HEIGHT TC) DlAJMEnriSR jBLA/ITCOjlS Ai(:%]HVII»E:lTriTW:M\f]ON][ME:K114 YOUNG PLANTATIONS OF LODGEPOLE PINE IN TinSt^J^DEIUTOKHfFtMWESTIIKnitBCTCIFIHUlTSHCCHJUNCBLl by Norman Jacob B.Sc., University of Northern British Columbia, 1999 B.A.Sc., University of Waterloo, 1979 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in NATURAL RESOURCES AND ENVIRONMENTAL STUDIES © Norman Jacob, 2003 TIME IJNIT/ERJSITTYtZNF NCJRTTIEBJN IJItlllKSII C:C)LUn\4BI/l April, 2003 All rights reserved. This work may not be reproduced in whole or in part, by photocopy or other means, without the permission of the author. National Library of Canada Bibliothèquue nationale du Canada Acquisitions and Bibliographic Services Acquisitions et services bibliographiques 395 Wellington Street Ottawa ON K1A0N4 Canada 395, rue Wellington Ottawa ON K1A0N4 Canada YourfHe Voire réiérerKO OurSte Notre référence The author has granted a non­ exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic formats. L’auteur a accordé une licence non exclusive permettant à la Bîbhothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfîche/fihn, de reproduction sur papier ou sur format électronique. The author retains ownership of the copyright in this thesis. Neitiier the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author’s permission. L’auteur conserve la propriété du droit d’auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation. 0 612 80655-3 - Canada - APPROVAL Name: Norman Jacob Degree: Master of Science Thesis Title: HEIGHT TO DIAMETER RATIO AS A COMPETITION INDEX IN YOUNG PLANTATIONS OF LODGEPOLE PINE IN THE VANDERHOOF FOREST DISTRICT OF BRITISH COLUMBIA Examining Committee: ChainjDr. Robert Tait Dean of Graduate Studies UNBC Supervisor: Dr. Chris Opio Assistant Professor, Forestry Program UNBC Committee Member: Dr. Arthur Fredeen Associate Professor, Forestry Program UNBC I Committee Member: Dr. Jian Wang Assistant Professor, Faculty of Forestry & the Forest Environment Lakehead University External Examiner: Dr. Phil Comeau Associate Professor, Department of Renewable Resources University of Alberta Date Approved: ABSTRACT Height to diameter ratio (HDR) has been proposed as an alternative to conventional procedures for assessing competition between crop trees and other vegetation. However, in order to use HDR as a competition index, forest managers need to understand how HDRs respond to the removal of above ground competing vegetation (i.e., brushing treatments), how HDRs vary from the time of planting to when competition becomes a problem, and how HDRs vary within a growing season. They also need to be able to measure HDRs against an independent criterion such as stem volume. Trends in HDRs of lodgepole pine {Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) were investigated: (i) during a three year period (1998-2001) following the initiation of treatments; (ii) between the time of planting (1990-1995) and destructive sampling (2000); and (Hi) within a growing season on study sites in the sub-boreal spruce (SBS) biogeoclimatic zone in the central interior of British Columbia. Item (z) included investigation of trends in stem volume during the same period. Item (ü) involved destructive sampling and stem analyses. Competition from trembling aspen (Populus tremuloides Michaux), and other deciduous tree species was a concern on these sites. The study used a completely randomized, one-factor experimental design, with replication of measurements over time. The removal of competing vegetation with a brush saw was the factor. The design consisted of four levels of brushing, replicated three times on each site. It was found that significant (p < 0.05) reductions in HDRs were possible with brushing, and a brushing radius between 1.0 m and 1.25 m was optimal. Reference HDRs (i.e., ranges of HDRs) 40-49, 40-51, 45-54, and 38-47 were determined for the four sites. In describing the variation in HDRs prior to treatments, no consistent pattern in HDRs was found. It was determined that stem volume responded positively (p < 0.05) where 1.0 m and 1.25 m treatments were undertaken. Ranges of stem volume were defined within which reference HDRs were recommended. Variations in HDRs from early May to October were determined on two sites. It was ascertained that HDRs stabilize after mid-August, and that HDR measurements are most reliable when taken at this time. It was concluded that the use of HDRs can complement conventional vegetation competition assessment procedures. ii TABLE OF CONTENTS Page Abstract ii Table of contents iii List of tables v List of figures vi Acknowledgements viii Introduction 1 1. Literature review 1.1 Growth of lodgepole pine 1.2 Resource allocation under competitive stress 1.3 Competition indices and competitive thresholds 1.4 Height to diameter ratio as a competition index Literature cited 6 6 7 9 13 16 2. Inter-seasonal variations in height to diameter ratios in lodgepole pine following variable removal of competing vegetation Abstract 2.1 Introduction 2.2 Materials and methods 2.2.1 Study sites 2.2.2 Experimental design 2.2.3 Measurements 2.2.4 Analysis 2.3 Results 2.3.1 CanFor-Bednestii-Little Bobtail Lake site 2.3.2 Fraser Lake-101 km site 2.3.3 Fraser Lake-137 km site 2.3.4 Fraser Lake-1116 km site 2.4 Discussion 2.5 Conclusions and recommendations Literature cited 20 3. Retrospective analysis of inter-seasonal variations in height to diameter ratios in lodgepole pine prior to removal of competing vegetation Abstract 3.1 Introduction 3.2 Materials and methods 3.2.1 Study sites 3.2.2 Experimental design 3.2.3 Measurements 3.2.4 Analysis m 20 21 23 23 25 26 27 32 33 35 36 38 41 49 51 54 54 55 56 56 57 57 60 3.3 Results 3.3.1 CanFor-Bednestii-Little Bobtail Lake site 3.3.2 Fraser Lake-101 km site 3.3.3 Fraser Lake-137 km site 3.3.4 Fraser Lake-1116 km site 3.4 Discussion 3.5 Conclusions and recommendations Literature cited 4. 5. 6. Variations in stem volume of lodgepole pine following variable removal of competing vegetation Abstract 4.1 Introduction 4.2 Materials and methods 4.2.1 Study sites 4.2.2 Experimental design 4.2.3 Measurements ' 4.2.4 Analysis 4.3 Results 4.3.1 CanFor-Bednestii-Little Bobtail Lake site 4.3.2 Fraser Lake-101 km site 4.3.3 Fraser Lake-137 km site 4.3.4 Fraser Lake-1116 km site 4.4 Discussion 4.5 Conclusions and recommendations Literature cited Intra-seasonal variations in height to diameter ratios in lodgepole pine following variable removal of competing vegetation Abstract 5.1 Introduction 5.2 Materials and methods 5.2.1 Study sites 5.2.2 Experimental design 5.2.3 Measurements 5.2.4 Analysis 5.3 Results 5.3.1 CanFor-Bednestii-Little Bobtail Lake site 5.3.2 Fraser Lake-101 km site 5.4 Discussion 5.5 Conclusions and recommendations Literature cited Synopsis and management implications of height to diameter ratios in lodgepole pine study Literature cited Appendix A. Randomized plot layouts for all study sites IV 63 65 67 68 70 73 75 77 79 79 80 81 81 82 82 86 93 95 97 98 99 102 108 110 112 112 113 114 114 115 115 117 120 121 123 125 128 130 132 139 140 LIST OF TABLES Page Table 2.1. MANOVAs with repeated measures showing the factors affecting mean HDRs and percent changes in HDR between 1998 and 2001 for all study sites 33 Table 2.2. Mean HDRs and standard errors of the mean (SEM) from 1998 to 2001 for all study sites 40 Table 2.3. Recommended reference HDRs with vegetation complexes, EEC classifications, and ranges of percent cover of competing vegetation and mean diameter within which reference HDRs apply for all study sites 48 Table 3.1. MANOVAs with repeated measures showing the factors affecting mean HDR ibS and percent changes in HDR ib between 1991 and 1999 for all study sites 65 Table 3.2a. Mean HDRibS and standard errors of mean (SEM) from 1994 to 1999 for CanFor-Bednestii-Little Bobtail Lake and Fraser Lake-101 km sites 68 Table 3.2b. Mean HDR ibS and standard errors of mean (SEM) from 1991 to 1999 for Fraser Lake-137 km and -1116 km sites 72 Table 4.1. Primary regression models for stem volume for all study sites 93 Table 4.2. MANOVAs with repeated measures showing the factors affecting mean stem volumes and percent changes in stem volume between 1998 and 2001 for all study sites 95 Table 4.3. Mean stem volumes and standard errors of mean (SEM) from 1998 to 2001 for all study sites 101 Table 4.4. Recommended reference HDRs with vegetation complexes, BEG classifications, and ranges of percent cover of competing vegetation and mean stem volume within which reference HDRs apply for all study sites 107 Table 5.1. MANOVAs with repeated measures showing the factors affecting mean HDRs and percent changes in HDR from May to October 1999 for CanForBednestii-Little Bobtail Lake site, and September 1999 and July to September 2000 for Fraser Lake-101 km site 121 Table 5.2. Mean HDRs and standard errors of mean (SEM) from May to October 1999 for CanFor-Bednestii-Little Bobtail Lake site, and September 1999 and July to September 2000 for Fraser Lake-101 km site 125 LIST OF FIGURES Page Figure 2.1. Location of the study sites 24 Figure 2.2. Percent changes in HDR between 1998 and 2000 for CanFor-BednestiiLittle Bobtail Lake site 34 Figure 2.3. Percent changes in HDR between 1998 and 2000 for Fraser Lake101 km site 36 Figure 2.4. Percent changes in HDR between 1998 and 2000 for Fraser Lake137 km site 37 Figure 2.5. Percent changes in HDR between 1999 and 2001 for Fraser Lake1116 km site 39 Figure 3.1. Schematic diagram of crop tree as measured for stem analysis at CanForBednestii-Little Bobtail Lake site 59 Figure 3.2. Percent changes in HDR,g between 1994 and 1999 for CanFor-BednestiiLittle Bobtail Lake site 66 Figure 3.3. Percent changes in HDR,g between 1995 and 1999 for Fraser Lake101 km site 67 Figure 3.4. Percent changes in HDR,g between 1991 and 1999 for Fraser Lake137 km site 69 Figure 3.5. Percent changes in HDR,g between 1991 and 1999 for Fraser Lake1116 km site 71 Figure 4.1a. Schematic diagram of crop tree as measured for volume calculations without additional measurement for 1999 for CanFor-Bednestii-Little Bobtail Lake and Fraser Lake-101 km sites 84 Figure 4.1b. Schematic diagram of crop tree as measured for volume calculations with additional measurement for 1999 for CariFor-Bednestii-Little Bobtail Lake and Fraser Lake-101 km sites 85 Figure 4.2. Percent changes in stem volume between 1998 and 2000 for CanForBednestii-Little Bobtail Lake site 96 Figure 4.3. Percent changes in stem volume between 1998 and 2000 for Fraser Lake101 km site 97 Figure 4.4. Percent changes in stem volume between 1998 and 2000 for Fraser Lake137 km site 99 Figure 4.5. Percent changes in stem volume between 1999 and 2001 for Fraser Lake1116 km site 100 VI Figure 5.1. Percent changes in Ht)R between May and October 1999 for CanForBednestii-Little Bobtail Lake site 122 Figure 5.2. Percent changes in HDR for September 1999 and July to September 2000 for Fraser Lake-101 km site 123 Vll ACKNOWLEDGEMENTS I thank my supervisor, Dr. Chris Opio, for our many discussions with regard to all aspects of scientific research; for the considerable time he spent helping me to produce coherent and communicable chapters and manuscripts; his commitment to seeing the project through to completion; and the friendship he offered. I am grateful to the committee members, Drs. Art Fredeen and Jian Wang, for suggesting that I attempt to describe the development of crop and non-crop vegetation on the study sites prior to undertaking brushing treatments. Their suggestions prompted me to undertake retrospective and stem volume analyses. I thank Dave Coopersmith who first introduced me to the usefulness of height to diameter ratios. I further acknowledge the assistance he provided with the initial experimental design and logistics of site installations, and for introducing me to the project collaborators. I am grateful for the many useful discussions I had with the industrial collaborators on the project, Brian Walker of West Fraser Mills Ltd., and Vince Day and John Brockley of Canadian Forest Products Ltd. I also acknowledge the assistance they provided in selecting the study sites. I thank West Fraser Mills Ltd., Canadian Forest Products Ltd., the Science Council of British Columbia, and the University of Northern British Columbia for the financial support they provided. I thank Sonya Saloma, Rob Bourcier, Colin Lacey, and others for assisting in many long hours of data collection; Richard Reich and Dr. Kathy Lewis for advice on stem analyses; Irisol Alaniz and Avens Dawson for assisting with the stem analyses; Dr. Oscar Garcia for assistance he provided with the assessment of stem volume; and Dieter Ayers, Dr. Peter MacMillan, and Symbiotech Research Inc. for advice on statistical analyses. Lastly, but not least, I deeply appreciate the support and encouragement my wife Carolyn provided to me while I worked on my thesis. vm INTRODUCTION Forest managers are concerned about vegetation control in young conifer plantations. Competition from shrubs, herbaceous plants and deciduous trees can limit light, nutrient and water availabilities, thereby restricting crop tree growth. There are, however, few quantitative methods that can assist forest managers in deciding what level of removal of competing vegetation is silviculturally necessary, and economically justifiable (Richardson et al. 1999). Several competition indices (e.g., DeLong 1991; Comeau et al. 1993; Wagner and Radosevich 1991b, 1998; Morris and McDonald 1991) have been developed for use in conifer plantations. However, many of these indices are static (i.e., they are based on a single set of measurements taken at one point in time), and have overlooked seasonality effects and underlying competition processes (Burton 1993, Maclsaac and Navratil 1996). In this respect, an operationally important measure of competition is the “free growing” assessment procedure in British Columbia (BC). This procedure requires, depending on the biogeoclimatic zone, a conifer to brush ratio of 125% or 150% within a i m radius cylinder surrounding the crop tree (BC Ministry of Forests 1996, 2000). The recommendations by Burton (1993) are that competition indices should capture the changes in crop trees and growth of competing vegetation by any of the following methods: (i) repeated calculations of competition index over time; (ii) comparison and analysis of various attributes of plant growth; and (iii) simulation of the simultaneous growth of both crop trees and non­ crop vegetation. To be useful to forest managers, a competition index should: (i) be simple to measure and easily repeatable; (ii) involve analysis of both the growth and development of competing vegetation and crop tree over time; and (iii) involve assessment of seasonality effects on the growth and development of crop trees over time (Burton 1993). Furthermore, a competition index should not involve too much field work to measure, and should be a tool that assists forest managers in making ecologically and economically appropriate forest stand management decisions (Opio et al. 2003). At the same time as our understanding of what makes a good competition index has developed (e.g.. Burton 1993, Richardson et al. 1999), the “free growing” assessment procedure in BC has come under increased scrutiny (Davis 1998). An alternative or supplement to the free growing assessment procedure has been sought, and height to diameter ratio (HDR) of the crop tree is thought to address many of these concerns (Mustard and Harper 1998, Coopersmith and Hall 1999). HDR is an individual tree-based index, calculated by dividing the height of crop tree either by the diameter at the root collar or diameter at breast height (DBH). It is commonly thought that HDR is primarily influenced by the availability of light (Chen 1997). However, height and diameter growth are influenced by availability of nutrients and moisture, planting position, elevation, growing season, litter depth, slash, slope, aspect, tree species, age, seasonal climate, site preparation, stock type and provenance (Zimmerman and Brown 1971, MacDonald et al. 1990, Kozlowski and Pallardy 1997, Mustard and Harper 1998). Height and diameter growth are also influenced by wind and bending of the stem (Henry and Aarssen 1999, Ruel et al. 2000). Typical silvicultural measures to control competition in conifer plantations include release and thinning treatments. In this connection, HDR has been used in Germany as a competition index for undertaking thinning in second-growth stands of Douglas-fir (Pseudotsuga menziessi (Mirb.) Franco var. menziessi) m à western hemlock (Tsuga heterophylla (Raf.) Sarg) since 1970 (Smith 1986). HDR has been used both as a predictor of when to commercially thin spruce plantations (Abetz 1976) and as an index to predict post-thinning plantation damage (Merkel 1975). Ruel et al. (2000) and Ruel (1995) also describe these uses of HDR. In evaluating changes in HDR, it is important to note that height growth of many conifers (e.g., lodgepole pine, white spruce) is determinate and has a higher priority of carbon allocation, whereas cambial growth is indeterminant and exhibits a lower priority function. The latter is dependent on the current year’s resources, whereas the former is relatively independent of these resources (Waring and Pitman 1985; Wagner and Radosevich 1991a, 1991b; Harrington and Tappeiner 1991; Williams et al. 1999). This suggests that HDR would be greatest in shaded areas and decrease as the availability of light increased. HDR s response to the availability of light demands particular attention when we are dealing with shade-intolerant species such as lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) (Lotan and Critchfield 1990). Lodgepole pine occurring in areas with the sub-boreal spruce (SBS) dry warm (dw) and dry cool (dk) biogeoclimatic classification comprise a major portion of the forested land base in BC (De Long et al. 1993). It is a species of commercial interest, and subject to adverse competition from trembling aspen (Populus tremuloides Michaux), among other species in the early stages of its growth. Thus, young SBS dw and dk lodgepole pine plantations were chosen for the study sites addressed in this thesis. Our understanding of HDR as a potentially useful competition index that could assist forest managers in making decisions about the control of competing vegetation in conifer plantations is limited. Deficiencies in our knowledge prevent the wider application of HDRs. First, forest managers do not have locally calibrated reference HDRs (i.e., HDR thresholds) that are applicable to the various areas for which they may wish to use this tool. Second, reference HDRs that are available have generally been defined in terms of HDR itself, or HDRs that prevailed where levels of non-crop vegetation were deemed acceptable. There is a shortcoming in using HDR independently from other criteria. Both stem diameter and volume potentially could fill this void. Third, it is not known how HDRs respond to different levels of light that are produced by the variable removal (i.e., brushing with a brush saw) of above ground competing vegetation, and what would be an optimum brushing radius. Fourth, information is lacking on how HDRs vary between the time of planting of crop trees and the time of brushing of experimental plots. This knowledge would help forest managers to identify the time period when brushing should be undertaken in plantations. Fifth, it is known that HDRs vary within a growing season, however, this pattern has not been defined in precise terms to be able to avoid incurring significant errors in measurements. There is a need to define the period at the beginning and end of the growing season when changes in HDRs are sufficiently negligible to be able to take reliable measurements. In response to the second deficiency in our knowledge, HDR thresholds for lodgepole pine are defined in the thesis with reference to percent change in HDR (i.e., %AHDR defined with respect to the time of site installation). A suitable reference HDR is determined as being the point at which HDRs are stabilizing (i.e., %AHDRs from one year to the next are negligible), and a base level HDR is being achieved. However, this method for recommending reference HDRs needs to incorporate an independent criterion or other factor(s) to be measured in conjunction with HDR. Possible criteria to be used in conjunction with HDR include stem volume, wood quality, and other attributes of the tree or site. Stem volume is of prime consideration because it is readily quantifiable, and an accepted measure of productivity (i.e., growth) in the early stages of stand development. For example, Wagner et al. (1999) used a stem volume index (i.e., stem diameter^ x height) to determine the critical period of interspecific competition. However, HDRs are poorly correlated with stem volume, thus it is unreasonable to assume that a functional relationship can be established between reference HDRs and stem volume. A reasonable alternative is to use stem volume as a supportive tool, in conjunction with HDR thresholds. This requires that an HDR threshold be determined for a specific range of stem volume. In order to use stem volume in conjunction with reference HDRs, models of stem volume need to be developed; for this, destructive sampling and stem analyses need to been undertaken. In the absence of this criterion, however, stem diameter is a consideration. Stem diameter is known to be highly correlated with stem volume, and is the key determinant of stem volume in most stem volume equations (Husch et al. 1993). In addressing these concerns, reference HDRs are first determined with respect to ranges of stem diameter. Where stem volume equations are not available, it seems that diameter can be used to delimit (or constrain) the application of reference HDRs. Second, HDR thresholds are re-defined with respect to ranges of mean stem volume for which the reference HDRs are recommended. The task of determining mean stem volumes to be used in conjunction with reference HDRs is addressed in the thesis. The purpose of this thesis is to provide information on HDR that may assist in determining the feasibility of HDR as a competition index. The main objectives of the HDR studies (Chapters 25) reported in the thesis were to (1) determine how HDRs respond to different levels of removal (i.e., brushing) of above ground competing vegetation applied to crop trees, over time; (2) recommend reference HDRs that apply to plantations similar to the study sites (i.e., similar ranges of diameter, BEG classifications, and percent cover of competing vegetation); (3) determine which brushing radius (i.e., 0.75 m, 1.0 m, or 1.25 m radius) is best to brush plantations similar to the study sites (items 1-3 are addressed in Chapter 2); (4) describe trends in HDRs between the time of planting of crop trees and the time of brushing of experimental plots; (5) examine the relationships between pre-treatment (before brushing) and post-treatment (after brushing) HDRs; (6) identify the time period when brushing should be undertaken in plantations similar to the study sites items 4-6 are addressed in Chapter 3); (7) develop regression models of stem volume, and apply these to the field-based measurements; (8) use the stem volumes obtained to determine ranges of mean stem volumes within which reference HDRs are meant to be applied; (9) determine how stem volume increment responds to various brushing treatments in the period 1998-2000 (or 1999-2001) (items 7-9 are addressed in Chapter 4); (10) determine the pattern of variations in HDRs through the 1999 growing season at one study site, and the 2000 growing season at another study site; and (11) determine when in the growing season changes in HDR are negligible for no brushing and a range of brushing thresholds, and HDR measurements may be taken without incurring significant error (items 10-11 are addressed in Chapter 5). Chapter 6 examines the management implications of the research that was undertaken. CHAPTER 1 LITERATURE REVIEW 1.1 Growth of lodgepole pine Lodgepole pine comprises a major portion of the forested land base in British Columbia (BC). A variety of lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) found in the interior of BC tends to grow in even-aged (pure or mixed) stands (Lotan and Critchfield 1990). It is a fast growing pioneer species (Eis et al. 1982). Lodgepole pine is very intolerant of shade and competition (Armit 1966). This is indicated by the fact that lodgepole pine reaches its saturation of photosynthesis at > 543 W/m^, an extremely high level compared to other conifer species (Bassman and Koch 1996). The strategy of lodgepole pine is to germinate quickly following disturbance and provide rapid early height growth, while being able to adapt to a variety of conditions (Lotan and Critchfield 1990). Lodgepole pine is resistant to frost injury, and can grow on very nutrient poor sites and sites with extreme water conditions (Klinka et al. 1989). This tolerance to adverse environmental factors can give lodgepole pine a competitive advantage over other pioneer species (Lotan and Critchfield 1990). However, as a shade intolerant species, lodgepole pine tends to outgrow its competitors and is susceptible to etiolation. Where this condition occurs, most of the net biomass production is invested in stem height growth, and shoot/root and height/diameter ratios increase correspondingly (Kimmins 1997). It seems that etiolation occurs where light is a limiting resource. The concept of a “single limiting resource” describes much interspecific competition (Tilman 1996). The shade intolerance of lodgepole pine makes light a limiting resource in many competition situations (Mustard and Harper 1998). Trembling aspen (Populus tremuloides Michaux) is an early serai species with many of the same strategies as lodgepole pine (Perala 1990). It also attempts to outgrow its competitors via rapid early height growth. On nutrient poor sites, or overly wet or dry sites, lodgepole pine is not impeded by aspen (Lotan and Critchfield 1990). However, where lodgepole pine is planted on relatively “good” sites, competition from aspen may be deleterious. Lodgepole pine growing in mesic (i.e., medium nutrient and moisture) areas of the sub-boreal spruce (SBS) dry warm (dw) and dry cool (dk) biogeoclimatic (BEG) classification of BC (DeLong et al. 1993) have been a concern to forest managers. For example, lodgepole pine planted in areas with SBS dw3 (01) site series have often been adversely affected by competition from aspen (B. Walker, pers. comm. 1998). While the removal of above ground competing vegetation (i.e., brushing) is frequently necessary to maintain crop tree vigour on mesic sites, it is not known what level of brushing is biologically necessary, and economically viable (Weingartner and Basham 1979, Richarson et al. 1999). It is felt that HDR is advantageous for assessing deleterious competition in such areas (Mustard and Harper 1998, Coopersmith and Hall 1999). 1.2 Resource allocation under competitive stress Source-sink theory explains the relative variability in height and diameter growth due to: (i) stresses felt by the tree, and (ii) the time of the growing season (Zimmermann and Brown 1971, Waring and Schlesinger 1985, Oliver and Larson 1996). Waring and Schlesinger (1985) postulate a hierarchy for normal carbon allocation in a tree. In order from highest to lowest priority, these are (I) buds and new foliage, (ii) new roots, (iii) canopy storage, (iv) diameter growth, and (v) protective chemicals. Item (i) includes apical extension of the branches (i.e., lateral growth) and leader (i.e., vertical growth). Carbon resources are allocated to those parts of the tree that are most likely to increase the tree’s chances of survival (Waring and Schlesinger 1985). According to this theory, the cambium has a lower priority in the allocation of resources; thus it is only after apical extension has been satisfied that radial growth will proceed. In terms of source-sink theory, the cambium is said to be a weaker sink than the apical meristem. The relatively higher priority apical meristemic growth has vis a vis cambial growth may be reflected in an inflation in HDR. In trees that exhibit preformed (or fixed) growth, such as lodgepole pine, apical meristemic growth is largely determined (i.e., the number of stem units are determined) by environmental conditions felt by the tree in previous growing seasons. Apical extension is a product of factors prevalent over a relatively long time period, and is relatively unaffected by competition (Lanner 1985, Navratil and Maclsaac 1993). Extension of the units (i.e., produced during the previous growing season) is under the control of the current year’s growing conditions. Cambial growth, by comparison, is entirely a product of environmental influences felt by the tree in the current growing season. Radial expansion is strongly affected by competition. Change in HDR between years integrates the tree’s response to environmental factors prevalent in previous years with the tree’s response to environmental factors prevalent in the current year. At the same time, HDR is the net result of the history of a tree’s growth over its life. The pigment phytochrome plays a crucial role in the allocation of resources (i.e., carbohydrates) between height and diameter growth. Phytochrome (i.e., P, which absorbs red light and converts to P&, and Pft which absorbs far-red light and converts to P J regulates development of changes in the red:far-red wavelength ratio of light (Ross et al. 1986). The red:far-red ratio is lower when the tree is shaded than when it is open-grown (Ross et al. 1986). When exposed to light, both forms of phytochrome absorb photons of the respective wavelengths until equilibrium is established (Ross et al. 1986, Salisbury and Ross 1992). The ratio of PfriP, is an important determinant of the rate of growth of the shoot of the plant. When the red:far-red ratio is lower (i.e., when the tree is shaded), the ratio of Pfr:Pr is lower, and stem elongation is stimulated (Salisbury and Ross 1992). Phytochrome is a key regulator of the relative strength of the apical meristem as a sink. In response to restricted light, shade intolerant species will reduce lateral branch growth and increase height growth, in an attempt to outgrow the reduced light conditions (Williams et al. 1999). Lodgepole pine will have fewer branches per whorl in low light situations, and more branches per whorl in high light situations (Williams et al. 1999). When diameter growth is limited, so is the formation of protective chemicals (Waring and Pitman 1985). Since allocation of these chemicals is posited to have a lower priority than diameter growth, trees growing relatively less in diameter than height may have their vigour adversely affected. Thus, inflated HDRs may indicate susceptibility to disease (Waring 1987). Apical growth proceeds from the initiation of extension, approximately in April or May, to the cessation of extension, approximately in June or July. Cambial growth proceeds concurrently with apical growth early in the growing season but continues much later in the growing season. Depending on the insolation levels and temperatures prevalent, cambial growth may continue into the fall. Cambial growth will generally proceed from the bud downward to the base of the tree (Zimmermann and Brown 1971, Waring and Schlesinger 1985, Oliver and Larson 1996). The pattern is one of growth in height (i.e., the numerator in HDR) proceeding faster than growth in diameter (i.e., the denominator in HDR) in the earlier part of the growing season (i.e., bringing about an increase in HDR), and one of growth in height falling behind growth in diameter in the later part of the growing season (i.e., bringing about a decrease in HDR). Opio et al. (2003) have determined that the periods of relative stability for lodgepole pine vis a vis height and diameter growth occur at the beginning and end of the growing season. Growth in diameter is cumulative (Eis et al. 1982). Therefore, trees grown under lightrestricted conditions when released from competition will not make up the stem volume “lost” in comparison with trees initially grown under light-abundant conditions. Determining when (i.e., years after planting) plantations need to be brushed has a major impact on yield. Wagner et al. (1999) applied a “critical period concept” to several northern conifer species to gauge when non-crop vegetation must be controlled to prevent notable losses in crop yield. 1.3 Competition indices and competitive thresholds There is an interest in developing quantitative methods that can assist forest managers in deciding what level of removal of competing vegetation is silviculturally necessary, and economically justifiable (Richardson et al. 1999). Both competition indices and competition thresholds address this concern. These decision tools are different, and it is important to distinguish between them. The following definitions have been used. A competition index measures competitive pressure on an individual tree (MacDonald et al. 1990). A competition threshold is the density of competition that results in a loss in crop species yield, or at which measures must be taken to avoid a competition problem in the future (Wagner 1994, cited in Mustard and Harper 1998). Competition indices typically fall into one of two categories: those calculated using stand level measures and those based on individual tree measurements (Burton 1993, Maclsaac and Navratil 1996). Competition indices based on individual tree measurements incorporate a variety of information on neighbouring trees including: available growing space around the crop tree crown, neighbour tree density or basal area, neighbour tree diameter, distance to neighbour, and measured or inferred shade produced by the neighbour (Maclsaac and Navratil 1996). These measurements are intended to give an indication of the abundance, proximity, and stature of neighbouring trees (Burton 1993). Several competition indices (e.g., DeLong 1991; Comeau et al. 1993; Wagner and Radosevich 1991b, 1998; Morris and MacDonald 1991) have been developed for use in conifer plantations. Examples of competition indices are: the basal diameter ratio competition index calculated by dividing the basal diameter of the tallest aspen within 1.8 m of the crop tree by the basal diameter of the crop tree; and the individual tree index calculated by summing the products of percent cover and height for all non-crop species within a 1.26 m radius plot and dividing by the height of the crop tree (Comeau et al. 1993). Richardson et al. (1999) used a modeling technique to develop, under New Zealand conditions, a range of competition indices, with some sensitivity to both the growth and development of competing vegetation and radiata pine (Pinus radiata D. Don) over time. Their “best” indices included measures of height of competing vegetation relative to tree height, proximity of the competing vegetation to the tree, and abundance of competing vegetation. 10 The “free growing” criterion in BC (BC Ministry of Forests 1996, 2000) is a competition threshold. It is defined as “a stand of healthy trees of commercially valuable species, the growth of which is not impeded by plants, shrubs, or other trees” (BC Ministry of Forests 1996, p. 9). Current standards for assessing the impact of competing vegetation on crop trees require, depending on the biogeoclimatic zone, a conifer to brush height ratio of 125% or 150% within a i m radius cylinder surrounding the tree (BC Ministry of Forests 1996, 2000). Beginning in 1998 or earlier, there has been increasing criticism of the free growing criterion. It is felt that the procedure places too much emphasis on light availability in assessing the future viability of crop trees (Davis 1998). The crop tree to brush ratio cannot account for the complicated interplay of other factors that limit crop tree growth (Davis 1998). Static and dynamic approaches to vegetation competition assessment may be undertaken at either the level of the individual tree or at the level of the whole stand. Historically, most approaches to competition assessment have been static approaches. Most of the indices that Burton (1993) reviewed may be criticized for being static approaches to competition assessment. They provide a “snapshot” of how a crop tree is growing relative to non-crop vegetation. HDR is potentially a dynamic approach to competition assessment, but it is important that HDR be used in a manner that allows it to capture relative changes in crop tree and non-crop vegetation growth. This requires that repeated measurements of the same trees need to be taken over a number of years. Individual tree based indices and a whole stand approach are used in the assessment of brush problems in conifer stands (Wagner et al. 1991b). This distinction is important because “understanding the influence of interspecific competition on stand dynamics ... requires analysis at the individual plant level” (Firbank and Watkinson 1987, cited by Wagner and Radosevich 1991b). Individual tree based indices may measure competition directly as with HDR (Opio et al. 2000) and the stem volume index (Wagner et al. 1999), where competition is measured entirely in terms of characteristics of the crop tree itself; or indirectly, where competition is measured in terms of the relationship between the crop tree and neighbouring vegetation. Burton’s (1993) review of static II competition indices addresses primarily indirect measures of competition which depend upon a fixed neighbourhood search radius. Thus, Burton’s critique of such indices needs to be interpreted with respect to HDR, a competition index that is not addressed in his review. There are limitations inherent to static indices of plant competition. In his survey of these limitations. Burton (1993) lists the following attributes of competition indices. First, most indices of competition are static and do not reflect seasonal changes or differences in growth trajectories. Second, competition indices may not capture the stresses that are important. Third, the application of many competition indices involves the use of a fixed neighbourhood search radius which has its limitations. Fourth, some measure of the abundance and size of competitors is made (i.e., judgements are made about what is important). Fifth, many measures of plant competition ignore competitors beneath some fixed or relative stature (i.e., implies very strong asymmetries in competition). Burton (1993) offers a critique of the above common attributes of competition indices in terms of the following points. First, system dynamics cannot be represented by one-time measurements. Second, competition can be intense without being important. Third, the competitive arena is not constant in size. Fourth, below ground constraints are considered only in so far as they are represented by above ground attributes. Fifth, there can be pronounced species and growth-form differences in competitive effects and responses. Sixth, there are often unwarranted assumptions of additivity in neighbourhood effects and crop-tree responses. Burton (1993) makes the following recommendations for replacing or refining the use of static competition indices. First, use crop trees as site-specific “phytometers” (i.e., a control crop tree that is grown under optimal site conditions). The “control” tree is then compared against “standard” trees grown at various levels of competition. Second, project tree growth in concert with vegetation dynamics. In this respect. Burton proposes use of either (i) repeated calculations of a competition index through time, (ii) plant growth analysis attributes such as “relative production rate” (Brand et al. 1987, cited in Burton 1993), or (iii) simulation of the simultaneous growth of both crop and non­ 12 crop plants. Third, locally calibrate and verify competition indices with respect to the site, crop tree, and non-crop vegetation. 1.4 Height to diameter ratio as a competition index HDR is being proposed as an additional tool within the set of current procedures for assessing the vigour and free growing status of young plantations in BC (Mustard and Harper 1998). At the time the HDR research project began, HDR was beginning to be used in the Prince George Forest Region “when determining if competition within the one metre radius cylinder of a tree is acceptable during a free growing survey” (BC Ministry of Forests 1998). Maximum HDRs which were to be used in the application of this procedure were 50 for lodgepole pine and 60 for white spruce. HDR has been used as a vegetation management tool in the Fraser Lake area since 1998 or earlier (B. Walker, pers. comm. 1998). HDR has been used in Germany as a thinning tool in second-growth stands of Douglas-fir {Pseudotsuga menziessi (Mirb.) Franco var. menziessi) and western hemlock {Tsuga heterophylla (Raf.) Sarg) since at least 1970 (Smith 1986). It has also been used both as a predictor of when to commercially thin spruce plantations (Abetz 1976), and as a tool to predict post-thinning plantation damage (Merkel 1975). In Europe, HDRs have been used to indicate the need for thinning, resistance to windthrow, and estimate stem form (Coopersmith and Hall 1999). Ruel et al. (2000) and Ruel (1995) also describe these uses of HDR. HDR is calculated by dividing the height of the crop tree by the diameter at root collar (BC Ministry of Forests 1998, Opio et al. 2000), or other height of the tree such as breast height (DBH). DBH or 130 cm above the root collar is a common point of measurement on trees several metres in height (Husch et al. 1993). HDR differs from most other competition indices in that HDR is a direct measure of competition whereas most others are indirect measures. HDR is believed to integrate many factors that have influenced the growth of a tree over the course of its life, and determine its vigour. A 13 healthy tree should have a lower HDR than a tree limited by competitive stress. Williams et al. (1999) found that HDRs decreased with increasing light. Mustard and Harper (1998) found that generally, a higher HDR indicated lower vigour. These results seem to be consistent with the hierarchy of normal carbon allocation proposed by Waring and Schlesinger (1985). Following the belief that HDR is primarily an indicator of a tree’s response to limited availability of light, Froese (2000) makes the following relative statements. Restricted light resources can either: (i) reduce height growth and increase lateral branch growth, in order to maximize light intercepted (shade tolerant species); or (ii) reduce lateral branch growth and increase height growth, in an attempt to “outgrow” the vegetation that produces the reduced light conditions (shade intolerant species) (Williams et al. 1999). In the context of Froese’s (2000) statements, we would expect species such as white spruce (Picea glauca (Moench) Voss. sp. glauca) and lodgepole pine to respond differently to various limitations on light created by different brushing radii produced around crop trees on a study site. According to Froese’s (2000) hypotheses, white spruce should respond to limitations on light by reducing height growth relative to lateral branch growth, while lodgepole pine should respond to limitations on light by increasing height growth relative to lateral branch growth. Under conditions of restricted light resources, white spruce will tend to increase its HDR by a lesser amount than will lodgepole pine. White spruce will tend to increase in height more slowly than will lodgepole pine, resulting in relatively less of an increase in HDR over the same period of time. Despite the suggestion that light availability is typically the limiting resource in lodgepole pine and the primary influence on the relative growth of height and diameter, there are many complex influences on HDR. Height and diameter growth are influenced by planting position, elevation, growing season, litter depth, slash, slope, aspect, tree species, age, seasonal climate, site preparation, stock type and provenance (Zimmerman and Brown 1971, MacDonald et al. 1990, Kozlowski and Pallardy 1997, Mustard and Harper 1998, Opio et al. 2000). HDR may be potentially influenced by many factors. It is important to gain a better understanding of the various factors that influence HDR 14 and determine its feasibility as a competition index. The above-stated influences on HDR suggest caution with respect to the use of standards (i.e., reference HDRs or HDR thresholds). HDR addresses many of the concerns Burton (1993) has with competition indices. First, with repeated measurements, HDR can represent system dynamics such as seasonality and differences in growth trajectories. Second, since HDR is a direct measure of competition it is not misled by measures of neighbouring vegetation that are not important, avoids problems associated with the fixed neighbourhood search radius, and avoids assumptions about the abundance and size of competitors that are counted. Third, HDR accounts for below ground constraints as far as they are represented by above ground attributes (i.e., relative changes in height and diameter). HDR can be used in a manner that addresses most of Burton’s (1993) recommendations as follows: 1. Tracking the ratio of “standard seedling performance to control seedling performance” (Burton 1993) is easy to do with HDR, although was not specifically recommended in the thesis. Opio et al. (2000) used relative percent change (RC%) to compare treatment mean HDR and control mean HDR. Furthermore, Opio et al. (2003) used percent change in HDR (%AHDR) to track changes in HDR from the time of the site installation. 2. Burton (1993) describes ways of accounting for the dynamics of growth of subject trees and non­ crop vegetation. Repeated measures of HDR capture much of this subject tree-non-crop vegetation dynamics. 3. Reference HDRs (i.e., HDR thresholds) recommended in the thesis have been “locally calibrated” to species, vegetation complex, BEC classification, percent cover of competing vegetation, and ranges of stem diameter or volume in specified years following planting of the crop trees. The HDR thresholds remain to be verified (i.e., validated) on similar sites. 15 Literature cited Abetz, P. von. 1976. Beitrazum Baumwachstum. Der Forst und Holzwirt 19, 389-398. Armit, D. 1996. Silvics and silviculture of lodgepole pine in the north central interior of British Columbia: a problem analysis. BC Forest Service Research Note 40, Victoria, BC. Bassman, J., and Koch, P. 1996. Physiology. In Lodgepole pine in North America. Vol. Conifers. Edited by P. Koch. Forest Products Society, Madison, Wisconsin, pp. 81-137. BC Ministry of Forests 1996. Forest Practices Code of British Columbia Act. B.C. Ministry of Forests, Victoria, BC. BC Ministry of Forests 1998. Draft vegetation competition guidelines in the Prince Goerge Forest Region to assist districts in determining free growing on areas with basic silvicultural obligations for the 1998 field season. Prince George Forest Region, Ministry of Forests, Prince George, BC. BC Ministry of Forests 2000. Establishment to Free Growing Guidebook, Prince George Forest Region (Revised edition. May 2000). Ministry of Forests, Victoria, BC. Burton, P.J. 1993. Some limitations inherent to static indices of plant competition. Can. J. For. Res. 23: 2141-2152. Chen, H.Y.H. 1997. Interspecific responses of planted seedlings to light availability in interior British Columbia: survival, growth, allometric patterns, and specific leaf area. Can. J. For. Res. 27: 1383-1393. Comeau, P.G., Braumandlt, T.F., and Xie, C.Y. 1993. Effects of overtopping vegetation on light availability and growth of Engelmann spruce {Picea engelmanii) seedlings. Can. J. For. Res. 23: 2044-2048. Coopersmith, D., and Hall, E., 1999. Experimental project 1077 - the Siphon creek mixedwood trial: the use of a simple height-to-diameter ratio to predict the growth success of planted white spruce seedlings beneath aspen canopies. Research Note No. PG-17. Prince George Forest Region. Prince George, BC. Davis, I. 1998. Non-crop Vegetation, detrimental or not?: redefining Free Growing. B.C. Ministry of Forests. Victoria, BC. DeLong, S.C. 1991. The light interception index: a potential tool for assisting in vegetation management decisions. Can. J. For. Res. 21: 1037-1042. DeLong, S.C., Tanner, D., and lull, M.J. 1993. A Field Guide for Site Identification and Interpretation for Southwest Portion of Prince George Forest Region. Land Management Handbook 24, Research Branch, Ministry of Forests, Province of British Columbia, Victoria, BC. Eis, S., Craigdallie, D., and Simmons, C. 1982. Growth of lodgepole pine and white spruce in the central interior of British Columbia. Can. J. For. Res. 12 (3): 567-575. 16 Firbank, L.G., and Watkinson, A.R. 1987. On the analysis of competition at the level of the individual plant. Oecologia (Berlin) 71: 308-317. Froese, K. 2000. Height to diameter ratios as an indicator of competitive stress in lodgepole pine {Pinus controrta var. latifolia) under four brushing treatments. BSc. (NRM-forestry major) Professional Report. University of Northern British Columbia, Prince George, BC. Harrington, T.T., and Tappeiner, J.C.B. 1991. Competition affects shoot morphology, growth, duration and relative growth rates of Douglas-fir saplings. Can. J. For. Res. 21: 474-481. Henry, H.A.L., and Aarssen, L.W. 1999. The interpretation of stem diameter-height allometry in trees: biomechanical constraints, neighbourhood effects, or biased regressions? Ecology Letters 2 (2), 89-97. Husch, B., Miller, C.I., and Beers, T.W. 1993. Forest Mensuration. Krieger Publishing Co. Malabar, Florida. Kimmins, J.P. 1997. Forest Ecology: A Foundation for Sustainable Management 2"‘‘ ed. Prentice Hall, Upper Saddle River, NJ. Klinka, K., Krajina, V., Ceska, A., and Scagel, A. 1989. Indicator plants of coastal British Columbia. Univ. of British Columbia, Vancouver, BC. Kozlowski, T.T., and Pallarady, S.G. 1997. Physiology of Woody Plants. Academic Press Inc., Toronto. Lanner, R.M. 1985. On the insensitivity of height growth to spacing. For. Ecol. Manage. 13: 143-148. Lotan, I.E., and Critchfield, W.B. 1990. Lodgepole pine. In Silvics of North America, Vol. Conifers. Edited by R.M. Bums, and B.H. Honkala. USDA For. Serv., Agric. Hndbk. 654, Washington, DC. MacDonald, B., Morris, D.M., and Marshall, P.L. 1990. Assessing components of competition indices for young boreal plantations. Can. J. For. Res. 20: 1060-1068. Maclsaac, D.A., and Navratil, S. 1996. Competition dynamics in juvenile boreal hardwood-conifer mixes. In Silviculture of temperate and broadleaf conifer mixers. Edited by P. Comeau and K.D. Thomas. BC Min. For. Victoria, pp. 23-34. Merkel, V.O. 1975. Schreebmch in Fictenbestand bei 40jahriger Auslese durchforstung. Allgemeinen Forst Zeitschrift 30, 663-665. Morris, D.M., and McDonald, G.B. 1991. Development of a competition index for young conifer plantations established on boreal mixedwood sites. For. Chron. 67: 403-410. Mustard, J., and Harper, G. 1998. A Summary of the Available Information on Height to Diameter Ratio. B.C. Ministry of Forests. Victoria, BC. Navratil, S., and Maclsaac, D A. 1993. Competition index for juvenile mixed wood stands of lodgepole pine and aspen in West-Central Alberta. Forestry Chronicle, Northwest Region, Forest Management Note 5. 17 Oliver, C D., and Larson, B.C. 1996. Forest Stand Dynamics. McGraw-Hill, New York. Opio, C., Jacob, N., and Coopersmith, D. 2000. Height to diameter ratio as a competition index for young conifer plantations in northern British Columbia, Canada. For. Ecol. Manage. 137: 245-252. Opio, C., Diest, K. van, and Jacob, N. 2003. Intra-seasonal changes in height to diameter ratios for lodgepole pine in the central interior of British Columbia, Canada. West. J. Appl. For. 18 (1): 5 2^9. Perala, D.A. 1990. Quaking aspen. In Silvics of North America, Vol. Conifers. Edited by R.M. Bums, and B.H. Honkala. USDA For. Serv., Agric. Hndbk. 654, Washington, DC. Richardson, B., Kimberley, M.O., Ray, J.W., and Coker, G.W. 1999. Indices of interspecific plant competition for Pinus radiata in the central north island of New Zealand. Can. J. For. Res. 29: 898-905. Ross, M.S., Flanagan, L.B., and La Roi, G.H. 1986. Seasonal and successional changes in light quality and quantity in the understory of boreal forest ecosystems. Can. J. Bot. 64: 2792- 2799. Ruel, J-C. 1995. Understanding windthrow: silvicultural implications. For. Chron. 71: 434-445. Ruel, J-C., Messier, C., Doucet, R., Claveau, Y., and Comeau, P. 2000. Morphological indicators of growth response of coniferous advance regeneration to overstorey removal in the boreal forest. For. Chron. 76: 633-642. Salisbury, F.B., and Ross, C.W. 1992. Plant Physiology. Wadsworth Publishing Co., Belmont, CA. Smith, D.M., Larson, B.C., Kelty, M.J., and Aston, P.M.S., 1997. The Practice of Silviculture: Applied Forest Ecology. John Wiley and Sons, New York. Smith, J.H.G., 1986. Projections of stand yields and values to age 100 for Douglas-fir, western red cedar, and western hemlock and implications for management from the UBC forest spacing trials. Faculty of Forestry, University of British Columbia. Vancouver, BC. Tilman, D. 1996. Mechanisms of plant competition. In Plant Ecology. Edited by M.J. Crowley. Blackwell Science, Oxford, UK. pp. 239-261. Wagner, B. 1994. To treat or not to treat: what is the threshold? The VMAP Report, Vegetation Management Alternatives Program. Fall 94. Vol. 3 No. 2. Wagner, R.G., and Radosevich, S.R. 1991a. Interspecific competition and other factors influencing the performance of Douglas-fir saplings in the Oregon Coast Range. Can. J. For. Res. 21: 821-828. Wagner, R.G., and Radosevich, S.R. 1991b. Neighborhood predictors of interspecific competition in young Douglas-fir plantations. Can. J. For. Res. 21: 829-835. Wagner, R.G., and Radosevich, S R., 1998. Neighbourhood approach for quantifying interspecific competition in coastal Oregon Forests. Ecol. Appl. 8: 779-794. 18 Wagner, R.G., Mohammed, G.H., and Noland, T.L. 1999. Critical period of interspecific competition for northern conifers associated with herbaceous vegetation. Can. J. For. Res. 29: 890-897. Waring, R.H. 1987. Characteristics of trees predisposed to die. BioScience 37, 569-574. Waring, R.H., and Pitman, G. 1985. Modifying lodgepole pine stands to change susceptibility to maintain pine beetle attack. Ecology 66 (3): 889-897. Waring, R.H., and Schlesinger, W.H. 1985. Forest Ecosystems: Concepts and Management. Academic Press, Toronto, Ont. Weingartner, D.H., and Basham, J.T., Editors. 1979. Forest management and research needs in the boreal mixedwood forest of Ontario. Spruce-Fir-Aspen Forest Research Committee. Canadian Foretry Service, Sault Ste. Marie, and the Ontario Ministry of Natural Resources, Toronto. Williams, H., Messier, C., and Kneeshaw, D.D. 1999. Effects of light availability and sapling size on the growth and crown morphology of understorey Douglas-fir and lodgepole pine. Can. J. For. Res. 29: 222-231. Zimmerman, M.H., and Brown, C.L. 1971. Trees: Structure and Function. Springer-Verlag New York Inc., New York. 19 CHAPTER 2 INTER-SEASONAL VARIATIONS IN HEIGHT TO DIAMETER RATIOS IN lTM>GEIKRLEI%NnSFCMJUDVnDM;i^lRlABIJ2RE3HCnA&L()F COMPETING VEGETATION Abstract The use of height to diameter ratio (HDR) as a forest vegetation management tool is beginning to gain acceptance as part of the free growing assessment procedure in British Columbia (BC). Wider acceptance of HDR requires a better understanding of how HDRs reflect various physiological and environmental constraints to growth. Trends in HDRs of young lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelra.) crop trees were investigated for three years following the removal (i.e., brushing) of above ground competing vegetation on four study sites in the Bednestii Lake and Fraser Lake areas of central BC. The design consisted of four levels of bmshing (0.0 m or control (no brushing), 0.75 m, 1.0 m and 1.25 m brushing radii), replicated three times on each site. The study involved a completely randomized, one-factor experimental design, with replication of measurements over time. Brushing treatment was the factor. It was found that: (i) brushing treatment, year of measurement, and interaction had a significant (p < 0.01) overall effect on the mean HDRs; (ii) for 1.0 m and 1.25 m radius brushing treatments, the mean HDR (and mean percent change in HDR) decreased significantly (p < 0.05) with time; (Hi) in the second and third years of measurements, the mean percent change in HDR for trees brushed to a narrower brushing radius was significantly (p < 0.05) greater than the mean percent change in HDR for trees brushed to a wider brushing radius; (iv) reference HDRs defined by the ranges of 40-49, 40-51,45-54, and 38-47 were recommended within ranges of percent cover aspen and alder, and mean diameter for specified years for the four sites; and (v) the optimum brushing radius for all sites, and that was recommended for bmshing similar sites, was in the range 1.0-1.25 m. 20 2.1 Introduction HDR is an individual tree-based index, calculated by dividing the height of the crop tree either by the diameter at the root collar or diameter at breast height (DBH). It is commonly thought that HDR is primarily influenced by the availability of light (Chen 1997). As the availability of light increases, the HDR is thought to decrease (Wilhams et al. 1999). For example, HDRs were measured over a number of years by Williams et al. (1999) and Konopka et al (1987, cited in Wang et al. 1998). These researchers found that HDRs increased after crop trees were planted, stabilized for a time, and then decreased. Change in the availability of light is often presumed to be the dominant contributor to such inter-seasonal changes in HDRs, however, other factors may have contributed to the described variation. Height and diameter growth are influenced by many factors in addition to availability of light. Litter depth, slash, slope, aspect, tree species, age, seasonal climate, site preparation, stock type, provenance, and site quality may affect HDR (Burton 1993, Maclsaac and Navratil 1996, Davis 1998, Mustard and Harper 1998). Wind, heavy snow loads, and bending of the stem influence height and diameter growth (Zimmerman and Brown 1971, Henry and Aarssen 1999, Ruel et al. 2000). Opio et al. (2000) determined that planting position has a significant effect on HDR. When light is not the limiting resource, then various patterns in height and diameter growth may complicate the interpretation of HDR (Mustard and Harper 1998, Ruel et al. 2000). For example, nutrient deficiency may result in a lower HDR as more resources are allocated to root growth (Mustard and Harper 1998), and a nutrient surplus may result in a higher HDR (Waring 1987). This limitation, however, may be overcome with the development of locally calibrated standards, the use of another indicator along with HDR, and repeated measurements of HDR over a number of years. Deficiencies in our knowledge need to be addressed. First, there is a need for locally calibrated reference HDRs (i.e., HDR thresholds) that are applicable to the young lodgepole pine {Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) plantations for which forest managers wish to 21 use this tool. Second, reference HDRs that are available have generally been defined in terms of HDR itself, or HDRs that prevailed where levels of non-crop vegetation were deemed acceptable. There is a shortcoming in using HDR independently from other criteria. Third, there is a lack of knowledge of how HDRs respond to the light produced by different levels of removal (i.e., brushing) of above ground competing vegetation, and what would be an optimum brushing radius. Possible criteria to be used in conjunction with HDR include stem volume, wood quality, and other attributes of the tree or site. Stem volume is attractive as a criterion to be used with HDR because it is readily quantifiable, and an accepted measure of productivity in the early stages of stand development. For example, Wagner et al. (1999) used a stem volume index (i.e., stem diameter^ x height) to determine the critical period of interspecific competition. In order to use stem volume in conjunction with reference HDRs, models of stem volume need to be developed; for this, destructive sampling and stem analyses need to be undertaken. In the absence of this criterion, however, stem diameter, basal area (Husch et al. 1993), and stem volume index (Wagner et al. 1999) are alternate criteria. Stem diameter is a criterion suitable for use with HDR because it is known to be highly correlated with stem volume, and is the key determinant of stem volume in most stem volume equations (Husch et al. 1993). To increase our knowledge about the use of HDR as a competition index, two approaches have been taken in this chapter. First, HDR thresholds are defined with reference to percent change in HDR (i.e., %AHDR is defined with respect to the time of site installation); a suitable reference HDR is determined as being the point at which HDRs stabilize (i.e., %AHDRs from one year to the next are negligible), and a base level HDR is being achieved. Second, in the absence of stem volume criteria, recommended reference HDRs are delimited within ranges of stem diameter obtained during specified years following planting (i.e., the post-treatment measurement period). The objectives of this chapter are: (i) to determine how HDRs of young lodgepole pine respond to different brushing radii, over time; (ii) to recommend reference HDRs that apply to plantations similar to the study sites and to define these within ranges of diameter, biogeoclimatic 22 ecosystem (BEC) classifications, and percent cover of competing vegetation; and (in) to determine which brushing radius (i.e., 0.75 m, 1.0 m, or 1.25 m radius) is best to brush plantations similar to the study sites. To address the above objectives, the following hypotheses were tested. First, for a given treatment, the mean HDR decreases with time. Second, for a given treatment, mean percent change in HDR decreases with time. Third, in any one year, the mean percent change in HDR for trees brushed to a narrower brushing radius is greater than the mean percent change in HDR for trees brushed to a wider brushing radius. 2.2 Materials and methods 2.2.1 Study sites The study involved four lodgepole pine plantations (sites) in the sub-boreal spruce (SES) biogeoclimatic zone, in the Vanderhoof Forest District (lat. N 53‘47’ to N 54" 03’, and long. W 123" 32’ to W 124" 51’) (Fig. 2.1). Using the BC biogeoclimatic ecosystem classification (EEC) as a basis for comparison, these plantations included two sites in the dry warm (dw) subzone, variant 3, site series 01 (Little Bobtail Lake and 101 km sites); and one site each in the dry cool (dk) subzone, site series 01 (137 km site) and 05 (1116 km site) (DeLong et al. 1993). At the time of plot installations (1998 and 1999) these plantations ranged between four and ten years of age; they were planted between 1990 and 1995. The sites were selected from areas in which competition from trembling aspen (Populus tremuloides Michaux), sitka alder (Alnus crispa ssp. sinuata (Regel) Hulten), and paper birch (Betula papyrifera Marshall) were severe. The sites ranged in elevation from approximately 755 to 854 m above sea level. 23 W 120 NGO British W l20” c i f Prince ^ ^ 1 5 4 , G eorg& . C o lu m b ia 114 icouver CanFor-Bednestii: Fraser Lake: C l: Little Bobtail Lake site FI: 101 km site F 2 : 137 km site F 3 : 1116 km site Fig. 2.1. Location of the study sites 24 Lodgepole pine {Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) stock types planted on the sites were PSB 211A (1+0) (Little Bobtail Lake site), PCT 313B (1+0) (101 km site), PSB 211 (1+0) (137 km site), and PSB 211A (1+0) (1116 km site) with initial HDRs estimated to be approximately 44.0, 48.6, 83.3, and 44.0, respectively. The trees were planted on the sites at an approximate range of 1270-1460 stems/ha. Mean annual precipitation in the Prince George Forest Region, where the CanFor-Bednestii and Fraser Lake sites are located, ranges from 427 to 648.5 mm. Some of the areas in the Vanderhoof Forest District are the warmest of the SBS variants (SBS dw3) (i.e.. Little Bobtail Lake and 101 km sites), while other areas in this Forest District are somewhat cooler (SBS dk) sites (i.e., 137 km and 1116 km sites). Winter precipitation in the areas of the Forest District studied is relatively low, with winter snowpacks generally less than 2 m in depth (DeLong et al. 1993). The sites were prepared for planting by windrow burning followed by disc trenching. The crop trees were planted on raised spots created by the trenching, with the majority planted on the middle and top positions (McMinn and Hedin 1990). One site was disc trenched two years prior to planting (i.e.. Little Bobtail Lake site); and all other sites were disc trenched one year prior to planting. 2.2.2 Experimental design The experimental design was a completely randomized one-factor design, with replication of measurements over time. Brushing within the prescribed radius, was the factor (Sit 1995, Zar 1996). The design consisted of four levels of brushing (0.0 m or control (no brushing), 0.75 m, 1.0 m and 1.25 m brushing radii) replicated three times on each study site (12 plots/site). Randomized plot layouts for all study sites are depicted in Appendix A. A relatively homogeneous aspen-dominated area (stratum) within the cut block was selected for sampling on each site. Where alder or birch was the dominant form of persistent competing vegetation, then a relatively homogeneous alder- or birchdominated area was selected (i.e., 1116 km site). A total of 12 plots, each 11.28 m in radius 25 (0.04 ha), were randomly located within a 120 m x 90 m (1.08 ha) study site, on the selected stratum of the cut block. A buffer of about 7.44 m between plots was thereby established. Each plot had approximately 50 trees (600 trees/site). All crop trees within a plot were sampled late in the growing season (typically after mid-August) on each site, in each of the study years. Treatment plots were brushed within the specified radius prior to the initial HDR measurements being taken. The brushing procedure was similar to that described by Smith et ai. (1997), but with a minor difference. Instead of employing a conventional cylindrical approach, a branch overhanging the perimeter of the brushing radius was retained as long as the stem of the competing vegetation was located outside the prescribed radius. The brushing radii of crop trees were measured between the centre of the bases of the crop tree and competing vegetation. This departure from the conventional approach was taken primarily for reasons of experimental control. Each treatment plot was bmshed at the time the plot was placed (between June 10 and July 29) in 1998 or 1999. Treatment plots were brushed according to one of three bmshing radii (0.75 m, 1.0 m, and 1.25 m). Control plots (0.0 m) were not bmshed. Treatment plots were re-bmshed between June 12 and July 6 in 1999 and 2000 (2000 and 2001 for the site established in 1999) to maintain the bmshing radii as they were initially established. The period between the initial bmshing and initial HDR measurements was relatively short, and no treatment effect was noticeable in the first year of HDR measurements (Opio et al. 2000). 2.2.3 Measurements Mean percent cover and height, and distribution, of aspen, alder, other deciduous shmbs, and herbaceous plants were visually estimated within each quadrant of a plot prior to bmshing the sites (between June 10 and July 29, in 1998 and 1999). Herbaceous plants were identified and listed in order of their relative abundance, also prior to bmshing. Thus, an approximation of the level of competition that existed on the plots at the time of the assessments was provided. 26 Total height, diameter and leader length measurements were taken between August 7 and September 20 for all crop trees in a plot in each of three years (1998-2000 for the Little Bobtail Lake, 101 km, and 137 km sites; and 1999-2001 for the 1116 km site). All crop trees on a site were measured within a two to five day interval late in the growing season, in each year of the study. Crop trees were typically re-measured (in 1999 and 2000, or 2000 and 2001) within two weeks of the date when HDR measurements were initially taken (in 1998 or 1999). This complete set of HDR measurements was undertaken at a point in time (typically after mid-August) when it is generally believed that most of the height and diameter growth for the year has occurred (Mustard and Harper 1998). Total height (cm) measurements were made with a height pole, total height being defined as the distance between the root collar (upper side of the slope at the top of the mineral soil) and the tip of the bud. Diameter (cm) measurements were made, at 1 cm above the root collar or above the swelling of the root collar (DiameterRc), with electronic callipers or a diameter tape (i.e., larger trees at the 137 km and 1116 km sites) to the nearest 0.01 cm. The locations of diameter measurements were identified on the tree stem with a painted line. Two diameter measurements (i.e., cross-sectional distances) were taken of each tree; one transected the stem at the wide diameter, and another transected the stem at the narrow diameter. The two diameter measurements were combined to produce an average diameter. Damage to trees caused by disease, insects, or mechanical damage was noted at the time of measurement, in each year of the study. Malformations of the stem such as multiple leaders were also noted at this time. 2.2.4 Analysis Total height and diameter measurement data for each site were screened by means of the following procedure. First, lodgepole pine trees where growth in height in a year (1999 and 2000, or 2 0 0 0 and 2 0 0 1 ) was less than 1 0 cm due to browsing, mechanical breakage, disease, or tree dying 27 were dropped from the data sets. Second, crop trees where growth in diameter in a year was nil (or where a reduction in diameter took place), due for example, to root collar weevil infestation, were dropped from the data sets. The above criteria were chosen because they gave an approximate indication of major damage due to browsing, mechanical breakage, disease, or tree dying without rejecting those diseased or damaged trees that were performing relatively well despite the disease or damage they suffered. The initial measurements (1998 or 1999) of total height, diameter, and HDR were tested for normality using the Kolmogorov-Smimov and Lilliesfors tests, and homoscedasticity (homogeneity of variance) using the Brown-Forsythe and Levene tests (Zar 1996, StatSoft 1999). The KolmogorovSmimov and Lilliesfors tests for normality indicated that HDRs, total heights, and diameters for all sites were normally distributed. The Brown Forsythe and Levene tests indicated that homoscedasticity for HDRs, total heights, and diameters for all sites (1998 and 1999) was achieved. HDRs were individually calculated for each tree in each of the three years measurements were taken: HDR rc = Height! DiameterRc where Height = total height (cm) measurement previously described, and Diameterrc = diameter (cm) measured approximately 1 cm above the root collar. The mean HDRrc for each plot within a site was calculated for use in the analysis. The mean diameterRc for each plot within a site was calculated for the purpose of defining ranges of mean diameterRc within which reference HDRs were recommended. Relative growth rate and simple ratios are among the techniques used to compare growth characteristics across years (Kozlowski and Pallardy 1997). Percent change in HDR (%AHDR) combines both these approaches. Although other approaches were possible (Opio et al. 2000, Opio et al. 2003), it was advantageous to use %AHDRs, and normalize HDRs to the year of the initial treatments. Percent change in HDR was calculated for each tree: %A^DR, = [(FfDR, - FfDRgg)/ 28 * 100 where %AHDif; = percent change in HDR between 1998 (year of installation) and 1999 and 2000, or 1999 (year of installation) and 2000 and 2001; = HDR in 1999 and 2000, or 2000 and 2001 (years following year of installation). The mean %AHDR for each plot within a site was calculated for use in the analysis. Three hypotheses, adapted from Froese’s (2000) statement of these hypotheses, were tested. First, for a given treatment, the mean HDR decreases with time (called "HDRgg > HDR9 9 > HDRqo hypothesis”). Second, for a given treatment, mean percent change in HDR decreases with time (called "%AHDRgg > %AHDR% > %AHDRqo hypothesis”). Third, in any one year, the mean percent change in HDR for trees brushed to a narrower brushing radius is greater than the mean percent change in HDR for trees brushed to wider brushing radius (called “%AHDRo,om > %AHDRo,7 5 m> %AHDRi.om > %AHDRi.2 5 mhypothesis”). With respect to the second hypothesis, %AHDRs were calculated relative to the HDR in 1998 (i.e., %AHDRgg=0). With respect to the third hypothesis, each of %AHDRo.om, %AHDRo,7 5 n» %AHDRi.om, and %AHDR 1.2 5 mwere calculated relative to the HDR in 1998, and the hypothesis tested for 1999 and 2000 only. Multivariate analysis of variance (MANOVA) with repeated measures was performed using Statistica® on each site to test whether the mean HDRs of crop trees were significantly (p < 0.05) different among treatments, dates of measurement, and interaction between these two factors (StatSoft 1999). The MANOVA model used was: HDRijk = /r + treatmenti + datej + (treatment*date)ij + Sÿ* where HDRijk = plot mean HDR, n = grand mean HDR, treatmenti = brushing radius (0.0 m or control, 0.75 m, 1.0 m, 1.25 m), datej = year of measurement (1998, 1999 and 2000; or 1999, 2000 and 2 0 0 1 ), (treatment*date)y = interaction between treatment and date of measurement, and £p = experimental error (Johnson and Wichem 1992, p. 263-4). Factors such as stock type, biogeoclimatic classification, and planting density were constant for a study site, thus they were not included in the analysis. Opio et al. (2000) had previously 29 determined that planting position had a significant (p < 0.05) effect on mean HDRs, therefore this factor was not included in the present study. Comparisons between study sites were conducted on the basis of vegetation and other site characteristics. MANOVA was used instead of the usual analysis of variance (ANOVA) because HDRs of repeated measurements were highly correlated. MANOVA is commonly applied to repeated measures data with several correlated dependent variables (von Ende 1993, p. 117-8). The repeated measures MANOVA provided an overall assessment of the mean HDRs of crop trees; and whether the factors, treatment and date, had a significant effect on the mean HDRs. If the MANOVA determined that one or more of the factors had a significant overall effect, then investigating more specific effects by means of Tukey HSD post hoc procedures was justified. The Tukey HSD post hoc test was performed on each site to test whether the mean HDRs were significantly (p < 0.05) different between specific years of measurement (1998, 1999 and 2000; or 1999, 2000 and 2001). Differences between mean HDRs, broken down by treatment and year of measurement, were assessed by means of the HDRgg > HDR9 9 > HDRqo hypothesis. The Tukey HSD test was performed on each site to test whether the %AHDRs were significantly (p < 0.05) different between specific treatments (0.0 m or control, 0.75 m, 1.0 m, and 1.25 m brushing radii) and specific years of measurement (1998, 1999, and 2000; or 1999, 2000, and 2001). Differences between mean %AHDRs, broken down by treatment and year of measurement, were assessed by means of the %AHDR98 > %AHDR99 > %AHDRqo and %AHDRo.om > %AHDRo7 5 m> %AHDR; om> %AHDR 1 .2 5 m hypotheses. The procedures described above helped to indicate whether the mean HDRs were stabilizing (i.e., changes in HDR from one year to the next were becoming negligible), and a base level HDR was being achieved. Three years of HDR measurements made it possible to observe from the graphs of mean %AHDRs whether a faster response to treatment was being realized in the period 1998-1999, or in the period 1999-2000 (alternately, 1999-2000, or 2000-2001). Recommendations for reference HDRs (or HDR thresholds) were determined by testing the third hypothesis in the final year of 30 measurements. Reference HDRs were determined on the basis of mean HDRs ± standard error of the mean (SEM) for the 1.0 m and 1.25 m brushing treatments in 2000 (or 2001). Recommendations for reference HDRs were qualified by the following procedure: if the response to treatment was slowing down in the second period (i.e., slope of mean %AHDR was decreasing or flattening out), then mean HDRs would probably not decline too much further than the levels achieved at the time of the 2000 measurements. A “satisfactory estimate” of the reference HDR was thereby determined. Conversely, if the response to treatment was getting faster (or not declining) in the second period (i.e., slope of mean %AHDR was increasing or remaining constant), then it seemed that mean HDRs could decline significantly further than the levels obtained at the time of the 2000 measurements. In the latter case, a “tentative estimate” of the reference HDR was produced. Based on this procedure, recommendations were obtained for reference HDRs for each site. In order to delimit the application of reference HDRs to areas to which they should be applied, the recommendation for a reference HDR was linked to a specific BEG classification, vegetation complex, percent cover of competing vegetation, and the size of trees on a site. Reference HDRs were recommended within ranges of percent cover of competing vegetation (e.g., ranges of percent cover aspen 5 years after planting), and mean diameter of the crop trees (e.g., ranges of mean diameter 5, 6 , and 7 years after planting) on a site. The reference HDRs recommended in this chapter were made to depend upon diameter because diameter has been viewed as being the major determinant of stem volume, and is the basis for reference HDRs recommended in Chapter 4 (Husch et al. 1993). Thus, percent change in diameterRc {%ADiam) was calculated for each tree. Any impact that brushing treatments had on diameter (and stem volume) increment was measured by calculations of percent difference between mean diameter for no brushing and mean diameter for treatments in 2 0 0 0 {%ADiam2 ooo)\ and percent difference between mean percent change in diameter for no brushing and mean percent change in diameter for treatments in 2000 (%A^Diam2 ooo)- An exact description of these calculations is too lengthy to be 31 included in the thesis. They are analogous to those for %AVol2 ooo and %A^Vol2 ooo described in Chapter 4 (Section 4.2.4). The optimum brushing radius around crop trees was determined from the mean %AHDRs in the final year of measurement. Where there was a significant (p < 0.05) separation in mean %AHDRs between brushing treatments in 2 0 0 0 (or 2 0 0 1 ), one brushing radius was recommended over another (i.e., the 1.25 m brushing radius produces lower HDRs than the 1.0 m bmshing radius). Where there was a poor separation in mean %AHDRs between the 1.0 m and 1.25 m treatments in 2000, a range of bmshing radii was recommended (i.e., 1.0 m to 1.25 m bmshing radius produces lower HDRs than no bmshing). For operational reasons, the narrowest bmshing radius (0.75 m) was not recommended for any of the sites. The influence of competing vegetation on HDRs was assessed by means of ANOVAs and regressions. The ANOVAs were used to assess the homogeneity/heterogeneity or uniformity of non­ crop vegetation at the sites prior to bmshing treatments, and the regressions were used to explore for possible relations between HDR and percent cover and/or height of competing species. Homogeneity/heterogeneity of competing vegetation may indicate the relative nutrient-richness of a site, with the more homogeneous or uniform site indicating greater nutrient-richness. 2.3 Results The MANOVAs on mean HDRs (Table 2.1) generally indicated that bmshing treatment, year of measurement, and the interaction between bmshing and year had a significant overall effect for all the study sites. An exception was the 1116 km site, where treatment was found to be marginally non­ significant (p = 0.07). No exceptions were found for year, and interaction between treatment and year at any sites. The MANOVAs on percent changes in HDR (Table 2.1) indicated approximately the same outcome, where treatment was found to have a significant overall effect for all the sites. 32 Table 2.1. MANOVAs with repeated measures showing the factors affecting mean HDRs and Study site Effect Mean HDR F Value (Pr > F) df Effect Percent change in HDR F Value (Pr > F) CanFor-Bednestii Little Bobtail Lake Brushing Year Brushing* Year 3 14.85 0 .0 0 4.14 0.05 2 102.54 0 .0 0 86.50 0 .0 0 6 4.19 0 .0 1 4.02 0 .0 1 Brushing Year Brushing* Year 3 Fraser Lake 101 km 137 km 1116 km 2 6.64 334.63 0.00 50.35 378.98 0.00 34.19 0 .0 0 42.20 6 0 .0 0 Brushing Year Brushing* Year 3 7.60 0 .0 1 2 2^66 0.00 4J8 31.51 6 3.48 0 .0 2 4.17 0 .0 1 Brushing Year 3 3.60 0.07 9H1 0.01 2 27.61 0 .0 0 25.19 0 .0 0 Brushing*Year 6 6.75 0 .0 0 6^3 0 .0 0 0 .0 1 0 .0 0 0.03 0 .0 0 . "Percent changes in HDR were calculated with reference to HDRs in 1998 for Little Bobtail Lake site (n=486/site), 101 km site (n=555/site), and 137 km site (n=487/site); and HDRs in 1999 for 1116 km site (n=509/site). Mean HDRs and mean percent changes in HDR were calculated for each plot {n=12/site). Note; MANOVAs were run on 1998-2000 data for Little Bobtail Lake, 101 km, and 137 km sites; and 1999-2001 data for 1116 km site (n=12/site). 2.3.1 CanFor-Bednestii-Little Bobtail Lake site The Tukey HSD post hoc test indicated that the first hypothesis did not hold for the control (i.e., HDR9 8 = HDRgg ~ HDRoo). A variation of the hypothesis, HDRgg > HDR(9 9 ,oo) (i.e., HDR9 9 « HDRoo), appeared to hold for all of the treatments (Table 2.2). The Tukey HSD test confirmed this perception, indicating that mean HDRgg was significantly (p < 0.01) greater than both mean HDR9 9 and mean HDRoo for all of the treatments (except the control). 33 The Tukey HSD test indicated that the second hypothesis did not hold for the control (i.e., %AHDRgg = %AHDRgg ~ %AHDRoo). A version of the hypothesis, %AHDR g > %AHDR( ,oo) 9 99 .(i.e., %AHDR9 9 ~ %AHDRoo), appeared to hold for all of the treatments (Fig. 2.2). The Tukey HSD test confirmed this observation indicating that mean %AHDR9 gwas significantly (p < 0 .0 1 ) greater than both mean %AHDRgg and mean %AHDRoo for all of the treatments (except the control). The post hoc procedures indicated that the third hypothesis did not hold. Rather, a variation of this hypothesis - %AHDRo.om > %AHDR(i.om, 1 ,2 5 m) (i.e., %AHDR,.om ~ A%HDR|.2 5 m) - appeared to hold in 1999 and 2000 (Fig. 2.2). The post hoc procedures confirmed this perception, indicating that mean %AHDRo.om was significantly (p = 0.04) greater than mean %AHDRi.2 5 m in 1999; and mean %AHDRo,om was significantly (p < 0 .0 1 ) greater than both mean %AHDRi,om and mean %AHDRi 2 5 m in 2 0 0 0 . No significant differences were found between mean %AHDRo.om and mean %AHDRo.7 5 min 1999 or 2000, or between any of the treatments in 1999 or 2000. .S I : -o— 0 . 0 0 m % -10 § u -D - 0.75 m -15 -A— 1 . 0 0 m I -20 -X— 1.25 m -25 -30 1998 2000 1999 Year Fig. 2.2. Percent changes in HDR between 1998 and 2000 for CanFor-BednestiiLittle Bobtail Lake site. Percent changes in HDR were calculated with reference to HDRs in 1998 (n=486/site), and mean percent change in HDR was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. 34 The Tukey HSD test indicated that mean %AHDRs were significantly (p <0.01) different between the control and both 1.0 m and 1.25 m treatments in 2000; and mean %AHDRs were not significantly different between the treatments in 2000. 2.3.2 Fraser Lake-101 km site The Tukey HSD post hoc test indicated that the first hypothesis did not hold for the control (i.e., HDRgg ~ HDR 9 9 ~ HDRoo). Table 2.2 indicated that the hypothesis might hold for most comparisons for all of the treatments. The Tukey HSD test confirmed this perception, indicating that mean HDRgg was significantly (p < 0.01) greater than both mean HDR 9 9 and mean HDRoo for the 0.75 m treatment. Mean HDRgg was significantly (p < 0.01) greater than mean HDR 9 9 , which was significantly (p < 0.01) greater than mean H D R qo for both the 1.0 m and 1.25 m treatments. The Tukey HSD test indicated, except for the control, that the second hypothesis held in almost all comparisons. Fig. 2.3 indicated that %AHDR9 g > %AHDRg9 > %AHDRqq held for most comparisons in all the treatments. The Tukey HSD test confirmed this perception indicating that mean %AHDR9 8 was significantly (p < 0.01) greater than both mean %AHDRg9 and mean %AHDRoo for the 0.75 m treatment. Mean %AHDR9 g was significantly (p < 0.01) greater than mean %AHDRç9 , which was significantly (p < 0.01) greater than mean %AHDRoo for both the 1.0 m and 1.25 m treatments. The Tukey HSD test indicated that the third hypothesis held throughout the comparisons. Mean %AHDRo.om was significantly (p < 0.01) greater than mean %AHDRo.7 5 m, which was significantly (p < 0.02) greater than mean %AHDRi.om, which was significantly (p < 0.01) greater than mean %AHDRi,2 5 min 1999 and 2000. Fig. 2.3 accurately reflected the Tukey test: this result for the %AHDRo.on, > %AHDRo.7 5 m> %AHDR],om > %AHDRi,2 5 mhypothesis was unique among all the sites investigated. The Tukey HSD test indicated that mean %AHDRs were significantly (p < 0.02) different between the control, 0.75 m, 1.0 m, and 1.25 m treatments in 2000 (and 1999), with the 1.25 m brushing radius producing the lowest HDRs. 35 I : •o— 0.00 m & -10 0.75 m -0 -15 -û— 1 . 0 0 m .S i 1.25 m -20 -25 -30 1998 1999 2000 Year Fig. 2.3. Percent changes in HDR between 1998 and 2000 for Fraser Lake-101 km site. Percent changes in HDR were calculated with reference to HDRs in 1998 (n=555/site), and mean percent change in HDR was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. 2.3.3 Fraser Lake-137 km site The Tukey HSD post hoc test indicated that the first hypothesis did not hold for the control or 0.75 m treatment (i.e., HDR^g = HDR9 9 ~ HDRoo). A variation of the hypothesis, HDRgs > HDR^gg, oo) (i.e., HDR9 9 ~ HDRoo), appeared to hold for both the 1.0 m and 1.25 m treatments (Table 2.2). The Tukey HSD test confirmed this perception, indicating that mean HDR 9 8 was significantly (p < 0.01) greater than both mean HDR 9 9 and mean HDRqofor both the 1.0 m and 1.25 m treatments. The Tukey HSD test indicated that the second hypothesis did not hold for the control or 0.75 m treatment. A variation of the hypothesis, %AHDR9 g > %AHDR(9 9 ,oo) (i.e., %AHDRo9 ~ %AHDRoo), appeared to hold for the 1.0 m and 1.25 m treatments (Fig. 2.4). The Tukey HSD test supported this perception indicating that mean %AHDR9 gwas significantly (p < 0 .0 1 ) greater than both mean %AHDR9 9 and mean %AHDRoofor the 1.0 m and 1.25 m treatments. 36 0 £ c 0 .0 0 m - ° - 0.75 m m -s -15 - 1 .0 0 c y -20 1.25 m ^-25- -30 1998 1999 2000 Year Fig. 2.4. Percent changes in HDR between 1998 and 2000 for Fraser Lake-137 km site. Percent changes in HDR were calculated with reference to HDRs in 1998 (n=487/site), and mean percent change in HDR was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. The post hoc procedures indicated that the third hypothesis did not hold. Variations of the hypothesis, however, appeared to hold (Fig. 2.4): %AHDRo.om > %AHDR(t,om. i.25m) (i.e., %AHDRi.om = %AHDRi,25m) appeared to hold in 1999; and %AHDR(o,om, 0.75m) > %AHDR(i,om. 1.25m) (i.e., %AHDRo.om ~ %AHDRo.75n, and %AHDRi.om = %AHDR 1.25m) appeared to hold in 2 0 0 0 . The post hoc procedures upheld this observation, indicating that mean %AHDRo.om was significantly (p = 0.040.05) greater than both mean %AHDR, om^nd mean %AHDRi.25m in 1999; mean %AHDRo.om was significantly (p < 0 .0 1 ) greater than both mean %AHDRi,om and mean %AHDR 1.25m in 2 0 0 0 ; and mean %AHDRo.75m was significantly (p = 0.02-0.03) greater than both mean %AHDRi.on,and mean %AHDR,.23m in 2000. The Tukey HSD test indicated that mean %AHDRs in 2000 were significantly different between the control and both the 1.0 m and 1.25 m treatments (p < 0.01); and between the 0.75 m treatment and both the 1.0 m and 1.25 m treatments (p = 0.02-0.03). However, mean %AHDRs in 37 2000 were not significantly different between the 1.0 m and 1.25 m treatments (or between no bmshing and the 0.75 m treatment). 2.3.4 Fraser Lake-1116 km site The Tukey H S D post hoc test indicated that the first hypothesis did not hold for the control or 0.75 m treatment (i.e., H D R 9 9 ~ HDRqo ~ H D R o i ) . However, the hypothesis appeared to hold in large measure for the 1.0 m and 1.25 m treatments, suggesting that H D R 9 9 > HDR(qo,oi) (i.e., HDRqo ~ H D R o i ) (Table 2.2). The Tukey H S D test confirmed this perception, indicating that mean H D R 9 9 was significantly (p < 0.03) greater than both H D R o o and H D R o i for the 1 . 0 m and 1.25 m treatments. The Tukey HSD test indicated that the second hypothesis did not hold for the control or 0.75 m treatment (i.e., %AHDR9 9 ~ %AHDRqo ~ %AHDRqi). A variation of this hypothesis - %AHDR9 9 > %AHDR(oo,oi) (i.e., %AHDRoo ~ %AHDRqi) - appeared to hold for the 1 . 0 m and 1.25 m treatments (Fig. 2.5). The Tukey HSD test confirmed this perception, indicating that mean %AHDR9 9 was significantly (p = 0 .0 1 ) greater than mean %AHDRoo for the 1 . 0 m treatment; and mean %AHDR9 9 was significantly (p < 0.01) greater than both mean %AHDRqq and mean % A H D R o i for the 1.25 m treatment. The post hoc procedures indicated, overall, that the third hypothesis did not hold. However, a curtailed version of the hypothesis appeared to hold (Fig. 2.5): % A H D R o .o m > % A H D R (i,o m , 1.25m) (i.e., % A H D R i ,o m ~ % A H D R i,2 5 m ) in 2 0 0 0 and 2 0 0 1 ; and % A H D R o ,7 5 m > % A H D R i . 25m i n 2 0 0 1 . The post hoc procedures generally confirmed this perception, indicating that mean % A H D R o .o m was significantly (p < 0 .0 1 ) greater than mean % A H D R i 25m in 2 0 0 0 ; mean % A H D R o .o m was significantly (p < 0 .0 1 ) greater than both mean % A H D R i.o m and mean % A H D R i 25m in 2 0 0 1 ; and mean %AHDRo.7 5 n, was significantly (p = 0 .0 1 ) greater than mean % A H D R 1,25m in 2 0 0 1 . The Tukey HSD test indicated that mean %AHDRs in 2001 were significantly (p < 0.01) different between the control and both the 1.0 m and 1.25 rri treatments, and between the 0.75 m and 38 the 1.25 m treatments. Mean %AHDRs in 2001, however, were not significantly different between the 1.0 m and 1.25 m treatments. i .5 -5 0 .0 0 & ^ 0 0.75 m I,. { m 1 .0 0 m 1.25 m -20 -25 -30 1999 2000 2001 Year Fig. 2.5. Percent changes in HDR between 1999 and 2001 for Fraser Lake-1116 km site. Percent changes in HDR were calculated with reference to HDRs in 1999 (n=509/site), and mean percent change in HDR was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. 39 Table 2.2. Mean HDRs and standard errors of mean (SEM)" from 1998 to 2001 for Year of measurement Study site Brushing radius (m) 1998 1999 2000 2001 60.4 (5.0) 51.7 (4.5) 60.9 (4.8) CanFor-Bednestii Little Bobtail 0.00 64.1 (5.0) Lake* 0.75 56.7 (4.7) {n=486/site) 1.00 1.25 64.4 (5.0) 53.8 (4.6) 56.1 (4.7) 50.6 (4.4) 55.0 (4.6) 46.9 (4.2) 44.6 (4.1) n.m. n.m. n.m. n.m. 61.5 (4.9) 62.9 (4.9) 56.8 (4.7) 59.9 (4.8) 61.2 (4.9) 58.2 (4.7) 50.0 (4.4) 49.2 (4.4) 59.0 (4.8) 55.6 (4.6) 46.4 (4.2) 44.5 (4.1) n.m. n.m. n.m. n.m. 64.0 (5.0) 55.9 (4.7) 53.6 (4.5) 53.9 (4.6) 63.0 (4.9) 54.3 (4.6) 49.8 (4.4) 63.5 (4.9) 54.6 (4.6) n.m. n.m. n.m. n.m. n.m. n.m. n.m. n.m. 46.8(4.2) 45.9 (4.2) 44.6 (4.1) 44.5 (4.1) Fraser Lake 101km" 0 .0 0 (n=555/site) 0.75 1 .0 0 1.25 137 km'' 0 .0 0 {n=487/site) 0.75 1 .0 0 1.25 1116 km' 0.00 (n=509/site) 0.75 1 .0 0 1.25 50.2 (4.4) 49.4 (4.4) 49.4 (4.4) 46.5 (4.2) 44.6(4.1) 42.6 (4.0) 42.0 (4.0) 47.8 (4.3) 45.2 (4.2) 42.8 (4.1) 41.9(4.0) “SEM are presented in brackets following mean HDRs. *'^Exact dates of measurement for ^Little Bobtail Lake site were August 7-10, 1998, August 19-23, 1999, and August 11-15, 2000; "101 km site were August 14-19, 1998, September 16-20, 1999, and August 22-25, 2000; '^137 km site were August 12-17, 1998, August 30-September 3, 1999, and September 5-11, 2000; and "1116 km site were September 7-11, 1999, September 12-15, 2000, and September 4-7, 2001. Note: n.m. indicates no measurements were taken. 40 2.4 Discussion Mean HDRs (Table 2.2) in 1998 and 1999 were at considerably higher levels for Little Bobtail Lake, 101 km, and 137 km sites (53.6-64.4) than for the 1116 km site (44.5-46.8). These results seem to be confirmed in ANOVAs run on control portions (n ~ 36/site) of the same data sets (Jacob and Opio 2003). These ANOVAs indicated overall that HDRs were not significantly different between Little Bobtail Lake and 101 km sites, and were significantly (p < 0.01) different between 137 km and 1116 km sites. It seemed that the HDRs for 1116 km site were substantially lower than those for all other sites, however this comparison was not tested. The more severe brushing treatments (1.0 m-1.25 m brushing radii) appreciably reduced mean HDRs for Little Bobtail Lake, 101 km, and 137 km sites by 2000 and 2001 (to levels between 44.5 and 55.0). However, these levels of mean HDR remained above those for the 1116 km site (41.9-42.8) in 2000 and 2001. Sitka alder (20-30% cover) and paper birch (10-15% cover) were the dominant non-crop vegetation at the 1116 km site, whereas trembling aspen (5-45%, 30-40%, and 0-30% cover) was the dominant non-crop vegetation at Little Bobtail Lake, 101 km, and 137 km sites. These differences in competing vegetation suggest that the presence of the nitrogen-fixing species of alder had the effect of reducing mean HDRs at the 1116 km site. The positive effect sitka alder has on lodgepole pine’s and other species’ growth has been demonstrated by various researchers (Sachs and Comeau 1991, Simard and Heineman 1996, Sanborn et al. 1997). The substantial presence of alder at the 1116 km site seems to explain the difference in HDRs between the two sites with larger trees (i.e., 137 km and 1116 km sites). The mean percent decline in HDRs was greater (i.e., %AHDR lesser) in the period 1998-1999 than in the period 1999-2000 for most sites (Figs 2.2-5). However, this was not the case for the 101 km site, where the mean %AHDR continued to decline in the period 1999-2000. One explanation for the difference in mean %AHDR between the Little Bobtail Lake and 101 km sites may be the difference in relative homogeneity/heterogeneity of competing vegetation (i.e., prior to bmshing treatments). Greater homogeneity or uniformity in percent cover of trembling aspen may indicate that 41 the 101 km site was nutrient-richer than the Little Bobtail Lake site. Nutrient-richness may have been reflected in a greater homogeneity of competing vegetation at the 1 0 1 km site. The greater homogeneity in non-crop vegetation at the 101 km and 1116 km sites was demonstrated by ANOVAs on deciduous percent cover at these sites. The converse was observed for the Little Bobtail Lake and 137 km sites. At the more heterogeneous sites (i.e., Little Bobtail Lake and 137 km sites) regressions were obtained that showed HDRs increasing with increasing aspen cover (R^= 0.70, F (l, 10) = 23.4, p < 0.001; and R^ = 0.69, F (l, 10) = 22.4, p < 0.001, respectively). The presence of sitka alder at one site, the relative homogeneity/heterogeneity of competing vegetation (i.e., differences in nutrient-richness) between sites, and the relative sizes of the trees between sites were among the factors which may have influenced the outcomes obtained with respect to the three hypotheses that were tested. Factors other than availability of light affect HDRs (Zimmerman and Brown 1971, MacDonald et al. 1990, Kozlowski and Pallardy 1997, Mustard and Harper 1998). Mean HDRs (Table 2.2) and Tukey HSD post hoc tests indicated that some variation of the first hypothesis held for the treatments but not for the control, for all sites. This would be expected because even a minimal level of brushing treatment generally increases the amount of light that reaches the crop tree. Thus, brushing increases the relative allocation of resources to diameter increment, and lowers the HDR (Waring and Pitman 1985, Waring and Schlesinger 1985). Most commonly, it was found that mean HDRs decreased significantly in the second year (i.e., HDR 9 9 ) but not in the third year of measurements (i.e., HDRoo): for the Little Bobtail Lake site, HDRgg > HDR(9 9 , 0 0 ) (i.e., HDR 9 9 = HDRqo) held for all the treatments; and for the 137 km and 1116 km sites, HDR 9 8 > HDR(9 9 ,oo), held for the 1.0 m and 1.25 m treatments. The 101 km site was exceptional, in that the HDRgg > HDR 9 9 > HDRqo hypothesis described the pattern for all the treatments. Mean %AHDRs (Figs. 2.2-5) and Tukey HSD tests generally indicated, except for the control, that some variation of the second hypothesis held for all the sites. The results for the second hypothesis were consistent with the results for the first hypothesis. It was found that mean %AHDRs 42 were significantly lower in the second year of measurements (i.e., %AHDR%), but that mean %AHDRs in the second year were not significantly different from %AHDRs in the third year of measurements (i.e., H D Rqo). For example, for the Little Bobtail Lake site, %AHDR@g > %AHDR(%, oo) (i.e., %AHDR9 9 « %AHDRoo) held for all the treatments; and for the 137 km site, %AHDRc,g > %AHDR(9 9 ,oo) held for the 1.0 m and 1.25 m treatments. The 101 km site was notable, in that the %AHDR9 8 > %AHDR9 9 > %AHDRoo hypothesis described the pattern for all the treatments. The overall pattern of greater response in the second year than in the third year of measurements seems to be confirmed by Ruel et al. (2000) who observed that stem diameter and structural root growth usually increase immediately following release (i.e., exposure of trees to light via removal of the overstory), whereas height growth is slower to react to release. The decline in HDRs at the 101 km site, sustained into the third year of measurements, may be due to the greater homogeneity of competing vegetation (i.e., greater nutrient-richness) at this site. The greater response to brushing at both the Little Bobtail Lake and 101 km sites seems to be due to the smaller trees at these sites, which were more able to respond to brushing interventions than were the larger trees at the 137 km and 1116 km sites (Jacob and Opio 2003). A complete description of the assessments of competing vegetation prevailing prior to bmshing treatments is too lengthy for inclusion in the thesis, however, is available in unpublished notes. The third hypothesis held in its entirety for one site, and in curtailed forms for the remaining sites. The Tukey HSD post hoc tests supported the full statement of the hypothesis for the 101 km site in 1999 and 2000; a unique result among all the sites. However, the Tukey HSD tests caused us to adopt only tmncated versions of the hypothesis for most other sites. For example, for the Little Bobtail Lake site, %AHDRo,om > %AHDR(, om, 1.2 5 m) (i.e., %AHDRi.om ~ %AHDRi,25m) held for the most part in 1999 and 2000; for the 137 km site, %AHDR(o.om, 0 .7 5 m) > %AHDR(i.om, 1 .2 5 m) (i.e., %AHDRo,om = %AHDRo.7 5 n, and %AHDR].om ~ %AHDR 1 .2 5 m) generally held in 1999 and 2 0 0 0 ; and for the 1116 km site, %AHDRo.om > %AHDR(i.om, 1 .2 5 m) (i.e., %AHDRi,om = %AHDRi.2 Sm) generally held in 2000 and 2001. Overall, the mean percent decline in HDRs was greater where the bmshing 43 radius was wider (i.e., the availability of light was greater). Increasing the availability of light has the immediate effect of increasing the allocation of resources to growth in diameter, a response also described by Waring and Pitman (1985). The observed patterns in the data indicated that; (i) the effect of brushing on HDRs was greater in the second year (i.e., one year after the sites were initially brashed) than in the third year (and potentially in subsequent years) of measurements; (ii) sites with sitka alder and paper birch (i.e., 1116 km site) as the dominant non-crop vegetation had lower mean HDRs prior to brushing than did sites dominated by trembling aspen (i.e., 44.5-46.8 vs. 53.6-64.4); (Hi) HDRs for the site (i.e., 101 km site) with greater homogeneity of competing vegetation, possibly reflecting greater nutrient-richness, seemed more likely to continue to decline after the second year of measurements than were HDRs for the sites (i.e.. Little Bobtail Lake and 137 km sites) where competing vegetation was more heterogeneous; (iv) the effect of brushing on %AHDRs was greater for sites with smaller trees (i.e., -12.8% and -16.2% for the 1.0 m and 1.25 m treatments for the Little Bobtail Lake site, and -18.0% and -25.0% for the 1.0 m and 1.25 m treatments for the 101 km site) than for sites with larger trees (i.e., -7.8% and -7.9% for the 1.0 m and 1.25 m treatments for the 137 km site, and -3.3% and -5.4% for the 1.0 m and 1.25 m treatments for the 1116 km site); (v) the effect of the 0.75 m bmshing radius on HDRs was indistinguishable from no bmshing for sites with larger trees (i.e., 137 km and 1116 km sites); and (vi) HDRs could rise again after temporarily falling in the second or third year of measurements (i.e., 2 0 0 0 - 2 0 0 1 for 1116 km site). Five management implications can be derived from the above discussion. First, the impact of bmshing interventions on growth of smaller trees can be measured in a relatively short span of time using %AHDR. Second, factors such as the presence of sitka alder, and homogeneity/heterogeneity (i.e., nutrient-richness) of a site need to be taken into account when planning bmshing interventions. Third, bmshing interventions need to be undertaken earlier (i.e., < 4-5 years after planting), rather than later (i.e., 9-10 years after planting) in the life of a plantation. Fourth, bmshing radii need to be 44 1.0 m or greater to have sufficient effect. Fifth, the medium-long term effect of brushing interventions may not be stable. The determination of HDR thresholds was based on mean HDRs (± SEM) for the 1.0 m and 1.25 m bmshing treatments in the third year of measurements. HDRs might rise again in subsequent years due to the delayed response of height growth to release (Ruel et al. 2000). The magnitude of the release response and the concurrent growth of non-crop trees in the neighbourhood of the crop tree are not known. The availability of light to the crop tree may decline in subsequent years. Other factors such as root competition may also affect a shift in HDR in subsequent years. In the absence of better information, recommendations for reference HDRs were based on HDRs in the final year of measurements. Table 2.3 presents these recommendations along with qualifications and constraints for their use. If information becomes available from a re-measurement of the sites (e.g., three years after HDR measurements were last taken), then the presently recommended reference HDRs should be re-evaluated. The mean HDRs (± SEM) for the 1.0 m and 1.25 m treatments for the Little Bobtail Lake site (Table 2.2) were 50.4-59.6 and 40.5-48.7, respectively (significant difference p < 0.01). Hence, a reference HDR of 40-49 was recommended. This HDR threshold was meant to apply on SBS dw3 (01) plantations where aspen percent cover was approximately in the range of 5-45% and height was approximately in the range of 1.0-3.5 m, 5 years after planting (1998); and mean diameters were approximately in the range of 1.70-2.00 cm, 5 years after planting 1998), 2.40-3.10 cm, 6 years after planting (1999), and 3.10-4.20 cm, 7 years after planting (2000). The mean %AHDRs (Fig. 2.2) indicated that the slopes for the 1.0 m and 1.25 m treatments were decreasing between the periods 1998-1999 and 1999-2000, suggesting that mean %AHDRs are likely to hold at their 2000 levels. Thus, the reference HDR produced (40-49) was qualified as being a “satisfactory estimate”. The mean HDRs (± SEM) for the 1.0 m and 1.25 m treatments for the 101 km site (Table 2.2) were 42.2-50.6 and 40.5-48.7, respectively (not significantly different). Hence, a reference HDR of 40-51 was recommended. This HDR threshold was meant to apply on SBS dw3 (01) plantations 45 where aspen percent cover was approximately in the range of 30-40% and height was approximately in the range of 1.0-3.5 m, 4 years after planting (1998); and mean diameters were approximately in the range of 1.50-1.90 cm, 4 years after planting (1998), 2.20-3.00 cm, 5 years after planting (1999), and 3.00-4.10 cm, 6 years after planting (2000). The mean %AHDRs (Fig. 2.3) indicated that the slopes for the 1.0 m and 1.25 m treatments were not levelling off in any appreciable amount between the periods 1998-1999 and 1999-2000, and HDRs might continue to decrease below what they were in 2000. Thus, the reference HDR obtained (40-51) was qualified as being a “tentative estimate”. The mean HDRs (± SEM) for the 1.0 m and 1.25 m treatments for the 137 km site (Table 2.2) were both 45.0-53.8 (not significantly different). Hence, a reference HDR of 45-54 was recommended. This HDR threshold was meant to apply on SBS dk (01) plantations where aspen percent cover was approximately in the range of 0-30% and height was approximately in the range of 1.0-3.5 m, 9 years after planting (1998); and mean diameters were approximately in the range of 4.306.00 cm, 9 years after planting (1998), 5.40-7.70 cm, 10 years after planting (1999), and 6.308.90 cm, 11 years after planting (2000). The mean %AHDRs (Fig. 2.4) indicated that slopes for the 1.0 m and 1.25 m treatments were decreasing between the periods 1998-1999 and 1999-2000, providing reasonable assurance that mean %AHDRs would remain approximately at their 2000 levels. Thus, the reference HDR range of 45-54 was qualified as being a “satisfactory estimate”. The mean HDRs (± SEM) for the 1.0 m and 1.25 m treatments for the 1116 km site (Table 2.2) were 38.7-46.9 and 37.8-45.9, respectively (not significantly different). Hence, a reference HDR of 38-47 was recommended. This HDR threshold was meant to apply on SBS dk (05) plantations where alder percent cover was approximately in the range of 20-30% (birch percent cover was approximately 10-15%) and height was approximately in the range of 1.5-2.5 m (alder and birch), 10 years after planting (1999); and mean diameters were approximately in the range of 6.90-7.50 cm, 10 years after planting (1999), 8.20-8.80 cm, 11 years after planting (2000), and 9.30-9.90 cm, 12 years after planting (2001). The mean %AHDRs (Fig. 2.5) appeared to be rising in the second period of 46 measurement (2000-2001) for the control and 0.75 m treatments, but were relatively stable (remaining level) for the 1,0 m and 1.25 m treatments. Thus, the reference HDR range of 38-47 was qualified as being a “tentative estimate”. Recommendations (and rationale) for optimum bmshing radii for each site are as follow. Mean %AHDRs for the Little Bobtail Lake site in 2000 were not significantly different between the treatments. Therefore, the range of 1.0-1.25m bmshing radius was recommended. Mean %AHDRs for the 101 km site in 2000 (and 1999) were significantly different between no bmshing, 0.75 m, 1.0m, and 1.25 m treatments, with the 1.25 m bmshing radius producing the lowest HDRs. Therefore, the 1.25 m brushing treatment was specifically recommended. Mean %AHDRs for the 137 km site in 2000 were significantly different between no bmshing and both 1.0 m and 1.25 m treatments; and between the 0.75 m treatment and both 1.0 m and 1.25 m treatments. However, mean %AHDRs in 2000 were not significantly different between the 1.0 m and 1.25 m treatments (or between no bmshing and the 0.75 m treatment). Therefore, the range of 1.0-1.25 m bmshing radius was recommended. Mean %AHDRs for the 1116 km site in 2001 were significantly different between no bmshing and both the 1.0m and 1.25m treatments, and between the 0.75 m and 1.25 m treatments. However, mean %AHDRs were not significantly different between the 1.0 m and 1.25 m treatments. Therefore, the range of 1.0-1.25 m bmshing radius was recommended. 47 Table 2.3. Recommended reference HDRs with vegetation complexes, BEC classifications, and ranges of percent cover of competing vegetation and mean diameter within which reference HDRs Study site Veg. complex/ Ref. Pet. cov. comp. BEC class." HDR^ veg.^ Mean diameter (cm)'' 3rd 1 st 2 nd year year year CanFor-Bednestii Little Bobtail Lake aspen SBS dw3 (01) years after planting maximum minimum 5 49 40 5 6 7 2 .0 0 3.10 1.70 2.40 4.20 3.10 4 40 30 4 1.90 5 6 3.00 4.10 1.50 2.20 3.00 9 30 9 10 11 6 .0 0 0 4.30 7.70 5.40 6.30 10 10 11 12 30 (15) 20 (10) 7.50 8.80 8.20 9.90 45 5 Fraser Lake 101 km aspen SBS dw3 (01) aspen 137 km SBS dk (01) alder/birch 1116 km SBS dk (05) years after planting maximum minimum 51 40 years after planting maximum minimum 54 45 years after planting maximum 47 minimum 38 6.90 8.90 9.30 “Veg. complex/BEC class, refers to vegetation complex and biogeoclimatic ecosystem classification within which reference HDRs are meant to be applied. *Ref. HDRs (reference HDRs) are meant to be applied on plantations of the specified age, vegetation complex, BEC classification, and ranges of percent cover of competing vegetation and mean diameter. Reference HDRs for Little Bobtail Lake and 137 km sites were quahfied as being "satisfactory estimates"; reference HDRs for 101 km and 1116 km sites were qualified as being "tentative estimates". Tct. Cov. comp. veg. refers to percent cover of competing vegetation (i.e. aspen or alder/birch) estimated before treatments were conducted. Percent cover was visually estimated in July or August 1998 for Little Bobtail Lake, 101 km, and 137 km sites; and in June 1999 for 1116 km site. '^Mean diameter refers to diameter outside bark at root collar. Diameters along with total heights were measured after mid-August 1998, 1999, and 2000 for Little Bobtail Lake, 101 km, and 137 km sites; and mid-August 1999, 2000, and 2001 for 1116 km site. 48 2.5 Conclusions and recom mendations A systematic pattern in the mean HDRs and percent changes in HDR (%AHDR) was evident in testing the hypotheses. For most sites, a variant of the first hypothesis, HDRgg > HDR(9 9 ,oo) (i.e., HDR99 ~ H D Rqo), held for some or all of the treatments. Except for the 101 km site, the effect of brushing on HDRs was greater in the second year (i.e., one year after the sites were initially brushed) than in the third year of measurements. The 101 km site was unique in that HDRs continued to decline into a third year of measurements for all treatments: the pattern HDRgg > H D R 9 9 > H D Rqo was confirmed. The relatively greater homogeneity of competing vegetation at the 101 km site (i.e., possibly reflecting greater nutrient-richness) may have contributed to the sustained decline in HDRs at this site. A variant of the second hypothesis, %AHDR9 g > %AHDR(9 9 ,oo> (i.e., %AHDRg9 ~ %AHDRqo), held for the 1.0 m and 1.25 m treatments for most sites. The effect of brushing on HDRs was greater for the sites with smaller trees (i.e.. Little Bobtail Lake and 101 km sites) than for the sites with larger trees (i.e., 137 km and 1116 km sites). HDRs could rise again after temporarily falMng in the second and third years of measurements (i.e., 2 0 0 0 - 2 0 0 1 for 1116 km site). For most sites, variations of the third hypothesis, % A H D R o .o m > % A H D R (o .7 5 m , i.om, 1.25m) (i.e., % A H D R o .7 5 m ~ % A H D R i,o m ~ % A H D R 1,25m) or % A H D R (o .o m , 0.75m) > % A H D R ( i,o m , 1.25m) (i.e., % A H D R o .7 5 m ~ % A H D R i,o m and % A H D R i.o m ~ % A H D R i,2 5 m ) held in the second and/or third year of measurements (1999 and/or 2000, or 2000 and/or 2001). For sites with larger trees, the effect of the 0.75 m brushing radius was indistinguishable from no brushing (i.e., 137 km and 1116 km sites). Using change in slope of %AHDR as a criterion, a “satisfactory estimate” of the reference HDR was obtained for the Little Bobtail Lake (40-49) and 137 km (45-54) sites; and a “tentative estimate” of the reference HDR was obtained for the 101 km (40-51) and 1116 km (38-47) sites. Reference HDRs were recommended for specific vegetation complexes, BEC classifications, ranges of percent cover competing vegetation, and ranges of mean diameter during specified years following planting. 49 The optimum brushing radius for all sites was found to be in the range of 1.0-1.25 m: specifically 1.25 m brushing radius for the 101 km site, and the range of brushing radii 1.0-1.25 m for all other sites. Two patterns seemed apparent when considering the vegetation complex and maturity of trees at the sites. First, with relatively heterogeneous and possibly nutrient-poorer sites (i.e.. Little Bobtail Lake and 137 km sites), mean HDRs were higher for treatments/plots where aspen percent cover was higher (i.e., a pattern not evident with the more homogeneous and possibly nutrient-richer 101 km and 1116 km sites). Second, mean %AHDRs were lower for the sites with larger trees (i.e., 137 km and 1116 km sites) than for the sites with smaller trees (i.e.. Little Bobtail Lake site and 101 km sites). Recommendations that follow from the research presented in Chapter 2 are that: (1) retrospective analyses of stem discs be conducted to provide a dynamic view of how HDRs varied in the period prior to installation of the sites (i.e., prior to brushing treatments); (2) volume equations be developed to permit volume to be used in place of diameter as a parallel criterion (or constraint) for the application of the reference HDRs (and to validate the use of diameter in place of volume where volume equations are not available); (3) HDR measurements be taken approximately six years after the initial installations (i.e., 2003 or 2004) to check the relative stability of the HDR levels for which the reference HDRs were determined; we need to know whether HDRs will remain at (or depart from) the levels obtained at the end of the current study, and to have the opportunity to reconsider the presently recommended reference HDRs. 50 Literature cited Burton, P J. 1993. Some limitations inherent to static indices of plant competition. Can. J. For. Res. 23: 2141-2152. Chen, H.Y.H. 1997. Interspecific responses of planted seedlings to light availability in interior British Columbia: survival, growth, allometric patterns, and specific leaf area. Can. J. For. Res. 27: 1383-1393. Davis, I. 1998. Non-crop Vegetation, detrimental or not?: redefining Free Growing. B.C. Ministry of Forests. Victoria, B.C. DeLong, S.C., Tanner, D., and lull, M.J. 1993. A Field Guide for Site Identification and Interpretation for Southwest Portion of Prince George Forest Region. Land Management Handbook 24, Research Branch, Ministry of Forests, Province of British Columbia, Victoria, BC. Ende, C.N. von 1993. Repeated Measures Analysis: Growth and Other Time-Dependent Measures. In Design and Analysis of Ecological Experiments. Edited by S.M. Scheiner, and J. Gurevitch. Chapman and Hall, London, pp. 113-137. Froese, K. 2000. Height to diameter ratios as an indicator of competitive stress in lodgepole pine (Pinus contorta var. latifolia) under four brushing treatments. BSc. (NRM-forestry major) Professional Report. University of Northern British Columbia, Prince George, BC. Henry, H.A.L., and Aarssen, L.W. 1999. The interpretation of stem diameter-height allometry in trees: biomechanical constraints, neighbourhood effects, or biased regressions? Ecology Letters 2 (2), 89-97. Husch, B., Miller, C.I., and Beers, T.W. 1993. Forest Mensuration. Krieger Publishing Co. Malabar, Florida. Jacob, N., and Opio, C. 2003. Correction factors for mid-growing season measurements of height to diameter ratios in young lodgepole pine and white spruce plantations in the Vanderhoof Forest District of British Columbia. Report prepared for West Fraser Mills Ltd. Univ. of Northern British Columbia, Prince George, BC. Johnson, R.A., and Wichem, D.W. 1992. Applied Multivariate Statistical Analysis. Prentice Hall, Englewood Cliffs, NJ. Kozlowski, T.T., and Pallarady, S.G. 1997. Physiology of Woody Plants. Academic Press Inc., Toronto. MacDonald, B., Morris, D M., and Marshall, P.L. 1990. Assessing components of competition indices for young boreal plantations. Can. J. For. Res. 20: 1060-1068. Maclsaac, D.A., and Navratil, S. 1996. Competition dynamics in Juvenile boreal hardwood-conifer mixers. In Silviculture of temperate and broadleaf conifer mixers. Edited by P. Comeau and K.D. Thomas. BC Min. For. Victoria, pp. 23-34. 51 McMinn, R.G., and Hedin, I.P. 1990. Site Preparation: Mechanical and Manual. In Regenerating British Columbia’s Forests. Edited by Lavender, D.P., Parish, R., Johnson, C.M., Montgomery, G., Vyse, A., Willis, R.A., and Winston, D. University of British Columbia Press, Victoria, B.C. pp. 150-163. Mustard, J., and Harper, G. 1998. A Summary of the Available Information on Height to Diameter Ratio. B.C. Ministry of Forests. Victoria, BC. Opio, €., Diest, K. van, and Jacob, N. 2003. Intra-seasonal changes in height to diameter ratios for lodgepole pine in the central interior of British Columbia. West. J. Appl. For. 18(1): 52-59. Opio, C., Jacob, N., and Coopersmith, D. 2000. Height to diameter ratio as a competition index for young conifer plantations in northern British Columbia, Canada. For. Ecol. Manage. 137: 245-252. Ruel, J-C., Messier, C., Doucet, R., Claveau, Y., and Comeau, P. 2000. Morphological indicators of growth response of coniferous advance regeneration to overstorey removal in the boreal forest. For. Chron. 76: 633-642. Sachs, D., and Comeau, P.G. 1991. Determination of nitrogen fixation by sitka alder at two sites in the southern interior of British Columbia. FRDA Project C03, Res. Br., BC Ministry of Forests, Victoria, BC (unpublished report). Sanborn, P., Brockley, R., and Preston, C. 1997. Ecological roles of sitka alder in a young lodgepole pine stand. Forest Research Note #PG-10. BC Ministry of Forests, Prince George, BC. Sit, V. 1995. Analyzing ANOVA Designs. Biometrics Information Handbook No. 5. BC Ministry of Forests Research Program, Victoria, BC. Simard, S., and Heineman, J. 1996. Nine-year response of lodgepole pine and the dry alder complex to chemical and manual release treatments on an ICH m kl site near Kelowna. FRDA Report 259. Can. For. Serv. and BC Ministry of Forests, Victoria, BC. Smith, D.M., Larson, B.C., Kelty, M.J., and Aston, P.M.S. 1997. The Practice of Silviculture: Applied Forest Ecology. John Wiley and Sons, NY. StatSoft Inc. 1999. STATISTICA® Version 5.5. Tulsa, OK. Wagner, R.G., Mohammed, G.H., and Noland, T.L. 1999. Critical period of interspecific competition for northern conifers associated with herbaceous vegetation. Can. J. For. Res. 29: 890-897. Wang, Y., Titus, S.J., LeMay, V.M. 1998. Relationships between tree slenderness coefficients and tree or stand characteristics for major species in boreal mixedwood forests. Can. J. For. Res. 28: 1171-1183. Waring, R.H. 1987. Characteristics of trees predisposed to die. BioScience 37: 569-574. Waring, R.H., and Pitman, G. 1985. Modifying lodgepole pine stands to change susceptibility to mountain pine beetle attack. Ecology 66(3): 889-897. 52 Waring, R.H., and Schlesinger W.H. 1985. Forest Ecosystems: Concepts and Management. Academic Press Inc., Orlando, Florida. Williams, H., Messier, C., Kneeshaw, D.D. 1999. Effects of light availability and sapling size on the growth and crown morphology of understory Douglas-fir and lodgepole pine. Can. J. For. Res. 29: 222-231. Zar, J.H., 1996. Biostatistical Analysis. Prentice-Hall, Inc., Englewood Cliffs, NJ. Zimmerman, M.H., and Brown, C.L. 1971. Trees: Structure and Function. Springer-Verlag New York Inc., New York. 53 CHAPTER 3 RETROSPECTIVE ANALYSIS OF INTER-SEASONAL VARIATIONS IN HEIGHT TO DIAMETER RATIOS IN LODGEPOLE PINE PRIOR TO REMOVAL OF COMPETING VEGETATION Abstract Height to diameter ratio (HDR) is one of the competition indices being considered as a basis for determining when to remove (i.e., bmsh) above ground competing vegetation from young conifer plantations. However, little is known about changes in HDRs after planting of crop trees and prior to undertaking brushing in the plantations. Trends in HDRs between the time of planting of lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) and the start of bmshing were investigated on four study sites in the central interior of British Columbia. The relationship between pre-treatment (before bmshing) HDRs and post-treatment (after bmshing) HDRs was also investigated. The study involved a completely randomized, one-factor experimental design, with replication of measurements over time. Bmshing treatment was the factor. The design consisted of four levels of bmshing (0.0 m or control (no bmshing), 0.75 m, 1.0 m and 1.25 m bmshing radii), replicated three times on each site. HDR measurements were retrospectively obtained for randomly selected crop trees, from the time of planting of crop trees to ± e time of destmctive sampling and laboratory analysis. The results indicated that there were no consistent patterns in HDRs, from the time of planting to the initiation of bmshing treatments. The multivariate analysis of variance (MANOVA) with repeated measures indicated that year of measurement had a significant (p < 0.05) effect on mean HDRs and percent changes in HDR. The results suggest that determining the best time to undertake bmshing in lodgepole pine plantations requires further monitoring of HDR after planting. 54 3.1 Introduction HDR is the net result of the history of a crop tree’s growth over its life (Navratil and Maclsaac 1993,Lanner 1985). Change in HDR between years integrates the tree’s response to environmental factors prevalent in previous years with the tree’s response to environmental factors prevalent in the current year (Mustard and Harper 1998). Measurements of height and diameter of crop trees following the variable removal (i.e., brushing) of above ground competing vegetation may be used to determine HDR s response to competition over time (Opio et al. 2000). Thus, it is thought possible to recommend reference HDRs (i.e., HDR thresholds) for lodgepole pine {Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) on the basis of measurements taken, for example, during a three year post-treatment period (Opio et al. 2000). An assessment of vegetation conditions prevailing at the time of plot installations (Section 2.4 (Chapter 2)) indicated differences in species composition, mean percent cover and height of non­ crop trees, and heterogeneity/homogeneity of competing vegetation. This provided a static assessment of competition. However, the perspective offered by such a static view is limited (Burton 1993). Competition between crop trees and non-crop vegetation is a dynamic process (Burton 1993). Assessing the dynamics of crop tree-non-crop vegetation interactions from the time of plantation establishment is desirable, however, it is frequently not practical or logistically possible. It is possible to recover much of the history of crop tree-non-crop vegetation competition via retrospective analysis of HDRs between the time of planting of the crop trees and the initiation of brushing treatments (Lieffers et al. 1996, Hogg and Schwartz 1999). HDRs may record much of the changing pattern of crop tree-non-crop vegetation interaction. HDR is a potentially useful competition index that could assist forest managers in making decisions about the control of competing vegetation in conifer plantations. However, a lack of understanding of how HDRs vary between the time of planting of crop trees and the time of brushing restricts the application of this vegetation management tool. The purpose of the present study is to 55 investigate the pre-treatment (before brushing) period in the history of the four study sites, and to relate these findings to the post-treatment (after brushing) period. The objectives of Chapter 3 are: (i) to describe trends in HDRs of lodgepole pine between the time of planting of crop trees and the time of brushing of experimental plots, (ii) to examine the relationships between pre-treatment and post-treatment HDRs, and (Hi) to identify the time period when brushing should be undertaken in plantations similar to those in the study. 3.2 Materials and methods 3.2.1 Study sites A full description of the four plantations (sites) is provided in Chapter 2 (Section 2.2.1). Briefly, the sites included portions of the sub-boreal spruce (SBS) dry warm (dw) variant 3, and dry cool (dk) subzones in the Vanderhoof Forest District (lat. N 53"47’ to N 54' 03’, and long. W 123' 32’ to W 124' 51’) (Fig. 2.1 (Chapter 2)). Lodgepole pine were planted on the sites at an approximate range of 1270-1460 stems/ha. At the time of plot installations (1998 and 1999) these plantations ranged between four and ten years of age; they were planted between 1990 and 1995. The sites were selected from areas in which competition from trembling aspen {Populus tremuloides Michaux), sitka alder (Alnus crispa ssp. sinuata (Regel) Hulten), and paper birch (Betula papyrifera Marshall) were severe. The sites ranged in elevation from approximately 755 to 854 m above sea level. Mean annual precipitation across the sites extended from 427 to 648.5 mm. Winter precipitation is relatively low, with winter snowpacks generally less than 2 m in depth (DeLong et al. 1993). The sites were prepared for planting hy windrow burning followed by disc trenching. The crop trees were planted on raised spots created by the trenching, with the majority planted on the middle and top positions (McMinn and Hedin 1990). One site was disc trenched two years prior to planting (i.e.. Little Bobtail Lake site); and all other sites were disc trenched one year prior to planting. 56 3.2.2 Experimental design A full description of the experimental design is provided in Chapter 2 (Section 2.2.2). Briefly, for the purpose of data analysis and comparison, and achieving the smdy objectives, the pre­ treatment experimental design (i.e., four treatments applied to retrospective data) was made the same as the post-treatment design (i.e., four treatments applied to experimental data). The experimental design was a completely randomized one-factor design, with replication of measurements over time. Brushing treatment was the factor (Sit 1995, Zar 1996). The design consisted of four levels of brushing (0.0 m or control (no bmshing), 0.75 m, 1.0 m and 1.25 m bmshing radii) replicated three times on each site (12 plots/site). Illustrations of the plot layouts are found in Appendix A. A total of 12 plots, each 11.28 m in radius (0.04 ha), were randomly located within the 120 m x 90 m (1.08 ha) study site, on the selected stratum of the cut block. Destmctive sampling of randomly chosen 6 crop trees within a plot was undertaken for all sites (72 trees/site) in May 2000. 3.2.3 Measurements Heights and diameters of each crop tree (72 trees/site) were retrospectively obtained for each year between the time (1990-1995) of planting of the trees, and the last year of measurements (1999). The sampled trees were cut down with a hand saw. Stump height and heights of the whorls were measured with a height pole and/or logger’s tape as follows. First, the height of the stump was measured from above the top of the mineral soil (measured on the upper side of the slope). Second, the heights of the whorls were measured at a point immediately beneath the whorl, on the felled tree. On each study site, the year when each whorl was put on the tree was determined by counting down from the top whorl (see an example in Fig. 3.1). The total height of the tree was obtained from measurements taken on the same tree late in the previous year (late-August to September 1999). The heights of what seemed to be intermediate whorls (or inter-nodes) were recorded, for use in later corrections in the dating of the whorls. 57 Stem discs of approximately 2 cm in thickness were cut from the stem with a hand saw. In addition, discs were taken from a point 130 cm above the root collar on trees greater than 130 cm height. At points where there was ambiguity about the age of a particular whorl, an additional disc would be cut at the whorl in question (usually amounting to one or two additional discs per tree). All discs were labeled, placed in plastic freezer bags and stored in a laboratory freezer (maintained at -20°C). The stem discs were prepared for measurement by sanding, and cutting with a dissecting knife. The diameters of the stem discs (more precisely, the cross-sectional distances across circular-shaped to elliptical-shaped discs) were then measured outside the bark (i.e., to the nearest 0.01 cm) using electronic calipers. Two measurements were taken: one transected the disc at the wide diameter, and another transected the disc at the narrow diameter. The two diameter measurements were combined to produce an average diameter and radius. The average radius was used to locate two paths, 90° or more apart, on the disc. The paths extended from the centre of the pith to points on the outside bark that equaled the average radius (Lieffers et al. 1996, Hogg and Schwartz 1999). The stem discs were observed under a dissecting microscope. A lead pencil was used to indicate the latewood part of the tree rings, along the two paths. This procedure facilitated the viewing of the discs using the WinDENDRO™ digital scanner (Regent Instruments 2000). Tree ring widths were measured (i.e., to the nearest 0.01 cm) between the latewood ends of the tree rings, along the two paths. The sum of tree ring widths, measured from the centre of the pith to the outside of the tree ring, indicated the radius for that year. The average of two radii for a year was used to calculate inside bark diameter for that year. Thus, inside bark diameters were calculated for each year between the year that the tree was planted and the last year measured by the destructive sampling. 58 1998 (top whorl) Top whorl disc 1997 1996 1995 1994 *■ Root collar disc Diameter^ j ■■ - • ................... Diameter,; ........... Diaiiieterq* Diameter, ---Diameter, Diameter» Fig. 3.1. Schematic diagram of crop tree as measured for stem analysis at CanFor-BednestiiLittle Bobtail Lake site. Total heights were estimated by heights of whorls: 1998 indicates whorl to which height was measured for total height in 1998, 1997 indicates whorl to which height was measured for total height in 1997, and so forth. 59 3.2.4 Analysis For each crop tree, HDR based on diameter inside bark (HDR ib) was calculated for all years except 1999 as follows: HDRib = Height„horil D iam eterib where Height„hod = total height (cm) as determined by the height of the whorl, and D ia m ete r ib = diameter (cm) inside bark at root collar (Fig. 3.1). HDRib in 1999 was calculated from total height as measured on the hve standing tree since the height of the whorl was not available for 1999. A regression on heights of whorls over total heights for 1998 was run. The regression results indicated a possible inflation of the mean H D R jb for 1998 of approximately 0.7 (i.e., based on mean H D R ib), which seemed negligible. Thus, no correction was applied to total heights in 1999, and they were treated as being equivalent to 1999 heights of whorls. The diameter for each year between the time of planting of the tree and its destructive sampling date was calculated from tree ring widths: DiameterjBi = 2 * E RWy where D iam eter ibi = diameter inside bark along path i (1 or 2), and RWy = ring width along path i (1 or 2), for year j (e.g., 1994-1999 for Little Bobtail Lake site). Diametermi and Diametermz were averaged to obtain the diameter inside bark (Diameter^) used in the above calculation of HDRibs: DiameterjB = (Diam eter, bi + D iam eter, bz)! 2 The mean HDRib of the 6 crop trees selected for measurement in each plot (i.e., plot mean HDRib based on height^hori and diameter^ measurements) was calculated, and measurement time and site were recorded for use in the analysis. Normality of H D R ib data was tested using Kolmogorov-Smirnov and Lilliesfors tests (StatSoft 1999); and homoscedasticity of HDRib data was tested using Brown Forsythe and Levene tests (StatSoft 1999). Normality and homoscedasticity were generally achieved. The concern for the small sample size (6 trees/plot) in the study suggested use of a non-parametric procedure. However, the existing non-parametric procedures (e.g., Freidman’s method and Scheier-Ray-Hare test) were not 60 applicable to the experimental design (D. Ayers, pers. comm. 2002; Sokal and Rohlf 1995, pp. 440447). Thus, HDRib data were rank-transformed. For each site, a multivariate analysis of variance (MANOVA) with repeated measures was conducted: first, based on non-transformed HDR ib; and second, based on rank-transformed HDR ib data using Statistica® (StatSoft 1999). Conover (1980, p. 337) advises that the parametric analysis will be valid if the ANOVA on rank-transformed data produces “nearly identical results” as the ANOVA on non-transformed data. A repeated measures MANOVA was performed on each site using Statistica® to test whether the mean HDRs of crop trees were significantly (p < 0.05) different among treatments, dates of measurement, and interaction between these two factors (StatSoft 1999). The MANOVA model used was: HDRiBijk = M + treatmenti + datej + (treatment *date) y + Syn where HDRmjk = plot mean H D R ib, R = grand mean H D R ib, treatment, = bmshing radius (0.0 m or control, 0.75 m, 1.0 m, 1.25 m), datej = year of measurement (each year from 1994 to 1999 for Little Bobtail Lake site; each year from 1995 to 1999 for 101 km site; and 1991, 1993, 1995, 1997, and 1999 for 137 km and 1116 km sites), (treatment*date)y = interaction between treatment and date of measurement, and syk= experimental error (Johnson and Wichem 1992, p. 263-4). Factors such as stock type, biogeoclimatic classification, and planting density were constant for a study site, thus they were not included in the analysis. Opio et al. (2000) had previously determined that planting position had a significant (p < 0.05) effect on mean HDRs, therefore this factor was not included in the present study. Comparisons between study sites were conducted on the basis of vegetation and other site characteristics. MANOVA was used instead of the usual analysis of variance (ANOVA) because HDRs of repeated measurements were highly correlated. MANOVA is commonly applied to repeated measures-data with several correlated dependent variables (von Ende 1993, p. 117-8). The repeated measures MANOVA provided an overall assessment of the mean HDR ibs of crop trees; and whether 61 the factors, treatment and date, had a significant effect on the mean HDR ibS. If the MANOVA determined that one or more of the factors had a significant overall effect, then investigating more specific effects by means of Tukey HSD post hoc procedures was justified. The MANOVA results based on rank-transformed data were not substantially different from that based on non-transformed data; thus, the analysis was based on the non-transformed H D R ib data. There were two reasons for this choice. First, the MANOVAs based on rank-transformed data and the MANOVAs based on non-transformed data provided very similar results. Second, it was felt that the determination of percent change in HDR ib would provide very useful results, which depended on the use of non-transformed data. Relative growth rate and simple ratios are among the techniques used to compare growth characteristics across years (Kozlowski and Pallardy 1997). Percent change in HDRœ combines both these approaches. Although other approaches were possible (Opio et al. 2000, Opio et al. 2003), it was advantageous to use % AHDRibS, and normalize HDR^s to the year of the initial treatments. This approach seemed readily understandable, and facilitated comparisons both within the retrospective data analysis and with the experimental data analysis. Percent change in HDR ib was calculated for each tree as follows: * 100 where %AHDRiBi= percent change in H D R ib between 1998 (year when treatments begun, 1999 for 1116 km site) and 1994, 1995, 1996, 1997, 1998, and 1999 (i.e.. Little Bobtail Lake site); HDRmi = H D R ib in 1994, 1995, 1996, 1997, 1998, and 1999 (primarily HDRibS prior to treatments (1994 to 1997), but also HDR^s following treatments (1998 and 1999)). The mean % AHDRib for each plot within a site was calculated for use in the analysis. The Tukey HSD post hoc test was performed on each site to test whether the mean HDR^s were significantly (p < 0.05) different between specific treatments (0.0 m or control, 0.75 m, 1.0 m, and 1.25 m brushing radii), and between specific years of measurement (each year from 1994 to 1999 for Little Bobtail Lake site; each year from 1995 to 1999 for 101 km site; 1991, 1993, 1995, 1997, 62 1999 for 137 km and 1116 km sites). The Tukey HSD post hoc test was also performed on each site to test whether the mean percent changes in H D R jb (% A H D R ibs) were significantly (p < 0.05) different between specific years of measurement (each year from 1994 to 1999 (i.e., Little Bobtail Lake site)). The MANOVAs and Tukey HSD post hoc tests for the Little Bobtail Lake and 101 km sites were run using data from all the years for which the data were available (each year from 1994 to 1999 for Little Bobtail Lake site, and each year from 1995 to 1999 for 101 km site). Due to the need to reduce the number of dependent variables (von Ende 1993, p. 117-8), the 137 km and 1116 km sites were run using every second year (1991, 1993, 1995, 1997, and 1999) of H D R jb measurements. Determining when (years after planting) to undertake brushing in lodgepole pine plantations was based on examining %AHDRjbS. Where the pattern of declining % AHDRjbS in the first two years after planting was followed by a significant (p < 0.05) rise thereafter (i.e., up to the initiation of treatments), then brushing was recommended. 3,3 Results The MANOVAs on mean HDRjbS (Table 3.1) indicated that brushing treatment did not have a significant overall effect for 101 km and 1116 km sites. However, treatment was only marginally non-significant (p < 0.10 and p < 0.08, respectively) for Little Bobtail Lake and 137 km sites. Year of measurement had a significant overall effect for all study sites. Only the 101 km site had a significant interaction between brushing and year. The MANOVAs on percent changes in H D R jb (Table 3.1) indicated approximately the same outcome as those on mean HDRjbS. Bmshing treatment did not have a significant overall effect for 101 km and 137 km sites. However, treatment was only marginally non-significant (p < 0.10 and p < 0.06, respectively) for Little Bobtail Lake and 1116 km sites. Year of measurement had a significant overall effect for all sites. Little Bobtail Lake arid 101 km sites had a significant interaction between bmshing and year. 63 Theoretically, for a given treatment, %AHDRmS at the time of planting are usually positive numbers (due to relatively higher HDRs in the first year after planting), %AHDRib9 8 S always equal zero, and %AHDRib9 9 S are typically negative numbers (in response to brushing treatments). % A H D R ib S typically decrease from a maximum value in the first year after planting (e.g., 1994) to zero in the first year of treatments (e.g., 1998). % A H D R ibS may be negative or positive in the years between planting and 1998. In any given year, %AHDRm for the control and %AHDRm for the treatments may fall in any order. In comparisons between treatments in a given year, there were no consistent patterns in %AHDRibS (Figs. 3.2-5) for any of the study sites. The Tukey HSD post hoc tests confirmed these observations, generally indicating no significant differences in HDRibS or % AHDRibS between treatments, for any of the sites. However, at the Little Bobtail Lake site, HDR jbS for the 0.0 m and 1.25 m treatments in 1999 were significantly (p < O.OI) different. Furthermore, % AHDRibS for the 1.0 m and 1.25 m treatments in 1995 were significantly different (p < 0.01). Similarly, at the 101 km site, HDRibS for the 0.0 m and 1.25 m treatments in 1999 were significantly (p = 0.02) different, and %AHDRibS for the 1.0 m and 1.25 m treatments in 1995 were significantly (p = 0.03) different. At the 137 km site, for example, HDRiB(o.om, o.75m) was significantly (p = 0.02) greater than HDRiB(l,Om, 1.25m) (1-6., HDRœO.Om ~ HDRffi0.75m> H DRiBl.O m ~ H D R m 1.25m) in 1991. 64 Table 3.1. MANOVAs with repeated measures showing the factors affecting mean HDRms" and percent changes in HDRm* between 1991 and 1999 for all study sites Study site Effect Mean HDR jb F Value (P r> F ) df Effect Percent change in HDRjb F Value (Pr>F) CanFor-Bednestii Little Bobtail Lake 2.88 0.10 31.22 1.40 0.00 0.19 0.61 0.63 13.39 0.00 0.01 Brushing Year Brushing* Year 3 5 15 Brushing Year Brushing* Year 3 4 12 Brushing Year Brushing* Year 3 4 12 3.22 45.21 Brushing Year 3 4 2.42 Brushing*Year 12 1.11 2.88 32.90 1.89 0.10 0.00 0.05 1.37 17.27 0.32 Fraser Lake 101 km 137 km 1116 km 2.79 1.43 3.13 0.08 0.00 0.20 0.14 0.03 0.39 2.83 0.00 0.01 1.33 43.20 0.98 0.33 0.00 3.75 3.57 0.06 0.02 1.19 0.33 0.49 “HDRibS were calculated from heights of whorls and inside bark diameters (n=72/site). ^Percent changes in HDRib were calculated with reference to HDRibS in 1998 for Little Bobtail Lake, 101 km, and 137 km sites; and HDR^s in 1999 for 1116 km site. Mean HDRs and mean percent changes in HDR were calculated for each plot (n=12/site). Note; MANOVAs were run on 1994-1999 data for Little Bobtail Lake site, and 1995-1999 data for 101 km site {n=12/site)\ and 1991, 1993, 1995, 1997, and 1999 data for 137 km and 1116 km sites {n=12/site). 3.3.1 CanFor-Bednestii-Little Bobtail Lake site Comparisons within treatments indicated that HDRibS (Table 3.2a) declined steadily from the time of planting (1994) to the initiation of treatments (1998). HDRib9 4 was generally > HDRgw. The Tukey HSD post hoc test confirmed this pattern (p = 0.01) in comparisons between 1994 and 1998, and 1994 and 1999. 65 & 0.00 m c 0.75 m 1.00 m ! 1.25 m -10 -20 1994 1995 1996 1997 1998 1999 Year Fig. 3.2. Percent changes in H D R ib between 1994 and 1999 for CanFor-Bednestii-Little Bobtail Lake site. HDRibS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. Percent changes in FIDRib were calculated with reference to HDRibS in 1998 {n=72/site), and mean percent change in H D R jb was calculated for each plot {n=12/site). In order to improve clarity, error bars are not presented in the figure. Similarly, comparisons within treatments with respect to percent changes in H D R ib (Fig. 3.2) indicated that %AHDRibS declined steadily from the time of planting (1994) to the initiation of treatments (1998). The Tukey HSD post hoc test confirmed this pattern (p = 0.02) in comparisons between 1994 and 1997, 1994 and 1998, and 1994 and 1999 (all treatments except 1.0 m treatment). Prior to treatments (1995), %AHDRjBi,2 5 mwas significantly (p = 0.01) greater than %AHDRiBi,om; and following treatments (1999), no significant differences were observed between treatments. 66 3,3.2 Fraser Lake-101 km site Comparisons within treatments indicated that HDRgS for the 0.75 m and 1.0 m treatments (Table 3.2a) declined dramatically in the two years following planting (1995-1996) but then rose (S 0.00 m 0.75 m 1.00 m 1.25 m I 1995 1996 1997 1998 1999 Year Fig. 3.3. Percent changes in HDRœ between 1995 and 1999 for Fraser Lake-101 km site. HDRibS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. Percent changes in HDRib were calculated with reference to HDRibS in 1998 (n=72/site), and mean percent change in HDR® was calculated for each plot {n=12/site). In order to improve clarity, error bars are not presented in the figure. steadily thereafter up to the initiation of treatments (1998). The Tukey HSD post hoc test confirmed the first part of this pattern; HDRiBo.rsm and HDRiBi.om decreased significantly (both p < 0.01) from 1995 to 1996, but failed to rise significantly thereafter (i.e., 1996-1998). With respect to percent changes in H D R ib within treatments (Fig. 3.3), %AHDRibS for the 0.75 m and 1.0 m treatments declined dramatically in the two years following planting (1995-1996), but then rose steadily thereafter up to the initiation of treatments (1998). The Tukey HSD post hoc 67 test confirmed the first part of this pattern: %AHDRiBo.7 5 mand %AHDRiBi.om decreased significantly (both p < 0.01) from 1995 to 1996, but failed to rise significantly thereafter (i.e., 1996-1998). Prior to treatments (1995), %AHDRiBi.om was significantly (p = 0.03) greater than %AHDRib 1 .25 m; and following treatments (1999), no significant differences existed between treatments. Table 3.2a. Mean HDR#s* and standard errors of mean (SEM)* from 1994 to 1999 for CanForBednestii-Little Bobtail Lake and Fraser Lake-101 km sites Year of measurement Study site Brushing radius (m) 1994 1995 1996 1997 91.9 (5.9) 81.7 (5.6) 80.3 (5.6) 80.7 (5.6) 7 9 j(5 j) 74jl(54) 72.8 (5.3) 78,1 (5.5) 78.4 (5.6) 66.6 (5.1) 69.7 (5.2) 69.0(5.1) 75.1 (5.4) n.m. n.m. n.m. n.m. 6L2(5H) 68J(5H) 71.6 (5.3) 63.1 (5.0) 63.9 (5.0) 52.5 (4.5) 55.0 (4.6) 58.6 (4.8) 66.5 (5.1) 59.9 (4.8) 62.1 (4.9) 66.8 (5.1) 1998 1999 CanFor-Bednestii Little Bobtail Lake“ (n=72/site) 0 .0 0 0.75 1 .0 0 L25 72.8 (5.3) 71.6 (5.3) 67.9 (5.1) 63.8 (5.0) 60.6 (4.8) 72.8 (5.3) 70.0 (5.2) 64.5 (5.0) 66.3(5.1) 5&9(48) 53.3 (4.5) Fraser Lake 1 0 1 km'' (n=72/site) 0.00 0.75 1.00 1.25 67.5 (5.1) 64.5 (5.0) 60.7 (4.8) 62.6 (4.9) 56.5 (4.7) 66.5 (5.1) 53.7 (4.5) 67.5 (5.1) “HDR ibS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. *SEM are presented in brackets following mean HDR^s. “Trees were planted in 1994. “^Trees were planted in 1995. Note: n.m. indicates no measurements were taken. 3.3.3 Fraser Lake-137 km site Comparisons within treatments indicated that HDRjbS for all treatments (Table 3.2b) declined dramatically in the four years following planting (1991-1993), and remained relatively constant thereafter. The Tukey HSD post hoc test confirmed the pattern HDRœgi > HDR 1B9 2 >HDRib9 3 68 (p < 0.04) for all treatments; however, no significant differences were found between HDRœga and HDRibS in later years. 0.00 m 0.75 m -6— 1.00 m 1.25 m -10 -20 1991 1992 1993 1994 1995 1996 1997 1998 1999 Year Fig. 3.4. Percent changes in HDRib between 1991 and 1999 for Fraser Lake-137 km site. HDRgS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. Percent changes in HDRib were calculated with reference to HDRibS in 1998 {n.-72/site), and mean percent change in HDRm was calculated for each plot (n=12/site). Percent changes in HDRib for 1990 (the year of planting of trees) were not presented because MANOVAs were run on 1991, 1993, 1995, 1997, and 1999 data. In order to improve clarity, error bars are not presented in the figure. With respect to percent changes in HDRm within treatments (Fig. 3.4), % AHDRibS declined dramatically in the two to four years following planting (1991 to 1993 or 1994), and remained relatively constant thereafter. The pattern % AHDRib9i > % AHDRib92> % A H D R ib93 > % AHDRib94 appeared to hold. The Tukey HSD post hoc test confirmed this pattern (p < 0.03) in comparisons between 1991 and 1993, 1991 and 1995, and 1991 for all treatments. No significant differences were found between % AH DRib93, % AHDRib95, % AHDRib97, and % AHDRib99- 69 It was discovered that, prior to treatments (1991 and 1993, for example) and following treatments (1999), no significant differences existed between % AHDRibS for any treatments. 3.3.4 Fraser Lake-1116 km site Comparisons within treatments appeared to indicate that HDR ibS for the 0.0 m, 0.75 m, and 1.0 m treatments (Table 3.2b) declined dramatically in the four years following planting (1991-1993), and remained relatively constant thereafter. HDRibS for the 1.25 m treatment remained relatively stable throughout the measurement period. The Tukey HSD post hoc test failed to confirm this pattern, indicating no significant differences between HDR ibS in any of the years. With respect to percent changes in HDR ib within treatments (Fig. 3.5), indications seemed to be that %AHDRibS for the 0.0 m, 0.75 m, and 1.0 m treatments declined dramatically in the four years following planting (1991-1993). %AHDRibS for the 0.0 m treatment appeared to rise appreciably thereafter (1993-1995), whereas %AHDRœS for the 0.75 m and 1.0 m treatments remained relatively constant thereafter. %AHDRibS for the 1.25 m treatment remained relatively stable throughout the measurement period. The Tukey HSD post hoc test failed to confirm this pattern, indicating no significant differences between %AHDRibS in any of the years. It was found that prior to treatments (1991 and 1993, for example), no significant differences existed between % A H D R ibs for any treatments. 70 — 0.00 m - a - 0.75 m 1.00 m J= -X— 1.25 m -10 -20 1991 1992 1993 1994 1995 1996 1997 1998 1999 Year Fig. 3.5. Percent changes in HDRib between 1991 and 1999 for Fraser Lake-1116 km site. HDRibS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. Percent changes in HDRib were calculated with reference to HDRœS in 1999 (n=72/site), and mean percent change in HDRm was calculated for each plot {n=12/site). Percent changes in HDRm for 1990 (year of planting of trees) were not presented because MANOVAs were run on 1991, 1993, 1995, 1997, and 1999 data. In order to improve clarity, error bars are not presented in the figure. 71 Table 3.2b. Mean HDR ibs" and standard errors of mean (SEM)* from 1991 to 1999 for Fraser Lake-137 km and -1116 km sites Year of measurement Study site Brushing radius (m) 1991 1992 1993 1994 1995 Fraser Lake 137 km" 0.00 {n—72/site) 0J5 1.00 1J3 1116 km" (n=72/site) 0.00 0J5 1.00 125 84.8 (5.7) 85.7 (5.8) 70.6 (5.2) 70.9 (5.2) 742(52) 68.5 (5.1) 79.4 (5.5) 69.6 (5.2) 64.1 (5.0) 56.8 (4.7) 66.0 (5.0) 58.1 (4.7) 46.8 (4.2) 61.3 (4.9) 59.4 (4.8) 51.3 (4.4) 41.7 (4.0) 64.4 (5.0) 67.2 (5.1) 62.3 (4.9) 61.7 (4.9) 52.6 (4.5) 55.1 (4.6) 56.4 (4.7) 58.2 (4.7) 40.1 (3.9) 42.6(4.1) 45.7 (4.2) 53.5 (4.6) 51.2 (4.4) 48.2 (4.3) 49.4 (4.4) 56.2 (4.7) 51.9 (4.5) 50.8 (4.4) 52.1 (4.5) 49.6 (4.4) 50.7 (4.4) 50.1 (4.4) 51.6 (4.5) Year of measurement Study site Bmshing radius (m) 137 km" 0.00 (n=72/site) 0.75 1.00 125 1116 km" {n-72/site) 0.00 0.75 1.00 1.25 1996 66.1 (5.0) 60.4 (4.8) 53.7 (4.5) 56.4 (4.7) 1997 1998 1999 66.8 (5.1) 70.2 (5.2) 72.6 (5.3) 60.2 (4.8) 62.1 (4.9) 61.2 (4.9) 54.1 (4.6) 54.3 (4.6) 53.7 (4.6) 55.8 (4.6) 59.2 (4.8) 58.0 (4.7) 48.5 (4.3) 49.4 (4.4) 51.1 (4.4) 49.8 (4.4) 50.0 (4.4) 49.8 (4.4) 51.2 (4.4) 49.9 (4.4) 51.1 (4.4) 49.8 (4.4) 50.0 (4.4) 47.6 (4.3) 47.6 (4.3) 46.2 (4.2) 46.4 (4.2) 46.2 (4.2) “HDRffiS were calculated from heights of whorls and inside bark diameters of destructively sampled trees. *SEM are presented in brackets following mean HDRœS. ‘^Trees were planted in 1990. 72 3.4 Discussion The retrospective analyses provided a dynamic view of HDRgS from the time of planting to the initiation of brushing treatments. However, they did not establish precisely what vegetation conditions might have prompted the resulting patterns of changing HDR ibS. The history of site preparation on the four study sites probably contributed to the development of the vegetation complexes on these sites. The Little Bobtail Lake site was planted two years after being disc trenched, whereas the 101 km, 137 km, and 1116 km sites were planted one year after being disc trenched. A regeneration delay may have been an important factor at the Little Bobtail Lake site (Wood and von Althen 1993). Competition from herbaceous and deciduous vegetation may have unfolded at the study sites as follows. Herbaceous plants and grasses probably colonized these sites in the first and second years after disc trenching. Shrubs and deciduous trees are likely to have become competitive factors in the third and fourth years after disc trenching. This pattern resembles that described by Wagner et al. (1999) in their definition of the critical period of inter-specific competition. In the first year after planting the seedlings at the Little Bobtail Lake site may have already felt competition from herbaceous plant and grass competitors, and perhaps also from shrubs and deciduous trees. The seedlings at the 101 km, 137 km, and 1116 km sites would likely have had a year without any significant competition, whereas those at the Little Bobtail Lake site may have experienced significant competition in their first year after planting. Deciduous trees are likely to have become the primary sources of competition in the third and fourth years after planting on all sites (Wood and von Althen 1993, Wagner et al. 1999). At Little Bobtail Lake and 137 km sites, we were able to obtain regressions that showed HDRs increasing with increasing aspen percent cover (Section 2.4 (Chapter 2)). Similar relationships between HDRs and aspen cover were not obtained from the more homogeneous 101 km and 1116 km sites. In Section 2.4 (Chapter 2) it was suggested that a more homogeneous site (i.e., greater 73 uniformity in aspen percent cover) may be nutrient-richer than a more heterogeneous site (i.e., lesser uniformity in aspen percent cover). Mean HDRœS of the seedlings were relatively high one growing season after planting (i.e., 80.3-91.9 for Little Bobtail Lake site, 63.1-71.6 for 101 km site, 70.6-85.7 for 137 km site, and 46.861.3 for 1116 km site) (Tables 3.2a-b). This initial rise in HDRs after planting may indicate a nutrient surplus, which may result in higher HDRs (Waring 1987). It is difficult to compare these values with the mean HDRs (i.e., outside bark) of the initial planting stock (i.e., 44.0 for Little Bobtail Lake site, 48.6 for 101 km site, 83.3 for 137 km site, and 44.4 for 1116 km site) (Section 2.2.1 (Chapter 2)). However, for one to two years after planting, HDR jbS generally descended (Tables 3.2a-b). This general decline HDRs may indicate a nutrient deficiency, where more resources are allocated to root growth (Mustard and Harper 1998). Thereafter, a variety of patterns of HDRms were observed. In running the MANOVAs on mean HDR^s for the analyses on the retrospective data (i.e., 6 trees/plot), mean HDRs were grouped the same as they were in running the MANOVAs for the analyses on experimental data (approximately 50 tree/plot): by brushing treatment. However, treatments were only begun in the year prior (e.g., 1998) to the last year (e.g., 1999) for which the MANOVAs for the retrospective analyses were run. Therefore, the retrospective data sets were run with only one year’s (e.g., 1999) data where a treatment effect was possible; and the overall effect introduced by treatments was negligible. However, year of measurement continued to be a significant overall factor because HDRigS changed over time irrespective of the presence or absence of treatments. The presence of significant interaction at the 101 km site (not evident at the other sites) may be due to the influence of nutiient-richness or other factor not tested in the MANOVA. No significant differences in % AHDRibS were found in the first, third, or fifth years after planting; or in the second or third years after treatments at any of the study sites. At best, significant differences were found in the % A H D R ibs in the pre-treatment period (in comparisons between two treatments only), or significant differences were found in the %AHDRibS in the post-treatment period (data not presented), but never both. These results made it difficult to draw statistical comparisons. 74 In graphical comparisons between the ordering of % AHDRibS in the first, third, or fifth years after planting (prior to treatments) and the ordering of %AHDRibS in the second (1999) or third (2000) years after treatments, however, two consistencies emerged across all the sites. First, prior to treatments (first, third, or fifth years after planting), % AHDRibS fell in any order (e.g., % A H D R m 1.25m > % A H D R ffio.75m > % A H D R iB o.om > % A H D R [B i.om )- Second, following treatments (1999 or 2 0 0 0 ), %AHDRjbS always conformed to the pattern % A H D R iBo.om > % A H D R iB 075m > % A H D R iB i,om > %AHDRiBi,2 5 m- These observations indicate that the post-treatment pattern of % A H D R ibs was not dependent upon the pre-treatment pattern of %AHDRibS. In other words, % AHDRibS of trees prior to treatments did not affect the response of the trees to brushing treatments. This result indicated that brushing treatments are a powerful tool for manipulating H D R s . The pattern of declining HDRibS and % AHDRibS in the first two years after planting followed by a significant rise thereafter (i.e., up to the initiation of treatments), would justify brushing interventions in the third or fourth year after planting. There is one example (i.e., 101 km site) that nearly meets the requirement for brushing intervention. However, the apparent rise in HDRs two years after planting is not significant. Wagner et al. (1999) were able to make general recommendations for when during the first five years after planting to brush plantations of five conifer species. In Section 2.4 (Chapter 2), it was possible to broadly recommend that lodgepole pine plantations be brushed < 4-7 years after planting. However, based on an inspection of HDRs between the time of planting of crop trees and brushing of experimental plots, it seems difficult to provide a more specific recommendation for when (years after planting) to brush these plantations. 3.5 Conclusions and recommendations Retrospective analyses of HDRs augmented the overall HDR research project. A variety of pre-treatment patterns of HDRibS were found. These were described in terms of four patterns. H D R ibs; (i) declined steadily from the time of planting (e.g., 1994) to the initiation of treatments (1998); (ii) declined dramatically in the two years following planting (e.g., 1995-1996), but then rose 75 steadily to the initiation of treatments (1998); (Hi) declined dramatically in the four years following planting (e.g., 1991-1993), and remained relatively constant thereafter; or (zV) remained relatively constant over the entire measurement period (1991-1999). Each site was approximately described by one of these patterns. Thus, rather than finding one pattern that repeated itself between study sites, we found a great deal of dissimilarity between the patterns. No relationship was found between the pre-treatment and post-treatment patterns of HDRibS. Whatever the pre-treatment pattern of HDRibS, the post-treatment pattern was consistently %AHDRiBo.om > %AHDRiBo.7 5 m> %AHDRiBi.om > % A H D R ib 1.25m- The post-treatment pattern of HDRibS was not dependent upon the pre-treatment pattern of HDRm s. The HDRmS of trees prior to treatments did not seem to affect the response of the trees to brushing. The contrary was found to hold: bmshing treatments were found to be a powerful tool for manipulating HDRs. No pre-treatment pattern of % AHDRibS emerged as a general mle. The second of the patterns described, where % AHDRibS declined dramatically in the two years following planting (e.g., 19951996), but then rose steadily to the initiation of treatments (1998), was only inadequately the pattern for one site. Thus, based on the results of this chapter, no specific recommendation was made as to the appropriate time (years after planting) to bmsh lodgepole pine plantations. In Chapter 2, it was recommended that forest managers bmsh lodgepole pine plantations < 4-7 years after planting. This chapter does not provide more specific recommendations. It is not known what specific factors produced the four patterns of pre-treatment HDRmS. Further research should aim at tracking changes in the vegetation complex along with changes in HDRibS. Further monitoring of changes in HDRs from the time of planting is required to determine the best time to bmsh lodgepole pine plantations. 76 Literature cited Burton, P J. 1993. Some limitations inherent to static indices of plant competition. Can. J. For. Res. 23: 2141-2152. Conover, W.J. 1980. Practical Nonparametric Statistics 2"^ ed. John Wiley and Sons, Inc., New York. DeLong, S.C., Tanner, D., and lull, M.J. 1993. A Field Guide for Site Identification and Interpretation for Southwest Portion of Prince George Forest Region. Land Management Handbook 24, Research Branch, Ministry of Forests, Province of British Columbia, Victoria, BC. Ende, C.N. von 1993. Repeated-measures analysis: growth and other time-dependent measures. In Design and analysis of ecological experiments. Edited by Samuel M. Scheiner and Jessica Gurevitch. Chapman & Hall, New York. pp. 113-137. Hogg, E.H., Schwarz, A.G. 1999. Tree-ring analysis of declining aspen stands in west-central Saskatchewan. Nat. Resour. Can., Can. For. Serv., North. For. Cent., Edmonton, Alberta. Inf. Rep. NOR-X-359. Johnson, R.A., and Wichem, D.W. 1992. Applied Multivariate Statistical Analysis. Prentice Hall, Englewood Cliffs, NJ. Kozlowski, T.T., and Pallarady, S.G. 1997. Physiology of Woody Plants. Academic Press Inc., Toronto. Lanner, R.M. 1985. On the insensitivity of height growth to spacing. For. Ecol. Manage. 13: 143-148. Lieffers, V.J., Stadt, K.J., and Navratil, S. 1996. Age structure and growth of understory white spruce under aspen. Can. J. For. Res. 26: 1002-1007. McMinn, R.G., and Hedin, I P. 1990. Site Preparation: Mechanical and Manual. In Regenerating British Columbia’s Forests. Edited by Lavender, D.P., Parish, R., Johnson, C.M., Montgomery, G., Vyse, A., Willis, R.A., and Winston, D. University of British Columbia Press, Victoria, B.C. pp. 150-163. Mustard, J., and Harper, G. 1998. A summary of the available information on height to diameter ratio. BC Min. For. Victoria, BC. Navratil, S., and Maclsaac, D A. 1993. Competition index for juvenile mixed wood stands of lodgepole pine and aspen in West-Central Alberta. Forestry Chronicle, Northwest Region, Forest Management Note 5. Opio, C., Diest, K. van, and Jacob, N. 2003. Intra-seasonal changes in height to diameter ratios for lodgepole pine in the central interior of British Columbia, Canada. West. J. Appl. For. 18 (1): 52 ^ 9 . Opio, C., Jacob, N., and Coopersmith, D. 2000. Height to diameter ratio as a competition index for young conifer plantations in northern British Columbia, Canada. For. Ecol. Manage. 137: 245-252. 77 Regent Instruments Inc. 2000. WinDendro™ Version 6.5. Quebec, Que. Wagner, R.G., Mohammed, G.H., and Noland, T.L. 1999. Critical period of interspecific competition for northern conifers associated with herbaceous vegetation. Can. J. For. Res. 29: 890-897. Waring, R.H. 1987. Characteristics of trees predisposed to die. BioScience 37, 569-574. Wood, I.E., and Althen, F.W. von 1993. Establishment of white spruce and black spruce in boreal Ontario: effects of chemical site-preparation and post-planting weed control. For. Chron. 69: 554-560. Sit, V. 1995. Analyzing ANOVA Designs. Biometrics Information Handbook No. 5. BC Ministry of Forests Research Program, Victoria, BC. Sokal, R.R., and Rohlf, F.J. 1995. Biometry 3rd ed. W.H. Freeman and Co., New York. StatSoft Inc. 1999. STATISTICA® Version 5.5. Tulsa, OK. Zar, J.H. 1996. Biostatistical Analysis. Prentice-Hall, Inc., Englewood Cliffs, NJ. 78 CHAPTER 4 VARIATIONS IN STEM VOLUME OF LODGEPOLE PINE FOLLOWING VARIABLE REMOVAL OF COMPETING VEGETATION Abstract Height to diameter ratio (HDR) is one of the competition indices being considered as a criterion for deciding when to remove (i.e., brush) above ground competing vegetation from young conifer plantations. Reference HDRs (thresholds) have often been defined in terms of HDR itself, or HDRs that prevailed where non-crop vegetation was deemed acceptable. However, this method for recommending reference HDRs seems insufficient. Reference HDRs need to be determined with respect to an independent criterion. Thus in the present study, ranges of mean stem volume (i.e., obtained in specified post-treatment years) are determined, within which reference HDRs are meant to be applied. Trends in stem volume and HDR of lodgepole pine (Pinus conforta Dougl. ex Loud, var. latifolia Engelm.) crop trees under four treatments (0.0 m or control (no brushing), 0.75 m, 1.0 m and 1.25 m brushing radii), on each site, were monitored over a period of three years in Bednestii Lake and Fraser Lake areas of central British Columbia (BC). It was hypothesized that in any one year, the increment in stem volume of trees brushed to a wider brushing radius would be greater than the increment in stem volume of trees brushed to a narrower brushing radius. Regression models of stem volume were developed from height and diameter measurements retrospectively obtained from 72 randomly selected crop trees, for the year (1999) preceding the destructive sampling and laboratory analysis of stem discs (May 2000) for each site. A model (i.e., one for each site) was applied to all trees on the site for the three year post-treatment period of the study (1998-2000 or 1999-2001). The results indicated that stem volumes were significantly (p < 0.05) affected by brushing treatments two years after the initial brushing treatments (2000) with volume increment increasing in proportion to the intensity of brushing. Reference HDRs (i.e., ranges of HDRs) 40-49, 40-51, 45-54, and 38-47 were recommended within ranges of deciduous percent cover, and mean stem volume for specified years for the four study sites. 79 4.1 Introduction Stem volume and biomass accumulation are measures of total growth; the one measure, a primary concern to forest managers, is closely related to the other, a primary concern to forest biologists (Kimmins 1997). Indices of growth such as the stem volume index (i.e., stem diameter^ x height) (Wagner et al. 1999), and basal area (Husch et al. 1993) are valued by forest managers because they may be used to approximate stem volume. Reference HDRs (i.e., HDR thresholds) have generally been defined in terms of HDR itself, or HDRs that prevailed where levels of non-crop vegetation were deemed acceptable. In Chapter 2, locally calibrated HDR thresholds were determined with reference to percent change in HDR (i.e., %AHDR defined with respect to the time of site installation). There, a suitable reference HDR was determined as being the point at which HDRs were stabilizing (i.e., %AHDRs from one year to the next were becoming negligible), and a base level HDR was being achieved. However, this method for recommending reference HDRs needed to incorporate an independent criterion or other factor(s) to be measured in conjunction with HDR. Possible criteria to be used in conjunction with HDR include stem volume, wood quality, and other attributes of the tree or site. Stem volume is of prime consideration because it is readily quantifiable, and an accepted measure of productivity in the early stages of stand development (Husch et al. 1993). However, HDRs are poorly correlated with stem volume, thus it is unreasonable to assume that a functional relationship can be established between reference HDRs and stem volume. A reasonable alternative is to use stem volume as a supportive tool, in conjunction with HDR thresholds. This requires that a reference HDR be determined for a specific range of stem volume. The present chapter builds on the procedure for defining reference HDRs used in Chapter 2. Stem diameter has been found to be highly correlated with stem volume. This is evident by the use of diameter (i.e., radius-squared) as a primary input in stem volume equations (Husch et al. 1993) Thus, in the absence of stem volume criteria, recommendations for reference HDRs (in Chapter 2) can be delimited within ranges of mean diameters. However, in the present chapter, HDR thresholds 80 are re-defined with respect to ranges of mean stem volume for which the reference HDRs are recommended. The task of determining mean stem volumes to be used in conjunction with reference HDRs is addressed. The present chapter also builds on the retrospective analyses undertaken in Chapter 3. Geometric models of stem volume are developed based on random and destructive sampling of 72 trees/site. Regression models of stem volume are developed from the geometric models, and applied to the complete data sets of field-based measurements. The main objectives of this chapter are: (i) to develop regression models of stem volume, and to apply these to the field-based measurements; (ii) to use the stem volumes obtained to determine ranges of mean stem volumes within which reference HDRs are meant to be applied; and (Hi) to determine how stem volume increment responds to various levels of removal of above ground competing vegetation (i.e., brushing treatments) in the period 1998-2000 (or 1999-2001). In addressing the second and third objectives, the following hypotheses were tested. In any one year: (i) the mean stem volume for trees brushed to a wider brushing radius is greater than the mean stem volume for trees brushed to a narrower brushing radius; and (ii) the mean percent change in stem volume for trees brushed to a wider brushing radius is greater than the mean percent change in stem volume for trees brushed to a narrower brushing radius. 4.2 Materials and methods 4.2.1 Study sites A full description of the four plantations (sites) is provided in Chapter 2 (Section 2.2.1). Briefly, the sites included portions of the sub-boreal spruce (SES) dry warm (dw) variant 3, and dry cool (dk) subzones in the Vanderhoof Forest District (lat. N 53'47’ to N 54' 03’, and long. W 123' 32’ to W 124' 51’) (Fig. 2.1 (Chapter 2)). Lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) were planted on the sites at an approximate range of 1270-1460 stems/ha. At the time of plot installations (1998 and 1999) these plantations ranged between four and ten years of age; they 81 were planted between 1990 and 1995. The sites were selected from areas in which competition from trembling aspen {Populus tremuloides Michaux), sitka alder (Alnus crispa ssp. sinuata (Regel) Hulten), and paper birch {Betula papyrifera Marshall) were severe. The sites ranged in elevation from approximately 755 to 854 m above sea level. Mean annual precipitation across the sites extended from 427 to 648.5 mm. Winter precipitation is relatively low, with winter snowpacks generally less than 2 m in depth (DeLong et al. 1993). The sites were prepared for planting by windrow burning followed by disc trenching. The crop trees were planted on raised spots created by the trenching, with the majority planted on the middle and top positions (McMinn and Hedin 1990). One site was disc trenched two years prior to planting (i.e.. Little Bobtail Lake site); and all other sites were disc trenched one year prior to planting. 4.2.2 Experimental design A full description of the experimental design is provided in Chapter 2 (Section 2.2.2). Briefly, the experimental design was a completely randomized one-factor design, with replication of measurements over time. Brushing within the prescribed radius, was the factor (Sit 1995, Zar 1996). The design consisted of four levels of brushing (0.0 m or control (no brushing), 0.75 m, 1.0 m and 1.25 m bmshing radii) replicated three times on each study site (12 plots/site). A total of 12 plots, each 11.28 m in radius (0.04 ha), was randomly located within a 120 m x 90 m (1.08 ha) study site, on the selected stratum of the cut block. The plot layouts are illustrated in Appendix A. Destmctive sampling of six crop trees within a plot was undertaken for all sites (72 trees/site) in May 2000. The six trees were randomly selected from among all the trees in a plot. 4.2.3 Measurements For a complete description of measurements on destmctively sampled (retrospective) trees, refer to Section 3.2.3 (Chapter 3). Measurements were taken on 6 trees/plot (72 trees/site) on all 82 study sites. Measurements of particular concern with respect to the retrospective trees were height of the cut or stump (approximately 1 cm above root collar); 130 cm above the root collar (137 km and 1116 km sites only); additional height measurements at approximately 25 cm (5 trees for Little Bobtail Lake site), 30 cm (16 trees for 101 km site), 30 cm (20 trees for 137 km site), and 35 cm and 85 cm (16 trees for 1116 km site) above the root collar; height of top whorl (total height 1998); total height (1999); and inside bark diameters (diameterœs measured to the nearest 0.01 cm) at these points of measurement (1998 and 1999). The additional measurements (approximately 1 to 2 per tree) were used to clarify questions about the age of a particular whorl, and improve the accuracy of the geometric models of stem volume. Measurements used in stem volume calculations are indicated in schematic diagrams of crop trees (see an example in Figs. 4.1a-b). Total heights were measured on live standing trees in late-August to September 1999; all other heights (1998 and 1999) were measured on felled trees in May 2000; all diameterœs (cm) were measured on tree stem discs in the laboratory (1998 and 1999 diameterœs retrospectively obtained in May 2000). A complete description of measurements on live standing (inter-seasonal) trees is found in Section 2.2.3 (Chapter 2). Measurements of particular concern with respect to the inter-seasonal trees were total heights (cm) and outside bark diameters (cm) at root collar (diameterogs) measured on live standing trees in each of three years (1998-2000 for Little Bobtail Lake, 101 km, and 137 km sites; and 1999-2001 for 1116 km site). 83 Ht 1 t Fig. 4.1a. Schematic diagram of crop tree as measured for volume calculations without additional measurement for 1999 for CanFor-Bednestii-Little Bobtail Lake and Fraser Lake-101 km sites. Diagram depicts geometric model (GM la) used in volume calculations. GM la is described by the following_ formula: Vol.w add. meas,. = [1/3 * Ti *Ri Ht,] + [1/3 * 7[ * Htz * (R,^ + R2 ^ + R, * Rz)] + [n * Rz" * Ht;] where Vol,» = volume without additional measurement (i.e., without additional measurement described in Fig. 4.1b), R, = radius at top whorl, Rz = radius at root collar. Ht, = length of leader, Htz = length of stem between top of root collar and top whorl, and Ht3 = height of root collar (n=72/site). All radii were measured inside bark. Height of top whorl was measured to a point on the stem immediately beneath the top whorl. Height of root collar was measured from the top of the mineral soil (or root) to 1 cm above the root collar (or above the swelling of the stem). 84 4k •s i Figure 4.1b. Schematic diagram of crop tree as measured for volume calculations with additional measurement for 1999 for CanFor-Bednestii-Little Bobtail Lake and Fraser Lake-101 km sites. Diagram depicts geometric model (GM lb) used in volume calculations. GM lb is described by the following formula: Volw add. meas. = [1/3 * %*R/ » Ht]] 4- [1/3 » m * Htz * (Ri + Rz + R] * Rz)] 4[1/3 * 7t * Ht3 * (Rz^ 4 - Rs^ 4-Rz * R 3 ) ] + [%* R]^ * Ht4 ] where VoLadd. meas. = volume with additional measurement approximately 25 cm (Little Bobtail Lake site (n=5/site)) and 30 cm (101 km site {n=16/site)) above the root collar, R; = radius at top whorl, Rz = radius at point of additional measurement, R3 = radius at root collar, Hti = length of leader, Htz = length of stem between point of additional measurement and top whorl, Hts = length of stem between top of root collar and point of additional measurement, and Ht4 = height of root collar. All radii were measured inside bark. Height of top whorl was measured to a point on the stem immediately beneath the top whorl. Height of root collar was measured from the top of the mineral soil (or root) to 1 cm above the root collar (or above the swelling of the stem). The ratios of stem volumes, VoL add. meas./ VoLo add. meas., from the 101 km site (n=16/site) were used to develop a regression equation which, in turn, was used to correct Volwo add. meas. (n=72/site) for the Little Bobtail Lake and 101 km sites (called GM 2). 85 4.2.4 Analysis Geometric and regression models of tree stem volume were developed from measurements on destructively sampled trees. Seventy-two trees/site (called a partial data set) were sampled from approximately 600 trees/site (called a complete data set). Geometric models (GMs) of stem volume were developed from heights and diam eter's obtained from the partial data sets. Regression models (RMs) of stem volume based on total heights and diameteroas (i.e., measurements on the same 72 trees/site taken prior to their destructive sampling) were developed from the geometric models. The RMs were used to calculate stem volumes for the complete data sets (1998-2000 for Little Bobtail Lake, 101 km, and 137 km sites; and 1999-2001 for 1116 km site). GMs based on 1999 data were developed using formulae for simple geometric forms obtained from Beyer (1991, p. 110-111). The GMs were developed from 1999 data but applied to 1998-2000 data for Little Bobtail Lake, 101 km, and 137 km sites (or 1999-2001 data for 1116 km site). The calculations for tree stem volume fell into four groups: stem volumes calculated without (GM la) and with (GM lb) an additional measurement at approximately 25 cm (Little Bobtail Lake site), 30 cm (101 km and 137 km sites), and 35 cm and 85 cm (1116 km site) heights above the root collar; and without (Little Bobtail Lake and 101 km sites) and with (137 km and 1116 km sites) a measurement at 130 cm above the root collar. An example of GM la is presented in Fig. 4.1a, and GM lb in Fig. 4.1b. GM lb helped to determine the extent that the shape of the stem departed from a conical form (i.e., was more neiloid or paraboloid). Thus, GM lb was used to assess whether a correction equation should be applied to the GM la stem volumes being examined. Stem volumes calculated for the smaller trees (i.e.. Little Bobtail Lake and 101 km sites) were corrected using a correction regression developed from a subset (n=16/site) of the destructively sampled trees (n=72/site) from the 101 km site. For these sites, stem volumes that were corrected using a regression equation from the 101 km site (i.e., a corrected version of GM la called GM 2) became the inputs (DVs) for the RMs. The principle reason for not using a subset of the destructively 86 sampled trees from the Little Bobtail Lake site to correct GM la for this site was the insufficient sample size (n=5/site for Little Bobtail Lake site). Stem volumes calculated for the larger trees (i.e., 137 km and 1116 km sites) were left uncorrected. For these sites, stem volumes that were used in their uncorrected form (GM la) became the inputs (DVs) for the RMs. The principle reason for not correcting GM la for these sites was that GM lb was found not to be significantly (p < 0.05) different from GM la for the subsets of the destructively sampled trees tested (n=20/site for 137 km site and n=16/site for 1116 km site). Specific choices with respect to the application (or non-application) of correction regressions involved graphical assessment, use of r-tests for dependent samples, and logistical considerations too lengthy to present in the thesis. The stem volumes obtained from the GMs (either GM 2 or la) became the input (DVs) for RMs of stem volume (i.e., developed from total height and diameteroa measurements (IVs) on the same trees used to develop the GMs). The latter measurements (n=72/site) came from the same trees that constituted the partial data sets, however were taken prior to the destructive sampling (1999). RMs of stem volume were developed from the partial data sets. These included primary RMs used in the analysis and secondary RMs used to test the error associated with the primary RMs. All RMs had the general form (Schumacher and Hall 1933): logioVolume =flo + ^ i * logjoDiameter + p z* logioHeight where logioVolume = logarithm (base 10) of stem volume determined by either (i) GM 2 (stem volumes corrected using the correction regression equation from 101 km site), or (ii) GM la (stem volumes left uncorrected). In the primary RMs, logioVolume = logarithm of stem volume (1999), as determined by either (i) or (ii) above. For each of (i) and (ii), Po, Pi, and Pz = coefficients determined by the regressions; logioDiameter = logarithm of diameteroa; and logioHeight = logarithm of total height. Both diameteroB and total height were measured on live standing trees in 1999. The primary RMs were 87 applied to diameteross and total heights for the complete data sets for 1998-2000 (Little Bobtail Lake, 101 km, and 137 km sites), and 1999-2001 (1116 km site). The error associated with the application of the primary RM to years preceding (e.g., 1998) and following (e.g., 2000) the year for which the model was developed (e.g., 1999), was assessed graphically and by comparison between values. The secondary RMs were used to assess other sources of error associated with the primary RMs. For example, regressions were run: (i) on uncorrected instead of corrected stem volumes (i.e., for sites whose primary RMs were developed from corrected stem volumes); (ii) on 1998 instead of 1999 data; and (Hi) using diam eter's instead of diameterofiS. A full description of the procedures used to assess the error associated with the regression models is too lengthy for inclusion in the thesis, however is available in unpublished notes. Data screening based on the removal of trees with height increment/year < 10 cm and diameter increment/year < 0 . 0 0 cm, was undertaken for the complete data sets (i.e., as described in Chapter 2 (Section 2.2.4)). However, testing of stem volumes (i.e., non-transformed stem volumes (1998 or 1999) obtained by application of the primary RMs to the complete data sets) for normality and homoscedasticity remained to be undertaken. The Kolmogorov-Smimov and Lilliesfors tests (StatSoft 1999) indicated stem volumes obtained from the primary RMs (1998 or 1999) were generally normally distributed among plots. The Lilliesfors test indicated that stem volumes for the Little Bobtail Lake site were non-normal for about half of the plots. The Brown-Forsythe test (StatSoft 1999) indicated homoscedasticity for stem volumes for all sites (1998 and 1999). The Levene test (StatSoft 1999) indicated that stem volumes for the 1116 km site were non-homogeneous. The concern for the small sample size ( 6 trees/plot) in the study suggested use of a non-parametric procedure. However, the existing non-parametric procedures (e.g., Freidman’s method and Scheier-Ray-Hare test) were not applicable to the experimental design (D. Ayers, pers. comm. 2002; Sokal and Rohlf 1995, pp. 440-447). Since normality and homoscedasticity appeared marginal for at least one site, data sets were logtransformed and the logioVolume data tested for normality and homoscedasticity. The Kolmogorov- 88 Smirnov and Lilliesfor tests for normality, and the Brown-Forsythe and Levine tests for homoscedasticity, were conducted on log-transformed data. The transformed data resulted in a marginal reduction in the numbers of non-normal plots, however, the improvement was not uniform across all sites. The Kolmogorov-Smimov test indicated only minor reductions in the numbers of non-normal plots. The Lilliesfors test indicated a major reduction in the number of non-normal plots for the Little Bobtail Lake site, however, only marginal (or nil) reductions in the numbers of non­ normal plots for the other sites. Normality and homoscedasticity was only marginally improved by the log-transformation of the data, therefore, stem volumes were rank-transformed and multivariate analysis of variance (MANOVA) with repeated measures run on rank-transformed stem volumes. The results were compared with MANOVAs on non-transformed stem volumes and log-transformed stem volumes. The MANOVAs on rank-transformed stem volumes were used to validate the MANOVAs on nontransformed stem volumes. Conover (1980, p. 337) advises that the parametric analysis will be valid if the ANOVA on rank-transformed data produces “nearly identical results” as the ANOVA on nontransformed data. Stem volumes (cm^) of crop trees were calculated from the logioVolumes produced by the primary regression models for each tree; logioVolume Volume =10 The mean stem volumes for each plot (Vol) within a study site were calculated for use in the analysis. In secondary analyses (not to be confused with the secondary RMs): (i) mean logioVolumes for each plot within a site were calculated for use in an alternate analysis; and (ii) stem volumes for all trees at a site x 3 years (each year from 1998 to 2000 for Little Bobtail Lake, 101 km, and 137 km sites; and each year from 1999 to 2001 for 1116 km site) were ranked; and mean stem volume ranks calculated for use in an alternate analysis. The secondary analyses were run in conjunction with the primary analysis, and an overall best means of analyzing the data across all the sites selected. If no 89 substantial improvement could be attributed to either the repeated measures MANOVA on logioVolumes (i.e., no substantial differences were found between the MANOVAs on nontransformed and log-transformed stem volumes) or stem volume ranks (i.e., no substantial differences were found between the MANOVAs on non-transformed and rank-transformed stem volumes), then it seemed reasonable to use the results obtained by the repeated measures MANOVA on nontransformed data. In addition, the repeated measures MANOVA on mean stem volumes had the advantage of facilitating the calculation of percent changes in the stem volume (which were analyzable by post hoc procedures). Relative growth rate and simple ratios are among the techniques used to compare growth characteristics across years (Kozlowski and Pallardy 1997). Percent change in stem volume combines both these approaches. Although other approaches were possible (Opio et al. 2000, Opio et al. 2003), it was advantageous to use %AVols, and normalize stem volumes to the year of the initial treatments. Percent change in stem volume was calculated for each tree; %A VoZ, = [(VoZ, - VoZgg)/ VbZw] * 100 where %A VbZ, = percent change in stem volume between 1998 (year when treatments begun, 1999 for 1116 km site), and 1999 and 2000 (2000 and 2001 for 1116 km site); VbZ, = volume in 1999 and 2000 (2000 and 2001 for 1116 km site). The mean %AVol for each plot within a site was calculated for use in the analysis. MANOVAs and Tukey HSD post hoc tests were run on the mean %AVols. Pearson product-moment correlation coefficients were calculated comparing stem volume and diameteroB, and %AVol and %ADiam (i.e., diameteron and %ADiam as determined in Section 2.2.4 (Chapter 2)) for 2000 for all sites (Zar 1996, StatSoft 1999). The mean Vols and %AVols were used directly in the repeated measures MANOVAs. The primary part of the method described here involves the testing of two hypotheses adapted from Chapter 2 (Section 2.2.4). These hypotheses are the following. In any one year (i) the mean stem volume for trees brushed to a wider brushing radius is greater than the mean stem volume for trees 90 brushed to a narrower brushing radius (called “Voli.2 5 m> Voli.om > Volojsm > Volo.om hypothesis”); and (ii) the mean percent change in stem volume for trees brushed to a wider brushing radius is greater than the mean percent change in stem volume for trees brushed to a narrower brushing radius (called “%AVoli.25m > %AVoli.om > ^AVolojsm > %AVolo.om hypothesis”). A repeated measures MANOVA was performed using Statistica® on each site to test whether the mean stem volumes were significantly (p < 0.05) different among treatments, dates of measurement, and interaction between these two factors (StatSoft 1999). The repeated measures MANOVA model used was: Volijk = // + treatmenti+ datej + (treatment*date)ij + % where Volyk = plot mean stem volume, ju = grand mean stem volume, treatmenti = brushing radius (0.0 m or control, 0.75 m, 1.0 m, 1.25 m), datCj = year of measurement (1998, 1999, 2000), (treatment*date)ij = interaction between treatment and date of measurement, and Sijk= experimental error (Johnson and Wichem 1992, p. 263-4). Factors such as stock type, biogeoclimatic classification, and planting density were constant for a study site, thus they were not included in the analysis. Comparisons between study sites were conducted on the basis of vegetation and other site characteristics. MANOVA was used instead of the usual analysis of variance (ANOVA) because stem volumes of the repeated measures were highly correlated. MANOVA is commonly applied to repeated measures data with several correlated dependent variables (von Ende 1993, p. 117-8). The repeated measures MANOVA provided an overall assessment of mean stem volumes; and whether the factors, treatment and date, had a significant (p < 0.05) effect on the mean stem volumes. If the repeated measures MANOVA determined that one or more of the factors had a significant overall effect, then more specific effects could be investigated by means of post hoc procedures. The Tukey HSD post hoc test was performed on each site to test whether the mean stem volumes were significantly (p < 0.05) different between specific treatments (0.0 m or control, 0.75 m, 1.0 m, and 1.25 m brushing radii). Differences between mean stem volumes, broken down by 91 treatment, were assessed by means of the Voli.2 5 m> Voli.om > Volo.7 5 m> Volo.om hypothesis. The Tukey HSD test was performed on each site to test whether the mean %AVols were significantly (p < 0.05) different between treatments (0.0 m or control, 0.75 m, 1.0 m, and 1.25 m brushing radii). Differences between mean %AVols, broken down by treatment, were assessed by means of the %AVoli,25m > %AVoli.0 m> %AVolo.7 5 m> %AVolo.om hypothesis. The procedures described above helped to determine whether brushing treatments resulted in significant (p<0.05) increases in stem volume. Thus, absolute differences in mean stem volumes between no brushing and the treatments were calculated for 2 0 0 0 ; %A Vb/2 0 0 0 = [(V oli - Volo.Om )/ Volo.Om] * 100 where %AVol2 ooo = difference in mean stem volume between no brushing and the treatments in 2 0 0 0 ; Voli = mean stem volume for 0.75 m, 1 . 0 m, or 1.25 m brushing radius in 2 0 0 0 ; and Volo.om = mean stem volume for no brushing in 2000. Since measures of %AVol2 ooo could be misleading, differences in percent change in mean stem volume between no brushing and treatments were calculated for 2 0 0 0 : %A^yoW = [(%AVoZ, - %AVoZo.oJ/ %AVoZka«] * 100 where %A^Vol2ooo = difference in mean percent change in stem volume between no brushing and treatments in 2000, %AVb/, = mean percent change in stem volume for 0.75 m, 1.0 m, and 1.25 m brushing radius in 2 0 0 0 , and %AVolo.om = mean percent change in stem volume for no brushing in 2000. The procedures described assisted in the definition of ranges of absolute stem volumes (based on volumes in 2000) within which reference HDRs were recommended. 92 4.3 Results The primary RMs, those applied to the complete data sets (i.e., approximately 600 trees/site), are described in Table 4.1. The primary analyses were conducted using these RMs. Table 4.1. Primary regression models" for stem volume for all study sites Geometric Study site Model* Coefficients for Regression Model F-value p-level 0.980 1704.7 <0.001 1 .2 1 2 0.980 1700.5 < 0 .0 0 1 L6% L267 0.989 2987.4 <0.001 1.766 L075 0^178 1459.3 < 0 .0 0 1 Po Pi Pi 2 -0.609 1T02 L522 707 W 2 41807 1.475 137 kn/ la -1.084 1116 km* la 41619 CanFor-Bednestii Little Bobtail Lake'' Fraser Lake “Primary regression models were developed from *stem volumes that were either corrected using the correction regression from 101 km site (GM 2), or stem volumes that were uncorrected (GM la). GMs 2 and la, and the primary RMs were based on 1999 data. RMs la used diameteroes and heights as TVs. RMs la were applied to the complete data sets. ‘Regression models have the general form: logioVolume = Po + pi * logioDiameteron + Pz* logioHeight (units for volume are cm^, units for diameter and height are cm). ''^Ranges of diameteron and height from which RMs were developed are: ‘^1.03-4.60 cm diameteron and 70-187 cm height; '"1.35-4.11 cm diameteron and 76-196 cm height;^2.27-12.20 cm diameteron and 134-610 cm height; and *3.72-11.70 cm diameteron and 185478 cm height. Note: n=72/site. In comparing the results from the primary analysis (i.e., based on non-transformed stem volumes) with those from the secondary analyses (i.e., based on log-transformed and ranktransformed stem volumes), a strong endorsement for either of the secondary analyses was not obtained. A full description of the secondary analyses vis a vis the primary analysis is not presented 93 in the thesis, however is available in unpublished notes. Briefly, it was found that year of measurement was significant (p < 0 .0 0 1 ) for the primary and both secondary analyses for all sites. Although the results for the Little Bobtail Lake site were ambiguous for brushing treatment and interaction; the outcomes for the 101 km and 137 km sites were uniform for all the effects tested. Furthermore, the results for the 1116 km site were the same for brushing treatment, but ambiguous for interactions. Neither did the tests for normality and homoscedasticity indicate a substantial and consistent improvement with the log-transformed data. Considering also the substantial benefit to be gained by being able to conduct post hoc tests on plot mean %AVols data (not available from either logioVolumes or volume ranks data), the results produced and conclusions drawn are those based on the primary analysis (i.e., that conducted on non-transformed stem volumes). The repeated measures MANOVAs on mean stem volumes (Table 4.2) indicated that year of measurement had a significant (p < 0.001) overall effect. Beyond this trivial outcome, brushing treatment and interaction between brushing treatment and year generally did not have a significant effect. The 101 km site was an exception with both brushing treatment (p < 0.05) and interaction (p < 0 .0 0 1 ) being significant. The repeated measures MANOVAs on percent change in stem volume (Table 4.2) also indicated that year of measurement had a significant (p < 0.001) overall effect. The results differed from the MANOVAs on mean stem volumes in that both brushing treatment and interaction proved to be significant for two additional sites. Brushing treatment was significant (all p < 0.05) for the Little Bobtail Lake, 101 km, and 1116 km sites. Interaction was significant (all p < 0.01) for the same sites. 94 Table 4.2. MANOVAs with repeated measures showing the factors affecting mean stem volumes Study site Effect Mean stem volume (Pr > F) F Value df Effect Percent change in stem volume F Value (P r> F ) CanFor-Bednestii Little Bobtail Lake'' 0.34 6 1.29 245.30 1.97 0.13 1550.85 6.07 3 5.29 0.03 6.78 0.00 6 500.60 6.83 1114.38 7.31 3 1.45 0.30 2 66.44 0.00 0.98 306.73 6 1.59 0 .2 1 1 .0 1 0.87 5.24 0.00 0.99 1235.55 3.99 Brushing Year Brushing* Year 3 Brushing Year Brushing*Year 2 0 .0 0 5.74 0.02 0 .0 0 0.00 Fraser Lake 101 km* 137 km“ 1116 km“ Brushing Year Brushing*Year Brushing Year Brushing* Year 2 2 0.23 352.64 6 0.14 3 0 .0 0 0 .0 1 0 .0 0 0 .0 0 0.45 0.00 0.45 0.03 0.00 0 .0 1 “Percent changes in stem volume were calculated with reference to stem volumes in 1998 for Little Bobtail Lake site (n=486/site), 101 km site (n=555/site), and 137 km site {n=487/site)\ and stem volumes in 1999 for 1116 km site {n=509/site). *'“Stem volumes were calculated; ''with 101 km site correction, and “without correction. Note; MANOVAs for all sites were run on primary regression model stem volumes (Table 4.1). MANOVAs were run on 1998-2000 data for Little Bobtail Lake, 101 km, and 137 km sites; and 1999-2001 data for 1116 km site (n=12/site). 4.3.1 CanFor-Bednestil-Little Bobtail Lake site An assessment of the first hypothesis did not indicate a difference between mean stem volumes (Table 4.3) obtained by the different brushing radii in 1998 and 1999. Mean stem volumes in 2000, however, did seem to be greater for the 1.0 m and 1.25 m brushing radii. It seemed that Voli^m > Vol(0 .7 3 m. 0 .0 m) (i c., Volojsm = Volo.om) would hold. The Tukey HSD post hoc test confirmed 95 this impression, indicating that mean Voh zw w a s significantly (p = 0 .0 1 ) greater than both mean Volo.7 3 m(P < 0.05) and mean Volo.om in 2000. Consideration of the second hypothesis (Fig. 4.2), did not indicate any difference in mean %AVols obtained by different brushing radii in 1999. However, the results suggested that % AVoli.2 5 m > %AVol(i.om, 0 .7 5 m) > %AVolo.om (1.6., %AVoli.om ~ %AVolo.7 5 m) would hold in 2000. The Tukey HSD test partially confirmed this impression, indicating that mean %AVoli.2 5 mwas significantly (p = 0 .0 2 ) greater than mean %AVoli.om, which was significantly (p = 0.04) greater than mean %AVolo.om in 2 0 0 0 . Mean %AVolo.7 5 mwas significantly (p = 0 .0 1 ) greater than mean %AVolo,om in 2 0 0 0 . 600 E -3 500 ^(/] 400 0 .0 0 .2 m -B - 0.75 m & 300 1 .0 0 "S 200 m 1.25 m I ^ 100 1998 2000 1999 Year Fig. 4.2. Percent changes in stem volume between 1998 and 2000 for CanForBednestii-Little Bobtail Lake site. Percent changes in stem volume were calculated with reference to stem volumes in 1998 (n=486/site), and mean percent change in stem volume was calculated for each plot {n=12/site). In order to improve clarity, error bars are not presented in the figure. In comparisons with no brushing in 2 0 0 0 (Volo.om= 522.0 cm and %AVolo.om= 357.4%), mean stem volume was 41.9% greater (p = 0.01) for the 1.25 m treatment (Voli,2 5 m= 739.9 cm^); and mean %AVols were 35.3% greater (p < 0.01) for the 1.25 m treatment (A%Vol],2 5 m= 483.5%), 16.5% 96 greater (p = 0.04) for the 1 . 0 m treatment (A % V o li,o m = 416.4%), and 19.3% greater (p = 0 .0 1 ) for the 0.75 m treatment (A%Volo.7 5 m= 426.5%) in 2000. 4.3.2 Fraser Lake-101 km site An assessment of the first hypothesis did not indicate a difference between mean stem volumes (Table 4.3) obtained by the different brushing radii in 1998 and 1999. Mean stem volumes in 2000, however, did seem to be greater for the 1.0 m and 1.25 m brushing radii. It seemed that Vol(i,25m. l.Otn) > Vol(o.75tn, 0 .0 m) (1-6., Voli,25m ~ Voli.Om and Volo.75m ~ Volo.Qm) W O U ld hold. The Tukcy HSD post hoc test confirmed this impression, indicating that both mean Voli.2 5 m and Voli.om were significantly (p < 0 .0 1 ) greater than both mean Volo.7 5 m and Volo.om in 2 0 0 0 . 600 I 500 I I« KX) —0 .0 0 m .s -s — 0.75 m S, 300 -6 — 1 . 0 0 m I 200 1.25 m " I 100 1998 2000 1999 Year Fig. 4.3. Percent changes in stem volume between 1998 and 2000 for Fraser Lake101 km site. Percent changes in stem volume were calculated with reference to stem volumes in 1998 (n=555/site), and mean percent change in stem volume was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. 97 Consideration of the second hypothesis (Fig. 4.3), did not indicate any difference in mean %AVols obtained by different brushing radii in 1999. However, the results seemed to indicate that %AVol(i.25m, 1,0 m) > %AVol(i.Om, o.7 5 m) (t-G., %AVoli.25m ~ %AVol].Om &Itd %AVolo.7 5 m~ %AVolo,Om) would hold in 2000. The Tukey HSD test confirmed this impression, indicating that both mean %AVoli.2 5 mand %AVoli,om were significantly (p < 0 .0 1 ) greater than mean %AVolo.7 5 m and %AVolo.om in 2 0 0 0 . In comparisons with no brushing in 2 0 0 0 (V olo,om = 468.0 cm^ and % A V olo.om = 449.8%), mean stem volumes were 54.1% greater (p < 0.01) for the 1.25 m treatment (Voli.2 5 m= 721.1 cm^), and 55.6% greater (p < 0.01) for the 1.0 m treatment (V o li,o m = 728.0 cm^); and mean %A Vols were 37.6% greater (p < 0.01) for the 1.25 m treatment (%AVoli.2 5 m= 619.0%), and 24.1% greater (p < 0 .0 1 ) for the 1 . 0 m treatment (% A V o li.o m = 558.3%) in 2 0 0 0 . 4.3.3 Fraser Lake-137 km site An assessment of the first hypothesis did not indicate a difference between mean stem volumes (Table 4.3) obtained by the different brushing radii in 1998. Mean stem volumes in 1999 and 2000, however, did seem to be greater for the 1.0 m and 1.25 m brushing radii. The impression was that Vol(i,25m. I.Om) > Vol(0 .7 5 m,0 .0 m) (I.G., %AVoli.25m ~ %AVoli.Om and %AVolo.7 5 m~ %AVolo.Om) would hold in 1999 and 2000. The Tukey HSD post hoc test indicated that mean Yoh om was significantly (p = 0.03) greater than mean Volo.7 5 min 1999. The same test indicated that both mean Voli.2 3 mand Voli.om were significantly (p < 0 .0 1 ) greater than mean Volo.7 3 m, and mean Vol, omwas significantly (p = 0 .0 1 ) greater than mean Volo.om in 2 0 0 0 . Consideration of the second hypothesis (Fig. 4.4), did not indicate any difference in mean %AVols obtained by different brushing radii in 1999. However, the results seemed to indicate that %AVol(i.2 5 m, l.Om,0 .7 5 m) > %AVolo.Om (i.G., %AVoli.2 5 m~ %AVoli.Om ~ %AVolo.7 5 m) WOUld hold in 2000. The Tukey HSD post hoc test contradicted this perception, indicating no significant differences between mean %AVols in 2000. 98 In comparisons with no brashing in 2 0 0 0 (Volo.om= 5074 cm^ and %AVolo.om= 161.4 %), mean stem volumes were 59.2% greater (p = 0 .0 1 ) for the 1 . 0 m treatment (Voli.om= 8080 cm^) in 2 0 0 0 . Mean stem volumes were 72.7% greater (p = 0 .0 1 ) for the 1.25 m treatment (Voli.2 5 m= 7054 cm^) than with the 0.75 m treatment (Volo.7 5 m= 4085 cm^) in 2000. Mean stem volume for the 1.25 m treatment (Voli.2 5 m= 7054 cm^) was not significantly different from mean stem volume for no brushing (Volo.om= 5074 cm^) in 2000. It was found that mean %AVols (%AVolo,om= 161.4% to %AVoli.2 5 m= 212.5%) were not significantly different from one another in 2000. I 600 ^ 500 I 400 - ^ 0 .0 0 m c -B - 0.75 m & 300 -A— 1 . 0 0 m Xi " ! 1.25 m 200 100 1998 2000 1999 Year Fig. 4.4. Percent changes in stem volume between 1998 and 2000 for Fraser Lake137 km site. Percent changes in stem volume were calculated with reference to stem volumes in 1998 (n=487/site), and mean percent change in stem volume was calculated fro each plot {n=12/site). In order to improve clarity, error bars are not presented in the figure. 4.3.4 Fraser Lake-1116 km site An assessment of the first hypothesis did not indicate a difference between mean stem volumes (Table 4.3) obtained by the different brushing radii in 1999, 2000, or 2001. The Tukey HSD 99 post hoc test confirmed this perception, indicating no significant differences between any of the mean stem volumes, in any of the years of measurement. Consideration of the second hypothesis (Fig. 4.5), did not indicate a difference in mean %AVols obtained by different brushing radii in 1999. However, the results seemed to indicate that %AVol(i.25m, l.Om, 0 .7 5 m) > %AVolo.om (i.e., %AVoli.2 5 m~ %AVol|.om ~ %AVolo.7 5 m) would hoM in 2 0 0 0 . The Tukey HSD post hoc test confirmed this perception, indicating that mean %AVoli.2 5 m, %AVol|.om, and %AVolo.7 5 mwere all significantly (p < 0 .0 1 ) greater than mean %AVolo,om in 2 0 0 0 . 600 1 -S 500 — 0.(X) m I 400 s - o - 0.75 m & 300 1 .0 0 I 1.25 m " 200 100 1999 2001 2000 Year Fig. 4.5. Percent changes in stem volume between 1999 and 2001 for Fraser Lake1116 km site. Percent changes in stem volume were calculated with reference to stem volumes in 1999 (n=509/site), and mean percent change in stem volume was calculated for each plot {n=12/site). In order to improve clarity, error bars are not presented in the figure. 100 m Table 4.3. Mean stem volumes" and standard errors of mean (SEM)* from 1998 to 2001 for all Year of measurement Study site Brushing radius (m) 1998 1999 2000 2001 111.9 (6 .6 ) 268.2 (10.3) 262.4 (10.1) 306.7 (11.0) 360.1 (11.9) 522.0 (14.4) 537.5 (14.5) 608.6 (15.5) 739.9 (17.0) n.m. n.m. n.m. n.m. 224.5 (9.3) 258.9(10.1) 337.3(11.4) 314.6(11.0) 468.0 (13.5) 538.2 (14.5) 728.0 (16.8) 721.1 (16.7) n.m. n.m. n.m. n.m. 3451.7 (37.9) 2726.0 (34.4) n.m. n.m. n.m. n.m. 10071.5 (63.2) 9214.4 (59.6) 9478.2 (60.5) 9837.6 (62.1) CanFor-Bednestii Little Bobtail Lake" (n=486/site) 0 .0 0 0.75 1 .0 0 1.25 106.6 (6.5) 122.0 (6.9) 137.5 (7.4) Fraser Lake km'' {n=555/site) 1 0 1 0.75 84.0 (5.7) 99.5 (6.2) 1 .0 0 116.6 (6.7) 1.25 106.1 (6.4) 0 .0 0 137 km" 0 .0 0 (n=487/site) 0.75 1 .0 0 1.25 1116 k n / (n=509/site) 0.00 0.75 1 .0 0 1.25 1956.4 (28.4) 1508.0 (25.8) 2951.1 (33.7) 2548.4 (33.2) 4694.8 (44.7) 5074.4 (46.0) 4084.5 (42.2) 8080.3 (55.8) 7053.9 (54.8) n.m. n.m. n.m. n.m. 5146.3 (45.3) 4274.3 (40.6) 4357.6 (41.0) 4481.1 (41.9) 7474.2 (54.5) 6557.4 (50.2) 6755.0 (51.1) 6983.1 (52.3) 5445.0 (45.8) "Mean stem volumes (cm^) were calculated with primary regression models (Table 4.1): "Little Bobtail Lake and '^101 km sites with 101 km site correction, and "137 km and^l 116 km sites without correction. *SEM are presented in brackets following mean stem volumes. "'^Exact dates of measurement for "Little Bobtail Lake site were August 7-10, 1998, August 19-23, 1999, and August 11-15, 2000; '^101 km site were August 14-19, 1998, September 16-20, 1999, and August 22-25, 2000; "137 km site were August 12-17, 1998, August 30-September 3, 1999, and September 5-11, 2000; and^l 116 km site were September 7-11, 1999, September 12-15, 2000, and September 4-7, 2001. Note: n.m. indicates no measurements were taken. Mean stem volumes (i.e., ranging from 9214 cm^ to 10072 cm^) were found not to be significantly different from one another in 2 0 0 1 . In comparisons with no brushing in 2 0 0 1 (Volo.om = 101 10072 cm^ and %AVolo,om= 106.3%), mean %AVols were 24.1% greater for the 1.25 m treatment (%AVoIi.2 5 m= 131.9%), 29.9% greater for the 1 . 0 m treatment (%AVoli,om= 138.1%), and 25.8% greater (p < 0.01) for the 0.75 m treatment (%AVolo.7 5 m= 133.7%) in 2001 (p < 0.01). 4.4 Discussion In the repeated measures MANOVAs on mean stem volumes, year of measurement was significant for all sites. It would not be expected that brushing treatments on the sites with larger crop trees would significantly affect stem volume. The larger trees were already well established and over-topping most non-crop vegetation, thus inter-specific competition would not be a factor for these trees (Oliver and Larson 1996). Brushing on 137 km and 1116 km sites did not significantly affect stem volume. It might be expected that brushing treatments would have a significant effect for the sites with smaller trees; they did at the 101 km site but did not at the Little Bobtail Lake site. This outcome may be due to greater nutrient-richness at the 1 0 1 km site, a possibility that is indicated by the greater homogeneity in competing vegetation at this site. The “better” (i.e., 101 km) site is likely to produce trees with greater stem volume (Eis et al. 1982, Harrington and Tappeiner 1991). Interaction between year and treatment was significant for the 101 km site, indicating the influence of a factor (possibly nutrient-richness) not tested in the MANOVA. Interaction was not significant for the other sites. In the repeated measures MANOVAs on percent changes in stem volume (%AVol), year of measurement was significant for all sites. Brushing treatment was significant for all sites, except the 137 km site. It seemed surprising that brushing treatment was significant at the 1116 km site, since brushing was not significant for either site with larger trees in the MANOVA on mean stem volumes. This outcome may be explained by the substantial presence of nitrogen-fixing alder at the 1116 km site. The presence of alder would have improved the nutrient quality of this site, which possibly affected the greater homogeneity (i.e., uniformity) in competing vegetation at the 1116 km site (Sachs and Comeau 1991, Simard and Heineman 1996, Sanborn et al. 1997). Interaction was significant for 102 most sites, indicating the effect of a factor such as nutrient-richness not tested in the MANOVA. Interaction was not significant for the 137 km site. It is useful to compare the above outcomes with MANOVAs on mean diameteroBS (Diams) and percent changes in diameteroB (%ADiams). The repeated measures MANOVAs on mean diameterofiS produced similar results as those on mean stem volumes: the only differences were the significant interactions indicated for the Little Bobtail Lake site, and 137 km and 1116 km sites. The repeated measures MANOVAs on percent changes in diameteros produced identical results as those on the %AVols. A full description of the MANOVAs on mean diameteroas and %ADiams was too lengthy for inclusion in the thesis, however is available in unpublished notes. The similarity in results between the repeated measures MANOVAs on mean stem volumes and mean diameteroBS, and %AVols and %ADiams, is what we would expect since diameter (or basal area) is a primary component in most stem volume equations (Husch et al. 1993). With respect to the data sets used in the study, stem volume and diameteros were found to be highly correlated (Pearson r = 0.94-0.95 for 2000 for all sites), and %AVols and %ADiams were found to be highly correlated (Pearson r = 0.88-0.92 for 2000 for all sites) (Zar 1996, StatSoft 1999). These similarities are a support for the use of mean diameteroBS in place of mean stem volumes for sites where stem volume equations are not available. Variations of the first hypothesis were found to hold for all the sites, except the 1116 km site, in 2000. These varied from Vol 1 ,2 5 m> Vol(o,7 5 m,0 .0 m) (i.e., Volo.vsm ~ Volo.om) at the Little Bobtail Lake site, to Vol(i.25m, l.Om) > Vol(o,7 5 m.0 . 0 m) (i.e., Vol|.25m ~ Vofi.om and Volo.7 5 m~ Volo.om) at the 101 km site, to Voli.om > Volo.7 5 m> Volo.omat the 137 km site. Variations of the second hypothesis were found to hold for the Little Bobtail Lake and 101 km sites in 2000; but not for the 137 km and 1116 km sites. Where significant patterns in the data were found, these varied from %AVoli,2 5 m> %AVol(i.om,o.7 5 m) > %AVolo.om (i.e., %AVol|.om = %AVolo.7 5 m) at the Little Bobtail Lake site, to %AVol(i.25m, i.om) > %AVol(0 .7 3 m.0 An) (i.e., %AVoli^m = %AVoIi.om and %AVolo.7 5 m= %AVolo.om) at the 1 0 1 km site. 103 The outcomes with respect to the two hypotheses seem to be the result of the crop tree obtaining more light the wider its brushing radius, and radial growth being greater the greater the availability of light (Eis et al. 1982). Absolute stem volume and increment in stem volume are very much dependent on diameter and increment in diameter (Husch et al. 1993). Therefore, stem volume and increment in stem volume will be greater the more severe the brushing treatment. The Voli,25m > Vol, On, > Volo.7 5 m> Volo.Om and %AVoli,25m > %AVol,,Om > %AVolo,7 5 m> %AVolo.on, hypotheses were partially confirmed by comparisons with mean diameteroes (Diams) and percent changes in diameteros (%ADiams) undertaken for inter-seasonal analyses in 2000 (Chapter 2). For example, the patterns Diami.2 5 m> Diamo om, 0 .7 3 m) > Diamo.om (i.e., Diam,.om = Diamo.7 3 J for the Little Bobtail Lake site, Diam(i.25m, i.om) > Diam(o.7 5 m,o.om) (i.e., Diam,.2 5 m~ Diam,,om and Diamo,7 5 m ~ Diamo.orn) for 1 0 1 km and 137 km sites, and Diam,.2 5 m> Diam(o.7 5 m.o.om) (i.e., Diamojsm ~ Diamoom) for the 1116 km site were found to hold. Both hypotheses seemed more likely to hold for the sites with smaller trees (i.e.. Little Bobtail Lake and 101 km sites) than for the sites with larger trees (i.e., 137 km and 1116 km sites). However, these results were not confirmed by the differences in absolute stem volumes between no brushing and the 1.25 m treatment in 2 0 0 0 (%AVol2ooo): +41.7% (p = 0 .0 1 ) and +54.1% (p < 0 .0 1 ) for the sites with smaller trees, and +39.0% and -2.3% for the sites with larger trees (both values not significant); or differences in percent changes in stem volume between no brushing and the 1.25 m treatment in 2000 (% a V o 12 ooo): +35.3% (p < 0.01) and +37.6% (p < 0.01) for the sites with smaller trees, and +31.7% (not significant) and +24.1% (p = 0.01) for the sites with larger trees. The following recapitulates the recommendations for reference HDRs stated in Section 2.4 (Chapter 2). The only change that has been made is to delimit recommended reference HDRs within ranges of mean stem volumes (instead of ranges of mean diameters). Reference HDRs along with BEC classifications, ranges of percent cover of aspen or alder, and ranges of mean stem volumes of the crop trees are summarized for all sites in Table 4.4. The HDR thresholds have not changed. 104 The reference HDR previously recommended for the Little Bobtail Lake site was 40-49 (Chapter 2 (Section 2.4)). This HDR threshold was meant to apply on SBS dw3 (01) plantations where aspen percent cover was approximately in the range of 5-45% and height was approximately in the range of 1.0-3.5 m, 5 years after planting (1998); and mean diameteroBS were as stated in Chapter 2 (Section 2.4). The currently recommended ranges of mean stem volumes are approximately 100150 cm^ for plantations 5 years after planting (1998), 250-400 cm^ for plantations 6 years after planting (1999), and 500-750 cm^ for plantations 7 years after planting (2000). The reference HDR previously recommended for the 101 km site was 40-51 (Chapter 2 (Section 2.4)). This HDR threshold was meant to apply on SBS dw3 (01) plantations where aspen percent cover was approximately in the range of 30-40% and height was approximately in the range of 1.0-3.5 m, 4 years after planting (1998); and mean diameterons were as stated in Chapter 2 (Section 2.4). The presently recommended ranges of mean stem volumes are approximately 50-150 cm^ for plantations 4 years after planting (1998), 200-350 cm^ for plantations 5 years after planting (1999), and 450-750 cm^ for plantations 6 years after planting (2000). The reference HDR previously recommended for the 137 km site was 45-54 (Chapter 2 (Section 2.4)). This HDR threshold was meant to apply on SBS dk (01) plantations where aspen percent cover was approximately in the range of 0-30% and height was approximately in the range of 1.0-3.5 m, 9 years after planting (1998); and mean diameteroes were as stated in Chapter 2 (Section 2.4). The currently recommended ranges of mean stem volumes are approximately 1500-3000 cm^ for plantations 9 years after planting (1998), 2500-5500 cm^ for plantations 10 years after planting (1999), and 4000-8000 cm^ for plantations 11 years after planting (2000). The reference HDR previously recommended for the 1116 km site was 38-47 (Chapter 2 (Section 2.4)). This HDR threshold was meant to apply on SBS dk (05) plantations where alder percent cover was approximately in the range of 20-30% (birch percent cover was approximately 1015%) and height was approximately in the range of 1.5-2.5 m (alder and birch), 10 years after planting (1999); and mean diameteroaS were as stated in Chapter 2 (Section 2.4). The presently 105 recommended ranges of stem volumes are approximately 4000-5500 cm^ for plantations 10 years after planting (1999), 6500-7500 cm^ for plantations 11 years after planting (2000), and 9000-10000 cm^ for plantations 1 2 years after planting (2 0 0 1 ). Graphical inspection of the scatter plots produced by the primary RMs seemed to indicate that the RMs developed from 1999 data provided an acceptable estimate of the 1998 and 2000 (2000 and 2 0 0 1 for the 1116 km site) stem volumes. values describing the variations between logioVolumes predicted for 1998 and the regression lines based on 1999 data indicated errors of 5.5 % (Little Bobtail Lake site), 9.1 % (101 km site), and 2.0 % (137 km site) when the primary RMs were applied to 1998 heights and diameterons. R^ values describing the regression lines based on 1999 data indicated errors in the range of 1.1-2.2% when the primary RMs were applied to 1999 heights and diameterons. It was not possible to determine R^ values describing the variations between logioVolumes predicted for 2000 (and 2001 for 1116 km site), and the regression lines based on 1999 data. However, it seems probable that the errors incurred by applying the primary RMs to 2000 (and 2 0 0 1 ) heights and diameteroas would not be too much higher than those indicated for the application of the primary RMs to the 1998 data. The above ranges of error may be compared against the errors found in applying the primary RMs to 1999 heights and diameteroas for Little Bobtail Lake, 101 km, 137 km, and 1116 km sites. Mean stem volumes for the complete data sets (i.e., 300.6 cm^, 284.6 cm^, 4326 cm^, and 4592 cm^) were 13.2 %, 1.1 %, 1.2 % and 2.9% greater than mean stem volumes for the partial data sets (i.e., 265.5 cm^, 281.6 cm^, 4275 cm^, and 4462 cm^). With the exception of the Little Bobtail Lake site, these comparisons seem to validate the primary RMs as they were applied to 1999 height and diameteron measurements. 106 Table 4.4. Recommended reference HDRs with vegetation complexes, BEC classifications, and ranges of percent cover of competing vegetation and mean stem volume within which reference Study site Veg. complex/ Ref. Pet. cov. comp. BEC class." HDR* veg." Mean stem volume (cm^)'^ 1st 2nd 3rd year year year CanFor-Bednestii Little Bobtail Lake aspen SBS dw3 (01) years after planting Maximum Minimum 49 40 5 5 6 7 45 150 100 400 250 750 500 4 150 5 6 50 350 200 750 450 10 5500 2500 8000 4000 5 Fraser Lake aspen 101 km SBS dw3 (01) aspen 137 km SBS dk (01) alder/birch 1116 km SBS dk (05) years after planting maximum minimum 4 40 40 30 54 30 45 0 9 3000 1500 10 10 11 12 30 (15) 20 (10) 5500 7500 1 0 0 0 0 4000 6500 9000 51 years after planting maximum minimum 9 years after planting maximum 47 minimum 38 11 “Veg. complex/BBC class, refers to vegetation complex and biogeoclimatic ecosystem classification within which reference HDRs are meant to be applied. *Ref. HDRs (reference HDRs) are meant to be applied on plantations of the specified age, vegetation complex, BEC classification, and ranges of percent cover of competing vegetation and mean stem volume. Reference HDRs for Little Bobtail Lake and 137 km sites were qualified as being "satisfactory estimates"; reference HDRs for 101 km and 1116 km sites were qualified as being "tentative estimates". Tct. cov. comp. veg. refers to percent cover of competing vegetation (i.e. aspen or alder/birch) estimated before treatments were conducted. Percent cover was visually estimated in July or August 1998 for Little Bobtail Lake, 101 km, and 137 km sites; and in June 1999 for 1116 km site. 'Wean stem volumes were calculated from diameter outside bark at root collar and total height measurements taken after mid-August 1998, 1999, and 2000 for Little Bobtail Lake, 101 km, and 137 km sites; and mid-August 1999, 2000, and 2001 for 1116 km site. 107 Inspection of mean stem volumes produced by the secondary and primary RMs indicated an overall consistency between the two sets of RMs. However, a possible error of ±10% remained for certain sites, and years. This level of error seemed consistent with the level of error estimated by graphical and statistical comparisons between logioVolumes (i.e., produced by the primary RMs) in the various years of measurement. It seems probable that the overall error of the primary RMs (Table 4.1) would be < 10% if applied to similar size lodgepole pine growing on similar sites (SBS dw3 (01) for Little Bobtail Lake and 101 km sites, SBS dk (01) for the 137 km site, and SBS dk (05) for the 1116 km site). A complete presentation of the assessment of error was too lengthy to be included in the thesis, however is available in unpublished notes. 4.5 Conclusions and recommendations A systematic pattern in the mean stem volumes and percent changes in stem volume (%AVol) was evident in testing the two hypotheses. The hypotheses held to a considerable extent in 2000 for study sites with smaller trees, and to a lesser extent in 2 0 0 0 (or 2 0 0 1 for the 1116 km site) for sites with larger trees. The first hypothesis did not hold for the 1116 km site, and the second hypothesis did not hold for the 137 km or 1116 km sites. Mean stem volumes and %AVols increased over time at all sites. Variations of the first hypothesis were found to hold for Little Bobtail Lake, 101 km, and 137 km sites in 2 0 0 0 . These varied from Voli.2 5 m> Vol(o.7 5 m, o.om) (i.e., Volo.7;m~ Volo.om) at the Little Bobtail Lake site, to Vol(i.2 5 m, i.om) > Vol(o.7 5 m,o.om) (i.e., Voh zsm ~ Vofi.om and Volojsm ~ Volo.om) at the 101 km site, to Voli om> Volojsm > Volo.omat the 137 km site. Variations of the second hypothesis were found to hold for the Little Bobtail Lake and 101 km sites in 2000. These varied from %AVoli.2 5 m> %AVol(i.0 m, 0 .7 5 m) > %AVolo,om (1-6., %AVoli.om ~ %AVolo.7 5 tn) at the Little Bobtail Lake site, to %AVol(i.2 5 m, l.Om) > %AVol(o.7 5 m.o.om) (i.e., %AVol,.25m ~ %AVoli,om and %AVolo.7 5 m= %AVolo.om) at the 101 km site. 108 The above results seemed to indicate that the two hypotheses were more likely to hold for sites with smaller trees (i.e., Little Bobtail Lake and 101 km sites) than for sites with larger trees (i.e., 137 km and 1116 km sites). However, this conclusion was not confirmed by differences in absolute stem volumes between no brushing and the 1.25 m brushing treatment, or differences in percent changes in stem volume between no brushing and the 1.25 m treatment in 2000. These anomalies may be due to differences in nutrient-richness between the four sites, as reflected in differences in homogeneity in competing vegetation. The similarity in the patterns observed between mean stem volumes and diameteroeS, and between mean %AVols and %ADiams, indicated that ranges of mean diameteroB may be used to delimit the application of the reference HDRs for areas where stem volume equations are not available. Where stem volume equations are available, these should be used to determine ranges of mean stem volumes within which reference HDRs are recommended. The primary regression models obtained are recommended for use in delimiting the ranges of application of the reference HDRs previously recommended (i.e., in Chapter 2) for specific vegetation complexes, BEC classifications, ranges of percent cover competing vegetation, and ranges of mean diameter during specified years following planting. The only change that was made in the present chapter was to delimit recommended reference HDRs within ranges of mean stem volumes (i.e., instead of ranges of mean diameters). Extensive comparisons between stem volumes obtained by various primary and secondary regressions (i.e., statistical tests, graphical inspections, and comparisons between means) indicated a probable maximum error in the models obtained of < 1 0 %. It is recommended that in future research: (i) the accuracy of stem volume equations be improved by obtaining stem discs from 30 cm height above the root collar on all destructively sampled trees; and (ii) the determination of stem volumes be extended back into the period prior to treatments by obtaining stem discs from the whorls from which total heights were estimated. 109 Literature cited Beyer, W.H. 1991. Standard Mathematical Tables and Formulae 29'*' ed. CRC Press, Inc. Boca, CA. Conover, W.J. 1980. Practical Nonparametric Statistics 2"** ed. John Wiley and Sons, Inc., New York. DeLong, S.C., Tanner, D., and lull, M.J. 1993. A Field Guide for Site Identification and Interpretation for Southwest Portion of Prince George Forest Region. Land Management Handbook 24, Research Branch, Ministry of Forests, Province of British Columbia, Victoria, B.C. Eis, S., Craigdallie, D., and Simmons, C. 1982. Growth of lodgepole pine and white spruce in the central interior of British Columbia. Can. J. For. Res. 12 (3): 567-575. Ende, C.N. von 1993. Repeated Measures Analysis: Growth and Other Time-Dependent Measures. In Design and Analysis of Ecological Experiments. Edited by S.M. Scheiner, and J. Gurevitch. Chapman and Hall, London, pp. 113-137. Harrington, T.T., and Tappeiner, J.C.B. 1991. Competition affects shoot morphology, growth, duration and relative growth rates of Douglas-fir saplings. Can. J. For. Res. 21: 474-481. Husch, B., Miller, C.I., and Beers, T.W. 1993. Forest Mensuration. Krieger Publishing Co. Malabar, Florida. Johnson, R.A., and Wichem, D.W. 1992. Applied Multivariate Statistical Analysis. Prentice Hall, Englewood Cliffs, NJ. Kimmins, J.P. 1997. Forest Ecology: A Foundation for Sustainable Management 2"'* ed. Prentice Hall, Upper Saddle River, NJ. Kozlowski, T.T., and Pallarady, S.G. 1997. Physiology of Woody Plants. Academic Press Inc., Toronto. McMinn, R.G., and Hedin, I.P. 1990. Site Preparation: Mechanical and Manual. In Regenerating British Columbia’s Forests. Edited by Lavender, D.P., Parish, R., Johnson, C.M., Montgomery, G., Vyse, A., Willis, R.A., and Winston, D. University of British Columbia Press, Victoria, B.C. pp. 150-163. Oliver, C D., and Larson, B.C. 1996. Forest Stand Dynamics. McGraw-Hill, New York. Opio, C., Jacob, N., and Coopersmith, D. 2000. Height to diameter ratio as a competition index for young conifer plantations in northern British Columbia, Canada. Forest Ecology and Management 137: 245-252. Sachs, D., and Comeau, P.G. 1991. Determination of nitrogen fixation by sitka alder at two sites in the southern interior of British Columbia. FRDA Project C03, Res. Br., BC Ministry of Forests, Victoria, BC (unpublished report). Sanborn, P., Brockley, R., and Preston, C. 1997. Ecological roles of sitka alder in a young lodgepole pine stand. Forest Research Note #PG-10. BC Ministry of Forests, Prince George, BC. 110 Schumacher, F.X., and Hall, F.S. 1933. Logarithmic expression of timber-tree volume. Jour. Agric. Res. 47; 719-34. Simard, S., and Heineman, J. 1996. Nine-year response of lodgepole pine and the dry alder complex to chemical and manual release treatments on an ICH mkl site near Kelowna. FRDA Report 259. Can. For. Serv. and BC Ministry of Forests, Victoria, BC. Sit, V. 1995. Analyzing ANOVA Designs. Biometrics Information Handbook No. 5. BC Ministry of Forests Research Program, Victoria, BC. Sokal, R.R., and Rohlf, F.J. 1995. Biometry 3rd ed. W.H. Freeman and Co., New York. StatSoft Inc. 1999. STATISTICA® Version 5.5. Tulsa, OK. Wagner, R.G., Mohammed, G.H., and Noland, T.L. 1999. Critical period of interspecific competition for northern conifers associated with herbaceous vegetation. Can. J. For. Res. 29: 890-897. Zar, J.H. 1996. Biostatistical Analysis. Prentice-Hall, Inc., Englewood Cliffs, NJ. Ill CHAPTER 5 INTRA^EASONAL VARIATIONS IN HEIGHT TO DIAMETER RATIOS IN LODGEPOLE PINE FOLLOWING VARIABLE REMOVAL OF COMPETING VEGETATION Abstract Height to diameter ratio (HDR) has been proposed as an alternative to conventional procedures for assessing competition between crop trees and other vegetation. However, HDRs vary throughout the growing season due to variations in the rate of change in height and diameter. There is an interest, therefore, in determining variations in HDR within a growing season (i.e., intra-seasonal changes) and the time of the year when measurements of HDR should be taken for operational purposes. HDR measurements were taken at approximately monthly intervals during the 1999 and 2000 growing seasons at two 6 year old lodgepole pine {Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) plantations (sites) in the central interior of British Columbia. The study involved a completely randomized, one-factor experimental design, with rephcation of measurements over time. The removal (i.e., brushing) of above ground competing vegetation was the factor. The design consisted of four levels of brushing (0.0 m or control (no brushing), 0.75 m, 1.0 m and 1.25 m brushing radii), replicated three times on each study site. The results indicated that HDRs increased from May or June to July and then decreased until August, remaining level thereafter. HDRs in the periods August-October 1999 for the first site, and August-September 2000 for the second site were found to be equal to or lower than those obtained in May 1999 (or September 1999 (used to approximate May 2000 measurements)). The highest HDR values were observed in control plots. The multivariate analysis of variance (MANOVA) with repeated measures indicated that brushing treatment and date of measurement had significant (p < 0.05) effects on the intra-seasonal changes in HDR. The results suggest that HDR measurements should be taken either after mid-August, or before mid-May when changes in HDR are negligible. 112 5.1 Introduction Lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) and other species of conifers are known to make one rapid flush of extension growth in spring but continue to grow in diameter until late summer or early fall (Lotan and Critchfield 1990). Apical growth (i.e., growth in height) proceeds from the initiation of extension, approximately in April or May, to the cessation of extension, approximately in June or July. Cambial growth (i.e., growth in diameter) proceeds concurrently with apical growth early in the growing season but continues much later in the growing season (Zimmerman and Brown 1971, Waring and Schlesinger 1985, Oliver and Larson 1996). Depending on the insolation levels and temperatures prevalent, cambial growth may continue into the fall. The pattern is one of growth in height proceeding faster than growth in diameter in the earlier part of the growing season, and growth in height falling behind growth in diameter in the later part of the growing season. Using HDR as a tool in choosing a level of vegetation control requires that HDR measurements be taken at a time when intra-seasonal (i.e., monthly) changes in HDR are negligible. In order to make HDR an operationally useful tool, the appropriate time within a growing season to take HDR measurements must be determined and intra-seasonal variations in HDR considered in interpreting the HDRs. This knowledge would allow forest managers to know when the least amount of change in HDR may occur within the growing season and when it is desirable to take HDR measurements. This information would contribute toward establishing trends in HDR over several years and offer forest managers a basis for making decisions on when and how to treat problems with competing vegetation in young conifer plantations. The focus of this chapter is the examination of intra-seasonal changes in HDR. The objectives are to determine: (i) the pattern of variations in HDRs through the 1999 growing season at one study site, and the 2 0 0 0 growing season at another study site, and (ii) when in the growing season changes in HDR are negligible for no removal of above ground competing vegetation (i.e., brushing) and a range of bmshing thresholds, and HDR measurements may be taken without incurring major error. 113 Three hypotheses were used to address these objectives: (i) in any one month, the mean percent change in HDR for trees brushed to a narrower brushing radius is greater than the mean percent change in HDR for trees brushed to a wider brushing radius; (ii) for a given treatment, mean percent change in HDR increases from the start (i.e.. May) to the mid-point (i.e., June-July) of the growing season; and (Hi) for a given treatment, mean percent change in HDR decreases from the mid­ point (i.e., June-July) to the end (i.e., August-October) of the growing season. 5.2 Materials and methods 5.2.1 Study sites The study involved two of the four plantations (sites) described in Chapter 2 (Section 2.2.1). The CanFor-Bednestii-Little Bobtail Lake site is located at lat. 53° 47' 24" N, long. 123° 31' 48" W; and the Fraser Lake-101 km site is located at lat. 54° 03' 01" N, long. 124° 38' 54" W. The sites included portions of the sub-boreal spruce (SBS) dry warm (dw) variant 3 subzone in the Vanderhoof Forest District (Fig. 2.1). Lodgepole pine were planted on the sites at approximately 1270 and 1390 stems/ha. At the time of installation (1998) these sites were four and five years of age; they were planted in 1994 and 1995. The sites were selected from areas in which competition, primarily from trembling aspen {Populus tremuloides Michaux), was severe. The sites ranged in elevation from approximately 755 to 854 m above sea level. Mean annual precipitation across the sites extended from 427 to 648.5 mm. Winter precipitation is relatively low, with winter snowpacks generally less than 2 m in depth (DeLong et al. 1993). The sites were prepared for planting by windrow burning followed by disc trenching. The crop trees were planted on raised spots created by the trenching, with the majority planted on the middle and top positions (McMinn and Hedin 1990). One site Was disc trenched two years prior to planting (i.e.. Little Bobtail Lake site); and all other sites were disc trenched one year prior to planting. 114 5.2.2 Experimental Design A complete description of the experimental design is provided in Chapter 2 (Section 2.2.2). Briefly, the experimental design was a completely randomized, one-factor design, with replication of measurements over time. Brushing treatment was the factor (Zar 1996). The design consisted of four levels of brushing (0.0 m or control (no brushing), 0.75 m, 1.0 m, and 1.25 m brushing radii) replicated three times on each study site (12 plots/site). Randomized plot layouts are depicted in Appendix A. A relatively homogeneous aspen- and alder-dominated area (stratum) within the cut block was selected for sampling for each site. A total of 12 plots, each 11.28 m radius (0.04 ha), were located within a 120 x 90 m (1.08 ha) study site on the selected stratum of the cut block on each study area. Brushing treatments were randomly assigned to plots. A buffer of about 7.44 m between plots was thereby established. Each plot had approximately 50 crop trees (600 trees per 12 plots). A random sample of 12 crop trees within a plot (144 trees per 12 plots) was taken in 1999 and 2000 for the purpose of assessing intra-seasonal variations in HDR in response to the four levels of brushing treatments. All treatment plots were brushed within the specified radius shortly after the plots were established. No brushing occurred within the control plots. 5.2.3 Measurements All planted lodgepole pine trees (crop trees) were identified and tagged in each plot. Crop trees were brushed in a manner, which Smith et al. (1997) have described as weeding or cleaning, but with a minor difference. Instead of employing a conventional cylindrical approach, a branch overhanging the perimeter of the brushing circle was retained as long as the stem of the competing vegetation was located outside the prescribed radius. The brushing radii of crop trees were measured between the center of the bases of the crop tree and the competing vegetation. This departure from the conventional approach was taken primarily for reasons of experimental control. For the Little Bobtail Lake site, plots were brushed June 30-July 1, 1998, and re-brushed June 21-22, 1999 and June 20, 2000. For the 101 km site, plots were brushed July 8-9, 1998, and re­ 115 brushed June 30-July 1, 1999 and June 14, 2000. Percent cover, distribution, and average height were estimated within each quadrant of a plot for aspen, alder, other deciduous shrubs, and herbaceous plants prior to brushing the sites. Herbaceous plants were identified and listed in order of thenrelative abundance also prior to brushing. Height, diameter, and leader length measurements were taken on 12 crop trees randomly selected from approximately 50 crop trees in each plot (on both sites) at approximately monthly intervals during the 1999 and 2000 growing seasons. At the Little Bobtail Lake site these measurements were taken in May, June, July, August, and October 1999. At the 101 km site these measurements were taken in September 1999 (used to approximate May 2000 measurements which were not available), and July, August, and September 2000. Additional measurements were taken at the 101 km site in May, June, and July 1999, however the absence of an August 1999 measurement prompted us to use the measurement data from 2 0 0 0 in the analysis. For the Little Bobtail Lake site, measurements relevant to the determination of intra-seasonal variations in HDR in 1999 were taken May 7, June 7, July 9, August 19-23, and October 24 of that year. Due to logistical constraints, measurements were not taken at the Little Bobtail Lake site in September 1999. For the 101 km site, measurements relevant to the determination of intra-seasonal variations in HDR in 2000 were taken September 16-20, 1999, and July 17, August 22-25, and September 26, 2000. Due to logistical constraints, measurements were not taken at the 101 km site in May and June 2000, and we found it necessary to substitute measurements taken in September 1999 for May 2000 measurements. In order to verify that September 1999 measurements approximated May 2000 measurements, other measurements taken August 14-19, 1998, and May 10, 1999 were compared. Height and leader length (cm) measurements were made with a height pole. Height was defined as the distance between the root collar (upper side of the slope at the top of the mineral soil) and the tip of the bud. Diameter (cm) measurements were made at approximately 1 cm above the root collar (above the swelling at the root collar) with electronic calipers. Diameter was measured to the 116 nearest 0.01 cm. The locations of diameter measurements were identified on the tree stem with a painted line. Damage to crop trees caused by disease, insects, and mechanical brushing devices was noted at the time of measurement. Abnormalities in the crop trees (such as multiple stems) were noted at this time. The position of each planted tree (crop tree) in relation to the raised planting spot, or berm, was also noted. The position was described as being either on the top, in the middle (mid-point), or on the bottom (in the scalp). 5.2.4 Analysis HDR was calculated for each tree by dividing the tree height (cm) by the root collar diameter (cm). The mean HDR of the 12 crop trees selected for measurement in each plot was calculated, and measurement time and site were recorded for use in the analysis. Normality of HDR data was tested using Kolmogorov-Smimov and Lilliesfors tests (StatSoft 1999); and homoscedasticity of HDR data was tested using Brown Forsythe and Levene tests (StatSoft 1999). Normality and homoscedasticity were generally achieved. The concern for the small sample size (12 trees/plot) in our study suggested use of a non-parametric procedure. However, the existing non-parametric procedures (e.g., Freidman’s method and Scheier-Ray-Hare test) were not applicable to our experimental design (D. Ayers, per s. comm. 2002; Sokal and Rohlf 1995, pp. 440-447). Thus, HDR data were ranktransformed. For each site, a multivariate analysis of variance (MANOVA) with repeated measures was conducted: first, based on non-transformed HDR; and second, based on rank-transformed HDR data using Statistica® (StatSoft 1999). Conover (1980, p. 337) advises that the parametric analysis will be valid if the ANOVA on rank-transformed data produces “nearly identical results” as the ANOVA on non-transformed data. A repeated measures MANOVA was performed using Statistica® on each site to test whether the mean HDRs of crop trees were significantly (p < 0.05) different among treatments (brushing 117 radii), months of measurement, and interaction between these two factors (StatSoft 1999). The MANOVA model used was: HDRijk= jU + treatmenti + monthj + (treatment*month)ij + % where HDRijk= plot mean HDR; // = grand mean HDR; treatmenti = brushing radius (0.0 m or control, 0.75 m, 1.0 m, 1.25 m); monthj = morAh of measurement (May, June, July, August, October 1999 for Little Bobtail Lake site, and September 1999, and July, August, September 2000 for 101 km site); (treatment*month)ij = interaction between treatment and month of measurement; and % = experimental error (i.e., error term for three plots or replicates for each treatment and month) (Johnson and Wichem 1992, p. 263-4). Comparisons between study sites were conducted on the basis of vegetation and other site characteristics. HDR measurements for the 101 km site for 2000 were used in place of those for 1999 because measurements for August 1999 were not available. Although measurements were available for May, June, July and September 1999, the absence of an August measurement for 1999, prevented an accurate description of the crucial period when HDRs decline dramatically, occurring between July and September. It seemed more important to present results that accurately described changes in HDRs during the period July to September, than those during the period May to July. It was not possible to compare HDRs for September 1999 and May 2000. However, an analysis of variance (ANOVA) indicated that mean HDRs for August 1998 and May 1999 were not significantly (p > 0.05) different for treatment, month, and interaction. Thus, it seemed that HDRs for September 1999 provided a reasonable approximation of HDRs for May 2000. MANOVA was used instead of the usual ANOVA for the overall analysis because HDRs of repeated measurements were highly correlated. MANOVA is commonly applied to repeated measures-data with several correlated dependent variables (von Ende 1993, p. 117-8). The repeated measures MANOVA provided an overall assessment of the mean HDRs of crop trees; thereby, it was possible to consider, on average, whether the factors, treatment and month, had a significant effect on 118 the mean HDRs. If the MANOVA determined that one or more of the factors had a significant overall effect, then investigating more specific effects by means of Tukey HSD post hoc procedures was justified. The MANOVA results based on rank-transformed data were very similar to those based on non-transformed data; thus, the choice was made to base the analysis on the non-transformed HDR data. There were two reasons for this decision. First, the MANOVAs based on rank-transformed data and the MANOVAs based on non-transformed data provided very similar results. Second, it was felt that the determination of percent change in HDR would provide useful results, which depended on the use of non-transformed data. Relative growth rate and simple ratios are among the techniques which can be used to compare growth characteristics across years (Kozlowski and Pallardy 1997). Percent change in HDR combines both these approaches. Although other approaches were possible (Opio et al. 2000, Opio et al. 2003), it was advantageous to use %AHDRs, and normalize HDRs to the month when growth in height and diameter had not yet begun (i.e.. May in the year for which the intra-seasonal analysis was conducted, or September in the previous growing season). This approach seemed readily understandable, and facilitated comparisons both within the intra-seasonal analysis and with the interseasonal analysis. Percent change in HDR (%AHDR) was calculated for the Little Bobtail Late site for each tree (12 trees/plot), for each month in the 1999 growing season as follows: » 100 where %AHDRi = percent change in HDR between May 1999 and June, July, August, and October 1999; and HDRmonth <= HDR in June, July, August, or October 1999. %AHDR was calculated for the 101 km site for each tree (12 trees/plot), for September 1999 and the latter months in the 2000 growing season as follows: = [(FlDR^w,, - ^ D R ^ )/ 119 * 100 where % A H D R i = percent change in HDR between September 1999 (used to approximate May 2000) and July, August, and September 2000; and HDRmonth / = HDR in July, August, or September 2000. The mean %AHDR for each plot within a site was calculated for use in the analysis. Three hypotheses, adapted from Froese’s (2000) earlier statement of these hypotheses, were tested: (i) in any one month, the mean percent change in HDR for trees brushed to a narrower brushing radius is greater than the mean percent change in HDR for trees brushed to a wider brushing radius (called “%AHDRo,om > %AHDRo.75m > %AHDRi,om > %AHDRi 25mhypothesis”; (ii) for a given treatment, mean percent change in HDR increases from the start to the mid-point of the growing season (called “%AHDRMay < %AHDRju„ < %AHDRjui hypothesis”); and (m) for a given treatment, mean percent change in HDR decreases from the mid-point to the end of the growing season (called %AHDRjui > %AHDRAug > %AHDRsep-oct hypothesis”). 5.3 Results Table 5.1 presents the results of the MANOVA on mean HDRs for Little Bobtail Lake and 101 km sites. For both sites, brushing treatment and month of measurement had a significant overall effect on mean HDRs. Interaction was non-significant (p = 0.08) for the Little Bobtail Lake site, and significant for the 101 km site. The results of the MANOVAs on percent change in HDR were nearly the same as those on mean HDRs, with only treatment for the Little Bobtail Lake site not being significant. 120 Table 5.1. MANOVAs with repeated measures showing the factors affecting mean HDRs and percent changes in HDR from May to October 1999 for CanFor-Bednestii-Little Bobtail Lake site," and Study site Effect df Effect Mean HDR F Value (Pr > F) Percent change in HDR F Value (Pr > F) CanFor-Bednestii Little Bobtail Lake Brushing Month Brushing*Month 3 4 12 Brushing Month 3 3 Brushing*Month 9 7.39 83.71 1.87 0.01 0.00 0.08 2.63 83.39 1.85 5.89 210.77 5.70 0.02 0.00 9.74 189.20 0.(X) 0 .0 0 6.42 0 .0 0 0 .1 2 0.00 0.08 Fraser Lake 101 km 0 .0 1 "Due to logistical constraints, measurements for the Little Bobtail Lake site were not taken in September 1999. MANOVAs were run on measurement taken May 7, June 7, July 9, August 19-23, and October 24, 1999. Percent changes in HDR were calculated with reference to HDRs measured May 7, 1999 (n=144/site), and mean percent change in HDR was calculated for each plot {n=12/site). *Due to logistical constraints, measurements for the 101 km site were not taken in May or June 2000. September 1999 measurements were used to approximate May 2000 measurements. MANOVAs were run on measurement taken September 16-20, 1999, and July 17, August 22-25, and September 26, 2000. Percent changes in HDR were calculated with reference to HDRs measured September 1620, 1999 (n-I44/site), and mean percent change in HDR was calculated for each plot {n=12/site). 5.3.1 CanFor-Bednestii-Little Bobtail Lake site In comparisons between treatments. Fig. 5.1 suggests that the first hypothesis might partially hold in June-October 1999. The Tukey HSD post hoc test indicated only that mean %AHDRs for the control were significantly greater than both mean %AHDRs for the 1.0 m treatment in July-October 1999 (p < 0.03), and mean %AHDRs for the 1.25 m treatment in August-October 1999 (p = 0.02). Although mean HDRs were found to be significantly (p < 0.05) different between many treatments (e.g., H D R o .o m > H D R (0 .7 5 m , 1.25m) (I.B., H D R fl.v sm = H D R i.2 5 m ) ) , tio cousisteut pattern was found in these comparisons. 121 8 3 î 2 { ■7 - « - 0 .0 0 m 0.75 m -A— 1 . 0 0 m 1.25 m 12 May-99 Jim-99 Jul-99 Aug-99 Sep-99 Oct-99 Month Fig. 5.1. Percent changes in HDR between May and October 1999 for CanFor- BednestiiLittle Bobtail Lake site. Due to logistical constraints, data were not taken in September 1999. Exact dates of measurement were May 7, June 7, July 9, August 19-23, and October 24, 1999. Percent changes in HDR were calculated with reference to HDRs in May 1999 {n=144/site), and mean percent change in HDR was calculated for each plot (n=I2/site). In order to improve clarity, error bars are not presented in the figure. In comparisons within treatments. Fig. 5.1 indicates that variations of the second hypothesis, %AHDR(May9 9 , Jun9 9 ) < %AHDRjui9 9 (i-G., %AHDRMay9 9 = %AHDRju„9 g), and third hypothesis, %AHDRju199 > %AHDR(Aug9 9 , oct99) (i.G., %AHDRAug9 9 ~ %AHDRoct99), might hold. The Tukey HSD post hoc test partially confirmed the second hypothesis, indicating that mean %AHDRs in May and June 1999 were significantly (p <0.01) less than mean %AHDRs in July 1999 for no brushing, and mean %AHDRs in June 1999 were significantly (p = 0.05) less than mean %AHDRs in July 1999 for no brushing and the 0.75 m treatment. The same test confirmed the alternate version of the third hypothesis, indicating that mean %AHDRs in July 1999 were significantly (p < 0.01) greater than mean %AHDRs in both August and October 1999 for no bmshing and all treatments. Mean %AHDRs in August 1999 were not significantly different from mean %AHDRs in October 1999. Comparisons of mean HDRs within treatments produced similar results as those for mean %AHDRs. The Tukey test indicated that mean HDRs in May and June 1999 were significantly 122 (p < 0.01) less than mean HDRs in July 1999 for no brushing, and mean HDRs in July 1999 were significantly (p < 0.01) greater than mean HDRs in both August and October 1999 for no brushing and all treatments. Mean HDRs in August 1999 were not significantly different from mean HDRs in October 1999. 5.3.2 Fraser Lake-101 km site In comparisons between treatments, Fig. 5.2 suggested that a version of the first hypothesis, %AHDRo,om > %AHDRo.7 5 m> %AHDR;,om > %AHDRi.2 5 m, held in Juiy-September 2 0 0 0 . The Tukey HSD post hoc test partially confirmed this pattern, indicating that mean %AHDRs for the 0.0 m treatment were significantly (p < 0.02) greater than mean %AHDR for both 1.0 m and 1.25 m treatments in July-September 2000. The same test indicated that mean %AHDRs for the 0.75 m treatment were significantly (p < 0.01) greater than mean %AHDR for the 1.25 m treatment in Pi C 0.00 m & 0.75 m I 1 .0 0 I m 1.25 m -12 Sep-99 Jun-00 Jul-00 Aug-00 Sep-00 Month Fig. 5.2. Percent changes in HDR for September 1999 and July to September 2000 for Fraser Lake-101 km site. Due to logistical constraints, data were not taken in May or June 2000. September 1999 measurements were used to approximate May 2000 measurements. Exact dates of measurement were September 16-20, 1999, and July 17, August 22-25, and September 26, 2000. Percent changes in HDR were calculated with reference to HDRs in September 1999 (n=144/site), and mean percent change in HDR was calculated for each plot (n=12/site). In order to improve clarity, error bars are not presented in the figure. 123 August-September 2000. The Tukey test indicated that the pattern HDRo.om > HDRojsm > HDRi.om > HDR;,25 mwas Significant (p < 0 .0 1 ) in all comparisons. In comparisons within treatments. Fig. 5.2 indicates that a variation of the second hypothesis, %AHDRsep99 < % AHDRjuioo, and third hypothesis, % AHDRjuioo > % AHDR(Augoo, octoo) (i.e., %AHDRAugoo ~ % A H D R octoo), might hold. The Tukey H S D post hoc test partially confirmed the second hypothesis, indicating that mean % A H D R s in September 1999 (equivalent to May 2000) were significantly (p = 0.01) less than mean % A H D R s in July 2000 for no brushing. The same test confirmed the alternate version of the third hypothesis, indicating that mean % A H D R s in July 2000 were significantly (p < 0.01) greater than mean % A H D R s in both August and September 2000 for no brushing and all treatments. Mean % A H D R s in August 2000 were not significantly different from mean % A H D R s in September 2000. Comparisons of mean HDRs within treatments produced similar results as those for mean %AHDRs. The Tukey test indicated that mean HDRs in September 1999 (equivalent to May 2000) were significantly (p < 0.01) less than mean HDRs in July 2000 for no brushing, and mean HDRs in July 2000 were significantly (p < 0.01) greater than mean HDRs in both August and September 2000 for no brushing and all treatments. Mean HDRs in August 2000 were not significantly different from mean HDRs in September 2000. 124 Table 5.2. Mean HDRs and standard errors o f mean (SEM)" from May to October 1999 for CanForBednestii-Little Bobtail Lake site,* and September 1999 and July to September 2000 for Fraser Lake1 0 1 km site‘s Study site Jun-99 Jul-99 or Jul-00 Aug-99 or Aug-00 Sep-00 Oct-99 61.0 (4.8) 56.2 (4.6) 61.8 (4.9) 53.0 (4.5) 62.2 (4.9) 55.9 (4.6) 60.6 (4.8) 53.0 (4.5) 66.5 (5.1) 59.2 (4.8) 63.5 (4.9) 55.3 (4.6) 60.6 (4.8) 53.2 (4.5) 56.5 (4.7) 48.9 (4.3) n.m. n.m. n.m. n.m. 59.2 (4.8) 52.4 (4.5) 56.0 (4.6) 47.8 (4.3) 0.75 60.8 (4.8) 58.2 (4.7) 1 .0 0 51.0(4.4) 1.25 49.6 (4.4) n.m. n.m. n.m. n.m. 63.0 (4.9) 59.2 (4.8) 58.9 (4.8) 59.0 (4.8) 55.7 (4.6) 55.4 (4.6) 51.1 (4.4) 47.6 (4.3) 47.3 (4.3) 48.8 (4.3) 44.8 (4.2) 44.6(4.1) n.m. n.m. n.m. n.m. Brushing radius (m) May-99 or Sep-00 CanFor-Bednestii Little Bobtail Lake {n=144/site) 0 .0 0 0.75 1.00 1.25 Fraser Lake 1 0 1 km (n=144/site) 0 .0 0 “SEM are presented in brackets following mean HDRs. *Due to logistical constraints, measurements for Little Bobtail Lake site were not taken in September 1999. Exact dates of measurement were May 7, June 7, July 9, August 19-23, and October 24, 1999. “Due to logistical constraints, measurements for 101 km site were not taken in May or June 2000. September 1999 measurements were used to approximate May 2000 measurements. Exact dates of measurement were September 16-20, 1999, and July 17, August 22-25, and September 26, 2000. Note: n.m. indicates no measurements were taken. 5.4 Discussion The relative variability in height and diameter growth, due to the time of the growing season, is explained by source-sink theory (Zimmerman and Brown 1971, Waring and Schlesinger 1985, Oliver and Larson 1996). Carbon resources are allocated to those parts of the tree that are most likely to increase the tree’s chances of survival (Waring and Schlesinger 1985). According to this theory, the cambium (i.e., growth in diameter) has a lower priority in the allocation of resources; thus it is only after apical extension (i.e., growth in height) has been satisfied that radial growth will proceed. 125 The relatively higher priority height growth has vis a vis diameter growth may be reflected in a mid­ growing season inflation in HDRs. Study site was not included as a factor in the repeated measures MANOVAs. However, ANOVAs that Jacob and Opio (2003) ran on control portions (n ~ 36/site) of the same data sets indicated that HDRs for the two sites were not significantly different. In evaluating the three hypotheses, it is probable that the response to brushing treatments (initially undertaken between June 30 and July 9, 1998) was variable between the two study sites. First, we were viewing responses to brushing treatments over different growing seasons: 1999 for the Little Bobtail Lake site, and 2000 for 101 km site. Thus, %AHDRs were being observed one year (1999) and two years (2000) after the initial treatments. HDRs typically responded more strongly in the first year than in the second year after initial brushing (Chapter 2 (Section 2.3)). Seasonality may also have been a factor in the different responses observed. Second, responses to treatments were being viewed for sites where the initial densities of competing vegetation in treatment plots, initial growth rates of crop trees, micro-sites, and other factors differed. ANOVAs and regressions (Chapter 2 (Section 2.4)) indicated that the 101 km site was more homogeneous (i.e., more uniform) with respect to aspen percent cover than was the Little Bobtail Lake site. The greater homogeneity of competing vegetation at the 101 km site, may indicate greater nutrient-richness at this site. Third, the growing periods for the sites assessed probably differed. These would have been variably affected by elevation, number of frost free days, aspect, slope, light availability, water and mineral nutrients, site preparation, planting position, and compaction of soils (Zimmerman and Brown 1971, Mustard and Harper 1998). Consideration of the first hypothesis indicates that key separations in the data were between mean %AHDRo.om, %AHDRo.75m, and %AHDR(i.om, 1.25m) in August and October 1999 for the Little Bobtail Lake site; and mean %AHDR(o.om, 0,75m) and %AHDR(i.om, 1.25m) in August and September 2 0 0 0 126 for the 101 km site. The first hypothesis did not hold in its entirety, however a substantial separation in the data between the control and various treatments remained. In evaluating the second hypothesis, it seems that: (i) significant increases in HDR were obtained only for no brushing, (ii) %AHDRs peaked approximately in July for both sites, and (Hi) %AHDRs peaked at a substantially higher level for the Little Bobtail Lake site (+9.1%) than the 101 km site (+3.7%). Consideration of the third hypothesis indicates that: (i) significant decreases in HDR were obtained for all treatments, (ii) %AHDRs in August were not significantly different from %AHDRs in September or October, and (Hi) although %AHDRs for the Little Bobtail Lake site peaked at a substantially higher level than that for the 101 km site, %AHDRs appeared to equalize by the end of the growing season. An abbreviated version of the second hypothesis (%AHDRMay-jun < %AHDRjui (i.e., %AHDRMay ~ %AHDRjun)) held for the control only. For the Little Bobtail Lake site, at least, this indicated considerable latitude as to when HDR measurements could be taken early in the growing season. Certainly, HDR measurements could be taken up to mid-May, and possibly later. It seems that the downward trend in HDRs between 1998 and 2000 (as expressed by inter-seasonal changes in HDRs addressed in Chapter 2 (Section 2.3)) dampened the inflation in HDRs (i.e., %AHDRs in July 1999 and 2000) for the 1.0 m and 1.25 m treatments, but not for the control. The maximum inflation in HDRs occurred specifically for the control, an important outcome that needs to be considered if it is found operationally necessary to take HDR measurements in June or July. The considerably greater inflation in HDRs at the Little Bobtail Lake site than at the 101 km site (i.e., %AHDRs for the control in July 1999 vs. July 2000), may have been due to a seasonality effect since mean %AHDR in July 1999 for the 101 km site (part of the analysis not reported in this chapter) was 10.6%. This outcome illustrates how seasonality would be a factor were it necessary to take HDR measurements in June or July. The maximum inflation in HDRs appeared to occur in mid-July (i.e., measurements were taken July 9, 1999 and July 17, 2000 for the two sites). However, exactly when this peak occurs 127 probably varies depending on the year measurements are taken. The occurrence of this peak depends on growing conditions in the previous growing season (i.e., affecting height increment), and present growing season (i.e., affecting both height and diameter increment). Certainly, HDR measurements should not be taken in June or July without the inflation in HDRs being taken into account. A simplified version of the third hypothesis (i.e., %AHDRj„i> % AHDRAug-sep oroct (i.e., %AHDRAug ~ %AHDRsep-oct)) held for the control and all treatments. This indicated substantial latitude as to when HDR measurements could be taken later in the growing season. No significant error would be incurred if HDR measurements were taken after mid-August. However, seasonality may affect how much further diameter increment continues past mid-August in the growing season. The downward trend in HDRs between growing seasons (i.e., 1998-1999 and 1999-2000) may have been a factor in producing the asymmetry between the second and third hypotheses: a significant rise in HDRs for the control only vs. a significant decline in HDRs for the control and all treatments. 5,5 Conclusions and recommendations The main purpose of the research described in this chapter was to investigate intra-seasonal changes in HDR. Results from the 1999 and 2000 data suggest that field personnel would be able to take HDR measurements in both early spring (up to mid-May) and early fall (after mid-August) without needing to account for the inflation in HDRs occurring in June and July. Early spring and early fall are the time periods when changes in HDRs are negligible. If forest managers find it operationally necessary to take HDR measurements when HDRs are inflated (i.e., June and July), then it will be necessary to correct the inflated measurements to equivalent end of growing season measurements. Results of this study suggest that seasonality influences intra-seasonal variations in HDRs. In addition, size (absolute height and diameter) of trees is a factor that influences the degree of inflation in HDRs. Iri order for forest managers to be able to 128 take HDR measurements outside the ideal periods of measurement, better information on how HDRs vary within a growing season is required. The results of the intra-seasonal part of the study suggest that trends in HDRs should be observed: (i) for a longer period (e.g., six years instead of one or two years), and (ii) at more frequent intervals between measurements (e.g., every two weeks instead of monthly) in order to more adequately gauge intra-seasonal variations in HDRs. This information is necessary in order that HDRs may become a practical tool for forest managers to make brushing decisions. It is recommend that: (i) HDR measurements be taken on two or more sites similar to those that were studied, on trees both similar to and larger in size than those we have studied, at two week intervals, over two or more years; and (ii) conversion factors be developed that permit HDR measurements taken in June or July to be converted to the equivalent end of season measurements. 129 Literature cited Conover, W J. 1980. Practical Nonparametric Statistics 2"** ed. John Wiley and Sons, Inc., New York. DeLong, S.C., Tanner, D., and lull, M.J. 1993. A Field Guide for Site Identification and Interpretation for Southwest Portion of Prince George Forest Region. Land Management Handbook 24, Research Branch, Ministry of Forests, Province of British Columbia, Victoria, B.C. Ende, C.N. von 1993. Repeated-measures analysis: growth and other time-dependent measures. In Design and Analysis of Ecological Experiments. Edited by S.M. Scheiner, and J. Gurevitch. Chapman & Hall, New York. pp. 113-137. Froese, K. 2000. Height to diameter ratios as an indicator of competitive stress in lodgepole pine (Pinus contorta var. latifolia) under four brushing treatments. BSc. (NRM-Forestry Major) Professional Report. Univ. of Northern British Columbia, Prince George, BC. Jacob, N., and Opio, C. 2003. Correction factors for mid-growing season measurements of height to diameter ratios in young lodgepole pine and white spruce plantations in the Vanderhoof Forest District of British Columbia. Report prepared for West Fraser Mills Ltd. Univ. of Northern British Columbia, Prince George, BC. Johnson, R.A., and Wichem, D.W. 1992. Applied Multivariate Statistical Analysis. Prentice Hall, Englewood Cliffs, NJ. Kozlowski, T.T., and Pallardy, S.G. 1997. Physiology of Woody Plants. Academic Press Inc., Toronto. Lotan, I.E., and Critchfield, W.B. 1990. Lodgepole pine. In Silvics of North America, Vol. Conifers. Edited by R.M. Bums, B.H. Honkala, B.H. USDA For. Serv., Agric. Hndbk. 654, Washington, DC. McMinn, R.G., and Hedin, I.P. 1990. Site Preparation: Mechanical and Manual. In Regenerating British Columbia’s Forests. Edited by Lavender, D.P., Parish, R., Johnson, C M., Montgomery, G., Vyse, A., Willis, R.A., and Winston, D. University of British Columbia Press, Victoria, B.C. pp. 150-163. Mustard, J., and Harper, G. 1998. A summary of the available information on height to diameter ratio. BC Ministry of Forests. Victoria, BC. Oliver, C D., and Larson, B.C. 1996. Forest Stand Dynamics. McGraw-Hill, New York. Opio, C., Diest, K. van, and Jacob, N. 2003. Intra-seasonal changes in height to diameter ratios for lodgepole pine in the central interior of British Columbia. West. J. Appl. For. 18(l):52-59. Opio, C., Jacob, N., and Coopersmith, D. 2000. Height to diameter ratio as a competition index for young conifer plantations in northem British Columbia, Canada. For. Ecol. and Manage. 137: 245-252. Smith, D.M., Larson, B.C., Kelty, M.J., and Aston, P.M.S. 1997. The Practice of Silviculture: Applied Forest Ecology. John Wiley & Sons, New York. 130 Sokal, R.R., and Rohlf, F J. 1995. Biometry 3rd ed. W.H. Freeman and Co., New York. StatSoft Inc. 1999. STATISTICA® Version 5.5, Tulsa, OK. Waring, R.H., and Schlesinger, W.H. 1985. Forest Ecosystems: Concepts and Management. Academic Press, Toronto, Ont. Zar, J.H. 1996. Biostatistical Analysis. Prentice-Hall, Inc. Upper Saddle River, NJ. Zimmerman, M.H., and Brown, C.L. 1971. Trees: Structure and Function. Springer-Verlag New York Inc., New York, 131 CHAPTER 6 SYNOPSIS AND MANAGEMENT IMPLICATIONS OF HEIGHT TO DIAMETER RATIOS IN LODGEPOLE PINE STUDY The purpose of this thesis is to provide information on height to diameter ratio (HDR) in young lodgepole pine (Pinus contorta Dough ex Loud. var. latifolia Engelm.) plantations that will help in determining the feasibility of HDR as a competition index. Through this undertaking, knowledge of the management implications of HDR was obtained. The following synopsis of research findings addresses implications of the use of HDRs for lodgepole pine, and suggests directions for future research. The presentation in this chapter is organized by the studies (i.e.. Chapters 2-5) from which the management imphcations are derived. The following discussion derives from the first study (Chapter 2). In determining how HDRs of lodgepole pine respond to different levels of removal (i.e., brushing) of competing vegetation applied to crop trees over time, a systematic pattern in the mean HDRs and percent changes in HDR (%AHDR) was evident. For most sites, the pattern HDRgg > HDR^g, oo) (i.e., HDR 9 9 ~ HDRqo) was apparent for some or all of the brushing treatments. The pattern %AHDR9 g > %AHDR(9 9 ,oo> (i.e., %AHDRq9 ~ %AHDRqo) held for the 1 . 0 m and 1.25 m treatments for most sites. These results indicate that (i) the impact of brushing interventions can be measured in a relatively short span of time (i.e., one year after the treatment) in young lodgepole pine stands, and (ii) the effect of brushing on HDR with the smaller lodgepole pine investigated in the study (i.e., mean heights of 100-190 cm) becomes appreciable with brushing radii > 1 . 0 m. A noticeable rise in HDRs (i.e., 2000-2001) following the initial dechne in HDRs (i.e., 19992000) was observed for one site with larger trees (i.e., 1116 km site). This suggested that the medium term effect (i.e., > 10-15 years old) of brushing interventions may not be stable (i.e., a possible reversal of the initial decline in HDRs may occur 10-15 years after the trees were planted). The possibility of a reversal in HDRs as they were measured in 2000 and 2001 led to the research recommendation that HDR measurements be taken approximately six years after the initial 132 installations (i.e., 2003 or 2004). This information would provide a check on the relative stability of the HDRs from which reference HDRs (i.e., HDR thresholds) were determined. This information could also be used for the assessment of older stands. For most sites, the patterns %AHDRo.om > %AHDR(o.75m, i.om, 1.25m) (i.e., %AHDRo.7Sm = %AHDRi.Om ~ %AHDRi,25m) or %AHDR(O.Om, 0.7Sm) > %AHDR(i.Om, 1.25m) (i.e., %AHDRo.75m~ %AHDRi.om and %AHDRi.om ~ %AHDRi.25m) held in the second and/or third year of measurements (1999 and/or 2000, or 2000 and/or 2001). For sites with larger trees, the effect of the 0.75 m brushing radius was indistinguishable from no brushing. The effect of brushing on HDRs was greater for the sites with smaller trees (i.e.. Little Bobtail Lake and 101 km sites) than for the sites with larger trees (i.e., 137 km and 1116 km sites). This outcome should indicate to forest managers that brushing interventions need to be undertaken earlier (i.e., < 4-5 years after planting), rather than later (i.e., 9-10 years after planting) in the life of a plantation. This recommendation is consistent with the findings of Wagner et al. (1999), Wood and von Althen (1993), and Newton and Freest (1988). Reference HDRs that apply to plantations similar to the study sites were recommended. Using change in slope of %AHDR as a criterion, a “satisfactory estimate” of the reference HDR (i.e., ranges of HDRs) was obtained for the Little Bobtail Lake (40-49) and 137 km (45-54) sites; and a “tentative estimate” of the reference HDR was obtained for the 101 km (40-51) and 1116 km (38-47) sites. See Table 2.3 (Chapter 2). It is not certain that HDRs will hold at levels from which the HDR thresholds were determined. Thus, an important research recommendation is that reference HDRs be re-evaluated in order to be able to assess a possible reversal in the downward trends presently observed. Reference HDRs were recommended for specific vegetation complexes, BEG classifications, ranges of percent cover competing vegetation, and ranges of mean diameter during specified years following planting. The ranges of mean diameters were broadly defined as 1.704.20 cm for Little Bobtail Lake site, 1.50-4.10 cm for 101 km site, 4.30-8.90 cm for 137 km site, and 6.90-9.90 cm for 1116 km site. See Table 2.3 (Chapter 2). The optimum brushing radius for all sites was found to be in the range of 1.0-1.25 m. 133 Generally, the 0.75 m brushing treatment had no effect on HDRs. In returning to the study sites two years after completion of field measurements (i.e., September 2002), it was apparent that the maximum brushing radius was not equivalent to a total removal of non-crop species. Depending on how the brushing is undertaken in an operational setting, 1.00-1.25 m brushing treatments may leave behind approximately 700-1400 stems/ha of aspen and/or birch (i.e., growing among approximately 1270-1460 stems/ha of crop trees). This should indicate to forest managers that brushing to a 1.0 m or 1.25 m radius would in time produce mixedwood plantations. Two patterns seemed apparent when considering the vegetation complex and maturity of trees at the sites. First, at relatively heterogeneous (i.e., possibly nutrient-poorer) sites such as Little Bobtail Lake and 137 km sites, mean HDRs were higher for treatments/plots where aspen percent cover was higher (i.e., a pattern not evident for the more homogeneous and possibly nutrient-richer 101 km and 1116 km sites). Second, the absolute value of mean %AHDRs was lower for the sites with larger trees (i.e., 137 km and 1116 km sites) than for the sites with smaller trees (i.e.. Little Bobtail Lake site and 101 km sites). This may be due (/) to the fact that larger trees have responded to the reduction in competition with neighbouring non-crop vegetation by lowering their HDRs, and (ii) larger trees are slower to respond to changing competition conditions as are introduced by brushing interventions. The implications for management of the patterns of heterogeneity/homogeneity and size of trees on sites are that (1) factors such as nutrient quality of a site (i.e., reflected in homogeneity/ heterogeneity) need to be taken into account when planning brushing interventions, and (2 ) a general downward trend in HDRs seems to be indicated between 5 and 10 years after establishment of the plantations. The following is a synopsis of results from the second study (Chapter 3). Retrospective analyses of HDRs augmented the overall HDR research project by describing trends in HDRs between the time of planting of crop trees and the time of brushing. A variety of pre-treatment patterns of HDRs (i.e., determined from measurements of inside bark diameter (H D R ibs)) were 134 evident. HDR^s either: (i) declined steadily from the time of planting (e.g., 1994) to the initiation of treatments (1998); (ii) declined dramatically in the two years following planting (e.g., 1995-1996), but then rose steadily to the initiation of treatments (1998); (in) declined dramatically in the four years following planting (e.g., 1991-1993), and remained relatively constant thereafter; or (iv) remained relatively constant over the entire measurement period (1991-1999). Each site was approximately described by one of these patterns. Thus, rather than finding one pattern that repeated itself between study sites, a substantial variation in patterns was observed. Possible relationships between pre-treatment (before brushing) and post-treatment (after brushing) HDRs were examined, however, no relationship was found. Whatever the pre-treatment pattern of HDR^s, the post-treatment pattern was consistently %AHDRiBo.om > %AHDRiBo.7 5 m> %AHDRiBi.om > % AHDRib 1.25m- The HDR^s of trees prior to treatments did not seem to affect the response of the trees to brushing. The contrary was observed: brushing treatments were found to be a powerful tool for manipulating HDRs. The results certainly indicated that the removal of competing vegetation (i.e., brushing) reduced HDRs. The extent to which this reduction in competition led to a reduction in the competition for light requires further research (Mustard and Harper 1998). The time period (years after planting) when brushing should be undertaken in plantations similar to the study sites was not determined in the second study. No pre-treatment pattern of % A H D R ibS emerged as a general rule. The second of the patterns described, where % A H D R ibS declined dramatically in the two years following planting (e.g., 1995-1996), but then rose steadily to the initiation of treatments (1998), was only inadequately the pattern for one site. Thus, no definite management recommendation emerged from this study as to the appropriate time to brush plantations. Based on the results of Chapters 2, however, it seemed that brushing interventions should be undertaken < 4-5 years after planting. This recommendation seems consistent with the findings of Wagner et al. (1999), Wood and von Althen (1993), and Newton and Freest (1988). It is not known what specific factors produced the four patterns of competition development (i.e, patterns of pre- 135 treatment HDR ibs). Thus in future research, it is recommended that changes in the vegetation complex be tracked along with changes in HDRs. The following synopsis and consideration of management implications derives from the third study (Chapter 4). This study was the outcome of regression models of stem volume that were developed from retrospective analyses of stem discs, and applied to the field-based measurements. Investigation of the response of stem volume increment to brushing treatments was undertaken on the basis of these regression models. A systematic pattern in the mean stem volumes and percent changes in stem volume (%AVol) was evident. In examining the response of stem volume increment to various bmshing treatments in the period 1998-2000 (or 1999-2001), it was determined that trees brashed to a wider bmshing radius generally had a greater increment in stem volume than trees bmshed to a narrower bmshing radius. This pattern held to a considerable extent in 2000 for sites with smaller trees, and to a lesser extent in 2000 (or 2001 for the 1116 km site) for sites with larger trees. Nevertheless, mean stem volumes and %AVols increased over time at all sites. Variations of the pattern Voli^m > Voli.om > Volojsm > Volo.om were found to hold for Little Bobtail Lake, 101 km, and 137 km sites in 2000. These varied from Voli.asm > Vol(o.75m,o.om) (i.e., Volo.75m ~ Volo.om) at the Little Bobtail Lake site, to Vol(i.25m, i.om) > Vol(o.75m,o.om) (i.e., Vol,.25m~ Voli.om and Volojsm ~ Volo.om) at the 1 0 1 km site, to Vol,,om > Volo.7 5 m> Volo.omat the 137 km site. Variations of the pattern %AVoli,25m > %AVoli.om > %AVolo.75m> %AVolo.om were found to hold for the Little Bobtail Lake and 101 km sites in 2000. These varied from %AVoli.25m > %AVol(i.om, 0,75m) > %AVolo.om (i.e., %AVoli.om = %AVolo.75m) at the Little Bobtail Lake site, to %AVol(i.25m, i.om) > %AVol(0.75m,0.0m) (i.e., %AVoli.25m ~ %AVoli,om and %AVolo,75m~ %AVolo.om) at the 1 0 1 km site. The observed patterns indicated that a bmshing effect on stem volume was more likely to hold for sites with smaller trees (i.e.. Little Bobtail Lake and 101 km sites) than for sites with larger trees (i.e., 137 km and 1116 km sites). This may be due to the fact that larger trees are slower to respond to changing competition conditions that are introduced by bmshing interventions. 136 Ranges of mean stem volumes were determined within which reference HDRs are meant to be applied. The ranges of mean stem volume were broadly defined as 100-750 cm^ for Little Bobtail Lake site, 50-750 cm^ for 101 km site, 1500-8000 cm^ for 137 km site, and 4000-10000 cm^ for 1116 km site. See Table 4.4 (Chapter 4). The similarity in the patterns observed between mean stem volumes and diameters, and between mean %AVols and percent changes in diameter, indicated that ranges of mean diameter may be used in operational situations to delimit the apphcation of the reference HDRs for areas where stem volume equations are not available. Where stem volume equations are available, these should be used to determine ranges of mean stem volumes within which reference HDRs are recommended. The use of diameter and stem volume as indices to be used with HDR seems consistent with Wagner et al.’s (1999) use of the stem volume index along with HDR. Extensive analysis of the error in regression models of stem volume indicated a probable maximum error of < 10% in the stem volumes produced by these models. This level of error could be improved in future research by obtaining stem discs from 30 cm height above the root collar on all destructively sampled trees. Another research recommendation is that the determination of stem volumes be extended back into the period prior to treatments by obtaining stem discs from the whorls from which total heights were estimated. The following discussion derives from the fourth study (Chapter 5). Its main purpose was to investigate intra-seasonal changes in HDR. The pattern of variations in HDRs was determined for the sites with smaller trees: Little Bobtail Lake site through the 1999 growing season, and 101 km site through the 2000 growing season. Results from the 1999 and 2000 data suggested that field personnel would be able to take HDR measurements in both early spring (up to mid-May) and early fall (after mid-August) without needing to account for the inflation in HDRs occurring in June and July. Early spring and early fall are the time periods when changes in HDRs are negligible. The maximum inflation in HDRs occurred specifically for the control, an important outcome that needs to be considered if it is found operationally necessary to take HDR measurements in June 137 or July. If forest managers find it necessary to take HDR measurements when HDRs are inflated, then it will be necessary to correct the inflated measurements to equivalent end of growing season measurements. It was found that seasonality influences intra-seasonal variations in HDRs. The considerably greater inflation in HDRs at the Little Bobtail Lake site (i.e., 9.1% in July 1999) than at the 101 km site (i.e., 3.7% in July 2000), may have been due to seasonal differences since mean %AHDR in July 1999 for the 101 km site was closer to that for the Little Bobtail Lake site (i.e., 10.6% in July 1999). In addition, size (i.e., absolute height and diameter) of trees is a factor that influences the degree of inflation in HDRs: the inflation in HDRs will be less for larger trees than for smaller trees. In order for forest managers to be able to take HDR measurements outside the ideal periods of measurement, better information on how HDRs vary within a growing season is required. The results of the intra-seasonal analyses suggest that trends in HDRs should be observed: (i) for a longer period (e.g., six years instead of one or two years), and (ii) at more frequent intervals between measurements (e.g., every two weeks instead of monthly) in order to more adequately gauge intra-seasonal variations in HDRs. This information is necessary in order that HDRs may become a practical tool for forest managers to make brushing decisions. Recommendations for future research are that: (i) HDR measurements be taken on two or more sites similar to those we have studied, on trees both similar to and larger in size than those we have studied, at two week intervals, over two or more years; and (ii) conversion factors be developed that permit HDR measurements taken in June or July to be converted to equivalent end of growing season measurements. 138 Literature cited Mustard, J., and Harper, G. 1998. A Summary of the Available Information on Height to Diameter Ratio. B.C. Ministry of Forests. Victoria, BC. Newton, M., and Freest, D.S. 1988. Growth and water relations of Douglas-fir {Pseudotsuga menziesii) seedlings under different weed control regimes. Weed Sci. 36: 653-662. Wagner, R.G., Mohammed, G.H., and Noland, T.L. 1999. Critical period of interspecific competition for northem conifers associated with herbaceous vegetation. Can. J. For. Res. 29: 890-897. Wood, I.E., and Althen, F.W. von 1993. Establishment of white spruce and black spruce in boreal Ontario: effects of chemical site-preparation and post-planting weed control. For. Chron. 69: 554-560. 139 M 1.25 m 0.0 m 0.0 m 0.75 m 1.25 m 1.0 m ■> < 3.72 m <■-7.44 m ■-> 1.0 m 1.25 m 0.0 m 0.75 m 1.0 m CanFor-BednestüLittle Bobtail Lake site Slope: 4-12% Aspect: SW <-- 11.28m 120 m 0.0 m 0.75 m 1.25 m 1.0 m 1.0 m 1.25 m 0.0 m 0.75 m 10 0.75 m 0.0 m 1.0 m 1.25 m Fraser Lake-101 km site Slope; 0-6% Aspect: NW <-• 7.44 m 11.28m 120 m Appendix A. Randomized plot layouts for all study sites. Levels of removal of competing vegetation, 0.0 m or control (no bmshing), and 0.75 m, 1.0 m, and 1.25 m bmshing radii, are indicated inside plots. 140 V — ^ /O.O m \ 1 "— ^ l.Qm^ y ^ /y0.75 m \ ^ » V ____ / ____/ / 1.0m \ s. / 1.25 m \ / 1.25 m ^ ------ . /O.O m \ /0 .7 5 m \ a .. .9 . ^ __ 1 1 / ------ N . /0 .7 5 m \ 12y----- ^/ — X 0.0 m \ f 0.75 m^ \ — X y/ 1.0 m / 1 .25 m 'v / o . Q m \ Fraser Lake-137 km site Slope: 0-16 % Aspect: SE Lv ..Y... : ■> '< 3.72 m > a <■-7.44 m <•■ 11.28 m ••> ■120 m / 1.0 m ^ \ /1 .2 5 m \ f -------- / 8 ___ /0 .7 5 m \ » ^ /— X /O.O m \ \ — 6 y ____ / 1 .25 m ^ y — X Xl.O m \ ? a N .i /M.25 m : a > < 3.72 m ^ — X / 0.75 m \ 1 1 / ----- X / 1.0m \ 12y----/0 .0 m \ w 0 0 Fraser Lake-1116 km site Slope: 13-31% Aspect: N 7.44 m < 11.28 m 120 m Appendix A. Randomized plot layouts for all study sites (continued). Levels of removal of competing vegetation, 0.0 m or control (no brushing), and 0.75 m, 1.0 m, and 1.25 m brushing radii, are indicated inside plots. 141