UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Natural regeneration in a cutover (clearcut) and the adjacent old-growth stand on the outer central coast… McClarnon, Christine Fietkau 2000

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

Item Metadata

Download

Media
831-ubc_2000-0119.pdf [ 5.86MB ]
Metadata
JSON: 831-1.0099496.json
JSON-LD: 831-1.0099496-ld.json
RDF/XML (Pretty): 831-1.0099496-rdf.xml
RDF/JSON: 831-1.0099496-rdf.json
Turtle: 831-1.0099496-turtle.txt
N-Triples: 831-1.0099496-rdf-ntriples.txt
Original Record: 831-1.0099496-source.json
Full Text
831-1.0099496-fulltext.txt
Citation
831-1.0099496.ris

Full Text

N A T U R A L R E G E N E R A T I O N IN A C U T O V E R ( C L E A R C U T ) A N D T H E A D J A C E N T O L D - G R O W T H S T A N D O N T H E O U T E R C E N T R A L C O A S T O F B R I T I S H C O L U M B I A by C H R I S T I N E F I E T K A U M C C L A R N O N B.ScF. University of Toronto, 1989 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F S C I E N C E in F A C U L T Y O F G R A D U A T E S T U D I E S Department of Forestry (Forest Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 1999 © Christine Fietkau McClarnon 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. 1 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 ToreST\~-| ^ P o r ^ f Scit ntfi*.^ The University of British Columbia Vancouver, Canada Date DE-6 (2/88) ABSTRACT Given increased interest in harvesting lowland cedar-hemlock rainforests in the Very Wet Hypermaritime Coastal Western Hemlock subzone of British Columbia, this observational study was initiated to seek better understanding of how these forests regenerate naturally, both in the absence of, and following, harvesting. Plots were established in a clearcut and an immediately adjacent old-growth stand, within which small trees (less than 1.3 m) were examined. Small tree presence, height, and age distributions were constructed according to stand type, species, reproductive origin, stand edge distance (in the clearcut), microtopography, substrate, soil moisture and nutrient regime, and non-crop vegetation. Potential future crop trees per hectare were also estimated. Based on this study: 1) Ingress could not be relied upon to regenerate the clearcut within 5-years, whereas advanced regeneration, some of which lacked apical dominance of shoot terminal leaders, appeared to be important. 2) Proportionally more veglings than seedlings reached 5-years of age and 30 cm in height. In the clearcut, proportionally more Tsuga heterophylla than Thuja plicata reached 30 cm in height, and advanced regeneration consisted of proportionally more Tsuga heterophylla than Thuja plicata. 3) Proportionally more small trees occurred on Lignomor, mineral, and moss substrates than on other substrates in both stand types. 4) Microtopographic influence was related to substrate and soil moisture and nutrient regime, and differed between stand types. In the clearcut, increased soil nutrient regime on inclines/mounds positively influenced small tree presence and height growth beyond 30 cm, while increased soil ii moisture regime on inclines/mounds negatively influenced height growth beyond 30 cm. In the old-growth, proportionally fewer small trees reached 30 cm in height on inclines/mounds, and proportionally fewer reached 5-years of age on wet organic or moss substrates on inclines/mounds. 5) In the old-growth, increased soil nutrient regime negatively influenced small tree presence, and increased soil moisture regime positively influenced survival beyond 5 years. In the clearcut, increased soil nutrient regime negatively influenced small tree height growth beyond 30 cm, yet positively influenced height growth beyond 30 cm on Lignomors; increased soil moisture regime negatively influenced survival beyond 5-years; and increased soil nutrient regime negatively influenced Tsuga heterophylla survival beyond 5-years. 6) Blechnum spicant and Calamagrostis nootkatensis were more extensive in the clearcut, and Blechnum spicant negatively influenced small tree presence. Grass/sedge presence may also limit regeneration. TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES vii ACKNOWLEDGMENTS viii 1. INTRODUCTION 1 1.1 Background 1 1.2 Objectives 4 2. MATERIALS AND METHODS 4 2.1 Terms of Reference 4 2.2 Description of Study Area 5 2.3 Field Methods 8 2.4 Analysis 10 2.4.1 Description of the study design 10 2.4.2 Potential future crop trees per hectare estimates 11 2.4.3 Response and explanatory variables 12 2.4.4 Logistic regression 12 2.5 Descriptions of the explanatory variables 16 2.5.1 Stand type 16 2.5.2 Block 17 2.5.3 Plot Contiguity 18 2.5.4 Origin 18 2.5.5 Species 19 2.5.6 Stand edge distance in the clearcut 19 2.5.7 Microtopography 20 2.5.8 Substrate 21 2.5.9 Soil moisture and nutrient regimes 22 2.5.10 Non-crop vegetation 22 3. RESULTS 25 3.1 General nature of small tree data 25 3.1.1 Small tree height and age profiles by stand type, for Block, reproductive origin, species, microtopography, substrate, soil moisture and nutrient regime, and non-crop vegetation cover 25 3.1.2 Small tree abundance 33 3.1.3 Estimates of the numbers of taller potential future crop trees per ha 34 3.2 Chi-Square test results for the influence of plot contiguity on small tree presence, age and height distributions 35 3.3 Stepwise logistic regression results for small tree presence, height and age classes 36 3.3.1 Model adequacy for small tree presence, height and age class regressions 36 iv 3.3.2 Variables detected to influence small tree presence, height and age class distributions, according to selected logistic regression models 37 3.3.3 Small tree presence class distributions 40 3.3.4 Small tree height class distributions 43 3.3.5 Small tree age class distributions 50 4. DISCUSSION 56 4.1 Factors that influence small tree presence, height and age distributions in the study area ....56 4.1.1 Block and plot contiguity 59 4.1.2 Genetic factors 61 4.1.3 Clearcut stand edge effects 63 4.1.4 Microtopography 64 4.1.5 Substrates 66 4.1.6 Soil moisture and nutrient regimes 70 4.1.7 Non-crop vegetation 71 4.2 Statistical analysis 76 5. CONCLUSIONS 79 LITERATURE CITED 83 APPENDIX!: SUBSTRATES 96 APPENDIX 2: VEGETATION 101 APPENDIX 3: MODEL ADEQUACY STATISTICS 103 v LIST OF TABLES Table 1. Terms of reference 5 Table 2. Classification of response variables for logistic regression 13 Table 3. Summary of dummy variables in small tree presence, height class, and age class regressions... 14 Table 4. Proportions of plots located in each site and stand type 16 Table 5. Species occurrence in study plots 19 Table 6. Stand edge distance classes and plot frequencies 20 Table 7. Percentage of plots occurring in different microtopography groups 20 Table 8. Distribution of soil nutrient and moisture regimes by stand type 22 Table 9. Occurrence of non-crop vegetation groups throughout study area 23 Table 10. Non-crop vegetation cover classes and categories 24 Table 11. Relative occurrence of vegetation cover categories for each vegetation group over all plots in the study 24 Table 12. Summary of small tree height and age on each Block within each stand type 26 Table 13. Summary of small tree height and age for each reproductive origin within each stand type 27 Table 14. Summary of small tree height and age for each species within each stand type 27 Table 15. Summary of small tree height and age on each microtopography within each stand type 28 Table 16. Summary of small tree height and age on each substrate within each stand type 29 Table 17. Summary of small tree height and age for each soil moisture and nutrient regime within each stand type 31 Table 18. Summary of small tree height and age for non-crop vegetation presence within each stand type 32 Table 19. Univariate statistics describing the number of small trees per plot by stand type 33 Table 20: Small tree (per plot) abundance by stand type 34 Table 21. Estimated number of taller potential future crop trees/ha in study area by stand type 34 Table 22. Results of Chi-Square tests for the influence of plot contiguity on small tree presence, height and age distributions 35 Table 23. Model acceptance statistics for small tree presence, height and age class regressions 37 Table 24. Variables that influence small tree presence, height and age class, according to selected logistic regression models 38 Table 25. Substrates found throughout the study area 97 Table 26. Relative occurrence of plant species identified in the study plots. 101 LIST OF FIGURES Figure 1. Location of Illahie Inlet 6 Figure 2. Field map of study area, showing sampling design and plot layout 9 Figure 3. Percentage of area occuppied by substrate groups within each stand type 21 Figure 4. Small tree presence in the clearcut, according to microtopography and stand edge distance. ...41 Figure 5. Small tree presence, by stand type, according to soil nutrient regime and microtopography. ...41 Figure 6. Small tree presence, by stand type and substrate cover class, according to substrate group 42 Figure 7. Small tree presence, by stand type and non-crop vegetation cover class, according to non-crop vegetation group 43 Figure 8. Percentage of small trees => 30 cm by stand type and reproductive origin 44 Figure 9. Percentage of small trees => 30 cm by stand type and species 44 Figure 10. Percentage of small trees => 30 cm, by stand type and microtopography 45 Figure 11. Percentage of small trees => 30 cm, by stand type and microtopography according to soil moisture and nutrient regime classes 45 Figure 12. Percentage of small trees => 30 cm, by stand type and substrate group 47 Figure 13. Percentage of small trees => 30 cm, by stand type and species, according to substrate group 47 Figure 14. Percentage of small trees => 30 cm in clearcut, by soil nutrient regime according to substrate group 48 Figure 15. Percentage of small trees => 30 cm by stand type according to soil moisture and nutrient regime 48 Figure 16. Percentage of small trees => 30 cm, by stand type, species and non-crop vegetation cover class, according to non-crop vegetation group 49 Figure 17. Percentage of small trees => 5-years by stand type and reproductive origin 50 Figure 18. Percentage of small trees => 5-years by stand type and species 50 Figure 19. Percentage of small trees => 5-years, by stand type and microtopography, according to substrate group 51 Figure 20. Percentage of small trees => 5-years by stand type and substrate group 52 Figure 21. Percentage of small trees => 5-years, by stand type according to soil moisture and nutrient regime classes. 53 Figure 22. Percentage of small trees => 5-years, by stand type and soil nutrient regime classes, according to species 54 Figure 23. Percentage of small trees => 5-years, by stand type and soil nutrient regime class, according to substrate group 54 Figure 24. Percentage of small trees => 5-years, by stand type and evergreen shrub cover class, according to Lignomor versus other substrates 55 Figure 25. Percentage of small trees => 5-years, by stand type and grass/sedge cover class, according to wet organic versus other substrates 56 ACKNOWLEDGMENTS Many thanks to the following individuals and institutions who contributed advice, services, and/or financial assistance to this project. I am grateful for all support received. Dr. K. Klinka (University of British Columbia) kindly agreed to supervise this project, and provided helpful guidance in setting up the project, along with copious editorial comments. Dr. V. LeMay, Dr. C. Prescott, and Dr. T. Ballard (all from the University of British Columbia) kindly agreed to serve on the committee and all provided helpful editorial comments on the draft. Dr. J. Worrall also provided useful editorial comments on the manuscript. Dr. V. LeMay provided rigorous criticism of the statistical analysis. Dr. J. Zhou (University of Victoria) provided helpful general comments on statistical analysis of skewed data, and W. Bergerud (Research Branch, Ministry of Forests) provided helpful advice on how to use SAS, along with general comments on the applicability of logistic regression techniques to forestry data. Dr. C. Prescott provided useful background literature on SCFflRP and other regeneration studies on Vancouver Island. Mid Coast Forest District granted an 8 month unpaid leave of absence to complete work associated with graduate residency, and also provided maps, metal flags, sample collection bags, waterproof field cards, and cruise information. J. Charles (Pacific Phytometric) provided invaluable assistance with data collection throughout difficult field conditions. L.Woodham (Coast Forest Management) also assisted with plot layout and data collection. L. Kaytor (Hagensborg) and R. Taylor (Coast Forest Management) assisted with small tree measurements. Floatplanes from Wilderness Airlines were used to access the field site during the first phase of data collection. The boat of E. Wilson (Hagensborg) was used for lodging during the second phase of data collection. Mid Coast Forest District provided the use of the Forest Patrol (boat) to complete interpretive work in the study area. All of these services were financed by the Mid Coast Forest District. Ministry of Forests (Silviculture Practices Branch) provided unpaid partial leaves of absence to complete graduate coursework. D. Trotter (Ministry of Forests, Green Timbers Nursery) provided polycorders to assist with seedling data capture. A. Banner and J. Pojar (Ministry of Forests, Smithers) provided background reference material on North American hypermaritime forests. M. Milum (Ministry of Forests, Timber Supply Branch) and M. Tsoi (Ministry of Forests, Forest Practices Branch) provided data for the context of this study. Mill and Timber provided the use of camp, food and lodging during the first phase of data collection. I. Leung (forestry consultant) provided maps, silvicultural history, and general information about the study area. Elan Data Management helped input data from field cards onto an Excel spreadsheet (I financed this service). Dr. J. Dobry (University of British Columbia) and D. Epp (Hagensborg) assisted in the preparation of base discs for microscopic tree ring counts. R. Daniels (Mid Coast District) constructed the metal frame used to define quadrat plots. My husband, John, cheerfully provided enthusiasm, inspiration, patience, and moral support since the beginning of this project through to its completion. My mother provided moral support, good meals, and financial help during my leaves of absence from my employer. H. Chen, G. Kayahara, B. Brett and P. Varga provided helpful advice throughout this project. I would like to thank Dr. Klinka, Dr. LeMay, Dr. Prescott, and Dr. Ballard for setting examples of high standards of workmanship, standards which I hope one day to achieve as I gain more experience in my forestry career. Finally, I would like to thank the Mid Coast District for having provided me with the opportunity not only to pursue this study, but also to work in one of the world's most beautiful and pristine natural environments. It was the experience of a lifetime. 1. Introduction 1.1 Background The unique temperate rainforest of Canada is situated along the British Columbia (BC) outer coast in the Very Wet Hypermaritime Coastal Western Hemlock subzone (CWHvh) (Meidinger and Pojar 1991; Green and Klinka 1994; Banner et al. 1993). During the last 10 years, approximately 5000 ha of this forest have been harvested annually (BC Ministry of Forests (MOF) 1999a). The central (CWHvh2) variant of this subzone (Green and Klinka 1994) is located in the Mid Coast Forest District. In the Mid Coast Forest District, lowland forests occurring within the CWHvh2 variant fall predominantly within the MOF Resource Inventory Branch "Height Class 3" classification, which represents a stand height interval of 19.5 - 28.4 m (BC MOF 1997). These stands are referred to as "economically marginal" (BC MOF 1993). Eighteen percent of the volume of wood cut annually (180 000 m3y_1) in the Mid Coast Timber Supply area is expected to come from economically marginal stands (BC MOF 1999b). Small tree establishment and growth depend on both genetic and site attributes. Genetic attributes determine tolerance (as defined by Klinka et al. 1990) of the tree, and requirements for specific ranges of the site attributes. Genetic attributes include species and parental origin. Site attributes include light, moisture, nutrients, air temperature, and humidity (Klinka et al. 1990) as well as exposure to strong winds. The choice of which silvicultural system1 to use in any given stand is based on the interaction of numerous genetic and site factors (Klinka et al. 1990; Klinka 1 As defined by BC Ministry of Forests 1991 1 et al. 1994) and is therefore complex. Given historical lack of interest in harvesting from this area, coupled with the absence of stand level management objectives, little research has been conducted to determine which silvicultural systems and treatments are best suited to the CHWvh2 variant, including the overall impact of harvesting. Nevertheless, studies elsewhere along the entire length of the outer coast from Alaska, USA, to southern Vancouver Island have indicated that poor nutrition (Weetman et al. 1989a; Weetman et al. 1989b) and excess moisture (Zach 1950; Andersen 1955; Neiland 1971; Ugolini and Mann 1979) may affect regeneration in outer coastal forests. Some evidence indicates that ingress may be important in regenerating western hemlock (Tsuga heterophylla [Raf] Sarg.), over a number of years (MacBean 1941) and that vegetative layering may be important for establishing western redcedar (Thujaplicata Donn ex D. Don in Lamb.) (Habeck 1968 and 1978; Schmidt 1955; DeLong 1997; Boyd 1959). Although western hemlock (Hw) and western redcedar (Cw) may have different substrate preferences, with decayed wood and thick forest floors being preferred for Hw establishment (Green and Klinka 1994) and mineral substrates being preferred for Cw establishment (Soos and Walters 1963), Hw appears to be less tolerant to high water tables than Cw (Minore and Smith 1971). The forests in the study area are dominated by Cw-Hw forests, similar in species composition to Cw-Hw forests in the nearby Very Wet Maritime Coastal Western Hemlock (CWHvm) subzone (Meidinger and Pojar 1991; Green and Klinka 1994; Banner et al. 1993), which have been noted to undergo growth check after 8-14 years (Weetman et al. 1989a; Weetman et al. 1989b). Poor growth of conifers on salal-dominated Cw-Hw clearcuts may be attributable to the combined effects of inherently low forest floor nutrient availability in older Cw-Hw forests and salal competition for scarce nutrients and nutrient immobilization in salal (Messier 1993). Although forests in the CWHvh2 variant also contain a high component of salal, and the 2 vigor of salal appears to decrease with increasing latitude along the outer coast (A. Bannerz pers. comm. 1999), the impact of salal on forest regeneration has not been clearly determined in the CWHvh2 variant. Prescott et al. (1996) have previously shown that, in the CWHvm subzone: 1) Cw litter contains less nitrogen and more decay-resistant material than litter produced from other species, and produces forest floors with low rates of nitrogen (N) mineralization; 2) forest floors in Cw-Hw forests are wetter and have less soil fauna than in western hemlock-amabilis fir (Abies balsamea (L.) Mill.) (Hw-Ba) forests found under the same climatic conditions, leading to incomplete N decomposition and mineralization; and 3) salal understories in Cw-Hw forests interfere with N mineralization through tannin production. Alban (1969) found that organic horizons were thicker under Cw than under Hw. Cw litter, which has higher lignin concentrations than litter under Hw-Ba forests, may contribute to humus accumulation in Cw-Hw forest floors (Prescott et al. 1995; Keenan et al. 1996). High rainfall in southeast Alaska has been considered a hindrance to optimum tree growth given its tendency to cause excess groundwater (Zach 1950; Andersen 1955), even to the extent that bog forests have developed in maritime areas (over a period of several hundred years) from productive forests initially occurring on well-drained soils (Ugolini and Mann 1979). This process occurs on sites in which drainage is impeded, including gentle slopes (Banner et al. 1986; Neiland 1971, Zach 1950; Andersen 1955). Mean rainfall in Alaska can range between 120-550 cm annually (Neiland 1971), which is comparable to the rainfall received in the study area (see Section 2.2). Features present within the lowland CWHvh2 variant, such as high rainfall, abundant vegetation, mild winters and cool summers, low topographic relief, and strong 2 BC Ministry of Forests, Prince Rupert Region, Smithers 3 podzolization may promote widespread long-term paludification, which could ultimately affect species composition and growth. Pattern (species distribution in relation to site attributes, substrate, and competing vegetation) and growth of natural regeneration established in old-growth stands and on clearcuts are not completely understood. How does natural regeneration on a clearcut differ from that occurring in a forest understory in this area? Does lesser vegetation (shrubs, herbs, or bryophytes) hinder natural regeneration when the natural canopy is removed? How does natural regeneration vary with increasing distance from the stand edge? Can advanced regeneration and ingress be relied upon to meet regeneration commitments? This study uses a recent clearcut and an adjacent old-growth forest as a preliminary basis from which to examine these questions. 1.2 Objectives The purpose of this study is to characterize and compare regeneration presence, height, and age distributions of naturally occurring small trees <1.3 m in height on a 5-year-old cutover and on an adjacent old-growth stand, including relationships among genetic (i.e., reproductive origin; and, species, primarily Cw and Hw) and site (i.e., stand edge distance, microtopography, substrate, soil moisture and nutrient regime, and non-crop vegetation) characteristics. 2. Materials and Methods 2.1 Terms of Reference Certain terms in this thesis are not consistent with definitions used by other organizations and are therefore defined in Table 1. This thesis focuses only on small trees less than 1.3 m in height. The terminology as defined below allows for distinction between reproductive origin; e.g., 4 seedlings and veglings, versus trees of the two different height classes; e.g., mini small trees and taller small trees throughout this thesis, whereas, for example, USD A (Burns and Honkala 1990) considers all trees less than 1 metre in height "seedlings" with no distinction placed on reproductive origin. Table 1. Terms of reference. Term Description "small tree" Any tree equal to or less than 1.3 metres in height. "mini small tree" Any small tree (seedling or vegling) less than 30 cm in height, including germinants. "taller small tree" Any small tree equal to or exceeding 30 cm in height. "seedling" Any small tree deemed to originate from seed (Gove 1986). "vegling" Any small tree deemed to originate vegetatively (Parker and Johnson 1987; Minore 1990). "younger small tree" Any small tree less than 5-years of age. "older small tree" Any small tree equal to or exceeding 5-years of age. "advanced regeneration" Any small tree equal to or exceeding 5-years of age in the clearcut. "ingress" Any small tree less than 5-years of age in the clearcut (natural regeneration only). "potential future crop (PFC) tree" Any small tree likely to meet provincial stand stocking requirements. 2.2 Description of Study Area The study area is located in Illahie Inlet, immediately south of Elizabeth Lagoon (Figure 1), in the central outer coast of British Columbia at approximately 127°50'W longitude and 51°35' N latitude and within the CWHvh2 variant (Green and Klinka 1994; Banner et al. 1993). This area is within a system of lowlands known as the "Coastal Trough and Lowlands" (Luttmerding et al. 1990), which is part of a structural depression formed by down-warped sequences of Cretaceous and Tertiary rocks, most of which is submerged, but portions of which rise above sea level to form a series of coastal lowlands (Clayton et al. 1977). 5 Figure 1. Location of Illahie Inlet. Scale: 1:600,000. Arrow points to the location of the study area. Illahie Inlet is located directly south of Elizabeth Lagoon (identified on map) and directly north of Fish Egg Inlet (not identified on map). 6 The climate is characterized by cool summers, mild winters with little snow, frequent precipitation with seasonal fog blankets (Green and Klinka 1994; Banner et al. 1993), and prolonged and violent winter storms each year (Phillips 1990). Precipitation exceeds 250 cm each year and average winter monthly totals may exceed 40 cm (Phillips 1990; Meidinger and Pojar 1991), causing permanent soil saturation. Soil temperature is cold to moderately cold cryoboreal, under maritime influence, with mean annual temperatures of 2-8°C, and mean summer temperatures of 8-15°C (Department of Agriculture 1975). Soil is perhumid, under maritime influence, and moist all year with no significant moisture deficit (Department of Agriculture 1975; Banner et al. 1988). Soil organic content is high. Humo-Ferric Podzols (Department of Agriculture 1972), Humic Podzols (Canada Soil Survey Committee 19783), and soils of the Organic Soil Order, including Humisols and Folisols (Canada Soil Survey Committee 1978), are common throughout the area. The forest in the study area resembles an "upland scrub forest" (as described by Banner et al. 1987) and is typical of outer coastal lowland bog forests (Banner et al. 1986 and 1993) in the North Coast Pacific Oceanic Wetland Subregion (Banner et al. 1988). Forest productivity is limited by exposure to Pacific winds and gales (Clayton et al. 1977). Very old, slow growing, deformed, and stunted trees comprise a moderately open canopy forest containing many snags. Major tree species growing in this area are western redcedar, western hemlock, and yellow-cedar (Chamaecyparis nootkatensis [D.Don in Lamb.] Spach.) (Yc). Lodgepole pine (Pinus contorta Dougl. ex Loud.) (PI) and western yew (Taxus brevifolia Nutt.) (Tw) form a minor component. The plant communities and forest floor appear to have similar characteristics of the cedar-hemlock 3 Although this publication was used at the time this study was initiated, the Soil Classification Working Group (1998) have since updated this edition. 7 forests described by Prescott (1996) in the neighbouring CWHvm subzone, with the exception that salal (Gaultheria shallon Pursh) is somewhat less abundant. Site and floristic characteristics are as the "intermediate communities" described by Neiland (1971), which are intermediate between the steep, well drained coniferous forests and poorly drained bog forests along the Pacific coast occurring between 36" to 60° N latitude. 2.3 Field Methods Illahie Inlet is remote and difficult to access, with no fresh water supply. Data were collected over a 10-day period in August, 1994 and over a 4-day period in October, 1994. Three Blocks were delineated (Figure 2). Within each Block were two stand types: i) a 5-year-old clearcut, and ii) the immediately adjacent old-growth stand. Stand types and Blocks are further described in Section 2.5. Transect lines were installed; and quadrats, 1 m2 in size, were systematically located along each transect line (Figure 2). Moreover, eight groups of 25 contiguous quadrats (in 5x5 squares) were installed, in each Block-stand type combination (Figure 2). Each plot was flagged at plot center with a numbered flag. A lxl m metal frame was used to define each plot boundary. 8 Figure 2. Field map of study area, showing sampling design and plot layout. Scale: 1:4000. Dots enclosed by open-circles indicate approximate locations of sampled plots. Filled squares (approx. 2 mm2, n=8) denote locations of 25 contiguous 5x5 plot groups. Total number of plots sampled = 537. Thick broken line ( • » • » • • • » ) depicts boundary between West, South, and Flat Blocks. Thick cross-hatched line ( 4 » 4 * 4 * < t ) depicts boundary between stand types. 9 For each quadrat (see Section 2.5 for detailed descriptions of site attributes), the following types of data were collected: 1. Actual soil moisture and nutrient regimes, and microtopography (mound, depression, flat, incline). 2. Distance from the closest stand-edge in the clearcut quadrats (approximated from ocular estimates and reconfirmed with measurements using the map). 3. Ground surface substrates (APPENDIX 1) and their cover. A depth of <30 cm, approximating Cw and Hw rooting depth (Eis 1974; Cade-Menun 1996), was used for substrate examination. 4. Plant species identification (APPENDIX 2) and their cover. 5. Age, height (from root collar to terminal shoot apex), species, and reproductive origin (seedling or vegling) of each small tree, along with the substrate on which it grew. Age determination was consistent with Parker and Johnson (1987). Age estimation of older small trees required destructive sampling (cutting base discs) and counting annual rings. Height measurements for veglings were from the point on a branch from which small rootlets began, to the top of the terminal shoot. Franklin's (1961) guidelines were used to distinguish among species of recent germinants. 2.4 Analysis 2.4.1 Description of the study design This study is primarily descriptive (Jackson 1988), observational (Littell et al. 1996) and exploratory (Milliken and Johnson 1989). Unlike designed experiments and sample surveys, in which data are collected according to fully controlled pre-defined units, data in an observational 10 study are collected on units that are available (Littell et al. 1996). One site was studied, given the lack of availability of other logged sites in the area at the time the study was initiated. This type of study is suitable where the object(s) of interest is influenced by numerous factors in uncontrolled natural environments that are costly to research, is useful in areas where few studies have been conducted on the object of interest, and in developing guidelines for future studies (Milliken and Johnson, 1989). 2.4.2 Potential future crop trees per hectare estimates Estimates of the numbers of taller potential future crop (PFC) trees per ha (see Section 3.1.3) were based on a counting system in which only one taller PFC tree was assumed to eventually consume the growing space of a quadrat; therefore, only one 1 taller PFC tree was counted within any quadrat. This is consistent with the "nearest-neighbour" concept in stand structure analyses, in which distance between adjacent trees tends to increase as tree size increases (Moeur, 1997). Where both younger and older taller PFC trees were present in a plot, the younger one was counted, assuming better vigor in younger, faster growing small trees. These estimates are more liberal than tree-per-hectare estimates methods used by the BC MOF (1995), in which only "well-spaced" trees which are based on the minimum inter-tree distance4 specified in silviculture prescriptions are counted in a survey. Further details of methods for estimating PFC trees/ha are described in Table 21. 4 The minimum inter-tree distance reported in the prescription on the study area was 2.0 m, with a target inter-tree distance ranging between 3.12 - 3.58 m (courtesy of Mil l and Timber prescription files). This means that a tree must be spaced at least 2.0 m from its nearest neighbour before it can be counted in a legal regeneration survey 11 2.4.3 Response and explanatory variables In this study, the response variables (Hosmer and Lemeshow 1989; Stokes et al. 1995) for small trees were presence, height, and age. Explanatory variables (Hosmer and Lemeshow 1989; Stokes et al. 1995) were Block, plot contiguity in relation to other plots, plot to stand edge distance (clearcut only), reproductive origin, species, microtopography, substrate, soil moisture and nutrient regimes, and non-crop vegetation. 2.4.4 Logistic regression Small trees occurred on 35% of the substrates sampled. From the 3350 small trees sampled, 10% were at least 30 cm in height, and 25% were at least five years in age. Given the non-normal height and age distributions along the different genetic and site parameters, and variable sample sizes of most of the variables (see Section 3.1.1), traditional analysis of variance and general linear models methods could not easily be used to analyze the data. Since small tree presence, survival and growth performance under any given light conditions should be interpreted in relation to other factors such as seed bed, microsite variability, differences in age and height, and morphological adaptations to changing light environments (Carter and Klinka 1992), an analysis technique that could simultaneously analyze a number of important genetic and site variables acting together, was sought to determine how regeneration pattern (in terms of presence, height and age distributions) differed between the clearcut and old-growth. Given that similarly distributed forestry data have been previously successfully analyzed with logistic regression techniques (Bergerud 1996), data for the response variables were reconfigured into dichotomous ordinal classes, and stepwise logistic regression techniques (Hosmer and Lemeshow 1989; Agresti 1996; Stokes et al. 1995) were used to determine which 12 of the explanatory variables were most important in influencing the response variables in each stand type, after a satisfactory model was constructed. Additional Chi-Square tests were used to analyze the effects of plot contiguity separately since this variable was not adequately modeled in the regressions. A maximum of two classes for each response variable (Table 2) allowed meaningful cross-classifications with the relatively large number of explanatory variables in the study. Each response variable was regressed twice: 1) for the old-growth, without the "stand edge distance" variable; and 2) for the clearcut, with the "stand edge distance" variable. SAS statistical software (SAS Institute Inc. 1994b) was used to analyze the data. Table 2. Classification of response variables for logistic regression. Response variable Class Text reference Parameter description presence class 0 no trees number of trees = 0 1 at least 1 tree number of trees > 0 height class 0 mini small tree 0 < height < 30cm 1 taller small tree height => 30cm age class 0 younger small tree 0 < age < 5 1 older small tree age => 5 Each explanatory variable was initially categorized into either polytomous or dichotomous ordinal classes. Since logistic regressions are not easily performed on nominal data, all nominal explanatory variables were reconfigured into dummy variables (Table 3) after methods described by Stokes et al. (1995). Prior to final categorization of the explanatory variables, numerous iterations of regressions were attempted using various classes and combinations of variables. Data were recategorized, either by reducing the number of classes, or by combining certain variables with others (when possible), when too many cell counts for given variables were "0", and some variables were eliminated when it was not possible to combine them with other variables. 13 Table 3. Summary of dummy variables in small tree presence, height class, and age class regressions. Regression Variable Comparison dummy Variables Reference Dummy Variable presence, height, age contiguous no yes presence, height, age Block West, South Flat presence, height, age microtopography incline/mound depression/flat height, age species hemlock, other cedar height, age origin vegetative seed Statistics used to determine model adequacy (APPENDIX 3) for each regression were: 1) -2 log likelihood; 2) residual Chi-Square; 3) max-rescaled R2; 4) Hosmer-Lemeshow goodness-of-fit test; 5) rank correlation measures of observed responses and predicted probabilities; and, 6) sensitivity (the ability to predict an event correctly) and specificity (the ability to predict a non-event correctly) from the classification table (SAS Institute 1997). Pearson and.deviance statistics (SAS Institute 1997) were not used to determine model adequacy given data sparseness. All explanatory variables (main effects only) were initially regressed against the response variables. A regression output was produced indicating which variables were most important. These variables were then included in a stepwise regression along with some possible interactions5. For each regression, between 10-14 models were attempted before a final model was chosen. Model fit was considered adequate (for the purpose of this study) when the following conditions occurred: 1) the -2 log likelihood statistic produced a p value less than 0.05; 2) the residual Chi-Square produced a p-value greater than 0.05; 3) concordance measures exceeded 75%; 4) the Hosmer-Lemeshow goodness-of-fit statistic produced a p-value greater than 0.5; 5) sensitivity and specificity exceeded 70%; and 6) at least 80% of the explanatory variables were within Wald confidence limits range which did not include "1". In all cases, no 5 The SAS frequency procedure was used to eliminate those cross-classifications in which the following conditions (Stokes et al. 1995) did not hold: 1) at least 10 subjects per group; 2) 80% of predicted cell counts at least 5, and 5) all other expected cell counts exceeding 2, with no 0 counts. 14 model could be fitted which was adequate according to all of these criteria. Therefore, the model which satisfied most of the criteria was chosen. After each model was fitted, it was inspected to determine whether any dummy variables were missing (sometimes the stepwise regressions eliminated one dummy variable in a given set while retaining another). This occurred with all six regressions. To address this, the dummy variables which were eliminated from each model were then forced back into one further (and final) regression, this time performing the regression in a single step, using the variables in the final step of the preceding stepwise regression, plus the missing dummy variables. When it was not possible to force a dummy variable back into the regression (due to an insufficient number of samples), all dummy variables used to represent a given variable were eliminated and a final output was produced. Parameter estimates were not examined until an adequate model was fitted to the data according to the aforementioned criteria. Once model adequacy was confirmed, statistics used to determine which variables were most important include: 1) Wald confidence intervals; and, 2) Wald Chi-Squares and associated p-values (SAS Institute 1997). Finally, once the important variables were determined, their odds ratios6 (SAS Institute 1997) were used to interpret their relationships to the response variable. Where dummy variables were used, the regression odds ratios of the comparison dummy variables were always in relation to a reference dummy variable7 (Table 3); however it must be noted that odds ratios for the dummy variables were not true odds ratios and the regression output had to be interpreted with care for these variables (V. LeMay 6 "the change in the odds for any increase of one unit in the corresponding risk factor" 7 Throughout this document, the comparison dummy variable always precedes the reference dummy variable with a colon (e.g., West:Flat). 8 .Department of Forest Resources Management, University of British Columbia, Vancouver 15 pers. comm.). All models were followed up by graphing any results which would be used to draw conclusions, since Hosmer and Lemeshow (1989) have cautioned that it is not uncommon for logistic regression models to detect relationships which are non-existent and always recommend follow-up graphs. This would prevent conclusions from being drawn based on the regressions alone. 2.5 Descriptions of the explanatory variables This section provides an overview of each explanatory variable fitted to the logistic regressions for the three response variables described in Section 3.3. Each explanatory variable is described, followed by a rationale of how and why the explanatory variable was further classified for logistic regression analysis. Stand type, although not an explanatory variable, is described in this section, as it was the basis of stratification used to produce separate regressions for the clearcut and old-growth for each response variable. 2.5.1 Stand type Some differences between the clearcut and old-growth stand types (Figure 2) included full light exposure in the clearcut (although the old-growth forest had a fairly open canopy with both small-crowned and dead-topped cedar, with patches of areas subjected to windthrow interspersed throughout), higher proportion of ground vegetation in the clearcut, and higher proportion of substrates occupied by slash in the clearcut. The number of plots established differed by stand type (Table 4). Table 4. Proportions of plots located in each site and stand type. B L O C K (% plots) STAND TYPE (% plots) clearcut (n=305) old-growth (n=232) West (n=291) 55 34 21 Flat (n=104) 19 10 9 South (n=142) 26 13 13 16 2.5.2 Block The site was divided into three Blocks (Figure 2) and was included in the regression as an explanatory variable since the three aspects were different, and since one Block was flatter and wetter than the other two. Some differences in harvesting methods (e.g., amount of site disturbance, amount of slash left on site) may also have impacted the regeneration pattern among the three Blocks. The three Blocks were named according to their relative situations to one another: West Block was located on the western part of the site, South Block was located on the southern part of the site, and Flat Block was the flat area located on the northern part of the site. West Block is on a predominantly moist to very moist, nutrient poor to medium (Green and Klinka 1994), imperfectly drained, moderate to strong slope (Luttmerding et al. 1990), facing east and was identified as site series "01" according to the grid described in Green and Klinka (1994). Flat Block is level (Luttmerding et al. 1990), poorly drained, and predominantly very moist to wet, nutrient poor (Green and Klinka 1994) and was identified as a mosaic of site series "12" and "11" (Green and Klinka 1994). South Block is on a predominantly moist to very moist, nutrient poor to medium (Green and Klinka 1994), imperfectly drained, moderate to strong slope (Luttmerding et al. 1990), facing north on one side and south on the other, and was identified as site series "01". Considerable slash was left on the West and South Blocks. The number of plots established differed across the three Blocks (Table 4). Two dummy variables were created to describe the three Blocks, each of which was coded "1" for West and South Blocks, respectively. Zeros for both variables indicated the plot was located in Flat Block. 17 2.5.3 Plot Contiguity Eight contiguous groups of plots, which were laid out in squares of 5x5 quadrats, producing a 25 m square, were established to determine variability attributable to proximity and size of plots. For example, a plot located immediately adjacent to another plot was expected to have similar variation in the response variable compared to another plot located some distance away. One group of contiguous plots was established in Flat Block clearcut, Flat Block old-growth, South Block clearcut, and South Block old-growth. Two groups of contiguous plots were established in West Block clearcut, and West Block old-growth (since West Block was considerably larger than the other two Blocks). In the regressions, one dummy variable was created to describe plot contiguity, and was coded "1" if not contiguous to others. A zero for this variable indicated the plot was contiguous to others. Even so, modelling plot contiguity in this way was not adequate, since the model did not indicate to which plot(s) any particular plot was contiguous. Plot contiguity can be difficult to model unless spatial statistics are used (V. LeMay, pers. comm.) in the regressions. Since spatial statistics were not used in the regressions, separate Chi-Square tests were used to determine significance in responses within each unique stand type-Block combination. 2.5.4 Origin Reproductive origin was included only in the height and age regressions since origin data were too sparse for regeneration presence regression analysis (e.g., from substrate groups within plots which had no trees). Two tree origins were assigned: 1) seed origin (3149 seedlings), and 2) vegetative origin (191 veglings). Small tree origin in the clearcut was primarily expected to be from seed, whereas origin of older or taller small trees in the old-growth was primarily expected 18 to be vegetative (Schmidt 1955; Habeck 1978). Although veglings were expected only in Cw (K. Klinka9, pers. comm.), a small number of Hw, Yc and Tw appeared to have established from branches on boles of fallen trees and were identified as veglings. One dummy variable was created to describe the two origin classes, and was coded "1" for vegetative origin. A zero for this variable indicated seed origin. 2.5.5 Species Species variable (Table 5) was included only in the small tree height and age regressions, since species data were too sparse for small tree presence regression analysis (e.g., from substrates within plots which had no trees). Two dummy variables were created to describe the "Hw" and "Other" classes (Table 5), and were coded "1" for each class respectively. Zero for this variable indicated "Cw". Table 5. Species occurrence in study plots. Six conifer species were found in the study area. Three classes were formed for analysis: "Cw" (for western redcedar, comprising 2444 specimens), "Hw" (for western hemlock, comprising 776 specimens), and "Other" (for Lodgepole pine, Sitka spruce, yellow-cedar, and western yew, comprising 130 specimens). *Species codes are according to BC MOF (1998). **Yc was arbitrarily assigned to the "Other" class, given its generally low occurrence in the study area (similar to the other species in the "Other" class). Species (code*) % occurrence (N=3350) Class Cw 73.0 Cw Hw 23.2 Hw PI 1.1 Other Ss 0.1 Other Yc** 2.4 Other Tw 0.3 Other 2.5.6 Stand edge distance in the clearcut This study considered stand edge distance only in the clearcut. Three ordinal 50 metre distance classes, which occurred well within maximum seed dispersion ranges reported by Owens 9 Department of Forest Sciences, University of British Columbia, Vancouver 19 and Molder (1984a; 1984b; 1984c) of species occurring within the study area, were used (Table 6). Table 6. Stand edge distance classes and plot frequencies. The numbers of plots occurring in each of the stand edge classes is depicted. Class Distance (of plot) in metres to nearest stand edge Number of plots 1 <50 2 50 - <100 3 =>100 304 160 73 2.5.7 Microtopography Since depressions comprised only 4% of the plots, while mounds comprised 13% of the plots (Table 7), logistic regression analysis became difficult, given that some classes had very low cell frequencies. To overcome this difficulty, depressions were grouped with flat microsites since the drainage of a flat area was considered almost as poor as a depression in this hypermaritime climate, whereas the mounds were grouped with the inclines since some drainage might be expected on these microsites. One dummy variable was created to describe the two microtopography classes, and was coded "1" for incline/mound. A zero for this variable indicated the plot was located on a depression/flat. Table 7. Percentage of plots occurring in different microtopography groups. Key: d=depression f=flat df=depression and flat i=incline m=mound im=incline and mound Microtopography Plot occurrence (%) (N=537) _ _ f 55 df 59 i 28 m 13 im 41 20 2.5.8 Substrate Ten substrate groups were typed out (APPENDIX 1). To determine whether tree presence was influenced by amount of any given substrate, plot area coverage by each substrate was broken into four classes for the small tree presence regressions: "0" for no plot coverage of a given substrate; "1" for plot coverage less than 10%; "2" for plot coverage between 10 and less than 40%; and "3" for plot coverage equal to, or exceeding 40%. The substrate group occupying the greatest proportion of space in the clearcut plots was undecomposed coarse wood (28%, compared with 3% in the old-growth) (Figure 3). Other substrate groups occupying substantial plot area in the clearcut were wetland organic (24%), upland organic (20%), and Lignomor (16%) (Figure 3). By contrast, the old-growth plots were comprised primarily of upland organic (43%), Lignomor (23%), wetland organic (18%) and moss (12%, compared with 5% in the clearcut) (Figure 3). The "burned", "rock", "unknown", and "water" groups were not included in the regression analysis, since each comprised less than 1% of the given stand type, and could not reasonably be combined with other groups. D I C L E A R C U T B O L D - G R O W T H 111 111 111 , U L . , i i 111 Bur L i g M i n Mos Roc Ucw Unk Upl Wat Wet Substrate Groups Figure 3. Percentage of area occuppied by substrate groups within each stand type. Proportions are based on substrate occupation within a quadrat plot summed over all plots within a given stand type. Key: Bur=burned Lig=Lignomor Min=mineral Mos=moss Roc=rock Ucw=undecomposed coarse wood Unk=unknown Upl=upland organic Wat=water Wet=wetiand organic. 21 2.5.9 Soil moisture and nutrient regimes Given that soil moisture and nutrient influences on small tree growth are well accepted (Green and Klinka 1994; Pluth and Corns 1983; Jones 1986; Bergeron et al. 1992; Meades and Roberts 1992), soil actual10 moisture and nutrient regimes were typed out according to Green and Klinka (1994). In the clearcut, plots were predominantly moist, followed by very moist, whereas in the old-growth, plots were predominantly moist (Table 8). Plots were predominantly nutrient poor in both stand types (Table 8). Given low occurrences among some soil moisture and nutrient regimes, regimes were grouped together into two ordinal classes for the regressions. For soil moisture regime (SMR), Class 1 represents regimes sd, f, and m, and Class 2 represents vm and w. For soil nutrient regime (SNR), Class 1 represents regimes a and b, and Class 2 represents c and d. Table 8. Distribution of soil nutrient and moisture regimes by stand type. Shown are the percentage of plots that were identified with given snr and smr. SNR key: b=poor, c=medium, d=rich. SMR key: sd=slightly dry, f=fresh, m=moist, vm=very moist, w=wet. aearOTtXn=304)_ i P ] d -^^ . . (?=?? .? ) SMR: sd f m vm w total snr % { m vm w total snr% SNR: 1 b 0.3 9.5 35.5 23.0 16.5 84.8 j 52.6 15.5 13.4 c 0 4.3 6.9 0.7 1.0 12.9 15.5 0.4 0 d 0 0.7 0.3 1.0 0.3 i 2.3 j 0.4 0.9 1.3 total smr % 0.3 14.5 42.7 24.7 17.8 j 68.5 16.8 14.7 2.5.10 Non-crop vegetation Seventy-five plants were identified in the study plots (APPENDIX 2). Each plant was coded (Table 26 of APPENDIX 2) based on the first four letters of the genus name, followed by the first four letters of the species name. Pojar and Mckinnon (1994), Vitt et al. (1988), Lyons 81 5 15 9 2 6 1 0 As classified in British Columbia - see Green and Klinka 1994 22 (1976), Klinka et al. (1989), Taylor (1973), Brown (1979), and Britton and Brown (1970) were used to assist in plant identification and verification. Each plant was classified according to one of six vegetation groups (Table 9). Bryophytes predominated the plant life in the old-growth plots, covering 62% of the area over 232 plots, whereas ferns and bryophytes were most notable in the clearcut, with 27% and 23% occupation over 305 plots, respectively (Table 9). Deciduous shrubs had the least coverage in the clearcut, and deciduous shrubs and grasses/sedges had the least coverage in the old-growth, all of which occupied 2% or less of the non-crop vegetation coverage in the given stand type (Table 9). Of vegetation most likely to compete with the small trees, Blechnum spicant (L.) Roth and Gaultheria shallon Pursh. occurred most frequently in the plots (426 and 476 occurrences respectively out of the 537 plots) (Table 26). However, total coverage, which is the frequency of occurrence of the species weighted by the percentage of space taken up in each plot, of B. spicant was almost twice as great as that of G. shallon (23% versus 13% total coverage respectively) (Table 26). Table 9. Occurrence of non-crop vegetation groups throughout study area. Vegetation group % coverage over all plots in % coverage over all plots in jCLEARCUT (n=305) OLD-GROWTH (n =232) bryophyte 23 62 deciduous shrub 2 1 evergreen shrub 17 12 fern 27 19 forb 19 4 grass-sedge 12 2 The vegetation groups were reclassified as ordinal data (Table 10) for regression analysis. Although four categories were recognized, they had to be reclassified into two vegetation cover classes to overcome the difficulty of regressing sparse data in some of the classes. Class 0 23 represents no or sparse coverage by the vegetation group in a given plot. Class 1 represents moderate or heavy coverage. Table 10. Non-crop vegetation cover classes and categories. Vegetation cover categories were based on the percentage occupancy of the plot area by the vegetation group. Vegetation Percentage of total plot area occupied by the vegetation Vegetation cover cover class group (Vegetation cover category) category description 0 0 No coverage 0 less than 30 Sparse coverage 1 between 30 and 50 Moderate coverage 1 greater than 50 Heavy coverage No vegetation cover was noted on four plots in the study area. Bryophytes and evergreen shrubs had greater presence in the plots than any other vegetation group (Table 11). About 10% of the plots did not have bryophytes or evergreen shrubs. Grasses and sedges, and deciduous shrubs, occupied the least number of plots - only 25% (approx.) of the plots contained any grasses or sedges or deciduous shrubs. Next to bryophytes, ferns had the most moderate to heavy coverage, in at least 20% of plots. For most vegetation groups, the most frequent vegetation cover category was sparse coverage. Table 11. Relative occurrence of vegetation cover categories for each vegetation group over all plots in the study. Frequency represents the number of plots which contained a given vegetation cover category. N=533. Vegetation Group Relative frequency Relative frequency Relative frequency of Relative frequency of no coverage of sparse coverage moderate coverage of heavy coverage Bryophyte 42 328 48 115 Deciduous Shrub 388 142 3 0 Evergreen Shrub 51 447 28 7 Fem 107 324 55 47 Forb 134 373 12 14 Grass and Sedge 398 99 23 13 24 3. Results 3.1 General nature of small tree data This section provides some broad descriptions of small tree height, age and abundance data. The skewed, sparse and variable nature of the data is illustrated for each stand type. Some rough estimates on the stocking status (BC MOF 1995) of the study area are also provided. 3.1.1 Small tree height and age profiles by stand type, for Block, reproductive origin, species, microtopography, substrate, soil moisture and nutrient regime, and non-crop vegetation cover Shapiro-Wilk normal probability p values indicated non-normality in the data reported in Table 12 through Table 18. Given that the standard deviation exceeds the means of all small tree height and age variables (Table 12 through Table 18), medians are more meaningful descriptors for trends in height (since the height data are decimal numerals), whereas mode is useful to describe trends in age (since the age data are whole numerals). In general, height medians were similar across all Blocks (about 4- or 5-cm), regardless of stand type, with the exception of West Block in the clearcut, which had a median height of about 8 cm (Table 12). Modes for ages were similar in all Blocks regardless of stand type (about one and two years of age) (Table 12). 25 Table 12. Summary of small tree height and age on each Block within each stand type. Statistics Key: max=maximum mm=nunirnum p value=normal probability associated with Shapiro-Wilk statistic std dev=standard deviation. Variable Stand type Block # plots # trees mean std dev p value max median min mode HEIGHT (cm) clearcut West 180 850 17.7 21.9 0.00 123.0 8.3 0.6 5.0 Flat 198 12.7 16.0 0.00 85.0 5.5 0.9 3.5 South 358 11.5 18.0 0.00 170.0 5.0 0.3 4.0 old-growth West 111 823 8.6 11.8 0:00 91.0 5.0 0.3 2.0 Flat 693 10.8 16.4 0.00 150.0 4.8 0.1 1.7 South 428 9.5 16.9 0.00 110.0 3.6 0.5 1.7 AGE (years) clearcut West 71 850 5.4 7.8 0.00 89.0 3.0 1.0 2.0 Flat 197 5.6 7.9 0.00 58.0 3.0 1.0 2.0 South 358 4.7 10.2 0.00 100.0 2.0 1.0 2.0 old-growth West 71 823 4.0 4.9 0.00 66.0 3.0 1.0 2.0 Flat 693 6.6 13.1 0.00 119.0 3.0 1.0 1.0 South 428 4.9 10.5 0.00 95.0 2.0 1.0 1.0 HEIGHT (cm) clearcut all Blocks 305 1406 15.4 20.4 0.00 170.0 6.8 0.3 4.0 old-growth all Blocks 232 1944 9.6 14.8 0.00 150.0 4.5 0.1 2.5 AGE (years) clearcut all Blocks 305 1405 5.2 8.5 0.00 100.0 3.0 1.0 2.0 old-growth all Blocks 232 1944 5.1 9.8 0.00 119.0 3.0 1.0 1.0 Regardless of stand type, about 94% of all small trees were seedlings, compared with 6% veglings. Height medians for a given reproductive origin were always greater in the clearcut than the old-growth; however, height medians were always higher for veglings (at least five fold) than for seedlings, regardless of stand type (Table 13). Modal age for veglings was nine years in the old-growth compared with five in the clearcut, whereas seedling age modes were similar at one and two years, respectively (Table 13). Regardless of stand type, over 70% of the small trees were Cw, over 20% were Hw, and about 4% were other species (PI, Ss, Yc, and Tw). For this reason, this thesis focuses primarily on Cw and Hw. In the clearcut, the median height was higher for Hw than Cw (about 24 versus 6 cm, respectively), whereas in the old-growth, the differences were not as pronounced (5.4 versus 4, respectively) (Table 14). Modal age differences were more pronounced in the clearcut than the old-growth, with Hw older than Cw (5 versus 2 years respectively) in the clearcut, and about the same age in the old-growth (2 versus 1 year respectively) (Table 14). 26 Table 13. Summary of small tree height and age for each reproductive origin within each stand type. Statistics Key: max=maximum min=nunimum p value=normal probability associated with Shapiro-Wilk statistic variable stand type origin # trees mean std dev p value max median min mode HEIGHT (cm) clearcut seedling 1318 13.5 18.0 0.00 122.0 6.3 0.3 4.0 vegling 84 41.8 27.4 0.00 123.0 33.5 3.5 30.0 old-growth seedling 1831 7.9 12.0 0.00 110.0 4.2 0.1 2.5 vegling 107 34.9 22.8 0.00 100.0 27.0 4.5 23.0 AGE (years) clearcut seedling 1317 4.6 7.6 0.00 91.0 3.0 1.0 2.0 vegling 84 13.3 14.1 0.00 100.0 8.0 1.0 5.0 old-growth seedling vegling 1831 107 3.9 22.9 6.2 22.6 0.00 95.0 0.00 110.0 2.0 12.0 1.0 3.0 1.0 9.0 Table 14. Summary of small tree height and age for each species within each stand type. Statistics Key: max=maximum nun=minimum p yalue=normal probability associated with Shapiro-Wilk statistic std dev=standard deviation. Species key: Cw=western redcedar Hw=western hemlock Other=lodgepole pine, Variable Stand type Species # trees mean std dev p value max median min mode HEIGHT (cm) clearcut Cw 1073 10.5 14.1 0.00 170.0 5.9 0.6 4.0 Hw 278 32.8 29.3 0.00 123.0 23.7 0.3 1.5 Other 55 22.2 18.9 0.00 75.0 16.8 2.6 11.0 old-growth Cw 1371 7.2 10.5 0.00 91.0 4.0 0.1 2.0 Hw 498 13.0 18.3 0.00 110.0 5.4 0.3 5.0 Other 75 29.5 29.1 0.00 150.0 22.6 0.7 27.0 AGE (years) clearcut Cw 1073 3.7 4.5 0.00 58.0 2.0 1.0 2.0 Hw 277 10.6 15.3 0.00 100.0 6.0 1.0 5.0 Other 55 7.7 8.3 0.00 37.0 4.0 1.0 2.0 old-growth Cw 1371 4.2 7.6 0.00 110.0 2.0 1.0 1.0 Hw 498 6.2 11.1 0.00 95.0 3.0 1.0 2.0 Other 75 16.3 22.1 0.00 119.0 8.0 1.0 1.0 Height medians in the clearcut appear to be similar albeit slightly lower (about 6 cm) on depressions/flats than on inclines/mounds (about 8 cm), and similar (about 5 cm) on both microtopography groups in the old-growth (Table 15). Within each stand type, mode ages are similar regardless of microtopography (2 years in the clearcut and 1 year in the old-growth). In the clearcut, the highest median height was 7.5 cm on Lignomor and upland organic substrates, with all others similar at about 6.3 cm, except for undecomposed wood on which it was 3.6 cm (Table 16). In the old-growth, the highest median height was considered to be 5 cm on Lignomor 27 substrates.11 Mode age (2 years) was consistent on all substrates in the clearcut except on undecomposed wood, where the mode age was 1, and mode age was 1 on all substrates (not including mineral) in the old-growth, except for Lignomor, which had a mode age of 2 (Table 16). Table 15. Summary of small tree height and age on each microtopography within each stand type. Statistics Key: max=maximum min=minimum p value=normal probability associated with Shapiro-Wilk statistic std dev= standard deviation. variable stand type microtopography # trees mean std dev p value max median min mode HEIGHT (cm) clearcut depression/flat 715 15.1 20.9 0.00 122.0 6.0 0.6 4.0 incline/mound 691 15.8 19.9 0.00 170.0 8.0 0.3 5.0 old-growth depression/flat 1204 10.4 15.5 0.00 150.0 5.0 0.1 5.0 incline/mound 740 8.3 13.3 0.00 109.0 4.0 0.3 2.0 AGE (years) clearcut depression/flat 714 5.4 9.7 0.00 91.0 3.0 1.0 2.0 incline/mound 691 5.1 7.0 0.00 100.0 3.0 1.0 2.0 old-growth depression/flat 1204 5.5 10.2 0.00 119.0 3.0 1.0 1.0 incline/mound 740 4.6 9.2 0.00 106.0 2.0 1.0 1.0 1 1 A 25 cm median height on mineral substrate was based on a sample size of 4, therefore could not be, with certainty, considered the highest. 28 Table 16. Summary of small tree height and age on each substrate within each stand type. Statistics Key: max=maximum value min=minimum value p value=normal probability associated with Shapiro-Wilk statistic std dev=standard deviation. Substrate key: Lig=Lignomor Min=mineral Mos=moss variable stand type substrate # trees mean std dev p value max median min mode group HEIGHT (cm) clearcut Lig 460 18.8 23.6 0.00 123.0 7.5 0.3 4.0 Min 287 9.3 9.3 0.00 69.0 6.3 0.6 9.0 Mos 117 11.7 15.6 0.00 110.0 6.2 1.0 6.2 Ucw 16 10.5 16.2 0.00 64.0 3.6 0.9 1.8 Upl 231 18.6 24.8 0.00 170.0 7.5 0.6 4.5 Wet 293 14.8 18.7 0.00 107.0 6.4 0.6 3.5 old-growth Lig 931 10.9 15.5 0.00 110.0 5.0 0.3 5.0 Min 4 28.4 23.2 0.39 56.0 24.9 7.6 7.6 Mos 512 6.7 10.0 0.00 88.0 3.8 0.5 2.0 Ucw 4 4.0 3.9 0.06 9.8 2.5 1.4 1.4 Upl 226 11.1 18.0 0.00 87.0 3.1 0.8 1.7 Wet 267 9.1 15.8 0.00 150.0 4.5 0.1 5.0 AGE (years) clearcut Lig 459 6.4 10.9 0.00 100.0 3.0 1.0 2.0 Min 287 2.6 1.6 0.00 11.0 2.0 1.0 2.0 Mos 117 4.3 6.3 0.00 56.0 3.0 1.0 2.0 Ucw 16 3.3 3.5 0.00 13.0 2.0 1.0 1.0 Upl 231 5.5 7.9 0.00 66.0 3.0 1.0 2.0 Wet 293 6.0 8.6 0.00 82.0 3.0 1.0 2.0 old-growth Lig 931 5.2 9.1 0.00 110.0 3.0 1.0 2.0 Min 4 31.3 50.0 0.01 106.0 8.0 3.0 3.0 Mos 512 3.9 5.9 0.00 65.0 2.0 1.0 1.0 Ucw 4 1.8 1.5 0.00 4.0 1.0 1.0 1.0 Upl 226 6.8 12.8 0.00 72.0 2.0 1.0 1.0 Wet 267 5.6 13.0 0.00 119.0 3;0 1.0 1.0 Slightly less than 30% of the small trees in the clearcut occurred on each of fresh, moist and very moist SMRs, compared with slightly less than 20% on wet regimes. The highest median height in the clearcut occurred on fresh and moist regimes (about 8 cm), and median heights on the remaining regimes were about 6 cm (not including slightly dry regimes, which had a sample size of 5) (Table 17). Whereas 71% of small trees in the clearcut occurred on poor SNRs (followed by 21% on mediuth, and 8% on rich regimes), the highest median height occurred on medium regimes (9.0 cm), then on poor regimes (6.8 cm), and lastly on rich regimes (3.9 cm) (Table 17). In the old-growth 65% of the small trees occurred on moist SMRs, followed by 24% 29 on wet regimes and 11% on very moist regimes. The median heights were similar on all SMRs in the old-growth, between 4 and 5 cm (Table 17). Similarly, median heights were about 4.5 cm on both poor and medium SNRs (excluding rich, which had a sample size of 1) in the old-growth (Table 17). Modal age in the clearcut was about two years for all SMRs and SNRs (not including slightly dry or rich, which had negligible sample sizes), whereas in the old-growth it was about one, except on moist and medium regimes (which had modal ages of 2 years) (Table 17). Generally, median heights were slightly higher (by about 1-3 cm) in the clearcut versus the old-growth for a given non-crop vegetation cover class regardless of non-crop vegetation group (Table 18). Median heights were generally similar for all non-crop vegetation groups, ranging between about 6-9 cm in the clearcut, and about 3-5 cm in the old-growth, being slightly higher under no/sparse non-crop coverage compared with moderate/heavy coverage. Notable exceptions include a higher median height under moderate/heavy (at 21.6 cm) compared with no/sparse (at 6.8 cm) deciduous coverage in the clearcut; however, this was based on a sample size of 10, therefore cannot be considered conclusive. Median heights were also slightly higher (only about 1-2 cm) under moderate/heavy evergreen shrub and fern coverage, compared with no/sparse coverage in the clearcut. Age modes for small trees under any vegetation cover were generally 2 in the clearcut, and dropped to about 1 in the old-growth. Any greater height and age modes were part of very small samples. 30 Table 17. Summary of small tree height and age for each soil moisture and nutrient regime within each stand type. Statistics Key: max=maximum min=minimum p value=normal probability associated with Shapiro-Wilk statistic std dev=standard deviation. Soil moisture regime key: SMR=soil moisture regime refresh m=moist sd=slightly dry vm=very moist w=wet. Soil nutrient regime key: SNR=soil nutrient regime b=nutrient poor c=nutrient medium d=nutrient rich. regime type variable stand type regime # trees mean std dev p value max median min mode SMR HEIGHT (cm) clearcut sd 5 2.7 0.6 0.78 3.4 2.7 1.8 1.8 f 380 13.8 16.2 0.00 110.0 8.0 0.6 5.0 m 411 21.0 26.5 0.00 170.0 8.5 0.3 1.5 vm 365 11.6 15.9 0.00 107.0 5.6 0.6 2.9 w 243 14.2 17.8 0.00 93.0 6.0 1.0 3.5 old-growth m 1256 8.8 14.0 0.00 110.0 4.1 0.3 5.0 vm 217 12.6 17.3 0.00 100.0 4.9 0.7 1.5 w 471 10.2 15.2 0.00 150.0 5.1 0.1 2.5 AGE (years) clearcut sd 5 1.2 0.4 0.00 2.0 1.0 1.0 1.0 f 380 4.5 5.7 0.00 66.0 3.0 1.0 2.0 m 411 6.7 11.1 0.00 100.0 3.0 1.0 2.0 vm 364 4.8 8.7 0.00 84.0 2.0 1.0 2.0 w 243 4.5 5.3 0.00 47.0 3.0 1.0 2.0 old-growth m 1256 4.6 8.6 0.00 106.0 2.0 1.0 2.0 vm 217 6.9 11.6 0.00 110.0 3.0 1.0 1.0 w 471 5.8 11.7 0.00 119.0 3.0 1.0 1.0 SNR HEIGHT (cm) clearcut b 996 16.9 22.5 0.00 170.0 6.8 0.3 4.0 c 290 13.8 14.0 0.00 94.0 9.0 0.6 6.0 d 118 5.5 4.8 0.00 29.2 3.9 0.6 3.0 old-growth b 1913 9.6 14.8 0.00 150.0 4.5 0.1 5.0 c 30 8.7 7.9 0.00 26.4 4.6 1.7 1.7 d 1 7.5 7.5 7.5 7.5 7.5 AGE (years) clearcut b 995 6.0 9.7 0.00 100.0 3.0 1.0 2.0 c 290 3.9 3.4 0.00 19.0 3.0 1.0 2.0 d 118 2.1 0.9 0.00 5.0 2.0 1.0 2.0 old-growth b 1913 5.2 9.9 0.00 119.0 3.0 1.0 1.0 c 30 2.9 2.0 0.00 8.0 2.0 1.0 2.0 d 1 4.0 4.0 4.0 4.0 4.0 31 Table 18. Summary of small tree height and age for non-crop vegetation presence within each stand type. Statistics Key: max=maximum min=mimmum p value=normal probability associated with Shapiro-Wilk statistic std dev=standard deviation. Non-crop vegetation key: veg group=non-crop vegetation group veg cover=non-crop vegetation quadrat plot coverage BRYO=bryophyte DECI=deciduous shrub EVER=evergreen shrub FERN=fern FORB=forb GRSD=grass/sedge. No/sparse coverage represents less than 30% above-ground occupation of a given vegetation group within plot. Mod/heavy coverage represents 30% or greater above-ground occupation of a given vegetation group within plot. veg group variable stand type veg cover # trees mean std dev p value mas median min mode B R Y O HEIGHT (cm) clearcut no/sparse mod/heavy old-growth no/sparse mod/heavy 1215 191 ""635" 1309 15.8 12.9 9.6 20.9 16.7 "l4.0" 15.1 0.00 0.00 "6V66" 0.00 170.0 92.0 Ti'6'.o" 150.0 AGE (years) clearcut no/sparse mod/heavy old-growth no/sparse mod/heavy 1214 5.2 8.2 191 5.7 10.3 ""635"4.76.8" 1309 5.4 11.0 0.00 100.0 0.00 84.0 "O.OO66.0" 0.00 119.0 7.0 5.7 "5.0" 4.1 3.0 3.0 T o " 3.0 0.3 0.9 "0.3" 0.1 1.0 1.0 T 6 " 1.0 4.0 4.5 T 6 " 2.5 2.0 2.0 T o 1.0 D E C I HEIGHT (cm) clearcut AGE (years) no/sparse mod/heavy old-growth no/sparse clearcut no/sparse mod/heavy old-growth no/sparse 1396 15.4 20.4 10 19.6 5.8 "l9449.6i '4 .8" 1395 5.3 8~5~ 10 4.2 1.0 7 9 4 4 5 . T 9 ' T 0.00 170.0 0.06 27.0 T o o 1 5 0 . 6 " T o o 16T6" 0.01 5.0 T o o - " " "ii'9.6" 6.8 21.6 " T T 4.5 " T o " 0.3 8.3 T F T o " 2.0 T o " 4.0 21.6 " " ' i l ' " T o ' 5.0 " T o E V E R HEIGHT (cm) clearcut AGE (years) no/sparse mod/heavy old-growth no/sparse mod/heavy clearcut no/sparse mod/heavy old-growth no/sparse mod/heavy 1308 15.1 19.8 98 19.8 26.3 " 1 8 4 9 9 J 1 4 . 8 " 95 8.0 14.3 1307 5.2 8 T 98 6.4 6.5 "18495 .29 .9 '" 95 4.2 8.0 0.00 170.0 0.00 118.0 T o o 1 5 0 . 0 " 0.00 109.0 T 6 o - ~ i 6 T 6 ~ 0.00 35.0 T o o l ' T ' o ' 0.00 60.0 6.8 7.5 .......... 4.4 T o " 3.0 "3.6" 3.0 0.3 1.5 T i " 0.7 T o " 1.0 T 6 " 1.0 4.0 5.0 "i'.'o 1.5 T o " 3.0 " T o 1.0 F E R N HEIGHT (cm) clearcut AGE (years) no/sparse mod/heavy old-growth no/sparse mod/heavy clearcut no/sparse mod/heavy old-growth no/sparse mod/heavy 1198 15.0 20.0 208 17.9 22.4 "i '74'39.914*7" 201 7.1 14.8 lf97 5.1 8~F 208 5.9 10.3 " i ' 7 4 3 5 . 3 l O . ' o " 201 3.6 7.9 0.00 170.0 0.00 122.0 'Too1 5 0 . 0 " 0.00 110.0 T o o - T 6 0 T " 0.00 91.0 "6.60'i'i'9.6" 0.00 66.0 6.5 8.7 "4.5" 3.8 T o " 3.0 "3.'6" 2.0 0.3 0.6 T i " 0.5 T o " 1.0 T o " 1.0 4.0 3.0 ...„.„. 5.0 T o ' 2.0 "i'.'o 2.0 F O R B HEIGHT (cm) clearcut AGE (years) no/sparse mod/heavy old-growth no/sparse clearcut no/sparse mod/heavy old-growth no/sparse 1313 15.3 20.2 93 16.4 22.6 " i ' 9449 .6 ' lTF 1312 5.1 8~5~ 93 6.8 8.1 "1944"5"'i9.8'" 0.00 170.0 0.00 122.0 ' T o o 1 5 0 . 6 " "6~oo TobTj -0.00 47.0 o . o o i ' i ' 9 . 6 " 6.9 6.2 "4.5" T o " 3.0 "3.6" 0.3 1.8 T T " T o " 1.0 "i'.'o" 4.0 4.0 '2.5' T o " 2.0 "i'.'o continued... 32 variable stand type veg cover # trees mean std dev pvalue max median min mode Table 18 continued... HEIGHT (cm) clearcut no/sparse 1303 15.4 20.6 0.00 170.0 6.8 0.3 4.0 mod/heavy 103 15.2 16.7 0.00 85.0 6.4 0.9 3.2 old-growth no/sparse 1943 9.6 14.8 0.00 150.0 4.5 0.1 2.5 mod/heavy 1 17.7 17.7 17.7 17.7 17.7 AGE (years) clearcut no/sparse 1303 5.1 8.5 0.00 100.0 3.0 1.0 2.0 mod/heavy 102 6.6 8.6 0.00 58.0 3.0 1.0 2.0 old-growth no/sparse 1943 5.1 9.8 0.00 119.0 3.0 1.0 1.0 mod/heavy 1 7.0 7.0 7.0 7.0 7.0 3.1.2 Small tree abundance The mean number of trees per plot in the clearcut was 4.6 compared to 8.4 in the old-growth; however, the standard deviations of the samples in both stand types were higher than the means (Table 19). More plots in both stand types had zero trees per plot (mode=0 for both strata), compared with all other tree-per-plot frequencies. However, when comparing no trees per plot, to any number of trees per plot, 70% of the plots in the clearcut had trees in them, compared to 60% in the old-growth. Nineteen percent of the old-growth plots contained more than 10 trees per plot, compared with 9% of the clearcut plots (Table 20). Chi-Square tests cannot be applied to these data to determine whether differences in per plot frequencies could be detected among the sample populations, since 82% of the cells contained frequencies of less than five (SAS Institute Inc. 1994a; Conover 1980; Agresti 1996). Table 19. Univariate statistics describing the number of small trees per plot by stand type. Variable = number of small trees per plot Stand type Statistical parameter Clearcut Old-growth N 305 232 Mean 4.6 8.4 Std Dev 9.8 21.9 Skewness 5.3 6.3 Kurtosis 36.6 50.8 W: Normal Q?r<W) 0.49 (0.0001) 0.43 (0.0001) Max 96 222 Med 2 1 Min 0 0 Mode 0 0 33 Table 20: Small tree (per plot) abundance by stand type. Small tree abundance 0 1 2 3 4 5 6 7 8 9 10 11-222 (number of trees per plot): Stand type Percentage of plots with given small tree abundance by stand type clearcut (n=305) 30 17 13 7 6 4 4 3 4 1 2 9 old-growth (n=232) 41 10 9 7 4 3 1 2 2 1 1 19 3.1.3 Estimates of the numbers of taller potential future crop trees per ha Most of the taller potential future crop (PFC) trees in the clearcut (and all of them in the old-growth) were at least 5-years in age; therefore, most of the regeneration in the clearcut was attributable to advanced regeneration (Table 21). Only 131 taller PFC trees/ha in the clearcut were estimated to be younger (ingress), whereas none in the old-growth were estimated to be younger (Table 21). Table 21. Estimated number of taller potential future crop trees/ha in study area by stand type. Shown are estimates of the number of younger taller PFC trees versus older taller PFC trees per ha (rounded to nearest 1). *Estimated number of younger taller PFC trees per ha was based on the assumption that 1 younger taller PFC tree per plot will consume the growing space of that plot and was calculated by the following formula: (c+a) younger taller PFC trees/m2xl0000 m2/ha. **Estimated number of older taller PFC trees per ha was based on the assumption that 1 older taller PFC tree per plot will consume the growing space of that plot, and was — i — ' . i — ' - 1 . i - - 11 ' f 1 . . . / i - - \ i i— — r>rr/~< . / 2 .. lnnnn 2 //. Formula Parameter CLEARCUT OLD-Label GROWTH a Number of Plots (a=b+c+d+e+j) 305 232 b Number of Plots with no trees 93 94 c Number of plots with at least 1 younger taller small tree (including 4 0 plots with older taller small trees, older mini small trees and younger mini small trees) d Number of plots with at least 1 older taller small tree (including plots 89 55 with older mini small trees and younger mini small trees, in the absence of any younger taller small trees) e Number of plots with at least 1 older mini small tree (including plots 37 31 with younger mini small trees, in the absence of younger taller small trees and older taller small trees) f Number of plots with only younger mini small trees 82 52 'Estimated number of younger taller PFC trees per ha 131 0 (ingressj **Estimated number of older taller PFC trees per ha 2920 2370 (advanced regeneration^  34 3.2 Chi-Square test results for the influence of plot contiguity on small tree presence, age and height distributions According to Chi-Square tests, plot contiguity, excluding aii other factors, influenced some of the distributions at a significance level of p<0.05 (Table 22). For small tree presence distributions, plot contiguity was significant only in the old-growth of Flat Block (p=0.010). For small tree height distributions, plot contiguity was significant in all Blocks, with the exception of West Block, old-growth (p=0.361). For small tree age distributions, plot contiguity was significant in West Block, clearcut (p=0.001), Flat Block, old-growth (p=0.007), and South Block, old-growth (p=0.001). Despite these findings, all plots whether contiguous or non-contiguous, were included in the regressions, because of sample size constraints, and the influence of plot location on the variability of the data had to be considered when interpreting the results. Table 22. Results of Chi-Square tests for the influence of plot contiguity on small tree presence, height and age distributions. Response variable Block Stand type Pearson Chi-Square p-value small tree presence West clearcut 5.2 0.076 (presence versus no presence) West old-growth 5.0 0.084 Flat clearcut 1.0 0.325 Flat old-growth 6.7 0.010 South clearcut 0.7 0.387 South old-growth 0.9 0.339 small tree height distributions West clearcut 62.4 0.001 (mini versus taller small trees) West old-growth 2.0 0.361 Flat clearcut 5.0 0.025 Flat old-growth 6.6 0.010 South clearcut 5.3 0.022 South old-growth 9.6 0.002 small tree age distributions West clearcut 40.0 0.001 (younger versus older small trees) West old-growth 4.0 0.137 Flat clearcut 0.0 0.917 Flat old-growth 7.3 0.007 South clearcut 2.3 0.133 South old-growth 16.5 0.001 35 3.3 Stepwise logistic regression results for small tree presence, height and age classes This chapter summarizes the most noteworthy results of small tree presence, height and age class regressions, by stand type, in relation to Block, plot contiguity, species (height and age class regressions only), reproductive origin (height and age class regressions only), stand edge distance (clearcut only), microtopography, substrate, SMR, SNR, and non-crop vegetation. 3.3.1 Model adequacy for small tree presence, height and age class regressions Model acceptance statistics (Table 23) indicate only marginal model fits for all regressions. They were marginal, since each regression did not satisfy all of the minimum requirements for model adequacy described in Section 2.4. Strong models would have had Hosmer-Lemeshow p-values and concordance, specificity and sensitivity values exceeding 0.9. with maximum rescaled R2 values exceeding 0.8. In this study, this was not the case. Maximum rescaled R2values generally hovered between 0.35 and 0.4 for all regressions. 36 Table 23. Model acceptance statistics for small tree presence, height and age class regressions. Variables included in the models are listed in Table 24 Statistic or Measure i Value p-value DF | Value p-value DF PRESENCE CLASS Old-growth J Clearcut Hosmer-Lemeshow GF ! 11.9 0.16 8 6.5 0.60 8 -2 LOG L 146.7 0.0001 14 293.5 0.0001 13 Concordant (%) i 81.5 j 82.9 Sensitivity (%) ! 74.5 0.350 (cutpoint) J 82.5 0.350 (cutpoint) Specificity (%) 72.0 0.350 (cutpoint) 69.7 0.350 (cutpoint) Max Rescaled R 2 j_0.38 [_0.43 HEIGHT CLASS Old-growth j Clearcut Hosmer-Lemeshow GF i 10.9 0.14 7 j 11.2 0.19 8 - 2 L O G L ! 292.5 0.0001 11 i 319.9 0.0001 16 Concordant (%) ! 85.0 j 84.2 Sensitivity (%) 74.3 0.1 (cutpoint) 83.7 0.1 (cutpoint) Specificity (%) i 84.6 0.1 (cutpoint) ! 74.4 0.1 (cutpoint) Max Rescaled R 2 i 0.34 ! 0.36 AGE CLASS Old-growth ! Clearcut Hosmer-Lemeshow GF ! 7.8 0.45 8 ! 11.0 0.14 7 -2 LOG L 533.3 0.0001 20 441.9 0.0001 18 Concordant (%) 78.4 82.3 Sensitivity (%) ! 75.6 0.25 (cutpoint) ! 73.1 0.25 (cutpoint) Specificity (%) j 67.5 0.25 (cutpoint) j 80.5 0.25 (cutpoint) Max Rescaled R 2 ! 0.35 | 0.39 3.3.2 Variables detected to influence small tree presence, height and age class distributions, according to selected logistic regression models Each model represents a set of variables acting together to influence small tree presence, height, or age class distributions. The odds ratios are the given likelihood of a small tree being present, taller, or older (according to the particular regression), given the influence of the other variables in the model Small tree presence, height and age class models were all influenced by Blocks, and three out of the six models were affected by plot contiguity (Table 24). Although the Block and plot contiguity effects, and their interactions with other variables, were important, they are not further described other than in Table 24, since they were not objects of interest in this study. 37 Table 24. Variables that influence small tree presence, height and age class, according to selected logistic regression models. Variables are ranked according to Wald Chi-Square statistic (from highest to lowest) given in the final step of the regression. Variables with Chi-Square p-values <0.05 are presented in normal font, and are deemed to contribute significantly to the resulting model. Variables with Chi-Square values =>0.05 are presented in italics, and are deemed to render the model less stable. Data have been rounded to first digit after the decimal. Note that odds ratios presented for dummy variables are not true odds ratios. General key: CL=confidence limit IM:DF=incline/mound:depression/flat microtopography int.=interaction NC:C=non-contiguous:contiguous plots SED=stand edge distance SMR=soil moisture regime SNR=soil nutrient regime Veg:Seed=vegling:seedling. Block key: Flat=Flat Block South=South Block West=West Block. Non-crop vegetation presence key: BRYObryophyte DECI=deciduous shrub EVER=evergreen shrub FERN=fern GRSD=grass/sedge. Species key: Cw=western redcedar Hw=western hemlock Other=species other than Cw or Hw. Substrate group key: Lig=Lignomor Min=mineral Mos=moss Ucw=undecomposed coarse wood Upl=upland organic Wet=wetlapd organic. Variable Wald Chi- Odds Upper/ Variable Wald Chi- Odds Upper/ Square p- Ratio Lower Square p- Ratio Lower value 95% CL value 95% CL PRESENCE C1ASS Old-srowth Clearcut Lig 0.0001 4.8 3.2/7.2 Lig 0.0001 2.9 2.2/3.9 Upl 0.0001 2.3 1.7/3.3 Wet 0.0001 1.9 1.5/2.4 WestFlat 0.0001 0.1 0.1/0.3 Mos 0.0001 2.4 1.7/3.5 Mos 0.0001 2.6 1.8/3.9 Min 0.0001 2.4 1.7/3.6 Wet 0.0001 2.1 1.5/2.9 Upl 0.0001 1.8 1.4/2.3 Min 0.002 3.6 1.6/8.1 Ucw 0.0001 0.4 0.3/0.7 West:Flat x IM:DF 0.01 0.4 0.2/0.8 SED x IM:DF 0.002 0.5 0.37 0.8 GRSD 0.02 0.1 0.01/0.6 SED 0.007 1.5 1.1 /2.0 SoutkFlat 0.03 0.3 0.1/0.9 IM:DF x SNR int. 0.01 2.1 1.2/3.8 SNR 0.1 0.5 0.2/1.1 FERN 0.03 0.6 0.4/0.9 West-Flat x NC:C int. 0.1 1.9 0.9/3.8 West-.Flat 0.1 0.6 0.3/1.1 South.Flat x NC:C int. 0.1 0.5 0.2/1.2 GRSD 0.1 0.6 0.3/1.2 EVER 0.2 0.5 0.2/1.4 South:Flat 0.9 0.9 0.5/1.9 South.Flat x M.-DF 0.1 1.1 0.5/2.6 HEIGHT CLASS Old-erowth Clearcut Veg:Seed 0.0001 13.6 8.1/23.1 Hw:Cw 0.0001 6.8 4.4/10.5 HwCw 0.0001 5.7 3.6/9.0 Veg:Seed 0.0001 8.1 4.4/14.7 OthenCw 0.0001 7.1 3.7/13.9 SNR 0.0001 0.04 0.02/0.1 South:Flat x FERN int. 0.0001 0.01 0.003 / 0.07 Other:Cw 0.0001 9.2 4.0/20.8 IM:DF 0.0001 0.3 0.1/0.5 SNR x M : D F int. 0.0001 6.8 3.0/15.5 West:Flat x FERN int. 0.0001 19.0 4.5/78.6 West:Flat 0.0001 4.4 2.1/9.4 South:Flat x NC:C int. 0.0001 10.6 3.2 / 34.5 SMR x M : D F int. 0.0004 0.3 0.1/0.6 West:Flat 0.0002 0.1 0.02/0.3 Lig x SNR int. 0.0007 7.8 2.4/25.8 West:FlatxNC:Cint. 0.006 5.8 1.6/20.8 West:FlatxNC:Cint. 0.001 0.4 0.2/0.7 SMR 0.007 0.4 0.2/0.7 |Lig 0.003 0.1 0.02/0.5 South:Flat 0.02 0.2 0.1/0.8 OtherCw x Upl int. 0.02 0.2 0.03/0.8 SED x Wet int. 0.03 0.6 0.3/0.9 Wet x SNR int. 0.05 2.6 1.0/6.7 1 Hw:Cw x Upl int. 0.2 0.5 0.2/1.5 \South:Flatx NC.C int. 0.5 0.8 0.3/1.7 J South:Flat 0.6 0.8 0.4/1.8 continues... 38 Variable Wald Chi- Odds Upper/ Variable Wald Chi- Odds Upper/ Square p- Ratio Lower Square p- Ratio Lower value 95% CL I value 95% CL Table 24 continued... AGE CLASS Old-growth Clearcut Veg:Seed 0.0001 128.9 43.7/380.2 Veg:Seed 0.0001 16.7 7.8/35.9 South:FlatxNC:Cint. 0.0001 22.4 7.4/68.3 i Min 0.0001 0.1 0.1/0.2 South:Flatx FERN int. 0.0001 0.06 0:02/0.2 West:Flat x IM:DF int. 0.0001 5.0 3.1/8.2 Hw:Cw 0.0001 2.5 1.7/3.8 Hw:Cw 0.0001 28.3 8.9/89.7 BRYO x SMR int. 0.0001 1.9 1.5/2.6 SMR 0.0002 0.5 0.3/0.7 Wet 0.0009 0.5 0.3/0.7 Upl x SNR int. 0.005 0.6 0.4/0.8 South:Flat 0.001 0.02 0.001 / 0.2 DECI 0.007 0.1 0.03/0.6 Mos x IM:DF int. 0.001 0.4 0.2/0.6 Lig x EVER int. 0.008 0.4 0.2/0.8 West:Flat x Lig int. 0.01 2.5 1.2/5.0 Hw:Cw x Upl int. 0.009 0.3 0.1/0.7 Lig x EVER int. 0.02 3.4 1.6/7.3 Wet x GRSD int. 0.01 2.6 1.2/5.5 Hw:Cw x BRYO int. 0.03 0.5 0.3/0.9 OthenCw 0.02 8.7 1.3/56.3 South:Flat x Upl int. 0.03 15.3 1.3/178.5 Hw:Cw x SNR int. 0.03 0.3 0.1/0.9 Other:Cw 0.03 4.6 1.2/18.3 Lig x IM:DF int. 0.04 0.6 0.3/1.0 Wet x IM:DF int. 0.03 0.1 0.01/0.9 South.Flat 0.1 0.5 0.2/1.1 West:FlatxNC:Cint. 0.04 1.7 1.0/2.6 South:Flat x M:DF int. 0.2 1.6 0.8/3.6 South:Flat x Lig int. 0.1 7.1 0.7/74.7 Other.CwxSNR int. 0.2 0.5 0.1/1.6 West:Flat x FERN int. 0.2 1.9 0.6/5.6 I Other.Cwx Upl int. 0.9 0.9 0.2/3.9 West:Flat 0.5 0.8 0.4/1.6 WesV.Flat 0.9 1.0 0.6/1.7 WesV.Flat x Upl int. 0.9 1.1 0.3/3.7 Other.CwxBRYO int. 0.9 1.1 0.2/5.0 In the clearcut model, variables which most influenced small tree presence class were Block, stand edge distance, microtopography, SNR, all six substrates, and fern presence (Table 24); in the old-growth model, they were substrates (all with the exception of undecomposed coarse wood) and grass/sedge presence. In the clearcut model, variables which most influenced small tree height class were Block, plot contiguity, stand edge distance, species, origin, microtopography, SMR, SNR, and Lignomor and upland and wetland organic substrates; in the old-growth model, they were species, origin, microtopography, Block, plot contiguity, SMR, and fern presence (through interactions with Block). In the clearcut model, variables which most influenced small tree age class were Block, species, origin, substrates (all except for undecomposed wood and moss), microtopography, SMR, SNR, and non-crop vegetation presence (evergreen and deciduous shrub, grass/sedge); in the old-growth model, they were 39 origin, species, Block, plot contiguity, microtopography, substrate (wetland and upland organic, moss, and Lignomor through interactions with Block), and non-crop vegetation (fern through interactions with Block, bryophyte, and evergreen shrub). In most cases, when individual explanatory variables were plotted against small tree presence class (see Section 3.3.3), height class (see Section 3.3.4), and age class (see Section 3.3.5), the distributions appeared to coincide generally with a positive or negative relationship12 also indicated by the odds ratio for that variable in the model. However, some of these influences, although statistically significant, were negligible when the distributions were visually examined; and, in some cases, variables which were not included in the models, appeared to influence small tree presence, height and age. Given the numerous dummy variables in the models, along with the model weaknesses, the odds ratios could not be used to interpret the precise magnitude of the relationships between the response and explanatory variables in this thesis. Even so, the influences indicated by the odds ratios were valid only under the influence of the other variables in each model. Sections 3.3.3 through 3.3.5 describe small tree presence, height and age distributions of some selected explanatory variables. 3.3.3 Small tree presence class distributions In the clearcut, small tree presence decreased slightly on inclines/mounds compared with depressions/flats, as stand edge distance increased (Figure 4); and, small trees were more prevalent on inclines/mounds located on nutrient medium/rich microsites (Figure 5). Although 1 2 This means that if the odds ratio exceeded 1, then the relationship was positive. If the odds ratio was less than 1, then the relationship was negative. If the odds ratio was equal to 1 (or very near to 1), then no relationship was apparent between explanatory and response variable. 40 SNR was not statistically significant in the old-growth model (Table 24); small tree presence decreased with increasing SNR (regardless of microtopography) (Figure 5). 45 40 35 4. 30 25 4 20 15 10 5 o 4 1 2 depression/flat CLEARCUT 3 Number of small trees > 0 n=247 n=i27 IBl l l l i i i incline/mound Figure 4. Small tree presence in the clearcut, according to microtopography and stand edge distance. Key: 1- 0<=Stand edge distance<50m 2- 50m<=Stand edge distance<100m 3- Stand edge distance=>100m. 50 -I 45 -40 • 35 -8 g 30 -25 -O 20 15 -10 5 -0 -n=412 1 Number of small trees > 0 n=234 n=U4 Figure 5. Small tree presence, by stand type, according to soil nutrient regime and microtopography. Soil nutrient regime key: vp=very poor p=poor m=medium r=rich. 41 Of the substrates, Lignomors appeared to have the strongest positive influence on small tree presence in both stand types, followed by mineral (clearcut only) and moss substrates (Figure 6). Undecomposed coarse wood substrates appeared to have the strongest negative influence on small tree presence in the clearcut (no similar observation could be made for this substrate in the old-growth given its low occurrence there) (Figure 6). 100 o •Lig. " Min. A Mos. »Ucw. •Upl. - €> - Wet. CLEARCUT OLD-GROWTH Substrate plot coverage by stand type Figure 6. Small tree presence, by stand type and substrate cover class, according to substrate group. Symbols indicate the percentage of samples within a given stand type, substrate group and cover class, which contain at least 1 small tree. Trend lines are not shown for samples sizes <10. Key: Cover class (percentage of substrate coverage within quadrat): l=0<cover<10% 2=10%<=cover<40% 3=cover=>40%. Substrate: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetland organic Of the non-crop vegetation, increased fern coverage appeared to have the most negative influence on small tree presence in both stand types (Figure 7), yet was a significant predictor only in the clearcut model (Table 24). While grass/sedge presence was significant in the old-growth model (Table 24), the sample size was low. In the clearcut, increased grass/sedge appeared to influence small tree presence negatively (Figure 7), but the relationship was not significant in the clearcut model, given the other variables (Table 24). Although deciduous shrubs were not shown by the regressions to influence small tree presence (due to sample size constraints), the percentage 42 of small trees present declined under moderate/heavy deciduous shrub presence (Figure 7). Small tree presence increased slightly under both moderate/heavy bryophyte and forb coverage in the clearcut, markedly for bryophytes in the old-growth (Figure 7). 20 — O - BRYO ~ D* DECI A — - E V E R < *"**FERN IU FORB — O - - GRSD . • (> : ,— ; 1 1 <30% =>30% <30% =>30% CLEARCUT OLD-GROWTH Non-crop vegetation plot coverage by stand type Figure 7. Small tree presence, by stand type and non-crop vegetation cover class, according to non-crop vegetation group. Symbols indicate the percentage of samples within a given stand type, vegetation cover group and cover class, which contain at least 1 small tree. Trend lines are not shown for samples sizes <10. Less than 30% non-vegetation coverage represents zero to sparse coverage. Greater than or equal to 30% coverage represents moderate to heavy coverage. The steeper the slope of the trend line, the greater the influence on small tree presence with increased non-crop vegetation coverage. Increased fern (both stand types) and grass/sedge (clearcut only) coverage appears to have the most negative impact on small tree presence, relative to the other non-crop vegetation groups. Key: BRYO=bryophyte DECI=deciduous shrub EVER=evergreen shrub FERN=fern FORB=forb GRSD=grass and sedge. 3.3.4 Small tree height class distributions In both stand types, the proportions of taller veglings exceeded 45%, whereas those proportions for seedlings were only 12% in the clearcut and 5% in the old-growth (Figure 8). The proportions of taller Hw and other species exceeded the proportions of taller Cw in both stand types (Figure 9). Proportionally fewer taller small trees were found on inclines/mounds in the old-growth (Figure 10); whereas in the clearcut, SNR and SMR appeared to influence small tree height according to microtopography. For example, proportionally more taller small trees 43 were found on inclines/mounds, compared with depressions/flats, on nutrient medium/rich quadrats; and proportionally fewer were found on inclines located on very moist/wet quadrats (Figure 11). S m a l l tree height => 30 cm 60 50 40 30 20 10 seedling C L E A R C U T vegl ing seedling O L D - G R O W T H vegling Figure 8. Percentage of small trees => 30 cm by stand type and reproductive origin. Only proportions of small trees => 30 cm in height are depicted. Not shown, but implicit, are the proportions of small trees < 30 cm in height, which can be calculated by subtracting the % Occurrence from 100% for each category. S m a l l tree height => 30 cm 45 40 35 30 25 20 15 10 5 0 n=55 n=1371 Cw C L E A R C U T Cw O L D - G R O W T H Figure 9. Percentage of small trees => 30 cm by stand type and species. Only the proportions of small trees => 30 cm in height are depicted. Not shown, but implicit, are the proportions of small trees < 30 cm in height, which can be calculated by subtracting the % Occurrence from 100% for each category. Key: Cw=westera redcedar Hw=western hemlock Other=lodgepole pine, western yew, Sitka spruce, and yellow-cedar. 44 16 n"715 Small tree height => 30 cm =fi9J 11^1204 n=740 depression/flat CLEARCUT incline/mound depression/flat OLD-GROWTH incline/mound Figure 10. Percentage of small trees => 30 cm, by stand type and microtopography. Only the proportions of small trees => 30 cm in height are depicted. Not shown, but implicit, are the proportions of small trees < 30 cm in height, which can be calculated by subtracting the % Occurrence from 100% for each category. Small tree height => 30 cm 8 O Soil moisture or nutrient regime class Figure 11. Percentage of small trees => 30 cm, by stand type and microtopography according to soil moisture and nutrient regime classes. Symbols indicate the percentage of small trees => 30 cm in height, within a given stand type, microtopography group and soil moisture or nutrient regime class. Trend lines are not shown for samples sizes <= 5. The steeper the slope of the trend line, the greater the influence of soil moisture or nutrient regime on small tree height class with change in microtopography . Key for soil moisture regime classes: l=slightly dry, fresh and moist 2=very moist and wet. Key for soil nutrient regime classes: l=very poor and poor 2=medium and rich. 45 In the old-growth model, no substrate was detected to influence small tree height distributions (Table 24) and no substantial trends were apparent in the raw data; nevertheless, upland organic substrates appeared to have a slightly higher proportion of taller small trees than Lignomor and wetland organic substrates, and moss substrates had the lowest proportions of taller small trees (Figure 12). Similarly, no substantial trends were apparent for substrate in the clearcut, although the proportions of taller small trees were generally higher in all of the clearcut substrates, compared with the old-growth substrates, and that Lignomor13, upland organic and wet organic substrates had higher proportions of taller small trees than moss and mineral substrates. These relationships were not reflected in the clearcut model, in which substrates were important primarily in conjunction with other site factors (Table 24). Although species-substrate interactions did not contribute significantly14 to either stand type model, Hw appeared to have proportionally more taller small trees than Cw on upland organic, wetland organic, Lignomor, and moss substrates in the clearcut; whereas in the old-growth, this was true only on moss substrates (Figure 13). In the clearcut, the proportion of taller small trees increased, relative to other substrates, on Lignomor and wet organic substrates with increasing SNR (Figure 14), a relationship which was also apparent in the clearcut model (Table 24). SNR negatively influenced small tree height class in the clearcut model (Table 24); and proportionally fewer taller small trees occurred with increasing SNR in the clearcut (Figure 15). Although SMR was shown 1 3 The odds ratio for Lignomor (Table 24) suggests a lower likelihood of finding taller small trees on Lignomor, compared with other, substrates. Although Figure 12 shows higher proportions of taller small trees on Lignomors than on any other substrate in the clearcut; when all substrates are pooled together, 47% of the taller small trees occur on Lignomors, whereas 53% occur on other substrates. 1 4 Although the odds ratio indicated a positive relationship between Other: Cw and upland organic substrates (Table 24), the corresponding dummy variable pair, Hw:Cw x upland organic interaction was not statistically signficant. Similarly, in the age class distributions, a significant interaction was shown between Hw:Cw and upland organic substrates, and a non-significant interaction between Other: Cw and upland organic substrates. 46 to negatively15 influence small tree height class in the old-growth model (Table 24), the raw data indicated that the proportion of taller small trees increased slightly as SMR increased (Figure 15). 50 -, 45 40 35 30 25 20 15 10 5 0 Small tree height => 30 cm n=U7 n=931 n-267 Ui Old'grow 111 Figure 12. Percentage of small trees => 30 cm, by stand type and substrate group. Only the proportions of small trees => 30 cm in height are depicted. Not shown, but implicit, are the proportions of small trees < 30 cm in height, which can be calculated by subtracting the % Occurrence from 100% for each category. Not graphed are substrate groups which had <= 5 taller small trees. Sample sizes are for all small trees within a given stand type - substrate group combination. Key: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wedand organic. > ill tret height - > 30 ca C w O L D - G R O W T H Figure 13. Percentage of small trees => 30 cm, by stand type and species, according to substrate group. Symbols indicate the percentage of small trees => 30 cm in height, within a given species and substrate group. Trend lines are not shown for samples sizes <=5. The steeper the slope of the trend line, the greater the influence of substrate group on small tree height class for a given species. Key for species: Cw=western redcedar Hw=western hemlock. Key for substrate groups: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetland organic. Note that trend lines appear to be similar in the clearcut, whereas trend lines increase from Cw to Hw on moss and Lignomor substrates in the old-growth. 1 5 Given that the model was slightly unstable (with a relatively low Hosmer-Lemeshow Goodness-of-fit p-value of 0.14), the odds ratio for soil moisture regime was therefore deemed to have been confounded either by other variables in the model, or by the absence of important variables from the model. Note that the negative trend line for soil moisture regime in the clearcut (Figure 15) also has a slight slope and that this variable was not detected to be statistically significant in the clearcut model. ^ Small tree height => 30 cm, CLEARCUT ONLY Soil nutrient regime class according to substrate group Figure 14. Percentage of small trees => 30 cm in clearcut, by soil nutrient regime according to substrate group. Symbols indicate percentage of small trees => 30 cm in height, within given soil nutrient regime class and substrate group (= "given") relative to all other substrate groups (= "other"). Trend lines are not shown for samples sizes <=10. Steeper trend line slope indicates greater soil nutrient regime and substrate group influence on small tree height class. Soil nutrient regime key: vp-p=very poor and poor m-r=medium and rich. Substrate group key: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetland organic. Trend lines increase for Lignomor and upland organic substrates in very poor/ poor quadrats, and increase for Lignomor and wetland organic substrates in medium/rich quadrats. Small tree height => 30 cm Soil moisture/nutrient regime class by stand type Figure 15. Percentage of small trees => 30 cm by stand type according to soil moisture and nutrient regime. Symbols indicate the percentage of small trees => 30 cm in height, within a given stand type and soil moisture or nutrient regime class. Trend lines are not shown for samples sizes <=5. Steeper trend line slope indicates greater influence of soil moisture or nutrient regime on small tree height class. Note that only soil nutrient regime in the clearcut shows strong trend. Trend for soil moisture regime in both stand types is weak. Soil moisture regime key: l=slightly dry, fresh and moist 2=very moist and wet. Soil nutrient regime key: 1 =very poor and poor 2=medium and rich. 48 In either stand type model, competing vegetation was not shown to influence height distributions (except for a Block-fern interaction in the old-growth). Height distributions were difficult to analyze by non-crop vegetation cover, given the lack of suitable sample sizes for analysis; however, the trend lines in the clearcut indicate that the proportion of Hw under any non-crop vegetation cover was taller than Cw (Figure 16). Increased deciduous coverage appeared to coincide with an absence of taller Cw and Hw in both stand types (Figure 16). Small tree height =>30cm <30% =>30% <30% >^30% <30% =>30% <30% =o30% Cw Hw Cw Hw CLEARCUT OLDOROWTH Vegetation cover by species and stand type Figure 16. Percentage of small trees => 30 cm, by stand type, species and non-crop vegetation cover class, according to non-crop vegetation group. Symbols indicate the percentage of small trees => 30 cm in height, within a given stand type, vegetation cover group and cover class, for Cw and Hw only. Trend lines are not shown for samples sizes <10. Less than 30% non-vegetation coverage represents zero to sparse coverage in a quadrat plot. Greater than or equal to 30% coverage represents moderate to heavy coverage in a quadrat plot. The steeper the slope of the trend line, the greater the influence on small tree height class with increased non-crop vegetation coverage. Key: BRYO=bryophyte DECI=deciduous shrub EVER=evergreen shrub FERN=fern FORB=forb GRSD=grass and sedge. 49 3.3.5 Small tree age class distributions In both stand types, whereas less than 25% of seedlings were older, about 90% of the veglings were older (Figure 17). Proportions of older Hw and other species exceeded those of Cw (Figure 18). Age => 5 years 100 90 80 70 60 50 40 30 20 10 0 n=S4 n= 131 7 >>>>HWHOXW"X-: u 1031 SS:':s:-::$^SSx i i l i l l ! seedling C L E A R C U T . vegling seedling O L D - G R O W T H vegling Figure 17. Percentage of small trees => 5-years by stand type and reproductive origin. Only proportions of small trees => 5-years in age are depicted. Not shown, but implicit, are the proportions of small trees < 5-years, which can be calculated by subtracting the % Occurrence from 100% for each category. A g e => 5 years I 3 u w o n-277 n - 5 4 " ° 7 S n=498 n=1371 C w C L E A R C U T H w C w O L D - G R O W T H H w Other Figure 18. Percentage of small trees => 5-years by stand type and species. Only proportions of small trees => 5-years in age are depicted. Not shown, but implicit, are the proportions of small trees < 5-years, which can be calculated by subtracting the % Occurrence from 100% for each category. Key: Cw=westem redcedar Hw=westem hemlock Other=lodgepole pine, Western yew, Sitka spruce, and yellow-cedar. 50 As a main effect, microtopography was not detected to influence small tree age class models, nor was it considered to be important in any interactions16 in the clearcut model; however, in the old-growth model, microtopography interactions with moss and wet organic substrates influenced small tree age class. Proportions of older small trees on depressions/flats exceeded those on inclines/mounds on moss and wet organic, compared with other, substrates (Figure 19). Small tree age => 5 years -Lig 'Min " Mos "Ucw -Upl Wet 90 80 1 I a u depression/flat CLEARCUT incline/mound depression/flat OLD-GROWTH incline/mound Figure 19. Percentage of small trees => 5-years, by stand type and microtopography, according to substrate group. Symbols indicate percentage of small trees => 5-years in age, by stand type, microtopography class and substrate group. Trend lines are not shown for categories with sample sizes < 10. Steeper trend line slope indicates greater influence on small tree age distribution with the given change in microtopography class and substrate group. Key: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetland organic. Trend line appears to be of similar slope and direction in the clearcut, indicating little difference in trend for older age distributions among substrate groups and microtopography class combinations (i.e., all substrates appear to have higher proportions of older small trees on inclines/mounds than on depressions/flats); however, trend lines in the old-growth appear slightly steeper and in opposite direction for moss and wetland organic substrates, indicating that, compared with other substrates, proportionally fewer older small trees occur on inclines/mounds on moss and wetland organic substrates. 1 6 The interaction between Lignomor substrate and microtopography (Table 24) was deemed unreliable because of low cell sample sizes. 51 Substrates were important both as main effects and interactions, which differed according to stand type. In the old-growth model, wet organic substrates negatively influenced age class, whereas in the clearcut model mineral substrates negatively influenced age class (Table 24). Although probabilities of finding older small trees were not found to be statistically significant for upland organic and moss, compared with other, substrates in the old-growth model, the proportions of older small trees found on these substrates were, in fact, similar to those on the wetland organic substrates (Figure 20). Lignomor substrates had the highest proportion of taller small trees in both stand types. Small tree age => 5 years 50 -1 : — 45 _ Ckarcut Old-growth Figure 20. Percentage of small uees => 5-years by stand type and substrate group. Only proportions of small trees => 5-years in age are depicted. Not shown, but implicit, are the proportions of small trees < 5-years, which can be calculated by subtracting the % Occurrence from 100% for each category. Not graphed are substrate groups which had <= 5 older small trees. Sample sizes are for all small trees within a given stand type - substrate group combination. Key: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetiand organic. In the clearcut, the substrate group containing the lowest relative proportions of older trees is mineral; compared with moss, upland organic and wetiand organic substrate groups in the old-growth. Although SMR did not influence age distributions in the old-growth model, proportions of older small trees increased with increasing SMR (Figure 21). In the clearcut model, SMR negatively influenced age class (Table 24); and proportions of older small trees decreased slightly 52 with increasing SMR in the clearcut (Figure 21). SNR did not influence age distributions in the old-growth model; however, proportions of older small trees decreased with increasing SNR (Figure 21). In the clearcut model, no statistically significant relationship17 was detected between SNR and species, although proportions of older Hw appeared to decrease more than Cw with increasing SNR (Figure 22). Interactions between SNR and upland organic substrates negatively influenced age class in the clearcut model, which is also apparent by a decrease in the proportions of older small trees with increasing SNR (Figure 23). Despite the existence of no statistically significant relationship between wetland organic substrates and SNR in the model, proportions of older small trees increased with increasing SNR on wetland organic substrates (Figure 23). Smal l tree age => 5 y e a n C L E A R C U T O L D - G R O W T H Soi l moisture/nutrient regime class by stand type Figure 21. Percentage of small trees => 5-years, by stand type according to soil moisture and nutrient regime class. Symbols indicate percentage of small trees=> 5-years in age, by stand type and soil moisture or nutrient regime class. Not shown, but implicit, are proportions of small trees < 5-years, which can be calculated by subtracting the % Occurrence from 100% for each category. Steeper trend line slope indicates greater influence of soil moisture or nutrient regime on small tree age class. All trend lines appear to be somewhat weak, having gradual slopes. In both stand types, the percentage of older small trees decreased slighUy with increasing soil nutrient regime. In the clearcut, the percentage of older small trees decreased slightly with increasing soil nutrient regime, however increased in the old-giowth. Soil moisture regime key: l=slightly dry, fresh and moist 2=very moist and wet. Soil nutrient regime key: 1 =very poor and poor 2=medium and rich. 1 7 Although a negative relationship was apparent between Hw:Cw and SNR; the relationship between SNR and the corresponding dummy variable Other:Cw contained the value 1 within its confidence limits (Table 24). Small tree age => 5 years Soil nutrient regime class by stand type Figure 22. Percentage of small trees => 5-years, by stand type and soil nutrient regime classes, according to species. Symbols indicate percentage of small trees => 5-years in age, by stand type and soil nutrient regime class, for given species. Trend lines are not shown for categories in which sample sizes <10. Steeper trend line slope indicates greater influence of soil nutrient regime on small tree age class for a given species. In the clearcut, the trend line is steeper for Hw than Cw, suggesting that nutrient regime influences Hw age distributions moreso than for Cw. Species key: Cw=westem redcedar Hw=westem hemlock Other=lodgepole pine, western yew, Sitka spruce, and yellow-cedar . Soil nutrient regime key: vp-p=very poor and poor m-r=medium and rich. Small tree age => 5 years Lig Min Mos Ucw • -Upl - O - Wet | a - * — vp-p CLEARCUT vp-p OLD-GROWTH Soil nutrient regime class by stand type Figure 23. Percentage of small trees => 5-years, by stand type and soil nutrient regime class, according to substrate group. Symbols indicate percentage of small trees => 5-years in age, by stand type, soil nutrient regime and substrate group. Trend lines not shown for samples sizes <=10. Steeper trend line slope indicates greater soil nutrient regime and substrate group influence on small tree height class. Soil nutrient regime key: vp-p=very poor and poor m-r=medium and rich. Substrate group key: Lig=Lignomor Min=Mineral Mos=Moss Ucw=Undecomposed coarse wood Upl=Upland organic Wet=Wetland organic. Note that whereas trend lines increase (indicating a positive relationship) for wetland organic substrates in the clearcut, they decrease (indicating a negative relationship) for upland organic and Lignomor substrates with increasing soil nutrient regime. 54 Some non-crop vegetation appeared to influence age distributions in both stand types. Whereas Lignomor substrates interacted positively with evergreen shrub presence to influence age class in the old-growth model, this interaction had a negative influence in the clearcut model (Table 24) (Figure 24). In the clearcut model, although low odds indicated reduced probability for finding older small trees with increasing deciduous shrub presence (Table 24), these odds were calculated based on a sample size of 10; therefore, this relationship was deemed to be unreliable. Similarly, high odds, which indicated increased probability for finding older small trees on wetland organic substrates subject to moderate/heavy grass/sedge presence, were also deemed to be unreliable given a lower confidence limit of 1.2 (Table 24) and the almost parallel trend lines between substrate groups (Figure 25). Small tree age => 5 years • Other •——"Lignomor <30% =>30% <30% =>30% CLEARCUT OLD-GROWTH Evergreen shrub coverage Figure 24. Percentage of small trees => 5-years, by stand type and evergreen shrub cover class, according to Lignomor versus other substrates. Symbols indicate percentage of small trees => 5-years in age, by stand type and evergreen shrub cover class, according to substrate group. Steeper trend line slope indicates greater evergreen shrub coverage influence on small tree age distribution on a given substrate group. Only two substrate groups are compared: Lignomor substrates versus all other substrates pooled together. In the clearcut, the positive trend line for other substrates, compared to the slightly negative trend line for Lignomor substrates indicates that as evergreen shrub presence increases, older small trees are more prominent on other substrates than on Lignomor substrates. In the old-growth, the negative trend line for other substrates, compared to the slightly positive trend line for Lignomor substrates indicates that as evergreen shrub presence increases, older small trees are more prominent on Lignomor substrates than on other substrates. 55 Smal l tree age => 5 years, C L E A R C U T O N L Y • — O t h e r - B - Wet organic 45 I 20 I 5 0 -I — <30% = > 3 0 % Grass/sedge coverage Figure 25. Percentage of small trees => 5-years, by stand type and grass/sedge cover class, according to wet organic versus other substrates. Symbols indicate the percentage of small trees => 5-years in age, by stand type and grass/sedge cover class, according to substrate group. Steeper trend line slope indicates greater influence of increased grass/sedge coverage on small tree age distribution on a given substrate group. Here, only two subsuate groups are compared: wet organic substrates versus all other substrates pooled together. The almost parallel trend lines for both substrate groups indicates very littie difference in proportions of older small trees within each substrate group as grass/sedge presence increases. The percentage of older small trees generally increases as grass/sedge presence increases, with a slighfly greater increase (indicated by a slighdy steeper slope) on wet organic substrates. 4. Discussion 4.1 Factors that influence small tree presence, height and age distributions in the study area Although small tree presence distributions were used to infer how a given explanatory variable influenced small tree occurrence, it was the age and height class distributions that provided a way of inferring relative survival and growth, respectively, of small trees relative to given explanatory variables. Abundance was not modelled in this study; however, the higher abundance of small trees in the old-growth (mean =>8 trees per plot) than in the clearcut (mean < 5 trees per plot) (Table 19) may be attributable to more seed trees in the old-growth. The lower mean trees per plot in the clear-cut may partially be attributable to mortality during harvest 56 operations, whereas a higher overall incidence of plots containing 10 or less trees in the clear-cut compared with the old-growth may represent an increase in conditions (i.e., more light) suitable for seed germination in the clearcut over the old-growth. Although Cw and Hw can become established under an understory (Minore 1990; Packee 1990), and Cw and Hw germination tends to be best under partial shade (Sharpe 1974; Soos and Walters 1963), even in coastal forests (Hetherington 1965), shade tolerance does not imply shade necessity (Mitchell and Arnott 1995). The lower percentage, in the clearcut, of plots containing no trees, than in the old-growth, (30% versus 41% respectively) (Table 20), suggests an overall better distribution of trees throughout the clearcut versus the old-growth. In this study, growth and survival of small trees appears to be low after germination. Other studies in coastal temperate rainforests have cited similar findings for Cw (Gregory 1957; Boyd 1959; Hetherington 1965) and Hw (Alaback and Tappeiner II 1991; Hetherington 1965). The prolific seed production rates of Cw and Hw are well recognized (James 1959; Boyd 1959; Minore 1990; Packee 1990), albeit cyclic (Hetherington 1965), and the seeds germinate quite readily where adequate moisture (Owens and Molder 1984a; 1984b) and partial shade (Hetherington 1965) are available. Given the prolific seed germination in this environment, and that seedbed may have little influence in quantity and rate of germination (Soos and Walters 1963), regeneration setbacks due to failure of seed germination can be ruled out, and appear to be more associated with factors which prevent seedling establishment. Age pattern between stand types can be difficult to compare in an uncontrolled setting, since harvesting operations destroy a number of previously established small trees, and simultaneously improve seedbeds for post-harvest recruitment. For example, before harvest, a large proportion of this forest was comprised of unmerchantable overmature trees with heart rot. 57 Much of the coarse, high volume slash (common in overmature old-growth logging operations) left on the clearcut after logging was likely deposited over advanced regeneration. Also, heavy trees which were yarded through logging corridors with cables not only would have damaged or killed some of the advanced regeneration, but conversely also redistributed the upper forest floor, and in some cases exposed mineral soil, especially on shallow Folisols, creating ideal seedbeds for ingress. The 5-year timespan over which this study occurred may have been somewhat limiting, in that seedling recruitment in the clearcut was not measured according tb its fullest potential, since seedlings can recruit periodically for at least 10 years after harvesting (MacBean 1941). Nevertheless, the results can be useful given the 6-year recommended legal small tree establishment period (regeneration delay) for natural regeneration in the CWHvh2 variant (BC MOF 1995). The findings of this study support a long-held idea that advanced regeneration is particularly important in the early establishment of logged coastal Cw-Hw forests (Stoodley 1925). Whereas an estimated 131 taller potential future crop (PFC) trees/ha have been established in the study area through ingress, 2920 estimated taller PFC trees/ha are advanced regeneration (Table 21). Although these rough estimates indicate that the study area likely meets the minimum numbers of trees per ha to be considered fully stocked before the 6-year regeneration delay period stocking standards, many of the trees exceeding 30 cm in height are comprised of advanced regeneration of dubious growth habit. Given that tree vigour data were not collected, the proportion of advanced regeneration which escaped harvesting damage to become future commercially well-formed trees cannot be determined from this study. Oliver and Larson (1996) discussed how Hw can lose their terminal shoots under conditions of high shade, and where horizontal shoot growth is favoured over vertical shoot 58 growth. King (1991) suggested that shade tolerant understory coniferous species, including Cw and Hw, have morphologies, for example small and thick needles, long needle retention times, and wide crowns, that increase light interception and persistence in the understory at the expense of height growth. In fact, Cw can maintain extensive horizontal growth of upper lateral branches even after release from high shade (Oliver and Larson 1996). Although Hw advance regeneration can form a dominant component of a regenerating stand if not destroyed (Williamson and Ruth 1976; Deal et al. 1991) and suppressed Hw has been observed to resume vertical height growth again after release (Oliver 1979), trees which completely lose epinastic control18 after long periods of suppression under high shade, can develop crooks or forks as lateral branches turn upward and compete for the terminal position (Oliver and Larson 1996). Although Cw response to overstory release can be favourable (Graham 1982), Oliver et al. (1988) cautioned that release of Cw from suppression by an overstory does not appear to be promising for good quality wood production. Collection of data related to apical dominance of the terminal shoot (i.e., terminal shoot length compared with nearest lateral shoots, growth angle of terminal shoot and nearest lateral shoots) may therefore contribute useful information in future studies on regeneration pattern of Cw and Hw in the CWHvh2 variant. 4.1.1 Block and plot contiguity Any conclusions drawn from this study, are confined to the study area alone, given that only one site was examined. This was an unavoidable limitation, given the lack of sites of suitable age available in the CWHvh2 variant at the time this study commenced. Although the study area was divided into three Blocks, which were included in the regressions because some differences 1 8 Control exerted by a tree's terminal bud over the length and orientation of lateral branches (Oliver and Larson, 1996) 59 were expected due to the drainage, aspect and harvest differences between the Blocks, no special discussion is applied to their influence in the models (other than to say that they did), given the low number of Blocks. Some of the variability in the data was influenced by a natural distortion in the distribution of site factors and small trees, causing, for example, a somewhat "all or nothing" seedling or vegling production incidence. For example, as many as 222 small trees were counted within a quadrat plot, likely attributable to the nearby presence of a tree which recently shed seeds or cones. Adjacent plots may have had small trees originating from the same parent tree. Similarly, plots which contained one vegling, often contained other veglings from the same parent tree, which may also have produced veglings in adjacent plots. This pattern is likely related to what is known as "neighbourhood effects" (Frelich et al. 1998). These effects are viewed to be any process mediated by canopy trees that affects the replacement probability by the same or other species at the time of canopy mortality and includes seed rain, stump and root sprouting, alteration of the physical or nutrient status of the forest floor to favour or disfavour germination and establishment of a given species, and the influence of the canopy on local temperature. According to this concept, random groups of species on uniform environments can interact to form spatially distinct communities, which can account for unexplained variability in studies attempting to relate environmental parameters to forest composition. Plot contiguity may be useful for examining neighbourhood effects, since plots located immediately contiguous to one another were expected to have similar patterns in small tree presence, height and age distributions (whatever they might be), given similar site conditions and proximity of seed sources, and given similar local site disturbance from harvesting in the clearcut, or windthrow in the old-growth. The consistent influence of plot contiguity on small tree height distributions, and to a lesser extent age 60 distributions (Table 22) likely contributes to weakness in the presence, height and age distribution models. However, the true magnitude of the effect of plot contiguity on response variables can only be determined with improvements to how plot contiguity is modelled, which must be done in such a way that contiguous plots are spatially identified. 4.1.2 Genetic factors 4.1.2.1 Origin The detection of reproductive origin as being a primary influence on height and age class distributions in this study, both as a main effect and in interactions with other factors, was not surprising. Many plants tend to favour vegetative over seed reproduction in environments not conducive to seedling establishment (Fenner 1985); and vegling development in Cw has been widely documented (Boyd 1959; DeLong 1997; Habeck 1968 and 1978; Schmidt 1955). Although 94% of all the small trees in this study were seedlings, compared to only 6% veglings, the oldest and tallest small trees were veglings. Median vegling height in the clearcut was 33.5 cm, and median age was 8-years, compared with 6.3 cm and 3-years, respectively for seedlings. Proportions were similar in the old-growth, where median height and age for veglings was, respectively, 27.0 cm and 12 years, compared with 4.2 cm and 2 years for seedlings, respectively. About 90% of the veglings were at least 5-years of age, compared to less than 25% of the seedlings in both stand types, suggesting that proportionally fewer seedlings survive to age 5. Nevertheless, since veglings are not as numerous as seedlings in the study area, and Cw vegling growth is slow (Boyd 1959), veglings cannot be expected to contribute to major renewal of this forest in short time periods. 61 Seedlings are considered to be fully established when they become fully independent of their seed reserves (Fenner 1985). In this study, determination of whether or not a vegling was fully independent of its parent was difficult, since veglings remain attached to their parent trees for an indeterminate length of time. Some Hw, Tw, and Yc vegling growth was noted in this study, often originating from branches of fallen trees, where the branches grew vertically, perpendicular to the fallen bole, and in the case of Tw, where terminal leaders exerted epinastic control over laterals (similar Hw vegling leaders were without apparent epinastic control). Whether or not veglings from these species can be expected to survive or sever independence from their parent trees could not be determined from this study. 4.1.2.2 Species Species influences were important to both height and age class distributions in this study, both by themselves and as part of interactions. Although Cw predominated in overall small tree abundance regardless of stand type (over 70%, versus Hw at over 20% and other species at 4%), Hw had the highest percentage of taller and older small trees in clearcut, followed by other species. Other species had higher proportions of taller and older small trees in the old-growth, followed by Hw. Given the shade tolerance of Hw and Cw (Kobe and Coates 1997; Krajina 1965; Krajina 1969; Minore 1979), and that both species have been observed to germinate to maximum levels at 75% shade (Minore 1979), factors other than stand type are causing the greater presence in Cw relative to Hw in both stand types. Although both Cw and Hw have been shown to express low tolerance to heat and frost (Minore 1979; Krajina 1969), neither of these factors is chronically limiting in this area. Cw tolerates excess moisture better than Hw (Minore 1979); however, this does not explain the relatively higher proportions of taller and older Hw, 62 compared with Cw. This may be related to substrate, and is further discussed in Section 4.1.5. The higher proportions of older Hw, compared with Cw, in the clearcut over the old-growth is difficult to interpret and may simply indicate higher mortality in Cw during harvest operations as well as a higher number of Cw germinants relative to Hw germinants. 4.1.3 Clearcut stand edge effects Increasing stand edge distance has been reported to influence small tree presence negatively (MacBean 1941; Garman 1951; Gashwiler 1969; Hetherington 1965; Clark 1970), and small trees have been observed to be most dense at distances close to the stand edge (Stoodley 1925; MacBean 1941). During the reconnaissance phase of the study, recruitment in both Cw and Hw appeared to be prolific near the stand edge particularly on West and South Block. The negligible differences in small tree presence classes attributable to stand edge distance, in this study, may be attributable to: 1) The "all or nothing" approach used in classifying presence. For example, all plots containing at least one tree were considered to have small tree presence regardless of how many trees they contained. 2) Plot distribution. The expectation was that more plots located closer to the stand edge would have contained small trees. However, coarse woody accumulations were also prevalent close to the stand edge (either from windthrow or harvesting residue). Large numbers of snags and fallen trees arising from windthrow and/or harvesting damage are not uncommon near stand edges (Chen et al. 1992). 3) Most plots within this study were easily located within maximum seed dispersal distances of Cw (120m) and Hw (600-1150m) reported by Owens and Molder (1984a, 1984b). The 63 coastal weather systems in this area generate frequent and violent windstorms (Phillips 1990), which could also possibly increase the seed dispersal distances (Oliver and Larson 1996). 4) Stand edge variability. In their review of biological responses at forest edge, Chen et al. (1992) also cite forest type, edge age, orientation, and formation (e.g., clearcut vs. natural bog), patch shape and size, and topographic features as some major influences, some of which may have contributed to small tree presence distributions in this study. 4.1.4 Microtopography A limiting factor in this area is impeded drainage caused by the rich accumulation of organic matter from abundant vegetation. Since the area receives so much rainfall, and poorly drained sites provide conditions which are oxygen limiting to plant roots (Marschner 1986) caused by greatly reduced oxygen exchange in permanently saturated soils (Ponnamperuma 1972), survival was not expected to be high for older small trees on depressions and flat areas. Microtopography was not expected to greatly influence germinant presence, since oxygen may not be as limiting a factor for germination as for growth (Leadem 1996). On wet sites, small trees were expected to be found on elevated microsites (Ehrenfeld 1995; Banner et al. 1993; Green and Klinka 1994). In fact, only one out of the six regressions in this study detected microtopography to be important, on its own, in influencing a response variable, although all six regressions detected relationships through interactions with other factors. Since accumulating organic matter can waterlog areas other than steep slopes (Zach 1950), it is possible that certain types of inclines or possibly even mounds in this area may not be adequately aerated for survival and/or optimum height growth. This may, in part, explain the 64 lower proportions of taller small trees found on inclines located with increasing soil moisture regime in the clearcut, or the lower proportions of older small trees found on inclines/mounds located on moss and wet organic substrates in the old-growth. Although tree height growth is generally greater on inclines and mounds than on depressions and flat areas (Beatty and Stone 1986), proportions of taller small trees in the clearcut were virtually the same regardless of microtopography group, and the median height was only slightly higher on inclines/mounds (8 cm) than on depressions/flats (6 cm). In this study, increased soil nutrient regime appeared to coincide with increased proportions of taller small trees on inclines and mounds. Beaudry and Banner (1990) suggested that productivity of outer coastal lowland Cw-Hw sites could be improved with harvesting and site preparation treatments (e.g., mounding) that improve soil aeration following the premise that disturbances such as windthrow activity and harvesting with site preparation tend to retard humification and podzolization and stimulate nutrient cycling. Some preliminary research indicates that mounding may improve growth and productivity, especially of Hw and PI, of poorly aerated coastal wet sites, although improvements to Cw productivity could not be discerned (Puis 1998; Banner and Puis 1999). 65 4.1.5 Substrates The study results generally supported the expectation that regeneration would be prolific on exposed mineral, moss, upland organic and Lignomor substrates and minimal on undecomposed coarse wood (Neiland 1971; Owens and Molder 1984a, 1984b; Leadem et al. 1997; Packee 1990; Minore 1990; Williamson 1979; Deal etal. 1991; Soos and Walters 1963), with similar trends in small tree presence in both stand types. Although the percentage of samples containing at least one tree would normally be expected to increase with increased coverage of any given suitable substrate, such a trend was only slightly noticeable in this study (Figure 6), likely influenced by small sample sizes for some of the substrate group - cover class combinations. Height was generally expected to be relatively higher on well-aerated exposed mineral and upland organic, compared with other, substrates (Packee 1990; Hennon 1992, Minore 1983). Given that height growth is positively correlated with age under given stand conditions for a given species (Oliver and Larson 1996; Daniel et al. 1979; Mitchell and Polsson 1988; Davis and Johnson 1987), similar distributions for age class was expected as for height class on similar substrates for a given species and stand type. Although these expectations were generally satisfied for height distributions, with the exception that mineral substrates had the lowest proportion of taller small trees regardless of stand type, the slight differences in distributions of older small trees among substrates between the two stand types may simply have been indicative of the impact of harvest operations on the clearcut, contributing not only to small tree mortality but also causing changes to substrate distributions themselves, as has already been discussed in Section 4.1. 66 4.1.5.1 Exposed mineral substrates In this study, low proportions of older small trees on mineral samples in the clearcut may be a result of mineral substrates being recently created by harvesting (or windthrow in the old-growth) disturbance. The exposed mineral substrates may not have existed long enough for older small tree establishment. Since mineral substrate sample size was small, no strong conclusions can be drawn about the influence of mineral substrates on the presence, height and age distributions in this study. 4.1.5.2 Lignomors Lignomor substrates (often loosely referred to as "decayed or "decaying wood") occupy a significant portion of the forest floor in coastal forests. Keenan et al. (1993) found approximately 60% of the forest floor mass in a CHWvm forest was comprised of "decaying wood". In this study, Lignomors just exceeded 16% of clearcut substrates and 23% of old-growth substrates. Decomposed and partially decomposed nurse logs and overturned stump mounds appear to be important substrates for seed germination and small tree establishment, in clearcuts and old-growth (MacBean 1941; Turner and Franz 1985; Harmon and Franklin 1989). Therefore, the positive influence in this study, of Lignomors on small tree presence in both stand types, was fully expected. In some Pacific Northwest old-growth forests, logs, stumps and large wood fragments comprised from 6-14% of the forest floor but account for as much as 98% of tree regeneration (cited by Harmon 1987). In this study, Lignomors had greater small tree presence than other substrate groups. The relatively greater increase in proportions of taller small trees on Lignomor substrates with increased soil nutrient regime may be related to the degree of decomposition of, and hence nutrient release by the Lignomor, which may perhaps be more decomposed with 67 increasing nutrient richness of the microsite. Klinka et al. (1995) have hypothesized that the influence of decayed wood on vegetation and soil is site-specific and varies with soil moisture and nutrient regime, other soil properties, and climate. Alternatively, Lignomor substrates may somehow temper the effects of competing non-crop vegetation. 4.1.5.3 Undecomposed coarse woody debris Given that undecomposed, coarse woody substrates are unfavorable seedbeds (Stoodley 1925; MacBean 1941; Hetherington 1965; Williamson and Ruth 1976), the low occurrence of small trees on these substrates was expected. This is important, because undecomposed coarse wood occupied almost 30% of the substrates in the clearcut plots (Figure 3). In coastal rainforest clearcuts, undecayed coarse wood can sustain average summer temperatures between 23°and 35°C compared with only 17.5°C for the soil (Marra and Edmonds 1996). Undecomposed logs dry under exposure to sunlight during the long daylength, causing new germinants to desiccate. Although 72 small trees were noted growing on undecomposed coarse wood, in some cases other materials (i.e., moss, litter, exposed mineral soil) lightly covered the undecomposed wood surfaces. Even so, some older apparently undecayed logs which were typed out as "undecomposed wood" were likely already in the initial stages of decay in the outermost 1 cm surface, although this was difficult to detect occularly. This could be sufficient for the anchorage of germinant rootlets. The outer surface of one substrate recorded as undecomposed coarse wood, in which the roots of a 13-year old, 64 cm tall Hw seedling were anchored was likely already in transition to Lignomor. The otherwise complete absence of taller or older small trees on undecomposed coarse wood substrates suggests that any small trees occurring on these substrates were recent germinants and not likely to survive. 68 4.1.5.4 Moss Although certain surface mosses can, in fact, retard regeneration as they dry out during the mid-summer (Godman 1953), this is not true for Sphagnum mosses. Sphagnum mosses, which are indicative of wet soils (Klinka et al. 1989), and can hold greater than 20 times their dry weight in water (Vitt et al. 1988), prevailed over other mosses in the study area. Sphagnum seedbeds are known to be highly sterile and acidic (Baker 1989; Daubenmire 1974). The strong positive relationship between small tree presence and moss substrates in both clearcut and old-growth may simply have reflected the availability of moisture required for early survival of Cw germinants (Boyd 1959). In the old-growth less than 5% of the small trees on moss substrates were at least 30 cm in height, compared to about 10% in the clearcut; yet, over 20% survived beyond 5-years in both stand types, suggesting better survival than growth on moss substrates. 4.1.5.5 Wetland and upland organic substrates The marginal positive relationship between wetland organic substrates and small tree presence, regardless of stand type, was unexpected, since wet organic substrates were expected to have a negative influence on small tree presence (see Section 4.1.6). However, given that most of the small trees (almost 80% in the old-growth, and about 65% in the clearcut) on wetland organic substrates were less than 5-years in age, survival was not expected to be high, despite the initial recruitment. Of interest is the role nutrient regime plays on these substrates. As soil nutrient regime increased on wet organic substrates, the proportion of taller small trees increased. The marginal positive relationship between upland organic substrates and small tree presence indicates only a marginally beneficial recruitment environment against the effects of high moisture and plant nutrient competition compared with other substrates. In the clearcut, the 69 lower proportions of older small trees found on upland organic substrates with increasing nutrient richness may simply reflect some mortality in the older age classes caused by harvest disturbance. 4.1.6 Soil moisture and nutrient regimes By themselves, soil moisture and nutrient regime were not detected to influence small tree presence; however, some notable interactions with microtopography have already been discussed (Section 4.1.4). The lower proportions of older Hw, relative to Cw, with increasing soil nutrient regime in the clearcut may be related to low nutritional requirements of Hw, which regenerates best on mor humus or very acid decaying coniferous wood, with poor survival on nutrient rich sites (Krajina 1969). The lower proportions of taller small trees with increasing soil nutrient regime in the clearcut is somewhat unexpected, and may be related to increased non-crop vegetation competition, or post-harvest mortality (for small trees older than 5-years). In the clearcut, the slight trend indicating a decrease in proportions of taller and older small trees with increasing soil moisture regime may be related to seasonal soil anaerobic conditions under prolonged periods of heavy rain (Rowell 1981; Fausey and Lai 1990). Short periods of heavy rain during warm periods (not uncommon in the CWHvh2 variant) when respiration rates are rapid can also create anaerobic conditions (Rowell 1981). Added to potential anaerobic conditions caused by rain on the entire study area, already one third of the plots are located on very moist to wet microsites, which are located on water tables. Root mortality under anaerobic conditions, although difficult to quantify (Fogel 1990), may be important in the study area, since the extent of root mortality likely affects height growth, and up to a certain threshold, survival. Any differences in root longevity (Black et al. 1998) of the tree species growing in the study area may affect their unique survival and growth strategies in this environment. The 70 preponderance of older veglings in the study area suggests that veglings may tolerate seasonally saturated conditions, The extent to which veglings still attached to their parent trees rely on the parent tree for establishment until they reach a certain age or height may influence their success in this environment. The unexpected opposite trend in the old-growth, in which higher proportions of taller and older small trees were observed with increasing moisture regime, may possibly reflect a relatively less wet site than in the clearcut, given that water tables have been reported to increase in clearcuts (reviewed by Armson 1979). Both Cw and Hw tolerate winter water tables more than 15 cm below the soil surface (Minore and Smith 1971). Despite that Hw appears to be intolerant of water tables less than 15 cm deep whereas Cw appears to maintain growth where stagnant winter water table is less than 15 cm deep (Minore and Smith 1971), this study did not detect any relationships between species and soil moisture regime in either height or age class distributions. 4.1.7 Non-crop vegetation In the clearcut, small tree presence, height and age distributions were expected to be inhibited by ferns (based on observations during the site walkthrough), grasses and sedges (Lieffers and Stadt 1994; Scholes and Archer 1997), and evergreen shrubs (MacBean 1941; Bennett 1996; Chang et al. 1996). Similarities in mode, median, and mean ages for the no/sparse . vegetation coverage of any vegetation group may be influenced by the presence of other vegetation groups. Part of the difficulty in detecting relationships of above-ground vegetation cover to regeneration presence, height growth, and survival, is that even though any one vegetation group may have had no/sparse coverage in any plot, any other vegetation group may have had moderate/heavy coverage, and may have influenced the small tree presence, height and 71 age distributions. For example, the slight increase in small tree presence with increased bryophyte and forb coverage may not so much reflect a relationship with those vegetation groups, as perhaps less coverage of other more competitive non-crop groups. Not all of the competitive effects of non-crop vegetation may be adequately correlated with above-ground vegetation cover because plant competition also consists of below-ground competition for nutrients, water, and oxygen (Casper and Jackson 1997). This may also be, in part, why some of the expected relationships were either non-existent or weak in this study, and why unexpected relationships occurred. Much in the same way that canopy gaps are important for individual tree species light requirements (Daniels and Klinka 1996; Wright et al. 1998; Runkle 1981), and emergence of competing vegetation (Klinka etal. 1996), below-ground root gaps (Casper and Jackson 1977) are likely important for nutrient and oxygen distribution in this ecosystem, especially following the germinant stage, when a tree's nutritional requirements increase with growth. Given Cw and Hw shade tolerance and plentiful water supply, considerable plant competition occurring in the study area may also be below-ground competition for nutrients and oxygen. Below-ground non-crop vegetation competition in the root zone can adversely affect Cw height growth, moreso than above-ground competition which limits light penetration (Adams and Mahoney 1991). Below-ground competition from live trees, a factor not considered in this study, also influences small tree presence (Moeur 1997). Given the potential importance of below-ground competition, the use of above-ground quadrat plant coverage estimates in this study may not have adequately compared the competitive effects of root systems. For example, a 30% above-ground coverage, within the plot, of non-crop vegetation, translated into a corresponding unknown percentage root coverage underground, may or may not have been important to the trees growing within that quadrat, since root systems of older or taller trees may 72 extend well beyond the plot area (Eis 1974; Leaphart and Grismer, 1974; Rigg and Harrar, 1931). Better quantification of below-ground root systems of competing vegetation may improve future regeneration studies in this environment. Grasses, especially Pacific reedgrass (Calamagrostis nutkaensis (J.S. Presl in CB. Presl) Steud.), were observed to restrict regeneration in the area generally. Cw and Hw were sometimes observed on scalps or machine ruts in the wet grassy areas in Flat Block clearcut, during the reconnaissance phase of the study. Gouges and rutts appeared to break apart dense grass root mats and create adequate seedbed and rooting substrate for small tree establishment. However, given low numbers of plots which contained grasses/sedges (primarily in Flat Block, the smallest of the Blocks), no statistically significant relationships with presence, height, or age distributions could be detected for grass/sedge competition, although trend lines indicated decreased small tree presence with increased grass/sedge coverage in the clearcut. Pacific reedgrass is distributed along the west coast of North America in Hypermaritime climates on moist to wet sites (Klinka et al. 1989; Hubbard 1969), along with several other reedgrasses (Pojar and Mackinnon 1994). Although information on the competitive effects of Pacific reedgrass in Hypermaritime climates is sparse, they, like other Calamagrostis species, produce creeping rhizomes (Hubbard 1969) as well as large tough clumps (Pojar and Mackinnon 1994). Given its similarity in growth habit to other reedgrasses, and given the studies on reedgrasses, for example Calamagrostis canadensis (Michx.) Beauv., which have demonstrated rapid growth patterns under increased light intensity and site disturbance (Powelson and Lieffers 1991 and 1992; Rivard et al. 1990), the potential for Pacific reedgrass to behave similarly should be monitored. This growth habit is especially significant in areas where clearcutting is the silvicultural method of choice for commercial tree 73 production. Lieffers and Stadt (1994) have therefore recommended the use of partial canopy retention such as shelterwood in areas containing understory reedgrasses. Deer fern was observed to limit small tree presence in this study. Trend lines in both stand types indicated reduced small tree presence with increased fern coverage. Its thick and leathery foliage forms an impenetrable layer on the forest floor, creating a barrier between germinant roots and soil. Since deer fern reproduces not only via spores, but also via creeping or erect rhizomes (Pojar and MacKinnon 1994; Soltis et al. 1988; Taylor 1973), and proliferates in coastal Hypermaritime areas on acidic substrates (Klinka et al. 1989), it can be a potential competitor as it spreads rapidly to occupy seedbeds otherwise suitable for small trees. The effect of deciduous shrubs on presence, height, and age patterns could not be determined because of sampling zeros (Agresti 1996) in this study. Salmonberry (Rubus spectabilis Pursh) was the only potentially competitive deciduous shrub present, to a minor extent, in the study area. Nevertheless, of the small trees growing under deciduous shrub presence, less than 1% were comprised of older small trees under moderate/heavy presence, and less than 1% were taller under moderate/heavy presence. No mini small trees were found under moderate/heavy presence. Salmonberry's aggressive growth habit, persistence, and occupation of clearcuts is well documented (Tappeiner et al. 1991, Tappeiner and Zasada 1993; Haeussler et al. 1990). After stand disturbance, a very large rhizome bud bank capable of shoot production allows the establishment of a persistent salmonberry cover, hindering succession to other tree/shrub communities (Tappeiner et al. 1991). This may take as long as 15-25-years (Tappeiner et al. 1991). For this reason, the mere presence of salmonberry should be monitored. Of the evergreen shrubs, salal had the greatest plot coverage (about 13%). Other studies have indicated that salal presence reduces tree growth (Prescott et al. 1996; Chang et al. 1996; 74 Messier 1993; Weetman et al. 1989b), apparently more for Hw than Cw (Fraser et al. 1995; Messier et al. 1990). Some possible reasons for the failure of this study to detect any influence of evergreen shrub presence, including salal, on small tree presence include: 1) The low occurrence of mineral substrates throughout the study area. Salal emergence has been observed to be higher on exposed mineral soils than on intact organic layers, especially under partial shade (Tappeiner and Zasada 1993). 2) Possible reduced vigor in salal the further north in latitude from northern Vancouver Island. Salal is less dominant and less vigorous on the north coast (Prince Rupert Region) than on northern Vancouver Island (A. Banner pers. comm.). Given the study area's location between the north coast and northern Vancouver Island, salal vigor may possibly be somewhat reduced from that reported on northern Vancouver Island. 3) The relatively low historical disturbance, compared with northern Vancouver Island. Given long-term harvest disturbance on Vancouver Island, salal has become well established, whereas disturbance to the forests in the study area is relatively recent. 4) Age of the clearcut in the study area. Studies on Cw and Hw growth check (Prescott et al. 1996; Chang et al. 1996; Weetman et al. 1989b; Messier 1993) focused on older clearcuts (e.g., > 10 years). Possibly not enough time has elapsed for salal to become established and have any effect on the trees in the study area. Given these reasons, this study's results cannot lead to conclusions that salal may not have the potential to negatively affect height growth of Cw and Hw. The apparently opposite trends of increased evergreen shrub influence on proportions of older small trees found on Lignomors, compared with other substrates, between stand types, appears not so much to reflect change in distributions of older small trees found on Lignomors (since the trend lines are of very minor 75 slope in both elearcut and old-growth), as it does for the other substrates which have steeper trend lines (Figure 24). Unlike on other substrates, the proportions of older small trees observed on Lignomors both in old-growth and clearcut changes very little regardless of evergreen shrub coverage. This suggests that evergreen shrub coverage has little effect on small trees growing on Lignomor substrates compared with small trees growing on other substrates. 4.2 Statistical analysis Although logistic regression techniques appeared, at first, to be promising in analyzing this data, the end result produced only weak models which cannot be relied upon to predict accurately the probability of small trees belonging to a given small tree presence, height, or age class. Nevertheless, some potentially important general relationships among regeneration parameters (small tree presence, height and age) were identified, as already discussed. However, any positive or negative influences of explanatory variables on response variables which were indicated by odds ratios could only be interpreted in relation to the other variables in each respective model. Sparse data affected all six regressions, necessitating the removal of variables that contained sampling or structural zeros (Agresti 1996), which can cause severe bias in estimators of odds ratios and poor Chi-Squared approximations for goodness-of-fit statistics (Agresti 1996). To overcome this difficulty in future studies, the number of plots within each Block would have to be increased; however, this would result in increased research costs. Recent advances in exact methods in logistic regression have facilitated analysis of sparse data (Agresti 1996; Hollander and Wolfe 1999; Stokes et al. 1995). SAS/STAT software used in this project, employs methods based on large sample statistics, and does not yet include computational methods for performing exact logistic regressions on sparse data (Stokes et al. 1995). CYTEL's LogXact software 76 (http://vvWvV.cytel.com), recommended by Stokes et al. (1995) and Hollander and Wolfe (1999), may be promising for future studies of this nature, especially where data collection costs are high. Low R 2 s in all six regressions indicate that the proportion of variation explained by the variables in the models is low (Nagelkerke 1991), and are indicative of the difficulty of conducting research in natural environments subject to many uncontrolled influences. Such difficulties have been attributed to historical factors which cannot easily be measured, for example, timing of good seed crops, weather conditions following disturbance, sporadic pest outbreaks (e.g., deer) in addition to unmeasured site variables, suboptimal mathematical models, and imprecise measurements (McCune and Allen 1985). However, given the interrelations of different physiological and morphological aspects of plant growth in their natural settings (Lambers and Poorter 1992), and that natural field conditions are difficult to reproduce in a laboratory (Matzner et al. 1998), studies such as this can be useful in determining general relationships between plant growth and site attributes in their natural environments. Factors (e.g., age of parent, prevailing winds, cone and seed trajectory paths, weather, pathogens) that influence cone production and viability (Hetherington 1965; Clark 1970), including predation (Sharpe 1974; Boyd 1959), and where cones and seeds land; and factors that favor the creation of parent trees for veglings (e.g., branch growth habit, disturbance - reviewed by Edwards and Leadem 1988), were not considered in this study, but are also important for predicting small tree presence. Variation attributable to parent source could be better accounted for by assigning parent identifier labels, during the field collection, to germinants originating from a nearby parent tree, or veglings originating from one parent, and include parent identifier as an additional variable during the data analysis. Future studies could also include measuring the effect of climate on regeneration by analyzing weather data, along with tree ring analysis (to determine 77 which cold periods checked growth, and which warm periods stimulated cone and seed production) for the regeneration study period. Although no data were collected on organisms besides plants, faunal and fungal activity may also contribute to unexplained variability in small tree presence, height and age structures. A variety of fungi, annelids, arthropods, and insects were observed within the 30 cm depth of the substrates, in particular throughout the old-growth plots. Faunal diversity in coastal maritime forests has recently begun to be defined (Battigelli et al. 1994; Fons and Klinka 1998) and may be important, given that the morphology of folic materials of coastal Humic, Lignic, and Ffistic Folisols may be affected by faunal activity, especially mites (Fox et al. 1994). Canopy gap size (not measured in this study) has been found to interact with substrate type and species to influence small tree presence and growth (Wright et al. 1998; Gray and Spies 1997). Seedlings in Cw-Hw old-growth forests tend to be located in or near gaps between crowns of live overstory trees (Moeur 1997), and seedling presence is generally greater in gaps than under the canopy (Spies et al. 1990). How a tree dies and falls can impact the gap size, and ultimately the amount of light available to the forest floor, including new substrates created by the tree (Beatty and Stone 1986; Putz et al. 1983; Lertzman and Krebs 1991). For example, canopy gaps created by dead standing trees, after disintegration, are small and narrow with little soil disturbance, whereas trees in which the entire root bole becomes dislodged tend to create larger gaps with considerable soil disturbance (Putz et al. 1983; Lertzman and Krebs 1991; Beatty and Stone 1986). In some cases, gaps created by dead-standing trees favor the growth of seedlings and saplings present at the time of canopy tree death relative to plants that germinate after gap formation (Putz et al. 1983; Turner and Franz 1985). Gap size has not been found to affect Hw understory presence (Spies et al. 1990), and Hw appears to have a remarkable ability to acclimate 78 to shade by morphologically altering patterns of carbon allocation (Mitchell and Arnott 1995). Nevertheless, Hw height growth in coastal forests has been observed to be positively correlated with light transmission through openings in the forest canopy (Alaback and Tappeiner II 1991; Deal et al. 1991). Cw has both higher root growth and shoot growth rates, as well as a longer growing season, than Hw (Minore 1979) and it appears that while Hw is dependent on canopy gap events for growth to future canopy, Cw is not (Daniels and Klinka 1996), suggesting a slight competitive advantage of Cw over Hw in the old-growth. R2 values may have improved somewhat in the old-growth models had gap size and consequent light penetration to seedbeds been factored into the regressions. 5. Conclusions The most significant conclusions about some influences of genetic and site factors on small tree presence, height and age distributions in the study area are: 1. Ingress cannot be relied upon to regenerate this clearcut within five years. Advanced regeneration appeared to be important for small tree establishment in the clearcut; however, an unknown percentage of the advanced regeneration lacked apical dominance of the shoot terminal leaders. How these trees will respond to release, and whether they will be adequate for commercial production, is yet unknown. 2. Species and origin, strongly influenced small tree age and height distributions. Proportionately more veglings than seedlings survived beyond 5-years or reached 30 cm in height. In both stand types, a higher proportion of Hw than Cw reached 30 cm in height. Advanced regeneration in the clearcut consisted of proportionally more Hw than Cw. 79 Higher proportions of small trees occurred on Lignomor, mineral, and moss substrates than on other substrates regardless of stand type. Small tree survival on undecomposed substrates was virtually nil. The importance of microtopography was related to substrate group and soil moisture and nutrient regime, and differed between stand types. In the clearcut, small tree presence and height growth beyond 30 cm were positively influenced by increased SNR on inclines/mounds; however, height growth beyond 30 cm was also negatively influenced by increasing SMR on inclines/mounds. In the old-growth, proportionally fewer small trees reached 30 cm in height on inclines/mounds, and proportionally fewer reached 5-years in age on wet organic or moss substrates located on inclines/mounds. Soil moisture and nutrient regime influences were less pronounced in the old-growth (except for a decrease in small tree presence with increasing SNR; and an increase in proportion of small trees which reached 5-years in age, with increasing SMR), and were related to microtopography, species, and substrate in the clearcut. In the clearcut, increased SNR negatively influenced small tree height growth beyond 30 cm, yet positively influenced height growth beyond 30 cm on Lignomor, compared with other substrates. Increased SMR negatively influenced survival beyond 5-years. Hw survival beyond 5-years was negatively influenced by increased SNR. Deer fern and reedgrass presence was more extensive in the clearcut than in the old-growth. In the clearcut, small tree presence was negatively influenced by increased fern presence. Trends indicate that increased grass/sedge presence may also be potentially limiting for small tree establishment and survival. To determine whether any of these conclusions apply generally to the CWHvh2 variant, the number of sites would have to be replicated, and true hypothesis tests constructed. These conclusions can serve as the basis for constructing hypotheses for future studies. Weak logistic regression models in this study may be attributable to the following: 1 . Other important variables; for example, climate temperature and precipitation history, cone crop cycles, faunal and fungal predation, light penetration through canopy gaps, and proximity of parent tree (cone-bearing or vegetative), were not included as part of the study design. 2. Competing vegetation was modeled based on above-ground cover, while ignoring potential below-ground coverage, which may have extended beyond a plot or penetrated into a plot from vegetation outside the plot. 3. Not enough samples were collected for the variable combinations, resulting in both structural and sampling zeros. 4. Plot contiguity likely significantly influenced height and age distributions, but was not adequately modeled in the regressions. Although logistic regression can be a powerful technique for analyzing multivariate data from a natural, uncontrolled ecosystem, it proved to be difficult and unreliable in this study. Despite that, the influence (positive or negative) of the variables in the models appeared to generally coincide with trends shown in graphs of the given small tree presence, height, or age class distributions. The use of this technique could be improved in the future by increasing the number of Block replicates within each site, increasing the number of plots within each Block (thereby increasing the sample sizes of the explanatory variables), adding other potentially important variables, using spatial statistics to model plot contiguity, improving non-competing 81 vegetation data modelling, and using exact logistic methods to analyze data with small sample sizes. Alternatively, techniques other than logistic regression should be considered to analyse height and age data. Generalized linear models19 is one possibility (V. LeMay, pers. comm.). 1 9 For complete details of this technique, SAS Institute (1994b) recommends consulting: McCullagh, P. and Nelder, J.A. 1989. Generalized linear models. Chapman and Hall. London. LITERATURE CITED Adams, D. L. and Mahoney, R.L. 1991. Effects of shade and competing vegetation on growth of western redcedar regeneration. WJAF 6(l):21-22. Agresti, A. 1996. An introduction to categorical data analysis. John Wiley and Sons, Inc. New York. 290pp. Alaback, P.B. and Tappeiner II, J.C. 1991. Response of western hemlock (Tsuga heterophylla) and early huckleberry (Vaccinium ovalifolium) seedlings to forest windthrow. Can. J. For. Res. 21:534-539. Alban, D.H. 1969. The influence of western hemlock and western redcedar on soil properties. Soil Sci. Soc. Am. Proc. 33(3):453-457. Andersen, H.E. 1955. Climate in Southeast Alaska in relation to tree growth. USDA For. Serv. Alaska For. Res. Cent. Station Paper No. 3. Juneau. 11pp. Armson, K.A. 1979. Forest soils. University of Toronto Press. Toronto. 390pp. Baker, H.G. 1989. Some aspects of the natural history of seed banks. In: Ecology of soil seed banks. M. A. Leek, V.T. Parker, and R.L. Simpson, (eds). Academic Press Inc. San Diego. 9-21. Battigelli, J.P., Berch, S.M. and Marshall, V.G. 1994. Soil fauna communities in two distinct but adjacent forest types on northern Vancouver Island, British Columbia. Can. J. For. Res. 24:1557-1566. Banner, A., and Puis, L. 1999. Excavator mounding to enhance productivity in coastal cedar-hemlock ecosystems near Prince Rupert, B.C.: some preliminary results. Unpublished manuscript. 3pp. Banner, A., Pojar, J. and Trowbridge, R. 1986. Representative wetland types of the northern part of the Pacific Oceanic wetland region. Research Report RR85008-PR. BC MOF. Victoria. 45pp. Banner, A., Pojar, J. and Kimmins, J.P. 1987. The bog-forest complex of north-coastal British Columbia. In: "Proceedings: Symposium '87 Wetlands/Peatlands. August 23-27, 1987. Edmonton." CD.A. Rubec and R.P. Overend (eds). International Peat Society. Canadian National Committee. Ottawa. p483-490. Banner, A., Hebda, R.J., Oswald, E.T., Pojar, J. and Trowbridge, R.. 1988. Wetlands of Pacific Canada. In: "Wetlands of Canada." National Wetlands Working Group. Canada Committee on Ecological Land Classification. Polyscience Publication Inc. Ottawa. 307-346. 83 Banner, A., Mackenzie, W., Haeussler, S., Thomson, S., Pojar, J. and Trowbridge, R. 1993. A field guide to site identification and interpretation for the Prince Rupert Forest Region. LMH#26. Parts 1 and 2. BC MOF. Victoria. BC Ministry of Forests. 1991. Silvicultural systems: their role in British Columbia's forest management. B.C. MOF. Victoria. 42pp. BC Ministry of Forests. 1993. Mid Coast Timber Supply Analysis. Integrated Resources Branch. B.C. Ministry of Forests. Victoria. 54pp. BC Ministry of Forests. 1995. Forest Practices Code of British Columbia Establishment to Free Growing Guidebook. Vancouver Forest Region. Victoria. 130pp. BC Ministry of Forests. 1997. Relational data dictionary 2.0a. Resource Inventory Branch. Victoria. Internet HTML version. BC Ministry of Forests. 1998. BC tree code list. Version 4.0. Vegetation Inventory Standards and Procedures Manual. Resource Inventory Branch. Victoria. Internet HTML Version. BC Ministry of Forests. 1999a. Special Data Summarization, May 10, 1999: Silviculture Information Access, ISIS/MLSIS Database. Information Current to April 29, 1999. Forest Practices Branch. Victoria. BC Ministry of Forests. 1999b. Timber Supply Review. Mid Coast timber supply area analysis report. June 1999. Province of British Columbia. Victoria. 114pp. Beaudry, P. and Banner, A. 1990. Enhancement of productivity in coastal cedar-hemlock ecosystems. Working Plan. SMFRA Project 15.2-Northern B.C. Ccast. BC MOF. Smithers. 12pp. Beatty, S.W. and Stone, E.L. 1986. The variety of soil microsites created by tree falls. Can. J. For. Res. 16(3):539-548. Bennett, J. 1996. Fertilization and salal removal for improving conifer regeneration on CH sites. In: "SCFTTRP: Salal Cedar Hemlock Integrated Research Program. Research Update #1: December 1996". CE. Prescott (ed). Fac. Forestry. UBC. Vancouver. Bergeron, J.F., Saucier, J.P., Robitaille, A., and Robert, D. 1992. Quebec forest ecological classification program. 1992. Forestry Chronicle. 68(l):53-63. Bergerud, W. 1996. Introduction to regression models: with worked forestry examples. Biom. Info. Hand. 7. Res. Br. B.C. MOF. Victoria, B.C. Work. Pap. 26/1996. Black, K.E., Harbron, C.G., Franklin, M., Atkinson, D. and Hooker, J.E. 1998. Differences in root longevity of some tree species. Tree Physiology- 18:259-264. 84 Boyd, RJ.Jr. 1959. Silvics ofWestern Redcedar. USDA Forest Service. Intermountain Forest and Range Experiment Station. Misc. Pub. No. 20. 14pp. Britton, N.L., and Brown, A.B. 1970. An illustrated flora of the Northern United States and Canada. Volumes 1, 2, and 3. Dover Publications Inc., New York. Brown, L. 1979. Grasses, an identification guide. Houghton Mifflin Co. New York. 240pp. Burns, R.M., and Honkala, B.H. (eds). 1990. Glossary, in: "Silvics of North America. Vol. 1, Conifers". Agriculture Handbook 654. USDA. Washington. 635-645. Cade-Menun, B. 1996. Phosphorus forms of podzolic soils and their use by western red cedar. In: SCHIRP: Salal cedar hemlock integrated research program. CE. Prescott. (ed.). Research Update #1. Fac. For. University of British Columbia. 44-45. Canada Soil Survey Committee. 1978. The Canadian system of soil classification. Publication 1646. Canadian Government Publishing Center. Ottawa. 164pp. Carter, R.E. and Klinka, K. 1992. Variation in shade tolerance of Douglas fir, western hemlock, and western redcedar in coastal British Columbia. For. Ecolog. Manage. 55:87-105. Casper, B.B., and Jackson, R.B. 1997. Plant competition underground. Annu. Rev. Ecol. Syst. 28:545-70. Chang, S.X., Weetman, G.F. and Preston, CM. 1996. Understory competition effect on tree growth and biomass allocation on a coastal old-growth forest cutover site in British Columbia. Forest Ecology and Management. 83:1-11. Chen, J., Franklin, J.F., and Spies, T.A. 1992. Vegetation responses to edge environments in old-growth Douglas-fir forests. Ecological Applications. 2(4):387-396. Clark, M.B. 1970. Seed production of hemlock and cedar in the interior wet belt region of British Columbia related to dispersal and regeneration. Research Note No. 51. BC For. Serv. Victoria. 11pp. Clayton, J.S., Ehrlich, W.A., Cann, D.B., Day, J.H., and Marshall, I.B. 1977. Soils of Canada. Vol. 1: Soil report. Research Branch. Canada Department of Agriculture. Ottawa. 243pp. Conover, W.J. 1980. Practical nonparametric statistics. John Wiley and Sons, Inc. New York 493pp. Curran, M.P. and Dunsworth, B.G. 1988. Coastal western redcedar regeneration: problems and potentials. In: Western redcedar - does it have a future? Smith, N.J. (Ed.). Conference Proceedings. University of British Columbia. 20-32. 85 Daniel, T.W., Helms, J.A., and Baker, F.S. 1979. Principles of silviculture. 2nd ed. McGraw-Hill Book Company. New York. 500pp. Daniels, L.D., and Klinka, K. 1996. The dynamics of old-growth Thuja z Tsuga forests near Vancouver, British Columbia. Radiocarbon. 1996:379-393. Daubenmire, R.F. 1974. Plants and environment: a textbook on plant autecology. 3rd ed. John Wiley and Sons. New York. 422pp. Davis, L.S. and Johnson, K^N. 1987. Forest management. 3rd ed. McGraw-Hill Book Company. New York. 790pp. Deal, R.L., Oliver, CD., and Bormann, B.T. 1991. Reconstruction of mixed hemlock-spruce stands in coastal southeast Alaska. Can. J. For. Res. 21:643-654. DeLong, D.L. 1997. A retrospective investigation of advanced western redcedar regeneration in the ICHwkl, ICHmw2, and ICHmwl of the Nelson Forest Region. Experimental Project 1174. Res. Br. BC Min. For. Victoria. Work Pap. 25/1997. 18pp. Department of Agriculture. 1972. Soils of Canada. 1:5,000,000 map. Surveys and Mapping Branch. Department of Energy, Mines and Resources. Ottawa. Department of Agriculture. 1975. Soil climates of Canada. 1:10,000,000 map. Surveys and Mapping Branch. Department of Energy, Mines and Resources. Ottawa. Edwards, D.G.W., and Leadem, C.L. 1988. The reproductive biology of western red cedar with some observations on nursery production and prospects for seed orchards. In: Western redcedar—does it have a future? Smith, N.J. (Ed.), conference Proceedings, University of British Columbia, Faculty of Forestry. 102-113. Ehrenfeld, J.G. 1995. Microtopography and vegetation in Atlantic white cedar swamps: the effects of natural disturbances. Can. J. Bot. 73:474-484. Eis, S. 1974. Root system morphology of western hemlock, western redcedar, and Douglas-fir. Can. J. For. Res. 4:28-38. Fausey, N.R. and Lai, R. 1990. Soil wetness and anaerobiosis. In: "Advances in Soil Science. Vol. 11. Soil Degradation." Springer-Verlag. New York. pl73-186 Fenner, M. 1985. Seed ecology. Chapman and Hall. New York. 151pp. Fogel, R. 1990. Root turnover and production in forest trees. Hortscience. 25(3):270-273. Fons, J., and Klinka, K. 1998. Temporal variations of forest floor properties in the Coastal Western Hemlock zone of southern British Columbia. Can. J. For. Res. 28(4):582-590. 86 Fox, C.A., Preston, C M . , and Fyfe, CA. 1994. Micromorphological and 1 3 C NMR characterization of a Humic, Lignic, and Histic Folisol from British Columbia. Can. J. Soil Sci. 74:1-15. Franklin, J.F. 1961. A guide to seedling identification for 25 conifers of the Pacific NortHwest. USDA For. Serv. Pacific Northwest Forest and Range Experiment Station. Portland. 65pp. Fraser, L.H., Chanway, CP., and Turkington, R. 1995. The competitive role of Gaultheria shallon on planted western hemlock and western redcedar saplings on northern Vancouver Island. For. Ecol. Manage. 75:27-39. Frelich, L.E., Sugita, S., Reich, P.B., Davis, M.B., and Friedman, S.K. 1998. Neighbourhood effects in forests: implications for within-stand patch structure. J. Ecology. 86(1): 149-162. Garman, E.H. 1951. Seed production by conifers in the coastal region of British Columbia related to dissemination and regeneration. Technical Publication T.35. Dept. Lands and Forests. BC Forest Service. Victoria. 47pp. Gashwiler, J.S. 1969. Seed fall of three conifers in west-central Oregon. For. Sci. 15(3):290-295. Godman, R.M. 1953. Moss retards regeneration in Southeast Alaska. USDA For. Ser. Alaska For. Res. Cent. Technical Note 18. Juneau. 1pp. Godman, R.M. and Gregory, R.A. 1953. Physical soil characteristics related to site quality in climax stands of Southeast Alaska. USDA For. Serv. Alaska For. Res. Cent. Tech. Note 17. 1 pp. Gove, P.B. (ed). 1986. Webster's third new international dictionary of the English language unabridged. Merriam-Webster Inc. Springfield. 2663pp. Graham, R.T. 1982. Influence of tree and site factors on western redcedar's response to release: a modeling analysis. USDA Forest Service. Intermountain forest and Range Experiment Station. Research Paper: LNT-296. 19pp. Gray, A.N. and Spies, T.A. 1997. Microsite controls on tree seedling establishment in conifer forest canopy gaps. Ecology. 78(8):2458-2473. Green, R.N., and Klinka, K. 1994. A field guide for site identification and interpretation for the Vancouver Forest Region. LMH#28. BCMOF. Victoria. 285pp. Green, R.N., Trowbridge, R.L. and Klinka, K. 1993. Towards a taxonomic classification of humus forms. Forest Science Monograph 29. Society of American Foresters. Bethesda, MD. 87 Gregory, R.A. 1957. Some silvicultural characteristics of western redcedar in southeast Alaska. Ecology. 38(4): 646-649. Habeck, J.R. 1968. Forest succession in the Glacier Park cedar-hemlock forests. Ecology. 49(5):872-880 Habeck, J.R. 1978. A study of climax western redcedar (Thuja plicata Donn.) forest communities in the Selway-Bitterroot Wilderness, Idaho. Northest Sci. 52(l):67-77. Haeussler, D., Coates, D. and Mather, J. 1990. Autecology of common plants in British Columbia: a literature review. FRDA Report 158. BC Min. For. Res. Br. Victoria. 272pp. Harmon, M.E. 1987. The influence of litter and humus accumulations and canopy openness on Picea sitchensis (Bong.) Carr. and Tsuga heterophylla (Raf.) Arg. seedlings growing on logs. Can. J. For. Res. 17(12):1475-1479. Harmon, M.E. and Franklin, J.F. 1989. Tree seedlings on logs in Picea-Tsuga forests of Orgeon and Washington. Ecology. 71(l):48-59. Hennon, P.E. 1992 Survival and growth of planted Alaska-cedar seedlings in southeast Alaska. Tree Planters' Notes 43(3):60-66. Hetherington, J.C. 1965. The dissemination, germination and survival of seed on the west coast of Vancouver Island from western hemlock and associated species. Research Note No. 39. Research Division. BC For. Serv. Victoria. 22pp. Hollander, M. and Wolfe, D.A. 1999. Nonparametric statistical methods. John Wiley and Sons. New York. 787pp. Hosmer, D.W., and Lemeshow, S. 1989. Applied logistic regression. John Wiley and Sons, Inc. New York. 307pp. Hubbard, W.A. 1969. The grasses of British Columbia. Handbook No. 9. British Columbia Provincial Museum. Victoria. 205pp. Jackson, W. 1988. Research methods: rules for survey design and analysis. Prentice-Hall Canada Inc. Scarborough. 290pp. James, G.A. 1959. Seed production in a scrub stand. USDA For. Serv. Technical Note No. 43. Alaska Forest Research Center. Juneau. 3pp. Jones, R.K. 1986. Soil survey and its use in forest management in Ontario. In: Site classification in relation to forest management. COJFRC Symposium Proceedings. August 27-29,1985. Sault Ste. Marie. O-P-14. Great Lakes Forestry Centre. Canadian Forestry Service. Sault Ste. Marie. 29-37. 88 Keenan, R.J., Prescott, C.E., and Kimrnins, J.P. 1993. Mass and nutrient content of woody debris and forest floor in western redcedar and western hemlock forests on northern Vancouver Island. Can. J. For. Res. 23:1052-1059. Keenan, R.J., Prescott, CE., Kimrnins, J.P., Pastor, J. and Dewey, B. 1996. Litter decomposition in western redcedar and western hemlock forests on northern Vancouver Island, British Columbia. Can. J. Bot. 74:1626-1634. King, D.A. 1991. Tree allometry, leaf size and adult tree size in old-growth forests of western Oregon. Tree Physiology. 9:369-381. Klinka, K., Krajina, V.J., Ceska, A., and Scagel, A.M. 1989. Indicator plants of coastal British Columbia. University of British Columbia Press. Vancouver. 288pp. i Klinka, K., Carter, R.E., and Feller, M.C 1990. Cutting old-growth forests in British Columbia: ecological considerations for forest regeneration. Northest Environmental Journal. 6(2):221-242. Klinka, K., Carter, R.E., and Kayahara, G.J. 1994. Forest reproduction methods for coastal British Columbia: principles, criteria, and a stand selection guide. Forestry Chronicle. 70(5):569-577. Klinka, K., Lavkulich, L.M., Wang, Q., and Feller, M.C. 1995. Influence of decaying wood on chemical properties of forest floors and surface mineral soils: a pilot study. Ann. Sci. For. 52:523-533. Klinka, K., Chen, H.Y.H., Wang, Q., and deMontigny, L. 1996. Forest canopies and their influence on understory vegetation in early-seral stands on west Vancouver Island. NortHwest Sci. 70(3): 193-200. Kobe, RK. and Coates, K.D. 1997. Models of sapling mortality as a function of growth to characterize interspecific variation in shade tolerance of eight tree species of nortHwestern British Columbia. Can. J. For. Res. 27:227-236. Krajina, V.J. 1965. Biogeoclimatic zones and biogeocoenoses of British Columbia. In: Ecology of Western North America. Vol.1. V.J. Kranina (ed.). University of British Columbia. Vancouver. 1-17. Krajina, V.J. 1969. Ecology of forest trees in British Columbia. In: Ecology of Western North America. V.J. Krajina and R.C. Brooke (eds). Vol 2(1): 1-146. University of British Columbia. Vancouver. Lambers, H., and Poorter, H. 1992. Inherent variation in growth rate between higher plants: a search for physiological causes and ecological consequences. Adv. Ecol. Res. 23:187-261. 89 Leadem, C. 1996. A guide to the biology and use of forest tree seeds. LMH30. BCMOF. Victoria. 20pp. Leadem, C.L., Gillies, S.L., Yearsley, K., Sit, V., Spittlehouse, D.L., and Burton, PJ. 1997. Field studies of seed biology. LMH40. BCMOF. Victoria. 196pp. Leaphart, CD. and Grismer, M.A. 1974. Extent of roots in the forest soil mantle. J. For. 6:358-359. Lertzman, K.P. and Krebs, CJ. 1991. Gap-phase structure of a subalpine old^ -growth forest. Can. J. For. Res. 21:1730-1741. Lieffers, V.J. and Stadt, K.J. 1994. Growth of understory Picea glauca, Calamagrostis canadensis, and Epilobium angustifolium in relation to overstory light transmission. Can. J. For. Res. 24:1193-1198. Littell, R.C., Milliken, G.A., Stroup, W. W., and Wolfinger, RD. 1996. SAS system for mixed models. SAS Institute Inc., Gary, NC. 633pp. Luttmerding, H.A., Demarchi, D.A., Lea, E.C, Meidinger, D.V., and Void, T. (eds). 1990. Describing ecosystems in the field. 2nd ed. MOE Manual 11. BC Ministry of Environment. Victoria. 213pp. Lyons, CP. 1976. Trees, shrubs, and flowers to know in British Columbia. J.M. Dent and Sons (Canada) Ltd. Toronto. 194pp. MacBean, A.P. 1941. A study of the factors affecting the reproduction of western hemlock and its associates in the Quatsino Region, Vancouver Island. British Columbia Department of Lands. Forest Branch. Victoria. 37pp. Marra, J.L. and Edmonds, R.L. 1994. Coarse woody debris and forest floor respiration in an old-growth coniferous forest on the Olympic Peninsula, Washington, USA. Can. J. For. Res. 24:1811-1817. Marra, J.L. and Edmonds, R.L. 1996. Coarse woody debris and soil respiration in a clearcut on the Olympic Peninsula, Washington, U.S.A. Can. J. For. Res. 26:1337-1345. Marschner, H. 1986. Exposed mineral nutrition of higher plants. Academic Press. London. 674pp. Matzner, E., Pijpers, M . , Holland, W., and Manderscheid, B. 1998. Aluminum in soil solutions of forest soils: influence of water flow and soil aluminum pools. Soil Cience Society of America Journal. 62(2):445-453. McCune, B., and Allen, T.F.H. 1985. Will similar forests develop on similar sites? Can. J. Bot. 63:367-376. 90 Meades, WJ. and Roberts, B.A. 1992. A review of forest site classification activities in Newfoundland and Labrador. Forestry Chronicle. 68(l):25-33. Meidinger, D. and Pojar, J. 1991. Ecosystems of British Columbia. B.C. MOF. Special Report Series No. 6. 95-112. Messier, C. 1993. Factors limiting early growth of western redcedar, western hemlock and Sitka spruce seedlings on ericaceous-dominated clearcut sites in coastal British Columbia. For. Ecol. Manage. 60:181-206. Messier, C, Kimmins, J.P., Bunnell, F.L., and McCann, R.K. 1990. Understanding salal as a competitor. In: Vegetation management - an integrated approach. Proceedings of the fourth annual vegetation management workshop. Vancouver. E. Hamilton (compiler). FRDA Report 109. Victoria. p40-42. Milliken, G.A., and Johnson, D.E. 1989. Analysis of messy data: non-replicated experiments. Volume 2.. Chapman and Hall. New York. 199pp. Minore, D., and Smith, CE. 1971. Occurrence and growth of four northwestern tree species over shallow water tables. USDA For. Serv. Research Note. PNW-160. Portland. 9pp. Minore, D. 1979. Comparative autecological characteristics of northwestern tree species - a literature review. USDA For. Serv. General Technical Report PNW-87. 73pp. Minore, D. 1983. Western redcedar - a literature review. USDA Forest Service. General Technical Report PNW-150. 70pp. Minore, D. 1990. Thuja plicata Donn ex D. Don. in: "Silvics of North America. Vol.1, Conifers". Burns, R.M., and Honkala, B.H. (eds). Agriculture Handbook 654. USDA. Washington. 590-600. Mitchell, A.K. and Arnott, J.R. 1995. Effects of shade on the morphology and physiology of amabilis fir and western hemlock seedlings. New Forests. 10:79-98. Mitchell, K.J. and Polsson, K.R. 1988. Site index curves and tables for British Columbia: coastal species. FRDA Report 037. BC MOF. Research Branch. Victoria. 29pp. Moeur, M. 1997. Spatial models of competition and gap dynamics in old-growth Tsuga heterophylla/Thujaplicata forests. For. Ecol. Managem. 84:175-186. Nagelkerke, N.J.D. 1991. A note on a general definition of the coefficient of determination. Biometrika. 78: 691-692. Neiland, B.J. 1971. The forest-bog complex of Southeast Alaska. Vegetatio. 22:1-63. 91 Oliver, CD. 1979. Growth response of suppressed hemlocks after release. In: Proceedings -Western Hemlock Management Conference. W. A. Atkinson and R. J. Zasoski (eds). College of Forest Resources. University of Washington. 266-272. Oliver, CD. and Larson, B. 1996. Forest stand dynamics. John Wiley and Sons, Inc. New York. 520pp. Oliver, CD., Nystrom, M.N., and DeBell, D.S. 1988. Coastal stand silvicultural potential for western redcedar. In: Western redcedar - does it have a future? Smith, N.J. (Ed.). Conference Proceedings. University of British Columbia. Faculty of Forestry. Vancouver. 39-46. Owens, J.N., and Molder, M. 1984a. The reproductive cycles of western redcedar and yellow-cedar. BC MOF. Victoria. 29pp. Owens, J.N., and Molder, M. 1984b. The reproductive cycles of western and mountain hemlock. BC MOF. Victoria. 35pp. Owens, J.N., and Molder, M. 1984c. The reproductive cycle of Lodgepole pine. BCMOF. Victoria. 29pp. Packee, E.C. 1990. Tsuga heterophylla (Raf.) Sarg. in: "Silvics of North America. Vol.1, Conifers". R.M. Burns and B.H. Honkala (eds). Agriculture Handbook 654. USDA. Washington. 613-622. Parker, T. and Johnson, F.D. 1987. Branching and terminal growth of western redcedar. NortHwest Science. 61(1):7-12. Phillips, D. 1990. The climates of Canada. Environment Canada. Ottawa. 176pp. Pluth, D.J. and Corns, I.G.W. 1983. Productivity of conifers in western Canadian boreal forests in relation to selected environmental factors. I.U.F.R.O. Symposium on forest site and continuous productivity. USDA Forest Service. General Technical Report. PNW-163. 101-111. Pojar, J. and MacKinnon, A. (eds). 1994. Plants of Coastal British Columbia. B.C. MOF and Lone Pine Publishing. Vancouver. 527pp. Ponnamperuma, F.N. 1972. The chemistry of submerged soils. In: Advances in Agronomy. 24:29-96. Academic Press. Amsterdam. Powelson, R.A. and Lieffers, V.J. 1991. Growth of dormant buds on severed rhizomes of Calamagrostis canadensis. Can. J. Plant Sci. 71:1093-1099. Powelson, R.A., and Lieffers, V. J. 1992. Effect of light and nutrients on biomass allocation in Calamagrostis canadensis. Ecography. 15:31-36. 92 Prescott, CE. 1996. Introduction. In: SCHIRP: Salal cedar hemlock integrated research program. CE. Prescott. (ed.). Research Update #1. Fac. For. University of British Columbia. Pi-Prescott, C.E., Weetman, G.F., DeMontigny, L.E., Preston, CM., and Keenan, R.J. 1995. Carbon chemistry and nutrient supply in cedar-hemlock and hemlock-amabilis fir forest floors. In: Carbon forms and functions in forest soils. WW. McFee and J.M. Kelly (eds). Soil. Sci. Soe. Am. Wisonsin. 377-396. Prescott, C.E., Weetman, G.F., and Barker, J.E. 1996. Causes and amelioration of nutrient deficiencies in cutovers of cedar-hemlock forests in coastal British Columbia. Forestry Chronicle. 72(3):293-302. Puis, L.J. 1998. Enhancement of productivity in a coastal cedar-hemlock ecosystem: 5th year results. B.Sc. F. Graduating Essay. Fac. For. University of British Columbia. 23pp. Putz, F.E., Coley, P.D., Lu, K., Montalvo, A., and Aiello, A. 1983. Uprooting and snapping of trees: structural determinants and ecological consequences. Can. J. For. Res. 13:1011-1020. Rigg, G.B. and Harrar, E.S. 1931 The root systems of trees growing in Sphagnum. Am. J. Bot. 6(18):391-397. Rivard, P.G., Woodard, P.M., and RotHwell, R.L. 1990. The effect of water table depth on white spruce (Picea glauca) seedling growth in association with the marsh reed grass (Calamagrostis canadensis) on wet exposed mineral soil. Can. J. For. Res. 20:1553-1558. Rowell, D.L. 1981. Oxidation and reduction. In: "The chemistry of soil processes". Edited by D.J. Greenland and M.H.B. Hayes. John Wiley and Sons Ltd. New York. 401-461. Runkle, J.R. 1981. Gap regeneration in some old-growth forests of the eastern United States. Ecology. 62(4): 1041-1051. Salonius, P.O. 1983. Effects of organic-mineral soil mixtures and increasing temperature on the respiration of coniferous raw humus material. Can. J. For. Res. 13:102-107. SAS Institute Inc., 1994a. SAS/STAT User's Guide, Version 6, Fourth Edition, Volume 1. SAS Institute Inc. Cary, NC. 943 pp. SAS Institute Inc., 1994b. SAS/STAT User's Guide, Version 6, Fourth Edition, Volume 2. SAS Institute Inc. Cary,NC. 846 pp. SAS Institute Inc., 1997. SAS/STAT Software: Changes and Enhancements through Release 6.12. SAS Institute Inc. Cary, NC. 1162pp. 93 Schmidt, R.L. 1955. Some aspects of western redcedar regeneration in the coastal forests of British Columbia. Research Notes. No. 29. British Columbia Forest Service. Department of Lands and Forests. Victoria. 10pp. Scholes, R.J., and Archer, S.R. 1997. Tree-grass interactions in savannas. Annual Review of Ecology and Systematics. 28:517-544. Sharpe, G.W. 1974. Western redcedar. College of Forest Resources. University of Washington. 144pp. Soil Classification Working Group. 1998. The Canadian system of soil classification. Research Branch. Agriculture and Agri-Food Canada. Publication 1646 3rd ed. NRC Research press. Ottawa. 187pp. Sokal, R.R., and Rohlf, F.J. 1981. Biometry. 2nd Ed. W.H. Freeman and Company. New York. 859pp. Soltis, P., Soltis, D.E., and Holsinger, K.E. 1988. Estimates of intragametophytic selfing and interpopulational gene flow in homosporous ferns. Amer. J. Bot. 75(11):1765-1770. Soos, J. and Walters, J. 1963. Some factors affecting the mortality of western hemlock and western redcedar germinants and seedlings. Research Paper No. 56. Faculty of Forestry. University of British Columbia. Vancouver. 12pp. Spies, T.A., Franklin, J.F., and Klopsch, M. 1990. Canopy gaps in Douglas-fir forests of the Cascade Mountains. Can. J. For. Res. 20:649-658. Stokes, M.E., Davis, C.S., and Koch, G.G. 1995. Categorical data analysis using the SAS system. SAS Institute Inc. Cary, NC. 499pp. Stoodley, G.E. 1925. Reproduction studies in hemlock-cedar and hemlock-balsam-spruce types. BC Forest Service Report 42. Victoria. 31pp. Tappeiner, J.C, and Zasada, J.C. 1993. Establishment of salmonberry, salal, vine maple, and bigleaf maple seedlings in the coastal forests of Oregon. Can. J. For. Res. 23(9): 1775-1780. Tappeiner, J., Zasada, J., Ryan, P., and Newton, M. 1991. Salmonberry clonal and population structure: the basis for a persistent cover. Ecology. 72(2):609-618. Taylor, T.M.C. 1973. The ferns and fern-allies. Handbook No. 12. British Columbia Provincial Museum. 172pp. Turner, D.P., and Franz, E.H. 1985. Size class structure and tree dispersion patterns in old-growth cedar-hemlock forests of the northern Rocky Mountains (USA). Oecologia. 65:52-56. 94 Ugolini, F.C. and Mann, D.H. 1979. Biopedological origin of peatlands in Southeast Alaska. Nature. 281(4):366-368. Vitt, D.H., Marsh, J.E., and Bovey, R.B. 1988. Mosses, lichens and ferns of Northwest North America. Lone Pine Publishing. Edmonton. 296pp. Weetman, G.F., Fournier, R., Barker, J., Schnorbus-Panaozzo, E., and Germain, A. 1989a. Foliar analysis and response of fertilized chlorotic Sitka spruce plantations on salal-dominated cedar-hemlock cutovers on Vancouver Island. Can. J. For. Res. 19(12): 1501-1511. Weetman, G.F., Fournier, R., Barker, J., and Schnorbus-Panozzo, E. 1989b. Foliar analysis and response of fertilized chlorotic western hemlock and western redcedar reproduction on salal-dominated cedar-hemlock cutovers on Vancouver Island. Can. J. For. Res. 19(12): 1512-1520. Williamson, R.L. 1979. Natural regeneration of western hemlock. In: Proceedings - Western Hemlock Management Conference. W.A. Atkinson and R.J. Zasoski (eds). College of Forest Resources. University of Washinton. 166-169. Williamson, R.L. and Ruth, RH. 1976. Results of shelterwood cutting in western hemlock. USDA For. Serv. Res. Pap. PNW-201. Portland. 25pp. Wright, E.F., Coates, K.D., and Bartemucci, P. 1998. Regeneration from seed of six tree species in the interior cedar-hemlock forests of British Columbia as affected by substrate and canopy gap position. Can. J. For. Res. 28(9): 1352-1364. Zach, L.W. 1950. A northern climax, forest or muskeg? Ecology 31:304-306. 95 APPENDIX 1: SUBSTRATES All substrates found in all plots are listed in Table 25. Each substrate was assigned a broad substrate group (as shown in the table) to facilitate data analysis. A depth of 30 cm beneath the surface of the forest floor was used to classify each substrate according to a humus form or soil order. Soil orders were classified to a depth of 30 cm where the humus form was non-existent (as in the case of exposed organic soils). Distinguishing between humus forms and soil orders was sometimes difficult, and a certain amount of subjectivity is inherent in the classifications of substrate data. For example: in determining whether an organic substrate was a "Humisol" versus a "Humimor", an LFH was looked for. It the substrate did not have a discernible LFH layer, and had only an O (organic soil) layer, then it was classified as a Humisol. If it had an LFH layer over the 0 layer, then it was classified a Humimor. Diagrams on page 17 of Green et al. 1993, and pages 84 and 85 of the Canadian Soil Survey Committee (1978) were used to assist in making the distinctions between soil orders and humus forms. The same principles applied when classifying Fibrisols versus Fibrimors, Mesisols versus Mesimors, and so forth. Given that both wetland mors or soils were both assigned to the "wetland organic" group, these classification difficulties were not expected to affect the results. The "upland organic" group was named thus to distinguish non-saturated organic substrates from the saturated organics in the "wetland organic" group. Also, a certain degree of subjectivity exists between what was identified as "Lignomor" versus what was identified "undecomposed coarse wood", since ocular methods to identify substrates in some cases led to difficulties in distinguishing between the first stages of a Lignomor, versus mostly undecomposed wood beginning initial stages of decomposition. A substrate was identified as a Lignomor if it had 96 litter and partly decomposed wood (Fw) layers over a humus layer which consisted of primarily (>35% of volume of solids) of wood in various stages of decomposition (Green et al. 1993). The stage of decay (i.e., partly decayed or fully decayed) in the Lignomor was not distinguished. Those coarse woody substrates in which the cambium was mostly intact, with or without bark, and with or without a litter layer, were identified as undecomposed coarse wood. In some large logs, if the outer surface 1 cm was beginning to show inconsistent evidence of decay (that is, not all portions of the log were beginning to decay), it was still identified as undecomposed coarse wood, if beyond the outer surface, the wood was impenetrable to the insertion of a pen. Although most substrates were assigned to the substrate group which best described the most predominant type of substrate, the "mineral" group was the only substrate group which contained substrates other than predominantly pure exposed mineral soils. For example, if a predominantly "Lignomor" or "Resimor" was overtopped with a layer of exposed mineral soil, then the substrate was grouped with the "mineral" group. The rationale for this grouping is based on the premise that mineral soil, contacting LF and H layers of organic horizons, appears to stimulate microbial decomposition of organic matter and plant nutrient release into soil solution (Salonius, 1983), thereby improving overall substrate nutrition. Table 25. Substrates found throughout the study area. Substrate Substrate Codes Substrate Descriptions Substrate Group sphagnum/Fibrimor 1 sphagnum/Mesimor la sphagnrnn/Humisol lb sphagnum moss lc sphagnum/Resimor Id sphagnum/Hydromor le fibric sphagnimor lg Fibrimor (see Green et al. 1993, page 19) dominated by thick layer Moss (greater than 10 cm) Sphagnum mosses Mesimor (see Green etal. 1993, page 20) dominated by thick (greater Moss than 10 cm) layer Sphagnum mosses Humisol (see Canadian Soil Survey Committee, 1978, page 89) Moss overtopped with a thick (greater than 10 cm) layer of Sphagnum mosses Thick layer of Sphagnum mosses exceeding 30 cm in depth Moss Resimor (see Green et al. 1993, page 18) overtopped by thick (greater Moss than 10 cm) layer of Sphagnum mosses Hydromor (see Green et al. 1993, page 19) overtopped by thick Moss (greater than 10 cm) layer of Sphagnum mosses Fibrimor (see Green et al. 1993, page 19) dominated by tluck (greater Moss than 10 cm) layer of Sphagnum mosses 97 Substrate Substrate Codes Substrate Descriptions Substrate Group mixed moss mor lh Fibrimor (see Green etal. 1993, page 19) dominated by thick (greater than 10 cm) layer of mixed mosses Moss Mesimor 2 Mesimor (see Green et al. 1993, page 20) Wetland organic fibric Mesimor 2a Mesimor. Top tier comprised primarily of grasses. Lower tier comprised of decomposed, and partially decomposed, unidentifiable fibres, (see Green et al. 1993, page 20) Wetland organic Mineric Mesimor 2b Mesimor intermixed with exposed organic mineral soil particles (see Green etal. 1993, page 13 and 20) Wetland organic slash-Mesimor 2c Mesimor (see Green et al. 1993, page 20) overtopped by slash Wetland organic Fibrimor 3 Fibrimor (see Green et al. 1993, page 19) dominated by fibres predominantly other than Sphagnum Wetland organic Fibrimor-Folisol 3a shallow Fibrimor (see Green et al. 1993, page 19) over Folisol (see Canadian Soil Survey Committee, 1978, page 91) Wetland organic Mineric Fibrimor 3b Fibrimor intermixed with exposed organic mineral soil particles (see Green etal. 1993, page 13 and 19) Wetland organic Mineric Fibrisol 3c Fibrisol (see Canadian Soil Survey Committee, 1978, page 87) intermixed with exposed organic mineral soil particles (see Green et al. 1993, page 13) Wetland organic moss Fibrimor 3d Fibrimor (see Green etal. 1993, page 19) dominated by mixed mosses Wetland organic sphagni-Fibrimor 3e Fibrimor (see Green et al. 1993, page 19) dominated by decomposing Wetland organic Sphagnum moss fibres humic Fibrimor 3f Fibrimor (see Green et al. 1993, page 19) containing some well Wetland organic decomposed organic matter (less than 50% of the profile) exposed mineral 4 exposed mineral soil (little or no organic matter present) Mineral podzol 4a non-exposed podzol (see Canadian Soil Survey Committee, 1978, Mineral page 93) under very shallow humus (<10cm) not further classified as humic, ferro-humic, or humo-ferric p-humic podzol 4b ploughed (disturbed through harvesting activity), humic podzol (see Mineral Canadian Soil Survey Committee, 1978, page 25 and 96) turbic exposed 4c turbic (windthrow disturbance within last 3-5-years) exposed mineral Mineral mineral soil (little organic matter present) xmin-rutt w litter 4d exposed mineral soil machine rutt overtopped with litter Mineral xmin-rock 4e exposed mineral soil over bedrock Mineral pellia-min 4/ exposed mineral soil overtopped by liverwort (Pellia neesiana) Mineral pmin 4g ploughed (disturbed by harvesting) exposed mineral soil (essentially same as #4c) Mineral g min 4h granitic exposed mineral soil Mineral Lignomor 5 Lignomor (see Green et al. 1993, page 18) Lignomor Lignomor-exposed 5a Lignomor (see Green et al. 1993, page 18) over exposed mineral soil Lignomor mineral Lignomor-moss 5b Lignomor (see Green et al. 1993, page 18) overtopped by mosses (<10cm thick) other than Sphagnum Lignomor turbic Lignomor 5c overturned, disturbed (through unknown events) Lignomor (see Green etal. 1993, page 13 and 18) Lignomor sphagni/Lignomor 5d Lignomor (see Green et al. 1993, page 18) overtopped by Sphagnum mosses (<10cm thick) Lignomor r>Lignomor-G-min 5e ploughed (disturbed through harvesting activity) Lignomor (see Mineral Green et al. 1993, page 13 and 18) overtopped by granitic exposed mineral soil resi-Lignomor 5f Lignomor overtopped by thin (<10cm) Resimor (see Green et al. 1993, page 18) Lignomor p-Lignomor 5g ploughed (disturbed through harvesting activity) Lignomor (see Green et al. 1993, page 13 and 18) (same as 5c) Lignomor lignic Folisol 5h Folisol (see Canadian Soil Survey Committee, 1978, page 91) Upland organic overtopped by lignic material (see Green et al. 1993, page 13) turbic Lignomor-G- 5i overturned, disturbed (through unknown events) Lignomor (see Green Mineral min et al. 1993, page 13 and 18) overtopped by granitic exposed mineral soil 98 Substrate Substrate Codes Hydromor 6 Resimor 7 Resi-Folisol 7a Resimor/exposed 7b mineral (podzol) sphagni-Resimor 7c Resimor/bedrock 7d t-Resimor/min 7e t-Resimor 7f fibric Resimor 7g lignic Resimor 7h sphagni- Resimor- 7i gleysol Resimor-blech-gaul 7j litter Resimor/G min 7k p-Resimor 71 p-Resimor-exposed 7m mineral (podzol) UCWD 8 p-UCWD-G min 8a moss-UCWD log 8b p-UCWD/humic 8c felled stump 8d UCWD/water 8e Folisol 9 Mesi-Folisol 9a Folisol-G-min 9b p-Folisol-G-min 9c p-Folisol-G- 9d min/slash tibric Folisol-rock 9e Substrate Descriptions Substrate Group Hydromor (see Green et al. 1993, page 19) Resimor (see Green et al. 1993, page 18) Folisol (see Canadian Soil Survey Committee, 1978, page 91) overtopped by thin (<10cm) Resimor (see Green et al. 1993, page 18) Resimor (see Green et al. 1993, page 18) over podzolic (see Canadian Soil Survey Committee, 1978, page 93) exposed mineral soil Resimor (see Green et al. 1993, page 18) overtopped by thin layer (less than 10 cm) of Sphagnum mosses bedrock overtopped by thin (<10cm) Resimor (see Green et al. 1993, page 18) overturned, disturbed (through unknown events) Resimor (see Green etal. 1993, page 13 and 18) overtopped by exposed mineral soil overturned, disturbed (through unknown events) Resimor (see Green etal. 1993, page 13 and 18) Resimor (see Green et al. 1993, page 18) dominated by fibric materials (i.e., fine roots) Resimor (see Green et al. 1993, page 13 and 18) dominated by decaying wood shallow (less than 20 cm) Resimor (see Green et al. 1993, page 18) overtopped by thin layer (less than 10 cm) of Sphagnum mosses over gleysol Resimor (see Green et al. 1993, page 13 and 18) dominated by deerfern and salal litter Resimor (see Green et al. 1993, page 18) over granitic exposed mineral soil ploughed (by harvesting activity) Resimor (see Green et al. 1993, page 13 and 18) ploughed (by harvesting activity) Resimor (see Green ci al. 1993, page 13 and 18) overtopped by podzolic (see Canadian Soil Survey Committee, 1978, page 93) exposed mineral soil undecomposed coarse woody material (slash, fallen trees) ploughed (by harvesting activity) undecomposed coarse woody material overtopped by granitic exposed mineral soil log overtopped by moss ploughed (by harvesting activity) undecomposed coarse woody material overtopped by humic (well decomposed organic) material felled stump undecomposed coarse woody material over water Wetland organic Upland organic Upland organic Upland organic Upland organic Upland organic Mineral Upland organic Upland organic Upland organic Upland organic Upland organic Upland organic Upland organic Mineral Undecomposed Coarse Wood Mineral Undecomposed Coarse Wood Undecomposed Coarse Wood Undecomposed Coarse Wood Undecomposed Coarse Wood Upland organic Upland organic Folisol (see Canadian Soil Survey Committee, 1978, page 91) Folisol (see Canadian Soil Survey Committee, 1978, page 91) overtopped by Mesimor (see Green et al. 1993, page 20) Folisol (see Canadian Soil Survey Committee, 1978, page 91) overtopped by granitic exposed mineral soil ploughed (by harvesting activity) Folisol (see Canadian Soil Survey Committee, 1978, page 91) overtopped by granitic exposed mineral soil ploughed (by harvesting activity) Folisol (see Canadian Soil Survey Committee, 1978, page 91) overtopped by granitic exposed mineral soil and slash shallow Folisol (see Canadian Soil Survey Committee, 1978, page 91) Upland organic dominated by litter (fine roots, leaves, needles, etc.) over bedrock (greater proportion of L layer compared to F and H) Mineral Mineral Mineral 99 Substrate Substrate Substrate Descriptions Substrate Codes Group resi -Folisol/G-min 9f Folisol (see Canadian Soil Survey Committee, 1978, page 91) Mineral containing greater portion of F layer compared with L and H, over granitic exposed mineral soil Upland organic p-Folisol 9g ploughed (by harvesting activity) Folisol (see Canadian Soil Survey Committee, 1978, page 91) min Folisol/bedrock 9h shallow Folisol (see Canadian Soil Survey Committee, 1978, page 91) Mineral dominated by exposed mineral soil over bedrock humic Folisol 9i Folisol (see Canadian Soil Survey Committee, 1978, page 91) Upland organic containing greater portion of H layer compared with L and F Moss moss (sphagnum) 10 Sphagnum moss (less than 30 cm deep) over water over water Wetland organic Humimor 11 Humimor (see Green et al. 1993, page 17) Humisol 11a Humisol (see Canadian Soil Survey Committee, 1978, page 89) Wedand organic P- Ub ploughed (by harvesting activity) Humisol (see Canadian Soil Survey Wetland organic Hunusol/Humimor/sl Committee, 1978, page 89) mixed with Humimor (see Green et al. ash 1993, page 17) and slash p-Humisol He ploughed (by harvesting activity) Humisol (see Canadian Soil Survey Wetiand organic Committee, 1978, page 89) p-Humhnor inverted lid ploughed (by harvesting activity) Humimor (see Green et al. 1993, Wetland organic humus mound page 17) Wetland organic fibric Humisol lie fibric Humisol (see Canadian Soil Survey Committee, 1978, page 90) sphagnum-Humhnor llf Humimor (see Green et al. 1993, page 17) overtopped by Sphagnum Wedand organic mosses Humimor - blech Ug Humimor (see Green et al. 1993, page 17) overtopped by deer fern Wetland organic litter litter Wetland organic Mesisol 12 Mesisol (see Canadian Soil Survey Committee, 1978, page 88) hemimor 13 hemimor (see Green et al. 1993, page 16) Upland organic hemimor-p 13a ploughed (by harvesting activity) hemimor (see Green et al. 1993, Upland organic page 16) Upland organic litter/needles 14 leaf/twig litter and needles dead grass 14a dead grass Wetland organic blechnum litter 14b deer fern litter Upland organic dead moss litter 14c dead moss litter Upland organic deadtaxus 14d undecomposed yew log Undecomposed Coarse Wood gaultheria litter 14e salal litter Litter standing water 15 standing water Water moving water 15a moving water Water charcoal, burned logs 16 burned logs Burned ustic Fibrimor 16a burned Fibrimor (see Green et al. 1993, page 13 and 19) dominated Burned by fibres predominantly other than Sphagnum exposed rock 17 exposed rock Rock Saprimoder 18 Saprimoder (see Green et al. 1993, page 25) Wetland organic live tree 19 live tree Undecomposed Coarse Wood unknown 20 unrecorded substrate Unknown 100 APPENDIX 2 : VEGETATION Table 26. Relative occurrence of plant species identified in the study plots. "Count" represents the number of plots in which a given species occurred. "Percentage" represents the percentage of space, over all plots, occupied by a given species. ; VEGCODE Plant Identification (Genus, Species) Count Percentage Group blec spic Blechnum spicant 426 23.305 fern cala cana Calamagrostis canadensis 16 2.084 grass-sedge cala nutk Calamagrostis nootkatensis 8 0.553 grass-sedge calamagro Calamagrostis species 28 2.831 grass-sedge care aene Carex aenea 1 0.174 grass-sedge care dewe Carex deweyana 22 0.125 grass-sedge care kell Carex kelloggii 2 0.009 grass-sedge care obnu Carex obnupta 10 0.127 grass-sedge carex Carex species 46 0.585 grass-sedge clad chlo Cladonia chlorophaea 1 0.073 bryophyte clad maci Cladonia macilenta 3 0.081 bryophyte clad port Cladina porentosa 2 0.006 bryophyte clad rang Cladina rangiferina 1 0.003 bryophyte copt aspl Coptis asplenifolia 76 0.495 forb corn cana Cornus canadensis 265 2.249 farb dicr cirr Dicranoweisia cirrata 1 0.006 bryophyte dicr poly Dicranum polysetum 13 0.084 bryophyte dicr scop Dicranum scoparium 60 0.356 bryophyte dicranum Dicranum species 61 0.460 bryophyte dros rotu Drosera rotundifolia 20 0.151 forb empe nigr Empetrum nigrum 1 0.006 evergreen faur cris Fauria crista-galli 1 0.015 forb gaul shal Gaultheria shallon 476 12.984 evergreen good oblo Goodyera oblongifolia 6 0.017 forb herb adun Herbertus aduncus 18 0.414 bryophyte hete proc Heterocladium procurrens 4 0.052 bryophyte hook luce Hookeria lucens 2 0.015 bryophyte hylo sple Hylocomium splendens 314 5.178 bryophyte kalm occi Kalmia occidentalis 1 0.012 forb kind oreg Kindbergia oregana 144 1.702 bryophyte ledu groe Ledum groenlandicum 30 0.252 evergreen linn bore Linnea borealis 215 5.444 forb lyco clav Lycopodium clavatum 12 0.058 evergreen lyco obsc Lycopodium obscurum 3 0.026 evergreen lyco sela Lycopodium selago 1 0.003 evergreen lysi amer Lysichiton americanum 57 0.689 forb maia dila Maianthemum dilatatum 197 1.586 forb menz ferr Menziesia ferruginea 100 0.868 deciduous shrub mnium Mnium species 1 0.003 bryophyte orth lyel Orthotrichum lyellii 5 0.035 bryophyte pell nees Pellia neesiana 7 0.292 bryophyte plag pore Plagiochila porelloides 36 0.836 bryophyte plag undu Plagiothecium undulatum 80 0.495 bryophyte platjung Platydictya jungermanniodes 4 0.015 bryophyte 101 VEGCODE Plant Identification (Genus, Species) Count Percentage Group pleu schr Pleurozium schreberi 10 0.177 bryophyte poa Poa species 11 0.177 grass-sedge poly comm Polytrichum commune 4 0.023 bryophyte polyjuni Polytrichum juniperinum 38 0.535 bryophyte poly pili Polytrichum piliferum 1 0.029 bryophyte pter aqui Pteriium aquilinum 1 0.015 fern rhiz glab Rhizomnium glabrescns 19 0.148 bryophyte rhyt lore Rhytidiadelphus loreus 324 15.919 bryophyte rubupeda Rubus pedatus 26 0.278 forb rubu spec Rubus spectabilis 15 0.208 deciduous shrub scap bola Scapania bolanderi 48 0.596 bryophyte scap undu Scapania undulata 1 0.015 bryophyte sorb site Sorbus sitchensis 7 0.026 deciduous shrub spha capi Sphagnum capillifolium 1 0.058 bryophyte spha fuse Sphagnum fuscum 51 1.398 bryophyte spha papi Sphagnum papillosum 62 1.421 bryophyte sphagnum Sphagnum species 137 12.092 bryophyte stre ampl Streptopus amplexifolius 2 0.006 forb stre rose Streptopus roseus 1 0.006 forb thuj plic Thuja plicata 7 0.654 evergreen tiartrif Tiarella trifoliata 1 0.006 forb trie lati Trientalis latifolia 56 0.272 forb tsug hete Tsuga heterophyla 5 0.313 evergreen unkn 10 Not identified 1 0.029 bryophyte unkn 50 Not identified 1 0.029 bryophyte unkn moss Moss not identified 4 0.304 bryophyte vacc alas Vaccinium alaskaense 4 0.020 deciduous shrub vacc ovat Vaccinium ovatum 23 0.075 evergreen vacc parv Vaccinium parvifolium 35 0.179 deciduous shrub vacc viti Vaccinium vitis-idaea 38 0.179 evergreen vera viri Veratrum virid 14 0.067 forb 102 APPENDIX 3: MODEL ADEQUACY STATISTICS The -2 log likelihood statistic tests the null hypothesis that the joint effects of all of the explanatory variables in the model are zero, but was mainly used to compare different models during the model building stage. Lower values of the statistic indicate a more desirable model. The residual Chi-Square test is a measure of goodness of fit which compares the model-predicted cell proportions with the observed proportions (Stokes et al. 1995; SAS Institute 1997). When p>0.05, adequate fit of the data by the model may be assumed. The Hosmer-Lemeshow goodness of fit statistic is often used when a model contains one or more continuous predictors, or for sparse data. Small p-values for the statistic indicate a lack of fit of the model. The maximum rescaled R2 (Nagelkerke 1991) is used to determine to what extent the model is represented by the data, but in this case was used to compare various models to one another, since the value itself was small in each regression (possibly a result of not including in the study other potentially important explanatory variables). Rank correlation measures were used as a general indicator of how well the overall observed values in the data compared with the expected values (since no p-values were attached to them). For example, a concordance of 80% indicates that 80% of the observed data were as expected, based on the model fit. Sensitivity and specificity were used to determine classification accuracy. Cutpoints for evaluating sensitivity and specificity were based on a random selection of 1/3 of the data in the datasets used for each of the six regressions. Frequency counts were taken on this random set of data and the proportion of events and non-events were determined. The proportion of events was then used as the cutpoint for determining appropriate sensitivity and specification. Wald confidence intervals were examined during the model building stage to assist in removing unimportant variables. 103 Variables were removed which contained the value "1" within the interval. A value of 1 indicates that the likelihood for observing any given response relative to all possible responses is the same for a given explanatory variable. 104 

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.831.1-0099496/manifest

Comment

Related Items