ANALYTICAL STUDY OF PLANT/ENVIRONMENT INTERACTIONS IN THIMBLEBERRY AND DEVIL'S CLUB By ROSEMARY MASON B.Sc. (Hons.), University of Victoria, 1965 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR A DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Botany Department We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 1990 ® Rosemary Mason, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of /2o~rZ( / ^ y The University of British Columbia Vancouver, Canada Date Se f>k W ? € r Si I, / ? ? Q DE-6 (2/88) ii Abstract The morphology, phenology and stem demography of devil's club and thimbleberry were examined to elucidate their niche utilization strategies. The study was conducted in the Kitimat River valley in west central B.C. during the 1986 and 1987 growing seasons. Thimbleberry was sampled in a grded alder site and a nondrded alder site, whereas devil's club was sampled in an old growth forest. The variation in the plant characters, as summarized by principal components axes, was apportioned within and among clones, between sites, years, and species. Except for the thimbleberry vegetative phenology, within-sites differences accounted for most variation and variation between-sites often exceeded that between years. Moreover, between-spedes differences accounted for less variation than within-species differences for morphology and phenology. The variation in plant characters was also examined in relation to canopy cover, soils and adjacent vegetation using multivariate methods. The rate of vegetative development for devil's club in 1986 increased as canopy cover decreased; other environmental measures were uncorrelated with devil's club. Both vegetative and reproductive rates of development increased with disturbance due to girdling and increasing moisture for the combined grded and ungrded thimbleberry data set. Similarly, morphdodcal size was greater fa the combined thimbleberry data set with increasing moisture and dsturbance. Environmental correlations were reflected differently within-sites, however, with rates of development, plant size and the number of flowers decreasing with increasing moisture at the nongirded thimbleberry she. The relationship between plant characters was also assessed. Phenology and morphology were correlated for both devil's dub and thimbleberry; stem development began earlier and was more rapid with increasing stem size. Demography and phenology were unrelated. Both spedes displayed different niche utilization strategies; thimbleberry being more flexible than devil's club. In contrast to devil's dub, thimbleberry is morphologically and iii phenoioglcaily responsive to disturbance ana is mizomatous ratner man stoioniTerous. stems and lateral branches also had several phonological and developmental possibilities. This flexibility imparted an advantage to thimbleberry in the fluctuating conditions of its earlier successional niche. The differing correlation structure between and within thimbleberry sites suggests that several scales of observation are necessary to clarify plant-environment relationships. Moreover, as envronmental characters interact differently with plants from site to site, management must be site specific. Alder girdling may be a judicious management technique at drier sites, where thimbleberry is not as prolific under an open canopy. iv Table of Contents Abstract ii List of Tables vii List of Figures x Acknowledgement xii Chapter I INTRODUCTION 1 1.1 OBJECTIVES 2 1.2 PHENOLOGY, MORPHOLOGY AND DEMOGRAPHY 2 1.3 VARIATION IN PHENOLOGY, MORPHOLOGY AND DEMOGRAPHY .. 4 1.4 FACTORS CORRELATED WITH VARIATION 5 1.5 THIMBLEBERRY 11 1.6 DEVIL'S CLUB 12 Chapter II STUDY AREAS 14 2.1 DESCRIPTION OF STUDY AREA 14 2.1.1 GEOGRAPHIC LOCATION 14 2.1.2 EDAPHIC ENVIRONMENT 14 2.1.3 DISTURBANCE HISTORY 16 2.2 CLIMATE 17 Chapter III SAMPLING METHOD & DATA COLLECTION 19 3.1 DESIGN 19 3.2 CANOPY COVER 21 3 3 SOILS 21 3 ^ PLANT NEIGHBOURS '"!]]!!!!!!!!!!!*.".'.'.!'.'.!!!!!!!!!.'!!!!!!!!!!! 22 3.5 MORPHOLOGY 23 3.6 PHENOLOGY 25 3.7 DEMOGRAPHY 26 Chapter IV ANALYTICAL TECHNIQUES 27 4.1 TECHNIQUES 27 4.1.1 PRINCIPAL COMONENT ANALYSIS (PCA) 27 4.1.2 ANOVA 28 4.1.3 CANONICAL CORRELATION (CANCORR) 29 4.1.4 WEIGHTED MULDTIMENSIONAL SCALING (WMDS) 30 Chapter V DATA ANALYSIS 33 5.1 GRAPHICAL ANALYSES 33 5.2 ENVIRONMENT 33 V 5.3 MORPHOLOGY 35 5.4 PHENOLOGY 37 5.5 DEMOGRAPHY 38 5.6 CANCORR AND WMDS 41 5.7 COMPUTATION 42 Chapter VI RESULTS 43 6.1 ENVIRONMENTAL RESULTS 43 6.1.1 CANOPY COVER 43 6.1.2 SOILS VARIABLES 44 6.1.3 PLANT NEIGHBOURS-PRESENCE/ABSENCE 51 6.1.4 PLANT NEIGHBOURS-FREQUENCY 60 6.2 MORPHOLOGY 68 6.2.1 UNIVARIATE ANALYSES AND GRAPHS 68 6.2.2 PRINCIPAL COMPONENT ANALYSES 71 6.2.3 PARTITIONINGOFVARIATION 77 6.2.4 MORPHOLOGICAL-ENVIRONMENTAL RELATIONSHIPS 79 6.3 PHENOLOGY 81 6.3.1 UNIVARIATE ANALYSES AND GRAPHS 81 6.3.2 PRINCIPAL COMPONENT ANALYSES 89 6.3.3 PARTITIONINGOFVARIATION 92 6.3.4 PHENOLOGICAL-ENVIRONMENTAL RELATIONSHIPS 95 6.4 DEMOGRAPHY 99 6.4.1 UNIVARIATE ANALYSES AND GRAPHS 99 6.4.2 PRINCIPAL COMPONENT ANALYSES 106 6.4.3 PARTITIONINGOFVARIATION 111 6.4.4 DEMOGRAPHIC-ENVIRONMENTAL RELATIONSHIPS 112 6.5 MORPHOLOGY-PHENOLOGY-DEMOGRAPHYINTERACTIONS 115 Chapter VII DISCUSSION 119 7.1 MORPHOLOGY 119 7.2 PHENOLOGY 124 7.3 DEMOGRAPHY 127 7.4 NICHE UTILIZATION AND LIFEHISTORY STRATEGY 130 7.5 MANAGEMENT 134 7.6 FURTHER RESEARCH 135 LITERATURE CITED 136 APPENDIX 1 - MANN-WHITNEY U TESTS FOR THE DIFFERENCES IN MORPHOLOGY BETWEEN THIMBLEBERRY SITES IN 1986 AND 1987, BETWEEN YEARS (1986 AND 1987), AND WITHIN PLANTS BETWEEN STEMS AND BRANCHES 151 vi APPENDIX 2- MANN-WHITNEY U TESTS FOR DIFFERENCES IN GENERATIVE AND VEGETATIVE PHENOLOGIES BETWEEN THIMBLEBERRY SITES IN 1986 AND 1987, BETWEEN YEARS (1986 AND 1987) AND WITHIN PLANTS BETWEEN STEMS AND BRANCHES APPENDIX 3 - CANCORR RELATIONSHIPS BETWEEN MORPHOLOGICAL PHENOLOGICAL AND DEMOGRAPHIC CHARACTERS FOR TB, KIT, WED AND DC AND SOILS, FREQUENCY AND PRESENCE/ABSENCE OF PLANT NEIGHBOUR MATRICES APPENDIX 4 - WMDS RELATIONSHIPS BETWEEN MORPHOLOGICAL. PHENOLOGICAL AND DEMOGRAPHIC CHARACTERS FOR TB, KIT, WED AND DC AND SOILS, FREQUENCY AND PRESENCE/ABSENCE OF PLANT NEIGHBOUR MATRICES v i i List of Tables 1. Phonological, morpholoojcal and demographic sampling in 1966 and 1987 24 2. Phenological codes 25 3. Analytical methods for environmental data 34 4. Analytical methods for morphological data 36 5. Analytical methods for phenological data 39 6. Analytical methods for demographic data 40 7. TB soils - PCA axes relationships 45 8. KIT soils - PCA axes rdationsh 47 9. WED soils - PCA axes relationships 48 10. DC soils - PCA axes relationships 49 11. TB presence/absence of plant neighbours - PCA axes relationships 52 12. KIT presence/absence of plant neighbours - PCA axes relationships 55 13. WED presence/absence of plant neighbours - PCA axes relationships 56 14. DC presence/absence of plant neighbours - PCA axes relationships 56 15. TB frequency of plant neighbours - PCA axes relationships 60 16. KIT frequency of plant neighbours - PCA axes relationships 63 17. WED frequency of plant neighbour - PCA axes relationships 65 18. DC frequency of plant neighbour - PCA axes relationships 66 19. Eigenvalues of the PCAs used in ANOVA only 71 20. Morphology - PCA axes relationships for TB data in 1967 72 21. Morphology-PC A axes relationships for KIT data in 1987 73 22. Morphdogy-PCA axes relationships for WED data in 1987 75 viii 23. Morphology-PCA axes relationships for DC data in 1987 76 24. Morphological variation apportioned between years, species, sites and within and among quadrats by ANOVA 78 25. Canopy cover-morphological PCA relationships 79 26. Summary of relationships between TB, KIT, WED and DC morphological variables and envronmental characters 79 27. Phenology - PCA relationships for TB data 89 28. Phenology - PCA relationships for KIT data 90 29. Phenology - PCA relationships for WED data 90 30. Phenology - PCA relationships for DC data 91 31. Percentage of phenological variation apportioned between years, species, sites and within and among quadrats for devil's club and thimbteberry 93 32. Canopy cover - phenology relationships 95 33. Summary of relationships between TB, KIT, WED and DC phenological variables and environmental characters 96 34. Regression analysis of thimbleberry stem longevity and stem density 105 35. Regression analysis of devil's club stem age and density 105 36. Eigenvalues of 1986 and 1987 PCA for the number of stems surviving to each age class 106 37. Demography - PCA axes fa TB data for the number of stems dying during each age class 107 36. Demography - PCA axes for KIT data for the number of stems dying during each age class 108 39. Demographic - PCA axes for WED data for the number of stems dying during each age class 109 40. Demography - PCA axes for DC data for the number of stems in each age class 110 41. Demographic variation apportioned between years, sites and within and between quadrats by ANOVA 111 42. Canopy cover - demographic relationships 112 ix 43. Canopy cover - age distribution relationships 113 44. Summary of relationships between TB, KIT, WED and DC demographic variables and environmental characters 114 45. Summary of relationships between TB, KIT, WED and DC morphological, phendoacal and demographic variables 115 X list of Figure? 1. Map of tiie study area showing the location of each site in relation to the river system and each other 15 2. Weather in the study areas from January to August 31,1986 and 1987 18 3. Map of quadrats and transects in each study site 20 4. WMDS and CANCORR analyses 42 5. Percentage canopy cover over 63 quadrats at KIT, WED and DC in July, 1987 43 6. Graph of axes I and II of the soils PCA fa TB showing individual quadrats .... 46 7. Graph of axes I and II of the soils PCA fa DC showing individual quadrats .. 50 8. Graph of axes I and II of the presence/absence of plant neighbours PCA fa TB showing individual quadrats 54 9. Graph of axes I and II of the presence/absence of plant neighbours PCA fa DC showing individual quacrats 59 10. Graph of axes I and II of the frequency of plant neighbours PCA fa TB showing individual quacrats 62 11. Graph of axes I and II of the frequency of plant neighbours PCA fa DC showing individual quacrats 67 12. Number of thimbleberry flowers per quadrat at WED and KIT on June 22, 1987 68 13. Number of devil's club racemes per quadrat in 1986 and 1987 70 14. Total number of stems per quadrat fa devil's club on June 30,1986 and 1987 70 15. Vegetative phonological codes at KIT and WED in 1986 and 1987 81 16. Vegetative phonological codes fa devil's club in 1967 and 1987 63 17. Reproductive phenoJogical codes at KIT and WED in 1986 and 1987 85 xi 18. Reproductive phendogjcal codes fa devil's dub in 1986 and 1987 87 19. Survivorship of thimbleberry stems produced in 1986 and 1987 at the WED and KIT sites 100 20. Percentage of stems dying within distinct time periods at KIT and WED sites in 1986 and 1987 101 21. Cumulative number of devil's dub stems older than the time indicated on the x-axis 103 22. Percentage of devil's club stems in distinct age classes on June 30,1987 .... 104 xii Acknowledgement I would like to acknowledge then continuing support and friendship of my supervisor, Gary Bradfield, during the formulation, analysis and writing of the thesis. My committee, Jack Maze, Roy Turkington and Jim Pojar, were also especially helpful at different times throughout the process. Jack Maze was encouraging, challenging and warm throughout. Roy Turkington was especially constructive with editing and writing suggestions in the final manuscript. Jim Pojar, deserves special mention, for his understanding, patience and positive input throughout 'the whole process. r Lacey Samuels offered encouragement and friendship and with Mishtu Banerjee and Len Dyck provided stimulating discussion and exchange of new and interesting ideas. Finally I am most grateful to and appreciative of Rick Caswell, for his excellent field assistance, warmth, understanding and caring during this entire thesis. I. INTRODUCTION Rubus parvfflorus Nutt. (thimbleberry) and Oplopanax horridus (Smith) Miq. (devil's club) are two clonal iteroparous shrubs which occur in different successional stages in B.C. forests. Thimbleberry is found in disturbed areas and early to mid-successional sites whereas devil's club is found in climax forests. The dfferences observed between species, such as thimbleberry and devil's club, are often ascribed to drfferent environmental conditions found in early and late successional seres along a successional gradient (Lee et al. 1986). Variation observed within and among populations and years within a species is linked to other factors. Variation between two genetically similar populations at drfferent sites implies that there are environmental differences between sites which maintain this variation (Van Cauteren and Lefebvre 1986; Huenneke 1987; Matlock 1987; Maddox et al. 1989; Moot et al. 1989; Cartsson and Callaghan 1990). Variation within populations has been linked to genetic and environmental influences (Primack 1980, 1985; Huenneke 1987; Menges 1987; Schlichting 1989). Genetic variability can account for differences between clones of a single population living in a common environment (Huenneke 1987). Variation observed within individuals, which is manifested as plasticity, has been seen as evidence of a heterogeneous environment (Primack 1985; Kephart 1987; Schlichting 1989). Temporal variation, which may also be manifested as plasticity, has also been observed and is often explained by climatic heterogeneity (Kjellson 1985; Rathke 1988). Observations of variability at these differing levels suggest a study quantifying the variation within and among populations and individuals. Investigating factors which maintain that variability may clarify genetic and environmental relationships (Koehl 1989). In this thesis I examine rdationships within and among individuals and populations of thimbleberry and devil's club to gain insight Into the major factors shaping the development of these two species in the Krtimat valley. As ecologically drfferent species, it was of interest to determine whether they responded drfferentJy to drfferent environmental conditions. My study was also motivated by a need within the B.C. Forest Service for information about life histories of these two species which are important understory components. By comparing problem with 2 nonproblem species, the study win elucidate me method ot niche utilization employed Dy eacn. I also had the personal goal to learn more about multivariate statistical techniques and plant ecology. 1.1 OBJECTIVES The objectives of this study are: 1) to document the morphology, phenology and demography of thimbleberry and devil's club in the Kitimat River valley in west central B.C. during the 1986 -1987 growing seasons. 2) to apportion variation in morphology, phenology and demography of devil's club and thimbleberry within and among clones, between sites, species and years and 3) to ascertain to what extent canopy cover, soils and adjacent vegetation are correlated with phenology, demography and morphology at these dfferent levels of organization. 1.2 PHENOLOGY. MORPHOLOGY AND DEMOGRAPHY Phenology, morphology and demography were examined for both species in 1986 and 1987 at each level in the above hierarchy. Phenology is the study of periodic phenomena in an organism's life in response to environmental factors (Funk and Wagnalls 1978). As plants are sedentary organisms changes in timing of responses such as flowering and fruiting may be an effective strategy in a fluctuating environment and may be critical in a harsh environment (Gill and Mahall 1986). More information on phenology can be found in Rathcke and Lacey (1985). Morphology is the study of the form and shape of plants and animals (Funk and Wagnalls 1978) and has also been related to function and strategy in differing environments (Barnes 1986; Lee et al. 1986; Hancock and Pritts 1987). Morphological relationships have been described with techniques such as plant growth analysis (Hunt 1982; Hardwick 1984; Hunt 1984; Jolliffe and Courtney 1984), morphometries (Pimentel 1979) and allometry (White 1981; Joirffe and Courtney 1984; Welter 1987). 3 Rant demographic studies consist of actuarial measures. When correlated with biotic and abiotic factors these measures may explain community (Harper 1980; Crawley 1986; Moot et al. 1989) or population (Hutchings 1976,1986; Harper and Bell 1979; Blom 1988) dynamics. Ecological demographers document the growth of genets or genetically distinct individuals (Harper and Bell 1979) by enumerating genets and their constituent parts, ramets or modules (Hutchings 1976). Modules which nave been selected for study include leaves (Clark 1980; Lovett Doust 1981; Lovett Doust and Eaton 1982; White 1985; Gamier and Roy 1988), buds (Maillette 1982), rhizomes (Bell 1976), shoots (White 1980), branches (McGraw and Antonovics 1983) and flowers (Lovett Doust and Eaton 1982). Male and female plants within a population have also been counted (AKiende and Harper 1989). Commonly, stems are considered the most meaningful subunit for demographic analysis of clonal individuals (Sarukhan and Harper 1973; Hartnett and Bazzaz 1985). Reviews of demographic theory are provided by Harper and White (1974), White (1979), Harper (1980), Blom (1988) and Balogh and Grigal (1988). A comparative study of plant demography and plant growth analysis is described by Hunt (1978), Bazzaz and Harper (1979), Hunt and Bazzaz (1980) and JoWfe and Courtney (1984). Differences within and among species, populations and individuals have often been described for phenology, morphology and demography. Recent studies attempt to link dfffering phenological, morphological and demographic patterns observed in various environments using life-history strategies though overall this approach has not proven conclusive (Steams 1977; Grime 1979; Southwood 1988). 4 1.3 VARIATION IN MORPHOLOGY. PHENOLOGY AND DEMOGRAPHY Variability or variation within- and among-groups has seldom been quantified. Scagel and Maze (1984) found that within individual variance exceeded the amount of between individual variance among two Stipa species and an intermediate when they apportioned morphological variation with ANOVA. Differences between the three groups accounted for the most variation (Scagel and Maze 1984). Other studies have described the morphological variation of individuals and clones in different populations (Aarssen and Turkington 1985; Barnes 1986; Herndon 1989; Maddox et al. 1989). Substantial variation was found for clonal structure both within and between populations (Maddox et al. 1989). Morphological measures of individual traits, such as leaf length or awn length, revealed variation within populations to be large (Aarssen and Turkington 1985; Barnes 1986; Herndon 1987). These studies support earlier research which also described large amounts of morphological variation among (Burdon 1980) and within (Fowler et al. 1983) populations. Reinartz and Popp (1987) reported that morphological variation may also be substantial within and between individuals. Clonal variation was large for shoot height, above and below ground biomass, and the pattern of root connections of northern prickly ash both within and among individuals (Reinartz and Popp 1987). The extent of temporal and spatial variation has been quantified for demography. Demographic variation between years has been reported to be greater than variation occurring between populations in a single year (Weiss 1981; Mack and Pyke 1983; Waite 1984). Angevine (1983) found spatial demographic variation to be greater among populations within a species than between related species. At a finer scale of observation, greater demographic variation was found within a population than between populations at different sites (Huenneke 1987). Spatial demographic variation has also been observed within and between individuals, with genetic (Burdon et al. 1983; Ennos 1985) and environmental (Werner 1985) influences invoked to account for the variation. Several studies have quantified the relative amounts of variation stemming from genetic and environmental effects (Sarukhan et al.1984; Blom and Lotz 5 1985; Cook 1985; Billington et al. 1990). In all cases cited environment accounted for most of the variation (Oka 1976; Sarukhan et al. 1984; Blom and Lotz 1985; Cook 1985). In a recent study, Billington et al. (1990) reported that environment accounted for up to 100 percent of the demographic variation in adjacent grass populations. 1.4 FACTORS CORRELATED WITH VARIATION Factors which maintain variation have been measured at many different scales of observation. Phenological studies have compared similar communities, such as Asian rainforests (Newton 1988) or temperate bogs (Wieder et al. 1984). in widely divergent areas to seek common patterns. At this large scale of observation ii is dfficult to assess factors which control phenological cffferences, due to the interactive nature of biotic and abiotic influences, though climatic differences are often cited (Primack 1985). Within-community studies comparing phenological development of dtfering strata have often been descriptive. Examples include phenological comparisons between the shrub layer and the canopy layer (Ralhan et al. 1985), between deciduous and evergreen trees (Gill and Mahall 1986) or the effects of canopy cover on understory plants (Niison 1986). Studies examining constancy or changes in phenological pattern between years have also been done (Helenurm and Barrett 1987). When several or all species in the community were observed, a sequential pattern of flowering or a divergence of flowering times has been reported (Pojar 1974; Heinrich 1976). Competition for pollinators seems to be the most widely accepted explanation for this asynchrony. Other studies nave stated that nonrandom distributions of phenological data indicate that competition has occurred (Rathke 1984, 1988; Wieder et al. 1984; Kjeilson 1985). Flowering phenology has also been viewed as a secondary trait related to fruit and seed production. According to this idea fruiting phenology is selected for times when seed dispersers are prevalent or seed predators are rare and flowering phenology is subject to these selection pressures (Stiles 1980; Primack 1985). Other explanations of flowering phenology have been purely physiological focussing on the availability of resources within the environment (Heinrich 1976; Kawano et al. 1982; Primack 1985). 6 Bioflc and abiotic influences nave been linked 10 demograpnic venation wrtnin communities or guilds though interpretations applied to these influences are often conflicting (Cook 1979; Bakker et al. 1980; Huenneke 1983; Beatty 1984; Verkaar and Shenkeveid 1984; Sakai and Sluak 1985; Dunn 1986; Huenneke and Sharitz 1986; Van der Toorn and Pons 1988; Balogh and Grigal 1988) Demographic comparisons between species focus on early versus late serai species in an overall successional development (Bierzychudek 1982; Whitney 1986). Morphological comparisons between species are also made at drfferent points along a successional gradient. Though morphological traits are correlated with environmental and ontogenetic differences between species (Antos and Zobd 1984; Barnes 1986; Lee et al. 1986), reproductive effort is most often quantified. Harper et al. (1970) argue that reproductive effort is expected to be maximized by annual species found early in succession, when competition is minimal, whereas perennial species, which predominate in later successional seres, minimize reproductive effort to devote higher proportions of energy to vegetative structures which may confer a competitive advantage (Harper et al. 1970). Since this idea was first espoused ft has been elaborated by several others (GadgJI and Sdbrig 1972; Abrahamson and Gadgjl 1973). Although several studies have measured reproductive effort, few have accounted for below ground biomass. It is also unclear whether patterns in reproductive allocation are phenotypic or genotypic Hi origin (Hancock and Prftts 1987). Demographic (Law et al. 1977; Whitney 1986; Forte 1989) and morphological (Gamier and Roy 1988) patterns were also compared within genera though in many cases no overriding trends were discernible and environmental factors accounting for the variation observed were not identifiable. Environmental causes have been inferred for maintaining morphological, phonological and demographic variation among populations of a single species. Environmental influences were arroborated when drfferent phonological races found in drffering environments were examined in between population studies (Van Cauteren and Lefebvre 1986). Other studies have suggested environmental variables which maintain morphological variation among 7 populations though the relationships were not measured (Aarssen and Turkington 1985; Maddox et al. 1989). Environmental influences have also been suggested by adjacent dissimilar populations found at opposite ends of a gradent, such as moisture, that morphologically integrade where they overlap on the gradient (Barnes 1986). Finally, comparative studies between drfferent populations have provided dues to the environmental correlates of demographic variation. In some cases, no differences in mortality were observed between sites (Gamier and Roy 1988), whereas in other cases sdls properties (Sakai and Sluak 1985; Matlock 1987; Moot et al. 1989) and moisture (Sakai and Sluak 1985; Moot et al. 1989; Carisson and Callaghan 1990) were found to affect longevity. Phendogjcal variation among indviduals has been documented to examine the genetic component of phonological raits (Primack 1980; Primack 1985; Somers and Grant 1981; Kephart 1987). Successful selection of early and late developing subspedes in agriculture is evidence of the strong genetic component to phendogy (Rathcke and Lacey 1985). Mcrphdogjcal studes at the level of the indviduaJ or done have frequently dealt with three ideas. First, morphdogical variation has been explained as an adaptation to dffering environmental factors such as light or sdl moisture (Prtefka et al. 1980; Gawler et al. 1987; Menges 1987). Secondy. the effects of physiological integration on the development of large long-lived donal indviduals has been examined (Hartnett and Bazazz 1983; Slade and Hutchings 1987a; Hutchings 1987b; Schmid and Bazazz 1987; Lau and Young 1988). Physidogical integration implies that ramets in favorable areas may translocate resources to other ramets of the same done in stressed areas and increase the longevity or growth of the redpient ramets. This physidogical integration may buffer ramets within indement areas and facilitate rapid growth of these ramets if condtions improve. Several studes have documented this translocation of resources and examined donal architecture through a spatially heterogeneous area (Hartnett and Bazazz 1983; Slade and Hutchings 1987a; Slade and Hutchings 1987b; Lau and Young 1988). These studes found that rhizomes between stem ramets tended to be fewer and longer through areas of nutrient stress (Slade and Hutchings 1987a; Slade and Hutchings 1987b) and water stress (Hartnett and Bazazz 1983). Clumps of 8 stem ramets may therefore be widely separated or densely packed within a site. Architectural differences may affect competitiveness and/a cooperation between individuals (Hartnett and Bazazz 1985; Schmid and Bazazz 1987). Thirdly relative resource allocations between sexual and vegetative reproductive have been documented (Rtelka et al. 1980; Eriksson 1985; Jurik 1985; Loehie 1987; Cid-Benevento 1987). Sexual and vegetative reproduction may maintain and spread a clonal individual under differing conditions. Abrahamson (1980) considered vegetative reproduction advantageous in early successional sites as it may more rapidly fill a site than sexual reproduction. In later successional sites, where density and competition are greater, sexual reproduction was thought to be a more effective method of clone maintenance (Abrahamson 1980). This idea has since been expanded to include other condtions than competitive intensity and density. Loehie (1987) predicts that 1) sexual reproduction will increase under condtjons which the plant perceives as favorable to germination, 2) where nutrients are available sexual reproduction will be less costly and will increase, 3) vegetative reproduction will increase relative to sexual reproduction when the opportunities diminish for sexual progeny, 4) as the value of vegetative reproduction diminishes sexual reproduction will increase and 5) when both vegetative and sexual reproduction value decrease concurrently, the ratio of sexual to vegetative reproduction will increase. Other researchers have examined the relationship between environment and reproductive allocation though patterns observed are not always consistent and predictable (Piteika et al. 1980; Eriksson 1985; Jurik 1985; Cid-Benevento 1987). Under condffions of increasing density sexual reproduction may increase in relation to vegetative reproduction (Piteika et al. 1980; Eriksson 1985). Other studies have found that sexual reproduction may increase with respect to vegetative reproduction under more favorable conditions such as 9 diminished density (Jurik 1985) or increased water, light or nutrient availability (Loehie 1987). Increased light was also found to increase the amount of sexual reproduction in comparison to vegetative reproduction by Cid-Benevento (1987), though vegetative reproduction was also found to increase. Environmental heterogeneity has been inferred from variation within-individuals or plasticity (Primack 1985; Rathcke and Lacey 1985; Schlichting 1986; Kephart 1987). Morphological patterns within individuals have been linked with this environmental heterogeneity (Oberbauer and Strain 1986; Svenssen and Callaghan 1988; Schlichting 1989). Svennson and Callaghan (1988) found that small scale changes in vegetation and micro-environment were associated with small scale changes in the morphology of Lycopodium annotinum . Canopy position and light have also been correlated with leaf thickness and weight in Pentaclethra macroloba (Oberbauer and Strain 1986). Finally Schlichting (1989) demonstrated changes in the correlation structure of weight, number and sizes of leaf, flower, root and stem parts within Phlox individuals due to drfferent nutrient, moisture and clipping treatments. Demographic studies within-individual plants have focused on stem ramets within clones (Sakai and Sluak 1985). Earlier workers assumed that stem ramets experienced constant risks of mortality regardless of age or size (Harper 1978). Mae recent studies, howeva, have questioned this assumption and suggested that ramet mortality is altered by clonal integration (Cook 1979; Cook 1985; Eriksson 1988), positional effects (Eriksson 1988; Douglas 1989) and survival of sibling ramets on the same stolon (Eriksson 1988). Environmental factors have frequently been cited as responsible fa phenological, morphological and demographic variation at all spatial and temporal scales. Climate is considered to maintain phenological variation between years (Rathcke 1988) and within seasons (Kjellson 1985). Annual dffaences in climate also affect pollinator availability which impacts on flowering phenology, whereas seasonal patterns that affect germination determine optimum time of seed dispersal (Rathcke and Lacey 1985). Physiological studies have accentuated drought stress or water potentials as determinants of phenological strategy 10 (Jackson and Bliss 1984; Gill and Manaii 1900) tnougn temperature is Deiieved to De important for woody shrubs (Ueth 1974; Reader 1983). Interactions of photoperiod, temperature and moisture were also necessary to induce flowering (Rathcke and Lacey 1985). Physiological studies have characterized the morphological responses to differing environmental conditions such as humidity, water stress and temperature (Mooney 1980; Roy and Mooney 1982,1989). Light, canopy cover, substrate, soil moisture and the dominant vegetation have also been correlated with size and number of leaves and the size and number of inflorescences in field studies (Piteika et al. 1985; Gawler et al. 1987). Other studies (Givnish 1982; Menges 1987) predicted and subsequently corroborated morphological patterns in differing environmental conditions. Givnish (1982) predicted leaf height and the proportion of the total biomass devoted to leaves in differing light regimes and Menges (1987) constructed a model of leaf thickness, area and height and the percentage of biomass devoted to roots under differing regimes of canopy opening, drainage, soil texture and plant density. Many studies have examined the environmental factors affecting demography though the large number of factors and their interactions do not form many consistent patterns. Descriptive studies have linked soils texture and moisture (Hobbs and Mooney 1987), disease (Cook 1979), environmental patchiness (Cook 1979), fire regime (Piatt et al. 1988) and competition (Groves et al. 1990) with demography. Experimental studies involving manipulations have linked fire regime with demography (Hartnet 1987) as well as nutrient enrichment (Noble et al. 1979), predation (Noble et al. 1979; Bazely and Jefferies 1989) and trampling (Noble et al 1979). Other influences that have been identified include soil nutrients (Beatty 1984; Kelly 1989a), soil texture (Beatty 1984) and water level and sediment depth (Beatty 1984; Huenneke and Sharitz 1986). Cook (1979) also stated that herbivores, pathogens and drought stress are important causes of mortality in many communities. Several studies have accentuated the effects of drffering canopy cover or light conditions on demographic patterns (Huenneke 1983; Sakai and Sluak 1985; Dunn 1986; Balogh and Grigal 1988). In several shrub species abundance is controlled by regeneration which may dfffer under dffering canopies (Balogh and Grigal 1988): under a closed coniferous canopy 11 regeneration was inversely related to site quality and moisture; under an aspen canopy regeneration was inversely related to soil nutrients. Canopy openings allowing greater light penetration into forest stands were also correlated with increased density and survival of Lonicera (Luken 1988). In another study increased survival during cold winters was reported for Plantago growing in the shelter of an overstory canopy (Moot et al. 1989). Kelly (1989b) showed that shading by taller neighbours and density of these neighbours affects recruitment of three annuals and two biennials in the grass chalklands of England. Kelly (1989a) also stated that daylength and temperature were also linked with variation in germination times. Low light and vegetation density were also linked to demography by Cook (1979), Verkaar and Shenkeveld (1984) and Van der Toorn and Pons (1988). 1.5 THIMBLEBERRY Thimbleberry is a common shrub throughout B.C. Its range extends to 55' N latitude in the interior (Haeussler and Coates 1986), though ft is found as far north as Alaska along the coast (Hulten 1968). The southern limit of its distribution is southern California (Hitchcock and Cronquist 1973). Thimbleberry is morphologically quite variable and several subspecies have been described (Hulten 1968). The plant is composed of dumps of canes or stems which are connected by a network of rhizomes. Each stem lives from one to three years, bearing fruit and becoming lignified after the first year of growth. The dedduous stems are glandular and hairy before lignrfication occurs. Thimbleberry leaves are palmate, from 5 to 30 cm in length and feel soft like tissue paper. The showy, white flowers are borne in loose cymes (Hitchcock and Cronquist 1973). They produce a thimble-sized, soft, orange-red fruit which is an aggregation of drupes. Thimbleberry is frequently found on sites following dsturbances. It is espedally successful at cdonizing if present before dsturbance (Bs 1981). Halpern (1989) found thimbleberry to be the dominant plant spedes approximately 10 years after logging and to have greater longevity than other early succession cdonizers. Although not plentiful in later 12 successional seres, tnimbieberry is present In natural gaps in the forest, in west central B.C. , the plant is widespread, occurring in many forested ecosystems (Coupe et al. 1982). Thimbleberry is perceived by foresters to be a serious competitor for light and nutrients, thus hindering the growth of young trees (Haeussler and Coates 1986). Many reports have focused on forest management of the species (Gratkowski 1971; Stewart 1973; Stewart 1974; Coates and Haeussler 1986). Biomass regression equations have also been prepared to provide inventory information (Alaback 1980; Aiaback 1986). Like many weedy species (Baker 1974; Holzner 1982), thimbleberry may be apomictic. Agamospermy is common in many Rubus species (Grant 1981; Richards 1984), though no information is available on the breeding system for thimbleberry. Richards (1984) states that few members of the genus are known to be fully sexually diploids. Of the thousands of Rubus species most are polyploid and apomictic (Nybom 1987). A subject of other thimbleberry papers is plant/insect interactions, though these are primarily entomological (Briggs et al. 1982; Gilbert and Gutierez 1973; Jones 1983; McNicoi et al. 1983). A comprehensive review of thimbleberry is found in Haeussler and Coates (1986). 1.6 DEVIL'S CLUB In B.C., devil's club is distributed from Vancouver Island, north along the coast to Alaska and in the interior wet belt (Lyons 1976). Its range also extends into Oregon (Hitchcock and Cronquist 1973). The stems and the underside of the large leaves have many thorns. This large plant is often up to 3 meters high and 5 meter long decumbent stems are common. Leaves are large, up to 35 cm wide, maplelike in shape and 7 - 9 lobed. The white flowers are produced on an elongate raceme and yield showy scarlet berries. Devil's club is found in old growth forests and does not persist in logged areas. Klinka et al. (1989) state that it commonly occurs in very moist, water receiving sites with nitrogen rich soils. 13 No information on reproductive biology is available though most members of the Araliaceae are known to be outcrossing and fully sexual and some are dioecious (Flanagan and Moser 1985). Devil's club flowers are hermaphroditic (Hitchcock and Cronquist 1973). Little research has been done on devil's club except for ethnobotanical studies (Turner 1982) and a pharmacological report (Smith 1983). Biomass regression equations, which predict above-ground biomass from stem basal diameter, were calculated by Alaback (1980, 1986) and Yarie and Mead (1989) to assist in field inventories. 14 II. STUDY AREAS 2.1 DESCRIPTION OF STUDY AREA 2.1.1 GEOGRAPHIC LOCATION The three study sites are located in west central British Columbia about 8 kilometers east of the townsite of Krtimat along Highway 25. This area is 54*10' latitude and 128' 35' longitude. All three sites are close to the Krtimat River or its estuary. The devil's club site, referred to as DC throughout this document, is located within a B.C. Forest Service reserve on the east bank of the river in an area of old growth forest designated as L6201 and L6202 by the Ministry of Forests (Figure 1). One thimbleberry site, designated as WED throughout this thesis, is situated between the Wedeene and the Little Wedeene rivers close to their confluence with the Krtimat River at L6144 and L6141 (Figure 1). The other thimbleberry site, KIT. abuts a small arm of the Krtimat River at L6235 and L6235 (Figure 1). KIT and WED will be collectively referred to as TB. The study sites were chosen because of the abundance of the study species, their close proximity and the differing canopy cover at each thimbleberry site. 2.1.2 EDAPHIC ENVIRONMENT Much of the bedrock within the area is composed of crystalline rocks. Throughout the valley large deposits of unconsolidated glacial material formed from the breakdown of these hard rocks are common. Along the river and its tributaries glaciofluvial and fluvial materials form floodpiains and terraces made up of sequential deposition of these materials (Haeussler et al. 1984). The study sites were formed by this sequential deposition and may still experience flooding. Channels of water and standing pools are common throughout the sites. All three sites are edapfucally similar. Soils are generally Orthic Humo-Ferric Podzols with thick Ae horizons. Due to the heavy rainfall and the coarse nature of these soils the Ae horizons lose aluminum and iron to deeper reddish brown Bf horizons. Decaying wood and other organic materials, impregnated with fungal hyphae, blanket the ground forming a surface LFH layer often greater than 10 cm thick (Valentine et al. 1978; Haeussler et al. 1984). 15 Figure 1. Map of the study area showing the location of each site in relation to the river system and each other. (— « gravel logging roads) KIT - gjdled alder thimbleberry site, WED - nongrdled alder thimbleberry site. DC - devil's dub site 16 2.1.3 DISTURBANCE HISTORY Although one or two trees adjacent to the river were felted, the devil's club site remains relatively undisturbed. The only other signs of human disturbance include a large hole in one of the oldest cedars and a fishermen's path which follows the river at the edge of the site. An area adjacent to the DC site is presently used by the townspeople as a source for firewood and occasionally trees were felled onto the B.C. Forest Service reserve. Within the reserve the DC site encompasses an area of about 60 square meters. Disturbance is more noticeable in the the thimbleberry sites. The KIT thimbleberry site was logged in 1969 and subsequently burned. The WED thimbleberry site was logged in 1969, burned and planted with spruce in 1975. Regeneration of Alnus rubra (alder) proceeded naturally on both sites and a dense canopy of this species rapidly filled Hi both areas. In 1985 alder were girdled leaving the trees standing but effectively dead at the KIT site. 17 2.2 CLIMATE The study area, which is characterized as sub-oceanic or coastal transitional, is located in the Northern Drier Maritime Subzone of the Coastal Hemlock Biogeoclimatic Zone (CWHfl) (Krajina 1965,1969; Haeussler et al. 1984). This zone has wet and mild weather compared to other areas at similar latitudes. This climate is predominantly coastal but is influenced by continental factors. Warm maritime winds deposit rain onto the westward facing slopes of the Coast Mountains and by the time they reach the CWHf zone have lost much of their moisture. As a result this area is drier than the outer coast and may have long dry spells during which fire hazards are common (Haeussler et al. 1984). In the winter the ground is covered by a thick layer of snow and does not usually freeze. When cold spells due to the influx of continental air do occur their duration is short (Haeussler et al. 1984). Weather patterns for precipitation, hours of sunshine and number of degree days from January to August of 1986 and 1987 are displayed in Figure 2. This number of degree days greater than 0* C for each time period is calculated as: i-n J(mean T*j > 0*C) i-1 where n is the number of days in the time period, i is the day number and mean T is the average temperature for day i greater than 0* C. The summer rainfall was high in both years and there was no two week period without rain. Although spring temperatures dfffered little between years, 1987 had more rain than 1986 from March to mid-June, whereas from mid-June to August 1987, was more sunny, warmer and had less rain than 1986. 18 Figure 2. Weather in the study areas from January to August 31,1986 and 1987 (Environment Canada, 1986 & 7). a) number of degree days 400 DATE 1987 1986 b) number of hours of sunshine 200 150 OO rr O UJ 100 X oo T 1 1 i 1 r j i i i i i_ VN ^ *^ ^ <^ ^ DATE c) rainfall (mm) 1987 1986 200 ^ 4? # 17 quadrats and <5 quadrats removed from KIT, WED, DC raw data matrices All species in > 38 quadrats and <4 quadrats removed from TB raw data matrix Frequency variables include lists species in quadrats coded fromO- 4 based on number of quadrat quarters in ¥ai i | j :o fl PCA • covariance matrix among »JCANC0RR1 variables •KIT, WED, TB, DC Euclidean distance matrix among quadrats -KIT, WED, TB, DC mm Presence/Absence j l PCA - covariance matrix among variables JCANCORRI variables include lists of al species coded 0 or 1 based] |-KIT, WED. TB. DC on presence/absence Euclidean distance matrix among quadrats -KIT, WED, TB, DC mm 35 The influence of plant neighbors was examined at different scales as different environmental attributes may be reflected at each scale. As presence/absence data of plant neighbors represents a coarser scale of measurement than cover or frequency of vegetation, more coarse grained environmental gradients may be reflected at this scale (Allen and Wyleto 1983). Allan and Starr (1982) also indicate that coarse gained presence/absence data may reflect influences due to time scales which are not apparent in finer-scaled frequency data. 5.3 MORPHOLOGY The morphological variables and analyses are displayed in Table 4. Data were divided into groups for analysis. Except for the ANOVA between species, devil's club and thimbleberry were analyzed separately. The two species were sufficiently different that morphological descriptions of each one required different variables. These divisions also reflect differences in annual sample sizes . In 1986,10 quadrats only were sampled whereas 21 were sampled in 1987. Except for the ANOVAs which apportioned variation among quadrats and years, data from 1987 were used. Input variables used in simple correlation, ANOVA and CANCORR were generated by PCA. All PCAs were performed on correlation matrices as measurement scales differed between variables. Correlations were calculated between PCA scores and the 1986 and 1987 canopy cover data for DC, KIT, WED and TB. ANOVA was used to partition PCA scores among years, sites, species and quadrats. Variation in thimbleberry data was partitioned using a nested ANOVA as years, sites and quadrats could be simultaneously considered. Devil's club was not included in the nested ANOVA as only differences between thimbleberry sites were of interest. ANOVA was also used to partition variation among quadrats fa the number of stem leaves in 1987, the stem lengths in 1987 and leaf length and width in 1986. Most quadrats contained at least 5 replicates of these variables whereas other mophological traits wae not as numerous within quadrate. As sample sizes within quadrats wae unequal these variables wae 36 not summarized with PCA Dirt were analyzed individually. Finally Euclidean distance matrices among quacrats were calculated from the standardized morphological data and analyzed using WMDS. Matrices of PCA axes scores and Euclidean distance matrices were produced fa the KIT, WED, DC and TB data sets. Table 4. Analytical methods fa morphological data. Arrows indicate the sequence of analyses from raw variables through summary variables to ANOVA, WMDS, CANCORR a simple correlation. Sampling time is the second week of July unless otherwise indicated. (When KIT, WED, TB a DC are listed, a separate analysis is performed fa each) KIT « girdled alder thimbleberry site, N • 21 in 1987,10 in 1986 WED • nondrded alder thimbleberry she, N - 21 in 1987,10 in 198 DC - devil's club ate, N - 21 in 1987,10 in 1986 TB - KIT and WED thimbleberry data, N • 42 in 1987, 20 in 1986 1986 & 1987 TB variables include: * flowers per branch @ Aug.8, "flowering stems @ Aug. 8," 2-year-old stems @ June 30," 1-year-old stems @ June 30, " leaves/1-year-old stems, " leaves/2-year-old stems.1-year-oJd stem length, 2-year-old stem length, 1-year-old stem basal diameter 2-year-old stem basal diameter, brandi length, branch basal diameter PCA -correlation matrix among variables ANOVA - to apportion TB variation among years and sites -Y • A + B(A) + E where A - years. B - sites, E -erra a within sites 8c Y • PCA axis scaes 1986 & 1987 DC variables indude: " stems @ June 30," branches/stem, " leaves/stem, stem length, stem dameter, branch length, brand diameter - correlation matrix amonc variables AR5VA - to apportion DC variation among years -Y = A + E where A - years E » erra a within years 8c Y - PCA axis scaes 1987 TB 8c DC variables indude: " branches per stem, " flowers/stem (TB) a bunches (DC)," stems, " leaves/branch, " leaves, stem length, stem dameter, branch diameter, branch length PCA - correlation matrix amonc variables ANOVA - to apportion variation among spedes - Y - A + E where A - spedes, E » erra 8 Y • PCA axis scaes Table 4 continued 37 DC&TB 1987 *stem leaves 1987 stem length 1986 leaf length 1986 leaf width ANOVA - to apportion DC & TB variation among quadrats Y - A + E where A - quadrats. E » error & Y - raw data 1987 TB. KIT. WED variables indude: * brandies/stem <§> 30/6 * flower/ inflorescence @ 8/7, * flowering stems @ 8/7* flowers/stem @ 8/7, # 2-year-old stems @ 30/6, * 1-year-old stems @ 30/6, # leaves, * branch leaves, * leaves/1-year-old stem, * leaves/ 2-year-old stem, 1-year-old stem length, 2-year-old stem length, 1-year-old stem diameter, 2-year-old stem diameter, branch length, branch diameter PCA - correlation matrix among variables -KIT, WED, TB correlation with canopy cover from 1986 & 1987 -KIT, WED, TB Eudidean distance matrix among quadrats KIT, WED, TB mm 1987 DC variables indude: » branches/stem, * flower bunches/ stem, * stems, * stem leaves, * branch»- correlation matrix leaves, stem length increment, stem length, stem diameter .branch length increment, branch diameter, branch length PCA among variables correlation with canopy cover from 1986 & 1987 CANCORR 4Eudidean distance matrix among quadrats 5.4 PHENOLOGY The sequence of phenological analysis is displayed in Table 5. The slope of a linear regression line was calculated as input data to all subsequent analyses. The slope of the line was an effective summary variable as it gave an estimate of the rate of change of phenological development with time. Simple linear regression was found to yield higher r2 values than log-38 log or log-normal relationships in all cases. Slopes were calculated tor eacn year and Tor Dotn vegetative and generative data at each quadrat for 1986 and 1987 and for stems within each quadrat in 1987. PCA axes scores of covariance matrices and Euclidean distance matrices of slopes of regression lines were used in subsequent CANCORR and WMDS. The slopes of regression lines were also correlated with canopy cover in 1986 and 1987. Finally ANOVA was used to partition generative and vegetative slopes of regression lines between years, sites, species, quadrats and stems within quadrats. Nested ANOVA was used to partition variation among years and sites fa thimbleberry as this model allowed simultaneous comparisons between drfferent levels in the hierarchy. Devil's club was not induded in the nested analysis as site differences fa thimbleberry only were of interest. Separate ANOVAs were used to partition variation fa both generative and vegetative data. Generative data may be less plastic than vegetative data and both vegetative and generative were considered functionally separate. ANOVA was also used to apportion variation among quacrats using both stem and branch phendogies. Variation within quacrats was apportioned separately from other sfata as the number of stems and branches within quacrats were not equal. 5.5 DEMOGRAPHY The effect of density on thimbleberry longevity and the rate of stem production was examined with regression analysis of the number of stems in a quadrat and the percentage of the total number of stems in a quadrat surviving to each time period. The time periods were those used in PCA analysis (Table 6). The sequence of all further analyses is dsplayed in Table 6. PCA axes scaes of covariance matrices were used as input data in ANOVA and CANCORR; whereas Euclidean dstance matrices were used in WMDS. ANOVA was used to partition variation between years and sites fa thimbleberry and within and between quacrats fa both thimbleberry and devil's dub PCA axes scaes. Comparisons between the two spedes were not possible as devil's dub numbers dd not fluctuate significantly during the course of the study. 39 Table 5. Analytical methods for phenological data. Arrows indicate the sequence of analyses from raw data through summary variables to CANCORR, ANOVA and WMDS. (When KIT, WED, TB or DC are listed, a separate analysis is performed for each) KIT - girdled alder thimbleberry site, N = 21 WED = nongirdled alder thimbleberry site, N = 21 TB « KIT and WED thimbleberry data, N - 42 DC = devil's club site, N = 21 Euclidean distance matrices among quadrats -KIT, WED.TB, DC mm Vegetative and generative codes at each quadrat Regression Y • mX + E where X number of days since Jan 1 when sampling occurred, Y = phenological code. E = error & m = slope slope = summary variable I ICANCORR PCA - slopes of 1986 & 7 vegetative 8c generative data are input variables - covariance matrix among variables -KIT, WED.TB, DC ANOVA - all are done for vegetative & generative i) - to apportion TB variation among years & sites Y - A + B(A) + E where Y - slope, A - years, B = sites & E = error or within sites -TB ii) - apportion DC variation between years Y = A + E where Y - slope, A - years DC iii) - to apportion variation between species Y = A + E where Y - slope, A - species & E = error KB&DC correlation with canopy cover from 1986 8c 1987 40 Table 5 continued Vegetative and generative codes for branches and stems in 1987 Y - mX + E where X» number of days since Jan 1 when sampling occurred, Y « phonological code, E - error & m = slope slope • summary variable SR5VA - to apportion variation among quacrats = A + E where A • quadrats and E = error The number of stems living fo each age class from cohorts produced in May 1986 and 1987 were used as input data to ANOVAs which apportioned thimbleberry variation between years and sites. This measure incorporates both density effects and life expectancy between quacrats through time. Age dstributions were partitioned among quacrats for both species. Age distributions represented the variation at one time only and allow comparison between devil's club and thimbleberry. The age distributions did not, however, have equal sample sizes within quacrats so devil's dub age data were grouped to satisfy PCA requirements. Table 6. Analytical methods fa demographic data. Arrows indcate the sequence from raw data through summary variables to ANOVA, CANCORR and WMDS. (When KIT, WED, TB a DC are listed, a separate analysis is performed fa each) KIT • girded alder thimbleberry site, N • 21 WED - nongrded alder thimbleberry she, N - 21 TB - KIT and WED thimbleberry data, N - 42 DC - devil's dub she, N - 21 1986 & 1987 TB variables - * of stems livin< to age dass where age dasses are: dayO, 1-55 days. 56-105 days, 106-450 days, > 451 days covariance matrix amon variables ANOVA - to partition TB variation among years and sites -Y»A + B(A) + E where A - years, B - sites, E - erra and Y - PCA axis scaes 1986 & 7 TB, WED, KIT variables - * of stems livin to age dass where age dasses are: 1986 - 0-55, 56-105, 106-360, 361-415. 416-465, 466-810, >811,1987 - 0-55.56-105,1 OS-450. >451 PCA covariance matrix amon variables ^CANCORR orrelation with canopy cover from 1986 & 1987 Euclidean dstance matrices among quadrats WMDS Table 6 continued 41 TB - age of all stems in quadrat @ 30/6/87 ' ANOVA - to partition TB variation among quadrats - Y • A + E where A » quadrats, E - error and Y - raw data DC - age of all stems in quadrat @ 30/6/87 ANOVA - to partition DC variation among quadrats - Y • A + E where A • quadrats, E = error and Y - raw data 1 DC - age distribution of stems * of stems of 0-5,6-10,11-15, 16-21 and 22-28 years old PCA - covariance matrix among variables JcANCORR @i luclidean distance matrix mono quadrats mm 5.6 CANCORR ANP WMDS Relationships between all environmental variables and plant characters and between the matrices of plant characters were examined with WMDS and CANCORR (Figure 4). PCA axes scores were used as input data to CANCORR and WMDS was performed on pars of Euclidean distance matrices (Tables 3 - 6). The first 10 axes generated by PC As were used in CANCORR, though interpretation was based on the first three only. Canonical axes were only interpreted when canonical correlation coefficients were greater than .75 and the redundancy value for each matrix was greater than 1 divided by the number of variables in that matrix. All correlations of less than .50 were not considered large enough, either between PCA axes and raw data or canonical axes and PCA axes, to warrant interpretation. The analyses outlined in Figure 4 were performed for KIT, WED, DC and TB. WMDS is used to corroborate CANCORR and WMDS results are reported only where they clarify discrepancies in the CANCORR results. 42 5.7 COMPUTATION All analyses were done using the computing facilities at the University of British Columbia. The computing packages used included UBC ANOVAR (Greig and Osterlin 1978) for analysis of variance, MIDAS (Fox and Gure 1976) for all simple correlation analyses and some PCA analyses, SYSTAT (Wilkinson 1988) and SYGRAPH (Wilkinson 1988) for all graphical presentation of the data and SAS (SAS Users Guide 1986) for all WMDS and CANCORR. The ALSCAL procedure which is incorporated into SAS was used for all WMDS (Young and Lewyckyj 1979). Some PCA analyses were also performed using a program written by Gary Bradfieid of the Botany Department. All packages, except SYSTAT and SYGRAPH were available on the mainframe. SYSTAT and SYGRAPH were used on the Madntosh Plus personal computer in the Botany department. Other packages and subroutines developed at the University of British Columbia were also used. Figure 4. WMDS and CANCORR analyses. Arrows indcate all parwise comparisons analyzed with WMDS and CANCORR. Analyses were performed using WED, KIT, TB and DC data sets. KIT - girded alder thimbleberry site, N - 21 WED • nongrded alder thimbleberry site, N - 21 TB - KIT and WED thimbleberry data, N»42 DC - devil's dub site, N - 21 43 VI RESULTS 6.1 ENVIRONMENTAL RESULTS 6.1.1 CANOPY COVER All sites range from little or no cover to a dense canopy overhead (Figure 5). The WED and DC sites have the most dense canopy cover with most quadrats having over 60 percent cover. Due to the alder grdling at the KIT site, most quadrats have less than 30 percent cover. The higher cover values at KIT quadrats 3, 8 and 14 are due to adjacent willows, which were not girdled. Quadrats 18 and 5 are shaded by elderberry. Figure 5. Percentage canopy cover over 63 quadrats at KIT, WED and DC in July 1987. Numbers refer to the quadrats at each site. LU. 100 90 80 70 60-50 40 30 20 10 3 * 18« 5 « 1 4 * 2 ; 1 2 21* 16 « 2 0 1 • 11 1 9 tm 2 6 * 3 7 3 8 44 46 4 9 ; 25 3 0 | 3 3 2 7 2 3 7 8 2 9 2 8 45 5 9 : f | 3 1 2 3 . 39 o x . 5 4 * 4 8 1 2 3 . 2 4 ' 2 2 * 42 • 41 • 36« 34« 3 5 * 40 • -33-*-55 1 4 7 5 1 * 5 7 * 58 * 50« 62 • 53 63 < 61 ' 60 • KIT W E D DC 44 6.1.2 SOILS VARIABLES Axis I, of the TB soils PCA, may represent a nutrient gradient (Table 7). The soils are more basic with a higher percentage of sand at the positive end of the axis and are acidic, have more nitrogen, organic material and coarse fragments at the negative end of the axis (Table 7). Axis II suggests a textual gradient or the time since the site was flooded and mineral soil was covered with a layer of silt and clay (Table 7). The percentage of clay and silt are negatively correlated with axis II whereas the depth to mineral soil, the percent organic matter and the percentage of sand in the soil are positively correlated with the axis. Axis III may be a gradient of drainage conditions or the amount of time that quadrats are submerged (Table 7). The percentage of clay and the depth to mineral soil are positively correlated with the axis (Table 7). As smaller clay sized particles retain water and limit drainage, biological processes may be inhibited and organic matter may accumulate on the surface. There is some separation between the two sites on both axes (Figure 6). The KIT scores are mostly positive for axis I and negative for axis II (Figure 6). This suggests that KIT may have less rich soils and may have been flooded more frequently than WED (Figure 6). The WED site is further from the river and has a blanket of rich alder leaves covering the ground, which supports this interpretation. 45 Table 7. TB soils - PCA axes relationships. Correlation coefficients relating soils variables to PCA axes l-lll. Axis 1 II III % variation 47.78 18.33 13.99 soils variables PH .7313 -0863 .0539 %coarse -.7825 .3885 -.0809 %sand .6942 .6520 -.0999 %silt -6311 -.6363 -.2357 %clay -.0965 -.3349 .8882 organic matter -8663 .3409 .0264 nitrogen -.9189 .0063 -.1492 depth to mineral soil -.4372 .4992 .4820 The KIT soils PCA axes are similar to the TB data set (Table 8). Axis I may be a nutrient gradient. Total soil nitrogen, the percentage of organic matter and the percentage of sift are negatively correlated with the axis whereas the percentage of sand and pH are positively correlated with the axis (Table 8). Axis II represents a textual gradient or the time since flooding. The percentage of silt is negatively correlated with the axis and the percentages of coarser fragments and clay and the depth to mineral soil are positively correlated with the axis (Table 8). Axis III suggests a drainage gradient or the duration of flooding. The percentage of clay is positively correlated with the axis, whereas, the percentage of coarse fragments are negatively correlated with the axis (Table 8). 46 Figure 6. Graph of axes I and II of the soils PCA for TB showing individual quadrats. 3 8 » 3 0 • 40 20 U 2 6 . 3 6 • 10 • 42 *28 • 4 • 35*2 • 41 • 8 • 14 • 16 0768 - 0 . 7 2 6 3 - 0 . 3 7 5 7 - 0 . 0 2 6 2 0 . 3 2 5 3 0 . 6 7 6 8 - 0 . 9 0 1 9 - 0 . 5 6 1 0 A X I S 1 - 0 . 2 0 0 5 0 . 1 5 0 0 0 . 5 0 0 6 47 Table 8. KIT soils - PCA axes relationships. Correlation coefficients relating soils variables to PCA axes Mil. Axis 1 II III % variation 42.79 20.84 13.55 soils variables PH .7250 -.0978 -.4170 %coarse -.4674 .6439 -.5359 %sand .8479 .1526 -.1780 %silt -.6564 -.6614 .2186 %clay -.0914 .5730 .5306 organic matter -.7784 .3758 -.2135 nitrogen -.9133 .0064 -.1095 depth to mineral soil .2844 .5596 .4520 Axis I of the WED soils PCA suggests a nutrient gradient (Table 9). The percentages of soil nitrogen, organic matter, coarse fragments, sift and clay are negatively correlated with the axis, while the pH and sand content are positively correlated (Table 9). Axis II again suggests a textural gradient or the time since the site was flooded. The percent sift and day are negatively correlated with the axis and the percent sand is positively correlated with the axis (Table 9). Axis III suggests a drainage gradient or a gradient of standing water, being positively correlated with the depth to mineral soil (Table 9). 4 8 Table 9. WED soils - PCA axes relationships. Correlation coefficients relating soils variables to PCA axes Mil. Axis 1 II III % variation 59.96 18.92 10.47 soils variables PH .8346 -.3089 .2380 %coarse -.7898 .3767 -.3823 %sand .8030 .5190 .1158 %silt -.7886 -.5441 -.1002 %clay -.5112 -.6580 .3830 organic matter -.8695 .3231 .0118 nitrogen -.8867 .0989 -.3449 depth to mineral soil -.6379 .4050 .5879 Axis I of the DC soils PCA may represent a drainage gradient. The depth to mineral soil and the clay content are positively correlated with the axis and the sand content and the percentage of organic fragments are negatively correlated (Table 10). The pH is also postively correlated with axis I (Table 10); however, it is consistently acidic at this site and varies little. Most of the soils samples also contained a high percentage of organic matter from fallen trees and branches. Axis II is correlated with the soil nitrogen content and the percent coarse fragments and organic matter (Table 10). The underlying biological significance of the correlations to Axis II is not clear. Axis III suggests the time since the time site, which is located adjacent to the river, was flooded. The percentage silt has a strong negative correlation with this axis (Table 10). 49 Table 10. DC soils - PCA axes relationships. Correlation coefficients relating soils variables to PCA axes l-lll. Axis 1 II III % variation 32.25 22.06 18.89 soils variables PH .5778 .2860 -.4400 %coarse -.3988 .6240 .1185 %sand -.8243 -.3483 -.0486 %SJIt .2226 .0480 -.9018 %clay .6350 .2835 .6330 organic matter -.6801 .5930 .1095 nitrogen -.3813 .7139 -.2689 depth to mineral soil .5888 .4777 .0544 Most quacrats to the northeast of the ditch running between the two transect lines have positive scaes on axis I of the DC soils PCA, suggesting that they may have poorer crainage (Figure 7). Those quadrats closer to the river have more negative scaes (Figure 7). These quadrats are somewhat less boggy and may be higher in elevation (Figure 7), though this was not measured. Transects are dissimilar (Figure 7). AXIS 2 P ? ? P ? ? ? ? ? P ? ? ? ? ? ? ? ? ? ? P ? ? P 0 0 0 0 ° o o o o o o o o o o o o o o o o o o o o o o o o o o o m m m in m m m m ni m o o o o o o o o o o 5 1 6.1.3 PLANT NEIGHBOURS-PRESENCE/ABSENCE Although the PCA of the total thimbleberry data set of presence/absence data represents only 30 percent of the total variation on the first three axes (Table 11), these axes can nonetheless be ascribed with environmental processes. Axis I may be a gradient of disturbance due to girdling and a fluctuating water table. The axis is positively correlated with Sambucus racemosa. Osmorhiza chilensis and Epilobium watsonii (Table 11). Sambucus racemosa is characteristic of disturbed sites whereas Osmorhiza chilensis is frequently indicative of a fluctuating water table (Klinka et al 1989). The axis is negatively correlated with Dryopteris assimilis. Tiarella unifoliata and Cornus canadensis (Table 11), which are usually found in moist woods in the Skeena area (Coupe et al 1982). Axis II is negatively correlated with Equisetum arvense. Thelypteris phegopteris. Galium triflorum and Athyrium filix-femina (Table 11), which are characteristic of moist, nutrient rich areas (Klinka et al. 1989). Axis III may be a disturbance gradient representing changing moisture regimes and nitrogen content. The axis is positively correlated with Epilobium anaustifolium (Table 11), characteristic of disturbed sites (Klinka et al. 1989) and Cornus sericea (Table 11), found in flooded areas (Klinka et al. 1989). Streptopus roseusand Gymnocarpium drvopteris are negatively correlated with the axis (Table 11) and common in stable, moist, rich sites (Klinka et al. 1989). The KIT quacrats are more positive on axis I than the WED quacrats (Figure 8). This acrees with the previous interpretation of the axis as a disturbance gradient. There is no separation between the sites on axis II (Figure 8). Quacrats 21 and 38, at KIT and WED, have most negative scores and both are located in wet, sunken areas. Quacrats 1 and 36, which are most positive, are located in open areas which may be subject to flooding and crying. This suggests that axis II may be interpreted as a disturbance gradient due to moisture. Maianthemum dilatatum and Ribes laxiflorum are positively correlated with this axis (Table 11) and both are characteristic of flooded sites and shaded moist stream banks (Klinka et al. 1989). 52 Table 11. TB plant neighbour© - PCA axes relotionehipo. Correlation coefficients relating the presence/absence of plant neighbours to PCA axes I-III. Axis I II III % variation 12.19 9.40 8.33 species Rubus spectabilis .0854 .2837 .0800 Sambucus racemosa .5836 -.0510 -.0843 Galium triflorum .3215 -.5328 .0414 Stachys mexicana .2714 .0376 .3529 Osmorhiza chilensis .5129 .2123 -.0035 Cinna latifolia .2155 .2725 .1789 Epilobium watsonii .5024 .2251 .0419 Agrostis capillaris (tenuis) .3440 -.1157 .1807 Picea sitchensis -.3586 .2739 -.0516 Aruncus dioicus .3342 -.1595 -.2461 Epilobium anaustifolium -1255 .2600 .5834 Chamaecyparis nootkatensis .3194 -.4551 .0202 Ribes bracteosum -.3301 -.0844 .1160 Dryopteris assimilis -.5591 -.0492 .2067 Bromus vulaaris .3570 -.2046 .0640 Salix sitchensis .4335 .2583 .4013 Oplopanax horridus -.0270 .1951 -.4560 Streptopus roseus -.3283 .2372 -.5260 Athyrium filix-femina .0394 -.5268 -.0014 Gymnocarpium drvopteris -.1835 -.0395 -.5340 Equisetum arvense -.0269 -.7446 -.1356 Smilacina racemosa -.0019 .1305 -.2269 Tiarella trifoliata -.0089 -.3162 -.2312 53 Table 11 continued Tiarella unifoliata -.5379 -.4020 -.3290 Viola glabella -.3148 -.1537 .4333 Tellima orandiflora .4133 -.1849 -1962 Maianthemum dilatatum -.4351 .3535 -.2194 Clintonia uniflora -.3917 .0948 -.3827 Circaea alpina -.2330 -.3989 .0624 Thelypteris phegopteris -.1894 -.5918 .1824 Ribes laxtflorum -.3474 .4359 .0374 Cornus sericea -.1820 -.0900 .5035 Alnus rubra .0251 -.2684 -.2450 Vaccinium ovalifolium -.4858 -.2419 .4546 Cornus canadensis -.6626 -.0859 .3205 Axis I of the KIT presence/absence of plant neighbors PCA may represent a disturbance gradient. It is positively correlated with devil's club, Streptopus roseus. Gymnocarpium dryopteris. Tiarella unifoliata and Tiarella trifoliata (Table 12), which tend to occur in rich, moist, less disturbed areas (Klinka et al. 1989). The axis also has a weak negative correlation with Epilobium watsonii (Table 12), which is an introduced species found in open and disturbed areas. Axis II is positively correlated with Galium triflorum. Athyrium filix-femina. Bromus vulgaris and Equisetum arvense and negatively correlated with Piceae sitchensis (Table 12). No obvious interpretation emerges due to the ecological dissimilarity between these species (Klinka et al. 1989). This distribution of Picea sitchensis. however, could be due to planting; though no records of this were available. Axis III has no strong positive correlations (Table 12) and was not interpreted. 54 Figure 8. Graph of axes I and II of the presence/absence of plant neighbors PCA for TB showing individual quadrats. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0 0 0 0 0 -0 •0 ™ -0 . w -0 ~ -0 . - - 0 . -0 . -0 . -0 . -0 . •0 . •0 . -0 . -0 . •0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . -0 . •0 . 51490 49485 47480 45476 43469 41464 39459 37454 35449 33444 31438 29433 27428 25423 23418 21412 19407 17402 15397 13392 11388 93813E-01 73761E-01 53709E-01 33657E-01 13605E-01 64464E-02 28498E-01 46550E-01 66602E-01 86654E-01 10671 12676 14681 16686 18691 20696 22702 24707 26712 28717 30722 32728 34733 36738 38743 40748 42754 44759 46764 48789 50774 52779 54785 56790 58795 • 18 • 33 31 • • i g • 12 • 42 • 28 +29 +37 --I 1 1 1 1 -0.4483 -0.1981 -0.5733 -0.3232 . » , , . . -0.0731 0.0520 0.3021 0.5522 0.1771 0.4272 55 Table 12. KIT plant neighbours - PCA axes relationships. Correlation coefficients relating the presence/absence of plant neighbours to PCA axes I-III. Axis I II III % variation 19.67 18.63 11.83 species Sambucus racemosa -.3970 .3668 -.1649 Galium triflorum -.1548 .5626 -.5009 Osmorhiza chilensis -.1832 -.3430 -.7068 Epilobium watsonii -.4622 -.2936 .2127 Agrostis papillaris (tenuis) -.3195 .4803 .0899 Picea sitchensis -.0248 -.5994 .3235 Epilobium ancrustifolium -.1238 .1974 -.4163 Bromus vulgaris .1421 .5891 -.4879 Oplopanax horridus .6937 -.3376 -.2420 Streptopus roseus .5574 -.4177 -.2635 Athyrium filix-femina .1424 .6429 .4820 Gymnocarpium drvopteris .6788 -.0104 .2274 Equisetum arvense .2461 .7097 .0820 Tiarella trtfoliata .6685 -.0115 -.0618 Tiarella unifoliata .7725 .2604 .0660 " Tellima crandiflora .0456 .3212 .0132 56 Only axis I of the WED PCA for the presence/absence of plant neighbours, can be fully interpreted. Axis III has no strong negative correlations while the species both negatively and positively correlated with axis II (Table 13) commonly co-occur and do not differ ecologically (Klinka et al. 1989). Axis I is positively correlated with Galium triflorum and Alnus rubra and negatively correlated with Epilobium anqustifotium (Table 13) and may represent a moisture gradient. Both Cornus canadensis and Epilobium angustifolium are only found in open, drier areas of the site and Equisetum arvense and Alnus rubra are commonly found in areas of standing water. Alnus is also an indicator of a fluctuating water table (Klinka et al. 1989). Table 13. WED plant neighbours - PCA axes relationships. Correlation coefficients relating the presence/absence of plant neighbours to PCA axes I-III. Axis l II III % variation 19.84 15.20 13.00 species Sambucus racemosa .4287 .0547 -.2309 Galium triflorum .6096 .1469 .4766 Aorostis captllaris (tenuis) .2565 .1690 .3811 Picea sitchensis -.1399 -.4197 -.2917 Epilobium angustifolium -.7787 .2349 -.1067 Dryopertis assimilis -.3990 .4933 .4947 Oplopanax horridus -.0814 -.6145 .1860 Streptopus roseus -.1938 -.4485 .6517 Athyrium filix-femina .4027 .2207 -.1371 Gymnocarpium dryopteris .2007 -.6562 .1813 Equjsetum arvense .5369 .2915 .4892 Tiarella trifoliate .2378 .6460 .0546 57 Table 13 continued -.3852 -.0143 -.0494 Tsuaa heterophvlla Clintonia uniflora .1106 -.5627 .5963 Alnus rubra .7621 .1135 -.1649 Vaccinium ovalifolium -.4758 .4834 .2704 Cornus canadensis -.5467 .2591 .4352 Although the first three axes of the devil's club presence/absence data summarize over 55 percent of the total variation in the data set, only axis I can be ascribed with any underlying pattern (Table 14). Axes II and III have no strong negative correlations with any species (Table 14). Axis I may be a moisture gradient from damp to very moist to wet conditions. Rubus spectabilis. Maianthemum dilatatum and Cornus sericea are all indicators of very moist to wet sites with either fluctuating groundwater or flood conditions (Klinka et al. 1989). The axis is negatively correlated with Gymnocarpium dryopteris (Table 14), which is indicative of fresh to very moist sites, as used by Klinka et al. (1989). Most quacrats are widely separated on axes I and II (Figure 9). Like the graph of the soils scores (Figure 7), quadrats 60, 61 and 63 are at the wet end of the PCA axis and quadrats 46 and 47 are at the negative crier end (Figures 7 and 9). The first axes of these two PCAs are not equivalent, however, as relationships between all quacrats on both ojaphs are not identical (Figures 7 and 9). 58 Table 14. DC plant neighbours - PCA axes relationships. Correlation coefficients relating the presence/absence of plant neighbours to PCA axes I-III. Axis I II III % variation 27.23 18.13 10.16 species Rubus spectabilis .7378 -.4237 .4150 Galium triflorum .4870 .6877 -.1108 Osmorhiza chilensis -.0415 .7604 .0406 Acrostis capillaris (tenuis) .6314 .2450 -1358 Dryopteris assimilis -.4705 .3535 .5595 Actaea rubra .4376 .0990 -.1886 Gymnocarpium oryopteris -.7630 -.2359 -.1837 Streptopus roseus -.2106 .2546 -.0111 Tiarella unifoliata .6930 -.4262 -.1148 Viola glabella -.1866 .2869 .0285 Maianthemum dilatatum .6776 .2762 .2235 Clintonia uniflora -.4406 .0767 -.1213 Circaea alpina .4320 -.0638 .5034 Streptopus amplexifolius -.0090 .1340 .6875 Matteuccia struthiopteris .4514 .7035 -.2773 Cornus serjcea .5788 -.4367 -.2479 59 Figure 9. Graph of axes I and II of the presence/absence of plant neighbors PCA for DC showing individual quacrats. 0. 0. 0. 0. 0. 0. 0 0 0. 0. 0 0 0. 0. 0. 0. 0. 0 0 0. 0. 0 0 0 0 0 0 0 0 0 0 -0 -0 -0 -0 •0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 •0 -0 -0 -0 -0 -0 -0 -0 35023 33885 32708 3155) 30393 29236 28079 26922 25764 24607 23450 22293 21135 19978 18821 17864 16506 15349 14192 13035 11878 10720 95630E-01 8405SE-01 72486E-01 60913E-01 49341E-01 37768E-01 261B8E-01 14824E-01 30S12E-02 85212E-02 20094E-01 31668E-01 43238E-01 54811E-01 66383E-0I 77956E-0! 89528E-01 10110 11267 12425 13582 14739 15896 17053 18211 19368 20525 21682 22840 23997 .26154 26311 27469 .28626 •62 •63 .81 '43 -0.0365 AXIS 1 60 6.1.4 PLANT NEIGHBOURS-FREQUENCY Like the presence/absence PCA fa the TB data, the PCA axes fa the frequency data reflect disturbance. Axis I of the TB PCA of the frequency of plant neighbours data, is positively correlated with Sambucus racemosa and Epilobium watsonii and negatively correlated with Tiarella unrfoliata. Clintonia unrflora and Conus canadensis (Table 15) and suggests a disturbance gradient. Axis II may also be a disturbance gradient. Epilobium angustifolium and Cornus canadensis are positively correlated with the axis (Table 15) and are indicative of disturbance and poor nutrient conditions respectively (Klinka et al. 1989). The axis is negatively correlated with Streptopus roseus. and Tiarella trifoliate (Table 15), which are characteristic of rich, moist, somewhat stable areas (Klinka et al. 1989). As axis III has no positive correlations (Table 15), an ecological gradient was not ascribed to the axis. The graph of axes I and II scores corroborates the interpretation that axis I is a disturbance gradient and may be due to girdling (Figure 10). Most quadrats, numbered from 1 to 21 at the KIT a ungjdled alder site, have positive scores and most quadrats, numbered from 22 to 42 at the WED a ungjdled alder site, have negative scores. Although axis II was also interpreted as a disturbance gradient, it is not a gradient due to girdling. Both KIT and WED quadrats have positive and negative scores on the axis (Figure 10). Table 15. TB plant neighbours - PCA axes relationships. Correlation coefficients relating the frequency of plant neighbours to PCA axes I III. Axis I II III % variation 12.21 8.80 7.91 species Rubus spectabilis .4160 -.0378 -.2049 Sambucus racemosa .5355 -.3650 .2805 Galium triflorum .2516 .0930 -.5276 Stachvs mexicana .3035 .2529 -.1424" Osmorhiza chilensis .3621 -.3255 .2141 61 Table 15 continued Cinna latifolia .2682 .0118 .1410 Epilobium watsonii .5148 -.0612 .0784 Acrostis capillaris (tenuis) .4727 .1477 -.4046 Picea sitchensis -.3167 -.0215 .1197 Aruncus dioicus .1513 -.4686 -.0590 Epilobium anqustifolium .0755 .6146 .1801 Chamaecyparis nootkatensis .2411 -.0454 -.0804 Ribes bracteesum -3318 .2678 .0453 Dryopteris assimilis -.4366 .0319 .1031 Bromus vulaaris .3106 .1818 -.0924 Salix sitchensis .4977 .1439 .0065 Oplopanax horridus -.2871 -.3843 .3111 Streptopus roseus -.3703 -.5361 .2861 Athyrium filix-femina -.0591 -.0671 -.4242 Gymnocarpium drvoDteris -.4391 -.4823 .0907 Equisetum arvense -.2537 -.1628 -.7616 Smilacina racemosa .0678 .0187 .2891 Tiarella trifoliata -.1572 -.5540 .0706 Tiarella unifoliata -6904 -.2777 -.2931 Viola glabella .2170 -.4121 -.0419 Tellima crandiflora .3297 -.1399 -.0029 Tsuga heterophylla -.3330 .4145 .1135 Maianthemum dilatatum -.3267 .1631 .4595 Clintonia uniflora -.5065 -.0540 .0890 Circaea alpina -.1981 -.0642 -.3927 62 Table 15 continued Thelypteris pheqopteris -.2368 .0071 -.3634 Ribes laxiflorum -.1527 .3487 .4555 Cornus sericea .0024 .2939 -.1479 Alnus rubra -.2634 -.2741 -.4263 Vaccinium ovalifolium -.3666 .2743 -.2424 Cornus canadensis -.5536 .5168 -.0645 Figure 10. Graph of axes I and II of the frequency of plant neighbors PCA for TB showing individual quadrats. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. -0. -0. -0. -0. <* -0. oi -0. S-o. < -0. -0. -0. -0. -0. -0. -0. -0. -0. -0. -0 -0 -0 -0 -0 -0. -0 -0 -0. -0. -0 -0 -0. -0. -0 -0 56028 53530 51032 48534 46036 43638 41040 38542 36044 33546 31048 28550 26052 23554 21056 18557 16059 13561 11063 85654E-01 60674E-01 35693E-01 10713E-01 14267E-01 39247E-01 64228E-01 89208E-01 11419 13917 16415 18913 21411 23909 28407 28905 31403 33901 36399 38897 41395 43893 46391 48889 51387 53885 56383 58881 61379 63877 66375 68873 71371 73869 76367 78865 81363 • 33 •42 • 40 • 37 • 17 OO 27.28 18^ 31 • 38 • 6 -0.7354 -0.4456 -0.1558 0.1339 0.4237 0.7135 -0.5905 -0.3007 A X I S , -0.0109 0.2788 0.5886 63 Axis I of the KIT frequency PCA may, like axis I of the TB frequency data, represent a disturbance gradient. Streptopus roseus. devil's club, Gymnocarpium dryopteris. Tiarella trifoliata and Tiarella unifoliata. which are negatively correlated with the axis (Table 16), tend to occur in rich, moist, less disturbed areas (Klinka et al. 1969). Acrostis capillaris (tenuis) is generally found at open, more disturbed quacrats at KIT. Axis II is positively correlated with Galium triflorum. Epilobium anaustifolium and Bromus vulgaris and negatively correlated with Sambucus racemosa (Table 16) and may represent a ligjit gradient of illumination. Bromus is found in quacrats with less cover at this site and Epilobium is shade intolerant while Sambucus is known to be shade tolerant (Klinka et al. 1989). Axis III has no strong negative correlations (Table 16) and was therefore not interpreted. Table 16. KIT plant neighbours - PCA axes relationships. Correlation coefficients relating the frequency of plant neighbours to PCA axes scores I-III. Axis I II III % variation 31.01 14.33 12.39 species Sambucus racemosa .0669 -.5211 .5119 Galium triflorum .4897 .6031 .2689 Osmorhiza chilensis -.0506 .0098 -.2953 Epilobium watsonii .2904 -.4081 -.2893 " Acrostis capillaris (tenuis) .7158 .1431 .1094 Picea sitchensis -.3338 -.0026 -.1899 Epilobium anaustifolium .4121 .6946 .4461 Bromus vulgaris .2747 .5470 .4804 Oplopanax horridus -.6557 .4664 -.3577 Streptopus roseus -.7550 .0793 -.1237 Table 16 continued 64 Athyrium filix-femina -.0155 -.4442 .5610 Gvmnocarpium dryopteris -.7266 .1230 .1250 Equisetum arvense -.1027 -.1935 .4544 Tiarella trifoliata -.7690 .0083 .4185 Tiarella unifoliata -.8198 .1149 .3241 Tellima grandiflora .2299 -.4407 -.1264 Axis I,of the WED data set, is a moisture gradient. The axis is negatively correlated with Epilobium angustifolium (Table 17), which is characteristic of open moist, disturbed sites and positively correlated wrth Alnus rubra. Equisetum arvense and Galium triflorum (Table 17). Equisetum arvense and Alnus rubra co-occur in areas of standing water at the WED site. Alnus rubra also shades and cools the site reducing surface evaporation and is an indicator of a fluctuating water table (Klinka et al. 1989). Biological significance cannot be readily ascribed to axis II. The axis is positively correlated with devil's dub, Streptopus roseus. Gymnocarpium dryopteris and Clintonia unrflora and negatively correlated with Athvrium filix-femina (Table 17). All these spedes co-occur. Klinka et al. (1989) also states that Gymnocarpium dryopteris and Streptopus roseus are indicative of damp, very moist soils, devil's dub and Athyrium filix-femina are indicative of very moist to wet soils, and Clintonia uniflora is indicative of moderately dry to damp soils. Axis III may be a light gradient of illumination. It is positively correlated with Aorostis tenuis (tenuis) and Epilobium angustifolium (Table 17), a shade intolerant spedes (Klinka et al. 1989) and negatively correlated with Sambucus racemosa (Table 17), a shade tolerant spedes (Klinka et al. 1989). 65 Table 17. WED plant neighbour - PCA axes relationships. Correlation coefficients relating the frequency of plant neighbours to PCA axes scores I-III. Axis I II III % variation 24.50 15.91 13.43 species Sambucus racemosa .0686 -.0742 -.8424 Galium triflorum .5780 -.1158 .2318 Acrostis capillaris (tenuis) .4459 -.3469 .6402 Picea sitchensis -.1382 .3635 .0003 Epilobium anaustifolium -.7949 -.0125 .5105 Drvopertis assimilis -.1055 -.0024 -.0629 Oplopanax horridus .3095 .6046 -.0884 Streptopus roseus .2356 .6007 .1514 Athvrium filix-femina .2786 -.6690 -.3210 Gymnocarpium crvopteris .3697 .8120 .1419 Equisetum arvense .7074 -.2385 .4144 Tiarella trifoliata .1820 -.4410 -.0250 Tsuaa heterophvlla -.2277 .1532 -.3255 Clintonia uniflora .2526 .5182 -.1290 Alnus rubra .7595 .1162 .1152 Vaccinium ovalifolium -.0763 -.1215 .1048 Cornus canadensis -.4103 .0032 .1977 The first three axes of the devil's club frequency of plant neighbours data represent almost sixty-five percent of the total variance in the data (Table 18). Axis I is a moisture gradient which is positively correlated with Gymnocarpium dryopteris (Table 18), an indicator of damp to very moist conditions (Klinka et al.1989). Galium triflorum. Actaea rubra. Maianthemum 66 dilatatum. Cornus sericea and Acrostis capillaris (tenuis) are negatively correlated with this axis (Table 18). Both Cornus sericea and Maianthemum dilatatum are tolerant of a fluctuating water table or flood conditions and indicate very moist to wet conditions (Klinka et al. 1989). Axis II is strongly correlated with Dryopteris assimilis (Table 18), an indicator of rich damp to very moist sites (Coupe et al. 1982, Klinka et al. 1989). Circaea alpina is aso correlated with the axis (Table 18) and is found on moist, nutrient medium sites (Klinka et al. 1989). As this axis has no negative correlations greater than .5, it was not interpreted further (Table 18). Axis III has positive correlations with Osmorhiza chilensis and Smilacina stellata and is negatively correlated with Rubus spectabilis (Table 18). These species, however, are ecologically similar (Klinka et al. 1989) and a biological interpretation was not ascribed to the axis. Table 18. DC plant neighbour - PCA axes relationships. Correlation coefficients relating the frequency of plant neighbours to PCA axes I -III. Axis I II III % variation 36.36 14.68 13.43 species Rubus spectabilis -.3625 -.1487 -.7717 Galium triflorum -.7618 .0906 .0848 Osmorhiza chilensis -.2362 .2648 .5465 Acrostis capillaris (tenuis) -.7527 -.2176 -.3105 Dryopteris assimilis .2197 .9365 .0020 Actaea rubra -.6274 -.1904 .4649 Gymnocarpium cryopteris .8485 -.2115 .0275 Smilacina stellata -.0805 -.0302 .5514 Tiarella unifoliata -.2662 -.1623 -.4137 Viola dabella -.2093 .0176 .3972 Tsuaa heterophylla .0705 .3043 .0656 Maianthemum dilatatum -.8504 -.1068 -.2139 67 Table i s continued Clintonia uniflcra .2543 -.1435 .2580 Circaea ajpina -.2004 .6742 -.3596 Streptopus amplexrfolius .1202 .3608 .0103 Matteuccia struthiopteris .1949 -.0015 .0417 Cornus sericea -.7692 .0764 .4569 Like the DC presence/absence graph (Figure 9), quadrats 61 to 63 and quadrats 46 and 47 are at opposite extremes of the moisture gradient on the graph of the axes scores (Figure 11). The quadrat relationships shown in the graph, however, do not suggest an ecological interpretation of axis II. Figure 11. Graph of axes I and II of the frequency of plant neighbors PCA fa DC showing individual quadrats. 0.87137 0.84688 0.82239 0.79790 0.77341 0.74892 0.72443 0.69995 0.67546 0.65097 0.62648 0.60199 0.57750 0.55301 0.52852 0.50403 0.47955 0.45506 0.43057 0.40608 0.38159 0.35710 0.33261 0.30812 0.28364 0.25915 0.23468 0.21017 v ^v D A T E PHENOLOGICAL CODE PHENOLOGICAL CODE _L CO 01 -vl CD -x 83 Figure 15 continued d) WED-1987 LU Q O o _l < o CD O _ i o z LU X 0_ jP-&>^ A ^ A * ^ - A f p ' Unlike thimbleberry, the slopes of regression lines of devil's dub vegetative phenology differed between years (Appendix II) and phenology may have initially proceeded more rapidly in 1986 than 1987 (Figure 16). Devil's dub stems and branches were not vegetatively different (Appendix II). Figure 16. Vegetative phenological codes fa devil's dub in 1987 and 1987. The number within the crdes represent the number of quadrats with the code on the y-axis at the sampling time. a) DC-1986 LU Q O O _ i < o o o _ i o LU X CL jfr x 4 5 0 D A Y S " • 1 0 5 - 4 5 0 D A Y S • 5 5 - 1 0 5 D A Y S J • 0 - 5 5 D A Y S Q U A D R A T Figure 20 continued 102 b) WED 1987 100 oc LU CD Z) z LU h-00 8 0 6 0 4 0 2 0 c) KIT 1986 or LU CO 2 LU 00 d) WED 1986 100 cc LU CO z> z LU r-00 8 0 6 0 4 0 2 0 0 1 m > 4 5 0 D A Y S • 1 0 5 - 4 5 0 D A Y S • 5 5 - 1 0 5 D A Y S • 0 -55 D A Y S 22 24 2 6 2 8 3 0 32 34 Q U A D R A T 36 3 8 4 0 42 1 0 0 8 0 6 0 4 0 2 0 0 : 8 ti MI M m r-i 8 : •I 5 E 1 3 5 7 9 11 13 15 17 19 21 Q U A D R A T n m M I 22 24 2 6 2 8 3 0 3 2 3 4 3 6 3 8 4 0 4 2 Q U A D R A T S > 8 1 0 g 4 6 6 - 8 1 0 | 4 1 6 - 4 6 5 | 3 6 1 - 4 1 5 1 1 0 6 - 3 6 0 gj 5 6 - 1 0 5 | 0 - 5 5 S > 8 1 0 I 4 6 6 - 8 1 0 gj 4 1 6 - 4 6 5 | 3 6 1 - 4 1 5 I 1 0 6 - 3 6 0 i 5 6 - 1 0 5 I 0 -55 103 In the devil's club site, only six stems were produced in 1987 and only one stem was still alive by August. This large rate of early mortality and the graph of the age distribution (Figure 21) suggest a Deevey Type III curve (Hutchings 1976). A type III distribution implies that youngest plants have the greatest risks of mortality. This risk of mortality declines with age so that some very old individuals will be found in the population (Hutchings 1976). A type III curve is common for longjived species (Balogh and Grigal 1988). The cumulative distribution, however, shows stem ages at one time only (Figure 21) and is not identical to a Deevey survivorship curve which shows the change in cohort numbers through time (Hutchings 1976). Devil's dub populations do not necessarily have the same age distribution at all times as factors determining longevity may change. It was also impossible to determine the age of this longjived population accurately and it may be hundreds of years old. Indeed aging stems by counting terminal bud scars probably underestimates each stem by at least five years as these scars are obliterated by stem thickening. Figure 21. Cumulative number of devil's dub stems older than the time indicated on the x-axis. 150 120 90 -cn LU h-00 LL O CC 60 LU CQ W 30 'Si 10 15 20 YEARS 25 30 0 104 in comparison with tnimbieberry, the aevirs CIUD population consists or a mixture or ages (Figure 22). The largest overall proportion of stems are less than five years old indicating that although the risks of mortality are high early in life sufficient numbers are produced to ensure that many stems reach reproductive maturity. Few stems were observed to be greater than 22 years old, indicating the maximum time period before stems become stdoniferous and fragment. Figure 22. Percentage of devils' club stems in distinct age classes on June 30,1987. CC LU CO LU \— 00 100 80 60 40 h 20 0 I / / / / / / / I I E3 2 2 - 2 8 H 16-21 • 11 -15 • 6 - 1 0 • 1 - 5 43 45 47 49 51 53 55 57 59 61 63 Q U A D R A T There appears to be little relationship between the density of thimbleberry at each quadrat and the percentage of stems surviving to each age dass. As density increases a concomitant change in the longevity at each quadrat does not occur (Table 34). In most cases the slope of the regression is not significantly different from zero (Table 34), indicating no relationship. In those cases where the regression is significant (1986 - all periods from 106 -810 days) (Table 34), the variation in the data accounted fa by the regression is less than 30 percent. 105 Table 34. Regression analysis of thimbleberry stem longevity and stem density. Y=a+bx where y = percentage of the total number of stems in quadrat surviving to the time period, a = intercept, b = slope and x = total number of stems in quadrat. r.P<05) Time period (days) r-square slope significance (1986) 0-55 .07 -.014 .10 (1986) 55-105 .07 -.014 .10 (1986) 105-360 .14 -.033 .01* (1986) 361-415 .26 -.015 .00* (1986) 416-465 .26 -.052 .00* (1986) 466-810 .10 -.033 .04* (1987) 0-55 .06 -.013 .10 (1987) 56-105 .03 -.003 .26 (1987) 106-450 .03 -.004 .25 The regression of the proportion of the total number of devil's dub stems in each dass against the total number of stems in the quadat suggests that density has little or no effect on the proportion of stems of each age dass observed (Table 35). As the density of stems increases the proportion of stems in each age dass does not also increase (Table 35). Table 35. Regression analysis of devil's dub stem age and density. Y=a+bx where Y = ratio of number of stems in age dass to total number of stems in quadat, a = intercept, b = slope and x = total number of stems in quadat. (*, p<.05) Age dass (years) r-square slope significance 1-5 .33 .035 .00* 6-10 .12 -.019 .12 11-15 .02 -.004 .60 16-21 .01 -.003 .61 22-28 .05 -.011 .32 106 6.4.2 PRINCIPAL COMPONENT ANALYSES The PCA of the number of thimbleberry stems alive in each age class in 1986 and 1987 was used to generate axes scores for use in ANOVA only. The PCA summarized over 99 percent of the variation in the data on the first three axes, (Table 36). No biological interpretation was ascribed to these axes. Table 36. Eigenvalues of 1986 and 1987 PCA for the number of stems surviving to each age class. Axis I Axis II Axis III Eigenvalues 66.90 31.99 .73 The PCA of the number of stems dying in the age classes in 1986 and 1987 was used in subsequent CANCORR analysis, though only 54.81 percent of the variation in the total data were explained by the first three axes (Table 37). Axis I reiterates the relationship observed graphically and groups together those quadrats with the shortest living stems in 1986 and longer lived stems in 1987, all of which are vegetative and not sexually reproductive (Table 37). Axis II implies that quacrats with the longest living stems in 1986 produce shorter living stems in 1987 (Table 37). Axis I and II suggest at tradeoff between younger vegetative and older generative stems and indicate that the plant may alternately channel resources~between vegetative and generative stem production. Axis III has no strong positive correlations and is negatively correlated with stem deaths during the period from July to August, 1987 (1986 416-465, 1987 56-105 days), the period where least deaths occur whether these stems were produced in 1986 and 1987 (Table 37). 107 Table 37. Demography - PCA axes for TB data for the number of stems dying during each age class. Correlation coefficients relating demographic data to PCA axis I - III. Axis I II III % variation 22.59 18.75 13.47 demographic variables (days) (1986) 0-55 .7844 -.1013 .1101 (1986) 106-360 -.0299 -.1217 .3179 (1986) 361-415 .7699 -.0014 .0117 (1986) 416-465 .1497 .1669 -.7688 (1986) 466-810 -.0005 .7069 -.7260 (1986) >810 .5506 -.3978 .1878 (1987) 0-55 .2144 .7417 .2930 (1987) 56-105 .1659 .1929 -.6575 (1987) 106-450 .0202 .7551 .2875 (1987) >450 .8065 .0806 -.0419 PCA of the KIT data represents approximately 60 percent of the variation on the first three axes (Table 38). Axis I of the KIT PCA, like axis I of the TB PCA, groups together those stems dying early in 1966 and 1987, before they can flower and those stems living longest, perhaps because they have also not yet flowered. Axis II is correlated with those stems produced in 1986 and 1987 which die by the end of the summer of 1987 (1986 416-465 days and 1987 56-105 days), the period when stem mortality is at a minimum (Table 38). Axis III is negatively correlated with the stems produced in May, 1986 which die by August 1987 (416-465 days) and positively correlated with those stems from 1987 which die by the end of June, 1987 (0-55) days (Table 38). These opposite correlations suggest that reproductive and vegetative 108 stems are not produced at the same time, assuming two year old stems which die by the end of the summer were reproductive. Table 38. Demography - PCA axes for KIT data for the number of stems dying during each age class. Correlation coefficients relating demographic data to PCA axes I - III. Axis I II III %variation 29.34 15.46 13.86 Demographic variables (days) (1986) 0-55 .8451 .0327 .2632 (1986) 106-360 -.2022 -.3099 .1540 (1986) 361-415 .8491 .2917 .1522 (1986) 416-465 .0709 .5411 -.6937 (1986) 466-810 -.4316 .3863 .0465 (1986) >810 .6313 -.3334 -.2176 (1987) 0-55 .0474 .4642 .6203 (1987) 56-105 .2247 .6464 -.3244 (1987) 106-450 -.2908 .4180 .4921 (1987) >450 .8547 -.0525 .0851 The PCA of the WED summarizes 58 percent of the total variation in the data on the first three axes (Table 39). It is most strongly correlated with the stems from 1986 which die after August,1987 (466-810 days) perhaps after flowering and producing fruit, and those vegetative stems produced in 1987 which die within two months of their emergence or after June,1987 (0-55 days)(Table 39). Axis I is also correlated with those stems produced in 1987 which live past the first summer (106-450 and >450 days). Axis II is correlated with stems produced in spring, 1986, which die in all age categories before July, 1987, or before they can produce fruit. The stems from spring. 1987. are most strongly correlated with axis III. The stems which die during the time period from August, 1987, to August, 1988, (106-450 days) may have died between 1 0 9 August and June, 1980, and can De assumed to De vegetative as Truit did not ripen until mid-July in 1987. Those stems dying from June to August, 1987, were most probably reproductive as stems which produce fruit common died shortly afterwards. This axis may therefore be a continuum showing the differing longevities of vegetative to reproductive stems. Table 39. Demographic - PCA axes for WED data for the number of stems dying during each age class. Correlation coefficients relating demographic data to PCA axes I -III. Axis I II III %variation 24.24 19.36 14.82 Demographic variables (days) (1986) 0-55 -.1990 .6001 -.0732 (1986) 106-360 -.3150 .6035 .1233 (1986) 361-415 .2046 .8998 .0729 (1986) 416-465 .2634 .1967 .3081 (1986) 466-810 .8283 -.0041 .1637 (1986) >810 -.2773 .4813 -.0848 (1987) 0-55 .7542 .0723 -.4665 (1987) 56-105 .2234 -.2414 .6754 (1987) 106-450 .6532 .0977 -.5737 (1987) >450 .6048 .2414 .5701_ The devil's club PCA of the number of stems of each age category recorded in quadrats on June 30,1987, explains 80 percent of the total variation on the first three axes (Table 40). Axis I is negatively correlated with the oldest class, those stems aged from 22 to 28 years (Table 40). Stems greater than 16 years old tend to be stoloniferous and those younger than 10 years old are not reproductive. Axis II may, therefore, be a vegetative gradient for those two vegetative states. Axis III may represent the differences between vegetative and generative 110 stems, certainly that is the most noticeable difference between these two age classes correlated with the axis. Table 40. Demography - PCA axes for DC data for the number of stems in each age class. Correlation coefficients relating demographic data to PCA axes I - III. Axis I II III %variation 31.47 30.92 18.33 Demographic variables (years) 0-5 -4856 .7846 .1465 6-10 .3962 .4771 -.6568 11-15 . 5427 .3919 .6739 16-21 .1104 .7329 -.0935 22-28 -.9349 .1087 .0257 I l l 0.4.3 PARTITIONING OF VARIATION The ANOVA results for thimbleberry indicate that a substantial proportion of the total variation is accounted for by annual differences (Table 41). This agrees with trends suggested by PCA which indicated that vegetative and generative patterns were not maximized every year. The variation accounted fa by differences between sites is less than 10 percent and suggests a relatively small effect due to large scale environmental differences (Table 41). The genetic component in demographic hetaogeneity is also small as suggested by the proportion of variation accounted fa by among-quadrat diffaences (Table 41). Most variation is accounted fa by differences between stems within quadrats (Table 41), indicating that small scale environmental a developmental influences are an important component of longevity, and may be the major impetus determining age structure within populations. Although almost 20 pacent of the variation in devil's dub stem ages on June 30, 1987, is accounted fa among quadrats, most variation is within quadrats a between stems, suggesting that small scale environmental a developmental influences are a large component in population age structure (Table 41). Table 41. Demographic variation apportioned between years, sites and within and among quadrats by ANOVA. Thimbleberry Devil's dub Years 29.10 n/a Sites 8.27 Erra 62.63 Within and among quadrats quadrats erra 7.10 92.09 17.18 82.82 112 6.4.4 DEMOGRAPHIC-ENVIRONMENTAL RELATIONS Correlations between canopy cover and the longevity of thimbleberry stems are all less than .50 and indicate little relationship (Table 42). Table 42. Canopy cover - demographic relationships. Correlations between the number of stems dying in each time period and the canopy cover in 1986 and 1987 for TB, KITandWED.(*,p<.05). Time period KIT WED TB (days) 1986 1987 1986 1987 1986 1987 (1986) 0-55 .11 .19 -.11 106-360 -.05 .23 -.05 361-415 -08 -.21 .15 .27 .14 .15 416-465 -.20 -.05 .26 .21 -.01 .01 466-810 -.22 .07 .45* .45* -.13 .04 >810 .11 .10 -.35 -.06 -.23 -.10 (1987) 0-55 -.41 .21 "-.18 56-105 -.28 -.26 -.27 106-450 -.28 .30 -.06 >450 .07 .23 -.11 113 canopy cover also seems to nave little effect on the longevity or devils cluD stems, as correlations are small (Table 43). Table 43. Canopy cover - age distribution relationships. Correlation between the age distribution of stems and the canopy cover in 1987 for devil's club. (*,p<.05). correlation .18 .06 .00 -.07 -.01 CANCORR suggests little or no relationship exists between all the demography and environmental matrices (Table 44). Although the canonical correlation coefficients are frequently large, redundancies often do not support any overlap between point swarms or correlation structure is not conclusive (Appendix Illl). CANCORR also implies a relationship between the WED demography-soils and the WED and KIT demography-frequency of plant neighbours matrices, but WMDS refutes these relationships (Table 44)(Appendices III and IV). A relationship is also indicated between the phenology and the presence/absence of plant neighbours at KIT; however, the second PCA axis which is correlated with the first canonical axis (Appendix III) could not be interpreted (Table 12). age class (years) 1-5 6-10 11-15 16-21 22-28 114 Table 44. Summary of relationships between TB, KIT, WED and DC demographic variables and environmental characters. (+) relationship occurs, (-) no relationship, (?) possible relationship indicated by CANCORR. PLANT NEIGHBOURS SOILS FREQUENCY PRESENCE/ABSENCE CANCORR WMDS CANCORR WMDS CANCORR WMDS TB -KIT ? - ? WED ? - ? -115 6.5 MORPHOLOGY-PHENOLOGY-DEMOGRAPHYINTERACTIONS Morphology and phenology appear to be related at each level in the nested hierarchy for both thimbleberry and devil's dub, as suggested by CANCORR (Table 45). CANCORR also suggests a relationship between demography and morphology for both devil's club and the thimbleberry data set (Table 45), though WMDS refutes these rdationships at KIT and WED (Appendix IV). In no case, though, is phenology related to demography (Table 45), suggesting that longevity has little relationship to the rate of development. Table 45. Summary of relationships between TB, KIT, WED and DC morphological, phendogical and demographic variables. (+) relationship occurs, (-) no relationship. IB-MORPHOLOGY PHENOLOGY KJI-MORPHOLOGY PHENOLOGY WED-MORPHOLOGY PHENOLOGY DC-MORPHOLOGY PHENOLOGY PHENOLOGY CANCORR WMDS DEMOGRAPHY CANCORR WMDS CANCORR suggests that both vegetative and reproductive developmental rates in 1966 and vegetative developmental rates in 1967 are more rapid at those quacrats with larger stems with more, larger branches and leaves and more flowers for the TB data set. The first canonical 116 axes for the TB morphology-phenology CANCORR are correlated with the first morphological and phenological axes (Appendix III). Axis I of the morphological PCA data is correlated with all morphological variables except the number of stems per quadrat and the number of branch leaves (Table 20). The first phenological axis is correlated with the reproductive phenology in 1986 and 1987 and the vegetative phenology in 1986 (Table 27). For the KIT morphology-phenology data, CANCORR indicates that quadrats with more flowers have more rapid developmental rates. In those quadrats, where phenology does not proceed as rapidly, there are fewer flowers. The first canonical axis is correlated with the first phenological PCA axis and the third morphological PCA axis (Appendix III). The first phenological PCA axis is correlated with all the phenological variables (Table 28), whereas the third morphological PCA axis is a reproductive axis and is correlated with the number of flowers per inflorescence and per quadrat and the number of two-year-old stems per quadrat (Table 21). As those quadrats with no flowers have no change in reproductive phenological code, the slope of the regression line between the phenological code and time is zero (Table 5). Phenological summary slopes would be larger than zero at those quadrats with flowers, regardless of how rapidly they matured. Only two-year-old stems produced flowers and they also had the most rapid rate of vegetative development. Quadrats with most two-year-old stems would therefore be expected to have greatest overall developmental rates or slopes of the regression. At WED, the CANCORR indicates a relationship between phenology and vegetative and reproductive morphological features. Those quadrats with more flowers and larger amounts of vegetative tissue have most rapid developmental rates and at those quadrats where there are fewer flowers and smaller vegetative parts, the developmental rates are less rapid. The first canonical axis pair is strongly positively correlated with the first morphological and phenological PCA axes (Appendix III). The first morphological PCA axis is, like axis I of the TB data, a gradient of vegetative tissue size and is also less strongly correlated with the number of flowering stems and the number of flowers per inflorescence (Table 22). The first phenological axis is correlated with all the phenological variables, especially the 1987 phenological variables (Table 29). 117 CANCORR Deiween tne TD morpnoiogy-demograpny matrices indicates a reiationsnip between the number of stems in each quadrat and the number of stems in most demographic time periods. The first morphological canonical axis is correlated with the second morphological PCA axis (Appendix III), a gradient of the number of stems per quadrat (Table 20). The first demographic canonical axis is correlated with the first and second PCA axes (Appendix III). These PCA axes are correlated with most stem age classes, except those from July to August in either 1986 or 1987, and from August, 1986, to May, 1987, (Table 37), the time periods when fewest stems die(Figure 20). As the number of stems dying is related to the number of stems in each quadrat, this relationship is not surprising though perhaps it does attest to the validity of the method. It also reiterates the regression analysis between each demographic age dass and the stem density (Table 34). The regression indcates that as the number of stems per quadat increases more stems de but the proportion dying in any one age dass does not change (Table 34). According to CANCORR (Appendix III), as stem number per quadat increases, no stem age class emerges when stems de. Phendogy is unrelated to demography (Table 30), suggesting that the rate of annual periodic development is uncorrelated to the longevity of stems and may also be subject to dfferent environmental contraints. As suggested for morphology and demography, a relationship between longevity and reproductive phendogy may be expeded as stems usually de after flowering and flowering stems tend to be oldest in the population. Regardless, CANCORR does not strongly indicate a relationship (Appendx III). The type II survivorship curve for thimblebery also suggests that the risk of mortality is relatively constant through time for each cohort (Figure 37). Developmental rates increase in devil's dub quadats where there are more flowers and stems are larger with more vegetative tissue. CANCORR further indcates that plants with smaller stems and branches with fewer leaves may have slower developmental rates. The first canonical axes are correlated with the first morphdogical and phendogical PCA axes (Appendx III). The first morphdogical PCA axis is correlated with stem sizes and the number and size of branches and the number of branch leaves and the number of flowers per quadat 118 (Table 23). The first phenological axis is correlated with the vegetative and reproductive phenological codes in 1986 and 1987 (Table 30). As with the thimbleberry data, quadrats lacking flowers generate a zero slope from the regression of phenological slope against time (Table 5). Reproductive developmental rates will therefore be perceived as more rapid at those quadrats with flowers, regardless of the rate at which flowers and fruit mature. Although, larger devil's club stems mature more rapidly or larger stems may initiate growth earlier than smaller stems, no correlation is apparent for the stem length increment and the rate of development. Branches, though, increase more in length at quadrats with more rapid developmental rates and larger stems (Appendix III). As branches represent the next generation of stems and the means of clonal expansion, their increased growth may indicate that conditions are most favorable for vegetative reproduction at sites where developmental rates are maximized. CANCORR also indicates a relationship between devil's dub morphology and demography. The first canonical axis is correlated with the second morphological and demographic PCA axes (Appendix III). The second morphological axis is negatively correlated with the number of branches per stems and the stem length and positively correlated with the number of stems per quadrat (Table 23). The second demographic axis is correlated with those stems between 0 and 5 years of age and 16 and 21 years of age (Table 40). This correlation links together those stems, aged 16 to 21 years, which have become stoloniferous and along which branches have rooted to become independent stems estimated to be between 0 and 5 years old. Although the 16 to 21 year old stems would have the greatest overall stem lengths, nonetheless, the average stem length per quadrat would be low due to the large number of newer, shorter stems. 119 VII DISCUSSION 7.1 MORPHOLOGY The environmental characteristics of differing successional niches have been credited with maintaining morphological differences between some species (Lee et al. 1986). Morphological differences are evident between thimbleberry and devil's club with perhaps a palmate leaf shape and a clonal growth habit being the most obvious traits that the two species share. Differences in the successional niche occupied by each species are also apparent. Although thimbleberry is found in clearings in old growth forests with devil's dub, it is most abundant at earlier successional stages. This suggests that differing successional environments may correlate with morphdogical characters unique to devil's dub and thimbleberry. Lee et al. (1986) attributed variation in leaf size and shape and plant architecture between two spedes of Pdygonum to differing successional habitats. Thimbleberry morphologies are correlated with disturbance caused by girding and moisture, whereas devil's dub morphology is uncorrected with any other environmental variables selected (Table 26). Devil's dub may be buffered by large trees in the dd growth forest against fluctuations in dimate and, therefore, may be living in a stable environment. Though few studes have compared vegetative morphdogy between spedes in differing successional environments (Bierzychudek 1982), reproductive effort has frequently been linked to succession (Hancock and Prrtts 1987; Stearns 1977).-Harper (1970) indcates that spedes found earlier in succession devote larger proportions of energy to reprodudion than those species found later in succession. Thimbleberry and devil's dub suggest that those factors which determine reproductive allocation are complex. Both are vegetatively and sexually reproductive, perennial and iteroparous; as well thimbleberry rhizomatous growth is extensive and devil's dub roots are also large and deep (personal observation). However, their flowering responses dffer. Devil's dub doesn't flower in the first year and each stem may flower thereater, whereas thimbleberry stems produce fruits once in their second or third year of growth and then 120 die. This diversity of reproductive possibilities suggests that several differing factors may determine reproductive allocation and that reproductive allocation is a complex species-specific phenomena. Both species, however, fit into life history arguments which suggest that earlier successional species flower sooner after germination than later successional species (Stearns, 1977). Differences between the two thimbleberry populations may be due to environmental differences at each site (Jurik 1985; Van Cauteren and Lefebvre 1986; Maddox et al. 1989; Schwaegerie and Levin 1990). Some site differences are apparent, with WED having greater canopy cover (Figure 7), richer soils and showing less sign of flooding than at KIT (Table 6). In 1987, morphological patterns also differed between sites. KIT had greater numbers of flowers, fruits, leaves and stems per quadrat, longer stems and a larger proportion of fruit per flower produced than WED (Appendix I). Disturbance, due to moisture differences and alder girdling, was correlated with morphological differences between sites. These environmental characters were reflected differently between sites, though. At WED drier quadrats contained plants with more flowers, larger stems and more branches (Appendix III); whereas for the overall TB data set more disturbed, moist quadrats contained larger stems with more leaves and branches (Appendix III). At KIT, where disturbance was greater, moist sites produced more flowers and branches on larger stems; whereas at WED, where disturbance is less, moist sites produce fewer flowers and branches on smaller stems. Relationships between morphology and environment were also found to differ between sites by Schwaegerie and Levin (1990) and between and among sites by Gawler et al. (1987) for Phlox drummondii and Peducularis furbishiae respectively. Menges (1987) also found larger leaves and longer stems produced in canopy gaps for Laportea canadensis to differ between upland forests and floodplains. Most variation is accounted for within sites or populations, being greater than between sites for thimbleberry and between years for either thimbleberry or devil's dub. If, as assumed, each quadrat contains a single unique genet then it would appear that 121 morpnoiogy is strongly influenced Dy genetic differences between individuals and Dy small scale environmental heterogeneity. The number and size of leaves are mostly influenced by small scale environmental heterogeneity for both species (Table 24). Variation within other clonal populations for leaf size was also found to be substantial by Barnes (1986) and Herndon (1987). Leaf plasticity has also been observed with such examples as sun versus shade leaves (Schlichting 1986) or different nutrient, moisture and clipping treatments affecting the number and size of Phlox leaves (Schlichting 1989). Plasticity is common for leaf (Oberbauer and Strain 1986) and stem (Schlichting 1986; Piteika et al. 19865) morphology and may even be predictable in differing environments (Givnish 1982; Menges 1987). Morphological plasticity is also commonly observed in weedy species (Holzner 1982) and has been noted for thimbleberry (Hulten 1974). Most variation for stem length is apportioned within individuals for devil's club and equally divided within and between individuals for thimbleberry (Table 24). Thimbleberry stem length is influenced by both genetic differences and small scale environmental heterogeneity. Several factors may account for the differences observed. Devil's dub stems originate individually and are long-lived stoloniferous stems producing roots and vertical branches along their length. Thimbleberry stems are produced in dumps and remain erect. Each thimbleberry stem may therefore encounter less environmental variation than devil's dub, both spatially and temporally. The common origin of each thimbleberry is also assurance that each dump of stems within a quadrat is genetically identical. Devil's dub stems within a quadrat may not be genetically identical and may therefore also be influenced by a combination of genotypic and environmental heterogeneity. The large within-quadrat morphological term suggests that replicate environmental measures made within quadrats might have been informative. If small-scale environmental variability promotes plasticity within individuals as Levin (T986) suggests, this large within-quadrat term (Table 24), indicates that environmental 122 variation within quacrats may also be significant. Certainly soil characteristics are known to be highly variable, perhaps requiring up to 100 samples in a hectare to accurately estimate soils features (Courtin et al. 1983). Although the relative allocations to sexual and vegetative reproduction may be important to maintain or spread clonal plants under different environmental conditions (Pitelka et al. 1980; Abrahamson 1980; Loehle 1987), they are difficult to assess for thimbleberry and devil's club. In addition, both plants are long-lived and do not flower every year. Accounting for vegetative costs for years when no seed is produced is not straightforward. Although thimbleberry was found to increase sexual reproduction at KIT, where light levels are greater than WED, vegetative reproduction was also found to increase. Vegetative and sexual reproduction were also found to increase for Acalypha rhomboidea and Pilea pumila (Cid-Benevento 1987) with increasing light. Physiological integration may be an important strategy for both devil's club and thimbleberry, though devil's club does not maintain the same degree of interconnection between parts as does thimbleberry. Once devil's club stems become stoloniferous, branch and root at stem nodes, the stolon degenerates between nodes, though these connections may be several meters long before degradation occurs. Thimbleberry rhizomes tend to maintain connections between clumps of stems and appear to die at the tip of the growing rhizome (personal observation). Certainly the decree of rhizome branching, primary and secondary root formation and the distance between above ground stems is variable. Architectural differences such as these have been interpreted as a strategy to alleviate stress (Schmid and Bazazz 1987; Hartnett and Bazazz 1985) as stem ramets were found to be longer and fewer through areas of nutrient (Slade and Hutchings 1987a; Slade and Hutchings 1987b) and moisture (Hartnett and Bazazz 1983) stress. Morphological patterns observed for devil's dub and thimbleberry are similar to results reported by Scagel and Maze (1984); more variation is accounted for within individuals rather than between individuals. Environmental correlates reported here do 123 not parallel tnose OT otner studies. Devirs CIUD morpnoiogy was uncorreiated witn tne environmental variables, though thimbleberry showed similar correlations to those reported by Piteika et al. (1985) and Gawler et al. (1987) for other clonal species. Canopy openings (Piteika et al. 1985) and moisture (Gawler et al. 1987) were correlated with the number and size of leaves and the number of flowers. Disturbance and moisture were both positively correlated with these traits for thimbleberry. If disturbance equates to girdling of the alder and therefore canopy openings, the relationship between these studies is more apparent. No interaction, as thimbleberry displayed, between moisture and canopy openings, though was observed by Piteika et al. (1985) for Clintonia borealis or Gawler et al. (1987) for Peducularis furbishiae. 124 7.2 PHENOLOGY The different successional niches occupied by thimbleberry and devil's club may account for some of the phenological differences observed (Rathcke and Lacey 1985). Devil's club, as a member of a late successional community may be a stronger competitor than thimbleberry, which is found at earlier stages along a successional gradient (Grime 1979). It has a well defined peak of leaf production during the longest days of summer followed by the rapid production of flowers. Grime (1979) indicates that competitive species generally will produce leaves during periods of maximum productivity and then flower. In contrast, thimbleberry produced leaves early in the summer on two-year-old stems, which may then flower for up to two months. One-year-old stems produce leaves throughout the summer. The strategies of devil's club and thimbleberry are quite different and may indicate stronger differences between the two species, especially reproductively, than the ANOVA indicated. Little variation is accounted fa by species differences in the ANOVA of reproductive phenology (Table 31), these results may partly reflect the coding used. Each thimbleberry had several flowering stems which flowered continuously; whereas devil's club had only one a two flowering stems per quadrat with only one floral spike per stem. Thimbleberry was, therefae, potentially mae phenologically varied than devil's dub, although the average quadat value fa the two spedes was similar. Annual dimatic differences appeared to have some effect on thimbleberry reproductive phendogy though not on the vegetative phendogy. Flowering proceeded at a fasta rate in 1987 as compared to 1986 (Figure 17), but in addition far mae thimbleberry plants bore fruit at both sites in 1987 than in 1986 (Appendix I). Annual variation in numbers of fruit and flowers was also reported to be substantial by Duke (1990), and the relationship between variation in dimatic factors, such as frost (Rathcke 1988), temperature (Helenurm and Barrett 1987; Rathcke 1989) and rainfall (Jackson and Bliss 1984; Gill and Mahall 1986; Heideman 1989), on flowering phendogy have been recorded. 125 in contrast to tnimDieDerry, annual cnnerences tor aevii's CIUD were seen tor vegetative phenologies only (Figure 18), with phenological development proceeding more rapidly in 1986 than 1987. Annual differences observed for devil's club phenology may be due to interactions of canopy cover and climate, which would affect the rate of snow melt and the temperature at each quadrat (Table 32. Figure 2). Temperature was also linked to annual differences in leaf and flower phenology of Aralia nudicaulis (Flanagan and Moser 1985). Environmental differences between sites may maintain the phenological differences between two populations (Primack 1985). A large proportion of the phenological variation for thimbleberry is accounted for by site differences (Table 31), with canopy cover negatively correlated with the 1987 vegetative phenology and disturbance due to girdling and fluctuating water conditions positively correlated with all phenological variables (Table 33). Separate populations were also found to exhibit different phenological patterns in different environments by Van Cauteren and Lefebvre (1986) and Nilsen (1986). Differential responses to environmental factors were also observed at each site for thimbleberry. At WED, phenological development was negatively correlated with moisture and canopy cover due to alder; whereas at KIT phenological development was more rapid at open sites with more fluctuations in water conditions. Site differences and differential responses within sites to moisture were also reported by Heideman-(1989). Studies have indicated that phenology may be affected by pollinator availability (Rathcke and Primack 1985; Rathcke 1988) or the availability of seed dispersers (Rathcke and Primack 1985; Primack 1985). Devil's club was observed to have many different pollinators on the large conspicuous floral stems. The short period of flowering or the narrowing of devil's club's floral niche, may as Rathcke (1988) suggests for several other species, be a result of competition for pollinators. The availability of resources or the presence of seed dispersers, though, are equally likely influences on phenology. Devil's club fruit was favored by bears which distributed these large masses 126 of berries widely inctheir droppings. No other plant appeared to be a serious competitor for this mode of dispersal as none bore fruits as large, which offered an equivalent reward. Migrating salmon, however, which arrive in August, would certainly distract bears. In this case, the availability of seed dispersers may shape devil's club fruiting phenology. The large clusters of fruit may also be expensive to produce and resource availability may limit devil's club fruiting phenology. Fruits mature immediately following the longest days of summer. Perhaps increasing levels of photosynthate produced in late July are required for fruit to ripen successfully. Flanagan and Moser (1985) inferred that both resource limitation and the lack of pollinators were important factors in the phenology of Aralia nudicaulis Thimbleberry reproductive phenology may also be subject to these constraints. Although flowers are produced continuously for almost two months, nonetheless a peak in floral production is apparent. During this peak in late June (Figure 15), flowers are filled with many (>20 per flower) tiny mating beetles, which may be the main pollinator. Many other pollinators were also observed though none were as plentiful as the beetles. In addition, the availability of seed dispersers may determine when thimbleberry fruit ripen. The main period of fruit ripening is August (Figure 15), coinciding with the arrival of migrating birds. From mid-June on. 1987 was warmer and drier than 1986 (Figure 2) and far more fruit ripened that summer (Appendix I). Reproductive development was also more rapid in 1987 (Figure 17). Perhaps flowering proceeds only after a critical number of degree days is achieved, as has been suggested for many temperate woody species (Lieth 1974; Reader 1983; Rathcke 1985). 127 7.3 DEMOGRAPHY The stem demography for thimbleberry varied greatly over the time of the study, whereas devil's club changed little. The Type II survivorship curve for thimbleberry stems (Figure 19), suggests a constant risk of mortality throughout their lifespan (Hutchings 1976). The devil's dub cumulative distribution (Figure 20) implies that the risks of stem mortality diminish with age increasing the probability that some very old individuals will be found in the population (Hutchings 1976). Indeed devil's dub stems were far older than thimbleberry stems. While the dynamics of ramet production may differ greatly between the two spedes, it is possible that genet dynamics may not be too dissimilar. As the characteristic spedes of an earlier state of forest succession than devil's dub, thimbleberry may be expected to have lower survivorship (Bierzychudek 1982; Whitney 1986). However in areas of recurring disturbance, thimbleberry genets may also have long lifespans. Carlsson and Callaghan (1990) indicate that long term observations are necessary to more fully understand genet dynamics. Annual temporal variation accounts for a larger proportion of the variation in thimbleberry than spatial environmental heterogeneity between populations (Table 41). Given that demography examines changes through time, it is not surprising that temporal influences would be large. Indeed, greater demographic variation between years than between populations was also reported by Weiss (1981) for an annual, and by-Waite (1984) and Mack and Pyke (1983) for Plantago coronopus and Bromus tectorum respectively. Annual dimatic differences may partially explain the demographic fluctuations in stem recruitment for thimbleberry. As climate in 1987 was more condudve to flowering, increased stem mortalities in 1987 for both one- and two-year-old stems may be due to this greater floral production. Two-year-old stems generally die after flowering. Perhaps one-year-old stems have two alternatives; either to augment the resources of flowering stems or to flower in later years. Some vegetative stems produced in 1987 may have 128 transposed resources to flowering stems and subsequently died. This implies that there is a relation between the number of vegetative and flowering stems in a clump, though this was not tested. Alternatively, this pattern may arise from demographic periodicity which is not annual but longer term. In quacrats where most stems produced in 1986 died by the end of that summer, stem survivability appeared greater in 1987 (Figure 19, Table 37); moreover, quacrats where stems produced in 1986 had the longest lifespans, those produced in 1987 had the shortest lifespans (Table 37). Over the period of this study, thimbleberry may alternately maximize vegetative and reproductive strategies and demographies, producing short-lived vegetative stems in one year and longer-lived flowering stems in the next. Most of the demographic variation for thimbleberry was found within rather than between sites, suggesting that environmental influences may be operating on a relatively fine scale (Table 41). Spatial demographic variation was also found to be similarly distributed by Huenneke (1987). Billington et al. (1990), Blom and Lotz (1985), Sarukhan and Harper (1973), Matlock (1987) and Watson and Cook (1987) also found extensive demographic plasticity within clonal individuals. With the exception of the canopy cover at WED for the time interval from 466-810 days, none of the environmental variables were correlated with stem democraphy (Table 44). Several reasons for this can be advanced. As most demographic variation occurs within individuals, the scale of environmental sampling may have been too large. Perhaps replicate samples within quadrats of the environmental correlates would reflect environmental-demographic correlations more strongly. Stem demographics also fluctuate temporally and environmental variables may have failed to encapsulate these fluctuations. If the demographic periodicity was not synchronized with the periodicity of the environmental correlates, a relationship may not be perceived. In addition, if thimbleberry sites within each year and individuals within each site were not sufficiently 129 variable, then, as with Gamier ana Roy (1388), no environmental correlates at tne level of sites or individuals will be perceived. Finally abundance fa thimbleberry and devil's club may, as in other shrub species (Balogh and Grigal 1988), be controlled by shrub regeneration rather than mortality. Recruitment a a compound measure composed of recruitment and mortality may mae strongly reflect the environmental correlates assessed rather than a measure based primarily upon mortality. 130 7.4 NICHE UTILIZATION AND LIFEHISTORY STRATEGY As the niches occupied by devil's dub and thimbleberry differed, so did their life history strategy. Devil's club is located in an area of old growth forest which may be characterized as having longterm stability; in contrast thimbleberry is found where disturbance has occurred. Perhaps due to the influence of disturbance, thimbleberry's strategy may be more flexible in response to change than devil's club. Most morphological, phenological and demographic variation is apportioned within quadrats (Tables 24, 31 and 41) suggesting that both spedes are somewhat plastic. Unlike thimbleberry. however, this variation is not readily observable for devil's dub. implying that it may be less variable than thimbleberry. If the plasticity observed for thimbleberry is appropriate to the environment, it may, as Levin (1986) suggests for a phenotypically plastic strategy, be a rapid and flexible method of accommodating change in environmental conditions (Levin 1986). Environmental responses also illustrate thimbleberry's flexibility in that disturbance is correlated with both morphology and phenology (Table 26,33, Appendix III). In contrast devil's dub generally does not survive after disturbance and is shade tolerant, a trait more characteristic of a competitor of later successional stages (Grime 1979). Thimbleberry has been reported to be more successful under the open canopy of earlier succession (Barber 1976). This flexibility is also structurally apparent. Thimbleberry is rhizomatous-and may, as Abrahamson (1980) suggests for rhizomatous donal spedes, be able to position stem ramets in optimum conditions by growing through less favorable areas. Rhizomes were observed to consist of a complex network with dumps of stems separated by varying rhizome lengths. Devil's dub is stoloniferous, with branches produced on erect stems. Once these stems become horizontal, the branches root and become independent. Their position within the site may therefore, be limited by their placement on the stem. If the stolon grows though an unfavorable area, the branch may root in a less desirable location and be unsuccessful. It does like the thimbleberry stem 131 ramet have the advantage of being physiologically connected to the parent plant, however. Antos and Zobel (1984) also speculate that extensive rhizome systems enable plants to cross unfavorable habitat and provide flexibility in dealing with the heterogeneous forest floor. Other flexibility expressed by thimbleberry stems and their lateral branches, aside from varying the numbers of leaves produced and lengths grown (Appendix I), is their several developmental potentialities. They may, 1) transfer resources to flowering stems and subsequently die, 2) flower in the second year of growth, or 3) if conditions are not suitable they may branch and flower in later years. Annual demographic differences also suggest that clones may alternatively produce vegetative stems in one year and reproductive stems in the next (Table 37, Figure 19). Stems and branches were also observed to develop at different rates, with two-year-old stems developing most rapidly followed by one-year-old stems and branches (Appendix I). These same patterns were not observed for devil's club. Branches and stems were not phenologically different, neither did they differ in annual growth or the number of leaves produced (Appendices I and II). Unlike thimbleberry, devil's dub stems all shared the same fate, being initially vegetative and finally flowering and branching. This flexibility may allow thimbleberry to react to the rapidly changing site conditions that occur in early successional stages. Certainly the amount and size of vegetative and reproductive tissue was correlated with disturbance and mdsture (Appendix lilt. Phenological patterns also suggest somewhat more flexibility for thimbleberry. Devil's dub produces leaves during the longest days of summer and then flowers (personal observation), a competitive strategy as suggested by Grime (1979). Thimbleberry produces leaves all summer on one-year-dd stems and flowers for up to two months (personal observation). While the devil's dub strategy may produce more seed if conditions are optimum and may be advantageous in a temporally stable site, this single pulse of seed and leaf production may be risky in a fluctuationing environment. Thimbleberry, by producing flowers and leaves all summer, has a higher probability of 132 some seed production and some increase in vegetative tissue even if disturbance occurs. A strategy which allows simultaneous and longterm reproductive and vegetative growth may be more suited to an unpredictable growing season (Harper 1977; Rathcke and Lacey 1985). The differences in modular longevity between species may also be a reflection of the character of their differing niches. Both genets and stem ramets of devil's club are long-lived. Thimbleberry stems lived to be four years old (Table 20) and while genets may survive for lengthy periods (Halpern 1989), they are unlikely to live the hundreds of years which may be the case for devil's dub. Devil's club stems flower later in life than thimbleberry and also branch later. This also suggests that environmental fluctuations are perceived in a different time scale than thimbleberry. Although phenological differences were seen between years, perhaps other changes occur at a much larger time scale for devil's dub. For example perhaps flowering or branching are influenced by environmental conditions occurring over several seasons. The scale at which temporal fluctuations affect this spedes may be in the order of decades. Thimbleberry may also be influenced by temporal fluctuations which are not annual, but short-term, such as weather patterns or disturbance events, and longer-term, such as successional patterns. Only long-term monitoring would assess such phenomena, however. Regardess of niche differences certain similarities emerged between both spedes. Phenology and morphology were positively correlated; with larger flowering plants having more rapid developmental rates for both spedes. Slade and Hutchings (1989) found similar relationships between phenology and morphology with Glechoma hederaceae . They suggest that larger stems may initiate growth earlier than smaller stems and may also have a greater probability of flowering. Larger stems were certainly more likely to flower for both devil's club and thimbleberry (Tables 20 and 23). Morphdogy was also linked to demography, however this relationship may be somewhat trivial; number of thimbleberry stems in each quadat was correlated~with the 133 numDer OT stems dying in eacn time period ana Tor devils CIUD, me number OT stems and their length were correlated with me number of newly produced stems (Appendix III). The lack of any correlation between floral morphology and demography for thimbleberry was unexpected, as most stems died after flowering whether in their second or third year of growth. In addition, stems which eventually flowered generally had longest lifespans. Bradbury (1981), Hutchings (1983) and Gawler et al. (1987) found that flowering stems had longest lifespans for other clonal species. This has been interpreted as the strong commitment to sexual reproduction by a clone (Slade and Hutchings 1989). There was also no evidence to suggest devil's club stems have longer lives if they flower. The survivorship curve indicated that the risk of mortality declined with age and the rate at which this declines is not altered once stems are mature enough to flower (Figure 21). 134 7.5 MANAGEMENT Both devil's club and thimbleberry live longer than the two year scope of this study. Together with the annual differences observed, this implies that only a longterm study will achieve a deep understanding of the population dynamics and autecology of these two species. For example, longer-term studies may discern the demographic pattern of devil's club and determine whether the two year pattern observed for thimbleberry repeats. The differing environmental relationships inferred for thimbleberry within and between populations also suggests that stringent management rules may be unattainable. Environmental relationships may also be complex at other levels. Certainly the differing developmental possibilities of thimbleberry stems interacting with environment make any demographic prediction tenuous. The study does suggest, however, that alder girdling may be an effective management practice at some sites. It appeared to affect thimbleberry growth and phenology, though not simply through light relations, and also interacted with moisture conditions (Appendix III). Alder girdling may be a successful management technique for alder and thimbleberry at drier sites. Where flooding occurs or if the site is moist, however, opening up the canopy by girdling may encourage thimbleberry, especially if it persists at the site after logging (Halpern 1989). 135 7.6 FURTHER RESEARCH This study was descriptive and hypothesis testing was not performed. Rather the study sought preliminary data to construct a framework for suitable hypothesis generation. Hypotheses based on observation have the advantage of a meaningful relationship to the plant, while not being overly simplistic. Perhaps more importantly, descriptive studies provide criteria to discard unsuitable hypotheses before time and money are spent. Based on this study, future research could explore the following questions more fully: 1. How do increasing temperatures occurring under an open canopy increase devil's club's rate of development? 2. Do longer thimbleberry and devil's club stems begin to develop earlier and do they develop more rapidly? Does earlier development ensure that maximum stem length is attained in a season for thimbleberry? 3. What quantities and proportions of light, moisture and temperature interact to affect thimbleberry's morphology and phenology? 4. What is the periodicity of the demographic pattern for thimbleberry? Does the strategy of each thimbleberry dump of stems vary from vegetative to reprodudive in a repeatable pattern? Do thimbleberry vegetative stems adively transport resources to reprodudive stems? Is there a relationship between the number of reprodudive stems and the number of vegetative stems per quadat? Finally as with most studies mae questions are unanswered than answered. The large within quadat term in all ANOVA calculations (Tables 24,31,41) and the extent to which this variation is due to plasticity a development may be of interest to assess by examining within quadat environmental variability and the phenotypic response. and they all lived happily ever after.. 136 LITERATURE CITED Aarssen, L. W. and R. Turkington. 1985. Within-species diversity in natural populations of Holcus lanatus. Lolium perenne and Trifolium repens from four different-aged pastures. Journal of Ecology. 73:869-886. Abrahamson, W. G. 1980. Demography and vegetative reproduction. In: Solbrig, 0. T. ed. Demography and evolution in plant populations. Los Angeles: University of California; p. 89-106. Abrahamson, W. G. and M. Gadgil. 1973. Growth form and reproductive effort in goldenrods (Solidaao . Compositae). American Naturalist. 107:651-661. Alaback, P. B. 1980. Biomass and production of understory vegetation on serai sitka spruce-western hemlock forests of southeast Alaska. Corvallis: Oregon State University. Dissertation. . 1986. Biomass regression equations for understory plants in coastal Alaska: effects of species and sampling design on estimates. Northwest Science. 60:90-103. Allen, T.F.H. and T.B. Starr. 1982. Hierarchy perspectives for ecological complexity. Chicago-London: The University of Chicago Press. Allen, T.F.H. and E.P. Wyleto. 1983. A hierarchical model for the complexity of plant communities. Journal of Theoretical Biology. 101:529-540. Alliende, M.C. and J. L. Harper. 1989. Demographic styudies of a dioecious tree. I. Colonization, sex and age structure of a populationof Salix cinerea . Journal of Ecology. 77:1029-1047. Anderson, M.C. 1964. Studies on the woodland light climate. I. The photographic computation of light conditions. Journal of Ecology. 52:27-41. Angevine, M.W. 1983. Variations in the demography of natural populations of the wild strawberries Fragaria vesca and F. virgianum. Journal of Ecology. 71:959-974. Antos, J. A. and D. B. Zobel. 1984. Ecological implications of belowground morphology of nine coniferous forest herbs. Botanical Gazette. 145:508-517. Austin, M. P. 1987. Models for the analysis of species' response to environmental gradients. Vegetatio 69:35-45. . 1990. Community theory and competition in vegetation. In: Grace, J.B. and Tilman eds. Perspectives in Plant Competition. Academic Press; San Diego; p. 215-239. Baker, H. G. 1974. The ecology of weeds. Annual Review of Ecology and Systematics. 5:1-24. Bakker, J.P., M. Dekkar and Y. De Vries. 1980. The effectof different management practices on a grassland community and the resulting fate of seedlings. Acta Botanica Neerlandica. 29:509-522. 137 Balogh, J . 0 . and D. F. Grigal. 1988. Tall shrub dynamics in northern Minnesota aspen and conifer forests. Res. Pap. Nc-283. St. Paul MN: U.S. Department of Agriculture, Forest Service, North Central Forest Experiment Station. 15 p. Barber, H.W. 1976. An autecological study of salmonberry (Rubus spectabilis. Pursh) in western Washington. Washington: University of Washington. Dissertation. Barnes, P. W. 1986. Variation in the big bluestem (Andropogon gerardii) - sand bluestem (Andropogon hallii) complex along a local dune/meadow gradient in the Nebraska sandhills. American Journal of Botany. 73(2):172-184. Bazely, D.R. and R.L. Jefferies. 1989. Leaf and shoot demography of an arctic stoloniferous grass, Puccinellia phrvganodes in response to grazing. Journal of Ecology. 77:811-822. Bazzaz, F.A. and J.L. Harper. 1977. Demographic analysis of the growth of Linum usitatissimum. New Phytologist. 78:193-208. Beatty, S.W. 1984. Influence of microtopography and canopy species on spatial patterns of forest understory plants. Ecology. 65:1406-1419. Beckett, P.H.T. andR. Webster. 1971. Soil variability: a review. Soil and fertilizers. 34(1):1-15. Bell, A.D. 1984. Dynamic morphology: a contribution to plant population ecology. In: Dirzo, R. and J. M. Sarukhan eds. Perspectives in plant population ecology. Sunderland, Massachusetts: Sinauer Associates; p.48-65. Bierzychudek, P. 1982. Life histories and demography of shade-tolerant temperate forest herbs: a review. New Phytologist. 90:757-776. Billington, H.L., A.M. Mortimer and T. McNeilly. 1990. Survival and reproduction in adjacent grass populations: effects of done, time and environment. Journal of Ecology. 78:1-14. Blom, C.W.P.M. 1988. The realism of models in plant demography. Acta Botanica Neerlandica. 37(4) :421-438. Blom, C.W.P.M. and L.P. Lotz, 1985. Phenotypic plasticity and genetic differentiation of demographic characteristics in some Plantago spedes. In: Haeck, J. and J.W. Woldendorp eds. Structure and functioning of plant populations 2. Amsterdam: North Holland Publishing Company, p. 105-113. Bouyoucos, G.J. 1962. Hydrometer method improved for making particle size analysis of soils. Agronomy Journal. 54:464-465. Bradfield, G. E. and A. Campbell. 1986. Vegetation-elevation correlation in two dyked marshes of northeastern Vancouver Island: a multivariate analysis. Canadian Journal of Botany. 64(11):2487-2494. Briggs, J. B., J. Danek, M. Lyth and E. Keep. 1982. Resistance to the raspberry beetle Byturus tomentosus in Rubus spedes and their hybrid derivatives with Rubus idaeus cultivar Norfolk giant. Journal of Horticultural Sdence. 57(1):73-78. 138 Burdon, J. J. 1980. Intra-specific diversity in a natural population of Trifolium repens. Journal of Ecology. 68:737-757. Burdon, J.J., D.R. and Marshall and A.D.H. Brown. 1983. Demographic and genetic changes in populations of Echium plantagineum . Journal of Ecology. 71:667-679. Carlsson, B.A. and T. V. Callaghan. 1990. Effects of flowering on the shoot dynamics of Carex bigelowii along an altitudinal gradient in Swedish Lapland. Journal of Ecology. .78:152-165. Carroll, D. J. 1987. Some multidimensional scaling and related procedures devised at Bell Laboratories with ecological applications. In: Legendre P. and L. Legendre eds. Developments in numerical ecology. Springer-Verlag; Berlin-Heidelberg-New York-London-Paris-Tokyo: p. 65-138. Carroll, D.J. and Jih-Jie Chang. 1970. Analysis of individual differences in multidimensional scaling via an N-way generalization of "Eckart-Young" decomposition. Psychometrika 35(3):283-319. Carter, R. E. and L. E. Lowe. 1986. Lateral variability of forest floor properties under second-growth Douglas-fir stands and the usefulnes of composite sampling techniques. Canadian Journal of Forest Resources. 16:1128-1132. Chan, S.S., R. W. McCreight, J.D. Walstad and T. A. Spies. 1986. Evaluating forest vegetation cover with computerized analysis of fisheye photographs. Forest Science 32(4) :1085-91. Chang, D. H. S. and H. G. Gauch, Jr. 1986. Multivariate analysis of plant communities and environmental factors in Ngari, Tibet. Ecology. 67(6):1568-1575. Chazdon, R. L. and C. B. Field. 1987. Photographic estimation of photosynthetically active radiation: evaluation of computerized technique. Oecologia (Berlin). 73:525-532. Cid-Benevento, C. R. 1987. Relative effects of light, soils moisture availability and vegetative size on sex ratio of two monoecious woodland annual herbs: Acalypha rhomboidea (Euphorbiaceae) and Pilea pumila (Urticaceae) Bulletin of the Torrey Botanical Club. 114(3)293-306. Clark, S.C. 1980. Reproductive and vegetative performance in two winter annual grasses, Catapodium rigidum (L). CE. Hubbard and C. marinum (L.) CE. Hubbard. 2. Leaf-demography and its relationship to the production of caryopses. New Phytologist. 84:79-93. Coates, D. and S. Haeussler. 1986. A preliminiary guide to the response of major species of competing vegetation to silvicultural treatments. Land management handbook no. 9, ISSN 02291692. Victoria, B.C.: Ministry of Forests. Cook, R.E. 1985. Growth and development in clonal plant populations. In: Jackson, J.B.C., L.YV. Buss and R.E. Cook eds. Population biology and evolution of clonal organisms. New Haven, Connecticut: Yale University Press, p.259-296. 139 . 1979a. Patterns of juvenile mortality and recruitment in plants. In: Solbrig, O.T., S. Jain, G.B. Johnson and P.H. Raven eds. Topics in plant population biology. New York: Columbia University Press, p.2067-231. . 1979b. Asexual reproduction: further consideration. American Naturalist. 113:767-772. Coutin, P.. M. C. Feller and K. Klinka. 1983. Lateral variability in some properties of disturbed forest soils in southwestern British Columbia. Canadian Journai of Soil Science. 63:529-539. Coupe, R., CA. Ray, A. Comeau, M.V. Ketcheson and R.M. Annas. 1982. A guide to some common plants of the Skeena Area, British Columbia. Land Management Report, ISSN 0229-1662; no.4. Ministry of Forests, Victoria, B.C. Crawley, M.J. 1986. Life history and environment. In: Crawley, M.J. ed. Plant Ecology. Oxford-London-Edinburgh-Boston-Palo Alto-Melbourne: Blackwell Scientific Publications; p. 253-290 Douglas, D. A. 1989. Clonal growth of Salix setchelliana on glacial river gravel bars in Alaska. Journal of Ecology. 77:112-126. Duke, N.C 1990. Phenological trends with latitude in the mangrove tree Avicennia marina. Journal of Ecology. 78d:113-133. Dunn, CP. 1986. Shrub layer response to death of Ulmus americana in southeastern Wisconsin lowland forests. Bulletin of the Torrey Botanical Club. 113:142-148. Eis, S. 1981. Effect of vegetative competition on regeneration of white spruce. Canadian Journal of Forest Resources. 11:1-8. Ennos, RA. 1985. Measuring the effects of genetic variation on plant fitness. In: Haeck, J. and J.W. Woldendorp eds. Structure and functioning of plant populations. Amsterdam: North Holland Publishing Company, p. 51-64. Environment Canada, Atmospheric Environment Service, Monthly Record, Meteorological Observations in Canada. Eriksson, 0. 1988. Patterns of ramet survivorship in clonal fragments of the stoloniferous plant Potentilla anserina. Ecology. 69(3):736-740. . 1985. Reproduction and clonal growth in Potentilla anserina L. (Rosaceae): the relation between growth form and dry weight allocation. Oecologia (Berlin). 66:378-380. Flanagan, L. B. and W. Moser. 1985. Flowering phenology, floral display and reproductive success in dioecious Aralia nudicaulis L. (Araliaceae). Oecologia (Berlin). 68:23-28. Fone, A.L. 1989. A comparative demographic study of annual and perennial Hypochperis (Asteraceae). Journal of Ecology. 77:495-508. » Fowler, N., J. Zasada and J. L. Harper. 1983. Genetic components of morphological variation in Salix repens. New Phytologist. 95:121-131. M O Fox, D.J. and K.E. Guire. 1976. Documentation for MIDAS. 3d ed. University of Michigan, Ann Arbor: Statistical Research Lab. Funk and Wagnalls standard college dictionary, Canadian Edition. 1978. Vancouver, B.C.: Fitzhenry and Whiteside Limited. Gadgil, M. and 0. T. Solbrig. 1972. The concept of r- and K-selection: evidence from wild flowers and some theoretical considerations. American Naturalist. 106:14-31. Gamier, E. and J. Roy. 1988. Modular and demographic analysis of plant leaf area in sward and woodland population of Dactylis glomerta and Bromus erectus. Journal of Ecology. 76:729-743. Gawler, S. C , D. M. Waller and E. S. Menges. 1987. Environmental factors affecting establishment and growth of Peducularis furbishiae: a rare endemic of the St. John river valley, Maine. Bulletin of the Torrey Botanical Club. 114(3):280-292. Gilbert, N. and A.P. Gutierrez. 1973. A plant aphid parasite relationship. Journal of Animal Ecology. 42(2):323-340. Gill, D.S. and B. E. Mahall. 1986. Quantitative phenology and water relations of an evergreen and a deciduous chaparral shrub. Ecological Monographs. 56(2) :127. Gittins, R. 1985. Canonical analysis a review with applications in ecology. Bertin-Heidelberg-New York-Toyko: Springer-Verlag. Grant, V. 1981. Plant Speciation. New York: Columbia University Press. Gratkowski, H. 1971. Mid summer foliage sprays on salmonberry and thimbleberry. United States Forest Service, Pacific Northwest Forest and Range Experimental Station Research Notes. 171:1-15. Greig, M. and D. Osterlin. 1978. UBC ANOVAR analysis of variance and covariance revised. Vancouver: University of British Columbia Computing Centre. Greig-Smith, P. 1983. Quantitative Plant Ecology, 3d ed. Studies in Ecology, vol 9. Berkley: University of California Press. Grime, J.P. 1979. Plant strategies and vegetation processes. Chichester: J. Wiley and Sons. Haeussler, S. and D. Coates. 1986. Autecological characteristics of selected species that compete with conifers in British Columbia: a literature review. Land management report ISSN 0702-9861; no. 33. Victoria, B.C.: Ministry of Forests. Haeussler, S. J. Pojar, B.M. Geisler, D. Yole and R.M. Annas. 1984. A guide to the coastal western hemlock zone, northern drier maritime subzone (CWFf), in the Prince Rupert forest region, British Columbia. Land Management Report. ISSN 0702-9861; no. 21. Victoria, B.C.: Ministry of Forests. Halpern, C. B. 1989. Early successional patterns of forest species: interactions of life history traits and disturbance. Ecology. 70(3)704-720. H I Hancock, J. F. and M. P. Pritts. 1987. Does reproductive effort vary across different life forms and serai environments? a review of the literature. Bulletin of the Torrey Botanical Club. 114(1):53-59. Harper, J.L. 1980. Plant demography and ecological theory. Oikos. 35:244-253. . 1978. The demography of plants with clonal growth. In: Freysen, A.H.J, and J.W. Woldendrop eds. Structure and functioning of plant populations. Amsterdam: North-Holland Publishing Co. p.27-45. . 1977. Population Biology of Plants. London: Academic Press. Harper, J.L. and A.D. Bell. 1977. The population dynamics of growth of form in organisms with modular construction. In: Anderson, R.M., B.D. Turner and L.R. Taylor eds. Population dynamics 20th symposium of the British ecological society. Oxford: Blackwell Scientific Publications, p. 29-52. Harper, J.L. and J. White. 1974. The demography of plants. Annual of Ecology and Systematics. 5:419-463.* Harper, J. L., P. J. Lovell and K. G. Moore. 1970. The shapes and sizes of seeds. Annual Review of Ecology and Systematics. 1:327-356. Hardwick, R.C. 1984. Some recent developments in growth analysis - a review. Annals of Botany. 54(6):807-812. Hartgerink, A.P. and F.A. Bazzaz. 1984. Seedling-scale environmental heterogeneity influences individual fitness and population structure. Ecology. 65:198-206. Hartnett, D.C. 1987. Effects of fire on clonal growth and dynamics of Pityopsis graminifolia (Asteraceae). American Journal of Botany. 74(11):1737-1743. Hartnett, D.C. and F. A. Bazzaz. 1985. The genet and ramet population dynamics of Solidago canadensis in an abandoned field. Journal of Ecology. 73:407-413. Heideman, P.D. 1989. Temporal and spatial variation in the phenology of flowering and fruiting in a tropical rainforest. Journal of Ecology. 77:1059-1079. Heinrich, B. 1976. Flowering phenologies: bog, woodland and disturbed habitats. Ecology. 57:890-899. Helenurm, K. and S.C.H. Barret. 1987. The reproductive biology of boreal forest herbs. II. phenology of flowering and fruiting. Canadian Journal of Botany. 65:2047-2056. Herndon, A. 1988. Variation in resource allocation and reproductive effort within a single population of Liatris laevigata Nuttall (Asteraceae). American Midland Naturalist. 118(2):406-413. Hickman, J.C. 1975. Environmental demography of the sand dune annual Caklla edentula growing along an environmental gradient in Nova Scotia. Journal of Ecology. 69:615-630. 142 Hitchcock, C. L. and A. Cronquist. 1973. Flora of the pacific northwest an illustrated manual. Seattle, Washington: University of Washington Press. Hobbs, R.J. and H.A. Mooney. 1987. Leaf and shoot demography in Baccharis shrubs of different ages. American Journal of Botany. 74{7):1111-1115. Hobbs, R. J. and H. A. Mooney. 1985. Vegetative regrowth following cutting in the shrub Baccharis pilularis ssp. consanginea (DC). American Journal of Botany. 72(4):514-519. Holzner, W. 1982. Concepts, categories and characteristics of weeds. In: Holzner W. and M Numata eds. Biology and Ecology of weeds. The Hague: Dr. N. Junk; p. 3-20. Huenneke, L.F. 1987. Demography of a clonal shrub, Alnus incana ssp. rugosa (Betulaceae). The American Midland Naturalist. 117(1):43-55. . 1983. Understory response to gaps caused by the death of Ulmus americana in central New York. Bulletin of the Torrey Botanical Club. 110:170-175. Huenneke, L.F. and R. R. Sharitz. 1986. Microsite abundance and distribution of woody seedlings in a South Carolina cypress-tupelo swamp. The American Midland Naturalist. 1145(2):328-335. Hulten, E. 1968. Flora of Alaska and neighbouring territories a manual of vascular plants. Stanford, California: Stanford University Press. Hunt, R. 1978. Demography versus plant growth analysis. New Phytologist. 80:269-272. . 1982. Plant growth analysis: second derivatives and compounded second derivatives of splined plant growth curves. Annals of Botany. 50:317-327. Hunt, R. and F. A. Bazzaz. 1980. The biology of Ambrosia trifida L. V. response to fertilizer, with growth analysis at the organismal and sub-organismal levels. New Phytologist. 84:113-121. Hunt, R., J.W. Wilson, D. W. Hand and D. G. Sweeney. 1984. Integrated analysis of growth and light interception in winter lettuce. I. analytical methods and environmental influences. Annals of Botany. 54(6)743-757. Hutchings, M.J. 1986. The structure of plant populations. In: Crawley, M. J. ed. Plant ecology. OxforchLcriclcri-Edinburgh-Boston-Palo Alto-Melbourne:Blackwell Scientific Publications; p. 97-136. . 1983. Shoot performance and population structure in pure stands of Mercurialist perennis L. a rhizomatous perennial herb. Oecologia (Berlin). 58:260-264. . 1976. Plant population biology. In: P.D. Moore and S.B. Chapman eds. Methods in plant ecology. Oxford-London-Edinburgh-Boston-Palo Alta-Melbourne: Blackwell Scientific Publications; p. 377-436. Jackson, L.E. and L.C. Bliss. 1984. Phenology and water relations of three plant life forms in a dry tree-line meadow. Ecology. 65(4) :1302-1314. 1-43 Jeffers, J.N.R. 1976. An introduction of systems analysis with ecological applications. London, England: University Park Press. Jolliffe, P.A. and W.H. Courtney. 1984. Plant growth analysis: additive and multiplicative components of growth. Annals of Botany. 54:243-254. Jones, D. 1983. The influence of host density and gall shape on the survivorship of Diastrophus kincaidii Hymenoptera Cynipidae. Canadian Journal of Zoology. 61(9)2138-2142. Jurik, T. W. 1985. Differential costs of sexual and vegetative reproduction in wild strawberry populations. Oecologia tBerlin). 66:394-403. Kaul, R.B. 1985. Reproductive phenology and biology in annual and perennial Alismaticeae. Aquatic Botany. 22:153-164. Kawano, S., S. Hiratsuka and K. Hayashi. 1982. The productive and reproductive biology of flowering plants. V. Life history characteristics and survivorship of Erythronium japonicum . Oikos. 38:129-149. Kelly, D. 1989a. Demography of short-lived plants in chalk grassland. I. Life cycle variation in annuals and strict biennials. Journal of Ecology. 77:747-769. . 1989b. Demography of short-lived plants in chalk grassland. II. Control of mortality and fecundity. Journal of Ecology. 77:770-784. . 1989c. Demography of short-lived plants in chalk grassland. III. Population stability. Journal of Ecology. 77:785-798. Kephard, S.R. 1987. Phenological variation in flowering and fruiting of Asclepias. The American Midland Naturalist. 118(1):64-75. Kershaw, K. A. and J. H. Looney. 1985. Quantative and dynamic plant ecology, 3rd ed. London: Edward Arnold. Kirby, K.J. 1980. Experiments on vegetative reproduction in bramble (Rubus vestitus). Journal of Ecology. 68:513-520. Kirkpatrick, M. 1984. Demographic models based on size, not age, for organisms with indeterminate growth. Ecology. 65(6):1874-1884. Kjellson, G. 1985. Seed fall and phenological overlap in a guild of ant-dispersed herbs. Oecologia (Berlin). 68:140-146. Klinka, K., V.J. Krajina, A. Ceska and A.M. Scagel. 1989. Indicator plants of coastal British Columbia. Vancouver, B.C.;University of British Columbia Press. Koehl, M.A.R. 1989. Discussion from individuals to populations. In: Roughbarden, R., R. M. May, S. A. Levin eds. Perspectives in ecological theory. Princeton, N.J.;_Princeton University Press; p.39-53. 144 Krajina, V.J. 1969. Ecology of forest trees in British Columbia. Ecology of Western North America. 2:1-146. Kruskal, J. B. 1964a. Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika. 29(1):1-27. . 1964b. Nonmetric multidimensional scaling: a numerical method. Psychometrika. 29(2)1115-129. Kruskal, J.B. and M. Wish. 1978. Multidimensional Scaling. Beverly Hills, CA.: Sage Publications, Inc. Lau, R. R. and D. R. Young. 1988. Influence of physiological integration on survivorship and water relations in a clonal herb. Ecology. 69(1)215-219. Lavkulich, LM. 1981. Methods manual pedology laboratory. Vancouver, B.C.: Department of Soil Science, University of British Columbia. Law, R., D. Bradshawand P.D. Putwain. 1977. Life-history of Poa annua. Evolution. 31:233-246. Lee, H. S., A. R. Zangerl, K. Garbutt and F. A. Bazzaz. 1986. Within and between species variation in response to environment gradients in Polygonum pensylvanicum and Polygonum virginianum. Oecologia (Berlin). 68:606-610. Levin, D. A. 1986. Breeding structure and genetic variation. In: Crawley, M. J. ed. Plant Ecology. London, England: Blackwell Scientific Publications; p. 217-252. Lieth, H. ed. 1974. Phenology and seasonality modelling, vol 8. New York: Springer-Verlag. Loehie, C. 1987. Partitioning of reproductive effort in clonal plants: a benefit-cost model. Oikos. 49:199-208. Lovett Doust, J. and G.W. Eaton. 1982. Demographic aspects of flower and fruit production in bean plants, Phaseolus vulgaris L. American Journal of Botany. 69(7) :1156-1164. Lovett Doust, L. 1981. Population dynamics and local specialization in a clonaT perennial (Ranunculus repens) 2. Demography of leaves, and a reciprocal transplant-replant experiment. Journal of Ecology. 69:757-768. Luken, J.O. 1988. Population structure and biomass allocation of the naturalized shrub Lonicera maackii (Rupr.) Maxim, in forest and open habitats. The American Midland Naturalist. 119(2)258-267. Lyons, C. P. 1976. Trees, shrubs and flowers to know in British Columbia. Vancouver: J.M. Dent 8< Sons (Canada) Limited. Mack, R.N. and D. A. Pyke. 1983. The demography of Bromus tectorum variation in time and space. Journal of Ecology. 71:69-93. 145 Mader, D. L. 1963. Soil variability - a serious problem in soil-site studies in the northwest. Soil Science Society of America. 27:707-709. Maddox, 0. D., R. E. Cook, P. H. Wimberger and S. Gardescu. 1989. Clone structure in four Solidago altissima (Asteraceae) populations: rhizome connections within genotypes. American Journal of Botany. 76(2) :318-326. Maillete, L 1982. Structural dynamics of silver birch. 1. The fates of buds. Journal of Applied Ecology. 19:203-218. Matlock, G.R. 1987. Comparative demographies of four adjacent population sof the perennial herb Silene dioica (Caryophvllaceae). Journal of Ecology. 75:113-134. Maurer, B. A. 1987. Scaling of biological community structure: a systems approach to community complexity. Journal of Theoretical Biology. 127:97-110. McGraw, J.B. and J. Antonovics. 1983. Experimental ecology of Dryas octopetala ecotvpes. II. A demographic model of growth, branching and fecundity. Journal of Ecology. 71(3):899-912. McNicol, R. J., B. Williamson, D. L. Jennings and J. A. T. Woodford. 1983. Resistance to raspberry cane midge Resseliella theobaldi and its association with wound periderm in Rubus crataegifolium and its red raspberry derivatives. Annals of Applied Biology. 103(3):489-496. Menges, E. S. 1987. Biomass allocation and geometry of the clonal forest herb Laportea canadensis : adaptive responses to the environment or allometric constraints? American Journal of Botany. 74(4).551 -563. . 1986. Environmental correlates of herb species composition in five southern Wisconsin floodplain forests. The American Midland Naturalist. 115(1):106-117. . 1983. Plant strategies in relation to elevation and light in floodplain herbs. The American Naturalist. 122(4):454-473. Moot, J.H., J. Haeck, J. Van der Toorn and P.H. Vantienderen. 1989. Comparative demography of Plantago. I. Observations on eight populations of Plantaao lanceolata. Acta Botanica Neerlandica. 38(1):67-78. Newton, P.N. 1988. The structure and phenology of a moist deciduous forest in the central Indian highlands. Vegetatio. 75:3-16. Nilson, E.T. 1986. Quantitative phenology and leaf survivorship of Rhododendron maximum in contrasting irradiance environments of the southern Appalachian mountains. American Journal of Botany. 73(6):822-831. Noble, J.C.. A.D. Bell and J.L. Harper. 1979. The population biology of plants with clonal growth. I. The morphology and structural demography of Carex arenaria. Journal of Ecology. 67:983-1008. 146 Nybom, H. A. 1987. Demographic study of the apomictic blackberry, Rubus nessensis (Rosaceae). Nordic Journal of Botany. 7:365-373. Oberbauer, S. F. and B. R. Strain. 1986. Effects of canopy position and irradiance on the leaf physiology and morphology of Pentaclethra macroloba (Mimosaceae). American Journal of Botany. 73(3):409-416. Oka, H.I. 1976. Mortality and adaptive mechanisms of Oryza perennis strains. Evolution 30:380-392. O'Neill, R. V. 1989. Perspectives in hierarchy and scale. In: Roughgarden, J., R. M. May and S. Levin eds. Perspectives in ecological theory. Princeton, N.J.: Princeton University Press; p. 140-156. O'Neill, R.V., D. L. DeAngelis, J. B. Waide and T. F. H. Allen. 1986. A hierarchical concept of ecosystems. Princeton, N.J.: Princeton University Press. Pimentel, R. 1979. Morphometries. Dubuque: Randall/Hunt Publ. Co.. Pitelka, L. F., D. S. Stanton and M. 0. Peckenham. 1980. Effects of light and density on resource allocation in a forest herb, Aster acuminatus (Compositae). American Journal of Botany. 67(6):942-948. Piatt, W.J., G.W. Evans and S.L. Rathbun. 1988. The population dynamics of a long-lived conifer Pinus palustris. The American Naturalist. 131(4):491-525. Pojar, J. 1974. Reproductive dynamics of four plant communities of southwestern British Columbia. Canaidan Journal of Botany. 52:1819-1834. Pope, D.J. and P.S. Lloyd. 1975. Hemispherical photography, topography and plant distribution. In: Evans, G.C., R. Bainbridge and 0. Rackham, eds. Light as an ecological factor. 16th Symposium British Ecological Society. Oxford: Blackwell Scientific Publications, p.385-408. Primack, R.B. 1985. Patterns of flowering phenology in communities, populations, individuals and single flowers. In: White, J. ed. Handbook of vegetation science The population structure of vegetation. Dordrecht-Boston-Lancaster: Dr. W. Junk Publishers, p. 571-593. . 1980. Variation in the phenology of natural populations of montane shrubs in New Zealand. Journal of Ecology. 68:849-862. Ralhan, P.K., R.K. Khanna, S.P. Singh and J.S. Singh. 1985. Certain phenological characters of the shrub layer of Kumaun Himalayan forests. Vegetatio. 63:113-119. Rathcke, B. 1989. Flowering phenologies in a shrub community: competition and constraints. Journal of Ecology. 76:975-994. . 1988. Patterns of flowering phenologies: testability and causal inference using a randon model. In: Strong, D.R., D. Simberloff, L.G. Abele and A.B. Thistle eds. Ecological communities: conceptual issues and the evidence. Princeton: Princeton University Press, p. 383-393. M 7 Rathcke, B. and E.P. Lacey. 1985. Phenological patterns of terrestrial plants. Annual Review of Ecology and Systematics. 16:179-214. Reader, R.J. 1983. Using heat sum models to account for geographic variation in the floral phenology of two Ericaceous shrubs. Journal of Biogeography. 10:47-64. Reinartz, J. A. and J. W. Popp. 1987. Structure of clones of northern prickly ash (Xanthoxylum americanum). American Journal of Botany. 74(3):415-428. Richards, A. J. 1984. Plant breeding systems. London, England: George Allen 8c Unwin (Publishers) Ltd. Roy, J. and H. A. Mooney. 1987. Contrasting morphological and physiological traits of Heliotropium curassavicum L. plants from desert and coastal populations. Acta Oecologia (Oecologia Plantarum). 8(22):2:99-112. Sakai, A.K. and J.H. Sluak. 1985. Four decades of secondary succession in two lowland permanent plots in norther lower Michigan. American Midland Naturalist. 113:146-157. Sarukhan, J., M. Martinez-Ramos and D. Pihero. 1984. The analysis of demographic variability at the individual level and its population consequences. In: Dirzo, R. and J. Sarukhan eds. Perspectives on plant population ecology. Sunderland, Massachusetts: Sinauer Associates Inc., Publichers; p. 83-106. Sarukhan, J. and J.L. Harper. 1973. Studies on plant demography: Ranunculus repensL.. R bulbosus L, and R. acris L. I. Population flux and survivorship. Journal of Ecology. 61:675-716. SAS Users Guide: Statistics, Version 5 ed. 1985. Cary ,NC: SAS Institute Inc. SAS Technical Report P-161. Additional SAS/STAT procedures: CANCORR and ORTHOREG. 1985. Cary.NC: SAS Institute Inc. Scagel, R.K. and J. Maze. 1984. A morphological analysis of local variation in Stipa nelsonnii and S. richardsonii. Canadian Journal of Botany. 62:763-770. Schiffman, S. S., M. L.Reynolds and F.W. Young. 1981. Introduction to multidimensional scaling theory, methods and applications. New York-London-Toronto-Sydney-San Francisco: Academic Press. Schlichting, C. D. 1986. The Evolution of Phenotypic Plasticity in Plants. Annual Review of Ecology and Systematics. 17:667-93. . 1989. Phenotypic plasticity in Phlox II. plasticity of character correlations. Oecologia. 78:496-501. Schwaegerle, K.E. and D. A. Levin. 1990. Environmental effects on growth and fruit production in Phlox drummondii. Journal of Ecology. 78:15-26. Silva, J.F. and M. Ataroff. 1985. Phenology, seed crop and germination of coexisting grass species from a tropical savanna in western Venezuela. Acta Oecdogia(Oecologia Plantarum). 6(20)41-51. 148 Silvertown, J.W. 1985. Survival, fecundity and growth of wild cucumber, Echinovcvstis lobata. Journal do Ecology. 73:841-849. . 1982. Introduction to plant population ecology. New York: Longman Inc. Singer, M. J. and D. N. Munns. 1987. Soils, an introduction. New York: Macmillan Publishing Company. Slade, A.J. and M.J. Hutchings. 1989. Within- and between-population variation in ramet behaviour in the gynodioecious clonal herb, Glechoma hederacea (Latiatae). Canadian Journal of Botany. 67:633-639. Slade, A. J. and M. J. Hutchings. 1987a. The effects of nutrient availability on foraging in the clonal herb Glechoma hederacea . Journal of Ecology. 75:95-112. Slade, A. J. and M. J. Hutchings. 1987b. An analysis of the costs and benefits of physiological integration between ramets in the clonal perennial herb Glechoma hederacea . Oecologia (Berlin). 73:425-431. Smith, G. W. 1983. Artie pharmacognosia 2. devil's club Oplopanax horridus. Journal of Ethnopharmacology. 7(3) :313-320. Smith, T.M. and D.L. Urban. 1988. Scale and resolution of forest structural pattern. Vegetatio. 74:143-150. Sokal, R.R. and F.J. Rohlf. 1969. Biometry. San Francisco: W.H. Freeman and Co. Somers, G.F. and D. Grant. 1982. Influence of seed source upon phenology of flowering of Spartina alterniflora Loisel and the likelihood of cross pollination. American Journal of Botany. 68:6-9. Southwood, T. R. E. 1988. Tactics, strategies and templets. Oikos. 52:3-18. Stearns, S. C. 1977. The evolution of life history traits: a critique of the theory and review of the data. Annual Review of Ecology and Systematics. 8:145-71. Stewart, R. E. 1974a. Budbreak sprays for site preparation and release from 6 coastal brush species. United States Forest Service Research Paper PNW. 176:1-20. . 1974b. Foliage sprays for site preparation and release from 6 coastal brush species. United States Forest Service Research Paper PNW. 172:1-18. Stiles, E.W. 1980. Patterns of fruit presentation and seed dispersal in bird-disseminated woody plants in the eastern deciduous forest. American Naturalist. 116:670-86. Svensson, B. M. and T. V. Callaghan. 1988. Small-scale vegetation pattern related to the growth of Lycopodium annotinum and variations in its micro-environment. Vegetatio. 76:167-177. Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities. Princeton, New Jersey: Princeton University Press. 149 . 1986. Resources, competition and the dynamics of plant communities. In Plant ecology, Crawley, M. J. ed. Palo Alto, California: Blackwell Scientific Publications; p. 51-76. Turner, N. J. 1982. Traditional use of devil's club (Oplopanax horridus: Araliaceae) by native peoples in western North America. Journal of Ethnobotany. 2(1 ):17-38. Valentine, K.W.G., P.N. Sprout, T.E. Baker and L.M. Lavkulich. 1978. The soil landscapes of British Columbia. Victoria, B.C.: Resource Analysis Branch, Ministry of Environment. Van Cauteren, P. and C. Lefebvre. 1986. Morphological, phenological and cemical variationin woodland populations of the daffodil (Narcissus pseudonarcissus L). Acta Oecologia (Oecologia Plantarum). 7(21)327-337. Van der Toorn, J. and T. L. Pons. 1988. Seedling establishment of Plantago lanceolata L. and Plantago major L. between grass; an experimental investigation. II. Seedling survival and selection on germination. Oecologia (Berlin). 76:341-347. Verkaar, H.J. and A.J. Shenkeveld. 1984. On the ecology of short-lived forbs in chalk grassland: seedling development under low phton flux conditions. Flora. 175:135-141. Waite, S. 1984. Changes in the demography of Plantago coronopus at two coastal sites. Journal of Ecology. 72:809-826. Watkinson, A. R. 1985. Plant responses to crowding. In: White, J. ed. Studies on plant demography, a festschrift for John L Harper. London, England: Academic Press Inc. (London) Ltd; p. 275-290. Watkinson, A.R. and C.C. Gibson. 1985. Life-history variation and the demography of plant populations. In: Haeck, J. and J.W. Woldendorp, ed. Structure and functioning of plant populations 2. Amsterdam: North Holland Publishing Company; p. 105-113. Watson, M.A. and G.S. Cook. 1987. Demographic and developmental differences among clones of water hyacinth. Journal of Ecology. 75:439-457. Weiss, P.W. 1981. Spatial distribution and dynamics of the introduced annuarEmex australis in southeastern Australia. Journal of Applied Ecology. 18:849-864. Welter, D. E. 1987. Self-thinning exponent correlated with allometric measures of plant geometry. Ecology. 68(4) :813-821. Werner, P.A. 1985. Phenotypic variation and implications for reproductive success. In: Haeck, J. and J.W. Woldendorp, ed. Structure and functioning of plant populations 2. Amsterdam: North Holland Publishing Company; p. 1-26. Wieder, R.K., C. A. Bennett and G.E. Lang. 1984. Flowering phenology at Big Run Bog, West Virginia. American Journal of Botany. 71(2)203-209. White, J. 1985a. The thinning rule and its application to mixtures of plant populations, in: White, J. ed. Studies on plant demography, a festschrift fa John L. Harper. London, England: Academic Press Inc. (London) Ltd; p. 291-312. 150 . 1981. Demographic factors in populations of plants. In: Solbrig, O.T. ed. Demography and evolution in plant populations. Berkley: University of California Press; p.21-48. . 1981. The allometric interpretation of the self-thinning rule. Journal of Theoretical Biology. 89:475-500. . 1979. The plant as a metapopulation. Annual Review of Ecology and Systematics. 10:109-45. Whitney, G.G. 1986. A demographic analysis of Rubus idaeus and Rubus pubescens. Canadian Journal of Botany. 64:2916-2921. Wilkinson, L. 1988a. SYGRAPH. Evanston, IL: SYSTAT, Inc.. . 1988b. SYSTAT: the system for statistics. Evanston, IL: SYSTAT, Inc.. Yarie, J. and Mead, B, R. 1989. Biomass regression equations for determination of vertical structure of major understory species of southeast Alaska. Northwest Science. 63(5)221-231. Young, F. W. and R.Lewyckyj. 1979. ALSCAL user's guide, 3d. ed. Chapel Hill, N.C. : Psychometric Laboratory, Data Analysis and Theory Associates. Zar, J. H. 1974. Biostatistical analysis. Englewood Cliffs, N. J.: Prentice-Hall, Inc. 151 Appendix 1. Mann Whitney U tests for differences in morphology between thimbleberry sites in 1986 and 1987, between years (1986 ana 1987) and within plants between stems and branches. THIMBLEBERRY SITES YEARS 1987 1986 DC TB # flowers/quadrat .0019 1.000 .3609 .0240 # flowers/stem .1982 1.000 # flowering stems/quadrat .0004 1.000 # fruit/quadrat .0004 1.000 .0084 # stems/quadrat # 1-year-old stems/quadrat # 2-year-old stems/quadrat .1010 .0000 .3253 .2300 .7299 .0000 # leaves/branch .6636 .2378 .5194 .2480 * leaves/1 -year-old stem .0030 .1485 .0829 # leaves/2-year-old stem .0011 .3845 .5159 # leaves/stem .0004 .0311 .4494 .8181 1-year-old stem length .1146 .2568 .8887 2-year-old stem length .0000 .7622 .3186 branch length .0029 .6735 .5806 Thimbleberry branches and 1-year-old stems branches and 2-year-old stems 1-year-old stems and 2-year-old stems # stem leaves -1986 .0061 .0000 .0007 * stems leaves -1987 .0003 .0000 _ .0000 leaf length .0053 .0015 .0000 leaf width .0045 .0015 .0000 stem length -1987 .0000 .0000 .9923 stem length -1986 .0000 .0000 .4285 152 Devil's club * leaves -1987 »leaves -1986 length length grown stems and branches .0070 .0015 .0000 .8207 153 Appendix II. Mann-Whitney U tests for differences in generative and vegetative phenologies between thimbleberry sites in 1986 and 1987, between years (1986 and 1987), and within plants between stems and branches. THIMBLEBERRY SITES YEARS 1986 1987 DC TB reproductive phenology .0062 .0000 .0290 .0099 vegetative phenology .7914 .0000 .0099 ^ .8300 2-year-old stems vs branches 1- year-old stem vs 2- year-old stems 1-year-old stems vs branches stems vs branches vegetative phenology -KIT .0000 .0000 .0000 WED .0000 .0149 .0000 TB .0000 .0000 .0000 DC .0653 reproductive phenology KIT .0498 WED .0000 TB .0000 DC .7227 154 Appendix III. CANCORR relationships between morphological, phenological and demographic characters for TB, KIT, WED and DC and soils, frequency and presence/absence of plant neighbour matrices. Morphology A - morphological variance explained morphological variables B - morphological variance explained environmental variables C - environmental variance explained by environmental variables D - environmental variance explained by morphological variables TB KIT WED Canonical axes Correlation coefficient Redundancies A B C D I II .9654 .9095 .0915 .1412 .0853 .2021 .0914 .1011 .0851 .0837 I II .9654 .9095 .0915 .1412 .0853 .1168 .0914 .1011 .0851 .0837 I II .9856 .8614 .1715 .0707 .1666 .0626 .1116 .1040 .1084 .0921 DC I II .9908 .9732 .0920 .2373 .0904 .2247 .1591 .1885 .1562 .1785 Morphology -Frequency of TB KIT WED DC plant neighbours Canonical axes I II I II I II I Con-elation coefficient .8589 .7759 .9970 .9876 .9999 .9739 .9996 .9749 Redundancies A .3435 .0725 .1270 .0796 .0858 .1441 .0411 .1137 B .2534 .0437 .1263 .0776 .0858 .1366 .0411 .1081 C .1228 .0849 .0613 .1527 .0984 .1282 -.1018 .1136 D .0906 .0511 .0609 .1490 .0984 .1216 .1017 .1080 Correlation to PCA axis morphological axis I II III .8060 -.3221 .0555 .1485 .3769 -.0637 .1891 .3254 -.5350 .5212 -.0947 -.3798 environmental axis I II III .5043 -.0887 .6190 .3371 .1790 .0775 -.3507 .1971 -.2998 -.5599 .2821 -.0091 155 Morphology -Presence/ TB KIT WED DC absence of plant neighbours Canonical axes I II I II I II I Correlation coefficient Redundancies .8288 .7872 .9982 .9493 .9998 .9964 .9984 .9915 A .3258 .0839 .1252 .0837 .1376 .1044 .1169 .0769 B .2238 .0520 .1248 .0754 .1372 .1037 .1166 .0756 C .1321 .0821 .1217 .0729 .0932 .1402 .1051 .0918 D .0908 .0509 .1213 .0657 .0929 .1392 .1047 .0902 Correlation to PCA axes morphological axis I -7924 .2310 II -.1123 .1677 III -.0433 -.0200 environmental axis I 3946 .2725 II .7307 -.3341 III .1171 .0872 Phenology A - phenological variance explained phenological variables B - phenological variance explained environmental variables C - environmental variance explained by environmental variables D - environmental variance explained by phenological variables Phenology-soils Canonical axes Correlation coefficient Redundancies TB KIT WED DC .7322 .5154 .9234 .7426 .8279 .6845 .8973 .7993 A .1597 .2063 .1353 .4361 .3708 .3578 .1072 .2059 B .0856 .0548 .1154 .2405 .2336 .1667 .0863 .1315 C .0678 .1466 .1869 .1644 .0520 .1042 .0576 .2018 D .0899 .0390 .1594 .0906 .0356 .0488 .0464 .1289 156 Phenology -frequency of plant neighbours Canonical axes Correlation coefficient Redundancies TB KIT WED DC .8032 .7372 .9067 .8529 .9246 A .4249 .3057 B .2741 .1661 C .1470 .1249 D .0948 .0679 .5138 .1870 .4225 .1360 .1280 .0846 .1052 .0615 Correlation to PCA axis phenological axis environmental axis II .7611 .8914 II .8533 .4619 .1899 .3949 .1100 .1927 .0920 .1648 .2181 .3771 .2599 .2996 .1892 .0611 .1212 .0486 .0883 .8218 .5344 .9555 .0610 .8731 -.2408 II -.2846 .6842 -.1676 .7495 .0343 .7338 II .4286 -.4466 .0261 -.5647 .4854 .4112 .7425 -.0902 .5131 .0250 -.7720 .0840 II -.1874 .8082 .0371 -.5304 .1928 .2164 II .4710 .4863 .1581 .1345 .1482 .6802 Phenology -presence/ absence of plant neighbours Canonical axes Correlation coefficient Redundancies TB KIT WED .8227 .6983 .9150 .8108 .9083 II .8928 DC A .2728 B .1846 C .1513 D .1024 .4248 .2677 .4368 .2071 .2241 .2872 .1131 .1124 .0853 .0551 .0941 .0561 .1533 .5198 .1264 .4143 .1020 .1773 .0842 .1413 .9315 .8837 .1388 .1417 .1205 .1106 .0818 .1641 :0710 .1281 Correlation to PCA axis phenological axis environmental axis .5364 .7560 .1095 -.9363 II -.4906 .6298 .6490 .3099* II .6271 -.0839 .3237 -.1477 .8281 -.1311 -.2885 .8336 II .1215 .6354 -.0245 .1543 II -.0517 .4352 .3362 .2228 157 A - demographic variance explained phenological variables B - demographic variance explained environmental variables C - environmental variance explained by environmental variables D - environmental variance explained by demographic variables Demograpny TB KIT WED DC Demography -soils Canonical axes Correlation coefficient Redundancies A B C D Demography -frequency of plant neighbours Canonical axes Correlation coefficient Redundancies A B C D I II .8309 .6722 .0743 .1100 .0513 .0497 .2217 .2267 .1530 .1024 TB I II .8050 .6747 .1026 .0803 .0664 .0366 .1238 .1022 .0803 .0465 I II .9912 .9703 .0935 .0812 .0919 .0764 .0786 .3359 .0772 .3163 KIT I II .9922 .9809 .1061 .0885 .1044 .0852 .1386 .0858 .1365 .0826 I II .9979 .9841 .0968 .0732 .0964 .0709 .0948 .0981 .0944 .0950 WED I II .9733 .9611 .0655 .0796 .0621 .0735 .0908 .0886 .0860 .0819 I II .9412 .8164 .1066 .2731 .0945 .1820 .1169 .0827 .1035 .0551 DC I II .8609 .7381 .2784 .1692 .2064 .0922 .1175 .1557 .0871 .0848 158 Demography -presence/ TB KIT WED DC absence of plant neighbours Canonical axes I II I II I II I Correlation coefficient Redundancies .8371 .7038 .9995 .9718 .9991 .9799 .8874 .8095 A .0896 .0815 .1133 .1164 .0985 .1169 .2252 .1893 B .0628 .0404 .1132 .1099 .0983 .1123 .1773 .1240 C .1109 .1074 .1300 .0655 .0802 .1045 .0676 .0732 D .0777 .0532 .1299 .0619 .0800 .1003 .0532 .0480 Correlation to PCA axis demographic axis I -.2530 .2961 II .3311 .2424 III .6960 .2289 environmental axis I -.0037 -.2249 II .6911 -.1755 III -.2515 -.0837 Morphology - Phenology A - morphological variance explained morphological variables B - morphological variance explained phenological variables C - phenological variance explained by phenological variables D - phenological variance explained by morphological variables Morphology -phenology TB KIT WED DC Canonical axes I II I II I II I Correlation coefficient .9307 .7803 .9542 .8260 .9738 .8463 -.9620 .8623 Redundancies A .3204 .1432 .1156 .0424 .2551 .0647 .2101 .0641 B .2776 .0872 .1052 .0289 .2419 .0464 .1944 .0476 C .5290 .1936 .5065 .1128 .4784 .2000 .6190 .0810 D .4582 .1179 .4612 .0769 .4536 .1432 .5728 .0602 Correlation to PCA axis morphological .7647 axis I .4362 .1418 -.0561 .7071 -.1313 .6104 -.0919 II .2738 -.0786 -.4546 -.2646 .1783 .2182 .2845 -.1627 III .1321 -.0350 .6749 -.1388 -.5371 .3786 -.4476 -.5516 phenological .9660 axis I .1114 .9428 .0842 .9007 -.3961 .8979 -.2622 II .0238 -.7596 -.0665 .4895 .0671 .4883 ".2285 .1850 III .0279 .4974 -.3255 .0682 .0544 .2805 .3028 -.0602 159 Morphology-demography A - morphological variance explained morphological variables B - morphological variance explained demographic variables C - demographic variance explained by demographic variables D - demographic variance explained by morphological variables Morphology-demography TB KIT WED DC Canonical axes I II I II I II I II Correlation coefficient .9825 .8575 .9999 .9863 .9997 .9994 .9877 .9340 Redundancies A .1508 .0672 .0835 .1308 .1520 .1069 .1374 .0791 B .1456 .0494 .0835 .0173 .1519 .1067 .1340 .0690 C .1991 .0848 .0991 .1260 .1371 .1650 .2897 .1030 D .1992 .0626 .0991 .1226 .1370 .1648 .2826 .0898 Correlation to PCA axis morphological axis 1 .3798 -.2201 .5076 .0134 -.1822 .0467 II .7712 .1478 .0057 .5531 .6344 .0103 III -.0557 .2821 .1840 .4581 .0155 .7049 demography axis 1 .6945 -.1037 .5790 .7305 -.1548 .2090 II .6775 .3332 .2872 .2252 .9251 -.1157 III .0343 -.0985 -.2509 .2110 .1417 .1108 Phenology-demography A - phenological variance explained phenological variables B - phenological variance explained demographic variables C - demographic variance explained by demographic variables D - demographic variance explained by phenological variables Phenology -demography TB KIT WED DC Canonical axes I II I II I II I Correlation coefficient Redundancies .7147 .6665 .8825 .7804 .8516 .8287 .9551 .7574 A .5280 .1616 .2839 .1933 .3886 .1721 .1033 .2823 B .2697 .0718 .2211 .1178 .2818 .1182 .0943 .1619 C .1046 .1240 .0672 .1287 .0763 .1698 .0980 .1198 D .0534 .0551 .0524 .0784 .0554 .1167 .0894 .0687 160 Appendix VI. WMDS relationships between morphological, phenological and demographic characters for TB, KIT, WED and DC and soils, frequency and presence/absence of plant neighbour matrices. TB -morphology-frequency of plant neighbours relationships. Stress .195 r-square .828 Dimension I II III Importance .4348 .2016 .1047 Weights morphology .9101 .0403 .0389 frequency .2032 .6338 .4559 WED -morphology-frequency of plant neighbours Stress .169 r-square .815 -Dimension I II III IV Importance .3228 .2300 .1293 .1330 Weights morphology .7894 .0868 .1204 .5154 frequency .1500 .6726 .4940 .0193 TB - phenology-presence/absence of plant neighbours Stress .161 r-square .888 Dimension I II III Importance .5146 .2189 .1543 Weights phenology .9970 .0126 .0073 presence/absence .1876 .6615 .3555 WED - phenology-presence/absence of plant neighbours .150 Stress Dimension Importance Weights phenology presence/absence .5576 .9963 .3502 r-square II .2331 .0142 .6826 .773 III .0920 .0035 .4289 WED - demography-soils Stress Dimension I Importance .4604 Weights demography .0182 soils .9594 0.112 .2513 .7086 .0208 r-square III .1994 .6307 .0308 0.958 IV .0467 .2872 .1042 161 KIT - demography-frequency of plant neighbours .245 I .3228 Stress Dimension Importance Weights morphology phenology .7959 .1102 r-square II .8087 .0048 .8087 .807 III 0.0 .5607 0.0 WED - demography-frequency of plant neighbours Stress Dimension Importance Weights demography frequency .179 I .4836 .9817 .0580 r-square II .2488 .0256 .7044 .851 III .1187 .0277 .4868 KIT - morphology-demography Stress .209 Dimension I Importance .4658 Weights morphology .0514 demography .9639 r-square II .4217 .9182 .0169 .887 WED - morphology-demography Stress .136 Dimension I Importance .4810 Weights morphology .0736 demography .9781 r-square II .3484 .8321 .0671 .935 -III .1053 .4589 0.0 t