Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Stand structural characteristics and development patterns in old-growth interior cedar hemlock forests… MacKillop, Debra J. 2003

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

Item Metadata

Download

Media
831-ubc_2003-0603.pdf [ 7.81MB ]
Metadata
JSON: 831-1.0075102.json
JSON-LD: 831-1.0075102-ld.json
RDF/XML (Pretty): 831-1.0075102-rdf.xml
RDF/JSON: 831-1.0075102-rdf.json
Turtle: 831-1.0075102-turtle.txt
N-Triples: 831-1.0075102-rdf-ntriples.txt
Original Record: 831-1.0075102-source.json
Full Text
831-1.0075102-fulltext.txt
Citation
831-1.0075102.ris

Full Text

STAND S T R U C T U R A L C H A R A C T E R I S T I C S AND D E V E L O P M E N T P A T T E R N S IN O L D - G R O W T H INTERIOR C E D A R H E M L O C K F O R E S T S IN S O U T H E A S T E R N BRITISH COLUMBIA by D E B R A J . MacKILLOP B.E.S. , York University, 1997 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE R E Q U I R E M E N T S FOR THE D E G R E E OF M A S T E R OF S C I E N C E in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Forest Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 22, 2003 © Debra J . MacKillop In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Forest Sciences The University of British Columbia Vancouver, Canada Date: September 25. 2003 ABSTRACT Old-growth forests have been defined and described in many ways including stand age, stand structure, and stand population dynamics. This thesis uses these key approaches to describe the characteristics and stand development patterns of old-growth forests in the moist warm Interior Cedar Hemlock biogeoclimatic subzone of southeastern British Columbia. In total, 16 sites ranging from an estimated 130 to 630 years old were sampled in the Arrow and Kootenay Lake Forest Districts. Stand age plays an important role in defining and describing old growth. However, trees with internal decay were common in the study area, and in many cases, estimates were the only source of data. Age estimates using linear extrapolation and regression analysis were tested then applied to tree cores with internal decay. Using regression analysis, age-on-diameter relationships were assessed on a species-specific and a site-specific basis. Site-specific regression models provided the most accurate means of estimating missing rings. However, the data required for their development can be difficult and tedious to obtain, and although more accurate than other methods, can still be wrong by up to 100 years. Errors associated with linear extrapolation were tested using the mean growth rates from intact portions of tree cores to extrapolate ages to missing portions. However, if 40% or less of a tree's radius was used, estimates were unreliable and errors were as large as 172 years for Tsuga heterophylla ((Raf.) Sarg. (western hemlock) and 513 years for Pseudotsuga menziesii ((Franco Mirb) Douglas-fir). Where more than 40% of the tree radius was used for estimating missing ring counts, accuracy was improved, but decreased as tree size and age increased. Based on the standard deviation of the mean difference between estimates and counts, estimates for Tsuga cores over 50 cm dbh and with more than 40% of the radius included were generally accurate to within 55 years. Large Thuja plicata ((Donn ex D. Don) western redcedar) and Pseudotsuga I Larix occidentalis ((Nutt.) western larch) were accurate to within 37 and 83 years, respectively. With actual tree age estimates evaluated, old-growth forests were defined using stand ages from forest cover inventory planning data and measured ages from tree cores extracted in the field, from age class frequency distributions and tree size class frequency distributions, and by using principal components analysis (PCA) to analyze multiple stand structural attributes. The various methods of defining old growth led to different classifications of study stands. The P C A analysis was based on measures of live and dead trees as well as understory vegetation and coarse woody debris, and provided the most reliable definition of old growth. The P C A analysis resulted in an 'index of old-growthness' in which individual attributes can be measured in the field then compared to calculated thresholds to determine a 'score' of 'old-growthness' for older forests. To assess stand development patterns in old-growth stands, dendroecological techniques were used at six of the old-growth sites. Relative increases in growth were interpreted to reflect releases, while relative declining growth reflected suppression. Small-scale disturbances that create canopy gaps were the dominant disturbance in the old-growth forests studied here and both Tsuga and Thuja, the dominant trees in the study area, were found to rely on gaps for canopy ascension. Over 80% of the Thuja currently in the lower canopy had experienced one or more releases without reaching upper canopy positions. In contrast, almost half of the Tsuga trees currently in lower canopy positions have never released, but 77% of Tsuga trees that are currently in upper canopy positions required one or more release while in the understory to reach their dominant or codominant position. In contrast, 86% of Tsuga trees in lower canopy positions have undergone periods of suppression. These patterns of suppression and release were interpreted to mean that more understory releases are required for both Thuja and Tsuga trees to reach upper canopy positions. TABLE OF CONTENTS A B S T R A C T M T A B L E OF C O N T E N T S iiv LIST OF T A B L E S vi LIST OF F IGURES vii A C K N O W L E D G E M E N T S iix CHAPTER ONE: Introduction INTRODUCTION 1 S T U D Y A R E A 2 M E T H O D S 7 Field sampling 7 Data analysis 9 R E S U L T S and DISCUSSION 10 Site descriptions 10 Live trees 11 Standing dead trees 15 Coarse woody debris 16 C O N C L U S I O N S 21 CHAPTER TWO: Tree Age INTRODUCTION 22 M E T H O D S 24 Data processing 24 Growth to breast height 24 Estimating missing ring counts 24 R E S U L T S 27 Growth to breast height 27 Estimating the number of missing rings using linear extrapolation 28 Estimating the number of missing rings using regression 33 DISCUSSION 36 Growth to breast height 36 Comparing methods of estimating missing rings 38 Limitations 41 C O N C L U S I O N S 42 CHAPTER THREE: Definitions and descriptions of old-growth forests in moist warm Interior Cedar Hemlock stands INTRODUCTION 44 M E T H O D S 46 Tree ages 46 Stand Age 48 Population Dynamics 49 Stand Structure 50 R E S U L T S 51 Stand Age 51 Population Dynamics 53 Stand Structure 57 DISCUSSION 67 C O N C L U S I O N S 75 CHAPTER FOUR: Dynamics of Tsuga-dominated old-growth forests in southeastern British Columbia INTRODUCTION 76 M E T H O D S 78 R E S U L T S 81 Stand level growth changes 81 Individual species growth responses 83 Physiologic growth rates 86 DISCUSSION ' 89 Methodological limitations 91 Management Applications 93 C O N C L U S I O N S 95 BIBLIOGRAPHY 97 A P P E N D I X 1 : Double bark thickness ratios from PrognosisBC. 104 A P P E N D I X 2: Age class frequency distributions, stratified by age estimation method and species. 105 A P P E N D I X 3: Additional thresholds for use in an index of old-growthness. 109 LIST OF TABLES CHAPTER 1 Table 1. Elevation, aspect, slope, and biogeoclimatic information for each field site 10 Table 2. Stand ages and mean and maximum tree diameters for each field site 12 Table 3. Percentage of trees with pathogen indicators and with dead, broken, or forked tree tops at each field site 13 Table 4. Species composition by basal area and stem density for each field site 14 Table 5. Snag density by size class (dbh) for each field site 15 Table 6. C W D lengths, diameters, and density (by diameter class) 20 CHAPTER 2 Table 1. Summary statistics for the estimated number of years to grow to breast height using the sapling method, the ground method, and the average of both methods 29 Table 2. Summary statistics for differences between actual ring counts and estimates using the total core and the innermost 25 rings, stratified by species, tree size class, and percent intact radius 31 Table 3. Mean difference between estimated and counted tree rings using the total intact core and the innermost 25 rings. Significant differences indicate that estimated ages are not equivalent to counted ages 32 Table 4. Mean, minimum and maximum differences between inside bark radii and measured core lengths for Tsuga, Thuja, and Pseudotsuga /Larix trees <30 cm, 30-50 cm, and >50 cm dbh 33 Table 5. Site-specific regression equations between age and dbh based on all complete Tsuga cores not used to test linear extrapolation methods 35 Table 6. Summary statistics for differences between actual ring counts and estimates from regression models based on Tsuga trees from all sites and from individual sites, stratified by tree size class.... 36 Table 7. Percent error for different estimation methods 38 CHAPTER 3 Table 1. Mean, maximum and veteran ages and their associated species at each site 53 Table 2. Estimated sapling density, by age class 55 Table 3. Principal component matrix showing correlations between structural attributes and P C A axes.. 60 Table 4. Thresholds, measured values and old-growthness scores for each site 65 Table 5. Summary of old-growth classifications by site 68 Table 6. Mean and standard errors for selected attributes by stand age category 73 CHAPTER 4 Table 1. Precentage of understory growth changes in lower canopy Thuja trees and overstory changes in upper canopy Thuja trees. Cores with decay are included in summaries of upper canopy trees, but not for lower canopy trees 84 Table 2. Percentage of Tsuga trees in upper and lower canopy positions that experienced understory and overstory release and suppression events. Upper and lower canopy refer to current crown position; understory and overstory events are separated by a 40 cm dbh threshold. Cores with decay are not included 85 Table 3. Physiological growth rate categories for Tsuga 86 Table 4. Growth rates (mm / year) for Tsuga in the upper and lower canopy 87 LIST OF FIGURES CHAPTER 1 Figure 1. Map of the study area including the Arrow and Kootenay Lake Forest Districts. Field sites are identified by number 3 Figure 2. Nested plot design. The largest circle represents the large plot (25.23m radius). The smaller circle represents the 11.28m main plot. The three smallest circles represent seedling/sapling plots and the black lines represent one randomly located 25m long C W D transect and a second 25m long C W D transect at 90 degrees to the first 8 Figure 3. Snag density by species for each field site 17 Figure 4. Snag basal area by species for each field site 17 Figure 5. Snag density by decay class for each field site 18 Figure 6. Snag basal area by decay class for each field site 18 Figure 7. C W D stem density by decay class for each field site 19 Figure 8. C W D volume by decay class for each field site 19 CHAPTER 2 Figure 1. Average number of years for growth to breast height using the sapling method and the ground method for Tsuga and Thuja 27 Figure 2. Relationship between the percent intact radius and the accuracy of linear extrapolation using average ring widths from the total core and from the innermost 25 years before a break or decay for Tsuga, Thuja, and Pseudotsuga /Larix 30 Figure 3. Differences between inside bark radii and measured core lengths for Tsuga, Thuja, and Pseudotsuga /Larix trees <30 cm, 30-50 cm, and >50 cm dbh 33 Figure 4. Relationship between diameter at breast height and age (ring counts) for Russell and for all other sites combined 34 Figure 5. Difference between regression estimates and tree ring counts using regression models based on all field sites (a) and regression models based on individual sites (b) for Tsuga trees with dbh <30 cm, 30-50 cm, and >50 cm 35 Figure 6. Average ages for Tsuga cores with greater than 40% of the radius intact - estimated and counted 39 Figure 7. Average ages for Tsuga cores with less than 40% of the radius intact - estimated and counted. 39 CHAPTER 3 Figure 1. Mean, maximum, and veteran age estimates (corrected to breast height) as compared to Forest Cover age classes (shown as rectangular boxes) 52 Figure 2. Age class frequency distributions stratified by species for each site. IB = shade intolerant broadleaved species; TC = shade tolerant conifers; IC - shade intolerant conifers; Cw = western redcedar (Thuja); Hw = western hemlock (Tsuga) 56 Figure 3. Diameter class frequency distributions, stratified by species for all sample sites 58 Figure 4. Multivariate relationship between sample sites, based on structural attributes 61 Figure 5. Relationship between P C A and stand age 62 Figure 6. An index of old-growthness for moist warm Interior Cedar Hemlock forests in southwest B C . . . 64 CHAPTER 4 Figure 1. Proportion of trees experiencing suppression and release by decade. Sample sizes are recorded along the bottom of each graph and long arrows indicate the start of a reliable tree ring record. Short arrows reflect growth changes affecting at least 20% of a stand, including a minimum five trees. Solid bars represent major growth changes, while thatched bars reflect minor changes. . 82 Figure 2. Average yearly growth rate for three upper canopy Tsuga (Bremner #101, Pedro #46, and Ski #41 ) and one lower canopy Tsuga (Giveout #26), summarized by decade 88 ACKNOWLEDGEMENTS This thesis represents many years of hard work, and a little bit of procrastination. Funding was provided through grants from the Mountain Equipment Co-op Environment Fund, the Imajo Cedar Endowment Fund, the Natural Sciences and Engineering Research Council, and the British Columbia Science Council 's G R E A T scholarship. There are many people who have helped me to complete this project. Judy Rodrigues provided entertaining, amusing and skilled field assistance - may your travels bring you back to the mountains one day! Many other friends also helped to both motivate and distract me from completion - you all know who you are and I thank you for keeping things fun. Thank you also to my supervising committee - Dr. Cindy Prescott, Dr. Peter Marshall, Dr. Lori Daniels, Dr. Rachel Holt, and the infamous Mr. Andy MacKinnon. Special thanks goes to Rachel for her inspiration, and for reminding me of the importance of always staying true to what I believe, and to Cindy for taking me on and helping me through this, even though I know nothing about 'dirt'. I would like to offer additional thanks to Chris Steeger and Marlene Machmer at Pandion Ecological Research, Ltd in Nelson for sponsoring my G R E A T scholarship and for providing me with administrative, technical, and motivational support. Thanks also to my officemates in Nelson - Evan McKenzie, Tom Green, and Jakob Dulisse, as well as Devon Haag and Janice Anderson from my lab group at U B C . Tom Braumandl, Pam Dykstra and Emilee Fanjoy offered useful advice, maps and field equipment. Brian Cutts provided technical editing of a much earlier draft, and Dana Diotte made beautiful maps of my study area. A special thanks goes to my family for their support, and especially to my mom, Keren Freed, who proofread the final draft for me and helped in so many other ways! A final thanks goes to Norco Performance Bikes and Gerick's Cycle for keeping me busy on my mountain bike. Work is nothing without exciting distractions. CHAPTER ONE: Introduction The amount of old-growth forest remaining in British Columbia has declined considerably over the past century due to timber harvesting and other human activities (MacKinnon 1998). As old growth is replaced with managed forests, unique ecological functions, stand structural elements, and species assemblages are altered. These elements are particularly important for biodiversity conservation since species associated with old forests tend to have inflexible habitat requirements such as a need for large diameter trees, snags and logs, which, if removed can take centuries to replace (Bunnell and Kremsater 1991, Marcot 1997, MacKinnon 1998). In order to effectively manage for biodiversity conservation, forest managers require a clear understanding of old-growth characteristics and the pathways leading to their development. This study characterizes stand structure, age class distributions, and stand development patterns in mid-elevation old-growth cedar-hemlock forests in the West Kootenay region of southeastern British Columbia. The study objectives are twofold. First, throughout the study I evaluate methodologies commonly used to assess tree and stand ages, stand structure, and old-growth forest definitions. Second, I use these methods to describe the characteristics and development patterns of old-growth forests in moist warm variants of the Interior Cedar Hemlock (ICH) biogeoclimatic (BEC) zone in southeastern British Columbia (Braumandl and Curran 1992). Old-growth forests have been defined in many ways, including stand age, population dynamics, and stand structure. Age-based definitions are premised on pre-determined estimates of the time required for stands in a given ecosystem to develop expected characteristics of old growth. Definitions based on population dynamics provide insight into the processes shaping stand development and maintaining stand characteristics. Structural attribute definitions use the density of attributes such as large live and dead trees, plant species diversity, or decayed logs to identify stands with characteristics that are generally only found in older stands. Age, and population dynamics definitions relate to ecological processes functioning within a forest, while descriptions of old-growth structural attributes ensure that habitat and aesthetic characteristics are identified and maintained in conservation strategies. Regardless of definitions, old growth exists along a continuum of stand development. In forests with mixed species, multiple cohorts, and variable disturbance regimes, stand development patterns involve a range of ecological processes and lead to a variety of stand structures. Dendroecology, the study of stand history through tree rings, can be used to examine the scale of past disturbances and to evaluate the degree to which gap dynamics and larger-scale disturbances are shaping stand development and the resulting stand structures. The link between tree growth, disturbance and stand structure builds an understanding of how local old-growth conditions arise and provides a quantitative assessment of population dynamics. The document is structured such that each chapter builds on the findings of previous sections. This initial chapter provides background information on the study area, as well as the field methods used and a summary of stand structural characteristics for each study site. Chapter 2 assesses methods of determining tree ages with an emphasis on the errors associated with estimating ring counts in trees with internal decay and from coring trees above the root collar. Chapter 3 builds on tree ages estimated in the previous chapter and uses stand age, age structure, and stand structure to characterize and define old-growth forests in the moist warm ICH. In Chapter 4, dendroecological techniques are used to relate suppression and release patterns in individual trees to stand development patterns and to assess the ways in which individual species respond to canopy gaps. STUDY AREA Field sampling took place within the Arrow and Kootenay Lake Forest Districts. All sampling occurred in the Columbia - Shushwap Moist Warm Interior Cedar Hemlock (ICHmw2) and the Salmo Moist Warm Interior Cedar Hemlock (ICHmw4) biogeoclimatic variants (Figure 1 ). Mean annual temperatures for the study area range from 12.2 - 12.5 degrees Celsius, with mean temperatures of 25.5 - 26.7 degrees Figure 1. Map of the study area including the Arrow and Kootenay Lake Forest Districts. Field sites are identified by number. Celsius in the warmest month, and -0 .2 to -0 .8 degrees Celsius in the coldest month. Annual precipitation averages from 506.1 - 658.9 mm, with snowfalls of up to 225 cm per year (Canadian Climate and Water Information 2003). Moist warm ICH forests are generally found above the dry warm ICH subzone (ICHdw) in the southern portion of the region, and from valley bottoms to mid elevations in the northern half of the study area. Portions of the unit have been re-classified since field sampling took place. At the time of field sampling, the ICHmw2 included areas now classified as ICHmw4 and ICHdm (Dry Mild Interior Cedar Hemlock subzone) and was the largest ICH variant in the Nelson Forest Region (Braumandl and Curran 1992) with 921, 617 hectares of land. Three of the sites in this study are now within the ICHmw4 variant; however this new unit is very similar to the ICHmw2. The presence of bear grass (Xerophyllum tenax) dominated sites is the primary difference distinguishing the mw4 from the mw2 (Braumandl, pers. comm.). The ICHmw2 is found above the ICHdw and below the E S S F w d (wet cold Engelmann Spruce Subalpine fir) in the southern portion of the Nelson Forest Region. In central and northern areas of the region, it is found from valley bottoms to the ICHwkl (wet cool variant) or E S S F w d . The ICHmw2 is found as low as 500m (in areas where ICHdw does not occur) and, throughout the range, extends to 1550m in elevation on warm aspects and to 1450m on cool aspects. The ICHmw4 is only found above the ICHdw and is located below the ESSFwc5 , E S S F d m (dry moist) or E S S F w m (warm moist). The ICHmw4 ranges in elevation from 1250 - 1500m on warm aspects, and from 1075 - 1450m on cool aspects (Ketcheson et al. 2002). Western redcedar (Thuja plicata Donn ex D.,Don) and western hemlock (Tsuga heterophylla (Raf.) Sarg.) are considered 'climax' tree species on zonal sites throughout the ICH; Douglas"-fir (Psuedotsuga menziesii Franco Mirb), lodgepole pine (Pinus contorta var latifolia Engelm.), western white pine (Pinus monticola Dougl.), western larch (Larix occidentalis Nutt.) interior spruce (Picea englemannii (Parry) x glauca (Meonch) Voss), subalpine fir (Abies lasiocarpa (Hook) Nutt.), paper birch (Betula papyrifera Marsh.), black cottonwood (Populus balsimafera spp. trichocarpa Torr. & A. Gray), and trembling aspen (Populus tremuloides Michx.) are common serai or non-zonal species within the study area (Braumandl and Curran 1992; Ketcheson et al. 2002). The most abundant shrubs are black huckleberry (Vaccinium membranaceum) and falsebox (Paxistima myrsinities), and the most common understory herbs are queen's cup (Clintonia uniflora), twinflower (Linnea borealis), prince's pine (Chimaphila umbulata) and one-leaved foamflower (Tiarella unifoliata). Red-stemmed feather moss {Pleurozium shreberi), step moss (Hylocomium splendens) and pipecleaner moss (Rhytidiopsis robusta), as well as numerous lichens and liverworts, are common forest floor species. Natural disturbance patterns in the study area are characterized by infrequent stand replacing fires (Ministry of Forests and Ministry of Environment, Lands and Parks 1995) that leave a mosaic of climax and serai stands across the landscape (Braumandl and Curran 1992). For management purposes, the mean fire return interval is estimated at 200 years (Ministry of Forests and Ministry of Environment, Lands and Parks 1995), although actual return intervals likely vary considerably due to topography and aspect. Using multiple scenarios to estimate mean stand replacing disturbance intervals, Wong et al. (in press) determined that the potential'time between disturbances ranges from 160 years to 780 years. Historic fires are of natural origin (lightning), but extensive portions of the study area were burned by miners at the turn of the century. Fires typically follow topography and fuel distributions, burning uphill in finger-like patterns. Veteran, or legacy trees are often found either in groups or individually, with Thuja, Pseudotsuga and Larix as the most prevalent species (based on provincial inventory data). A preliminary study of post-fire residual stand structure found between zero and 200 live residual trees greater than 30 cm in diameter at breast height (dbh) and between zero and 58 dead trees of the same size on fire sites up to 30 years old (Steeger et al. 1998). This wide range reflects high variability in natural conditions. Mixed-species stands with multiple cohorts are common throughout the region due to the range of disturbance types and severities (Cameron 1996, Wong et al. in press). Although fire is the most widespread disturbance agent in the ICHmw2 and ICHmw4, avalanches and landslides are common in steep zones of the Columbia Mountains, windthrow occurs at various spatial scales, and insects and disease affect large areas of forest. Common insects and pathogens in the moist warm ICH include (from Boisvenue 1999): Laminated root rot (Phellinus weirii) Armilaria root rot (Armillaria ostoyae) Tomentosus root rot {Inonotus tomentosus) White pine blister rust (Cronartium ribicola) Mountain pine beetle (Dendroctonus ponderosae) Spruce beetle {Dendroctonus rufipennis) Douglas-fir bark beetle (Dendroctonus pseudotsugae) Dwarf mistletoe (Arceuthobium americanum) Dwarf larch misteltoe (Arceuthobium laricis) Hemlock sawfly (Neodiprion tsugae) Indian paint fungus (Echinodontium tinctorium) Hemlock looper (Lambdina fiscellaria tugubrosa) The Pinus monticola component of stands has been affected by Cronartium ribicola, a non-native blister rust. Loss of Pinus monticola can have a significant impact on stand structure at local scales. Eight percent of the Nelson Forest Region is protected in parks. Human settlements and development have reduced the area of native habitat, with low elevation areas most impacted (Utzig et al. 2003). Logging, flooding for hydroelectric dams, urban and rural development, and linear corridors (roads, powerlines, etc.) are the primary causes of habitat alteration and loss (Utzig et al. 2003). Logging has reduced overall habitat availability and changed historic landscape patterns and serai stage distributions (Utzig et al. 2003). Flooding of the Arrow Lakes, the Kootenay River, and the Duncan River for hydroelectric dams led to the loss of significant lower elevation, riparian, and wetland areas, and riparian and valley bottom areas along smaller tributaries such as the Slocan and Innonoaklin Rivers have been converted to farmland and townsites. The cumulative effects of these developments on biodiversity have not been examined. However, there are 18 red or blue listed species found in the ICHmw2 and ICHmw4 including mountain caribou (Rangifer tarandus caribou) and the northern leopard frog (Rana pipiens). METHODS Field sampling Sixteen sites in the Arrow and Kootenay Lake Forest Districts were sampled in this study. Sites were located along the West Arm of Kootenay Lake, at the Paulson pass, along the Arrow Lakes, and in the Slocan Valley (Figure 1 ). Site names, which were based on location or other distinctive feature, are shown in Figure l and are used throughout this thesis. The mean age of sample stands ranged from 137 to 470 years old. The British Columbia Ministry of Forests uses a nine-class system to identify forest cover age classes. Estimates of stand ages, based on interpretation of aerial-photographs, are provided for all provincially managed stands. Forest cover age classes (AC) are as follows: AC1 = 0-20 years; A C 2 = 21-40 years; A C 3 = 41-60 years; A C 4 = 61-80 years; A C 5 = 81-100 years; A C 6 = 101-120 years; A C 7 = 121-140 years; A C 8 = 141-250 years; A C 9 = 251+ years. Based on tree core data, one site was Age Class 7, six were Age Class 8, and nine stands were Age Class 9. Multi-aged stands were common, and all but one Age Class 8 stand contained 'veteran' trees that had survived past disturbances. Sample sites were randomly located at least 100 m from a forest edge and, to ensure consistent site classification, away from riparian zones, cliffs, and other topographic features. All sample sites were circum-mesic: twelve were zonal (site series 01 ), three were slightly drier than zonal (site series 04) and one was slightly wetter than zonal (site series 05; Braumandl and Curran 1992). The slightly drier and wetter sites were generally transitional to zonal. Trees were measured within a nested plot with three plot sizes (Figure 2). A 25.23 m radius (0.2 ha) was used for the large plot, while an 11.28 m radius (0.04 ha) was used for the main plot, and three small plots, used for seedlings and saplings, had radii of 3.99 m (0.005 ha) each. In the main plot, the diameter at breast height (dbh; 1.3 m), estimated height, damage (abiotic, pathogen and insect), crown class, and species were recorded for all live trees with a dbh greater than 7.5 cm. Tree cores were taken at breast height from all trees greater than 12.5 cm dbh and from at least three trees per species between 7.5 cm and 12.5 cm in diameter. All live trees greater than or equal to 50 cm at dbh were measured and cored in the large plot. Between three and six trees without internal decay were randomly selected and a second core was taken 20 cm above the estimated root collar location. The diameter, species, decay class, percent bark present, and top condition were recorded for all standing dead trees (snags) with a dbh greater than 7.5 cm in the main plot and greater than 20 cm in the large plot. The number of saplings (less than 7.5 cm dbh and greater than 1.3 m in height) and seedlings (less than 1.3 m tall) were tallied by species in three 3.99 m radius plots. Disks were taken at ground height and at breast height for between two and six saplings, depending on the density of regeneration. Lichen abundance on branches was estimated for each live tree as per Armleder et al. (1992). Lichen abundance estimates were recorded for both subcanopy and canopy levels. Pieces of coarse woody debris (CWD) greater than 7.5 cm in diameter were measured along two 25 m perpendicular transects. Diameters were taken at both ends of each piece to a minimum of 7.5 cm in diameter, and at the point of intersection with the transect. Length, decay class, and tilt angle (if greater than 25 degrees) were recorded for each piece. C W D volume was calculated using the following formula: Figure 2. Nested plot design. The largest circle represents the large plot (25.23m radius). The smaller circle represents the 11.28m main plot. The three smallest circles represent seedling/sapling plots and the black lines represent one randomly located 25m long C W D transect and a second 25m long C W D transect at 90 degrees to the first. SL Where V is volume in m 3 /ha, d is diameter (cm) of each piece of woody debris, and L is the length (m) of the transect. The number of C W D pieces per hectare (SPH) was estimated using the following formula: Where L is the transect length and /y is the estimated length of each piece (Marshall and Bugnot 1991 ). Vegetation and soil were used to determine the site series (see Braumandl and Curran 1992) for each plot and to provide additional information on stand structure and composition. All understory herb, shrub and moss species observed in the 25.23 m radius plot were recorded. The percent cover by layer was visually estimated for canopy trees (Layers A, A 1 , A2, A3), shrubs, seedlings, and saplings (Layers B, B1, B2); herbs (Layer C), and mosses (Layer D; Ministry of Environment, Lands and Parks and Ministry of Forests 1999). Thickness of the LFH layers and the humus form were measured at five randomly selected plot boundary locations. A shallow pit, ranging from 40 - 80 cm deep was dug at plot center to record soil texture, colour, coarse fragment content and size, parent material, and rooting depth. Data analysis Stand structural attributes were summarized on a per hectare basis for each stand. Attribute patterns were also assessed in relation to stand age. Stand ages used in this chapter are based on the highest ring count from a complete tree core found at each site (but do not include ring counts from veteran trees; see Chapter 3). This measure is intended to reflect the time since stand origin (time since the most recent stand-replacing disturbance), but may be underestimated because breakage and decay were common in tree cores sampled, and the oldest trees may not have had complete tree core samples. Stand age and tree age are discussed in detail in Chapters 2 and 3. SPH = 10000;r 2 L RESULTS and DISCUSSION Site descriptions Sampling included a broad range of site conditions that reflects the highly variable mountainous terrain of the ICH. Sample sites ranged in elevation from 950 m at Bremner to 1455 m at Glenmerry and in slope from less than 5% at Bremner to 88% at Hicks (Table 1 ). Eight sites were north facing, two were northeast, and one was west facing. Only one site (Clearcut) was south facing. Lichen abundance varied considerably and was not related to stand age. Plant species richness also varied across sites and the lowest number of plant species was observed at View where a dense overstory limited light at the forest floor. Relatively high species richness was observed in the three oldest stands, but the highest number of species was found at Hicks where there were small mesic and moist areas within the sample plot (Table 1 ). Three of the sample sites were in areas that were recently re-classified from ICHmw2 to ICHmw4 (Ketcheson et al. 2002), although Giveout was the only site where bear grass (Xerophyllum tenax), the primary indicator species for the ICHmw4, was found. Table 1. Elevation, aspect, slope, and biogeoclimatic information for each field site. Site Name BEC variant Site Series Elevation Aspect Slope% Number of plant species Green ICHmw2 01 1150 N 25 9 H i c k s ICHmw2 04 1213 N 88 29 Six Mile ICHmw2 04 1280 W 80 21 View ICHmw4 01 1410 N 42 4 Shields ICHmw2 01 1250 N 20 15 College ICHmw2 01 1320 N 51 15 Proctor ICHmw4 01 1350 NE 45 17 Russell ICHmw2 01 1450 N 65 9 Ski ICHmw2 01 1340 N 17 10 Kuskanax ICHmw2 05 1120 NE 42 21 Clearcut ICHmw2 04 1320 S 61 18 MacDonald ICHmw2 01 1380 N 48 14 Glenmerry ICHmw2 01 1455 NW 15 15 Giveout ICHmw4 01 1350 W 17 23 Pedro ICHmw2 01 1425 N 56 24 Bremner ICHmw2 01 950 N <5 18 Live trees A total of 743 trees were sampled at 16 sites. Tsuga was the leading species, by basal area, in 13 of 16 sample sites (Table 4). Thuja was dominant in Shields and Proctor, while Pseudotsuga was the primary species at Hicks. Broadleaved trees were uncommon on all sites, and only one Betula papyerifera and one Populus tremuloides were found at Green and Shields, respectively. Higher broadleaved densities would be expected in younger stands and earlier serai stages. Pinus contorta was also absent in all sites except View, although this is largely a function of site selection: sampling was restricted to circum-mesic sites where Pinus contorta is generally absent in moist warm ICH forests. Shade intolerant conifers were found in approximately half of the stands sampled, including both the younger and older sites. Basal area was highest at Glenmerry, where there was a relatively high density of large diameter trees and View, where there were numerous small diameter trees. Basal area was lowest at Shields and College. Stem densities were greatest at View and Green, where stem exclusion processes still influenced stand dynamics, and lowest in the older stands such as Giveout, Pedro, and Glenmerry where large diameter trees were dominant. Within the sample, mean dbh showed an approximately linear increase with stand age (r = 0.751; p < 0.001 ), although Six Mile and Hicks had several large diameter trees, but relatively young Pseudotsuga. The maximum dbh had a significant but weaker relationship with stand age (r = 557; p < 0.025). The largest diameter tree in the sample was a 122.2 cm dbh Thuja at Giveout (Table 2). Thuja and Pseudotsuga were the largest trees in 5 and 6 sites, respectively. Only Kuskanax, Giveout, Glenmerry, Six Mile, and College had trees 90 cm dbh or greater, but all sites except Hicks, Green, Russell , View, and Ski had trees over 70 cm dbh (diameter distributions are shown in Chapter 3). Table 2. Stand ages and mean and maximum tree diameters for each field site. Stand Mean StdDev Max Species with Site Name Age DBH DBH DBH MaxDBH Green 150 18.5 10.7 59.6 Fd Hicks 159 38.4 12.5 58.2 Cw Six Mile 185 41.2 '24.7 91.7 Fd View 185 17.1 9.7 66.5 Fd Shields 209 26.8 21.7 77.9 Cw College 220 27.3 20.8 90.0 Fd Proctor 256 36.5 22.1 78.9 Cw Russell 344 26.0 9.7 56.3 Pw Ski 439 32.6 22.6 69.0 Hw Kuskanax 441 46.0 28.5 95.4 Cw Clearcut 442 37.0 22.9 87.4 Fd MacDonald 500 42.8 21.2 78.1 Fd Glenmerry 526 50.9 25.2 99.4 Lw Giveout 553 60.5 26.6 122.2 Cw Pedro 559 48.4 23.8 87.5 Hw Bremner 630 48.0 26.0 88.5 Hw In a pilot study in the ICHmw2 (Holt et al. 1999), broken and dead tops were found to predict 'old-growthness'. In the sites studied here, 12% of all trees sampled had broken, dead or forked tops and the proportions increased with increasing stem diameter: only 10% of all trees under 50 cm dbh had broken, dead or forked tops, but 36.8% of trees 50-70 cm dbh, 49.1% of trees 70-90 cm dbh and 66.7% of trees over 90 cm dbh had 'problem' tops. Similar proportions were observed when all other pathogen indicators, including conks, blind conks, cankers, mistletoe, rotten branches, and scars, were combined: 22% of all trees under 50 cm dbh had one or more pathogen indicator while the same was true for 63% of all trees over 50 cm dbh. These figures are a function of senescence - as trees grow larger and older, they accumulate more pathogen and mechanical damage (Oliver and Larson 1996). Six Mile had the lowest incidence of pathogen indicators (1.5% of all trees), even among larger trees (see Table 3). However, the larger Pseudotsuga trees at Six Mile showed signs of Armillaria root decay. College, Green, and Shields had low pathogen densities (<20% of all trees), but these sites also had /4rm/7/an'a-infected trees as well as white pine blister rust. The low pathogen densities at Shields may be a function of the Thuja component, since Thuja generally has few pathogen indicators, although the incidence of internal decay is high. High levels of pathogen indicators also occurred at Hicks, where over 64% of the trees at Hicks had scars that were likely caused by moving rocks and snow damage, and at Glenmerry (46%) and Kuskanax (40%), where they were primarily related to treefall scars and conks. Table 3. Percentage of trees with pathogen indicators and with dead, broken, or forked tree tops at each field site. Site Name % frees with pathogens % trees with dead, broken or forked tops Green 10.6 3.4 Hicks 64.1 4.3 Six Mile 1.5 9.8 View 21.4 4.2 Shields 15.5 7.7 College 9.0 28.4 Proctor 25.3 1.1 Russell 21.1 30.1 Ski 32.4 8.0 Kuskanax 40.3 28.7 Clearcut 26.4 17.8 M a c D o n a l d 35.3 9.2 Glenmerry 46.4 21.8 Giveout 30.4 26.1 Pedro 29.1 39.5 Bremner 24.3 12.2 Table 4. Species-composition by basal area and stem density for each field site. Site Name Site Series BA/Ha SPH L S P * Hw Cw Fd Lw PI Pw 8/ Sx At Ep BA S P H %BA % S P H % B A % S P H %BA % S P H %BA % S P H %BA % S P H %BA % S P H %BA % S P H %BA % S P H %BA % S P H %BA % S P H Bremner 01 66.0 575 Hw Hw 93.4 64.3 6.6 35.7 Clearcut 04 48.9 645 Hw Cw 53.3 43.4 18.5 50.4 16.8 3.9 11.4 2.3 College 01 37.9 775 Hw Hw 49.3 61.3 9.5 16.8 14.0 1.3 4.5 0.6 14.3 3.9 8.4 12.9 4.4 3.2 Giveout 01 68.6 345 Hw Hw 72.8 91.3 23.8 7.2 3.3 1.4 Glenmerry 01 73.0 550 Hw Hw 55.9 80.0 33.7 18.2 10.4 1.8 Green 01 49.4 1740 Hw Hw 54.4 64.7 18.7 24.4 15.1 3.4 7.8 4.6 2.1 1.4 1.9 1.4 Hicks 04 61.5 585 Fd Fd 2.3 8.5 29.1 43.6 68.6 47.9 Kuskanax 05 65.6 645 Hw Hw 79.7 95.3 20.3 4.7 MacDonald 01 62.1 595 Hw Hw 84.0 93.3 16.0 6.7 Pedro 01 50.0 430 Hw Hw 68.1 67.4 26.6 31.4 5.3 1.2 Proctor 01 66.9 890 Cw Hw 46.7 71.9 50.0 24.7 3.3 3.4 Russell 01 46.3 830 Hw Hw 93.4 96.4 3.9 3.0 2.7 0.6 Shields 01 37.8 840 Cw Cw 1.8 3.0 76.8 89.9 3.7 0.6 14.5 3.6 3.3 3.0 Six Mile 04 60.6 665 Hw Cw 43.4 42.9 30.9 51.9 20.7 3.8 5.1 1.5 Ski 01 46.5 880 Hw Hw 93.1 77.3 6.9 22.7 View 01 71.6 2945 Hw Hw 58.2 76.4 2.4 0.2 22.4 7.3 7.1 4.2 1.9 0.8 5.6 7.6 2.5 3.4 * L S P = L e a d i n g S p e c i e s S p e c i e s c o d e s a r e : H w = Tsuga; C w = Thuja; F d = Pseudotsuga; L w = Lam; PI = Pinus contorta; P w = P. monticola; B l = Abies lasiocarpa; S x = Picea spp.; A t = Populus tremuloides; E p = Betula papyrifera. Standing dead trees Snag basal area varied across stand ages, but older stands had consistently high basal area for Tsuga snags (Figure 4). In general, the density of snags under 50 cm dbh decreased with stand age, while the density of snags over 50 cm dbh increased with age. The highest overall density of snags was found at View, where there were over 1000 snags per hectare under 20 cm dbh and no snags over 40 cm dbh (Table 5; Figure 3). Despite the lack of large stems, View also had the highest basal area per hectare. View had a high abundance of snags because of high site occupancy associated with the stem exclusion stage of forest development (Oliver and Larson 1996). The next highest snag density was observed at Green, although basal area was much higher for many other stands, particularly the oldest sites. The lowest snag density and basal area were found at Six Mile where one large (66.7 cm dbh) Pseudotsuga accounted for half of the total snag basal area. In general, Pseudotsuga, Larix, and Pinus monticola comprised a relatively small proportion of the sample size by stem density (23.1%), but a high proportion by basal area (43.4%), particularly at View, Shields, College, Russell , and Clearcut. Table 5. Snag density by size class (dbh) for each field site. Total Snag SPH SPH SPH SPH SPH SPH SPH SPH Site Name Density <20 20-30 30-40 40-50 50-60 60-70 70-80 80+ Green 375 300 50 5 15 5 0 0 0 Hicks 205 150 35 15 5 0 0 0 0 Six Mile 40 25 0 0 10 0 5 0 0 View 1145 1025 70 ' 50 0 0' 0 0 0 Shields 230 125 55 25 10 0 10 0 5 College 200 125 35 10 10 15 5 0 0 Proctor 210 150 40 15 5 0 0 0 0 Russell 285 125 110 40 5 5 0 0 0 Ski 75 50 5 0 10 5 0 0 5 Kuskanax 155 125 20 0 0 5 0 0 5 Clearcut 95 0 25 30 15 10 5 5 5 MacDonald 90 50 15 0 5 5 5 0 10 Glenmerry 110 50 25 5 5 10 5 10 0 Giveout 130 75 10 15 0 5 5 20 0 Pedro 45 25 5 0 0 0 10 5 0 Bremner 120 75 10 5 10 15 5 0 0 Snag distributions also varied by decay class across stand age (Figure 5; Figure 6). Highly decayed snags (DC8) comprised a considerable proportion of the older stands' basal area, and a minute amount of the younger stands' basal area. Many of these well-decayed snags had large diameters, and were relatively rare on a stems per hectare basis. For example, twp large decay class 8 snags at MacDonald (115 cm and 83.1 cm dbh) accounted for 63% of the total snag basal area on that site. Fire scars and charcoal suggests that they were from a pre-disturbance cohort (see age class distributions in Chapter 3). Windsnap and senescence, which are associated with heart rot and other pathogens, are the likely cause of large, highly decayed snags at Bremner, Giveout, and Clearcut where 58 - 84% of the snags had broken tops. Decay class 7 snags were abundant in both older and younger stands. In older stands, windsnap and senescence likely created the decay class 7 and 8 snags. However, in younger stands, particularly those with live veteran trees and signs of significant past disturbance, the abundance of decay class 7 snags relative to decay class 8 may be a function of time; given more time since disturbance, the large diameter veteran snags will progress towards decay class 8. Decay class 4 was the most common state found across all sites in the sample (by stem density and basal area), and was the most prevalent decay class at View, Proctor, and Russel l . Less decayed snags (DC2-4) were most common in the smaller diameter. classes, where they appear to have died due to suppression. Moderately decayed snags were spread across a range of diameters and species. Coarse woody debris Coarse woody debris provides important habitat for plants, lichens, invertebrates, and vertebrates V (Harmon 1986). C W D volumes were lowest at Hicks, View and Pedro, and highest at Ski , Clearcut and Six Mile (Table 6). However, View had the highest number of pieces per hectare, followed by Green, Hicks and Glenmerry. Most C W D was either Tsuga or of unknown species (too decayed to determine species without microscopic analysis), although the majority of C W D at Clearcut was Pseudotsuga. Proportions of known C W D species, by stem density, correspond reasonably well to those for snags and trees. Decay class 5 was the most prevalent class, while freshly fallen logs (DC2) were the most rare (see Figure 7; Figure 8). 1200 -i 1 Figure 3. Snag density by species for each field site. 20 T Figure 4. Snag basal area by species for each field site. Basal area per hectare • • S ID • • S o O D o o O a o o O o o o o w w "* 21 CD' c -1 <D O l 03 CD CQ Q . CD CO •< Q . CD O CD >< O . ' 03 V> V> CD co o CD Q . C/> «-*-' CD Stems per hectare u G> 00 Green • B B m El D 0 D O D D D O 0 O O O O O O Oi Co Figure 7. C W D stem density by decay class for each field site. Figure 8. C W D volume by decay class for each field site. The density (or volume) of large logs is critical because size, distribution, and orientation are considered to be more important to wildlife and plant species than volume (Bull et al. 1997). In addition, specific log sizes are generally correlated with wildlife use and longer, larger diameter logs provide habitat for a broader range of species (Bull et al. 1997). Larger and longer logs also have slower decay rates, and thus persist for longer periods of time (Harmon et al. 1986). In this study, average log lengths varied from 6.6 m at View to 17.1 m at Six Mile, although the longest logs were found at Bremner and MacDonald (32.9 m and 35 m). The largest diameter logs, based on the diameter of the large end, were found at Clearcut (74 cm) and MacDonald (71.5 cm), while the lowest mean diameter was found at View, where trees and snags were also smaller. Stands with high C W D volumes generally had several large diameter logs. For example, Ski , Clearcut, Kuskanax, and Six Mile had the highest C W D volumes and each site had over 75 large logs (>60 cm diameter) per hectare. Table 6. C W D lengths, diameters, and density (by diameter class). Site Name Length (m) Diameter (cm) Pieces per hectare - by diameter class Mean Max Mean Max 7.5-15 cm 15-30 cm 30-60 cm 60+ cm Green 11.3 26.4 15.5 29.0 260 947 109 18 Hicks 7.8 28.5 12.4 28.0 1173 74 28 0 Six Mile 17.1 28.0 36.2 65.0 0 41 74 79 View 6.6 18.1 11.1 23.0 1319 68 70 0 Shields 13.3 27.1 19.3 53.9 203 363 137 14 College 9.2 20.9 17.3 36.0 433 576 31 15 Proctor 10.1 27.8 25.5 66.0 95 618 380 68 Russell 9.5 19.6 17.5 31.8 710 418 100 0 Ski 15.3 32.7 28.9 60.0 0 586 205 138 Kuskanax 10.7 28.8 24.4 49.8 182 174 627 89 Clearcut 10.4 20.7 31.3 74.0 206 479 99 91 MacDonald 15.9 35.0 27.2 71.5 35 123 98 39 Glenmerry 8.7 19.0 27.8 58.0 205 680 327 17 Giveout 13.3 27.5 32.6 68.0 165 317 89 49 Pedro 13.0 29.1 18.5 35.0 131 69 292 0 Bremner 13.6 32.9 28.5 41.0 0 110 402 0 CONCLUSIONS There was considerable variation in stand structure in the 16 sites sampled for this study. In general, older stands had higher mean dbh values, higher densities of trees and snags over 50 cm dbh, and lower densities of trees under 30 cm dbh. However, exceptions were encountered at Six Mile (age = 185 years), where trees with a dbh over 50 cm were common, and at Hicks (age = 159 years), where densities of live stems under 30 cm dbh were more similar to older stands than younger stands. These types of differences in stand structure relate to differences in stand composition, site productivity, and disturbance histories (Oliver and Larson 1996). This chapter provided background information on the study area and the characteristics of each stand sampled. More detailed assessments of old-growth-related stand structure are presented in Chapter 3. CHAPTER TWO: Tree Age INTRODUCTION Tree ages are used in timber assessments and ecological studies. Ages are particularly important for studies of population dynamics, forest inventory data, disturbance histories and assessments of old-growth forest conditions (e.g. Tyrell and Crow 1994, Sano 1997, Kneeshaw and Burton 1998, Parish et al. 1999, Wong 1999). Age estimates can be obtained by counting tree rings from increment bore samples, although accurate ages are only obtained from trees where samples are extracted from the root-shoot interface, include a complete radius from bark to pith, and account for false and missing rings. Crossdating of tree cores can be used to correct for false and missing rings, and corrections can be applied, with various accuracy, to estimate growth from the root collar to sample height. However, heartwood decay is very common in Thuja plicata ((Donn. Ex D. Don) western redcedar) and Tsuga heterophylla ((Raf.) Sarg.) western hemlock) trees, and it is rare to find large, old trees that do not have hollow centers. Large Pseudotsuga menzeisii ((Franco Mirb) Douglas-fir) and Larix occidentalis ((Nutt.) western larch) are also commonly decayed. Decay is so prevalent in old-growth forests that accurate ages are not easily obtained and estimates are often the only option. This chapter addresses errors in age estimates due to missed piths and sampling above the root-shoot interface. There are no consistent approaches to estimating ages from cores with missed piths. In general, acceptable levels of error depend on study goals. Some authors have chosen not to include hollow-centered trees in studies of disturbance history and population dynamics (Parker and Peet 1994, Tyrell and Crow 1994, Abrams and Copenheaver 1997, Ishikawa et al. 1999, Parish et al. 1999) while others have developed various approaches for extrapolating ages. In an old-growth Abies-Acer-Quercus forest in Hokkaido, Japan, Abrams et al. (1999) counted rings and assigned minimum ages to badly broken or decayed cores (35 of 67 cores). In a study of age structure in old forests in northwestern British Columbia, Kneeshaw and Burton (1997) adjusted the ages of trees with decay by matching inner growth patterns with neighbouring trees of the same species and similar diameter class. However, cores with more than 2 cm missing were not given age estimates. Frehlich and Graumlich (1994) also considered cores complete if they terminated within 2 cm of the tree center, and estimated the number of missing rings using average growth rates from the five years closest to the pith. Alternatively, best-fit regressions of age and diameter have been used to estimate ages from trees with decay at their center (Duncan and Stewart 1991 ). Combinations of these approaches have also been applied. In coastal forests near Vancouver, Daniels and Klinka (1996) used the mean ring width of the 10 measured rings closest to the pith and the length of missing radius to estimate tree ages for cores that missed the pith, but where the missing radius was considerable, they used site- and species-specific regression models to predict age from tree diameter. As these studies indicate, various techniques have been applied to estimate ring counts from trees with incomplete cores. The primary focus of this chapter is on quantifying the error associated with two methods of estimating ring counts from cores where the pith was missed due to breakage or internal decay: (i) linear extrapolation from growth patterns on intact portions of tree cores; and (ii) regression analysis on size and age relationships. Linear extrapolation involves measuring growth rates on portions of intact tree cores, then extrapolating the average yearly growth onto the estimated length of missing core (Stephenson and Demetry 1999). In this study, ring count estimates are extrapolated using the growth rate from all rings, and from the innermost 25 tree rings prior to decay. Errors from coring trees at breast height are also addressed by measuring the difference between tree cores and sapling disks extracted at breast height and ground height. Ring count estimates from the analyses presented here are used as tree age estimates in subsequent chapters of this thesis. METHODS Data processing Breast height and ground-height cores were mounted on wooden supports, sanded, and counted under a binocular microscope with up to 60x magnification. Sapling disks were moistened to highlight individual rings and counted under the microscope. Cores were not crossdated, and false and missing rings are not addressed in this study because tree rings were examined by decade and in 50 year age class increments rather than on a year-by-year basis. Growth to breast height All saplings encountered during field sampling were either Tsuga or Thuja. Average growth to breast height was estimated for each species at each site using disks cut from saplings at 1.3 m above ground and at the estimated root collar ('Sapling Method', Wong 1999) and/or from cores taken at breast height and 20 cm above the estimated root collar ('Modified Ground Method', Wong 1999). Comparisons were made between ring counts at ground height and breast height for tree cores, and for ring counts at the estimated root-shoot interface and at breast height for disks. A total of 48 saplings were included in the analysis with one to five samples per site. For trees, it was difficult to obtain ground-height cores due to breakage and decay. When data from all sites were combined, ground-height tree cores were extracted from a total of two Thuja, 21 Tsuga, and three other species, but two or more basal cores were only extracted from seven stands. An additional 18 cores were sampled, but were either broken, decayed or missed the pith and could not be used to assess growth to breast height. Estimating missing ring counts When all trees sampled in the larger study were considered (see Chapter 1 ), the majority of trees less than 50 cm dbh were solid to the pith. However, 67% of Tsuga and 82% of Thuja over 50 cm dbh were rotten in the middle. The high rates of decay among large trees underlie the importance of determining reliable age estimates from tree cores with missing rings. Assessing the accuracy of linear extrapolation The accuracy of linear extrapolation techniques was evaluated using a subset of 94 tree cores from Tsuga (n = 54), Thuja (n = 16) and Pseudotsuga I Larix (n = 24) trees. Pseudotsuga and Larix have similar growth rates (Cameron 1996) and were pooled in order to increase sample sizes. Cores were selected from all sites and represented a range of diameters. To increase sample sizes, all cores either included pith or were estimated to be within five years of the pith. For Tsuga, all cores assessed and tested using linear extrapolation were within three years of the pith. Sample sizes were small for species other than Tsuga due to decay and species relative abundance at the study sites. For Thuja, most large trees were decayed. Each of the cores selected for analysis was hypothetically 'broken' in order to test the estimation methods. A random numbers table was used to determine a decade for the 'break' in each core. The average growth rates over the total core length and from the innermost 25 years of growth following the 'break' were then used to extrapolate estimated ages. Preliminary tests on Tsuga cores were conducted to determine whether the innermost 10, 25 or 50 years were to be used for extrapolation. Twenty-five years was chosen because it provided slightly more accurate results, although the difference was not statistically significant. The linear extrapolation approach assumes that the pith is in the geometric centre of the tree and that tree cores are equal in length to a tree's radius. Thus, to determine the radius of wood, bark thickness was estimated then subtracted from the total tree radius. Bark thickness was estimated using ratios from the PrognosisSC growth and yield model for southern British Columbia (A. Zumwari 1 , pers comm. 2001; Appendix 1). Once the estimated missing radius was determined, differences were quantified between actual ring counts and estimates using growth rates from the total core and the innermost 25 years. A univariate analysis of covariance using the General Linear Model in S P S S ( S P S S Inc. 1999) was used to determine whether there were differences between actual ring counts and estimated ring counts using the average growth rate from the total core and from the innermost 25 years. Diameter at breast height and percent intact core were entered as covariates. Errors in age estimates from the total core and the inner 25 years were summarized by the percentage of the intact radius included in the 'broken' core (<40% and >40%) and by dbh class (<30, 30-50, 50+ cm dbh). Trees were separated by size class because larger trees were expected to have more complex growth histories, and thus higher rates of error. Trees greater than 50 cm dbh were separated from smaller trees to account for the 50 cm dbh limit used in the nested plot design (Chapter 1). Paired t-tests were used to assess whether significant differences were observed between estimates and counts when data were stratified by the percent radius intact. Finally, tree core lengths were compared to the inside bark radius of all complete tree cores to test whether piths were located in the geometric centre of individual trees. Assessing the accuracy of age on dbh regressions Simple linear regression equations between dbh and age were developed for Tsuga trees using the remaining cores that were within five years of the pith. A new set of cores was selected for regression analyses so that the accuracy of regressions could be tested on the same subset of Tsuga cores used to assess the accuracy of extrapolation techniques. Regression models were developed for Tsuga trees from all sites, and on a site-specific basis where there were a minimum of seven complete cores remaining. Comparisons were made between breast height ring counts, extrapolated ages using the average growth rate from the total core and inner 25 years, and ages from regression equations. Regression analysis was used to estimate ring counts for Thuja trees on two sites, but was not applied to other species or other sites because sample sizes for complete cores were too small. 1 G r o w t h a n d Y i e l d B i o m e t r i c i a n , B r i t i sh C o l u m b i a M in i s t r y of F o r e s t s , R e s e a r c h B r a n c h . RESULTS Growth to breast height Thuja and Tsuga were the only saplings encountered during field sampling. They were also the most abundant trees sampled, and are therefore the only species with sufficient sample sizes to evaluate growth to breast height. The estimated time required for trees to grow from ground height to breast height varied across sites. Although sample sizes were small, growth to breast height for saplings ranged from eight years at View to 160 years at Russel l , and for trees from nine years at Proctor to 77 years at Giveout (Table 1 ). Where both Thuja and Tsuga were present in the sapling layer, Thuja took longer than Tsuga to reach breast height in all cases (Figure 1). On average, Tsuga trees (n = 33) and saplings (n = 16), and Thuja trees (n = 2) took approximately 25 years to reach breast height. Thuja saplings (n = 32) took an average of 50 years (std = 24), which was longer than the mean of 26 years for trees, although sample sizes were very low for trees. When trees and saplings were combined, the median number of years to breast height for Thuja was 40. Data for all other species is sparse. Reliable ground cores were taken from one each of Pseudotsuga, Larix, Pinus contorta and Abies. Ages to breast height for these ranged from 13 to 16 years with an average of 14 years (not shown). Tsuga Thuja Figure 1. Average number of years for growth to breast height using the sapling method and the ground method for Tsuga and Thuja. Estimating the number of missing rings using linear extrapolation When cores that were further than an estimated five rings from the pith were removed, only 190 Tsuga, 61 Thuja, and 30 Pseudotstuga /Larix complete cores remained in the sample. Of these, 94 cores were randomly selected to test linear extrapolation techniques, of which 14 Tsuga, 4 Thuja, and 14 Pseudotstuga /Larix were more than 50 cm in dbh. In randomly selecting samples, no cores were included from Hicks or Green. The subset of trees ranged from 7.5 - 92 cm dbh, with 14 Tsuga, four Thuja and 14 Pseudotsuga /Larix trees over 50 cm dbh. After cores were 'hypothetically' broken, the percentage radius intact ranged from 17% - 96% for Tsuga, 32% - 1 0 7 % for Thuja, and 16% - 1 3 3 % for Pseudotsuga / Larix. Where the radius intact exceeded 100%, the pith was considerably offset. Errors in age estimates were highest for Pseudotsuga / Larix, and lowest for Thuja. Analysis of covariance did not detect statistically significant differences between actual counts and estimates using either the total core or the innermost 25 rings for Tsuga (p = 0.466), Thuja (p = 0.522), or Pseudotsuga / Larix (p = 0.129), which suggests that both methods have similar precision levels. However, a significant relationship was observed between the percent radius intact, a covariate, and both estimation methods for all species (p<0.006). This trend can be seen in boxplots where accuracy decreased when cores included less than 40% of the radius (Figure 2). This pattern was particularly evident for Pseudotsuga I Larix when the innermost 25 rings were used and for Tsuga when both methods were used (Figure 2.). Sample sizes were small for Thuja (n = 16), although the trend was consistent with other species. In contrast, dbh was not a significant covariate for Thuja or Tsuga (p>0.2), although it was significant for Pseudotsuga I Larix (p = 0.017). Larger Pseudotsuga I Larix had less accurate estimates; however this was most pronounced in cores with less than 40% of the radius (Figure 2; Table 2). Table 1. Summary statistics for the estimated number of years to grow to breast height using the sapling method, the ground method, and the average of both methods. AVERAGE of SAPLING and SAPLING METHOD GROUND METHOD GROUNG METHODS Site Name Species n Mean SE Min Max n Mean SE Min Max n Mean SE Min Max Thuja 3 49 13 33 75 Bremner Tsuga Total 2 31 19 12 50 5 42 10 12 75 Thuja 4 66 7 56 86 Clearcut Pseudotsuga Total 1 15 15 15 5 56 11.5 15 86 Thuja 1 57 57 57 College Tsuga Total 1 2 32 45 13 32 32 32 57 Giveout Tsuga 3 61 9 45 77 Thuja 3 58 7 45 70 Glenmerry Tsuga Total 1 4 31 51 8 31 31 31 70 1 14 14 14 2 5 23 44 8.5 10 14 31 14 77 Hicks Thuja 1 32 32 32 Kuskanax Tsuga 2 28 5 23 32 1 10 10 10 3 22 6 10 32 Thuja 1 28 28 28 MacDonald Tsuga 3 26 6 16 37 1 31 31 31 4 27 4 16 37 Total 4 27 4 16 37 5 27 3 16 37 Thuja 2 44 26 18 69 Green Tsuga 2 27 10 17 36 2 29 1 28 29 4 28 4 17 36 Total 4 35 12 17 69 6 26 4 17 36 Thuja 2 87 16 71 102 1 30 30 30 3 68 21 30 102 Pedro Tsuga 1 47 47 47 Total 3 73 16 47 102 4 63 16 30 102 Thuja 3 18 4 11 25 Proctor Tsuga Total 3 10 1 9 13 6 14 2.5 9 25 Russell Tsuga 2 136 25 111 160 2 33 7 26 40 Shields Thuja 5 29 6 16 44 1 22 6 28 5 16 44 Thuja 3 49 17 18 76 Six Mile Tsuga Total 1 4 24 43 14 24 18 24 76 Thuja 4 72 5 58 80 Ski Tsuga 1 21 21 21 3 23 10 12 42 4 22 7 12 42 Total 5 61 11 21 80 8 '47 10 12 80 Abies 1 16 16 16 View Tsuga 2 9 1 8 9 3 22 6 13 34 5 17 5 8 34 Larix Pinus contorta 1 1 13 13 13 13 13 13 Total 2 9 1 8 9 6 18 3 13 34 8 16 3 8 34 All Sites Thuja 32 50 4 11 102 2 26 4 22 30 34 48 4 11 102 (excluding Tsuga 16 39 10 8 160 21 28 4 9 77 37 33 5 8 160 Russell) Total 60 43 4 8 160 23 26 3 9 77 75 39 3 8 160 In cores with less than 40% radius intact, tree age was consistently over-estimated for all species and all size classes. Estimates from cores with more than 40% of the radius intact were both higher and lower than actual ring counts, suggesting low precision and accuracy, but no bias. When less than 40% of the radius was used to extrapolate Tsuga ages, larger errors were observed. Using mean growth rates from the total intact core method, the average difference between the number of rings estimated and counted for trees greater than 50 cm dbh was 105 years (n = 3; std = 106). The same value was three years (n = 11 ; std = 55) when more than 40% of the radius was used (Table 2). The same patterns were observed in Thuja, and Pseudotsuga I Larix. These findings suggest that linear extrapolation is not a reliable method of estimating ring counts from cores with less than 40% of a tree's radius intact. Tsuga Thuja Pseudotsuga / Larix N = 9 9 8 8 8 8 9 9 1010 1010 <30 30-4040-5050-6060-70 70+ %Kadius intact N = 4 4 3 3 1 1 1 1 7 7 30-40 40-50 50-60 60-70 70+ % Radius Intact N = 2 2 7 7 5 5 5 5 6 6 <30 30-40 40-50 50-70 70+ % Radius Intact Figure 2. Relationship between the percent intact radius and the accuracy of linear extrapolation using average ring widths from the total core and from the innermost 25 years before a break or decay for Tsuga, Thuja, and Pseudotsuga / Larix. Comparing linear extrapolation methods Samples were stratified using the 40% intact core threshold, and paired t-tests were used to compare the actual number of rings counted to the various estimation methods used for each species or group of species. When less than 40% of the tree radius was used for ring count estimates, significant differences were found for Tsuga and Pseudotsuga I Larix where estimates were derived using growth rates from both the total core and the innermost 25 years, while no differences were found if over 40% of the radius was intact (Table 3). The mean difference (absolute value) for Pseudotsuga I Larix cores was lower using the inner 25 rings, but for Tsuga samples it was slightly lower using the total core. None of the estimates were significantly different for Thuja cores, regardless of intact core length. These findings suggest that any of the methods can be used for Thuja cores, but the innermost 25 rings is likely to provide more accurate estimates for Pseudotsuga and Larix, which are shade intolerant and grow quickly (Cameron 1996). There were virtually no differences between estimates using the total core and the inner 25 years for Tsuga, although the innermost 25 years produced a wider range of errors than the total core (Figure 2; Table 2). The total core is likely to provide better estimates for Tsuga than the innermost 25 years because Tsuga is shade intolerant and often experiences multiple periods of fast and slow growth (see Chapter 4). If the innermost 25 years of a broken core happens to be measured during a period of suppression, the missing ring count may be over-estimated. If the inner 25 years are measured during a period of rapid growth, ring counts may be under-estimated for missing portions of the core. Table 2. Summary statistics for differences between actual ring counts and estimates using the total core and the innermost 25 rings, stratified by species, tree size class, and percent intact radius. <40% >40% Size Species class Estimation Max Max Max Max (dbh) method n Mean Std under over n Mean Std under over Tsuga <30 Total core 5 52 52.9 9 142 17 9 28.1 -33 66 Inner 25 years 28 45.0 -20 77 6 25.3 -48 54 30-50 Total core 5 86 37.8 29 126 13 5 46.0 -98 91 Inner 25 years 68 31.2 34 110 11 52.8 -118 88 >50 Total core 3 105 106.2 -18 172 11 3 55.2 -83 113 Inner 25 years 77 83.7 -17 142 -7 72.4 -110 130 Thuja <30 Total core 1 55 55 55 5 8 28.8 -27 51 Inner 25 years 36 36 36 -1 30.7 -42 32 30-50 Total core 2 26 15.9 15 38 4 -6 52.2 -76 50 Inner 25 years 9 2.2 7 10 -13 48.9 -80 36 >50 Total core 1 150 150 150 3 2 52.4 -53 51 Inner 25 years 148 148 148 -11 37.4 -54 13 Pseudo-tsuga 1 <30 Total core 3 139 62.1 85 207 1 -32 -32 -32 Larix Inner 25 years 61 25.8 35 86 -35 -35 -35 30-50 Total core 3 284 141.8 121 379 4 46 45.5 0 99 Inner 25 years 282 229.5 59 518 25 26.4 0 49 >50 Total core 3 345 234.2 176 613 11 62 105.1 -128 238 Inner 25 years 261 242.5 86 538 14 83.2 -208 113 Setting confidence levels for estimates Although tree size was not a significant covariate for Tsuga and Thuja, larger trees of all species had higher variation in errors and higher mean differences than smaller trees (Table 2). This observation is important because it is often the larger trees for which age estimates are desired and because it directly relates to setting confidence levels for estimates. In this study, confidence levels have been stratified to account for the percent radius intact and the tree size. Error levels are based directly on the standard deviation of the mean estimate for each species, dbh class, and intact radius class (Table 3). The standard deviation was selected since it reflects the spread of the observations. Table 3. Mean difference between estimated and counted tree rings using the total intact core and the innermost 25 rings. Significant differences indicate that estimated ages are not equivalent to counted ages. Tsuga Pseudotsuga / Larix Thuja Percent Method Mean P. Mean P Mean P Intact Difference* Difference Difference <40% Total 77 0.001 -266 0.004 -64 0.118 25 Years 54 0.003 -216 0.021 -50 0.226 >40% Total -6 0.337 -53 0.060 -2 0.878 25 Years -4 0.601 -10 0.625 7 0.496 * Mean difference (years) between counted breast height age and each estimation method. ** Significance = 0.025 after multiple comparisons were adjusted using the Bonferoni correction factor. Assessing errors due to off-centre piths Extrapolation techniques assume that a tree's pith is at its geometric center, and large errors are likely when this assumption is false. Agee and Huff (1986) report that trees growing on hillsides often have piths that are located slightly to the uphill side of center. It is also common to find trees with non-concentric rings or horizontally offset piths. Of the 190 complete Tsuga cores sampled in this study, 63% were shorter than the geometric radius and had piths that were off-center to the uphill side of the tree, 14% had cores longer than the geometric radius and piths off-center on the downhill side, and 23% had a radius within 0.5 cm of the tree radius. After bark thickness was removed from the radius, Tsuga cores were, on average, 1.2 cm shorter than tree radii (n = 26; S E = 0.2). Larger Tsuga trees had a larger difference between measured core lengths and tree radii, with trees more than 50 cm dbh having cores an average of 3.7 cm shorter than the radius (Figure 3). For one Tsuga tree, the core length was 16.1 cm shorter than the radius. Thuja and Pseudotsuga /Larix trees between 30 and 50 cm dbh had the largest different between core lengths and radii (Table 4). T s u g a Thuja P s e u d o t s u g a / Larix N= 130 34 26 N = 49 7 5 N= 4 9 17 <30 30-50 50+ <30 30-50 50+ <30 30-50 50+ Diameter c l ass Diameter c l ass Diameter c l ass Figure 3. Differences between inside bark radii and measured core lengths for Tsuga, Thuja, and Pseudotsuga /Larix trees <30 cm, 30-50 cm, and >50 cm dbh. Table 4. Mean, minimum and maximum differences between inside bark radii and measured core lengths for Tsuga, Thuja, and Pseudotsuga /Larix trees <30 cm, 30-50 cm, and >50 cm dbh. Species Size Class Mean difference (cm) SE Maximum under- estimate Maximum over- estimate (cm) (cm) <30 0.5 0.1 -2.7 4.1 Tsuga 30-50 1.8 0.4 -4.4 8.0 >50 3.7 0.9 -3.6 16.1 All sizes 1.2 0.2 -4.4 16.1 <30 0.2 0.2 -4.1 3.3 Thuja 30-50 1.0 1.3 -3.6 7.5 >50 •1.1 2.8 -12.1 3.4 All sizes 0.2 0.3 -12.1 7.5 <30 0.5 0.7 -1.4 1.7 Pseudotsuga 30-50 1.4 0.5 -0.4 4.7 /Larix >50 0.9 0.7 -3.9 5.8 All sizes 1.0 0.4 -3.9 5.8 Estimating the number of missing rings using regression A linear regression was developed between age and diameter of 112 Tsuga trees with measured ring counts from cores that were within five years of the pith. Tree cores from Russell were excluded in this model because tree growth was consistently faster at all other sites (Figure 4). Site- and species-specific equations were developed for seven study sites (Table 5) where there were a minimum seven complete cores per site. Sample sizes were too small to develop site-specific equations for eight of the remaining sites. Regression models were not derived for Kuskanax because all of the Tsuga trees over 30 cm dbh had significant internal decay. Linear regression models were developed for each of the sites. Using curve estimation techniques ( S P S S Inc. 1999), the best fit at Russell was from quadratic and cubic models, but the linear approach was chosen because tree growth more closely approximates linear growth patterns than cubic or quadratic models. 600 10 20 30 40 50 60 70 D B H Figure 4. Relationship between diameter at breast height and age (ring counts) for Russell and for all other sites combined. Estimated ages were closest to actual ring counts when site-specific equations were applied (Figure 5; Table 6). Where regression models were developed, estimated ages were on average within +/-10 years of actual ring counts for Green, MâcDonald, Proctor, Ski and View; At Russell and Bremner, average differences between regression-based ages and actual ring counts were - 1 7 years and 32 years. Overall, the average difference between counts and estimates using site-specific regressions was eight years (n = 27; S E = 9.2). When absolute differences in age estimates were applied, estimates were within an average of 20 years for Green, MacDonald, Proctor, Russell , and View. However, average estimation errors were 40 years at Ski and 76 years at Bremner. Using Tsuga trees from all sites led to an average difference between counted and estimated ages of 10 years (n = 112; S E = 7.3) and an average of absolute differences of 76 years. In keeping with the pattern developed for linear extrapolations, confidence levels for estimates using regression are based on the standard error of the estimate, since it describes the spread of the observations. Table 5. Site-specific regression equations between age and dbh based on all complete Tsuga cores not used to test linear extrapolation methods. Sample size (n) for regression Model Coefficient of Determin-ation (R2) p-value Standard Error of the Estimate (Sy,) Mean difference in years (and Std) between estimate and count Mean ABSOLUTE difference in years (and Std) between estimate and count All Sites 112 yi = 49 + 4.45(xi) 0.552 <0.001 48.2 10.2 (77.2) 56.7 (52.9) (except Russell) Bremner 7 yi = 13 + 7.23(xi) 0.858 0.003 67.8 32.2 (94.8) 76.4 (56.8) Green 22 yi = 84 + 1.71(xi) 0.210 0.032 16.7 -8.5(10.9) 10.7 (8.5) MacDonald 8 yi = 121 + 2.02(xi) 0.793 0.003 10.8 0.4 (19.2) 15.2 (9.7) Proctor 12 y = 61 +0.81(xi) 0.698 0.001 22.7 -0.7 (10.5) 7.9 (5.7) Russell 18 yi = 245 + 2.02(xi) 0.364 0.008 20.8 17.1 (12.9) 17.1 (12.9) Ski 15 yi = 49 + 4.92(xi) 0.836 <0.001 45.8 2.1 (53.2) 39.8(21.3) View 21 yi = 111 + 1.38(») 0.217 0.033 12.3 -8.0(11.6) 9.4(10.1) a) b ) 30-50 Diameter class 30-50 Diameter class Figure 5. Difference between regression estimates and tree ring counts using regression models based on all field sites (a) and regression models based on individual sites (b) for Tsuga trees with dbh <30 cm, 30-50 cm, and >50 cm; Table 6. Summary statistics for differences between actual ring counts and estimates from regression models based on Tsuga trees from all sites and from individual sites, stratified by tree size class. Size class (dbh) Regression method n Mean Std Max under Max Over <30 Site-specific 22 -4.6 19.4 -30.2 63.0 All-sites 31 -7.0 44.4 -122.4 76.0 30-50 Site-specific 9 7.1 17.6 -32.1 29.0 All-sites 18 33.3 78.0 -100.4 157.8 >50 Site-specific 5 35.9 105.9 -100.4 171.2 All-sites 14 18.8 120.1 -253.4 185.7 All sizes Site-specific 36 3.9 42.1 -100.4 171.2 All-sites 63* 10.2 77.2 -253.4 185.7 * N o t e : a n a d d i t i o n a l n i n e c o r e s f r om G r e e n w e r e a d d e d to t he s u b s e t o f c o r e s t e s t e d for l i nea r e x t r a p o l a t i o n in o r d e r to tes t t he a c c u r a c y of s i t e - s p e c i f i c r e g r e s s i o n at G r e e n . Regression equations were attempted for Thuja trees from all sites combined and on a site-specific basis where at least seven complete cores were present. Site- and species-specific regression models were created for Six Mile (n = 9; y, = 51 + 2.0(Xj); R 2 = 0.623; S y . x = 25.4; p = 0.11) and Shields (n = 11 ; y = 73 + 1.8(Xi); R 2 = 0.566; S y x = 22.1 ; p = 0.008), but sample sizes were too small for the remaining sites and the regression was not significant for Green (n = 10; p = 0.054). The regression was not significant for all sites combined (n = 47; p = 0.138). Sample sizes were too small to make comparisons between extrapolation- and regression-based estimates for Thuja. DISCUSSION Growth to breast height Years to breast height estimates in this study varied widely, even within species and sites. For example, estimates of breast height growth ranged from 12 to 50 years for Tsuga trees in Bremner, and from 30 to 102 years for Thuja trees in Pedro (Figure 1 ). This degree of high variability in early height growth is expected in old-growth stands where advance regeneration dominates the understory. Palik and Pregitzer (1995) found that breast height ages did not adequately capture relative age differences among stems. In particular, they found that trees establishing in an initial post-disturbance cohort had faster early height growth due to greater resource availability than trees establishing later in the stand's development. Similarly, vegetative stems, like much of the Thuja regeneration in this study, had faster early growth rates, even in later cohorts, than non-vegetative stems. With growth corrections, breast height cores still produced inconsistent age estimates, and led to the assumption that stem exclusion periods (Oliver and Larson 1996) lasted for a significantly longer period. A study of nine logged stands between 11 and 32 years old in the ICH zone around Nelson also found high variability with growth to breast height ranging from a minimum of two years for paper birch to a maximum of 21 years for a spruce tree (Froese 2000). Thuja and Tsuga ranged from 3 - 1 8 years with a mean of 9.39 and 8.42 years. Froese found shorter periods of growth to breast height than reported in this study because his stands contained saplings that were generally grown in the open with minimal impacts from shade or other competitive factors. Suppressed early growth creates the largest barrier to accurate age estimates among shade tolerant regeneration. DesRochers and Gagnon (1997) also had difficulties finding the exact location of the root collar, particularly on black spruce (Picea mariana (Mill.) BSP) trees where adventitious rooting is common. Although adventitious rooting was not a factor in this study, ground level cores were taken from approximately 20 cm above the estimated root collar and underestimate actual ring counts. Wong (1999) assessed the difference between ground and breast height cores and concluded that even with 'good' estimates of growth to breast height, caution should be used in interpreting the resulting age estimates. Despite difficulties in obtaining accurate growth to breast height age estimates, breast height cores were justified in this study for several reasons. Extracting cores at the root-shoot interface is extremely time consuming due to difficulties in locating the root collar, as well as added challenges due to steep terrain. For example, it is often necessary to dig a hole around the base of a tree in order to extract a ground level core (Wong 1999). In addition, decay is often more prevalent at ground height and decreases at increasing tree heights. Errors associated with growth to breast height should be considered when interpreting results relating to tree ages. Comparing methods of estimating missing rings Stephenson and Demetry (1999) report that linear extrapolation often over-estimates missing ring counts. However, correlations between tree age and diameter are often weak (Stewart 1986, Daniels et al. 1995, Parish et al. 1999). Using site- and species-specific regression equations between dbh and age provided the most accurate estimates of tree age in the subset of Tsuga cores examined, particularly when cores with less than 40% of the radius were included (Figure 6; Figure 7). The regression equation derived from all Tsuga trees on all sites gave the poorest results, overestimating the age of younger trees and underestimating for older trees. Site-specific regressions account for variability in growing conditions and site productivity across sites and thus resulted in more accurate regressions than when Tsuga from all sites were combined. Table 7. Percent error for different estimation methods. Species Method Percent radius intact Percentage within 10% (+or-) Percentage within 25% Mr-) Tsuga Extrapolation Total core <40% 7 4 Extrapolation Total core >40% 37 61 Regression Site Specific 50 94 Thuja Extrapolation Inner 25 rings All cores 38 75 Pseudotsuga /Larix Extrapolation Inner 25 rings <40% 0 0 Extrapolation Inner 25 rings >40% 25 42 Age estimates using linear extrapolation had levels of accuracy close to those from site-specific regressions when mean growth rates were derived using more than 40% of the radius. With greater than 40% of the radius, estimates were statistically unbiased; below 40%, ring counts were biased upwards with over-estimates of individual tree ages as high as 170 years for Tsuga and 600 years for Pseudotsuga I Larix (Table 3). If over 40% of the radius was used, estimates were within 10% of the O O P o " o o o ® Ë 5 > <3 o o 2 o O Actual Ring Count • Extrapolation - Total Core A Regression - Site Specific ORegression - Ail Sites Figure 6. Average ages for Tsuga cores with greater than 40% of the radius intact - estimated and counted. 5 1 • Actual Ring Count • Extrapolation - Total Core A Regression - Site Specific ORegression - All Sites Q 2 .52 Figure 7. Average ages for Tsuga cores with less than 40% of the radius intact - estimated and counted. actual counts 37% of the time for Tsuga and 25% of the time for Pseudotsuga I Larix. Estimates were within 25% of the actual count for 61 % of the Tsuga and 42% of the Pseudotsuga ILarix. The 40% threshold was not as apparent for Thuja, but 38% of the estimates were within 10% of actual ring counts and 75% were within 25% of ring counts (Table 7). However, sample sizes were too small to be conclusive. Half of the estimates using site-specific regression models for Tsuga were within 10% of actual ring counts and 94% were within 25% of actual counts. The error rates are close to other estimation techniques reported in the literature. Stephenson and Demetry (1995) developed a method to address incomplete tree ring records in giant sequoias (Sequoiadendron giganteum (Lindl.) Buchh.) using a power function of the bole radius, the growth rate of the innermost 100 rings, the tree core and radius lengths, and the change in growth rates over time. Their estimates were within 10% of the actual age 62% of the time and within 25% of the actual ages 98% of the time, even if cores were from less than 20% of the radius. Errors in estimates from Stephenson and Demetry's (1995) method were generally due to uneven growth patterns, making the equation most applicable to 'pioneer' species such as Pseudotsuga /Larix and less useful for shade tolerant species such as Tsuga that generally have variable growth rates. The importance of accurate age estimates is linked to the ability to obtain accurate and complete cores, as well as the degree of accuracy desired for specific study goals. For example, in studies of old growth stand structure, where age is only one attribute among many, estimated ages are an important source of information since precise ages are difficult to obtain. Where disturbance history or age class frequency distributions are of interest, the magnitude of error associated with age estimates must be addressed. For example, Wong (1999) assessed the impact on age class distributions from using estimates of growth to breast height in Pseudotsuga-Pinus ponderosa (Dougl. Ex P. & C. Laws.) stands in south central British Columbia and found that age class distributions could be seriously misinterpreted unless the width of age classes was linked to the range of error observed in age estimates. For example, if Wong (1999) had used five-year age classes rather than 20-year age classes, she would have concluded that multiple moderate-severity disturbances had affected her study area rather than one moderate-severity disturbance. Limitations Site-specific regressions may be the most accurate method of estimating ages, but they are time consuming to develop because they require numerous complete cores, which are difficult to collect in old-growth Tsuga-Thuja stands. For example, in this study, 696 trees were cored and only 282 hit pith or were within five years of the pith. At a site level, only half of the study sites sampled contained more than 10 trees with complete cores - even where over 40 trees were cored - and no sites had more than eight complete cores over 50 cm dbh. Thus, where decay is prevalent, linear extrapolation may be a more time-efficient approach than attempting to sample enough complete cores to develop a regression model. Overall, the lowest variability and highest accuracy among extrapolated estimates was found for Thuja cores. The highest variability and lowest accuracy was observed for Pseudotsuga I Larix trees over 30 cm dbh and with less than 40% of the radius intact (Table 2). Pseudotsuga and Larix are generally long-lived serai species in the ICH zone. They have fast early growth and high longevity. Many older Pseudotsuga and Larix trees in this study had growth rates as high as 2 mm per year close to the pith, followed by a slow decline to less than 0.3 mm per year over the most recent 100 years. While no significant differences were observed between methods for Pseudotsuga and Larix, the 25 innermost rings provided the more accurate estimate in over 90% of the cases as well as lower standard deviations, and lower overestimates (Table 2). Using the innermost 25 rings is likely to provide more accurate calculations of missing ring counts for Pseudotsuga and Larix given that these shade intolerant species are expected to establish early and grow rapidly in larger openings, with growth declining over time (Cameron 1996). Tsuga is highly shade tolerant (Wright et al. 2000) and is able to remain in a slow-growing suppressed state in the understory for hundreds of years. In the sample assessed here, large errors in age were relatively rare for Tsuga and Thuja, but in some cases, extrapolations led to overestimates by as much as 94%. For example, a Tsuga at Six Mile with a breast height ring count of 170 years had an extrapolated estimate of 330 years because recent growth, which was used in the extrapolation, was much slower than early growth. In this study, maximum growth rates for Tsuga were as high as 4 mm per year or 4 cm of growth every 10 years, and as low as 0.05 mm per year or half a millimeter of growth over 10 years. Comparisons of extrapolations using the total core and the innermost 25 years indicated that the total core provides more consistent estimates of missing rings. In older stands, using the total intact core to estimate ages may provide balance for highly variable growth rates. However, researchers should be aware that large errors are probable with any estimation method and should evaluate the likelihood of any unusual seeming age estimates in relation to other tree cores in the stand. CONCLUSIONS Age estimates are never as good as actual counts, but they are the only source of data for many trees in the ICH. Site- and species-specific regressions provided the most accurate estimates for incomplete cores, followed by linear extrapolations on core samples that contained more than 40% of the radius. However, even with site- and species specific regressions, age estimates were only accurate to plus or minus 30 years, based on the standard error of the estimates from all of the sites where regressions were developed. For linear extrapolation with over 40% of the radius used, Tsuga cores over 50cm dbh were only accurate to within plus or minus 55 years, based on the standard deviation of the mean difference between estimated and counted ages. Error rates for large Thuja and Pseudotsuga!Larix were 37 and 83 years, respectively. Cores with less than 40% of the radius intact did not produce reliable estimates of missing rings, and, if used, should be applied with extreme caution. No significant differences were found between age estimates using the average growth rate from the total intact portions of a core or the innermost 25 rings, although trends suggest that using the total core will provide better estimates for Tsuga, which is more shade tolerant and commonly undergoes periods of suppression and release, while using the inner portion of the core is preferable for less shade tolerant species such as Pseudotsuga and Larix that generally have fast early growth. All ring count estimates using the methods described in this paper will have a degree of error since cores were extracted at breast height. Variable early growing and establishment conditions led to a wide range of ages to breast height within and across sites. Environmental variation may also obscure actual ages in regression analyses where some trees in a stand simply have faster growth rates than others due to site conditions or genetic advantages. With linear extrapolations, errors are inherent due to off center piths and elliptical or oddly shaped growth rings. Estimates will be imprecise where the actual distance to the pith is not equal to the tree radius. Despite the level of error, age estimates will still be an important part of stand-level research in the ICH. However, the levels of error presented here should serve as a caution in interpreting such estimates. CHAPTER THREE: Definitions and descriptions of old-growth forests in moist warm Interior Cedar Hemlock stands INTRODUCTION Empirical definitions of old-growth forests are important for management and conservation decisions. Old-growth forests have been described in many ways; however, there are few objectively derived thresholds for 'old growth' and definitions are all based on some degree of subjectivity (Hunter and White 1997, Wells et al. 1998). This chapter develops definitions and descriptions of old-growth forests in moist warm Interior Cedar Hemlock (ICH) forests using stand age, population dynamics, and stand structural attributes, which form the basis of most definitions (Pojar 1991, Spies and Franklin 1996, Wells et. al. 1998). Tree age is the simplest method for defining old-growth forests. Current forest policy in British Columbia defines old forest on the basis of forest cover age classes (Ministry of Forests and Ministry of Environment, Lands and Parks 1995, 1999). Age classes are based on the estimated mean age of the largest individuals of the leading species in a stand and are recorded on maps and in databases for all Crown-administered forest lands. Under this approach, all stands with an estimated average age greater than a specified threshold are considered old forest. For the moist warm ICH forests of southeastern British Columbia, 250 years (Age Class 9) is the 'old growth' threshold listed in the Forest Practices Code Biodiversity Guidebook (Ministry of Forests and Ministry of Environment, Lands and Parks 1995). Age-based definitions allow forest managers to use existing data sources to identify potential old-growth forests, but do not acknowledge that the rate of stand development varies widely depending on the species present, site productivity, and disturbance history in a stand (Thomas et al. 1988, Oliver and Larson 1996, Spies 1997), or that age estimates in the Forest Cover dataset are derived from aerial photograph interpretation rather than field studies and are often highly inaccurate (e.g. Holt et al. 2001, Wilson 2001). Describing stand age is relatively simple in single cohort or even-aged stands, but what is the age of an uneven-aged stand? Stand age can be characterized in many ways, including the average age of trees in the main canopy, the time since the last stand-replacing disturbance, or by the distribution of ages in a stand. Several authors have suggested that population biology and age structure analyses are the most appropriate means of defining and identifying old-growth forests (Oliver 1981, Hayward 1991, Oliver and Larson 1996). Using this approach, 'old growth' includes all stands where gap-phase replacement processes are the primary influence on forest structure and understory microclimate (Kneeshaw and Burton 1998, Wells et al 1998). Oliver and Larson (1996) divide stand development following major disturbances into four phases. Trees establish and compete for growing space during the stand initiation stage. During stem exclusion, new trees no longer invade and some existing trees die due to competition for resources. The understory re-initiation stage begins as overstory trees in the stand begin to senesce and die creating openings that lead to new growing space. True old growth is not reached until all trees that invaded following a stand-replacing disturbance have been replaced through small-scale processes by a new cohort of trees. Stands with a mixture of original and post-disturbance cohorts are referred to as transitional old growth. Using this functional approach, old-growth stands do not necessarily contain large, old trees. Similarly, stands with large, old trees are not necessarily 'old growth'. For example, many of the well-studied Pseudotsuga stands in Oregon and Washington state (e.g. Spies and Franklin 1988) are not true old growth, according to Oliver and Larson, but are merely comprised of large, long living, 'serai' species. Franklin et al. (2002) argue that the processes and functional attributes associated with each of the above phases occur at varying scales throughout all stages. They have divided stand development into eight stages: disturbance and legacy creation, cohort establishment, canopy closure, biomass accumulation/competitive exclusion, maturation, vertical diversification, horizontal diversification, and pioneer cohort loss. Studying population dynamics requires extensive time and resources. For this reason, quantitative approaches to defining old growth tend to focus on stand structural elements since they are relatively easy to measure and can be manipulated through forest management (Wells et al. 1998). In addition, the large-sized structural attributes associated with old growth are most important from a habitat perspective (Spies and Franklin 1988). These structural attributes are not readily available in younger forests or managed stands and are not easily or quickly created (Bunnell and Kremsater 1991). Because old-growth structures are generally linked to older stands, definitions based on structural attributes are premised on the assumption that forest structure is a surrogate for measuring ecological processes (Wells et al. 1998). Important structural attributes for habitat include: number of large diameter live trees; number and diameter of standing dead trees; volume and size of coarse woody debris (CWD); presence of live trees with broken or damaged tops; depth of litter and duff layers; plant species composition and diversity; patchiness associated with small scale disturbances; and abundance of arboreal lichens and bryophytes (Spies and Franklin 1988, Franklin and Spies 1991, Pojar 1991, Mehl 1992, Rebertus et al. 1992, Kneeshaw and Burton 1998, Franklin et al. 2002). Regardless of definition, old-growth forests exist within a continuum of forest development. In this chapter, old-growth forests in the moist warm ICH zone are described and defined based on stand age, population dynamics and stand structure. Definitions and thresholds for old growth under each of these paradigms are compared and contrasted using field data. Stands are classified, using each definition, as 'mature', 'old', or 'ancient'. The ancient category represents stands that have attributes or ages well beyond minimum thresholds developed for each definition. The comparisons and descriptions developed provide baseline data on the complexities of stand structural attributes and development processes. METHODS Tree ages Sixteen stands, ranging from stem exclusion.to old-growth stages, were sampled for this study, and in total, 744 trees (>7.5 cm dbh) were measured (see Chapter 1 for field methods). Of these, 607 had tree core samples extracted. All tree cores were sampled at breast height, and as indicated in Chapter 2, age to breast height varied widely in stands sampled. Analyses of tree cores and sapling disks taken at breast height and ground height suggested that, on average, Tsuga trees took 25 years to reach breast height, Thuja trees took approximately 30 years, and Pseudotsuga and Larix took approximately 15 years (see Chapter 2). To provide an unbiased estimate of growth to breast height and an age comparable to forest cover and other data, these averages were applied to trees in all stands where mean and maximum stand ages are discussed. Estimated years to breast height were added to the 282 trees with ring counts to within an estimated five years of the pith. An additional 115 trees had cores that were within an estimated 20 years of the pith using concentric ring overlays (Applequist 1958). The concentric ring overlay was deemed more accurate for estimating missing ring counts than extrapolation of average growth rates or regression of age on dbh since the approximate distance missing to pith was known. Age estimates, using regression and extrapolation as outlined in Chapter 2, were applied to 211 cores that were broken or decayed. Of the remaining 136 trees measured in the field, 118 were between 7.5 and 22.5 cm dbh and were not cored in order to increase time efficiency in the field, and 18 trees had cores that were too decayed and broken to be counted. In estimating ring counts for the trees with incomplete or missing cores, site- and species-specific regression equations were used where possible because they showed the lowest rates of error (Chapter 2). Where no cores were sampled and regression models were not developed, ages were estimated using a best-guess approach based on an examination of ages for similar sized trees (height and dbh) of the same species. These ages should be considered with caution. For cores with greater than 40% of the radius intact, the average growth rate from the total core was applied to Tsuga trees. The total core was also used to estimate Picea ages, because Picea is considered a slower growing, shade-tolerant species (Cameron 1996). The average growth rate of the innermost 25 years was used for species that typically have fast early growth: Pseudotsuga, Larix, Thuja, Pinus monticola and Populus tremuloides. When extrapolations were applied to cores with less than 40% of the radius intact, extremely high age estimates resulted in many cases. For example, the highest measured ring count in Kuskanax was 408 years for an 84.4 cm Tsuga, but the highest estimated count was 1397 years for a 64.5 cm diameter Thuja tree. Estimates like these place trees in a completely different cohort than the remainder of the trees in the stand. Since there was no evidence in the field to support such extreme ages, estimates for trees with under 40% of the radius were based on the lower extrapolation from the inner 25 rings and the total intact core. In many cases this still produced extreme ages. An upper age-class of 700 years or more was used to account for high age estimates. In general, cores with less than 40% of the radius intact should be evaluated with considerable caution. In this chapter, they were included in age class distributions, but were not used to determine maximum or average stand ages. Stand Age Stand ages were estimated using mean and maximum tree ring counts from dominant and codominant trees. Trees with age estimates from linear extrapolation or regression were not included in these measures. However, ring counts from incomplete cores with decay were often higher than those from the oldest complete cores in a stand, and were included in calculations of mean stand ages. Maximum ages from five stands (Giveout, Glenmerry, Kuskanax, MacDonald, and Shields) were from incomplete cores with only 50-86% of the radius intact. Thus, the ages presented here should be considered minimum reliable ages. Mean and maximum stand ages were used to classify stands into estimated stand development stages (mature, old, and ancient). Measured mean ages were also compared to forest coyer age classes to gauge the accuracy of forest cover inventory estimates. The mean stand age was based on cores from the leading major species (as per Ministry of Forests 2002), but the maximum age reported was taken from all species present. Major species comprise 20% or more of a stand, by basal area, while minor species are 10-19% and scattered species make up less than 10% of the stand (Ministry of Forests 2002). Veteran trees were considered separately from mean and maximum ages. Veterans are those that survived past disturbances and are larger and of an older age class than the stand as a whole. In simple stands, to be classified as a veteran, the total crown cover of remnant trees must not exceed 6% (approximately 25 stems per hectare), and trees must be at least 40 years older than the mean stand age. In complex, or uneven-aged stands, veteran trees must be at least 100 years older than the mean age of the main stand and must be much larger in diameter than other trees (Ministry of Forests 2002). Multi-layered stands do not generally have veteran trees but may have residuals that have survived past disturbances. In this study, the presence of fire scars was used to confirm that trees had survived stand-replacing disturbances. Some veterans did not have visible fire scars, but were still classified as veteran trees if there were signs of past fires throughout the stand. Population Dynamics Stand age only reflects a portion of the dynamics in a stand, particularly in uneven-aged stands where individual tree ages vary considerably. To provide a clearer picture of stand ages and development stages, age frequency distributions were created using estimated and counted breast height ages. Stem density by age class is commonly used to determine the number of cohorts in a stand and to classify stands into common stand types such as even-aged (single cohort), double cohort, and uneven-aged (three or more cohorts; Smith et al. 1997). The standard Yeverse-J' or negative exponential curve is often used to identify old-growth stands (e.g. Daniels 1994, Kneeshaw and Burton 1997, Orwig et al. 2001). In developing age class distributions, estimated ages were included to give a complete picture of the ages of all trees in a stand. Breast height ages were used to avoid high variability in early growth rates. Seedling ages were also excluded, so age class distributions should be interpreted to reflect the dynamics of trees at breast height and beyond. Saplings, defined as trees less than 7.5 cm dbh and more than 1.3 m tall, were not included in age class distributions, but their distributions were examined in determining the pattern of each age class curve. Saplings were not included because only a subset of sapling ages was sampled on each site, while the remaining ages were estimated for each stand using the ratio of saplings with known ages in a given age class to the total number of known ages sampled. Tree ages were stratified by species guild and by estimation method: actual ring count, site and species-specific regressions, and ring count extrapolations. Guilds are based on Cameron (1996) where IC = Shade Intolerant Conifers (Pseudotsuga, Larix, Pinus contorta, Pinus monticola); TC = Tolerant Conifers (Picea spp., Abies lasiocarpa); and IB = Intolerant Broadleafs (Populus tremuloides, Betula papyrifera). Tsuga and Thuja are Very Tolerant Conifers (VC), but were presented as individual species in this study because of their abundance and importance in the study area. Stand Structure Diameter distributions are often used as surrogates for age class distributions, and as with age class distributions, old-growth stands are expected to have an approximate reverse-J diameter distribution (Smith et al. 1997). Diameter class frequency distributions were assessed for each site, with trees stratified by 5 cm dbh classes. Saplings were not included in graphs, but their distributions were evaluated in determining overall patterns. Principal components analysis (PCA) was used to assess relationships between stand structural attributes and to explore whether any natural groupings (based on structure) occurred among the sites. P C A uses a correlation matrix of variables to find indices (principal components) that capture variation in different dimensions of the data. Each P C A axis is orthogonal (uncorrelated) with the others. PCA1 describes the maximum variation in the data and therefore describes the major patterns. P C A 2 is orthogonal to PCA1 and captures the next largest amount of variation in the data. Subsequent axes are also orthogonal to one another and describe diminishing proportions of the variation (Tabachnik and Fidell 1996). The results of P C A in this study are untested hypotheses that provide insight into stand structure. P C A does not assign clusters to the data, but the relationships between P C A and individual study sites were interpreted to identify underlying relationships. The analysis was first conducted using all variables related to tree and log size, decay class, and wildlife habitat values. Variables that had low correlations with all other variables in the dataset were excluded. Different models, containing different attribute sets were run using S P S S Factor ( S P S S lnc.1999).-The first principal component axis (PCA1) was graphed against the stand age to assess how stand age relates to P C A ordination. Although the analysis was intended to assess structural attributes, a relationship between age and structure was expected and was used to check whether P C A was representing 'old-growthness', or whether it measured some other value (such as differences in productivity among stands). Summary statistics of attribute values for each old-growthness grouping were used to develop a scorecard for ranking stands on the basis of their structural attributes. Values are presented on a per hectare basis throughout the analysis and in the scorecard. RESULTS Stand Age The field sample included age class 7 - 9 stands, where age class 7 includes stands 121-140 years old, age class 8 includes stands 141 - 250 years old, and age class 9 includes stands 251 years and older. The sites sampled here included one age class 7 stand (Green), six age class 8 stands, and nine age class 9 stands. According to age-based thresholds for old growth in the Biodiversity Guidebook, all age class 9 stands in the ICHmw area are 'old growth' (Ministry of Forests and Ministry of Environment, Lands and Parks 1995). These include the sites named Clearcut, Russell , Giveout, Pedro, and Ski - which had mean ages over 300 years old - and MacDonald, Kuskanax, and Bremner - which had average ages over 400 years old (see Table 1). Hicks, Shields, Six Mile, View, College, and Proctor had mean tree ages that place them in the mature category. When the mean age of the leading species was compared to forest cover estimates, 12 of 16 stands (75%) were correctly identified in the Forest Cover database. Giveout and Glenmerry were older than predicted, while Proctor and Green were younger than predicted (Figure 1 ). Using Forest Cover data, Proctor would be classified as old growth, but Glenmerry and Giveout would not. c CD CD O to o CO CD CO 5 CD > CD cn ô O o o 2 Q. CD E c _ÇD CD earcut ussell 'edro iveout O Q: O XL CO 2 CD o Q o co X CO c CO CO CD c: E CD CO Figure 1. Mean, maximum, and veteran age estimates (corrected to breast height) as compared to Forest Cover age classes (shown as rectangular boxes). If it is assumed that stands originated following stand-replacing fires, then maximum ages (which exclude veterans) represent the minimum time since stand initiation and best reflect true stand age. When maximum ages were considered, sample stands ranged from 150 years old in Green to 630 years in Bremner (although trees with incomplete cores are likely older). The presence of early serai species such as Larix and Pseudotsuga in the canopy of all stands except Bremner, Kuskanax, and Ski provide evidence of stand-replacing disturbances. Large shade intolerant conifers were found at MacDonald and Proctor, but only outside of sample plots. Using maximum ages to define old-growth forests, Green, Hicks, Shields, Six Mile, View, and College are mature, while Proctor, Russell , Ski , Clearcut, and Kuskanax are old. MacDonald, Glenmerry, Giveout, Pedro, and Bremner each had stand ages more than 500 years old and are classified as ancient in this study. Veteran trees were present at Green, Shields, View, Six Mile, and College and ranged in age from 295 to 370 years old (Table 1 ). At College, veteran trees were Thuja, Pseudotsuga and Larix. Veterans were Thuja and Pseudotsuga at Green, Pseudotsuga and Picea at Six Mile, Pseudotsuga and Larix at View and Thuja at Shields. Approximately half of the stand (by basal area) at Proctor was comprised of trees from an older cohort with scars from a turn-of-the century fire set by miners (H. Pinnell 1 , pers. comm.). Because of their relative abundance, these predominantly Thuja trees were classed as main canopy trees, not as veterans. The remainder of the stand consisted of a younger, post-fire and advance regeneration cohort of Tsuga, although one Thuja tree with extensive internal decay showed signs of surviving a major fire several hundred years ago. Table 1. Mean, maximum and veteran ages and their associated species at each site. Maximum Estimated Leading Maximum Oldest Veteran Veteran Estimated Oldest Site Name Mean Age Species* Age Species Age Species Age** Species Green 137 Hw 150 Pw 295 Cw 599 Fd Hicks 151 Fd 159 Fd 159 Fd Six Mile 162 Hw 185 Fd 296 Fd 328 Fd View 165 Hw 185 Hw 342 Lw 342 Lw Shields 158 Cw 209 Cw 370 Cw 244 Pw College 190 Hw 220 Cw 312 Cw 448 Fd Proctor 217 Cw 256 Cw 466 Cw Russell 315 Hw 344 Hw 346 Hw Ski 374 Hw 439 Hw 429 Fd Kuskanax 407 Hw 441 Cw 1427 Cw Clearcut 300 Hw 442 Fd 527 Cw Macdonald 401 Hw 500 Cw 965 Cw Glenmerry 290 Hw 526 Cw 998 Cw Giveout 346 Hw 553 Lw 1172 Cw Pedro 345 Hw 559 Lw 999 Hw Bremner 470 Hw 630 Hw 630 Hw * H w = T s u g a ; C w = T h u j a ; F d = P s e u d o t s u g a ; L w = Lar i x . ** A g e s f r om i n c o m p l e t e c o r e s w e r e not i n c l u d e d in the o t h e r m e a s u r e s o f a g e repo r t ed in th is t ab le . T h e a g e s repo r t ed in th is c o l u m n a r e the m a x i m u m a g e s fo r e a c h s i te u s i n g t he a g e s d e r i v e d f r om the m e t h o d s o u t l i n e d in C h a p t e r 2 . Population Dynamics Stem densities by age class Age frequency distributions exhibited several general patterns including reverse-J curves, bimodal and multi-modal distributions, and approximate bell-shaped curves, with variation in pattern at the scale of individual species (Figure 2). Even-aged stands are easily identified because stems are restricted to a few age classes. Individual species and/or guilds follow a range of patterns, even within a single stand, and include even-aged pulses as well as continuous establishment. 1 R e g i s t e r e d P r o f e s s i o n a l F o r e s t e r , H a r r o p - P r o c t o r C o m m u n i t y F o r e s t . Age class frequency distributions are shown in Figure 2 with tree ages stratified by species or species guild. Age sources - actual ring counts, linear extrapolation from cores with greater than and less than 40% of the radius intact, regression of age on dbh, and best-guess ages - are summarized in Appendix 2. Based on stem densities by age class, Bremner, Pedro, and Glenmerry were classified as 'ancient' stands because they had reverse-J shaped age class distributions. Giveout, MacDonald, Clearcut and Ski were 'old growth', while Kuskanax, Proctor, and Six Mile were transitional between 'mature' and 'old growth'. Giveout did not approximate a reverse-J curve because there were no saplings sampled in the subplots, although saplings were present in the stand. Ski and Kuskanax had distinctly bimodal distributions, and MacDonald and Clearcut were multimodal (Figure 2). It appears that there were several distinct cohorts in MacDonald: the oldest was more than 600 years old, with another between 300 and 550, a third between 100 and 250, and a sapling layer less than 100 years old. The presence of charred stumps, snags and larger trees suggested that a non-stand-replacing fire passed through MacDonald approximately 300 years ago. The group of trees over 600 years old were all from age estimates using extrapolations of ring counts with >40% of the core intact, and the break at 600 may be an artifact of inaccurate age estimations. The remaining stands - Green, Hicks, View, Shields, College and Russell -had even-aged, single cohorts and were classified as 'mature'. View, Green, and College contained veteran trees and there were high densities of relatively young trees and saplings in Shields and Proctor where fires damaged and killed numerous stems without creating stand-replacing conditions. The patterns described above outline the broader dynamics observed, but finer-scale patterns were evident within individual species and guilds. For example, the Tsuga component in Bremner closely resembled a reverse-J pattern with breast height age estimates ranging from 101-650 years old. However, Thuja trees were only found in the 101 to 350 year old range, and approximately 2000 saplings were present in the 51-150 age classes. Thuja dominated the sapling layer in Bremner, Clearcut, Glenmerry, Hicks, Pedro, Proctor, Shields, Six Mile, and Ski (Table 2) and was found in the older age classes in Glenmerry, Kuskanax, Proctor, Giveout, Pedro, and Clearcut. Intolerant conifers were only found in the oldest age class in Hicks, Green, Shields, Six Mile, Russel l , and Glenmerry, but were found in both the veteran and oldest post-disturbance age classes in College and View. Intolerant conifers such as Larix and Pseudotsuga require larger openings for establishment (Klinka et al. 1999). The restriction of intolerant conifers to older age classes in mixed species stands implies that establishment occurred following stand-replacing disturbances, which often leave veteran structures. In the younger, even-aged stands, the presence of intolerant conifers signified that the stands were still dominated by the initial cohort. However, more tolerant conifers, including Thuja and Tsuga, were among the older stems, which suggests concurrent establishment. Table 2. Estimated sapling density, by age class. Thuja Tsuga Total Saplings Site Name 1-50 51-100 101-150 1-50 51-100 101-150 All Ages Bremner 0 1333 667 0 0 0 2000 Pedro 300 300 0 200 0 0 800 Giveout 0 0 0 0 0 0 0 Glenmerry 400 200 0 200 0 0 800 MacDonald 400 0 0 3067 1533 0 5000 Clearcut 3000 0 - 0 0 0 0 3000 Kuskanax 0 0 0 0 400 0 400 Ski 900 1800 900 0 400 0 4000 Russell 0 0 0 300 300 0 600 Proctor 400 200 0 0 200 0 800 College 200 0 0 0 200 0 400 Shields 1200 1800 0 0 0 0 3000 Six Mile 600 0 0 200 0 0 800 View 0 0 0 0 200 200 400 Hicks 200 0 0 0 0 0 200 Green 0 1200 0 0 2200 0 3400 503 400 3C0 200 100 0 500 400 300 2D0 100 0 RarJo il. . o _ • a • - -3D 400 300 200 100 0 300 400 303 200 100 0 OverjUt • ••••• -Kusfcarac 1 • „ • _ 500 1 College 400-300-200 1 100 1 0 1 1 n _ _ 3 Gran • IB 0 H T C I • IC 1 H C w 1 • Hv 1 101- 201- 301- « V 5D1- 601- 7D> 193 293 393 493 SD 650 1 « l t ) 1 - 2 m - 3 0 1 - 4 3 1 - 5 0 1 - e O l - 7 n > 150 253 393 433 S5D 660 V93 1 0 1 - 2 0 1 - 3 D 1 - 4 3 1 - 5 0 1 - 6 0 1 - W > 133 253 350 493 550 650 V50 101- 201- 301- 431- 501- 601- 70> 150 233 393 450 593 661 Breast height ages Figure 2. Age class frequency distributions stratified by species for each site. IB = shade intolerant broadleaved species; TC = shade tolerant conifers; IC - shade intolerant conifers; Cw = western redcedar (Thuja); Hw = western hemlock (Tsuga). Stand Structure Diameter distributions Diameter distributions provide a third means of assessing stand dynamics by describing the range of tree size distributions in a stand. Diameter distributions have also been used as surrogates for age class distributions, although this is premised on the assumption that tree size approximates tree age, while size distributions approximate population dynamics. According to Smith et al. (1997), even-aged stands have diameter distributions that approximate the normal, bell-shaped curve and suggest that the majority of a stand has established following a major stand-replacing disturbance. Balanced uneven-aged stands have a reverse-J shaped curve, while irregular uneven-aged stands generally show up as various 'humps' along the diameter curve. These patterns only apply to pure stands or to mixes where all species have similar height and diameter growth. Where individual species within a stand have different growth rates, stratified mixtures occur and even-aged stands can approximate the reverse-J distribution. Shade tolerance further confounds the utility of size class distributions because smaller suppressed trees can accumulate in the understory (Lorimer 1985). When all species were examined together, reverse-J curves were observed for Bremner, Glenmerry, Pedro, Proctor, and Clearcut (Figure 3). Comparisons between age class and size class distributions indicated very similar patterns and curve shapes for Bremner, Glenmerry, and Pedro, and somewhat similar patterns for Clearcut. However, gaps were observed in some size classes for each of these sites. Age class distributions for Proctor showed a much larger concentration of trees in the 51 to 100 year age class as a result of the major fire approximately 100 years ago. Low stem densities in the 17.6 to 22.5 cm diameter class likely reflect this disturbance, although it was not evident from the diameter distributions alone. As with age class distributions, diameter distributions suggested at least two distinct cohorts at MacDonald, Ski , and Kuskanax, although the graph for Ski resembled an irregular uneven-aged reverse-J curve. As expected, the even-aged stands had approximately bell-shaped diameter distributions, although the left hand side of the curve is not developed at View and Green where stem exclusion processes were still affecting stand dynamics. 930 T 400 Jaoo fi.200 I 0 1 100 0 SB « 0 J 3 0 " S.2DO l 10D ami mi I l l n . l MxOonati l l l l l - l i . . . Green • IB B T C | B I C 1 R § B C w S • Hw l l l l l n 75-125 225-27.5 375425 52557.5 675-725 825875 7.5-125 225-27.5 375425 52557.5 67.5-725 825«7.5 75-125 225275 37.5425 52557.5 67.5725 825375 75-125 225275 375425 52557.5 67.5725 825675 Diameter at breast height Figure 3. Diameter class frequency distributions, stratified by species for all sample sites. A multivariate index of old growthness The diameter distributions shown above and the structural attribute summaries in Chapter 1 provide baseline data for individual attributes, but a multivariate approach is needed to assess old growth stand structure from a more comprehensive perspective that incorporates relationships between suites of attributes. In this study, principal component analysis was used to derive thresholds for old-growthness in moist warm ICH forests based on tree, snag, C W D , pathogen, and percent cover data. The relationships between variables and P C A axes are shown in the component matrix (Table 3). Larger loadings indicate that the variable is more highly correlated with the component and thus more representative of the axis. Scores higher than 0.71 reflect an 'excellent' correlation between the variable and the component; scores of 0.63, 0.55, and 0.45 are considered 'very good', 'good' and 'fair', respectively; scores of 0.32 are 'poor' and are at the lower limits of interpretability (Tabachnik and Fidell 1996). The first principal component explained 46.3 percent of the variation in the data and was highly correlated with the mean tree dbh, the density of trees over 50 cm dbh with dead, broken, or forked tops, the density of trees over 70 cm dbh, the estimated percent cover of the herbaceous understory layer, and the density of trees 50 to 70 cm dbh (Table 3). PCA1 was also negatively correlated with the density of trees and snags less than 30 cm dbh, and the density of snags 30 to 50 cm dbh. P C A 2 only explained 12.8% of the variation in the data and was negatively associated with the density of trees 30 to 50 cm dbh. The analysis also extracted a third, fourth, and fifth component, but these explained only 10.2, 8.1 and 7.9 percent of the variation and are not discussed further. Table 3. Principal component matrix showing correlations between structural attributes and P C A axes. PCA 1 2 Mean Tree DBH .914 -8.38E-02 Trees SPH<30 -.736 .651 Trees SPH 30-50 -.256 -.850 Trees SPH 50-70 .709 .110 Trees SPH>70 .802 .158 Trees >50 with DT/BT/FT .867 .342 Snags SPH<30 -.737 .581 Snags SPH30-50 -.688 7.772E-02 Snags SPH50-70 .463 -2.91 E-02 Snags SPH>70 .704 .273 Vol30-60CWD .580 -9.45E-02 CWD Volume>60 .442 -1.64E-02 %Cover Layer A (Trees) -.533 -1.77E-02 %Cover Layer C (Herbs) .775 .249 Extraction Method: Principal Component Analysis. P C A uses the underlying structure in the input data to arrange sample sites in multivariate space. The level of association between each site and each axis (the amalgamation of several variables) was graphed to determine underlying relationships between sample sites in the dataset. PCA1 was plotted against P C A 2 and the resulting graph shows a continuous gradient along both axes (Figure 4). Given that a continuous distribution of stands was sampled by age and serai stage, it is not surprising to find a correspondingly continuous distribution in 'multivariate space'. The attributes correlated with PCA1 are commonly associated with old-growth forests based on the literature. A PCA1 value of zero was used as an initial split between sites that were and were not 'old growth', because splitting the data where PCA1 equals zero has a statistical basis in that plots with a positive score are positively correlated with PCA1 and its associated variables, while those with a negative score are not. • ! ® j x * • i $ o X ; B O + •> O © ! • X > o Pedro MacDore -* © © X • O + -1 0 -3 -2 PCA1 Figure 4. Multivariate relationship between sample sites, based on structural attributes. The relationship between PCA1 and P C A 2 for all sites except View and Green is clearly linear. View (age = 185) had a high density of small stems (<30cm dbh) and a low PCA1 score (the highest absolute score) because the stand was still undergoing stem exclusion and high competition-induced mortality. From field observations and data analysis, it appears that Green also had some influences from the stem exclusion stage of stand development, although competition-induced mortality was much lower. Most stands less than 250 years old were negatively correlated with PCA1 (PCA1 <0). Conversely, most sites more than 250 years old were positively associated with PCA1 (PCA1>0; Figure 5). Russell (age = 344), which was comprised of old but small Tsuga trees, and Six Mile (age = 185), with several large but relatively young Pseudotsuga trees, are exceptions to this pattern. With the exception of Six Mile, Russell , and View, there was a strong, positive correlation between PCA1 and stand age (r = 0.766; p = 0.001; Figure 5). Unfortunately, the analysis was based on a small sample size and there are few sites with a stand age between 250 and 400 years old. However, an imperfect relationship between age (maximum age sampled) and old-growthness is expected since the rate of stand development will differ due to factors such as site productivity, partial disturbances during stand development, and species composition. X su > SuM.M, ^ Shields X o O RuattU © Proctor -fx *-^ Podro O MacDonald "X" Kuskana» O © Q Hick* Grwn. Glann-iony • > ° Grv»cvt O CoHeçO © 0 -}- Cleareul S BromrKf -3 PCA1 Figure 5. Relationship between P C A and stand age. The results from P C A were used to derive thresholds to identify and assess old-growth forests following the methodology developed by Holt et al. (1999). In conducting the P C A analysis, a range of structural attributes was used as input variables. While there was considerable variation in the relationship between sample sites and P C A 2 , the distribution of sites along PCA1 was generally constant when different suites of structural attributes were used. Several models could have been included, but the final model chosen was (i) that which explained most variation in the data, (ii) where the attributes associated with the main axes (PCA1 and PCA2) could be linked to expected patterns of old growth development based on theory and literature review of important old-growth attributes, and iii) where the largest number of variables were included while still meeting the first two criteria. Plots were separated into two categories. Those plots with a PCA1 score greater than zero were identified as 'old growth'. Those with a negative PCA1 score were considered 'mature'. Thresholds for 'old-growthness' were developed for several attributes (see Figure 6) using the midpoint between the mean of both groups, plus or minus each group's standard error of the mean. For an attribute that increases with 'old-growthness', the threshold was calculated as: [(mean of old growth - SE) + (mean of mature + SE)] / 2 No thresholds were provided where there was overlap between means and standard error of the means. For example, there were no thresholds for the density of snags 50 to 70 cm dbh, or the density of trees 7.5 to 12.5 cm dbh. View was not included in developing thresholds for the index of old-growthness because it had a significantly lower P C A score and was heavily influenced by stem exclusion processes. The results of this analysis are an 'index of old-growthness' (see Franklin and Spies 1991) whereby individual attributes can be measured in the field then compared to calculated thresholds to determine an overall index value. To use the index, field measurements for structural attributes that meet or exceed the threshold are given a score of one. Those that fail to meet the threshold are given a score of zero. Scores for all structural attributes measured are summed to produce a total score for a sampled stand. A sample index of old-growthness is shown in Figure 6. Additional thresholds were calculated for similar variables to provide some flexibility in data collection and are listed in Appendix 3. For example, thresholds are provided in Figure 6 for the volume of C W D by size class, but Appendix 3 has similar thresholds for C W D by stem density. Using thresholds for both volume and density of C W D by size class could effectively double count the significance of C W D in determining overall index scores. A summary of the measured values and the score allocated to each site using the index of old-growthness is provided in Table 4. Based on the scorecard presented in Figure 6, the maximum possible score for a site that met or exceeded each threshold is 26. Giveout and MacDonald received the highest scores (24 and 23), while Green (1), College (2), and View (3) received the lowest scores. MacDonald had a lower total score than Giveout because there was a high density of small diameter trees at MacDonald that may be associated with historic partial disturbances. Bremner and Glenmerry had very high scores (22), although both sites 'lost' points for having low C W D volumes. Ski (16), Clearcut (17), and Six Mile (19) had moderate scores that each reflected different combinations of structural attributes. Figure 6. An index of old-growthness for moist warm Interior Cedar Hemlock forests in southwest B C 2 Old Growth Scorecard for Moist Warm Interior Cedar Hemlock Forests Stand Name MaD Sheet Polygon Label Size (ha) BEC variant Site Series Slope Aspect Elevation Structural Attribute Measurement Threshold Score* TREES Maximum tree age > 342 years Mean tree diameter > 37 cm dbh Largest diameter Tree > 81 cm dbh Tree BA/Ha >56 m 2 Total trees < 710 sph Trees 12.5-17.5cm dbh < 106 sph Trees 17.5-30cmdbh < 181 sph Trees 30-50cm dbh < 151 sph Trees 50-70cm dbh > 49 sph Trees >70cm dbh > 19 sph Trees >50cm dbh with pathogens > 53 sph Trees >50cm dbh with DT/BT/FT > 25 sph SNAGS Total Snags < 166 sph Snags <20cm dbh < 100 sph Snags 20-30cm dbh < 29 sph Snags 30-50cm dbh. < 19 sph Snags >70cm dbh >3sph CWD CWD Volume > 309 m 3 CWD Density < 923 sph Volume CWD 7.5-15cm diameter < 6.2 m 3 Volume CWD 15-30cm diameter <48 m 3 Volume CWD 30-60cm diameter > 145 m 3 Volume CWD 60cm diameter >95 m 3 Volume CWD DC4+5 >213m 3 VEGETATION %Cover Layer C (Herbs) > 14.5 % %Cover Layer D (Moss) > 27.4 % Comments: TOTAL * Score = 1 if field measure meets or exceeds threshold; Score = 0 if field measure fails to meet threshold. 2 A d a p t e d f r o m Ho l t et a l . ( 1999 ) . Table 4. Thresholds, measured values and old-growthness scores for each site. Structural Attribute Threshold Green Hicks 6 Mile View Shields College Proctor Russell Ski Kuskanax Clearcut MacDonald Glenmerry Giveout Pedro Bremner TREES Maximum tree age > 342 years 150 0 159 0 185 0 185 0 209 0 220 0 256 0 344 1 439 1 441 1 442 1 500 1 526 553 1 559 ! 630 Mean tree diameter > 37 cm dbh 18.5 0 38.4 1 41.2 1 17.1 0 26.8 0 27.3 0 36.5 0 26 0 32.6 0 46 1 40 1 42.8 1 50.9 1 60.5 1 48.4 ! 48 < Largest diameter tree > 81 cm dbh 50.6 0 58.2 0 91.7 1 66.5 0 77.9 0 90 1 78.9 0 56.3 0 69 0 95.4 1 87.4 1 78.1 0 99.4 ! 122.2 1 87.5 88.5 Tree B A / H a >56 m* 49.4 0 61 1 60.6 0 71.6 1 37.8 0 37.9 0 66.9 1 46.3 0 46.5 0 65.6 1 48.9 0 62.1 1 73 1 68.6 1 50 66 1 Total trees < 710 sph 1740 0 585 1 665 1 2945 0 840 0 775 0 890 0 830 0 880 0 645 1 645 1 595 1 550 1 345 1 430 1 575 1 Trees 12.5-17.5 cm dbh < 106 sph 325 0 25 1 25 1 750 0 125 0 150 0 125 0 150 0 150 0 125 0 125 0 50 1 75 ! 75 1 75 100 ! Trees 17.5-30 cm dbh < 181 sph 650 0 200 0 150 1 1000 0 250 0 125 1 175 0 300 0 150 0 225 0 150 1 200 0 125 ! 50 1 125 ! 50 1 Trees 30-50 cm dbh <151 sph 100 1 325 0 175 0 50 1 75 1 175 0 250 0 325 0 50 1 100 1 150 1 225 0 100 ! 100 1 50 125 Trees 50-70 cm dbh > 49 sph 15 0 35 0 65 1 20 0 25 0 15 0 65 1 5 0 105 1 40 0 60 1 45 ' 1 110 ! 60 1 80 70 Trees >70 cm dbh > 19 sph 0 0 0 0 25 1 0 0 15 0 10 0 25 1 0 0 0 0 55 1 10 0 25 1 40 60 1 25 55 ! Trees >50 cm dbh with pathogens > 53 sph 15 0 25 0 25 0 5 0 35 0 20 0 80 1 0 0 95 1 75 1 35 0 65 1 120 90 1 75 100 Trees >50 cm dbh with DT/BT/FT > 25 sph 10 0 0 0 15 1 0 0 15 0 20 0 10 0 0 0 45 1 60 1 15 0 30 1 95 65 1 50 45 SNAGS Total Snags < 166 sph 375 0 205 0 40 1 1145 0 230 0 200 0 210 0 285 0 75 1 155 1 95 1 90 1 110 1 130 1 45 1 120 1 Snags <20 cm dbh < 100 sph 300 0 150 0 25 1 1025 0 125 0 125 0 150 0 125 0 50 1 125 0 0 1 50 1 50 75 1 25 75 « Snags 20-30 cm dbh < 29 sph 50 0 35 0 0 1 70 0 55 0 35 0 40 0 110 0 5 1 20 1 25 1 15 .- 1 25 ! 10 1 5 ! 10 Snags 30-50 cm dbh < 19 sph 20 0 20 0 10 1 50 0 35 0 20 0 20 0 45 0 10 1 0 1 45 0 5 1 10 15 1 0 ! 115 ! Snags >70 cm dbh > 3 s p h 0 0 0 0 0 0 0 0 5 1 0 0 0 0 0 0 5 1 5 1 10 1 10 1 10 20 1 5 0 0 Structural Attribute Threshold Green Hicks 6mile View Shields College Proctor Russell Ski Kuskanax Clearcut MacDonald Glenmerry Giveout Pedro Bremnei CWD C W D Volume > 309 m3 183 0 55 0 488 1 58 0 292 0 159 0 472 1 154 0 572 1 325 1 534 342 1 446 1 391 1 92 0 298 0 C W D Density < 923 sph 1333 0 1274 0 194 1 1457 0 717 1 1054 0 1161 0 1229 0 929 0 1072 0 875 \ 295 1 1228 0 620 1 491 1 513 1 Volume C W D 7.5-15 cm diameter < 6.2 IT|3 10.1 0 19 0 0 1 30.9 0 9.5 0 22.5 0 1.6 1 8.2 0 0 1 5.4 1 4.8 2.6 4.8 1 4.8 1 2.9 1 0 1 Volume C W D 15-30 cm diameter < 4 8 m 3 111 0 2.4 1 14.4 1 14 1 49 0 85 0 73 0 76 0 116 1 28 1 21 24 55 0 15 1 17 1 20 1 Volume C W D 30-60 cm diameter > 145 m3 54 0 334 0 134 0 13 0 161 1 19 0 289 1 70 0 149 0 169 1 150 153 302 1 102 0 72 0 278 1 Volume C W D B0 cm diameter >95 ms 8 0 0 0 339 1 0 0 72 0 32 0 107 1 0 0 306 1 122 1 357 163 83 0 269 1 0 0 0 0 Volume C W D DC4+5 >213m3 73.5 0 11.8 0 479 1 3.6 0 232 1 95 0 315 1 22.7 1 442 1 283 1 59.6 0 265 ! 239 1 157 0 56.8 0 512 1 %Cover Layer C (Herbs) > 14.5 % 1 0 10 0 30 1 1 0 15 1 10 0 6 0 1 0 7 0 45 1 2 0 20 45 1 65 1 15 1 10 0 %Cover Layer D (Moss) > 27.4 % 9 0 20 0 1 0 5 0 25 0 7 0 15 0 30 1 45 1 90 1 15 0 80 17 0 35 1 40 1 70 1 Total S C O R E Max = 26 1 5 19 3 6 2 9 3 16 21 17 23 22 24 21 22 DISCUSSION In this study, older forests were described and defined based on stand age, stand age structures, diameter distributions, and multivariate relationships between structural attributes. Each of these approaches is premised on quantifiable aspects of old-growth forests, but each also involves some degree of subjectivity. For example, using mean stand age, whether estimated or measured in the field, relies on a pre-determined threshold that does not account for differences in rates of stand development caused by site productivity, partial disturbances, or species complexes. In addition, thresholds for age-based approaches are influenced by the ease of using existing data sources. Using age class distributions provides more information, but requires a subjective assessment of patterns and distribution curves, as well as accurate measures of individual tree ages. Similarly, diameter distributions, although they provide valuable structural data, assume there is a strong relationship between tree size and age, which is often not the case. Structural attribute definitions based on analyses such as P C A use multiple attributes and are quantitative, but can be influenced by the number and type of input variables used. Definitions of old growth tend to be correlated with the objectives of those creating the definition (Rebertus et al. 1992, Hunter and White 1997). To provide a context for comparing the subjectivity of old-growth definitions, an explicitly subjective rating of old-growthness was applied to each stand (see Table 5). During field sampling, each site was given an initial subjective old-growthness rating (from 1-10) immediately upon reaching plot centre. The rating was based on a visual examination of stand characteristics and was largely intuitive. Although it has no 'scientific basis', the ratings were used for comparisons with stand age and statistically produced definitions of stand structure. Table 5 provides a summary of old-growth definitions for each site, using the various approaches outlined in this report. Only four stands (Green, Hicks, Clearcut, and Ski) are given the same definition using all of the methods employed. The remaining 12 sites varied in their classification. For example, using stand age data, Six Mile received a 'mature' rating, but diameter and age class distributions gave the site a 'transitional' rating, and P C A ranked it as 'old growth'. Glenmerry was classed as 'mature' using Forest Cover data, but was 'ancient' using field-based age data and age class distributions, while diameter distributions and P C A both led to 'old growth' ratings. Subjective ratings generally followed the patterns of P C A , although Six Mile was given a lower rating. Table 5. Summary of old-growth classifications by site. Site Name Measured Forest Measured Age Class Diameter PCA Index Subjective Stand Cover Age Age Data2 Distributions Distributions of Old- Old-Age1 Class Data (SPH) growthness growthness rating3 Green 150 Mature Mature Mature Mature Mature 2 Hicks 159 Mature Mature Mature Mature Mature 2 Six Mile 185 Mature Mature Transitional Transitional Old Growth 5 View 185 Mature Mature Mature Mature Young 1 Shields 209 Mature Mature Mature Old Growth Mature 4 College 220 Mature Mature Mature Old Growth Mature 4 Proctor 256 Old Growth Old Growth Transitional4 Old Growth Mature 6 Russell 344 Old Growth Old Growth Mature Mature Mature 5 Ski 439 Old Growth Old Growth Old Growth Old Growth Old Growth 8 Kuskanax 441 Old Growth Old Growth Transitional Transitional Old Growth 9 Clearcut 442 Old Growth Old Growth Old Growth Old Growth Old Growth 6 Macdonald 500 Old Growth Ancient Old Growth Transitional Old Growth 6 Glenmerry 526 Mature Ancient Ancient Old Growth Old Growth 8 Giveout 553 Mature Ancient Old Growth Old Growth Old Growth 8 Pedro 559 Old Growth Ancient Ancient Ancient Old Growth 8 Bremner 630 Old Growth Ancient Ancient Ancient Old Growth 8.5 1 B a s e d o n the r ing c o u n t o f the o l d e s t c o m p l e t e o r n e a r l y c o m p l e t e t ree c o r e s . D o e s not i n c l u d e a g e e s t i m a t e s f r om the m e t h o d s ou t l i ned in C h a p t e r 2 . 2 T h e o ld c a t e g o r y is b a s e d o n the 2 5 0 y e a r o l d t h r e s h o l d f r o m the B i o d i v e r s i t y G u i d e b o o k a s we l l a s a s e c o n d t h r e s h o l d at 5 0 0 y e a r s fo r A n c i e n t f o r e s t s . 3 B a s e d o n a s u b j e c t i v e a s s e s s m e n t c o n d u c t e d i m m e d i a t e l y u p o n e n t e r i n g the f ie ld plot ( s c a l e o f 1-10 wi th 10 b e i n g a p e r s o n a l ' i m p r e s s i o n ' o f h igh b i o d i v e r s i t y - v a l u e 'o ld g rowth ' ) . 4 T h e s e s t a n d s a r e c l a s s i f i e d a s t rans i t i ona l b e c a u s e t h e y c o n t a i n c h a r a c t e r i s t i c s r e s e m b l i n g bo th o l d - g r o w t h a n d m a t u r e f o r e s t s . Using forest cover inventory data, most sites were accurately identified (by age class) 3 , although two of those that were misclassified (Giveout and Glenmerry) had inventory ages that differed from field ages by hundreds of years. Wilson (2001 ) also found that inventory data was not a reliable means of identifying old-growth forests because of the high degree of error. He compared forest cover inventory data to field plots (from a Predictive Ecosystem Mapping project) in the Arrow Forest District (which occupies approximately half of the current study area) and found that forest cover data underestimated the age of stands more than 130 years old. Wong et al. (in press) also used inventory data from the Arrow Forest 3 F o r e s t C o v e r a g e c l a s s d a t a co r rec t l y p r e d i c t e d t he a g e c l a s s o f 12 out o f 16 (75%) s a m p l e s i t e s . District to estimate disturbance return intervals and found frequent errors in age estimates. However, they did not observe a bias towards over- or under-estimating ages. Regardless of bias, using the Biodiversity Guidebook definition associated with forest cover data, old growth includes all stands with an estimated mean age more than 250 years old (Ministry of Forests and Ministry of Environment, Lands and Parks 1995). As shown by the measured stand ages reported here, stands in the moist warm ICH reach ages much greater than this threshold, which is not accounted for using the Biodiversity Guidebook definition of old growth. In Table 5, 500 years was set as a second threshold to distinguish between old and very old stands, and MacDonald, Glenmerry, Giveout, Pedro, and Bremner were classified as very old. Stand age, however, can be characterized in many ways including the average age of trees in the main canopy, the maximum age encountered, or by the distribution of ages. Describing stand age is much simpler in single-cohort or even-aged stands than in uneven-aged stands that contain trees of many ages. Oliver and Larson's (1996) definition of the true old-growth phase of stand development states that stands must be comprised of trees that developed in the cohort(s) following the initial disturbance. This can be problematic in that ecosystems that experience frequent stand replacing disturbances may never achieve this development stage (e.g. Pinus conforta stands in British Columbia's central plateau). Although they may not contain true old growth, stands that are relatively old or contain relatively large structural attributes are likely to contribute unique and important habitat values. In other cases, stands may be older than the age of the oldest living trees - Goward and Arsenault (2000) call these stands 'antique'. Although it is extremely difficult to assess, it is possible that stands are older than the oldest trees at Bremner, Kuskanax, and Ski , since these sites had no signs of fire scarred trees, early serai species, or other indicators of past disturbance. Although a finer-scaled assessment of age class distributions would have allowed for better classifications of stand types, Bremner, Pedro, and Glenmerry had very clear reverse-J patterns and were defined as ancient stands using age class distributions. Giveout would have received the same classification, but the site lacked trees under 12.5 cm dbh, which may be an artifact of sampling with small subplots, since seedlings and saplings often follow irregular clumped distributions within a stand (Smith et al. 1997). MacDonald, Clearcut, Ski, Kuskanax, Proctor, and Six Mile were given moderate 'old growth' ratings using age class distributions. These ratings corresponded well with diameter distributions and P C A for Clearcut, Ski , and Six Mile, but past disturbances altered stand development trajectories at MacDonald and Proctor, creating a pattern characteristic of transitional old growth. Age class distributions may provide more information on past disturbance history than serai stage development. Age and size class distributions, however, only provide clues to stand development and serai stages. For example, normal age class distribution could be from a single large disturbance, a series of major and minor disturbances, slow colonization of a disturbed site, changes in seedling establishment due to climate change, or other causes (Lorimer 1985). Reverse-J size class distributions could either represent an even-aged stand with an accumulation of suppressed trees in the understory, or an old-growth stand with well developed canopy gaps; the size class distribution for old-growth stands could be reverse-J or relatively flat with nearly constant ingrowth over time (Parker and Peet 1984). Thus, old-growth definitions should not be based on age and size distributions alone. Partial disturbances, which include moderate-severity fire, wind, and insect damage, play an important role in determining the age and structural characteristics of moist warm ICH forests and confound simple age-based old-growth definitions. For example, there was evidence that stands such as Shields, Proctor, and MacDonald experienced moderately severe fires approximately 100, 150 and 300 years ago, but the majority of basal area was still from the pre-disturbance cohort. While these stands contained older (and larger) trees, allogenic processes played a larger role than autogenic processes, and the simplified model of stand development put forward by Oliver and Larson (1996) is not relevant. Conversely, two distinct cohorts are evident at Kuskanax, but there are no signs of fire or other major disturbance. While age and structure (PCA) definitions described Kuskanax as old growth, the presence of a younger cohort lead to the site being classified as transitional between mature and old growth (Table 5). Although unconfirmed, this pulse of younger (<200 years old), smaller (<37.5cm dbh) Tsuga trees likely developed within a treefall gap located within the sample plot. However, plot sizes and sampling intensity were too small to determine the size and nature of the disturbance leading to the gap's formation. While age and age class analyses are important components of old growth, age is not the sole attribute of an old-growth forest. Differences in site productivity and environmental conditions can lead to significant differences among stands of similar age and species composition, as was seen in comparisons between Russell (age = 344) and other sites. Productive sites like Six Mile (age = 185) often develop large-sized attributes characteristic of 'old growth' at an early age, while very old stands are often mistaken for mature or younger stands due to their simplified stand structure. For example, researchers recently 'discovered' an old-growth stand on Wachusett Mountain, an area visible from highly urbanized Boston, Massachusetts (Orwig et al. 2001). The stand in question lacks the 'stereotypic majestic' structural features normally associated with old-growth forests, yet it contains trees that approach the maximum known longevities for their species. Using age or population dynamics to define old growth, the forest on Wachusett Mountain is clearly old growth; using structural attributes, it is not. The lack of congruency between age-based and structure-based definitions of old growth creates difficulties for management. Species found within the ICH rely on old-growth forests for different reasons, and both stand age and structure are important elements of old growth. Black bears (Ursus americanus) that den in hollowed-out Thuja trees require large spaces at the base of trees, but do not necessarily require old trees (Bull et al. 1997). Mountain caribou (Rangifer tarandus caribou), an endangered species found within the study area, require forests of an old age for their primary food source Bryoria spp. and Alectoria sarmentosa lichens (Apps et al. 2001). The abundance of Bryoria is related to forest age due to substrate characteristics of old trees, and not to tree size or stand microclimate (Goward 1998). Similarly, antique stands in the ICH harbour at least 16 rare species of epiphytic cyanolichens that are only found in very old stands (Goward 1993, Goward and Arsenault 2000). Immature and managed stands have been found to reduce habitat for a range of species. Immature stands on southwestern Vancouver Island had considerably lower understory plant diversity when compared to old growth (Qian et al. 1997). The decline in plant diversity was attributed to the lack of old-growth structural features in immature stands. A study of small mammals in the Pacific Northwest found that although species composition was similar between managed, naturally young, and old-growth forests, old growth supported 1.5 times more individuals and biomass than managed forests (Carey and Johnson 1995). Similarly, size of old-growth retention areas can also impact biodiversity. A study of bird distribution in various sized old-growth patches in northern Sweden found that small fragments (<5ha) only provided habitat for generalist species, while larger patches (>10ha) had more rare species (Edenius and Sjoberg 1997). Whether defined on the basis of age or stand structure, stand development and degree of old-growthness always exist along a continuum, and each site has a unique history that has shaped its age, structure, and age class. Developing a comprehensive, conclusive, descriptive model that defines all old growth is neither possible nor practical because there are too many site-specific factors to incorporate. However, quantifying the range of old-growth conditions and their expected patterns across different site types can provide an important context for conservation decisions. The index of old-growthness presented here provides a quantitative approach for assessing old growth. Because it is based on multiple structural attributes, stands are neither included nor excluded on the basis of a single variable (Holt et al. 1999). This is particularly important because not all sites will have all old-growth associated structural attributes. The scorecard approach also allows forest managers to assess individual stands on the basis of selected attributes (see Figure 6). For example, management of old growth for a woodpecker species may require higher densities of large snags, while similar management for marten would emphasize large pieces of C W D (Bull et al. 1997). The index of old-growthness presented here is based on stand structure rather than population dynamics or stand age. Although each method of defining and describing old growth has its own merits, a structure-based approach is most appropriate in the moist warm ICH for many reasons. First, there is too much internal decay in ICH trees to assess age with confidence. Second, complex disturbance histories confound classification of stands using population dynamics. Using structure to assess old growth incorporates a range of stand types and habitat values. In addition, the index approach is a useful management tool. Total values can be used to compare general characteristics of a group of stands, while individual measured values and scores for certain attributes can be assessed in relation to specific habitat needs. For example, large diameter snag densities may be of interest for rare or endangered species management. If a stand receives a high score on the index of old-growthness for these attributes, and meets other old-growth criteria, a manager could select the site to meet conservation objectives. Using the midpoint between the means of both groups plus or minus one standard error to develop thresholds is a relatively conservative approach. As such, the threshold values in the index of old-growthness provide minimum values and are not intended to be 'absolute'. This is particularly important given the small sample size used in this study. Although there was a strong correlation between age and P C A 1 , there were no obvious distinctions between old and very old sites in the P C A analysis. A larger sample may or may not detect such a difference, but to explore this possibility, structural attributes were summarized by age category, with stands less than 250, 251 to 500 and greater than 500 years old grouped together. Older stands had higher densities of large trees and snags (>50 cm dbh), higher mean diameters, and lower densities of small trees (<30 cm dbh) and C W D pieces (Table 6), which were not identified in the P C A analysis. The distinction in structural attributes between ancient and old stands implies that management for old-growth forests in the ICH should incorporate a range of stand ages and structures in order to accommodate the range of ecological communities and the habitat needs of native species. Table 6. Mean and standard errors for selected attributes by stand age category. <250 years old 251-500 years old >500 years old Mean SE Mean SE Mean SE Mean dbh 28.2 4.1 35.6 3.3 50 2.9 Density of Trees <30cm dbh 1071 412.0 530 53.3 265 35.9 Density of Trees >50cm dbh 38 11.2 73 17.9 114 . 13.2 Density of CWD pieces 1005 194.0 1053 67.1 630 158.7 Density of Snags >50cm dbh 8 3.4 10 4.2 13 2.5 In a previous study, Holt et al. (1999) used P C A to separate sites in the ICHmw2 into 'not' old growth, 'recruitment' old growth and 'old growth'. In the 'old growth' category, they found a mean of 140 live stems per hectare greater than 50 cm dbh, while the 'recruitment' category had a mean of 29 cm. The value for 'old growth' in Holt et al.'s study was greater than that reported here for both stands over 500 years old and for 'high value' old-growth stands as identified by P C A . Quesnel (1996) used historic data to summarise the density of trees over 50 cm dbh in stands over 140-years-old. He found a mean of 77 stems per hectare, but his approach lumped mature and old forests together, which may explain the low mean value. In pre-treatment sampling of the Date Creek Silviculture Systems study in ICH forests of northwestern British Columbia, Coates et al. (1997) found that old-growth stands had lower stem densities, lower merchantable timber volumes, and higher C W D volumes than mature and younger stands. The sample sizes in this study reflect a trade-off between intensive sampling for age structures and population dynamics and extensive sampling for stand structural characteristics. A larger sample size with more stand structure plots would have captured a wider range of stand structural characteristics, particularly due to the complex nature of the ICH. The prevalence of mixed-species stands and partial disturbances creates a broad range of conditions that are modified by inter- and intraspecies competition, site productivity, anthropogenic interference, and other historic conditions. An effort was made in this chapter to summarize the range of conditions across mesic old forest sites. However, conditions will be particularly different on drier sites where Pseudotsuga, Larix and Pinus contorta are more dominant and on moister sites where Thuja is more abundant. Landscape position also plays an integral role in assessing the conservation value of old forests and may also be an important factor in determining which sites escape fire long enough to develop old-growth characteristics. This is especially true in topographically complex landscapes such as the moist warm ICH where natural fire breaks and local wind patterns can have a considerable effect on fire behavior. Additional considerations for ranking old-growth areas include: the size of the patch, position in relation to other old-growth areas, connectivity potential of the patch, and the state of the surrounding forest cover matrix. CONCLUSIONS Definitions of old growth based on stand age, age structure, and stand structure produced reasonably similar results. Although it was difficult to define old-growth forests using age structures, age and size class frequency distributions provided a general description of stand development patterns and highlighted the importance of partial disturbances in the ICH. Veteran trees were present in all but one stand less than 200-years-old, and remnant cohorts were evident in several older stands. Veteran trees and remnant cohorts alter linear models of stand development and provide varied habitat structures. The tree size and age relationships outlined here provide an indication of disturbance and stand development history, but are poor indicators of old growth when used alone. Stand structure may be the most appropriate approach to defining old growth in moist warm ICH forests because stand ages and age structures rely on age estimates from decayed trees. Conceptually, structural attribute definitions of old growth are intended to act as surrogates for functional old-growth processes (e.g. Franklin and Spies 1991, Franklin et al. 2002). In this study, age and structural attributes were highly correlated and the abundance of old forest attributes increased with stand age. Using P C A , a distinction was made between mature and old-growth forests, but differences between old and ancient stands were not detected. It may be possible to detect differences using a larger sample with a more complete range of stand ages. CHAPTER FOUR: Dynamics of Tsuga-dominated old-growth forests in southeastern British Columbia INTRODUCTION The dynamics of forest ecosystems are affected by climate, site productivity, competition among trees, species composition, and disturbances caused by insects, pathogens, wind, fire, and other factors (Oliver and Larson 1996, Antos and Parish 2002). The size, magnitude and timing of disturbances determine whether stands respond through complete canopy turnover, establishment of some new trees, or growth increases among existing trees. Many authors (e.g. Oliver and Larson 1996, Kneeshaw and Burton 1998, Wells et al. 1998, Daniels 2003) have suggested that fine-scale canopy gap disturbances are the primary processes shaping the development and dynamics of old-growth forests. Canopy gaps are an integral component of old-growth forest dynamics. Gaps result from the death of one or more overstory trees, are critical in maintaining a heterogeneous light environment at the ground level, and may also increase nutrient and moisture availability to surrounding trees (Denslow and Spies 1990). Canopy gaps have been studied extensively in tropical forests (e.g. Canham et al. 1990, Denslow and Gomez Diaz 1990), mixed-wood forests of eastern North America (e.g. Runkle 1982, Frelich and Graumlich 1994, Abrams et al. 1997, Abrams and Copenheaver 1997, Orwig et al. 2001), Douglas-fir (Pseudotsuga menzeisii Franco Mirb) forests of the Pacific Northwest (e.g. Stewart 1986, Spies and Franklin 1989, Spies et al. 1990, Gray and Spies 1996), and subalpine forests (Lertzman and Krebs 1991, Lertzman 1992, Parish et al. 1999, Antos and Parish 2002). The dynamics of Tsuga-Thuja (Tsuga heterophylla (Raf.) Sarg. - Thuja plicata (Donn ex D. Don)) forests have been studied in coastal British Columbia (Daniels and Klinka 1996, Daniels 2003) and in the northwest interior of the province (LePage 1995, Coates and Burton 1997, Coates 2000), but there are no published studies for British Columbia's southern interior. Previous studies have found a range of patterns relating to canopy gaps in coniferous forests. In Picea-Abies forests in southern British Columbia, 83% of the trees in the canopy achieved their position as a result of episodic disturbances (Parish et al. 1999). Small-scale, gap-related disturbances were less i important than openings from bark beetles that affected the entire stand. In Tsuga canadensis ((L.) Carrière) stands in eastern North America, 95% of canopy accession resulted from release events relating to canopy gaps (Zeigler 2002). Similarly, Daniels and Klinka (1996) found that old trees did not develop as a single cohort in coastal Tsuga-Thuja-Abies forests in the Vancouver area. Instead, establishment was continuous and small-scale gap-processes predominated throughout stand histories. Daniels and Klinka also found that in coastal forests, Tsuga appeared to depend on canopy gaps to advance to upper canopy positions. Thuja, however, showed signs of infrequent release, and not all Thuja trees had released to reach the canopy. This study focuses on the dynamics of Tsuga and Thuja in moist warm subzones of the Interior Cedar Hemlock Biogeoclimatic Zone in southeastern British Columbia (Braumandl and Curran 1992). Tsuga and Thuja are highly shade-tolerant species (Cameron 1996) and, following establishment, tree growth in Tsuga-dominated stands is particularly linked to inter-tree competition and disturbance. Fire is the predominant stand-replacing disturbance in British Columbia's southern interior forests (Ministry of Environment, Lands and Parks and Ministry of Forests 1995). Estimates of mean fire return intervals for the moist warm ICH subzone range from 250 years (Ministry of Environment, Lands and Parks and Ministry of Forests 1995) to between 97 and 458 years (Wong et al. in press), and old-growth forests are estimated to be as old as 700 years, although they may be much older (see Chapter 3). The frequency of large-scale disturbances, such as fire, controls the influence of small-scale disturbances, particularly in old-growth forests where intervals between major disturbances are very long (Spies et al. 1990). Where stand-replacing disturbance intervals are short (e.g. in boreal ecosystems), overstory trees are replaced before gap dynamic processes are fully established. In old-growth Tsuga-Thuja forests, canopy gap dynamics may dominate stand processes for hundreds of years and are fundamental to stand development processes for long periods between major disturbances. This is particularly important because as stands age, the influence of stand origin events fades and autogenic and partial, or medium-scale, disturbance patterns replace their effects (Antos and Parish 2002). Partial disturbances are allogenic processes that include large insect outbreaks, wind, and low to moderate severity fires. Such disturbances have a large influence on a stand, but leave pre-existing trees to dominate stand structure, composition, and ecological function (Antos and Parish 2002). Large, intense fires determine patch dynamics at landscape levels and control stand initiation processes (Pickett and White 1985), but the sizes, rates, and types of fine to moderate-scale disturbances at both stand and landscape levels are poorly understood. Because Thuja and Tsuga are highly shade tolerant (Cameron 1996), their relative growth rates are more closely linked to canopy position than climate or other factors. In this chapter, dendroecological techniques are used to determine the pattern of fine-scale disturbance in old-growth forests (Lorimer and Frelich 1989). The objectives are (1) to assess the general patterns of disturbance in old-growth ICH forests; (2) to compare the responses of Thuja and Tsuga to canopy gaps; (3) to determine whether Thuja and Tsuga depend on gaps for canopy ascension; and (4) to compare current and historic understory growing conditions. I hypothesize that both species have the ability to remain in the understory for hundreds of years in a suppressed state before advancing into the overstory through canopy gaps. The results of this research are important for understanding canopy gap dynamics and predicting the consequences of forest management practices such as partial cutting, species composition, incremental silviculture, and conservation planning. METHODS Six old-growth sites (Bremner, Giveout, Glenmerry, Kuskanax, Pedro, Ski) were selected from the broader study for analysis of tree ring records. This subset of sites was selected because, using the definitions in Chapter 3, each site exhibited old growth characteristics and did not show signs of past disturbances such as moderate severity fire. Patterns of release and suppression in the annual ring-width series were used to infer historic canopy disturbances. Releases were categorized as 'major' and 'minor' where a major release was defined by a 100% increase in growth while minor releases showed a 50-99% increase (Lorimer and Frelich 1989). Equivalent decreases in growth reflected suppression events. Releases and suppressions were further divided into sustained and temporary, with sustained release or suppression events lasting for 15 or more years and temporary releases lasting between 10 and 15 years. Growth rates were measured using digital calipers and were based on the 10 year period prior to and following a change in growth pattern. The year of change, magnitude, average growth rate, rate of change, and number of years between growth changes were recorded. With abrupt changes, the year of disturbance was based on the first noticeably different ring. The year of origin for gradual changes was estimated visually. Abrupt changes are obvious shifts in growth rates, but gradual changes occur slowly and are more difficult to isolate. Growth changes were summarized by decade and synchronous release events in multiple trees were interpreted to reflect disturbances affecting stands in the past. Heartwood decay is common in the ICH zone and it was not possible to extract complete cores from all sites. In addition, several cores broke during extraction in the field or had a number of years missing due to small pockets of decay. Breakage is particularly common in larger Tsuga trees where removing the core in sections often avoids compression within the increment borer. In most cases it was possible to fit the pieces together when mounting cores on wooden supports. Where this was not possible, or where decay was present part way through the core, radial growth analysis was limited to the intact portion of the core. Thirty cores had badly broken sections within the most recent 40 years and were excluded from ring series analysis. The final analysis was based on 266 trees, which included partial cores from 21 Thuja trees, 50 Tsuga trees, and three Larix trees as well as 15 Thuja and 28 Tsuga trees that were highly decayed and were used to assess changes in growth pattern by decade, but were limited to the decades following major breaks where dates were considered reliable. Trees were classified into upper and lower canopy positions based on their crown class as well as their diameter and age. Upper canopy trees included co-dominant and dominant trees, while the understory included suppressed and intermediate stems. Large diameter, old trees with low heights were'classed as upper canopy trees because their short stature was due to broken or dead tops rather than canopy position. Growth changes in upper canopy trees were further categorized as occurring while the tree was in a lower or upper canopy position. This was necessary to distinguish between releases in understory trees leading to canopy accession and those in overstory trees leading to canopy expansion. A lower diameter limit of 40 cm was used as a threshold for distinguishing between understory and overstory events. Because of the uneven-aged nature of the study sites, it was assumed that trees had reached the canopy by the time their diameters grew to this size (as per Lorimer and Frelich 1989). Zeigler (2002) used 35 cm dbh as a similar threshold in Tsuga canadensis stands in New York, but 40 cm was selected here because it coincides with the size of trees that are currently in the understory. The 40 cm limit is a conservative estimate for crown ascension since differentiation into crown classes is expected in even-aged stands prior to trees reaching such a large diameter. Tree diameters at the time of a growth change were estimated based on the length of each growth change and the tree radius at the time of the change. Changes in relative growth were summarized by decade for each site. The number of trees that were alive and that included reliable sections of core in each decade was calculated and growth responses were assessed as a proportion of trees with a growth change in a given decade. A minimum of five cores was necessary in a single decade to begin the record for each stand. The number of releases and suppressions were calculated by stand, species, and crown position. The approach used here assumed that tree growth rates increase in response to canopy disturbances. However, the analysis of suppression and release events was based on changes in relative growth rates rather than physiological, or actual growth rates. For example, under the criteria used here, a doubling of ring width from 0.01 mm per year to 0.02 mm per year would qualify as a major release, but the tree may still be in a physiologically suppressed state. Wright et al. (2000) quantified physiologic suppression for 11 tree species in the ICH in northwestern British Columbia and reported thresholds of 0.25 mm/year for Tsuga, and 0.3 mm/year for Thuja. In their study, physiological suppression was defined by survival potential, which was based on minimum light level tolerances under which 50% of saplings of a given species would survive for 20 years (6% of full light for Thuja; 5% for Tsuga). In this study, physiological suppression reflects very slow tree growth, but was not quantified in relation to light levels and mortality probabilities. Instead, mean growth rates were categorized as very fast, fast, moderate, slow and very slow based on quartile and extreme (upper and lower 5%) values for the average ring-widths from all periods of tree growth used in this study. RESULTS Stand level growth changes As indicated by age class distributions in Chapter 3, all six old-growth sites were uneven-aged and small-scale disturbances dominated stand dynamics. Tree ring records with at least five live trees per decade range in length from 420 years at Bremner to 330 years at Giveout. In total, there were 274 releases and 325 suppressions in 266 tree cores, including partial cores. The average number of periods of suppression and release initiations per tree varied between sites. The highest frequency of both growth patterns occurred at Giveout and the lowest at Ski . Trees at Pedro and Glenmerry also showed signs of frequent suppression and release; however, sample sizes for trees without decay were very low for Glenmerry'(7), Giveout (8), and Pedro (13). Bremner, Kuskanax, and Ski had larger samples (n = 19, 21, 26, respectively), but only three of the trees from Kuskanax were in the overstory, while the remaining 86% were in the lower canopy. In general, upper canopy trees at Kuskanax had fewer release events in the understory than trees at other sites (0.33 releases compared to the overall average of 1.34), while lower canopy trees at Pedro had more releases and suppressions than other sites (average of 2.14 releases and suppressions compared to the overall average of 0.84 and 1.34, respectively). These patterns are congruent with the frequency of stand level responses at both Kuskanax, where none were observed, and Pedro, where four were recorded (Figure 1). Figure 1 shows the number of growth changes in each decade, at each site. These changes reflect both canopy accession by understory trees and canopy expansion by overstory trees. Individual releases were interpreted to reflect small gaps caused by the death of a single tree, the toppling of a snag, or damage to large branches or adjacent treetops. Suppression may be in response to overtopping by adjacent trees or damage to a small number of trees from an adjacent treefall or other disturbance. Synchronous growth increases in five or more trees were considered stand-level responses and were interpreted to reflect partial disturbances, such as windthrow, insect damage, or moderate-intensity fire. For example, several periods of individual-tree and stand-level suppression and release are evident in Figure 1 throughout the growth history of all stands. A minimum of five trees per stand with synchronous 0.5 s 0 4 $ 0.2 * 0., 0 -0.1 | ^ , j> -0.3 8 -0.4 <" -0.5 Bremner Giveout u T I P ' N= 5 5 6 6 6 6 8 11 12 1 4 16 23 26 30 34 35 35 36 36 36 36 1700 1800 Decade 1700 1800 Decade 0.4 S3 œ 0.2 (2 0.1 0 •0.1 1> -0.2 I -0.3 I -0.4 <r> -0.5 Glenmerry 16 18 19 20 21 25 29 34 37 39 39 39 39 0.5 s 0 4 $ 0.3 I 0.2 0.1 0 & -03 §• -0.4 V) -0.5 Kuskanax N = 5 5 7 7 7 9 9 10 10 12 17 26 35 : Decade Decade Pedro N= 6 7 8 8 9 10 11 15 23 26 28 30 32 33 33 33 33 1700 1800 Decade Ski N ° 5 6 6 6 6 8 11 15 21 26 3 2 3 6 37 37 3 7 37 Decade Figure 1. Proportion of trees experiencing suppression and release by decade. Sample sizes are recorded along the bottom of each graph and long arrows indicate the start of a reliable tree ring record. Short arrows reflect growth changes affecting at least 20% of a stand, including a minimum five trees. Solid bars represent major growth changes, while thatched bars reflect minor changes. growth changes was selected in an attempt to separate partial disturbances from responses to canopy gaps within or in close proximity to the sample plot. Note that a high proportion of trees at all sites show signs of growth changes in their early years. These growth changes may reflect responses to canopy gaps, but the sample sizes are too small to infer causality. When major and minor growth changes were combined for all trees, stand level releases were evident in all stands except Kuskanax. At Kuskanax, several growth changes are apparent in the early tree ring record, but they only affected up to three trees. At Bremner, eight of 30 trees released during the 1840's. Three trees also became suppressed during this period, while another three trees released within five years of the 1840's. Synchronous releases such as this may result from increased resource availability after the death of one or more dominant canopy trees (Abrams et al. 1999). An assessment of the spatial distribution of trees as well as more extensive sampling across larger areas of individual stands are required to assess the full extent of synchronous growth changes. Similar large releases were observed in 1940 in Glenmerry, 1870 in Giveout, 1880 in Pedro, and 1870 at Ski . Pedro experienced the most periods of suppression with declining growth rates in six trees (of 26 and 28) in the 1850's and 1860's, as well as 10 trees (of 33) in the 1980's. At least one period of major suppression initiation occurred at all sites except Bremner, Kuskanax and Ski. Individual species growth responses Growth patterns were summarized for 28 Thuja and 123 Tsuga (n = 152; Table 1 and Table 2), although tree cores from 37 Tsuga and 20 Thuja trees were too short (due to decay) to assess the number of growth changes over time. Upper canopy Thuja trees without decay were rare (n = 1 ), so assessments were limited to overstory releases and were made using incomplete tree cores (n=16). Tsuga trees with decay were not included in assessments of suppression and release frequency. Nine of 11 Thuja trees currently in the lower canopy experienced one or more releases prior to sampling (Table 1 ). Eight of 11 understory Thuja trees sampled showed signs of suppression periods. The high percentage of releases and suppressions in the lower canopy tree ring record suggests that Thuja relies on multiple gaps for canopy ascension. Unfortunately, comparisons between current lower canopy Thuja and historic understory conditions of upper canopy trees could not be made because virtually all upper canopy Thuja had decayed, and provided incomplete tree cores. However, 82% of the Thuja trees currently in the upper canopy showed increases in growth while in the upper canopy, and 94% of the same trees had declining growth. The frequency of releases in the upper canopy suggests that Thuja continues to respond to canopy gaps, although these growth increases likely reflect canopy expansion rather than ascension. The frequency of declining growth in upper canopy trees likely reflects, in part, growth allotment to areas other than diameter increases. As trees grow larger, they tend to show a geometric decline in ring-width that compensates for their increase in overall diameter (Oliver and Larson 1996). Table 1. Precentage of understory growth changes in lower canopy Thuja trees and overstory changes in upper canopy Thuja trees. Cores with decay are included in summaries of upper canopy trees, but not for lower canopy trees. Lower Canopy Trees (n = 11) Upper Canopy Trees (n = 17) # Growth Understory Understory Overstory Overstory Changes Releases Suppressions Releases Suppressions 0 18 27 18 6 1 45 45 12 24 2 36 18 59 53 3+ 0 9 12 18 Tsuga trees also appear to rely on canopy gaps and subsequent releases to reach the upper canopy. For example, 49% of the lower canopy Tsuga sampled did not show any signs of release, although 77% of the upper canopy trees experienced at least one release while in the understory (Table 2). Of the 26 lower canopy trees that had never released, 76% showed signs of suppression initiations at least once in their growth history. Multiple understory release events were observed in 40% of the upper canopy Tsuga, but only 19% of the lower canopy trees. In addition, t-tests resulted in significant differences between the number of understory suppressions and releases in upper and lower canopy trees. The mean number of understory releases for lower canopy trees was 0.77, while upper canopy trees experienced an average of 1.34 releases (p = 0.019; df = 79). Conversely, the mean number of understory suppressions for lower and upper canopy trees was 1.3 and 0.84, respectively (p = 0.012; df = 79). This implies that additional gap events will be necessary for current lower canopy Tsuga trees to reach the overstory. The frequency of understory suppression events in lower and upper canopy Tsuga trees suggests that understory conditions have changed over time. Upper canopy trees experienced more releases and fewer suppressions in their early years. In contrast, most lower canopy trees (86%) experienced at least one suppression event, and 30% experienced more than one period of suppression. This suggests two possibilities. First, conditions in the lower canopy may be more suppressed now than when the current upper-canopy trees were in the understory, and second, those trees that were very suppressed when current overstory trees were in the understory have long since died and were lost from the tree-ring record; surviving trees are limited to those that experienced few suppression periods and were able to out-compete neighbouring trees for light and other resources. This indicates that, although canopy gaps were important for canopy ascension for upper-canopy trees, they are an even more important process for current understory trees, and multiple releases are required for Tsuga canopy ascension. Table 2. Percentage of Tsuga trees in upper and lower canopy positions that experienced understory and overstory release and suppression events. Upper and lower canopy refer to current crown position; understory and overstory events are separated by a 40 cm dbh threshold. Cores with decay are not included. Number of Lower Canopy Trees (n = 43) Upper Canopy Trees (n = 38) Upper Canopy Trees (n = 38) Growth Changes Understory Releases Understory Suppressions Understory Releases Understory Suppressions Overstory Releases Overstory Suppressions 0 49 14 24 39 68 55 1 33 56 37 39 21 34 2 12 19 24 18 8 5 3+ 7 11 16 3 3 5 Once Tsuga trees reach the upper canopy, growth rates appear reasonably consistent. Sixty-eight percent of all Tsuga sampled had not experienced overstory releases (canopy expansion), while 55% did not show signs of overstory suppression. On average, Thuja trees experienced considerably more suppression and release in the overstory than Tsuga, although the relative frequency of suppressions in the lower canopy stratum was similar for both species (Table 1 and Table 2). The number of major (52%) and minor (48%) releases among Tsuga upper canopy trees was approximately even, but 75% of the releases in lower canopy trees were major. The opposite trend was apparent for suppression of Tsuga trees. Most suppression periods in upper canopy Tsuga trees reflected minor growth changes (57%), but 68% of all lower canopy suppressions were major. Ninety-three percent of release events and 83% of suppressions lasted for 15 or more years. Among upper canopy trees, 57% of releases were abrupt, but 61% of suppressions were gradual, which may reflect age-related trends. For lower canopy trees, 72% of all releases and 54% of all suppressions were abrupt, which likely reflects responses to canopy gaps and minor disturbances. Similar patterns were observed for Thuja. Sixty-six percent of all releases in upper canopy trees were major, but 61% of all suppressions were minor. Among lower canopy trees, 60% of releases and 71% of suppressions were major. 88% of all releases and 91 % of all suppressions were sustained for at least 15 years. For upper canopy trees, 61% of all releases were abrupt and 52% of all suppressions were gradual. 76% of lower canopy releases and 52% of lower canopy suppressions were abrupt. Physiologic growth rates The above results relate to relative changes in growth rate, but do not reflect actual, physiologic tree growth. Thresholds for very slow, slow, moderate, fast, and very fast Tsuga growth are presented in Table 3. Trees are considered suppressed if their growth rates are less than 0.3 mm/year (i.e. slow or very slow growth). The 0.3 mm/year threshold is similar, but greater than Wright et al.'s (2000) threshold of 0.25 mm/year for Tsuga in northern British Columbia, although Wright et al. used light levels to determine their thresholds rather than sampled growth rates. Table 3. Physiological growth rate categories for Tsuga. Growth Rate Category Thresholds for growth rates Very Fast >1.3 mm / year Fast 0.9-1.3 mm/year Moderate 0.3-0.8 mm / year Slow 0.1-0.3 mm / year Very Slow <0.1 mm/year Growth rates of Tsuga were relatively similar among sample sites, but varied by lower and upper canopy trees (Table 4). All sites had 'moderate' mean growth rates for all trees combined, but lower canopy trees at Bremner had borderline 'slow' growth, while upper canopy tree growth at Pedro was 'fast'. Trees with 'very slow' growth were rare at all sites, and did not occur at Ski , Pedro, or Giveout, although trees with periods of growth at the threshold were found at Pedro and Giveout. On all sites except Kuskanax, upper canopy trees had faster maximum and mean growth rates than lower canopy trees, which is consistent with the earlier suggestion that trees in lower canopy positions are more suppressed now than they would have been in the past. However, at Kuskanax it appears that current understory growth rates are higher than historic rates. The highest average growth rate for all trees was measured at Giveout, where the fastest period of growth was found on an upper canopy tree. Ironically, the slowest growing tree was found in the understory at Giveout, which is indicative of variable growing conditions across both time and space. The slowest average growth for trees of both canopy positions was found at Bremner. When relative growth changes were examined in relation to physiological growth rates, most releases produced fast growth (50%), very fast growth (25%) or moderate growth (22%). Only one of 274 release events led to very slow growth. Similarly, most suppression events produced moderate (53%) or slow growth (36%). Only five of 325 growth reductions resulted in very fast growth, while seven of 325 suppressions led to very slow growth. Table 4. Growth rates (mm / year) for Tsuga in the upper and lower canopy. Site Name All Lower Upper Lower Upper Lower Tsuga Canopy Trees Canopy Trees Canopy Canopy Canopy Upper Canopy Mean Min Max Bremner 0.56 0.36 0.62 0.11 0.09 1.15 1.75 Giveout 0.61 0.48 0.64 0.08 0.10 1.10 2.08 Glenmerry 0.51 0.41 0.53 0.10 0.09 0.97 2.02 Kuskanax 0.55 0.58 0.53 0.11 0.09 1.14 1.19 Pedro 0.49 0.48 0.49 0.13 0.10 1.12 1.29 Ski 0.58 0.46 0.66 0.11 0.14 1.26 1.28 All Sites 0.55 0.47 0.57 0.08 0.09 1.26 2.08 Growth patterns also varied for individual trees. Only seven trees, all in lower canopy positions, had neither released nor become suppressed, and of these, only two were slow growing, while four had fast or very fast growth. In general, wide fluctuations from decade to decade and across centuries were common in the old-growth stands sampled in this study (see Figure 2). For example, tree number 101 at Bremner spent its first 200 years in a suppressed state before undergoing periods of major release, followed by suppression for the past 200 years. Tree number 21 at Giveout had a similarly slow start, but growth rates dropped after 250 years and the tree entered a period of severe suppression before growth rates increased in the late 1880s. Tree number 46 at Pedro and 41 at Ski had moderate early growth followed by fast growth over the past 100 to 150 years. These examples show the variability in growth that is common within individual Tsuga trees in old-growth ICH stands. Bremner #101 Pedro #46 1800 Decade Figure 2. Average yearly growth rate for three upper canopy Tsuga (Bremner #101, Pedro #46, and Ski #41) and one lower canopy Tsuga (Giveout #26), summarized by decade. DISCUSSION Oliver and Larson (1996) categorize stand development into four stages: stand initiation, stem exclusion, understory re-initiation, and old growth. True old growth is defined by an overstory of trees that developed beneath a canopy in the absence of allogenic processes. Stands with a mixture of overstory trees that established immediately after a disturbance and new canopy trees that established beneath the first canopy trees are referred to as transitional old growth. As seen from assessments of stand structure (see Chapter 3), the sites sampled here represent some of the oldest and most structurally diverse stands in the moist warm ICH. However, using Oliver and Larson's (1996) criteria for old growth, the presence of trees that established in response to the most recent stand-replacing disturbance places all sites into the transitional category. Unfortunately, few of the tree ring records acquired in this study extend back to the estimated time of stand origins. Small sample sizes and extensive internal decay have eliminated evidence of the stand initiation stage, as well as a significant portion of the stem exclusion stage in most stands. In addition, evidence of stand-origin (i.e. charcoal, fire-scarred trees) is sparse for the old-growth forests sampled here, but old, live Larix trees and snags at Giveout, Glenmerry, Ski and Pedro are good indicators of historic stand-replacing disturbances such as fire. Larix is shade intolerant and requires mineral soil or burnt forest floor for germination (Klinka et al. 1998), which also makes it a good indicator of stand-replacing disturbance. No Larix were sampled at Bremner where the oldest tree was an approximately 630 year old Tsuga. If this tree originated as part of an even-aged cohort following a stand-replacing disturbance, then stand origin occurred in 1370. However, the tree ring record for Bremner only includes five reliable tree cores beginning in 1580. Because there are few trees with such long tree ring records, at minimum, the first 210 years of stand development are missing from the tree ring record. Thus, the portion of the tree ring record presented in Figure 1 for Bremner likely represents the later stages of stand development. Patterns in Figure 1 show one stand-level response where 20% of living trees either released or became suppressed in 1840, along with several smaller responses, particularly since 1800. Dendrochronological methods are often used in studies of past climate change. However, no patterns of synchronous growth changes were found among the six old-growth stands in this study. This is not surprising, considering the shade tolerance of Thuja and Tsuga. Analysis of individual rings may show some patterns associated with regional climates, but relative crown position is more likely to drive ring width and growth patterns with these species. In a 135 year old fire origin stand in a moist cold ICH forest in northwestern British Columbia, LePage (1995) concluded that stand development patterns were most influenced by individual height growth characteristics and inter-tree competition. Overall tree growth was site specific and different mixes of species or different development histories led to different stand dynamics. Canopy trees provide the most information on stand dynamics because they exert the largest influence on the current, past, and future stand by dominating stand structure, controlling understory microclimates, and providing seed sources (Antos and Parish 2002). Current upper canopy trees frequently required more than one release event in the understory to reach their current dominant or codominant position. Multiple releases were observed in 40% of the upper canopy Tsuga, and were most common at Bremner and Giveout. Many of the upper canopy trees with multiple understory releases are older than the mean stand age, and may have established with the initial cohort. Because release events were recorded based on changes in relative growth, almost half of the trees with multiple releases experienced a mixture of moderate and major release events successively (i.e. in a step-wise pattern). Although they are recorded as multiple events, it is likely that many actually reflect long lag times in response to an opening or expansion of existing openings (e.g. expansion of root rot centres, death of trees damaged as neighbouring trees fell). Individual tree responses are critical to the effects of multiple and/or single canopy ascension events. Wright et al. (2000) found no effect from previous periods of suppression on growth rates when shade-tolerant saplings were exposed to light. After 20 years of suppression, Tsuga is predicted to have the fastest growth rate at all light levels when released, although Thuja is also expected to respond positively to openings from partial cutting or natural disturbances (Wright et al. 2000). Wright et al.'s (2000) results are based on physiological growth rates from logged sites and are consistent with the patterns found in the natural stands in this study. Recovery from severe suppression lasting over 100 years was relatively common among the older Tsuga trees sampled in the current study. For example, a 309 year old Tsuga at Giveout that established in 1691 became suppressed in 1763 and experienced an average of 0.08 mm of growth per year for 123 years. When it released in 1886, growth over the following 27 years averaged 0.40 mm/year, a five-fold increase (see Figure 2 for more examples). Methodological limitations The methods used here rely on several assumptions. First, past events are inferred. For example, variations in local climate can cause changes in tree ring patterns that may be confused with gap dynamics. Lorimer and Frelich (1989) report that severe droughts usually cause five or six years of narrow rings and suggest that growth events relating to canopy gaps be limited to at least 10 year periods. Under this approach, temporary changes in growth are not attributed to internal stand dynamics. This study only included analysis of suppression and release; trees that originated in gaps were not quantified with releases as has been done elsewhere (e.g., Lorimer and Frelich 1989, Parish and Antos 2002). In Chapter 3, age class distributions were used to assess the number of cohorts in a stand and to infer the frequency of past disturbances. However, age class distributions do not accurately reflect disturbance events, particularly when tree cores are extracted at breast height. Breast height cores are convenient to extract and may capture less decay than ground height cores; however, they do not reflect the dynamics of seedlings and saplings. For example, five tree core records from Kuskanax began in the 1870's or 1880's and had rapid early growth. Although there is no correspondence between these dates and the proportion of trees with releases in Figure 1, if the average time for growth to breast height of 25 years (from Chapter 2) is incorporated into the origin dates, these trees likely established during the 1850's when a stand level growth response was detected (see Figure 1). Thus, the rapid growth rates at the centre of the tree reflect conditions at the time that trees reached breast height, and are not related to conditions at the time of tree origin. There are also uncertainties surrounding recent growth of older trees. Many growth increases and reductions occurred once trees had reached the upper canopy (Table 1 and Table 2). However, it is difficult to attribute these growth changes to shifts in competitive positioning, because trees would have been well established in the overstory once growth changes were initiated. To account for this, growth responses were categorized as occurring while trees were in under- and overstory positions, with the assumption that overstory 'suppressions' were related to senescence, while 'releases' were attributed to canopy expansion within gaps. As trees age, the allocation of their resources shifts from growth to maintenance of biomass and reproductive parts (Oliver and Larson 1996). In addition, the basal area increment added each year would have to increase if ring widths are to remain constant (Visser 1995, Smith et al. 1997). These processes result in smaller tree rings with age, which could be misinterpreted to be suppression events. Trees also become more stressed as they age, and declines in the growth of older, larger trees are often due to pathogen attacks, stem breakage, or other forms of tree damage. Despite these expected patterns, the growth rate of many older trees fluctuated widely, with several periods of increased growth followed by growth reductions. For example, a 583 year old Larix tree at Giveout experienced a major growth decline lasting 23 years at age 253, which was followed by a 70 year release period, a second major growth decline at age 346, and a major release at age 506. This pattern of fluctuating growth was observed for all species on all sites (e.g. Figure 2), and is more likely indicative of changes in canopy structure than age-related growth. Increased growth on overstory trees likely manifests itself as canopy expansions rather than canopy accession. Canopy expansion occurs through lateral branch growth of existing trees. Spies et al. (1990) speculate that horizontal branch growth occupies more gap space in 'well-stocked' young stands than in later stages of stand development when trees have reached their maximum sizes and stands are more widely spaced. Young trees are more able to fill openings by expanding in a given direction than large trees, because as trees grow, respiratory demands increase and competition between growth centres intensifies in larger trees (Oliver and Larson 1996). This was not the case in this study, in which every overstory Thuja tree studied experienced at least one growth increase or decline after reaching the canopy. Even after reaching the upper canopy, overstory Thuja went through an average of 1.63 periods of increased growth and 2.06 periods of reduced growth. Tsuga trees had fewer canopy expansions in the overstory with an average of 0.74 increases and 1.0 reductions in growth following canopy ascension. Interpretation of growth changes on all cores could also be a function of the methodology used, rather than actual responses to growing conditions. Assessments of growth history were based on a single core per tree, but growth rings are rarely circular, and their widths often vary widely around the stem. This can have large impacts on interpretations of growth rates since a core taken from one side may show signs of a large release while another core taken from the opposite side of the same tree could show a minor release, a major suppression, or no change during the same period. Disks or multiple cores would improve accuracy, but are more destructive and time consuming to collect. Even with disks, trees do not necessarily form rings at all heights of the stem each year. It is common for suppressed trees to miss rings, particularly where irregular or lobate diameter growth occurs. Duncan (1989) reports that missing rings can constitute up to 10% of the total number of rings present where suppression is common, and that cores taken from the longest radii in irregular diameter trees generally have the fewest missing rings and the most accurate tree ring records. Summaries of releases and suppressions by decade were used rather than yearly information to account for this problem, although cross-dating cores would have increased the precision of yearly measurements. Management Applications Despite these methodological concerns, the data presented here clearly show that gaps are an important mechanism for canopy ascension among Thuja and Tsuga trees in moist warm ICH forests. This understanding is important in developing and implementing silviculture systems that reflect natural processes. For example, Coates and Burton (1997) report that treefall gaps are an 'essential component of stand and landscape level heterogeneity', whether naturally occurring or developed through harvesting, and they recommend using canopy gaps as a framework for understanding responses to a range of silviculture practices including partial cutting. Several recommendations have been put forth outlining specific strategies and rationales for partial cutting in ICH forests. Many of these relate to habitat conservation for mountain caribou, which is listed as an endangered species (e.g. Stevenson 1986, Hamilton 1997, Stevenson et al. 2001, Waters and Delong 2001). Studies of western hemlock looper (Lambdina fiscellaria lugubrosa) and its influence on stand dynamics have also resulted in recommendations for partial-cutting and variable retention silviculture as a means of operationalizing ecosystem management (Hoggett 2000). Despite these suggestions, very little partial cutting is occurring in the ICH. Concerns regarding the feasibility of partial cutting have focused on post-harvest risk of windthrow, increases in Armillaria root decay, and poor quality of advanced regeneration. Using interim results, Waters and Quesnel (2001 ) concluded that small patch cuts (1-2ha) in wet cool ICH forests north of Revelstoke do not result in increased windthrow, and that windthrow varies from year to year. They also found no statistical differences between blowdown in partially cut and unmanaged old forests. Similarly, Coates et al. (1997) found that windthrow was not a significant concern in a partially cut ICH stand in northwestern British Columbia. In general, windthrow increases when more than 50% of volume or basal area is removed, or when openings exceed two tree heights in size (Stevenson et al. 2001). Concern regarding increased damage and mortality from Armillaria has also been expressed. However, in two ICH stands in the Nelson Forest Region (ICHmw2 and ICHmkl) , DeLong et al. (2000) found that mortality and growth of seedlings were not affected three years after push-over logging in a shelterwood prescription. Partial disturbances from wind, low-severity fire, or insect infestations create complex stand development patterns whereby stands in later serai stages develop a mixture of early and late serai characteristics (Oliver and Larson 1996). Partial disturbances, in contrast to gap dynamics, create larger openings than gap dynamics, and affect a larger proportion of the stand at any given time. Antos and Parish (2002) found that autogenic processes were more important for establishment of Abies in Picea-Abies stands, while partial disturbance for agents such as bark beetles was key for canopy ascension. In contrast, they note that both establishment and canopy ascension were closely linked to partial disturbance for Picea. This study was limited to old-growth stands that had not experienced stand-level partial disturbances from wind, fire, insects, or diseases. However, partial disturbances are both common and important for stand dynamics in moist warm ICH forests (Boisvenue 1999). An additional two stands with significant partial disturbances were sampled as part of the larger study presented here. Although the data are not shown, low intensity fires at MacDonald and Proctor (see Chapter 3) led to major releases affecting 50% of the trees sampled over two or three decades as well as high densities of saplings and/or small trees in openings. While the current study shows that Thuja and Tsuga do not depend on large partial disturbance for canopy ascension, the importance of partial disturbance in the dynamics of moist warm ICH stands should not be ignored. CONCLUSIONS Small-scale disturbances that create canopy gaps are the dominant disturbance in the old-growth forests studied here. Tree ring records showed no sign of large partial disturbances from agents like fire, wind, insects, or disease. Although tree ring records extended for 330 to 420 years, stand-level responses indicative of larger disturbances were rarely detected. However, stand dynamics are highly related to individual site histories and it is likely that partial disturbance is common in the development of other old-growth stands in the ICH. Future research is required to assess the scale, frequency, and impacts on stand structure from partial disturbance. Thuja and Tsuga clearly rely on gaps for canopy ascension. The majority of canopy trees with complete tree ring records experienced at least one period of release prior to advancing to the upper canopy. Conditions in the current understory appear to be more suppressed than in the past. Current lower canopy trees have undergone more growth reductions and fewer increases than did the trees in the upper canopy while they were in understory positions. Upper canopy trees also have higher mean growth rates than lower canopy trees. t The variety observed among the six stands sampled here reflects the multiple pathways through which old-growth forests develop in the moist warm ICH. Disturbances of many different sizes, intensities and frequencies shape the structural characteristics of old-growth forests as well as the population dynamics, growth histories and serai stage development. Canopy gaps are an important part of this process. Bibliography Abrams, M.D. and C.A. Copenheaver. 1997. Temporal variation in species recruitment and dendroecology of an old-growth white oak forest in the Virginia Piedmont, USA. Forest Ecology and Management 124: 275-284. Abrams, M.D., C.A. Copenheaver, K. Terazawa, K. Umeki, M. Takiya, and N. Akashi. 1999. A 370-year dendroecological history of an old-growth Abies-Acer-Quercus forest in Hokkaido, Northern, Japan. Canadian Journal of Forestry Research 29: 1891-1899. Abrams, M.D., D.A. Orwig, and M.J. Dockry. 1997. Denroecology and successional status of two contrasting old-growth oak forests in the Blue Ridge Mountains, U.S.A. Canadian Journal of Forest Research 27: 994-1002. Agée, J.K. and M.H. Huff. 1986. The care and feeding of increment borers. Report # C P S U / U W 86-3, National Parks Service and University of Washington, Seattle, Washington. Antos, J . A., and R. Parish. 2002. Dynamics of an old-growth, fire-initiated subalpine forest in southern interior British Columbia: tree size, age, and spatial structure. Canadian Journal of Forest Research 32: 1935-1946. Applequist, M.B. 1958. A simple pith locator for use with off-center increment cores. Journal of Forestry 56: 141. Apps, C. D., B.N. McLellan, T.A. Kinley, and J .P Flaa. 2001. Scale-dependant habitat selection by mountain caribou, Columbia Mountains, British Columbia. Journal of Wildlife Management 65: 65-77. Armleder, H.M., S.K. Stevenson and S.D. Walker. 1992. Estimating abundance of arboreal forage lichens. Land Management Handbook Field Guide Insert 7. BC Ministry of Forests, Research Program. Boisvenue, C. 1999. Early height growth and regeneration: applicability of Prognosis components to the southern interior of British Columbia. MSc Thesis, Department of Forestry, U B C , Vancouver, BC . 65 p. Braumandl, T.F. and M.P. Curran. 1992. A field guide for site identification and interpretation for the Nelson Forest Region. British Columbia Ministry of Forests Research Branch, Victoria, B.C., Land Management Handbook. No 20. Bull, E.L., C . G . Parks, and T.R. Torgersen. 1997. Trees and logs important to wildlife in the interior Columbia River Basin. U S D A Forest Service General Technical Report PNW-GTR-391 . 55 p. Bunnell, F.L. and L.L. Kremsater. 1991. Sustaining wildlife in managed forests. Northwest Environmental Journal 6: 243-269. Cameron, I. 1996. The structure of mixed-species stands in the ICHmw subzones. Available at: http://www.for.gov.bc.ca/kamloops/research/dynamics/index.htm Canadian Climate and Water Information. 2003. Available at: http://www.msc.ec.gc.ca/climate/climate_normals_1990/index_e.cfm Canham, C D . , J .S . Denslow, W.J . Piatt, J .R. Runkle, T.A. Spies, and P.S. White. 1990. Light regimes beneath closed canopies and tree-fall gaps in temperate and tropical forests. Canadian Journal of Forest Research 20: 620-631. Carey, A .B . and M.L. Johnson. 1995. Small mammals in managed, naturally young, and old-growth forests. Ecological Applications 5(2):336-352. Coates, D., A. Banner, J . Douglas, J .D. Stenenton, P. LePage, and P. Bartemucci. 1997. The Date Creek silviculture systems study in the Interior Cedar-Hemlock Forests of Northwestern British Columbia: Overview and treatment summaries. BC Ministry of Forests, Land Management Handbook #38, Victoria, B C . Coates, K. D., and P. J . Burton. 1997. A gap-based approach for development of silvicultural systems to address ecosystem management objectives. Forest Ecology and Management 99: 337-354. Coates, K.D. 2000. Conifer seedling response to northern temperate forest gaps. Forest Ecology and Management 127: 249-269. Coates, K.D. and P .J . Burton. 1999. Growth of planted tree seedlings in response to ambient light levels in northwestern interior cedar-hemlock forests of British Columbia. Canadian Journal of Forest Research 29: 1374-1382. Daniels, L.D. 1994. Structure and regeneration of old-growth Thuja plicata stands near Vancouver, British Columbia. Masters thesis, University of British Columbia. 80 p. Daniels, L.D. 2003. Western redcedar population dynamics in old-growth forests: Contrasting ecological paradigms using tree rings. The Forestry Chronicle 79: 517-530. Daniels, L.D. and K. Klinka. 1996. The dynamics of old-growth Thuja-Tsuga forests near Vancouver, British Columbia. In J .S . Dean, D.M. Meko. and T.W. Swetnam, (eds). Tree Rings, environment and humanity, conference of the Tree-Ring Society, pp. 379-393. May 1994. Tucson, Arizona. Daniels, L.D., P.L. Marshall, R.E. Carter, and K. Klinka. 1995. Age structure of Thuja plicata in the tree layer of old-growth stands near Vancouver, British Columbia. Northwest Science 69(3):175-183. DeLong, D. 1997. A retrospective investigation of advanced western redcedar regeneration in the ICHwk l , ICHmw2, and ICHmwl of the Nelson Forest Region. British Columbia Ministry of Forests Research Branch. Working Paper 25. DeLong, D., P. Comeau, C. Prescott. 2000. Shelterwood harvesting in root disease infected stands in southeastern British Columbia: Three year results EP-1186. Extension Note 051, Nelson Forest Region, Nelson, B C . DeLong, D., P. Comeau, I. Cameron, D. Simpson, R. Adams, E. Fanjoy. 2000. How light levels affect tree seedling performance in the Interior Cedar-Hemlock zone: Preliminary Results EP-1191, Extension Note 050, Nelson Forest Region, Nelson, BC . Denslow, J .S . and A . E . Gomez Diaz. 1990. Seed rain to treefall gaps in a neotropical rain forest. Canadian Journal of Forest Research 20: 642-648. Denslow, J .S . and T.A. Spies. 1990. Canopy gaps in forest ecosystems: an introduction. Canadian Journal of Forest Research 20: 619. Duncan, R.P. 1989. An evaluation of errors in tree age estimates based on increment cores in Kahikatea (Dacrycarpus dacryioides). New Zealand Natural Sciences 16: 31-37. Duncan, R.P., and G.H. Stewart. 1991. A temporal and spatial analysis of tree age distributions. Canadian Journal of Forestry Research 21:1703-1710. Edenius, L. and K. Sjoberg. 1997. Distribution of birds in natural landscape mosaics of old-growth forests in northern Sweden: relations to habitat area and landscape context. Ecography 20: 425-431. Franklin, J .F. and T.A. Spies. 1991. The structure of natural young, mature and old-growth Douglas-fir forests in Oregon and Washington. In L.F. Rigeer, K.B. Aubry, A .B . Carey, and M.H. Huff. (Technical Coordinators). Wildlife and Vegetation of Unmanaged Douglas-fir Forests, pp. 91-109. U S D A Forest Service General Technical Report PNW-285, Portland, Oregon. Franklin, J .F. , T.A. Spies, R. Van Pelt, A .B . Carey, D.A. Thornburgh, D.R. Berg, D.B. Lindehmayer, M.E. Harmon, W.S. Keeton, D.C. Shaw, K. Bible, and J . Chen. 2002. Disturbances and structural development of natural forest ecosystems with silvicultural implications, using Douglas-fir forests as an example. Forest Ecology and Management 155: 399-423. Frelich, L.E. and L.J. Graumlich. 1994. Age-class distribution and spatial patterns in an old-growth hemlock-hardwood forest. Canadian Journal of Forest Research 24:1939-1947. Froese, R.E. 2000. Modelling juvenile height in mixed species, even aged Interior Thuja-Tsuga stands.MSc Thesis, Department of Forestry, U B C . 93 p. Goward, T. 1993. Crown of the ICH: epiphytic macrolichens of old growth forests in the interior cedar-hemlock zone. Bioline 11: 15-17. Goward, T. 1998. Observations on the Ecology of the Lichen Genus Bryoria in High Elevation Conifer Forests. Canadian Field-Naturalist 112: 496-501. Goward, T. and A. Arsenault. 2000. Inland old-growth rain forests: Safe havens for rare lichens? In L.M. Darling, (Ed.) Proceedings of a conference on the biology and management of species and habitats at risk. pp. 759-766. Kamloops, British Columbia, February 1999. Volume Two. B C Ministry of Environment, Lands and Parks and University College of the Caribou, Kamloops, BC . Gray, A .N . and T.A. Spies. 1996. Gap size, within-gap position and canopy structure effects on conifer seedling establishment. Journal of Ecology 84: 635-645. Hamilton, D. 1997. Billings/Bennett Creeks Caribou Project. Phase I: Literature review. A report prepared by Nanuq Consulting Limited for the Arrow Forest District, Castlegar, B C . 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, K.W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research 15: 133-302. Hayward, G.D. 1991. Using population biology to define old-growth forests. Wildlife Society Bulletin 19: 111-116. Hoggett, A. 2000. Western Hemlock Looper and forest disturbance in the ICHwk3 of the Robson Valley -Stage 2: The effects of Western Hemlock Looper and silviculture recommendations. In E F M P P Information Session Proceedings, March 30-31, McBride, B.C. Holt, R.F., D.J. MacKillop, and T.F. Braumandl. 2001. Defining old-growth forests in the ICHwkl B E C variant in the Nelson Forest Region. Unpublished report prepared for Revelstoke Community Forest Corporation, Downie Street Sawmills, and Pope and Talbot. 47 p. Holt, R.F., T.F. Braumandl and'D.J. Mackillop. 1999. An index of old-growthness for two B E C variants in the Nelson Forest Region. Prepared for L U C O , M O E L P , Nelson Forest Region. Hunter, M.L. and P. White. 1997. Ecological thresholds and the definition of old-growth forest stands. Natural Areas Journal 17: 292-296. Ishikawa, Y. , P. Krestov, and K. Namikawa. 1999. Disturbance history and tree establishment in old-growth Pinus tora/ens/'s-hardwood forests in the Russian Far East. Journal of Vegetation Science 10:439-448. Ketcheson, M.V., T. Dool, L. Bradley, and G. Kernaghan. 2002. Kootenay Lake Forest District Predictive Ecosystem Mapping (PEM) report. Unpublished report for Kootenay Lake Forest District, Nelson, B C . Klinka, K., J . Worall, L. Skoda, P. Varga, and V . J . Krajina. 1999. The distribution and synopsis of ecological and silvical characteristics of tree species of British Columbia's forests. Canadian Cartographs Ltd., Coquitlam, B C . Kneeshaw, D.D. and P .J . Burton. 1997. Canopy and age structure of some old sub-boreal Picea stands in British Columbia. Journal of Vegetation Science 8: 615-626. Kneeshaw, D.D. and P.J . Burton. 1998. Assessment of functional old-growth status: a case study in the Sub-Boreal Spruce Zone of British Columbia, Canada. Natural Areas Journal 18: 293-308. Kneewshaw, D.D. and P.J . Burton. 1997. Canopy and age structure of some old sub-boreal Picea stands in British Columbia. Journal of Vegetation Science 8: 615-626. LePage, P. 1995. The structure and development pattern of mixed-species forest stands in the Interior Cedar Hemlock zone: moist cold subzone of northwestern British Columbia. MSc Thesis, Oregon State University, Corvallis, Oregon. 91pp. Lertzman, K.P. 1992. Patterns of gap-phase replacement in a subalpine, old-growth forest. Ecology 73: 657-669. Lertzman, K.P. and C . J . Krebs. 1991. Gap-phase structure of a subalpine old-growth forest. Canadian Journal of Forest Research 21: 1730-1741. Lorimer, C . G . and L.E. Frelich. 1989. A methodology for estimating canopy disturbance frequency and intensity in dense temperate forests. Canadian Journal of Forest Research 19: 651-663. MacKinnon, A. 1998. Biodiversity and Old-Growth Forests. In Voiler, J . and S. Harrison, (eds). Conservation Biology Principles for Forested Landscapes, pp. 146-184. U B C Press, Vancouver, B C . Marcot, B.G. 1997. Biodiversity of Old Forests of the West: A Lesson from our Elders. In Kohm, K.A. and J .F . Franklin, (eds). Creating a Forestry for the 21st Century, pp. 87-106. Island Press, Washington D.C. Marshall, P.L. and J.L. Bugnot. 1991. Review of some designs for sampling coarse woody debris. Research Report prepared for the B C Ministry of Forests, Vancouver Region. 55 p. Marshall, P.L., M.D. Pitt, and H.L. Habgood. Estimating browse biomass using multiple regression and plotless density estimates. Journal of Wildlife Management 54: 180-186. Mehl, M.S. 1992. Old Growth Descriptions for the Major Forest Cover Types in the Rocky Mountain Region. In Old-Growth Forests in the Southwest and Rocky Mountain Regions: proceedings of a workshop, pp. 106-120. U S D A Forest Service General Technical Report RM-213. Ministry of Environment, Lands, and Parks and Ministry of Forests. 1998. Field manual for describing terrestrial ecosystems. Government of British Columbia, Victoria, BC . Ministry of Forests and Ministry of Environment Lands and Parks. 1995. Biodiversity Guidebook. Forest Practices Code of British Columbia, Victoria, BC . Ministry of Environment Lands and Parks and Ministry of Forests. 1999. Landscape Unit Planning Guide. Forest Practices Code of British Columbia, Victoria, B C . 101 p. Ministry of Forests. 2002. Establishing permanent sample plots in natural stands. Vegetation Resources Inventory. Ministry of Forests, Victoria B C . Available at: http://srmwww.gov.bc.ca/tib/standard/volume3/2k2/chap1.htm Oliver, C D . 1981. Forest development in North America following major disturbances. Forest Ecology and Managements: 153-168. Oliver, C D . and B.C. Larson. 1996. Forest stand dynamics, 2 n d Ed. McGraw-Hill Inc. New York, 467 p. Orwig, D.O., C.V. Cogbill, D.R. Foster, and J .F. O'Keefe. 2001. Variations in old-growth structure and definitions: forest dynamics on Wachusett Mountain, Massachusetts. Ecological Applications 11: 437-452. Palik, B.J. and K.S. Pregitzer. 1995. Variability in early height growth rate of forest trees: implications for retrospective studies of stand dynamics. Canadian Journal of Forestry Research 25: 767-776. Parish, R. and J.A. Antos. 2002. Dynamics of an old-growth, fire-initiated, subalpine forest in southern interior British Columbia: tree ring reconstruction of 2 year cycle spruce budworm outbreaks. Canadian Journal of Forest Research 32: 1947-1960. Parish, R., J.A. Antos, and M.-J. Fortin. 1999. Stand development in an old-growth subalpine forest in southern interior British Columbia. Canadian Journal of Forest Research 29: 1347-1356. Parker, A . J . and R.K. Peet. 1984. Size and age structure of conifer forests. Ecology 65: 1685-1689. Pickett, S.T.A., and P.S. White, (ec/s). 1985. The ecology of natural disturbance and patch dynamics. Academic press, New York. Pojar, J . 1991. Old Growth Forests in British Columbia. In H. Schrier, S. Brown, P. O'Reilly, and P. Meehan, (eds). Applications in a Changing World, Symposium Proceedings GIS '91. pp. 15-21. Vancouver, BC , February 12-15, 1991. F R D A Report 153, Forestry Canada, Victoria, B C . Quesnel, H.J. 1996. Assessment and characterization of old growth stands in the Nelson Forest Region. Technical Report TR -013. B.C. Ministry of Forests, Forest Sciences Section, Nelson, B.C. Qian, H., K. Klinka, and B. Sivak. 1997. Diversity of the understory vascular vegetation in 40 year-old anold-growth forest stands on Vancouver Island, British Columbia. Journal of Vegetation Science 8: 773-780. Rebertus, A . J . , T.T. Verblen, L.M. Roovers, and J . Nystrom-Mast. 1992. Structure and dynamics of old-growth Engelmann Spruce - Subalpine Fir in Colorado. In Old-growth forests in the Southwest and Rocky Mountain Regions: proceedings of a workshop, pp. 139-153. U S D A Forest Service General Technical Report RM-213. Rogers, B. and M. Jull. 2000. Fleet Creek ICH group selection research project: planted regeneration trial 1995-1999. Unpublished report. Runkle, J .R. 1982. Patterns of disturbance in some old-growth mesic forests of eastern North America. Ecology 63: 1533-1546. Smith, D.M., B.C. Larson, M.J. Kelty, P.M.S Ashton. 1997. The practice of silviculture: Applied forest ecology, 9 t h edition. John Wiley & Sons, Inc., New York. 537 p. Spies, T.A. 1997. Forest Stand Structure, Composition, and Function. In Kohm, K.A. and J .F . Franklin, (eds). Creating a Forestry for the 21st Century, pp. 11-30. Island Press, Washington D.C. Spies, T.A. and J .F . Franklin. 1988. Old growth and forest dynamics in the Douglas-Fir region of Western Oregon and Washington. Natural Areas Journal, 8: 190-201. Spies, T.A. and J .F . Franklin. 1989. Gap characteristics and vegetation response in coniferous forests of the Pacific Northwest. Ecology 70: 543-545. Spies, T.A. and J .F. Franklin. 1996. The diversity and maintenance of old-growth forests. In Szaro, R.C. and D.W. Johnston, (eds). Biodiversity in managed landscapes: theory and practice. Oxford University Press, New York. 778 p. Spies, T.A., J .F. Franklin, and M. Klopsch. 1990. Canopy gaps in Douglas-fir forests of the Cascade Mountains. Canadian Journal of Forest Research 20: 649-658. S P S S Inc. 1999. S P S S 10.0 for Windows. Steeger, C , R.F. Holt, M.-A. Beaucher, and H. Quesnel. 1998. Post disturbance residual stand structure and its value as wildlife habitat. A report for BC Environment, Wildlife Branch, Kootenay Region, Nelson, B C . 56 p. Stephenson, N.L. and A. Demetry. 1995. Estimating ages of giant sequoias. Canadian Journal of Forestry Research 25: 223-233. Stevenson, S.K. 1986. Review of forestry practices in caribou habitat in southeastern British Columbia, Canada. Rangifer, Special Issue No. 1: 289-295. Stevenson, S.K., H.M. Armleder, M.J. Jull, D.G. King, B.N. McLellan, and D.S. Coxson. 2001. Mountain Caribou in managed forests: Recommendations for managers. Second Edition. Wildlife Report No. R-26. British Columbia Ministry of Environment, Lands and Parks, Victoria, BC . Stewart, G H . 1986a. Forest development in canopy openings in old-growth Pseudotsuga forests of the western Cascade Range, Oregon. Canadian Journal of Forest Research 16: 558-568. Stewart, G H . 1986b. Population dynamics of a montane conifer forest, western Cascade range, Oregon, USA. Ecology 67: 534-544. Tabachnik, B.G. and L.S. Fidell. 1996. Using Multivariate Statistics, 3 r d Ed . HarperCollins College Publishers. 880 p. Thomas, J.W., L.F. Ruggiero, R.W. Mannan, J.W. Schoen, and R.A. Lancia. 1988. Management and conservation of old-growth forests in the United States. Wildlife Society Bulletin 16: 252-262. Tyrrell, L.E. and T.R. Crow. 1994. Structural Characteristics of old-growth Tsuga-hardwood forests in relation to age. Ecology 75: 370-386. Utzig, G. , K. Scott-May, R.F. Holt, M. Machmer, B. Lewis, C. Wallace, and M. Carver. 2003. Ecological conservation in the Southern and Central Columbia Mountains Ecosections: a context for developing objectives, strategies and monitoring indicators. Unpublished report prepared for the Ministry of Water, Air and Land Protection, Nelson, BC . 134 p. Visser, H. 1995. Note on the relation between ring width and basal area increments. Forest Science 41: 297-304. Waters, L. and H. Quesnel. 2001. Case Study: Patch cutting in old-growth forests to maintain mountain caribou habitat, 1997-99 research results. Extension Note 054, Forest Sciences, Nelson Forest Region, Nelson, B C . Waters, L. and R. Delong. 2001. Recruiting Caribou Habitat using Silviculture Treatments. Draft report prepared for the B C Ministry of Environment, Lands and Parks, Nelson, B C . Wells, R.W., K.P. Lertzman, and S .C . Saunders. 1998. Old-growth definitions for the forests of British Columbia, Canada. Natural Areas Journal 18: 279-292. Wilson, S .F . Stand structure: preliminary statistical analyses. Unpublished report prepared for the Arrow IFPA, Slocan, B C . 14p. Wong, C , B. Dorner, and H. Sandmann. In press. Estimating historical variability of natural disturbances in British Columbia. Chapter 3 in Land Management Report (in press). Canada. 74 p. Wong, C M . 1999. Memories of natural disturbances in Ponderosa pine - Douglas-fir age structure, southeastern British Columbia. Research project in fulfillment of a Masters Degree, Simon Fraser University, Burnaby, British Columbia. 107 p. Wright, E.F., C D . Canham, and K.D. Coates. 2000. Effects of suppression and release on sapling growth for 11 tree species of northern, interior British Columbia. Canadian Journal of Forestry Research 30: 1571-1580. Zeigler, S .S . 2002. Disturbance regimes of hemlock-dominated old-growth forests in northern New York, USA. Canadian Journal of Forest Research 32: 2106-2115. APPENDIX 1: Double bark thickness ratios from PrognosisBC Sx 0.956 Bl 0.937 Lw 0.851 Fd 0.867 PI 0.969 Pw 0.964 Hw 0.934 Cw 0.950 Py 0.890 Bg 0.915 Other 0.934 Source: (A. Zumwari, Growth and Yield Biometrician, British Columbia Ministry of Forests, Research Branch. Pers. com. 2001) APPENDIX 2: Age class frequency distributions, stratified by age estimation method and species or species guild. All species are combined in the left hand graph. Borna* I M . . I Tsuga I • • - I a • Thuja Intolerant Conifers Tolerant Conifers Pedro A g s a n - B I . Gveout I.I I . m r H a 1 . 1 , 1 , DBirrded • B4^rldaWCP/o HEtop± teMP/o •QortEd WD 101- 201- 301- 401- 331- 601- 70CH1-90 101- 201- 301-401- 501- 601- 7001-33 101- 201- 301- 401- 501- 601- 700133 10^ 201- 301-401- 5 0 1 - 6 0 1 - 7 ™ ^ ^ ^ 133233350463533693 150 2 5 0 3 5 3 4 5 0 5 3 3 6 9 0 133 2 5 0 3 3 3 4 3 3 5 5 0 6 5 0 150 2 5 0 3 3 3 4 3 3 9 3 3 6 3 3 1 3 3 2 3 3 3 3 3 4 5 0 9 3 3 6 3 3 a* l I 1 i i i i J i DEJiubJ DBtailaaW/c BBiailaaMy/c iGuted 1Œl)1-20r-:TJV4DV59V8Dr-7Dl<l)101-2^ tD 2TJ 2PD 43D 5SD 65) 19D 293 3BD 4D 3D SD 19D 2TJ 3Œ) 4D 5D GEO 1S 293 33) 43D 5D 650 13) 3D 3BD 433 5D 83 Siée _ L i  rai)V20V3Jl-4Jl-3>ej1-TOl3D10^ 1903333043093)61) 190 233 33343031)653 133233333433333633 133233333433333633 133233333433333633 Tsuga Thuja Intolerant Conifers Tolerant Conifers SxMIe l i Hcte I I J _ ia)f)i-2o»-33V43i-50V6Ji-7rjDi-raio^ 133 23) 333 433 5FD ŒD 1502033)45033)630 133 233 3B3 <S3 333 ŒD 15023333)43)33)653 13323333)43)33)63) APPENDIX 3: Additional thresholds for use in an index of old-growthness. Many of these thresholds are repetitions of similar attributes presented in the index of old-growthness. Structural Attributes Threshold Mean tree age > 268 years CWD DC1+2 < 231 sph CWD DC3 < 187 sph Snags 20-50 cm dbh < 48.2 sph Snags 30-40 cm dbh < 11.4 sph Snags <30 cm dbh < 133sph #DC Snags <4.54 Snag BA <30 cm dbh < 2.23 m2 Snag BA <20 cm dbh < 1.08 m2 Snag BA 20-30 cm dbh <1.15m2 Snag BA 30-40 cm dbh <0.9m2 Snag BA >70 cm dbh > 1.63 m2 Trees <30 cm dbh < 464 sph Trees 70-90 cm dbh > 16 sph Trees >90 cm dbh > 3 sph Trees >50 cm dbh > 71 sph Tree BA 50-70 cm dbh > 11.3 m2 Tree BA 70-90 cm dbh > 6.59 m2 Tree BA >90 cm dbh > 1.96 m2 Tree StdDevDBH >21 CWD 7.5-15 cm diameter (SPH) < 224 sph CWD 15-30 cm diameter (SPH) < 373 sph CWD >60 cm diameter (SPH) > 35 sph 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0075102/manifest

Comment

Related Items