UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Decay dynamics of coarsewood habitat in old-growth spruce and pine stands in the Rocky Mountain Foothills Jones, Eileen Laura 2009

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

Item Metadata

Download

Media
24-ubc_2010_spring_jones_eileen.pdf [ 63.31MB ]
Metadata
JSON: 24-1.0068827.json
JSON-LD: 24-1.0068827-ld.json
RDF/XML (Pretty): 24-1.0068827-rdf.xml
RDF/JSON: 24-1.0068827-rdf.json
Turtle: 24-1.0068827-turtle.txt
N-Triples: 24-1.0068827-rdf-ntriples.txt
Original Record: 24-1.0068827-source.json
Full Text
24-1.0068827-fulltext.txt
Citation
24-1.0068827.ris

Full Text

Decay dynamics of coarsewood habitat in old-growth spruce and pine stands in the Rocky Mountain Foothills by Eileen Laura Jones  B.Sc., Simon Fraser University, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate Studies (Geography)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2009 c Eileen Laura Jones 2009  Abstract This thesis presents research on the decay dynamics of coarsewood wildlife habitat in the foothills of the Rocky Mountains, west-central Alberta. The study sites were located in permanent sample plots in five Picea glauca and five Pinus contorta old-growth stands. I combined field sampling, dendrochronology, and permanent sample plot data to characterize snags and logs. I used a functional classification scheme to assess the potential wildlife habitat value of snags and logs. The study had two main objectives: (1) to quantify the magnitude of error in dendrochronological work on decayed wood and (2) to assess the accumulation and persistence of snags and logs and their potential functions as wildlife habitat. I used permanent plot data to verify the accuracy of year-of-death estimates obtained by crossdating snags and logs. I obtained YOD estimates from 71 snags and 54 logs. Most YOD dates occurred within the observed interval of death dates from the permanent plot data (54%–80%, grouped by species and coarsewood type) and most remaining dates preceded the interval of death. Overall, the magnitude of error in YOD estimates increased with time since death. I located 322 snags and 405 logs. Mean densities were 403 snags/ha and 506 logs/ha. Snags and logs in intermediate decay classes were the most ii  Abstract common, and I hypothesize that most snags reach decay class 4 or 5, rot at the base and fall over, rather than decaying completely in situ. Coarsewood persisted for many decades after death: estimated time since death of the oldest snag and log was 180 and 175 years, respectively. Time since death varied significantly across decay class, but the range of YOD dates in each decay class was so broad that decay class was not a reliable indicator of approximate time since death. Most observations of habitat functions were limited to one of five functional types. Less than 1% of snags and 4% of logs provided four or more habitat functions. Given the longevity of coarsewood in these stands, management plans must take a long-term view in order to maintain levels of coarsewood that are within the natural range of variability.  iii  Table of Contents Abstract  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iv  List of Tables  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ix  Preface  xi  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Co-Authorship Statement  . . . . . . . . . . . . . . . . . . . . . . xiv  1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.1  Introduction and Context . . . . . . . . . . . . . . . . . . . .  1  1.2  Coarsewood  4  . . . . . . . . . . . . . . . . . . . . . . . . . . .  1.2.1  Research Approaches  . . . . . . . . . . . . . . . . . .  4  1.2.2  Coarsewood Dynamics  . . . . . . . . . . . . . . . . .  8  1.3  Wildlife Habitat . . . . . . . . . . . . . . . . . . . . . . . . .  13  1.4  Thesis Context and Overview  . . . . . . . . . . . . . . . . .  17  1.5  References  . . . . . . . . . . . . . . . . . . . . . . . . . . . .  20 iv  Table of Contents 2 Verification of Year-of-Death Estimates . . . . . . . . . . . . 2.1  Introduction  2.2  Methods  2.3  . . . . . . . . . . . . . . . . . . . . . . . . . . .  27  . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  30  2.2.1  Field Methods . . . . . . . . . . . . . . . . . . . . . .  30  2.2.2  Dendrochronlogical Analyses . . . . . . . . . . . . . .  39  2.2.3  Year-of-Death Analyses . . . . . . . . . . . . . . . . .  41  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  43  2.3.1  Chronology Development . . . . . . . . . . . . . . . .  43  2.3.2  Comparison of YOD Estimates from Pairs of Cores  Results  and Pairs of Radii . . . . . . . . . . . . . . . . . . . .  47  Verification of YOD Dates Using PSP Data  . . . . .  51  . . . . . . . . . . . . . . . . . . . . . . . . . . . .  57  2.4.1  Within-Tree YOD Estimates . . . . . . . . . . . . . .  57  2.4.2  Verification of YOD Estimates . . . . . . . . . . . . .  58  2.4.3  Magnitude of Error in YOD Estimates  . . . . . . . .  60  2.4.4  Applications and Recommendations . . . . . . . . . .  62  2.3.3 2.4  2.5  Discussion  References  . . . . . . . . . . . . . . . . . . . . . . . . . . . .  3 Coarsewood Decay Dynamics and Wildlife Habitat 3.1  3.2  27  Introduction  64  . . . .  68  . . . . . . . . . . . . . . . . . . . . . . . . . . .  68  3.1.1  Ecological Context  . . . . . . . . . . . . . . . . . . .  3.1.2  Research Context and Objectives  68  . . . . . . . . . . .  71  . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  73  3.2.1  Field Methods . . . . . . . . . . . . . . . . . . . . . .  73  3.2.2  Dendrochronlogical Analyses . . . . . . . . . . . . . .  84  Methods  v  Table of Contents 3.2.3 3.3  3.4  3.5  Data Analyses . . . . . . . . . . . . . . . . . . . . . .  85  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  91  Results 3.3.1  Stand-Level Accumulation  3.3.2  Coarsewood Dynamics  3.3.3  Wildlife Habitat . . . . . . . . . . . . . . . . . . . . . 105  Discussion  91  . . . . . . . . . . . . . . . . .  97  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110  3.4.1  Stand-Level Accumulation  3.4.2  Coarsewood Dynamics  3.4.3  Wildlife Habitat . . . . . . . . . . . . . . . . . . . . . 121  3.4.4  Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 124  References  4 Conclusion 4.1  . . . . . . . . . . . . . . .  . . . . . . . . . . . . . . . 110  . . . . . . . . . . . . . . . . . 115  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133  Summary of Results . . . . . . . . . . . . . . . . . . . . . . . 133 4.1.1  Chapter 2: Verification of Year-of-Death Estimates  4.1.2  Chapter 3: Coarsewood Decay Dynamics and Wildlife  . 133  Habitat . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.2  Research Implications . . . . . . . . . . . . . . . . . . . . . . 135  4.3  Conservation Implications . . . . . . . . . . . . . . . . . . . . 136  4.4  Management Implications . . . . . . . . . . . . . . . . . . . . 138  4.5  4.4.1  Long-Term Planning  4.4.2  Old-Growth Forests and Natural Disturbances . . . . 139  References  . . . . . . . . . . . . . . . . . . 138  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141  vi  List of Tables 1.1  Snag decay classes. . . . . . . . . . . . . . . . . . . . . . . . .  7  1.2  Log decay classes. . . . . . . . . . . . . . . . . . . . . . . . . .  8  1.3  Classification of snag function as wildlife habitat. . . . . . . .  15  1.4  Classification of log function as wildlife habitat. . . . . . . . .  16  2.1  Physical attributes of old-growth white spruce and lodgepole pine sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  33  2.2  Tree status codes used by Hinton Wood Products. . . . . . .  34  2.3  Ecological attributes of old-growth white spruce and lodgepole pine sites. . . . . . . . . . . . . . . . . . . . . . . . . . .  34  2.4  Snag decay classes. . . . . . . . . . . . . . . . . . . . . . . . .  37  2.5  Log decay classes. . . . . . . . . . . . . . . . . . . . . . . . . .  38  2.6  Summary statistics for site-specific spruce and pine master ring-width chronologies. . . . . . . . . . . . . . . . . . . . . .  3.1  3.2  43  Physical attributes of old-growth white spruce and lodgepole pine sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  76  Tree status codes used by Hinton Wood Products. . . . . . .  77  vii  List of Tables 3.3  Ecological attributes of old-growth white spruce and lodgepole pine sites. . . . . . . . . . . . . . . . . . . . . . . . . . .  77  3.4  Snag decay classes. . . . . . . . . . . . . . . . . . . . . . . . .  79  3.5  Classification of snag function as wildlife habitat. . . . . . . .  80  3.6  Log decay classes. . . . . . . . . . . . . . . . . . . . . . . . . .  82  3.7  Classification of log function as wildlife habitat. . . . . . . . .  83  3.8  Average size of snags and logs in spruce- and pine-dominated sites and corresponding coarsewood density. . . . . . . . . . .  3.9  93  Hypothesis tests for the number of discriminating functions required to discriminate the decay classes for snags and logs. 104  3.10 Classification matrix of snag and log decay classes using canonical discriminant analysis. . . . . . . . . . . . . . . . . . . . . 104 3.11 Pearson’s correlation coefficients for site-level and individuallevel explanatory variables. . . . . . . . . . . . . . . . . . . . 105 3.12 Distribution of habitat functions. . . . . . . . . . . . . . . . . 106 3.13 Time since death of coarsewood wildlife habitat functions. . . 108 3.14 Distribution of snag habitat function types by decay class. . . 109 3.15 Distribution of log habitat function types by decay class. . . . 111 3.16 Distribution of log habitat function types by base type. . . . 112  viii  List of Figures 1.1  Picea snags in decay classes 3, 4, and 5. . . . . . . . . . . . .  6  1.2  Picea logs in decay classes 2 and 3. . . . . . . . . . . . . . . .  6  1.3  Decay classes of snags and logs. . . . . . . . . . . . . . . . . .  7  1.4  Decomposition pathways of live and dead trees. . . . . . . . .  9  1.5  Habitat functions of Picea snags and logs. . . . . . . . . . . .  14  1.6  The Rocky Mountain Foothills and old-growth spruce and pine sites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  18  2.1  Map of study sites. . . . . . . . . . . . . . . . . . . . . . . . .  35  2.2  Picea site-specific master ring-width chronologies. . . . . . . .  44  2.3  Pinus site-specific master ring-width chronologies. . . . . . .  45  2.4  Scatter plots of year-of-death dates for pairs of cores from Picea and Pinus snags. . . . . . . . . . . . . . . . . . . . . . .  2.5  Scatter plots of year-of-death dates for pairs of radii from Picea and Pinus logs. . . . . . . . . . . . . . . . . . . . . . .  2.6  50  The percentage of year of death estimates that occurred before, within, and after the observed interval of death dates. .  2.7  48  52  Scatter plots of the error in year of death estimates by time since death for Picea and Pinus snags and logs. . . . . . . . .  53 ix  List of Figures 2.8  Frequency of error in year of death estimate of Picea and Pinus snags and logs by decay class. . . . . . . . . . . . . . .  55  3.1  Map of study sites. . . . . . . . . . . . . . . . . . . . . . . . .  78  3.2  Species distribution of live trees, snags, and logs at each site.  92  3.3  Size-class distribution of snags, logs, and live trees at each site. 95  3.4  Decay-class distribution of snags by species at each site. . . .  96  3.5  Decay-class distribution of logs by species at each site. . . . .  98  3.6  Frequency distribution of time since death for Picea and Pinus coarsewood. . . . . . . . . . . . . . . . . . . . . . . . . .  3.7  99  Boxplots of time since death by decay class for Picea and Pinus snags and logs. . . . . . . . . . . . . . . . . . . . . . . 100  3.8  Depletion curves for Picea and Pinus snags and logs. . . . . . 103  3.9  Frequency of habitat functions. . . . . . . . . . . . . . . . . . 106  x  I want to tell you how land shows the power of linking back, of reconnecting, of remembering. Brian Swimme  xi  Acknowledgements This project would not have been possible without the support, wisdom, and encouragement of a tremendous number of people. Thank you to everyone that provided feedback on my ideas, taught me about tree rings, dispensed field advice, shared knowledge of the foothills, provided technical support, supplied equipment, shared data, funded my project, critiqued my writing, inspired my analysis, challenged my interpretations, supported my ideas, sanded my wood, counted my rings, wielded borers and chainsaws, got stuck in the mud, indulged in quad photo shoots, played in the forest, talked about science, talked about art, celebrated at Koerner’s, shared long hours at the library, tempered my stress levels, gave me a vibrant non-academic life, took me for tea, and left chocolate on my desk. I have enormous appreciation for the support and guidance of Lori Daniels. Thank you for giving me the freedom to explore my ideas but having the wisdom to reign me in. Thank you to my committee members, Valerie LeMay and R. Dan Moore, for your feedback and advice. Thank you to Dave Andison for helping me to better understand the context of my work. Thank you to my funders and collaborators — NSERC, the Foothills Research Institute, Hinton Wood Products, and the Alberta Newsprint Comxii  Acknowledgements pany — and to Rich McCleary, Fran Hanington, Tom Archibald, Glenn Buckmaster, Rick Bonor, and Steve Grimaldi for technical support. Thank you to Bill Bresnahan for keeping the crew safe, generously equipping the project, and toodling down the Athabasca. My research assistants were the backbone of this project. I shared a summer of adventures with Amy Nicoll, Julia Amerongen Maddison, and Evan Henderson. They deserve great thanks for helping me to pull my project together despite numerous obstacles. Abha Parajulee and Kylie McLeod provided great assistance in the lab and Raphael Chavardes was my master sander. Thank you to everyone in the UBC Tree-Ring Lab for your collaboration, friendship, and hijinks. Special recognition goes to Trevor Jones for initiating me into the lab. The UBC Geography Department provided a wonderful learning community. I am especially thankful for the camaraderie of the 2007 Physical cohort and to Christie Andrews, Tammy Elliott, and Melissa Ewan for being ladies who lunch. Emily Jane Davis and Sarah Zell provided tremendous friendship and emotional support. Finally, thank you to Peter, Victoria, and Ian Jones for your endless love and encouragement.  xiii  Co-Authorship Statement Versions of Chapters 2 and 3 are co-authored by Lori D. Daniels and will be submitted for publication. Dr. Daniels envisioned the research program and was instrumental in designing the project. She also provided substantial guidance for analysis and writing. I identified the specific research questions and co-designed the project. I carried out and supervised all field sampling and lab work. I also conducted all statistical analysis and wrote the manuscripts.  xiv  Chapter 1  Introduction When one walks through the rather dull and tidy woodlands. . . that result from modern forestry practises, it is difficult to believe that dying and dead wood provides one of the two or three greatest resources for animal species in a natural forest, and that if fallen timber and slightly decayed trees are removed the whole system is gravely impoverished of perhaps more than a fifth of its fauna. Elton, 1966, p. 279  1.1  Introduction and Context  Historically, coarsewood has been the bane of foresters working in commercially managed forests (Thomas, 2002). Snags and logs have been widely viewed through the lens of efficient wood-fibre production, rather than from an ecologically based perspective. Snags have been regarded as a safety risk in logging operations and a source of spot fires, and logs have been considered unsightly, a haven for pests, and the fuel that spreads wildfires. Although the habitat value of coarsewood has been recognized since the 1920s (Graham, 1925), researchers and forest managers became increasingly 1  1.1. Introduction and Context aware of the ecological roles of coarsewood in the 1970s (e.g. Cornaby and Waide, 1973; Cromack et al., 1975; Maser et al., 1978). In the last 30 years, research interests in coarsewood have expanded (see Harmon et al., 1986; Jonsson and Kruys, 2001; Laudenslayer Jr. et al., 2002) and progressive management practices now aim to maintain coarsewood in managed stands (e.g. Bergeron and Harvey, 1997; Laudenslayer Jr. et al., 2002). Terrestrial coarsewood serves myriad roles in forest ecosystems. Snags and logs provide a nutrient-rich substrate on which plants and fungus grow; they function as wildlife habitat, providing a substrate for concealment, breeding, and nesting; they provide habitat for invertebrates and microorganisms, acting as a food source and site for shelter and breeding; their decomposition has an important role in terrestrial nutrient and carbon cycles; and they serve geomorphic functions, influencing the transport and storage of soil and sediment (Harmon et al., 1986). The functional importance of coarsewood depends not only on the volumes of snags and logs, but also on the densities, size classes, decay stages, and tree species, and the spatial arrangement and relative distributions of these characteristics between snags and logs (Harmon et al., 1986). Understanding these ecological functions is important for the effective management of forests, as the removal of coarsewood can lead to unexpected ecosystem changes. In both managed and natural landscapes, coarsewood is a biological legacy that sustains functional continuity through time. Biological legacies are “organisms, organic materials, and organically generated environmental patterns that persist through a disturbance and are incorporated into the recovering ecosystem” (Franklin et al., 2000). Snags and logs serve as 2  1.1. Introduction and Context structural legacies. When they persist through a natural or anthropogenic disturbance, snags and logs maintain critical structures and provide structural diversity in the new community (Lindenmayer and Franklin, 2002). For example, young forests established by fire have an abundance of snags and logs from the previous community (Perry, 1994). And logs that persist through many disturbances are critical to sustaining populations of many endangered species in Scandinavia (Bader et al., 1995). Management scenarios that try to emulate natural disturbances must retain coarsewood as biological legacies to ensure continuity of their ecological functions from old to new communities. In this thesis, I focus on a single function of coarsewood: the role of decaying snags and logs as wildlife habitat. Wildlife species in boreal and subboreal forests worldwide depend on coarsewood as an integral part of their habitat: in Norway, for example, 37% of endangered forest-dwelling wildlife species, and an even larger proportion of threatened species, use snags and logs as habitat (Gundersen and Rolstad, 1998). The loss of coarsewood in these forests is believed to be one cause of the declining populations of endangered and threatened wildlife species. Many vertebrate wildlife species are specialized to certain coarsewood characteristics, such as position, species, size, and decay class (Harmon et al., 1986). In British Columbia (Keisker, 2000) and Alberta (Crampton and Barclay, 1998; Norton et al., 2000), both common and threatened wildlife use specialized coarsewood as habitat. In this thesis, I examine the decay dynamics of coarsewood wildlife habitat in old-growth stands. While I do not explicitly address the role of coarsewood as structural legacies, this research does have implications for understanding 3  1.2. Coarsewood the persistence of coarsewood habitat following natural and anthropogenic disturbances.  1.2 1.2.1  Coarsewood Research Approaches  Year-of-Death Estimates Studying coarsewood dynamics requires knowledge of the year in which a snag or log died. Temporal year-of-death estimates can be acquired directly and indirectly though (1) remeasurement of permanent sample plots (PSPs) and (2) dendrochronological analysis. Permanent sample plot studies provide direct observations of dead trees during plot remeasurements; however, such studies must persist on a time scale long enough to garner useful information and can therefore be difficult for studying long-term dynamics. In addition, PSPs require a dedicated land base and sufficient resources to ensure regular plot remeasurements over the long term. Furthermore, most permanent plot networks are designed to monitor timber growth and yield and few monitor long-term coarsewood dynamics. In contrast, dendrochronological studies can provide indirect reconstructions of stand dynamics. Year-of-death estimates are obtained by crossdating the ring-width series of an individual snag or log against a master chronology, which allows the assignment of a calendar year to the outermost ring on a sample (Fritts, 1976). This technique permits retrospective  4  1.2. Coarsewood studies of many centuries but assumes that the outermost ring on a sample indeed reflects the year of tree death. This is often not the case when samples are decayed or missing bark and sapwood (Daniels et al., 1997; Mast and Veblen, 1994). Dendrochronological studies have been widely used to understand coarsewood dynamics and the temporal development of wildlife habitat (e.g Daniels et al., 1997; DeLong et al., 2005, 2008; Storaunet and Rolstad, 2004), but dendrochronologists rarely have the opportunity to assess the quality of these estimates. As a result, the magnitude of error associated with crossdating dead and decayed wood is largely unknown.  Classification Systems As snags and logs decompose, they undergo successive structural changes. Snags and logs that have undergone similar morphological changes while decomposing can be grouped into decay classes (Figures 1.1 and 1.2). In this study, I use a nine-class system for snags and a five-class system for logs, based on structural integrity and the soundness of heartwood and sapwood (Figure 1.3; Tables 1.1 and 1.2; Maser et al., 1979; Thomas et al., 1979). Decay classes are sometimes used as a proxy for time since death; however, many studies have found that range of year-of-death dates within a single class is highly variable, indicating that decay class is not a reliable indicator of approximate time since death (Daniels et al., 1997; DeLong et al., 2008; Mast and Veblen, 1994).  5  1.2. Coarsewood  Figure 1.1: Picea snags in decay classes 3, 4, and 5 (left to right).  Figure 1.2: Picea logs in decay classes 2 and 3.  6  1.2. Coarsewood  Figure 1.3: Decay classes of snags and logs. When snags fall, they enter the log decay class indicated by the vertical lines. (Adapted from Maser et al., 1979.)  Table 1.1: Snag decay classes based on branch order and structural integrity and soundness of sapwood and heartwood (Thomas et al., 1979).  7  1.2. Coarsewood Table 1.2: Log decay classes based on structural integrity and soundness of sapwood and heartwood (Maser et al., 1979).  1.2.2  Coarsewood Dynamics  Coarsewood Pathways When a tree dies, it can follow one of three general pathways as it decomposes (Figure 1.4): 1. If a tree dies standing, it turns into a snag and may decay completely in situ; 2. A standing dead tree may persist for some time before falling over and decomposing as a log; 3. A tree may die from uprooting and create a log.  8  1.2. Coarsewood  Figure 1.4: Decomposition pathways of live and dead trees. Coarsewood is created by breakage and mortality of live trees. Snag fragmentation and breakage creates logs and internal decay processes decompose both snags and logs. Fragmentation and burial transform coarsewood to fine woody debris and soil. (Adapted from Harmon et al., 1986.)  Logs can also be created by broken tops and large branches from both snags and live trees. If a snag transitions into a log, its decay class as a snag determines its classification as a log (Figure 1.4). For example, snags in decay class 3 create decay class 1 logs (Figure 1.3).  Tree Mortality and Coarsewood Accumulation Coarsewood is generated in a forest ecosystem when a tree dies. Thus, patterns of coarsewood input follow tree mortality patterns. Episodic mortality due to large-scale disturbances has traditionally been considered the most important driver of boreal and sub-boreal ecosystem dynamics (Perry, 1994); however, recent studies of forest dynamics have highlighted the importance 9  1.2. Coarsewood of fine-scale local disturbances leading to gap formation in interior boreal and sub-boreal forests (McCarthy, 2001). Continuous mortality of individuals or small groups of trees is important for creating the structural complexity characteristic of old-growth stands (Franklin et al., 1987). In this study, I focus on coarsewood resulting from background mortality rather than from a large-scale or stand-replacing disturbance. Possible fine-scale mortality agents include senescence, competition, disease, insects, low-severity fires, wind, and environmental stress (Franklin et al., 1987; Hinton Wood Products, 2006). Episodic mortality can create large volumes of coarsewood that persist after stand turnover. When an early-succession stand emerges after a standreplacing disturbance, coarsewood volumes often follow a bimodal distribution: volumes are high immediately following the disturbance, decrease as the stand ages, and increase again in mature and old-growth stands (Clark et al., 1998). The total percentage of dead wood relative to living trees may be proportional to the productivity of a stand, with old-growth forests having higher numbers of dead trees (Nilsson et al., 2002). In addition, older forests tend to have larger snags and logs than younger stands. As stands age and selfthinning is supplemented by the death of canopy co-dominants, coarsewood size is similar to the mean diameter of live trees (Clark et al., 1998; Cline et al., 1980; Lee, 1998).  10  1.2. Coarsewood Snag Persistence and Decay Snag decay rates are variable and appear to be species-specific. Pinus snags, for example, have been observed to decay more quickly than Abies (Morrison and Raphael, 1993) and Picea (Alban and Pastor, 1993) snags. The time period that a snag spends within a given decay class is highly variable. Intermediate decay classes (e.g. decay class 4) are often the most abundant, suggesting that snags may decay more rapidly immediately following death and show slower decay rates in intermediate and advanced decay classes (Harmon et al., 1986). Overall, time since death increases with decay classes; however, the year-of-death dates within a decay class are often too variable to infer time since death from decay class (Daniels et al., 1997; DeLong et al., 2008; Mast and Veblen, 1994). Snags either decay completely in situ or fall to the ground as logs. Snag longevity, expressed as the probability of a snag surviving to a given age, has been shown to follow a reverse sigmoid cure (Garber et al., 2005). Snag fall rates are initially low following tree death, resulting in a lag time. Documented lag times in montane and boreal forests range from 1–90 years (Garber et al., 2005; Lee, 1998; Morrison and Raphael, 1993; Vanderwel et al., 2006). Some studies have found lag times to increase with increasing snag diameter (Garber et al., 2005; Raphael and Morrison, 1987), while other data suggest that snag persistence is not related to tree size (Lee, 1998).  11  1.2. Coarsewood Log Persistence and Decay Logs exhibit species-specific decay rates (Alban and Pastor, 1993). Picea logs, for example, have been observed to reach advanced stages of decay more quickly than Abies logs (DeLong et al., 2005) but published depletion rates suggest that they decay more slowly than Pinus logs (Alban and Pastor, 1993; Johnson and Greene, 1991; Laiho and Prescott, 1999, 2004). Site productivity, decomposer community, log size, and log position also influence log decay rates and the resulting length of time spent in a particular decay class (Alban and Pastor, 1993; Daniels et al., 1997; Kruys et al., 2002; Vanderwel et al., 2006). As with snags, intermediate decay classes (e.g. decay class 3) are often the most abundant (Harmon et al., 1986). Generally, time since death of logs increases with decay class (Maser et al., 1979); however, the length of time spent within a decay class can be highly variable. For example, Picea logs in decay class 4 have been reported to be as young as 22 years old (Jonsson, 2000) and as old as 145 years (Antos and Parish, 2002). Decay class poorly predicts time since death since many logs originate as snags (Daniels et al., 1997; Mast and Veblen, 1994). Snags decay more slowly than logs, likely due to desiccation and decreased decomposer activity and thus the lag time influences the overall decay rate of a snag cum log (Harmon et al., 1986; Johnson and Greene, 1991). Consequently, time since snag fall is a better predictor of log decay class than time since death (Daniels et al., 1997; DeLong et al., 2005; Storaunet and Rolstad, 2002).  12  1.3. Wildlife Habitat  1.3  Wildlife Habitat  Managing forests to maintain wildlife biodiversity faces a substantial challenge: how to accommodate a large number of species and their myriad habitat requirements. One approach is to shift perspectives; rather than thinking about the requirements of individual species, instead consider what kinds of habitat structures are present in a stand and the suite of species they can support. In line with this habitat-based perspective, the British Columbia Ministry of Forests published a report surveying the habitat requirements of 133 wildlife species in north-central British Columbia (Keisker, 2000). From this assessment, wildlife habitat needs were delineated into coarsewood habitat functional types, including 10 snag types (S1–S10) and six log types (L1–L6; Figure 1.5, Tables 1.3 and 1.4). Habitat functions L1, L2, and L3 can apply to the lower bole of snags as well as logs. This functional habitat classification is a useful tool to rapidly distinguish snags and logs with high potential to serve as wildlife habitat from those with low habitat value. Although the system was developed for the forests of central British Columbia, it is applicable to other forests with a similar composition and wildlife community. Using a classification system to assess the habitat value of a stand has limitations: it does not consider the number and spatial distribution of each habitat function required by the wildlife community. Habitat functional type is not necessarily congruent with decay class, and we cannot infer one from the other (Keisker 2000); however, recent studies of decaying habitats revealed some general relations among habitat type,  13  1.3. Wildlife Habitat  Figure 1.5: Habitat functions of Picea snags and logs: a red squirrel uses a log as an elevated runway (top left); a snag (top right) and a log (bottom), each with a small cavity.  14  1.3. Wildlife Habitat  Table 1.3: Classification of snag function as wildlife habitat (after Keisker, 2000)  15  1.3. Wildlife Habitat  Table 1.4: Classification of log function as wildlife habitat (after Keisker, 2000).  16  1.4. Thesis Context and Overview coarsewood age, and level of decay (DeLong et al., 2005, 2008). Potential wildlife habitat function was most common in fresh and intermediate decay classes for Picea snags and intermediate and old decay classes to Abies snags (DeLong et al., 2008). In two studies of interior coniferous forests, common snag habitat functions included potential substrates for cavity excavation (S1), cracks, loose, or furrowed bark (S6), large open nest supports (S8), small concealed spaces above ground level (L3), and large concealed spaces at the base of snags (L1; Tables 1.3 and 1.4; DeLong et al., 2008; Stevenson et al., 2006). As logs decayed, their function as wildlife habitat changed significantly: for the first fifteen years on the ground, logs remained relatively undecayed and served as elevated runways (L5; DeLong et al., 2005). They increased in habitat value as wood softened and vegetation began to grow on the wood. As logs decomposed further, the number of concealed spaces increased, providing habitat for reproduction, resting, and escape (L2, L3). In advanced stages of decay, logs collapsed and provided little habitat value (Table 1.4).  1.4  Thesis Context and Overview  This thesis presents research on the decay dynamics of coarsewood wildlife habitat in the foothills of the Rocky Mountains (Figure 1.6), west-central Alberta. The study sites were located in permanent sample plots in the Hinton Wood Products Forest Management Area of the Foothills Research Institute (FRI). This study is part of a larger collaborative project through the FRI Natural Disturbance Program on the recruitment, residence times, and de-  17  1.4. Thesis Context and Overview  Figure 1.6: The Rocky Mountain Foothills, north of Hinton, Alberta (top). Oldgrowth study sites dominated by Picea (left) and Pinus (right) trees.  18  1.4. Thesis Context and Overview cay rates of deadwood in small streams and pine- and spruce-dominated riparian and upland forests. The outcomes of the larger project will have value to applied forest management; the researchers are working with the FRI to incorporate our findings into forest management practises and policy. In the FRI landbase, progressive management practises take a landscapebased approach and are guided by natural and historical ranges of patterns and structures. Previous research on coarsewood in the FRI landbase is limited and, given the variability in coarsewood accumulation and persistence, management decisions should be based on region- and ecosystem-specific studies of coarsewood. In this study, I aim to characterize the accumulation, persistence, decay, and wildlife-habitat function of coarsewood in old-growth coniferous stands in the FRI landbase. Chapters 2 and 3 present the results of field research. Chapter 2 addresses error in dendrochronological work on decayed wood by comparing year-of-death estimates to permanent sample plot data. Chapter 3 analyses the accumulation and persistence of snags and logs and their potential functions as wildlife habitat. Chapter 4 provides a summary and discussion of research, conservation, and management implications.  19  1.5. References  1.5  References  Alban, D. H. and J. Pastor. 1993. Decomposition of aspen, spruce, and pine boles on two sites in Minnesota. Canadian Journal of Forest Research, 23:1744–1749. Antos, J. A. and R. Parish. 2002. Structure and dynamics of a nearly steady-state subalpine forest in south-central British Columbia, Canada. Oecologia, 130:126–135. Bader, P., S. Jansson, and B. G. Jonsson. 1995. Wood-inhabiting fungi and substratum decline in selectively logged boreal spruce forests. Biological Conservation, 72:355–362. Bergeron, Y. and B. Harvey. 1997. Basing silviculture on natural ecosystem dynamics: an approach applied to the southern boreal mixedwood forest of Quebec. Forest Ecology And Management, 92:235–242. Clark, D. F., D. D. Kneeshaw, P. J. Burton, and J. A. Antos. 1998. Coarse woody debris in sub-boreal spruce forests of west-central British Columbia. Canadian Journal of Forest Research, 28:284–290. Cline, S. P., A. B. Berg, and H. M. Wight. 1980. Snag characteristics and dynamics in Douglas-fir forests, Western Oregon. Journal of Wildlife Management, 44:773–786. Cornaby, B. W. and J. B. Waide. 1973. Nitrogen-fixation in decaying chestnut logs. Plant and Soil, 39:445–448.  20  1.5. References Crampton, L. H. and R. M. R. Barclay. 1998. Selection of roosting and foraging habitat by bats in different-aged aspen mixedwood stands. Conservation Biology, 12:1347–1358. Cromack, K., R. L. Todd, and C. D. Monk. 1975. Patterns of basidiomycete nutrient accumulation in conifer and deciduous forest litter. Soil Biology & Biochemistry, 7:265–268. Daniels, L. D., J. Dobry, K. Klinka, and M. C. Feller. 1997. Determining year of death of logs and snags of Thuja plicata in southwestern coastal British Columbia. Canadian Journal of Forest Research, 27:1132–1141. DeLong, S. C., L. D. Daniels, B. Heemskerk, and K. O. Storaunet. 2005. Temporal development of decaying log habitats in wet spruce-fir stands in east-central British Columbia. Canadian Journal of Forest Research, 35:2841–2850. DeLong, S. C., G. D. Sutherland, L. D. Daniels, B. Heemskerk, and K. Storaunet. 2008. Temporal dynamics of snags and development of snag habitats in wet spruce-fir stands in east-central British Columbia. Forest Ecology and Management, 255:3613–3620. Elton, C. S. 1966. The Pattern of Animal Communities. Methuen and Co. Ltd., London, UK. Franklin, J. F., D. Lindenmayer, J. A. MacMahon, A. McKee, J. Magnuson, D. A. Perry, R. Waide, and D. Foster. 2000. Threads of continuity. Conservation Biology in Practice, 1:8–17.  21  1.5. References Franklin, J. F., H. H. Shugart, and M. E. Harmon. 1987. Tree death as an ecological process. BioScience, 37:550–556. Fritts, H. 1976. Tree Rings and Cimate. Academic Press, New York, NY. Garber, S. M., J. P. Brown, D. S. Wilson, D. A. Maguire, and L. S. Heath. 2005. Snag longevity under alternative silvicultural regimes in mixedspecies forests of central Maine. Canadian Journal of Forest Research, 35:787–796. Graham, S. 1925. The felled tree trunk as an ecological unit. Ecology, 6:397–411. Gundersen, V. and J. Rolstad. 1998. Truete arter i skog [in Norwegian]. ˙ Norway. Technical Report 6/98, Norweigan Forest Research Institute, As, Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, J. D. Lattin, N. H. Anderson, S. P. Cline, N. G. Aumen, J. R. Sedell, G. W. Lienkaemper, K. Cromack, and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research, 15:133–302. Hinton Wood Products. 2006. Permanent Growth Sample Program Manual, version 16. West Fraser Mills, Hinton, AB. Johnson, E. A. and D. F. Greene. 1991. A method for studying dead bole dynamics in Pinus contorta var latifolia–Picea engelmannii forests. Journal of Vegetation Science, 2:523–530.  22  1.5. References Jonsson, B. G. 2000. Availability of coarse woody debris in a boreal oldgrowth Picea abies forest. Journal of Vegetation Science, 11:51–56. Jonsson, B. G. and N. Kruys, editors. 2001. Ecological Bulletins No. 49: Ecology of Woody Debris in Boreal Forests. Blackwell Science, Oxford, UK. Keisker, D. 2000. Types of wildlife trees and coarse woody debris required by wildlife of north-central British Columbia. Working Paper 50, B.C. Ministry of Forests, Victoria, B.C. Kruys, N., B. G. Jonsson, and G. Stahl. 2002.  A stage-based matrix  model for decay-class dynamics of woody debris. Ecological Applications, 12:773–781. Laiho, R. and C. E. Prescott. 1999. The contribution of coarse woody debris to carbon, nitrogen, and phosphorus cycles in three Rocky Mountain coniferous forests. Canadian Journal of Forest Research, 29:1592–1603. Laiho, R. and C. E. Prescott. 2004. Decay and nutrient dynamics of coarse woody debris in northern coniferous forests: a synthesis. Canadian Journal of Forest Research, 34:763–777. Laudenslayer Jr. W. F., P. J. Shea, E. Valentine, Bradley, C. P. Weatherspoon, and T. E. Lisle, editors. 2002. Proceedings of the Symposium on the Ecology and Management of Dead Wood in Western Forests, 2– 4 November 1999, Reno NV. General Technical Report PSW-GTR-181, U.S. Department of Agriculture Forest Service Pacific Southwest Research Station, Albany, CA. 23  1.5. References Lee, P. 1998. Dynamics of snags in aspen-dominated midboreal forests. Forest Ecology and Management, 105:263–272. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving Forest Biodiversity: A Comprehensive Multiscaled Approach. Island Press, Washington, D.C. Maser, C., R. Anderson, K. Cromack, J. T. Williams, and R. E. Martin. 1979. Dead and down woody material. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 78–95. U.S. Department of Agriculture Forest Service, Washington, DC. Maser, C., R, R. A. Nussbaum, and J. M. Trappe. 1978. Fungal–small mammal interrelationships with emphasis on Oregon coniferous forests. Ecology, 59:799–809. Mast, J. N. and T. T. Veblen. 1994. A dendrochronological method of studying tree mortality patterns. Physical Geography, 15:529–542. McCarthy, J. 2001. Gap dynamics of forest trees: a review with particular attention to boreal forests. Environmental Reviews, 9:1–59. Morrison, M. L. and M. G. Raphael. 1993. Modeling the dynamics of snags. Ecological Applications, 3:322–330. Nilsson, S. G., M. Niklasson, J. Hedin, G. Aronsson, J. M. Gutowski, P. Linder, H. Ljungberg, G. Mikusinski, and T. Ranius. 2002. Densities of large living and dead trees in old-growth temperate and boreal forests. Forest Ecology and Management, 161:189–204. 24  1.5. References Norton, M. R., S. J. Hannon, and F. K. A. Schmiegelow. 2000. Fragments are not islands: patch vs. landscape perspectives on songbird presence and abundance in a harvested boreal forest. Ecography, 23:209–223. Perry, D. A. 1994. Forest Ecosystems. Johns Hopkins University Press, Baltimore, M.D. Raphael, M. G. and M. L. Morrison. 1987. Decay and dynamics of snags in the Sierra Nevada, California. Forest Science, 33:774–783. Stevenson, S. K., M. J. Jull, and B. J. Rogers. 2006. Abundance and attributes of wildlife trees and coarse woody debris at three silvicultural systems study areas in the interior cedar-hemlock zone, British Columbia. Forest Ecology and Management, 233:176–191. Storaunet, K. and J. Rolstad. 2002. Time since death and fall of Norway spruce logs in old-growth and selectively cut boreal forest. Canadian Journal of Forest Research, 32:1801–1812. Storaunet, K. O. and J. Rolstad. 2004. How long do Norway spruce snags stand? Evaluating four estimation methods. Canadian Journal of Forest Research, 34:376–383. Thomas, J. 2002. Dead wood: from forester’s bane to environmental boon. In W. F. Laudenslayer Jr., P. J. Shea, E. Valentine, Bradley, C. P. Weatherspoon, and T. E. Lisle, editors, Proceedings of the Symposium on Ecology and Management of Dead Wood in Western Forests, 2–4 November 1999, Reno, Nevada, U.S. Forest Service General Technical Report PSW-  25  1.5. References GTR-181, pages 3–9. U.S. Department of Agriculture Pacific Southwest Research Station, Albany, CA. Thomas, J., R. Anderson, C. Maser, and E. Bull. 1979. Snags. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 60–77. U.S. Department of Agriculture Forest Service, Washington, DC. Vanderwel, M. C., J. R. Malcolm, and S. M. Smith. 2006. An integrated model for snag and downed woody debris decay class transitions. Forest Ecology and Management, 234:48–59.  26  Chapter 2  Verification of Year-of-Death Estimates1 2.1  Introduction  Coarsewood serves myriad ecological functions but its dynamics are not thoroughly understood (Harmon et al., 1986). In order to be effectively incorporated into stand dynamics models, we require species- and ecosystem-specific understandings of mortality and coarsewood decay rates and processes. Decay rates of spruce logs, for example, are highly variable. In one study of sub-boreal forests in British Columbia, Picea glauca x Picea engelmanii logs with sapwood and heartwood decay (decay class 4) were found to be an average of 59 ± 20 years since tree death, and these characteristics were found to persist up to 148 years (DeLong et al., 2005). In contrast, one study of Swedish boreal forests found that Picea abies logs begin to become incorporated into the forest floor (decay class 5) in a short 34 ± 12 years since tree death (Jonsson, 2000). Understanding coarsewood decay dynamics requires 1  A version of this chapter will be submitted for publication. Jones, E.L. and L.D. Daniels. Verification of dendrochronological year-of-death estimates using permanent sample plot data.  27  2.1. Introduction knowledge of snag and log persistence (i.e. time since tree death) and thus year of tree death. Year of death (YOD) can be estimated both directly and indirectly through (1) remeasurement of permanent sample plots (PSPs) and (2) dendrochronological analysis. Permanent sample plot studies provide direct observations of dead trees during plot remeasurements; however, resulting YOD dates are imprecise. The precision of YOD dates is restricted by the length of the remeasurement interval. YOD estimates from PSP studies are either then reported as an interval (e.g. Vanderwel et al., 2006) or assumed to be the midpoint of the remeasurement interval in which a tree died (e.g. Newberry et al., 2004). In many permanent sample plot networks, the target remeasurement interval is every five years (Lewis et al., 2004). Due to resource constraints, remeasurement intervals can vary considerably in length, both among plots in a network and within individual plots. For example, studies of permanent sample plots in Denmark revealed variation in remeasurement intervals of 10 to 103 years (Nord-Larsen, 2006)! In contrast, dendrochronological analysis provides precise year-of-death estimates, at an annual resolution. YOD estimates are obtained by crossdating the ring-width series of an individual snag or log against a master chronology, which allows the assignment of a calendar year to the outermost ring on the sample (Fritts, 1976). This technique assumes that the outermost ring indeed reflects the year of tree death, which is often not the case when samples are decayed or missing bark (Daniels et al., 1997; Mast and Veblen, 1994). In addition, the outermost ring may not reflect the year of tree death if a tree is suppressed and declining prior to death (Takaoka, 28  2.1. Introduction 1993). Dendrochronological studies have been widely used in studies of coarsewood dynamics to determine YOD dates for snags and logs (e.g. Daniels et al., 1997; DeLong et al., 2008; Storaunet and Rolstad, 2004), but dendroecologists rarely have the opportunity to assess the quality of these estimates. Many studies acknowledge the potential sources of error inherent in working with decayed wood but, to my knowledge, the magnitude of error associated with crossdating decayed wood has never been quantified. In this study, I use permanent sample plot data to verify the accuracy of year-of-death estimates of white spruce (Picea glauca) and lodgepole pine (Pinus contorta) snags and logs obtained by crossdating against site- and species-specific chronologies. To this end, I addressed two groups of research questions: 1. Within a single tree (snag or log), how similar are YOD dates from pairs of cores or pairs of radii? When there is a mis-match in YOD dates, what factors contribute to this discrepancy? 2. How do year-of-death estimates compare to the reported interval of death dates from the permanent sample plot data? What percent of YOD dates occur within, before, and after the reported interval of death? What is the magnitude of error of YOD estimates? How does error relate to stage of decay? Ultimately, my intent is to identify the range and magnitude of error of YOD estimates in pine and spruce snags and logs in order to facilitate a more accurate understanding of coarsewood dynamics. 29  2.2. Methods  2.2 2.2.1  Methods Field Methods  Study Area This study was conducted in the Hinton Wood Products Forest Management Area of the Foothills Research Institute near Hinton, Alberta (53◦ 24’09” N 117◦ 34’33” W), east of the Rocky Mountains. In this region, upland forests are characterized by closed-canopy coniferous stands, dominated by lodgepole pine (Pinus contorta), white spruce (Picea glauca), and black spruce (Picea mariana). The landscape is a mosaic of mixed-successional pine, pine–white spruce, and white spruce–black spruce stands, created by standinitiating fires with a historical return interval of 100 years (Beckingham et al., 1996).  Establishment and Monitoring of Permanent Sample Plots Between 1956 and 1961, North Western Pulp and Paper (now Hinton Wood Products) established a network of approximately 3200 stand inventory plots, also called permanent sample plots (PSPs), that included a range of forest types and stages of stand development (Hinton Wood Products, 2006). Plot clusters were established every two miles at the intersection of the Alberta Legal Survey grid section lines. Clusters contain four plots, each at a distance of 100.6 m from the cluster centre, at azimuths of 45◦ , 135◦ , 225◦ , and 315◦ . In 1988, the area in the FMA expanded and an additional 114 PSPs were established in 1991. A number of other PSPs have since been  30  2.2. Methods established to represent combinations of species compositions not included in the original network or to assess specific timber management opportunities. Most plots are 0.08 ha, though some recently established plots are 0.04 ha. During PSP establishment, all trees with a diameter at breast height (dbh) greater than the tagging limit were tagged and measured (Hinton Wood Products, 2006). The tagging limit has changed five times since 1956, ranging from 5.0 to 11.6 cm. The current limit is 7.0 cm. The target remeasurement interval for old-growth coniferous PSPs is every five years; however, as many as 33 years have lapsed between successive measurements (Table 2.1). At each remeasurement, numerous tree attributes are recorded, including species, dbh, height, height to live crown, crown fullness, crown radius, crown position, cause of death (if applicable), tree status (live, snag, stub, stump, log, or missing; Table 2.2), and types and severity of damage. Trees that are identified as snags, stubs, stumps, or logs are usually not measured during subsequent plot remeasurements, but their status is often recorded.  Study Sites This study focused on five old-growth white spruce and five old-growth lodgepole pine sites. In the Alberta foothills, old-growth characteristics develop approximately 160 years after a stand-replacing disturbance (Morgantini and Kansas, 2003). To be included in this study, the PSPs had to meet two criteria:  31  2.2. Methods 1. The last stand-replacing disturbance was at least 160 years prior to program establishment in 1956. 2. At the last remeasurement, sites were at least 65% Picea glauca or Pinus contorta by both basal area and stem density. Of the approximately 3200 PSPs, nine spruce and 18 pine plots met the above criteria. These plots were assessed for accessibility, and ten study sites (five spruce, five pine) were randomly selected. For clarity, “spruce” and “pine” refer to site type and “Picea” and “Pinus” refer to trees of those genera. The 10 study sites were from six clusters of PSP plots (Figure 2.1, Table 2.1). All sites are 0.08 ha in size and nine sites have a slope of less than 30% (Table 2.1). Site CS1 has a 61–70% slope. The sites cover a range of densities, from 19.63 m2 /ha to 54.1 m2 /ha basal area or 852 to 2259 trees/ha (Table 2.3). All sites have a mesic moisture regime, a low bush cranberry or Labrador tea ecosite classification (Beckingham et al., 1996), and stand origin dates of 1710 to 1770, based on interpretation of fire and forest cover maps (Table 2.1). The plots were established between 1958 and 1961. Six of the sites (clusters CP1, CS1, and CS3) were remeasured three times (including the initial measurement), three were remeasured four times (CP2, CS2), and one was remeasured five times (CP3).  Chronology Development To build species- and site-specific tree-ring chronologies, I extracted increment cores from living canopy-dominant trees at all six PSP plot clusters. 32  2.2. Methods  Table 2.1: Physical attributes of old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) sites. All data except location were provided by Hinton Wood Products.  33  2.2. Methods  Table 2.2: Tree status codes used by Hinton Wood Products (Hinton Wood Products, 2006).  Table 2.3: Ecological attributes of old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) sites. All data were provided by Hinton Wood Products. Ecosite classification is based on Beckingham et al. (1996).  34  2.2. Methods  Figure 2.1: Old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) study sites in the Hinton Wood Products Forest Management Area of the Foothills Research Institute. Natural subregions are based on Beckingham et al. (1996).  35  2.2. Methods Trees were sampled outside the boundaries of the PSPs to ensure the integrity of these long-term research plots and, when there were several study sites in a PSP cluster, I sampled trees between the plots when possible. Trees were selected to be large canopy dominants, free of injury and disease. I tried to include only trees with sound heartwood; however, this was impossible with some of the largest trees. Trees were cored ≤30 cm above the ground. I cored approximately 30 trees per site, depending on core quality.  Sampling and Classification of Snags Using the PSP records, I located all tagged trees and identified each as living or dead, and subsequently as a snag or log. Contrary to the classification used by Hinton Wood Products, I did not differentiate between snags with and without a broken stem (stub and snag, respectively; see Table 2.2). For each snag, I identified species, measured dbh, and assessed decay class. A nine-stage decay classification for snags was used, based on soundness of heartwood and sapwood and tree structural integrity (Table 2.4; Thomas et al., 1979). All snags included in this study were ≥7.0 cm dbh, as per the PSP tagging limit, and ≥1.0 m in height. In addition, all snags included in this study either had their original tag or I could establish their identity using the PSP data. I collected two increment cores from a stratified random subsample of trees. Up to four snags per decay class were sampled at each site, limited by the number of snags in each represented decay class. Where possible, cores included cambium and bark, and I avoided coring through basal scars and other stem anomalies. To account for potential missing rings, I noted 36  2.2. Methods Table 2.4: Snag decay classes based on branch order and structural integrity and soundness of sapwood and heartwood (Thomas et al., 1979).  missing bark and rotting heartwood and sapwood. I collected increment core samples from 75 snags.  Sampling and Classification of Logs I located all pieces of downed wood and assessed each for inclusion in the study. Tags were found on only 31% of the logs, and the study was restricted to these tagged logs. To be included, a log had to be ≥7.0 cm diameter at mid-length and ≥1.0 m in length. In addition, at least 50% of the log had to be within the plot boundaries. This last criterion meant that some tagged trees which fell outside the plot were excluded from study. Downed material included uprooted trees, snapped trees, snapped tops, and large branches. For each log, I identified species, measured diameter at the top, middle, and  37  2.2. Methods Table 2.5: Log decay classes based on structural integrity and soundness of sapwood and heartwood (Maser et al., 1979).  bottom of the log, measured length, and assessed decay class. A five-stage decay classification for logs was used, based on structural integrity and wood condition (Table 2.5; Maser et al., 1979). Using a chainsaw, I collected a single cross-sectional disc from each of a stratified random subsample of logs. Before cutting, the sample section was bound tightly with duct tape to ensure structural integrity during cutting, transportation, and processing. In addition, highly decayed samples were wrapped in plastic cling wrap and frozen. Up to four logs per decay class were sampled at each site, limited by the number of tagged logs in each represented decay class. Samples were taken from the most structurally sound section of the log and included cambium and bark where possible. To account for potential missing rings, I noted missing bark and rotting  38  2.2. Methods heartwood and sapwood. I collected cross-sectional discs from 57 logs.  2.2.2  Dendrochronlogical Analyses  Chronology Development I air dried the cores, mounted them on wooden supports, and sanded them using progressively finer sandpaper to 600 grit to view the rings using a microscope (Stokes and Smiley, 1996). Each core was visually crossdated and ring-width series were measured to the nearest 0.001 mm using a Velmex bench interfaced with MeasureJ2X measuring software (VoorTech Consulting, 2004). The ring-width series were statistically crossdated and master ring-width chronologies were developed using the program COFECHA (Grissino-Mayer, 2001; Holmes, 1983). COFECHA uses correlation analysis to compare each tree-ring series against all other series to ensure that calendar years are properly assigned to each ring. Accurately dated and highly correlated cores were combined in master chronologies to represent average tree growth at each site (Fritts, 1976). The chronologies were standardized to remove age-related growth trends using the program ARSTAN (Cook, 1985). I used a horizontal standardization method which normalized each ring-width series to show departures from the mean growth rate and averaged all series to represent long-term and short-term trends in tree growth at a site.  39  2.2. Methods Cores and Cross-Sectional Discs from Snags and Logs The snag cores were prepared according to the standard dendrochronological techniques described above (Stokes and Smiley, 1996) and ring widths were measured to the nearest 0.001 mm (VoorTech Consulting, 2004). For logs, discs in decay classes 1 and 2 were air dried then sanded with progressively finer sandpaper. More decayed samples were air dried, reinforced with hot glue, and sanded. Log ring widths were measured for each of two radii, which were selected based on series length and the presence of bark and intact sapwood. To estimate year of tree death, I used COFECHA (GrissinoMayer, 2001; Holmes, 1983) and TSAP-Win (Rinn, 2003) to statistically crossdate the ring-width series of the dead trees against the species- and sitespecific master chronologies, and assigned a calendar year to the outermost ring on the sample. To verify crossdating, I used the Math Graph function of TSAP-Win to visually compare the ring-width series of the individual cores and radii against the standardized ring-width series of the appropriate species- and site-specific chronologies, assessing the calendar dates of narrow rings in the samples relative to narrow rings in the chronologies. When there was a discrepancy in YOD dates from pairs of cores or pairs of radii, I used the more recent YOD date to estimate the year of tree death. Time since death was calculated as 2008 minus the YOD.  40  2.2. Methods  2.2.3  Year-of-Death Analyses  Comparison of YOD Estimates from Pairs of Cores and Pairs of Radii I compared the outermost ring date from pairs of cores and pairs of radii to identify crossdating errors due to sample quality and decay, missing bark, narrow outer rings, and asymmetrical cambial death. I labelled each outermost ring date in a pair; the more recent of the two dates became year 1 (core 1 or radius 1) and the least recent date became year 2 (core 2 or radius 2). I calculated Spearman correlation coefficients (ρ) for the pairs of dates. Simple linear regression of year 1 versus year 2 was used to used to estimate the slope parameters for the equation  year 2 = α + β ∗ year 1  (2.1)  where α and β are parameter estimates for the intercept and slope, respectively (SAS, 2002). I used the slope parameter estimate to interpret how the correlations of outer ring date pairs varied with snag and log age. If the correlation between pairs of dates did not increase or decrease with snag and log age, I would expect β = 1. In addition, if all pairs of dates were perfectly correlated, I would expect α = 0. Since I used regression descriptively and did not use any hypothesis tests, I did not check for assumptions of equal variance and normality. Parameter estimates are reported plus or minus the standard error.  41  2.2. Methods Verification of YOD Dates Using PSP Data For trees that died after plot establishment (1956–1961, see Table 2.1), I verified YOD dates estimated by crossdating by comparing them to the recorded interval of death (IOD) from the PSP data. I restricted this analysis to trees that died post plot establishment as I could not calculate the magnitude of error of YOD estimates for trees that died prior to plot establishment. IOD was defined as follows: the upper limit was the remeasurement year in which the tree was recorded as dead and the lower limit was the preceding remeasurement year. For example, if a tree was recorded as a snag in 1999 but was alive in 1966, its IOD would be 1966–1999. If a tree (now a snag or log) was alive during the last remeasurement year, the upper limit of the IOD was the year that the coarsewood was sampled, 2008, and the lower limit was the last remeasurement year (e.g. 1999). I compared the estimated YOD dates with the recorded IOD dates and compared the number of trees with YOD dates that occurred within, before, and after the IOD. I also calculated and compared the magnitude of error, defined as follows: 1. YOD precedes IOD: error = YOD – lower limit of IOD; 2. YOD succeeds IOD: error = YOD – upper limit of IOD; 3. YOD occurs within IOD: error = 0. I plotted the absolute value of error versus time since death to interpret how the magnitude of error changed with increasing snag or log age. Error  42  2.3. Results Table 2.6: Summary statistics for site-specific white spruce (CS1–CS3) and lodgepole pine (CP1–CP3) master ring-width chronologies.  by decay class was plotted to assess the magnitude of error associated with increasing stages of decay.  2.3 2.3.1  Results Chronology Development  I developed six new species-specific chronologies (Table 2.6). A total of 87 increment cores were used to create three site-specific white spruce master chronologies. Each of the chronologies included 28–30 trees, and were 96– 262 years in length. Two chronologies, CS1 and CS3, began around 1700, while CS2 began in 1889 (Figure 2.2). A total of 84 increment cores were used to create the three lodgepole pine master chronologies. Each of the chronologies included 26–32 trees, and were 111–171 years in length. All three chronologies began in the 1700s, ranging from 1718–1774 (Figure 2.3). All six chronologies ended in 2007. For all chronologies, the sample depth, which indicates the number of 43  2.3. Results  Figure 2.2: Picea site-specific master ring-width chronologies for plot clusters (a) CS1, (b) CS2, and (c) CS3. The top panels (i) depict the chronologies which were standardized with a horizontal line through the mean. The bottom panels (ii) illustrate sample depth, the number of samples included in the chronology at a particular point in time.  44  2.3. Results  Figure 2.3: Pinus site-specific master ring-width chronologies for plot clusters (a) CP1, (b) CP2, and (c) CP3. The top panels (i) depict the chronologies which were standardized with a horizontal line through the mean. The bottom panels (ii) illustrate sample depth, the number of samples included in the chronology at a particular point in time.  45  2.3. Results cores used in the chronology at any given year, peaked in the 1900s (Figures 2.2, 2.3). Two of the chronologies, CS1 and CS3, had the greatest sample depth at the end of the chronology (28 and 30 cores, respectively). The sample depth of the other four chronologies peaks in the mid-1900s: CS2 from 1933–2002 (30 cores), CP1 from 1930–1979 (26 cores), CP2 from 1921– 1929 (30 cores), and CP3 from 1964–1989 (21 cores). All of the sample depth curves resemble a horizontal asymptotic curve, with the exception of CP3. This may be explained by the age structure of the stand. While the other sites had a single cohort of canopy dominant trees, the trees used to construct the CP3 chronology came from two distinct cohorts, which did not always crossdate well with one another. Summary statistics for the chronologies indicate that trees within a site responded similarly to variation in environmental factors (Table 2.6). Series intercorrelation quantifies the strength of the signal common to all trees in a chronology and is the average correlation derived by correlating each individual series with all other series combined (Fritts, 1976). Series intercorrelation values ranged from 0.49–0.64, indicating that trees at each site responded similarly to coarse-scale climate and environmental variation. These high intercorrelation values also suggest that crossdating within each chronology was robust. Mean sensitivity, which measures ring-width variability within successive years in a chronology, ranged from 0.17–0.21, indicating relatively low variation in ring widths from one year to the next. First order autocorrelation, the statistical association of each value in a time series with the previous value, ranged from 0.80–0.86, indicating a high correlation between the ring-width measurements in successive years. 46  2.3. Results  2.3.2  Comparison of YOD Estimates from Pairs of Cores and Pairs of Radii  Snags I successfully crossdated and obtained YOD estimates for 71 of 75 snags. Of these 71 crossdated snags, YOD estimates were obtained for each of two cores for 99% of the trees (n = 70). YOD dates from pairs of Picea snag cores were more correlated than pairs of YOD dates from Pinus snags based on linear regression (ρ = 0.99 and ρ = 0.81, respectively; Figure 2.4). The regression lines for both species deviated from the expected y = x line. Parameter estimates for Equation 2.1 were as follows: • Picea snags: α = −77.43 ± 31.81, β = 1.04 ± 0.016; • Pinus snags: α = −34.84 ± 193.45, β = 1.01 ± 0.098. Since both β estimates were ≥1, this suggests that the correlation between pairs of cores increased in younger snags. In both species, errors in YOD estimates were attributed to combinations of three factors: (1) low core quality, often a result of decayed sapwood, (2) highly suppressed outer rings, and (3) asymmetrical cambial death. In pine snags, these factors more often resulted in mis-matched YOD dates than in spruce snags. When there was a discrepancy in YOD dates from pairs of cores, I used the more recent YOD date to estimate the year of tree death.  47  2.3. Results  Figure 2.4: Scatter plot of year-of-death dates for pairs of cores from (a) Picea snags (ρ = 0.99, n = 32 pairs of cores) and (b) Pinus snags (ρ = 0.81, n = 37 pairs of cores). The open circles represent pairs of cores with the same year-of-death date, which was obtained through crossdating. All other symbols depict pairs of cores in which the year-of-death dates did not match. The factor(s) explaining the mis-matches are indicated in the caption.  48  2.3. Results Logs I successfully crossdated and obtained YOD estimates for 54 of 57 logs. Of these 54 crossdated logs, YOD estimates were obtained for each of two radii for 96% of the logs (n = 52). YOD dates from pairs of radii from Picea logs and Pinus logs were highly correlated (ρ = 0.98 for both species; Figure 2.5). These results, however, must be interpreted cautiously. All logs included in this analysis had retained their tags, which meant that they had relatively sound outerwood. Logs with more decayed outerwood would potentially have a greater discrepancy in YOD dates from pairs of radii. As with the snags, the regression lines for both species deviated from the expected y = x line. Parameter estimates for Equation 2.1 were as follows: • Picea logs: α = −78.62 ± 52.89, β = 1.04 ± 0.027; • Pinus logs: α = 17.79 ± 67.47, β = 0.99 ± 0.034. The β estimates suggest that the correlations between pairs of radii increased in younger Picea logs but decreased in younger Pinus logs. In both species, errors in YOD estimates were attributed to combinations of four factors: (1) low sample quality resulting from decayed sapwood, (2) highly suppressed outer rings, (3) asymmetrical cambial dieback, and (4) missing bark. When there was a discrepancy in YOD dates from pairs of radii, I used the more recent YOD date to estimate the year of tree death.  49  2.3. Results  Figure 2.5: Scatter plot of year-of-death dates for pairs of radii from (a) Picea logs (ρ = 0.98, n = 29 pairs of radii) and (b) Pinus snags (ρ = 0.98, n = 23 pairs of radii). The open circles represent pairs of radii with the same year of death date, which was obtained through crossdating. All other symbols depict pairs of radii in which the year of death dates did not match. The factor(s) explaining the mis-matches are indicated in the caption.  50  2.3. Results  2.3.3  Verification of YOD Dates Using PSP Data  Snags Estimated snag YOD dates ranged from 1840–2006 for Picea snags and 1866–2002 for Pinus snags. Observed IOD dates ranged from before plot establishment to 2008 for both snag genera. Estimated YOD dates more often occurred within the IOD for Picea snags than for Pinus snags (Figure 2.6), suggesting that YOD estimates may be more accurate for Picea snags than for Pinus snags. For Picea snags, 78% of the estimated YOD dates occurred within the recorded IOD, 19% occurred before the recorded IOD, and 3% occurred after the recorded IOD. For Pinus snags, 54% of the estimated YOD dates occurred within the recorded IOD and 46% occurred before the recorded IOD (Figure 2.6). For spruce snags, crossdating success was highest in decay class 4: 62% of decay class 3, 83% of decay class 4, and 78% of decay class 5 dates occurred within the IOD (Figure 2.6). Crossdating success was highest in decay class 4 and decay class 5 for pine snags: 33% of decay class 3, 60% of decay class 4, and 60% of decay class 5 dates occurred within the IOD. Of the subset of Picea trees that died after plot establishment, seven of 23 Picea snags had YOD dates that did not occur within the IOD (Figure 2.7). Of these, six were within 10 years of the IOD, and one snag had a YOD date that preceded the IOD by 24 years. On average, YOD estimates preceded the IOD by −1.74±5.30 years (mean ± standard deviation). Since most (six of seven) erroneous YOD estimates occurred before the IOD, the corresponding time-since-death values are overestimated. The magnitude of  51  2.3. Results  Figure 2.6: The percentage of year of death estimates from crossdating, by decay class, that occurred before, within, and after the observed interval of death dates from permanent sample plot data. The panels depict (a) Picea snags (n = 32), (b) Pinus snags (n = 39), (c) Picea logs (n = 30), and (d) Pinus logs (n = 24).  52  2.3. Results  Figure 2.7: Scatter plots of the error in year of death (YOD) estimates by time since death for (a) Picea snags (n = 22), (b) Pinus snags (n = 35), (c) Picea logs (n = 23), and (d) Pinus logs (n = 19). The magnitude of error is calculated as the number of years the YOD precedes or succeeds the observed interval of death dates from permanent plot data, and is expressed as the absolute value. Time since death is 2008 minus the YOD. Symbols indicate if the estimated YOD precedes, succeeds, or occurs within the interval of death.  53  2.3. Results error of YOD estimates appeared to increase slightly with time since death (Figure 2.7). The greatest magnitude of error was in decay class 5 (−24 years), followed by decay class 3 (Figure 2.8). In contrast to Picea, 18 of 35 Pinus snags had estimated YOD dates that did not occur within the IOD (Figure 2.7). Of these, 12 were within 10 years of the IOD, and one snag had a YOD date that preceded the IOD by 26 years. On average, YOD estimates preceded the IOD by −5.46 ± 7.85 years. All of the erroneous YOD estimates preceded the IOD, again suggesting that the corresponding time-since-death values are overestimated. The magnitude of error of YOD estimates appeared to increase slightly with time since death, but there was a lot of variability in the data. The greatest magnitude of error was in decay class 4 (−26 years); however all three decay classes had similar ranges of error, despite differences in sample size (Figure 2.8).  Logs Estimated YOD dates ranged from 1833–2008 for Picea logs and 1906–2002 for Pinus logs. Observed IOD dates ranged from before plot establishment to 2008 for both log genera. Estimated YOD dates more often occurred within the IOD for Picea logs than for Pinus logs (Figure 2.6), again suggesting that YOD estimates may be more accurate for Picea logs than Pinus logs. For Picea logs, 80% of the estimated YOD dates occurred within the recorded IOD, 17% occurred before the recorded IOD, and 3% occurred after the recorded IOD. For Pinus snags, 58% of the estimated YOD dates occurred within the recorded IOD and 42% occurred before the recorded IOD (Figure 2.6). 54  2.3. Results  Figure 2.8: Frequency of error in year of death estimate of Picea and Pinus snags (a–c) and logs (d–g) by decay class. Sample size is indicated in parentheses. Error is calculated as the number of years the YOD preceded or succeeded the observed interval of death dates from permanent plot data.  55  2.3. Results For Picea logs, the majority of logs in a single decay class occurred within the IOD (Figure 2.6). Crossdating success decreased with increasing stages of decay: 100% of decay class 1, 86% of decay class 2, 75% of decay class 3, and 67% of decay class 4 occurred within the IOD. For Pinus logs, crossdating success was highest for logs in intermediate decay classes: 60% of decay class 2 and 71% of decay class 3 logs occurred within the IOD. There was only one Pinus log in decay class 1 and one log in decay class 4. For both these logs, the estimated YOD date occurred before the IOD. Of the subset of Picea trees that died after plot establishment, six of 23 Picea logs had estimated YOD dates that did not occur within the IOD. Of these, four were within 10 years of the IOD (Figure 2.7). One log had a YOD date that preceded the IOD by 17 years. On average, YOD estimates preceded the IOD by −1.39±5.08 years (mean ± standard deviation). Again, since most (five of six) erroneous YOD estimates preceded the IOD, the corresponding time-since-death values are overestimated. The magnitude of error of YOD estimates appeared to increase with time since death (Figure 2.7). For all logs in decay class 1 and most logs in decay class 2, the YOD dates occurred within the IOD (Figure 2.8). Decay class 4 had the widest range of error, from 0 to −17 years, but also had the largest sample size (n = 11). Ten of 20 Pinus logs had estimated YOD dates that did not occur within the IOD (Figure 2.7). Of these, eight were within 10 years of the IOD. One log had a YOD date that preceded the IOD by 36 years. On average, YOD estimates preceded the IOD by −4.05 ± 8.35 years. All of the erroneous YOD estimates preceded the IOD, again suggesting that the corresponding 56  2.4. Discussion time-since-death values are overestimated. The magnitude of error of YOD estimates appeared to increase with time since death (Figure 2.7). As with Picea logs, decay class 4 had the widest range of error, from 0 to −36 years, but also had the largest sample size (n = 13; Figure 2.8).  2.4 2.4.1  Discussion Within-Tree YOD Estimates  Overall, the YOD dates from pairs of cores and pairs of radii were highly correlated despite numerous mis-matches in YOD dates (Figures 2.4, 2.5). While poor sample quality was responsible for some discrepancies in pairs of YOD dates, highly suppressed outer rings also caused mis-matches in YOD dates. When samples had highly suppressed outer rings, most suppressed rings were just counted, not measured (and the number of unmeasured outer rings was added to the YOD date obtained through crossdating). Errors in counting these very narrow rings could have caused YOD mis-matches. In addition, some rings may have been locally absent during the period of suppression (Mast and Veblen, 1994). Suspected asymmetrical cambial mortality (Kelly et al., 1992) also caused mis-matches for a small number of pairs of YOD dates (snags n = 4; logs n = 5). Despite mis-matches in YOD estimates, paired YOD dates were still highly correlated. Consequently, I believe that decay, erosion, and fragmentation, and suppression and asymmetrical cambial mortality generally did not cause significant errors in aging spruce and pine coarsewood. Based on the high correlations for pairs of outer ring dates, I have confidence in my 57  2.4. Discussion crossdating and resulting YOD dates and believe that the most recent YOD date for any given pair closely reflects the calendar year of a tree’s outermost ring. The high correlations in paired YOD estimates was expected, as I was only working with samples that had not lost their tags, meaning that at least part of the tree was not substantially decayed. I would expect the snags and logs with missing tags to be more decayed, and may have a larger mis-match in YOD dates. Consequently, I would not expect to see such a high correlation in paired YOD dates across all snags and logs at my study sites. Instead, this analysis indicates my confidence in the YOD dates I obtained from crossdating and suggests that the YOD estimates from these decayed samples closely represent the calendar year of a tree’s outermost ring.  2.4.2  Verification of YOD Estimates  The Picea YOD estimates were consistently more accurate for all snags and logs than the Pinus estimates (Figure 2.6). This analysis included all trees, including those that died prior to plot establishment. These trees have a wide observed IOD (e.g. prior to 1958), which means that the likelihood of their YOD dates occurring within the IOD by chance was higher than for those trees with more a more narrow IOD (e.g. 1996–2002). Thus, these trees inflate the perceived success rate. YOD dates most often preceded the IOD for Pinus trees, but some Picea snag and log YOD estimates also preceded the IOD. Poor sample quality likely explains some of this error, but only 6% of Picea snags and 58  2.4. Discussion 15% of Pinus snags were missing bark or had decayed sapwood. Sample quality may be more important in YOD error for logs, as 33% of Picea logs and 54% of Pinus logs had poor quality samples; however, despite the discrepancy in sample quality between snags and logs, a similar percentage of spruce snags and logs and a similar percentage of pine snags and logs had YOD dates that preceded the IOD. In addition, the high correlation of YOD dates for pairs of cores and pairs of radii suggests that sample quality is not a significant problem, as I would expect more discrepancy in paired YOD dates if a significant number of outer rings were missing due to decay. Instead, it is possible that as the trees are declining, they are not allocating carbon to radial growth thus the calendar year of the outermost ring does not accurately reflect the year of tree death. This has been previously documented in declining fir trees (Abies spp.), with up to 33 rings missing from sections of suppressed outer rings at the base of the tree (Marchand, 1984; Parent et al., 2002). One Picea snag and one Picea log had YOD dates that succeeded the IOD, by three and nine years respectively. Since both YOD dates were within a decade of the IOD, I suspect that both trees were severely declining during the remeasurement in which they were first reported dead, and may have only had a small number of green needles that were not seen during assessment. When YOD estimates precede the IOD, the resulting time since death (i.e. snag or log age) is overestimated. Overestimates can occur when samples are missing rings due to decay, but this error can be reduced by sampling from the least decayed portion of the snag or log (Daniels et al., 1997) and 59  2.4. Discussion taking multiple core samples from snags, especially when there is missing bark (Mast and Veblen, 1994); however the YOD date obtained from crossdating may still precede the actual year of tree death, even when samples are of high quality. This overestimate is problematic for coarsewood modeling (e.g. DeLong et al., 2004), as the inaccuracy of input snag and log ages may result in inaccuracies in the models. Future studies using dendrometers to measure basal area increase (e.g. Clark et al., 2000) could be useful in understanding ring formation in declining Picea and Pinus trees.  2.4.3  Magnitude of Error in YOD Estimates  Analysis of trees that died after plot establishment quantifies the magnitude of error associated with YOD estimates and how it relates to time since death and decay class; however, this analysis is restricted to trees that died within the last 50 years. This means that error associated with older snags and logs is unknown. Error in YOD estimate increases with snag and log age (time since death). This result is unsurprising, as trees that have been dead for a longer period of time are potentially missing more outer rings due to decay than trees that have died relatively recently. Consequently, the further back in time that we reconstruct coarsewood dynamics, the more uncertain we are about both the accuracy and precision of our YOD estimates. Although this analysis was restricted to trees that died after plot establishment, if I extrapolate back in time, I infer that the magnitude of error increases and is greatest for the oldest snags and logs. Although the general trend is an increase in error with increasing time 60  2.4. Discussion since death, there is a lot of variation in the data. With a much larger sample size, it might be possible to better model the increase in error with time and predict the expected amount of error for trees that died prior to plot establishment. The range in error of YOD estimates is similar for snag decay classes 3, 4, and 5. This is somewhat contradictory with the previous results, as we expect those snags that are more decayed to have greater errors; however, the decay class is assessed for the entire snag, while individual samples are taken from the least decayed portion of the tree. This indicates that we can still obtain high quality samples from snags in more advanced stages of decay. This also suggests that other factors are driving the trend, including lack of ring formation during tree decline. For logs, the greatest range of error is in decay class 3, for both Picea and Pinus; however, this is also the decay class with the largest sample size. The sample sizes for decay classes 1, 2, and 4 are too small to make any inferences. Overall, YOD dates on average preceded the IOD by −1.74 ± 5.30 years for Picea snags, −5.46 ± 7.85 years for Pinus snags, −1.39 ± 5.08 years for Picea logs, and −4.05 ± 8.35 years for Pinus logs. These means could be used by other studies in similar stands as preliminary corrections for YOD estimates obtained from crossdating. Ideally, however, means should be calculated for each remeasurement interval and each decay class in order to create time-specific and decay-class-specific corrections.  61  2.4. Discussion  2.4.4  Applications and Recommendations  It is clear that there is error associated with YOD dates, and the magnitude of error varies between species and position (i.e. snag or log) and across time since death. This error ranged from −1.39 ± 5.08 years for Picea logs to −5.46 ± 7.85 for Pinus snags. Consequently, YOD estimates for snags and logs in these forests, and corresponding calculations of mortality rates, are better represented by a range of dates (e.g. YOD ± error) rather than by a single calendar year. To improve upon these results, a redesigned sampling scheme which stratifies samples by IOD in addition to decay class would enable an understanding of the variability in magnitude of error across the range of timesince-death dates. This approach would still require an initial assessment of the stand, as some trees will have died since the last plot remeasurement and some snags may have transitioned to logs. Depending on the age-class distribution of snags and logs, up to 10 samples per interval could be collected, for example. Having a larger stratified sample would ideally facilitate the creation of age- and decay-class-specific error corrections for YOD estimates. Researchers rarely have access to both PSP data and dendrochronological data; however I encourage other dendrochronologists conducting coarsewood studies in PSPs to compare their YOD estimates to observed IOD dates if the data are available. This comparison would facilitate an understanding of the variability in error of YOD dates across other species, stand types, and landscapes. Until we reach a better understanding of error in YOD dates in other systems, I hope that dendrochronologists will con-  62  2.4. Discussion tinue to explicitly acknowledge the potential ranges of error associated with coarsewood YOD estimates in our publications. Ultimately, the construction of species- and stand-based models of YOD error would allow transferability and incorporation into other studies.  63  2.5. References  2.5  References  Beckingham, J., I. Corns, and J. Archibald. 1996. Field guide to ecosites of west-central Alberta. Special Report 9, Canadian Forest Service Northwest Region, Edmonton, AB. Clark, N. A., R. H. Wynne, and D. L. Schmoldt. 2000. A review of past research on dendrometers. Forest Science, 46:570–576. Cook, E. 1985. A time series analysis approach to tree-ring standardization. Ph.D. thesis, University of Arizona, Tuscon, AZ. Daniels, L. D., J. Dobry, K. Klinka, and M. C. Feller. 1997. Determining year of death of logs and snags of Thuja plicata in southwestern coastal British Columbia. Canadian Journal of Forest Research, 27:1132–1141. DeLong, S. C., L. D. Daniels, B. Heemskerk, and K. O. Storaunet. 2005. Temporal development of decaying log habitats in wet spruce-fir stands in east-central British Columbia. Canadian Journal of Forest Research, 35:2841–2850. DeLong, S. C., S. A. Fall, and G. D. Sutherland. 2004. Estimating the impacts of harvest distribution on road-building and snag abundance. Canadian Journal of Forest Research, 34:323–331. DeLong, S. C., G. D. Sutherland, L. D. Daniels, B. Heemskerk, and K. Storaunet. 2008. Temporal dynamics of snags and development of snag habitats in wet spruce-fir stands in east-central British Columbia. Forest Ecology and Management, 255:3613–3620. 64  2.5. References Fritts, H. 1976. Tree Rings and Cimate. Academic Press, New York, NY. Grissino-Mayer, H. D. 2001. Evaluating crossdating accuracy: a manual and tutorial for the computer program COFECHA. Tree-Ring Research, 57:205–221. Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, J. D. Lattin, N. H. Anderson, S. P. Cline, N. G. Aumen, J. R. Sedell, G. W. Lienkaemper, K. Cromack, and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research, 15:133–302. Hinton Wood Products. 2006. Permanent Growth Sample Program Manual, version 16. West Fraser Mills, Hinton, AB. Holmes, R. 1983. Computer-assisted quality control in tree-ring dating and measuring. Tree-Ring Bulletin, 43:69–78. Jonsson, B. G. 2000. Availability of coarse woody debris in a boreal oldgrowth Picea abies forest. Journal of Vegetation Science, 11:51–56. Kelly, P., E. Cook, and D. Larson. 1992. Constrained growth, cambial mortality, and dendrochronology of ancient Thuja occidentalis on ciffs of the Niagara Escarpment — an eastern version of the bristlecone pine. International Journal of Plant Sciences, 153:117–127. Lewis, S. L., O. L. Phillips, D. Sheil, B. Vinceti, T. R. Baker, S. Brown, A. W. Graham, N. Higuchi, D. W. Hilbert, W. F. Laurance, J. Lejoly, Y. Malhi, A. Monteagudo, P. N. Vargas, B. Sonke, N. Supardi, J. W. 65  2.5. References Terborgh, and R. V. Martinez. 2004. Tropical forest tree mortality, recruitment and turnover rates: calculation, interpretation and comparison when census intervals vary. Journal of Ecology, 92:929–944. Marchand, P. 1984. Dendrochronology of a fir wave. Canadian Journal of Forest Research, 14:51–56. Maser, C., R. Anderson, K. Cromack, J. T. Williams, and R. E. Martin. 1979. Dead and down woody material. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 78–95. U.S. Department of Agriculture Forest Service, Washington, DC. Mast, J. N. and T. T. Veblen. 1994. A dendrochronological method of studying tree mortality patterns. Physical Geography, 15:529–542. Morgantini, L. E. and J. L. Kansas. 2003. Differentiating mature and oldgrowth forests in the Upper Foothills and Subalpine subregions of westcentral Alberta. Forestry Chronicle, 79:602–612. Newberry, J. E., K. J. Lewis, and M. B. Walters. 2004. Estimating time since death of Picea glauca x P. engelmannii and Abies lasiocarpa in wet cool sub-boreal spruce forest in east-central British Columbia. Canadian Journal of Forest Research, 34:931–938. Nord-Larsen, T. 2006. Modeling individual-tree growth from data with highly irregular measurement intervals. Forest Science, 52:198–208. Parent, S., H. Morin, and C. Messier. 2002. Missing growth rings at the 66  2.5. References trunk base in suppressed balsam fir saplings. Canadian Journal of Forest Research, 32:1776–1783. Rinn, F. 2003. TSAP-Win User Reference. Rinntech, Heidelberg, Germany. SAS. 2002. SAS. SAS Institute Inc., Cary, NC. Stokes, M. A. and T. L. Smiley. 1996. An Introduction to Tree-Ring Dating. University of Arizona Press, Tuscon, AZ, 2nd edition. Storaunet, K. O. and J. Rolstad. 2004. How long do Norway spruce snags stand? Evaluating four estimation methods. Canadian Journal of Forest Research, 34:376–383. Takaoka, S. 1993. The effect of missing rings on stand-age estimation of evenaged forests in northern Hokkaido, Japan. Ecological Research, 8:341–347. Thomas, J., R. Anderson, C. Maser, and E. Bull. 1979. Snags. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 60–77. U.S. Department of Agriculture Forest Service, Washington, DC. Vanderwel, M. C., J. R. Malcolm, and S. M. Smith. 2006. An integrated model for snag and downed woody debris decay class transitions. Forest Ecology and Management, 234:48–59. VoorTech Consulting. 2004. MeasureJ2X. VoorTech Consulting, Holderness, NH.  67  Chapter 3  Coarsewood Decay Dynamics and Wildlife Habitat2 3.1 3.1.1  Introduction Ecological Context  Sustainable management of forested landscapes requires the retention of key structural components that maintain ecological integrity (Lindenmayer and Franklin, 2002). Although coarsewood has historically received little attention in forest management and planning, the ecological value of snags and logs is now well understood by many forest ecologists and managers (Bunnell and Dunsworth, 2009; Harmon et al., 1986). Some authors estimate that, globally, approximately 20% of biodiversity is associated with coarsewood (Grove, 2001; Hunter, 1990). Terrestrial coarsewood serves myriad roles in forest ecosystems: snags and logs provide a nutrient-rich substrate on which plants and fungus to grow; they provide essential habitat for vertebrates, invertebrates, and microorganisms; and their decomposition has an integral role in terrestrial nutrient and carbon cycles (Harmon et al., 1986). 2  A version of this chapter will be submitted for publication. Jones, E.L. and L.D. Daniels. Decay dynamics of coarsewood habitat in old-growth Picea and Pinus stands.  68  3.1. Introduction Coarsewood can also serve as an important biological legacy following a disturbance, linking old and new communities by providing structural and functional continuity through time (Perry, 1994). Coarsewood provides essential habitat for wildlife species, which use snags and logs for perching, concealment, breeding, resting, and feeding (Harmon et al., 1986). The loss of these critical structures can lead to steep declines in wildlife populations. For example, a century of intensive timber management in Scandinavia has resulted in the depletion of 90% of large trees and snags and a corresponding reduction of large-diameter logs, resulting in a decline in wildife diversity and populations (Lindenmayer and Franklin, 2002). In Sweden, approximately 50% of endangered forestdwelling wildlife species depend on coarsewood for all or part of their life cycle (Berg et al., 1994). Thus, it is widely agreed that future sustainable forest management strategies must consider coarsewood as an important habitat element. In British Columbia (Keisker, 2000) and Alberta (Crampton and Barclay, 1998; Norton et al., 2000), both common and threatened wildlife depend on specialized coarsewood structures as habitat. Managing forests to maintain the habitat requirements of each individual species is daunting, but utilizing a habitat-based perspective, rather than a species-focused one, can provide an efficient approach. A recently developed classification scheme for boreal forests in British Columbia (Keisker, 2000) provides a time- and cost-effective means of quantifying the potential habitat value of a snag or log without extensively surveying each piece of coarsewood for evidence of use by a particular species. The system was developed based on the habitat 69  3.1. Introduction needs of the suite of known coarsewood-dependent wildlife species in these forests and delineates habitat needs into 10 snag and six log habitat function types. The habitat function types are grouped into overarching categories: reproduction/resting and foraging for snags, and reproduction/resting, foraging, and travel for logs. The wildlife-habitat function of coarsewood is governed by temporal processes; the types of wildlife habitat that snags and logs provide changes as they decay (DeLong et al., 2005, 2008). Snags decay more slowly than logs, likely due to increased dessication and reduced decomposer activity (Johnson and Greene, 1991; Storaunet and Rolstad, 2002). Decay rates can be species-specific due to different structural properties of wood (Alban and Pastor, 1993; Laiho and Prescott, 2004), but the decay rates of a single species can vary markedly across the landscape and across sites. For example, in contrasting studies of Picea abies logs in Scandinavian boreal forests, decay class 4 logs had been dead for an average of 87 ± 26 years in one study (Storaunet and Rolstad, 2002), but only 34 ± 12 years in another (Jonsson, 2000). In sub-boreal forests in east-central British Columbia, mean time since death of Picea glauca x Picea engelmanii logs was found to be 59 ± 20 years, but some logs were as old as 148 years (DeLong et al., 2005). Decay rates of snags and logs, and the resulting diversity of coarsewood structure and function, is governed by the drivers of decomposition. While decay is often examined in terms of time since death, three interacting categories of factors have been found to influence decay processes: the wood substrate itself, the environmental conditions, and the decomposer community. Tree species, bole size, amount of contact with the ground, en70  3.1. Introduction try points for pathogens, wood anatomy and structure, oxygen and carbon dioxide levels in the wood, and wood nutrient composition have all been identified as substrate-level factors that influence decay processes (Alban and Pastor, 1993; Blanchette, 1995; Foster and Lang, 1982; Harmon et al., 1986; Makinen et al., 2006; Naesset, 1999). Environmental factors are also important, including soil and air temperature, humidity, soil moisture, slope, aspect, and elevation (Harmon et al., 1986; Mattson et al., 1987; Naesset, 1999). The composition and activity of the decomposer community, including fungi, bacteria, insects and other invertebrates, strongly effects decomposition; changes in complex communities can result in heterogeneous decay rates over time and at different locations within a single piece of coarsewood (Blanchette, 1995; Boddy, 2001; Harmon et al., 1986; Mattson et al., 1987).  3.1.2  Research Context and Objectives  In the foothills of Alberta, progressive management initiatives aim to maintain healthy landscapes. Such practises take a landscape-based approach, guided by natural and historical ranges of patterns and structures. The retention of coarsewood and conservation of wildlife habitat is paramount to these goals. While previous research on coarsewood accumulation, persistence, decay, and wildlife-habitat function has been conducted in similar landscapes (e.g. DeLong et al., 2005, 2008; Morgantini and Kansas, 2003; Newberry et al., 2004; Stevenson et al., 2006; Storaunet and Rolstad, 2002), there has been little research on coarsewood in the Foothills Research Institute (FRI) landbase. Given the variability in coarsewood accumulation and persistence, sound management practises require region- and ecosystem71  3.1. Introduction specific studies of coarsewood. In addition, the use of Keisker’s (2000) classification system for wildlife habitat function is relatively new. Only a few published studies have applied it (DeLong et al., 2005, 2008; Stevenson et al., 2006) and more research is required to warrant transferability of results. This study presents research on coarsewood and wildlife habitat from the old-growth spruce- and pine-dominated forests in the FRI landbase. My goal is to provide an inventory and characterization of coarsewood and its decay processes with the intent of understanding wildlife-habitat provision of snags and logs in this managed landscape. To this end, I addressed three groups of research objectives: 1. For each stand type, quantify and characterize the stand-level accumulation of coarsewood, with regards to density, the distribution of decay classes, and size- and species-composition; compare the distribution of snags and logs relative to live trees; 2. Determine time since death of individual Picea and Pinus logs and test for differences among decay classes for both snags and logs; model depletion rates to determine if differences in fall for snags and decay for logs exist between Picea and Pinus; identify the drivers of decay for snags and logs and understand the relative influence of time since death and other factors known to influence decay; 3. Identify how habitat function relates to time since death of snags and logs and how function changes as coarsewood decays; determine if habitat function is related to decay class for snags and logs and whether function is associated with log origin. 72  3.2. Methods  3.2 3.2.1  Methods Field Methods  Study Area This study was conducted in the Hinton Wood Products Forest Management Area of the Foothills Research Institute near Hinton, Alberta (53◦ 24’09” N 117◦ 34’33” W), east of the Rocky Mountains. In this region, upland forests are characterized by closed-canopy coniferous stands, dominated by lodgepole pine (Pinus contorta), white spruce (Picea glauca), and black spruce (Picea mariana). The landscape is a mosaic of mixed-successional pine, pine–white spruce, and white spruce–black spruce stands, created by standinitiating fires with a historical return interval of 100 years (Beckingham et al., 1996).  Establishment and Monitoring of Permanent Sample Plots Between 1956 and 1961, North Western Pulp and Paper (now Hinton Wood Products) established a network of approximately 3200 stand inventory plots, also called permanent sample plots (PSPs), that included a range of forest types and stages of stand development (Hinton Wood Products, 2006). Plot clusters were established every two miles at the intersection of the Alberta Legal Survey grid section lines. Clusters contain four plots, each at a distance of 100.6 m from the cluster centre, at azimuths of 45◦ , 135◦ , 225◦ , and 315◦ . In 1988, the area in the FMA expanded and an additional 114 PSPs were established in 1991. A number of other PSPs have since been  73  3.2. Methods established to represent combinations of species compositions not included in the original network or to assess specific timber management opportunities. Most plots are 0.08 ha, though some recently established plots are 0.04 ha. During PSP establishment, all trees with a diameter at breast height (dbh) greater than the tagging limit were tagged and measured (Hinton Wood Products, 2006). The tagging limit has changed five times since 1956, ranging from 5.0 to 11.6 cm. The current limit is 7.0 cm. The target remeasurement interval for old-growth coniferous PSPs is every five years; however, as many as 33 years have lapsed between successive measurements (Table 3.1). At each remeasurement, numerous tree attributes are recorded, including species, dbh, height, height to live crown, crown fullness, crown radius, crown position, cause of death (if applicable), tree status (live, snag, stub, stump, log, or missing; Table 3.2), and types and severity of damage. Trees that are identified as snags, stubs, stumps, or logs are usually not measured during subsequent plot remeasurements, but their status is often recorded.  Study Sites This study focused on five old-growth white spruce and five old-growth lodgepole pine sites. In the Alberta foothills, old-growth characteristics develop approximately 160 years after a stand-replacing disturbance (Morgantini and Kansas, 2003). To be included in this study, the PSPs had to meet two criteria:  74  3.2. Methods 1. The last stand-replacing disturbance was at least 160 years prior to program establishment in 1956. 2. At the last remeasurement, sites were at least 65% Picea glauca or Pinus contorta by both basal area and stem density. Of the approximately 3200 PSPs, nine spruce and 18 pine plots met the above criteria. These plots were assessed for accessibility, and ten study sites (five spruce, five pine) were randomly selected. For clarity, “spruce” and “pine” refer to site type and “Picea” and “Pinus” refer to trees of those genera. The 10 study sites were from six clusters of PSP plots (Figure 3.1, Table 3.1). All sites are 0.08 ha in size and nine sites have a slope of less than 30% (Table 3.1). Site CS1 has a 61–70% slope. The sites cover a range of densities, from 19.63 m2 /ha to 54.1 m2 /ha basal area or 852 to 2259 trees per hectare (Table 3.3). All sites have a mesic moisture regime, a low bush cranberry or Labrador tea ecosite classification (Beckingham et al., 1996), and stand origin dates of 1710 to 1770, based on interpretation of fire and forest cover maps (Table 3.1). The plots were established between 1958 and 1961. Six of the sites (clusters CP1, CS1, and CS3) were remeasured three times (including the initial measurement), three were remeasured four times (CP2, CS2), and one was remeasured five times (CP3).  Sampling and Classification of Snags Using the PSP records, I located all tagged trees and identified each as living or dead, and subsequently as a snag or log. Contrary to the classification 75  3.2. Methods  Table 3.1: Physical attributes of old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) sites. All data except location were provided by Hinton Wood Products.  76  3.2. Methods  Table 3.2: Tree status codes used by Hinton Wood Products (Hinton Wood Products, 2006).  Table 3.3: Ecological attributes of old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) sites. All data were provided by Hinton Wood Products. Ecosite classification is based on Beckingham et al. (1996).  77  3.2. Methods  Figure 3.1: Old-growth white spruce (SA–SE) and lodgepole pine (PA–PE) study sites in the Hinton Wood Products Forest Management Area of the Foothills Research Institute. Natural subregions are based on Beckingham et al. (1996).  78  3.2. Methods Table 3.4: Snag decay classes based on branch order and structural integrity and soundness of sapwood and heartwood (Thomas et al., 1979).  used by Hinton Wood Products, I did not differentiate between snags with and without a broken stem (stub and snag, respectively; Table 2.2). For each snag, I identified species, measured dbh, assessed decay class, and categorized habitat functions. A nine-stage decay classification for snags was used, based on soundness of heartwood and sapwood and tree structural integrity (Table 3.4; Thomas et al., 1979). To assess habitat function, I modified a ten-stage classification scheme developed for boreal forests in east-central British Columbia (Table 3.5; Keisker, 2000). I added an eleventh function, based on functional characteristics observed in the field. The eleven types are not mutually exclusive, meaning that a snag can be classified as more than one function. All snags included in this study were ≥7.0 cm dbh, as per the PSP tagging limit, and ≥1.0 m in height.  79  3.2. Methods  Table 3.5: Classification of snag function as wildlife habitat (after Keisker, 2000)  80  3.2. Methods I collected two increment cores from a stratified random subsample of trees. Up to four snags per decay class were sampled at each site, limited by the number of snags in each represented decay class. Where possible, cores included cambium and bark, and I avoided coring through basal scars and other stem anomalies. To account for potential missing rings, I noted missing bark and rotting heartwood and sapwood. I collected increment core samples from 81 snags.  Sampling and Classification of Logs I located all pieces of downed wood and assessed each for inclusion in the study. To be included, a log had to be ≥7.0 cm diameter in the middle of the log and ≥1.0 m in length. In addition, at least 50% of the log had to be within the plot boundaries. This last criterion meant that some tagged trees which fell outside the plot were excluded from study. Similarly, untagged trees which fell into the plot were included in the data set. Downed material included uprooted trees, snapped trees, snapped tops, and large branches. For each log, I identified species, measured diameter at the top, middle, and bottom of the log, measured length, assessed decay class, and categorized habitat function type. The diameter measurement in the middle of log is herein after referred to as “diameter at mid-length.” A five-stage decay classification for logs was used, based on structural integrity and wood condition (Table 3.6; Maser et al., 1979). To assess habitat function, I used a six-stage classification scheme (Table 3.7; Keisker, 2000). As with snags, the six types are not mutually exclusive, meaning that a log can be classified as more than one function. In addition, I classified “base type” by identifying 81  3.2. Methods Table 3.6: Log decay classes based on structural integrity and soundness of sapwood and heartwood (Maser et al., 1979).  whether each log originated from a snapped snag, an uprooted tree, or had an unknown origin. Using a chainsaw, I collected a single cross-sectional disc from each of a stratified random subsample of logs. Before cutting, the sample section was bound tightly with duct tape to ensure structural integrity during cutting, transportation, and processing. In addition, highly decayed samples were wrapped in plastic cling wrap and frozen. Up to four logs per decay class were sampled at each site, limited by the number of tagged logs in each represented decay class. Samples were taken from the most structurally sound section of the log and included cambium and bark where possible. To account for potential missing rings, I noted missing bark and rotting heartwood and sapwood. I collected cross-sectional discs from 99 logs.  82  3.2. Methods  Table 3.7: Classification of log function as wildlife habitat (after Keisker, 2000).  83  3.2. Methods  3.2.2  Dendrochronlogical Analyses  The snag cores were air dried, mounted on wooden supports, and sanded with progressively finer sandpaper to 600 grit to view the rings using a microscope (Stokes and Smiley, 1996). Ring widths were measured to the nearest 0.001 mm using a Velmex bench interfaced with MeasureJ2X measuring software (VoorTech Consulting, 2004). For logs, discs in decay classes 1 and 2 were air dried then sanded with progressively finer sandpaper. More decayed samples were air dried, reinforced with hot glue, and sanded. Log ring widths were measured for each of two radii, which were selected based on series length and the presence of bark and intact sapwood. To estimate year of tree death for both snags and logs, I used COFECHA (Grissino-Mayer, 2001; Holmes, 1983) and TSAP-Win (Rinn, 2003) to statistically crossdate the ring-width series of the dead trees against the species- and site-specific master chronologies presented in Chapter 2, and assigned a calendar year to the outermost ring on the sample. To verify crossdating, I used the Math Graph function of TSAP-Win to visually compare the ring-width series of the individual cores and radii against the standardized ring-width series of the appropriate species- and site-specific chronologies, assessing the calendar dates of narrow rings in the samples relative to narrow rings in the chronologies. When there was a discrepancy in YOD dates from pairs of cores or pairs of radii, I used the more recent YOD date to estimate the year of tree death. Time since death was calculated as 2008 minus the YOD.  84  3.2. Methods  3.2.3  Data Analyses  All statistical analyses were performed using SAS 9.1.3 (SAS, 2002). In the following descriptions, the statistical procedure used is indicated in parentheses. All means are reported ± the standard deviation. When analysis of variance (ANOVA) was used, residuals plots were examined to evaluate the assumption of equal variance. The assumption of normality of residuals was assessed with the Shapiro-Wilk test (Neter et al., 1996) and visual evaluations of normality plots. For ANOVA tests, the reported F and p values are for Type III sums of squares. Type III sums of squares are calculated by comparing the full model to the reduced model that excludes the variable of interest and are invariant to the ordering of effects in the model. All statistical analyses had a significance level of α = 0.05, unless otherwise indicated.  Stand-Level Accumulation I calculated log volume as a conical paraboloid (Fraver et al., 2007):  V =  L (5Ab + 5Au + 2 Ab Au ) 12  (3.1)  where L is the log length, Ab is the cross-sectional area of the basal end, and Au is the cross-sectional area of the upper end of the piece. Since I also measured the diameter at mid-length of each log, I increased the accuracy of the volume estimates by summing separate volume calculations for the basal half and upper half of each log (Fraver et al., 2007). All stand-level summaries include the entire sample of snags and logs 85  3.2. Methods at each site. I used ANOVA (PROC GLM) to test the differences between spruce and pine sites of snag size (dbh), density, and total plot basal area, and log size (diameter at mid-length, length, and volume), density, and total plot volume. For each site, the relative density of live trees, snags, and logs by species, the percent distribution of snags by decay class, the percent distribution of logs by decay class, and the percent distribution of snags and logs by size class were plotted. Relative density by species was expressed as a percent of the total sample of live and dead trees (divided into snags and logs). The numbers of live trees were obtained from the permanent sample plot data; I used the list of live trees at the last remeasurement for each site and subtracted any trees that had since transitioned to snags or logs. Counts of snags and logs were based on field observations. Picea glauca and Picea mariana were pooled due to difficulties in distinguishing the species for decayed coarsewood. If a species could not be determined in the field, it was labelled as unknown. Snags and logs were grouped into size classes based on 10 cm increments of dbh for snags and of diameter at mid-length for logs.  Coarsewood Dynamics All analyses of coarsewood dynamics were based on the subsample of snags and logs used for dendrochronological analysis. To interpret coarsewood input over time, I created separate histograms of time since death for Picea and Pinus snags and logs. Differences in time since death for each decay class were tested separately for snags and logs using two-factor ANOVA (PROC GLM), which included decay class and species and an interaction 86  3.2. Methods term for the two factors. Time since death was cube-root transformed to meet the assumptions of normality and equal variance. In the results, the reported means and standard errors were transformed back to the original units. To assess rates of coarsewood depletion, I sorted the snags and logs separately for each species by time since death and summed the number of coarsewood pieces beginning with the earliest year of death. I divided the cumulative values by the total number of samples to construct cumulative relative distributions (Hyatt and Naiman, 2001), to which I fit linear (PROC REG) and nonlinear logarithmic, sigmoidal, exponential, power, logistic, and inverse (PROC NLIN) models. To select the best model, they were evaluated on four criteria (high F value, low p value, small mean squared error, and small relative standard error for the model parameters), as well as a visual comparison of the modelled and observed values. Preliminary analysis indicated that a sigmoidal decay model provided the best fit for snags and logs of both species. The model took the general form     1  yˆ = α +    1+e  t−β1 β2     (3.2)  where yˆ is the predicted proportion of snags or logs less than age t and α, β1 , and β2 are the estimated intercept and shape parameters for the sigmoidal function. This model describes the relationship between time since death and the abundances of snags and logs that were successfully crossdated. The steepness and the rate of change in the slope of the curve describe the net effects of recruitment, storage, and depletion of snags and logs over time. 87  3.2. Methods If coarsewood was recruited and depleted at the same rate, the expected shape of the curve would be a straight line with a constant, negative slope, and the length of the line would equal the length of time required for the coarsewood to be depleted via fall or decay for snags or via decay for logs (Hyatt and Naiman, 2001). To identify the drivers of decay, I used canonical discriminant analysis (PROC DISCRIM and PROC CANDISC). Since I did not have any direct measures of decay (e.g. wood density), I assumed that the decay classes represented a sequential series of decay. Discriminant analysis uses continuous variables to develop canonical discriminant functions to assign individual observations to classes, and indicates which variables are most important for differentiating among groups (i.e. decay classes). The data set was randomly divided to produce a subset for model-building and a subset for model cross-validation. The classification matrices of properly classified snags and logs for the cross-validation subsets are presented in the results. Model selection was iterative and proceeded separately for snags and logs. I used both site-level and snag- or log-level explanatory variables. For snags, the preliminary explanatory variables were time since death (years prior to 2008), species (Picea, Pinus), average annual growth (expressed as mean ring width; mm), diameter at breast height (cm), height class (suppressed, intermediate, intermediate–leaning, dominant, codominant), slope (%), aspect (◦ ), elevation (m.a.s.l.), stand density (basal area per hectare; m2 ), mean annual temperature (◦ C), and mean annual precipitation (mm). For logs, the preliminary explanatory variables were time since death (years prior to 2008), species (Picea, Pinus), average annual growth (expressed as 88  3.2. Methods mean ring width; mm), slope (%), aspect (◦ ), elevation (m.a.s.l.), stand density (basal area per hectare; m2 ), mean annual temperature (◦ C), mean annual precipitation (mm), diameter at mid-length (cm), base type (snapped, uprooted, unknown), length (m), and volume (m3 ). For all class variables (species, snag height, and log base type), I created a set of dummy variables that were interpreted as a group (Hardy, 1993). Values for slope and aspect were obtained from the PSP data; slope was originally grouped into classes (e.g. 11◦ –20◦ ) and aspect was expressed as a cardinal direction. To reduce the number of categorical variables, I assumed that aspect was the midpoint of the range (e.g. 11◦ –20◦ became 15◦ ) and converted the cardinal directions into degrees. Mean annual temperature and mean annual precipitation were modelled using ClimateBC (Wang et al., 2006), an application which uses position and elevation to interpolate climate normals from PRISM data (Daly et al., 2002) for a reference period of 1961–1990. I applied four criteria to select the best model: minimal number of variables to eliminate redundancy and avoid over-parameterization, maximal Mahalabois distances between decay classes, maximal variance explained by the significant discriminant functions, and minimal mis-classification errors for the cross-validated subset. Preliminary analysis indicated a small Mahalabois distance between decay classes 1 and 2 for logs and significant overlap between them. Due to this small Mahalabois distance and the small sample size for decay class 1 (n = 7), decay classes 1 and 2 were pooled for subsequent analyses. For both snags and logs, I ran a preliminary model with all of the above-listed variables. To reduce the number of variables, I evaluated the variables’ loadings on the significant canonical discriminant functions, 89  3.2. Methods using a cut-off level of 0.5 for snags and 0.3 for logs. The cut-off value for logs was set relatively low due to the overall low loadings of the variables. Time since death, elevation, temperature, and precipitation were retained for snags, and time since death, length, volume, and base type were initially retained for logs. I attempted to further reduce the two sets of variables but could not improve the snag model; however, several subsequent iterations of log models reduced the set of variables to produce the best predictive model. Note that none of the models met the assumption of homogeneity of covariance matrices among groups and I did not test for multivariate normality; however, meeting the assumptions is only strictly necessary when inferential statistics are used to discriminate amongst groups. Since this analysis was used descriptively, the presented results remain valid.  Wildlife Habitat To understand the temporal development of wildlife habitat, I tested for differences in mean time since death for each habitat function using twofactor ANOVA (PROC GLM), which included habitat function and species and an interaction term for the two factors. This analysis was restricted to the subset of coarsewood used for dendrochronological analysis. To relate the distribution of snag and log attributes to functions of coarsewood as habitat for wildlife, I used contingency tables and chi-squared goodness-of-fit relative to a random distribution (PROC FREQ). I included the entire population of snags and logs and conducted three sets of independent tests:  90  3.3. Results 1. Snag habitat function with decay classes (3, 4, 5, 6); 2. Log habitat function with decay classes (1, 2, 3, 4, 5); 3. Log habitat function with base types (snapped, uprooted, unknown). Tests were conducted separately for each site type and used an equal number of degrees of freedom for each function within a site type.  3.3 3.3.1  Results Stand-Level Accumulation  I located 322 snags and 405 logs. On average, there were 41 ± 30 snags and 35 ± 12 logs in the spruce-dominated sites and 24 ± 21 snags and 43 ± 6 logs in the pine-dominated sites. Snag densities ranged from 63–1138 stems per hectare (mean: 403 snags/ha, 8.65 m2 /ha; Table 3.8). Log densities ranged from 388–663 logs per hectare (mean: 506 logs/ha, 70.81 m3 /ha). There was no difference in the mean number of snags or logs between spruce and pine sites (snags: p = 0.33; logs p = 0.23). In six of the sites (SA, SB, SC, PB, PC, and PD), the total number of snags and logs was greater than the number of live trees (Figure 3.2, panels a–c and g–i). Note, however, that logs included snapped tops and large branches, meaning that some trees may have been sampled in both the snag and log categories. Spruce sites had a wider range of species than pine sites. Both had Abies balsamea, Picea glauca, Picea mariana, and Pinus contorta coarsewood, but Populus balsamifera, and Populus tremuloides were also found in spruce sites (Figure 3.2). Populus spp. were only found in site SB: 2% of live 91  3.3. Results  Figure 3.2: Species distribution of live trees, snags, and logs at each of ten spruce(a–e) or pine-dominated (f–j) sites. Density is expressed as a percentage of the total population.  92  3.3. Results Table 3.8: Average size of snags and logs in spruce- and pine-dominated sites and corresponding average log volume, snag basal area, and coarsewood density. Means are reported ± standard deviation and differences among site types were tested using ANOVA with α = 0.05 (snags: n = 322; logs: n = 393).  trees, 0.5% of snags, and 4% of logs were Populus spp. (not shown in Figure 3.2). In all sites, species could not be definitively identified for 1–28% of the coarsewood, mostly logs (“unknown” species in Figure 3.2). For most sites, the species distribution of snags and logs was similar to the distribution of live trees; however, in site PE (panel j), live A. balsamea trees composed 18% of the total population (or 41% of the live trees) but I did not find any snags or logs of that species. Similarly, in site PC (panel h), Picea spp. made up 37% of the total population (or 63% of the live trees), but only 9% of the population was Picea coarsewood. In contrast, Pinus coarsewood composed 19% of the population. Overall, in pine sites, the proportion of Picea coarsewood tended to be small relative to the proportion of Picea live trees (panels f–j). Snag and log diameters ranged from 7.0–45.2 cm (dbh for snags, diameter 93  3.3. Results at mid-length for logs; Table 3.8). Log diameter at mid-length (12.84 ± 5.71 cm) was slightly smaller than snag dbh (15.36 ± 6.16 cm); however, for most logs, diameter at mid-length was measured at a position higher than dbh. Mean diameters were not significantly different between the spruce and pine site types (snags: p = 0.17; logs: p = 0.36). Eighty-three percent of coarsewood had a diameter less than 20 cm; 57% were between 10.0 and 19.9 cm and 27% were less than 10.0 cm (Figure 3.3). Live trees had a mean dbh of 19.60 ± 7.03 cm (range: 7.60–64.5 cm). Log length and volume did not differ significantly between the spruce and pine site types (length: p = 0.07; volume: p = 0.21; Table 3.8). Length ranged from 1.80–29.20 m and volume from 0.01–2.95 m3 . In spruce sites, 49% of logs had a snapped base, 33% were uprooted, and 18% had an unknown origin. In pine sites, 43% of logs had a snapped base, 33% were uprooted, and 25% had an unknown origin. Plot snag basal area (m2 /ha) and log volume (m3 /ha) did not differ between site types (basal area: p = 0.65; volume: p = 0.21; Table 3.8). Snag basal area ranged from 1.30–18.92 m2 /ha and log volume from 31.19– 185.71 m3 /ha. Most snags were in decay class 4: 68% were in decay class 4, followed by 27% in decay class 5, 13% in decay class 3, and 0.3% in decay class 6 (Figure 3.4). Picea spp. and P. contorta snags were found at both site types. In addition, at spruce sites, A. balsamea, Populus spp. and “unknown” snags were found. Snags of an unknown species were restricted to decay classes 5 and 6. Most logs were in decay class 3 (60%); 27% were in decay class 4, 9% 94  3.3. Results  Figure 3.3: Size-class distribution of snags, logs, and live at each of ten spruce- (a– e) or pine-dominated (f–j) sites, expressed as a percentage of the total number of coarsewood pieces at a given site. For snags and live trees, diameter was measured at breast height (dbh); for logs, diameter was measured at the mid-point of the length.  95  3.3. Results  Figure 3.4: Decay-class distribution of snags by species at each of ten spruce- (a–e) or pine-dominated (f–j) sites, expressed as a percentage of the total number of snags at a given site.  96  3.3. Results were in decay class 2, 2% were in decay class 1, and 1% were in decay class 5 (Figure 3.5). Picea spp. P. contorta, and “unknown” logs were found at both site types. In addition, A. balsamea and Populus spp. logs were at spruce sites. Logs of an unknown species made up 33% of all logs and the relative proportion of “unknowns” increased with advanced stages of decay.  3.3.2  Coarsewood Dynamics  I successfully obtained year of death estimates for 77 of 81 snags (95%) and 90 of 91 logs (99%). The subsample used for dendrochronological analysis contained 37 Picea and 40 Pinus snags and 44 Picea and 46 Pinus logs. Estimated time since death, expressed in years prior to 2008, ranged from 2–180 years for Picea snags, 6–142 years for Pinus snags, 0–175 years for Picea logs, and 6–133 years for Pinus logs (Figure 3.6). Most trees died within the last 100 years: 81% of Picea snags, 98% of Pinus snags, 89% of Picea logs, and 91% of Pinus logs are estimated to have died since 1908. Time since death of snags was significantly different across decay classes, but not between species (decay class: F = 33.10, p < 0.0001; species: F = 1.08, p = 0.3019; decay class x species: F = 2.35, p = 0.1019; Figure 3.7, panel a). There was, however, a lot of overlap between classes with time since death ranging from 3–34 years for decay class 3 (mean: 15.47 ± 9.10), 2–168 years for decay class 4 (mean: 42.87 ± 36.44), and 40–180 years for decay class 5 (mean: 83.76 ± 36.06). Time since death of logs increased significantly with increasing stages of decay, but did not differ between species (decay class: F = 28.01, p < 0.0001; species: F = 1.15, p = 0.2865; decay class x species: F = 1.96, p = 0.1265; 97  3.3. Results  Figure 3.5: Decay-class distribution of logs by species at each of ten spruce- (a–e) or pine-dominated (f–j) sites, expressed as a percentage of the total number of logs at a given site.  98  3.3. Results  Figure 3.6: Frequency distribution of time since death for Picea and Pinus coarsewood.  99  3.3. Results  Figure 3.7: Boxplots of time since death by decay class for Picea and Pinus (pooled) (a) snags and (b) logs. Differences among means were tested using a two-factor ANOVA. In each panel, boxplots with different letters are significantly different at α = 0.05 (Tukey-Kramer’s test).  100  3.3. Results Figure 3.7, panel b). Decay classes 3 and 4 were not different from one another, but decay classes 1 and 2 differed from each other and from 3 and 4. Like the snags, there was a lot of overlap between classes with time since death ranging from 2–7 years for decay class 1 (mean: 3.33 ± 1.07), 0–76 years for decay class 2 (mean: 20.41 ± 18.83), 12–175 years for decay class 3 (mean: 60.58 ± 38.38), and 20–133 years for decay class 4 (mean: 79.00 ± 31.15). The sigmoidal decay model described in Equation 3.2 produced statistically significant models for Picea and Pinus snags and logs (Picea snags: F = 675.34, p < 0.0001, mean square error (MSE) = 0.0022, n = 37; Pinus snags: F = 1404.00, p < 0.0001, M SE = 0.0433, n = 40; Picea logs: F = 650.67, p < 0.0001, M SE = 0.0027, n = 44; Pinus logs: F = 5508.40, p < 0.0001, M SE = 0.0003, n = 46; Figure 3.8). The parameters for individual models are as follows: • Picea snags: α = 0.1139, β1 = 35.1307, β2 = 26.6128; • Pinus snags: α = 0.0687, β1 = 31.77957, β2 = 13.4188; • Picea logs: α = 0.0936, β1 = 35.7236, β2 = 27.3213; • Pinus logs: α = 0.0304, β1 = 51.8403, β2 = 19.2904. Depletion rates were similar for all coarsewood types, with 50% of the coarsewood depleted in 47 years for Picea snags, 36 years for Pinus snags, 46 years for Picea logs, and 54 years for Pinus logs. The models best fit the observed data for the first 100 years; observed and modelled depletion rates diverged for snags and logs that had been dead for more than a century. In addition, 101  3.3. Results the models poorly described Pinus snags and Picea and Pinus snags and logs that died in the first 10–15 years. Based on multivariate discriminant analysis, only a single discriminating function (Can1) was required to discriminate the decay classes for both snags and logs (Table 3.9). Can1 explained 98.89% of the variance for snags and 99.42% for logs. Discriminant analysis correctly grouped a test sub-set of snags and logs into their correct decay class with 63.81% accuracy for snags and 73.19% for logs (Table 3.10). Mis-classification error rates were highest for decay classes 4 and 5 for snags, likely due to the relatively short Mahalanobis distances between decay class 3–4 and 4–5 (3–4: D2 = 0.91; 3–5: D2 = 5.04; 4–5: D2 = 1.74). Mis-classification was greatest for decay class 4 in logs, again as a result of a relatively short Mahalanobis distance between decay classes 3–4 (2–3: D2 = 2.73; 2–4: D2 = 4.72; 3–4: D2 = 0.30). Note that in preliminary analyses, decay classes 1 and 2 for logs were grouped due to high mis-classification rates of decay class 1, which resulted from a small sample size and the small Mahalanobis distance between 1–2 (see Section 3.2.3). For both snags and logs, time since death was most strongly correlated with Can1 (snags: r2 = 0.98; logs: r2 = 0.99), suggesting that it was the most important variable for predicting and discriminating amongst decay classes (Table 3.11). Preliminary analyses also indicated that elevation, mean annual temperature, and mean annual precipitation were important for discriminating amongst snags decay classes, and that volume was important for discriminating log decay classes. Not surprisingly, elevation, temperature, and precipitation were strongly correlated for snags, but none 102  3.3. Results  Figure 3.8: Depletion curves for Picea (triangles, solid line) and Pinus (circles, dashed and dotted line) (a) snags and (b) logs that died between 1728 and 2008. Logs are divided by base type, which indicates whether each log originated from a snapped snag (white), an uprooted tree (black), or had an unknown origin (white with crosshair). Curves were fit with the sigmoidal function in Equation 3.2. All four models were significant.  103  3.3. Results  Table 3.9: Hypothesis tests for the number of discriminating functions required to discriminate the decay classes for snags and logs (Ho : none of the discriminating functions are needed; Ha : at least one discriminating function is needed). Values denoted with an asterisk (*) were significant at α = 0.05.  Table 3.10: Classification matrix of snag and log decay classes using canonical discriminant analysis for the cross-validation data subset. Bold values were correctly classified.  104  3.3. Results Table 3.11: Pearson’s correlation coefficients for site-level and individual-level explanatory variables and correlations for each variable with the canonical discriminant function.  of the other variables were correlated.  3.3.3  Wildlife Habitat  Of the 11 possible snag habitat functions (see Table 3.5), eight functions were observed in the field (Figure 3.9, panel a). Potential substrates for cavity excavation (S1-2) and trunks with cracks or loose or furrowed bark (S6) were the most common. All six log habitat functions (see Table 3.7) were observed in the field (Figure 3.9, panel b). Small concealed spaces (L2-3), exposed, raised travel lanes (L5), and feeding substrates (L6) were the most common. I observed 151 instances of snag habitat function and 706 instances of log habitat function. Only 33% of snags provided at least one function, while 94% of logs provided one or more functions (Table 3.12). Approximately one-half of the logs had two functions, while less than 10% of the snags  105  3.3. Results  Figure 3.9: Frequency of habitat functions of (a) snags and (b) logs and spruce and pine sites. (See Tables 3.5 and 3.7 for detailed descriptions of the habitat functions.) For snags, functions S1 and S2 were pooled due to similarity in function and a small sample size for function S1. Similarly, for logs, function L3 had a small sample size and was pooled with habitat function L2. Note the difference in y axes.  Table 3.12: Distribution of habitat functions across all observed snags and logs.  106  3.3. Results served multiple functions. Less than 1% of snags and less than 2% of logs had four or more functions. Mean time since death did not vary across habitat function type nor between stand types for both snags and logs (snags: habitat function – F0.05,5 = 0.55, p = 0.74; species – F0.05,1 = 1.08, p = 0.31; species x habitat function – F0.05,4 , p = 0.0636; logs: habitat function – F0.05,4 = 1.37, p = 0.89; species – F0.05,1 = 0.02, p = 0.89; species x habitat function – F0.05,3 = 0.54, p = 0.66; Table 3.13). Wildlife habitat function was provided by snags 9–142 years since death and logs 0–175 years since death. In general, snags in an intermediate stage of decay (decay class 4) most commonly provided habitat function; however, two of the functions with small sample sizes (existing cavities, feeding substrates) were equally as frequent in decay classes 4 and 5 (Table 3.14). Sample sizes were largest for decay class 4, but decay classes 3 and 5 had large enough sample sizes that the exact chi-square tests were able to account for the uneven distribution of observations across decay classes. Snag function as potential wildlife habitat varied significantly across decay class for three of the attributes tested for spruce sites and two of the attributes for pine sites. Potential substrates for cavity excavation were most common in decay class 4 in spruce sites but decay class 5 in pine sites. Trees with cracks or loose or furrowed bark were most common in decay class 4 in spruce sites and were equally as common in decay classes 4 and 5 in pine sites. Snags with the potential to serve as large open-nest supports were most common in decay class 4 in spruce sites and varied non-significantly in pine sites. As with snags, log habitat function was most commonly found in an 107  3.3. Results Table 3.13: Time since death of coarsewood wildlife habitat function for Picea in spruce-dominated sites and Pinus in pine-dominated sites for the subsample of data used for dendrochronological analysis (n = 77 snags, n = 90 logs).  108  3.3. Results  Table 3.14: Distribution of habitat function types (Keisker, 2000) by decay class for snags in spruce- and pine-dominated sites. Values denoted with an asterisk (*) were significant at α = 0.05 using an exact one-way chi-square test and are bold to highlight the differences across classes.  109  3.4. Discussion intermediate stage of decay (decay class 3; Table 3.15). Again, sample sizes were largest for decay class 3, but the other decay classes had large enough sample sizes that the exact chi-square tests were able to account for the uneven distribution of observations across decay classes. Log function as potential wildlife habitat varied significantly across decay class for three of the attributes tested for spruce sites and four of the attributes for pine sites. Small concealed spaces and exposed, raised travel lanes were most common in decay class 3 for both spruce and pine sites. Long concealed spaces were most common in decay class 4 for pine sites, but did not vary significantly across decay class for spruce sites. Habitat function as feeding substrate was most common in decay classes 3 and 4 for both site types. Log function as wildlife habitat varied significantly across base type for three of the attributes (Table 3.16). Large concealed spaces were most common in uprooted trees; in pine sites, large concealed spaces were created exclusively by logs with uprooted bases. Small concealed spaces and exposed, raised travel lanes were most commonly associated with snapped trees for both site types.  3.4 3.4.1  Discussion Stand-Level Accumulation  The amount and size of coarsewood accumulated at the stand level did not vary significantly between site types, with the exception of log length, suggesting that tree size, tree density, mortality rates, and decay rates are similar in old-growth spruce and pine sites in the foothills (Table 3.8). Accu110  3.4. Discussion  Table 3.15: Distribution of habitat function types (Keisker, 2000) by decay class for logs in spruce- and pine-dominated sites. Values denoted with an asterisk (*) were significant at α = 0.05 using an exact one-way chi-square test and are bold to highlight the differences across classes.  111  3.4. Discussion  Table 3.16: Distribution of habitat function types (Keisker, 2000) by base type for logs in spruce- and pine-dominated sites. Values denoted with an asterisk (*) were significant at α = 0.05 using an exact one-way chi-square test and are bold to highlight the differences across classes.  112  3.4. Discussion mulation of coarsewood was highly variable across sites, both in the number of snags and logs and in their basal area and volume. This magnitude of variability is consistent with many similar studies in a range of deciduous and coniferous boreal forests (e.g. Aakala et al., 2007; Clark et al., 1998; Jonsson, 2000; Lee, 1998; Morgantini and Kansas, 2003; Zielonka, 2006). Mean basal area of snags (8.65 m2 /ha) tended to be smaller than in comparable sub-boreal Picea-dominated forests. For example, snag basal areas of 12.1 m2 /ha and approximately 30 m2 /ha have been found in eastcentral British Columbia (Clark et al., 1998) and southern Poland (Zielonka, 2006), respectively; however Zielonka (2006) noted coarsewood levels were higher than previously reported for the study area. In contrast, the mean number of snags found (403 snags/ha) was higher than in comparable studies (Aakala et al., 2007; Clark et al., 1998; Jonsson, 2000; Lee, 1998), suggesting that snags in these forests are small compared to those in sub-boreal forests in other regions. Similarly, mean log volumes (70.81 m3 /ha) were lower and mean log numbers (506 logs/ha) were higher than in comparable studies (Clark et al., 1998; Zielonka, 2006), again suggesting that logs in these stands are small, as a result of small diameters or short lengths or both. Based on both the time-since-fire maps (see Table 2.1) and the treering record (see Chapter 2), all sites are late-successional and can be compared to the “maturing climax” stands described in Morgantini and Kansas (2003), ≥230 years since stand initiation. These stands are located in the Upper Foothills and Subalpine subregions (Beckingham et al., 1996) less than 200 km south of the study area. (The one exception is site SB. The tree-ring record suggests that it may not actually be as old as suggested by 113  3.4. Discussion the fire maps, nor as old as the rest of the sites.) Mean snag basal area (8.65 m2 /ha) was smaller and mean frequency (40 snags/ha) was higher than in the southern stands (10.8 m2 /ha and 314 snags/ha, respectively; Morgantini and Kansas, 2003). Mean log volume (70.8 m3 /ha) was much smaller than in the oldest southern stands (140.9 m3 /ha), but was similar to the volume found in stands 211–230 years old. Two comparable studies grouped their late-successional sites into similar age categories: 201–250 and ≥250 years in Clark et al. (1998) and 211–250 and ≥230 years in Morgantini and Kansas (2003). Both studies found similar levels of snag accumulation in the oldest and second oldest stands, while log accumulation was much higher in the oldest stands. Although the stands in this study (with the possible exception of SB) originated ≥230 years ago, it is possible that they have not yet reached maximal log accumulation. As the stands age, we may see a somewhat constant input of snags and an increase in log volume. Most snags and logs were in the 10.0–19.9 cm size class. Only three sites (SC, SE, PD) had large-diameter snags (≥40 cm dbh) and none of the sites had large-diameter logs (Figure 3.3). While there were also many live trees in the 10.0–19.9 size class, there were more live trees than dead ones in large size classes and live trees were overall larger in diameter. This suggests that many of the dead trees were not canopy co-dominants and mean diameter of snags and logs may increase as future deadwood is recruited from the pool of canopy co-dominants (Lee, 1998). Indeed, the mean diameters of snags and logs found in this study (16.79 and 12.81 cm, respectively) were smaller than the mean diameters of snags and logs (21.1 and 16.1 cm, respectively) 114  3.4. Discussion in the “maturing climax” stands of the Morgantini and Kansas (2003) study. In seven of the stands, the relative numbers of snags, logs, and live trees, and their corresponding species distributions, suggest a shifting stand composition (Figure 3.2). Species distributions in site PE indicate that the composition of the overstory is changing; all of the coarsewood was Pinus (or an unknown species), while only 50% of the live trees were Pinus. Similarly, PC had relatively little Picea coarsewood, but 63% of the live trees were Picea. These shifts are not surprising, as old-growth pine-dominated stands are uncommon in the foothills (Morgantini and Kansas, 2003). As a percentage of the total population, three spruce and three pine sites (SA–SC and PB–PD) had more snags and logs than live trees (Figure 3.2). Large accumulation of coarsewood is characteristic of old-growth stands (Clark et al., 1998; Morgantini and Kansas, 2003; Spies et al., 1988); however, the relative percentage of coarsewood is higher than previously found on this landscape (Morgantini and Kansas, 2003). While these results exclude live trees below the tagging limit, i.e. the regenerating understory, the overstories of these six stands may be declining.  3.4.2  Coarsewood Dynamics  Snags and logs persisted for many decades after death. While 90% of snags and logs died within the last hundred years, estimated time since death of the oldest snag and log was 180 and 175 years, respectively (Figure 3.6). There was more old Picea than Pinus coarsewood, but Pinus snags and logs still persisted up to 142 and 133 years, respectively. These time-since-death values may underestimate maximum coarsewood persistence as decomposing 115  3.4. Discussion snags and logs eventually become too decayed for dendrochronological work. This longevity was unexpected, as several comparable studies have reported shorter persistence times (DeLong et al., 2005, 2008; Johnson and Greene, 1991; Jonsson, 2000). In subalpine Pinus contorta–Picea engelmannii stands in the Rocky Mountains, approximately 400 km south of the study area, P. engelmannii snags stood up to 100 years after death (Johnson and Greene, 1991). In wet spruce–fir stands of east-central British Columbia, Picea glauca x Picea engelmannii hybrid snags have been found to have an average standing time of 16.5 years, and persist for a maximum of 66 years (DeLong et al., 2008). In these same forests, Picea logs persisted for up to 118 years. In old-growth Swedish spruce forests, Picea abies logs have been found to persist a maximum of 50 years since death (Jonsson, 2000). Although I expected coarsewood longevity to be similar to the above-mentioned studies, persistence of >100 years in Picea–Pinus forests is not unprecedented: a study of subalpine forests in central Colorado found Pinus contorta and Picea glauca logs to persist for approximately 125 and 140 years, respectively (Brown et al., 1998). The sites in this study, however, have an average elevation of approximately 1400 m.a.s.l. compared to 2990 m.a.s.l. in the Colorado study, and we would expect the higher temperatures at lower elevations to accelerate decomposition (Harmon et al., 1986). Snags and logs in intermediate decay classes were the most common; 68% of snags were in decay class 4 and 60% of logs were in decay class 3. Bias towards intermediate decay classes is common across a range of forest types (e.g. Daniels et al., 1997; Jonsson, 2000; Spies et al., 1988). In these stands, 116  3.4. Discussion persistence in log decay class 3 may be largely governed by the coarsewood dynamics and the transition of snags to logs. Most of the snags found on the landscape were in decay classes 3, 4, and 5; only 0.31% of the snags were in decay class 6, and I did not find any snags in decay class 7. This suggests that rather than decomposing completely in situ, most snags reach decay class 4 or 5, rot at the base and fall over, entering the pool of decay class 3 logs (see Chapter 1, Figure 1.3). Mean time since death differed significantly across all snag decay classes, and among log decay classes 1, 2, and 3–4 (as a group); however, there was wide variability in the time-since-death estimates for all decay classes (Figure 3.7). Snag decay class 4, for example, had the widest range: 166 years. Log decay class 3 had the second widest range of variability, 163 years, which may be explained by the transition of decay class 4 snags into decay class 3 logs. Earlier decay classes (3 for snags, 1 and 2 for logs) have a narrower range of variability, indicating that snags and logs transition through these stages of decay relatively rapidly. Similarly, results from Chapter 2 indicate that the error associated with YOD estimates increases with successive decay classes, suggesting that range of variability in more advanced stages of decay might be even larger than depicted in Figure 3.7. The range of YOD dates in each decay class was so broad for both species of snags and logs that decay class is not a reliable indicator of approximate time since death (Daniels et al., 1997; DeLong et al., 2008; Mast and Veblen, 1994). Dendrochronological studies provide precise estimates of YOD dates and thus reliable estimates of coarsewood persistence, but such studies are time-consuming and costly. Recent studies suggest that 117  3.4. Discussion reasonable time-since-death dates can be modeled using a comprehensive suite of structural characteristics (Newberry et al., 2004; Storaunet, 2004). If a temporal understanding of decay processes is required for coarsewood modeling and planning in managed forests, then decay class clearly cannot be used as a proxy for time since death; however, predictive models based on numerous structural characteristics may partially alleviate the need for intensive dendrochronological studies. It has been well demonstrated that time since death does not accurately reflect log decomposition and time since fall better predicts log decay class (Daniels et al., 1997; DeLong et al., 2005; Storaunet and Rolstad, 2002). Log time since death was not divided into time spent standing and time on the ground, which explains some of the variability in the time since death of decay classes. Snags decompose more slowly than logs, likely due to dessication and decreased activity of decomposers, so logs that once stood as snags have slower initial decay rates than logs that have been on the ground since death (Harmon et al., 1986; Johnson and Greene, 1991; Storaunet and Rolstad, 2002). In these stands, the pool of decay class 4 and 5 snags that fall as decay class 3 logs partially account for the range in decay class 3 time-since-death dates. This may also explain why the oldest observed log was in decay class 3 rather than decay class 4 (Figure 3.7). If most snags indeed fall and transition to logs rather than decay in situ, then the snag depletion curve approximates the snag fall curve (Figure 3.8). Ideally, indirect field-based methods could be used to better understand snag fall rates. I did not sample scars on trees that had been impacted by falling snags (see DeLong et al., 2005) in order to maintain the integrity of 118  3.4. Discussion ongoing research projects in the permanent sample plots. Similarly, other approaches to modelling could yield more precise fall rates (Storaunet and Rolstad, 2004), but the current analysis allows preliminary interpretation of snag fall. Picea and Pinus snags have half-lives of 47 and 36 years. Assuming that snags are recruited to the population at a constant rate and will continue to decay at the current rates, half of the current population of Picea snags will fall after 47 years and half of the Pinus snags will fall after 36 years. These half-lives are similar to values modeled for Abies balsamea and Picea mariana in boreal forests of eastern Quebec (30–35 and 35–40 years, respectively; Aakala et al., 2008), which have been found to have similar decay rates to Picea glauca but not Pinus contorta (Laiho and Prescott, 2004). If snags decay more slowly than logs, we would expect the half-lives and residence times of logs to be greater than those of snags. This is indeed the case for Pinus coarsewood (36 and 54 years), but not for Picea coarsewood (47 and 46 years). This discrepancy may be explained by the relative proportions of uprooted and snapped Picea logs that died in the last 50 years; 43% of these logs have uprooted bases, indicating that they have been on the ground for the entire time since death. All four depletion curves were best fit with a sigmoidal function, which has a changing slope (Figure 3.8). This suggests that as snags and logs decay, the rate at which they decompose changes; they decay more rapidly for the first 50–75 years and decompose at a slower rate as the slope of the curve decreases. This interpretation is supported by the relatively rapid transition of snags and logs through early decay classes, and longer and more varied 119  3.4. Discussion persistence in advanced stages of decay (Figure 3.7). Mean time since death did not vary significantly between species (Figure 3.7). Previous studies suggest that decay rates are strongly speciesdependent (Harmon et al., 1986; Laiho and Prescott, 2004), but in this study other factors were more important in determining decay class, which was used as a proxy for wood decomposition. Based on multivariate discriminant analysis (see Section 3.2.3), time was the preliminary driver of decay. Several other snag-, log-, and site-level variables were explored: elevation, annual temperature, and annual precipitation also influenced the variability in snag decomposition and log volume influenced log decomposition (Table 3.11). Based on the correlations between the variables and Can1, more decayed snags were older (greater time since death) and in colder, wetter sites at higher elevations than less decayed snags. Logs in advanced stages of decay were older and smaller than more intact logs. Both these results were expected. This analysis examined factors influencing snag-to-snag and log-to-log variability in decay. While climate (expressed as temperature, precipitation, and elevation) was not an important variable for logs, large-scale climatic variables are likely still significant in determining overall decay rates (Harmon et al., 1986). Studies of Picea engelmannii coarsewood in wet boreal stands on the west side of the Rocky Mountains had shorter persistence times than Picea in these stands, which are in the eastern rainshadow of the Rocky Mountains. Other factors may be important in influencing decay rates. Taylor and MacLean (2007), for example, identified cause of death as an important factor controlling decomposition. Additional data collec120  3.4. Discussion tion is required to better understand the factors controlling decay rates and dynamics of Picea and Pinus coarsewood in these forests.  3.4.3  Wildlife Habitat  As snags and logs decay, morphological structural changes (e.g. softening of sapwood, loss of branches, loosening of bark) correspond to changes in habitat function and thus affect the long-term availability of wildlife habitat. Although time since death did not vary significantly across habitat function or species (Table 3.13), several functions were significantly associated with particular decay class (Tables 3.14 and 3.15). Snags appear to increase in habitat value as they decay. Decay class 4 snags were the most likely to serve at least one habitat function; however, snags in decay class 5 were also important for functions with few observations, such as existing cavities and feeding substrates (Table 3.14). In addition, the snags with the highest wildlife value — those few that provided more than four functions — were from decay classes 4 and 5 (see Table 3.12). In spruce sites, snags in decay class 4 were most likely to serve as potential substrates for cavity excavation by woodpeckers, chickadees, and nuthatches (Keisker, 2000), whereas in pine sites this function was provided by more decayed snags (Table 3.14). These findings are consistent with similar studies in spruce–fir (DeLong et al., 2008) and cedar–hemlock (Stevenson et al., 2006) stands in interior British Columbia; both studies found an increase in suitable cavity substrates with relative snag age (inferred by branch order and decay class, respectively). In addition, as snags decayed, bark loosened and provided important concealment for creepers and bats (Keisker, 2000). 121  3.4. Discussion This function was primarily provided by decay class 4 snags in spruce sites and equally by decay classes 4 and 5 in pine sites (Table 3.14). Again, these results are consistent with Stevenson et al. (2006) and DeLong et al. (2008). In contrast, logs in an intermediate stage of decay appear to have the greatest habitat value (decay class 3; Table 3.15). Similarly, in pine sites, the logs that provided four or more habitat functions were in decay class 3, but in spruce sites such logs were in decay classes 2–4 (see Table 3.12). Feeding substrates were equally as abundant in decay classes 3 and 4 in both site types, providing a food source for insectivorous birds, mammals, and amphibians (Table 3.15; Keisker, 2000). Logs in decay class 3 were generally not elevated above the ground, but were still solid enough to provide overhead cover; these logs served as exposed, raised travel lanes for squirrels and other small mammals and also provided small concealed spaces for a variety of small mammals, amphibians, and reptiles (Keisker, 2000). As logs collapsed and began to incorporate into the soil (i.e. decay class 4), their value as wildlife habitat generally decreased. Once more, these observations are consistent with the findings of DeLong et al. (2005) and Stevenson et al. (2006). Not surprisingly, uprooted logs provided most of the large, concealed spaces (Table 3.16). Although I observed 151 instances of snag habitat function and 706 instances of log habitat function, most observations were limited to one of five functional types. Furthermore, six functions had less than 10 observations in each site type and three snag functions were not observed at all (Figure 3.9). In particular, there were few snags that provided existing cavities for secondary cavity nesters and large open-nest supports and hunting perches 122  3.4. Discussion for herons, owls, diurnal raptors, passerines, and kingfishers (Figure 3.9; Keisker 2000). There were no snags that provided large or very large cavities. Consequently, in these stands, there is no suitable roosting habitat for species such as barred owls, boreal owls, red squirrels, flying squirrels, and marten, all of which have been identified as indicator species for the sustainable management of FRI land base (Dempster, 1998; Keisker, 2000). Less than 1% of observed snags and less than 4% of observed logs provided four or more functions (Table 3.12). Both high-quality wildlife trees were large, in the 30–40 cm size class, and were in decay classes 4 and 5. Surprisingly, they were both Pinus trees and from a single site. There was a single large spruce tree (33 cm dbh) with three functions, which was also a high-quality wildlife tree. There were six logs with four or more functions, three in spruce sites and three in pine sites. Unlike the snags, the logs were a range of sizes, with mid-length diameters of 10–32 cm and lengths of 5–20 m. Four were in decay class 3 and there was one in each of decay classes 2 and 4. These observations suggest that size is more important for high-quality snag habitat than it is for logs; however, there were very few large logs in decay class 3, so the interaction between decay class and size class may have biased the results. Regardless, management strategies that aim to preserve large snags as wildlife trees may consequently create large high-quality habitat logs as decay class 4 snags transition to decay class 3 logs.  123  3.4. Discussion  3.4.4  Conclusion  Coarsewood provides critical biological legacies in dynamic landscapes; when they persist through a natural or anthropogenic disturbance, snags and logs maintain structural and functional diversity. Thus, understanding coarsewood persistence and the links between decomposition and function is pertinent for ecologically based landscape planning. The results presented in this chapter inspire future research directions and have important implications for conservation and sustainable forest management. These topics are addressed in Chapter 4.  124  3.5. References  3.5  References  Aakala, T., T. Kuuluvainen, L. De Grandpre, and S. Gauthier. 2007. Trees dying standing in the northeastern boreal old-growth forests of Quebec: spatial patterns, rates, and temporal variation. Canadian Journal of Forest Research, 37:50–61. Aakala, T., T. Kuuluvainen, S. Gauthier, and L. De Grandpre. 2008. Standing dead trees and their decay-class dynamics in the northeastern boreal old-growth forests of Quebec. Forest Ecology and Management, 255:410– 420. Alban, D. H. and J. Pastor. 1993. Decomposition of aspen, spruce, and pine boles on two sites in Minnesota. Canadian Journal of Forest Research, 23:1744–1749. Beckingham, J., I. Corns, and J. Archibald. 1996. Field guide to ecosites of west-central Alberta. Special Report 9, Canadian Forest Service Northwest Region, Edmonton, AB. Berg, A., B. Ehnstrom, L. Gustafsson, T. Hallingback, M. Jonsell, and J. Weslien. 1994. Threatened plant, animal, and fungus species in Swedish forests — distribution and habitat associations. Conservation Biology, 8:718–731. Blanchette, R. A. 1995. Degradation of the lignocellulose complex in wood. Canadian Journal of Botany, 73:S999–S1010. Boddy, L. 2001. Fungal community ecology and wood decomposition pro125  3.5. References cesses in angiosperms: from standing tree to complete decay of coarse woody debris. Ecological Bulletins, 49:43–56. Brown, P. M., W. D. Shepperd, S. A. Mata, and D. L. McClain. 1998. Longevity of windthrown logs in a subalpine forest of central Colorado. Canadian Journal of Forest Research, 28:932–936. Bunnell, F. L. and G. B. Dunsworth, editors. 2009. Forestry and Biodiversity: Learning How to Sustain Biodiversity in Managed Forests. UBC Press, Vancouver, B.C. Clark, D. F., D. D. Kneeshaw, P. J. Burton, and J. A. Antos. 1998. Coarse woody debris in sub-boreal spruce forests of west-central British Columbia. Canadian Journal of Forest Research, 28:284–290. Crampton, L. H. and R. M. R. Barclay. 1998. Selection of roosting and foraging habitat by bats in different-aged aspen mixedwood stands. Conservation Biology, 12:1347–1358. Daly, C., W. Gibson, G. Taylor, G. Johnson, and P. Pasteris. 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research, 22:99–113. Daniels, L. D., J. Dobry, K. Klinka, and M. C. Feller. 1997. Determining year of death of logs and snags of Thuja plicata in southwestern coastal British Columbia. Canadian Journal of Forest Research, 27:1132–1141. DeLong, S. C., L. D. Daniels, B. Heemskerk, and K. O. Storaunet. 2005. Temporal development of decaying log habitats in wet spruce-fir stands 126  3.5. References in east-central British Columbia. Canadian Journal of Forest Research, 35:2841–2850. DeLong, S. C., G. D. Sutherland, L. D. Daniels, B. Heemskerk, and K. Storaunet. 2008. Temporal dynamics of snags and development of snag habitats in wet spruce-fir stands in east-central British Columbia. Forest Ecology and Management, 255:3613–3620. Dempster, W. 1998. Indicators of sustainable forest management for the Foothills Model Forest. Technical report, Foothills Model Forest, Hinton, AB. Foster, J. R. and G. E. Lang. 1982. Decomposition of red spruce and balsam fir boles in the White Mountains of New Hampshire. Canadian Journal of Forest Research, 12:617–626. Fraver, S., A. Ringvall, and B. G. Jonsson. 2007. Refining volume estimates of down woody debris. Canadian Journal of Forest Research, 37:627–633. Grissino-Mayer, H. D. 2001. Evaluating crossdating accuracy: a manual and tutorial for the computer program COFECHA. Tree-Ring Research, 57:205–221. Grove, S. J. 2001. Extent and composition of dead wood in Australian lowland tropical rainforest with different management histories. Forest Ecology and Management, 154:35–53. Hardy, M. A. 1993. Regression with Dummy Variables. Sage Publications, Newbury Park, California. 127  3.5. References Harmon, M. E., J. F. Franklin, F. J. Swanson, P. Sollins, S. V. Gregory, J. D. Lattin, N. H. Anderson, S. P. Cline, N. G. Aumen, J. R. Sedell, G. W. Lienkaemper, K. Cromack, and K. W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research, 15:133–302. Hinton Wood Products. 2006. Permanent Growth Sample Program Manual, version 16. West Fraser Mills, Hinton, AB. Holmes, R. 1983. Computer-assisted quality control in tree-ring dating and measuring. Tree-Ring Bulletin, 43:69–78. Hunter, M. L. J. 1990. Wildlife, Forests, and Forestry: Principles of Managing Forests for Biological Diversity. Prentice Hall, Eaglewood Cliffs, N.J. Hyatt, T. L. and R. J. Naiman. 2001. The residence time of large woody debris in the Queets River, Washington, USA. Ecological Applications, 11:191–202. Johnson, E. A. and D. F. Greene. 1991. A method for studying dead bole dynamics in Pinus contorta var latifolia–Picea engelmannii forests. Journal of Vegetation Science, 2:523–530. Jonsson, B. G. 2000. Availability of coarse woody debris in a boreal oldgrowth Picea abies forest. Journal of Vegetation Science, 11:51–56. Keisker, D. 2000. Types of wildlife trees and coarse woody debris required  128  3.5. References by wildlife of north-central British Columbia. Working Paper 50, B.C. Ministry of Forests, Victoria, B.C. Laiho, R. and C. E. Prescott. 2004. Decay and nutrient dynamics of coarse woody debris in northern coniferous forests: a synthesis. Canadian Journal of Forest Research, 34:763–777. Lee, P. 1998. Dynamics of snags in aspen-dominated midboreal forests. Forest Ecology and Management, 105:263–272. Lindenmayer, D. B. and J. F. Franklin. 2002. Conserving Forest Biodiversity: A Comprehensive Multiscaled Approach. Island Press, Washington, D.C. Makinen, H., J. Hynynen, J. Siitonen, and R. Sievaneni. 2006. Predicting the decomposition of Scots pine, Norway spruce, and birch stems in Finland. Ecological Applications, 16:1865–1879. Maser, C., R. Anderson, K. Cromack, J. T. Williams, and R. E. Martin. 1979. Dead and down woody material. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 78–95. U.S. Department of Agriculture Forest Service, Washington, DC. Mast, J. N. and T. T. Veblen. 1994. A dendrochronological method of studying tree mortality patterns. Physical Geography, 15:529–542. Mattson, K. G., W. T. Swank, and J. B. Waidb. 1987. Decomposition of woody debris in a regenerating, clear-cut forest in the southern Appalachians. Canadian Journal of Forest Research, 17:712–721. 129  3.5. References Morgantini, L. E. and J. L. Kansas. 2003. Differentiating mature and oldgrowth forests in the Upper Foothills and Subalpine subregions of westcentral Alberta. Forestry Chronicle, 79:602–612. Naesset, E. 1999. Decomposition rate constants of Picea abies logs in southeastern Norway. Canadian Journal of Forest Research, 29:372–381. Neter, J., M. Kutner, C. Nachtsheim, and W. Wasserman. 1996. Applied linear statistical models. McGraw-Hill, Boston, Mass., 4th edition. Newberry, J. E., K. J. Lewis, and M. B. Walters. 2004. Estimating time since death of Picea glauca x P. engelmannii and Abies lasiocarpa in wet cool sub-boreal spruce forest in east-central British Columbia. Canadian Journal of Forest Research, 34:931–938. Norton, M. R., S. J. Hannon, and F. K. A. Schmiegelow. 2000. Fragments are not islands: patch vs. landscape perspectives on songbird presence and abundance in a harvested boreal forest. Ecography, 23:209–223. Perry, D. A. 1994. Forest Ecosystems. Johns Hopkins University Press, Baltimore, M.D. Rinn, F. 2003. TSAP-Win User Reference. Rinntech, Heidelberg, Germany. SAS. 2002. SAS. SAS Institute Inc., Cary, NC. Spies, T. A., J. F. Franklin, and T. B. Thomas. 1988. Coarse woody debris in Douglas-fir forests of western Oregon and Washington. Ecology, 69:1689– 1702.  130  3.5. References Stevenson, S. K., M. J. Jull, and B. J. Rogers. 2006. Abundance and attributes of wildlife trees and coarse woody debris at three silvicultural systems study areas in the interior cedar-hemlock zone, British Columbia. Forest Ecology and Management, 233:176–191. Stokes, M. A. and T. L. Smiley. 1996. An Introduction to Tree-Ring Dating. University of Arizona Press, Tuscon, AZ, 2nd edition. Storaunet, K. and J. Rolstad. 2002. Time since death and fall of Norway spruce logs in old-growth and selectively cut boreal forest. Canadian Journal of Forest Research, 32:1801–1812. Storaunet, K. O. 2004. Models to predict time since death of Picea abies snags. Scandinavian Journal of Forest Research, 19:250–260. Storaunet, K. O. and J. Rolstad. 2004. How long do Norway spruce snags stand? Evaluating four estimation methods. Canadian Journal of Forest Research, 34:376–383. Taylor, S. L. and D. A. MacLean. 2007. Dead wood dynamics in declining balsam fir and spruce stands in New Brunswick, Canada. Canadian Journal of Forest Research, 37:750–762. Thomas, J., R. Anderson, C. Maser, and E. Bull. 1979. Snags. In J. Thomas, editor, Wildlife Habitats in Managed Forests: the Blue Mountains of Oregon and Washington, Agriculture Handbook No. 553, pages 60–77. U.S. Department of Agriculture Forest Service, Washington, DC.  131  3.5. References VoorTech Consulting. 2004. MeasureJ2X. VoorTech Consulting, Holderness, NH. Wang, T., A. Hamann, D. Spittlehouse, and S. Aitken. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology, 26:383–397. Zielonka, T. 2006. Quantity and decay stages of coarse woody debris in old-growth subalpine spruce forests of the western Carpathians, Poland. Canadian Journal of Forest Research, 36:2614–2622.  132  Chapter 4  Conclusion 4.1  Summary of Results  4.1.1  Chapter 2: Verification of Year-of-Death Estimates  Chapter 2 addressed error in dendrochronological work on decayed wood by comparing year of death estimates to permanent sample plot data. The major findings are as follows: • High correlations of year-of-death dates for pairs of cores and pairs of radii suggest that sample quality is not a significant problem in obtaining accurate year-of-death estimates (see Chapter 2, Figures 2.4 and 2.5). Instead, it is possible that as the trees are declining, they are not allocating carbon to radial growth thus the calendar year of the outermost ring does not accurately reflect the year of tree death. • Most year-of-death estimates were within or preceded the observed interval of death from the PSP data (see Chapter 2, Figure 2.6). • Of the subset of trees that died after plot establishment, the magnitude of error in year-of-death estimates increased with time since death (see Chapter 2, Figure 2.7). 133  4.1. Summary of Results • Overall, YOD dates on average preceded the IOD by −1.74 ± 5.30 years for Picea snags, −5.46 ± 7.85 years for Pinus snags, −1.39 ± 5.08 years for Picea logs, and −4.05 ± 8.35 years for Pinus logs.  4.1.2  Chapter 3: Coarsewood Decay Dynamics and Wildlife Habitat  Chapter 3 analysed the accumulation and persistence of snags and logs and their potential functions as wildlife habitat. The major findings are as follows: • Snags and logs in intermediate decay classes were the most common (see Chapter 3, Figures 3.4 and 3.5). Persistence in log decay class 3 may be largely governed by the coarsewood dynamics and the transition of snags to logs. Most of the snags found on the landscape were in decay classes 3, 4, and 5. This suggests that rather than decomposing completely in situ, most snags reach decay class 4 or 5, rot at the base and fall over, entering the pool of decay class 3 logs (see Chapter 1, Figure 1.3). • Snags and logs persisted for many decades after death (see Chapter 3, Figure 3.6). While 90% of snags and logs died within the last hundred years, estimated time since death of the oldest snag and log was 180 and 175 years, respectively. There was more old Picea than Pinus coarsewood, but Pinus snags and logs still persisted up to 142 and 133 years, respectively. • Time since death varied significantly across decay class for snags and 134  4.2. Research Implications logs, but there was wide variation in time since death within decay classes (see Chapter 3, Figure 3.7). The range of YOD dates in each decay class was so broad for both species of snags and logs that decay class is not a reliable indicator of approximate time since death. • Most observations of habitat functions were limited to one of five functional types (see Chapter 3, Figure 3.9 and Tables 3.5 and 3.7). Few snags provided existing cavities for secondary cavity nesters and large open-nest supports and hunting perches (Keisker, 2000). There were no snags that provided large or very large cavities. • Less than 1% of observed snags and less than 4% of observed logs provided four or more habitat functions (see Chapter 3, Table 3.12). The two high-quality wildlife trees were large, in the 30–40 cm size class, and were in decay classes 4 and 5. The logs were from a range of sizes: 10–32 cm diameter at mid-length.  4.2  Research Implications  Using permanent sample plot data to verify year of death estimates revealed both good and bad news for dendrochronologists. It is clear that there is error associated with YOD dates, and the magnitude of error varies between species and position (i.e. snag or log) and across time since death; however, the magnitude of error is relatively small. Consequently, I urge other dendrochronologists to explicitly address the error in their studies. Although dendrochronological approaches to studying coarsewood provide  135  4.3. Conservation Implications precise YOD estimates at an annual resolution, the resulting YOD dates may not be accurate. One conservative approach would be to express YOD estimates as a range of dates, rather than as a single year. If there is no information available about the potential magnitude of error, expressing estimates on a decadal scale may be appropriate. Two research extensions would enhance our understanding of Picea and Pinus decay dynamics in boreal forests. (1) Future work could integrate data on mortality, snag and log decay, and snag fall into a model of coarsewood decay class transitions (e.g. Kruys et al., 2002; Morrison and Raphael, 1993; Vanderwel et al., 2006). Such a model would allow the projection of decay-class distributions of snags and logs over time. (2) These results could be linked to other work in the region on riparian and in-stream decay processes (Powell et al., 2009) to create a more complete ecosystem-level understanding of deadwood dynamics.  4.3  Conservation Implications  Coarsewood-dependent wildlife species are impacted — positively and negatively — by forest-management decisions; thus, from a wildlife-conservation perspective, coarsewood must be included in the integration of wildlife management and timber resource planning. In particular, attention must be given to the abundance and diversity of coarsewood that is suitable for habitat, not just the total number of snags and logs. Effective wildlifeconservation planning requires an understanding of coarsewood decay dynamics, how wildlife habitat function changes through time, and how these  136  4.3. Conservation Implications two dynamics interact to provide long-term habitat availability. In the ten old-growth sites I studied, a few habitat functions were common (see Chapter 3, Figure 3.9). If this pattern is representative of the landscape, then the less common habitat functions require special attention in conservation planning. For example, few snags provided existing cavities for secondary cavity nesters and large open-nest supports and hunting perches for raptors. No snags provided large or very large cavities. Consequently, in these stands, there is no suitable roosting habitat for species such as barred owls, boreal owls, red squirrels, flying squirrels, and marten, all of which have been identified as indicator species for the sustainable management of FRI land base (see Chapter 3, Table 3.5; Dempster, 1998; Keisker, 2000). Similarly, large concealed spaces and long concealed spaces were relatively uncommon among logs, which provide habitat for large mammals and ground-dwelling birds, and concealed travel routes from amphibians and small mammals, respectively (see Chapter 3, Table 3.7 and Figure 3.9). In particular, black bears, snowshoe hares, and ruffed grouse require large concealed spaces and long-toed salamanders require long concealed spaces. All of these species are also used as indicators for sustainable management of the foothills (Dempster, 1998). Comprehensive wildlife conservation planning in the foothills should include a survey of snag and log habitat functional types beyond my study sites. Coarsewood habitat functions can have a patchy spatial distribution (Keisker, 2000), so functional types that were uncommon in my sites may be abundant elsewhere and vice versa. It is important to note that I only considered conservation implications 137  4.4. Management Implications from the perspective of coarsewood-dependent wildlife. The accumulation and persistence of coarsewood has implications for the myriad invertebrates, plants, fungi, and microorganisms that depend directly on snags and logs for habitat and survival, as well as those species that benefit indirectly from the many ecological functions of coarsewood. While these species and functions are beyond the scope of my thesis, they require consideration in comprehensive conservation planning.  4.4 4.4.1  Management Implications Long-Term Planning  Snags and logs in the foothills can persist for a long time — one to two hundred years (see Chapter 3, Figure 3.6). Consequently, management plans need to take a long-term view if we wish to maintain levels of coarsewood that are within the natural range of variability. The decisions that are made now will determine the types and amounts of coarsewood that will be on the landscape a century from now. One practical application of this research is to create a decision-support tool. Combined with other projects from the foothills and continuing research through the Natural Disturbance Program, these results could be used to create models that predict coarsewood accumulation resulting from various management scenarios (e.g DeLong et al., 2005; Tinker and Knight, 2001). In turn, such models could be used to inform long-term management and create ecologically based targets of coarsewood retention.  138  4.4. Management Implications  4.4.2  Old-Growth Forests and Natural Disturbances  The stands considered in this study were very old for the landscape (Bonar et al., 2003; Morgantini and Kansas, 2003). They all established at least 160 years prior to program initiation in 1956, based on the PSP records. The tree ring record indicates minimum cohort establishment dates of 1680–1765 for four of the spruce sites and all the pine sites. (One spruce site may actually be younger than indicated by the fire maps; the tree-ring record for site SB only goes back to 1891.) These sites are well beyond the traditional harvesting age for the landscape, which is 90–110 years (Morgantini and Kansas, 2003). Forest management plans that commit to conserving a percentage of stands older than the rotation age often group together all “old” forest; however, plans that follow this approach miss the variability in structure that develops as stands age. In the foothills, stands that established 300 years ago are ecologically different from stands that are 150 years old (Morgantini and Kansas, 2003). The old-growth stands in this study have high levels of accumulated coarsewood; in six of the sites, the total number of snags and logs is greater than the number of live trees (see Chapter 3, Figure 3.2). In addition, within stands of a similar age, there is variability in terms of snag and log accumulation, size, age, species, decay distribution, and wildlife function (see Chapter 3, Figures 3.2, 3.3, 3.4, 3.5, and 3.9). In order to conserve stands with this structure, management plans must differentiate such old-growth sites from much younger “old” ones and consider the full range of structural and functional variability with old-growth stands.  139  4.4. Management Implications Natural disturbances in the foothills leave behind residual pieces of dead wood.  Even severe stand-replacing fires create a significant amount of  coarsewood, like hardened snags. Progressive management practices that aim to mimic natural disturbances necessarily leave large amounts of dead wood on the landscape. Conserving target levels of coarsewood may only preserve coarsewood function if management plans are attentive to the variability in accumulation and persistence both across stand ages and types and within stands of a similar age and composition. While my research focussed on coarsewood as wildlife habitat, planning must accommodate the full range of known functions (see sections 1.1 and 4.3 and the references therein). In addition, planning should consider a buffer for uncertainty, as there may be many functions that we have yet to identify and understand.  140  4.5. References  4.5  References  Bonar, R. L., H. Lougheed, and D. W. Andison. 2003. Natural disturbance and old-forest management in the Alberta Foothills. Forestry Chronicle, 79:455–461. DeLong, S. C., L. D. Daniels, B. Heemskerk, and K. O. Storaunet. 2005. Temporal development of decaying log habitats in wet spruce-fir stands in east-central British Columbia. Canadian Journal of Forest Research, 35:2841–2850. Dempster, W. 1998. Indicators of sustainable forest management for the Foothills Model Forest. Technical report, Foothills Model Forest, Hinton, AB. Keisker, D. 2000. Types of wildlife trees and coarse woody debris required by wildlife of north-central British Columbia. Working Paper 50, B.C. Ministry of Forests, Victoria, B.C. Kruys, N., B. G. Jonsson, and G. Stahl. 2002.  A stage-based matrix  model for decay-class dynamics of woody debris. Ecological Applications, 12:773–781. Morgantini, L. E. and J. L. Kansas. 2003. Differentiating mature and oldgrowth forests in the Upper Foothills and Subalpine subregions of westcentral Alberta. Forestry Chronicle, 79:602–612. Morrison, M. L. and M. G. Raphael. 1993. Modeling the dynamics of snags. Ecological Applications, 3:322–330. 141  4.5. References Powell, S., L. Daniels, and T. Jones. 2009. Temporal dynamics of large woody debris in small streams of the Alberta foothills, Canada. Canadian Journal of Forest Research, 39:1159–1170. Tinker, D. B. and D. H. Knight. 2001. Temporal and spatial dynamics of coarse woody debris in harvested and unharvested lodgepole pine forests. Ecological Modelling, 141:125–149. Vanderwel, M. C., J. R. Malcolm, and S. M. Smith. 2006. An integrated model for snag and downed woody debris decay class transitions. Forest Ecology and Management, 234:48–59.  142  

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.24.1-0068827/manifest

Comment

Related Items