Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Pattern and process in old-growth temperate rainforests of southern British Columbia Arsenault, André 1995

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

Item Metadata


831-ubc_1995-982470.pdf [ 15.27MB ]
JSON: 831-1.0088822.json
JSON-LD: 831-1.0088822-ld.json
RDF/XML (Pretty): 831-1.0088822-rdf.xml
RDF/JSON: 831-1.0088822-rdf.json
Turtle: 831-1.0088822-turtle.txt
N-Triples: 831-1.0088822-rdf-ntriples.txt
Original Record: 831-1.0088822-source.json
Full Text

Full Text

PATTERN AND PROCESS IN OLD-GROWTH TEMPERATE RAINFORESTS OF SOUTHERN BRITISH COLUMBIA. BY ANDRE ARSENAULT B.Sc., Universite du Quebec a Montreal, 1987 M . S c , Universite du Quebec a Montreal, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE FACULTY OF GRADUATE STUDIES (DEPARTMENT OF BOTANY)  We accept this thesis as conforming to the required, standard  THE UNIVERSITY OF BRITISH COLUMBIA APRIL 1995 (c) Andre Arsenault, 1995  In presenting  this  thesis  in  degree at the University of  partial fulfilment  of  the  requirements  for  an advanced  British Columbia, I agree that the Library shall make it  freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or  by  his  or  her  representatives.  It  is  understood  that  copying  or  publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of  )y^s<7,  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  ii ABSTRACT I examined the effects of natural and anthropogenic disturbances on the patterns of species composition, tree regeneration, and forest architecture in old-growth forests from southern coastal British Columbia.  This study was conducted in  the submontane portion of the Coastal Western Hemlock Zone in the Vancouver watersheds, Pacific Spirit Park, and Clayoquot Sound. In the greater Vancouver area, old forests (250 yr) exhibited greater structural and compositional heterogeneity than young (31-60 yr) and mature (61-80 yr) forests.  Size class  distributions of living and dead standing trees in the three age groups suggested both qualitative and quantitative differences in regeneration and mortality processes.  The canonical correlation  between structure and composition was high (Rc = 0.84) but a substantial amount of total variation the analysis.  remained unexplained by  PCA (principal component analysis) axis 1 of  species composition separated the lower elevation (warmer/drier) mature forests from the higher elevation (cooler/wetter) young and old forests.  PCA axis 1 of structure separated the young and  mature forests from the old forests. Stand history reconstructions in the Capilano watershed and Clayoquot Sound indicated that frequent small-scale disturbances (0.1-0.2 treefalls/year) and relatively slow growth rates explain the relatively open character and complex architecture of oldgrowth cedar-hemlock forests. All three study plots exhibited reverse-J tree size distributions considered indicative of climax or steady-state conditions; however, age structures showed important differences.  One plot originated from a fire 300 yr  iii ago while the other two were over 1000 years old and showed no signs of catastrophic disturbances. Gap-phase dynamics may influence patterns of tree regeneration if small-scale disturbance events are relatively close in space and time.  However, the spatial pattern of  understory trees was not significantly correlated with canopy structure.  Thus, the role of gap-phase dynamics appears to be  more important as a release mechanism for suppressed trees already established in the understory. The presence of standing water had a strong influence on the spatial pattern of understory plant communities.  In addition,  bryophyte species composition was related to stand structure and dynamics as evidenced by a succession gradient on wood. Except for Douglas-fir, which appears to require large scale disturbances for its regeneration, all other tree species examined are well adapted to a range of disturbances.  This  indicates that coastal silviculture could use a diversity of cutting methods analogous to natural disturbances instead of relying solely on clearcutting.  The estimated forest turnover  time varied between 375 and 1096 yr, based on the present proportion of area in canopy gaps (39%), and the estimated time to fill canopy gaps.  Tree ring signatures revealed 208 yr-326 yr  are needed for small disturbances to have a noticeable effect on tree growth for an area equivalent to the size of the study plots (1/2 ha). At the landscape level, the main catastrophic disturbance in the Capilano watershed before logging was fire.  Large scale  fires were associated with warm south facing slopes and occurred  iv at low frequencies (ca 600 years) . The fire regime also coincided with periods of low sunspot activity.  Logging has  significantly changed the ecology of disturbance and patch dynamics in the Capilano watershed.  A slow recovery process  following the impact from human disturbances and a trend towards unprecedented levels of forest fragmentation supports concerns regarding conservation in our coastal forests.  V  RESUME J'ai examine les effets des perturbations naturelles et anthropiques sur les patrons de composition vegetale, de regeneration des coniferes et d7architecture forestiere dans les forets anciennes de la cote ouest de la Colombie-Britannique. Cette etude fut realisee dans la portion inferieure de la Zone cotiere de la Pruche de 1'Ouest dans les bassins de drainage du grand Vancouver, le Pare Pacific Spirit et dans le detroit de Clayoquot. Dans le grand Vancouver, les forets anciennes, agees de 250 ans et plus, ont presente une heterogeneite de structure et de composition superieure aux jeunes forets (agees entre 30 et 60 ans) et aux forets matures (agees entre 61 et 80 ans). Les distributions de classes de diametre d'arbres vivants et d'arbres morts debout dans les forets d'ages differents revelent des divergences qualitatives et quantitatives quant aux processus de regeneration et de mortalite. La correlation cannonique entre la structure et la composition est elevee (Rc = 0.84) mais une partie substantielle de la variation totale est restee inexpliquee.  Le premier axe de l'ACP (analyse en composantes  principales) sur la composition vegetale designe un gradient ecologique qui est represents par les forets matures a basse altitude d'un cote et par les forets jeunes et anciennes, generalement  a altitude plus elevee, de 1'autre. Les forets  jeunes et anciennes, de par leur altitude plus elevee, ont un climat plus frais et humide comparativement aux forets matures. Le premier axe de l'ACP sur la structure a separe les forets de seconde venue des forets anciennes.  vi La reconstruction historique de peuplements dans la vallee de la riviere Capilano et dans le detroit de Clayoquot a demontre que les perturbations frequentes (0.1-0.2 chablis/annees) et une croissance relativement lente expliquent le caractere ouvert et 1'architecture complexe des forets anciennes dominees par la Pruche de l'Ouest et le Cedre de l'Ouest.  Une structure de  diametre en forme de J inverse dans les trois parcelles d'echantillonnage reflete un stade climax et un certain etat d'equilibre ecologique.  Cependant les structures d'age ont  revele des differences importantes.  Une foret etait issue d'un  feu, il y a deja plus de 300 ans. Les deux autres forets etaient tres vielles ne montrant aucun signe de perturbation majeure depuis plus de mille ans. Les chablis peuvent influencer la regeneration des coniferes si les perturbations sont relativement rapprochees dans le temps et l'espace.  Cependant la repartition spatiale des arbres dans  le sous-etage arborescent n'est pas correlee au couvert forestier.  La dynamique associee aux chablis joue quand meme un  role important en favorisant les reprises de croissance des individus supprimes dans le sous-etage arborescent. La repartition spatiale des communautes vegetales au niveau des etages inferieurs de la foret etait etroitement relies a la presence de petits ruisseaux.  De plus les communautes de  bryophytes sont intimement associees a la dynamique forestiere, car un gradient refletant la preferences pour du bois de differents niveaux de pourriture a pu etre identifie. A 1'exception du Sapin de Douglas, qui requiert de grandes perturbations pour assurer sa regeneration, les autres essences  vii etudiees peuvent s'adapter a plusieurs types de perturbations. Une plus grande diversite de techniques silvicoles, analogues aux perturbations naturelles, pourrait done etre utilisee avec succes aux lieu de pratiquer la coupe a blanc de facon absolue. estime  J'ai  que la periode de rotation naturelle des chablis varient  entre 375 ans et 1096 ans selon la proportion actuelle de la foret en chablis (39%) et selon des estimations de temps a remplir les chablis. Une analyse dendrochronologique a revelee qu'une periode de 208 a 326 ans est requise pour affecter la croissance des arbres sur une surface equivalente a celle des parcelles etudiees (1/2 ha). Au niveau du paysage, le feu etait la perturbation catastrophique dominante, avant 1'exploitation forestiere, dans la vallee Capilano.  Les grands feux etaient associes aux  versants sud et survenaient a des intervalles de 600 ans. Le regime des feux coincide egalement avec des periodes de faible activite des taches solaires.  L'exploitation forestiere a  modifier considerablement le paysage et le regime des perturbations de la vallee Capilano.  Le retablissement lent qui  s'opere apres une coupe forestiere et les niveaux de fragmentation du paysage sans precedent confirment certaines inqietudes pour la conservation de nos forets cotieres.  TABLE OF CONTENTS ABSTRACT RESUME TABLE OF CONTENTS LIST OF TABLES  ii V viii X  LIST OF FIGURES  xii  ACKNOWLEDGEMENTS  XV  DEDICATION 1. INTRODUCTION 1.1 Overview 1. 2 What is an old-growth forest? 1. 3 Tools for dynamic plant ecology 1.4 General concepts in dynamic plant ecology 1.5 Community dynamics in temperate rainforests 1. 6 Objectives  xvii 1 2 4 6 7 9  2. STUDY AREA 2.1 Physiography and geology 2 . 2 Surf icial geology and soils 2 . 3 Climate 2 . 4 Vegetation 2 . 5 Disturbance regime  12 13 13 15 17  3. METHODS 3.1 Structural-compositional variation in young, mature, and old temperate rainforests of southern coastal British Columbia 3.1.1 Sampling 3.1.2 Data analysis  19 19 22  3.2 Structure and dynamics of three old-growth cedar-hemlock forests in southern coastal British Columbia 3.2.1 Sampling 3.2.2 Forest architecture 3.2.3 Plant species composition 3.2.4 Dendrochronology 3.2.5 Disturbance regime 3.2.6 Spatial pattern of vegetation  24 24 24 26 27 28 29  3.3 The ecology of disturbance and patch dynamics in a temperate rainforest landscape 3.3.1 Sampling 3.3.2 Data analysis  32 32 34  ix 4.RESULTS 4.1 Structural-compositional variation in mature, and old temperate rainforests southern coastal British Columbia 4.1.1 Tree size structure 4.1.2 Structure-composition correlation: age classes 4.1.3 Structure-composition correlation: age classes  young, of 36 36 Combined 38 Separate 42  4.2 Structure and dynamics of three old-growth cedar-hemlock forests in southern coastal British Columbia 50 4.2.1 Size and age distributions of tree populations 50 4.2.2 Radial growth patterns 59 4.2.3 Disturbance history 68 4.2.4 Canopy architecture and disturbance etiologySO 4.2.5 Forest turnover times 92 4.2.6 Spatial pattern of vegetation 92 4.3 Landscape patterns of natural and anthropogenic disturbances 4.3.1 Stand age distribution in the Eastcap Creek landscape 4.3.2 Structural-compositional trends along a disturbance gradient 4.3.3 Successional pathways  112 112 116 116  5. DISCUSSION 126 5.1 Old-growth/second-growth comparisons 126 5. 2 Stand structure and dynamics 130 5.3 The ecology of disturbance and patch dynamics in a temperate rainforest landscape 150 6. CONTRIBUTIONS TO COASTAL FOREST MANAGEMENT 6.1 The decadent argument 6.2 Fire and water: Some comments on the ecology and management of Vancouver's watersheds 6.3 B.C.'s green gold: New opportunities in silviculture 6.4 The value of biological information  167 171  7 . CONCLUSIONS  173  8. BIBLIOGRAPHY  176  156 160  X  LIST OF TABLES 2.1 Selected climatic characteristics of the study area  14  3.2.1 Site characteristics of the three forest history reconstruction plots  25  4.1.1 Pearson correlations relating PCA axes of forest composition and structure to the first canonical axis of each variable set (CV1 and CV2) 40 4.1.2 Pearson correlations between PCA axes of forest structure and composition accounting for two-thirds of the overall variation  41  4.1.3 Pearson correlations relating plant species to the first two PCA axes of the composition data 44 4.1.4 Pearson correlations relating structural variables to the first PCA axis of the structural data 45 4.1.5 Structural-compositional linkages as shown by Pearson correlations between the first two PCA axes (I and II) of composition and structure within separate forest age classes 48 4.1.6 Proportions of total sums of squares along PCA I and II axes (r 2 values from ANOVA) ascribed to study area locations 49 4.2.1 Mean radial growth rates ± SD of saplings, poles and trees for each species 67 4.2.2 Summary statistics for canopy gap measures  86  4.2.3 Number of canopy gap makers ( dead trees > 30 cm dbh) of each species, by type of mortality 89 4.2.4 Number of canopy gap makers ( dead trees > 30 cm dbh) in each decay class, by type of mortality 90 4.2.5 Forest turnover rates estimated from percent area in canopy gaps and from disturbance frequency 93 4.2.6 Moran's I spatial autocorrelation index for saplings, poles, and trees in plot MERC102 95 4.2.7 Moran's I spatial autocorrelation index for saplings, poles, and trees in plot EC183 96 4.2.8 Moran's I spatial autocorrelation index for saplings, poles, and trees in plot EC163 97 4.2.9 Abundance of tree seedlings and saplings on various substrata in the three study sites 100  xi 4.2.10 Frequency of plant species in EC183, EC163, and MERC102  103  4.3.2 Ecological characteristics of 26 stands from the Capilano watershed 125 5.2.1 Comparison of canopy gap data from northern and southern temperate rainforests 141 5.3.1 Fire history in Vancouver's watersheds  151  6.3.1 Natural disturbances and forest harvesting in the Coastal Western Hemlock Zone 169  LIST OF FIGURES 3.1.1 Map showing location of four study areas  21  3.3.1 Map showing topography of the Eastcap creek watershed and adjacent watersheds 33 4.1.1 Relative densities of living trees in 10cm diameter size classes for young, mature, and old forests....37 4.1.2 Relative densities of dead trees (snags) in 10cm diameter size classes for young, mature, and old forests 4.1.3  PCA ordinations of plots based on compositional and structural data  4.1.4 Principal component analysis ordinations of the separate forest age-classes  39 43 47  4.2.1 Relative densities of trees in 10cm diameter size classes in MERC102 51 4.2.2 Relative densities of living trees in 50 year classes in MERC102  52  4.2.3 Relative densities of trees in 10 cm diameter classes in EC183  54  4.2.4 Relative densities of living trees in 10 year classes in EC183  55  4.2.5  Relative densities of trees in 10cm diameter size classes in EC163  57  4.2.6 Relative densities of living trees in 50 year classes in EC163  58  4.2.7 Scatter diagrams of age versus diameter  60  4.2.8 Examples of various radial growth patterns found in the temperate rainforests of southern coastal British Columbia 62 4.2.9 Temporal pattern of disturbance and regeneration in MERC102  69  4.2.10 Temporal pattern of disturbance and regeneration in EC183 70 4.2.11 Temporal pattern of disturbance and regeneration in EC163 71 4.2.12 Map of disturbance events in EC183  75  4.2.13 Spatio-temporal distribution of disturbance patches in MERC102 76  xiii 4.2.14 Spatio-temporal distribution of disturbance patches in EC183 77 4.2.15 Spatio-temporal distribution of disturbance patches in EC163 78 4.2.16 Spatial autocorrelation of 1940 growth releases and regeneration cohort in EC183  79  4.2.17 Forest architecture and the spatial distribution of tree ages in MERC102 82 4.2.18 Forest architecture and the spatial distribution of tree ages in EC183 83 4.2.19 Forest architecture and the spatial distribution of tree ages in EC163 84 4.2.20 Size distributions of canopy gaps for the three study areas 87 4.2.21 Size distributions of expanded gaps for the three study areas 88 4.2.22 Frequency distributions of the number of gap makers per gap in the three study areas 91 4.2.23 Spatial distribution of western hemlock and Pacific silver fir saplings density 98 4.2.24 Species rank curves for understory vascular plants in the three study areas 101 4.2.25 Spatial autocorrelation of first PCA axis of understory vascular plant data showing multispecies pattern for the three study areas 104 4.2.26 Distribution of bryophyte species richness along four fifty meters transects in old-growth cedar hemlock forests 106 4.2.27 Box plots on the first two PCA axes by substratum using 221 units and b) DCA ordination of 10 substrata 107 4.2.28 Species/substratum relationships and niche breadth for 44 species of bryophytes in old-growth cedarhemlock forests 109 4.3.1 The Eastcap creek space-time mosaic  113  4.3.2 Stand age distributions in the East-Cap creek landscape. (CWHvml)  114  4.3.3 Structural diversity measured along a disturbance gradient in the Capilano watershed 117  xiv 4.3.4 Vascular plant species richness along a disturbance gradient in the Capilano watershed 118 4.3.5 Size class ordination vectors in 26 stands from the Capilano watershed 119 6.2.1 Logging scenario for the Greater Vancouver watersheds  162  XV  ACKNOWLEDGEMENTS Many people contributed their time, ideas, and support during my thesis.  D. Gagnon began the process by introducing me  to temperate rainforests during a trip to British Columbia in 1988. I would like to thank all the members of my committee, G.E. Bradfield, K. Klinka, K.P. Lertzman, W.B. Schofield, and R. Turkington.  I am especially appreciative of the encouragement,  rigor and able supervision that G.E. Bradfield offered during this degree. I would also like to thank J.K. Agee, M. Feller, J.P. Kimmins, J. Maze, and I. Taylor for useful comments on earlier drafts. Field assistance was provided by P. Yurke, N. Zolbrod, K. Liang, L. Waterhouse, and T. Plath.  R. Belland and T. Goward  assisted me in the field and laboratory with the identification of cryptogams.  Collaboration with D. Scheidemann and R. Stolz on  Vancouver Island was enjoyable and valuable. Generous financial support from the Research Branch, and the Vancouver Forest Region of British Columbia's Ministry of Forests made this study possible. always helpful.  The staff from the Forest Service was  In particular I would like to thank J.  Parminter, E. Hamilton, A. MacKinnon, A. Nicholson, D. Seip, R. Green, and J. Pojar. Working space and valuable assistance in the preparation of tree cross-sections was provided by Forintek Canada Corporation. I particularly would like to thank L. Josza for sharing his  xv i knowledge and enthusiasm in dendrochronology, and D. Fullbrook for his assistance and advice in woodworking. MacMillan Bloedel Limited kindly provided financial support and accommodations in Tofino during part of the 1991 field season.  Bill Beese provided useful advice and accompanied me  during the selection of the Clayoquot Sound study site. The staff from Kennedy Lake Division were always helpful.  Crucial  logistic assistance was contributed by J. Smith, and G. Petruko. Experienced faller B. Machan was helpful in the retrieval of cross-sections.  I would also like to thank N. Malbon and his  family for their hospitality during my stay in Tofino. The Watershed Management Department of the GVWD gave me and my colleagues permission to conduct research in restricted areas of the watersheds and offered logistic support. particularly helpful.  D. Bonin was  I would like to thank everyone who helped  me retrieve cross-sections.  I would particularly like to mention  J. Wiley, D. Wiley, D. Jack, and D. Smardon. I gratefully acknowledge an NSERC scholarship, and a UBC fellowship. Finally I would like to thank my family for their support and encouragements throughout this adventure.  Special thanks to  Susan for her love, patience, and joie de vivre.  xvii  To my parents Leonard Arsenault and Cecile Halle  1 1.1  OVERVIEW Despite continuous changes in the distribution and abundance  of organisms, patterns of community organization emerge. As early as 1807 Humboldt described geographical patterns of vegetation linked to climatic variation.  While Humboldt's static  view was useful to understand a fundamental concept in the pattern and process of the world's vegetation, other scientists like Darwin and De Candolle emphasized the dynamic nature of communities (Mcintosh 1985).  Early theories of succession,  influenced by both the ideas of change and stability, attempted to explain how communities converged towards predicted levels of organization (Clements 1916; Tansley 1920; Whittaker 1953) or maintained overall stability via cyclic succession (Cooper 1913, Aubreville 1938; Watt 1925, 1947). Although the concept of initial floristic composition (Egler 1954) and the importance of life history traits (Drury and Nisbet 1973) cast serious doubts on the predictable nature of succession, Watt's reflection on pattern and process in plant communities dominates contemporary ecology (e.g. Pickett and White 1985; Remmert 1991).  Watt's major contribution lies in the  use of a multiple-scale approach and in the suggestion that disturbances play an important role in community structure. In forests, although disturbances may occur at a variety of scales, large-scale and small-scale disturbances are usually contrasted.  The boreal forest of northern Canada represents one  extreme where large frequent fires maintain the dominance of pioneer species in the landscape.  In contrast, large  disturbances are rare in certain tropical rainforests (Aubreville  2 1938; Richards 1952; Brokaw 1985), temperate deciduous forests (Jones 1945; Runkle 1982), southern temperate rainforests (Veblen 1985), and subalpine coastal forests (Lertzman 1989) where vegetation dynamics are driven by the death of one or a few trees creating canopy openings (gap phase dynamics) providing an opportunity for shade tolerant or gap specialist species to access the canopy.  This information illustrates the importance  of conducting studies in virgin forests without which our understanding of nature would be much poorer. As the conversion of old-growth forests to managed forests is accelerating in coastal British Columbia, concern is mounting over the loss of ecological values. Although floristically well known (Krajina 1965; Gagnon 1985; Schofield 1988; Klinka et al. 1989), little quantitative information exists on coastal forest structure and its relationship to biological diversity.  This  makes forest planning difficult because neither the nature of the loss nor the chance to mitigate ecological impacts are clearly understood when an old-growth forest is harvested.  Furthermore,  harvesting old-growth forests without prior description is tantamount, in some cases, to erasing over a thousand years of naturally recorded history.  In this thesis I will take a  multiple-scale approach to investigate pattern and process in temperate rainforests of southern coastal British Columbia. 1.2 What are old-growth forests? A strict definition of old-growth forest using a standard English dictionary would read as follows: A relatively  wooded area which has existed  for a relatively  long time.  large  The  two components old and forest are very much scale dependent which  3 explains the various interpretations given for the term.  There  are also human values associated with how people perceive old forests.  For example the words "pristine" and "decadent" clearly  illustrate a value gradient. In areas subject to catastrophic disturbances, forests reach the old-growth stage when the initial cohort of trees, which established following a major disturbance, have all died (Oliver and Larson 1990) and have been replaced by more shade tolerant species.  In many forests of the world this condition is rarely  attained as a result of recurring allogenic processes such as fire.  For example, as a rule, old-growth conditions do not  develop in boreal forests where fire return intervals are often less than 100 years, except in a few small patches.  In western  North America Douglas-fir forests can grow for a thousand years or more and not reach a strict autogenic condition.  Oliver and  Larson (1990) use the term "transition old-growth" for forests that have not met entirely the autogenic criteria but exhibit many structural features characteristic of old-growth forests. These emergent structural elements are: a reverse J shape diameter distribution, large standing live and dead trees, large fallen trees, and a multilayered canopy. Douglas-fir forests of Washington and Oregon and cedarhemlock forests of British Columbia usually reach old-growth conditions two hundred and fifty years after a stand destroying disturbance (Franklin et al. 1981; Arsenault and Bradfield 1995). This is a subjective age limit that corresponds to the reality that many virgin forests of this area are 250 years or older.  4 Late stages of forest development are often considered to be relatively stable in species composition (i.e. climax, Clements 1916), and productivity (i.e. steady state, Bormann and Likens 1981).  Although, in the past, many ecologists believed that  these later stages of forest succession had static stability, it has been shown that the dynamic nature of ecosystems prohibits this condition (Botkin 1979; Pickett and White 1985). 1.3 Tools for dynamic plant ecology The development of concepts and their subsequent applications depend upon the methodology used and its limitations.  The study of plant community dynamics provides an  apt example of this. Before sampling, one must consider which spatial and temporal scales will be examined, and should be ready to justify this choice.  Ideally all scales should be examined but after  considering factors such as research funding and human life-span, the scientist usually needs to focus on one particular scale. In selecting variables to measure, difficult choices often have to be made. It is important to remember that this choice acts as a filter of total information present in the "real world".  For example, if the decision is made to study the  changes in plant species composition over time in the Coastal Western Hemlock zone, will all taxa be examined? This is a difficult question to answer because some species are predominantly epiphytes and inhabit the forest canopy, while others require a microscope for identification and make quadrat sampling difficult. addressed.  The difficulties are real and must be  Conversely, the effect of filtering the information  5 by focusing on a particular taxon also needs to be considered. This problem also applies to other community attributes, both structural and functional. Other important considerations involve the sampling design, including the location, orientation, and size of plots. In forested landscapes or other complex vegetation, researchers will stratify samples within known vegetation types in order to minimize unexplained variations (Barbour et al. 1980). Three different field approaches have been used to tackle plant community dynamics. 1. SFT: Space for time substitution (e.g. Cowles 1901; Cooper 1939).  This method has been one of the most widely used and also  one of the most criticized.  The advantage of using plots  representing different ages on similar sites is obvious. However the drawback is in the assumption that the younger sites represent a stage that will develop into something similar to the older sites (Pickett 1989).  Environmental variability and chance  may provide the fatal flaw in using this method.  Nonetheless  this approach may contribute useful information concerning successional trends which could be tested with the next two methods. 2. PP: The permanent plot approach (e.g. Franklin and DeBell 1988) examines changes in the cross section of time ahead. Furthermore, this method permits entry to the realm of experimentation and hypotheses testing. The immediate advantage of this method decreases with the researcher/plant longevity ratio.  However, the set-up of permanent plots represents a  6 valuable legacy that a scientist can leave for future generations of researchers. 3. HR: The historical reconstruction approach (e.g. Henry and Swan 1974; Oliver and Stephens 1977; Payette et al. 1990; Taylor and Halpern 1991) is time-consuming but is usually rewarding.  It  is complementary to PP, as here you concentrate on the cross section of time before present.  The reconstruction is made using  living and dead plant material.  The use of recorded events in  tree rings can be particularly helpful.  Similar frustrations as  those shared by historians and geologists will be experienced because part of the dead plant material will have vanished by decomposition or combustion.  However, similar excitement will  also be felt as insight into the history unfolds. 1.4 General concepts in dynamic plant ecology The history of concepts in plant community dynamics is closely linked to how ecologists have defined or perceived plant communities.  Of particular influence has been the ongoing debate  between the discrete view and the continuum view.  The discrete  view of plant communities is closely associated with F.E. Clements, who described the changes in plant communities (succession) as analogous to the development of an organism (Clements 1916; Mcintosh 1985).  Directional change was predicted  until a stable state determined by the climate (climax) would be achieved.  Also important to this view is the replacement of a  plant association by others through an environmental transformation via the vegetation (relay floristics).  Odum  (1969) later adapted this view to the development of ecosystems.  7 The monoclimax (Clements 1916), polyclimax (Tansley 1920), and climax pattern (Whittaker 1953) hypotheses all have relay floristics in common.  However, the latter is unique in  attempting to bridge a gap between the discrete, and continuum concept (Gleason 1926).  Egler (1954) seriously damaged the  concept of relay floristics by demonstrating that initial floristic composition is often more important in determining succession in plant communities. Two major research themes emerged in the 1970's and early 1980's. 1) The importance of vital attributes (Connell and Slatyer 1977) of species in succession associated with a Gleasonian view of plant communities (Gleason 1926).  This trend was stimulated by  the severe review of succession made by Drury and Nisbet (1973). 2) The role of natural disturbance in plant community dynamics revived some old ideas developed by Cooper (1926), Aubreville (1938), and crystallized in Watt's seminal paper in 1947 Pattern and process in the plant community.  An excellent review of this  theme was made by White (1979). In fact, there appears to be much interest among ecologists to develop a theory concerning the role of disturbance and patch dynamics in the organization of communities and ecosystems (Pickett and White 1985; Remmert 1991; Levin et al. 1993). 1.5 Community dynamics in western North America. Coastal forests of north western North America exhibit a high degree of structural complexity.  This results from the  longevity and large size of the dominant conifers combined with relatively slow decomposition rates  (Waring and Franklin 1979)  8 and a variety of natural disturbances (e.g. fire, landslides, windstorms, pest outbreaks, avalanches, treefalls).  Two studies  have specifically addressed structural-compositional dynamics in cool, wet forests along the British Columbia coast. Gagnon (1985) used structural and abiotic variables to interpret compositional gradients in western hemlock - western redcedar Pacific silver fir forests of western Vancouver Island (see also Gagnon and Bradfield 1986, 1987).  Lertzman (1989) examined patch  dynamics processes in mountain hemlock forests of the southern mainland (see also Lertzman and Krebs 1991, Lertzman 1992, and Lertzman 1995). Alaback (1982) observed important understory biomass changes over a chronosequence of forest development in Sitka spruce forests in southeast Alaska.  More recently Deal et al. (1991)  showed the importance of windthrow in mixed hemlock-spruce stands in coastal Alaska. Compared to these wetter and more northerly regions, more is known about structural-compositional dynamics in the warmer and drier Douglas-fir dominated forests of Washington and Oregon (Franklin and Hemstrom 1981).  Franklin et al. (1981) related  ecological differences between old-growth and young Douglas-fir forests to differences in major structural components such as large standing trees (living and dead), and large fallen trees on land and in streams.  Overstory canopy structure was correlated  with patterns of tree regeneration (Spies et al. 1990; Stewart 1986a), and with patterns of understory species composition (Stewart 1988).  Other studies used structure to infer processes  9 of forest dynamics (Kuiper 1988; Spies and Franklin 1988; Stewart 1989) . Few studies of landscape dynamics in Pacific coastal forests have been undertaken.  Fire has received the most attention and  an extensive review of existing literature is provided in Agee (1993).  Fire rotation in Mount Rainier national park was  estimated to be 434 years (Hemstrom and Franklin 1982). Henderson et al. (1989) described fire history in a section of the Olympic Peninsula, where an important relationship was noted to exist between the fire regime and plant associations.  Schmidt  (1970) has described the extent of some fires in the Coastal Douglas-Fir Zone on Vancouver Island and Eis (1962) provided some fire dates from Vancouver's northshore mountains. 1.6 Objectives. In this thesis, I examine patterns of vegetation structure and composition in the temperate rainforests of southern coastal British Columbia at both landscape and stand levels and consider the dynamics generating these patterns.  The thesis is comprised  of three studies that build on each other in order to provide a comprehensive understanding on the role that natural disturbances and patch dynamics play in shaping vegetation of this region. This thesis contributes to the background that is reguired to evaluate the extent to which present land management practices are changing the ecology of British Columbia's rainforests. In the initial study, I compared compositional and structural variation among young, mature, and old-growth forests (> 250 yr) from the Greater Vancouver area.  This analysis was at  10 the landscape level where interest focused on comparing forest patches of different ages.  Three questions were addressed:  1) What are the compositional and structural characteristics of forests of different age classes? 2) What are the quantitative relationships between structural and compositional variation in these forests? 3) What hypotheses can be generated with respect to the role of disturbance and patch dynamics in this landscape? The second study examined smaller spatial scales to elucidate the nature of forest stand development.  The structure  and dynamics of three old-growth cedar-hemlock forests (> 250 yr) were analyzed using the historical reconstruction approach.  I  focused on the pattern of regeneration of tree species inferred from size and age distributions, the disturbance regime and its role on tree regeneration, and  the spatial organization of  vegetation structure and composition.  The following questions  guided this investigation: A. Structure of tree populations and radial growth patterns. 1) Did the forest stands originate from catastrophic disturbance, or did forest development follow gap phase regeneration, and/or continuous regeneration? 2) What is the age and size structure of the tree populations? 3) What are the major types of radial growth patterns observed in these forests? B. Disturbance regime and forest architecture. 1) What is the spatio-temporal pattern of disturbances observed in these forests?  11 2) What are the main causes of disturbance and how do they relate to present canopy structure? 3) How does canopy structure of old-growth cedar-hemlock forests compare with other forests from western North America and southern temperate rainforests? 4) What are the differences in forest rotation rates between oldgrowth cedar-hemlock forests and other temperate rain forests? C. Spatial pattern of understory plant communities. 1) Is the spatial distribution of understory trees patchy, reflecting a gap phase structure (Watt 1947) or random? 2) Are these forests mosaics of small, even-aged patches? (Oliver and Stephens 1977). 3) What is the spatial organization of understory vascular plants? 4) What is the spatial organization of bryophyte communities and how does it relate to forest stand dynamics? The third study moved back up on the spatial scale to the landscape level. The objective was to provide a higher resolution of pattern and process at the scale of the Capilano watershed, using the historical reconstruction approach and information gathered in the previous two studies. The questions examined were: 1) What is the natural disturbance regime at this scale? 2) Is logging changing the natural disturbance regime and stand diversity? 3) What are the patterns of structural diversity and species richness along disturbance and environmental gradients? 4) What are the successional trends on the dominant habitats?  12 2. STUDY AREA 2.1 Physical geography. The study area is located in southern coastal British Columbia and covers three distinct geographic regions; The coast mountains, the insular mountains of Vancouver Island and the Fraser Lowland.  Excellent descriptions of the physiography and  geology of the study area are given in Holland (1976) and Armstrong (1990).  All study plots (except one on Vancouver  Island) were situated in the Greater Vancouver Area.  Most of the  field work took place on the lower and mid slopes of Vancouver's northshore mountains between 150 and 1000 meters above sea level. The magnificent coast mountains of this region rise rapidly from sea level to ca. 1600 meters and are dissected by several deep river valleys. diorite.  The rocks are mostly granodiorite and quartz  They originate from multiple igneous intrusions of  rocks forming a complex often referred to as a composite batholith.  A few plots were also situated in UBC's Pacific  Spirit Park in the Fraser Lowland.  This area is a smooth plateau  50-100 meters above sea level bordered by steep cliffs , 45-90 m in height, on three sides.  Late Cretaceous and Tertiary  sedimentary rocks are covered by 300 meters of unconsolidated sediments deposited during the last glaciation (Armstrong 1990). The Vancouver Island plot is located on the central west coast of the Island between Clayoquot sound and Barkley sound. to the insular mountains of Vancouver Island.  It belongs  This area is  dissected by numerous fiords, and deep river valleys, making it a spectacular and very rugged mountainous landscape.  The mountain  13 range is composed of a heterogeneous group of sedimentary and volcanic rocks (Armstrong 1990). 2.2 Surficial geology and soils. The last Pleistocene glaciation, ending approximately 12000 years ago, contributed much of the present surficial material. This material includes glacio-fluvial, glaciomarine, fluvioglacial, and glacial till deposits.  In addition, marine deposits  and colluvial material also form part of the surficial deposit makeup in this region (Gagnon 1985).  Ferro-Humic and Humo-Ferric  Podzols are the dominant soil forms found in the study area (Vallentyne and Lavkulich 1978).  Also present are pockets of  Regosols on rocky outcrops and Gleysols or organic soils on badly drained sites. 2.3 Climate. Trewartha (1954) has classified the climate of the study area as humid temperate with cool dry summers and moderate wet winters.  The mean annual precipitation totals 2,228 mm for the  CWH (Klinka et al. 1991) and varies from 1,258 mm in Pacific Spirit Park (Table 2.1), at the driest end of the gradient, to 2,296 mm at Cleveland Dam in Vancouver's northshore mountains, to 3,288 mm at Tofino on Vancouver Island's west coast.  Summer fog  is an important feature at the Vancouver Island study plot. Snow accumulations are of greater significance on Vancouver's northshore mountains than on Vancouver Island or Pacific Spirit Park. 1991) .  The mean annual temperature is 7.9 °C (Klinka et al.  14 Table 2.1 Selected climatic characteristics for the four study areas in Greater Vancouver.  Climate station  Vancouver (UBC)  Elevation (m)  Capilano  Seymour  Coquitlam  Cleveland  Falls  Lake  87  157  244  161  Mean annual precipitation (mm)  1258  2297  3841  3616  Mean precipitation May-Sept (mm)  53  94  144  134  Mean precipitation of the driest month (mm)  37  66  87  86  Mean precipitation of the wettest month (mm)  208  380  613  569  Mean annual temperature (°C)  9.8  n.a.  n.a.  8.4  Mean temperature of the warmest month(°C)  16.9  n.a.  n.a.  16.6  Mean temperature of the coldest month (°C)  2.9  n.a.  n.a.  0.6  Source of data: Canadian Climate Normals, 1951-1980, Environment Canada n.a. (data not available)  15 2.4 Vegetation. Krajina (1965) recognized four biogeoclimatic zones in southern coastal British Columbia: The Coastal Douglas Fir Zone, the Coastal Western Hemlock Zone, the Mountain Hemlock Zone, and the Alpine-Tundra Zone. All of the study area occurred within the Coastal Western Hemlock zone, and primarily in the submontane variant.  This biogeoclimatic unit corresponds to the temperate  rain forest biome (Whittaker 1975; Alaback 1991) of which approximately two thirds occur on the northwest coast of North America.  There appears to be some confusion in the literature  concerning the classification of this biome.  Some authors placed  it in the temperate forest category and therefore lumped it with the deciduous broadleaved forests of Eastern North America (Smith 1980) while others included it with the boreal forest under the northern coniferous forest biome (Walter 1973). In the northern hemisphere the temperate rainforest extends along the Pacific Coast from the redwood forests in California and Oregon to the Sitka spruce forests of northern British Columbia and Alaska and is dominated by conifers belonging to the Pinaceae and Cupressaceae.  In South America the Valdivian  rainforests in the lee of the Andes of Chile and Argentina are dominated by broadleaved evergreens (Nothofagus).  In New  Zealand, Tasmania and mainland Australia southern beech and in some cases conifers of the Podocarpaceae and of the Araucariaccae (the monkey puzzle tree) dominate emergent strata of the temperate rainforests located on coastal areas.  In addition,  some of Australia's rainforests are dominated by another broadleaved evergreen, Eucalyptus.  Ovington (1983) described the  16 broadleaved evergreen rainforests as multilayered, rich in epiphytes, and with a well-developed understory.  Many old-growth  forests of western North America share these traits.  In addition  the dominant tree species of the northern and southern temperate rainforests are often evergreen and long lived. The coastal Western Hemlock zone is dominated by western hemlock fTsuga heterophylla^, western redcedar (Thuja plicata^, Pacific silver fir (Abies amabilis) and Douglas-fir (Pseudotsuga menziesii).  These tree species vary in abundance according to  their position along environmental gradients, the disturbance regime, and chance.  Other trees found in the study area include  the deciduous red alder (Alnus rubra), bigleaf maple (Acer macrophy1lum), and black cottonwood CPopulus trichocarpa) found on wet sites and/or in younger serai stages. Alaska yellow-cedar (Chaemacyparis nootkatensisl, usually found at the cold-wet end of the gradient, is more abundant in the Mountain Hemlock Zone. Western white pine (Pinus monticola), probably more abundant before the introduction of the white pine blister rust, can be found on warm exposed slopes and rocky outcrops. Western yew (Taxus brevifolia) is not rare but has a sparse distribution. Understory vegetation is often characterized by Vaccinium alaskaense, Vaccinium parviflorum, Gaultheria shallon, Menziesia ferruginea, Rubus spectabilis, Blechnum spicant, Polystichum muniturn, Pteridium aquilinum, Cornus canadensisf Coptis asplenifoliaf Rubus pedatus, Lysichitum americanum and Oplopanax horridum.  Bryophytes are common on the forest floor and occur on  a variety of substrata.  Some common species include  Plagiothecium undulatum, Rhytidiadelphus loreus, Kindbergia  17 oregana, Sphagnum girgensonii, Rhyzomnium glabrescens, Scapania bolanderi, Lepidozia reptans, and Calypogea muelleriana. Although lichens are present on the forest floor, mainly as Cladonia species, they occur mostly as epiphytes in the forest canopy forming complex communities with bryophyte species.  Except for  Polypodium qlycyrrhiza, most species of epiphytes are nonvascular plants. However, it is not rare to observe some trees (mostly western hemlock) and shrubs (mostly Vacciniuml growing on branches in the canopy. 2.5 Disturbance regime. Anthropogenic disturbances have been rather severe for the past hundred years, corresponding with the increase of immigrant human settlements along the coasts. Logging, establishment of settlements and other land uses have modified the temperate rain forest landscape of this region substantially.  Before the  arrival of Europeans, Native populations practiced selective logging and modified trees for spiritual and cultural purposes on a very small scale.  Some First Nations people also cleared land,  using fire, for the production of bulbous plants (Turner 1991). It is generally accepted that these modifications by First Nations people did not alter the forest landscape substantially. Southern coastal British Columbia is characterized by a wide variety of non-anthropogenic disturbances that reflect the unique biotic and abiotic attributes of the temperate rain forest biome. Large stand-destroying disturbances, such as fire and hurricanes, can be extensive but are infrequent.  Although infrequent, the  recurrent nature of forest fires in parts of the study area is believed to have had a role in shaping the composition and  18 structure of some forests partly based on the longevity of some conifers which may establish after fire (Franklin 1988).  Gagnon  (1985) suggested that the disturbance regime in coastal forests may be related to features of the landscape.  For example on  Vancouver Island forest fires are more important in the drier, warmer forests in the rain shadow, while extensive windthrows occur predominantly near the coast.  Steep slopes are more  susceptible to landslides and avalanches than is more gentle terrain.  Large tracts of old-growth forests within the Coastal  Western Hemlock and Mountain Hemlock Zones appear to change very slowly through the death of small groups of trees. of change is referred to as gap phase dynamics.  This process  19 3. METHODS 3.1 Structural-compositional variation in young, mature and old temperate rainforests. 3.1.1 Sampling Four study areas were selected near Vancouver, British Columbia, encompassing a range of forest age classes and climatic conditions (Fig.3.1.1).  Three of the study areas were located in  the watersheds of the Capilano, Seymour, and Coquitlam Rivers; the fourth was in Pacific Spirit Park, adjacent to the University of British Columbia (UBC).  All areas fell within the Pacific  Ranges physiographic unit (Holland 1976), and the Coastal Western Hemlock biogeoclimatic zone (Krajina 1965).  Climatic variability  was associated mainly with elevation, and precipitation in parts of the watersheds exceeded three times that at the UBC site (Table 2.1). Ninety-six 30m x 30m plots were selected throughout the study areas on the basis of locations determined by the census points used in a companion vertebrate biodiversity study (Fig 3.1.1).  Three age classes of forests were represented: 48 plots  were in forests that had not experienced any major disturbance for at least 250 yr (referred to here as "old" forests); 28 plots were in 60-80 yr old forests that had developed after logging between 1910 and 1930 ("mature" forests); 20 plots were located in regenerating forests that were logged 30-60 years ago ("young" forests).  The historical patterns of logging throughout the  study area have resulted in mature forests being at lower  20 elevations, hence experiencing a somewhat warmer and drier climate, than the young and old forests. Percent coverage was visually estimated for all understory herbaceous and woody plant species, including common forest-floor bryophytes and tree seedlings, within each 30m x 30m plot. All living trees within plots were recorded by species and diameter at breast height (dbh) measured at 1.3m above ground level. Trees with dbh <10cm were recorded as saplings. Dead trees were described in three categories:  snags (dead  standing trees >2m in height), stumps (dead trees <2m in height), and logs (fallen trees).  Snags and stumps were each recorded by  their dbh, and snags were further subjectively classified into one of four height categories based on percentage of original tree height remaining (>75%, 51-75%, 25-50%, and <25%), and on the basis of bole decomposition as either hard or soft. This classification was similar to others proposed for decomposing wood in forests (Fogel et al. 1973; Cline et al. 1980; Hunter 1990), but recognized that stem breakage may have resulted in a variety of snag heights for a particular decay category.  Logs  were sampled using the triangular transect method described by Trowbridge et al. (1986). With this method the species (where identification was possible), diameter, and decomposition category (same as for snags) were recorded for all logs intercepted along the edges of a 90m equilateral triangle located within the boundaries of each study plot. In addition to the vegetation data obtained for each plot, a variety of habitat descriptors including landform, slope and aspect, organic layer depth, and subjective estimates of drainage  21 Figure 3.1.1 Map showing location of four study areas. Squares indicate forest stands in which study plots were established.  22 class and humus forms (Klinka et al. 1981) also were recorded. 3.1.2 Data analysis Size class distribution graphs of living and dead trees in each of the three forest age classes provided a direct means of assessing major compositional and structural features in relation to stand age. I used principal component analysis (PCA) of the combined age classes to calculate summary vectors (i.e. PCA axes) that allowed quantitative comparisons of the major trends in forest compositional and structural variation.  The compositional PCA  was based on an unstandardized covariance matrix calculated from the percent cover data for understorey species and the percent importance values for tree species. The choice of covariance was justified on grounds that all variables were measured on a common (i.e. percentage) scale.  Tree importance values were calculated  as the average of relative density and relative basal area for each species. Relative values refer to the amount recorded for a species in a plot divided by the total and multiplied by 100. The structural PCA was based on a correlation matrix calculated from the twenty-one structural attributes listed in Table 4.1.4 on page 45. The relationship between the compositional and structural variation was assessed with canonical correlation analysis (CCA). Input variables for the CCA were the PCA axes accounting for twothirds of the total variation in the respective compositional and structural measurement domains.  The within age class structural-  compositional relationships were examined by calculating Pearson correlations between the first two PCA axes of the respective  23 data sets.  CCA was not used in this case because it was felt the  results would be unreliable given the small sample sizes. In order to assess the effects of study area locations on the within age class variation, analysis of variance (Anova) was used. Study area effects were determined as r2-values expressing the proportion of total sum of squares along the first two PCA axes that could be ascribed to location.  24 3.2 Structure and dynamics of three old-growth cedar-hemlock forests in southern coastal British Columbia. 3.2.1 Sampling Three ca. 1/2 ha forest history reconstruction plots were selected based on the following criteria. 1) Western redcedar was one of the dominant tree species. 2) The sites were representative of mesic virgin forests in the CWHvml variant showing no sign of catastrophic disturbance for at least the past 250 years. 3) The plots were located in areas scheduled to be logged a few months following plot establishment in order to retrieve tree cross sections with the help of professional fallers.  One 0.45 ha plot was located on the west coast of  Vancouver Island in timber license 44 managed by MacMillan Bloedel in the general area of Clayoquot Sound. The other two plots (0.5 ha) were situated in the Eastcap Creek water catchment area within the Capilano watershed, which is managed by the Greater Vancouver Water District (Table 3.2.1) 3.2.2 Forest architecture All trees exceeding 1 cm in diameter were mapped in 5x5 m contiguous quadrats, and their dbh, species, and growing substrate recorded. All were labelled with numbered aluminum tags. The positions of trees exceeding 5 cm dbh were recorded to the nearest 0.1 m.  Seedlings (trees < 1 cm dbh) were tallied by  species and growing substrate in 1 m 2 quadrats placed at the center of each 5x5 m quadrat.  For canopy trees (> 30cm dbh)  total height, height to crown base, and main branches were measured using a clinometer and meter tape.  Canopy projections  25 Table 3.2.1 Site characteristics of the three forest history reconstruction plots. TFL refers to Tree Farm License. MERC102  EC183  EC163  Geographical area  Clayoquot sound  Northshore mountains  Northshore mountains  watershed  Mercantile cr.  Eastcap cr.  Eastcap cr.  Land status  TFL 44  GVWD  GVWD  300  450  750  Aspect  40  166  180  Slope  30  30  40  Elevation  Trophotope  poor-medium  medium-rich  poor-medium  Hygrotope  Mesic  Mesic  Mesic  692  1002  Present  Absent  Maximum age yr. Charcoal  946 Absent  26 were measured for the same trees, using distances from tree base to canopy edge in four directions (north, south, east, and west). When necessary the canopy edge was determined using a clinometer. In addition to these quantitative measurements, a rough sketch of these trees was prepared in the field in order to facilitate forest profile drawings.  Forest profiles are valuable in  providing a three dimensional perspective of forest structure (See Richards 1983 and Oldeman 1983, 1989). Maps of canopy projections were made in order to measure the relative area occupied by canopy gaps, expanded gaps, and closed canopy.  For practical reasons the minimum height of the canopy  was determined to be related to trees 30 cm in diameter.  Canopy  gaps were measured as the area covered by the projection of canopy openings on the ground.  Expanded gaps refer to the area  delimited by the boles of trees on the border of the canopy openings.  These gap descriptors have also been used in numerous  studies of forest stand dynamics ( e.g. Runkle 1982; Veblen 1985; Lertzman 1989) . 3.2.3 Plant species composition Percent ground cover of vascular plants was estimated for all species present in the 5X5 m contiguous quadrats.  In the  Eastcap Creek drainage of the Capilano watershed bryophytes were recorded by percent ground cover and substrate in 40 5X5 m plots distributed on four 5X50 m transects.  Collections of epiphytic  lichens and bryophytes were obtained after the trees were cut, and identifications were made in the lab.  27 3.2.4 Dendrochronology During the logging operations I accompanied the fallers who generously made extra cuts on the fresh stumps to obtain crosssections for all labelled trees. In most cases the cross-sections were retrieved between 30 cm and 80 cm from the ground. All cross-sections were brought back to Forintek Canada Corporation wood preparation lab for surface preparation.  A first series of  cuts was made on the larger cross-sections to obtain representative bark to pith areas. A second series of transversal cuts achieved a smoother surface.  Then, all cross-  sections were sanded using electric belt sanders for the final preparation of smooth surfaces.  Tree rings were counted and  unusual growth patterns recorded with the aid of a binocular dissecting microscope (40x).  A more than doubling of the radial  growth rate (using 10 years comparisons), sustained for at least 10 years, was considered to reflect a growth release. The inverse situation was interpreted as a period of suppression.  Narrow  rings, indicative of periodic climatic stress, were also recorded.  The date of germination was assumed to be the date  obtained from the pith.  This probably underestimated the real  germination date which could be obtained only if the crosssection was retrieved at the germination point.  This is  difficult with large trees and certainly was not practical with the number of trees I examined.  I am confident that the level of  accuracy obtained in ageing the trees is adequate for the purposes of this study.  In some cases, cross-dating of  individuals was achieved; however, it would have been impossible to cross-date all individuals based on noncircular uniformity  of  28 annual rings and complacent growth patterns for many of the individuals examined.  This suggests that interpretation of tree  growth patterns in coastal forests of B.C. from tree cores alone should be used with caution. 3.2.5 Disturbance regime Chronological evidence of past disturbances was obtained by recording all sudden radial growth releases or suppressions observed for every cross-section collected.  Pulses of  regeneration, initial growth rates, and the ages of trees growing on logs were other clues that helped in reconstructing the temporal pattern of disturbances.  Physical evidence of past  disturbances was monitored in recording the position, dbh, height, species, decay category, direction of fall for uprooted trees, and cause of mortality (when determinable) for all dead trees >10 cm dbh. In addition an intensive search for any sign of past catastrophic disturbances, especially charcoal, was undertaken in every plot. Several disturbance regime variables were calculated in order to evaluate the rate of change of these old-growth forests. Turnover time (TT2) is the average time between the creation of canopy gaps at any point in the forest, calculated from the proportion of forest in canopy gaps and the time required by trees to fill these gaps (Tf-Qj) (Lertzman 1989).  In the  calculation of TT2, I used a range of Tf-Q^ values which were consistent with the growth patterns observed on each site. The treefall rotation index (TRI) (Payette et al. 1990) estimates the time required for treefalls to disturb an area equivalent to the size of the plot surveyed and requires data on  29 the spatio-temporal distribution of disturbance patches. The area of disturbance patches was delimited by groups of trees which were influenced by the same disturbance event reflected in their specific tree-ring signature associated with patterns of radial growth release (Payette et al. 1990).  The tree fall free  interval (TFI) corresponds to the average number of years between consecutive treefalls, and is calculated using the following formula: "{ E(t^ +1 - t^) }/N, where (t^ +1 - tj_) is the number of years between any two consecutive tree falls and N is the number of treefall intervals recorded in the forest plot" (Payette et al. 1990).  The tree fall frequency (TFF), the inverse of TFI,  was calculated as the average number of treefalls per year. Since information of old disturbances obtained from tree ring records is probably not complete I decided to use the 1760-1980 period for the calculations of TRI, TFI, and TFF. 3.2.6 Spatial Pattern of vegetation The spatial pattern of tree populations was examined by calculating spatial autocorrelation coefficients on the density of saplings, poles, and trees in the 25 m 2 quadrats.  Moran's I  coefficient was used to evaluate the intensity and scale of pattern and to test the null hypothesis (H 0 ): there is no significant spatial autocorrelation in the data (Moran 1950; Legendre and Fortin 1989).  The global significance of the  spatial autocorrelograms was determined using a Bonferroni corrected a of 0.05 calculated by dividing a by the number of distance classes.  Furthermore, an examination of the  distribution of significant coefficients and shape of the  30 correlogram allowed identification of the type of pattern, for example a gradient or a patchy distribution. General patterns of community organization were described using species rank curves (Whittaker 1975).  Multispecies pattern  of the vascular plant understory (excluding tree species) was investigated using a multivariate approach combined with spatial autocorrelation (Goodman 1979).  First I used a principal  components analysis on a correlation matrix of the 25 m 2 quadrats containing species percent ground cover values for each plot separately.  The second step involved the calculation of Moran's  I coefficient using the scores from the first axis of the ordinations. Patterns of bryophyte diversity from the Eastcap plots were examined by plotting species richness along each 50 m transect. Four transects were established, one in EC183 and in an adjacent old-growth cedar hemlock stand, and one in EC163 and in an adjacent stand.  In order to describe spatial pattern, the  frequency of occurrence of each species was calculated on 10 substrate classes. The importance of substrate on bryophyte community structure was tested by performing ordinations on quadrat-substrate units (n=221) and then using the scores of the first four axes in a Kruskal-Wallis non-parametric test using substrata as a classification criterion. In order to examine habitat utilization for each species, a graph on the relative frequency of occurrence was plotted and the following index of niche breadth (Smith 1980) was computed: B-L = LOG 10 S Nij - (1/2 Nij) S Nji LOG £ H^ LOG r  31 where  Nj^ = Value of species i in niche category j r = Number of niche categories  B^ ranges from near 0 for species occurring on only one substrate to 1 for a species equally distributed on all substratum. Ten niche categories, or substrate types, were defined during sampling.  These were: epiphytes on trees, epiphytes on  shrubs and saplings, fresh decaying wood (Recently dead; fine twigs still abundant), Dl decaying wood (Most bark present; main branches are present but twigs are not abundant; the wood is hard), D2hard decaying wood (Most bark is gone; main branches are gone; the wood is still hard), D2-D3 decaying wood (The wood is relatively soft; most bark is usually absent but in exceptions may be present), humus, wet depression, intermittent creek, mineral soil and rock.  32 3.3 The role of natural and anthropogenic disturbances in a temperate rain forest landscape in southern coastal British Columbia. 3.3.1 Sampling This study focused on the Capilano River watershed and, more particularly, on the Eastcap Creek water catchment area (Fig 3.3.1).  Using the GVWD forest cover maps, aerial photographs,  and previous experience in this area, patches of forest vegetation were identified according to stage of development, stand age, and vegetation cover type.  More detailed information  on stand age and disturbance history was obtained from the forest history reconstruction plots, counting tree rings, and examining initial growth patterns of large stumps from recent clear-cuts adjacent to virgin stands.  Thirty-two plots from the initial  study as well as 28 new plots were used to describe vegetation structure and composition.  Information collected from these  plots is similar to that described for the first study except that structural data were collected using .05 ha circular plots for trees and snags while compositional data was collected in a .09 ha circular plot.  Furthermore, instead of using triangular  transects to survey the coarse woody debris, it was found to be more efficient to use a 50 m transect in the middle of the sampling ring. recorded.  Any sign of past catastrophic disturbances was  33 Figure 3-3.1 Map showing the topography and major creeks of the Eastcap Greek landscape and adjacent watersheds. The area enclosed by a thick line was used in the description of the space-time mosaic. Elevation contours are 60 m apart.  I  34 3.3.2 Data analysis The reconstruction of the space-time mosaic was made using a map of the Eastcap creek watershed, showing the patches of forests in different stages of development.  The information for  these patches includes time elapsed since the last major disturbance, as well as the nature (fire, logging) and severity of the disturbance.  In order to estimate time since disturbance  I used actual tree ring counts obtained in the field.  In order  to confirm the occurrence of a forest fire I required the presence of charcoal on the forest floor or in the soil, and/ or fire scars with charcoal present, and fast initial growth patterns of the initial cohort of trees.  Caution is required in  the identification of fire scars on western redcedar because old individuals of this species often show scars that originate from various other causes. Nine natural scarring forces were identified in a study of redcedar on Meares Island (Eldrige et al. 1984).  These were fire, lightning, falling trees, breaking  branches, animals, fungi, sunscalding, standing water, and abrasion by falling or sliding rocks. In addition to the tree ages collected in the sixty plots, I also made counts on cut stumps in most of the clear cuts.  The  area of the forest polygons was obtained from the GVWD management department and enabled me to estimate the proportion of total area in the different stand development categories. Structural variation was summarized by using PCA on 26 structural variables.  The trends in structural diversity (first  PCA axis) were examined along disturbance and environmental gradients.  The structural variables included abundance and sizes  35 of live and dead trees as well as percent cover of understory strata. Successional trends observed on the different sites were assessed using succession vectors (Goff and Zedler 1972; Enright 1982), and by using graphs of the size distributions of live and dead trees.  Succession vectors were obtained by dividing the  trees in each plot into six diameter classes ( 1-10, 10-30, 3050, 50-70, 70 -100, and >100 cm).  A matrix containing densities  of six tree species in 320 size class samples was analysed by correspondence analysis.  The size class samples corresponding to  the same plots were linked from large trees to small trees. The assumption is that trees present in the understory will eventually replace present trees in the canopy.  36 4 Results 4.1 Structural-compositional variation in young, mature, and old temperate rain forests. 4.1.1 Tree size structure Characteristics of the three forest age classes based on size class distributions of living trees are shown in Fig. 4.1.1. Western hemlock was clearly the dominant species in young forests followed by Pacific silver fir and western redcedar.  Isolated  individuals of western yew and Douglas-fir occurred in a range of size classes. Mature forests were characterized by active regeneration of western hemlock and western redcedar in the smallest dbh class, and an increased importance of Douglas-fir in the larger size classes.  Pacific silver fir was a relatively  unimportant species in this age group owing to the warmer and drier sites (lower elevations than young and old) on which the mature forests occurred. Tree size distributions in old forests were markedly different from those in young and mature forests (Fig. 4.1.1). western hemlock, followed by Pacific silver fir and western redcedar were the dominant species occurring over a wide range of sizes.  Large numbers of individuals in the smallest size class  and a steep decline and subsequent flattening across larger size classes produced an overall reverse-J distribution that has been noted by others as an emergent structural property of old-growth forests (Oliver and Larson 1990). The size class distributions of dead trees in the three forest age classes are presented in Fig. 4.1.2.  Main stem  37 Figure 4.1.1 Relative densities of living trees in 10cm diameter size classes for young, mature, and old forests. Also shown are total tree densities (stems/ha) in each age class. Tree species symbols: •(Pacific silver fir), •*»(Alaska yellow-cedar) ,a(Douglas-fir), f (western yew), ^(western redcedar), o(w^stern hemlock).  YOUNG (820 stems/ha.)  85  >100  85  >100  85  >100  MATURE (687 stems/ha.)  25  45  65  OLD (1260 stems/ha.)  25  45 65 Size Class(cm)  38 breakage was a common feature of dead trees in all three age classes and the degree of lower bole decomposition (i.e. "hard" versus "soft") was directly related to top damage.  In young and  mature forests most dead trees occurred in the small size classes, characteristic of stands undergoing self-thinning. Large snags (>100cm dbh) in young and mature forests were probably dead at the time of the original logging and were left standing.  Old forests exhibited a different pattern of snag  distributions from young and mature forests.  Total numbers of  snags in old forests were lower, owing to the reduced effects of self-thinning, and those present were more evenly distributed across all size classes. 4.1.2 Structure-composition correlation: Combined age classes. The canonical correlation between composition and structure was strong (r=0.84) (Table 4.1.1); however, low redundancy values (<25%) indicated that a substantial amount of variation remained unexplained by the analysis.  The first two canonical axes were  strongly correlated with the first PCA axes of the compositional and structural data which, in turn, were strongly intercorrelated (Table 4.1.2). Few other correlations were detected among the PCA axes. The PCA ordinations of plots based on compositional and structural data for the combined age classes are shown in Fig. 4.1.3.  In both cases old forests were segregated from young and  mature forests along the first PCA axis.  Several species  associated with old forests were positively correlated with axis 1 of the compositional data including Pacific silver fir (tree, sapling and seedling stages), western redcedar (tree stage),  100 100%  YOUNG  80  MATURE  (99 snags/ha)  (81 snags/ha)  (12 snags/ha)  80  £• 60  60  S 40  40  20 0  OLD  100  20 15 25 35 45 55 65 75 85 95>100  100  •15l .25 l . 35 n ,45  15 25 35 45 55 65 75 85 95 >100 100  I  Ml  55 65 75 85 95 >100  100  75% 80  (28 snags/ha)  60  60  40  40  20 0  I  20 SSL.  (22 snags/ha)  80  l ' ^ !•  i  1  15 25 35 45 55 65 75 85 95>100  100  0  80 •  >—  •  (10  >  60  L  20  !•-•-•  0  -I  15 25 35 45 55 65 75 85 95 >100  15 25 35 45 55 65 75 85 95 >100  (26 snags/ha)  40 20 15 25 35 45 55 65 75 85 95*100 100 25°/  00  i X  100  80  =5- 60 c 40 20  0  (25 snags/ha)  I  0 15 25 35 45 55 65 75 85 95>100 Size Class(cm)  fil  O hj if. O (D cn  p> t hi 0 CD en H-  P> £ H- CD 3 h H- CD  40 +  0J 3 M  ft 0  n  SB  -—^ r  H CD  oua e n cn o O 3 0 o\° h •-N  20 •  •„ • •  0> H ^ £S !_,.  H N  15 25 35 45 55 65 75 85 95 >100  &  :  Ul & Q. CD U o\° 3 Q) > O Cfl 4 •< H- cn Ul f t •* O T 3 Ho\» ff CD 3" ^ h 0) cn hj 0) O CD H- O, a 3 ^•t 3 O. r l SB CD 0 t O i Q £» t 3 Ul CD O CD o\° 3 " 3 S—* O H  100 80  (38  60  40  40  il*  s 3  ft rt 0 H-iQ  60  20  U & r t a0 0 CD  60  15 25 35 45 55 65 75 85 95 >100  80  (12 si  80  60  cn  3 3  100  80  t  3" cn  en  50% (46 snags/ha)  ^  O (D iQ M) CD h O  H cr 3 01 0 (D Ho H CD vl  40  100  iQ C O O S hj p> 3 hi H 0 hi Q. H f £! 3 c cn T3 > O. H H (D 3 3 • W iQ S» 0 hi H £B H) CD £! f t CD (+ CD  ,-,-,0,  15 25 35 45 55 65 75 85 95 >100 Size Class(om)  (26 snags/ha)  "taamia&  15 25 35 45 55 65 75 85 95 >100 Size Class(cm)  •»  40 Table 4.1.1 Pearson correlations relating PCA axes of forest composition and structure to the first canonical axis of each variable set (CVl and CV2). Redundancies denote the amount of variation in the PCA axes explained by the canonical axes. The canonical correlation coefficient between CVl and CV2 is 0.84. Significant correlations (p<0.01) are denoted with *. Canonical axes COMPOSITION PCA axes  CVl  I  -0.77*  -0.65*  II  -0.22  -0.18  III  -0.00  -0.00  IV  -0.51*  -0.42*  V  -0.31*  -0.26*  0.03  0.02  VI  Redundancies  0.17  CV2  0.12  STRUCTURE PCA axes I  -0.82*  -0.98*  II  -0.04  -0.04  III  -0.01  -0.02  IV  -0.15  -0.18  Redundancies  0.18  0.25  41 Table 4.1.2 Pearson correlations between PCA axes of forest structure and composition accounting for two-thirds of the overall variation. Percent variance explained by axes is shown in parentheses. Significant correlations (p< 0.01) are denoted with *.  Composition PCA axes  I  (25%)  Structure PCA axes I II III (32%) (14%) (13%)  IV ( 8%)  0.67*  II (14%)  0.19  111(12%)  -0.02  IV ( 8%)  0.36*  V  0.26*  ( 7%)  VI ( 5%)  -0.04  -0.06  -0.11  -0.02  -0.03  -0.03  0.02  0.35* -0.00  0.01  0.16  -0.04  0.34*  0.01  0.42  -0.01  -0.17  0.15  0.10  42 Oplopanax horridum, Clintonia uniflora, and (Table 4.1.3).  Blechnum spicant  Species negatively correlated with axis 1 showed  increased importance at lower elevations where the mature forests were located.  These included Douglas-fir, Gaultheria shallon,  Pteridium aquilinum, Berberis nervosa, and Eurynchium oreganum. The second axis of the composition data represented a gradient from plots dominated by Douglas-fir (tree) and western hemlock (sapling) to plots dominated by western hemlock and Pacific silver fir (both trees). The first PCA axis of the structural data indicated a "sizeabundance" gradient in the major components of forest structure (Table 4.1.4).  Negative correlations between axis 1 and the  total densities of living and dead trees and the numbers and sizes of stumps resulted in young and mature forests being placed toward the left; conversely, positive correlations between axis 1 and structural variables expressing the sizes of living and dead trees (both standing and fallen), as well as with total cover of shrub-herb layers and sapling density have drawn old forests toward the right. The second axis of the structural data did not correlate strongly with any of the original structural variables. 4.1.3 Structure-composition correlation: Separate age classes. The results of the separate age class PCAs are summarized in Table 4.1.5 and Fig. 4.1.4.  Mature forests showed stronger  correlations between composition and structure than young and old forests where the correlations were not significant. The ANOVA results for study area differences within the separate forest age classes are summarized in Table 4.1.6.  The  first PCA axis of young forest structure showed the strongest  43 Figure 4.1.3 PCA ordinations of plots based on compositional and structural data. Plot symbol shapes denote study area locations: triangles, Capilano watershed; squares, Seymour watershed; circles, Coquitlam watershed; diamonds, University Endowment Lands. Plot symbol shadings denote forest age classes: open ('young'); hatched ('mature'); solid ('old').  COMPOSITION  CM CO X CO  < O  <y  fa,.. < * * K  0  AFS •  is* o • „AA A A H  „r  A^Y-,  _  -1  -0.5  •1.5  0.5  1.5  PCA a x i s 1  STRUCTURE i  I  A  CM  « < o a  •  A  X CO  A  •  A  0 RSI E E3 A  0  AA  A  G  »  »HH«S'  V*AA  A  •e-  A  0  A  •  A  A *  •  • •  •  • *  A  •* / \ *  A  oo  i  1.5  • A  H  A Q  El*  • A  I  -0.5  0.5  PCA a x i s 1  1.5  44 Table 4.1.3 Pearson correlations relating plant species to the first two PCA axes of the composition data. Percent variance explained by PCA axes is shown in parentheses. Only those species with significant correlations (p<0.01) are listed.  PCA axis 1 Abies amabilis (tree) Abies amabilis (sapling) Abies amabilis (seedling) Berberis nervosa Blechnum spicant Clintonia uniflora Cornus canadensis Eurynchium oreganum Gaultheria shallon Hylocomium splendens Lycopodium selago Maianthemum dilatatum Oplopanax horridum Plagiothecium undulatum Prunus emarginata (tree) Pseudotsuga menziesii (tree) Pteridium aquilinum Rhytidiadelphus loreus Rubus pedatus Sorbus sitchensis Sphagnum spp. Streptopus amplexifolius Thuja plicata (tree) Thuja plicata (sapling) Thuja plicata (seedling) Tiarella trifoliata Tsuga heterophylla (tree) Tsuga heterophylla (sapling) Vaccinium alaskaense  (25%) 0.45 0.72 0.59 -0.31 0.61 0.29 0.44 -0.46 -0.72  PCA axis 2 (14%)  0.30 0.35 0.32 0.44 0.38 -0.66 -0.44 0.34 0.41  0.30 0.52 0.28 0.35  0.34 0.46 0.46 0.39 0.51 0.66 0.65  0.41 0.27 -0.64 0.66  45 Table 4.1.4 Pearson correlations relating structural variables to the first PCA axis of the structural data. Only significant correlations (p<0.01) are shown.  Structural Variables  PCA axis 1 (31%)  Tree density  -0.56  Tree mean diameter  0.76  Tree max. diameter  0.76  Tree total basal area  0.71  Snag density Snag mean diameter  -0.64 0.51  Snag max. diameter Snag total basal area  0.30  No. of large trees(>80cm)  0.81  No. of large snags(>40cm)  0.36  No. of stumps  -0.63  Stump mean diameter  -0.60  Stump max. diameter  -0.67  Stump total basal area  -0.70  No. of logs Log mean diameter  0.50  Log max. diameter Log total basal area  0.33  Herb total cover  0.69  Shrub total cover  0.57  Sapling density  0.49  46  distinction between study areas (r^ = 0.53); however, the proportion of total structural variance explained along the axis was low (23%).  Axis 1 of mature forest composition showed a  moderate distinction between study areas (r2 = 0.43) and a higher proportion of variance explained (41%).  Study area differences  within old-growth forest composition and structure were practically negligible (r2 < 0.1) and proportions of variance explained were low (<25%).  47 Figure 4.1.4 PCA ordinations of plots in the separate forest age-classes based on species composition (top row) and structural data (bottom row). Symbol number indicate study area locations: Capilano watershed (1); Seymour watershed (2); Coquitlam watershed (3); Pacific Spirit Park (4).  1  I  I  1  CO CO  Q CO  -I  o  eo  eo *"e> (0  -  n  r  CO  **>  r-  coT ^  CO X  <  CO  «  CO  JW * 3  -  c9  CO  rT  «,  "  eo  ,_  CO CO  CO  1  1  •  i  Z SIXV  Z SIXV 1  •  1  i  ] CM  •<r  LLi DC 3  <f«  "  -  •tf CM - g  CM CM  <"  cM,r CM  CM  N  «  o CM  ™  X  <  DC CM  <N  CM  CM  X  CM CM  •»  CM  *  **  •*CM  CM  <  cV  «CM N  CM  CM CM  * •  CM  CM CM  CM  «  CM  *  CM  1  1  1  1  1  Z SIXV  Z SIXV  1  CD  -  z o  1  1  -  CO CO  CO CO  2  X  <  , - * •  >T- * " , -  * ~ <r-  *• 1  Z SIXV  z sixv  -  48 Table 4.1.5 Structural-compositional linkages as shown by Pearson correlations between the first two PCA axes (I and II) of composition and structure within separate forest age classes. Percent variance explained by axes is shown in parentheses. Significant correlations are denoted with *.  Young I II (23%)  (22%)  I (32%) 0.36 11(20%) 0.34  -0.33 <0.10  Structure Mature I II (24%)  Old I  II  (16%)  (24%)  (15%)  0.58* 0.13  (24%) -0.17 (15%) 0.27  -0.24 0.14  Composition (41%) -0.48* (15%) 0.52*  49 Table 4.1.6 Proportions of total sums of squares along PCA I and II axes (r2 values from ANOVA) ascribed to study area locations.  (a) Young forests (Capilano and Coquitlam) Composition  Structure  I  II  I  II  0.29  0.12  0.53  0.02  (b) Mature forests (Seymour and Pacific Spirit) Composition  Structure  I  II  I  II  0.43  0.08  <0.01  0.42  (c) Old forests (Capilano and Coquitlam) Composition  Structure  I  II  I  0.01  0.07  <0.01  II <0.01  50 4.2 Structure and dynamics of three old-growth cedar-hemlock forests in southern coastal British Columbia. 4.2.1 Size and age distributions of tree populations. Stand MERC102 The populations of each tree species in the Mercantile creek plot exhibited reverse-J size distributions (Fig. 4.2.1) indicating a certain level of compositional stability (i.e. no regeneration failure). With many small individuals in the understory, a decreasing number of larger trees, and a multilayered canopy, overall plot structure was similar to that of old-growth Douglas-fir forests from Washington state (Franklin et al. 1981).  Western hemlock had the highest density,  represented by many stems in the sapling category and a declining number in larger size classes. Hemlock canopy trees did not exceed 58 cm in diameter.  Western redcedar was second in  abundance, but dominated the canopy layer with several large specimens reaching 158 cm in diameter.  Alaska yellow-cedar was  most abundant in the first two size classes with declining numbers in larger size classes and only a few trees larger than 100 cm dbh.  Pacific silver fir and western yew populations were  mostly represented as understory trees (<30cm dbh). The age structure of each tree species showed a decrease in abundance with increasing age, a characteristic of uneven-aged populations (Fig. 4.2.2). The number of trees in the youngest age class were underestimated because seedlings (<lcm dbh) were not included.  The age distributions did not allow a breakdown of the  tree populations into distinct cohorts, suggesting a continuous  51 Figure 4.2.1 Relative densities of trees in 10cm diameter size classes in MERC102. N refers to the total density of live and dead trees.  J  in  in  • * T—  •<*•  in CN T—  m CM  ,—v  E  E  o  2~  c o  CO T3  1 o  2  5 -  ^ *:d"  E if)  m  Q  m <u <o "5  r/>  <u  £ CD  SP  2^  c  CO ^'•i [/••']  L_  a  |j  CO TO CO  :>  s^ ^J -£:  CW  i n (U CO <u  R  §<  -t-»  L_  D  85 idpoi  I  o _i  m o  I.O o *•—'  <D  1-  t-H  in CM  in  o  o  o  o  CD  "3-  CM  CD  (%) uaa |9d  • -  o  o CM  (%) U9Q p y  in  m •<* T—  m  in  CM T  -  CM  /^^1  I  E  -- m o  o *—' c o  1  m a. 00 T3 2  V)  .2  a  ~-4  ID  < }  II  ts E  r-  55  • •  ^  It)  1  —  o  o  o  CO  •*  CM  (%) usa 'lay  CO  Q  oo  t  in CM  ^^•M^ i  |  m CD .22  -3: S  i—1  I  l_  a  00 v-i  in  m  (%) "usa "isy  * 8I  I •• I  52 Figure 4.2.2 Relative densities of living trees in 50 year classes in MERC102. N refers to the total density of aged trees.  o o o  " o o CO  ,—, m <D tl>  >>  03  o o c CO o  «ms  1  ^ &. •5 G % P  </i co  II  Q. 73  /  /  / \>  o o • *  1—i—1—  10 o CM CM  in CO  (%) U9Q | 9 y  o t  in  ro  <1 o in oCM m T— CO CM  (%) uaa • p y  o  in  ro  </i  (D 05  o o CM  o CO  in  o cu  fc  in CO  2  —1—1—1—1—1—1—1—'—1—1—1——  o  m o in <o CM CM i i (%) uaa |ay  CO  o CO  m o  CM  CM  m c  (%) UBQ lay  m  <  53 regeneration process.  The range in ages illustrated the ancient  character of this forest, particularly for western redcedar which varied between 52 and 946 years. Western hemlock ranged between 50 and 620 years, yellow cedar between 58 and 548 years, Pacific silver fir between 84 and 486 years, and western yew between 60 and 33 3 years. Stand EC183 Although the overall size class structure of the Eastcap Creek plot EC183 was similar to that of the Mercantile creek plot, there was a substantial difference in the presence of Douglas-fir. This species showed a bimodal size distribution with peaks around 95 and 135 cm possibly illustrating the presence of two cohorts (Fig.4.2.3). However an examination of the age distribution revealed a distinctive cohort in the 290-300 year age class.  The size variation within this cohort arose from  radial growth variation probably reflecting a range of competitive release.  Western hemlock and western redcedar were  represented in all size classes and were the dominant canopy trees along with Douglas-fir.  Similar to the Mercantile creek  plot Pacific silver fir and western yew were primarily understory trees.  Again, tree size distributions of each species exhibited  reverse-J curves except in the case of Douglas-fir, which was the only species showing regeneration failure.  The density of stems  was particularly high in this stand especially for western hemlock (1508 stems/ha) and Pacific silver fir (532 stems/ha). The age structure profiles of western redcedar and western hemlock were similar, exhibiting a decrease in tree abundance  54 Figure 4.2.3 Relative densities of trees in 10 cm diameter classes in EC183. N refers to the total density of live and dead trees.  "-  ID  -- •v— CM in 00  lO  CD E  m "I  to  co  •s E CD to  ••  ^™-  m en  w////i^m I B  3 s>  • "  lO -- CD  3 to ?5  _i  a  c o  sz  Q.  CD CO CO o CO  E  (1)  •5  T—  II  ---  -- m "S3  s  2cc  r r  b  m  CO  1  .  1  1  1  o  o  CO  CD  '  H  >  O •si"  1  oo  ]  o  CN-g-  cu •s: Q  S?  Z  cc  • "  Sc 5j  ft,  o o  D  CO  E  •••  & ^3  z  CM  O  CO  wm~ S  s  CM CM II  to  o  - in - p  F (i  < z co co  E  LU  UJ  •'  '  i  o o  O CM  1  1 —- f -  'I"  o  o  00  CD  (%) U9Q Pcd  '  1  r~ — i — i — t  O • *  (%) U8Q  lO  O CM  py  m in  cr o CM  CM  m co E m o  i co c •""" 3  i n *—'  -4-*  "i  „ o  CO  Q  E  "g  in <D  0) to ^f CO CN| II  in Q. CM T3  CN-§"  ^j  -1—1  en  E  ^  I  'Is  CD  i n CD CO  e  £  5  b in  co  o o (%) U8Q |9ed  o  o  00  CD  O • *  (%) U3Q |9M  O CM  55 Figure 4.2.4 Relative densities of living trees in 10 year classes in EC183. N refers to the total density of aged trees.  t 00 O  o o  •s:  o oo  1  G la  "4= w  1m  q +3  oo  ^* a, a  <o <N  'g> fi  I  O O  o  o  00  CD  O  o  (%) U3Q "lay  (%) U8Q |8d  o  o  o o  o  o  00  CO  o  (%) U9Q isy  56 with age and a small peak in the 290-300 year age class which corresponded to the Douglas-fir cohort (Fig. 4.2.4).  The  synchronous establishment observed in the 1690 period probably reflected a pulse of regeneration following the same disturbance event.  Only two trees were older than 300 years (not shown in  Fig. 4.2.4).  One was a 692-year-old redcedar and the other a  426-year-old western hemlock. The range in ages (17-692 years) was considerably less than that at Mercantile creek. Stand EC163 Eastcap Creek plot EC163 was more similar to MERC102 than to EC183 in tree species composition and age structure. All tree populations exhibited a reverse J size structure.  The high  density of Pacific silver fir (718/ha) reflected the higher elevation of this stand, with most individuals in the sapling category and a rapidly decreasing population in the larger size classes.  Western hemlock had a similar size distribution;  however, the abundance of canopy trees was higher.  All Alaska  yellow-cedar individuals were in the sapling category, except for two canopy trees. Western redcedar was represented in a wide range of sizes from 1cm to 365cm.  Western yew, although  relatively abundant, was not tallied as a tree in this stand because stems had a layering growth habit making ageing and counting difficult. The tree populations all showed a wide range of ages.  As in  MERC102 western redcedar had the greatest range with trees varying between 36 years and 1002 years.  The age distribution of  western redcedar was very irregular, with several peaks and gaps. This pattern may reflect variations in successful tree  57 Figure 4.2.5 Relative densities of trees in 10cm diameter size classes in EC163. N refers to the total density of live and dead trees.  in  in  CD  CD CO  co m  in CM CO  CM CO  in  m  oo  00  Is  1 1  CO  c  I  CO  cu 0  « SP  £  in  CO  S*.  o en II  +•»  to D5 CD C  §<  4-J  m N  D  m o  in o o a. CM -a  I  D) O _l  £^  !in ^o  r"  3  E-H  i-  oo  a  H  in  m o oo  o  o  o  <D  •*  CM  1  o 00  (%) U8Q |3Jd  t  '  1  o  o  o  CD  •*  CM  (%) U9Q |9cJ  CD CO  lCD CO  in CM CO  in CM  to  in 00  in  CM <£>  00  E  c,? in o  in o CM-g in o o a. CM "a ••*•  I  " - -  m o  <s  CD  r-  >-  1  - 3 u> 8  00  I  CM T -  I  |  m2  •8  , 0 Q. S-CM T3  in CD  C CO  CN II  Q m  in 00  %  00 in  m  •*  m  o  o  00  CD  o (%)  'U9Q|9y  o  o  CM  00  CU CD  -s «D  o CD  o  o  -*  CM  (%) U9Q | 8 a  E b  Figure 4.2.6 Relative densities of living trees in 50 year classes in EC163. N refers to the total density of a trees.  CM  |v. (0  S.-S  GO  ^D  §1  <3  i  (%) U9Q p y  £• & 9  "5  r» CS  f-H -s;  m  o  in  o  CO  fO  CM  CM  If) CM  O CM  (%) U30 1 ^  -s « 1 1^ Q  ^t-  >  II  e  55  o  If) CO  O CO  If) CM  o CM  if) T-  (%) U9Q | 9 y  o  • *  if> CO  O CO  If) •<-  (%) U8Q | a y  59 establishment and/or variations in mortality.  The age  distributions of Pacific silver fir and western hemlock corresponded fairly well to their size distribution;  however,  the spread appeared wider in the age distributions.  Saplings of  Alaska yellow-cedar varied between 25 and 150 years, and two canopy trees were slightly older than 300 years. There appeared to be no distinct cohorts resulting from a single disturbance event for any of the tree populations in this forest. 4.2.2 Radial growth patterns. Although it is tempting to infer tree age from diameter, many studies warn against the impreciseness of this extrapolation (Stewart 1986b).  The correlation between diameter at breast  height and age at stump height was significant for all species (p<0.001), except for Douglas-fir.  The examination of  scattergrams (Fig. 4.2.7), however, revealed a substantial amount of variation in the linear relationship between age and diameter. This variation arises from: 1) A severe suppression in some members of a distinct cohort either from factors such as competition, pathogens, or abiotic conditions. 2) The effect of small-scale disturbances (gap-phase dynamics) which create new habitat for fast growing seedlings in the open and/or provoke a competitive release of trees already present. These results suggest that caution must be exercised when making age predictions from diameters in temperate rainforests of British Columbia.  Nevertheless, tree size distributions remain  60 .7 Scatter diagrams of age (years) versus diameter (cm) for Pacific silver fir, western redcedar, and western hemlock in MERC102, EC183, and EC163. R2, and the total number of trees are included in each diagram.  61 useful in assessing forest structure and in predicting general trends of change in forest composition. All tree species (except Douglas-fir and western yew in EC183) in the three plots showed an increase in mean radial growth rates from sapling stage to pole stage, and from pole stage to tree stage (Table 4.2.1).  One way of interpreting this  result is that the mean growth rate reflects an average measure of competitive release. showed different trends.  A similar study by Lusk and Ogden (1992) In the forests of Tongariro National  Park, New Zealand, the mean radial growth rates peaked in either the sapling or pole stages, but not one species peaked in the tree stage.  The authors interpreted their results as a  reflection of the stage at which the individual trees were most likely to be in the open, suggesting that the species which had the highest growth rates as poles were overtopped as saplings. Although every tree is unique in having recorded its own growth performance with respect to the macro and micro environment, it is possible to recognize several common growth trends (Fig 4.2.8). Eight of the most common growth trends observed in this study are: 1) Fast and sustained early growth with a gradual decrease. 2) Fast and sustained early growth followed by a severe suppression. 3) Suppressed saplings. 4) Suppressed early growth followed by one or more releases. 5) Saplings with moderately fast early growth. 6) Canopy trees suppressed throughout their existence.  62 Figure 4.2.8 Examples of various radial growth patterns found in the temperate rainforests of southern coastal British Columbia, (a) Radial growth increment of an Pacific silver fir from EC183, and (b) corresponding cross section. (c) Radial growth increment of a western redcedar from EC183, and (d) corresponding cross section, (e) Radial growth increment of a Western hemlock from EC183, and (f) corresponding cross section. Arrows represent beginning of a growth release unless indicated otherwise, (g) Radial growth increment of a Douglas-fir, (h) a western hemlock and (i) a western redcedar, established following a catastrophic fire in EC183. (j) Radial growth increment of a western redcedar from EC16 3 showing multiple growth releases. Radial growth increments were measured over decades. a)  20 R a d i a 1 i n c r e m e n t  15  10  5  m m 0 _ 18 95  i..  1905  1915  1925  1935  1945  1955  Decade midpoint  b)  1965  1975  1985  63  Figure 4.2.8 c o n t ' d .  c)  30  r a  d i a  25  20 I  n c r e m e n t  15  I  10  i  m m 0 1890  1900  1910  1920  1930  1940  1950  Decade midpoint  d)  1960  1970  1980  64  Figure 4.2.8  e)  cont'd.  25 r R  a  d i a I n c r e m e n t m m 1885  1895  1905  1915  1925  1935  1945  Decade midpoint  f)  1955  1965  1975  1985  Figure 4.2.8  cont'd.  Following fire  g) R a d a I i n c r e m e n t  r  \ ^A  A  m m  1111 8 8 8 8 0 1 2 3 5 5 5 5  1 8 4 5  1 8 5 5  1 8 6 5  1 8 7 5  1 8 8 5  1 8 9 5  1 9 0 5  1 9 1 5  1 9 2 5  1 9 3 5  Decade midpoint  R a d a I  25  r  Following fire  20  \ 15  n c r e m e n t m m  10  5h 0^ 1 6 9 5  1111 7 8 8 8 9012 5555  1 8 3 5  1 8 4 5  1 8 5 5  1 8 6 5  1 8 7 5  1 8 8 5  Decade midpoint  1 1 1 1 8 9 9 9 9 0 1 2 5 5 5 5  1 9 3 5  1 9 4 5  1 9 5 5  1 9 6 5  1 9 7 5  1 9 8 5  Figure 4.2.8 c o n t ' d .  Following fire R a d i a I i n c r e m e n t m m  1 7 0 5  1 7 1 5  1 7 2 5  1 7 3 5  1 7 4 5  1 7 5 5  1 7 6 5  1 7 7 5  1 7 8 5  1 7 9 5  1 8 0 5  1 8 1 5  1 8 2 5  1 8 3 5  1 8 4 5  1 8 5 5  1 8 6 5  1 8 7 5  1 8 8 5  1 8 9 5  1 9 0 5  1 9 1 5  1 9 2 5  1 9 3 5  1 9 4 5  1 9 5 5  1 9 6 5  1 9 7 5  1 9 8 5  Decade midpoint  111111111111111111111111111111111111111111111111111  445555555555666666666677777777778888888888999999999 890123456789012345678901234567890123456789012345678 555555555555555555555555555555555555555555555555555  Decade midpoint  Table 4.2.1 Mean diameter growth rates (mm/year) and ranges in tree age and diameter at breast height in the three plots. N t refers to the total number of trees; N c refers to number of cross-sections examined. Parenthesized values are standard deviations. Species  Age range DBH range  MERC10 2  NN t  t  NN c  Saplings Saplings  Poles  Trees  (years)  (cm)  Pacific silver fir  57  55  0.36 (0.12)  0.82 (0.2)  1.81 (0.39)  84-486  2-47  Alaska yellow-cedar  66  53  0.65 (0.26)  1.01 (0.26)  2.0  58-548  2-14  Western yew  21 21  20 20  0.33 (0.16) 0.33 (0.16)  0.70 (0.20)  1.26  60-333  1-42  Western redcedar  151 151  135 135  0.50 (0.15) (0.15) 0.50  0.78 (0.24)  1.48 (0.32)  52-946  1-15  Western hemlock  166 166  158 158  0.40 (0.16)  0.76 (0.22)  1.12 (0.32)  59-620  1-58  214  190  0.83 (0. ,37) (0.37)  1.64 (0, ,51) (0.51)  2.64 (0.75)  22-180  1-40  Douglas-fir  10 10  10 10  4.25 (0.99) 286-305  73-16  Western yew  10 10  4 4  29-160  3-25  Western redcedar  91  Western hemlock  c  (0.69)  EC183 Pacific silver fir  1.38 1.38  1.17 (0.46) (0.,46) 1.17  63 63  (0.,39) 1.13 (0.39)  (0,,42) 1.62 (0.42)  3.07 (1.04)  23-692  1-18  636  543  ,39) 0.91 (0. (0.39)  (0,.62) 1.62 (0.62)  2.46 (1.01)  17-426  1-13  340  292  0.44 (0.22)  0.84 (0.27)  1.53 (0.33)  24-413  1-40  EC163 Pacific silver fir Alaska yellow-cedar Western redcedar Western hemlock  15  13  0.54 (0.16)  1.53 (0.39)  69-422  1-77  45 45  37 37  0.48 (0.08) (0.08) 0.48  0.64 0.64  2.73 (1.05)  36-1002  2-360  157  138  0.44 (0.22)  0.73 (0.24)  1.43 (0.42)  22-608  1-10  68 7) Complacent canopy trees that showed no peculiar growth patterns. 8) Irregular growth patterns resulting in non-circular uniformity, usually accompanied by compression wood. The cross-sections that displayed non-circular uniformity reflect tree trunk movement as a result of unstable substrate, wind, and/or gaptracking.  The resulting growth patterns were  exceedingly complex, difficult to interpret, and illustrated the limitation of tree coring in coastal forests of British Columbia. Nonetheless, if growth releases were significant they could be interpreted even in sections with asymmetric growth (Fig. 4.2.8c). 4.2.3 Disturbance history. Radial growth release and regeneration The first notable instances of radial growth release occurred in the year 1150 for MERC102 (Fig. 4.2.9), 1730 for EC183 (Fig. 4.2.10), and 1540 for EC163 (Fig 4.2.11). Plots MERC102 and EC163 are most similar in that both have an extended history of growth releases and tree establishment, although MERC102 appears to show a more ancient disturbance history with several releases occurring between 1130 and 1540.  Another  similarity is the continuous nature of disturbance events and regeneration patterns through time. This illustrates the dynamic nature of these forests.  The increased number of releases and  tree establishment in recent history (last 300 years) reflects a better historical record for this period as a result of the greater number of trees that have survived to this age. Although  69 Figure 4.2.9 Temporal pattern of disturbance and regeneration in MERC102. Regeneration refers to the number of trees established in each decade. Evidence of disturbance history is represented by percentages of trees alive in 1991 that showed a growth release for each decade.  (%) 9SB9|9J 6U|M0L|S S99J1  O CM CD CD CO CO  O  00  J» CD  O O  I  o o>  o '•*-<  to 0) T— T3  o CO  CD O CD  Q  CO 0)  c CD CD  o CO  o  CD CM O  o o 00  CD  ^  CN  (Bl|/S99JJ_) U0!JBJ9U969^|  k  70 Figure 4.2.10 Temporal pattern of disturbance and regeneration in EC183. Regeneration refers to the number of trees established in each decade. Evidence of disturbance history is represented by percentages of trees alive in 1992 that showed a growth release for each decade.  (%) 9SB9|9J 6U|M0L|S S99J1  lO  o  0 <f) CO 0 0  g CO s—  CD  c 0  D) 0  o o CO  o o io o  o o o m o m  CM CM T <*(BM/S99J1) U0|JBJ9U969y  o  71 Figure 4.2.11 Temporal pattern of disturbance and regeneration in EC163. Regeneration refers to the number of trees established in each decade. Evidence of disturbance history is represented by percentages of trees alive in 1992 that showed a growth release for each decade.  {%) 9SB8|9J 6UjM0L|S S99J1  <D c/) CO 0 0  *  c  o  W—» CO l_  0  c  0 O) <D  o o o o o o o o N  (D  m  t  CO CM  r-  (eq/saeji) uojjejeueBey  72 the maximum age for tree species in the area is quite high (5002000 years old) most trees will die before this maximal age thereby destroying a record of the forest's history. The history of tree regeneration in the two older stands does not appear to be closely linked to the history of perturbations as perceived through the radial growth releases. This suggests that the process of natural disturbance did not directly influence tree establishment in these two forest stands. Stand EC183 shows a strikingly different temporal pattern of radial growth release and tree regeneration (Fig.4.2.10).  The  sequence of events in this plot is consistent with Oliver's (1981) description of stand development following a catastrophic disturbance event.  First, a distinct cohort composed of Douglas-  fir, western redcedar, and western hemlock was established in 1690. This was followed by a period of about 100 years of very little regeneration.  The fast initial growth rates of the 1690  cohort (Fig. 4.2.8g-i) were sustained for several decades confirming that these trees were growing in open conditions. Thus, it appears that a severe disturbance killed most of the trees, and was followed by the reestablishment of a rapidly growing forest which, within a decade or so, entered into a stage of canopy closure.  During that period extremely low light  conditions on the forest floor would have led to a very depauperate plant understory.  As the forest would have been very  dense in this stage many understory trees would have been suppressed and died.  The only evidence available to indicate  that this happened, is the presence of a few severely suppressed western hemlock trees from the 1690 cohort.  The few releases in  73 the 1700's provide an indication that the canopy was beginning to open coinciding with the onset of the understory reinitiation stage. In the most recent hundred years there appears to be a coupling in time and space of growth releases and tree establishment, suggesting that the recent disturbance regime is influencing tree regeneration contrary to the situation in the two older cedar-hemlock forests. A few important observations can be made concerning this coupling of growth release and regeneration.  First, most of the releases happened in a  relatively short period of time (1930-1960).  Second, many of the  trees that are part of these cohorts show a relatively fast radial growth intermediate between growth of a suppressed tree and the growth of trees in the 1690 cohort.  The proximity in  time of these disturbance events have probably contributed to the effect on tree regeneration. Spatio-temporal pattern of disturbance events To achieve a clearer picture of the spatio-temporal pattern of disturbance events, I measured the area covered by patches of trees in which synchronous release occurred in the same region of the plot (Fig. 4.2.12).  A similar approach was taken by Payette  et al. (1990) in a cold temperate forest of Quebec in an attempt to obtain a finer resolution record of gap dynamics.  I prefer to  use the term disturbance event rather than "gap" to refer to patches enclosed by trees with synchronized growth release. A disturbance event will probably affect canopy structure, either by increasing the size of an existing gap or by creating a new one.  74 The total number of disturbance event patches identified was 77 in MERC102, with a mean size of 81 ± 115 m 2 , 23 in EC183, with a mean size of 165 ± 251 m 2 , and 34 in EC163, with a mean size of 122 ± 191 m 2 .  Most of the disturbance events were in the  smallest two size classes (1-50 and 50-100 m 2 ) for all three plots.  The general size distribution was log-normal for MERC102  (Fig. 4.2.13) and EC163 (Fig. 4.2.14), and bimodal for EC183 (Fig. 4.2.15) which had a small peak of medium size disturbance events (200-350 m 2 ) .  In EC183 ten patches in the 1-50 m 2 size  class, and one large patch in the 1050-1100 m 2 size class occupied the largest surface area.  In EC163 and MERC102 the  largest surface area was comprised of patches in the first two size classes. The temporal distribution of patches indicates that disturbances occurred frequently throughout the history of stands MERC102 and EC163, although they were slightly more abundant in the last two centuries, probably reflecting a loss of information for the older disturbances.  In EC163, events in the  1710 and 1960 decades occurred over the largest surface area, whereas in MERC102 several patches in the 1870 decade covered the greatest area.  In EC183 events in the 1940s and 1950s largely  dominated both in number and surface area covered. In EC183 the spatial pattern of some release episodes also coincided with similar patterns in regeneration.  The spatial  autocorrelogram of the number of trees released and the number of trees established during the 1940 decade is shown in Fig. 4.2.16. The shape of both spatial autocorrelations is very close for the first eight distance classes.  Therefore, at small to medium  75 Figure 4.2.12 Map of disturbance patches in EC183. The spatial distribution of patches was deduced by mapping trees with growth releases. The date of disturbance is indicated.  vS  1  1  1  ^f  co  CN)  r  76 Figure 4.2.13 Spatio-temporal distribution of disturbance patches in MERC102.  „„„„ Size distribution of disturbed patches c  1200 i  „„ r50  40  Number  30  | Q>  125 225 325 425 525 Area class midpoint sq. meters  1000  800  1  625  Temporal distribution of patches  I Area  600  CD  400 o 4-* CO  200  1150 1240 i 330 1420 1510 1600 1690 1780 1870 1960 Decade  77 Figure 4.2.14 Spatio-temporal distribution of disturbance patches in EC183.  1200  Size distribution of disturbed patches  1000 CD  % CD E  ST CD CD u. CD  £  CD 0_  I Area  10  Number  800-1 6  ts CD Q.  600 CD  4  |  400 200  25  325 475 625 775 925 Area class midpoint sq. meters  1075  Temporal distribution of patches  2000  w  175  I Area  1500  10  Number  CD CD  E  .  % 1000 CD  CD  CD  CD 0-  4 500  1760 1790 1820 1850 1880 1910 1940 1970 Decade  |  78 Figure 4.2.15 Spatio-temporal distribution of disturbance patches in EC163.  1000  Size distribution of disturbed patches  16 14  800  I Area  i H Number  12  e  V)  CD CD  CD  E 600  10 o  «  8 "5 fa-  CO CD  CO Q.  ce £1  400  o  6  CO  E  CL  200  4 2 25  1600  125  225 325 425 525 625 Area class midpoint sq. meters  725 0  Temporal distribution of patches  1400 I Area  Number  1200 CD  1000 w co  800  CD  600 co  a.  400 200  1550 1600 1650 1700 1750 1800 1 850 1900 1950 Decade  79 Figure 4.2.16 All-directional spatial correlograms of number of trees established, and number of trees with growth releases during the 1940 decade in EC183. Both correlograms are globally significant at Bonferonnicorrected level a = 0.05/19 = 0.003.  -erelease cohort  0  2  4  6  8  10 12 14 16 18  DISTANCE CLASS  80 scales (e.g. 5-40 meters) the spatial pattern of trees which released in the 1940 decade and the number of trees which regenerated in the same decade were probably influenced by the same process. 4.2.4 Canopy architecture and disturbance etiology. Forest profiles and space-time mosaics The forest profiles clearly illustrated the complex architecture of the three study plots (Fig. 4.2.17-19).  Several  features were common to the three sites. 1) Multilayered canopy 2) Numerous canopy openings 3) Substantial degree of overlap in foliage 4) High degree of architectural variation within each forest. The most notable structural difference between the three sites was the height of canopy trees in EC183, which reached 70 meters.  This site was estimated to be approximately twice as  productive as the other two sites.  All of the tree species in  the understory are shade tolerant, which complicated the delineation of ecounits, or homogenous patches of trees with a similar regeneration history (Oldeman 1989).  The space-time  mosaics are a complex pattern of overlapping patches, reflecting the disturbance and regeneration processes. Canopy gaps The three plots were remarkably similar in the percentage of the forest area occupied by canopy gaps  (39% ± 2 % ) , expanded  gaps (38 ± 1 ) , and closed canopy (23 ± 2 ) . These forests have a  81 total gap area (canopy gap + expanded gap) exceeding 70%.  Canopy  gaps varied between 5 m 2 and 255 m 2 (median = 57 m 2 ) in MERC102, between 11 m 2 and 374 m 2 (median = 127 m 2 ) in EC183, and between 11 m 2 and 355 m 2 (median = 103 m 2 ) in EC163 (Table 4.2.2).  The  density of gaps was substantially higher in MERC102 with 53.2 gaps/ha compared to 30 gaps/ha in both Eastcap plots.  This may  be explained by higher wind disturbance and shallower soils at MERC102 and/or slower gap filling times.  The canopy gap size  distribution in MERC102 resembled a negative exponential with many small gaps and a steadily decreasing abundance with increasing size (Fig.4.2.20).  EC183 and EC163 had canopy gap  distributions which approached a log-normal with a more central mode. Although the expanded gap size distributions were more irregular than the canopy gap distributions, similar trends were revealed.  Stand MERC102 had, on average, smaller expanded gaps  (median = 124 m 2 ) compared with EC183 (median = 213 m 2 ) and EC163 (median = 228 m 2 ) (Fig. 4.2.21). Many studies of gap-phase dynamics apply the technique to measure gap size that was developed by Runkle (1982; see also Runkle 1992), which assumes that most gaps will approach an elliptic form.  The canopy projections for the three plots  examined here (Fig.4.2.21) revealed a very diverse array of gap geometry.  Diverse patterns were also revealed by Lertzman (1989)  in old-growth subalpine forests from the same general study area.  82 Figure 4.2.17 Forest architecture and the spatial distribution of tree ages in MERC102. Contours were made by using a negative exponential smoothing algorithm on tree ages.  lOr  >  •40  83 Figure 4.2.18 Forest architecture and the spatial distribution of tree ages in EC183. Contours were made by using a negative exponential smoothing algorithm on tree ages.  84 Figure 4.2.19 Forest architecture and the spatial distribution of tree ages in EC163. Contours were made by using a negative exponential smoothing algorithm on tree ages.  -30  t-o  >-  85 Disturbance etiology Most canopy gaps were associated with more than one gapmaker (dead trees > 30cm dbh) (Fig. 4.2.22).  The mean number of  gapmakers per gap was 3.6 ± 3.6 for MERC102, 3.9 ± 3.2 for EC183, and 2.4 ± 1.9 for EC163.  The density of gapmakers was  substantially higher for MERC102 (200/ha), compared to EC183 (122/ha) and EC163 (80/ha).  This may partially explain the  higher number of canopy gaps in the Vancouver Island forest. Western redcedar was consistently the dominant gapmaker, >39% , although it was equalled by western hemlock in stand EC183. There was a significant difference in disturbance etiology between the three sites.  Only in EC183 was there any evidence of  trees killed by a catastrophic fire (Table 4.2.3-4).  Evidence  from a few trees in an adjacent unburned patch, which responded to the fire with a growth release, indicated the event occurred in 1683-1685. This date concurred with the peak of regeneration in 1690. All of the charred trees were western redcedar, 50% of which were still standing.  The decay time for these large cedars  is probably >500 years after death, which is at least five times longer than the other tree species.  Some of these charred dead  cedars appeared to have been uprooted or snapped during more recent disturbance events that occurred between 1930-1960. The percentage of uprooted trees was fairly high in the three sites, ranging from 28% to 39% (38% to 52% if only known events are included). There appears to be no particular mode of mortality, excluding the fire-killed cedars, associated with a particular species (Table 4.2.4); however, in the two oldest stands  86 Table 4.2.2 Summary statistics for canopy gap measures CG= canopy gap EG= Expanded gap. N Min.  Area (m2) Max. Median  Mean  SD.  Plot MERC102 CG  24  5  255  57  76  72  EG  24  28  362  124  147  106  CG  15  11  374  127  133  104  EG  15  76  663  213  265  178  CG  15  11  355  103  122  84  EG  15  43  553  228  249  56  Plot EC183  Plot EC163  87  Figure 4.2.20 Size d i s t r i b u t i o n s of canopy gaps for the t h r e e study a r e a s .  25  75  125  175  225  275  325  375  225  275  325  375  225  275  325  375  12 10  EC183  </> ro D)  " 6 E  i 4  25  75  125  175  12 10  EC163  a 8 6  75  125  175  Canopy gap area class midpoint (Sq.m.)  88  F i g u r e 4 . 2 . 2 1 S i z e d i s t r i b u t i o n s of e x p a n d e d g a p s f o r t h e study areas.  25  125  225  325  425  525  625  25  125  225  325  425  525  625  25  125  225  325  425  525  625  Expanded gap area class midpoint (sq.m.)  three  89 Table 4.2.3 Number of dead canopy trees (> 30cm) in each decay class by type of mortality. Fire Snapped Standing Uprooted Unknown Stand MERC102 FRESH  0  0  3  3  0  Dl  0  3  8  6  0  D2HARD  0  12  6  15  5  D2-D3  0  4  0  15  20  Total  0  19  17  39  25  FRESH  0  0  1  0  0  Dl  0  5  2  0  0  12  3  1  12  2  D2-D3  2  1  1  7  12  Total  14  9  5  19  14  FRESH  0  1  0  3  0  Dl  0  3  2  1  0  D2HARD  0  0  9  1  0  D2-D3  0  2  1  6  11  Total  0  6  12  11  11  Stand EC183  D2HARD  Stand EC163  90 Table 4.2.4 Number of dead canopy trees (> 30cm) of each species by type of mortality. Fire Snapped Standing Uprooted Unknown Stand MERC102 Pacific silver fir  0  3  1  0  4  Alaska yellow-cedar  0  1  1  0  0  Western yew  0  0  0  1  0  Western redcedar  0  9  13  20  12  Western hemlock  0  4  2  0  1  Unknown  0  2  0  18  8  Total  0  19  17  39  25  0  1  0  3  0  14  1  2  5  2  Western hemlock  0  7  3  7  7  Unknown  0  0  0  4  5  14  9  5  19  14  Pacific silver fir  0  4  0  5  2  Alaska yellow-cedar  0  1  0  0  Western redcedar  0  0  4  1  Western hemlock  0 0  1 1  0  2  Unknown  0  0  0  2  6  Total  0  6  12  11  11  Stand EC183 Douglas-fir Western redcedar  Total Stand EC163  0 10 2  91 Figure 4.2.22 Frequency distributions of the number of gap makers per gap in the three study areas.  MERC102  •  EC183 EC163  ,  2  1  f  3 4 5 6 7 8 9 10 11 12 13 Number of canopy gapmakers  92 most of the trees that died upright were western redcedar (>76%). It was not possible to determine the cause of mortality in 23% to 28% of the cases.  Most of these were in advanced stages of decay  and, in many cases, a substantial portion of the original tree had disappeared making a correct assessment difficult (Table 4.2.4) . 4.2.5 Forest turnover times Estimates of forest turnover times and other disturbance regime parameters are provided in Table 4.2.5.  Tree fall  frequency calculated for the 1760-1980 period was almost identical between the two Eastcap plots (0.11 and 0.12 per year), whereas it was double for MERC102 (0.22/year).  Treefall rotation  estimates indicated that 212 years were required to disturb the 4500 m 2 area of MERC102, while 291 years and 362 years were required for the 5000 m 2 areas of EC183 and EC163, respectively. The forest turnover times estimated from the percent area in canopy gap are given for a range of times required to fill a canopy gap (Tf-j^i = 150 to 400 years) based on the observed range of ages in trees entering the canopy (30cm). range between 375 years and 1096 years.  Turnover times  According to the range  of Tfills and turnover times trees will remain in the canopy between 225 years to 696 years. 4.2.6 Spatial pattern of vegetation. Tree species The results of spatial autocorrelation analysis on the density of saplings, poles and trees for each species are shown in Tables 4.2.6-8.  If the regeneration of trees was influenced  93 Table 4.2.5 Disturbance regime parameters estimated from released patches and from percent area in canopy gaps. Tf-Qj and Tres refer to time required for trees to fill a gap and to time trees spend in the canopy (cf. Lertzman and Krebs 1991). Estimates of Tf^-Q represent the range of tree ages for stems entering the canopy (30 cm dbh).  MERC102  EC183  EC163  Tree fall frequency  0.22/year  0.11/year  0.12/year  Tree fall free interval  4.4  8.9 years  8.4 years  Tree fall rotation  201 years  268 years  326 years  years  MERC102 T  fill  150  200  250  300  350  400 years  T  res  225  300  375  450  525  600 years  375  500  625  750  875  1000 years  Turnover time EC183 T  fill  150  200  250  300  years  T  res  225  300  375  450  years  375  500  625  750  years  Turnover time EC163 T  fill  150  200  250  300  350  400 years  T  res  260  348  435  522  610  696 years  410  548  685  822  960  1096 years  Turnover time  94 by canopy structure we should find patches of understory trees that correspond to canopy gaps. The temporal distributions of disturbance and tree establishment suggest that such patches should be found in plot EC183, but not in plots EC163 and MERC102.  Moran's I index has, indeed, detected strong patterns  of autocorrelation for all understory tree species in EC183 (Table 4.2.7).  All of these correlograms are globally  significant at a = 0.05. Saplings and poles of western hemlock and western redcedar show a patchy distribution throughout the plot corresponding to a cyclic pattern of Moran's I. The density of Pacific silver fir represents a clear gradient, as Moran's I values change gradually from positive and significant in the first half of the distance classes, to negative and significant in the second half of the distance classes.  Fig. 4.2.23 captured  this gradient as there are very few Pacific silver fir in the left portion of the plot and there is a large patch in the right section of the plot.  In plots MERC102 (Table 4.2.8) and EC163  (Table 4.2.9) only the density of Pacific silver fir saplings and poles have a globally significant structure, showing significant autocorrelations in the first distance class. This result does not necessarily mean that Pacific silver fir is clumped in canopy gaps, but may reflect a pattern of seed dispersal, as this species has the heaviest seed of all the species studied. The relationship between canopy structure and abundance of understory trees was assessed by comparing densities for each species recorded under closed canopy, expanded gap, and canopy gap.  Kruskall-Wallis tests revealed there were no significant  Table 4.2.6 All-directional spatial correlograms of number of saplings, poles, and trees in plot MERC102.2 Moran's I spatial autocorrelation index was calculated from 180 adjacent 25 m quadrats. The first distance class corresponds to pairs of neighboring points on the sampling grid. Significant values (p<0.05) are indicated. Pacific silver fir Distance Class 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17  Western redcedar  Sapling Pole  Tree  Sapling  ns ns ns ns 0.06 ns ns ns ns ns ns ns ns ns ns ns ns  ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  0.06 ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  Western hemlock  Pole  Tree  ns ns -0.05 ns ns ns ns ns ns ns ns ns ns ns ns ns ns  0.07 ns ns ns ns -0.05 ns ns ns ns ns ns ns ns ns ns ns  Sapling ns ns ns ns -0.05 ns ns ns ns ns ns ns ns ns ns ns ns  Pole ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  Tree ns ns ns ns -0.04 ns ns ns ns ns ns ns ns ns ns -0.12 ns  Ul  H H M M 4^ CO t o H  1 1 i 1 1 1 1 1 0 0 0 0 0 0 0 0 3  * • •U l • •co *t o •H •H *. CO co - J t o  H  UD 00 - J O"! U l £>. CO t o M  o o  H3  able 4.2. tree from pair are  H  istance lass  H t-> H H KO 00 v j  o  1  3 3 o  CO CO CO  O^ U l  •o  3 3 3 CO CO CO  •o •H •CO •U l  -J Ul M  en  o  o o o o o  00 CD V  hd Ul  H-  o HHl H-  o o o o o 3 o o o o 3CO C3O 3CO C3O 3CO o o o o  V  o  H H M H H \ ) 00 t o 00  I-1 CD  1  1  1  1  1  1  o  -J  *-  1  H  o  1  1  . • •H •t o o o 00 00 .&. CO  o o  !£> 00  o  CO H-  < CD H  3 3  3  3  CO  CO CO  CO  1  3 o CO3 CO • to  CO  1  1  O  O  1  3 3 3 3 3 3 3 3 3 CO CO CO CO CO CO CO CO CO CO CO CO  o C3O 00  1  H  ^ H  H)  CD CD  H  Ul Ul  53 CD CO  H-  • o *>  3 3 CO CO  o o  • • o o <&CTl  V H-  rt  3  CD H 3  hj  CD  • M  o M  0-  oo  CD  H  3 3 3 3 3 3 3 CO CO CO CO CO CO CO  o a\  o  • o *>  3 3 CO CO  o  o CD Ul H  1 3 3 3 3 3 3 3 3 CO CO CO CO CO CO CO CO  o 3 o 3 CO CO  • o  • o  3 CO  o o  1  o C3O 3CO  3 CO  o  • o  o  o  3 3 CO CO  4=-  \1  1  1  3 3 3 3 o C3O CO CO CO CO • o  3 CO  3 CO  o  o 03  1  3 3 CO CO  Ul  .  H o co VO  3 CO  3 o to • o  \1  M  li CU  H- H- QJ  g  H - H - hj  3  =3 CD  H-  Ul  CO  •  H  o  to  o  rt (-• H3  o  o  3 CO  Ul  V  o CD  1  3 3 o o CO CO •  H)  Ul  CD H 3 3CD  3 3 3 3 CO CO CO CO  tJ M  o o  H CD CD  1  o o o o  rt H o rt co o  h3  1  3 3 3 3 CO CO CO CO  Ul  3 3 3 3 3 3 3 3 3 3 3 3 CO CO CO CO CO CO CO CO CO CO CO CO  H H CD CD  3  o 0  M  saplings poles, index wa calcula ass corr sponds t ficant v lues (p<  3 3 CO CO  3  3 3 CO CO  Ul  3 3 3 CO CO CO  M  umber of relation stance c d. Sign  3 CO 3 CO3  CO  (T(D w n  s of utoco rst d ng gr  4^ .p* CO  o C3O  1  .H •H H• •  o> o>  3 3 3 CO CO CO  o  o o 3 o o o o C3O • • CO M  3 CO  1  3 3 3 CO CO CO •  to  1  o o  !> H M I f t Hhi CD  rrelog spatia The he sam  •H H• O  o  3 3 3 3  -J  tiona spatial C183. Moran's nt 25 m2 quadr oring points o  3  H - CO CO 3 to Q, O O H H- H) O 3 O Ul 3 Ul t J rtlD & H (D H - C J . O f i i Q cu r t • 3" O  Ul  CD CO -.  O O f t O i CD 3 O OJ 0 J Ul  Table 4.2.8 All-directional spatial correlograms of number of saplings, poles, and trees in plot EC163. Moran's I spatial autocorrelation index was calculated from 200 adjacent 25 m 2 quadrats. The first distance class corresponds to pairs of neighboring points on the sampling grid. Significant values (p<0.05) are indicated. Pacific silver fir Distance Sapling Pole Class 1 0.19 0.14 2 0.10 ns 3 ns ns 4 -0.06 ns 5 ns -0.05 6 -0.05 ns 7 ns ns 8 -0.04 ns 9 ns 0.06 10 ns ns 11 ns 0.08 12 ns ns 13 ns -0.08 14 ns -0.09 15 ns ns 16 ns ns 17 ns ns 18 ns 0.17 19 ns ns  Tree ns ns ns ns ns ns ns ns ns ns ns ns ns ns -0.10 -0.10 ns -0.14 ns  Western redcedar Sapling 0.07 ns ns ns ns 0.03 ns ns ns ns ns ns ns ns -0.08 ns ns ns ns  Western '.hemlock  Pole  Tree  ns ns ns ns ns ns ns ns ns ns 0.06 ns ns ns ns ns ns ns ns  ns ns ns ns ns -0.05 0.06 ns ns ns ns ns ns ns ns ns ns ns ns  Sapling ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  Pole  Tree  ns ns ns ns ns ns ns ns ns ns ns -0.07 ns ns ns ns ns ns ns  -0.07 ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns ns  98 Figure 4.2.23 Spatial distribution of the density of western hemlock and Pacific silver fir saplings in plot EC183.  cP  99 relationships between any of the understory tree species and my canopy classification. The mean densities of tree seedlings in the three study plots are shown in Table 4.2.9.  Western hemlock was the most  abundant species on elevated substrata, regardless of age class. In EC183, western hemlock was 28 times more abundant than redcedar.  Redcedar and western hemlock seedlings (both first  year seedlings and older) were always more numerous on elevated substrata.  Pacific silver fir showed greater abundance on the  forest floor in the Eastcap plots but the situation was reversed in MERC102.  There were very few seedlings of western yew  (present only on logs) and yellow-cedar (equally present on logs and forest floor).  Douglas-fir was more abundant on the forest  floor but was present only as first year seedlings, illustrating again its regeneration failure. Understory vascular plants The organization of understory vascular plant communities (excluding tree species) was similar in the three plots. However, some differences appeared in physical structure and species composition.  The species importance curves (Fig. 4.2.24)  approached lognormal distributions, characterized by few dominant species, few rare species, and many species intermediate in abundance.  This suggested that similar processes may be involved  in shaping community structure.  The physiognomy of these  communities was different as the shrub layer in MERC102 was very tall reaching 3 meters while it was only 1.5 meters in EC163 and <1 meter in EC183.  The general species composition was similar  among the three stands as they shared many species. However,  100 Table 4.2.9 Mean density/m 2 (±SD) of tree seedlings on the forest floor and on elevated substrata in the three plots. Number of 1 m 2 quadrats examined in each plot shown in parentheses. >1 year old Elevated  Ground  Pacific silver fir  0.05 ± 0.24  0.01 ± 0.08  ALaska yellow-cedar  0.01 ± 0.11  0.01 ± 0.15  Western yew  0.01 ± 0.08  0.00 ± 0.00  Western redcedar  0.58 ± 1.91  0.18 ± 1.57  Western hemlock  0.94 ± 3.11  0.10 ± 0.86  MERC102 (n=169)  First year Elevated  Ground  EC183 (n=200) Pacific silver fir  0.00 ± 0.00  0.00 ±  Douglas-fir  0.07 ± 0.30  0.10 ± 0.31  0.00 ± 0.00  0.00 ± 0.00  Western redcedar  0.61 ± 2.36  0.05 ± 0.36  0.02 ± 0.23  0.01 ± 0.07  2.07 ± 6.87  0.29 ± 1.68  Western hemlock  17.40 ± 26.46 7.50 ± 15.20  EC163 (n=200) Pacific silver fir  0.00 ± 0.00  0.00 ± 0.00  1.02 ± 3.42  4.06 ± 6.00  Alaska yellow-cedar  0.00 ± 0.00  0.00 ± 0.00  0.00 ± 0.00  0.03 ± 0.22  Western redcedar  2.58 ± 9.72  0.88 ± 3.42  0.22 ± 1.19  0.11 ± 0.87  Western hemlock  9.69 ± 20.42 10.92 ± 20.41 4.42 ± 11.72  1.18 ± 5.12  101 Figure 4.2.24 Species rank curves for understory vascular plants in the three study areas.  100  O w <U  MERC 102  10  o c <u 0)  0.1 10  15 20 25 Species sequence  100  30  35  30  35  30  35  EC 183  c <D O CD SX  >. o c <u 3  cr  10 15 20 25 Species sequence  100 T*  EC163 c CD  o <u  % 10 c CD  o<D  10  15 20 25 Species sequence  102 there were some differences associated with site characteristics (Table 4.2.10).  In MERC102 the dominance of Gaultheria shallon  and the presence of Vaccinium ovatum were characteristic of forests near the coast on Vancouver Island (Gagnon 1985).  In  EC163 the abundance of Vaccinium alaskaense and the presence of Valeriana sitchensis were characteristic of higher elevation submontane forests.  In EC183 the presence of Asarum caudatum,  Polystichum munitum reflect the higher soil nutrient richness of this site. The spatial autocorrelograms of the first PCA axes of species composition were all globally significant (p<0.05) (Fig. 4.2.25).  The spatial trends in EC183 and EC163 were very  similar, changing from significant positive autocorrelations at small distance classes, to significant negative autocorrelations at intermediate distance classes, and back to significant positive autocorrelations at large distance classes.  This  pattern is characteristic of a patchy distribution (Legendre and Fortin 1989).  Vascular plants in MERC102 exhibited a spatial  gradient, as the autocorrelations gradually changed from significant and positive at small distance classes, to significant and negative at large distance classes.  The patchy  distribution of EC183 and EC163 reflected distinct communities associated with streams and small depressions characterized by Veratrum viride and Lysichiton americanum.  In MERC102 the  spatial gradient results from a distinct community, also with standing water located at only one end of the plot.  103 Table 4.2.10 Frequency of plant species in MERC102, EC183, and EC163. Number of 25 m 2 quadrats examined in each plot shown in parentheses. MERC102 (180) Adiantum Aralia Asarum Blechnum Boykinia Carex Chimaphila Clintonia Coptis Corallorhiza Cornus Drvopteris Dryopteris Drvopteris Gaultheria Gaultheria Goodyera Gymnocarpium Habenaria Linnaea Listera Lycopodium Lycopodium Lvsichitum Maianthemum Menziesia Montia Oplopanax Pyrola Polypodium Polystichum Rubus Rubus Rubus Smilacina Sorbus Streptopus Streptopus Taxus Tiarella Vaccinium Vaccinium Vaccinium Vaccinium Veratrum Viola Valeriana  pedaturn nudicaulis caudaturn spicant elata sp. menziesxi uniflora asplenifolia maculata canadensis expansa filix-mas phecropteris ovatifolia shallon obloncrif olia drvonteris dilatata borealis cordata clavaturn selacro americanum dilatatum f errucrinea sibirica horridum secunda crlycyrrhiza muniturn pedatus spectabilis ursxnus stellata sitchensis amplexifolius roseus brevifolia (shrub) trifoliata alaskaense ovalifolium ovaturn parvifolium viride crlabella sitchensis  99 1 1 41  EC183 (200) 2 4 2 86 8 3 62 12  99 1  72 37 14 1  99  4 10 14  EC163 (200)  90 6 4 11 94 57 2 98 12 4  68 1 2 93 56 9  16 3 10 2 9 12 5  1 1 33 37  26 26 6 1 8 18  4 90 47 3 97 1 14  66 10 88  46 20 2 8 36 2 8 15 44 2  85 18 7 15 43 20 17 100 13 84 9 9 6  104 Figure 4.2.25 All-directional spatial correlograms of first PCA axis of understory vascular plant data showing multispecies pattern for the three study areas. All correlograms are globally significant at Bonferonnicorrected level a = 0.05/19 = 0.003.  0.5 0.25 ~c CO  0  o  •0.25 •0.5 0  5  10 Distance Class  ^MERC102 —EC163  15  EC183  105 Bryophytes The spatial distributions of bryophyte species richness along four 50m transects from the Eastcap watershed are shown in Fig. 4.2.26.  The transects located in EC183 and in an adjacent  900 year old patch (also of fire origin) have strikingly similar patterns of species richness despite a 600 year stand age difference.  The transects located in EC163 and in an adjacent  patch were also similar in patterns of species richness except for a peak at the beginning of EC163B.  In general the peaks in  species richness correlated with the number of substrata, available and the presence of intermittent streams. Since species composition was similar among the four transects, I decided to combine the data for the speciessubstrata analyses.  In order to assess variation in species  composition within and among substratum categories, a PCA was performed on a matrix containing presence/absence data of the 44 species distributed in 221 plot-substratum units.  These units  correspond to individual substratum categories for each 25 m 2 . Some substrata were not present in every plot resulting in 221 instead of 500 plot-substratum units.  Box plots of the first two  axes of the PCA are shown in Fig. 4.2.27a.  Despite a substantial  amount of overlap in species composition among substrata KruskalWallis tests on the first four axes of the PCA revealed significant differences among substrates.  There is a high degree  of variation in species composition within most substratum units, except for the shrub-sapling unit, depression, and mineral substrates which have the lowest number of species. In order to examine the overall ecological distance in species composition between substrata a DCA ordination was made  106 Figure 4.2.26 Distribution of bryophyte species richness in 25 m 2 quadrats along four 50 m transects in old-growth cedar hemlock forests (solid line). The number of available substrates in each plot is represented by the lower dotted line. Arrows signify the presence of intermittent creeks. The age of the stand in which the transects were located is indicated.  m  *•? o -K> •^- <u  ID i T CO U <D  •og CO (0  £ ! <=» CM c o  O " CM 0)  o •"" 1 r-,b  CN  i-  •=-  CM  ssauqou sapeds  sseuipu sapods  /  /  -  /  o -2S CO JS  -a o CO  p •>  \ V  o o o >—1  \ . X.  A  1  CM  1-  •«-  sseuipu s a p s d s  o co  1  m c» CM c _g o <"  —^•'•V  5a  CM  CO CO L. 4-*  H—  in CM  i  —  i  —  O CM  i  —  i  —  IT) T-  •  —  CM d)  •/  i  —  i  O t-  ssauipu sepads  —  i  —  o • « (6  i  —  107 Figure 4-2.27 Box plots on the first two PCA axes by substratum using 221 units and b) DCA ordination of 10 substrata. l=Epiphytes on trees, 2=Epiphytes on shrubs and saplings, 3=Fresh decaying wood, 4=D1 decaying wood, 5= D2hard decaying wood, 6=d2-d3 decaying wood, 7=Humus, 8=wet depression, 9=intermittent creek,10=mineral soil and rock. -i  1  1  1  1  1  1  r—i  r  "T  I  I  I  I  ~T  I  I  '  i  i  I  T"  «) ~ 2 < 8  111  5! 1  1  £  n  T T  t  i  i  1  Z  3  i  4  i  6  i  1 T  6  i  0  7  0  i  B  T I  8  1 _  _2 i  10  8ub8tmte  i  .i  i  1  2  3  ' *  i 6  8  7  8  Substrata 1  *>)  10 1  -  3  ss x a)  < O Q  i  i  3  9  40  i  1  -1 -1  8_  7  0 -  0 DCA axis 1  -  ' 8  ' 10  108 on a matrix of the frequency of 44 species distributed in the 10 substratum categories (Fig. 4.2.27b).  There appeared to be a  successional gradient on wood along the first axis in the order: epiphytes -+ freshly fallen -* hard wood -*• soft wood -+• humus. The communities on creeks, depressions, and mineral substrata were more distinctive as they appeared well separated on the ordination. The frequency of occurrence on the ten substrata and niche breadths are shown for all 44 species in Fig. 4.2.28.  Niche  breadth varied from near 0, in the case of a species which occurred on only one substratum, to 1 in the case of a species equally present on all substrata.  Although many species had  fairly wide niche breadths, it was possible to recognize eight groups based on substrate preferences. The epiphyte group was characterized by Dicranum fuscescens, Douinia ovata, Frullania nisquallensis, Hypnum circinale, Isothecium myosuroides, and Ptilidium californicum.  Niche  breadth in this group varied from 0.52 to 0.88. The freshly fallen group contained Antritichia curtipendula, Metanekera menziesii, Porrella navicularisf and Ulota sp.  The  species in this group had a more restricted distribution with niche breadths varying from <0.01 to 0.28. The hard decaying wood group was composed of Bazzania denudata. Calypogeia trichomanis, Jamiesionnella automnalis, and Lepidozia reptans. All these species had fairly high niche breadths (>0.79) except for Calypogeia trichomanis which was only present on this substratum.  109 Figure 4.2.28 Species/substrate relationships and niche breadth for 44 species of bryophytes in old-growth cedar-hemlock forests. Niche breadth values vary from near zero, in the case of species restricted to one substrate type, to one in the case of ubiquitous species. The frequency of occurence of each species is given for the following substrata l=Epiphytes on trees, 2=Epiphytes on shrubs and saplings, 3=Fresh decaying wood, 4=D1 decaying wood, 5=D2hard decaying wood, 6=d2-d3 decaying wood, 7=Humus, 8=wet depression, 9=intermittent creek,10=mineral soil and rock.  Antitrichia curtlpendula  Bazzania denudata Niche breadth = 0.88  10o T Niche breadth=0.28  80  100 80  60  60  40  40  20  20  0  1  2  3  3  5  Calypogeia  6  7  8  910  80  80  • •1  1  100 80  2  3  3  1  2  5  6  7  3  3  5  6  7  8  9  8  9  |  3  3  5  6  7  8  9  Diplophylfum albicans  Dicranum fuscescens  Niche breadth = 0.81  10  1  2  3  3  5  6  7  8  9  II  10  •  1  0  1  60 40  20  20  2  3  3  6  -. 6  Heterocladium  7  0 8  9  10  5  6  7  8  9  n  1  2  .1  T o  10  ' T ' o ' c ' o ' T ' o ' o 10  2  3  1  40  20 0  1  2  3  • • ••1  3  5  6  7  8  20 9  10  0  2  3  1  2  3  5  6  7  8  9  3  5  6  7  . 8  1  5  6  7  8  9  10  • . 1  2  3  3  5  6  7  8  9  10  macounii  40 j 2  3  3  •  20 j  5  6  7  8  i  0 I—.—i—i—i—i—i—i—JLM. 1 2 3 3 5 6 7 8 9 10  9 10  Hypnum  circinale 100 -I  dieckeii  Niche breadth < 0.01  80  •  •  60  | | _ _  1 i; i • 2  3  3  5  6  7  8  9  |  40  i  llllll. 1.  20 1  10  Jamiesioniella automnalis  2  3  3  5  6  7  8  9  1 1  10  Leucolepis acanthoneuron Niche breadth < 0.01  „  Niche breadth = 0.80  100  100 80  T  60  40 { 0  3  Heterocladium  |  60 I 40  1  3  Niche breadth = 0.19  Niche breadth = 0.88  10  2  60 J  Hypnum  9  lunuiifoiia  Niche breadth = 0.30  0  10  n • • • „ •  20  10  Diplophylfum taxlfolitim  60 |  3  1  100  80 |  IIIL.  9  100 j 80 )  L 1  3  • I  0  scoparium  J J  lucens  100 80  8  20  0  _ 5• 6_ 7• 8 9 1110  7  llli.  20  10  40  Isofhecium myosuroides  60  9  40  Frullania nisquallensis  Niche breadth = 0.77  40  8  6  60  Niche breadth = 0.59  is  1  60  7  5  80  60  oreganum  i  Hylocomium splendens  T  6  3  80  1  100 80  5  3  Cephalozia  20 9  2  Niche breadth = 0.64  40  80 60 40 20  20  Niche breadth = 0.64  3  1  bicuspidata  60  80  10  3  •  0  10  T  100 j  40  3  8  9  . .Ill 1  80 60  3  7  8  .1. 1  100  40  2  6  Hookeria  60  1  5  20  7  •  Niche breadth = 0.43  T  0  3  6  T  procurrens  Niche breadth = 0.39  20  3  Eurynchium  40  1 .  2  5  80  Niche breadth = 0.60  60  1  • .  I.llll.  20  100 80  3  Niche breadth = 0.74  100 80  Douinia ovata Niche breadth = 0.52  100 |  3  Dicranum  80 40  2  Niche breadth = 0.57  T  0  1  100 |  20  Niche breadth = 0.55  20  0  40 •  2  40  .nil.  20  60  1  60  |_  Cepha/oz/a  Niche breadth = <0.01  10  40  piperi  80  •  40  10  20  60  100 80  L.llli I  •  40  60  "  • | 1 III  60  •III. 1  20  Buxbaumia  Niche breadth < 0.01  100  •  60  Calypogeia trichomanis 100  40  80  0  100 j  trichophyllum  T  muelleriana  Niche breadth = 0.77  60  Blephorastoma  Niche breath = 0.71  100  1 •  _  40  I.llll.. 1  2  3  3  5  6  7  8  20 9  10  0  1  2  3  3  5  6  7  8  9  •  10  110  Figure 4.2.28  Lepidozia  cont'd.  Lophocolea heterophylla Niche breadth = 0.79  reptans  Niche breadth = 0.79 100 80 60 40 20 0  100 T  T  III ._lUl. 11 2 3 3 5 6 7 8 9  80 60 40 20  -III.  1 2 3 3 5 6 7 8 9  10  80 60 40 20 1 2 3 3 5 6 7 8 9  10  1 2 3 3 5 6 7 8 9  100 j 80 60 40 20  III! .  1 2 3 3 5 6 7 8 9  80 1 60  10  ._ 1  10  l.lllll.l. 10  80 60 40 20  1 iill  LUlLd 1 2 3 3 5 6 7 8 9  10  100 T 80 60 40 20  1 2 3 3 5 6 7 8 9  Racomitrium 80  10  heterostichum  Niche breadth < 0.01  40  •.  20  10  12  -•  10  1  10  100 80 60 40 20  ; .  __•••  .1.  1 2 3 3 5 6 7 8 9  Ulota SP. Niche breadth = <0.01  u  1 2 3 3 5 6 7 8 9  •  iVlo'  Riccardia multifida 100 80 60 40 20  •_•!  3 3 5 6 !  Niche breadth = 0.49  Sphagnum girgensohnll Niche breadth = 0.41  l l -  T  60  1 2 3 3 5 6 7 8 9  10  T  •  100 80 60 40 20 0  Rhvtidiopsis robusta Niche breadth = 0.72  Niche breadth = 0.89 100 80 60 40 20 0  •  1 2 3 3 5 6 7 8 9  Scapania bolanderi  Scapania americana Niche breadth = 0.22  • III  I  Porella navicular's  100 T  1 2 3 3 5 6 7 8 9  10  -  10  Niche breadth = 0.25  100 j 80 60 40 20  H I Nil III  °LULJJX  _ 1  1 2 3 3 5 6 7 8 9  Racomitrium aauaticum Niche breadth < 0.01  100 j 80 60  100 ,  T  10  1 2 3 3 5 6 7 8 9  Rhvtidiadelohus loreus Niche breadth = 0.97  Rhizomnium alabrescens Niche breadth = 0.80  1 2 3 3 5 6 7 8 9  100 j 80 60 40 20  10  Niche breadth = 0.65  — , ,1, I.I. 1 • • I.I. 1 2 3 3 5 6 7 8 9 10  1 2 3 3 5 6 7 8 9  1 2 3 3 5 6 7 8 9  Ptilidium californicum  Pseudotaxiphyilutn elegans  100 80 60 40 20  Plaaiothecium undulatum Niche breadth = 0.94  mjliiL^  Niche breadth = 0.85  100 T 80 60 40 20 0  10  100 T  80 60 40 20  Niche breadth = 0.09  100 j 80 60 40 20  Niche breadth = 0.88  100 T  Pellia neesiana  menziesii  Niche breadth = <0.01  Plagiochila porelloides  Plagiotnnium insigne Niche breadth = 0.29  100 , 80 60 40 20  .  Metaneckera  10  1 2 3 3 5 6 7 8 9  10  Ill The soft decaying wood group was characterized by Blepharostoma trichophyllum, Buxbaumia piperi, Cephalozia bicuspidata, Cephalozia lunulifolia, Dicranum scopariumf Lophocloea heterophylla, Plagiochila porelloidesr Scapania bolanderi, Calypogeia muelleriana. and Plagiomnium insigne. Buxbaumia piperi and Plagiomnium insigne were rare with niche breadths of <0.01 and 0.29 respectively.  The other species were  more common with niche breadths greater than 0.59. The humus group was composed of Eurynchium oreganum, Hylocomium splendens, Plagiothecium undulatum, Rhytidiadelphus loreus, and Rhytidiopsis robusta. All species in this group had high niche breadths, varying from 0.60 to 0.97. The depression group contained only Sphagnum girghensonii, with a niche breadth of 0.41. The intermittent creek group was characterized by Diplophyllum albicans, Diplophyllum taxifolium, Hetrocladium macounii, Hypnum dieckeii, Leucolepis acanthoneuron, Pellia neessiana, Pseudotaxiphyllum elegans, Rhizomnium glabrescens, Riccardia multifida, and Scapania americana.  The  niche breadth of most species was low, (<0.50), except for Diplophyllum albicans (0.55) and Pseudotaxiphyllum elegans (0.80). The mineral group was composed of Heterocladium procurrens, Racomitrium aquaticum, and Rhacomitrium heterosticum. These species all had low niche breadths ranging from <0.01 to 0.39.  112 4.3 Landscape patterns of natural and anthropogenic disturbances. 4.3.1 The Eastcap Creek landscape mosaic and disturbance regime. The spatial distribution of ecosystem types for the area under discussion is illustrated in Fig. 4.3.1.  The montane and  submontane Coastal Western Hemlock zone form the landscape matrix composed of fairly continuous forest dissected by many creeks. The boundary between these two variants is arbitrary, based on field observations and stand boundaries from the GVWD maps.  In  reality, the submontane develops gradually into the montane. The Alpine and Mountain Hemlock zones occupy steep terrain at higher elevations.  Preliminary observations led me to believe  that the most common stand-initiating disturbances in these ecosystems were associated with colluvial material breaking from the rock faces and triggering landslide events. Although I have not detected previous fires in this zone, they probably occurred infrequently and covered fairly small areas. Avalanches do not appear to be very important in shaping forest ecosystems, but may be more important for alpine meadows and scree slope communities. In the Eastcap watershed portion of the Coastal Western Hemlock Zone, records of previous catastrophic disturbances could be detected only on a small portion of the whole matrix (Fig 4.3.2).  The matrix is, indeed, very old, with canopy trees  ranging between 300 to over 1000 years, and no physical evidence of previous large scale disturbances. The major kind of disturbance before the 1930's was fire. Fire dates did not seem to correspond with growth anomalies detected in trees throughout the watershed.  The 1690 and 1100  fires were the most severe, destroying most trees in their path.  113 Figure 4.3.1 The Eastcap creek space-time mosaic in 1992. l=CWHvml >900 yr with no sign of catastrophic disturbance, 2=CWHvml 900 yr catastrophic fire, 3=CWHvml 300 yr partial fire with 900 yr survivors, 4=CWHvml 300 yr catastrophic fire, 5=CWHvml 100 partial fire with 900 yr survivors, 6=CWHvml 60 yr partial fire. 7=CWHvml 60 yr clearcut, 8=CWHvml 15-20 yr clearcut, 9=CWHvml l-4yr logging, ll=CWHvm2 >900 yr with no sign of catastrophic disturbance, 12=CWHvm2 220 yr partial fire, 13=CWHvm2 300 yr partial fire, 14=CWHvm2 1-4 yr clearcut, 15=Mountain Hemlock Zone, 16=Alpine Zone. CWHvml and CWHvm2 refer to the submontane and montane portion of the Coastal Western Hemlock Zone. Partial fires refer to fires which were not lethal to all trees.  123" ol'3o'  114 Figure 4.3.2 Stand age distributions in the East-Cap creek landscape (CWHvml). Letters indicate type of disturbance; L=logging, P=partial fire, F=total fire, +=no catastrophic disturbance recorded.  800 -600  CD O  CD  u  -•—»  03  03  Q_  2 400 CD -Q  03 CD  E  <200 0  13  15L 60P 60L 100P300P300F900F900+  Stand age  115 The fact that these two fires were located in the same general area, suggests a fire rotation of approximately 600 years for this area.  A more recent fire partially burned a few hectares in  1890 on the same slope.  The location of these fires on the  warmest south-facing slopes (Fig 4.3.2) suggests a link between the topography of the area and the fire regime. occurred in 1870 in a montane CWH stand.  Another fire  This fire was partial  and it appears to have covered only a few hectares.  Disturbances  associated with debris flows are dispersed throughout this zone, but occupy a minor portion of the landscape.  No major diseases  or pests appear to have affected this area, although dwarf mistletoe is very common, and no major synchronous mortality could be detected using tree ring information. More recently logging has been the most prominent disturbance, both directly through harvesting and indirectly through associated fires.  This has produced essentially four  different types of patches. The first wave of logging in the 1930's affected only the southwestern portion of the study area and was much more extensive in the Capilano River valley and Sisters Creek watershed. A fire adjacent to these harvested patches produced partially burned old-growth stands of mixed forest structure.  In the 1970's three patches were harvested and  planted with Douglas-fir. These patches are presently in the pole stage of development and were recently thinned. Eighteen new patches were harvested throughout the watershed between 1986 and 1990.  Most have been planted and are still in the reinitiation  stage of development.  116 4.3.2 Structural-compositional trends along a disturbance gradient. Structural diversity represented by the first PCA axis of the matrix of structural variables (Fig. 4.3.3) indicated significant differences amongst the disturbance categories (KW test, p<0.01).  Structural diversity appeared to follow a  disturbance gradient, with low values in the 1937 logged forest group, to high values in the 900 year old forest group.  In 1690  and 1930 several stands were disturbed by fires of different severities.  Stands affected by partial disturbance (i.e not  killing all the stems) consistently had higher values on the structural diversity scale. Although species richness also differed significantly among disturbance categories, it did not follow an obvious trend along the disturbance gradient (Fig. 4.3.4).  The richest community was  the ancient cedar-hemlock forest with no sign of catastrophic disturbances for 900 years or more.  It is notable that all  stands with a previous history of partial disturbances consistently had lower species richness. 4.3.3 Successional pathways Succession vectors obtained from the ordination of standdiameter class combinations (N=132) for five tree species are shown in Fig. 4.3.5.  The information displayed represents  differences in species composition between stands and between diameter classes.  A line connecting decreasing diameter classes  for each stand represents a succession vector.  The direction of  the arrow indicates the succession trend, assuming trees in the understory will replace trees in the overstory.  All vectors come  117 Figure 4.3.3 Box plot of structural diversity (First PCA axis of forest structure matrix) measured along a disturbance gradient in the Capilano watershed. The disturbance gradient is represented by the calendar year of a disturbance event and its severity (t=total destruction, p=partial destruction, U=undetected disturbance).  <  o  1 -  >  CD  0 -  cc LLI >  Q _l < CC Z>  -1 -  h-  O Z)  cc co  -2 -  \-  -3 A09  0  V^6^0|?0^ DISTURBANCE GRADIENT  118 Figure 4.3.4 Box plot of vascular plant species richness along a disturbance gradient in the Capilano watershed. The disturbance gradient is represented by the calendar year of a disturbance event and its severity (t=total destruction, p=partial destruction, U=undetected disturbance).  ^0V^ Q ^ 0 ^^^B1°V^ 6 V^0V^ DISTURBANCE GRADIENT  119 Figure 4 3 5 Size class ordination vectors in 26 stands from the Capilano watershed. A line connecting decreasincr S f tand v e c ^ r " ?ne ? e a °of 5 S the "Present! a succession vector. The d?" direction arrow indicates the he P t i 0 n f tree f o ? C : ? a n d n i i r i f ;h h sPeciefLdicated e 7S a m e T f o r a 1 1 t°h e ™L el • ? diagrams. All vectors come from a single ordination but are displayed separately to improve clarity. ABAM=Pacific silver fir, C H N ? Y™ i l o w cedar, PSME^Douglas-fir, THPL=western redcedar TSHE=western hemlock.  1  1  1  1  l  1  -  03 CM  X  • » -  "a  -  z  < H c/)  frr  i  i t  i  i  i  ^ n « ' - 0 ' - N r )  I  I  I  Z SIXV  Z SIXV  Z SIXV I  I  1  I  1  1  I  I  1  1  CQ  X  1  1  1  z  Q  Z  <  -  -  <  I  -  11 1  < \V)  \ \  C/)  ~ Q  Z  1  < 1-  X  oT  <  \ \1  • "  -  X  I  o  <c  1  '  o 2  1  CO  CO  Q  I  UJ  \  •4—-i l  Z SIXV  .  .  CO  CM  1 T-  1 O  1 T-  Z SIXV  1  1  CM  CO  *f  <  1  1  1  CO  CM  *-  O  1  Z SIXV  1  1  120  Figure 4.3.5  cont'd.  *  •  i  i  i  i  -  -  to X  X  T*  <  <  *—  -  - Q Z  <  y~ CO  / • •  Z SIXV  Z SIXV  B  N  i '  i -  i O  Z SIXV  i '  -  i N  B  ^  121  Figure 4.3.5  cont'd.  to X  <  Z SIXV  Z SIXV 1  1  1  1  1 —I  1  1  1  1  1  1  1  -  -  X  <  CO  CO  x  X  <  "•zQ  I\  < rw  <  •tf  -  Q Z  <  -  1  I 1  1  1  1  1  1  Z SIXV  Z SIXV  r  1  • t f C O C M ' - O ' - C M C O  ^  B  i  i  N  '  -  i O  I '  Z SIXV  i -  N  1  n  *  122  Figure 4.3.5  cont'd.  1  1  -1—1  1  1  1  -  tn  x  x <  CO CM  <  Q  Z  <  V01  «*  1  1  1  1  00  CM  •r-  O  1  1  1  1  1  I  CM I  CO )  1  1  z sixv  Z SIXV 1  *" I  1  1  1  £2 <  • -  1  1  1  1  X  o  CM  CM  " Q Z  " Q  -  Z  <  < 1-  00  1  I  1  1  1  1  1  1  1  1  . 1  Z SIXV  1  1 CO  ( t n w r O i - w n Z SIXV  CM  T-  O  V  Z SIXV  123  Figure 4 . 3 . 5 c o n t ' d .  CO  x  X  <  Z SIXV  <  Z SIXV  124 from a single ordination but are displayed separately to improve clarity. The first two axes of the ordination accounted for 70 % of the total variation.  The location of species on the ordination  plane revealed two major zones.  In the left-central left portion  western hemlock, western redcedar, Alaska yellow-cedar, and Pacific silver fir are clustered together. Douglas-fir is isolated.  At the opposite end  Essentially two types of succession  vectors were recovered, long vectors parallel with the first axis and pointing to the left, and shorter vectors parallel with the second axis and pointing up or down.  Stand characteristics  listed in Table 4.3.2 reveal that the long vectors reflect some of the stands affected by fire (stands 6,6B,10,10B,11,17,19), and one logged in 1925 (stand 5).  These all had Douglas-fir in  larger size classes with no regeneration.  Most of the other  stands had remarkably similar vectors whether they were logged, burned, or showed no sign of disturbance.  Most of the logged  stands had very short vectors showing fewer strata, and greater similarity between strata.  Some short vectors parallel to the  second axis but pointing in different directions were considered too small to reveal substantial composition differences.  125 Table 4.3.2. Ecological characteristics of the Capilano forest stands. Plots Aspect Elevation Slope Disturbance date  type  STAND 2  4  62-74  320  0-24  1930  Harv.  STAND 3  4  44-86  300  25-30  1937  Harv.  STAND 4  4  60-180  380  23-38  1937  Harv.  STAND 5  4  255-260  250  0-38  1925  Harv.  STAND 15  1  160  345  0  1930  Harv.  STAND 18  1  160  435  10  1930  Harv.  STAND  6  4  166-170  400-560  30-43 1690  Fire  STAND  6B  5  145-160  450-720  40-86 1690  Fire  STAND  10  1  154  630  30  1690  Fire  STAND 11  1  180  585  45  1690  Fire  STAND 17  4  160-200  755-1225  40-80 1690  Fire  STAND 20  1  120  660  90  1690  Fire  STAND 12  3  20-350  350-450  15-65 1930  Fire  STAND 12B  1  350  465  30  1930  Fire  STAND 13  1  269  525  25  1870  Fire  STAND 16  1  150  405  15  1930  Fire  STAND 22  1  180  365  0  1930  Fire  STAND 19  1  140  755  40  1893  Fire  STAND 14  1  120  560  0  993  Undetected  STAND 10B  4  150  600  35  1090  Fire  STAND 21  1  80  1000  50  1780  Fire  STAND 7  4  90-138  640  57-85  1093  Undetected  STAND 8  4  310-354  500  17-40  1093  Undetected  STAND 9  2  320  535-550  69  1093  Undetected  STAND 9B  2  320  415-435  29-69  993  Undetected  STAND 23  2  180  750  40  993  Undetected  126 5. DISCUSSION 5.1 Old-growth/second-growth comparisons. The results have clearly shown prominent differences in compositional and structural organization between old-growth and second-growth forests in southern coastal British Columbia.  With  all age classes combined, the main compositional variation was related to climate such that higher elevation old (> 250 yr) and young (30-60 yr) forests, in cooler and wetter climatic regimes, were more similar to each other than to mature forests (60-80 yr) found at lower elevations.  In contrast, structural variation  among age classes appeared to be less influenced by climate than by the normal developmental sequences related to tree growth and biomass accumulation that occur in coastal forests recovering from disturbance. Despite the high canonical correlation coefficient, the overall results of CCA indicated that a large amount of structural variation was unrelated to species composition. This raises concerns if structure were to be managed for the purpose of attaining a certain species composition. This is not entirely in accord with Franklin et al. (1981) who suggested a strong influence of structure on species composition in Douglas-fir forests.  The combined plot ordination of species composition  (Fig. 4.1.3) displayed a considerable amount of variation within age classes. Although a full account of the variation was not possible, undoubtedly some was attributable to differences in site factors (e.g. wetter sites characterized by Sphagnum sp., and drier sites characterized by Pteridium aquilinum') , and some to differences in the actual ages of plots, particularly within  127 the old-growth group.  The results from the ANOVA on the within  age class PCAs (Table 4.1.6) showed that the mature forest group was influenced most by geographic location of the plots, illustrating the climatic differences between Pacific Spirit Park and the Seymour watershed.  Some variation has also resulted from  the nonlinear responses of species to complex gradients in abiotic (e.g. topographic, edaphic) and biotic (e.g. symbiotic, competitive) conditions (Bradfield and Scagel 1984).  Further  field and analytical study of these important ecological factors is required to clarify this. The structural diversity gradient from young to old forests is characterized by the density and size of trees, and density of large snags (Table 4.1.4).  It is worthy of note that, even  though old-growth clearly differs from second-growth along the gradient, a large amount of within-age class variation persists. This variation was moderately related to geographic location in the young forest group, probably as a result of different harvesting practises. It would have been ideal to have had all forests of different ages on similar sites in order to avoid the potential confounding of age and site effects in my analyses.  This was not  possible, however, owing to the scarcity of mature stands adjacent to old-growth in the general study area.  The difficulty  was partly overcome by analyzing the old-growth plots as a separate group.  Comparable in climate and site conditions but  spanning a broad range of ages reflecting natural disturbances (250-1000+ yr), old-growth forests displayed a relatively low correlation between their overall compositional and structural  128 variation (r2 < 0.1). This does not support the more popular view that old-growth coastal forests are highly integrated systems, but suggests a rather loose connection between the factors controlling structure and composition.  The effect of  sampling scale (900 m 2 plots) cannot be discounted as influencing my results and I suspect that smaller plots corresponding with, for example, the sizes of nurse logs or treefall gaps, might have detected the more subtle compositional-structural linkages known to exist in old-growth forests. Klinka et al. (1985) examined successional changes from regenerating clear cuts (<20 yr) to old-growth (>250 yr) stands, and identified two major vegetation associations that corresponded to the earliest and all subsequent serai stages of forest development.  Although I did not examine the very young  regenerating clear-cuts, our results also indicated that many species are shared between age classes.  Thus, it appears that  species compositional change per se is not a reliable predictor of successional development in temperate coniferous forests of southern-coastal British Columbia. A possible exception to the foregoing relates to the suggestion by Franklin et al. (1981) that although species may not be restricted to old-growth, some may need old-growth to maintain their viability.  This may be the case for western yew,  for example, whose mean cover was 7.7% in the old-growth forests that we examined, but less than 0.2% in the mature and young forests.  Western yew has been proposed as an indicator species  for coastal old-growth forests of Washington and Oregon (Spies 1991).  129 Even though the old-growth forests investigated here differ in kind from the old-growth Douglas-fir forests described by Franklin et al. (1981), many of the ecological characteristics were similar.  The large trees and snags, and a well developed  understory are clearly important characteristics associated with old-growth forests along the northern Pacific coast. Moreover, the differences in understory abundance between young, mature, and old-growth forests followed a similar pattern to those described by Alaback (1982) for coastal forests of southern Alaska.  Second-growth forests tend to have high tree and snag  densities, and numerous large stumps (remnants of logging) that provide regeneration sites for many colonizing plant species, and useful habitat for a variety of wildlife species including small mammals, amphibians and invertebrates.  Whereas many coniferous  seedlings and saplings (especially western hemlock) may be found growing on the tops of stumps in second-growth forests, it is uncertain what impact the elevated starting position will have on their long term survival.  Although Franklin et al. (1981)  described large logs as important components of old-growth forests, only minor differences between old-growth and secondgrowth were found in our study.  This disparity could have  resulted from the different sampling methods used to obtain the log data.  The triangle sampling technique used here provided a  rapid inventory of log frequency but did not allow an overall assessment of total log dimensions. The  reverse-J diameter distribution of old-growth forests  reflects near-climax forest types (Parker and Peet 1984).  This  distribution was quite similar to that of old-growth Douglas-fir  130 forests of the Pacific Northwest (Spies and Franklin 1988); however, the dominant trees of the watersheds study sites were mostly western redcedar instead of Douglas-fir.  This raised  interesting questions relating to the origins of these old-growth forests.  Eis (1962) speculated that many forest stands in this  area originated from fire, after observing similar ages of codominant and dominant trees.  If this were actually the case,  then western redcedar acted as a pioneer species similar to Douglas-fir in Douglas-fir forests. Although western redcedar has many attributes of a climax species it also has the ability to colonize disturbed sites (Parker and Johnson 1987).  Detailed  cohort detection analysis and forest history reconstructions provided in the next two sections will clarify this matter. 5.2 Stand structure and dynamics Do old-growth cedar-hemlock forests originate from catastrophic fire? In southern coastal British Columbia forest development generally occurs in large patches following catastrophic disturbances such as fire (Eis 1962; Schmidt 1970; Rowe 1972; Gagnon 1985) and hurricanes (Gagnon 1985), in intermediate patches after landslides, avalanches, small fires or windstorms (pers. obs.), or in small patches following local canopy disturbances (gap phase dynamics)  (Gagnon 1985; Lertzman 1989)  caused by a combination of allogenic and autogenic factors. The historical reconstruction of three coastal cedar-hemlock forests showed they can develop either after a catastrophic fire as in  131 stand EC183 or as a result of slow gap-phase dynamics as in stands EC163 and MERC102. The development of a cedar-hemlock forest stand following a catastrophic disturbance (Fig. 4.2.10) is similar to the classic model of Oliver (1981).  During the stand reinitiation phase the  trees established rapidly 2-5 years after the fire. An early non-woody plant community had little time to modify the site demonstrating that relay floristics (Clements 1916) is not necessary for the establishment of a forested community.  Three  tree species colonized the site: western hemlock, western redcedar and Douglas-fir. Although the fire killed almost all the trees on site, many charred cedars remained standing and may have provided sufficient shade to inhibit establishment of an extensive shrub community.  During the stem exclusion stage  (1700-1800) the dominant trees grew rapidly inhibiting the establishment of an understory.  It is reasonable to assume that  mortality of the smaller trees was probably high during this period.  This can be supported by examining the density of  standing dead trees in young forests (30-60 years old).  I  observed that in young forests many hemlock seedlings establish during the spring season but all die following periods of moisture stress. This may result from the seedlings being unable to develop enough fine roots in low light conditions to overcome summer drought. During the following stage the dominant canopy gradually opens and hemlock and redcedar which were already codominants in the canopy established in the understory.  Pacific silver fir and  western yew, not present during the first stage, also established  132 in this period.  The absence of Pacific silver fir from the fire  cohort may result from chance (no propagules made it in time before canopy closure) or from unsuitable environmental conditions following the fire inhibiting successful germination or development.  The first suggestion is more plausible because,  in this particular case, canopy closure probably occurred relatively quickly, making the temporal window for successful establishment relatively small.  Furthermore, I observed Pacific  silver fir growing on many sites as part of the initial cohort following logging, with or without burns. Western yew on the other hand is a special case. I cannot eliminate chance as a possibility, however the fact that this species is virtually absent in young forests and clear-cuts within the Capilano watershed suggests that partial shelter may be necessary for its successful establishment.  In other words, western yew  populations are maintained in the temperate rainforest landscape through the facilitation pathway (Connell and Slayter 1977). Similar observations for western yew were made in the Bitterroot Canyons of Montana by McCune and Allen (1985). In the next phase several dominant and codominant trees died, opening the canopy thereby improving environmental conditions in the understory.  As a result tree germination  increased and some saplings, already present, experienced growth releases.  As some members of the original cohort are still  present, Oliver and Larson (1990) would classify this stand as a transition old-growth forest.  According to their scheme, a  forest develops into old-growth after all the members of the original cohort following the catastrophic disturbance have died.  133 It is unlikely that this stand will ever reach this stage because the life-spans of western redcedar and Douglas-fir exceed 1000 years while the fire rotation estimated on this site is approximately 600 years. However, after 300 years of development, stand EC183 shares most of the characteristics of "true old-growth" forests from the area, perhaps the most important being the changes in forest structure resulting from gap phase dynamics. Size and age structure of tree populations The study of population dynamics of tree species is one of compromise (Harper 1977).  Ideally one would obtain complete  census data on births and deaths at many time intervals. A few studies in the northern temperate rainforests have used this procedure and have already provided preliminary data treating annual mortality rates.  However much longer time spans and  narrower time intervals are needed to gain an accurate picture of stand dynamics.  A more practical approach is to infer tree  population dynamics from age class distributions.  However, a  fully accurate interpretation of age class distributions requires information on the shapes of the mortality and recruitment curves (Johnson and Fryer 1989).  In some forests it is possible to  circumvent this problem slightly by cross-dating dead individuals.  This provides a more complete portion of the  recruitment information as well as information on mortality. Johnson and Fryer (1989) were able to acquire this information on small plots (100 m 2 ) in a fairly simple boreal forest landscape driven by catastrophic fire. However this approach also had limitations, since it is impossible to determine how many trees  134 were completely decayed.  The reconstruction of tree mortality  from tree ring records was possible for a single Douglas-fir. However most trees in coastal forests of B.C. have complacent growth or non-circular uniformity.  In these cases, cross-dating  may lead to erroneous interpretations.  Furthermore, the  disturbance regime in these forests is much more complex than in the boreal forest and requires larger plot sizes (ca. 1/2 ha), making excavations of dead trees impractical.  Successful  interpretations of tree population dynamics in southern and northern temperate rainforests were inferred from size and age distributions of trees. The size distributions of all tree populations in the three stands indicate that the cedar-hemlock forests are in a relatively stable composition state.  Douglas-fir is the only  tree species showing regeneration failure in stand EC183. Regeneration failure of Douglas-fir is not unexpected as its shade intolerant character is well known (Krajina 1969).  Spies  et al. (1990) estimated that Douglas-fir required fairly large openings (750-1000 m 2 ) to establish on mesic sites. This at least twice as large as the largest gap recorded in the present study. In general, the age structure of most tree species showed a decrease in abundance with increasing age, a characteristic of uneven-aged populations.  The range of tree ages is probably  among the widest found in plant communities.  The first age class  is an underestimation of the true density and should contain more individuals as most trees <lcm dbh would fit in this class. Most tree species have a reverse J age distribution although in stand EC183 one can also note a slight increase at the end of the  135 distribution corresponding to the 1690 fire cohort.  One  exception to this is Douglas-fir, which regenerated only during the first phase of development after the 1690 fire, hence creating only one distinct cohort.  Another exception is the  erratic temporal distribution of western redcedar in EC163. Although no distinct cohorts can be recognized in this case, the distribution has many peaks and troughs.  One may interpret the  hollows as periods of regeneration failure interrupted by periods of more successful regeneration; however the hollows probably correspond to patterns of mortality. Transition probabilities, based on overstory/understory comparisons, would predict a directional succession in which most tree species would be replaced by hemlock in MERC102 and EC183 and by Abies in EC163 because of the very high densities of these species in the understory.  Beliefleur (1981) predicted that  redcedar would be replaced by western hemlock using Markov chain models on tree census data from permanent plots in the CWH. However Beliefleur concluded that the models were inaccurate, as he considered redcedar to be a climax species for this region. Although the relative abundance of redcedar (Daubenmire and Daubenmire 1968; Gagnon 1985) and Alaska yellow cedar (Lertzman 1989) often does not appear sufficient for their maintenance in the community, their legendary longevity and wide regeneration niche allow them to coexist with the other more prolific species (Daubenmire and Daubenmire 1968; Gagnon 1985).  Western redcedar  and Alaska yellow cedar may not require a large number of young trees in order to be replaced since some individuals live between 1000 and 2000 years old; thus a single cedar tree may persist as  136 long as three generations of Pacific silver fir or western hemlock.  Furthermore, most of the understory trees will not  reach the canopy unless a partial fire removes a substantial amount of the canopy trees or many small-scale disturbances occur at approximately the same time (e.g. the 1930-1960 gap events in EC183).  In general only a few individuals will grow very slowly  into the canopy or will be released following a small disturbance event.  Hence, the role of chance in determining tree replacement  patterns is predominant making accurate predictions difficult. The regeneration niches (Grubb 1977) of western redcedar, Alaska yellow cedar, and western yew are more elaborate than other conifers in the study area as they are able to regenerate from seeds, fallen branches or layering stems. Although it is not possible to distinguish "veglings" from seedlings after a few years I have observed a small number of western redcedars growing from fallen canopy branches and from partially uprooted saplings in the three stands.  This may indicate that western redcedar has  higher survival rates in the understory and therefore more potential of reaching the canopy. The spatial distribution of seedlings related strongly to substrate type.  Redcedar and hemlock always had more seedlings  present on elevated substrata (nurse logs, tree buttresses, stumps, and root mounds).  Pacific silver fir, on the other hand,  followed the opposite trend with more seedlings on the forest floor except in MERC102 where the extremely high density of shrubs on the forest floor may prohibit seedling establihment. Gagnon (1985) obtained similar results in old-growth forests from west-central Vancouver Island.  Gagnon (1985) hypothesized that  137 this may reflect the greater sensitivity of cedar and hemlock to accumulating litter as a result of their small seed size. Similar spatial patterns occur in different forest types. In eastern North American deciduous forests, Betula alleghaniensis produces small seeds and is often restricted to nurse logs and stumps (Jones 1945; Porcier 1975; pers. obs.).  In a southern  temperate rainforest of Tasmania, Lusk and Ogden (1992) found the numbers of seedlings established on elevated surfaces to be inversely proportional to seed size.  They also mention other  studies from temperate rainforests of New Zealand in which similar observations were reported.  In alluvial Picea-Tsuga  forests of western North America the preference of seedlings for elevated substrata is probably a result of periodic floods (Gagnon 1985).  Harmon and Franklin (1989) also suggest that this  pattern may emerge from higher competition between seedlings and moss mats on the forest floor.  Although many seedlings germinate  on nurse logs the low nutrient quality of wood and the unstable nature of this microsite may be inappropriate for their survival. Trees growing on nurse logs demonstrated a wide variety of growth patterns. off" the log.  Some trees gradually grew away almost "sliding  Others grew firmly on top, developing a  substantial root system embracing the log.  The key for tree  survival on nurse logs is probably the development of an effective and stable root system.  It is not rare to observe  roots of trees following the entire length of the log (>20 meters) above ground before going below ground.  Species growing  on soft wood will eventually develop a stilted root system which will render these trees more susceptible to disturbances. Many  138 trees showed very complex radial growth patterns resulting from the swaying movement of their trunks, which may be related to their growth on a dynamic substratum (i.e decaying logs).  Logs  of western redcedar are more stable, as they take a longer time to decompose.  A 400 year old redcedar from MERC102 was growing  very well on a large cedar log of approximately the same size. The substantial range in ages of trees growing on individual logs illustrates well the relatively slow decay process in these northern temperate rainforests as well as the process of continuous regeneration. Hence the tree species of the three cedar-hemlock forests appear to be in dynamic eguilibrium resulting from their similar ecological amplitudes, differential longevity, and complementary regeneration niches. Although regeneration of Douglas-fir is inhibited in EC183, we may still consider the species to be in coexistence as individuals should persist until the next fire. These observations contradict previous beliefs that a linear succession should lead to western hemlock forests. Growth patterns The significant positive correlation between diameter at breast height and age at stump height for all tree species except Douglas-fir is confounded by numerous outliers.  Stewart (1986b)  made similar observations in old-growth Douglas-fir forests and suggested that, in this case, size class analysis alone was insufficient to interpret forest development.  The observations  from the present study show that information on age structure and growth patterns are essential for an historical interpretation of forest development.  The practical use of size class  139 distributions resides in the description of forest structure and in the projection of successional trends.  The extrapolation of  age from diameters is most successful for tree populations that show a wide range of ages, such as the redcedar populations from MERC102 and EC163.  Even for these populations, however,  extrapolation of age from diameter should be avoided. Radial growth of trees >30cm was substantially higher in EC183, reflecting the fast initial growth following the 1690 fire.  Understory trees also exhibited faster average growth in  this stand, reflecting both differences in the recent small scale disturbance regime and the more highly productive nature of this site.  The increasing mean radial growth trend from saplings to  poles to trees suggests that although some of the understory trees have released most of them remain under suppression.  Lusk  and Ogden (1992) obtained different results in an ancient southern temperate rainforests from Tasmania, where most tree species peaked in growth during the sapling or pole stage, suggesting that efficient competitive release occurred for most individuals before they reached 30 cm in diameter.  In EC183 the  variance in mean annual growth between overstory and understory trees is explained, in part, by the different regeneration processes involved (i.e. fire cohort vs. understory establishment and gap phase dynamics).  All tree species demonstrated an  equivalent ability to release from growth suppression.  All  species, except Douglas-fir, can live for a long time in the understory under suppressed conditions.  The degree of shade  tolerance of saplings may be increased through developing root connections with overstory trees (root grafting).  I have  140 observed roots of Pacific silver fir grafted onto large cedar roots in MERC102.  Similar observations were made by Eis (1970).  The disturbance regime. Forest architecture. The relative area in canopy gaps is remarkably similar among the three stands (39±2%) and is higher than most forest types, including all northern and southern temperate rainforests (Table 5.2.1), and temperate deciduous forests of eastern North America (Tyrrell and Crow 1994).  The overall gap area in the three  cedar-hemlock forests examined in this study (canopy gap + expanded gap = 77±2%) is also higher than in most temperate rainforest types. The coastal subalpine forests of Cypress Provincial Park have a similar overall canopy structure (canopy gap + expanded gap = 70%); however, 52 percent of the total open area is covered by expanded gaps (Lertzman 1989). Lertzman (1989) suggested that the high overall percent gap resulted from a low rate of filling of gaps at Cypress ranging between 50 and 200 years.  The time required to fill canopy gaps should be faster in  the cedar-hemlock forests as a result of a slightly longer growing season, and a shallow spring snowpack.  However, the  growth of most understory trees is quite slow, and while some are able to release following a disturbance, others will enter the canopy at a very slow growth rate. completely fill.  Moreover, some gaps may never  Even if some trees are approaching canopy  status within the gap, other disturbances may increase the size of the gap.  The number of gapmakers of different decay classes  associated with each gap and the spatiotemporal pattern of  Table 5.2.1 Comparison of canopy gap data for northern and southern temperate rainforests. gaps), EG(expanded gaps).  Northern rainforests  Percent area Average gap in gaps size nri CG Total CG EG 18 47 7 7 ? 14 24 7 13 26 85 7 18 42 19 7 17 7 137 7 40 77 57 124 40 80 127 213 37 75 103 228 18 70 41 203  Sitka spruce/western hemlock Sitka spruce/western hemlock Douglas-fir/western hemlock Douglas-fir/western hemlock Redwoods Redcedar/western hemlock Redcedar/western hemlock Redcedar/western hemlock Yellow cedar/western hemlock  Gapmakers per gap 7 7 >2 1 7 4 4 2 6  Stand age years 200 200 500 150 1200 1000+ 300 1000+ 1000 +  CG(canop  Source  Taylor 1990 Taylor 1990 Spies et al. 1990 Spies et al. 1990 Hunter and Parker 1993 this study this study this study Lertzman and Krebi3 19  Southern rainforests Nothofagus dombeyi Riicryphi a c o r d i f o l i a N n t h n f a g n s bp+.ul oi r l p s / n r y m i H.. p u m i l i x ) / N. h n t n l oirie.s. U. pumi1i n Podocarp Myrtaceous  fi  9 7 22 3-7 12-19 7 15.6 7 106 12.1 7 61 104 11.6 7 7 7 9 9 197 29  432 151- 200 208 173 268 7 ?  300 300 470+ 400+ 300+ 1150+ ?  Veblen Veblen Rebertus and Veblen Rebertus and Veblen Rebertus and Veblen Lusk and Ogden Armesto and Fuentes  19 19 19 19 19 19 19  142 disturbance events support this statement. Aubreville (1938) proposed the mosaic theory of regeneration after observing in a tropical rainforest, a difference in tree species composition between the arborescent and lower forest strata.  According to this theory, a forest is perceived as a  mosaic of patches composed of different species (Richards 1952). Changes in species composition were cyclic as different species replaced the dominant trees. Watt (1947) observed a similar phenomenon in many different types of plant communities and elaborated a theory of cyclic succession.  Recently, several  studies have identified groups of trees (gap specialist) which appear to depend on canopy gaps for their regeneration.  In  cedar-hemlock forests, tree species have fairly wide regeneration niches adapted to a range of canopy openings and thus do not form a gap specialist guild.  Perhaps the ability of all species to  release from suppression also accounts for the absence of a relationship between canopy structure and tree species composition.  Although gap-phase dynamics does not influence  species composition it has a strong effect on forest architecture as described in the forest profiles. Does gap-phase dynamics produce even-aged patches? Numerous studies of forest stand dynamics have utilized spatial statistics to locate patches of even-aged trees corresponding with patterns of regeneration (e.g. Williamson 1975; Bonnicksen and Stone 1981, 1982; Stewart 1986b; Duncan and Stewart 1991).  In most of these studies a significant spatial  pattern for trees of a certain size category was believed to result from gap-phase dynamics.  However, to my knowledge, no  143 effort was made to verify whether the area covered by these patches was homogeneous in tree size or age.  Duncan and Stewart  (1991) provide the only study that used actual tree ages in their spatial analysis. Although in my study spatial autocorrelation identified tree cohorts, these either overlapped substantially in space, or contained a range of tree ages exceeding 200 years.  In cedar-  hemlock forests, the process of continuous regeneration and the ability of most trees to release did not permit even-aged patches of trees to develop.  Although spatial statistics revealed some  patches of trees of similar size and age with a significant spatial pattern, a closer examination of these patches showed that they were not homogenous.  These results are consistent with  the observations made by Henry and Swan (1974), and Oliver and Stephens (1977).  The resulting space-time mosaic is a complex  arrangement of overlapping patches corresponding with the processes of regeneration and disturbances.  These results  indicate that the mosaic theory of regeneration (Aubreville 1938) and the theory of cyclic succession (Watt 1947), now combined under the mosaic-cycle concept of ecosystems (Remmert 1991), may not be as unifying as current plant ecology interprets them. Disturbance etiology As a result of an overlapping pattern of disturbance events, many canopy gaps were associated with more than one gap maker. This observation was made also for Douglas-fir (Spies et al. 1990) and mountain hemlock old-growth forests (Lertzman 1989). Gap expansion, therefore, may be a more important process than actual "creation" of new gaps.  Tree mortality in some forests  144 affected by decline also showed this trend (Klein and Perkins 1987), and it was believed to result from environmental changes, including sunscalding, associated with the opening of the canopy. In Abies balsamea forests in Eastern North America, wave patterns of tree mortality associated with wind expanded canopy openings and produced a noticeable pattern on the White Mountain landscape (Sprugel 1976).  In mountain hemlock forests Matson and Waring  (1984) described canopy gap expansion associated with mortality caused by pathogenic fungi.  I did not attempt to verify the role  of pathogens in the etiology of canopy gaps in this study. However, Dr. David Shaw (pers. comm.) observed a similar phenomenon caused by the fungus Armillaria in high graded cedarhemlock stands on the west coast of Vancouver Island.  In some  cases it may be difficult to establish whether these fungi were the principal causes of tree mortality or were present as secondary organisms.  This pattern of mortality limits the  determination of the age of a canopy opening as it may be the product of many events. Redcedar exerts a strong influence on stand structure because of its long life span and extremely slow decay rate. The conspicuous difference in decay rates between redcedar and other species may limit the use of decay class of gapmakers for a relative determination of the time of a disturbance event. Spatio-temporal pattern of disturbances The reconstruction of disturbance history showed substantial overlap of events in space.  This illustrates the importance of  using a dendro-ecological approach (Payette et al. 1990).  The  low number of disturbance events in the period from 1100 to 1700  145 probably reflects the loss of tree-ring records in live plant material.  Hence, the following discussion on rates of change  using the tree-ring record in old-growth cedar-hemlock forests is based on the 1760-1980 period. The concept of forest rotation was introduced by Heinselman (1973) to describe the time required to burn equivalent large patches in a forest landscape.  The internal dynamics of a forest  stand resulting from small disturbances was also described in terms of turnover rates for North American deciduous forests (Runkle 1982), tropical forests (Brokaw 1985), southern temperate rainforests (Veblen 1985), and for a subalpine forest of southern coastal British Columbia (Lertzman and Krebs 1991).  Two methods  for estimating turnover time were described and compared by Lertzman (1989):  1) The inverse of the rate of creation of  canopy gaps (TT1), which requires permanent plot sampling, and 2) the average time between the creation of canopy gaps at any point in the forest (TT2) calculated from the proportion of forest in canopy gaps and the time required by trees to fill these gaps (Tfj^).  A new method, the Treefall Rotation Index (TRI)  (Payette et al. 1990) estimates the time required for treefalls to disturb an area equivalent to the size of the plot surveyed. This method requires detailed information on the spatio-temporal pattern of disturbances. The forest rotations (TT2) estimated in old-growth cedarhemlock forests are high, ranging between 375 to 750 years for EC183, 375 to 1000 years for MERC102, and 411 to 1096 years for EC163.  The variation in turnover rates within a stand represents  the observed range of time required for trees to enter the canopy  146 (defined here as 30 cm dbh).  The rates of change in these  forests are slower than those of tropical rainforests (TT2 = 60375 years) and of North American hardwood forests (TT2 = 156-204 years) but comparable to southern temperate rainforests (TT2 = 345-794 years) and to subalpine coastal forests from British Columbia (TT2 = 658 years) (rates provided in Lertzman 1989). The estimated treefall rotations for the three cedar-hemlock forests were much faster than turnover rates estimated from canopy gap data (TT2) as they ranged between 212 to 362 years. This discrepancy results from different definitions of canopy gap used in the calculations of TRI and TT2.  The estimation of TT2  requires application of a structural definition of a canopy gap, usually defined as the area equivalent to the vertical projection of the canopy opening.  In order to define and measure a canopy  gap, one must first define where the canopy begins, which is often difficult in forests with a complex architecture (Brokaw 1982; Payette et al. 1990).  In the calculation of the treefall  rotation index, a canopy gap is defined by Payette et al. (1990) as "the surface area under the direct influence of a canopy opening, recorded by individuals showing the same, specific treering signature associated with patterns of suppression and release."  This index gives a sharper resolution of the effects  resulting from disturbances and is similar to the idea of expanded gap (Runkle 1982).  If the proportion in expanded gaps  is used to calculate TT2 the estimates are still higher than TRI. This suggests that canopy disrupting disturbances have effects beyond the measured expanded gaps.  147 The treefall rotation index is useful in order to measure the time it takes for an area to experience the influence of disturbances.  However, it does not measure in the strict sense,  the amount of time required for the turnover of living space, and is therefore complementary to TT2, and not really comparable. Unfortunately only one study has reported values of TRI. A TRI of 45 years was estimated by Payette et al. (1990) for a yellow birch-maple forest of south-central Quebec.  This value is four  to seven times lower than the TRI of old-growth cedar-hemlock forests of southern British Columbia. The Treefall frequency was highest in MERC102 and consistent with the higher density of canopy gaps in this stand.  The  proximity of MERC102 to the open coast suggests that this area would be more affected by strong winds than is the relatively sheltered Eastcap creek watershed.  Several studies have  documented the importance of wind as a disturbance factor on the coast of Vancouver Island (Gagnon 1985; Keenan 1993).  The tree  fall free interval (TFI) is short for the three stands (4.5-10 years) when compared to values of Tfj_±±.  Frequent disturbances  and slow canopy gap filling rates explain the open character of old-growth cedar-hemlock forests in southern coastal British Columbia. Understory plant communities Vascular plant communities in the understory exhibit similar importance curves, illustrating that common processes are involved in the three cedar-hemlock forests.  The lognormal  distributions indicate that a number of environmental factors have different effects on different species (Whittaker 1975).  148 This type of community organization appears to predominate in old-growth rainforests from southern coastal British Columbia (see vegetation tables in Gagnon 1985). The spatial pattern of the overall species composition was significantly autocorrelated.  The strongest autocorrelation was  in the first distance class, which may reflect patterns of seed dispersal and vegetative reproduction.  The patchy nature of the  understory plant communities was related strongly with the presence of standing water in creeks or small depressions. The association between substrata and bryophyte species is well known (Schofield 1985).  However, few studies have examined  bryophyte community structure and its relation to substrata in old-growth temperate rainforests.  Although the ordination of 221  plot-substratum units showed considerable variation, the speciessubstratum association was significant at the community level. Species can be divided into two main groups related to woody vs non-woody substrata.  In the group of species growing on wood  there appears to be a successional trend associated with the decay categories.  However, the nature of this relationship is  "noisy" as most species have wide niche breadth.  Schuster (1949)  suggested that the distribution of hepatics on decaying logs represented one of the "few obviously distinct successions" (Muhle and LeBlanc 1975).  This trend was  observed also in  temperate deciduous forests (Jovet and Jovet 1944; Muhle and Leblanc 1975) and in boreal forests (Soderstrom 1988).  However,  an emerging property of this succession appears to be the rather loose (noisy) connection between logs of different decay categories and species assemblage.  The high number of species  149 occurring on relatively small areas on decaying wood was also observed by Slack (1977) in spruce-fir forests.  She suggested  that this may reflect a lack of competition in bryophyte communities also observed in different environments such as the Artie (Schofield 1972).  The proportion of species with low niche  breadth was highest in the creek and on mineral substrata. Lessica et al. (1991) suggested that communities of lichens and bryophytes differed in forests of different ages, reaching their optimum in old-growth.  Although I made no quantitative  measurements in young forests, I have observed generally fewer species in logged forests.  The major groups of species absent  from clear-cuts and young forests in the Capilano watershed are the epiphytes and species growing on freshly fallen wood.  150 5.3 The ecology of disturbance and patch dynamics in a temperate rainforest landscape. What is the disturbance regime in the Eastcap creek watershed? The most important large scale disturbance before logging was fire.  Two large stand-destroying fires were associated with  the topography of the Eastcap landscape as they were only detected on the south facing slopes in the submontane variant. These fires occurred approximately 309 and 900 years ago. The 1685 fire was also detected in the Seymour and Coquitlam watersheds (Table 5.3.1).  On the Olympic Peninsula, Henderson et  al. (1989) found that the last of the major fires in the little ice age also occurred between 287 and 320 years ago. They suggested that this was correlated with a period of low sunspot activity which produced drier summers, hence favoring major fires.  This may be only a coincidence but the fire of 1100 AD  also corresponds to a period of low sunspot activity. Inspection of tree ages and growth patterns in wet pockets where the last fire was only partial revealed that the 1685 and 1100 year fires largely overlapped in space.  It is therfore  reasonable to suggest a fire rotation of approximately 600 years for this section of the watershed.  A Fire return interval of 230  years was estimated for the Douglas-fir zone in the United States (Agee 1993).  Thus, the fire rotation rate estimated here seems  reasonable for the cooler wetter cedar-hemlock forests.  This is  also consistent with fire return intervals estimated for wet western hemlock forest of the United States varying between 700 years and 3500 years (Agee 1993).  The partial fires that  Table 5.3.1 Fire history in Vancouver's Northshore mountains.  FIRE HISTORY IN VANCOUVER'S WATERSHEDS .1 154016901 17901 18601 18901'2 1900S-1920S'  MAJOR FIRES(109 ha.)  SEYMOUR WATERSHED  1690 x 18401 18901'2 1920S-1935 2 1961-1990 2 19922  SEVERE SEVERE SEVERE FIRES MAJOR FIRES(612 ha.) MAJOR FIRES(93 ha.) SMALL FIRE  CAPILANO  1690-3 1870S-1900S 2 1920S-1935 2 1936-1960 2 1961-present 2  SEVERE  COQUITLAM WATERSHED  DATA SOURCES  1: EIS (1962) 2: Anon.(1991) 3: This thesis  •7  MAJOR FIRES(612 ha.) MAJOR FIRES(85+ ha.) MAJOR FIRES(93 ha.)  152 occurred in 1890 and 1780 were relatively small and difficult to detect from aerial photographs.  It is most likely that more  intensive field sampling would reveal a few more of these small fires in the watershed.  Most of the area investigated had no  indication of major catastrophic disturbances.  This suggests  that the main disturbance in this landscape, before logging, is associated with small scale disturbances, as those described in the previous section.  Small landslides appeared to be relatively  infrequent, but I did not attempt to reconstruct their history. Is logging changing the natural disturbance regime and stand diversity? Logging has significantly changed the natural disturbance regime of the Eastcap Creek landscape.  More old-growth was  converted to younger types as a result of logging in the past 60 years than was converted naturally in the last 300 years.  This  has produced a landscape with higher stand diversity because there are more stands in different stages of development. Perhaps the most significant impact of logging on the Eastcap landscape pattern is not so much the total area converted, which is rather small when compared with the amount of old-growth remaining, but the spatial distribution of new patches.  The size  of forest patches affected by logging has steadily decreased from the 1930's to the 1980's.  In the 1930's there were six patches  with an average size of 18 ha converted by logging and a fire related to logging.  In the 1970's there were three patches  created with an average area of 18 ha. In the 1980's 18 new patches with a mean size of 8 ha were harvested.  These more  recent patches are distributed throughout the submontane and  153 montane variants of the watershed.  This high degree of  fragmentation appears unprecedented in the natural history of this landscape.  From a biological conservation perspective this  raises some concerns.  We lack a firm understanding of the  ecological processes involved at the landscape level. However, recent studies in the Pacific Northwest have demonstrated that forest fragmentation is threatening the survival of a few species such as the infamous northern spotted owl (Wilcove 1988).  This  bird is a resident of the Capilano watershed (pers. obs.) and may be more vulnerable in this region than elsewhere, since it represents the northern limit of its range (Campbell and Campbell 1986).  It seems probable that other organisms, yet to be  determined, are equally vulnerable. Patterns of structural diversity and species richness along disturbance and environmental gradients. The trend of structural diversity corresponds very well with the disturbance gradient described in the Capilano watershed. Time since disturbance is the most important factor, since the ancient and young forests occupy the opposite ends of the disturbance gradient.  Old growth stands with partial  disturbances have higher structural diversity, compared to stands that were totally disturbed during the same year.  The stand that  was partially burned in 1890 has a similar value of structural diversity to stands logged in 1930. The cohort that established at that time is going through a period of intense self-thinning, which produced high numbers of small dead trees; this probably explains its proximity to the younger stands.  New forestry  practices may produce stands which closely resemble the partially  154 disturbed stands described here.  Monitoring of biological  diversity in these stands may provide a test to evaluate whether new forestry practices will achieve their noble objectives. Structural diversity was weakly correlated with plant species richness (r2 = 0.22).  The pattern of vascular plant  species richness along the disturbance gradient does not follow the classical model of plant succession in which richness gradually increases until it reaches a plateau and then drops slightly (Bormann and Likens 1981).  Connell (1978) suggested  that species diversity is highest in areas of intermediate disturbance.  The trend, as shown in Figure  exact opposite.  4.3.4, indicated the  Following a partial fire, the shelter provided  by standing dead trees and survivors probably favors the establishment of tree species.  The new tree cohort grows  vigorously forming a dense understory canopy intercepting most of the light before it reaches the forest floor.  This prohibits the  establishment of additional understory plants. Although vascular plant species richness did not appear to be related to aspect in older forests, the mean number of species was substantially higher on north-facing (24 ± 3.5) clear-cuts than on east (14.5 ± 3.5) or south-facing clear-cuts (15 ± 4.2). This may result from the higher climatic stress related to sun exposure on east and south facing sites.  Blackberry, fire weed, and pearly  everlasting were the only species restricted to clear-cuts. The information from tree size classes indicates that neither the monoclimax (Clements 1916) nor polyclimax (Tansley 1920) theories of succession apply to the temperate rainforests treated here. The climax-pattern is more acceptable as it  155 recognizes the clinal variation of tree species. However, the role of chance in the initial floristic composition (Egler 1954) following a catastrophic disturbance may weaken any theories that include the concept of convergence.  Furthermore, the role of  chance in the regeneration of tree species throughout stand development may also seriously limit the use of predictive models.  156 6. CONTRIBUTIONS TO COASTAL FOREST MANAGEMENT 6.1 The decadent argument. The management of the world's renewable resources is in a serious crisis as it is proceeding headlong towards a state of non-sustainable development.  This situation proceeds from the  uncontrolled exponential increase in the human population, human greed, and from a poor understanding or appreciation of ecological processes.  Examples are plentiful, and range from the  collapse of the anchovy and cod fisheries to the conversion of tropical rainforests to infertile land.  Ecologists cannot do  much about the problem of human demographics, or human greed, however it is clearly their responsibility to ascertain that concepts used in the ecology and management of ecosystems are sound.  The purpose of this section is to demystify the concept  of "forest decadence" and explain why it should be discarded from an ecological or a silvicultural perspective in British Columbia's rainforests. Dying forests Sometimes forests undergo a severe and sudden mortality and dieback in the dominant plant canopy.  Many biotic and abiotic  processes are involved in the development of this phenomenon. Many forest declines are considered as pathologies triggered by a combination of stresses, such as air pollution, early frosts, insect defoliation.  However, Mueller-Dombois (1987) suggested  that some declines are not triggered by diseases, but reflect a natural phase in the forest life cycle, which is followed by a regeneration phase.  The trees affected in these stand level  157 diebacks are in the same cohort and an inciting factor, such as a severe drought, will synchronize the event.  No incidence of  forest decline has been reported for the old-growth cedar-hemlock forests of southern coastal British Columbia.  Tree mortality is  usually asynchronous, as revealed in temporal patterns of disturbance, and affects trees of different ages.  The only  recorded case of forest dieback reported in southern coastal British Columbia occurred in a Douglas-fir plantation as a result of planting the wrong species on the wrong site (Gagnon 1985). What is forest decadence? The conversion of old virgin forests to second-growth forests is often considered to be a positive transformation by silviculturalists (Smith 1982).  This perspective describes old  forests as decadent, senescent, decrepit systems which are dying and their conversion is viewed almost as a rejuvenation! recently Kimmins (1992)  More  acknowledged that foresters often used  "forest decadence" to justify the harvesting of old-growth forests.  Kimmins (1992) in Balancing Act stated that, although  the term "forest decadence" may not be appropriate from an aesthetic perspective, it is a useful concept for timber production objectives.  The ecological definition of "decadence"  provided in Balancing Act reads as follows: "forest ecosystems that have not been disturbed for a long time and whose functional processes stagnate".  There are two fundamental problems with  this definition: 1) Forest ecosystems are very dynamic and recurrent disturbances are an integral part of their ecology, and 2) ecological processes do not stagnate.  However, on some sites  forests may develop into non-forested bogs (Banner et al. 1983;  158 Kimmins 1987).  This succession usually referred to as  paludification is used as an example of forest decadence by Kimmins (1992).  Although I remain in disagreement with Kimmin's  ecological definition of decadence, one may see the natural process of forest development into bogs as one of forest degradation.  As another example of a stagnating ecosystem  Kimmins (1992) described an old-growth cedar-hemlock forest from northern Vancouver Island, similar in kind to MERC102.  I have  demonstrated with the historical reconstruction of stand MERC102 that these types of forests are subject to small disturbances that continuously affect the growth and structure of the forest. The growth rates of most dominant trees are low when compared with trees growing in earlier stages of development following a catastrophic disturbance.  However, the trees are still growing  and forest dynamics are far from stagnant. An examination of the relationships between organisms and their environment, the essence of ecology, in these old-growth cedar-hemlock forests provides living proof of the impertinence of the term "decadent".  The bark of trees is covered by numerous  species of lichens and bryophytes interacting positively and negatively.  Mats of Antritichia curtipendula and Herbetus  aduncus form islands on branches in the canopy sheltering communities of other bryophytes, and microarthropods.  These mats  eventually fall to the forest floor dispersing species and contributing to the biogeochemical cycles. As trees fall new habitats are created for epixylic cryptogams, tree seedlings, fungi, bacteria, slime molds, microtines, and arthropods. The energy stored in the fallen trees is transferred through an  159 impressive diversity of organisms before it is made available for plants.  Energy transfers also occur between plants directly  through root grafts, or indirectly through mycelia-root networks. The use of the term "decadent forest" in the ecology and management of the temperate rainforest landscape is incorrect at best.  A measure of ecosystem health using a single variable,  such as productivity, lacks validity.  A more comprehensive  approach would assess the global integrity of an ecosystem, taking into account the natural disturbance regime and the dynamic interactions between organisms and their environment.  160 6.2 Fire and water: Some comments on the ecology and management of Vancouver's watersheds. The watersheds of the Capilano, Seymour, and Coquitlam rivers located on Vancouver's northshore mountains, British Columbia, Canada, provide drinking water to a population of approximately 1.5 million people.  The management of these  watersheds has a long history of controversy where logging and preserving old-growth forested landscapes are both put forward as measures of protecting and enhancing water quality.  Logging was  allowed in the early 1900s until the 1930s when Mr. E. Cleveland interdicted this practice advocating its negative impact on water quality.  Thirty years later, logging was again permitted to  protect the watersheds from decadence, fire, and insects (Anon., 1991).  Present management practices used by the Greater  Vancouver Water District (GVWD) involve applying concepts in forest dynamics to select old-growth forest stands for logging. Unfortunately, although directing succession can be viewed as a worthy resource management tool (Luken, 1990), the forest dynamics arguments used by the GVWD to justify logging are largely untested. The GVWD has chosen a pro-active, low level forest landscape management strategy for two main reasons: 1) Anticipate the consequences of natural processes and disasters. 2) Include preventative measures to reduce adverse impacts on water quality. This management strategy endeavours to manipulate the forest landscape, using silvicultural practices, in order to increase  161 "forest cover stability" and resistance to insects, disease, and fire. Although these objectives are creditable, it can be argued that they cannot be attained by converting most of the old-growth forest landscape into second-growth forests.  The management  plan, if applied, proposes to dramatically alter the Coastal Western Hemlock zone portion of the watersheds landscape over the next two hundred years (Fig. 6.2.1). The cornerstone of the GVWD forest landscape strategy is based on the belief that the outcome of forest succession in the Coastal Western Hemlock zone will lead to decadent forests, "it is mainly because of this that forest management is required to create more stable forest landscapes" (McLennan 1991).  This  argument is based on the following untested assumptions from McLennan (1991): 1) Western redcedar, yellow cedar, and Douglas-fir are early successional species. 2) Douglas-fir and  western redcedar are more resistant to  disturbances than are western hemlock and Pacific silver fir. 3) Most sites in the Vancouver watersheds will eventually be dominated by western hemlock and Pacific silver fir. 4) Stable forests should contain more than 50% coverage of Douglas-fir, western redcedar, and/or yellow cedar. The following discussion examines these assumptions in light of the evidence collected during this study.  It is my hope that  this will lead to an informed consensus on the management of Vancouver's Northshore Watersheds.  162 Figure 6.2.1 Logging plan for the Greater Vancouver Watersheds adapted from Anon. (1991). The projected transformation of the landscape is illustrated by the changing proportions of the watersheds in three stand age categories. Young=l-40 yr, mature=41-120 yr, 01d=121 yr and older.  30000 2 25000 CO  ® 20000 S 15000 j | 10000 S 5000 <  0  1990  2030  2070 Time  2110  Young • Mature • Old  2180  163 First assumption: Western redcedar, yellow cedar, and Douglas-fir are early successional species. Accounts from the literature describe that western hemlock (Packee 1990) western redcedar (Minore 1990), Pacific silver fir (Crawford and Oliver 1990), and yellow cedar (Harris 1990) are capable of occupying all stages of forest development. Only Douglas-fir is a true pioneer being shade intolerant on most sites. Several types of successful growth strategies are possible.  Following a large disturbance where all trees are  killed, invading trees will show a very rapid growth as illustrated by Douglas-fir, western redcedar, and western hemlock individuals that established after the 1690 fire in the Capilano watershed.  This is a clear example of colonization by early  successional species. Another growth strategy is illustrated by trees growing on the same site but living in lower canopy strata. These include western redcedar, western hemlock, and Pacific silver fir that have established under a closed canopy growing very slowly; sudden bursts of growth represent release from suppression following a disturbance event. These individuals illustrate a late successional strategy.  Other trees germinate  shortly after a disturbance, and have intermediate growth patterns between pioneers and suppressed trees. Growth patterns from the main tree species in the watersheds indicate that all species (except Douglas-fir) are adapted to occupy different environments along a disturbance gradient. Classifying tree species according to successional status or shade tolerance may therefore be considered inappropriate.  164 Second Assumption: Douglas-fir and western redcedar are more resistant to disturbances. This assumption cannot be tested utilizing data from the watersheds except that we know several pest outbreaks have affected western hemlock and Pacific silver fir the most (Anon. 1991).  Crawford and Oliver (1990) and Packee (1990) confirm that  Pacific silver fir and western hemlock are highly susceptible to pests and diseases. On most sites they are also the least resistant to windthrow; however, Minore (1990) suggests that western redcedar on wet sites is highly susceptible to windthrow. A destructive crown fire, like the 1690 Capilano fire kills most trees in its path. Douglas-fir can survive lower intensity fires due to its thick bark.  Minore (1990) describes western redcedar  as the most fire sensitive tree species in coastal regions; however, Pacific silver and western hemlock appear highly sensitive to fire as well (Crawford and Oliver 1990; Packee 1990).  Therefore, the above assumption is probably true for  pests and diseases and windthrow on certain sites, but is false with respect to fire.  Third Assumption: Most sites in the Vancouver watersheds will eventually be dominated by western hemlock and Pacific silver fir. The Coastal Western Hemlock zone landscape of the watersheds is already numerically dominated by western hemlock and Pacific silver fir.  However, old-growth forests are characterized by  large western redcedar individuals which dominate the total basal area of many stands. Although, after comparing overstory tree  165 composition with the understory, one might be tempted to conclude that these ecosystems are evolving towards convergence of western hemlock and Pacific silver fir, I argue that there is convincing evidence to suggest that coexistence with western redcedar is the rule (See chapter 5). If western redcedar were to eventually be replaced by other species in the understory it should display a decreasing population at younger ages (Whipple and Dix 1979). Stem size distributions the Capilano watersheds reflect that the general population of western redcedar should maintain itself in the landscape, exhibiting mostly reverse J and multimodal distributions.  Even though some plots contain western redcedar  only in large size classes, this does not necessarily indicate a decline in the entire population. Fourth Assumption: Stable forests should contain more than 50% coverage of Douglas-fir, western redcedar, and Alaska yellowcedar. Forest stand stability depends on many factors independent of dominant tree species composition and which vary in time. This assumption appears to have been developed mostly based on the belief that the CWH landscape will eventually be completely dominated by western hemlock and Pacific silver fir; this was disputed earlier.  Although western hemlock is more susceptible  to insects and pathogens, the strategy of fragmenting the oldgrowth landscape with younger forests will probably increase the rate of pest dispersal and the chance of insect epidemics (Scholwalter and Means 1989).  Furthermore creating large  openings with clearcutting might make the forest edges less stable exposing them to higher risks of windthrow.  166 The ecological impacts resulting from the intense conversion of the old-growth landscape into younger forests by the year 2180 are largely unknown.  In an effort to manage fuel and protect  water quality it would be unfortunate to lose sight of the big picture.  Biological diversity and ecosystem function are  intimately related.  The manipulation of one will eventually have  an influence on the other.  I suggest that a more comprehensive  approach of ecosystem management should be adopted for Vancouver's watersheds.  167 6.3 British Columbia's green gold: New opportunities in silviculture. Silviculture is one of the world's oldest professions and many kinds of forest harvesting and site preparation techniques have been developed world wide.  In coastal British Columbia,  nearly 100% of harvested forests are clear-cut (Harding 1994). The main reasons for clearcutting in this region are short-term economic gains and tradition.  Public perception of clearcutting  is generally negative and, consequently, techniques of hiding harvested sites have been developed under the so called, "landscape improvement"  umbrella.  Although some of the negative  ecological impacts of clearcutting are well known (Bormann and Likens 1981; Keenan and Kimmins 1993), a recent national survey showed that most Canadian forestry specialists favor this technique (Foret Canada 1991).  In temperate rainforest  landscapes the most serious potential ecological impacts from clearcutting are soil erosion, loss of habitat, and forest fragmentation. The rapid rate of logging (200 to 250 thousand hectares per year (McKinnon 1994), and the changing attitude towards a more conservative form of land management (Grumbine 1994) are at the roots of a vigorous public debate in British Columbia.  In the  Pacific Northwest of the United States a similar situation has stimulated researchers to develop an alternative to clearcutting called new forestry (Franklin 1990).  The guiding principle  behind new forestry is to conserve structural elements characteristic of old-growth forests, such as a few large live and dead trees, in order to maintain biodiversity.  Green tree  168 retention and snag retention were successfully achieved in some Douglas fir forests east of the Willamette Valley (Puleo 1990). Although new forestry is still in its experimental stages, the attempt to integrate ecosystem management principles (Grumbine 1994) into forestry is promising. Nature provides examples of how temperate rainforests could be managed from a renewable resource perspective.  For example  tree species planted on innappropriate sites led to episodes of forest dieback in British Columbia (Gagnon 1985), and Germany. The B.C Forest Service now uses tree selection guidelines based on the ecological niche of tree species (Klinka and Feller 1984). There is a growing interest to use information on the natural disturbance regime in the management of forest landscapes (Hunter 1993; Attiwill 1994).  Should we also develop forest harvesting  guidelines based on the natural disturbance regime?  Although  forestry will never be able to mimic completely the complex and unpredictable nature of ecosystems, several parallels can be drawn between forest harvesting and the natural disturbance regime (Table 6.2.1).  Indeed, both harvesting and natural  disturbances can be described in terms of their size, severity, and spatio-temporal distributions. The results from the present study indicate that clearcutting should not be the predominant harvesting method for the Coastal Western Hemlock zone (Table 6.2.1).  Single tree  selection, and group selection creating canopy gaps similar to those described in the three cedar-hemlock forests, would most closely resemble the natural disturbance regime.  Shelterwood  169 Table 6.2.1  The natural disturbance regime and forestry in the Coastal Western Hemlock Zone. The information on size of harvesting systems is taken from (Klinka et al. 1990). Disturbance  Type  Size  Severity  Frequency  Clear-cutting  >lha  Very high  80 yr  Wide  Seed-tree  >lha  High  80 yr  Wide  Strip shelterwood  shelterwood height  <2X stand  Distribution  High  80 yr  Uniform shelterwood  >lha  Very high  80 yr  Rare  Group selection  •Ol-.lha  Low  7  Very rare  Single-tree selection  <.01ha  Low  7  Very rare  Avalanches  >lha  Low  7  High elevations  Earthquakes  >lha  Very high  7  7  Fire  >lha  Very high  600 yr  Warmer slopes  Natural  Gap-phase  .Ol-.lha  Low  400-1000 yr  Wide  Landslides  >lha  Very high  9  Steep slopes  Pathogens  >lha  Low-high  ->  7  Rockfalls  .Ol-.lha  Low  -?  Steep slopes  Windthrows  >lha  High  7  Exposed sites  170 logging could be used in patches of similar size to the partial disturbance observed and clearcutting could be used on south facing slopes.  Ideally, roads should be avoided, in order to  minimize soil erosion and forest fragmentation. Forest rotations of 80 years are usually advocated by foresters in coastal B.C.  This number is based on the idea that  beyond this point trees will have reached their maximum growth rates and are "ripe" for harvesting.  Natural rotations between  catastrophic disturbances estimated for the Eastcap Creek landscape ranged between 600 to more than 900 years. Natural rotations for gap phase driven ecosystems ranged between 400 to over 1000 years.  These rates are five to ten times higher than  those observed in boreal forests and suggest that these zones should be managed differently. The decision of integrating information from natural disturbances in forest management relies on the acquisition of knowledge of forest dynamics and on an environmental ethics consensus (Hunter 1993).  The variety of regeneration strategies  demonstrated by tree species growing naturally in coastal British Columbia indicate that it would be sensible to use more than one kind of forest harvesting method.  171 6.4 The value of biological information. Many different values are associated with forest ecosystems in coastal British Columbia.  The socioeconomic values are  undeniable since much of the wealth of this province derives directly or indirectly from logging.  The ecological values,  often ignored in the past, are becoming more widely understood as we observe the negative impacts of human use on nature.  The  spiritual values are important in the cultural and mythological lives of many of our First Nations people as well as the public at large.  It is interesting that although human society  recognizes the importance of knowledge the value of biological information contained in old-growth forests is overlooked. As time progresses, we are losing a precious opportunity to record biological information which could be lost forever.  Some  of this information has immediate application to our society; the discovery of the anticancer compound, taxol, in the bark of western yew provides an impressive example.  Other information,  such as the history of climate and past disturbances recorded in plant material, the distribution  and abundance of organisms, the  diversity of substances, forms and colors of organisms, and the relationships between organisms and their and environment represent a wealth impossible to calculate.  It is obvious from  the cheap price of wood that we do not appreciate what we are losing. A lot of material wasted during logging operations could be very useful for educational purposes.  For example, many subjects  including ecology, history, mathematics, and taxonomy could be taught using samples of tree cross-sections.  Plant specimens  172 should be collected and stored in herbaria and centers for biodiversity.  Many plants record information concerning their  environment during their lifespan.  Plant tissues accumulate  certain chemicals from the atmosphere, like carbon and heavy metals.  Tree-ring signatures reveal climatic anomalies and the  dynamic history of the forest.  Systematic collections of such  plant material would be invaluable to document these features. The collection of information on ecosystem attributes is crucial now if we, or our successors, wish to evaluate the extent of our impact on nature.  Forest ecologists from Sweden must of  necessity go to Siberia to study the natural dynamics of the boreal forest and how it influences patterns of biodiversity since such ecosystems have been destroyed in Sweden.  Furthermore  there is an intrinsic value to biological information which scholars should defend with the utmost vigor. I suggest that when a group of people decides to convert old-growth forests to younger forests, roads, electrical transmission lines, dwellings, etc. they have a responsibility to collect biological information before and during such transformations.  Universities, governments, industry, and  natural history groups must collaborate in this endeavor to increase our knowledge and to respect the value of biological information for our generation and the ones to follow.  173 7. CONCLUSIONS. In 1952 P.W. Richards wrote "The process of natural regeneration in tropical rainforests is no doubt exceedingly complex, and, though its practical importance to the forester is obvious, surprisingly little is known about it." A similar situation exists in temperate rainforests of British Columbia today.  I sincerely hope that this thesis may contribute to a  better understanding and management of this magnificent ecosystem. I examined the effects of natural and anthropogenic disturbances on the patterns of species composition, tree regeneration, and forest architecture in old-growth forests from southern coastal British Columbia.  Old-growth cedar-hemlock  forests share many similarities with old-growth Douglas-fir forests of western Oregon and Washington (see Franklin et al. 1981).  Large live and dead standing trees, a reverse J size  distribution, a multilayered canopy, and complex architecture are some of the common traits. Although the size class distribution of old cedar-hemlock forests resembles that of old-growth Douglas-fir, cohort detection analysis indicated otherwise.  Few  of the virgin temperate rainforests examined during this study showed sign of catastrophic fire. This suggests that these oldgrowth cedar-hemlock stands fit the "true" old-growth category from Oliver and Larson (1990), and small-scale disturbances plays a major role in their makeup.  In the case where catastrophic  fire does occur, the low frequency of these events enables gapphase dynamics to be an important process in shaping patterns of forest structure and composition.  As a result of slow gap-  174 filling rates, natural forest rotations are quite high (400-1000 yr). The strongest correlation between plant species composition and forest structure occurred when stands in stem exclusion stage, and understory reinitiation stage, were compared with oldgrowth forests.  The correlation was rather weak when only old-  growth forests were used.  This probably reflects the fact that  many species of the study area are ubiquitous and given the right conditions will develop.  Thus, as a forest ages the relationship  between structure and composition may become more noisy as chance will play a more important role.  On the other hand, certain  groups of plants which have narrower ecological niches will have a stronger relationship with structural elements in old-growth. Small intermittent streams exerted a strong influence on the spatial pattern of understory vegetation.  This structural  element, not usually considered important during logging operations, may prove to be just as important as big trees in the maintenance of biodiversity. Present forestry practices are dramatically changing the ecology of temperate rainforests both at the stand and landcsape levels.  If the often quoted 80 yr rotation is applied  continually in coastal British Columbia, managed forests will turnover 3 to 12 times more rapidly than unmanaged forests.  In  some cases forest management has been operating under the false assumption that old-growth forests are decadent and must be renewed.  I hope the present study into the dynamic ecology of  these ecosystems can convince forest planners otherwise.  175 The diversity of the natural disturbance regime in coastal British Columbia and the ability of most tree species to adapt to a wide range of perturbations gives us very interesting sivicultural opportunities.  It would be sensible to place less  emphasis on clear-cutting as a "regeneration" method as its impact on nature, especially in rainforest environments, can be extensive.  176 Bibliography Agee, J.K. 1993. Fire ecology of Pacific Northwests Forests. Island Press, Washington D.C. 493 pages. Alaback, P.B. 1982. Dynamics of understory biomass in Sitka spruce-western hemlock forests of southeast Alaska. Ecology 63:1932-1948. Alaback, P.B. 1991. Comparative ecology of temperate rainforests of the Americas along analogous climatic gradients. Revista Chilena de Historia Natural 64:399-412. Anon. 1991. Greater Vancouver watershed management evaluation and policy review. Unpublished report prepared by Economic and Engineering Services Inc. Armesto, J.J., and E.R. Fuentes. 1988. Tree species regeneration in a mid-elevation, temperate rain forest in Isla de Chiloe, Chile. Vegetatio 74:151-159. Armstrong, J.E. 1990. Vancouver geology. Edited by Roots, C. and C. Staargaard. Cordilleran Section, Geological Association of Canada, Vancouver. 128 pages. Arsenault, A. and G.E. Bradfield. 1995. Structural-compositional variation in three age classes of temperate rainforests of southern-coastal British Columbia. Canadian Journal of Botany 73: In Press. Attiwill, P.M. 1994. The disturbance of forest ecosystems: the ecological basis for conservative management. Forest Ecology and Management 63:247-300. Aubreville, A. 1938. La foret coloniale: Les forets de l'Afrique Occidentale Francaise. Annales Academie Sciences coloniale, Paris, 9:1-245. Banner, A., J. Pojar, and G.E. Rouse. 1983. Postglacial paleoecology and successional relationships of a bog woodland near Prince Rupert, British Columbia. Canadian Journal of Forest Research 13:938-947. Barbour, M.G., J.H. Burk, and W.D. Pitts. 1980. Terrestrial plant ecology. Benjamin/Cummings, Menlo Park, California. 604 pages. Beliefleur, P. 1981. Markov models of forest-type secondary succession in coastal British Columbia. Canadian Journal of Forest Research 11:18-29. Bonnicksen, T.M. and E.C.Stone. 1981. The giant Sequoia-mixed conifer forest community characterized through pattern analysis as a mosaic of aggregations. Forest Ecology and Management 3:307-328. Bonnicksen, T.M., and E.C. Stone. 1982. Reconstruction of a presettlement giant Sequoia-mixed conifer forest community using the aggregation approach. Ecology, 63(4):1134-1148.  177 Bormann, F.H., and G.E. Likens. 1981. Pattern and process in a forested ecosystem. Springer Verlag, N.Y. 253 pages. Botkin, D. B. 1979. A grandfather clock down the staircase: Stability and disturbance in natural ecosystems, Pages 1-10 in Forests: Fresh perspectives from ecosystem analysis, edited by R.H. Waring. Proceedings of the 40th Annual Biology Colloquium, Oregon State University Press, Corvallis. Bradfield, G.E., and A. Scagel. 1984. Correlations among vegetation strata and environmental variables in subalpine spruce-fir forests, southeastern British Columbia. Vegetatio 55:105-114. Brokaw, N.V.L. 1982. The definition of treefall gaps and its effect on measures of forest dynamics. Biotropica 14:158-160. Brokaw, N.V.L. 1985. Treefalls, regrowth, and community structure in tropical forests. Pages 53-69 in: The ecology of disturbance and patch dynamics, edited by S.T.A. Pickett and P.S. White. Academic Press, N.Y. Campbell, E.C. and R.W. Campbell. 1986. Status report on the spotted owl in Canada, Strix occidentalis caurina: occurrence, British Columbia. Fish and Wildlife Branch, Victoria, British Columbia. Clements, F.E. 1916. Plant succession: An analysis of the development of vegetation. Carnegie Institute, Washington, D.C. 512 pages. Cline, S.T., A.B. Berg, and H.W. Wight. 1980. Snag characteristics and dynamics in Douglas-fir forests, western Oregon. Journal Wildlife Management 44:773-786. Connell, J.H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302-1310. Connell, J.H. and R.D. Slatyer 1977. Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111:1119-1144. Cooper, W.S. 1913. The climax forest of Isle Royale, Lake superior and its development. Botanical Gazette 55:1-44, 115140, 189-234. Cooper, W.S. 1926. The fundamentals of vegetation change. Ecology 7: 391-413. Cooper, W.S. 1939. A fourth expedition to Glacier Bay, Alaska. Ecology 20:130-155. Cowles, H.C. 1901. The physiographic ecology of Chicago and vicinity; a study of the origin, development, and classification of plant societies. Botanical Gazette 31(3):170-177.  178 Crawford, P.D. and C D . Oliver 1990. Pacific silver fir Pages 1725 in: Silvics of North America, Volume 1. Conifers. Agriculture Handbook 654. Edited by R.M. Burns and B.H. Honkala. USDA, Forest Service, Washington, D.C Daubenmire, R., and J.B. Daubenmire 1968. Forest vegetation of eastern Washington and northern Idaho. Washington Agricultural Experiment Station Bulletin 60. 104 pages. Deal, R.L., C D . Oliver, and B.T. Bormann. 1991. Reconstruction of mixed hemlock-spruce stands in coastal southeast Alaska. Canadian Journal of Forest Research 21:643-654. Drury, W.H., and I.C.T. Nisbet. 1973. Succession. Journal of the Arnold Arboretum. 54:331-368. Duncan, R.P., and G.H. Stewart. 1991. The temporal and spatial analysis of tree age distributions. Canadian Journal Forest Research 21:1703-1710. Egler, F.E. 1954. Vegetation science concepts. I. Initial floristic composition: A factor in old-field vegetation development. Vegetatio 4:412-417. Eis, S. 1962. Statistical analysis of several methods for estimation of forest habitats and tree growth near Vancouver, B.C. Forestry Bulletin 4. Faculty of Forestry, University of British Columbia, Vancouver. 75 pages. Eis, S. 1970. Natural root grafts in conifers and the effect of grafting on tree growth. Pages 25-29 in: Tree ring analysis with special reference to northwest America: Proceedings of a conference on biology of tree-ring formation, methods of measurement of tree rings, methods of analysis, and uses of tree ring data. Edited by J.H.G. Smith and J. Worral. Faculty of Forestry, UBC, Vancouver, British Columbia. Eldrige, M., A. Eldridge and A.H. Stryd. 1984. Meares Island aboriginal tree study. Unpublished report prepared by Areas Associates. Enright, N.J. 1982. Recognition of successional pathways in forest communities using size-class ordination. Vegetatio 48:133-140. Fogel, R., M. Ogawa, and J.M. Trappe. 1973. Terrestrial decomposition: a synopsis. Coniferous Forest Biome Internal Report 135, University of Washington. Seatle Washington. 12 pages. Forcier, L.K. 1975. Reproductive strategies and the co-occurrence of climax tree species. Science 189:808-810 Foret Canada. 1991. L'Etat des forets au Canada. Rapport au Parlement 1990. Ministre des Approvisionnements et Services Canada 1991, Ottawa. 80 pages.  179 Franklin, J.F., and M.A. Hemstrom. 1981. Aspects of succession in the coniferous forests of the Pacific Northwest. Pages 212-229 in: Forest succession: concepts and applications, edited by H.H. Shugart and D.B. Botkin. Springer-Verlag, N.Y. Franklin, J.F., K. Cromack, W. Denison, A. Mckee, C. Maser, J. Sedell, F. Swanson, and G. Juday. 1981. Ecological characteristics of old-growth Douglas-fir forests. USDA. Forest Service Technical Report PNW-118 Pac. Northwest Forest and Range Exp. Stn., Portland, Oregon. 48 pages. Franklin, J.F. 1988. Pacific Northwest forests. Pages 104-130 in: North American terrestrial vegetation. Edited by M.G. Barbour and W.D. Billings. Cambridge University Press, N.Y. Franklin, J.F. and D.S. DeBell. 1988. Thirty-six years of tree population change in an old-growth Pseudotsuga-Tsuga forest. Canadian Journal of Forest Resarch 18:633-639. Franklin, J.F. 1990. Old-growth forests and the new forestry. Pages 1-19 in: Forests, Wild and managed: differences and consequences. Edited by A.F. Pearson, and D.A. Challenger, Student For Forestry Awareness, University of British Columbia, Vancouver. Gagnon, D. 1985. Forest vegetation of west-central Vancouver Island, British Columbia. Ph.D. Thesis, Dept. of Botany, The University of British Columbia, Vancouver, B.C. Gagnon, D. and G.E. Bradfield. 1986. Relationships among forest strata and environment in southern coastal British Columbia. Canadian Journal of Forest Research 16: 1264-1271. Gagnon, D. and G.E. Bradfield. 1987. Gradient analysis of west central Vancouver Island forests. Canadian Journal of Botany 65: 822-833. Gleason, H.A. 1926. The individualistic concept of the plant association. Torrey Botanical Club Bulletin 53:7-26. Goff, F.G. and P.H. Zedler 1972. Derivation of species succession vectors. Am. Midi. Nat. 87:397-411. Goodman, D. 1979. Applications of eigen vector analysis in the resolution of spectral pattern in spatial and temporal ecological sequences. Pages 139-155 in: Contemporary quantitative ecology and related ecometrics. Edited by G.P. Patil and M.L. Rosenzweig. International Co-operative Publishing House, Finland, M.D. Grubb, P.J. 1977. The maintenance of species richness in plant communities: the importance of the regeneration niche. Biological Review 52: 107-145. Grumbine, R.E. 1994. What is ecosystem management? Conservation Biology 8(1):27-38.  180 Harding, L.E. 1994. Threats to diversity of forest ecosystems in British Columbia. Pages 245-277 in: Biodiversity in British Columbia: Our Changing Environment. Edited by L.E. Harding and E. McCullum. Pacific Yukon Region Canadian Wildlife Service, Environment Canada. Harmon, M.E., and J.F. Franklin. 1989. Tree seedlings on logs in Picea-Tsuga forests of Oregon and Washington. Ecology 70(1):48-59. Harper, J.L. 1977. Population biology of plants. Academic Press, N.Y. 892 pages. Harris, A.S. 1990. Alaska-cedar. Pages 97-102 In: Silvics of North America, op. cit. Heinselman, M.L. 1973. Fire in the virgin forests of the Boundary Waters Canoe Area, Minnesota. Quaternary Research 3:329-382. Hemstrom, M.A. and J.F. Franklin 1982. Fire and other disturbances of the forest in Mount Rainier National Park. Quaternary Research 18:32-51. Henderson, J.A., D.H. Peter, R.D. Lesher and D.C. Shaw. 1989. Forested plant associations of the Olympic national forest. USDA Forest Service, PNR. R5 Ecol Technical Paper 001-88. 502 pages. Henry, J.D., and J.M.A. Swan. 1974. Reconstructing forest history from live and dead plant material- an approach to the study of forest succession in southwest New Hampshire. Ecology 55:772-783. Holland, S.S. 1976. Landforms of British Columbia: a physiographic outline. Bulletin 48. Dept. of Mines and Petroleum Resources, Victoria, B.C. 138 pages. Hunter, M.L. 1990. Forests, wildlife and forestry. Prentice Hall, N.J. 370 pages. Hunter, M.L. 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65:115-120. Hunter, J.C., and V.T. Parker. 1993. The disturbance regime of an old-growth forest in coastal California. Journal of Vegetation Science 4:19-24. Johnson, E.A. and G.I. Fryer. 1989. Population dynamics in lodgepole pine-engelmann spruce forests. Ecology 70(5): 1335-1345. Jones, E.W. 1945. The structure and reproduction of the virgin forest of the North Temperate zone. New Phytologist 44:130-148.  181 Jovet, S. and P. Jovet 1944. Peuplement bryologique des bois pourissants et rochers ombrages des environs de Samoens (HauteSavoie). Revue Bryologie Lichenologie 14:120-148. Keenan, R. 1993. Structure and function of western redcedar and western hemlock forests on northern Vancouver Island. Ph.D. Thesis, Faculty of Forestry, The University of British Columbia, Vancouver, B.C. Keenan, R. and J.P. Kimmins. 1993. The ecological impacts of clear-cutting. Environmental Review 1:121-144. Kimmins, J.P. 1992. Balancing Act. University of British Columbia Press, Vancouver. 244 pages. Kimmins, J.P. 1987. Forest ecology. Macmillan, N.Y. 531 pages. Klein, R.M. and T.D. Perkins. 1987. Cascades of causes and effects of forest decline. Ambio 16: 86-93. Klinka, K., R.N. Green, R.L. Trowbridge, and L.E. Lowe. 1981. Taxonomic classification of humus forms in ecosystems of British Columbia. Land Management Report 8. Ministry of Forests, Victoria British Columbia. 54 pages. Klinka, K., J. Pojar, and D.V. Meidinger 1991. Revision of biogeoclimatic units of coastal British Columbia. Northwest Science 65(1):32-47. Klinka, K., V.J. Krajina, A. Ceska, and A.M. Scagel. 1989. Indicator plants of coastal British Columbia. UBC Press, Vancouver, B.C. 288 pages. Klinka, K. and M.C. Feller. 1984. Principles used in selecting tree species for regeneration of forest sites in southwestern British Columbia. Forestry Chronicle 77-85. Klinka, K., A.M. Scagel, and P.J. Courtin. 1985. Vegetation relationships among some serai ecosystems in southwestern British Columbia. Canadian Journal of Forest Research 15:561569. Klinka, K., R.E. Carter, and M.C. Feller. 1990. Cutting oldgrowth forests in British Columbia: Ecological considerations for forest regeneration. Northwest Environmental Journal 6:221242. Krajina, V. J. 1965. Biogeoclimatic zones and classification of British Columbia. Ecology Western North America 1:1-17 Krajina, V.J. 1969. Ecology of forest trees in British Columbia, Ecology of Western North America 2:1-147. Kuiper, L.C. 1988. The structure of natural Douglas-fir forests in western Washington and western Oregon. Agricultural Wageningen Papers 88(5):1-47.  182 Legendre, P. and M. Fortin 1989. Spatial pattern and ecological analysis. Vegetatio 80:107-138. Lertzman, K. P. 1989. Gap-phase community dynamics in a subalpine old-growth forest. Ph.D. Thesis, Dept. Zool. University of British Columbia. 168 pages. Lertzman, K.P. 1992. Patterns of gap-phase replacement in a subalpine, old-growth forest. Ecology 73:657-669. Lertzman, K.P. and C.J. Krebs 1991. Gap phase structure of a subalpine old-growth forest. Canadian Journal of Forest Research 21:1730-1741. Lertzman, K.P. 1995. Forest dynamics, differential mortality and variable recruitment probabilities. Journal of Vegetation Science 5: In Press. Lessica, P., B. McCune, S.V. Cooper and W.S. Hong. 1991. Differences in lichen and bryophyte communties between oldgrowth and managed second-growth forests in the Swan Valley, Mountains. Canadian Journal of Botany 69:1745-1755 Levin, S.A., T.M. Powell, and J.H. Steele. 1993. Patch dynamics. Lecture Notes in Biomathematics 96. Springer-Verlag, N.Y. 307 pages. Luken, J. 0. 1990. Directing ecological succession. Chapman and Hall, N.Y. 251 pages. Lusk, C. and J. Ogden. 1992. Age structure and dynamics of a podocarp-broadleaf forest in Tongariro National park, New Zealand. Journal of Ecology 80:379-393. Matson, P.A. and R.H. Waring. 1984. Effects of nutrients and light limitation on mountain hemlock: susceptibility to laminated root rot. Ecology 65(5):1517-1524. McCune, B. and T.F.H. Allen. 1985. Forest dynamics in the Bitterroot Canyons, Montana. Canadian Journal of Botany 63:377-383. Mcintosh, R.P. 1985. The background of ecology. Cambridge University Press, N.Y. 3 83 pages. McKinnon, A. 1994. British Columbia's old-growth forests. Bioline 11:1-9. McLennan, D. 1991. Assessment of proposed 1992 cutting areas for Greater Vancouver Water District. Unpublished report prepared by Oikos Consultants. 15 pages. Minore, D. 1990. Western redcedar. Pages 590-600 In: Silvics of North America. Volume 1. Conifers, op. cit. Moran, P.A.P. 1950. Notes on continuous stochastic phenomena. Biometrika 37:17-23.  183 Mueller-Dombois, D. 1987. Natural dieback in forests. Bioscience 37:575-583. Muhle, H. and F. LeBlanc 1975. Bryophyte and lichen succession on decaying logs. Jour. Hattori Bot. Lab. 39:1-33. Odum, E. 1969. The strategy of ecosystem development. Science 164:262-270. Oldeman, R.A.A. 1983 Tropical rain forest, architecture, silvigenesis and diversity. Pages 139-150 In: Tropical rain forest: Ecology and management. Edited by S.L. Sutton, T.C. Whitmore, and A.C. Chadwick. Blackwell Scientific Publications, Oxford. Oldeman, R.A.A. 1989. Elements of silvology. Springer-Verlag, N.Y. Oliver, C D . , and E.P. Stephens 1977. Reconstruction of a mixedspecies forest in central New England. Ecology, 58:562-572. Oliver, C D . 1981. Forest development in North America following major disturbances. Forest Ecology and Management 3:153-168. Oliver, C D . , and B.C. Larson. 1990. Forest stand dynamics. McGraw Hill, N.Y. 467 pages. Ovington, J.D. 1983. Temperate broad-leaved evergreen forests. Elsevier, N.Y. 241 pages. Packee, E. 1990. Western hemlock. Pages 613-622 In: Silvics of North America. Volume 1. Conifers, op. cit. Parker, A.J. and R.K. Peet. 1984. Size and age structure of conifer forests. Ecology 65(5): 1685-1689. Parker, T. and F.D. Johnson. 1987. Distribution and characterisics of western redcedar groves in the mountains of Idaho, Washington, and Montana. Pages 136-144 In: Western red cedar -- does it have a future? edited by N.J. Smith. Faculty of Forestry, Universty of British Columbia, Vancouver, British Columbia. Payette, S., L. Filion, and A. Delwaide. 1990. Disturbance regime of a cold temperate forest as deduced from tree-ring patterns: the Tantare Ecological Reserve, Quebec. Canadian Journal of Forest Research 20:1228-1241. Pickett, S.T.A., and P.S. White (eds.). 1985. The ecology of natural disturbance and patch dynamics. Academic, N.Y. 472 pages. Pickett, S.T.A. 1989. Space-for-time substitution as an alternative to long-term studies. Pages 110-135 in: Long-term ecological studies in ecology:approaches and alternatives, edited by G.E. Likens.Springer-Verlag, N.Y. 214 pages.  184 Puleo, V. 1990. "Just do it". Pages 135-161 in: Forests, Wild and managed: differences and consequences. Edited by A.F. Pearson, and D.A. Challenger, Student For Forestry Awareness, University of British Columbia, Vancouver. Rebertus, A.J. and T.T. Veblen. 1993. Structure and tree-fall gap dynamics of old-growth Nothofaaus forests in Tierra del Fuego, Argentina. Journal of Vegetation Science 4:641-654. Remmert, H. 1991. The mosaic-cycle concept of ecosystems. Springer-Verlag, N.Y. 168 pages. Richards, P.W. 1952. The tropical rainforest. Cambridge University Press, Cambridge. 450 pages. Richards, P.W. 1983. The three dimensional structure of tropical rainforest. Pages 3-10 in: Tropical rainforest: ecology and management. op. cit. Rowe, J.S. 1972. Forest regions of Canada. Department of the Environment, Canadian Forestry Service, Publication No. 1300. Ottawa, Ontario. 172 pages. Runkle, J.R. 1982. Patterns of disturbance in some old-growth mesic forests of eastern North America. Ecology 63:1533-1546. Runkle, J.R. 1992. Guidelines and sample protocol for sampling forest gaps. General Technical Report PNW-GTR-2 83. Portland, OR:US Department of Agriculture, Forest Service, Pacific Northwest Research Station. 44 pages Schofield, W.B. 1972. Bryology in arctic and boreal North America and Greenland. Canadian Journal, of Botany 50:1111-1133. Schofield, W.B. 1985. Introduction to bryology. MacMillan, N.Y. 431 pages. Schofield, W.B. 1988. Bryogeography and the bryophytic characterization of biogeoclimatic zones of British Columbia, Canada. Canadian Journal of Botany 66:2673-2686. Schmidt, R.L. 197 0. A history of pre-settlement fires on Vancouver Island as determined from Douglas-fir ages. Pages 107-108 in: Tree ring analysis with special reference to northwest America: Proceedings of a conference on Biology of tree-ring formation, methods of measurement of tree rings, methods of analysis, and uses of tree ring data. Edited by J.H.G. Smith and J. Worral. Faculty of Forestry, UBC, Vancouver, British Columbia. Shuster, R. 1949. The ecology and distribution of Hepaticae in central and western New York. American Midland Naturalist. 42: 513-712.  185 Sholwalter, T.D., and J.E. Means 19 89. Pests link site productivity to the landscape. Pages 248-250 in: Maintaining the long-term productivity of Pacific Northwest forest ecosystems. Edited by D.A. Perry, R. Meurisse, B. Thomas, R. Miller, J. Boyle, J. Means, C.R. Perry, and R.F. Powers. Timber Press, Portland, Oregon. 256 pages. Slack, N. 1977. Species diversity and community structure in Bryophytes: New York States communities. New York State Museum Bulletin 428, Albany, N.Y. 70 pages. Smith, R.L. 1980. Ecology and field biology. Harper and Row, N.Y. 83 5 pages. Smith, D.M. 1982. The practice of silviculture. John Wiley and Sons, N.Y. 527 pages. Soderstrom, L. 1988. Sequence of bryophytes and lichens in relation to substrate variables of decaying forest wood in northern Sweden. Nordic Journal of Botany 8(1): 89-97. Spies, T.A., and J.F. Franklin. 1988. Old-growth and forest dynamics in the Douglas-fir region of western Oregon and Washington. Natural Areas Journal 8(3): 190-201. Spies, T.A., J.F. Franklin, and M. Klopsch. 1990. Canopy gaps in Douglas-fir forests of the Cascade Mountains. Canadian Journal of Forest Research 20:649-658. Spies, T.A. 1991. Plant species diversity and occurrence in young, mature, and old-growth Douglas-fir stands in western Oregon and Washington. Pages 91-109. in: Wildlife and vegetation of unmanaged Douglas-fir forests. Edited by Ruggiero, L.F., K.B. Aubry, A.B. Carey, and M.H. Huff. USDA, Forest Service, PNW-GTR-285, Portland, Oregon. Sprugel, D.G. 1976. Dynamic structure of wave-generated Abies balsamea forests in the northeastern United States. Journal of Ecology 64:889-912. Stewart, G.H. 1986a. Forest development in canopy openings in old-growth Pseudotsuqa forests of the western Cascade Range, Oregon. Canadian Journal of forest Research 16:558-568. Stewart, G.H. 1986b. Population dynamics of a montane conifer forest, western Cascade range, Oregon, USA. Ecology 67(2): 534-544. Stewart, G.H. 1988. The influence of canopy cover on understory development in forests of the western Cascade Range, Oregon, USA. Vegetatio 76:79-88. Stewart, G.H. 1989. The dynamics of old-growth Pseudotsuga forests in the western Cascade range, Oregon, USA. Vegetatio 82:79-94.  186 Tansley, A.G. 1920. The classification of vegetation and the concept of development. Journal of Ecology 8:118-149. Taylor, A.H. 1990. Disturbance and persistence of Sitka spruce {Picea 'sitchensis (Bong) Carr. ) in coastal forests of the Pacific Northwest, North America. Journal of Biogeography 17:47-58. Taylor, A.H. and C.B. Halpern. 1991. The structure and dynamics of Abies maanifica forests in the southern Cascade Range, USA. Journal of Vegetation Science 2:189-200. Trewartha, G.T. 1954. An introduction to weather and climate. McGraw-Hill, N.Y. 402 pages. Trowbridge, R., B. Hawkes, A. Macadam, and J. Parminter. 1986. Field handbook for prescribed fire assessments in British Columbia: Logging slash fuels. Land Management Handbook No. 11. B.C. Min. For. Victoria. Turner, N. 1991. Burning mountain sides for better crops: Aboriginal landscape burning in British Columbia. Archeology in Montana 32 : 57-74. Tyrrell, L.E., and T.R. Crow. 1994. Structural characteristics of old-growth hemlock-hardwood forests in relation to age. Ecology 75 (2): 370-386. Vallentyne, K.W.G., and L.M. Lavkulich. 1978. The soil orders of British Columbia. Pages 67-95 in: The soil landscapes of British. Columbia. Edited by K.W.G. Vallentyne, P.N. Sprout, T.E. Baker, and L.M. Lavkulich. Ministry of Environment, Victoria, British Columbia. 197 pages. Veblen, T.T. 1985. Forest development in tree-fall gaps in the temperate rain forests of Chile. National Geographic Research 1:162-183. Walter, H. 1973. Vegetation of the earth: in relation to climate and the. ecophysiological conditions. Translated from the second German edition by Joy Wieser. Springer-Verlag, N.Y. 23 6 pages. Waring, R.H. and J.F. Franklin. 1979. Evergreen coniferous forests of the Pacific Northwest. Science 204:1380-1386. Watt, A.S. 1925. On the ecology of British beechwoods with special reference to their regeneration. Part II. The development and structure of beech communities on the Sussex Downs. Journal of Ecology 13:27-73. Watt, A.S. 1947. Pattern and process in the plant community. Journal of Ecology 35:1-22. Whipple, A.S. and R.L. Dix. 1979. Age structure and successional dynamics of Colorado subalpine forests. The American Midland Naturalist 101(1) : 142-158.  187 Whittaker, R.H. 1953. A consideration of climax theory: the climax as a population and pattern. Ecological Monographs 23:41-78. Whittaker, R.H. 1975. Communities and ecosystems. MacMillan, N.Y. 385 pages. White, P.S. 1979. Pattern, process, and natural disturbance in vegetation. Botanical Review 45 (3): 229-299. Wilcove, D.S. 1988. Owls and old-growth. Trends in ecology and evolution. 3:113-114 Williamson, G.B. 1975. Pattern and serai composition in an oldgrowth beech-maple forest. Ecology 56:727-731.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items