.-H., Seely, B., Blanco, J.A., Welham, C., Kimmins, J.P. “Development and Preliminary Testing of a Tree Productivity -Climate Model with Dendroclimatological Data”.  113  translated after into changes in species distribution, species abundance and plant growth. However, most of these models are population-level models that do not include the representation of other ecosystem components such as soil, belowground biomass or understory, while these components could be the keys, in some situations, to explain how the forest ecosystem behaves under changing climate condition (Bi et al., 2007). Therefore, the effects of climate change on these components should also be considered. On the other hand, the demands for reduction of atmospheric CO2 emissions by practicing different forest management strategies are continually growing, so is the idea of carbon credits (i.e. the right to release carbon to the atmosphere that has been created by planting or other practises), and not many models deal with these management topics. Most models (i.e. carbon budget models) calculate how much carbon can be stored within the current forest ecosystem and in the future under double CO2 concentration conditions or other climate scenarios provided by global circulation models (GCMs) (IPCC, 2007). However, not many of the current carbon budget models or landscape, stand or tree growth models can tell us if there is a way to change future CO2 levels by changing current forest management practices. In contrast to many other models, FORECAST - which is a hybrid forest ecosystem management model - is designed to explore ecosystem processes and to apply knowledge thereof in the design of the management of forest stands for multiple values, including carbon storage (Seely et al., 2002; Wei et al., 2003). After the discussion of the impacts of climate change on forest ecosystem in chapter one and also the model simulation ability of these impacts now, I draw the summary flowchart of how climate change will affect forest ecosystem production (Figure 4.1). However, here I only consider two major climatic variables, temperature and precipitation, and I do not include some factors such as CO2 or O3. One reason is because we usually lack of historical data of how daily or monthly concentration change of these gases as an input for model simulation. Another reason is because these variables are more limited their influences on individual plant level (i.e. growth, water use efficiency, and carbon allocation within the plant) while my summary is more focus on the stand or ecosystem scale. But we have to keep in mind that these factors also play important roles in the whole global warming issue.  114  Figure 4. 1. Flow chart of climate change impacts on forest ecosystem production. It only focuses on temperature and precipitation variables (bold text), and it does not consider other climate change variables such as CO2 or O3.  In this Chapter I describe the present state of FORECAST and ForWaDy models. The latter one is an energy and water-balance model, which will link to FORECAST and will give it the ability to capture the hydrological dynamic within a forest ecosystem. Also in this chapter, I demonstrate a theoretic approach of how temperature and precipitation affect annual production of a single tree using STELLA modeling framework as a first step to understand growth response under different climate conditions. As one can see from Figure 4.1, the whole issue is very complex and there are a lot of interactions which we are lack of the knowledge to explain the processes at this stage. However, one can always start from the simple relationship and then add more and more component to make the structure more and more complex as our knowledge grow. Therefore I start from two basic climatic variables and try to explore how they affect net primary production of a tree. And then I compare the simulation results with tree ring data from chapter  115  three to see how well this simple model captures the reality.  4.2. Models of tree growth and climate 4.2.1. FORECAST Dr. J.P. Kimmins and his team initiated the development of FORECAST in the late 1970s as a stand level ecosystem management model with an annual time step (Kimmins et al., 1986). Prior to that time most forest models were based on either empirical growth and yield relationships or purely eco-physiological processes. Both of these two major categories of models have their advantages and weaknesses. FORECAST combines the best features of both types of model by using local growth and yield data to derive the potential rate of essential ecosystem processes (Kimmins et al., 1999; Seely et al., 2002), thereby creating a “hybrid” simulation tool. It is a generic ecosystem model that can be calibrated for any forest ecosystem worldwide (Wang et al., 1995; Wei and Kimmins, 1995; Morris et al., 1997; Wei et al., 2000; Seely et al., 2002; Welham et al., 2002), although, as with any forest model, the relative simplicity of boreal and northern temperate forests can generally be represented more easily than multi-species, complex structure, mature tropical rain forests. However, because FORECAST can be configured for a wide variety of levels of structural and functional complexity, it is potentially one of the best modeling approaches for application to all types of forest. Figure 4.2 presents the input /output files and program structure of FORECAST. For a detail model description, see Kimmins et al. (1999). In FORECAST, outputs from growth and yield models and/or field data, together with a variety of other empirical information from the literature, are entered into input files (TREEDATA, PLANTDATA, BRYODATA and SOILDATA). These input data are then used in a series of “setup” programs ( TREEGROW, PLANTGROW, BRYOGROW and SOILS, respectively, Figure 4.2) to calculate the historic rates and values of various ecosystem processes related to productivity of selected tree, shrub, herb and bryophyte species (the user controls which life forms are simulated by the data entered into the “setup” input files). These process rates and various other data from the “setup” stage are then used as drivers of the ECOSYSTEM program to simulate the possible futures for the forest ecosystem under different management strategies and/or natural disturbance regimes (Kimmins et al., 1999; Seely et al., 2002).  116  Two of the key ecosystem processes represented in FORECAST are photosynthesis and decomposition. The former drives the simulation of potential net primary productivity and the latter determines the availability of soil nutrients, which then limits the extent to which potential, light-limited productivity can be realized. The fundamental driving variable in FORECAST is shade-corrected foliage nitrogen efficiency (kilograms of total new plant biomass produced per kilogram of foliage nitrogen, corrected for shading), which is limited by both light and nutrient availability (Kimmins et al., 1999). The ECOSYSTEM program allows the user to set different harvesting (of trees and/or the other plant life forms being represented) and silvicultural strategies and then to test the consequences for plant biomass change.  Figure 4. 2. Files and program structure of FORECAST. (Kimmins et al., 1999)  Projections of stand growth and ecosystem dynamics are based upon a representation of the rates of key ecological processes regulating the availability of, and competition for, light and nutrient resources, and a representation of moisture competition (this is near completion using ForWaDy - see explanations below). The rates of these processes are calculated from a combination of historical bioassay data (biomass accumulation in component pools, stand density, etc.) and measures of certain ecosystem variables (decomposition rates, photosynthetic saturation  117  curves, for example) by relating ‘biologically active’ biomass components (foliage and small roots) to calculated values of nutrient uptake, the capture of light energy, and net primary production. Using this ‘internal calibration’ or hybrid approach (it combines experience of the past – “historical bioassay” – with understanding of key ecosystem processes to forecast possible future ecosystem forms and functions – process simulation), the model generates a suite of growth properties for each tree and plant species to be represented. These growth properties are subsequently used to model growth as a function of resource availability and competition. They include (but are not limited to): 1) photosynthetic efficiency per unit foliage biomass based on relationships between foliage biomass, simulated self-shading, and net primary productivity after accounting for litterfall and mortality; 2) nutrient uptake requirements based on rates of biomass accumulation and literature- or field-based measures of nutrient concentrations in different biomass components on different site qualities; and 3) light-related measures of tree and branch mortality derived from stand density and live canopy height input data in combination with simulated light profiles. FORECAST performs many calculations at the stand level but it includes a submodel that disaggregates stand-level productivity into the growth of individual stems with information input by the user on stem size distributions at different stand ages. Top height and DBH are calculated for each stem and used in a taper function to calculate total and individual gross and merchantable volumes. The approach of many models to represent climate change effects is to focus on the impacts of altered climatic variables on a limited set of forest ecosystem components and processes. However, to assess some of key ecosystem-level effects of climate change on forest growth and development as well as the effectiveness of potential adaptation/mitigation strategies, a model is needed that addresses the complexity of ecosystem structure and processes and how they might respond to altered temperature and precipitation regimes. FORECAST is such a model because it has both prediction and management simulation capabilities (Seely et al., 2002; Wei et al., 2003) and has been successfully validated against field data for growth, yield, ecophysiological and soil variables (Bi et al., 2007; Blanco et al., 2007; Seely et al., 2008). FORECAST, like all models, has its shortcomings and limitations. The present version is not able to represent climate change and moisture competition. Also, the soil processes in FORECAST are relatively simplified compared with soil-focused models like CENTURY (Parton et al., 1993) The ability to explore  118  options for climate change mitigation through forest management, and the ability to forecast possible forest futures under climate change have become essential. Consequently, the Forest Ecosystem Simulation Research Group at UBC decided to add the hydrological and moisture limitation module ForWaDy and climate change capabilities to FORECAST. They are going to develop a new version - FORECAST Climate - and to use this to research the possible consequences of the global trend of climate change for forest ecosystems.  4.2.2. ForWaDy (Forest Water Dynamics) model ForWaDy is a forest hydrology model used to simulated forest water dynamics under given climate and forest stand structure conditions. It uses a daily time step to capture precipitation events (Seely et al., 1997). It was written in STELLA Research v7.0 software (High Performance Systems Inc., 2002), which is particularly suitable for modelling ecosystem-level processes (Constanza and Gottlieb, 1998). Figure 4.3 shows the structure of this model. ForWaDy uses an energy budget approach to calculate potential evapotranspiration (PET) as a function of climate (solar radiation, mean air temperature, precipitation and snow depth), stand structure and soil texture (Seely et al., 1997). It simulates precipitation interception by the vegetation canopy and competition between plants for water in the soil under different forest stand conditions, and calculates water demand by different canopy layers and within different soil layers. After calculating the difference between water supply and water demand of the tree, ForWaDy provides a tree water stress index: TDI (transpiration deficit index), which will be used as a modifier of tree growth in FORECAST. The advantages of this model are that it is written in a user-friendly language (i.e. STELLA) and it does not have a high input data requirement to run the model. Also, the processes within the model come from well-tested existing models or equations where possible (Seely et al., 1997) and it has been successfully tested at Shawnigan Lake and Montane Alternative Silviculture Systems sites (Seely and Welham, 2008). A detailed description is presented in Seely et al. (1997). ForWaDy also has recently been used successfully to analyze the possible linkages between western red cedar dieback on Vancouver Island and climatic factors (Seebacher 2007).  119  Rain  Snow  Evaporation  Sublimation Canopy Interception  ForWaDy  Snow throughfall  Air temp melt  Throughfall  Snowpack Radiation melt  Litter layer  Transpiration Deficit Index  Runoff Infiltration Humus layer  Canopy Transpiration Demand  Forest floor percolation  Canopy Transpiration  Soil A Understory Transpiration  Soil A percolation  Soil B Soil B percolation  Subsoil  Interflow  drainage Outflow  Figure 4. 3. Flow chart of ForWaDy compartments and processes (Seely et al., 1997).  4.3. Development of a STELLA tree productivity – climate model In order to explore how climate affects tree growth, I used STELLA v7.0 (High Performance Systems Inc., 2002) to write a simple model containing tree-growth-climate relationships. STELLA is a user-friendly, visual modelling language that facilitates the construction of models using basic building blocks. The objective of the research reported in this section is to explore the relationship between climate and tree growth, to be used as a preparation to implement the 120  idea into FORECAST Climate, and to check if our approach to simulate the influence of climate on tree-growth is consistent with current knowledge.  4.3.1. Tree productivity-climate model description In this tree productivity-climate model (Figure 4.4, C.1), there were climate inputs of two ecological zones (ESSF and IDF). These two zones represented of high and low elevation ecosystems, respectively. Based on previous research on relationships of temperature/precipitation and growth rate (Daubenmire, 1974; Salisbury and Ross, 1992; Cai and Dang, 2002; among others), I developed three different potential optimum growth responses to changing temperature and also three to changing precipitation. These curves represent different generic tree species with different climatic adaptations and also as the ranges of the possibility for plant to grow under extreme or moderate conditions (Figure 4.5). In the temperature part (upper panel of Figure 4.5), curve T1 is the species reaches its maximum productivity at low temperature range while T2 and T3 are the species have their maximum productivity at medium and high temperature range, respectively. While the lower panel of Figure 4.5 shows three different kinds of growth response to precipitation levels: drought tolerant (curve P1), intermediate moisture response (curve P2), and more water demanding (curve P3). The user can choose between these different curves of growth response to various temperature and moisture combinations by using the selection tool in the model (Figure 4.4, C.1), A frost effect is also considered in this model. I borrowed the concept from Levitt (1980) and Dang et al. (1992) to make an additional NPP multiplier based on the number of frost days within the growing season. I calculated the frost days (i.e., daily minimum temperature < 0 °C) from May to September (growing season in Okanagan Valley), and then transferred each month’s accumulated frost days into a frost index (Figure 4.6) for each month during the growing season, which is also a multiplier modifying the user defined base rate of net primary production (NPP). I put these growth-climate responses curves into a STELLA-written model (Figure C.1), creating a factorial experiment with nine different response scenarios (T1P1: cold dry optimal to T3P3: hot wet optimal). Average monthly climate data from the Westwold weather station for the period 1922-1997 were generated for two ecological zones with MTCLIM 4.3 (explained in Chapter Three). These two different sets of climate data were corresponding to high elevation site (the ESSF ecological zone) and low elevation site (the IDF ecological zone). After this  121  process, the tree productivity-climate model uses these monthly meteorological data to calculate the multipliers of monthly temperature and precipitation effect. Then these multipliers are fed into the monthly net primary production function, modifying the base NPP rate to calculate how much potential NPP is produced every month based on the nine combinations of different temperature/precipitation response scenarios. The output of modified monthly NPP is then used to calculate current growing season (May to September), early summer (June and July), and annual NPP. These summaries of projected NPP, for the three periods, were subsequently compared against the tree ring residual chronologies from Chapter Three using regression as the validation of model performance. Meanwhile, because from Chapter Three results, tree ring chronologies are strongly correlated with previous year climate, I also compare those chronologies with previous projected NPP, and also with the average of projected NPP from two conjunctive years.  122  Climate Scenarios  Temperature record Species optimum temperature scenarios  Precipitation record  Frost index  Ti  Species optimum precipitation scenarios  Pi  Calculate monthly relative potential NPP Output annual NPP  Figure 4. 4. Flow chart of tree productivity-climate model. Squares represent the input functions and output data. Three oval-shaped objectives are climatic data inputs. Ti and Pi are the temperature and precipitation multipliers to modify monthly base rate NPP. Three thick black lines are the major control factors of potential monthly NPP.  123  Temperature multiplier 1  Multiplier  0.8  T2  T1  0.6  T3  0.4 0.2 0 0  5  10  15 20 25 Temperature (degree C)  30  35  Multiplier  Precipitation multiplier 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0  P1  0  P2  5  P3  10 15 Precipitation (cm)  20  Figure 4. 5. Relationship between growth rate multiplier and air temperature (upper panel) and precipitation (lower panel) for three hypothetical species of each climate variables.  124  25  Frost multiplier 1  Multiplier  0.9 0.8 0.7 0.6 0.5 0.4 0  5  10  15  20  25  30  Frost Days Figure 4. 6. Relationship between frost days of the month and frost index for the hypothetical species.  4.3.2. Simulation results Figure 4.7 shows the seasonal temperature and precipitation effects on NPP of different combined scenarios which were calculated with climate data from high and low elevation ecological zones, respectively. For the high elevation site (Figure 4.7 upper panel), the figure from left to right shows a gradient of increasing moisture limitation by declining growing season precipitation index from P1 to P3. Patterns for all nine combinations (T1P1 to T3P3) were presented in Appendix C (Figure C.2 and C.3). T1P1 represents species adapted to a cold and dry environment, while T3P3 represents species which has optimum growth rate at warm and wet growing condition. From Figure 4.7 and additional results from Appendix C, the model seems more sensitive to precipitation than temperature. The fluctuation of the differences among temperature adaptation scenarios is smaller than among precipitation adaptation scenarios (i.e. the maximum value for temperature is always close to 1 for all the scenarios while the maximum values for precipitation varies).  125  Figure 4. 7. Seasonal limiting factors changing for different combinations of temperature and precipitation growth adaptation scenarios at high elevation (climate for ESSF zone, upper panel) and low elevation (climate for IDF zone, bottom panel) for three consecutive years. The thin solid line is the temperature multiplier and the dot line is precipitation multiplier. The thick solid line is the potential NPP (calculated with a base rate of 1). T1 represents temperature adaptation scenario for low temperature. P1 to P3 represent moisture adaptation to low precipitation (i.e. optimum growth at dry condition) to high precipitation (i.e. optimum growth at wet condition)  Figure 4.7 shows that from year to year, the temperature patterns are similar while the precipitation patterns vary. And it also shows that even under different temperature scenarios (Figure C.2, C.3), the length of the growing season only has little change. Descriptive statistics of projected potential NPP in June and July (early summer, Table 4.1) and tree ring residual chronologies from Chapter Three11 (Table 4.2) are listed. Among different temperature and precipitation scenarios, those contained P1 scenario (drought optimum) were the ones had the most similar average values with observed tree ring index (Figure 4.8). They also have the smallest and most similar standard deviation and value range 11  All the comparisons were compared using a relative scale, not the actual values.  126  compare to other scenarios. The comparing results of current year projected NPP (i.e. growing season, early summer and annual potential NPP) and tree ring residual chronologies are showed in both here and Appendix C. Only the significant ones (α = 0.05), which were from the chronologies of Douglas-fir, are listed here (Figures 4.9 to 4.14, Table 4.3, 4.4). Those which are not significant (i.e. hybrid spruce and lodgepole pine) are listed in Appendix C. Because results for current growing season and current annual NPP are all not significant for hybrid spruce and lodgepole pine, graphic results are not provided in this thesis to avoid excessive figures. From Figures 4.9 to 4.12, among three different precipitation scenarios, the P1 scenario (optimum growth happens in dry condition) shows a better match for Douglas-fir chronologies in both MS and IDF zones compare to P2 and P3 scenarios (moderate soil water requirement and highly water demanded, respectively). And same results are showed in both climate input scenarios (i.e. high and low elevation climate). In these four figures, they also show that the simulation results fluctuated more compare with tree ring chronologies. In addition, after 1985, the simulated NPP results do not match very well with tree ring chronologies. Therefore, I compared the same data set again but excluded the data after 1985. Results showed important increases of the ability of model performance (i.e. r2 increased) and also the significant relationships (α = 0.05) between simulation results and tree ring chronologies increased, although the variability explained remained low (under 26%). These results are showed in Figures 4.13, 4.14 and Tables 4.3, 4.4. Within each figure of Figure 4.8 to 4.11, if we look from top to bottom, they show that as the precipitation adaptation scenario changes (i.e. from adaptation to a drier climate (P1) to adaptation to a more moisture favourable one (P3)), the fluctuation of simulated NPP becomes bigger. Meanwhile, if we only compare the differences between different temperature adaptation scenarios versus the differences between different precipitation adaptation scenarios, the difference is bigger in precipitation part (e.g. T1P1 vs. T1P2 and T1P3) than between temperature part (e.g. T1P1 vs. T2P1 and T3P1). Within each chronologies (i.e. MS Douglas-fir and IDF Douglas-fir) but different elevational climate data (i.e. Figure 4.9 vs. 4.10 and Figure 4.11 vs. 4.12), it seems that with low elevation climate data (Figure 4.10 and 4.12), model sensitivity to temperature adaptation scenarios decreases as the precipitation adaptation scenarios become moister (i.e. in P3 scenarios, all  127  three temperature adaptation scenarios results overlay with each other), but this does not happen when using climate data from the high elevation site (Figure 4.9 and 4.11). For the other two species, the results are not as clear as for Douglas-fir; therefore, I put their results in Appendix C. Table 4. 1. Descriptive statistics of projected June and July potential NPP for high and low elevation climate data inputs. Data are in % of maximum potential NPP. See text for explanations for temperature and precipitation scenarios.  High elevation climate data Scenarios  Average  Std  T1P1  76.1  17.1  74.2  T1P2  58.0  20.3  T1P3  29.9  T2P1  low elevation climate data  range Average  Std  range  60.8  19.7  84.0  86.7  40.8  20.0  90.9  23.4  98.9  18.8  19.7  98.2  74.1  16.9  72.3  62.5  19.6  83.3  T2P2  63.9  20.5  84.3  41.9  20.2  90.5  T2P3  32.3  23.7  98.7  18.8  19.7  98.2  T3P1  73.4  16.2  72.7  68.1  19.6  80.9  T3P2  64.6  19.6  82.4  43.6  20.2  90.0  T3P3  36.5  24.6  98.4  19.0  19.8  98.2  Table 4. 2. Descriptive statistics of lodgepole pine (Pl), hybrid spruce (Sx) and Douglas-fir (Fd) in different BEC zones (ESSF, MS and IDF). Data are in % of maximum tree ring index.  Residual chronologies  Average  Std  range  ESSF Pl  76.6  10.1  57.1  MS Pl  77.0  9.5  51.8  IDF Pl  71.9  11.4  58.6  ESSF Sx  76.2  7.6  41.4  MS Sx  78.8  7.9  48.4  MS Fd  76.0  9.9  48.0  IDF Fd  76.5  10.4  44.7  128  Figure 4. 8. Average (dots) and standard deviation (lines) of projected potential NPP in June and July (left panels) and tree ring index (right panels). Horizontal ticks mark the minimum values. Data are in % of maximum potential NPP or maximum tree-ring index. See main text for the description of temperature and precipitation scenarios.  129  Figure 4. 9. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at the MS zone. The thin lines are simulation results for the combination of temperature and precipitation adaption scenarios applied for the high elevation site (ESSF). Missing points in the simulation results are due to lack of climate data inputs.  130  Figure 4. 10. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at the MS zone. The thin lines are simulation results for the combination of temperature and precipitation adaption scenarios applied for the low elevation site (IDF). Missing points in the simulation results are due to lack of climate data inputs.  131  Figure 4. 11. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology in the IDF zone. The thin lines are simulations results for the combination of temperature and precipitation adaption scenarios applied for the high elevation site (ESSF). Missing points in the simulation results are due to lack of climate data inputs.  132  Figure 4. 12. Simulation results comparing Jun and July NPP with tree ring index data. The thick lines are Douglas-fir chronology at IDF zone. Thin lines are simulations results for the combination of temperature and precipitation adaption scenarios applied for the low elevation site (IDF). Missing points in the simulation results are due to lack of climate data inputs.  133  Comparing Tables 4.3 and 4.4 (which report regression analysis results from the relationship between the tree ring chronologies and the STELLA simulations described above), there are more significant relationships when comparing model results with Douglas-fir chronologies from IDF zone than from MS zone. Table 4.3 is the comparison between model simulations and MS zone chronology using the full climate series (1922-1997). The significant relationships only happen in P1 scenarios (drought-adapted species) and only with June and July NPP. None of the other periods (i.e. May to September or annual) nor other precipitation adaptation scenario (P2 and P3) have any significant relationships. However, even for the significant relationships, the model can only explain up to 7% of the variance of tree ring chronology when the full climate data series were used. The relationships between model results and chronologies calculated with climate data from low elevation (IDF) are a little bit stronger than the relationships from high elevation climate (Table 4.2). However, I also noticed that the matches are not well between tree-ring residual chronologies and simulation results for the data points after 1985. I re-compared the data again with the data before 1985 and both tables show explaining ability (r2) and number of those which have significant relationships increased, doubling to reach 14% of the variability. In Table 4.4 (model simulation results vs. IDF Douglas-fir chronology), significant relationships are found not only for the June and July NPP, but also with summer and annual NPP. The strength of the relationships for the different periods, however, is different. For June and July NPP, all the temperature and precipitation scenarios in both high and low elevation environments showed significant relationships with Douglas-fir tree ring chronology, while for growing season NPP (May to September), there were more significant relationships when using climate data from a low elevation environment, and they only happen in the P3 scenario (optimum growth in wet environments) and one P2 scenario. As for the annual NPP regression with tree-ring chronology part, only P3 in low elevation environment showed significant relationship. Finally, if we compare the model performance (r2) with IDF Douglas-fir chronology with MS Douglas-fir chronology, the previous has higher values.  134  Table 4. 3. Linear regression results between simulation outputs vs. tree ring chronologies for Douglas-fir in the MS zone. Two data sets were used; one is from 1922 to 1985 while the other one is from 1922 to 1997. r2 improvement was calculated from the difference of the two data sets. Jun & Jul, May to Sep and annual represent the sum of NPP for different periods calculated from the tree productivity-climate model. Bold font in slope column represents statistically significant relationships (p<0.05).  Projected NPP Climate period Scenarios  ESSF  IDF  Average  T1P1 T1P2 T1P3 T2P1 T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 T2P2 T2P3 T3P1 T3P2 T3P3  Data from 1922 to 1985 Jun & Jul Slope r2 0.177 0.096 0.129 0.099 0.074 0.122 0.165 0.085 0.133 0.093 0.074 0.112 0.170 0.086 0.127 0.080 0.074 0.098 0.154 0.121 0.121 0.137 0.063 0.123 0.165 0.085 0.149 0.107 0.063 0.123 0.526 0.058 0.124 0.132 0.063 0.123 0.142 0.104  May to Sep Slope r2 0.123 0.030 0.114 0.060 0.091 0.123 0.093 0.015 0.115 0.049 0.081 0.095 0.090 0.011 0.112 0.037 0.075 0.075 0.136 0.073 0.139 0.113 0.088 0.129 0.126 0.057 0.140 0.114 0.088 0.129 0.125 0.045 0.133 0.100 0.090 0.135 0.109 0.077  Annual Slope r2 0.100 0.016 0.124 0.055 0.100 0.111 0.065 0.006 0.110 0.036 0.088 0.087 0.061 0.005 0.099 0.025 0.082 0.070 0.160 0.068 0.158 0.096 0.081 0.086 0.150 0.057 0.169 0.108 0.087 0.095 0.143 0.044 0.160 0.095 0.093 0.106 0.113 0.065  Improved r2 after Data from 1922 to 1997 removing last 12 years Jun & Jul May to Sep Annual Jun & May Annual 2 2 to Sep Slope r Slope r Slope r2 Jul 0.451 0.069 0.154 0.012 0.181 0.011 0.026 0.018 0.005 0.609 0.052 0.363 0.025 0.410 0.028 0.047 0.035 0.027 1.213 0.042 0.885 0.039 0.797 0.032 0.080 0.084 0.079 0.454 0.068 0.092 0.005 0.127 0.005 0.017 0.010 0.001 0.563 0.053 0.282 0.019 0.308 0.018 0.040 0.030 0.018 1.120 0.040 0.762 0.030 0.739 0.028 0.072 0.065 0.059 0.441 0.069 0.059 0.002 0.099 0.004 0.017 0.009 0.001 0.523 0.051 0.223 0.015 0.244 0.013 0.029 0.022 0.012 1.027 0.040 0.668 0.027 0.681 0.026 0.058 0.048 0.044 0.621 0.063 0.368 0.030 0.379 0.029 0.058 0.043 0.039 0.862 0.054 0.540 0.037 0.453 0.027 0.083 0.076 0.069 1.379 0.030 0.916 0.028 0.671 0.018 0.093 0.101 0.068 0.558 0.055 0.308 0.023 0.356 0.026 0.030 0.034 0.031 0.836 0.052 0.537 0.038 0.495 0.033 0.055 0.076 0.075 1.379 0.030 0.935 0.030 0.703 0.020 0.093 0.099 0.075 0.526 0.058 0.244 0.018 0.291 0.021 0.000 0.027 0.023 0.787 0.050 0.483 0.032 0.470 0.030 0.082 0.068 0.065 1.384 0.030 0.980 0.033 0.756 0.024 0.093 0.102 0.082 0.819 0.050 0.489 0.025 0.453 0.022 0.054 0.053 0.043  135  Table 4. 4. Linear regression results between simulation outputs vs. tree ring chronologies for Douglas-fir in the IDF zone. Two data sets were used; one is from 1922 to 1985 while the other one is from 1922 to 1997. r2 improvement was calculated from the difference of the two data sets. Jun & Jul, May to Sep and annual represent the sum of NPP for different periods calculated from the tree productivity-climate model. Bold font in slope column represents statistically significant relationships (p<0.05).  Projected NPP Climate period Scenarios  ESSF  IDF  Average  T1P1 T1P2 T1P3 T2P1 T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 T2P2 T2P3 T3P1 T3P2 T3P3  Data from 1922 to 1985 Jun & Jul Slope r2 0.262 0.185 0.216 0.245 0.113 0.251 0.200 0.110 0.216 0.219 0.117 0.242 0.168 0.074 0.123 0.153 0.123 0.241 0.236 0.254 0.180 0.266 0.093 0.238 0.239 0.243 0.182 0.264 0.094 0.238 0.242 0.225 0.185 0.258 0.094 0.238 0.171 0.219  May to Sep Slope r2 0.185 0.059 0.154 0.097 0.110 0.162 0.135 0.027 0.157 0.079 0.101 0.131 0.105 0.013 0.150 0.060 0.101 0.118 0.195 0.132 0.184 0.173 0.111 0.179 0.191 0.115 0.186 0.176 0.111 0.179 0.196 0.099 0.178 0.157 0.113 0.186 0.148 0.119  Annual Slope r2 0.156 0.035 0.167 0.088 0.124 0.148 0.089 0.011 0.152 0.062 0.112 0.123 0.048 0.003 0.136 0.041 0.112 0.113 0.235 0.130 0.228 0.176 0.117 0.156 0.226 0.115 0.228 0.173 0.124 0.167 0.224 0.095 0.219 0.157 0.131 0.183 0.157 0.110  Improved r2 after removing last 12 years Annual Jun & May Annual to Sep Slope r2 Jul 0.148 0.007 0.039 0.024 0.028 0.415 0.029 0.100 0.040 0.059 0.946 0.047 0.153 0.081 0.101 0.062 0.001 0.012 0.011 0.010 0.307 0.019 0.081 0.033 0.043 0.835 0.037 0.147 0.069 0.086 0.003 0.000 0.006 0.006 0.003 0.226 0.012 0.045 0.024 0.029 0.758 0.034 0.141 0.061 0.079 0.421 0.037 0.101 0.054 0.093 0.602 0.050 0.143 0.081 0.126 1.117 0.054 0.154 0.100 0.102 0.388 0.033 0.100 0.052 0.082 0.593 0.049 0.145 0.085 0.124 1.154 0.058 0.154 0.098 0.109 0.326 0.027 0.078 0.044 0.068 0.541 0.041 0.146 0.082 0.116 1.199 0.063 0.153 0.100 0.120 0.558 0.033 0.105 0.058 0.077  Data from 1922 to 1997 Jun & Jul Slope r2 0.641 0.146 0.993 0.145 1.821 0.098 0.532 0.098 0.890 0.138 1.683 0.095 0.429 0.068 0.742 0.108 1.583 0.100 0.945 0.153 1.279 0.123 2.268 0.084 0.885 0.143 1.237 0.119 2.268 0.084 0.821 0.147 1.154 0.112 2.273 0.085 1.247 0.114  136  May to Sep Slope r2 0.257 0.035 0.535 0.057 1.258 0.081 0.165 0.016 0.425 0.046 1.072 0.062 0.097 0.007 0.340 0.036 0.954 0.057 0.577 0.078 0.829 0.092 1.494 0.079 0.494 0.063 0.818 0.091 1.509 0.081 0.415 0.055 0.728 0.075 1.553 0.086 0.751 0.061  Figure 4. 13. Regressions between tree ring indexes from the chronology of Douglas-fir at IDF zone and potential early growing season NPP (i.e. June and July NPP) calculated with climate from a high elevation site (ESSF). All the regressions improved after removing the data after 1985 (open circles). Values of r2 and slopes are provided in the Table 4.2 upper panel.  137  Figure 4. 14. Regressions between tree ring indexes from the chronology of Douglas-fir at IDF zone and potential early growing season NPP (i.e. June and July NPP) calculated with climate from a low elevation site (IDF). All the regressions improved after removing the data after 1985 (open circles). Values of r2 and slopes are provided in the Table 4.2 lower panel.  138  The results of comparing all chronologies with previous projected NPP, and also with the average of projected NPP from two conjunctive years show most of them do not have significant relationships between tree ring chronologies and model simulation results (not listed here). Only seven pairs of comparison are found having significant relationship (Table 4.5). All of them come from the group which compares previous year projected NPP from June and July with hybrid spruce chronologies from ESSF and MS zones and with low elevation (IDF) climate data inputs. Among these seven pairs, most significant relationships happen under P1 scenario (drought adapted), except the one in the ESSF hybrid spruce chronology group. Unlike Table 4.3 and 4.4 results, take out the last twelve years does not increase model performance ability. Table 4. 5. Linear regression results between previous year simulation outputs vs. tree ring chronologies for hybrid spruce in both ESSF and MS zones. Two data sets were used; one is from 1922 to 1985 while the other one is from 1922 to 1997. r2 change was calculated from the difference of the two data sets. Sum of previous Jun & Jul NPP was calculated from the tree productivity-climate model. Here only presents the one with statistically significant relationships (p<0.05).  Data from 1922 to 1985 T1P1 T2P1 ESSF Spruce vs. IDF climate T3P1 T3P2 T1P1 MS Spruce T2P1 vs. IDF climate T3P1  Data from 1922 to 1997  Slope  r2  Slope  r2  0.129 0.129 0.115 0.126 0.126 0.135 0.124  0.100 0.100 0.086 0.078 0.095 0.110 0.100  0.110 0.119 0.114 0.101 0.112 0.126 0.124  0.086 0.095 0.091 0.066 0.092 0.113 0.114  139  Change in r2 after removing last 12 years 0.014 0.005 -0.005 0.012 0.003 -0.003 -0.014  4.4. Discussion and conclusions My model seems able to capture the seasonal variation of temperature and precipitation influences on NPP. One can see that during the growing season, precipitation is the limiting factor that controls growth, and it limits tree growth most at low elevation (IDF), where conditions are drier than in the high elevation (ESSF). By looking at all chronologies figures, the primary variability comes from precipitation, from the summary tables for the full series of climate data (1922-1997), either P1 (trees adapted to dry conditions) or P3 (adapted to a wet environment) are found to be significantly related to tree ring chronologies. This means that the precipitation range I chose is probably adequate for Douglas-fir in this area. Temperature is the second source of variability, but in the model results it does not make a significant difference among different scenarios. This could be because the three temperature adaptation scenarios which I defined do not vary greatly. With more extreme temperature scenarios the results would show a greater temperature effect. However, such wide temperature adaptation scenarios would not be consistent with our understanding of biological relationships (Salisbury and Ross, 1992; Cai and Dang, 2002). There are no significant relationships between current year simulated NPP and lodgepole pine and hybrid spruce chronologies, probably because 1), the sample size used to create the chronologies was not big enough for lodgepole pine, compared to the number of trees used for Douglas-fir chronologies, and 2), my dendroclimatology results (Chapter Three) and previous work (Nigh et al., 2004; Green and Miyamoto, 2006) showed, spruce is less sensitive to climate than Douglas-fir. Predictions for June and July NPP were most related to tree ring chronology probably because that is the period when early wood is produced, and early wood occupied a large fraction of annual ring width. Also, in May, August, and September, some biomass production is not going to the stems, but rather to new buds and roots (in the beginning of growing season) (Margolis et al., 1995; Kimmins, 2004), or allocated to reserves for next year or cones (at the end of the growing season) (Levitt, 1980; Havranek and Tranquillini, 1995; Kagawa et al., 2006). As for the annual NPP, it includes the production in wintertime, which does not contribute much radial ring growth and therefore provides little new information. As it seems from my results and previous studies, the detailed relationship between NPP and tree ring width is still not clear at this stage, and the few studies reporting on this topic have not found conclusive results, some 140  suggestions and new approaches are undergoing (Rocha et al., 2006; Litton et al., 2007; Girardin et al., 2008). From the results described above, I found that at low elevation (IDF zone), precipitation plays a more important role in determining tree growth than temperature. I also found that Douglas-fir chronology in the IDF zone is more significantly related to predicted NPP than in MS zone. These results agreed with my Chapter Three results, which showed that Douglas-fir growth in the IDF zone is more limited by growing season precipitation than temperature. In addition, previous studies had found that at low elevations, water stress during the growing season could also be a limiting factor for tree growth during dry years (Zhang et al., 2000; Case and Peterson, 2005). The simulation of NPP at these two sites seemed to be an acceptably prediction of what we expect in different forest ecosystems. At high elevation, where temperature is usually the limiting factor affecting tree growth, my model showed that temperature clearly changes the length of growing season. Only after the temperature is high enough for plants to grow does precipitation becomes the major limiting factor. On the other hand, at low elevation in the Okanagan Valley, the climate condition tends to be hot and dry. As long as temperature is adequate for trees to grow, precipitation plays an important role in controlling tree growth through soil moisture deficit. If the tree is adapted to a wetter environment, it will be more strongly affected by water stress during the growing season. The magnitude of tree ring chronology fluctuations is smaller than my simulated NPP fluctuations, which may imply that tree growth response to climate is not big or there are other factors should be included, which is consistent with Chapter Three results that climate variables only be able to explain at most 50% of the variability from tree ring width. This is a simple model with only three variables, but it shows significant results when comparing with Douglas-fir chronologies even though the model only explained a small part of the variability of Douglas-fir chronologies. As Scheffer (1998) suggests, simple models are best used for studying the properties of individual mechanisms, like the relationship between NPP and tree ring width. Pace (2003) also said that “…many of the more interesting questions in ecosystem science are about key processes that determine the similarities and differences within and among systems”. Therefore, I consider that this model is a good first approach in this specific area to cover the objective of this research. However, climate change will affect many  141  complex factors in the forest ecosystem and make impacts through complex interactions at the ecosystem level (Figure 4.1). Therefore, this simple model is not adequate to represent the key ecosystem processes, which is probably why it explained such a small percentage of the variation in ring width. Simple models can only answer simple questions. When dealing with more complex issues, like the ones related to ecosystem response to climate change, it is expected that more complex models will be needed for forest management (Kimmins et al., 2008). Therefore, the reader should keep in mind that the modeling described here is just a small part of the more complex ecosystem-level model FORECAST Climate. All things considered, it seems that tree ring width is affected by more factors than just temperature and precipitation. Although my tree productivity-climate model only captured a small proportion of the variability in tree ring width, it is significant and this is an indication that my model used a valid approach to capture part of the influence of climate change on tree growth. However, the model needs to be improved to increase its ability to explain the variability of tree growth. I expect that by linking this simple model to the rest of FORECAST Climate, which will include simulation of the effect of climate change on litterfall decomposition, nutrient cycling and inter- and intra- species competition for light, nutrient and moisture, the final prediction of tree growth will be more accurate. However, this has to be tested in future work.  142  4.5. References Bi J., J.A. Blanco, J.P.Kimmins, Y. Ding, B. Seely and C. Welham. 2007. Yield decline in Chinese Fir plantations: A simulation investigation with implications for model complexity. Canadian Journal of Forest Research. 37: 1615-1630. Blanco, J.A., B. Seely, C. Welham, J.P. Kimmins, and T.M. Seebacher. 2007. Testing the performance of FORECAST, a forest ecosystem model, against 29 years of field data in a Pseudotsuga menziesii plantation. Canadian Journal of Forest Research. 37: 1808-1820. Botkin, D.B., J.G. Janak and J.R. Wallis. 1972. Some ecological consequences of a computer model of forest growth. Journal of Ecology. 60: 849-872. Botkin, D. B. 1993. Forest Dynamics: An Ecological Model. Oxford University Press, New York. Pp.309. Cai, T and Q-L. Dang. 2002. Effects of soil temperature on parameters of a coupled photosynthesis–stomatal conductance model. Tree Physiology. 22: 819-827. Case, M.J. and D.L. Peterson. 2005. Fine-scale variability in growth-climate relationship of Douglas-fir, North Cascade Range, Washington. Canadian Journal of Forest Research. 35: 2743-2755. Costanza, R. and S. Gottlieb. 1998. Modelling ecological and economic systems with STELLA: Part II. Ecological Modelling. 112: 81-84. Dang, Q.L., V.J. Lieffers and R.L. Rothwell. 1992. Effects of summer frosts and subsequent shade on foliage gas exchange in peatland tamarack and black spruce. Canadian Journal of Forest Research. 22: 973-979. Daubenmire, R.F. 1974. Plants and Environment. A Textbook of Plant Autecology. 3rd Edition. John Wiley & Sons, New York. Pp.422. Girardin, M.P., F. Raulierb, P.Y. Berniera and J.C. Tardif. 2008. Response of tree growth to a changing climate in boreal central Canada: A comparison of empirical, process-based, and hybrid modelling approaches. Ecological Modelling. 213:209-228. Green S. and Y. Miyamoto. 2006. Characterizing the growth responses of three co-occurring northern conifer tree species to climate variation across a range of conditions. A summary report for a study funded by the B.C. Forest Science Program FSP project #Y061107. Prince George, BC. Pp.6. Hamann, A. and T.L. Wang. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology. 128: 211-221. Havranek, W.M. and W. Tranquillini. 1995. Physiological Processes during Winter Dormancy and Their Ecological Significance. p.95-124. in Smith, W.K. and T.M. Hinckley. 1995. Ecophysiology of Coniferous Forests. Academic Press, New York. Pp.338. High Performance Systems Inc., 2002. STELLA Research software, v 7.0. Hanover, New Hampshire, USA. Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007:The Physical Science Basis. Summary for Policymakers. Available at: http://hosted.ap.org/specials/interactives/_documents/climate_report.pdf. Retrieved January 14th, 2009. Kagawa, A., A. Sugimoto and T.C. Maximov. 2006. 13CO2 pulse-labelling of photoassimilates reveals carbon allocation within and between tree rings. Plant, Cell and Environment. 29:1571-1584.  143  Kimmins, J.P., K.A. Scoullar, R.E. Bigley, W. Kurz, P.G. Comeau and L. Chatarpaul. 1986. Yield prediction models: the need for a hybrid ecosystem-level approach incorporating canopy function and architecture. In: Fujimory, T., Whitehead, D. (Eds.), Crown and Canopy Structure in Relation to Productivity, FFPRI, Ibaraki, Japan, p. 26-48. Kimmins, J.P., 1988. Community organization: methods of study and prediction of the productivity and yield of forest ecosystems. Canadian Journal of Botany. 66: 2654-2672. Kimmins, J.P., D. Mailly and B. Seely. 1999. Modelling forest ecosystem net primary production: the hybrid simulation approach used in FORECAST. Ecological Modelling. 122: 195-224. Kimmins, J.P., 2004. Forest Ecology. A Foundation for Sustainable Management and Environmental Ethics in Forestry, 3rd Edition. Prentice Hall, New Jersey. Pp.611. Kimmins, J.P., J.A. Blanco, B. Seely, C. Welham and K. Scoullar. 2008. Complexity in modelling forest ecosystems: How much is enough? Forest Ecology and Management. 256: 1646-1658. Korol, R.L., S.W. Running, and K.S. Milner. 1995. Incorporating intertree competition into an ecosystem model. Canadian Journal of Forest Research. 25:413-424. Levitt, J. 1980. Responses of Plants to Environmental Stresses. Volume 1. Chilling, Freesing, and High Temperature Stresses. 2nd Edition. Academic Press, New York. Pp.497. Litton, C.M., J.W. Raich and M.G. Ryan. 2007. Carbon allocation in forest ecosystems. Global Change Biology. 13: 2089-2109. Loehel, C. and D. LeBlanc. 1996. Model-based assessments of climate change effects on forests: a critical review. Ecological Modelling. 90:1-31. Margolis, H., R. Oren, D. Whitehead and M.R. Kaufmann. 1995. Leaf area dynamics of conifer forestsForests. p.181-224 in Smith, W.K. and T.M. Hinckley. 1995. Ecophysiology of Coniferous Forests. Academic Press, New York. Pp.338. McMurtrie, R. E., D. A. Rook and F. M. Kelliher. 1990. Modelling the yield of Pinus radiata on a site limited by water and nitrogen. Forest Ecology and Management. 30: 381-413. McMurtrie, R. E. and J. J. Landsberg. 1992. Using a simulation model to evaluate the effects of water and nutrients on the growth and carbon partitioning of Pinus radiata. Forest Ecology and Management. 52: 243-260. Morris, D.M., J.P. Kimmins and D.R. Duckert. 1997. The use of soil organic matter as a criterion of the sustainability of forest management alternatives: a modeling approach using FORECAST. Forest Ecology and Management. 94: 61–78. Nigh, G.D., C.C. Ying and H. Qian. 2004. Climate and Productivity of Major Conifer Species in the Interior of British Columbia, Canada. Forest Science. 50: 659-671. Pace, M.L. 2003. The Utility of Simple Models in Ecosystem Science. p.49-62. in Canham, C.D., J.J. Cole and W.K. Lauenroth. Models in Ecosystem Science. Princeton University Press, New Jersey. Pp.476. Parton, W.J., J.M.O. Scurlock, D.S. Ojima, T.G. Gilmanov, R.J. Scholes, D.S. Schimel, T. Kirchner, J-C. Menaut, T. Seastedt, E.G.Moya, A.Kamnalrut and J. I. Kinyamario. 1993. Observations and modeling of biomass and soil organic matter dynamics for the grassland biome worldwide. Global Biogeochem. Cycles.7: 785-809. Pastor, J. and Post, W. M. 1985. Development of a Linked Forest Productivity-Soil Process Model. U.S. Dept. of Energy, ORNL/TM-9519. Pp.108  144  Pearson, R.G. and T.P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: are climate envelope models useful? Global Ecology and Biogeography. 12: 361-71. Rocha, A.V., M.L. Goulden, A.L. Dunn and S.C. Wofsy. 2006. On linking interannual tree ring variability with observations of whole-forest CO2 flux. Global Change Biology.12: 1378-389. Running, S.W. and J.C. Coughlan. 1988. A General model of forest ecosystem process for regional applications. I. Hydrologic balance, canopy gas exchange and primary production processes. Ecological Modelling. 42: 125-154. Salisbury, F.B. and C.W. Ross. 1992. Plant Physiology. 4th Edition. Wadsworth Publishing Company, Belmont, California. Pp.82. Scheffer, M. 1998. Ecology of Shallow Lakes. Chapman and Hall. New York. Seebacher T.M. 2007. Western redcedar dieback: possible links to climate change and implications for forest management on Vancouver island, B.C. MSc dissertation. University of British Columbia, Vancouver, B.C. Pp.136. Seely, B., P. Arp and J.P. Kimmins. 1997. A forest hydrology submodel for simulating the effect of management and climate change on stand water stress. Empirical and process-based models for forest tree and stand growth simulation. Oeiras, Portugal. Seely, B., C. Welham. 2008. Factorial analysis of soil cover, competing vegetation, aspect, and climate effects on tree water stress in Oil Sands reclamation using the ForWaDy model. Unpublished report. Seely, B., C. Welham and H. Kimmins. 2002. Carbon sequestration in a boreal forest ecosystem: results from the ecosystem simulation model, FORECAST. Forest Ecology and Management. 169: 123-135. Seely B., C. Hawkins, J.A. Blanco, C. Welham and J.P. Kimmins. 2008. Evaluation of an ecosystem-based approach to mixedwood modelling. Forest Chronicle. 84: 181-193. Shugart, H.H. 1984. A Theory of Forest Dynamics. Springer-Verlag, New York. Wang, J.R., P. Comeau and J.P. Kimmins. 1995. Simulation of mixedwood management of aspen and white spruce in Northeastern British Columbia. Water, Air and Soil Pollution. 82, 171-178. Wei, X. and J.P. Kimmins. 1995. Simulations of the long-term impacts of alder-Douglas-fir mixture management on the sustainability of site productivity using the FORECAST ecosystem model. In: Comeau, P., Thomas, K.D. (Eds.), Silviculture of Temperate and Boreal Broad-Leaved-Conifer Mixtures. Research Branch, Ministry of Forests, Victoria, BC, Canada. Wei, X., W. Liu, M. Waterhouse and M. Armleder. 2000. Simulation on impacts of different management strategies on long-term site productivity in lodgepole pine forests of the central interior of British Columbia. Forest Ecology and Management. 133: 217-229. Wei, X., J.P. Kimmins and G. Zhou. 2003. Disturbances and the sustainability of long-term site productivity in lodgepole pine forests in the central interior of British Columbia—an ecosystem modeling approach. Ecological Modelling. 164:239-256. Welham, C., B. Seely and J.P. Kimmins. 2002. The utility of the two-pass harvesting system: an analysis using the ecosystem simulation model FORECAST. Canadian Journal of Forest Research. 32: 1071-1079. Zhang Q.-B., R.J. Hebda, Q.-J. Zhang and R.I. Alfaro. 2000. Modeling tree-ring growth responses to climatic variables using artificial neural networks. Forest Science 46: 229-239.  145  5. Concluding remarks 5.1. Main findings of this work In order to forecast and organize future forest management plans, reliable decision-support tools are needed. The current trends of climate change will likely alter the amounts and patterns of precipitation and average and seasonal temperatures (IPCC, 2007), which it is thought will translate into changes in tree growth, bringing uncertainty to forest management and planning. This will be especially important in British Columbia, where predicted increases of more than 2°C over the next 100 years in cool temperate forest could greatly change growth patterns of many valuable tree species. For this reason, accurate understanding/estimation of the relationships between tree growth/productivity and climate is now, and will be in the near future, one of highest priorities in forest research (BC Ministry of Forests, 2004). If adequately quantified, these relationships could be incorporated into existing forest management models, providing these decision-support tools with the capability to simulate possible climate change consequences. In this project there were three objectives: 1) to evaluate the performance of the MTCLIM model by testing three pairs of weather stations in the Okanagan Valley in the Southern Interior of BC, 2) to explore the relationships between climate and tree breast-height growth obtained from tree ring width measurements for three tree species, Pseudotsuga menziesii, Pinus contorta and Picea glauca x engelmanni (Douglas-fir, lodgepole pine and hybrid spruce, respectively), in the Tree Farm License 49 (TFL 49) in the Okanagan Valley along an elevational gradient, and 3) to build a simple net primary production submodel and use the tree ring data to examine the performance of the submodel. In future work (not discussed here), the submodel will subsequently be incorporated in the ecosystem-level forest management model FORECAST to calibrate climate limitations to tree radial growth. The pairs of weather stations used to test the climate model MTCLIM were Hedley and Hedley Mine, Vernon and Silver Star, and McCulloch and Big White. The lengths of the simulation paired data sets were 59 years, 27 years and 21 years, respectively. From the statistical tests, MTCLIM performed well at a monthly time scale when simulating temperature, but it had some systematic bias at an annual time scale. In wintertime (i.e. October to February), the model’s predicted values were lower than the observed values for both  146  maximum and minimum temperature. For the rest of the year (i.e. March to September), the model’s simulated values were higher than the observed values for both maximum and minimum temperature. However, the difference in the minimum temperature is not as big as for maximum temperature part. This is a common result in climate simulations. For example, Hunter and Meentemeyer (2005) used PRISM in California and reported a seasonal bias in their simulation. They concluded that it was because the model performed differently in each month based on the number/amount of the original inputs of temperature and precipitation. Hamann and Wang (2005) summarized current models and how they typically over- or under-estimate temperature-related (e.g., maximum, minimum and average air temperature) variables by 0.5–1°C, with regional deviations as high as 2–4°C where weather station coverage is poor. The reason is likely that the model does not adequately reflect the complexity of reality due to more factors affecting temperature and precipitation than just elevation and topography, such as the influence of large water bodies, complex terrain, cold air drainage, and rain shadow effects. Therefore, when we use this model’s output to explore climate/tree growth response and also as input to an ecosystem model in such conditions, it may result in a misinterpretation of the climate/tree growth relationships or bias the results from an ecosystem model. MTCLIM did not perform as well for the simulation of precipitation as for temperature, but the error was still within an acceptable range and similar to that of other studies. For example, deviations of predicted mean annual precipitation were typically about 100 mm for the PRISM model and 400 mm for the Rehfeldt model (Hamann, and Wang, 2005); the monthly deviation in my study was -11.25 to 9.26 mm. These are about one to two percent biases compared with annual total precipitation. Several papers have reported that MTCLIM does not perform well in dry or tropical and subtropical areas (Kimball et al., 1997; Almeida and Landsberg, 2003), but in the present study, it seems able to perform well compared to the results of the previous authors. However, caution must be exercised when using the precipitation results for this dendrochronological research; especially since the growth response of the three species in this study were limited by water supply. In MTCLIM, precipitation isohyets and lapse rate are very important and are the main driving factors affecting model prediction (Creed et al., 1996; Chiesi et al., 2002). In general, the lapse rate is a very stable value, which does not vary much regionally (Trewartha, 1954; Lowry, 1969; Dodson and Marks, 1997; Barry and Chorley, 1998). However, the precipitation isohyet varies more because of large water bodies, complex terrain,  147  rain shadow and upward and downward air mass movements, and a more accurate estimation depends on station density (Custer et al., 1996). In addition to the accuracy of the lapse rate and precipitation isohyet, I also found that as the sample size (number of simulated years in the base data set) increases, the simulation accuracy increased as well. This is quite common for climate-generating models (McKeeny et al., 2006), because the more information we have to run the simulation, the better the model can adjust and make better predictions. Because MTCLIM’s simulation ability appeared to be within an acceptable range for my study, I concluded that it was satisfactory for generation of weather data for my tree ring research experimental sites. In the tree ring research, tree cores were sampled in 2004 in two separate transects running across three different ecological zones: ESSF, MS and IDF. These cores were used to date tree-rings by year of growth and used to create master chronologies of variation of tree-ring width for about 100 years (1900 to 2003). Residual chronologies were generated by standardizing and detrending master chronologies to remove non-climate-related influences on growth. These residual chronologies were then correlated to monthly mean temperature and monthly precipitation from previous April to current October obtained from the weather station sited in Westwold (interior BC). Climate records were adjusted to the specific BEC zones using the climate model MTCLIM, using the data from Westwold. In general, the climate signal for all the species and sites was moderate to weak (i.e. r2 = 0.12 to 0.48 in Table 3.7). However, it is common that climatic variables rarely capture more than 60% of the tree-ring variability, and 40 to 50% is more common (Hughes, 2002). This is an indication that although important for tree growth, climate influence could be significantly modified by other site-specific factors such as seasonal nutrient availability, microtopographic position and individual tree history. Therefore, a more complex growth model than solely considering temperature and precipitation factors is needed to forecast future tree growth and species distributions. Some landscape-level, climate-vegetation simulation or regional models may thus overestimate the potential impacts of climate change because most of them only consider climate effects on species distribution and abundance. Results of correlation and regression analysis also showed that relationships between chronologies and climate are site and species-specific, something that has also been shown for the same species that I studied by other researches (Zhang and Hebda, 2004; Green and Miyamoto, 2006; Savva et al., 2006; Pichler and Oberhuber, 2007; Su et al., 2007). However, my research is one of the few studies done with this  148  specific combination of BEC zones and species in interior BC. In spite of this site specificity, it is remarkable that the results follow the general pattern reported from similar researches done in the Pacific Northwest area (see Table 3.9 and 3.10 for references). The best master chronologies were produced for Douglas-fir. This species was also the most sensitive to climate and it had the strongest climate signal. Therefore, if we want to forecast future Douglas-fir growth and distribution, and the effects of management thereon, any forest management model used should have climate components. Douglas-fir in the MS and IDF zones showed similar patterns for both temperature and precipitation variables, and they showed that growth is limited by water stress during the growing season. Previous studies have also found that, at low elevation, water stress during the growing season could be a limiting factor for tree growth during dry years (Zhang et al., 2000; Case and Peterson, 2005). From my study, it seems that Douglas-fir was more sensitive to temperature than precipitation in both MS and IDF zones: the longer the growing period, the better the growth (i.e. positive correlated with previous April, May). Summer drought was more important for the IDF zone than the MS zone. This has direct consequences for the accuracy of simulating the climate change effects of growth in the IDF zone, and as the model development chapter showed, I found that Douglas-fir chronology in the IDF zone was more significantly related to predicted NPP than in the MS zone. This supports the finding that Douglas-fir growth in the IDF zone is more limited by growing season precipitation than in the MS zone. Results for lodgepole pine were more difficult to interpret and less reliable due to the relatively small sample size used to create the chronologies, caused by a series of operational difficulties explained in the main text. However, I found that, irrespective of which zone it is growing in, lodgepole pine prefers higher temperatures during the growing season and a longer growth period if the water supply is sufficient. In addition, the same climatic factor can have different effects depending on the BEC zone. For example, winter snowfall in the IDF zone is good for water supply in summer, whereas snowfall in the MS zone could cause the damage to branches or buds and reduce the growth in the following year. Hybrid spruce was the species with the least sensitive response to the climate signal. If we want to simulate the growth response to climate change of this species and manage it, a climate component in the model seems not to be very important, contrasting with the bigger climatic influence observed for Douglas-fir. However, despite its low sensitivity, I still found some  149  indications that hybrid spruce has a similar response to climate in ESSF and MS, and it is more sensitive to precipitation than to temperature. In both BEC zones where it grows, the results showed that for spruce, water stress during the growing season is the most important limiting factor related to climate: the chronologies in both zones showed a positive correlation with precipitation in the previous growing season (i.e. previous April to September), but a negative correlation with monthly mean temperature for that same period. My results suggest that spruce is more sensitive to soil moisture than temperature, as was recently observed by Savva et al. (2006), but in contrast to most of the current relate literature, which states that spruce is more sensitive to temperature than to soil moisture (MacDonald et al., 1998), In order to simulate spruce response to climate change, these relationships need to be examined in the area of interest. After establishing the weak relationships between tree ring growth and climate, I then used the chronologies as a validation tool for my model. I used different temperature/growth and precipitation/growth optimum scenarios at two different elevation sites to explore how NPP responds to different climate conditions, and how similar the pattern of yearly values of NPP were to the chronologies. From this I drew three conclusions: 1. The relationship between ring width and temperature has changed markedly over the last 40 years (Wilmking et al., 2004; Rocha et al., 2006). Using different climate/growth scenarios helps to explore the growth performance of different species under climate change scenarios. 2. A very simple simulation of NPP at the two sites proved to be a weak but significant predictor of what we could expect for Douglas-fir forest ecosystems. When simulating potential NPP for high elevation sites, where temperature is usually the limiting factor affecting tree growth, my model was able to reproduce the pattern of temperature change and how it affects the length of growing season. My model also showed that once the temperature is high enough for plants to grow, precipitation becomes the major limiting factor. On the other hand, at low elevation sites, climate tends to be hot and dry. As long as the temperature regime is adequate for tree growth, precipitation plays an important role in controlling tree growth. Furthermore, if the simulated tree species is adapted to wetter environments, water stress during the growing season will be a major limiting factor. 3. The magnitude of tree ring chronology fluctuations was smaller than that of the simulated NPP fluctuations, which may imply that a simple direct tree ring growth response to climate is  150  not strong, or that other factors are modulating the influence of climate. This is consistent with other tree ring analysis results because tree rings are likely to indicate deficits (for example, of water or of degree-days) more faithfully than surpluses, since climate control of tree-ring growth works through the most limiting factor (Hughes, 2002). And again, in tree-ring research, it is common that climate accounts for only 40 to 50% of the tree-ring variance. There is other 50 to 60% of variance that is contributed from non-climate factors. So when we use models that only consider climate variables to simulate the impacts of climate change, we have to be careful interpreting their simulation results, because they may exaggerate the consequences of climate change. The mathematical equations of the relationships of my three species’ radial growth to climate used in my tree productivity-climate model were developed in this thesis to predict tree radial growth from predicted climate. Although the variance of tree growth explained by climate was low to moderate at best, these equations could be included in a more complex ecosystem model, such as FORECAST-Climate, that already accounts for other sources of variability (i.e. light and nutrient availability, inter-species competition) as a way to improve the predictions of future tree growth under changing climate conditions. Currently, there is no decomposition or other soil processes considered in my tree productivity-climate model, but these components are important in estimating net ecosystem production or determining the site carbon budget (Creed et al., 1996; Cienciala and Tatarinov, 2006). In their research, they showed that it is important to consider soil, hydrology and climate together to make reliable predictions when exploring forest ecosystem production/climate issues. On the other hand, according to Vanclay and Skovsgaard (1997), a way to validate how close a complex model’s results are to reality is to test each small component of that model separately. I followed this recommendation, using my STELLA model to test part of the processes simulated in FORECAST-Climate and the results are promising. They provide confidence that part of FORECAST-Climate is realistic, but obviously more tests with the other processes have to be done. Some ecosystem models link different submodels to make the simulation (e.g. TRIPLEX, Peng et al., 2002). Our approach to link FORECAST with ForWaDy and other growth components to improve the original prediction ability follows this same model development philosophy.  151  5.2. Strengths and weaknesses of this research There are both strengths and weaknesses in the methodologies I used in this research. The weaknesses are:  1. In the climate-generating model section, the performance of MTCLIM depends strongly on previous knowledge of the local conditions (i.e., lapse rate and precipitation) and the length of available input climate records. However, data sets are often of limited duration, and this is becoming a common issue in this kind of modeling exercise. 2. In my tree ring research section, most of the time the tree-ring chronologies only responded to limited specific seasonal ‘windows’; they did not respond directly to a single monthly or even seasonal climate variable (Hughes, 2002), which makes the modeling of tree growth-climate relationships difficult. 3. Tree ring research is based on the assumptions that a) the factors that controlled the formation of tree rings in the past will continue to act in the same way in the future; and b), the techniques used to remove non-climatic variability in ring width, such as that caused by tree age/size trend and interactions with neighbours, will leave the climate signal intact. However, these non-climatic variables also limit the faithful representation of climate variations on centennial and longer time scales in many cases (Hughes, 2002), and several studies have found that the relationship between ring width and temperature has changed markedly over the last 40 years in some boreal forests (Wilmking et al., 2004; Rocha et al., 2006); it can be expected that these changes will continue under climate change. 4. Global warming in the late 20th century caused trees to grow faster (Wilmking et al., 2004); traditional detrending methods will decrease sensitivity during statistical analysis (Melvin and Briffa, 2008). However, most of my approaches in this research followed traditional techniques; I may have lost some information about the last years during the detrending process. 5. Wilmking et al. (2004) and Seebacher (2007) found that within a tree population, individual trees performed differently. Therefore, the traditional master chronology approach may not represent the real response of the whole population to past climate variation at my experimental sites. 6. In Chapter Four, the temperature/growth and precipitation/growth optimum scenarios are very simple and are based on hypothesized species; they may not be representative of real 152  species adaptations.  However, there are also several strengths in my research:  1. MTCLIM simulation results showed that not only the mean value but also the distribution matched the observed climate variables, giving me confidence in its performance. 2. This is the first time MTCLIM has been used for Okanagan region of BC, as far as I know. Its satisfactory performance allows us to use this model in this region and maybe in other regions where it has never been used before. As a result, it may help to solve the problem for many ecosystem models’ climate input requirements. 3. I used several statistical approaches to validate MTCLIM performance. Some of them have rarely been used before in this area but all of them were in agreement that MTCLIM performed acceptably. It is a good demonstration of how we can validate model performance not only for climate models but also for other kinds of models. 4. The tree ring signal usually has large geographic-scale patterns of common year-to-year variability. My results were consistent with other studies and have similar general patterns to others within the Pacific Northwest region (references are listed in Tables 3.9 and 3.10). 5. Some of the linear models I tried with just four or five variables explained almost 50% of the variability of tree-ring width, which can be considered very effective in this kind of research (Hughes, 2002). 6. My results showed that species respond differently at different elevations, which is also consistent with several tree ring studies at different altitudes (Zhang and Hebda, 2004; Case and Peterson, 2005, 2007; Savva et al., 2006). I have provided two summarized tables of these researches in Pacific Northwest region (Tables 3.9 and 3.10). 7. The simulation results suggested that even with changing climate, the variation of growth has not been as great as is commonly thought. This reminds us to interpret the results of climate change impact models with caution. 8. For Douglas-fir, my tree productivity-climate model agreed with the dendroclimatology results, even though it explained only a small part of the tree ring variability. 9. Considering the above, the validation of my simple tree productivity-climate model helps us to understand how a complex model (i.e. FORECAST Climate) can represent individual  153  ecophysiological processes. 10. And last but not least, this research is one of the few studies that compare tree ring results with ecosystem productivity, a new and promising research area (Rocha et al., 2006).  5.3. Conclusions 1. If sample size and input parameters are adequate, especially the lapse rate and precipitation isohyet, MTCLIM can give quite reliable weather data for remote locations in mountainous terrain. 2. All three species (Douglas-fir, lodgepole pine and hybrid spruce) had significant correlations with previous year climate variables. Previous growing season water stress affects current year growth, but the degree and specific months vary between species and zones. 3. Current growing season temperature and water stress also affect tree growth, but the magnitude of the effect is not as great as associated with the previous year’s climate. 4. In addition to growing season climates, winter precipitation also plays an important role in determining tree growth in the following year. Depending on species and the variables considered, it either acts as a water resource for the following year (e.g. lodgepole pine in ESSF and MS zone in the PCA analysis) or as a snow damage factor (lodgepole pine in the ESSF and MS zones in correlation and multiple regression. There may also be a negative soil temperature consideration, but this was not examined in my study). 5. Results showed that tree ring growth is affected by more factors than just temperature and precipitation, and that simulation of tree growth under climate change needs to include more variables than I was able to consider. A model like FORECAST Climate is needed. 6. Although my tree productivity-climate model only captured a small part of the variability in tree ring growth, the regression relationships were significant between simulated and observed, supporting the assertion that this is a valid approach to model the influence of climate change on tree growth. The linkage of this tree productivity-climate model to FORECAST Climate will expand the former to include additional key ecosystem processes. 7. My research indicates that the bio-envelope approach to modeling the impacts on species distributions may overestimate the effects of global warming. However, this does not reduce the value of such models as a means of alerting the public to the risks to forests of climate change.  154  5.4. Future research My research has shown several opportunities for future research. 1. In the climate modelling part, my results showed that MTCLIM can give us quite reliable weather data to use in mountainous terrain. Therefore, when the researchers need climate data from the site where has no weather station, they can consider to use this model to generate the climate data needed. However, because MTCLIM’s performance is strongly depended on the length of simulation years and adequate lapse rate and precipitation isohyet, this information has to be prepared before using it. Also, sensitivity analysis needs to be done to know how predictions change based on the specified error and bias (Hamann and Wang, 2005) Future research should also examine the effect on model output of combining data from several weather stations rather than just single station. 2. It should be considered the possibility of modifying MTCLIM to account for the influence of large water bodies. MTCLIM lacks the ability of capturing the influences of lakes (e.g. the Okanagan Lake), complex terrain, and atmospheric inversions in determining temperature and moisture, including cold air drainage and rain shadow effects (Glassy and Running, 1994; Hunter and Meentemeyer, 2005), and therefore we should keep these limitations in mind and interpolate model output with caution. Using the data sets in my study area, the effects of such factors on the model could be investigated, maybe by modifying the lapse rates depending on the distance to a water body of a given size. 3. My final selection of sampling sites represented a compromise between several different considerations, and this may have resulted in a weaker response to climate than sampling according to traditional methods. Also as Kienest et al. (2007) pointed out, PCA works particularly well if a larger network of tree ring sites is used. Future research could usefully expand the data base by sampling additional transects over a broader geographical area and expanding the sampling to drier sites on which the trees would be expected to be more sensitive to variations in precipitation. 4. X-ray densitometer analysis works better on larger diameter cores than I used. Future work with larger cores could be done to test the error that I may have introduced by using the smaller cores. Also, further work should examine other densitometer outputs. It would be valuable to examine such as early wood/late wood width and density and their ratios. 5. In this thesis I employed standard tree ring research methodologies. However, Melvin and 155  Briffa (2008) found that the late 20th century global warming caused trees to grow faster, with the consequence that the traditional detrending methods decrease the sensitivity of statistical analysis. They proposed a signal-free detrending method, which can improve the performance of statistical analysis. This is new methodology should be investigated to see if the new residual chronologies will increase the sensitivity to climate. 6. Because tree response to climate varies from tree to tree within a population (Welmking et al., 2004), the future study should consider to analyze individual trees rather than only master chronologies (Seebacher, 2007) when trying to explore the relationship between tree rings and climate. 7. Several studies have shown that the relationship between ring width and the physical environment is not always straightforward. Also ring width is nearly insensitive to climatic variability in north-eastern North America (Rocha et al., 2006). However, the tree productivity-climate model in this study has shown weak but significant ability to predict tree growth using climate variables. Also, increasing the complexity of the model could improve its performance (Kimmins et al., 2008). In the same time, other papers (FORMAT, 2009) suggested that the annual mass approach (i.e. annual mass = ring width × ring density) is more related to NPP, it will also be a possible approach for future work. 8. Non-climate factors such as the soil chemistry and biology can be important in modulating the effect of climate change on tree growth. I expect that by linking my simple tree productivity-climate model to the rest of FORECAST Climate, which will include simulating the effect of climate change on litter decomposition, nutrient cycling and inter- and intra- species competition for light, nutrient and moisture, the final prediction of tree growth will be more accurate. This should, and will be, tested in future work. 9. The details of carbon allocation changes under different temperature and soil moisture changes are still not clear, but I also expect that knowledge of this ecosystem dynamic will help ecosystem models make more accurate predictions about impacts of climate change. These relationships should be studied in more detail. 10. According to Cannell (1995) and Canadell and Raupach (2008), we can preserve more carbon in the forest under climate change by proper management. However, carbon-based management strategies will be different from those of the past because climate change will cause changes in the rate of growth (carbon storage), decomposition and respiration (carbon loss). We  156  cannot simply use the traditional method, but must consider the consequences of these new factors. Simulations of alternative forest management approaches should be therefore carried out using climate-equipped, process-based, ecosystem management models, and the most plausible results tested in the field.  157  5.5. References Almeida A.C. and J.J. Landsberg. 2003. Evaluating methods of estimating global radiation and vapor pressure deficit using a dense network of automatic weather stations in coastal Brazil. Agricultural and Forest Meteorology. 118: 237-250. BC Ministry of Forests. 2004. Forest Investment Account. Forest Science Program. Strategic plan 2004 - 2008. Victoria, BC. Pp.16. Barry, R.G., and R.J. Chorley. 1998. Atmosphere, Weather and climate 7th Ed. Routledge. New York. NY. Pp.409. Canadell, J.G. and M.R. Raupach. 2008. Managing forest for climate change mitigation. Science. 320: 1456-1457. Cannell, M. 1995. Forests and the Global Carbon Cycle in the Past, Present and Future. European Forest Institute Research Report 2. Joensuu, Finland. Pp.66. Case, M.J. and D.L. Peterson. 2005. Fine-scale variability in growth-climate relationship of Douglas-fir, North Cascade Range, Washington. Canadian Journal of Forest Research. 35: 2743-2755. Case, M.J. and D.L. Peterson. 2007. Growth-climate relations of lodgepole pine in the North Cascades National Park, Washington. Northwest Science. 81: 62-75. Chiesi, M., F. Maselli, M. Bindi, L. Fibbi, L. Bonora, A. Raschi, R. Tognetti, J. Cermak and N. Nadezhdina. 2002. Calibration and application of FOREST-BGC in a Mediterranean area by the use of conventional and remote sensing data. Ecological Modelling. 154: 251-262. Cienciala, E. and F.A. Tatarinov. 2006. Application of BIOME-BGC model to managed forests 2. Comparison with long-term observations of stand production for major tree species. Forest Ecology and Management. 237: 252-266. Creed, I.F., L.E. Band, N.W. Foster, I.K. Morrison, J.A. Nicolson, R.S. Semkin and D. S. Jeffries. 1996. Regulation of nitrate-N release from temperate forests: A test of the N flushing hypothesis. Water Resources Research. 32: 3337-3354. Custer, S.G., P. Farnes, J.P. Wilson and R.D. Snyder. 1996. A Comparison of hand and spline-drawn precipitation maps for mountainous Montana. Water Resources Bulletin. 32: 393-405. Dodson, R. and D. Marks. 1997. Daily air temperature interpolated at high spatial resolution over a large mountainous region. Climate Research. 8: 1-20. FORMAT - Forest Modelling Assessment and tree-rings. 2009. Sensivity of tree-growth to climate change and Growth modelling from past to future. Available at: http://medias.obs-mip.fr/format/index.html Retrieved January 12, 2009. Glassy, J.M. and S.W. Running. 1994. Validating diurnal climatology logic of the MT-CLIM model across a climatic gradient in Oregon. Ecological Applications. 4: 248-257. Green S. and Y. Miyamoto. 2006. Characterizing the growth responses of three co-occurring northern conifer tree species to climate variation across a range of conditions. A summary report for a study funded by the B.C. Forest Science Program FSP project #Y061107. Prince George, BC. Pp.6. Hamann, A. and T.L. Wang. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology. 128: 211-221. Hughes, M.K. 2002. Dendrochronology in climatology - the state of the art. Dendrochronologia. 20: 95-116.  158  Hunter, R.D. and R.K. Meentemeyer. 2005. Climatologically aided mapping of daily precipitation and temperature. Journal of Applied Meteorology. 44: 1501-1510. Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007: The Physical Science Basis. Summary for Policymakers. Available at: http://hosted.ap.org/specials/interactives/_documents/climate_report.pdf Retrieved Feb. 1st, 2009. Kienast F., O.Wildi and S. Ghosh. 2007. On Selected Issues and Challenges in Dendroclimatology. A Changing World. Challenges for Landscape Research. p.113-132. Kimball, J.S., S.W. Running and R. Nemani. 1997. An improved method for estimating surface humidity from daily minimum temperature. Agricultural and Forest Meteorology. 85: 87-98. Kimmins, J.P., J.A. Blanco, B. Seely, C. Welham and K. Scoullar. 2008. Complexity in modelling forest ecosystems: How much is enough? Forest Ecology and Management. 256: 1646-1658. Lowry, W.P. 1969. Weather and Life: An Introduction to Biometeorology. Academic Press, New York. Pp.305. MacDonald, G. M., J. M. Szeicz, J. Claricoates and K. A. Dale. 1998. Response of central Canadian treeline to recent climatic changes. Annals of the association of American Geographers. 88: 183-208. McKeeny, D.W., J.H. Pedlar, P. Papadopol, and M.F. Hutchinson. 2006. The development of 1901-2000 historical monthly climate models for Canada and the United States. Agricultural and Forest Meteorology. 138: 69-81. Melvin, T.M. and K.R. Briffa. 2008. A ‘‘signal-free’’ approach to dendroclimatic standardization. Dendrochronologia. 26: 71-86. Peng, C., J. Liu, Q. Dang, M. J. Apps and H. Jiang. 2002. TRIPLEX: a generic hybrid model for prediction forest growth and carbon and nitrogen dynamics. Ecological Modelling. 153:109-130. Pichler, P. and W. Oberhuber. 2007. Radial growth response of coniferous forest trees in an inner Alpine environment to heat-wave in 2003. Forest Ecology and Management. 242: 688-699. Rocha, A.V., M.L. Goulden, A.L. Dunn and S.C. Wofsy. 2006. On linking interannual tree ring variability with observations of whole-forest CO2 flux. Global Change Biology. 12: 1378-1389. Savva, Y., J. Oleksyn, P.B. Reich, M.G. Tjoelker, E.A. Vaganov and J. Modrzynski. 2006. Interannual growth response of Norway spruce to climate along an altitudinal gradient in the Tara Mountains, Poland. Trees. 20: 735-746. Seebacher, T.M. 2007. Western Redcedar Dieback: Possible Links to Climate Change and Implications for Forest Management on Vancouver Island, B.C. MSc Thesis, University of British Columbia, Vancouver, BC. Pp.125. Su, H.X., W.G. Sang, Y.X. Wang and K.P. Ma. 2007. Simulating Picea schrenkiana forest productivity under climatic changes and atmospheric CO2 increase in Tianshan Mountains, Xinjiang Autonomous Region, China. Forest Ecology and Management. 246: 273-284. Trewartha, G.T. 1954. An Introduction to Climate. McGraw-Hill, New York. Pp.402. Vanclay, J.K. and J.P. Skovsgaard. 1997. Evaluating forest growth models. Ecological Modelling. 98: 1-12.  159  Wilmking, M., G.P. Juday, V.A . Barber and H.S.J. Zaldw. 2004. Recent climate warming forces contrasting growth responses of white spruce at treeline in Alaska through temperature thresholds. Global Change Biology. 10: 1-13. Zhang, Q.B. and R.J. Hebda. 2004. Variation in radial growth patterns of Pseudotsuga menziesii on the central coast of British Columbia, Canada. Canadian Journal of Forest Research. 34: 1946-1954. Zhang Q.-B., R.J. Hebda, Q.-J. Zhang and R.I. Alfaro. 2000. Modeling tree-ring growth responses to climatic variables using artificial neural networks. Forest Science. 46: 229-239.  160  Appendix A. Supplementary Information for the validation of MTCLIM A.1. Geographic weather stations information in the preliminary study  Before being ready to test the MTCLIM model, I did a preliminary study in the Okanagan Valley by using similar pairs of weather stations but different base stations (i.e. weather in bold case in Table A.1) and model’s default lapse rate (i.e. 6°C /1000m for maximum temperature and 3°C /1000m minimum temperature) and precipitation isohyets (i.e. 10cm). Table A. 1. Geographic information for the weather stations used to test MTCLIM in the preliminary study. Weather stations in bold case were replaced in the final study. Station  Latitude  Longitude Elevation (m) Record year  Simulation years 1909 – 1958, 1972-1999 (54 years)  Hedley  N 49.357° W120.077°  517  1904 - 2002  Hedley NP Mine  N 49.369° W 120.022°  1,707  1904 - 2002  Vernon North  N 50.344° W 119.271°  512  1990 - 2002  1,572  1994 - 2002  N 49.783° W 119.715°  351  1973 - 1981  N 49.733° W 118.933°  1,841  1981 - 1999  Vernon Silver Star Lodge N 50.358° W 119.056° Peachland Big White  161  1994, 1997 – 2000 (5 years) 1974 – 1981 (8 years)  Figure A. 1. Location of the weather stations used in the preliminary study to test MTCLIM in the Okanagan Valley area (Interior B.C.) (Google Inc., 2009).  A.2. Calculation of lapse rates and precipitation isohyets for the experimental sites After my preliminary validation approach of MTCLIM, I realized that the lapse rate and precipitation isohyet values are the key factors for this climate generate model. Therefore, in order to calculate the right regional lapse rate, I picked up 12 pairs of weather stations which had enough data and came with pairs (total has 72 weather stations close to my experiment site) and used their data of maximum and minimum temperatures recorded from Environment Canada to calculate the lapse rate for both variables.  162  Temperature (Degree C)  Lapse rate for maximum temperature  y = -0.0061x + 16.246 2  r = 0.8736  20 15 10 5 0 0  500  1000  1500  2000  2500  Elevation (m)  Figure A. 2. Relationship between altitude and average monthly maximum temperature in the Okanagan Valley. The slope of the regression function is the lapse rate of monthly maximum temperature in this area.  Temperature (Degree C)  Lapse rate for minimum temperature  y = -0.0042x + 4.0107 2  r = 0.7619  6 4 2 0 -2 -4 -6 0  500  1000  1500  2000  2500  Elevation (m)  Figure A. 3. Relationship between altitude and average monthly minimum temperature in the Okanagan Valley. The slope of the regression function is the lapse rate of monthly minimum temperature in this area.  As for precipitation, because there were not many higher elevation weather stations with enough precipitation data, I only picked 14 stations out of 72 which are at elevation higher than 100 m a.s.l. and I built a function to calculate the precipitation isohyet for my experimental site. Based on Figure A.4, the precipitation isohyet at given elevation can be calculated as: Precipitation (mm) = 0.0003 × elevation2 (m) - 0.3223 × elevation (m) + 458.98  163  2  y = 0.0003x - 0.3223x + 458.98  Precipitation (mm)  Precipitation isohyet  2  r = 0.8238  1200 1000 800 600 400 200 0 0  500  1000  1500  2000  Elevation (m)  Figure A. 4. Relationship between altitude and annual total precipitation in the Okanagan Valley.  A.3. Validation results of the preliminary study Table A. 2. Descriptive statistics of model performance for the simulation of average monthly maximum temperatures for each year in Hedley NP Mine, Vernon Silver Star Lodge and Big White (n = 12).  Simulation years  Hedley NP Mine  Vernon Silver Star Lodge  Big White  1909 – 1958, 1972-1999 (54 years)  1994, 1997 – 2000 (5 years)  1974 – 1981 (8 years)  record Annual T (Average (±SD))  12  simulation  record  simulation  7.69 ± 9.51 6.97 ± 11.42 5.03 ± 9.73 7.43 ± 10.67  record  simulation  12  4.42 ± 10.05  -  r  0.99  0.99  0.94  R2  0.97  0.98  0.89  MER (°C)  0.39  -1.35  1.90  MAE (°C)  2.40  2.01  2.41  ME  0.88  0.92  0.82  Theil’s U  0.26  0.23  0.82  No years with full 12 months records were available at Big White.  164  Table A. 3 Descriptive statistics of model performance for the simulation of average monthly minimum temperature in Hedley NP Mine, Vernon Silver Star Lodge and Big White. (n = 12).  Simulation years  Hedley NP Mine  Vernon Silver Star Lodge  Big White  1909 – 1958, 1972-1999 (54 years)  1994, 1997 – 2000 (5 years)  1974 – 1981 (8 years)  record  simulation  record  Annual T (Average (±SD)) -2.46 ± 7.91 -1.27 ± 8.02 -1.70 ± 8.02  simulation  record  simulation  1.36 ± 7.25  -13  0.59 ± 7.62  r  0.92  0.99  0.90  R2  0.97  0.97  0.81  MER (°C)  -1.07  -2.11  -2.88  MAE (°C)  1.35  2.11  2.88  ME  0.93  0.88  0.88  Theil’s U  0.25  0.36  0.37  Table A. 4. Descriptive statistics of model performance for the simulation of monthly total precipitation in Hedley NP Mine, Vernon Silver Star Lodge and Big White (n = 12).  Simulation years  Annual precipitation (Average (±SD)) r  Vernon Silver Star Lodge  Big White  1909 – 1958, 1972-1999 (54 years)  1994, 1997 – 2000 (5 years)  1974 – 1981 (8 years)  record 515.31 ± 136.01  simulation 340.19 ± 89.49  record 730.99 ± 190.62  simulation 498.44 ± 76.93  record -13  simulation 356.75 ± 38.42  0.86  0.41  0.93  R  0.75  0.17  0.86  MER (mm)  19.57  27.94  69.56  MAE (mm)  19.57  33.90  69.56  ME  0.15  0.69  0.18  Theil’s U  0.42  0.55  0.68  2  13  Hedley NP Mine  No years with full 12 months records were available at Big White  165  A.4. Regression results of monthly precipitation data in three sites.  Figure A. 5. Regression between observed and predicted values of monthly P in Hedley Mine.  166  Figure A. 5. (cont.). Regression between observed and predicted values of monthly P in Silver Star (top panel) and in Big White (bottom panel).  167  Appendix B. Additional statistical output for relations tree ring width – climate variables B.1. Principal Component Analysis B.1.1. Lodgepole pine a. ESSF Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  3.6210 2.8414 2.5883 2.4922 2.4235 2.1445 1.9431 1.8082 1.5568 1.4956 1.3859 1.2847 1.2149 1.1506 1.0847 1.0206 0.9048 0.8333  9.5291 7.4773 6.8113 6.5585 6.3777 5.6434 5.1136 4.7583 4.0967 3.9358 3.6472 3.3808 3.1970 3.0279 2.8545 2.6859 2.3810 2.1928  168  Cumulative Proportion 9.5291 17.0064 23.8177 30.3762 36.7538 42.3973 47.5108 52.2691 56.3659 60.3017 63.9489 67.3296 70.5266 73.5546 76.4091 79.0949 81.4759 83.6687  b. MS Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  3.6718 2.8437 2.6315 2.5445 2.4722 2.1749 2.0209 1.9132 1.5683 1.5103 1.3923 1.3044 1.2174 1.1921 1.1246 1.0381 0.9066 0.8339  9.4150 7.2916 6.7474 6.5244 6.3389 5.5765 5.1818 4.9056 4.0214 3.8725 3.5701 3.3446 3.1215 3.0568 2.8835 2.6617 2.3246 2.1382  Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  3.6229 2.8415 2.5910 2.4902 2.4240 2.1449 1.9439 1.8066 1.5567 1.4974 1.3863 1.2856 1.2166 1.1481 1.0842 1.0181 0.9035 0.8346  9.5341 7.4777 6.8186 6.5530 6.3791 5.6446 5.1156 4.7543 4.0965 3.9404 3.6481 3.3832 3.2016 3.0212 2.8532 2.6791 2.3777 2.1963  Cumulative Proportion 9.4150 16.7066 23.4540 29.9784 36.3173 41.8939 47.0756 51.9812 56.0026 59.8751 63.4452 66.7898 69.9113 72.9681 75.8516 78.5133 80.8380 82.9761  c. IDF  169  Cumulative Proportion 9.5341 17.0118 23.8303 30.3834 36.7624 42.4070 47.5226 52.2769 56.3734 60.3138 63.9619 67.3452 70.5467 73.5680 76.4212 79.1003 81.4780 83.6743  B.1.2. Hybrid spruce a. ESSF Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  Cumulative Proportion 0.1685 0.2368 0.3026 0.3648 0.4211 0.4710 0.5187 0.5610 0.5999 0.6359 0.6696 0.7013 0.7311 0.7589 0.7855 0.8092 0.8312 0.1685  2.9509 2.6621 2.5686 2.4244 2.1945 1.9461 1.8617 1.6510 1.5144 1.4055 1.3134 1.2377 1.1623 1.0849 1.0372 0.9231 0.8580 2.9509  0.0757 0.0683 0.0659 0.0622 0.0563 0.0499 0.0477 0.0423 0.0388 0.0360 0.0337 0.0317 0.0298 0.0278 0.0266 0.0237 0.0220 0.0757  Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  Cumulative Proportion  3.6292 2.8427 2.7048 2.5736 2.4241 2.1637 1.9549 1.8082 1.5730 1.4991 1.3921 1.3084 1.2300 1.1517 1.1213 1.0607 1.0185 0.8657  0.0931 0.0729 0.0694 0.0660 0.0622 0.0555 0.0501 0.0464 0.0403 0.0384 0.0357 0.0335 0.0315 0.0295 0.0288 0.0272 0.0261 0.0222  0.0931 0.1659 0.2353 0.3013 0.3634 0.4189 0.4690 0.5154 0.5557 0.5942 0.6299 0.6634 0.6950 0.7245 0.7533 0.7804 0.8066 0.8288  b. MS  170  B.1.3. Douglas-fir a. MS Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  Cumulative Proportion  3.6240 3.0663 2.6043 2.5180 2.4238 2.1509 2.0878 1.8859 1.5975 1.4972 1.3923 1.3425 1.2166 1.1738 1.1145 1.0436 0.9269 0.8480  0.0929 0.0786 0.0668 0.0646 0.0621 0.0552 0.0535 0.0484 0.0410 0.0384 0.0357 0.0344 0.0312 0.0301 0.0286 0.0268 0.0238 0.0217  0.0929 0.1715 0.2383 0.3029 0.3650 0.4202 0.4737 0.5221 0.5630 0.6014 0.6371 0.6716 0.7027 0.7328 0.7614 0.7882 0.8119 0.8337  Principal Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  Eigenvalue  Proportion  Cumulative Proportion  3.6561 2.9257 2.6826 2.5454 2.4249 2.1562 2.0109 1.8935 1.6336 1.5035 1.4674 1.3043 1.2285 1.1533 1.0901 1.0187 0.9196 0.8478  0.0937 0.0750 0.0688 0.0653 0.0622 0.0553 0.0516 0.0486 0.0419 0.0386 0.0376 0.0334 0.0315 0.0296 0.0280 0.0261 0.0236 0.0217  0.0937 0.1688 0.2375 0.3028 0.3650 0.4203 0.4718 0.5204 0.5623 0.6008 0.6385 0.6719 0.7034 0.7330 0.7609 0.7870 0.8106 0.8324  b. IDF  171  B.2. Correlations B.2.1. Lodgepole pine a. ESSF Variable  Correlation  Signif Prob -1  January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January  -0.0311 0.0408 0.0212 -0.0390 -0.0368 0.1133 -0.0372 0.0967 0.1423 0.1008 0.0092 -0.0054 -0.2140 0.1051 0.2015 -0.3122 -0.3074 0.0452 0.0843 -0.0887 -0.0195 -0.0175 -0.0381 -0.0505 -0.1786 -0.2183 -0.3632 -0.0311 0.1909 0.1100 -0.0666 -0.3951 0.0661 0.0851 0.0408 0.0408 0.2084 -0.0706 -0.0036 -0.0818 -0.0935 -0.0958 -0.0746 0.1394 -0.0671 0.0036 0.0092 -0.0366 0.0462 0.1712 -0.0843 -0.0350 0.0287 -0.0617 -0.0095 0.3208  0.7969 0.7355 0.8606 0.7468 0.7605 0.3467 0.7583 0.4225 0.2365 0.4029 0.9391 0.9643 0.0845 0.3833 0.0920 0.4955 0.2147 0.7904 0.5691 0.7349 0.9064 0.8993 0.7727 0.7768 0.2051 0.0939 0.0032 0.7969 0.1109 0.3611 0.5812 0.0006 0.5842 0.4806 0.7355 0.7355 0.0812 0.5585 0.9762 0.4977 0.4382 0.4266 0.5363 0.2462 0.5783 0.9760 0.9393 0.7617 0.7021 0.1534 0.4843 0.7720 0.8120 0.6118 0.9379 0.0047  172  Plot Corr (Limits -1 to +1) 0  +1  Variable  Correlation  Signif Prob -1  PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  0.1961 0.1708 0.2261 0.1099 0.1763 -0.1089 -0.0274 -0.1029 -0.0877 -0.1128 0.0073  Plot Corr (Limits -1 to +1) 0  +1  0.0896 0.1401 0.0496 0.3447 0.1277 0.3493 0.8140 0.3763 0.4511 0.3318 0.9502  b. MS Variable  Correlation  Signif Prob -1  January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March  0.2314 0.1587 0.2942 0.0070 -0.0023 0.0774 0.0004 0.0978 0.1243 0.1049 0.1326 0.1128 -0.2551 0.0633 0.3102 -0.0547 0.1577 0.1960 0.0772 0.0528 0.0743 0.0713 0.0714 0.0528 -0.1119 -0.2280 -0.3689 0.2314 0.3425 0.2463 0.0465 -0.2008 0.0067 -0.0710 0.1587 -0.0718 0.1077 0.1604 0.0625 -0.1777 -0.2729 -0.1559  0.0522 0.1861 0.0128 0.9537 0.9848 0.5209 0.9970 0.4172 0.3017 0.3838 0.2705 0.3488 0.0387 0.6000 0.0085 0.8732 0.4321 0.1918 0.5679 0.7666 0.6079 0.5918 0.5718 0.7666 0.4297 0.0798 0.0027 0.0522 0.0035 0.0384 0.7001 0.0931 0.9558 0.5564 0.1861 0.5518 0.3713 0.1815 0.6045 0.1382 0.0213 0.1943  173  Plot Corr (Limits -1 to +1) 0  +1  Variable CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  Correlation  Signif Prob  -0.0438 0.1471 0.0694 0.1700 -0.0948 -0.0929 -0.0997 0.1600 0.0333 -0.0534 0.2348 -0.0617 -0.0095 0.2517 0.2228 0.1747 0.2562 0.1600 0.2398 -0.0060 0.0370 -0.0385 -0.0729 -0.0477 -0.0279  0.7171 0.2209 0.5653 0.1565 0.4317 0.4407 0.4082 0.1824 0.7826 0.6580 0.0487 0.6118 0.9379 0.0283 0.0530 0.1311 0.0255 0.1673 0.0369 0.9591 0.7510 0.7415 0.5312 0.6826 0.8108  Correlation  Signif Prob  -1  Plot Corr (Limits -1 to +1) 0  +1  -1  Plot Corr (Limits -1 to +1) 0  +1  c. IDF Variable January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January  0.1429 0.0645 0.1888 -0.1392 -0.1369 -0.0910 -0.1315 -0.0745 -0.0551 -0.0783 0.0813 0.0450 -0.2619 -0.0089 0.4147 0.3207 0.0429 0.0784 0.0813 0.0583 0.0345 0.0869 0.0450 -0.2014 -0.3517 -0.2650 -0.2619 0.1429  0.2346 0.5933 0.1148 0.2468 0.2549 0.4506 0.2744 0.5367 0.6482 0.5161 0.5002 0.7095 0.0336 0.9411 0.0003 0.0530 0.7581 0.5315 0.5002 0.6698 0.7849 0.4743 0.7095 0.1441 0.0051 0.0329 0.0336 0.2347  174  Variable  Correlation  Signif Prob -1  PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  0.3169 0.2098 -0.0428 -0.1605 -0.1932 -0.1400 0.0646 -0.0201 -0.0107 0.1499 0.1131 -0.1987 -0.3453 -0.2614 0.0341 0.1518 0.0388 0.1793 -0.0886 -0.1152 -0.0975 -0.1070 -0.0836 -0.2152 0.1876 -0.0617 -0.0095 0.0301 0.0752 0.0714 0.1107 0.1256 0.0765 -0.0323 0.0329 0.0199 -0.0652 -0.0621 0.0735  0.0071 0.0791 0.7228 0.1812 0.1065 0.2443 0.5925 0.8676 0.9293 0.2123 0.3476 0.0967 0.0032 0.0277 0.7780 0.2063 0.7482 0.1346 0.4627 0.3387 0.4188 0.3743 0.4880 0.0716 0.1172 0.6118 0.9379 0.7964 0.5188 0.5401 0.3410 0.2795 0.5114 0.7819 0.7779 0.8646 0.5759 0.5941 0.5282  175  Plot Corr (Limits -1 to +1) 0  +1  B.2.2. Hybrid spruce a. ESSF Variable  Correlation  Signif Prob -1  January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January  0.0547 -0.0613 0.0848 -0.0614 0.2009 0.2065 0.1875 0.2025 0.1893 0.2123 -0.0648 -0.0352 -0.1768 -0.0053 0.1826 0.4058 0.1874 0.0840 0.0297 0.2265 -0.0066 0.0979 -0.0092 -0.0469 0.0058 -0.1853 -0.1619 0.0547 0.0385 0.0045 0.1098 -0.0379 0.2637 0.0578 -0.0613 -0.0613 0.1636 0.0835 -0.0930 -0.1080 0.0082 0.0420 0.0375 -0.0834 -0.2751 0.0487 0.1069 0.0377 -0.1302 0.0191 0.0729 -0.0574 0.0929 -0.0617 -0.0095 0.1755  0.6507 0.6116 0.4821 0.6108 0.0930 0.0841 0.1174 0.0903 0.1138 0.0754 0.5914 0.7706 0.1555 0.9650 0.1275 0.3663 0.4565 0.6211 0.8412 0.3819 0.9683 0.4769 0.9441 0.7923 0.9676 0.1563 0.2011 0.6507 0.7497 0.9701 0.3620 0.7536 0.0263 0.6322 0.6116 0.6116 0.1729 0.4886 0.4407 0.3699 0.9457 0.7278 0.7563 0.4894 0.0202 0.6865 0.3749 0.7552 0.2792 0.8747 0.5455 0.6344 0.4408 0.6118 0.9379 0.1433  176  Plot Corr (Limits -1 to +1) 0  +1  Variable  Correlation  Signif Prob -1  PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  0.0930 0.1771 0.2177 0.0283 0.0520 -0.0032 -0.0391 -0.0379 -0.0249 0.0566 0.1507  Plot Corr (Limits -1 to +1) 0  +1  0.4403 0.1396 0.0682 0.8149 0.6666 0.9787 0.7459 0.7536 0.8365 0.6393 0.2095  b. MS Variable  Correlation  Signif Prob -1  January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February  0.0450 0.0447 0.0646 -0.1186 0.0858 0.0370 0.0657 0.1076 0.0808 0.1226 -0.1136 0.0701 0.3105 0.1138 0.0423 -0.0908 0.2145 0.0907 0.0393 -0.0728 -0.0654 0.0679 0.0610 -0.0027 -0.0358 0.2402 0.2757 -0.0948 0.1111 0.0450 0.1818 -0.0473 0.1125 0.0834 -0.0651 0.0208 0.0447 0.0025 -0.1622 -0.0040 -0.0513  0.7095 0.7111 0.5924 0.3244 0.4769 0.7593 0.5861 0.3717 0.5029 0.3083 0.3454 0.8376 0.1149 0.4514 0.7545 0.4517 0.2385 0.5310 0.7676 0.5644 0.6017 0.7028 0.6677 0.9834 0.7788 0.0520 0.0250 0.4316 0.3565 0.7095 0.1293 0.6951 0.3503 0.4893 0.5898 0.8636 0.7111 0.9837 0.1765 0.9736 0.6708  177  Plot Corr (Limits -1 to +1) 0  +1  Variable  Correlation  Signif Prob -1  CMI March CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  -0.0541 -0.0205 0.0125 -0.1699 -0.1151 -0.1339 -0.0328 0.0117 -0.0637 -0.0880 0.0233 0.0410 -0.1290 0.1013 -0.1056 -0.0761 -0.0257 -0.0330 -0.0485 -0.0548 -0.0331 -0.0830 0.0632 -0.1721 -0.1205 0.0113  0.6540 0.8654 0.9177 0.1567 0.3391 0.2656 0.7859 0.9229 0.5980 0.4654 0.8469 0.7343 0.2837 0.4005 0.3808 0.5284 0.8316 0.7844 0.6878 0.6498 0.7841 0.4915 0.6003 0.1511 0.3170 0.9255  178  Plot Corr (Limits -1 to +1) 0  +1  B.2.3. Douglas-fir a. MS Variable  Correlation  Signif Prob -1  January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January PDO February  0.3046 -0.0493 0.2087 0.1170 -0.0544 0.0187 -0.0289 0.0319 0.0617 0.0491 0.2194 0.2024 -0.1585 -0.0479 0.2431 -0.1812 -0.0974 0.0813 0.0624 0.0234 0.0040 0.0891 0.2374 0.0988 -0.0719 -0.2396 -0.2965 0.3046 0.1381 0.1492 -0.0747 -0.1900 0.0561 -0.0829 -0.0493 -0.0251 0.1132 0.1983 0.0388 -0.2078 -0.0999 -0.0492 0.0600 0.1717 0.1379 0.1779 0.0336 -0.0169 -0.1156 0.1725 -0.0004 0.0784 0.0622 -0.0617 -0.0095 0.3346 0.2875  0.0098 0.6831 0.0807 0.3314 0.6522 0.8772 0.8109 0.7915 0.6092 0.6840 0.0660 0.0905 0.2038 0.6917 0.0411 0.5939 0.6289 0.5912 0.6446 0.8988 0.9779 0.5020 0.0569 0.5781 0.6124 0.0652 0.0174 0.0098 0.2506 0.2142 0.5357 0.1125 0.6424 0.4920 0.6831 0.8356 0.3474 0.0974 0.7482 0.0821 0.4072 0.6838 0.6190 0.1522 0.2515 0.1378 0.7806 0.8887 0.3371 0.1503 0.9976 0.5157 0.6066 0.6118 0.9379 0.0043 0.0150  179  Plot Corr (Limits -1 to +1) 0  +1  Variable PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  Correlation  Signif Prob  0.1994 0.2490 0.1548 0.2399 -0.0191 -0.0029 -0.1176 -0.0420 0.0189 -0.0883  0.0954 0.0363 0.1974 0.0439 0.8742 0.9810 0.3288 0.7282 0.8756 0.4642  Correlation  Signif Prob  -1  Plot Corr (Limits -1 to +1) 0  +1  -1  Plot Corr (Limits -1 to +1) 0  +1  a. IDF Variable January Temperature August Temperature Annual Mean Temperature Annual Total Precipitation GDD at 5 from May to August GDD at 5 from June to August Annual Accumulative GDD above 5 °C GDD at 10 from May to August GDD at 10 from June to August Annual Accumulative GDD above 10 °C Precipitation from May to August Precipitation from June to August PP12 to P2 TM tm P from May to August < 150 mm P from May to August < 200 mm P from May to August < 250 mm P from May to August < 300 mm P from June to August < 150 mm P from June to August < 200 mm P from June to August < 250 mm P from June to August < 300 mm PP12to P2 < 150 mm PP12 to P2 < 200 mm PP12 to P2 < 250 mm PP12 to P2 < 300 mm PET January PET February PET March PET April PET May PET June PET July PET August PET September PET October PET November PET December CMI January CMI February CMI March  0.2390 -0.0226 -0.0074 0.2861 -0.2396 -0.1645 -0.2560 -0.1992 -0.1440 -0.2154 0.3863 0.3116 0.0556 -0.1516 0.0372 0.3437 0.4385 0.3869 0.3863 0.3145 0.3235 0.3613 0.3116 0.1097 0.0519 0.0295 0.0556 0.0055 0.2390 0.1116 0.0340 -0.1660 -0.2413 -0.1962 -0.2627 -0.0224 -0.0866 -0.1722 0.2410 -0.1165 -0.1285 -0.0830  0.0447 0.8519 0.9512 0.0156 0.0441 0.1704 0.0312 0.0959 0.2310 0.0713 0.0009 0.0082 0.6575 0.2071 0.7578 0.0373 0.0009 0.0013 0.0009 0.0183 0.0086 0.0021 0.0082 0.4298 0.6889 0.8155 0.6575 0.9634 0.0447 0.3543 0.7783 0.1664 0.0427 0.1010 0.0269 0.8530 0.4728 0.1509 0.0429 0.3332 0.2854 0.4912  180  Variable  Correlation  Signif Prob -1  CMI April CMI May CMI June CMI July CMI August CMI September CMI October CMI November CMI December Annual CMI Annual PET Previous Annual CMI Previous Annual PET PDO January PDO February PDO March PDO April PDO May PDO June PDO July PDO August PDO September PDO October PDO November PDO December  0.0708 0.1864 0.3605 0.3245 0.2784 0.0252 -0.0511 -0.0203 0.0625 0.0491 0.2313 -0.0617 -0.0095 0.1196 0.0846 0.0069 0.0369 0.0578 0.1382 -0.0123 -0.0004 -0.0503 -0.0821 -0.0043 -0.0857  0.5572 0.1195 0.0020 0.0058 0.0187 0.8348 0.6719 0.8669 0.6048 0.6843 0.0523 0.6118 0.9379 0.3203 0.4828 0.9544 0.7602 0.6320 0.2503 0.9192 0.9971 0.6769 0.4959 0.9718 0.4773  181  Plot Corr (Limits -1 to +1) 0  +1  B.3. Regressions B.3.1. Lodgepole pine a. ESSF Variable  r2  Signif Prob  Previous Apr. T  0.1123  0.0046  Previous Nov. T  0.0884  0.0124  Previous Dec. T  May T  0.156139  0.0006  Accum. PP12 to P2  Variable Previous Aug. T  r2  Signif Prob  0.0732  0.0235  0.0756  0.0212  .  < 300 mm  182  0.1319  0.0032  Variable  r2  Signif Prob  Variable  r2  Signif Prob  PET May  0.1561  0.0006  PDO January  0.1029  0.0047  PDO April  0.0511  0.0496  183  b. MS Variable  r2  Signif Prob  Variable  r2  Signif Prob  Annual T mean  0.0866  0.0128  PP12 to P2  0.0651  0.0387  Min of Tmin  Previous Aug. T  0.0962  0.1057  0.0085  Previous May T  0.0060  Previous Dec. T  184  0.08241  0.0896  0.0160  0.0118  Variable  r2  Signif Prob  Variable  r2  Signif Prob  Feb. T  0.1173  0.0035  Mar. T  0.0607  0.0384  Previous Aug. P  Accum. PP12 to  0.1484  0.1361  0.0010  Previous Oct. P  0.0027  PET Feb.  P2 < 300 mm  185  0.0566  0.1173  0.0474  0.0035  Variable  r2  Signif Prob  Variable  r2  Signif Prob  PET Mar.  0.0607  0.0384  CMI Feb.  0.0745  0.0213  Annual PET  0.0551  4.0273  PDO January  0.0633  0.0283  PDO April  0.0657  0.0255  PDO June  186  0.0575  0.0369  c. IDF Variable  r2  Signif Prob  Variable  r2  Signif Prob  Accum PP12 to P2  0.0686  0.0336  Min of Tmin  0.1720  0.0003  Previous Jun. T  0.0728  0.0239  Feb. T  0.1004  0.0071  Feb. P  0.0753  0.0206  P from May to  0.1028  0.0530  Aug < 150 mm  187  Variable  r2  Signif Prob  Variable  r2  Signif Prob  Accum PP12 to  0.1237  0.0051  Accum PP12 to  0.0702  0.0329  P2 < 200 mm  Accum PP12 to  P2 < 250 mm  0.0686  0.0336  PET Feb.  0.1004  0.0071  0.1192  0.0032  CIM Mar.  0.0683  0.0277  P2 < 300 mm  CIM Feb.  188  B.3.2. Hybrid spruce a. ESSF Variable  r2  Signif Prob  Variable  r2  Signif Prob  Previous Jul. T  0.1370  0.0022  Previous Aug. T  0.1409  0.0019  Previous Jun. T  0.0695  0.0263  Previous Jul. P  0.0865  0.0165  Previous Aug. P  0.1071  0.0073  Jun. P  0.057297  0.0444  189  Variable  r2  Signif Prob  Variable  r2  Signif Prob  PET June  0.0695  0.0263  CMI June  0.0757  0.0202  b. MS Variable  r2  Signif Prob  Variable  r2  Signif Prob  Accum Jun. P to  0.0760  0.0250  Previous Jul. T  0.0624  0.0431  0.0645  0.0397  Previous Aug. P  0.0596  0.0483  August P  Previous Jul. P  190  B.3.2. Douglas-fir a. MS Variable  r2  Signif Prob  Variable  r2  Signif Prob  Jan. T  0.0928  0.0098  Min of Tmin  0.0591  0.0411  Previous Apr. T  0.0962  0.0113  Previous May T  0.1383  0.0021  Previous Nov. T  0.0747  0.0264  Previous Dec. T  0.1066  0.0075  191  Variable  r2  Signif Prob  Variable  r2  Signif Prob  PET Jan.  0.0928  0.0098  Accum PP12 to  0.0879  0.0174  P2 < 300 mm  PDO January  0.1120  0.0043  PDO February  0.0829  0.0150  PDO April  0.0620  0.0363  PDO June  0.0576  0.0439  192  b. IDF Variable  r2  Signif Prob  Variable  r2  Signif Prob  Jan. T  0.0571  0.0447  Annual P  0.0819  0.0156  GDD above 5°C  0.0574  0.0441  GDD above 10°C  0.0655  0.0312  0.0971  0.0082  May to Aug.  P May to Aug.  May to Aug.  0.1492  0.0009  P Jun. to Aug.  193  Variable  r2  Signif Prob  Variable  r2  Signif Prob  Previous Apr. T  0.0723  0.0290  Previous Jul. T  0.0682  0.0342  Previous Nov. T  0.0642  0.0402  May T  0.0583  0.0426  Jul. T  0.0699  0.0269  May P  0.1156  0.0037  194  r2  Signif Prob  Variable  r2  Signif Prob  Jun. P  0.0965  0.0084  Jul. P  0.0664  0.0300  Accum P May to  0.1181  0.0373  Accum P May to  0.1923  0.0009  0.1492  0.0009  Variable  Aug < 150 mm  Accum P May to  Aug < 200 mm  0.1497  0.0013  Accum P May to  Aug < 250 mm  Aug < 300 mm  195  Variable  r2  Signif Prob  Variable  r2  Signif Prob  Accum P June to  0.0989  0.0183  Accum P June to  0.1047  0.0086  0.0971  0.0082  0.0582  0.0427  Aug < 150 mm  Accum P June to  Aug < 200 mm  0.1306  0.0021  Accum P June to  Aug < 250 mm  PET Jan.  Aug < 300 mm  0.0571  0.0447  PET May  196  Variable  r2  Signif Prob  Variable  r2  Signif Prob  PET Jul.  0.0689  0.0269  PET Nov.  0.0581  0.0429  CMI May  0.1299  0.0020  CMI Jun.  0.1053  0.0058  CMI Jul.  0.0775  0.0187  197  Appendix C. Review of Models linking tree growth and climate and additional graphical output of the model test C.1. Stand level climate change models: a review Fourteen stand level model used for predicting climate change effects have been reviewed and compared. Although nowadays there are many simulation models capable to access climate change impacts, I have focused my review on those whose conceptual models or model structures are similar to FORECAST (Tables C.1 and C.2).  C.1.1. PnET PnET is a monthly process stand dynamics model. It uses monthly time steps because the developers assumed the aggregation of daily data into months would not cause a significant loss of information. This assumption has been tested and proven before (Aber and Federer, 1992). The model structure focuses on water and carbon balances. It deals with climate change via temperature and precipitation (water balance), but it does not include atmospheric CO2. Actually, the physiological process used to produce biomass is similar to CENTURY model (described below, Parton et al., 1993). It also has a similar carbon and water balance structure to that of FOREST-BGC (Running and Coughlan, 1988; Coughan and Running, 1997, described below) and BIOMASS (McMurtrie et al., 1990; described below) models, with the exception that their time steps are different (Aber and Federer, 1992). The central concept behind the PnET model is that photosynthesis is a function of foliage N, and water use efficiency (WUE) is a function of vapour pressure deficit (VPD) (Aber and Federer, 1992). Therefore, the function of maximum net photosynthesis per unit leaf area (NetPsnmax, µmolCO2m-2sec-1) and foliage N content (N %) is: NetPsnmax = -5.98 + 4.86 * N% Meanwhile, they assumed basal respiration of foliage is 10% of the maximum net photosynthesis rate, so maximum gross photosynthesis (GrossPsnmax) is 1.1 times maximum net photosynthesis. In this model, they assumed the actual gross photosynthesis (GrossPsn) would be affected by temperature (DTemp), water stress (Dwater) and vapour pressure deficit (DVPD). Therefore, the actual gross photosynthesis is:  198  GrossPsn = GrossPsnmax * DTemp * Dwater * DVPD where DTemp, Dwater and DVPD are modifiers scaled from 0 to1. However, there is no decomposition component in this model. It is a simple physiological process model with no representation of forest management.  C.1.2. Forest-BGC and Tree-BGC Tree-BGC is a variant of the Forest-BGC model. Most parts of these two models are very similar except the scales are different (e.g. one is a tree-level model and the other is a stand-level model). Forest-BGC is a stand level, process-based, mixed time scale (daily and yearly) ecosystem model (Running and Coughlan, 1988; Korol et al., 1995). It is used to predict stand growth and to provide site quality index (Korol et al., 1995). The key processes considered in this model are the effects of carbon, nutrient and water cycling in forest ecosystems. Short-wave radiation, air temperature, dew point and precipitation are daily input data used to drive the model (Running and Coughlan, 1988). The model calculates daily canopy photosynthesis (PSN; kg CO2 day-1) by multiplying CO2 diffusion gradient ( ∆CO2 ; kg m-3), radiation and temperature-controlled mesophyll CO2 conductance (CM; m s-1). PSN = [( ∆CO2 CC CM) / (CC + CM)] LAI DAYL. The other parameters of this algorithm are CC: canopy conductance (m s-1), LAI: leaf area index (m2 m-2), and DAYL: day length for a flat surface (s). The mesophyll CO2 conductance (CM) is calculated from three modifier functions: nitrogen (CMn), light (CMq) and temperature (CMt). These modifiers are all scaled from 0 to 1. CMn = 67.0 LEAFN CMq = (Q - Q0) / (Q + Q0.5) CMt = (TMAX –TAIR) (TAIR-TMIN) / TMAX2 LEAFN is leaf nitrogen concentration (fraction dry wt). Q is canopy average radiation (kJ m-2day-1). Q0 is photosynthesis light compensation point (kJ m-2day-1). Q0.5 is radiation level where CMq is 0.5 of maximum (kJ m-2day-1). TMAX and TMIN are high and low temperature (°C) at photosynthesis compensation points. TAIR is daylight average air temperature. Based on those values, the model calculates daily canopy photosynthesis, then subtracts the value of night canopy respiration (calculated from night average temperature and LAI) and gets daily net  199  canopy carbon fixation. The reason this model considers respiration is because it is a key process of the carbon budget. The daily maintenance respiration of stem and root biomass is calculated from compartment size and average air and soil temperature under a Q10 = 2.3 assumption. Rl,s,r = α exp ( 0.085 TEMP) Cl,s,r where Rl,s,r is maintenance respiration of leaf, stem and root compartments (kg day-1); α is scaling factor for leaf, stem and root compartments (0.00015, 0.0010 and 0.0002 kg-1day-1kg-1); 0.085 is also a scaling factor given Q10 = 2.34. TEMP (°C) here represents night and daily average air temperature and soil temperature. The night time average temperature is used for leaf respiration, daily average is used for stem respiration, and soil temperature is used for root respiration. Cl,r is carbon storage either in leaf part or root part. Cs is stem respiration calculated by the function Cs = exp (0.67 ln (stem carbon storage)) The yearly growth respiration is calculated as a fixed fraction of the carbon allocated to leaf, stem and root compartment. The coefficients are from literature review and are independent of temperature. Unlike PnET, Forest- BGC considers nutrient cycling and therefore it has a decomposition component. The annual litter decomposition function is: DECOMP = (-3.44 + 0.100AET) – ((0.0134 + 0.00147 AET) LIG) where DECOMP is annual percent weight loss of fresh litter (% year-1) and LIG is initial litter lignin concentration (% dry wt). Actual annual evapotranspiration (AET; mm year-1) is calculated from daily model. The developers of this model point out the shortcoming of Forest-BGC as the fact that the canopy is homogeneous. Therefore, the leaf area index is proportional to the depth of the canopy and to some degree, it may not capture the water and carbon budgets accurately (Running and Coughlan, 1988). Because of the lack of a management component, it cannot be a management tool for foresters. However, it is a suitable research tool to predict the impact of climate change. However, the model is offering a link between input data and GIS databases which is good for future data collection. This model has either developed a series of submodels (Tree-BGC, Fire-BGC) or is combined with other models (PnET-BGC) to overcome its weakness and has been widely used for predicting climate change or disturbance effects.  200  Tree-BGC, a variant of FOREST-BGC model, is also a stand level, process-based, mixed time scale (daily and yearly) ecosystem model. The purpose of the model is the same as Forest-BGC: to calculate carbon, water and nitrogen flows in forest ecosystems (Korol et al., 1995). The only difference of these two models is that in Tree-BGC, all the simulated processes are based on individual tree physiological characters, and it focuses on light competition and ignores decomposition. To scale up the simulation results from individual tree level to the stand level, Tree-BGC has to make an important assumption that the reactions of individual photosynthesis processes under different constraining factors at tree level are same at stand level (Korol et al., 1995). Most structures of Tree-BGC are very similar to the ones in Forest-BGC. Therefore, each tree daily canopy photosynthesis (PSNi; kg CO2 day-1) is calculated as:   PARi PSNi = PSN   ∑ PARi      where PSN is stand daily canopy photosynthesis (kg CO2 day-1); PARi is individual tree’s photosynthetically active radiation (MJ m-2), and the stand daily canopy photosynthesis is the sum of tree daily canopy photosynthesis. Not only the photosynthesis, but also the maintenance respiration has been modified compared to Forest-BGC. The maintenance respiration of each stem (MRs; kg C) is multiplied by stem respiration coefficient (f; kg C -1 day-1kg-1); temperature (T) controlled function and respiration volume (RV; m3ha-1) which is the sum of phloem and live sapwood volume. MRs = f e (0.085T) RV The maintenance respiration of leaf (MRLi; kg C) and root (MRri; kg C) are allocated to each tree (i) proportional to its leaf and root carbon. Each tree’s yearly maintenance respiration (MRi; kg C year-1) is calculated by following the function: MRi = MRLi MRsi + MRri Tree-BGC does not consider decomposition because it focuses only on competition for light.  C.1.3. BIOMASS BIOMASS is a daytime step, stand-level process model, used for simulating forest carbon and water-balance and used for predicting forest growth (McMurtrie et al., 1990; McMurtrie and  201  Landsberg, 1992). The two main components of the model are the canopy assimilation and water balance. General speaking, canopy assimilation is a function of an elaborated simulation of stomata processes (involving radiation, CO2 concentration, temperature, soil water, etc.) and foliage nitrogen content (McMurtrie et al., 1990; McMurtrie and Landsberg, 1992; McMurtrie et al., 1992; McMurtrie and Wang, 1993). In McMurtrie et al. (1989), they use respiration in the calculation of dry-matter production, allocating it to different parts of the tree and estimating litterfall. There is no decomposition component in this model. BIOMASS is a daily time step model, and it is easy to calibrate it processes by getting daily input data from a standard meteorological station (McMurtrie et al., 1990). It separates the canopy vertically into three homogenous layers and simulates the details of stomata process for each layer. It is totally a physiological process-based model and shares the strengths and shortcomings of all completely mechanistic models: including the requirement of many input data (McMurtrie et al., 1990). Because it simulates the details of the stomata to control photosynthesis, inputs include CO2 concentration, temperature and soil moisture. Consequently it is a powerful tool for predicting climate change impact as long as the calculated rates of all the physiological process relationships remain the same. In the water balance component, BIOMASS considers impacts of different silviculture strategies on the dynamics of soil water, so for areas where moisture is the major determent of growth, BIOMASS can be used as a management tool.  C.1.4. LINKAGES The LINKAGES model is designed to help to understand the ecosystem carbon and nitrogen storage and cycling under climate and soil moisture constraints (Pastor and Post, 1985). It can be seen as an offspring of JABOWA model (Botkin, 1993). The model time step is yearly, but simulating of the effects of temperature and precipitation are based on monthly data (Pastor and Post, 1985). The model contains two parts: the environmental and the tree species population components. The environmental component includes three subcomponents: TEMPE (temperature), MOIST (soil moisture) and DECOMP (decomposition) which are used to determine the site conditions. The population component also has three subroutines: BIRTH, GROW and KILL. These are used  202  to calculate the population dynamics. These two groups are connected by GMULT (modifier for optimal birth rate, annual stem growth and mortality) (Pastor and Post, 1985). Because the model structure and concepts are inherited from JABOWA, refer to the models I have described above, LINKAGES focuses more on how stand structure changes than how stand productivity changes. Sunlight is the driving variable for stand dynamics (Pastor and Post, 1985). In the TEMPE subroutine, LINKAGES uses a random number generator algorithm to generate daily temperature based on each month’s mean and stand deviation, and sums the number of degree days for the year. In MOIST, it uses Thornthwaite and Mather’s water-budgeting method to calculate actual evapotranspiration as the input to DECOMP. LINKAGES also considers soil physical characters (depth, texture), monthly temperature and rainfall to calculate the dry days of the year as an input to the GMULT subroutine. In the DECOMP subroutine, it calculates mass loss, nitrogen immobilization and mineralization, lignin decay and CO2 loss from decomposing litter cohorts and humus. As I mentioned above, the simulation objective of LINKAGES is different from the other models, which I reviewed. Unlike other models’ calculation of either GPP or NPP, LINKAGES calculates annual diameter and height increment as a function of site and climate variables (Pastor and Post, 1985). Because it follows the ideas of JABOWA, it is more like a plant dynamics model than a stand production model and it does not contain any management tools. Therefore, it is more of a research model than an applied model. However, because many stand production dynamic simulation models in use today use the concepts in LINKAGES, it is worth considering.  C.1.5. G’DAY G’DAY is more a plant-soil model than a stand simulation model (Medlyn et al., 2000). It describes how photosynthesis and nutrient factors interact with each other (Comins and McMutrie, 1993). The model is designed to predict the forest growth response to elevated atmospheric CO2 concentrations and temperature. It predicts the response from decadal to century time scales (Medlyn et al., 2000). The earlier version of G’DAY was linked with CENTURY (Parton et al., 1993). The latest version uses the BEWDY model (Medlyn, 1996) to replace the plant production calculation of CENTURY, but it still keeps other components of this soil model (i.e. soil carbon and nutrient dynamic components). This is because they think  203  BEWDY is more mechanistic and it considers the temperature and CO2 effects on plant photosynthesis and respiration better than CENTURY (Medlyn et al., 2000). In developing G’DAY, the authors considered two approaches to represent plant respiration biomass loss because this part is still a big debate among ecosystem modellers on how to deal with this process (Medlyn et al., 2000). In model 1, they separate respiration into maintenance respiration (Rm) and growth respiration (Rg). Maintenance respiration is assumed to be proportional to the non-structural nitrogen content of the plant. The growth respiration is calculated by a ratio (Yg; between 0 and 1) of the difference between potential photosynthesis (or gross primary production, growth canopy photosynthesis; Pg) and maintenance respiration (Rm). Therefore, net primary production (NPP) is the result after potential photosynthesis minus maintenance respiration minus growth respiration: NPP = (1 – Yg) (Pg – Rm) In model 2, the authors assumed that respiration is a constant fraction of growth canopy photosynthesis (Pg): NPP = f Pg where f is the carbon use efficiency factor, independent of temperature and atmospheric CO2 concentration (Medlyn et al., 2000). Gross primary production (Pg) is calculated from the BEWDY model in which the photosynthesis rate depends on canopy leaf area index, the intensity of beam and diffuse radiation, leaf N content, air temperature and CO2 concentration. Detail can be found in Medlyn (1996). There is no decomposition rate function in the model description, but decomposition is implicit in each component of the nitrogen cycle. The decomposition rate is temperature dependent. G’DAY is an annual time step model dealing with atmosphere CO2 and temperature effects. No management tools are included in this model, but it does predict long-term forest production as an index of the impact of climate change.  C.1.6. 3-PG (Physiological Principles in Predicting Growth) 3-PG is based on similar ideas to LINKAGES and other models developed later about how forest stands grow. It is a physiological process stand growth model that uses monthly weather data as input (Landsberg and Waring, 1997). The model is based on well-established  204  physiological principles and empirical data and therefore does not need much local calibration to predict forest growth. General speaking, it uses absorbed photosynthetically active radiation to calculate gross primary production (PG) and then uses the ratio (Cpp) of net primary production (PN) to gross primary production (Cpp = 0.45 ± 0.05) to calculate net primary production. The authors believe that the ratio is a constant. 3-PG employs data and functions of growth effects under different growing condition from the literature to create a simple relationship between root growth and turnover rate to estimate the below-ground carbon allocation. For the above ground part, it uses carbon allometric ratios, age effects and the 3/2 power law to constrain tree growth patterns and stand dynamics (Landsberg and Waring, 1997). Gross primary production is the product of φ p.a.u . (utilizable, absorbed photosynthetically active radiation) and α c (canopy quantum efficiency coefficient = 0.03 mol C (mol photon)-1 = 1.8 g C MJ-1). The model uses α c as a constant that is based on three reference papers and concludes that the maximum canopy quantum efficiency for the forest is about 0.03 mol C (mol photon)-1 and does not vary much. The utilizable, absorbed photosynthetically active radiation  φ p.a.u . is calculated from modifiers that come from monthly means of day-time vapour pressure deficit, soil water, temperature, and tree age: PG =  φ p.a.u.  × αc  3-PG does not have a strong nutrient component; the only consideration of nutrients in 3-PG is that nutrient availability will affect root growth therefore change carbon allocation (Landsberg and Waring, 1997). Although 3-PG is not as complicated as other models (BIOMASS, G’DAY etc.), it incorporates important ideas about how forest stands produce biomass. The model does not consider canopy complexity, does not have a water balance component, and does not attempt to be a management tool, but it contains accordantly physiological processes which have been proven good enough to produce accurate prediction for their experimental site (Landsberg and Waring, 1997).  205  C.1.7. CENTURY, TREEDYN3, and TRIPLEX  Combining the strengths of 3-PG, CENTURY and TREEDYN3, TRIPLEX was built as a meta-model of existing powerful models, which avoided the difficult early model development period. The developers of TRIPLEX believe that linkage of existing models as a meta-model instead of spending time and money to develop a completely new model to represent the ecosystem is a global trend (Peng et al., 2002). As I have already introduced 3-PG, here I will introduce CENTURY and TREEDYN3, and then briefly describe the TRIPLEX model. CENTURY is a terrestrial biogeochemistry model. It focuses on the plant-soil linkage, which therefore is the target of the simulation rather than the forest stand. It has a detailed soil nutrient component (Parton et al., 1993). It represents the relationship between climate, forest management, soil characters, plant productivity and decomposition. It incorporates key process relating to carbon assimilation and turnover from existing models. It contains three main components: soil organic C model, N submodel and an aboveground production model. The soil organic matter submodel contains three major components: active soil organic matter, a slow organic matter pool, and a passive stable organic component. This well developed submodel, which is used in many other models (G’DAY and TRIPLEX), uses temperature and moisture as two of the factors, which control decomposition rate. In the temperature part, it uses mean monthly soil temperature as the input; in the soil moisture part, the input is the ratio of stored soil water plus monthly precipitation to potential evapotranspiration. The decomposition model is as follows: dC I = K I LC AC I dt  dC I = K I ATm C I dt dC I = K I AC I dt  I = 1, 2  I = 3 I = 4, 5, 6, 7, 8  Tm = (1 − 0.75T )  LC = e ( −3Ls ) CI and KI stand for carbon in different pools and the maximum decomposition rate (year-1) of that pool; I = 1: surface material (K1 = 3.9); 2: soil structure material (K2 = 4.9); 3: active soil organic matter (K3 = 7.3); 4: surface microbes (K4 = 6.0); 5: surface metabolic material (K5 = 14.8); 6: 206  soil metabolic material (K6 = 18.5); 7: slow soil organic matter (K7 = 0.2) and 8: passive organic matter (K8 = 0.0045). A is the combined effect of soil moisture and soil temperature. Tm is the soil texture effect (silt plus clay content) in the active soil organic matter component. Ls is the structural material and Lc is the impact of lignin content. The nitrogen submodel is similar to the soil C submodel. Organic N is the product of the carbon and the N: C ratios of the soil stable component that receives the C. CENTURY can simulate plant production for different ecosystems (i.e. grasslands, agricultural crops, forests and savannah). However, this research mainly focuses on grasslands. The general idea is that above-ground production is a function of soil temperature, available water and self-shading factor. But it also relates the soil nutrient supply (N, P and S). Unlike most of the physiological models, CENTURY does not consider detailed solar radiation effects. The model developers did not consider the effects of changes in the plant community (Parton et al., 1993). Because the time step is monthly, it is not sensitive to daily rainfall patterns and there is a lag effect between nutrient effects and photosynthetic storage in plant. CENTURY is not considered to be a tool for foresters and there is no representative of silviculture strategies in this model. TREEDYN3 is a process model, which predicts tree growth, carbon and nitrogen dynamic in a single species, even-aged forest stand (Bossel, 1996). But it also has a description of stand structure. The model is different from other models in that it introduces diurnal and seasonal variation in physiological processes (i.e. photosynthesis; seasonal dynamic of respiration, phenology and soil processes) and it considers energy and mass balance of carbon and nitrogen flow (Bossel, 1996). The reason for using diurnal and seasonal scales is because these physiological processes are sensitive to daily and seasonal variation. TREEDYN3 is designed to explore the effects of climate change, air pollution, and different forest management strategies (Bossel, 1996). The photosynthate storage A is the result of net photosynthetic production (αprod) and assimilate relocation (αreloc) minus the assimilate consumption from growth (αgrow), respiration (αresp) and death (αdead). dA = αprod + αreloc – αresp – αgrow – αdead dt  For details of each part, please see Bossel (1996).  207  The respiration submodel calculates respiration consumption from following function: αresp =     h  kTr σ L 1 −  L + σ w bW + σ F τ F F  + kTs σ R R 24      where kTr and kTs are temperature modifiers of air and soil temperatures. σL, σW, σF and σR are the respiration rate of leaf, wood, fruit and fine roots. L is leaf mass, b is the proportion of respiring wood volume (sapwood) and τF respiration period when there is fruit, and R is fine root mass. The authors considered respiration because they think it is a limiting factor for tree growth. Litter and humus decomposition (cGE, cSE) are from the following two functions: CGE = (1 – χ) ρdec kTs CG CSE = ρmin kTs Cs where ρdec andρmin are normal decomposition rate and specific humus mineralization rate, χ is the humification ratio, and CG is the carbon in litter. The TREEDYN3 model has many different features from other models. First, it is the only model considering mass and energy balance of carbon and nitrogen flows as a constraint. Second, it follows the trend of model development; it’s a hybrid model. Third, it introduced diurnal and seasonal variation. Meanwhile, it is also a management tool for foresters who are considering thinning and harvest effects on forest yield (Bossel, 1996). The major shortcoming of the model is that it is only suitable for even age artificial forest stands, because during the simulation, all trees are of uniform size. When silviculture strategies are simulated, it does not represent the result as the reality. But it’s still a good tool for predict long-term effects of climate change, air pollution and managements. TRIPLEX is a hybrid, monthly-time step, stand model used for predicting forest growth and yield and ecosystem carbon and nitrogen dynamics. As noted above, it integrates three well-developed process-based models: 3-PG (Landsberg and Waring, 1997), CENTURY (Parton et al., 1993) and TREENYD3 (Bossel, 1996). It borrows the soil submodel from CENTURY, and growth and yield components from 3-PG and TREENYD3. It has four major submodels: forest production submodel, soil C and N dynamics submodel, forest growth and yield submodel and soil water balance submodel (Peng et al., 2002). The TRIPLEX model uses the approach from 3-PG to calculate gross primary productivity (GPP), GPP = k × Im × LAI × fa × ft × fw × fd  208  where GPP is a function of monthly received photosynthetically active radiation; PAR (Im), leaf area index (LAI), forest age (fa), monthly mean temperature (ft), soil drought (fw), percentage of frost days in a month (fd) and a conversion constant (k). It combines the idea that net primary production (NPP) is a fixed proportion of gross primary productivity (GPP), and NPP is affected by nutrient availability. NPP = CNPP fr GPP CNPP is a fixed fraction (0.47 ± 0.04) and fr is the modifier indicating available N. As a result, there is no respiration component in this model. The decomposition part adapts the approach of CENTURY, but it also adds some additional components. Ri = Ki Ci Md Td Ri’ = min ( Ri,  K i S N ( Bs Bt ) ( pB s − pBt − (1 − p) Bt Re )  )  where Ri and Ri’ are potential decomposition and actual decomposition of each carbon pool respectively; Ki, Ci, Md and Td are maximum decomposition rate, carbon stock in particular pool, soil moisture and temperature modifier respectively. In the restriction function, decomposition is gotten from the lower value between potential decomposition and restricted decomposition. In this function, SN is soil mineral N, Bs and Bt are C: N ratio of source and target C pools, p is the proportion of decomposed C which flows into other pools and Re is the fraction of soil organic N generated from C decomposition process which flows into the mineral N pool. The approach developed in TRIPLEX is new in that it combines existing models instead of building a new model to predict the climate change effects. The difficulty with this approach is the need to combine different time scales. However, comparing the simulation results with observed data suggests good model performance. As TREEDYN3 incorporates silviculture strategies, TRIPLEX can be used as a management tool.  C.1.8. Canadian models: BEPS, EASS, ECOSYS and CN-CLASS BEPS (Boreal Ecosystem Productivity Simulator; Liu et al., 1997; Chen et al., 1999) was developed at the Canadian Centre for Remote Sensing (CCRS) and the University of Toronto for short-term carbon cycle simulations. This model has been used with remotely sensed estimates of leaf area index (LAI) and land cover, and with Soil Landscapes of Canada (SLC), forest  209  inventory and gridded meteorological data to make regional and national estimates of NPP, NEP and net biome productivity (NBP) (Chen et al., 2003). CO2 fixation in BEPS is constrained by leaf stomatal conductance, calculated empirically from canopy temperature, humidity and global radiation (Humphreys et al., 2003). ECOSYS was developed at the University of Alberta as a detailed, comprehensive model of terrestrial ecosystems (Grant, 2001). This model provides a means to anticipate ecosystem behaviour under different environmental conditions (soils, climates and managements). EDOSYS has been used to estimate the impacts of climate, land use practices and soil management on primary productivity, soil and atmospheric quality and associated resource requirements (e.g. water, fertilizer) of diverse terrestrial ecosystems (Grant et al., 2001). Water deficits in ECOSY constrains CO2 fixation through non-stomatal effects. EASS (ecosystem–atmosphere simulation scheme) is a remote sensing-based ecosystem model, developed at the University of British Columbia (Chen et al., 2007). EASS has the following characteristics: (i) satellite data are used to describe the spatial and temporal information on vegetation, and in particular, the use of a foliage clumping index, in addition to leaf area index to characterize the effects of three-dimensional canopy structure on radiation, energy and carbon fluxes; (ii) energy and water exchanges and carbon assimilation in the soil–vegetation–atmosphere system are fully coupled and are simulated simultaneously; (iii) the energy and carbon assimilation fluxes are calculated with stratification of sunlit and shaded leaves to avoid shortcomings of the “big-leaf” assumption. CLASS (Verseghy, 2000) was developed by the Meteorological Service of Canada (MSC) for coupling with the Canadian General Circulation Model (CGCM) in regional climate–ecosystem interactions. CLASS has participated in the International Project for Intercomparison of Land–Surface Parameterization Schemes (PILPS). Versions of the CLASS biospheric component (C-CLASS) are being developed at McMaster University (C-CLASSm) (Arain et al., 2002, 2006)) and the University of Alberta (C-CLASSa) (Zhang et al., 2004). In C-CLASSa, soil water deficits effects constrained CO2. In CCLASSm, CO2 fixation was constrained directly by soil water content.  210  Table C. 1. Comparison of different ecosystem processes, climate input included and main features in several stand-level models. Scale Model PnET FOREST BGC  Model  Time  stand monthly daily stand and yearly  Climate input  Physiological Process  Temp. Moist. CO2  Photosynthesis GPP NPP  Respiration  Decomposition  Driving Function Foliage LAI [N]  Nutrient Cycling C  N  Y  Y  -  2  1  Y (proportion)  -  -  Y  Y  -  Y  Y  Y  1  2  Y  Y  Y  -  Y  Y  TREE-BGC  tree to stand  daily  Y  Y  -  1  2  Y  -  Y  -  Y  Y  BIOMASS  stand  Daily to monthly  Y  Y  Y  1  2  Y  -  Y  ?  Y  -  LINKAGE  tree to monthly stand  Y  Y  -  -  1  -  Y  ?  -  Y  Y  stand  Y  -  Y  1  2  Y (2 models)  Y  Y  Y  G’DAY  yearly  Y  Y  -  Y  Y  -  -  -  Y  Biomass  2  Y  Y  Affect radiation  Y  Y  Y  1  2  -  Y  Y  -  Y  Y  -  -  1  -  Y  -  Y  Y  Y  -  -  Y  -  Y  -  Y  Y  Y  3-PG  stand monthly  Y  Y  -  CENTURY  stand monthly  Y  Y  -  TREEDYN3  stand  Monthly and seasonal  Y  -  -  1  stand monthly  Y  Y  Y  FORECAST  stand  yearly  -  -  FORECAST Climate  stand  daily  Y  Y  TRIPLEX  Y  Canopy quantum efficiency coefficient  1  2  Potential production  211  Table C.1. Comparison of different ecosystem processes, climate input included and main features in several stand-level models. (Continued).  Model  Stomata level  Canopy Spatial Structures  Ecosystem level Soil  Plant  Forest  Management tool  GIS input  PnET  -  Y (several layers)  -  -  Y  -  -  FOREST - BGC  Y  -  Y  -  Y  -  Y  TREE – BGC  -  Light Shade  -  -  Y  -  -  BIOMASS  Y  LINKAGE  -  G’DAY  Y  3-PG  -  CENTURY  layers  Model Link  Reference  MAGIC Aber and Federer, 1992 (Cosby et al., 1985) -  Running and Coughlan, 1988 Korol et al., 1995  Branch of Korol et al., 1995 FOREST - BGC  -  -  Y  Y  -  -  -  Y  -  -  Y  -  Y  -  -  -  -  -  Y  -  Y  -  Landsberg and Waring, 1997  -  -  Y  Y  -  -  -  Parton et al., 1993  TREEDYN3  -  layers  -  -  Y  -  -  Bossel, 1996  TRIPLEX  -  -  Y  -  Y  Y  Y  CENTURY4.0, TREEDYN3 3-PG  FORECAST  -  layers  Y  Y  Y  Y  -  -  FORECAST Climate  -  layers  Y  Y  Y  Y  -  FORECAST  Consider shade effect Sunlit Shade  Y  212  -  McMurtrie et al., 1989, McMurtrie et al., 1990, McMurtrie and Landsberg, 1992, McMurtrie et al., 1992, McMurtrie and Wang, 1993  Modified JABOWA Pastor and Post, 1985 (Botkin, 1993) CENTURY Medlyn, 1996, Medlyn et al., 2000 BEWDY Comins and McMutrie, 1993  Peng et al., 2002 Kimmins et al., 1999 Seely et al., personal communication  C.1.9. References Aber, J.D. and C.A. Federer. 1992. A generalized, lumped-parameter model of photosynthesis, evapotranspiration and net primary production in temperate and boreal forest ecosystems. Oecologia. 92:463-474. Aber, J.D., S.V. Ollinger, and C.T. Driscoll. 1997. Modelling nitrogen saturation in forest ecosystems in response to land use and atmospheric deposition. Ecological Modelling. 101: 61-78. Arain, M.A., F. Yuan and T.A. Black. 2006. Soil–plant nitrogen cycling modulated carbon exchanges in a western temperate conifer forest in Canada. Agricultural and Forest Meteorology. 140: 171-192. Arain, M.A., T.A. Black, A.G. Barr, P.G. Jarvis, J.M. Massheder, D.L. Verseghy and Z. Nesic. 2002. Effects of seasonal and interannual climate variability on net ecosystem productivity of boreal deciduous and conifer forests. Canadian Journal of Forest Research. 32: 878-891. Bossel H. 1996. TREEDYN3 forest simulation model. Ecological Modelling. 90: 187-227. Botkin, D. B. 1993. Forest Dynamics: An Ecological Model. Oxford University Press, New York. Pp. 309. Chen, J.M., W. Ju, J. Cihlar, D. Price, J. Liu, W. Chen, J. Pan, T.A. Black and A. Barr. 2003. Spatial distribution of carbon sources and sinks in Canada’s forests based on remote sensing. Tellus B. 55: 622-642. Chen , B., J.M. Chen and W. Ju. 2007. Remote sensing-based ecosystem–atmosphere simulation scheme (EASS)—Model formulation and test with multiple-year data. Ecological Modelling. 209: 277-300. Comins, H.N. and R.E. McMutrie. 1993. Long-term response of nutrient limited forests to CO2 enrichment; equilibrium of plant-soil models. Ecological Applications. 3: 666-681. Cosby, B. J., G. M. Hornberger and J. N. Galloway. 1985. Modeling the effects of acid deposition: assessment of a lumped parameter model of soil and streamater chemistry. Water Res. Res. 21: 51-63. Coughlan, J.C. and S.W. Running. 1997. Regional ecosystem simulation: A general model for simulating snow accumulation and melt in mountainous terrain. Landscape Ecology. 12: 119-136. Grant, R.F., N.G. Juma, J.A. Robertson, R.C. Izaurralde and W.B. McGill. 2001. Long term changes in soil C under different fertilizer, manure and rotation: testing the mathematical model ecosys with data from the Breton Plots. Soil Science Society of America Journal. 65: 205-214. Humphreys, E.R., T.A. Black, G.J. Ethier, G.B. Drewitt, D.L.Spittlehouse, E.-M. Jork, Z. Nesic, Z. and N.J. Livingston. 2003. Annual and seasonal variability of sensible and latent heat fluxes above a coastal Douglas-fir forest, British Columbia, Canada. Agricultural and Forest Meteorology. 115: 109-125. Kimmins, J.P., D. Mailly and B. Seely. 1999. Modelling forest ecosystem net primary production: the hybrid simulation approach used in FORECAST. Ecological Modelling. 122: 195-224. Korol, R.L., S.W. Running, and K.S. Milner. 1995. Incorporating intertree competition into an ecosystem model. Canadian Journal of Forest Research. 25:413-424.  213  Landsberg, J.J. and R.H. Waring. 1997. A generalized model of forest productivity using simplified concepts of radiation-use efficiency, carbon balance and partitioning. Forest. Ecology and Management. 95:209-228. McMurtrie, R.E., J.J. Landsberg and S. Linder. 1989. Research priorities in field experiments of fast growing tree plantation: implication of a mathematical model. In: Pereira and J. J. Landsberg. Biomass Production by Fast-Growing Trees. Kliwer, Dordrecht, The Netherlands. p.181-207. McMurtrie, R.E., D.A. Rook and F.M. Kelliher. 1990. Modelling the Yield of Pinus radiata on a site limited by water and nitrogen. Forest Ecology and Management. 30: 381-413. McMurtrie, R.E. and J.J. Landsberg. 1992. Using a simulation model to evaluate the effects of water and nutrients on the growth and carbon partitioning of Pinus radiata. Forest Ecology and Management. 52: 243-260. McMurtrie, R.E., R. Leuning, W.A. Thompson and A.W. Wheeler. 1992. A model of canopy photosynthesis and water use incorporating a mechanistic formulation of leaf CO2 exchange. Forest Ecology and Management. 52: 261-278. McMurtrie, R.E. and Y.–P. Wang. 1993. Mathematical models of the photosynthetic response of tree stands to rising CO2 concentrations and temperature. Plant, Cell and Environment. 16: 1-13. Medlyn, B.E. 1996. The representation of photosynthetic productivity in an ecosystem model used to assess plant response to climate change. Ph.D. thesis, University of New South Wales, Sydney, Australia. Medlyn, B.E., R.E. McMutrie, R.C. Dewar and M.P. Jeffreys. 2000. Soil processes dominate the long-term response of forest net primary productivity to increased temperature and atmospheric CO2 concentration. Canadian Journal of Forest Research. 30:873-888. Parton, W.J., J.M.O. Scurlock, D.S. Ojima, T.G. Gilmanov, R.J. Scholes, D.S. Schimel, T. Kirchner, J-C. Menaut, T. Seastedt, E.G. Moya, A. Kamnalrut and J.I. Kinyamario. 1993. Observations and modeling of biomass and soil organic matter dynamics for the grassland biome worldwide. Global Biogeochem. Cycles 7: 785-809. Pastor, J. and Post, W.M. 1985. Development of a Linked Forest Productivity-Soil Process Model. U.S. Dept. of Energy, ORNL/TM-9519. Peng, C., J. Liu, Q. Dang, M.J. Apps and H. Jiang. 2002. TRIPLEX: a generic hybrid model for prediction forest growth and carbon and nitrogen dynamics. Ecological Modelling. 153:109-130. Running, S.W. and J.C. Coughlan. 1988. A General model of Forest ecosystem process for regional applications. I. Hydrologic balance, canopy gas exchange and primary production processes. Ecological Modelling. 42: 125-154. Verseghy, D.L. 2000. The Canadian Land Surface Scheme (CLASS): its history and future. Atmosphere-Ocean 38: 1-13. Yuan, F., M.A. Arain, T.A. Black, K. Morgenstern. 2007. Energy and water exchanges modulated by soil–plant nitrogen cycling in a temperate Pacific Northwest conifer forest. Ecological Modelling. 201: 331-347. Yuan, F., M.A. Arain, A.G. Barr, T.A. Black, C.P.-A. Bourque, C. Coursolle, H.A. Margolis, J.H. McCaughey and S.C. Wofsy. 2008. Modeling analysis of primary controls on net ecosystem productivity of seven boreal and temperate coniferous forests across a continental transect. Global Change Biology. 14: 1765-1784.  214  Zhang, Y., R.F. Grant, L.B. Flanagan, S.S. Wang and D.L. Verseghy. 2004. Recent developments and testing of a carbon-coupled Canadian land surface scheme in a water-stressed northern temperate grassland. Ecological Modelling. 181: 591-614.  215  C.2. Tree Productivity -Climate model  Figure C. 1. Tree productivity-climate model structure written in STELLA.  216  C.3. Additional graphical output for the test of the tree productivity-climate model C.3.1. Graphical comparisons of seasonal limiting factors changing for different combinations of temperature and precipitation growth scenarios.  Figure C. 2. Seasonal limiting factors changing for different combinations of temperature and precipitation growth scenario at high elevation (climate for ESSF zone) for three consecutive years. The thin solid line is the temperature multiplier seasonal change and the dot line is precipitation multiplier seasonal change. The thick solid line is potential NPP. T1 to T3 represent three different temperature optimum growth scenarios from low to high temperature. P1 to P3 represent three different precipitation optimum growth scenarios from dry to wet condition.  217  Figure C. 3. Seasonal limiting factors changing for different combinations of temperature and precipitation growth scenario at low elevation (climate for IDF zone) for three consecutive years. The thin solid line is the temperature multiplier seasonal change and the dot line is precipitation multiplier seasonal change. The thick solid line is potential NPP. T1 to T3 represent three different temperature optimum growth scenarios from low to high temperature. P1 to P3 represent three different precipitation optimum growth scenarios from dry to wet condition.  218  C.3.2. Graphical comparisons of NPP predictions and tree-ring chronologies.  Figure C. 4. Simulation results of Jun and July NPP vs. tree ring index data. The thick lines are lodgepole pine chronology at ESSF zone. Thin lines are simulations results for the combination of temperature and precipitation scenarios applied for low elevation site (upper panel) and low elevation site (lower panel).  Figure C. 5. Simulation results of Jun and July NPP vs. tree ring index data. The thick lines are lodgepole pine chronology at MS zone. Thin lines are simulations results for the combination of temperature and precipitation scenarios applied for low elevation site (upper panel) and low elevation site (lower panel).  219  Figure C. 6. Simulation results of Jun and July NPP vs. tree ring index data. The thick lines are lodgepole pine chronology at IDF zone. Thin lines are simulations results for the combination of temperature and precipitation scenarios applied for low elevation site (upper panel) and low elevation site (lower panel).  Figure C. 7. Simulation results of Jun and July NPP vs. tree ring index data. The thick lines are hybrid spruce chronology at ESSF zone. Thin lines are simulations results for the combination of temperature and precipitation scenarios applied for low elevation site (upper panel) and low elevation site (lower panel).  220  Figure C. 8. Simulation results of Jun and July NPP vs. tree ring index data. The thick lines are hybrid spruce chronology at MS zone. Thin lines are simulations results for the combination of temperature and precipitation scenarios applied for low elevation site (upper panel) and low elevation site (lower panel).  221  C.3.3. Results of regressions of NPP predictions vs. tree-ring chronologies Table C. 2. Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the ESSF zone. June & July, May to September and annual represent the sum of NPP for different periods calculated from STELLA. Bigger font in the slope column represents positive relationships. None of the relationships was statistically significant (P<0.05).  Climate data  Scenario T1P1 T1P2 T1P3 T2P1 High elevation T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 Low elevation T2P2 T2P3 T3P1 T3P2 T3P3  ESSF Pl Jun & Jul May to Sep Slope r2 Slope r2  Annual Slope  r2  -0.026  0.000  -0.101  0.005  -0.032  0.000  -0.097  0.001  -0.048  0.000  0.000  -0.111  0.000  0.050  0.000  0.031 0.116  0.026  0.000  -0.127  0.009  -0.056  0.001  -0.052  0.000  -0.065  0.001  -0.004  0.000  -0.148  0.001  0.014  0.000  0.072  0.000  0.098  0.003  -0.089  0.006  -0.029  0.000  0.001  0.000  0.000  -0.083  0.002  -0.020  0.000  -0.145  0.001  0.038  0.000  0.001  -0.067  0.001  -0.045  0.000  -0.063  0.000  -0.023  0.000  -0.057  0.000  -0.012  0.000  0.094 0.010 0.038 0.015  -0.094  0.002  -0.078  0.002  -0.009  0.000  -0.083  0.001  -0.025  0.000  0.000  -0.057  0.000  -0.010  0.000  0.034 0.050  -0.047  0.000  0.089  0.002  -0.017  0.000  -0.110  0.001  -0.041  0.000  0.018  0.000  -0.053  0.000  0.007  0.000  0.070  0.000  222  0.000 0.000 0.000  0.000  Table C. 3 Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the MS zone. June & July, May to September and annual represent the sum of NPP for different periods calculated from STELLA. Bigger font in the slope column represents positive relationships. None of the relationships was statistically significant (P<0.05).  Climate data  Scenario T1P1 T1P2 T1P3 T2P1 High elevation T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 Low elevation T2P2 T2P3 T3P1 T3P2 T3P3  MS Pl Jun & Jul May to Sep 2 Slope r Slope r2 0.016 -0.069 0.002 0.225 0.017 0.000 0.360 0.011 0.023 0.006 0.949 0.360 0.016 -0.093 0.005 0.230 0.018 0.000 0.347 0.008 0.021 0.003 0.848 0.255 0.018 -0.065 0.003 0.239 0.017 -0.018 0.000 0.315 0.019 0.002 0.735 0.193 0.024 0.001 0.401 0.061 0.028 0.005 0.654 0.209 0.024 0.011 1.290 0.582 0.019 0.000 0.343 0.029 0.026 0.005 0.618 0.200 0.024 0.010 1.290 0.577 0.020 0.000 0.327 0.024 0.023 0.003 0.562 0.157 0.024 0.011 1.292 0.605  223  Annual Slope 0.041 0.142 0.408 0.010 0.107 0.327 0.029 0.078 0.284 0.171 0.258 0.365 0.157 0.277 0.414 0.144 0.245 0.466  r2 0.000 0.003 0.008 0.000 0.002 0.005 0.000 0.001 0.004 0.005 0.008 0.005 0.005 0.009 0.006 0.005 0.007 0.008  Table C. 4. Linear regression results between simulation outputs vs. tree ring chronologies for Lodgepole pine in the IDF zone. June & July, May to September and annual represent the sum of NPP for different periods calculated from STELLA. Bigger font in the slope column represents positive relationships. None of the relationships was statistically significant (P<0.05).  Climate data  Scenario T1P1 T1P2 T1P3 T2P1 High elevation T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 Low elevation T2P2 T2P3 T3P1 T3P2 T3P3  IDF Pl Jun & Jul May to Sep 2 Slope r Slope r2 0.013 -0.128 0.012 0.163 0.016 -0.047 0.001 0.284 0.009 0.001 0.469 0.129 0.008 -0.150 0.019 0.128 0.011 -0.123 0.005 0.216 0.010 0.001 0.477 0.087 0.007 -0.129 0.017 0.114 0.004 -0.183 0.014 0.128 0.014 0.000 0.502 0.066 0.018 0.000 0.277 0.010 0.011 0.001 0.323 0.080 0.004 0.002 0.395 0.199 0.016 -0.018 0.000 0.252 0.010 0.001 0.312 0.079 0.004 0.002 0.395 0.205 0.014 -0.066 0.002 0.219 0.011 0.001 0.304 0.054 0.004 0.003 0.399 0.227  224  Annual Slope  r2  -0.158  0.012  -0.075  0.001  0.036  0.000  -0.173  0.014  -0.161  0.007  0.003  0.000  -0.145  0.011  -0.210  0.014  -0.009  0.000  -0.004  0.000  0.038 0.056  0.000  -0.027  0.000  0.039 0.096  0.000  -0.062  0.001  0.025 0.126  0.000  0.000  0.001  0.001  Table C. 5. Linear regression results between simulation outputs vs. tree ring chronologies for hybrid spruce in the ESSF zone. June & July, May to September and annual represent the sum of NPP for different periods calculated from STELLA. Bigger font in the slope column represents positive relationships. None of the relationships was statistically significant (P<0.05).  Climate data  Scenario T1P1 T1P2 T1P3 T2P1 High elevation T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 Low elevation T2P2 T2P3 T3P1 T3P2 T3P3  ESSF Sx Jun & Jul May to Sep 2 Slope r Slope r2 0.001 -0.006 0.000 0.080  Annual Slope 0.243 0.208 0.080 0.253 0.255 0.194 0.285 0.293 0.272 0.020  r2 0.011  -0.110  0.001  -0.037  0.000  -0.553  0.005  -0.116  0.000  0.206 0.004  0.008  0.001  0.000  0.046 0.003  -0.381  0.003  -0.027  0.000  0.284 0.163  0.017  0.004  -0.240  0.001  0.096 0.056 0.072  -0.189  0.003  -0.136  0.002  -0.442  0.008  -0.173  0.002  -0.030  0.000  -1.282  0.015  -0.560  0.006  -0.195  0.001  0.134  0.002  -0.105  0.002  0.001  -0.389  0.007  -0.153  0.002  0.075 0.007  -1.282  0.015  -0.558  0.006  -0.255  0.002  -0.056  0.000  -0.085  0.001  0.002  0.308  0.004  -0.121  0.001  0.108 0.052  -1.280  0.015  -0.558  0.006  -0.281  0.002  0.003  225  0.000  0.001 0.000  0.004 0.000 0.013 0.007 0.001 0.018 0.011 0.002 0.000  0.000  0.000  Table C. 6. Linear regression results between simulation outputs vs. tree ring chronologies for hybrid spruce in the MS zone. June & July, May to September and annual represent the sum of NPP for different periods calculated from STELLA. Bigger font in the slope column represents positive relationships. None of the relationships was statistically significant (P<0.05).  Climate data  Scenario T1P1 T1P2 T1P3 T2P1 High elevation T2P2 T2P3 T3P1 T3P2 T3P3 T1P1 T1P2 T1P3 T2P1 Low elevation T2P2 T2P3 T3P1 T3P2 T3P3  MS Sx Jun & Jul May to Sep 2 Slope r Slope r2  Annual Slope r2 0.016 0.000  0.107  0.002  -0.128  0.005  -0.153  0.002  -0.371  0.016  -0.238  0.005  -0.912  0.014  -0.751  0.016  -0.658  0.013  0.137  0.004  -0.074  0.002  0.052  0.001  -0.092  0.001  -0.286  0.012  -0.122  0.002  -0.719  0.010  -0.700  0.015  -0.554  0.009  0.149  0.005  -0.015  0.000  0.102  0.002  -0.023  0.000  -0.216  0.008  -0.049  0.000  -0.571  0.007  -0.629  0.014  -0.488  0.008  -0.178  0.003  -0.320  0.014  -0.233  0.006  -0.558  0.013  -0.406  0.012  -0.323  0.008  -1.784  0.030  -0.969  0.019  -0.781  0.015  -0.132  0.002  -0.294  0.013  -0.196  0.005  -0.509  0.011  -0.397  0.012  -0.306  0.007  -1.784  0.030  -0.981  0.019  -0.812  0.016  -0.070  0.001  -0.227  0.009  -0.119  0.002  -0.416  0.008  -0.379  0.012  -0.266  0.006  -1.778  0.029  -0.992  0.020  -0.824  0.017  226  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0067154/manifest

Comment

Related Items