UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mapping the distribution of conifer tree species in response to environmental changes across western… Mathys, Amanda Sarah 2017

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

Item Metadata

Download

Media
24-ubc_2017_november_mathys_amanda.pdf [ 7.04MB ]
Metadata
JSON: 24-1.0357125.json
JSON-LD: 24-1.0357125-ld.json
RDF/XML (Pretty): 24-1.0357125-rdf.xml
RDF/JSON: 24-1.0357125-rdf.json
Turtle: 24-1.0357125-turtle.txt
N-Triples: 24-1.0357125-rdf-ntriples.txt
Original Record: 24-1.0357125-source.json
Full Text
24-1.0357125-fulltext.txt
Citation
24-1.0357125.ris

Full Text

MAPPING THE DISTRIBUTION OF CONIFER TREE SPECIES IN RESPONSE TO ENVIRONMENTAL CHANGES ACROSS WESTERN NORTH AMERICA USING A PHYSIOLOGICALLY BASED APPROACH  by Amanda Sarah Mathys  B.Sc. (Honours), Queen’s University, 2008 M.Sc., The University of British Columbia, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  October 2017  © Amanda Sarah Mathys, 2017   ii Abstract Over the past decade, changes in climate have been sufficient to affect both the composition and function of forest ecosystems. The extent that projected climate change will continue to impact tree species vulnerabilities remains unclear and has been mainly assessed based on simple relationships between the distribution of mature trees and climate variables. The objective of this thesis was to assess the effects of regional climate and soil variations on the current and future distribution of 20 major conifer tree species across western North America and determine the impacts of changing environmental variables on tree species vulnerabilities. The spatial variation in properties of soil water availability and soil fertility was combined in the process-based model 3-PG to provide detailed projections of species shifts in response to changes in environmental conditions. The relative importance of limitations imposed on photosynthesis by suboptimal temperatures, frost, solar radiation, soil water and vapor pressure deficits was ranked in a decision tree analysis based on tree species occurrences across the region. The baseline distributions of the tree species were predicted with an average accuracy of 84% (κ = 0.79), based on their recorded presence and absence on 43,404 field survey plots. Inclusion of soil properties was crucial to improving the overall accuracy of the species distribution models and 75% of the species directly responded to changes in the soil water input. At the ecoregion level, this thesis identified the highest vulnerability of the 20 tree species analyzed to occur within the North American Deserts, particularly in the Thompson-Okanagan Plateau of British Columbia (BC). Comparison of areas suitable for tree species range expansion with a large empirical dataset on tree seedling occurrences in BC agreed on average 79%, serving as indicators of early species responses to climate shifts in the province. Outcomes of this thesis demonstrate species-  iii specific responses to current and future climatic variations and can contribute to informing forest management for climate change adaptation.   iv Lay Summary Global changes in climate are affecting forest ecosystems and the distribution of tree species. Across western North America, increased disturbances caused by wildfires, insect attack and diseases are already posing stress on tree species that may lead to changes in their distribution. This thesis aimed to map how tree species distributions are shifting with climate change and how vulnerable they are to extreme climate events.  Tree species growth is not only affected by climate variations but also by soil properties. I used soil maps to model the current and future distribution of 20 common tree species of western North America. I found that tree species are sensitive to the amount of available soil water indicating its importance for improved mapping of tree species distributions. I identified ecological regions with a high species vulnerability to focus forest conservation efforts and explored emerging patterns of species shifts from tree seedling distributions.     v Preface The research project and questions presented in this thesis were developed by myself in consultation with my supervisory committee. I completed all of the data analysis, scientific writing and revisions of the thesis chapters and manuscripts and prepared them for final publication in scientific journals. The co-authors provided valuable input in the research project development, data analysis as well as editorial comments. The following thesis chapters appear in peer-reviewed publications:  A version of Chapter 3 has been published in:  Mathys, A., Coops, N.C., Waring, R.H., 2014. Soil water availability effects on the distribution of 20 tree species in western North America. Forest Ecology and Management 313, 144–152.  A version of Chapter 4 has been published in:  Mathys, A.S., Coops, N.C., Waring, R.H., 2017. An ecoregion assessment of projected tree species vulnerabilities in western North America through the 21st century. Global Change Biology 23, 920–932.  A version of Chapter 5 has been submitted for publication:   Mathys, A.S., Coops, N.C., Simard, S.W., Waring, R.H., Aitken, S.N., 2017. Predicting tree species responses to climate shifts: Integrating an empirical regeneration database with physiological modeling.    vi In addition, a portion of this thesis research has been published as part of a NASA funded project in collaboration with Dr. Richard Waring at Oregon State University. I undertook the majority of the data analysis and synthesized results of the publication:  Waring, R.H., Coops, N.C., Mathys, A., Hilker, T., Latta, G., 2014. Process-based modeling to assess the effects of recent climatic variation on site productivity and forest function across western North America. Forests, 5, 518–534.                    vii Table of Contents  Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ........................................................................................................................ vii List of Tables ..................................................................................................................................x List of Figures ............................................................................................................................... xi List of Symbols and Abbreviations .......................................................................................... xiii Acknowledgements ......................................................................................................................xv 1. Introduction ............................................................................................................................... 1 1.1 Background and motivation ........................................................................................... 1 1.2 Species distribution models ........................................................................................... 4 1.3 The 3-PG model ............................................................................................................. 9 1.4 Research needs ............................................................................................................. 13 1.5 Research overview and objectives ............................................................................... 15 2. Study area and data sources .................................................................................................. 18 2.1 The Pacific Northwest.................................................................................................. 18 2.2 Data sources ................................................................................................................. 21 2.2.1  Climate and soil data ............................................................................................... 21 2.2.2  Tree species plots .................................................................................................... 22 3. Soil water availability effects on the distribution of 20 tree species in western North America ........................................................................................................................................ 24   viii 3.1 Introduction .................................................................................................................. 24 3.2 Methods........................................................................................................................ 27 3.2.1  3-PG model parameterization .................................................................................. 27 3.2.2  Predicting species distributions with decision tree models..................................... 28 3.2.3  Sensitivity analysis.................................................................................................. 30 3.3 Results .......................................................................................................................... 30 3.4 Discussion .................................................................................................................... 39 4. An ecoregion assessment of projected tree species vulnerabilities in western North America through the 21st century ............................................................................................. 43 4.1 Introduction .................................................................................................................. 43 4.2 Methods........................................................................................................................ 46 4.2.1  Ecological regions of western North America ........................................................ 46 4.2.2  Current and future climate ...................................................................................... 49 4.2.3  Modeling approach ................................................................................................. 49 4.3 Results .......................................................................................................................... 51 4.4 Discussion .................................................................................................................... 59 5. Changes in seedling and tree distributions indicate emerging shifts in forest composition....................................................................................................................................................... 67 5.1 Introduction .................................................................................................................. 67 5.2 Methods........................................................................................................................ 70 5.2.1  Study area ................................................................................................................ 70 5.2.2  Sampling species distributions ................................................................................ 72 5.2.3  Predicting environmental constraints on species distributions ............................... 73   ix 5.3 Results .......................................................................................................................... 74 5.4 Discussion .................................................................................................................... 81 6. Conclusion ............................................................................................................................... 87 6.1 Research innovations ................................................................................................... 87 6.2 Limitations of research ................................................................................................ 92 6.2.1  Limitations in 3-PG modeling framework .............................................................. 92 6.2.2  Non-climatic factors controlling species shifts ....................................................... 94 6.3 Research applications in forest management ............................................................... 95 6.4 Directions for future research ...................................................................................... 97 References ...................................................................................................................................100 Appendices ..................................................................................................................................135 Appendix A Tree species distribution maps for 20 major conifer species across western North America .................................................................................................................... 135 A.1  Climatically suitable species distributions under baseline conditions (1950-1975) 135 A.2  Tree species vulnerability maps in 2000-2009 compared to 1950-1975 ................ 141    x List of Tables Table 3.1 The accuracy of decision tree models and Kappa (κ) statistic derived for each of 20 tree species based on mean monthly climate conditions during the period 1950-75. The 95% confidence intervals are shown in brackets. The species distribution maps produced from these models are presented in Appendix A.1. ........................................................................................ 33 Table 3.2 Current predicted ranges and percent change in species distributions when increasing or decreasing ASWC by 50% compared with baseline values with variable ASWC maps (Coops et al., 2012) and constant FR (0.5). ............................................................................................... 37 Table 4.1 Level III Ecoregions within study region of western Canada and United States. ........ 48 Table 4.2 The 20 tree species and number of presence and absence plots utilized in this thesis. 51 Table 4.3 Percent and absolute changes in the potential distribution of dominant species by ecoregion from 2075-2100 compared to their current range. ....................................................... 55 Table 5.1 Decision tree model accuracies for the trees during the period 1950-1975 and for the seedlings during 2000-2009. Brackets represent the 95% confidence intervals. .......................... 74 Table 5.2 Projected species expansion and stress and agreement with area deemed suitable or unsuitable for regeneration growth (2000-2009). ......................................................................... 81    xi List of Figures Figure 1.1 Diagram of the components used in the 3-PG forest growth model. Net primary production (NPP) and gross primary production (GPP) are driven by light absorption by the canopy (APAR). Constraints on photosynthesis are imposed by vapor pressure deficits (VPD), air temperature (Temp), available soil water and frost. NPP is partitioned into both above-ground (NPPA) and below-ground (NPPB) components based on soil fertility. Modified from Waring et al. (2010). ...................................................................................................................................... 12 Figure 1.2 Conceptual flow diagram of thesis chapters and outcomes of research questions. ..... 17 Figure 2.1 The study area in western United States and Canada consisting of 11 States and two provinces. ...................................................................................................................................... 20 Figure 3.1 Predicted differences in LAI (m2 m-2) when available soil water content varied across the region from 0-300 mm, depending on local climate and soils, compared to when it was held constant at 200 mm. In both simulations, soil fertility was held constant at a median level of 0.5. The gray background depicts the digital elevation model of non-forested areas in the study region. ........................................................................................................................................... 32 Figure 3.2 Occurrence of soil water modifiers (annual, spring, summer, fall, winter) in decision tree models for each of 20 tree species native to the Pacific Northwest Region. ......................... 34 Figure 3.3 Importance of each soil water modifier in determining the mean % variance of the tree species. .......................................................................................................................................... 35 Figure 3.4 Changes in the distributions of two species with varying available soil water content (ASWC) during the period 1976-2009. The expansion of a) western white pine range for ASWC at 50% (red) and baseline condition (green) and b) Douglas-fir range at baseline conditions (red) and 150% ASWC (green) and the current range of c) western white pine and d) Douglas-fir at baseline ASWC values. ................................................................................................................. 38 Figure 4.1 Ecological regions (level III) in western Canada and United States as defined by the U.S. Environmental Protection Agency (EPA) and Environment Canada (Ecological Stratification Working Group (ESWG), 1996). The major ecoregions (level I) are outlined in bold. .............................................................................................................................................. 47 Figure 4.2 Current and future potential suitable habitat of lodgepole pine, Douglas-fir and whitebark pine from (a) 1976-2009 and (b) 2075-2100 under the RCP2.6 emission scenario. ... 52 Figure 4.3 Projected tree species abundances averaged from 2075 to 2100 under different emission scenarios for Douglas-fir, lodgepole pine, whitebark pine and sitka spruce. ................ 53 Figure 4.4 Decision tree models of Douglas-fir, lodgepole pine and whitebark pine ranking the relative importance of four environmental constraints (0 = no photosynthesis and 1 = optimum   xii conditions for photosynthesis) that determine the presence and absence of tree species (ranked between 0 and 1). Temp = temperature modifier. ......................................................................... 54 Figure 4.5 Areas deemed unsuitable (purple) or suitable (green) for (a) Douglas-fir, (b) lodgepole pine, (c) Sitka spruce and (d) whitebark pine under three future emission scenarios in 2075-2100 compared to their current distributions (grey) in 1976-2009. ....................................................... 57 Figure 4.6 Predicted number of species stressed throughout the study regions for the three emission scenarios (a) RCP2.6, (b) RCP4.5 and (c) RCP8.5 in 2075-2100. ................................ 59 Figure 5.1 Study area in British Columbia and location of field survey plots of the mature trees (purple) and seedlings (green). ..................................................................................................... 72 Figure 5.2 Mean seasonal variation in climate modifiers of lodgepole pine, Douglas-fir, subalpine fir and western larch during baseline conditions in 1950-75 (0 = no growth and 1 = optimum conditions for growth). .................................................................................................. 76 Figure 5.3 Shifts in mean (dots), median (line) and interquartile range (box) between the two periods (1950-75 and 2000-09) for four seasonally-defined climatic growth modifiers averaged across all species. .......................................................................................................................... 77 Figure 5.4 Changes in seasonal climate growth modifiers for Douglas-fir, lodgepole pine, subalpine fir and western larch between 1950-1975 and 2000-2009. The error bars represent the standard error. Negative values indicate that the modifier is becoming less limiting. ................. 78 Figure 5.5 Potential for range expansion (green) and vulnerability (purple) of a) lodgepole pine and b) western larch based on 2000-2009 climatic conditions compared to conditions during 1950-1975 and suitable habitat for seedlings of c) lodgepole pine and d) western larch. ............ 80    xiii List of Symbols and Abbreviations Symbol / Abbreviation Definition APAR Absorbed photosynthetically active radiation (MJ m-2) ASWC Available soil water holding capacity (mm) BC British Columbia BEC Biogeoclimatic Ecological Classification C Carbon CanESM2 Canadian Earth System Model CEC Commission for Environmental Cooperation CO2 Carbon dioxide (mol mol-1) CCCma Canadian Climate Centre’s Modeling and Analysis DBH Diameter at breast height (cm) DEM Digital elevation model (m) DTREG Decision Tree Regression software EPA Environmental Protection Agency ESWG Ecological Stratification Working Group GCM General Circulation Model LAI Leaf area index (m2 m-2) LiDAR Light Detection and Ranging FIA Forest Inventory and Analysis FR Soil fertility GPP Gross primary production (mol m-2 or Mg ha-1 time-1) IPCC Intergovernmental Panel on Climate Change   xiv  Symbol / Abbreviation Definition MODIS Moderate Resolution Imaging Spectroradiometer NARR North American Regional Reanalysis NPP Net primary production (mol m-2 or Mg ha-1 time-1) PNW Pacific Northwest Region Ra  Autotrophic respiration (mol m-2 or Mg ha-1 time-1) RCP Representative Concentration Pathways RESULTS Reporting Silviculture Updates and Land Status Tracking System SRTM Shuttle Radar Topography Mission STATSGO State Soil Geographic database US United States VPD Vapor pressure deficit (kPa) 3-PG Physiological Principles Predicting Growth model α Canopy quantum efficiency (mol C mol-1 photons) κ Cohen's kappa coefficient   xv Acknowledgements I am greatly thankful to my supervisor Dr. Nicholas Coops for his guidance, knowledge and generosity throughout my PhD degree. I would like to thank my supervisory committee, Dr. Suzanne Simard and Dr. Sally Aitken for sharing their expertise in forest ecology and for their valuable input that improved the quality of this thesis. Many thanks to Dr. Richard (Dick) Waring at the Oregon State University for his continuous support and teachings in tree physiology and for his guidance in my research and career path over the years. I am also thankful to my former supervisor Dr. Andy Black, who taught me a great deal in research that prepared me well for a PhD degree.  Dr. Todd Schroeder (USFS) and Dr. Andreas Hamann (University of Alberta) provided tree species plot data in the United States and British Columbia, and the Alberta Environment and Sustainable Resource Development provided species occurrence data in Alberta. I am thankful to Dan Turner, Matt LeRoy and Caroline Wood from the BC Ministry of Forests, Lands and Natural Resource Operations for their support in accessing regeneration plot information from the RESULTS database. I appreciate the support of Dr. Robbie Hember at the University of British Columbia for producing estimates of solar radiation for the study and would like to remember Dr. Thomas Hilker and thank him for his research contributions and mentorship. I am also grateful to the IRSS community for their support and friendships over the years.  I appreciate funding provided by the Alexander Graham Bell Canada Graduate Scholarship from the Natural Sciences and Engineering Research Council of Canada (NSERC), an NSERC TerreWEB Scholarship and the Faculty of Forestry Strategic Recruitment Scholarship to Mathys.   xvi The research project was funded by the National Aeronautics and Space Administration (NASA Grant NNX11A029G) from the Biodiversity and Ecological Forecasting program to Waring and an NSERC Discovery Grant to Coops.   1 1. Introduction 1.1 Background and motivation Forests are complex adaptive systems composed of many parts and functions that interact at different spatial and temporal scales through non-linear relationships, exchange carbon (C), water and energy through biogeochemical cycles and are sensitive to disturbances and perturbations (Levin, 1998; Puettmann et al., 2009). The ecological resilience of forests depends on their ability to endure disturbances while providing a similar set of ecosystem services (Messier and Puettmann, 2011). Historic records of species distribution patterns and forest composition indicate that the dynamic nature of forest ecosystems is in part driven by climate variations on a regional scale (e.g. Salisbury, 1926; Ellenberg, 1988; Salajanu et al., 2010).  Species migrations in response to changing climate conditions have long occurred throughout the Quaternary, when the glaciers retreated providing greater habitat for tree species (Davis and Shaw, 2001). Paleoecological studies document species migrations through inferred mechanisms such as seed dispersal where seedlings established on sites that became climatically suitable, or through population expansion from glacial refugees (McLachlan and Clark, 2004; Davis and Shaw, 2001). Species migrations are often linked with historical global climate variations of glacial and interglacial cycles as shown in pollen and macrofossil records (Aitken et al., 2008; Davis and Shaw, 2001). Following the last glaciation, tree species began to migrate to higher latitudes and elevations at rates of up to 200 m year-1 (Davis and Shaw, 2001), although genetic evidence suggests that these rates were likely overestimated (McLachlan and Clark, 2004; Aitken et al., 2008). Responses varied by species with differing climate tolerances leading to changes in forest composition as certain species increased and others declined in abundance   2 (Davis and Shaw, 2001; Webb et al., 1993). For example in southwestern United States, bristlecone pine (Pinus longaeva) contracted its range in response to climate warming while both ponderosa pine (Pinus ponderosa) and western juniper (Juniperus occidentalis) expanded as conditions became more suitable for these species (Webb et al., 1993). In eastern North America, eastern white pine (Pinus strobus), a species that was widely distributed in the early Holocene, has decreased in abundance since the last 4000 years, as seedling establishment became less suitable with changing precipitation patterns and fire regimes (DeHayes et al., 2000).   The widely studied changes in species distributions around the world indicate that species range shifts will likely continue to occur with climate change in the future (DeHayes et al., 2000; Melillo et al., 1995). Despite the high migration rates estimated from pollen records, the rate of climate change as projected in climate models far exceeds climate shifts experienced in the past (Davis and Shaw, 2001). Tree species would need to migrate at a much faster pace (100-250 km per decade) than has been documented during the postglacial period in the Holocene to keep up with climate change (Aitken et al., 2008; Ordonez and Williams, 2013). With no records of a similar phenomenon in the past, there is an urgency to track tree species responses to rapid climate change.   Global changes in climate are already affecting forest ecosystems and their vulnerability to disturbance events (Raffa et al., 2008; Westerling et al., 2006). In its latest report, the Intergovernmental Panel on Climate Change (IPCC) states that since the 1950s, unprecedented changes have occurred with climate warming (IPCC, 2014). Global atmospheric carbon dioxide (CO2) concentrations now exceed 400 ppm since 2015 compared to pre-industrial levels of 260   3 ppm (WMO, 2016; Canadell et al. 2007). On average, air temperatures have increased by 0.85 °C globally from 1880 to 2012 (IPCC, 2014). In western North America, the climate has become warmer and drier over the past 30 years (Waring et al., 2014) with a decline in snow cover during the wintertime (Mote et al., 2005). These changes have led to recent alterations in established tree species distribution and growth patterns (IPCC, 2014; Breshears et al., 2005; Allen et al., 2010; Worrall et al., 2013). Climate change is predicted to continue in the future with elevated greenhouse gas emissions to the atmosphere and has been attributed to anthropogenic interferences (IPCC, 2014). Changes to both temperature and precipitation are expected to accelerate in western North America, especially in northern and interior regions (Mote et al., 2003), at rates that exceed historic climate change at the end of the last glacial period (Davis and Shaw, 2001). With such projections, it becomes important to determine how climate shifts are affecting current tree species distributions and use this knowledge to predict how they will respond to future climate change scenarios.  Western North America is a suitable geographic region to examine shifts in tree species distributions as it contains a high diversity of ecoregions that have been exposed to increased forest disturbances in recent years. Disturbances caused by fires, insects and diseases (Ramsfield et al., 2016; Woods et al., 2010; Anderegg et al., 2015; Westerling et al., 2006) are impacting forest composition and affecting the forests’ role in mitigating climate change by altering the forest carbon (C) balance (Amiro et al., 2010; Brown et al., 2012, Mathys et al., 2013). Increased bark beetle activity such as the mountain pine beetle Dendroctonus ponderosae (Hopkins) outbreak is a major disturbance that has been linked with climate change (Raffa et al., 2008; Carroll et al., 2006; Safranyik and Wilson, 2006; Heineman et al., 2010). Water stress and   4 wildfire occurrences across western North America have also been associated with warmer summers and changing precipitation patterns (Westerling et al. 2006; Wheaton, 2005; Allen et al., 2010). Increased drought in the southwest of the United States has been associated with a rise in tree mortality. (Ganey and Vojta, 2011) and has been described as a major emerging climate effect of this century (Cayan et al. 2010; Cook et al. 2015). Climate variations also impact fire frequencies and severities as drier conditions make forests more susceptible to fires, and wetter conditions can lead to a buildup of fuels in arid, lower elevation forests (Graham et al., 2004). In North America, changes in climate are increasing the length of the growing season (Zhou et al., 2001). A decrease in snow cover and earlier snowmelt are reducing the soil water supply during the summer and impacting tree species that are adapted to these conditions, particularly at high elevations (IPCC, 2014). The impacts of climate change will vary among species across the ecoregions of western North America and a better understanding is needed on how forests are beginning to respond to these altered environmental conditions. In the next section, I provide an overview of species distribution models that are employed in combination with regional climate patterns and local soil and vegetation types. 1.2 Species distribution models The geographic distribution of tree species has been of interest to forest ecologists and biogeographers for many centuries (Guisan and Thuiller, 2005; Elith and Leathwick, 2009). Sampling species occurrences over large spatial and temporal scales can pose significant challenges, as large distances need to be covered on a continuous basis with terrain that can be difficult to access. Instead, species distribution models along with remote sensing techniques can be employed to predict suitable habitat for species based on their recorded occurrences combined   5 with knowledge on their preferred environmental setting (Elith and Leathwick, 2009). Species distribution models can provide ecological knowledge for conservation purposes and to manage forest resources. They often function by quantifying relationships between species and environmental variables (Guisan and Thuiller, 2005). Climate is the main driver used in species distribution models and only rarely are local site conditions, such as soils, employed despite their recognized importance in influencing tree growth and distributions (Syphard and Franklin, 2009; Iverson and Prasad, 1998).   Species distribution models can be divided into 1) correlative and 2) process-based models. The majority of studies utilize correlative models, also known as ecological niche models, habitat models or bioclimatic envelope models to map species distributions. Correlative models form simple relationships between current species occurrences and environmental variables such as climate (Elith and Leathwick, 2009; Guisan and Thuiller, 2005). They do not require detailed knowledge on individual species growth requirements and generally perform well when predicting the current distribution of species based on the availability of adequate survey plots and chosen predictor variables (Pearson and Dawson, 2003). However, it is still challenging to apply these models over time and space when the relationships between species and the environmental variables may vary (Elith and Leathwick, 2009). The model correlations are based on the observed tree species distributions established under past climatic conditions that may differ from the climate experienced today. Empirical models thus form relationships that may not hold in the future, particularly with rising concentrations of atmospheric CO2 (Pearson and Dawson, 2003; Elith and Leathwick, 2009). The extrapolation of these correlations to non-tested   6 systems and novel conditions can introduce significant bias by omitting important mechanisms that drive the processes of natural systems (Adams et al., 2013).  Correlative models employ a variety of statistical methods or machine-learning techniques such as regression tree analysis or random forest algorithms to map tree species distributions (Iverson et al., 2008; Rehfeldt et al., 2014) or climate envelope models to predict suitable species habitat (Gray and Hamann, 2013; McKenney et al. 2007). For example, Hamann and Wang (2006) combined climate envelope modeling with the Biogeoclimatic Ecological Classification (BEC) system that groups distinct forest ecosystems in British Columbia based on common vegetation, climate and site conditions (Meidinger and Pojar, 1991). Using this ecosystem-based modeling approach, they predicted the geographic range of the dominant species of conifers. McKenney et al. (2007) also used the climate envelope approach to investigate how climate will change the future distribution of 130 tree species in North America. Both studies concluded that species were predicted to expand northward and to higher elevations with losses experienced in their southern ranges.  Recent studies have focused on incorporating genetic effects in species distribution models. Universal transfer functions were developed for lodgepole pine (Pinus contorta) based on population-specific climate responses from common garden experiments (Wang et al., 2010; O’Neill et al., 2008). Such comprehensive experiments however, are only available for the most widely studied species (Aitken and Bemmels, 2016). Rehfeldt et al (2014) also modeled the realized climate niche of different populations of Douglas-fir (Pseudotsuga menziesii) and   7 ponderosa pine. Accounting for genetic variation among species allowed them to develop seed transfer guidelines for adaptive forest management practices to climate change.   Sophisticated machine-learning algorithms have also been recently adopted in species distribution models (Elith and Leathwick, 2009; Iverson et al., 2008; Rehfeldt et al., 2014; Coops et al., 2009; Wang et al., 2012). Iverson et al. (2008) applied a random forest modeling tool that constructs decision trees by identifying the most important predictor variables to subset the data into branches ending at a terminal node with the predicted value. In contrast to the more simple regression tree analysis, this approach builds a large number of decision trees with a subset of the predictor variables and averages the dependent variables leading to greater prediction accuracy (Prasad et al., 2006). Such methods, originally developed by Breiman et al. (1984), have since also been employed in ecological studies (Verbyla, 1987) and have the advantage of applying the model outputs to the landscape level. Their application in providing ecological insight however is more limited (Elith and Leathwick, 2009).   The second type of species distribution model uses a process-based approach that incorporates mechanistic processes and functions and their interactions within a system and the surrounding environment (Kearney and Porter, 2009; Mäkelä et al., 2000). Process-based models apply physiological principles based on field experiments to simulate species growth and can predict how tree species might respond to new combinations of environmental conditions (Guisan and Thuiller, 2005; Cuddington et al., 2013). Such models have been described as more robust in extrapolating across species and regions and predicting potential distributions of individual species in the future (Adams et al., 2013; Kearney and Porter, 2009; Cuddington et al., 2013).   8 Process-based models are deemed more suitable to address complex questions on forest stand dynamics and the effects of climate change on species growth (Messier and Puettmann, 2011), however they require detailed information than can be challenging to obtain. Incorporating physiological responses of species allows to both predict climate effects on species distribution patters as well as to identify the underlying mechanisms controlling growth of individuals. As an example, western white pine is limited by soil water shortage conditions and the warmer, drier climate of recent years has led the species to become more susceptible to white pine blister rust leading to increased mortality (Harvey et al., 2008). Knowledge of physiological processes allows these models to be extrapolated to untested forest ecosystems (Weiskittel et al., 2010).  A number of process-based models link leaf area index (LAI) and forest productivity with C cycling process and allocation to stems, leafs and roots (Mäkelä et al., 2000; Crookston et al., 2010). The models operate at different scales from leaf or tree to ecosystem and regional levels, with model complexity increasing at higher temporal and spatial resolutions (Nightingale et al., 2004). The majority of process-based models function at the forest plot or stand level at spatial scales ranging from meters to 1 km2 (Nightingale et al., 2004). For example, productivity models incorporate a nutrient balance to derive properties of stand dynamics based on carbon-nitrogen interactions (Ågren, 1996; Thornley and Cannell, 1996; Nissinen and Hari, 1998; Kimmins et al., 1999). Forest gap models (JABOWA, SORTIE/BC) focus on the interaction between trees within a stand (Ludlow et al., 1990; Nikinmaa, 1992; Bartelink, 2000; Coates et al., 2003). The annual productivity of the stand is assumed to be constant and C is allocated among the trees using a shading model. These forest gap models serve well to assess mixed forest and heterogeneous stands but to a lesser extent in predicting productivity (Mäkelä et al., 2000).   9 The majority of process-based models incorporate knowledge on physiology and biological feedback mechanisms that are aspatial in nature (Nightingale et al., 2004). Only a few of these models have been designed to project both tree growth and mortality across regional scales (Nightingale et al., 2004) and those that do often assume homogenous patterns in soil water storage (Crookston et al., 2010) and soil fertility (Piao et al., 2013). In the next section, I describe the process-based 3-PG (Physiological Principles Predicting Growth) model used in this thesis.  1.3 The 3-PG model The 3-PG model was developed by Landsberg and Waring (1997) to combine empirical tree growth parameters measured by foresters with carbon and water balance models. It is a simple process-based forest growth model based on physiological principles and has been tested and validated across many forest types around the world (Landsberg and Sands, 2011; Landsberg et al., 2003; Mäkelä et al., 2000; Nightingale et al., 2004). The model provides a balance between complex, fine-scale process-based models and those functioning at annual time steps. It can also be upscaled from the stand to regional scale and used in association with remote sensing data. The 3-PG model uses climate and soil data to calculate photosynthesis, transpiration and growth allocation of tree species at the landscape level (Figure 1.1). It simulates tree responses to climate by determining how seasonal variations in LAI limit the photosynthetic capacity and the rate at which soil moisture can be withdrawn from the rooting zone. The model uses a number of biophysical factors and simplifying assumptions that have been developed based on experimental studies (Landsberg et al., 2003). The basic model assumptions include: (a) climate data at monthly intervals are sufficient for the modeling purposes, (b) net primary production (NPP) and autotrophic respiration (Ra) are constant fractions of gross primary production (GPP), (c)   10 maximum canopy conductance becomes constant when LAI > 4.0 and (d) the proportion of photosynthate allocated below-ground decreases linearly with nutrient availability to a minimum of ~25% of NPP (Landsberg and Sands, 2011).   The model performance has been thoroughly tested and the model outputs agreed well with field measurements of NPP, stem, foliage and root mass, LAI and basal area across different forest ecosystems in North America, Australia, Europe, Africa, South America and China (e.g. Landsberg et al., 2003; Gonzalez-Benecke et al., 2016; Almeida et al., 2004; Law et al., 2000; Zhao et al., 2009). Flux tower measurements of NPP and GPP in a ponderosa pine stand correlated well with 3-PG derived estimates (Law et al., 2000). Almeida et al. (2004) calibrated the model at Eucalyptus grandis Hill ex Maiden plantations in Brazil and was able to accurately determine growth outputs such as total stand volume, diameter at breast height (DBH) and soil water balance under different climate conditions. Gonzalez-Benecke et al. (2016) found that model parameters obtained for loblolly pine (Pinus taeda L.) stands were robust when tested throughout their range in both the US and South America for a variety of stand growth characteristics such as bole volume (r2 = 0.71) and survival (r2 = 0.95). They concluded that one set of tree parameters was applicable to forest stands throughout the geographic range of loblolly pine that encompass different stand ages and characteristics (Gonzalez-Benecke et al., 2016).  The 3-PG model calculates forest growth, GPP, NPP, LAI, transpiration, and litterfall on a monthly basis. Absorbed photosynthetically active radiation (APAR) is estimated from global solar radiation and LAI. The actual portion of APAR used for photosynthesis is constrained by environmental factors that are applied to the model, ranging from 0 being most limiting to 1   11 representing no constraint on photosynthesis. The environmental modifiers include (a) average daily temperature, (b) soil water limitations, (c) mean atmospheric vapor pressure deficits (VPD) during the daytime and (d) the frequency of frost. The soil water modifier is calculated as the ratio of the amount of plant available water in the rooting zone (ASW) to the soil water holding capacity (ASWC), the maximum amount of water the soil can hold at field capacity. ASW is determined with a monthly soil water balance that compares the total monthly precipitation, evaporation from the canopy and transpiration loss calculated with the Penman–Monteith equation (Landsberg and Waring, 1997; Monteith, 1965). Any excess water that is greater than the maximum soil water holding capacity is accounted as runoff or soil drainage. The rate of NPP is reduced in the model when evapotranspiration exceeds soil water availability. Soil fertility is defined as a site property and influences the canopy quantum efficiency (α) and the proportion of NPP allocated to roots (Landsberg and Sands, 2011). The 3-PG model defines the fertility-dependent growth modifier as a function of the soil fertility rating, where the poorest soils were ranked zero and the most fertile as one (Landsberg and Waring, 1997). The most limiting climate variable on photosynthesis is defined each month based on maximum deviation from optimum conditions. Further details on the 3-PG model as well as access to the software are available at www.3pg.forestry.ubc.ca.         12                               Figure 1.1 Diagram of the components used in the 3-PG forest growth model. Net primary production (NPP) and gross primary production (GPP) are driven by light absorption by the canopy (APAR). Constraints on photosynthesis are imposed by vapor pressure deficits (VPD), air temperature (Temp), available soil water and frost. NPP is partitioned into both above-ground (NPPA) and below-ground (NPPB) components based on soil fertility. Modified from Waring et al. (2010).   13 1.4 Research needs Current species distribution models that assess the environmental effects on species distributions often focus on individual tree responses. In this section, three research needs are identified that can contribute to the current knowledge on mapping tree species distributions and their vulnerability to climate shifts. They involve: 1) a more detailed analysis of the effects of soil properties in species distribution models, 2) how species vulnerabilities will vary among the ecoregions of western North American, and 3) what the emerging climate effects are on tree seedling distributions.  There is consensus in the literature that the variation in soil parameters influences tree species growth and distributions, and that the inclusion of soil properties can improve model predictions (Syphard and Franklin, 2009; Iverson and Prasad, 1998). Furthermore, there has been increased attention recently on the effects of drought on species growth and survival, which is becoming a major characteristic of the Anthropocene (Allen et al., 2015; McDowell et al. 2008). Despite the importance of water stress on species growth, current species distribution models generally focus on climate as a predictor and often omit knowledge of local soil characteristics or apply very coarse scale soil maps (Rehfeldt et al., 2014; McKenney et al., 2007; Iverson et al., 2008). Soil properties determine the water available to trees and the soil type influences the amount of rainfall that evaporates or infiltrates into the soil and the amount of plant available water stored in the rooting zone. A better understanding of the interaction between tree species and soil moisture can thus improve the ability to maintain healthy, productive forests. Inclusion of spatially variable soil properties in species distribution models would provide more nuanced species responses to environmental changes.   14 Tree species responses to climate change will likely vary among the ecoregions of western North America. While current studies have focused on climate change effects on the distribution of individual trees species, an improved understanding of species vulnerabilities at the ecoregion level is also essential (Coops and Waring, 2011, Rehfeldt et al., 2014, Iverson et al., 2008). Ecoregions are distinct geographical units, classified based on common landforms, climate and major vegetation types, and are widely used in forest ecology and conservation (CEC, 1997). They reflect species distributions and assessing species shifts by ecoregion is useful to examine geographically where to expect increased species stress. Within ecosystems, shifts in forest composition may occur following disturbances through a transition in dominant tree species or through changes in species abundance. Conifer forest stands have been projected to become increasing replaced by shrubs and grasslands in certain parts of western North America (Jiang et al., 2013). Incorporating process-based simulations in the species projections would allow to not only identify those ecoregions with the greatest expected species expansions or contractions, but also to recognize some of the causes of species stress. Identifying ecoregions where species shifts are likely to occur or where the least change is expected from current to future climates can contribute to focusing efforts to the most crucial geographic areas when developing adaptive forest management strategies such as assisted migration efforts. Recently established regeneration reflects if and how species are beginning to respond to changing environmental conditions. Most studies that predict shifts in species distributions with climate change base their predictions on the occurrence of mature tree species alone (Coops and Waring, 2011; McKenney et al., 2007; Hamann and Wang, 2006; Rehfeldt et al., 2014). Clearly, there are added benefits to monitoring the distribution of tree seedlings given the complexities of   15 species migration dynamics (Malcolm et al., 2002). The lack of seedling data employed in species distribution models limits the ability to assess the effects of climate change on forest regeneration and an improved understanding is needed on the interacting variables that favour or limit growth of regeneration at large spatial scales (Blanco et al., 2009; Weiskittel et al., 2011; Parmesan et al., 2011). Recent climate shifts provide an opportunity to observe emerging changes in species distributions by examining the occurrence of recently established tree seedlings. There is a need to continuously monitor these species responses, as seedlings will likely be exposed to altered climate conditions once they mature to trees.  1.5 Research overview and objectives The main objective of this thesis is to predict the current and future distribution of 20 major conifer tree species in western North America, and to identify those environmental variables that are most limiting to the growth of these species. In particular, I focus on soil water as an important environmental factor that affects species distributions and make model improvements by accounting for the spatial variation in soil properties. To address the main research objective, this thesis focuses on the following research questions:  1) How does soil water availability affect the distribution of 20 conifer tree species? 2) How do projected tree species vulnerabilities vary across the ecoregions of western North America?  3) Are climate shifts leading to emerging changes in tree seedlings distributions across British Columbia?    16 The flow and progression of the thesis chapters that answer these research questions are shown in Figure 1.2 and detailed below.  Chapter 2 describes the study region within the Pacific Northwest Region of North America as well as the main datasets used in this thesis that drive the species distribution models. Chapter 3 addresses the first research question though a sensitivity analysis that examines the importance of varying soil properties in the tree species distributions. It also provides improved baseline projections of 20 dominant tree species in western North America using spatial soil maps.  Chapter 4 applies the species models developed in Chapter 3 to project tree species vulnerabilities to the end of the 21st century. It identifies those ecoregions in the study region that are predicted to have the highest number of species stressed, thus answering the second research question of the thesis. Chapter 5 focuses on British Columbia as a diverse region for a range of species responses as identified in Chapter 4. It investigates the third research question by comparing the suitable habitat of recently established tree seedlings under current climate conditions with the vulnerability of mature tree species developed in Chapter 2.  Chapter 6 provides a summary of the key research findings and innovative knowledge drawn from the thesis and identifies the path forward for future research.       17         Improved predictions of baseline tree species distributions accounting for variations in climate and soil Projected tree species distribution maps and knowledge on regions suitable for species expansion or contraction Species vulnerability model assessment using early indicators of tree seedling occurrences Figure 1.2 Conceptual flow diagram of thesis chapters and outcomes of research questions.   18 2. Study area and data sources  2.1 The Pacific Northwest The Pacific Northwest Region (PNW), which spans across western Canada and the United States, contains a number of diverse ecoregions with varying climate and landforms (Figure 2.1). This diversity in landforms, climate and vegetation gives rise to a wide variety of soil types (Franklin and Dyrness, 1973).  Extending from Alaska to northern California, the Marine West Coast Forest, is the most productive PNW zone (Whittaker, 1961). These temperate coastal forests contain tree species such as Sitka spruce (Picea sitchensis), western hemlock (Tsuga heterophylla), and coastal Douglas-fir (var. menziesii), with western redcedar (Thuja plicata), grand fir (Abies grandis), Alaska yellow-cedar (Chamaecyparis nootkatensis) and coast redwood (Sequoia sempervirens) abundant in certain areas as well. The climate in this ecoregion is generally mild with high annual precipitation. Soils can vary from infertile, well-drained shallow soils to nutrient-rich fens with high organic matter content (Valentine et al., 1978).   In the Interior, the Northwest Forested Mountains has a drier climate and is the second most productive zone. Species such as interior Douglas-fir (var. glauca), western hemlock, noble fir (Abies procera), western larch (Larix occidentalis) are well distributed in this area. The subalpine environment contains species such as lodgepole pine, whitebark pine (Pinus albicaulis), subalpine fir (Abies lasiocarpa), and Engelmann spruce (Picea engelmannii). The soils range from nutrient-poor to moderately rich depending on both parent material and soil formation rates (CEC, 1997).   19 The North American Deserts, found predominantly in southwestern US and to a lesser extent in eastern WA, OR and southern BC, has an arid to semi-arid climate caused by the rain shadow of the Sierra Nevada and Cascade Mountains. Species such as ponderosa pine are widely distributed, in addition to pinyon pines (Pinus edulis) and junipers. Only about 2% of the ecoregion is covered by forests (McLaughlin, 1986). This ecological zone contains some very dry soils with low organic matter content associated with sparse vegetation (CEC, 1997).                    20                              Figure 2.1 The study area in western United States and Canada consisting of 11 States and two provinces.   21 2.2 Data sources 2.2.1 Climate and soil data Climate data required to drive 3-PG include minimum and maximum air temperature, precipitation, solar radiation, vapor pressure deficit and frost on a monthly basis. Mean monthly temperature and precipitation values were obtained from ClimateWNA (http://www.climatewna.com/), a program that interpolates long-term measurements from weather stations spatially. PRISM (Parameter-elevation Regressions on Independent Slopes Modell, dataset Norm71m) records were downscaled to 1 km through bilinear interpolation of the climate data and adjustments of temperature in mountainous terrain (Wang et al., 2016; Daly et al., 2008). The elevation adjustments were undertaken using a dynamic local regression function that relates monthly temperature with latitude, longitude and elevation (Hamann and Wang, 2005). A 90-m digital elevation model (DEM) from the Shuttle Radar Topography Mission (SRTM) was resampled to one km to obtain climate data at a given elevation from the ClimateWNA program. Average monthly VPD during the daytime were calculated with the assumption that daytime water vapor concentrations are the same as those at mean minimum temperature (Kimball et al., 1997). Maximum VPD was obtained as the difference between the saturated vapor pressure at the mean maximum and minimum temperatures. The average monthly daytime VPD was then derived as two thirds of maximum VPD, in order to obtain a mean daytime value instead of one derived from the daily extremes (Waring, 2000). The number of days per month with subfreezing temperatures (≤ 2 °C) was calculated from empirical equations with mean minimum temperature (Coops et al., 1998). Total incoming shortwave radiation was acquired on a monthly basis by combining spatial information from the North   22 American Regional Reanalysis (NARR) with topographic variations (Hember et al., 2017; Fu and Rich, 2002; Schroeder et al., 2009).  Soil properties required in the model include available soil water-holding capacity and soil fertility. These properties were previously obtained by taking advantage of their link with climate, forest productivity and maximum leaf area index (LAImax). Soil maps at 1-km resolution were derived for these two properties by using this link and adjusting LAImax values to correspond with those detected by MODIS satellite imagery (Coops et al., 2012). In areas where LAImax < 3.0, it was assumed the soil water was more limiting than fertility, based on field observations in the US and New Zealand (Runyon et al., 1994; Waring et al., 2008). Soil fertility increased in importance on sites where LAImax > 3.0. Accounting for regional variations in soil properties helps to refine estimates of suitable environments for different tree species. The validity of these soil maps remains to be tested but to date provided the most detailed soil information available throughout the entire study region.  2.2.2 Tree species plots Tree species presence and absence information were acquired from various sources across the PNW. In British Columbia, species data were derived from the centroids of stand-level polygons located in protected forested areas and from vegetation resource inventory plots collected across all forest lands using a three-phase, photo- and ground-based sampling design (Schroeder et al., 2010). In Alberta, species occurrence data were provided by Alberta Environment and Sustainable Resource Development. For the Canadian plots, only mature trees were considered and the spatial accuracy of the plot coordinates was approximately ±500 m. Tree species data in   23 the USA were acquired from Forest Inventory and Analysis (FIA) survey plots from the US Forest Service. As described in Schroeder et al. (2010), a permanent sampling grid was used to record FIA data at a density of about one plot per 2400 ha (Bechtold and Patterson, 2005). The samples included only trees of DBH > 2.54 cm and no seedlings. I used the publically available FIA data coordinates, which have been swapped between similar plots and reduced in spatial accuracy. All tree species data for Canada and USA, acquired from a total of 43,404 field plots, were compiled in a database for further analysis.   24 3. Soil water availability effects on the distribution of 20 tree species in western North America1 3.1 Introduction Most ecologists and botanists recognize that the distribution of tree species within a geographic region is shaped by both climate and soil properties (Syphard and Franklin, 2009). Increased climate change on a global scale is altering the hydrological cycle (IPCC, 2007) and affecting the amount of water available for tree growth. Differences in soil depth and water holding capacity are becoming increasingly important to identify with changes in climatic conditions (Ganey and Vojta, 2011; Peterman et al., 2013).   In the Pacific Northwest, the regional climate has become warmer since 2000 compared to a cooler phase in 1950–1975 (Waring et al., 2014). Such climate alterations are affecting forest ecosystems and their vulnerability to changes in disturbance regimes (Raffa et al., 2008; Westerling et al., 2006). In southwestern parts of the United States, extended periods of drought have been observed, leading to higher rates of tree mortality of pinyon pine–juniper forests caused by low water content in the soils (Peterman et al., 2013). The recent severe drought in California from 2012 to 2015 led to a forest canopy water loss of >30% as detected by satellite imagery (Asner et al., 2016). Ganey and Vojta (2011) reported a die off in mixed-conifer and ponderosa pine species in Arizona during a drought event in 1997–2007. They found a high mortality of 85% quaking aspen (Populus tremuloides) and 28% white fir (Abies concolor) and                                                  1 A version of chapter 3 has been published. Mathys, A., Coops, N.C., Waring, R.H., 2014. Soil water availability effects on the distribution of 20 tree species in western North America. Forest Ecology and Management 313, 144–152.   25 attributed this to elevated temperatures associated with climate change as well as insect attack. Extensive aspen decline has also occurred in western Canada in response to regional drought and insect defoliation (Hogg et al., 2008; Worrall et al., 2013; Krishnan et al., 2006). Soil water deficits during the growing season are also prevalent in low elevation areas causing a decline in productivity (Latta et al., 2010). Water stress is known to affect plant growth by limiting photosynthesis and transpiration, and can lead to mortality under severe conditions due to hydraulic failure from extreme drought or carbon starvation during extended periods of drought (Reichstein et al., 2007; Adams et al., 2013; Bradford et al., 2017; McDowell et al., 2008). There has been an expressed need to better understand the link between tree responses associated with increased exposure to drought conditions and use this to predict areas where species shifts will occur (Allen et al., 2010). Available soil water content is an essential requirement for successful tree establishment, growth and productivity, providing a means to quantify tree responses to hydrological changes (Weltzin et al., 2003).  Foresters and ecologists have long recognized species-specific requirements with regard to soils. For example, ponderosa pine, a widely distributed species in the Pacific Northwest, is known to be drought tolerant and can effectively compete in well-drained sandy soils, however is very nutrient demanding (Tarrant, 1953; Klinka et al., 1999). Douglas-fir grows in a wide variety of soils although it prefers sandy loams with good drainage and high nitrate content (Farrar, 1995; Klinka et al., 1999). In contrast, western redcedar is able to survive anaerobic conditions and has a wide ecological amplitude, tolerating soils with low moisture and nutrient content (Klinka et al., 1999; Harlow and Harrar, 1950). Knowledge about how soil water availability influences tree   26 species establishment and growth is important for forest managers seeking to conserve biodiversity while also increasing forest resources.   Whilst the importance of soil attributes on tree growth and species distributions is well known, inclusion of this information into predictive models is less common (Syphard and Franklin, 2009) and even rarer in species models designed to assess the impact of climate change (Rehfeldt et al., 2009; McKenney et al., 2007). Instead, species distribution models often utilize climate data and focus on climatic controls on species occurrence. McKenney et al. (2007) used the climate envelope approach to model the climate niche of 130 tree species in North America. The study did not utilize soil data due to limited maps available on a continental scale. Rehfeldt et al. (2009) modeled the distribution of quaking aspen in western USA using only climate parameters and noted that inclusion of soil factors improves the accuracy of model predictions. The challenge for including soil information is great due to the low spatial resolution of available digitized data sets, however attempts to improve the situation continue through the development of a global digital soil map (Sanchez et al., 2009). In North America, the State Soil Geographic (STATSGO) database is the source that most species distribution modellers utilize (Iverson et al., 2008; Coops and Waring, 2001). Iverson et al. (2008) mapped species habitat in the eastern United States using soil properties derived from STATSGO, indicating that soil parameters influenced tree species distributions. However, they did not provide any further explanation on how soils might assist or constrain species range shifts under climate change. Coops and Waring (2001) also employed STATSGO to derive soil water content in Oregon. They found soil water to have an important impact on forest growth especially during summer drought conditions. All of the above cited authors agreed that variation in soil parameters influences the predicted   27 distribution and growth of tree species and that coarse-resolution maps of soil properties were insufficient and need to be refined.  Recognizing the need for more accurate spatial information, a new soil map was produced from the relationship between soils, climate and forest productivity (Coops et al., 2012). The spatial variation in ASWC and soil fertility (FR) were inferred at 1 km resolution by optimizing the predictive maximum leaf area index (LAImax) derived with a process-based growth model with values acquired from satellite measurements.  In this study, I utilized these derived layers of soil properties to model the occurrence of tree species across the Pacific Northwest. The sensitivity of model predictions to variations in soil water availability was also assessed by analyzing species predictions when ASWC was increased and decreased by 50%. Finally, the implications of recent climatic change on species distributions were evaluated by comparing shifts in ranges under stable and variable soil water conditions.  3.2 Methods 3.2.1 3-PG model parameterization Monthly mean climate from 1950–1975 acquired from ClimateWNA was used in the 3-PG model, which was run for 50 years following establishment of a stand populated with 300 seedling ha-1. The long-term climate record was chosen to match the field survey plots of tree species occurrences described in section 2.2.2. The 3-PG model was parameterized for Douglas-fir, a widely distributed species across the study region, using existing forestry yield tables and   28 field-based observations (Waring and McDowell, 2002). The relative importance of the seasonal modifiers on other species was then described in relation to how they limited photosynthesis of Douglas-fir. The soil maps developed by Coops et al. (2012), accounting for variations in soil fertility and moisture on a regional scale, were also used in the model, unlike most previous studies, where soil properties were assumed to be uniform or are omitted. The soil water map provided information on ASWC, which in turn was used to calculate the soil water modifier in 3-PG using a monthly soil water balance as described in section 1.3. The term soil water availability used in the thesis refers to the soil water modifier, which poses limits to the amount of plant available water for tree growth. ASWC was allowed to vary across the region from 0 to 300 mm based on local climate and site characteristics. In this chapter, I analyzed the effect of soil water availability on LAImax by generating spatial layers of LAI with varying soil variables. I compared the simulation results of baseline conditions using spatially variable maps of ASWC (Coops et al., 2012) with FR values held constant at 0.5 to modeled values of LAImax when both soil variables were held constant (FR = 0.5, ASWC = 200 mm). The maximum photosynthetic efficiency was set to 0.04 mol C mol-1 photon-1 when FR was held constant.  3.2.2 Predicting species distributions with decision tree models The 3-PG process-based growth model outputs were combined with a decision tree analysis to map species distributions following methods of Coops et al. (2009). The seasonal averages of the climatic and soil modifiers were related to the field-based observations of species occurrences (section 2.2.2) by extracting this information at each of the 43,404 plots distributed within the PNW. Using Decision Tree Regression (DTREG, Sherrod, 2010) software, I determined the relative importance of the four variables for each tree species and predicted their presence and   29 absence across the landscape. The decision tree model ranked the importance of the seasonal and annual environmental constraints and established thresholds between 0 (no growth) and 1 (unlimited growth) that determined whether a species would be present on a site based on their recorded occurrences.  Model accuracies were calculated as a weighted value of the percentage of plots where the species occurrence was correctly assigned, proportional to the number of plots corresponding to the two categories, presence and absence. For species that had a low presence value, the categories were balanced by increasing the weight of data rows of the minority category so as to make the sum of all of the target categories equal. Accuracy was assessed through a 10-fold cross-validation technique, where the data were divided into 10 random groups of equal size. Nine of the groups were used in the model and tested against the remaining 10% of the data. The procedure was repeated 10 times and the overall accuracy was determined from the average of the 10 results in a process known as k-fold partitioning (Breiman et al., 1984). This cross-validation technique is a recognized approach to validate the model outputs derived from machine learning algorithms (Hastie et al., 2009). Ideally, an independent dataset would be used for validation, however such information on tree species occurrences was not available for the full study region. The Kappa (κ) statistic was calculated to evaluate the accuracy of the species models. When the predicted values were in agreement with the observed plots, κ = 1 and when there was no agreement, κ = 0. If one category has a significantly larger class size, then κ will be less than one, in which case κ can be rescaled to match the observed marginal frequencies (Ben-David, 2008). Species distribution results were also visually compared to field survey plots and to existing range maps (Little, 1971).   30 3.2.3 Sensitivity analysis A sensitivity analysis was conducted to determine the effects of varying ASWC across the study area on the distributions of selected tree species. ASWC was increased and decreased by 50% from baseline conditions (resulting in values of 0-150 and 0-450 mm, respectively) while fertility was held constant at 0.5 (medium soil nutrient regime). Simulations were run using contemporary monthly climate data (1976–2009), which includes the available decade of reprocessed MODIS LAImax values to produce the spatial soil layers (Coops et al., 2012). The environmental constraints were then applied to the decision tree analysis to generate the resulting tree species models. The extent to which the spatial distribution of each tree species was altered was assessed by comparing consecutive runs of the species distribution models.  3.3 Results Comparison of predicted LAImax resulting from different values of soil properties helped identify where a more accurate assessment of ASWC appears critical to predict the distribution of tree species. The difference between LAI max with constant FR (0.5) and when both variables were held constant is shown in Figure 3.1. The greatest difference in LAI max occurred in parts of Marine West Coast Forests in Washington and Oregon, with increases of >0.8 m2 m-2 observed when using the spatial map of ASWC. In these areas, values of ASWC > 200 mm are required to maintain LAI max commensurate with forest productivity (Waring et al., 2008). LAI max values were >0.8 m2 m-2 lower in the North American desert zone to the east of the Cascade and Sierra Mountains when using variable compared with fixed values of ASWC. Low rainfall in these areas combined with shallow soils lead to sparse vegetation and low productivity. Over much of   31 the study area, LAI was at least 0.2 m2 m-2 higher when ASWC was allowed to vary over the landscape.  The accuracies of the predicted species distributions using the variable soil layers from Coops et al. (2012) are displayed in Table 3.1. The species distribution models were validated using the cross-validation technique described in section 3.2.2., where a random sample of observations was used as validation data and tested against the remaining observations used as training data. The procedure was repeated 10 times to determine the accuracy of the model predictions (Breiman et al., 1984). The overall accuracy of the species models averaged 84% with a kappa value of 0.79. Generally, the overall presence accuracy was slightly higher (88%) than the absence accuracy (81%) of the species. Sitka spruce had the highest overall accuracy of 94% (κ = 0.92) with a presence accuracy of 96%. The lowest model accuracy of 70% (±0.4) was for lodgepole pine with a κ value of 0.62 indicating only moderate agreement of model predictions with the plot values.            32                          Figure 3.1 Predicted differences in LAI (m2 m-2) when available soil water content varied across the region from 0-300 mm, depending on local climate and soils, compared to when it was held constant at 200 mm. In both simulations, soil fertility was held constant at a median level of 0.5. The gray background depicts the digital elevation model of non-forested areas in the study region.   33 Table 3.1 The accuracy of decision tree models and Kappa (κ) statistic derived for each of 20 tree species based on mean monthly climate conditions during the period 1950-75. The 95% confidence intervals are shown in brackets. The species distribution maps produced from these models are presented in Appendix A.1.  Species Latin name Presence Accuracy (%) Absence Accuracy (%) Overall Accuracy (%) Κ Lodgepole pine Pinus contorta 68 (±1.0) 78 (±0.4) 70 (±0.4) 0.626 Douglas-fir Pseudotsuga menziesii 74 (±0.7) 80 (±0.5) 78 (±0.4) 0.599 Pinyon pine Pinus edulis 84 (±1.9) 96 (±0.2) 90 (±0.3) 0.910 Subalpine fir Abies lasiocarpa 95 (±0.6) 62 (±0.5) 79 (±0.4) 0.856 Engelmann spruce Picea engelmannii 84 (±1.1) 72 (±0.4) 78 (±0.4) 0.887 Whitebark pine Pinus albicaulis 91 (±1.6) 81 (±0.4) 86 (±0.3) 0.805 Western larch Larix occidentalis 88 (±1.7) 83 (±0.4) 85 (±0.3) 0.742 Pacific silver fir Abies amabilis 90 (±1.8) 87 (±0.3) 88 (±0.3) 0.790 White fir Abies concolor 93 (±1.1) 80 (±0.4) 87 (±0.3) 0.839 Grand fir Abies grandis 85 (±1.7) 79 (±0.4) 81 (±0.4) 0.675 Noble fir Abies procera 93 (±3.4) 90 (±0.3) 91 (±0.3) 0.858 Alaska yellow-cedar Callitropsis nootkatensis 95 (±1.6) 90 (±0.3) 92 (±0.3) 0.895 Utah juniper Juniperus osteosperma 95 (±1.0) 82 (±0.4) 89 (±0.3) 0.890 Sitka spruce Picea sitchensis 96 (±1.9) 92 (±0.3) 94 (±0.2) 0.922 Western white pine Pinus monticola 81 (±2.7) 70 (±0.4) 75 (±0.4) 0.570 Western redcedar Thuja plicata 87 (±1.2) 79 (±0.4) 83 (±0.4) 0.718 Western hemlock Tsuga heterophylla 95 (±0.8) 80 (±0.4) 87 (±0.3) 0.883 Mountain hemlock Tsuga mertensiana 85 (±2.5) 88 (±0.3) 87 (±0.3) 0.754 Ponderosa pine Pinus ponderosa 91 (±0.7) 73 (±0.5) 82 (±0.4) 0.779 Quaking aspen Populus tremuloides 82 (±1.0) 71 (±0.5) 77 (±0.4) 0.895  Figure 3.2 presents the number of times the soil water modifiers were included in a decision tree model for each species. ASWC explained some of the variation for 75% of the tree species, where 30% were extremely sensitive and 45% somewhat sensitive to changes in water availability. Of these, western hemlock had the most sensitivity to a soil water modifier, with all four seasons contributing to predicting its presence across the PNW. The occurrence of western redcedar was also highly affected by ASWC during the summer, winter, and averaged annually. Douglas-fir and ponderosa pine, the two most widely distributed species, were relatively insensitive to simulated variation in seasonal ASWC values, with the variable appearing only   34 once in their respective decision tree models. ASWC was absent in the models of Engelmann spruce, whitebark pine, grand fir, Sitka spruce and Alaska yellow-cedar, indicating that other environmental constraints were predicted to have a greater influence on the occurrence of these species (Figure 3.2).             Figure 3.2 Occurrence of soil water modifiers (annual, spring, summer, fall, winter) in decision tree models for each of 20 tree species native to the Pacific Northwest Region.  The importance of each soil water modifier in determining the distribution of the tree species is indicated in Figure 3.3. Out of the 20 species, 15 had at least 1 variable involving soil water in their respective decision tree model. Of these, 6 to 38% of the variance was explained by soil water related variables. Soil water deficits during the summer were the most limiting on species occurrences, accounting for 38% of the variance in the species range-prediction models. Drought during this time of the year provided a major constraint on the distribution of western redcedar and western hemlock. The amount of moisture in the soil during the wintertime was also an important constraint, reducing the distribution of some of the species by 21%. Because the 01234Soil water modifer  35 accumulation of a snowpack was not accounted for when temperatures remained below freezing, the 3-PG model predicts that soils will remain at field capacity in much if not most of the subalpine forest zone in the winter. As a result, soils with high winter water contents are surrogates for areas where a heavy snowpack normally accumulates, which contributes to defining the range of lodgepole pine, subalpine fir, and four other snow-adapted species. ASWC annually and during the spring and fall played a lesser role in defining species distributions, contributing between 6% and 9% of the prediction’s accuracy (Figure 3.3). Below I discuss the results in relation to Douglas-fir and western white pine, two tree species with different soil water tolerances.   Figure 3.3 Importance of each soil water modifier in determining the mean % variance of the tree species.  Changing the soil water holding capacity had varying effects on species distribution models. Western white pine, a species that grows on productive sites and has a low tolerance to drought 0 10 20 30 40Soil water summerSoil water winterSoil water springSoil water annualSoil water fall% Variance  36 conditions, exhibited a contraction of its range, with a 50% reduction in baseline ASWC in Idaho, Montana, and the Interior of BC (Figure 3.4a). Decreasing ASWC by half reduced the overall range of western white pine by 6% (Table 3.2), whereas increasing ASWC by 50% had no measurable effect. Among the water deficits, those in the spring were most important in defining the range of western white pine. The range of the tree species extends from the coast of BC to the east of the Cascade and Sierra Mountains (Figure 3.4c).  Increasing ASWC by 50% in the sensitivity analysis, would result in the expansion of predicted Douglas-fir distributions along the coast of BC and Vancouver Island, while simultaneously decreasing it in parts of California and Oregon, where such changes may favor other species (Figure 3.4b). Douglas-fir is moderately drought tolerant and does not grow well on water-logged soils. Increasing ASWC by 50% reduced its overall range by 13%, whereas decreasing ASWC compressed its range by only 3% (Table 3.2). Coastal Douglas-fir is well represented in the Marine West Coast Forests of BC, Oregon, and Washington, and extends southward to California. Interior Douglas-fir is also found in the Northwest Forested Mountains to the east of the Rocky Mountains in the United States and Canada (Figure 3.4d).  Western hemlock produced the greatest change in the sensitivity analysis when ASWC was increased by 50%, with its range reduced by 21% of baseline ASWC values. Both of the subalpine species, Engelmann spruce and whitebark pine displayed no change in distribution with varying ASWC and other environmental constraints were ranked as more limiting in the decision tree models of these species. In general, the tree species were more sensitive to the presence or absence of excess soil moisture conditions (representing heavy winter snowpack)   37 because they normally experience little soil water deficit at high elevations (Runyon et al., 1994). Sensitivity analysis of western redcedar showed the largest potential change in species distribution. The species responded with range expansion to both increased and decreased ASWC, which may be explained by its ability to tolerate a wide range of soil moisture regimes (Klinka et al., 2009). The range expansion under reduced ASWC seems counterintuitive and may indicate that the simulated reduced soil water availability was above a threshold required to impose stress within the native range of the species.   Table 3.2 Current predicted ranges and percent change in species distributions when increasing or decreasing ASWC by 50% compared with baseline values with variable ASWC maps (Coops et al., 2012) and constant FR (0.5).  Species Total range (km2) Decrease ASWC 50% Increase ASWC 50% Western white pine 527,791 -6 0 Douglas-fir 915,579 -3 -13 Western hemlock 475,097 14 -21 Pinyon pine 140,244 17 9 Ponderosa pine 115,130 21 9 Western redcedar 478,509 27 -9 Engelmann spruce 965,623 0 0 Whitebark pine 788,490 0 0          38                            Figure 3.4 Changes in the distributions of two species with varying available soil water content (ASWC) during the period 1976-2009. The expansion of a) western white pine range for ASWC at 50% (red) and baseline condition (green) and b) Douglas-fir range at baseline conditions (red) and 150% ASWC (green) and the current range of c) western white pine and d) Douglas-fir at baseline ASWC values.   39 3.4 Discussion Species distribution models are most useful to not only predict but also to provide an improved understanding of potential shifts in range. As demonstrated in this chapter, the degree that soil water deficits occur seasonally has a measureable effect on the accuracy of predictions (Table 3.1) for a majority of the 20 native tree species. I also showed in predicting LAImax that ASWC becomes a critical variable to include in models of forest productivity (Coops et al., 2012). Climate change affects soil moisture properties and depending on the soil order, texture and depth impacts their ability to retain and release water. Changes in temperature and precipitation patterns can affect the soil water balance, which as this chapter has shown, changes the predictions of species distributions. In combination, climate and soils also may alter the vulnerability of forests to disturbance from insects, pathogens and wildfires (Westerling et al., 2006; Raffa et al., 2008).  In this study, the important link between forest productivity and soil water storage capacity was demonstrated, reinforcing the growing need for quantitative information on soil properties. Soil water has an impact on LAI, which in turn affects photosynthetic uptake and evapotranspiration of the trees (Landsberg and Waring, 1997). The soil maps derived by Coops et al. (2012) and used in this study are a step forward in providing continuous spatial information across large regions. The extent to which the approach can be refined is unknown but worth pursuing as remote sensing techniques become more sensitive.  The inclusion and employment of the soil map in the species distribution models, helped to slightly increase the model accuracy of the tree species compared to a previous study by Coops   40 et al., (2011), where soil properties were held constant (accuracy = 81%, κ = 0.74). Studies have found that combining climate, soil, and satellite imagery produced the highest model accuracies than when omitting any of these variables (Salajanu et al., 2010; Iverson et al., 2008). Meyer and Thuiller (2006) also reported that inclusion of variables at different spatial scales can increase the overall accuracy of model predictions. Sensitivity analysis in this chapter revealed that the distribution of many of the tree species is variable with soil water holding capacity. The range of western white pine contracted slightly with reduced ASWC, while increasing water availability had no effect. It used to dominate the Interior Northwest forests of North America; however, drought has made western white pine prone to disease and its distribution has been greatly reduced due to white pine blister rust (Harvey et al., 2008). In this analysis, Douglas-fir was more vulnerable to excess soil moisture in the rooting zone than to drought conditions. Field measurements and model predictions confirm the species to be tolerant to drought by accessing soil water in deeper soil layers and in fractured bedrock (Waring et al., 2008). The species generally prefers soils with good drainage and is generally absent in waterlogged sites (Farrar, 1995). Ponderosa pine was also sensitive to ASWC, with varying soil moisture content altering its spatial distribution. This species is known to be drought tolerant and grows best in loamy soils with good drainage and high nutrient content. Reduction of ASWC projected expansion of the Ponderosa pine range, indicating that increased water stress associated with climate change may enable drought-adapted species such as ponderosa pine and Douglas-fir to thrive in a changing environment, depending on factors such as available seed sources and establishment success.  Changes in ASWC did not affect all tree species. Engelmann spruce was insensitive to increases and decreases in soil moisture. This species is frequently found in saturated soils with a high   41 water table and has a moderate tolerance to drought (Alexander and Shepperd, 1990). The species’ tolerance to summer frost and air temperatures during the fall and annually were more important in the decision tree model in determining its presence over the landscape. In this analysis, pinyon pine was also not predicted to be affected by reduced available soil water conditions. A decline in pinyon pine in southwestern USA has been previously attributed to increased drought exposure and bark beetle attack (Peterman et al., 2013). Soil water availability did not seem to be the main cause of predicted species decline in our study, and long term increases in vapor pressure deficits associated with a warming trend may play an equally important role (Waring et al., 2011). In addition, soil factors such as soil orders and mycorrhizal colonization as well as species interactions with disturbance regimes that were excluded from this analysis are also important to consider (Pickles et al., 2015; Simard, 2009).  It is recognized that climate is the main control on species distributions at broad spatial scales (Salajanu et al., 2010). Temperature variables are important indicators of climate change and its impact on species distributions. Elevated minimum winter air temperatures have been associated with increased bark beetle activity (Raffa et al., 2008), leading to widespread tree mortality such as the mountain pine beetle attack on lodgepole pine trees in BC. Tree species such as whitebark pine were previously shown to be more vulnerable to changes in air temperature than to changes in precipitation and the ASWC (Schrag et al., 2008). In this study, whitebark pine was also unaffected by ASWC and its distribution was mainly determined by suitable climate conditions. Nonetheless, results from this chapter indicate that water availability to trees also plays an important role in the decision tree models and can provide further insight into species-specific requirements. Climate and soil water content are spatially interrelated and it can be challenging   42 to examine their importance on model predictions separately. For example, an increase in temperature is likely associated with greater runoff from snowmelt and early soil water recharge, which in turn could cause soil water depletion earlier in the growing season (Hamlet et al., 2007). Adaption to changes in heavy snowpack is likely an important constraint on the distribution of subalpine species, which is challenging to model climatically and is currently not included in the 3-PG model and thus may diminish the effects of drought. Chapter 3 informs where future field surveys will benefit in validating the model predictions and where forest and soil management is required for climate change adaptation, through efforts such as minimizing soil disturbances, removing competing vegetation to increase the soil water supply and maintaining soil fertility for optimum forest growth.   43 4. An ecoregion assessment of projected tree species vulnerabilities in western North America through the 21st century2 4.1 Introduction The ecological regions of western North America have diverse environmental conditions and floristic communities. Species that coexist in these ecoregions have distinct adaptations to climate, geology and competing vegetation (Franklin and Dyrness, 1973). Forest ecosystems in western North America have been exposed to increasingly severe drought, a decline in snow cover, and lengthening in the growing season, all of which can result in stresses on tree species, changes in forest productivity and permanently alter forest composition (Zhou et al., 2001; Mote et al., 2005; Ganey and Vojta, 2011; IPCC, 2014; Waring et al., 2014; Allen et al., 2015). How vulnerable are tree species in the different ecoregions to the recent changing climatic conditions and those projected in the future?  The IPCC has reported observed large-scale changes to terrestrial systems with recent climate change that are predicted to accelerate with elevated greenhouse gas emissions from anthropogenic sources (IPCC, 2014). Increases in average annual air temperature of 3 °C are projected by 2080 compared to 1970–1999 (Mote and Salathé, 2010). The accelerated temperature increases are predicted to be particularly acute in northern and interior regions of western North America (Mote et al., 2003). The rapid projected climate change will exceed the capacity of many species to migrate to areas with more favorable conditions (Davis and Shaw,                                                  2 A version of chapter 4 has been published. Mathys, A.S., Coops, N.C., Waring, R.H., 2017. An ecoregion assessment of projected tree species vulnerabilities in western North America through the 21st century. Global Change Biology 23, 920–932.   44 2001; Aitken et al., 2008). Major changes in species distributions will likely be accompanied by disturbances caused by fires, insect attacks and diseases that will affect the carbon, water and energy balances of affected ecosystems. Recognition of those ecoregions that are most vulnerable to climate change would permit land managers to concentrate their efforts where protection or a change in management is most needed.  A number of studies have employed species distribution models to assess the impact of climate change over terrestrial environments (Hamann and Wang, 2006; McKenney et al., 2007; Coops and Waring, 2011; Notaro et al., 2012; Rehfeldt et al., 2012; Jiang et al., 2013). In the southwest United States, Notaro et al. (2012) predicted that the number of tree species in decline would likely exceed those projected to expand. The predicted declines were expected to be most prominent in high-elevation conifer forests by the end of the 21st century. Rehfeldt et al. (2012) predicted an expansion of desert biomes, as well as a northward expansion of temperate and montane forests. Jiang et al. (2013) also projected a decline in conifer tree species of up to 24% across western North America, especially in the latter half of the 21st century. Forest stands would increasingly be replaced by shrubs and grasslands (Jiang et al., 2013). They highlighted the need to incorporate process-based simulations to understand why species are becoming vulnerable to climate change.  The majority of the above-cited studies use empirical approaches to form statistical relationships between current climatic variables and species occurrences and use these to project suitable climate space in the future. These approaches have important associated limitations (section 1.2) and an alternative is to use process-based models that incorporate biological responses based on   45 experimental data. Process-based models offer a means to predict how tree species might respond to new combinations of environmental conditions (Guisan and Thuiller, 2005). They can also provide an understanding of the underlying mechanisms controlling species growth and identify critical environmental limitations. This allows to not only predict how species respond to climate change but also to determine the climate constraints that limit tree growth such as temperature extremes, frost, drought and vapor pressure deficit.   In this chapter, I used the forest growth model, 3-PG, to infer how tree species are likely to respond to changing climatic conditions. The model is described in detail in in Chapter 1.1. The model provides a balance between complex, fine-scale process-based models and those functioning at annual time steps. This study incorporated 1-km resolution soil maps depicting available soil water-holding capacity and soil fertility (Coops et al., 2012) to provide more realistic representation of spatially diverse site conditions. The majority of studies currently omit soil properties or hold values constant, although it is generally recognized that variation in soil parameters influence tree species growth and distributions and their inclusion can improve model predictions (Iverson and Prasad, 1998; Nightingale et al., 2007; Syphard and Franklin, 2009). In Chapter 3, I predicted the current distribution of 20 common tree species across western North America with an average accuracy of 84%. In this chapter, these species distribution models were extended to predict and map the vulnerability of ecoregions to climate change in the 21st century. The IPCC defines vulnerability to climate change as a measure of a systems exposure, sensitivity and their adaptive capacity to respond to an experienced stress (IPCC, 2014). The modeling approach in this thesis accounted for the degree of exposure of the tree species to climate variation and their sensitivity by mapping their tolerance to climate effects such as   46 drought. It also addressed the adaptive capacity of the system, how well it responds to the experienced change, by posing limits to the migration rates of a species. To do this, a maximum migration rate of 200 m/year was applied to the species distribution models, based on paleoecological records of dispersal distances and long-term studies on migration rates of different species (Davis, 1989; Beckage et al., 2008; Coops et al., 2016). The adaptive capacity of a forest ecosystem also depends on genetic adaptation, reproduction, demographic and life history traits. These factors can increase the complexity and uncertainty of projecting tree species responses to climate change. The focus of this study was to identify areas with suitable climate and soil conditions for species rather than predict actual changes to their distribution in the future. Annual monthly climate data to the end of the 21st century were used to model suitable sites for tree species using updated climate models in concert with the most recent IPCC (2014) projections. This chapter predicted tree species vulnerabilities in the future by comparing the areas that are projected to become climatically suitable or unsuitable to the current species ranges. Finally, a constraint was applied on the rate that species can migrate to areas with more favorable climatic conditions.  4.2 Methods 4.2.1 Ecological regions of western North America Western North America, which spans across Canada and the United States, contains a number of diverse ecoregions that are shaped by unique landforms, soils, climate and vegetation (Franklin and Dyrness, 1973). The major ecoregions include the Marine West Coast Forest, the Northwest Forested Mountains and the North American Deserts, as described in Chapter 2.1. The major ecoregions are further subdivided into more detailed (level III) ecological regions   47 (http://archive.epa.gov/wed/ecoregions/web/html/na_eco.html#Level III), allowing them to be differentiated at a level appropriate for designing regional environmental management strategies (CEC, 1997). The 41 ecoregions included in this thesis are shown in Figure 4.1 and Table 4.1.                    Figure 4.1 Ecological regions (level III) in western Canada and United States as defined by the U.S. Environmental Protection Agency (EPA) and Environment Canada (Ecological Stratification Working Group (ESWG), 1996). The major ecoregions (level I) are outlined in bold.   48 Table 4.1 Level III Ecoregions within study region of western Canada and United States.   Ecoregion III  Name   10.1.1 Thompson-Okanogan Plateau 10.1.2 Columbia Plateau 10.1.3 Northern Basin and Range 10.1.4 Wyoming Basin 10.1.5 Central Basin and Range 10.1.6 Colorado Plateaus 10.1.7 Arizona/New Mexico Plateau 10.1.8 Snake River Plain 11.1.3 Southern and Baja California Pine-Oak Mountains 13.1.1 Arizona/New Mexico Mountains 5.4.1 Mid-Boreal Uplands and Peace-Wabaska Lowlands 5.4.2 Clear Hills and Western Alberta Upland 6.1.4 Wrangell and St. Elias Mountains 6.1.5 Watson Highlands 6.1.6 Yukon-Stikine Highlands/Boreal Mountains and Plateaus 6.2.1 Skeena-Omineca-Central Canadian Rocky Mountains 6.2.10 Middle Rockies 6.2.11 Klamath Mountains 6.2.12 Sierra Nevada 6.2.13 Wasatch and Uinta Mountains 6.2.14 Southern Rockies 6.2.15 Idaho Batholith 6.2.2 Chilcotin Ranges and Fraser Plateau 6.2.3 Columbia Mountains/Northern Rockies 6.2.4 Canadian Rockies 6.2.5 North Cascades 6.2.6 Cypress Upland 6.2.7 Cascades 6.2.8 Eastern Cascades Slopes and Foothills 6.2.9 Blue Mountains 7.1.4 Pacific Coastal Mountains 7.1.5 Coastal Western Hemlock-Sitka Spruce Forests 7.1.6 Pacific and Nass Ranges 7.1.7 Strait of Georgia/Puget Lowland 7.1.8 Coast Range 7.1.9 Willamette Valley 9.2.1 Aspen Parkland/Northern Glaciated Plains 9.3.1 Northwestern Glaciated Plains 9.3.3 Northwestern Great Plains 9.4.1 High Plains 9.4.3 Southwestern Tablelands   49 4.2.2 Current and future climate Monthly climatic data averaged from 1976 to 2009 served as a baseline for this study and were obtained from ClimateWNA as described in Chapter 2.2. I also applied future climate projections to model potential species distributions. Future monthly projections from 2010 through 2100 were obtained from the Canadian Climate Centre’s Modeling and Analysis (CCCma) CanESM2 model for three emission scenarios (low (RCP2.6), medium (RCP4.5) and high (RCP8.5)) as described in the IPCC Fifth Assessment Report (AR5). Although the CCCma model starts with a lower air temperature value during the 20th century and projects more accelerated warming than other GCMs, it is considered, the most appropriate model for the Pacific Northwest Region (Mote et al., 2005). Over the last century, the mean annual air temperature across the region has increased by 0.8 °C. By 2080, the Canadian model projects an additional 3 °C average annual increase and wetter summers, particularly in BC, compared to records from 1970 to 1999 (Spittlehouse, 2008; Mote and Salathé, 2010). These data were incorporated in the 3-PG model to evaluate how environmental stresses might affect the distribution of tree species in the future.  4.2.3 Modeling approach The 3-PG model was first run to obtain predictions on the seasonal constraints to species growth following the procedures in Chapter 3. Monthly baseline climate (1976–2009) drove the model from seedling establishment to a stand age of 50 years. Spatial variations in available soil water-holding capacity and soil fertility were included (Coops et al., 2012; Mathys et al., 2014). In the previous chapter, decision tree models were developed by combining the climate modifiers derived from 3-PG with recorded occurrences of the 20 tree species on 43,404 field plots (Table 4.2). The models predicted species occurrence and absence with an overall accuracy of 84% and   50 a kappa statistic of 0.79 (Mathys et al., 2014). In this chapter, the analysis was expanded by comparing the availability of suitable habitats in the future using climate data averaged from 2075 to 2100 with that projected under baseline conditions (1976–2009). I also estimated within the current range of a tree species, where it was likely to experience unfavorable conditions (defined at a vulnerability threshold of 70% for the years sampled) using monthly averaged climate data for each year from 2075 to the end of the 21st century. Similarly, the potential areas for range expansion were identified by mapping where favorable conditions might occur ≥ 70% of the time. To provide more realistic predictions of potential migration, a maximum rate of dispersal of 200 m per year (Coops et al., 2016) was imposed on tree species expansion based on paleobotanical records of species migrations (Davis, 1989; Beckage et al., 2008).   Within each ecoregion, the percentage change in suitable locations of 20 tree species between 2075 and 2100 was contrasted with their current range (1976–2009). The number of species stressed was calculated under changing climate by ecoregion (EPA level 3) to determine those forested areas in western North America that are highly vulnerable to projected shifts in climatic conditions. To do this, the number of pixels classified as stressed was counted for all species and averaged across each ecoregion unit. I only considered ecoregions that currently have at least three conifer species present to eliminate those ecoregions that contained a high diversity of species excluded from this study (Figure 4.1). Among the 20 dominant conifer tree species, I focused on the vulnerability of two of the most widely distributed, Douglas-fir and lodgepole pine. Whitebark pine was also selected because although its distribution is more limited, it is a keystone species in the subalpine zone providing an important food source for grizzly bears and contributing to watershed stability (Burns and Honkala, 1990).   51 Table 4.2 The 20 tree species and number of presence and absence plots utilized in this thesis.  Species Latin name Presence Absence % Present % Absent Total (n) Lodgepole pine Pinus contorta 9236 34168 21.3 78.7 43404 Douglas fir Pseudotsuga menziesii 12959 30445 29.9 70.1 43404 Piñon pine Pinus edulis 1387 42017 3.2 96.8 43404 Subalpine fir Abies lasiocarpa 5577 37827 12.8 87.2 43404 Engelmann spruce Picea engelmannii 4476 38928 10.3 89.7 43404 Whitebark pine Pinus albicaulis 1161 42243 2.7 97.3 43404 Western larch Larix occidentalis 1489 41915 3.4 96.6 43404 Pacific silver fir Abies amabilis 1119 42285 2.6 97.4 43404 White fir Abies concolor 2178 41226 5.0 95.0 43404 Grand fir Abies grandis 1816 41588 4.2 95.8 43404 Noble fir Abies procera 218 43186 0.5 99.5 43404 Utah juniper Juniperus osteosperma 1738 41666 4.0 96.0 43404 Sitka Spruce Picea sitchensis 374 43030 0.9 99.1 43404 Western white pine Pinus monticola 826 42578 1.9 98.1 43404 Alaska yellowcedar Callitropsis nootkatensis 740 42664 1.7 98.3 43404 Western redcedar Thuja plicata 2992 40412 6.9 93.1 43404 Western hemlock Tsuga heterophylla 3083 40321 7.1 92.9 43404 Mountain hemlock Tsuga mertensiana 773 42631 1.8 98.2 43404 Ponderosa pine Pinus ponderosa 5701 37703 13.1 86.9 43404 Quaking aspen Populus tremuloides 6190 37214 14.3 85.7 43404  4.3 Results Comparison of the current distribution of three tree species Douglas-fir, lodgepole pine and whitebark pine as defined from 1976 to 2009 to their projected suitable habitat from 2075 to 2100 using the most optimistic emission scenario that simulates the lowest GHG emissions (RCP2.6) is shown in Figure 4.2. The tree species are well distributed throughout the study region, with all three currently present in the Northwest Forested Mountains. Douglas-fir also occurs in the Marine West Coast Forest, with its range extending from British Columbia to California and further in the southwest. The species models projected that in another century, both lodgepole pine and whitebark pine will have much more constricted ranges in the Northwest   52 Forested Mountains. Douglas-fir on the other hand is likely to respond favorably to the projected climatic conditions with an expansion throughout this and other ecoregions. Figure 4.3 displays variations between the projected species distributions in 2075-2100 under the different emission scenarios.                    Figure 4.2 Current and future potential suitable habitat of lodgepole pine, Douglas-fir and whitebark pine from (a) 1976-2009 and (b) 2075-2100 under the RCP2.6 emission scenario.   53  Figure 4.3 Projected tree species abundances averaged from 2075 to 2100 under different emission scenarios for Douglas-fir, lodgepole pine, whitebark pine and sitka spruce.   The decision tree models developed for Douglas-fir, lodgepole pine and whitebark pine identified the relative importance of environmental constraints (ranked between 0 = no growth and 1 = optimum conditions for growth) in determining the presence and absence of the tree species (Figure 4.4). Accordingly, Douglas-fir is present on sites with favorable air temperatures during the spring (>0.23) and limited by both fall drought (>0.31) and annual frost (>0.48). Whitebark pine and lodgepole pine are limited by unfavorable high temperatures in the summer. Lodgepole pine also appears to depend on sufficient soil water recharge in the winter (>0.97), through snowpack accumulation and tolerates sites with frost below 0.41 in the fall. Soil water deficits imposed an important environmental constraint in two of the decision tree models in Figure 4.4.   0200000400000600000800000100000012000001400000160000018000002000000Douglas-fir Lodgepole pine Whitebark pine Sitka sprucekm2RCP 2.6RCP 4.5RCP 8.5  54                 Ecoregions with the largest projected decline in the distribution of dominant species are shown in Table 4.3. The ecoregions with the greatest species vulnerability are mostly located within the Northwestern Forested Mountains, including the Watson Highlands (6.1.5), the Southern Rockies (6.2.14) and the Idaho Batholith (6.2.15). The potential reduction in species distributions ranged from 1% to 64% (Table 4.3) Both lodgepole pine and western white pine showed the greatest decline in their predicted future distributions, whereas subalpine fir and Douglas-fir were most likely to become stressed within the ecoregions.  Figure 4.4 Decision tree models of Douglas-fir, lodgepole pine and whitebark pine ranking the relative importance of four environmental constraints (0 = no photosynthesis and 1 = optimum conditions for photosynthesis) that determine the presence and absence of tree species (ranked between 0 and 1). Temp = temperature modifier.   55  Table 4.3 Percent and absolute changes in the potential distribution of dominant species by ecoregion from 2075-2100 compared to their current range.  Ecoregion III code Ecoregion name Species Change in distribution Percent change (%) Net change (km2) 9.3.3 Northwestern Great Plains  Western white pine  -64 -47367 6.1.5 Watson Highlands Lodgepole pine -64 -75938 10.1.7 Arizona/New Mexico Plateau Ponderosa pine  -58 -17550 6.2.14 Southern Rockies Subalpine fir -35 -28063 6.2.13 Wasatch and Uinta Mountains Subalpine fir -22 -4882 6.2.15 Idaho Batholith Douglas-fir  -19 -10130 6.2.10 Middle Rockies Subalpine fir -8 -8396 10.1.4 Wyoming Basin Douglas-fir -1 -590 6.2.2 Chilcotin Ranges and Fraser Plateau Subalpine fir -1 -779  The areas where individual tree species were projected to expand or contract their range under different emission scenarios are shown in Figure 4.5. Douglas-fir displayed increased stress in the southern part of its distribution within the Northwest Forested Mountains and North American Deserts in the United States (Figure 4.5a). Areas within the northern part of its distribution were projected to become climatically suitable for species expansion, both in the Marine West Coast Forest along the coast of BC as well as on the Canadian side of the Northwest Forested Mountains in BC and Alberta, although this would ultimately depend on the establishment success and survival of the species. For the RCP 2.6 scenario, Douglas-fir had the lowest vulnerability ranking, with a 22% reduction in its current range (1976–2009). The RCP 8.5 scenario would, according to the model, cause up to a 59% reduction in the current range of Douglas-fir, whereas the RCP 4.5 scenario projected the greatest area for expansion (22%).  Both lodgepole pine and whitebark pine were predicted to become stressed throughout their entire current distribution under all emission scenarios (Figure 4.5b, d). Areas deemed unsuitable   56 for lodgepole pine ranged from 67% (RCP 2.6) to 81% (RCP 8.5) of its current range. These values serve as indicators of species vulnerabilities to environmental limitations and can be interpreted as a potential decline in species abundance rather than their disappearance from the landscape.  Sitka spruce, a species in close proximity to the Pacific Coast throughout the Marine West Coast Forest, is likely to be more limited in the southern part of its range in Washington, Oregon and California, while likely to expand northwest along the coast of BC under future environmental conditions (Figure 4.5c). About 80% of its current range remained suitable for the species under the RCP 2.6 scenario. The RCP 8.5 scenario projected both the greatest stress (39%) on Sitka spruce in its current range and most potential for range expansion (9%). Other species within the study region showed a variety of responses to future climatic projections (not shown). Many subalpine species were projected to become increasingly stressed where they occur within the Northwest Forested Mountains. These include subalpine fir and mountain hemlock (Tsuga mertensiana), which both are predicted to decline throughout their current ranges, although subalpine fir should remain present in parts of BC, Alberta and the Rocky Mountains in the United States. Decision tree analysis showed subalpine fir tolerates sites with low fall temperature and high summer frost while mountain hemlock had a high tolerance to annual frost and lower tolerance to spring VPD. While ponderosa pine may become highly stressed where it occurs within the North American Deserts, within the Northwest Forested Mountains some areas favored its expansion. Species such as western hemlock, western redcedar and Alaska yellow-cedar retained their habitat within the Marine West Coast Forest but were predicted to experience increased stress in parts of the Northwest Forested Mountains.   57                        (a) (b) (c) RCP2.6RCP4.5RCP8.5RCP2.6RCP4.5RCP8.5no changeRange expansion Stressed in current range R  .  R  .  R  .  R  .   .5  .5  nge (d) Figure 4.5 Areas deemed unsuitable (purple) or suitable (green) for (a) Douglas-fir, (b) lodgepole pine, (c) Sitka spruce and (d) whitebark pine under three future emission scenarios in 2075-2100 compared to their current distributions (grey) in 1976-2009.   58 The highest number of species was characterized as likely to be stressed within the North American Deserts (Figure 4.6). The Thompson-Okanagan Plateau (10.1.1) ecoregion was projected to have the highest vulnerability in the study area, with an average of six out of eight species stressed for all emission scenarios (66% of the original number of species present). Other ecoregions with high stress values included the Wyoming Basin (10.1.4), the Columbia Plateau (10.1.2), the Northern Basin and Range (10.1.3), the Colorado Plateaus (10.1.6), the Snake River Plain (10.1.8), the Central Basin and Range (10.1.5) and the Arizona/New Mexico Plateau (10.1.7) with averages of five out of seven species predicted as stressed in the future and relative values of 54–70%. Ecoregions within the Marine West Coast Forests appeared to be the least vulnerable to projected climate shifts from 2075 to 2100 compared to their current occurrence in 1976–2009. These include the Coast Range (7.1.8) and Coastal Western Hemlock-Sitka Spruce Forests (7.1.5) where less than 35% (two of six) of the species present were predicted to be vulnerable to future climatic conditions.             59             4.4 Discussion The projections of tree species vulnerabilities to climate shifts varied among ecoregions in this study. Over the coming century, the largest number of species stressed among the 20 included in the analysis was predicted to occur in the North American Deserts to the east of the Sierra Nevada and Cascade Mountain ranges in the western United States and British Columbia, Canada. During the growing season, this ecoregion experiences high air temperatures and a limited water supply, especially in the southern part (CEC, 1997). With projected future climate change, tree species that currently inhabit this environment will likely be reduced to those that can adapt to even more extreme conditions.  Figure 4.6 Predicted number of species stressed throughout the study regions for the three emission scenarios (a) RCP2.6, (b) RCP4.5 and (c) RCP8.5 in 2075-2100.   60 The majority of tree species analyzed that currently occupy the desert ecoregion were projected to be subject to increased stress, although ponderosa pine and Utah juniper were shown to retain suitable habitat in some areas. The Thompson-Okanagan Plateau in the Interior of BC, was projected to have the highest number of species stressed. With the climate there already among the warmest and driest in BC, endemic forest disturbances such as drought, wildfire and insect attacks are also set to increase in frequency and intensity (Wheaton, 2005; Haughian et al., 2012). In relative terms, the desert ecoregion ranked among the top ten in terms of species stress and ecoregions within the Northwest Forested Mountains experienced a high relative vulnerability as well.  Within the Northwest Forested Mountains, a range of responses were projected to occur. Some species were projected to become increasingly stressed, while others remained resilient and may even expand into new habitat that becomes climatically suitable. Montane and subalpine species such as lodgepole pine and whitebark pine were found to be especially vulnerable to climate change, with both species showing overall range contractions, consistent with Wang et al. (2012) climatic niche model projecting the largest reductions in extent for subalpine forest ecosystems. Species such as Douglas-fir, meanwhile, responded with both a contraction and expansion in their distribution within the Northwest Forested Mountains. The relative species stress in this ecoregion was as high as within the North American Deserts, indicating that a large proportion of original species present are vulnerable in both of these major ecoregions.  The Marine West Coast Forest, an ecoregion with a wet, mild climate contributing to the highest forest productivity in the study area, displayed the lowest species vulnerability. Although the   61 Marine West Coast Forest climate is projected to change the least among the ecoregions studied, increased precipitation could lead to more flooding and slope failures (Haughian et al., 2012). The projected distribution of Douglas-fir was maintained throughout most of the Marine West Coast Forest with a potential of expansion northward in British Columbia. Conditions remained suitable for Sitka spruce as well, except in the southern part of the ecoregion in Oregon, Washington and California where the species is projected to experience increased stress.  One of the benefits of the 3-PG model is that it cannot only predict where increased species vulnerabilities are likely but also can identify the environmental changes that cause stress (Landsberg et al., 2003). For instance, the decision tree models ranked spring temperature as the most limiting constraint on growth of Douglas-fir followed by annual frost. In the Interior Douglas-fir zone of British Columbia, Douglas-fir seedlings were found to be sensitive to frost during the growing season and air temperature was recognized as an important limitation on seedling survival (Heineman et al., 2003). Douglas-fir is a species known to be well adapted to both the mild, wet climate of the Marine West Coast Forest as well as to drier conditions in the Northwest Forested Mountains, although it is still vulnerable to excessively warm and dry conditions (Klinka et al., 1999). Douglas-fir projections from this chapter resulted in a similar trend to other climate niche studies that predicted habitat to become increasingly suitable in the northern parts of their range (Gray and Hamann, 2013; Rehfeldt et al., 2014). In this study, greater stress was projected in southern part of their range in the US, which may be partly explained by the greater warming trend projected in the Canadian GCM. Results from this chapter also showed that Sitka spruce was most limited by frost conditions during the fall. It grows well in waterlogged soils but has a low tolerance to both extreme temperatures and frost   62 and does not tolerate drought well (Klinka et al., 1999). This relatively shade tolerant species has a good chance to dominate sites unsuitable for other species (Burns and Honkala, 1990). Hybridization of the species with white spruce (Picea sitchensis × P. glauca) increased its frost tolerance and could improve the adaptive capacity of the population in a changing environment (Hamilton and Aitken 2013; Aitken and Bemmels, 2016).  In the Northwest Forested Mountains ecoregion, the decision tree model ranked summer temperatures as the most limiting modifiers for lodgepole pine, indicating that the projected increase in air temperatures with climate change would be the chief stressor on the species. Soil water conditions in the winter and frost in the fall also affected predicted lodgepole pine distribution. In recent years, an increase in both winter temperatures and summer drought have led to conditions that increased the vulnerability of lodgepole pine to the mountain pine beetle outbreak that has killed an unprecedented number of trees in western Canada and the United States (Carroll et al., 2006; Safranyik and Wilson, 2006). The large areas classified as stressed for the species can be interpreted in terms of increased physiological limitations imposed on the species that may lead to a decline in their abundance. Wang et al. (2010) projected that the height of lodgepole pine populations in BC was greater when planted with optimal populations at a given site than for current populations growing in their native provenances. While building decision tree models for specific populations could provide additional insight on the climatic adaptations of different subspecies, the focus of this thesis was to assess tree species responses at the species-level and focus on 20 different species rather than a few select populations.    63 Whitebark pine, meanwhile, had a highly variable response to altered climatic conditions, with results in this chapter showing that the most important climate constraint on whitebark pine was summer temperature. The current decline of the whitebark pine throughout its distribution is of great concern, and government agencies have devised conservation strategies to help protect and restore the species in both Canada and the United States (Haughian et al., 2012; Hansen and Phillips, 2015). Concern over whitebark pine decline is in part due to its role as a keystone species, providing an important food source for animals and facilitating forest growth in alpine habitats (Burns and Honkala, 1990). Whitebark pine has been affected by both white pine blister rust and the mountain pine beetle, with those trees weakened by summer drought and inhabiting high altitudes most affected (Tomback and Resler, 2007; Haughian et al., 2012). Similar to findings from this chapter, other studies project a continued decline of whitebark pine over the next century with climate warming (Schrag et al., 2008; Chang et al., 2014; Warwell et al., 2006; McLane and Aitken, 2012). Climate-induced disturbances such as an increase in the frequency and size of fires as well as changing species competitors may also impact whitebark pine in Montana (Loehman et al., 2011). Knowledge of the climate and soil factors that stress whitebark pine as detailed in this study can help to develop conservation strategies to reduce the threats of climate change.  Projections from this chapter agreed with other studies that predicted increased stress of species in the southern part of their ranges coupled with potential range expansions to higher latitudes and altitudes (Iverson and Prasad, 1998; Hamann and Wang, 2006; McKenney et al., 2007). Similar to findings from this chapter, a number of studies that have reported increased species die-off in the North American Deserts in recent years, although these studies generally focus on   64 areas within the southern United States rather than considering the full range of tree species (Mueller et al., 2005; Ganey and Vojta, 2011; Moore et al., 2016). Recent tree mortality has been observed in the mixed-conifer forests of Arizona that are part of the North American Deserts ecoregion (Ganey and Vojta, 2011). Ma et al. (2015) similarly reported that in Australia, the greatest vulnerability of ecosystems to drought occurs in semi-arid regions. In general, studies have projected that there would be an overall expansion of desert biomes while temperate forest regions will expand northward, becoming replaced by shrublands in their southern ranges (Rehfeldt et al., 2012; Jiang et al., 2013). An analogous trend was also predicted in the mountains of Colorado and Utah, where conifer species would become partially replaced by grasslands except at the highest elevations where tree cover was expected to increase (Notaro et al., 2012). Studies elsewhere in the world also indicated increased vulnerability of subalpine species to climate warming (Thuiller et al., 2005; Rehfeldt et al., 2012; Hansen and Phillips, 2015). Indeed, an overall decline in suitable habitat for subalpine and montane species such as whitebark and lodgepole pine has been projected by the end of the 21st century in a number of studies that used bioclimate envelope models (Rehfeldt et al., 2012; Gray and Hamann, 2013; Bell et al., 2014).  Unlike many of the above empirical studies, the 3-PG model allowed for the identification of the underlying climate constraints that cause species vulnerabilities. In the case of subalpine species, warmer winters, earlier snowmelt and drier summers all contribute to tree species stress. Soil properties also have a well-recognized influence on the growth and distribution of tree species although they have still seen a limited inclusion in species distribution models (Iverson and Prasad, 1998; Syphard and Franklin, 2009; Mathys et al., 2014). By incorporating spatially   65 variable soil information in the species models, additional knowledge can be gained on areas that are suitable for range expansion and where stress is expected to increase a species’ vulnerability. The soil water modifier was an important environmental constraint in this study that helped account for much of the variation in the majority of the decision tree models. Among the 20 conifer species evaluated in this study, 75% responded to changes in soil water-holding capacity, with up to a 27% change in their distribution when varying this property in the models (Mathys et al., 2014).  Adding spatial constraints to the rate of migration provides a more realistic picture of the implications of rapid changes in climate. It is clearly unrealistic, without human assistance, to accommodate the conclusion that many tree species would need to migrate 100–250 km per decade to keep up with the rate of projected climate warming (Iverson and Prasad, 1998; McKenney et al., 2007; Gray and Hamann, 2013), a distance that exceeds historic postglacial migration rates during the Quaternary (Davis and Shaw, 2001; Aitken et al., 2008). In recognition of the inability of species to migrate rapidly, foresters are considering assisting the movement of species into more suitable areas, while expanding, where possible, their genetic diversity (McLachlan et al., 2007; Hoegh-Guldberg et al., 2008). The limits imposed on species migrations in this study, therefore, helped to classify locations that may be suitable for natural range expansion. The suitable habitat of some of the tree species can be 50–90% lower when migration is constrained compared to an assumption of uninhibited migration (Coops et al., 2016). Curtailing the rates and distances of tree species migration allowed identification of areas that are not only projected to become climatically suitable for each of the species assessed, but also where species are unlikely able to reach from their current locations. The predictions of   66 future tree species distributions from this study can serve as guidelines to distinguish areas where increased stress is projected to occur from those that are more resilient to future climate change. These results can help to focus assisted migration efforts on suitable locations that are beyond the natural dispersal limits of tree species.   67 5. Changes in seedling and tree distributions indicate emerging shifts in forest composition3 5.1 Introduction Forests in British Columbia (BC) encompass 60 million hectares and provide many important ecosystem services. Some of these ecological benefits include providing habitat for wildlife and maintaining a clean water and air supply for human survival (Spittlehouse, 2008). Many local communities also depend on these forests for timber supply as well as the cultural and spiritual values they provide. Maintaining these services requires an improved understanding on how BC forests are responding to a changing climate. Over recent decades the climate has been shifting and affecting forest ecosystems (IPCC, 2014). Warmer temperature and changing precipitation patterns are often associated with increased forest disturbances (Woods et al., 2010; Westerling et al., 2006; Raffa et al., 2008; Wheaton, 2005; Heineman et al., 2010). For example, increased bark beetle activity such as the unprecedented recent mountain pine beetle attack on forests in BC has been linked with climate change (Raffa et al., 2008). A drought in Interior BC in 2001-2003 led to severe fire occurrences (Wheaton, 2005). While such disturbances pose stress on tree species, they also create potential for species range expansions to areas that are becoming climatically suitable.  British Columbia is a potential hot spot for conifer migrations as many species have their northern range limit in the province. At the same time, BC contains ecoregions that harbor some                                                  3 A version of chapter 5 has been submitted for publication. Mathys, A.S., Coops, N.C., Simard, S.W., Waring, R.H., Aitken, S.N., 2017. Predicting tree species responses to climate shifts: Integrating an empirical regeneration database with physiological modeling.   68 of the most vulnerable species to climate change in western North America (Mathys et al., 2017). With increases in fire, insect and disease outbreaks, it is crucial to understand what effects climate change is having on forest composition. One way to investigate this is to assess the occurrence of tree seedlings, as they can serve as early indicators of species range expansions or contractions. Comparing tree species presences over time may provide insights as to those species, most likely to be affected by a changing climate and the types of forests that might emerge. Differences between the habitats that mature trees and seedlings occupy indicate where a species might be expected to contract or expand its range. The forest regeneration phase is of particular interest to foresters as the successful establishment of both newly arriving and existing species depends on their regeneration potential and ability to persist under changing environmental conditions (Bose et al., 2016). Furthermore, forest management practices have the most influence over species composition trajectories at the seedling phase.   Most studies use empirical models based on simple relationships between species occurrences and climate without addressing the mechanistic factors that lead to species vulnerabilities (Kerr et al., 2007). Process-based models that incorporate physiological principles tested in field experiments allow an integrated assessment of the influence of climatically induced limitations on the distribution of tree species and of how changes in these variables might affect species growth. This chapter employed the process-based model 3-PG as described in Chapter 1.3. Following procedures in Chapter 3, a decision tree analysis combined the modeled environmental constraints from 3-PG with field observations of species occurrences, thus providing insights on the external factors that influence the distribution of species, both as seedlings and mature trees. Using a process-based approach thus allows for the prediction of   69 areas suitable for species expansion or contraction based on those environmental factors that limit species growth.  Information on forest regeneration was obtained from RESULTS (Reporting Silviculture Updates and Land Status Tracking System), a large empirical database provided by the BC Ministry of Forests, Lands and Natural Resource Operations. RESULTS collects, manages and distributes information on disturbances, silvicultural operations and forest cover in BC (BC Ministry of Forests, Lands and Natural Resource Operations, 2014). The forest cover inventory report used in this study is part of the Silviculture and Land Status Tracking dataset and contains spatial tree species regeneration data from across the province. Obtaining this empirical dataset on regeneration cover in BC allowed for contrasting the distribution of seedlings to that of mature trees to gain an idea of emerging forest responses to climate change and to compare these responses with model predictions of the effects of climate shifts on tree species (Chapter 3).  Chapter 3 mapped the baseline distribution of tree species using climate and soil information with an average accuracy of 84%. The objective of this study was to determine emerging shifts in species distributions linked to changing environmental limitations on species growth. To accomplish this, I applied an innovative approach to relate the recorded change in distribution of tree seedlings established since 2000 with areas predicted as climatically suitable for tree species range expansions or contractions in BC. The modeled environmental controls on species distributions during a cooler, wetter period from 1950-75 were compared with a more recent warmer and drier one from 2000-2009 (Waring et al., 2014), and model results were tested against the large empirical forest regeneration database from RESULTS. This provides an   70 improved understanding of the stresses tree species are exposed to under recent climate change compared to earlier baseline conditions in the past.  5.2 Methods 5.2.1 Study area British Columbia is the most biologically diverse province in Canada, with distinct ecosystems ranging from coastal forests to alpine tundra (Meidinger and Pojar, 1991). It is part of the Pacific Northwest region with borders extending from south of the Yukon and Northwest Territories to Washington, Idaho and Montana in the US, and Alberta to the east (Figure 5.1). The complex topography ranges from the Coast Mountains and Southern Rockies to the Interior Plateau and Great Plains (Valentine et al., 1978). The Pacific Ocean and mountain ranges greatly influence the diverse climate and precipitation patterns throughout the province (Meidinger and Pojar, 1991). The climate varies from mild and wet along the coast to cold winters and hot summers in the Interior of BC. The Coast Mountains cast a rain shadow leading to very dry conditions in the south-central Interior of BC. Coniferous forests are the dominant vegetation type throughout the province. In this study, I focused on four widely distributed and economically important coniferous species in the province: Douglas-fir, lodgepole pine, western larch and subalpine fir.   Douglas-fir is distributed along the Marine West Coast Forest ecoregion of BC where the coastal variety grows in temperate forests that are among the densest and most productive in BC (Klinka et al., 1999). Interior Douglas-fir populations exist in montane forests in a continental climate within the Northwest Forested Mountains. They also grow in the dry forests of the Thompson-  71 Okanagan Plateau often with a grassy understory. At higher elevations, interior Douglas-fir can occur in mixed forests together with lodgepole pine.  Lodgepole pine grows in montane and subalpine forests throughout the Northwest Forested Mountains in BC. It is the one of the most widely distributed species in Canada and an important timbre source (Klinka et al., 1999). Lodgepole pine is a pioneer species that often regenerates in even-aged stands following fires or less frequently in mixed-species forests. It can also occur in cool, temperate climates of southern BC in association with western larch.  Western larch is present in montane forests of Southeastern BC in both the Thompson-Okanagan Plateau and Columbia Mountains. Here it grows predominantly in mixed-species stands and is nearly absent on very moist sites (Klinka et al., 1999). This deciduous conifer is of high ecological and economic value and considered the most productive species of the Larix genus in North America (Rehfeldt and Jaquish, 2010).   Subalpine fir is another important timber species that occurs mainly in subalpine forests. It is distributed throughout the Northwestern Forested Mountain in BC in continental climates and is nearly absent in warmer and drier climates (Klinka et al., 1999). Subalpine fir is a dominant species at higher elevations and tolerates cold conditions with a heavy snowpack.       72                5.2.2 Sampling species distributions Information on the presence or absence of different species of seedlings, representing individuals <1.3 m in height, were acquired from the RESULTS database by the BC Ministry of Forests, Lands and Natural Resource Operations. The data were obtained from forest openings throughout BC, representing disturbances, silvicultural activities and free growing stands, collected and submitted by individuals funded by government programs and administered by the RESULTS online database. The data were selected from the RESULTS forest cover inventory table by counting every species recording of the uneven-aged regeneration layer as a seedling Figure 5.1 Study area in British Columbia and location of field survey plots of the mature trees (purple) and seedlings (green).   73 presence. Plot locations were estimated from centroids of polygons representing forest openings with an accuracy of about 500 m. A total of 21,097 plots were surveyed since 2000. information on tree species, now well-established sapling and pole-stage trees, were also obtained as described in Section 2.2.2. Distributions of the survey plots for trees and seedlings are presented in Figure 5.1.  5.2.3 Predicting environmental constraints on species distributions I predicted the environmental constraints on species growth using the 3-PG model, following procedures detailed in previous chapters. The predicted suitable or unsuitable areas for tree species were compared with the presence and absence of seedling data from RESULTS recorded since 2000. For establishment conditions of current mature trees, I used the climate growth modifiers produced from a baseline climate record from 1950-1975, and for seedlings I used a more recent period from 2000-2009 to represent conditions experienced by seedlings in the RESULTS regeneration database. The seasonal averages of the climate modifiers were extracted for each of the field plots to build separate decision tree models for trees and for seedlings in BC (DTREG software; Sherrod, 2010) following the approach described in Section 3.2.2. These species models allowed mapping the presence and absence of trees and seedlings based on the relative importance of seasonal modifiers as derived from the decision tree analysis. Constraints on growth were predicted for Douglas-fir and then applied to other species by describing how their tolerances deviated from optimal conditions of Douglas-fir. A new set of tree species models were produced that differ from those used in Chapters 3 and 4, using surveyed tree occurrences within BC to reflect the climate constraints experienced in the province. The model   74 accuracies were evaluated using a 10-fold cross-validation technique (Breiman et al., 1984) and a kappa statistic as described in Section 3.2.2 (Table 5.1).  I predicted species range expansions or contractions by comparing the potential distribution of trees in 2000-2009 to their baseline distribution in 1950-1975 using the same vulnerability threshold of 70% from Chapter 4. In this study, species stress was defined by the environmental limitations posed on photosynthesis. A limit of 200 m/year was imposed on the distance a tree species can migrate (Mathys et al., 2017; Davis, 1989; Beckage et al., 2008). The percent agreement of projected species range expansions and contractions were finally compared with the suitable habitat of seedlings as sampled from the RESULTS database. Areas predicted as suitable for tree range expansion were compared with the tree seedlings presences and areas deemed unsuitable with the seedling absences.  Table 5.1 Decision tree model accuracies for the trees during the period 1950-1975 and for the seedlings during 2000-2009. Brackets represent the 95% confidence intervals.   Presence accuracy (%) Overall accuracy (%) κ Species Tree Seedling Tree Seedling Tree Seedling Subalpine fir 90 (±2.0) 81 (±1.1) 62 (±1.0) 72 (±0.6) 0.8147 0.6634 Douglas fir 78 (±1.6) 91 (±0.5) 72 (±0.9) 75 (±0.6) 0.615 0.7079 Western larch 94 (±3.4) 86 (±1.2) 83 (±0.8) 69 (±0.6) 0.9299 0.7631 Lodgepole pine 78 (±1.7) 79 (±0.8) 61 (±1.0) 67 (±0.6) 0.515 0.4398   5.3 Results The importance of the seasonal 3-PG modifiers, averaged across all plots, differed among species during baseline conditions in 1950-1975 (Figure 5.2). These importance values differ   75 from the previous chapter as the growth constraints were considered specifically for tree species established within the province of BC. Both Douglas-fir and lodgepole pine were constrained mainly by suboptimal air temperature and frost in the winter when these modifiers restricted growth by over 85%. Lodgepole pine is widely distributed in Interior BC and adapted to cold winters (Klinka et al., 1999). The species was not severally limited by soil water and evaporative demand in the summer; these variables were predicted to restrict photosynthesis by 38% and 29%, respectively. On sites occupied by Douglas-fir, growth was also constrained by suboptimal temperatures and frost in the spring and fall, but by less than 70% because major limitations were imposed in the winter. During the summer time, soil water constraints imposed a 39% reduction and VPD a 33% reduction on potential growth. In higher elevation-forests, subalpine fir is adapted to low temperatures in the spring and fall that can restrict growth by over 82%. Limitations by drought and evaporative demand are <30% where subalpine fir occurred in BC during baseline conditions. The native range of western larch is in southeastern BC in the Cordilleran region, where soil water deficits during the summer limited photosynthesis by up to 55%. Frost in the spring and fall can reduce growth of the species by over 20%.            76                  Figure 5.3 shows the seasonal modifiers that displayed the greatest mean difference between the two time periods (1950-1975 and 2000-2009) at the plot occurrences of all species averaged across their full range in BC. The most pronounced difference was summer soil water that became more constraining for the seedlings under current climate conditions than for the earlier period when the mature trees established. Summer VPD was also more constrained in 2000-2009 than during baseline conditions. Springtime temperatures and frost became less limiting on the species between these periods, although on average they remained the most constraining environmental factors.  Figure 5.2 Mean seasonal variation in climate modifiers of lodgepole pine, Douglas-fir, subalpine fir and western larch during baseline conditions in 1950-75 (0 = no growth and 1 = optimum conditions for growth).   77                The species-specific changes in average seasonal modifiers are shown in Figure 5.4. Both lodgepole pine and subalpine fir had the greatest change in modifier values experiencing less constraints from spring temperature and frost, but increased limitations of summer soil water and evaporative demand in 2000-2009 compared to 1950-1975. Habitats favorable for these two species registered the largest relative increase in the effect of drought in the summer, whereas those areas dominated by western larch experienced the least. In regard to limitations by spring frost, conditions improved the most for subalpine fir and changed the least for Douglas-fir. All species experienced a rise in spring temperatures across their current ranges, which was most Figure 5.3 Shifts in mean (dots), median (line) and interquartile range (box) between the two periods (1950-75 and 2000-09) for four seasonally-defined climatic growth modifiers averaged across all species.   78 rapid for subalpine fir and lodgepole pine and least for Douglas-fir. In contrast, summer vapor pressure deficits increased the most for lodgepole pine and the least in areas favourable for western larch.                 The vulnerability of tree species to changing climate in 2000-2009 compared to their previous baseline condition are shown in Figure 5.5, illustrating climatically-suitable areas for species expansion as well as where a species may experience increased stress. Lodgepole pine was predicted to become increasingly stressed throughout its historic range with changing climate Figure 5.4 Changes in seasonal climate growth modifiers for Douglas-fir, lodgepole pine, subalpine fir and western larch between 1950-1975 and 2000-2009. The error bars represent the standard error. Negative values indicate that the modifier is becoming less limiting.   79 conditions (Figure 5.5a). Approximately 4% of its baseline distribution was classified as no longer suitable with some areas in Interior BC (1%) becoming climatically suitable for range expansion (Table 5.2). Western larch displayed the greatest potential for range expansion (10%) of all species analysed in this study (Figure 5.5b; Table 5.2). The species was projected to expand northwest from its historic range in Interior BC as conditions improved climatically and displayed only limited stress under recent climate conditions. The suitable habitat of recently established seedlings from 2000-2009 is shown for lodgepole pine (Figure 5.5c) and western larch (Figure 5.5a).  To assess model performance, the areas of predicted species range expansions and contractions were compared with habitat favorable for seedlings (Table 5.2). In general, there was good agreement between areas where trees were projected to expand or contract their range and the recorded distribution of tree seedlings from the RESULTS database. The areas deemed suitable for expansion agreed 79% on average with successful seedling establishment and areas classified as stressed had an average agreement of 77% with areas where the seedlings were absent.           80                        Figure 5.5 Potential for range expansion (green) and vulnerability (purple) of a) lodgepole pine and b) western larch based on 2000-2009 climatic conditions compared to conditions during 1950-1975 and suitable habitat for seedlings of c) lodgepole pine and d) western larch.   81 Table 5.2 Projected species expansion and stress and agreement with area deemed suitable or unsuitable for regeneration growth (2000-2009).   Expansion (%) Stress (%) Species Projected Agreement Projected Agreement Douglas-fir 5 92 1 99 Lodgepole pine 1 62 4 82 Western larch 10 67 2 99 Subalpine fir 0 97 16 25 Average 4 79 6 77  5.4 Discussion Estimates of species vulnerabilities from this chapter agree well in most cases with observed seedling establishment (Table 5.2). Differences in the distribution and climate exposures of seedlings compared to those of trees established last century provide some evidence that species expansions are already occurring in response to climate change (Lenoir et al., 2009). Results from this chapter indicate that warmer temperatures, reduced frost and increased drought in recent years have caused shifts in forest composition in BC.   Both lodgepole pine and subalpine fir experienced the greatest changes in environmental constraints on sites where they currently occur and also had the most areas classified as stressed in the model predictions. In the past, sites occupied by these species experienced only limited soil water deficits; that situation is now changing. Other studies note that climate change is particularly felt at high altitudes, where tree species are most sensitive to changing conditions (Lenoir et al., 2009; Kullman, 2007). Although lodgepole pine is a widely distributed species adapted to a range of environments (Klinka et al., 1999), warmer temperatures and reduced frost can create an environment for both competing species and damaging biotic agents to thrive. In   82 recent year, lodgepole pine forests have experienced unprecedented mountain pine beetle outbreaks in BC (Carroll et al., 2006; Safranyik and Wilson, 2006; Mather et al., 2010; Heineman et al., 2010) as well as Dothistroma needle blight that has been associated with climate change (Woods et al., 2010). Such disturbances are affecting the forest health of lodgepole pine stands in Southern Interior BC (Mather et al., 2010; Heineman et al., 2010) and may limit lodgepole pine habitat throughout BC in the future (Monserud et al., 2008; McKenney et al., 2007; Mathys et al., 2017; Hamann and Wang, 2006).  Sites occupied by subalpine fir experienced the greatest increase in temperatures since 2000 of all species analysed in this study. This species is well adapted to frost and generally occurs on wetter sites (Klinka et al., 1999), but today is experiencing greater soil water deficits. This poses stress to subalpine fir, as it is generally absent on dry sites and has higher transpiration rates compared to other species (Kaufmann et al., 1982). The distribution of subalpine fir seedlings displayed the lowest agreement with areas we predicted as stressed. The lower number of tree plots compared with those for seedlings may have led to a higher accuracy of the regeneration predictions (Table 5.1). In any case, the analysis in this chapter suggests that the projected stress at these sites is not yet sufficiently severe to induce mortality of the species. Nonetheless, subalpine fir has been increasingly prone to insect attack in recent years. Maclauchlan (2016) reported an increase in subalpine fir susceptibility to the western balsam bark beetle over the past two decades, with up to a 70 % increase in mortality recorded in some parts of southern BC. As conditions become increasingly unsuitable for subalpine species an opportunity is provided for their replacement by more temperate species such as Douglas-fir and western larch, if migration can occur fast enough.    83 Seedling results from this chapter indicate that Douglas-fir has the opportunity and capacity to expand northward. Environmental conditions in BC are improving for the species as soils become drier and frost less frequent during the active growing season (Klinka et al., 1999). Other studies also projected an expansion of suitable habitat for more southern species such as Douglas-fir and ponderosa pine northward and upward (Rehfeldt et al., 2014; Gray and Hamann, 2013). Their projections displayed an even greater northward expansion of Douglas-fir range within the Northwest Forested Mountains than results from this study although at a lower frequency (<5%) (Gray and Hamann, 2013; Rehfeldt et al., 2014).  Western larch displayed the greatest capacity to expand its range of the four species studied. Increasing air temperatures and reduced snowpack are likely providing favorable conditions for the species to migrate in BC. The suitable habitat of western larch has been previously predicted to expand northward in BC (Rehfeldt and Jaquish, 2010) and management practices now allow planting small amount of western larch north and west of its current distribution (Jaquish, 2010). Areas classified as suitable for western larch expansion in this chapter agreed well with Rehfeldt and Jaquish (2010) mapped guidelines for seed transfers although this chapter also predicted areas further northwest in the Thompson-Okanagan Plateau to become suitable.   This study applied an innovative approach that identifies emerging species shifts by comparing the recorded species distribution and responses over contrasting periods. Empirical studies that have analysed the distribution of trees and seedlings generally report migrations of species to higher altitudes and latitudes (Woodall et al., 2009; Monleon and Lintz, 2015; Lenoir et al., 2009), although caution has been expressed to use the distribution of two life stages as evidence   84 of species range shifts (Sittaro et al., 2017). In the dry portions of western US, the distribution of seedlings was predicted to decline, especially in subalpine areas (Bell et al., 2014). Unlike these empirical studies, the approach in this chapter allowed to not only assess divergences in tree and seedling distributions but also to compare the physiological attributes that are limiting the species today and in the past.   Using a process-based approach helped to gain an understanding of some of the causes of species stress that may be useful for developing forest management practices. All species analyzed in this study are becoming increasingly limited by soil water deficits, as environmental factors such as low temperatures are becoming less constraining with climate change. This greater exposure to summer drought will be important to monitor to determine species ability to tolerate such changing conditions. Species that are more drought-adapted such as Douglas-fir will have a competitive advantage as warming trends continue. Experimental studies in BC have found that seedling survival and growth of a number of species was greatest in forest openings with the highest light availability and greatest soil moisture contents (Wright et al., 1998; Vyse et al., 2006). Protection will be required especially for young trees that still need to develop resistance to environmental stresses such as drought (Niinemets, 2010; Fredericksen et al., 1996). Thinning treatments in forest stands can increase the soil water supply to the remaining trees in areas that are drought prone. Site preparation techniques such as removing competing understory cover can also reduce water stress especially at the seedling stage (Heineman et al., 2003).  Studies have highlighted temperature changes as important drivers that can lead to species expansions to colder habitats (Monleon and Lintz, 2015) and species contractions under   85 excessively warm conditions (Bell et al., 2014). This chapter identified temperature along with frost to be the most limiting factors on species growth both under current and baseline conditions. Provenance trials support this finding, as low temperatures have driven adaptation of temperate and boreal species with a weaker differentiation of drought detected among populations (Aitken and Bemmels, 2016; Alberto et al., 2013). As low temperatures become less constraining, they could facilitate migrations of certain species. In general, earlier dates of last spring or first fall frost increase the available growing season length, but warm conditions can also lead to earlier snowmelt, reduced soil water storage and greater transpiration.   The greatest predicted change in frost occurred during the springtime on sites occupied by the mature trees compared to those of the seedlings although some studies also highlight the importance of fall frost (Alberto et al., 2013; Aitken and Bemmels, 2016). In this study, the spring frost modifier was an important predictor in the decision tree model for most species, explaining its importance in determining the presence or absence to the species. This importance of spring frost could be partly explained by the 3-PG model being driven by a radiation budget, with the greatest incoming radiation occurring in the springtime thus increasing the importance of frost on photosynthesis at this time of the year. Spring frost can affect the accumulation of leaf area and reduce growth in the same year (Vitasse et al., 2009). Nonetheless, it is important to consider the interacting phenological processes during both the spring and fall rather than analyzing them separately (Way, 2011). Early fall frost can cause injury leading to reduced growth in the following year (Norby et al., 2003; Skomarkova et al., 2006). Details on the phenological events of bud flush and bud set are complex mechanisms that require knowledge of species genetic differences and were beyond the scope of this study.   86 This chapter highlights how environmental limitations on growth are affecting different species as climatic conditions are changing in BC. Knowledge of the crucial stressors on tree species and where to expect potential for species range expansions can be useful for forest managers developing mitigation practices to climate change. Continuous monitoring of forests would help to provide an improved understanding of climate change impacts on tree mortality and regeneration. Including forest disturbances in species distribution models should improve predictions of stress that are expected to increase with climate change (Malmström and Raffa, 2000) by for example incorporating projections of fire occurrences (Waring and Coops, 2015) or insect attack and disease (Woods et al., 2010; Mather et al., 2010). Furthermore, differentiating soil water access for seedlings and trees would refine the model predictions, as this chapter highlighted the increased importance of drought on species growth. For example, young ponderosa pine trees have been found to be more sensitive to drought earlier in the season compared to older trees (Irvine et al., 2002). Accounting for differences in rooting depth with tree age can contribute to improving predictions of species responses from regeneration to mature tree stage.   87 6. Conclusion 6.1 Research innovations Across western North America, major changes in tree species distributions are currently underway. Forest communities depend on trees for their ecological and economic benefits and outcomes of this thesis can contribute to a better understanding on tree species responses to climate change. The main goal of this PhD thesis was to provide improved projections of current and future tree species distributions under a changing climate and to identify the physiological mechanisms that are driving species’ stress. Some of the important contributions of this thesis that advance current knowledge are detailed below.    Highlighting the importance of soil water information for more accurate mapping of tree species distributions (Chapter 3)  Providing improved current and future species distribution models that account for variations in climate and soil (Chapter 3,4,5)  Identifying ecological regions in western North America where tree species are most vulnerable to climate change (Chapter 4)  Assessing emerging shifts in tree species distributions with new environmental conditions using an innovative approach (Chapter 5)  The next section provides answers to each of the following research questions that contributed to the outcomes of this thesis:     88 1) How does soil water availability affect the distribution of 20 conifer tree species? 2) How do projected tree species vulnerabilities vary across the ecoregions of western North America?  3) Are climate shifts leading to emerging changes in tree seedlings distributions across British Columbia?   Question 1: How does soil water availability affect the distribution of 20 conifer tree species? In Chapter 3, I identified an important research gap in current species distribution studies that generally focus on species responses to climate variations without addressing differences in local soil properties (Syphard and Franklin, 2009). Current models often lack adequate soil data and an understanding of the mechanisms controlling species distributions. The importance of soil water availability in particular was investigated by incorporating a recently developed soil map that accounts for variations in ASWC (Coops et al., 2012) to calculate the soil water modifier in the species distribution models. To accomplish this, I undertook a sensitivity analysis that assessed changes in productivity and distributions with varying soil water inputs. ASWC was increased and decreased by 50% from the originally mapped values to evaluate the effects on predicted species distributions. Soil water availability helped explain the variation in the distribution of 75% of the tree species. It was found that 30% of the species were very to extremely sensitive to changes in ASWC, while 45% were somewhat sensitive. These outcomes indicate that knowledge of soil properties generally improves overall accuracy of species distribution models. The sensitivity analysis of Chapter 3 identified the species most sensitive to changes in water availability, and indicated where additional information on soil properties would be most critical   89 to verify. By allowing soil properties as well as climate to vary across western North America, the species models predicted the occurrence of 20 tree species with an average accuracy of 84% (κ = 0.79).   Results of Chapter 3 highlight the importance of soil water availability in successful species occurrence over the landscape. They emphasize the sensitivity of certain tree species to changes in ASWC and others that are adapted to fluctuations. Uncertainties in data inputs can have a large impact on predicted distributions of certain species as shown in the sensitivity analysis. Improved knowledge of species requirements in terms of water availability and tolerance to drought will be beneficial to improving the accuracy of model predictions in a changing climate scenario. This will help forest managers and ecologists to maintain forest productivity and adopt silvicultural practices that can sustain forest resources in western North America.  Question 2: How do projected tree species vulnerabilities vary across the ecoregions of western North America? Chapter 4 addressed this research question by applying the species models developed in Chapter 3 to project species vulnerabilities in the future using variable climate and soil properties. An improved understanding is needed on species variations to projected climate change in the ecoregions throughout their entire range. Most studies apply empirical methods to predict species distributions with limited knowledge of the physiological controls that can limit species growth (Pearson and Dawson, 2003). In Chapter 4, the 3-PG model was employed to provide estimates of responses of 20 tree species to changes in environmental conditions and to evaluate the extent that species are resilient to shifts in climate over the rest of this century. Projected suitable areas   90 for tree species were compared to their current ranges based on observations at >40 000 field survey plots and tree species were classified as vulnerable if environmental conditions projected in the future appeared outside that of their current distribution ≥70% of the time. A migration constraint of <200 m yr-1 was applied that limits species dispersal to provide more realistic projections on species distributions. Based on these combinations of constraints, the greatest changes in the distribution of dominant tree species were predicted to occur within the Northwest Forested Mountains and the highest number of tree species stressed will likely be in the North American Deserts. Projected climatic changes appeared especially unfavorable for species in the subalpine zone, where major shifts in composition may lead to the emergence of new forest types. A decline of these species will free sites for the invasion of tree species that are better adapted to changing climatic conditions.  The model predictions presented in Chapter 4 serve as a basis to predict where the potential for tree migration is strongest. Imposing limits on the rates that species can migrate provides insights into their ability to respond to climate change naturally and where forest management practices may be necessary to assist their migration. Chapter 4 also provided mechanistic understandings of the climate and soil constraints that affect species vulnerabilities, thus allowing forest managers to better devise adaptation strategies to climate change. In other ecoregions, changes in species community composition are expected with new forest types emerging as some species become more abundant than others (Williams et al., 2004). While the projections remain to be confirmed, they provide a framework for understanding where to focus conservation efforts to build forest resilience and which species are likely to persist in a future climate.    91 Question 3: Are climate shifts leading to emerging changes in tree seedlings distributions across British Columbia? The composition and health of British Columbia’s forests have been manifesting signs of change over the last half-century associated with altered climate conditions. Projected shifts in tree species distributions are often derived from the long-term responses of mature trees, however under a dynamic climate, recently established seedlings may more closely reflect recent changes in climate conditions. Chapter 5 compared shifts in distribution of tree species in BC in reference to how they might perform as seedlings between two contrasting time periods: 1950-1975 and 2000-2009. This chapter first predicted recent expansion and contractions of tree species based on deviations from their baseline distribution from 1950-1975, following the approach to assess species vulnerabilities in Chapter 4. Chapter 5 then applied an innovative approach that compared the species projections with an extensive empirical database from 21,097 regeneration plots in BC (RESULTS), sampled since 2000. Areas suitable for species range expansions corresponded with successful seedling establishment on average 79%, indicating that species migrations are beginning to occur with climate change in BC. Decision tree analysis based on species occurrences provided insight into species-specific changes in environmental limitations experienced during the two periods. In agreement with previous research, I found that as suboptimal temperatures and frost become less constraining, increased drought and atmospheric humidity deficits are becoming more prevalent as limiting factors. Tree responses varied by species, with habitats favourable for lodgepole pine experiencing the largest relative increase in summer drought and areas dominated by western larch experiencing the least.    92 Chapter 5 highlights how environmental limitations on growth are affecting different species as climatic conditions are changing in BC. The most important contribution of Chapter 5 is perhaps the approach itself, where the distribution of two life stages in a species are compared in reference to their appearance or absence on recorded plots over contrasting periods, representing subtle, but progressive shifts in climatic conditions.   6.2 Limitations of research The predictions of current and future tree species distributions provided in this thesis can serve as guidelines to distinguish areas where increased stress might be expected to occur from those that are more resilient to climate change. The extent of the distributions themselves should not be assumed accurate due to uncertainties surrounding different climate model projections, the rate of disturbances, habitat fragmentation and the availability of seed sources (Corlett and Westcott, 2013). The extent that tree species ranges will change depends on complex, interactive processes such as species interactions, genetic adaptation, CO2 fertilization effects, disturbance regimes and land use change.   6.2.1 Limitations in 3-PG modeling framework Notable limitations in the modeling framework used in this thesis include employing data inputs at monthly time steps, which limits the ability to accurately assess the timing of phenological events and of hydrological processes that occur over shorter time intervals. Almeida and Sands (2015) fine-tuned the 3-PG model with a more detailed soil water balance that uses daily climate data. Applying this model requires more data inputs on soil physical properties and also   93 significantly increases the computational requirements given the spatial and temporal scale of the species analysis in this thesis.  The 3-PG model was parameterized only for Douglas-fir and the physiological limitations on other species were then described in relation to how they depart from optimal conditions for photosynthesis of Douglas-fir. Ideally, the parameterization would be carried out for each species individually, however the comprehensive data required for this task is only available for the most widely studied species. The version of the 3-PG model used in this thesis is also only applicable for conifer trees, although these are the dominant tree species found in the study region of western North America. Recently, the model has been modified to account for deciduous and mixed species forests by adjusting for differences in light absorption and within-canopy vertical gradients in climate (Forrester and Tang, 2016).  Changes in snowpack are also important and are currently only indirectly deduced with the 3-PG model. Days with low temperatures and precipitation that exceeds losses through evapotranspiration are indicators of a heavy snowpack accumulating in certain locations. These sites can be identified in the decision tree models as they rarely experience a high vapor pressure deficit, low soil moisture or temperatures favorable for growth in the wintertime. These conditions are especially important for subalpine species that normally occur on sites with the presence of a heavy snowpack in the winter.      94 6.2.2 Non-climatic factors controlling species shifts Disturbances caused by insect, fires, and disease are processes interrelated with climate and soil and are currently not considered in the species models. Incorporating these into future species distribution models is beneficial when determining the presence of tree species and the effects of changing climate and soil conditions (Malmström and Raffa, 2000). Species migrations can also be affected by land use change caused by humans as well as natural barriers, which were not included in this thesis. The effects of land cover change vary by ecosystem type, with coastal species and those in early successional stages being more adapted to human-caused habitat fragmentation than slower growing, more shade tolerant species (Opdam and Wascher, 2004). Attempts have been made to incorporate land cover data into species distribution models to assess the effects of fragmented landscapes on species migrations (Iverson et al., 2004; Thuiller et al., 2005; Coops et al., 2016).  In this thesis, limits were imposed on the distance a species can migrate by selecting a maximum migration constraint of 200 m per year (Mathys et al., 2017; Coops et al., 2016; Davis, 1989; Beckage et al., 2008). While a more mechanistic understanding of the constraints on species migration would be desirable, the additional detail required would likely increase the complexity and uncertainty of the model predictions (Thuiller et al., 2008). Many studies chose to omit using species dispersal distances due to limited data available on the rate that species might disperse (Parmesan and Yohe, 2003), the extent the landscape has become fragmented or experienced a change in land use (Opdam and Wascher, 2004; Pearson and Dawson, 2005). Choosing a simple approach allowed to address species migration rates as a measure of adaptive capacity by accounting for the distances they can potentially migrate.   95 Tree seedling distributions are not only affected by growth but also by germination and establishment success (Blanco et al., 2009), factors that are currently omitted from the 3-PG model. Nonetheless, the tree seedling inventory used in this thesis reflected the occurrences of those species that had successfully become established and survived in their natural environment. Continuous monitoring of the seedlings would allow testing their performance under future climatic stress. The RESULTS database focused on areas that have undergone silvicultural activities and that have been planted with seedlings. It would be valuable to also have information available on natural regeneration in BC to compare with the model predictions. Still, the current dataset was useful in providing insights on the actual inventory of regeneration in the province as influenced by forest management practices.  6.3 Research applications in forest management The results of this thesis have important applications in forest management. Without extraordinary effort by foresters and geneticists, the low natural migration rates (2 km per decade) will prevent most species of reaching areas favorable to their growth and survival. In the ecoregions where increased stress is most likely, it may thus be advisable to plant tree species that are better able to persist in a future climate. Knowledge on the species and ecoregions that are most vulnerable to climate shifts enables land managers to concentrate their efforts to provide refugia and to assist tree species migration. Outcomes to this thesis can contribute to existing assisted migration efforts in Canada and the United States that aim to increase genetic diversity by planting species or populations most adapted to future climates (O’Neill et al., 2008; Aitken et al., 2008; Rehfeldt et al., 2010). Target locations within ecoregions were identified that have the greatest probability of current and future species declines or expansions and where   96 management effort can focus on to increasing forest resilience and reduce their susceptibility to insect and pests. These species vulnerability maps are provided in Appendix A.2 and are publically available at www.databasin.org. Decision frameworks also need to include risk assessments to address the uncertainties, concerns and benefits related to assisted migration efforts (Hewitt et al., 2011; Hoegh-Guldberg et al., 2008; Ricchiardi and Simberloff, 2009). Increasing habitat connectivity can help species migrate to new areas that are projected to favor their growth (Jongman et al., 2011). Planting a mixture of species rather than monocultures can contribute to avoiding stress and mortality within forest stands under changing environmental conditions that are projected to increase in the future.  Outcomes of this thesis can also contribute to forest management practices by providing physiological explanation for species responses to climate change that are beginning to occur. Drought has been identified as a major limitation on species growth and management efforts can target those ecoregions with the greatest water limitations. Thinning treatments in forest stands can increase the soil water supply to the remaining trees in areas that are drought prone. Site preparation techniques such as removing competing understory cover can also reduce water stress especially at the seedling stage (Heineman et al., 2003).   In general, applying a management strategy at the landscape-level rather than the traditional stand-level strategy is crucial especially when responding to climate change effects on forest ecosystems (Mah et al., 2012). Such a landscape management strategy considers the interactions between species, disturbance events and climate shifts and accounts for the uncertainties associated with species responses to future climate conditions. Measuring the success of the   97 adaptive strategy requires an improved monitoring system that tracks continuous changes in species composition as well as in forest health and productivity.  6.4 Directions for future research Projections of species distributions can be continuously improved with access to quality datasets that drive the models. Current knowledge of climate change impacts on forest ecosystems is often based on model projections and implementing forest monitoring programs could provide field based evidence of climate effects on tree mortality and regeneration. This could be achieved by establishing permanent sample plots at large spatial and temporal scales to continuously track tree species responses to climate change and gain a better understanding of forest health and productivity. Such monitoring programs would benefit from cooperation across political boundaries to provide consistent datasets throughout the full range of a species. It may be challenging to implement such a monitoring program at a macroscale and future work may benefit from employing public participation techniques as a means to gather vast field collection while also engaging the public on this important issue.  Access to independent datasets would allow for improved validation of the model predictions in addition to the cross-validation technique used in this thesis. Expanding the approach in Chapter 5 that compared projected species vulnerabilities with an independent seedling database in BC, to across North America would enable to monitor continuous changes in forest composition in the field. Tree species vulnerabilities could also be compared with forest disturbances detected through remote sensing, similar to the approach by Waring et al. (2011). Furthermore, the performance of the 3-PG model could be assessed by conducting a sensitivity analysis with all   98 environmental constraints on species growth, similar to the soil water modifier assessment in Chapter 3.  Increasing the quality of soil maps at wide spatial scales can also improve the accuracy of species model predictions. As remote sensing technologies continue to advance, more accurate remotely sensed estimates of LAI could be achieved, which in turn provide better estimates of soil properties. Airborne Light Detection and Ranging (LiDAR) can capture the full spectrum of LAI and quantify values exceeding 7 m2 m-2, unlike the MODIS-derived estimates used in this thesis that cannot identify LAI values that extend to 10 m2 m-2 and higher. (Lefsky et al., 2002). The soil maps discussed in Chapter 3 were likely unable to portray the most productive and fertile sites in western North America and increased accessibility to LiDAR data would contribute to providing more accurate regional soil information.  There is potential to make model refinements by distinguishing species responses at different life stages and by including forest disturbances in the species projections. For example, soil water access for seedlings and trees could be differentiated, as this thesis highlights the increased importance of drought on species growth. Irvine et al. (2002) found that young ponderosa pine trees were more sensitive to drought earlier in the season compared to older trees. Accounting for differences in rooting depth with tree age can contribute to improving predictions of species responses from regeneration to mature tree stage.   Including forest disturbances in species distribution models should also improve predictions of stress that are expected to increase with climate change (Malmström and Raffa, 2000) by for   99 example incorporating projections of fire occurrences (Waring and Coops, 2015) or insect attack (Woods et al., 2010). In general, a better understanding of the environmental thresholds that cause tree mortality would improve the ability to recognize species that are most vulnerable to climate change (Park et al., 2014; Allen et al., 2015).  Advancements in the field will clearly benefit from collaborations between experimental and modeling studies across the fields of ecology, physiology, genetics, ecosystem modeling and global change research. A multitude of approaches are encouraged to improve models, including those incorporating interactions with rising atmospheric concentrations of CO2 (Waring and Gao, 2016) and genetic effects (Wang et al., 2010). It is also desirable to seek more representative data on species growth, distribution, and site characteristics. The approach introduced in this thesis that relates recorded species distributions during contrasting periods (Chapter 5) provides potential to track regeneration responses as climate change progresses. I hope to see this approach extended and tested more widely in western North America and beyond.   100 References Adams, H.D., Williams, A.P., Xu, C., Rauscher, S.A., McDowell, N. G., 2013. Empirical and process-based approaches to climate-induced forest mortality models. Frontiers in Functional Plant Ecology 4, 438. http://doi.org/10.3389/fpls.2013.00438  Ågren, G.I., 1996. Nitrogen productivity of photosynthesis minus respiration to calculate plant growth. Oikos 76, 529-535.  Aitken, S.N., Yeaman, S., Holliday, J.A., Wang, T., Curtis-McLane, S., 2008. Adaptation, migration or extirpation: climate change outcomes for tree populations. Evolutionary Applications 1, 95–111.  Aitken, S.N., Bemmels, J.B., 2016. Time to get moving: Assisted gene flow of forest trees. Evolutionary Applications 9, 271–290.  Alberto, F.J., Aitken, S.N., Alía, R., González-Martínez, S.C., Hänninen, H., Kremer, A., Lefèvre, F., Lenormand, T., Yeaman, S., Whetten, R., Savolainen, O., 2013. Potential for evolutionary responses to climate change - evidence from tree populations. Global Change Biology 19(6), 1645–1661.  Alexander, R.R., Shepperd, W.D., 1990. Picea engelmannii Parry ex Engelm., In: Burns, R.M., Honkala, B.H. (Eds.), Silvics of North America, Volume 1, Conifers. Agriculture Handbook 654. USDA Forest Service, Washington DC, pp. 187–203.   101 Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H., Gonzalez, P., Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.H., Allard, G., Running, S.W., Semerci, A., Cobb, N., 2010. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecology and Management 259 (4), 660–684.  Allen, C.D., Breshears, D.D., McDowell, N.G., 2015. On underestimation of global vulnera- bility to tree mortality and forest die-off from hotter drought in the Anthropocene. Ecosphere 6, 1–55.  Almeida, A.C., Landsberg, J.J., Sands, P.J., 2004. Parameterisation of 3-PG model for fast-growing Eucalyptus grandis plantations. Forest Ecology and Management 193, 179-95.  Almeida, A.C., Sands, P.J., 2015. Improving the ability of 3-PG to model the water balance of forest plantations in contrasting environments. Ecohydrology 9, 610-630.  Amiro, B.D., Barr, A.G., Barr, J.G., Black, T.A., Bracho, R., Brown, M., Chen, J., Clark, K.L., Davis, K.J., Desai, A.R., , Dore, S., Engel, V., Fuentes, J.D., Goldstein, A.H., Goulden, M.L., Kolb, T.E., Lavigne, M.B., Law, B.E., Margolis, H.A., Martin, T., McCaughey, J.H., Misson, L., Montes-Helu, M., Noormets, A. Randerson, J.T., Starr, G., Xiao, J., 2010. Ecosystem carbon dioxide fluxes after disturbance in forests of North America. Journal of Geophysical Research 115, G00K02.    102 Anderegg, W.R., Hicke, J.A., Fisher, R.A., Allen, C.D.,Aukema,J., Bentz, B., Hood, S., Lichstein, J.W., Macalady, A.K., McDowell, N., Pan, Y., Raffa, K., Sala, A., Shaw, J.D., Stephenson, N.L., Tague, C., Zeppel, M.., 2015. Tree mortality from drought, insects, and their interactions in a changing climate. New Phytologist 208, 674–683.  Asner, G.P., Brodrick, P.G., Anderson, C.B., Vaughn, N., Knapp, D.E., Martin, R.E., 2016. Progressive forest canopy water loss during the 2012–2015 California drought. Proceedings of the National Academy of Sciences 113, E249–E255.  Bartelink, H.H., 2000. Effects of stand composition and thinning in mixed-species forests: a modeling approach applied to Douglas-fir and beech. Tree Physiology 20, 399–406.  B.C. Ministry of Forests, Lands and Natural Resource Operations, 2014. Results information submission specifications, ed. 4a, Online at www.for.gov.bc.ca/HFP/publications/00220/RISS_gf_Edition_4a_2014_May.pdf (19 January, 2017).  Bechtold, W.A., Patterson, P.L., 2005. The enhanced forest inventory and analysis national sample design and estimation procedures. General Technical Report SRS-80. USDA Forest Service, Southern Research Station, Ashville, NC.    103 Beckage, B., Osborne, B., Gavin, D.G., Pucko, C., Siccama, T., Perkins, T., 2008. A rapid upward shift of a forest ecotone during 40 years of warming in the Green Mountains of Vermont. Proceedings of the National Academy of Sciences 105, 4197–4202.  Bell, D.M., Bradford, J.B., Lauenroth, W.K., 2014 Mountain landscapes offer few opportunities for high-elevation tree species migration. Global Change Biology 20, 1441–1451.  Bell, D.M., Bradford, J.B., Lauenroth, W.K., 2014. Early indicators of change: divergent climate envelopes between tree life stages imply range shifts in the western United States. Global Ecology and Biogeography 23, 168–180.  Ben-David, A., 2008. Comparison of classification accuracy using Cohen’s weighted kappa. Expert Systems with Applications 34, 825–832.  Blanco, J., Welham, C., Kimmins, J., 2009. Guidelines for modeling natural regeneration in boreal forests. Forest Chronicles 85, 427–439.  Bradford, J.B., Bell, D.M., 2017. A window of opportunity for climate-change adaptation: easing tree mortality by reducing forest basal area. Frontiers in Ecology and the Environment 15, 11-17.  Breiman, L., Friedman, J.H., Olshen, R.A., Stone, C.G., 1984. Classification and regression trees. Wadsworth International Group, Belmont, CA.    104 Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., Romme, W.H., Kastens, J.H., Floyd, M.L., Belnap, J., Anderson, J.J., Myers, O.B., Meyer, C.W., 2005. Regional vegetation die-off in response to global-change-type drought. Proceedings of the National Academy of Sciences of the United States of America 102, 15144–15148.  Brown, M.G., Black, T.A., Nesic, Z., Fredeen, A.L., Foord, V.N., Spittlehouse, D.L., Bowler, R., Burton, P.J., Trofymow, J.A., Grant, N.J., Lessard, D., 2012. The carbon balance of two lodgepole pine stands recovering from mountain pine beetle attack in British Columbia. Agricultural and Forest Meteorology 153, 82–93.  Burns, R.M., Honkala, B.H., 1990. Silvics of North America, Volume 1, Conifers. Agriculture Handbook 654. USDA Forest Service, Washington DC, USA.  Canadell, J.G., Le Quéré, C., Raupach, M.R., Field, C.B., Buitenhuis, E.T., Ciais, P., Conway, T.J., Gillett, N.P., Houghton, R.A., Marland, G., 2007. Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intensity, and efficiency of natural sinks. Proceedings of the National Academy of Sciences of the United States of America 104, 18866–18870.  Carroll, A.L., Regniere, J., Logan, J.A., Taylor, S.W., Bentz, B.J., Powell, J.A., 2006. Impacts of climate change on range expansion by the mountain pine beetle. Mountain Pine Beetle Initiative Working Paper 2006–14, Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre.   105 Cayan, D.R., Das, T., Pierce, D.W., Barnett, T. P., Tyree, M., Gershunov, A., 2010. Future dryness in the southwest US and the hydrology of the early 21st century drought. Proceedings of the National Academy of Sciences of the United States of America 107, 21271–76.  Chang, T., Hansen, A.J., Piekielek, N., 2014. Patterns and variability of projected biocli- mate habitat for Pinus albicaulis in the Greater Yellowstone Ecosystem. PLoS One 9, e111669.  Coates, K.D., Canham, C.D., Beaudet, M. Sachs, D.L., Messier, C., 2003. Use of a spatially explicit individual-tree model (SORTIE-BC) to explore the implications of patchiness in structurally complex forests. Forest Ecology and Management 186, 279-310.  Commission for Environmental Cooperation, 1997. Ecological regions of North America: toward a common perspective. Commission for Environmental Cooperation, Montreal, Quebec, Canada.  Cook, B.I., Ault, T.R., Smerdon, J.E., 2015. Unprecedented 21st century drought risk in the American Southwest and Central Plains. Science Advances 1, e1400082.  Coops, N.C., Waring, R.H., Landsberg, J.J., 1998. Assessing forest productivity in Australia and New Zealand using a physiologically-based model driven with averaged monthly weather data and satellite derived estimates of canopy photosynthetic capacity. Forest Ecology and Management 104, 113–127.    106 Coops, N.C., Waring, R.H., 2001. Assessing forest growth across southwestern Oregon under a range of current and future global change scenarios using a process model, 3-PG. Global Change Biology 7, 15–29.  Coops, N.C., Waring, R.H., Schroeder, T.A., 2009. Combining a generic process productivity model and a statistical classification method to predict presence and absence of tree species in the Pacific Northwest USA. Ecological Modelling 220, 1787–1796.  Coops, N.C., Waring, R.H., 2011. Estimating the vulnerability of fifteen tree species under changing climate in Northwest North America. Ecological Modelling 222 (13), 2119–2129.  Coops, N.C., Waring, R.H., Hilker, T., 2012. Prediction of soil properties using a process-based forest growth model to match satellite-derived estimates of leaf area index. Remote Sensing of Environment 126, 160–173.  Coops, N.C., Waring, R.H., Plowright, A., Lee, J., Dilts, T.E., 2016. Using remotely-sensed land cover and distribution modeling to estimate tree species migration in the Pacific Northwest Region of North America. Remote Sensing 8, 65.  Corlett, R.T., Westcott, D.A., 2013. Will plant movements keep up with climate change? Trends in Ecology & Evolution 28, 482–488.     107 Crookston, N.L., Rehfeldt, G.E., Dixon, G.E., Weiskittel, A.R., 2010. Addressing climate change in the forest vegetation simulator to assess impacts on landscape forest dynamics. Forest Ecology and Management 260, 1198–1211.  Cuddington, K., Fortin, M., Gerber, L.R., Hastings, A., Liebhold, A., O’Connor, M., Ray, C., 2013. Process-based models are required to manage ecological systems in a changing world. Ecosphere 4, 1–12.  Daly, C., Halbleib, M., Smith, J.I., Gibson, W.P., Doggett, M.K., Taylor, G.H., Curtis, J., Pasteris, P.P., 2008. Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. International Journal of Climatology 28, 2031–2064.  Davis, M.B., 1989. Lags in vegetation response to greenhouse warming. Climatic Change 15, 75–82.  Davis, M.B., Shaw, R.G., 2001. Range shifts and adaptive responses to Quaternary climate change. Science 292, 673–679.  DeHayes, D.H., Jacobson, G.L., Schaber, P.G., Bongarten, B., Iverson, L.R., Dieffenbacker-Krall, A., 2000. Forest responses to changing climate: lessons from the past and uncertainty for the future. In: Mickler, R.A., Birdsey, R.A., Hom, J.L. (Eds.), Responses of Northern Forests to   108 Environmental Change. Springer-Verlag, New York, NY, (ecological studies series), pp. 495–540.  Ecological Stratification Working Group, ESWG, 1996. A national ecological framework for Canada. Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources Research and Environment Canada, State of Environment Directorate, Ottawa/Hull.  Elith, J., Leathwick, J.R., 2009. Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics 40, 677–697.  Ellenberg, H., 1988. Vegetation ecology of Central Europe, fourth ed. Cambridge University Press, Cambridge.  Farrar, J.L., 1995. Trees in Canada. Natural Resources Canada, Canadian Forest Service, Headquarters, Ottawa, copubl. by Fitzhenry and Whiteside Limited, Markham, Ontario.  Forrester, D.I., Tang, X., 2016. Analysing the spatial and temporal dynamics of species interactions in mixed-species forests and the effects of stand density using the 3-PG model. Ecological Modelling 319, 233–254.  Franklin, J.F., Dyrness, C.T., 1973. Natural vegetation of Oregon and Washington. United States Forest Service General Technical Report PNW-8. Portland, Oregon, USA.   109 Fredericksen, T.S., Steiner, K.C., Skelly, J.M., Joyce, B.J., Kolb, T.E., Kouterick, K.B., Ferdinand, J.A., 1996. Diel and seasonal patterns of leaf gas exchange and xylem water potentials of different-sized Prunus serotina Ehrh. trees. Forest Science 42, 359–365.  Fu, P., Rich, P.M., 2002. A geometric solar radiation model with applications in agriculture and forestry. Computers and Electronics in Agriculture 37, 25–35.  Ganey, J.L., Vojta, S.C., 2011. Tree mortality in drought-stressed mixed-conifer and ponderosa pine forests, Arizona, USA. Forest Ecology and Management 261, 162–168.  Gonzalez-Benecke, C.A., Teskey, R.O., Martin, T.A., Jokela, E.J., Fox, T.R., Kane, M.B., Noormets, A., 2016. Regional validation and improved parameterization of the 3-PG model for Pinus taeda stands, Forest Ecology and Management 361, 237-256.  Graham, R.T., McCaffrey, S., Jain, T.B. (tech. eds.), 2004. Science basis for changing forest structure to modify wildfire behavior and severity. General Technical Report RMRS-GTR-120. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, CO.  Gray, L.K., Hamann, A., 2013. Tracking suitable habitat for tree populations under climate change in western North America. Climatic Change 117, 289–303.  Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more than simple habitat models. Ecology Letters 8 (9), 993–1009.   110 Hamann, A., Wang, T.L., 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agriculture and Forest Meteorology, 128, 211– 221.  Hamann, A., Wang, T., 2006. Potential effects of climate change on ecosystem and tree species distribution in British Columbia. Ecology 87, 2773–2786.  Hamilton, J.A., Aitken, S.N., 2013. Genetic and morphological structure of a spruce hybrid (Picea sitchensis x P. glauca) zone along a climatic gradient. American Journal of Botany 100, 1651–1662.  Hamlet, A.F., Mote, P.W., Clark, M.P., Lettenmaier, D.P., 2007. Twentieth-century trends in runoff, evapotranspiration, and soil moisture in the western United States. Journal of Climate 20, 1468–1486.  Hansen, A.J., Phillips, L.B., 2015. Which tree species and biome types are most vulnerable to climate change in the US Northern Rocky Mountains? Forest Ecology and Management 338, 68–83.  Harlow, W.M., Harrar, E.S., 1950. Textbook of Dendrology. McGraw-Hill Book Company Inc., New York.  Harvey, A. E., Byler, J.W., Mcdonald, G.I., Neuenschwander, L.F., Tonn, J.R., 2008. Death of an Ecosystem: Perspectives on Western White Pine Ecosystems of North America at the End of   111 the Twentieth Century. General Technical Report RMRS-GTR-208. USDA Forest Service, Rocky Mountain Research Station, Fort Collins CO.  Hastie, T., Tibshirani, R., Friedman, J.H., 2009. The elements of statistical learning: data mining, inference, and prediction (2nd edn). Springer-Verlag, New York.  Haughian, S.R., Burton, P.J., Taylor, S.W., Curry, C., 2012. Expected effects of climate change on forest disturbance regimes in British Columbia. BC Journal of Ecosystems and Management 13, 1–24.  Heineman, J., Hope, G., Simard, S., Vyse, A., Lloyd, D., Miege, D., 2003. The effects of site preparation and harvesting practices on planted seedling productivity and microenvironment in southern interior dry, grassy IDF forests. Technical Report 009. B.C. Ministry of Forests, Research Branch, Victoria, BC.  Heineman, J.L., Sachs, D.L., Mather, W.J., Simard, S.W., 2010. Investigating the influence of climate, site, location, and treatment factors on damage to young lodgepole pine in southern British Columbia. Canadian Journal of Forest Research 40, 1109-1127.  Hember, R.A., Coops, N.C., Spittlehouse D.L., 2017. Spatial and temporal variability of potential evaporation across North American forests. Hydrology 5, doi: 10.3390/hydrology4010005.    112 Hewitt, N., Klenk, N., Smith, A.L., Bazely, D.R., Yan, N., Wood, S., MacLellan, J.I., Lipsig- Mumme, C., Henriques, I., 2011. Taking stock of the assisted migration debate. Biological Conservation 144, 2560–72.  Hoegh-Guldberg, O., Hughes, L., McIntyre, S., Lindenmayer, D.B., Parmesan, C., Possingham, H.P., Thomas, C.D., 2008. Assisted colonization and rapid climate change. Science 321, 345–346  Hogg, E.H., Brandt, J.P., Michaelian, M., 2008. Impacts of a regional drought on the productivity, dieback, and biomass of western Canadian aspen forests. Canadian Journal of Forest Research 38, 1373-1384.  IPCC, 2007. Climate change 2007: the physical science basis. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Contribution of Working Group I to the Fourth Assessment, Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom/New York, NY, USA, p. 996.  IPCC, 2014. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. In: Field, C.B., Barros, V.R., Dokken, D.J., Mach, K.J., Mastrandrea, M.D., Bilir, T.E., Chatterjee, M., Ebi, K.L., Estrada, Y.O., Genova, R.C., Girma, B., Kissel, E.S., Levy, A.N., MacCracken, S., Mastrandrea, P.R., White, L.L. (Eds.), Contribution of Working Group II   113 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK., pp. 271–359.  Irvine, J., Law, B.E., Anthoni, P.M., Meinzer, F.C., 2002. Water limitations to carbon exchange in old-growth and young ponderosa pine stands. Tree Physiology 22, 189–196.  Iverson, L.R., Prasad, A.M., 1998. Predicting abundance of 80 tree species following cli- mate change in the eastern United States. Ecological Monographs 68, 465–485.  Iverson, L.R., Prasad, A.M., Matthews, S.N., Peters, M., 2008. Estimating potential habitat for 134 eastern US tree species under six climate scenarios. Forest Ecology and Management 254, 390–406.  Jaquish, B., 2010. Review of British Columbia’s contemporary and projected western larch seed planning zones in light of climate change. B.C. Ministry of Forests and Range, Kalamalka Forestry Centre, Research and Knowledge Management Branch, Vernon, B.C.  Jiang, X., Rauscher, S.A., Ringler, T.D., Lawrence, D.M., Williams, A.P., Allen, C.D., Steiner, A.L., Cai, D.M., McDowell, N.G., 2013. Projected future changes in vegetation in western North America in the twenty-first Century. Journal of Climate 26, 3671–3687.  Jongman, R.H.G., Bouwma, I.M., Griffioen, A., Jones-Walters, L., Van Doorn, A.M., 2011. The pan European ecological network: PEEN. Landscape Ecology 26, 311–326.   114 Kaufmann, M.R., Edminster, C.B., Troendle, C.A., 1982. Leaf area determination for three subalpine trees species in the central Rocky Mountains. Research Paper RM-238. USDA Forest Service, Rocky Mountain Forest and Range Experiment Station, Fort Collins CO.  Kearney, M., Porter, W., 2009. Mechanistic niche modelling: combining physiological and spatial data to predict species’ ranges. Ecology Letters 12, 334–350.  Kerr, J.T., Kharouba, H.M., Currie, D.J., 2007. The macroecological contribution to global change solutions. Science 316, 1581–1584.  Kimball, J.S., Running, S.W., Nemani, R., 1997. An improved method for estimating surface humidity from daily minimum temperature. Agricultural and Forest Meteorology 85, 87–98.  Kimmins, J.P., Mailly, D., Seely, B., 1999. Modelling forest ecosystem net primary production: the hybrid simulation approach used in FORECAST. Ecological Modelling 122, 195-224.  Klinka, K., Worrall, J., Skoda, L., Varga, P., Krajina, V.J., 1999. The distribution and synopsis of ecological and silvical characteristics of tree species in British Columbia’s forests (2nd edn). Canadian Cartographics Ltd, Coquitlam, BC, Canada.  Krishnan, P., Black, T., Grant, N., Barr, a, Hogg, E., Jassal, R., Morgenstern, K., 2006. Impact of changing soil moisture distribution on net ecosystem productivity of a boreal aspen forest during and following drought. Agriculture and Forest Meteorology 139, 208–223.    115 Kullman, L., 2007. Tree line population monitoring of Pinus sylvestris in the Swedish Scandes, 1973-2005: implications for tree line theory and climate change ecology. Journal of Ecology 95, 41-52.  Landsberg, J.J., Waring, R.H., 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., Waring, R.H., Coops, N.C., 2003. Performance of the forest productivity model 3-PG applied to a wide range of forest types. Forest Ecology Management 172, 199–214.  Landsberg, J., Sands, P., 2011. Physiological Ecology of Forest Production: Principles, Processes and Modeling. London Academic Press, Oxford, UK, Elsevier Inc.  Latta, G., Temesgen, H., Adams, D., Barrett, T., 2010. Analysis of potential impacts of climate change on forests of the United States Pacific Northwest. Forest Ecology and Management 259, 720–729.  Law, B.E., Waring, R.H., Anthoni, P.M., Aber, J.D., 2000. Measurements of gross and net productivity and water vapor exchange of a Pinus ponderosa ecosystem, and an evaluation of two generalized models. Global Change Biology 6, 155–168.    116 Lefsky, M.A., Cohen, W.B., Parker, G., Harding, D.J., 2002. Lidar remote sensing for ecosystem studies. BioScience 52, 19–30.  Lenoir, J., Gégout, J.-C., Pierrat, J.-C., Bontemps, J.-D., Dhôte, J.-F., 2009. Differences between tree species seedling and adult altitudinal distribution in mountain forests during the recent warm period (1986–2006). Ecography 32, 765–777.  Levin, S.A., 1998. Ecosystems and the biosphere as complex adaptive systems. Ecosystems 1, 431-436.   Little, E.L., Jr., 1971. Atlas of United States Trees, Volume 1: conifers and important hardwoods. U.S. Department of Agriculture Miscellaneous Publication 1146, vol. 9. www.esp.cr.usgs.gov/data/atlas/little (accessed 10 August 2013).  Loehman, R.A., Corrow, A., Keane, R.E., 2011. Modeling climate changes and wildfire interactions: effects on whitebark pine (pinus albicaulis) and implications for restoration, glacier national park, Montana, USA. In: Keane, R.E., Tomback, D.F., Murray, M.P., Smith, C.M. (Eds.), The Future of High-Elevation, Five-Needle White Pines in Western North America: Proceedings of the High Five Symposium RMRS-P-63. USDA Forest Service, Rocky Mountain Research Station, Missoula MT, pp. 176–189.  Ludlow, A.R., Randle, T.J. Grace., J.C., 1990. Developing a process-based model for Sitka spruce. In: Dixon, R.K., Meldahl, R.S., Ruark, G.A., Warren, W.G. (Eds.), Process Modeling of   117 Forest Growth Responses to Environmental Stress. Timber Press Inc., Portland, OR, pp. 249–262.  Ma, X., Huete, A., Moran, S., Ponce-Campos, G., Eamus, D., 2015. Abrupt shifts in phenology and vegetation productivity under climate extremes. Journal of Geophysical Research Biogeosciences 120, 2036–2052.  Maclauchlan, L., 2016. Quantification of Dryocoetes confusus-caused mortality in subalpine fir forests of southern British Columbia. Forest Ecology and Management 359, 210–220.   Mah, S., Astridge, K., DeLong, C., Wickland, C., Todd, M., McAuley, L., LePage, P., Coates, D., Heemskerk, B., Banner, A., Hall, E., 2012. A landscape-level species strategy for forest management in British Columbia: exploration of development and implementation issues. Technical Report 067. B.C. Ministry of Forests, Lands and Natural Resource Operations, Victoria, B.C.  Mäkelä, A., Landsberg, J., Ek, A.R., Burk, T.E., Ter-Mikaelian, M., Ågren, G.I., Oliver, C.D., Puttonen, P., 2000. Process-based models for forest ecosystem management: current state of the art and challenges for practical implementation. Tree Physiology 20, 289–298.  Malcolm, J.R., Markhma, A., Neilson, R.P., Garaci, M., 2002. Estimated migration rates under scenarios of global climate change. Journal of Biogeography 29, 835–849.    118 Malmström, C., Raffa, K., 2000. Biotic disturbance agents in the boreal forest: considerations for vegetation change models. Global Change Biology 6, 35–48.  Mather, W.J., Simard, S.W., Heineman, J.L, Sachs, D.L., 2010. Decline of young lodgepole pine in southern interior British Columbia. Forestry Chronicle 86, 484-497.  Mathys, A., Black, T.A., Nesic, Z., Nishio, G., Brown, M., Spittlehouse, D.L., Fredeen, A.L., Bowler, R., Jassal, R.S., Grant, N.J., Burton, P.J., Trofymow, J.A., Meyer, G., 2013. Carbon balance of a partially harvested mixed conifer forest following mountain pine beetle attack and its comparison to a clear-cut. Biogeosciences 10 (8), 5451–5463.  Mathys, A., Coops, N.C., Waring, R.H., 2014. Soil water availability effects on the distribution of 20 tree species in western North America. Forest Ecology and Management 313, 144–152.  Mathys, A.S., Coops, N.C., Waring, R.H., 2017. An ecoregion assessment of projected tree species vulnerabilities in western North America through the 21st century. Global Change Biology 23, 920–932.  McDowell, N., W. T. Pockman, C. D. Allen, D. D. Breshears, N. Cobb, T. Kolb, Plaut, J., Sperry, J., West, A., Williams, D.G., Yepez, E.A., 2008. Mechanisms of plant survival and mortality during drought: why do some plants survive while others succumb to drought? New Phytologist 178, 719–739.    119 McKenney, D.W., Pedlar, J.H., Lawrence, K., Campbell, K., Hutchinson, M.F., 2007. Potential impacts of climate change on the distribution of North American Trees. BioScience 57 (11), 939–948.  McLachlan, J. S., Clark, J. S., 2004. Reconstructing historical ranges with fossil data at continental scales. Forest Ecology and Management 197 (1–3), 139–147.   McLachlan, J.S., Hellmann, J.J., Schwartz, M.W., 2007. A framework for debate of assisted migration in an era of climate change. Conservation Biology 21, 297–302.  McLane, S.C., Aitken, S.N., 2012. Whitebark pine (Pinus albicaulis) assisted migration potential: Testing establishment north of the species range. Ecological Applications 22, 142–153.   McLaughlin, S.P., 1986. Floristic analysis of the southwestern United States. Western North American Naturalist 46, 46–65.  Meidinger, D., Pojar, J., 1991. Ecosystems of British Columbia. Special Report Series No. 6. BC Ministry of Forests, Victoria, BC.  Melillo, J.M., Borchers, J., Chaney, J., Fisher, H., Fox, S., Haxeltine, A., Janetos, A., 1995. Vegetation/ecosystem modeling and analysis project (VEMAP): comparing biogeography and biogeochemistry models in a continental-scale study of terrestrial ecosystem responses to climate change and CO2 doubling. Global Biogeochemical Cycles 9, 407–437.   120 Messier, C., Puettmann, K. J., 2011. Forests as complex adaptive systems: implications for forest management and modelling. L’Italia Forestale E Montana 66 (3), 249–258.  Meyer, C.B., Thuiller, W., 2006. Accuracy of resource selection functions across spatial scales. Diversity and Distributions 12 (3), 288–297.  Monleon, V.J., Lintz, H.E., 2015. Evidence of tree species’ range shifts in a complex landscape. PLoS ONE 10 (1), e0118069.   Monserud, R.A., Yang, Y., Huang, S., Tchebakova, N., 2008. Potential change in lodgepole pine site index and distribution under climate change in Alberta. Canadian Journal of Forest Research 38 (2), 343– 352.  Monteith, J.L., 1965. Evaporation and environment, In: Fogg, G.E. (ed), The state and movement of water in living organisms. Symposia of the Society for Experimental Biology 29. Cambridge University Press, Cambridge, pp. 205–234.  Moore, G.W., Edgar, C.B., Vogel, J., Washington-Allen, R.A., March, R.G., Zehnder, R., 2016. Tree mortality from an exceptional drought spanning mesic to semiarid ecoregions. Ecological Applications 26, 602–611.  Mote, P.W., Parson, E.A., Hamlet, A.F., Ideker, K.N., Keeton, W.S., Lettenmaier, D.P., Mantua, N.J., Miles, E.L., Peterson, D.W., Peterson, D.L., Slaughter, R., Snover, A.K., 2003. Preparing   121 for climate change: the water, salmon, and forests of the Pacific Northwest. Climatic Change 61, 45– 88.  Mote, P.W., Hamlet, A.F., Clark, M.P., Lettenmaier, D.P., 2005. Declining mountain snowpack in western North America. American Meteorological Society 86, 1–39.  Mote, P.W., Salathé, E.P., 2010. Future climate in the Pacific Northwest. Climate Change 102, 29–50.  Mueller RC, Scudder CM, Porter ME, Trotter RT, Gehring CA, Whitham TG., 2005. Differential tree mortality in response to severe drought: evidence for long-term vegetation shifts. Journal of Ecology, 93, 1085–1093.  Nightingale, J.M., Phinn, S.R., Held, A.A., 2004. Ecosystem process models at multiple scales for mapping tropical forest productivity. Progress Physical Geography 28, 241–281.  Nightingale, J.M., Coops, N.C., Waring, R.H., Hargrove, W.W., 2007. Comparison of MODIS gross primary production estimates for forests across the USA with those generated by a simple process model, 3-PG. Remote Sensing of Environment 109, 500–509.  Niinemets, Ü., 2010. Responses of forest trees to single and multiple environmental stresses from seedlings to mature plants: past stress history, stress interactions, tolerance and acclimation. Forest Ecology and Management 260, 1623–1639.   122 Nikinmaa, E., 1992. Analyses of the growth of Scots pine; matching structure with function. Acta Forestalia Fennica 235.  Nissinen, A., Hari, P., 1998. Effects of nitrogen deposition on tree growth and soil nutrients in boreal Scots pine stands. Environmental Pollution 102, 61–68.  Norby, R.J., Hartz-Rubin, J.S., Verbrugge, M.J., 2003. Phenological responses in maple to experimental atmospheric warming and CO2 enrichment. Global Change Biology 9 (12), 1792–1801.  Notaro, M., Mauss, A., Williams, J.W., 2012. Projected vegetation changes for the American Southwest: combined dynamic modeling and bioclimatic-envelope approach. Ecological Applications 22, 1365–1388.  O’Neill, G. A., Hamann, A., Wang, T., 2008. Accounting for population variation improves estimates of the impact of climate change on species’ growth and distribution, Journal of Applied Ecology 45, 1040–1049.   Opdam, P., Wascher, D., 2004. Climate change meets habitat fragmentation: linking landscape and biogeographical scale levels in research and conservation. Biological Conservation 117, 285–297.    123 Ordonez, A., Williams, J.W., 2013. Climatic and biotic velocities for woody taxa distributions over the last 16 000 years in eastern North America. Ecology Letters 16, 773-781.  Park, A., Puettmann, K., Wilson, E., Messier, C., Kames, S., Dhar, A., 2014. Can boreal and temperate forest management be adapted to the uncertainties of 21st century climate change? Critical Reviews in Plant Sciences 33, 251–285.  Parmesan, C., Yohe, G., 2003. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, 37–42.  Parmesan, C., Duarte, C., Poloczanska, E., Richardson, A.J., Singer, M.C., 2011. Overstretching attribution. Nature Climate Change 1, 2–4.  Pearson, R.G., Dawson, T.P., 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecology and Biogeography 12, 361–371.  Pearson, R.G., Dawson, T.P., 2005. Long-distance plant dispersal and habitat fragmentation: identifying conservation targets for spatial landscape planning under climate change. Biological Conservation 123, 389–401  Peterman, W., Waring, R.H., Seager, T., Pollock, W.L., 2013. Soil properties affect pinyon pine - juniper response to drought. Ecohydrology 6 (3), 455–463.   124 Piao, S., Sitch, S., Ciais, P., Friedlingstein, P., Peylin, P., Wang, X., Zeng, N., 2013. Evaluation of terrestrial carbon cycle models for their response to climate variability and to CO2 trends. Global Change Biology 19, 2117–2132.  Prasad, A., Iverson, L.R., Liaw, A., 2006. Newer classification and regression tree techniques: bagging and random forests for ecological prediction. Ecosystems 9, 181–199.  Puettmann, K., Coates, D., Messier, C., 2009. A critique of silviculture: managing for complexity. Island Press, Washington.  Raffa, K.F., Aukema, B.H., Bentz, B.J., Carroll, A.L., Hicke, J.A., Turner, M.G., Romme, W.H., 2008. Cross-scale drivers of natural disturbances prone to anthropogenic amplification: the dynamics of bark beetle eruptions. Bioscience 58, 501– 517.  Ramsfield, T. D., Bentz, B. J., Faccoli, M., Jactel, H., Brockerhoff, E. G., 2016. Forest health in a changing world: Effects of globalization and climate change on forest insect and pathogen impacts. Forestry, 89 (3), 245–252.   Rehfeldt, G.E., Ferguson, D.E., Crookston, N.L., 2009. Aspen, climate, and sudden decline in western USA. Forest Ecology and Management 258, 2353–2364.    125 Rehfeldt, G.E., Jaquish, B.C., 2010. Ecological impacts and management strategies for western larch in the face of climate-change. Mitigation and Adaptation Strategies for Global Change 15, 283–306.   Rehfeldt, G.E., Crookston, N.L., Saenz-Romero, C., Cambell, E.M., 2012. North American vegetation model for land-use planning in a changing climate: a solution to large classification problems. Ecological Applications 22, 119–141.  Rehfeldt, G. E., Jaquish, B. C., López-Upton, J., Sáenz-Romero, C., St Clair, J. B., Leites, L. P., Joyce, D.G., 2014. Comparative genetic responses to climate for the varieties of Pinus ponderosa and Pseudotsuga menziesii: realized climate niches. Forest Ecology and Management 324, 126–137.  Reichstein, M., Ciais, P., Papale, D., Valentini, R., Running, S., Viovy, N., Cramer, W., Granier, A., Ogee, J., Allard, V., Aubinet, M., Bernhofer, Chr., Buchmann, N., Carrara, A., Grunwald, T., Heimann, M., Heinesch, B., Knohl, A., Kutsch, W., Loustau, D., Manca, G., Matteucci, G., Miglietta, F., Ourcival, J.M., Pilegaard, K., Pumpanen, J., Rambal, S., Schaphoff, S., Seufert, G., Soussana, J.F., Sanz, M.J., Vesala, T., Zhao, M., 2007. Reduction of ecosystem productivity and respiration during the European summer 2003 climate anomaly: a joint flux tower, remote sensing and modelling analysis. Global Change Biology 13, 634–651.  Ricciardi, A., Simberloff, D., 2009. Assisted colonization is not a viable conservation strategy. Trends in Ecology and Evolution 24, 248–253   126 Runyon, J., Waring, R.H., Goward, S.N., Welles, J.W., 1994. Environmental limits on net primary production and light-use efficiency across the Oregon transect. Ecological Applications 4, 226–237.  Safranyik, L., Wilson, B., 2006. The mountain pine beetle: a synthesis of biology, management, and impacts on lodgepole pine. Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, BC.  Salajanu, D., Sensing, F.R., Jacobs, D.M., 2010. Contribution of climate, soil, and MODIS predictors when modeling forest invasive species distribution using forest inventory data. ASPRS 2010 Annual Conference San Diego, California. April 26–30, 2010.  Salisbury, E.J., 1926. The geographical distribution of plants in relation to climatic factors. The Geographical Journal 57, 312–335.  Sanchez, P.A., Ahamed, S., Carré, F., Hartemink, A.E., Hempel, J., et al., 2009. Digital soil map of the world. Science 325, 680–681.  Schrag, A.M., Bunn, A.G., Graumlich, L.J., 2008. Influence of bioclimatic variables on tree-line conifer distribution in the Greater Yellowstone Ecosystem: implications for species of conservation concern. Journal of Biogeography 35 (4), 698–710.    127 Schroeder, T., Hember, R., Coops, N., Liang, S., 2009. Validation of solar radiation surfaces from MODIS and reanalysis data over topographically complex terrain. Journal of Applied Meteorology and Climatology 48, 2441–2458.  Schroeder, T.A., Hamann, A., Coops, N.C., Wang, T., 2010. Occurrence and dominance of six Pacific Northwest conifer species. Journal of Vegetation Science 23, 586– 596.  Sherrod, P.H., 2010. DTREG manual. Online at  https://www.dtreg.com/uploaded/downloadfile/DownloadFile_5.pdf (14 December 2016).  Simard SW. 2009. The foundational role of mycorrhizal networks in self- organization of interior Douglas-fir forests. Forest Ecology and Management 258, S95–S107.  Sittaro, F., Paquette, A., Messier, C., Nock, C.A., 2017. Tree range expansion in eastern North America fails to keep pace with climate warming at northern range limits. Global Change Biology, doi:10.1111/gcb.13622  Skomarkova, M.V., Vaganov, E.A., Mund, M., Knohl, A., Linke, P., Boerner, A., Schulze, E.D., 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.    128 Spittlehouse, D.L., 2008. Climate change, impacts, and adaptation scenarios: climate change and forest and range management in British Columbia. Technical Report 045. B.C. Ministry of Forests & Range, Research Branch, Victoria, BC.  Syphard, A.D., Franklin, J., 2009. Differences in spatial predictions among species distribution modeling methods vary with species traits and environmental predictors. Ecography 32 (6), 907–918.  Tarrant, R.F., 1953. Soil moisture and the distribution of Lodgepole and Ponderosa Pine: A review of the literature. USDA Forest Service Pacific Northwest Forest and Range Experiment Station, Research Paper No. 8.  Thornley, J.H.M., Cannell, M.G.R., 1996. Temperate forest responses to carbon dioxide, temperature, and nitrogen: a model analysis. Plant, Cell & Environment 19, 1331–1348.  Thuiller, W., Lavorel, S., Araújo, M.B., Sykes, M.T., Prentice, I.C., 2005. Climate change threats to plant diversity in Europe. Proceedings of the National Academy of Sciences 102, 8245–8250.  Thuiller, W., Albert, C., Araujo, M.B., Berry, P.M., Cabeza, M., Guisan, A., Hickler, T., Midgley, G.F., Paterson, J., Schurr, F.M., Sykes, M.T., Zimmermann, N.E., 2008. Predicting global change impacts on plant species’ distributions: future challenges. Perspectives in Plant Ecology, Evolution and Systematics 9, 137–152.    129 Tomback, D.F., Resler, L.M., 2007. Invasive pathogens at alpine treeline: consequences for treeline dynamics. Physical Geography 28, 397–418.  Valentine, K.W.G., Sprout, P.N., Baker, T.E., Lavkulich, L.M., 1978. The soil landscapes of British Columbia. B.C. Ministry of Environment. Resource Analysis Branch.  Verbyla, D. L., 1987. Classification trees: a new discrimination tool. Canadian Journal of Forest Research 17, 1150– 1152.  Vitasse, Y., Delzon, S., Bresson, C.C., Michalet, R., Kremer, A., 2009. Altitudinal differentiation in growth and phenology among populations of temperate-zone tree species growing in a common garden. Canadian Journal of Forest Research 39, 1259–1269.  Vyse, A., Ferguson, C., Simard, S. W., Kano, T., Puttonen, P., 2006. Growth of Douglas-fir, lodgepole pine, and ponderosa pine seedlings underplanted in a partially-cut, dry Douglas-fir stand in south-central British Columbia. Forestry Chronicle 82, 723–732.  Wang, T., O’Neill, G.A., Aitken, S.N., 2010. Integrating environmental and genetic effects to predict responses of tree populations to climate. Ecological Applications 20, 153–163.  Wang, T., Campbell, E.M., O’Neill, G.A., Aitken, S.N., 2012. Projecting future distributions of ecosystem climate niches: Uncertainties and management applications. Forest Ecology and Management 279, 128–140.   130 Wang, T., Hamann, A., Spittlehouse, D., Carroll, C., 2016. Locally downscaled and spatially customizable climate data for historical and future periods for North America. PLoS One 11, e0156720.  Waring, R.H., 2000. A process model analysis of environmental limitations on growth of Sitka spruce plantations in Great Britain. Forestry 73, 65–79.  Waring, R.H., McDowell, N., 2002. Use of a physiological process model with forestry yield tables to set limits on annual carbon balances. Tree Physiology 22, 179–188.  Waring, R., Nordmeyer, A., Whitehead, D., Hunt, J., Newton, M., Thomas, C., Irvine, J., 2008. Why is the productivity of Douglas-fir higher in New Zealand than in its native range in the Pacific Northwest, USA? Forest Ecology & Management 255, 4040–4046.  Waring, R.H., Coops, N.C., Running, S.W., 2011. Predicting satellite-derived patterns of large-scale disturbances in forests of the Pacific Northwest Region in response to recent climatic variation. Remote Sensing of Environment 115, 3554–3566.  Waring, R.H., Coops, N.C., Mathys, A., Hilker, T., Latta, G., 2014. Process-based modeling to assess the effects of recent climatic variation on site productivity and forest function across western North America. Forests, 5, 518–534.    131 Waring, R.H., Coops, N.C., 2015. Predicting large wildfires across western North America by modeling seasonal variation in soil water balance. Climate Change 135 (2), 325–339.   Waring, R.H., Gao, L. 2016. Recent reduction in the frequency of frost accounts for most of the increased growth of a high elevation spruce forest in northwestern China. Trees 30, 1225-1236.  Warwell, M.V., Rehfeldt, G.E., Crookston, N.L., 2006. Modeling contemporary climate profiles of whitebark pine (Pinus Albicaulis) and predicting responses to global warming. Proceedings of the Conference Whitebark Pine: A Pacific Coast Perspective R6-NR-FHP-2007-01. USDA Forest Service Rocky Mountain Research Station, Moscow, Idaho.  Way, D.A., 2011. Tree phenology responses to warming: spring forward, fall back? Tree Physiology 31 (5), 469-471.  Webb, R.S., Anderson, K.H., Webb, T. I., 1993. Pollen response-surface estimates of late-Quaternary changes in the moisture balance of the northeastern United States. Quaternary Research 40, 213-227.  Weiskittel, A.R., Maguire, D.A., Monserud, R.A., Johnson, G.P., 2010. A hybrid model for intensively managed Douglas-fir plantations in the Pacific Northwest, USA. European Journal of Forest Research 129, 325–338.    132 Weiskittel, A.R., Hann, D.W., Kershaw, J.A. Jr, Vanclay, J.K., 2011. Forest growth and yield modeling. John Wiley & Sons, Chichester, UK.  Weltzin, J.F., Loik, M.E., Schwinning, S., Williams, D.G., Fay, P.A., Haddad, B.M., Harte, J., Huxman, T.E., Knapp, A.K., Lin, G.H., Pockman, W.T., Shaw, M.R., Small, E.E., Smith, M.D., Smith, S.D., Tissue, D.T., Zak, J.C., 2003. Assessing the response of terrestrial ecosystems to potential changes in precipitation. BioScience 53, 941– 952.  Westerling, A.L., Hidalgo, H.G., Cayan, D.R., Swetnam, T.W., 2006. Warming and earlier spring increase western U.S. forest wildfire activity. Science 313, 940– 943.  Wheaton, E., 2005. Canadian droughts of 2001 and 2002: comparing the 2001 and 2002 droughts with other droughts in western Canada. Saskatchewan Research Council Publication, 11602-9E03.  Whittaker, R.H., 1961. Vegetation history of the Pacific Coast States and the “central” significance of the Klamath region. Madrono 16, 5–23.  Williams, J., Shuman, B., Webb, T., Bartlein, P., Leduc, P., 2004. Late-quaternary vegetation dynamics in North America: scaling from taxa to biomes. Ecological Monographs 74, 309–334.    133 Woodall, C.W., Oswalt, C.M., Westfall, J.A., Perry, C.H., Nelson, M.D., Finley, A.O., 2009. An indicator of tree migration in forests of the eastern United States. For. Ecol. Manage. 257, 1434–1444.  Woods, A. J., Heppner, D., Kope, H. H., Burleigh, J., Maclauchlan, L., 2010. Forest health and climate change: A British Columbia perspective. The Forestry Chronicle 86, 412–422.  World Meteorological Organization, WMO, 2016. The state of greenhouse gases in the atmosphere based on global observations through 2015, WMO Greenhouse Gas Bulletin 12.   Worrall, J.J., Rehfeldt, G.E., Hamann, A., Hogg, E.H., Marchetti, S.B., Michaelian, M., Gray, L.K., 2013. Recent declines of Populus tremuloides in North America linked to climate. Forest Ecology & Management 299, 35–51.  Wright, E.F., Coates, K.D., Bartemucci, P., 1998. Regeneration from seed of six tree species in the interior cedar-hemlock forests of British Columbia as affected by substrate and canopy gap position. Canadian Journal of Forest Research 28, 1352–1364.  Zhao, M., Xiang, W., Peng, C. and Tian, D., 2009. Simulating age-related changes in carbon storage and allocation in a Chinese fir plantation growing in southern China using the 3-PG model. Forest Ecology and Management 257, 1520-1531.    134 Zhou, L.M., Tucker, C.J., Kaufmann, R.K., Slayback, D., Shabanov, N.V., Myneni, R.B., 2001. Variations in northern vegetation activity inferred from satellite data of vegetation index during 1981 to 1999. Journal of Geophysical Research 106, 20069–20083.    135 Appendices Appendix A  Tree species distribution maps for 20 major conifer species across western North America A.1 Climatically suitable species distributions under baseline conditions (1950-1975)                      136                          137                          138                          139                          140                          141 A.2 Tree species vulnerability maps in 2000-2009 compared to 1950-1975                         142                          143                          144                          145                          146                        

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.24.1-0357125/manifest

Comment

Related Items