Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Plant succession after active layer detachment slides, in high Arctic tundra, Fosheim Peninsula, Ellesmere… Desforges, Manon 2001

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

Item Metadata

Download

Media
831-ubc_2001-0019.pdf [ 8.28MB ]
Metadata
JSON: 831-1.0089829.json
JSON-LD: 831-1.0089829-ld.json
RDF/XML (Pretty): 831-1.0089829-rdf.xml
RDF/JSON: 831-1.0089829-rdf.json
Turtle: 831-1.0089829-turtle.txt
N-Triples: 831-1.0089829-rdf-ntriples.txt
Original Record: 831-1.0089829-source.json
Full Text
831-1.0089829-fulltext.txt
Citation
831-1.0089829.ris

Full Text

P L A N T SUCCESSION AFTER ACTIVE L A Y E R D E T A C H M E N T SLIDES, IN HIGH ARCTIC TUNDRA, FOSHEIM PENINSULA, E L L E S M E R E ISLAND, C A N A D A by M A N O N DESFORGES B.A. , University of Ottawa, 1990 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Geography We accept this thesis as conforming to the required standard UNIVERSITY OF BRITISH C O L U M B I A December 2000 ©Manon Desforges, 2000 UBC S p e c i a l C o l l e c t i o n s - T h e s i s A u t h o r i s a t i o n Form I n p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f t h e r e q u i r e m e n t s f o r an advanced degree a t t h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , I a gree t h a t t h e L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p u r p o s e s may be g r a n t e d by t h e head o f my department o r by h i s or h e r r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department o f The U n i v e r s i t y o f B r i t i s h C olumbia Vancouver, Canada Date , J k a 2oeo Abstract Vegetation succession patterns and processes were studied after active layer detachment slides. These natural disturbances were grouped into four age categories in the valleys Hot Weather Creek and Big Slide Creek on the Fosheim Peninsula, Ellesmere Island (80°N). A time-space substitution was used to build a surrogate revegetation sequence in order to evaluate primary succession in the scar and secondary succession in the toes of the slides. Vegetation cover and environmental characteristics were measured using haphazard sampling along transects laid across the slide, during the summer of 1994. TWLNSPAN and detrended canonical correspondence analysis were used together to examine vegetation patterns in relation to environmental variables. Sexual reproduction effort was measured from harvested seed and seed bank samples to assess it role as one mechanism driving plant succession. The change in viable seed production for species of two life history groups (ruderal and late-sere grasses and forbs) and the change in seed bank density and composition were examined in relation to the above-ground vegetation and terrain age. Variation in species composition was accounted for by a combination of terrain age and environmental factors. In the scar, where environmental conditions improved over time, primary succession was directional with eventual replacement only in the oldest terrain. The succession followed four main stages of dominance: ruderal grass and forb -> late sere grasses and forb -> shrubs, late sere forbs and grasses -> shrubs and cushion plants. In the toe, which experienced and reduction in soil moisture over time, a retrogressive succession was observed with reduced density and diversity of species, and shift to species better adapted to surviving drought. n Large seed production by ruderal species in all terrain ages of the scars suggest that the absence of these species in the oldest terrain does not result from a reduction of resource allocation to reproductive efforts due to possible inter-species interaction. However, the slight reduction of seed production in the more severe environment of the aging toes indicates the important role of physical environmental factors. Total seed bank densities were large and more comparable to temperate environments or disturbed habitats, confirming the importance of sexual reproduction efforts in these ecosystems. Dissimilarity between the germinable seed bank and the extant vegetation composition throughout the successional sequence was expressed by the large number of seeds of late sere grasses in the young scars and the large number of seeds of ruderal species in the old scars. This suggests that seed dispersal occurs at a scale beyond the local vegetation cover, most likely as a result of winter seed rain. This study agrees with succession models that incorporate levels of environmental severity. In habitats with low environmental stress, patterns of community change, with eventual species replacement, were possibly governed by the same sort of processes described in more temperate environments and classical models of succession, but with a much slower rate of change. However, with increases in environmental stress, succession shifted from species replacement to species establishment and survival. in Table of Contents Abstract ii Table of Contents iv List of Tables vii List of Figures ix Acknowledgements xii Chapter 1: Literature review and research objectives 1 1.1 Introduction 1 1.2 Research objectives 3 1.3 The theory of succession 5 1.4 Succession models for marginal environments 7 Chapter 2: Study sites, measurements, and methods 11 2.1 Study area 11 2.2 Active layer detachment slides 13 2.2.1 Definition 13 2.2.2 Time-space substitution 15 2.2.3 Age categories of the active layer detachment slides 15 2.2.4 Specific study sites 19 Chapter 3: Plant Succession - Vegetation Patterns 23 3.1 Introduction and Rational 23 3.2 Material and methods 25 3.2.1 Study site 25 3.2.2 Vegetation and Environmental measurements 25 3.2.3 Additional Environmental Variables 28 3.2.4 Statistical Analysis 30 3.2.4.1 Community Classification 30 iv 3.2.4.2 Ordination Analysis/Vegetation pattern and vegeation-environment relation 31 3.2.4.3 Environmental conditions over time 33 3.3 Results 33 3.3.1 Community classification - Vegetation association 33 3.3.2 Community types 40 3.3.3 Patterns of primary succession 43 3.3.3.1 Species richness: 54 3.3.4 Primary vs. Secondary succession 55 3.3.5 Soil characteristics on terrain of increasing age 66 3.3.5.1 Soil moisture 66 3.3.5.2 Depth to permafrost 72 3.3.5.3 Soil texture 75 3.3.5.4 Soil pH and conductivity 75 3.3.5.5 Soil temperature 76 3.4 Discussion 1 3.4.1 Patterns of succession 80 3.4.2 Application to succession models 83 3.4.2.1 Processes controlling succession 84 Chapter 4: The role of sexual reproduction in plant succession in high arctic ecosystems.. 88 4.1 Introduction and rational 88 4.2 Study area 91 4.3 Material and methods 91 4.3.1 Seed harvest 91 4.3.2 Seed bank 93 4.4 Results 100 4.4.1 Seed Harvest 100 4.4.2 Germinable seed bank 107 4.4.2.1 Seed bank change over time, in composition, density and diversity 107 4.4.2.2 Comparison of seed bank with above ground vegetation 113 4.4.2.3 Factors explaining the variation in seed bank density and diversity 114 v 4.5 Discussion 118 4.5.1 Seed harvest and viability 118 4.5.1.1 Effects of time, life history groups and succession type (primary vs secondary) on seed viability 118 4.5.1.2 Rate of germination 120 4.5.2 Seed bank 121 4.5.2.1 Change over time and comparison with above ground vegetation 124 4.5.3 Factors controlling seed bank densities 126 Chapter 5: Conclusions 128 References cited 132 Appendices 141 vi List of Tables Table 2.1: Morphological features of the surveyed slides for BSC. Measurement units for slide length, width and depth are in meters, and in degrees for slope angle 21 Table 2.2: Morphological features of the surveyed slides for HWC. Measurement units for slide length, width and depth are in meters, and in degrees for slope angle 22 Table 3.1: Environmental factors measured for each zone of the slides and control communities: 26 Table 3.2: Distribution of slides in each age category by community groups at the second level of division in TWTNSPAN 34 Table 3.3: Summary of TWTNSPAN sample groups. Numbers are the average of all samples in the TWTNSPAN group and represent cover classes and also corresponds to the cut-level used: *=<0.1%, 1=0.1-0.5%, 2=0.5-1%, 3=1-3%, 4=3-7%, 5=7-20% '. 36 Table 3.4: Summary of ordination statistics for a) the data set combining HWC and BSC quadrats of the scars and b) HWC data set with the quadrats for the scars and toes of the slides 44 Table 3.5: Weighted correlations between the environmental variables and D C C A axes 1 and 2 for the combined data set of HWC and BSC (weight = sample total) 45 Table 3.6: Weighted correlation between the environmental variables for the combined data set of HWC and BSC 45 Table 3.7: Number of samples for each disturbance type at HWC by age category. Each sample represents the average of quadrats, for zones of homogeneous environmental conditions within the slide of control community 55 Table 3.8: Distribution of slides by community group at the second level of division and terrain age classes for the HWC data set 57 Table 3.9: Distribution of slides by community group at the third level of division and terrain age classes for the HWC data set 57 Table 3.10: Weighted correlations between the environmental variables and D C C A axes 1 and 2 for the HWC data set only (weight = sample total) 62 Table 3.11: Weighted correlation between the environmental variable for the HWC data set only 62 vii Table 3.12: Mean (SD) air and soil surface temperature differences between young and old disturbances and between scar and control environments. Averages represent hourly temperature measurements recorded from June 26 t h to July 5 t n, 1994 77 Table 4.1: Number of sites (slides) by age category and geographic location from which seed bank samples were taken 93 Table 4.2: Seed bank species included in the four groups 95 Table 4.3: Percent germination of harvested seeds a) by life form group and b) by individual species within the ruderal group 100 Table 4.4: List of species found in the germinable seed bank samples and total number of seedlings counted by species and for each geographic location 108 Table 4.5: Comparison of germinable seed bank for disturbed and undisturbed communities, at HWC and BSC 110 viii List of Figures Figure 2.1: Study area of the Fosheim peninsula on Ellesmere Island locating the three study sites of 1) Black Top Creek, 2) Hot Weather Creek and 3) Big Slide Creek 12 Figure 2.2: Sections within a typical active layer detachment slide, and schematic example of the transects used for sampling 14 Figure 2.3: Active layer detachment slides at Hot Weather Creek 16 Figure 2.4: Active layer detachment slides at Big Slide Creek. 17 Figure 3.1: Community classification dendrogramme derived from TWLNSPAN, for the combined data sets of HWC and BS 35 Figure 3.2 a-f: Mean (+SE) environmental conditions for TWINSPAN community groups for the HWC and BSC combined data set 37 Figure 3.3: Distribution of the terrain ages in each TWINSPAN community group 39 Figure 3.4: D C C A species ordination for the combined vegetation data set, with environmental variables represented by arrows 48 Figure 3.5: D C C A ordination diagram showing the TWINSPAN group membership with environmental variables represented by arrows. The outlines delimit the spread of the groups at the second level of division 49 Figure 3.6: D C C A ordination showing TWINSPAN sample group centroids labeled with their community type (see section 3.3.2 for the community descriptions). The bars are 95% confidence intervals. The ellipses represent the 95% confidence interval for the age category centroid of the HWC and BSC data sets 50 Figure 3.7: Changes in vegetation cover (%) over time, for individual species 53 Figure 3.8 Species richness comparing HWC and BSC location 54 Figure 3.9: Community classification dendrogramme derived from TWINSPAN, for the data set of HWC 56 Figure 3.10a-j): Mean (+ SE) environmental conditions for each TWINSPAN community group 58 Figure 3.11: D C C A ordination of the HWC data set showing TWINSPAN sample group centroids labeled with their community type 63 ix Figure 3.12: D C C A species ordination for the HWC data set, with environmental variable scores represented by arrows 64 Figure 3.13: D C C A species scores for the HWC data set, averaged by disturbance type (S=scar T=toes, C=Control, Cx=Control in exposed conditions) and age category (l=Young, 2=Intermediate age, 3=01d 4=Very old 5=Undisturbed controls). The ellipses represent 95% confidence intervals. The centroids of the TWINSPAN community groups are drawn as white box 65 Figure 3.14: Changes in soil moisture for selected active layer detachment slides (ALDS) throughout the growing season 68 Figure 3.15: Relation between soil moisture index classes and total vegetation cover 69 Figure 3.16: Mean (+ SE) soil moisture (a) and permafrost depth (b) for disturbances of 2 age categories (young and old) and 2 succession types (primary = scars, secondary = toes) derived from the larger number of A L D S sampled on July 15th 70 Figure 3.17: Changes in active layer depth throughout the growing season for the two succession types (primary = scars, secondary = toes) 73 Figure 3.18: Relation between active layer depth and absolute soil moisture at 5 cm depth, for 4 dates during the growing season 74 Figure 3.19: Soil texture for the scars and associated undisturbed community for three age categories: a) all samples plotted; b) Averaged by disturbance type (S=Scar, Cntr=Associated undisturbed controls) and age categories (l=young, 2=intermediate, 3=old) 78 Figure 3.20: Soil pH for young, intermediate and old scar, compared to their respective undisturbed communities 79 Figure 3.21: Soil conductivity for young, intermediate and old scar, compared to their respective undisturbed communities. Soil conductivity was used as an indicator of soil salinity 79 Figure 4.1: Germinable seed bank at Big Slide Creek, comparing samples from a) scar and scar edge and b) from the exposed dry undisturbed controls and the more moist undisturbed controls 99 Figure 4.2: Germinable seed bank at HWC, comparing samples from scar and scar edge over time 99 Figure 4.3: Percent of germinated seeds from the seed head harvest by age categories and life history groups. Age categories are: y = young; i = intermediate; o = old; undist.= undisturbed 102 x Figure 4.4: Germinated seed by individual species for the Ruderal life form group. Age categories are: y = young; i = intermediate; o = old; undist.= undisturbed 103 Figure 4.5: Relation between germinated seeds and vegetation cover, a) Total seeds harvested by site, b) Seeds of the two life groups 105 Figure 4.6: Rates of germination for individual species by age category and disturbance type (Scar,Toe) 106 Figure 4.7: Mean (+SE) germinable seed bank and species diversity in relation to age category for a) Hot Weather Creek and b) Big Slide Creek locations. Age categories include Y=young; M=intermediate aged; 0=old; VO^very old; Undist=undisturbed control terrain. As well, for each age category, the relative importance of the 4 functional type is shown (rud gr = ruderal graminoids; late gr= late sere graminoids; rud forb = ruderal forbs; late forb = late sere forbs) I l l Figure 4.8: Mean (+/- SE) germinable seed bank in scars and undisturbed tundra for a) Hot Weather Creek and b) Big Slide Creek 112 Figure 4.9: Mean (+/- SE) germinable seed bank and total vegetation cover by age category for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the separate data set of b) HWC only and c) BSC only 115 Figure 4.10 Species density (mean number of species +/- SE) in the germinable seed bank and the corresponding vegetation cover by age categories for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the separate dataset of b) HWC and c) BSC only 116 Figure 4.11: Germinable seed bank in relation to plant cover for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the two geographic locations separately: b) HWC and c) BSC only 117 Figure 4.12: Distribution patterns of the two life form strategies (ruderal vs late sere) over time for both the seed bank and the extant vegetation. This is the result of the combined dataset for the two geographic location and the two functional types (monocotolydons and dicotolydons) 125 xi Acknowledgements A Marek mon fils, je te souhaite de toujours savoir trouver en toi le courage de perseverer dans tes buts et l'accomplissement de tes reves. Je te souhaite d'etre explorateur et de decouvrir autant que les grands explorateurs de l'Arctic, que ce soit dans l'espace ou la decouverte de l'autre. I owe many thanks to those who supported and encouraged me throughout so many years. Many thanks to Marc Boulerice, who sacrificed his knee sowing seeds and collecting more seed bank samples in the field, but mostly to the best friend who never stopped believing in me. I look forward to many more years of exploring and growing together. Much appreciation to Greg Henry for his inspiration, guidance and patience with "duh french", and constant sense of humour. Au Nord Greg! Thank you to Tony Lewkowicz for his personal maps and air photos of Hot Weather Creek and Big Slide Creek and to U B C Botanical Gardens Nursery for the greenhouse space. Thank you to the friends who lived this with me: the geo500 group who sweated together over that first year; Wendy Hales in particular who dried many tears of discouragement and joy; Kate and Elizabeth for the stress relief talks at the kitchen counter; Katherine and family who provided a room during trips down from Whitehorse; Glenda whom I guided in the first years and whom I followed when writing this work; Ross McLachlan for his humour and financial help with many jobs over the years! But most of all, thank you to my parents for providing a home of love and security and the best years of my life at the cottage where this love for nature began. To my father who gave me that desire to explore and discover new words. To my mother who mirrored me in emotions and thoughts and always understood and listen to my every joy and sadness. Merci papa, maman! Research was funded by a Natural Sciences and Engineering Research Council of Canada (NSERC) grant to G. Henry, with logistic support from the Polar Continental Shelf Project of Natural Resources Canada. Funding for my term of study at U B C was provided by an NSERC Post Graduate Scholarship. xii Chapter 1: Literature review and research objectives 1.1 Introduction Vegetation succession has long been recognised as an important concept for understanding ecological systems, and for improving our ability to predict or control the impact of disturbances (natural or human) in the landscape. High arctic ecosystems, because of their low species diversity, are particularly sensitive ecosystems, susceptible to damage or change due to small or large scale events, such as industrial expansion and global warming. Unfortunately, succession patterns and processes are poorly understood in these environments. High arctic ecosystems are characterised by limited heat energy, short growing seasons, poor soil nutrient and low water availability (Chapin and Shaver, 1985; Chapin, 1987; Svoboda and Henry, 1987; Edlund and Alt, 1989; Shaver and Kummerow, 1992). This generally results in low productivity and sparse or intermittent vegetation cover, which is in contrast to many subarctic and temperate environments, from which most models of succession are derived. A number of ecological studies have been conducted in the Arctic, but many addressed revegetation, recolonisation and restoration of low arctic tundra following disturbances such as those resulting from the oil and gas exploration and exploitation (Bliss and Wein, 1972; Chapin and Chapin, 1980; Walker and Walker, 1991; Forbes,1993). Although hypotheses regarding successional processes were generated from these studies (Walker & Chapin III, 1987), and known succession theories were tested using these tundra restoration projects (Cargill and Chapin, 1987; Denslow, 1980), it remains that low arctic ecosystems differ from high arctic ones. Thicker organic soil development and continuous vegetation cover are typical of subarctic environments, and vegetation recovery proceeds at higher rates there than in the High Arctic (Hernandez, 1973). Consequently, 1 succession theories derived from these environments are not necessarily valid in the High Arctic. In the High Arctic, many vegetation studies have focused on individual species and their responses to environmental manipulations (Chapin and Shaver, 1985; Henry and Molau, 1997; Robinson et al, 1998; Johnstone and Henry, 1997; Arft et al, 1999). Most ecological studies at the community level have focused predominantly on describing the local plant assemblage, community composition and spatial patterns in relation to environmental characteristics (Bliss et al, 1984; Sohlberg and Bliss, 1984; Bergeron and Svoboda, 1989; Muc et al, 1994; Bliss et al. 1994). Very few studies of succession in the High Arctic are available, and most of these focused almost exclusively on secondary succession and natural recovery of vegetation following anthropogenic disturbances (Babb and Bliss, 1974; Barrett and Shulten, 1975; Forbes, 1993a, 1993b)). These studies of secondary succession typically occurred at small spatial and temporal scale or were conducted in very productive sedge moss tundra, which represents only a small portion of the High Arctic. Primary succession has rarely been observed in the High Arctic because of the slow rate of revegetation in such nutrient poor habitats (Prach, 1993). Revegetation on dated historical disturbance sites can, however, provide surrogate data for primary succession studies in such areas of slow revegetation. Jones (1997), for example, studied the patterns and processes of succession in the glacial foreland of the Twin Glacier at Alexandra Fjord, Ellesmere Island, and Bliss and Gold (1994) evaluated effects of cyanobacteria on the initiation of succession at Truelove Lowland on Devon Island, Nunavut. Unfortunately, the paucity of dated disturbed land has resulted in fewer studies of primary succession and of the processes controlling successional change in the High Arctic, than in other biomes (Bliss and Peterson, 1992). 2 On the Fosheim Peninsula of Ellesmere Island, as in most of the western Queen Elisabeth Islands, active layer detachment slides scar much of the landscape. Many of these slides on the Fosheim Peninsula have been mapped and dated using aerial photography and historical climatic records (Lewkowich ,1992). This offers an excellent surrogate revegetation sequence and thus a good opportunity to contribute more quantitative data on the patterns and processes of both primary and secondary succession in a high arctic environment. The Fosheim Peninsula is a polar oasis that contains a wide variety of plant communities, including large areas defined as polar semi-deserts according to Bliss's classification (1979). Although vegetation is sparse in these areas and the communities are less productive than those of the tundra sedge and grass meadow environments, polar semi-deserts support a range of wildlife including Peary caribou, muskox, wolves, lemmings, weasels and several species of nesting birds. The susceptibility of the High Arctic to environmental change and the importance of the polar semi-desert within the High Arctic warrants further understanding of this ecosystem. 1.2 Research objectives The first objective of this research is to describe the pattern of vegetation succession and environmental change in a high arctic ecosystem after a natural disturbance: active layer detachment slides (chapter 3). The ultimate purpose is to see i f succession in this ecosystem operates in a manner similar to other ecosystems from which various successional models have been derived. This will be done by: 1) evaluating the changes in vegetation communities in disturbances of increasing ages; 3 2) describing the main environmental factors associated with the vegetation community changes occurring over time; 3) investigating the effect of disturbances on the vegetation of the polar semi-desert landscape of the Fosheim peninsula. The second objective is to assess the role of sexual reproductive effort as one mechanism driving plant succession (chapter 4). Ultimately, succession is a result of fluctuations between 1) invasion, 2) establishment and survival, and 3) death of established species. Studying sexual reproduction dynamics in these ecosystems allows for investigation of the invasion and colonisation potential within the succession equation. The role of sexual reproduction dynamics will be studied by: 1) evaluating how the germinable seed pool available in the soil (seed banks) changes over time, and how it relates to above ground vegetation patterns and environmental conditions; and 2) exploring how the viability of seeds from different life groups harvested at the end of a growing season changes throughout the successional sequence, and how this change compares to the changes in community composition. Before presenting these first 2 objectives, the study sites and the method of time-space substitution will be presented and discussed in Chapter 2. The disturbance investigated permitted the evaluation of both primary and secondary succession in a High Arctic ecosystem (further discussed in chapter 2). The uniqueness of this study is that primary succession was evaluated in a large vegetated landscape. Most primary succession studies have focused on shorelines or glacial forelands. These 4 environments may not lend themselves as well to generalisation since they are favoured by higher moisture availability. Although the exposed surfaces use to study primary succession on the Fosheim Peninsula were depleted of vegetation and seeds following active layer detachment slides, it is recognised that the slide areas may not strictly represent primary succession, since potential seed sources are not far away, given the proximity of the surrounding vegetation. The remainder of this chapter will outline the various theories and models of succession to which the results of this research will be compared. 1.3 The theory of succession Vegetation communities exist in a dynamic state of continuous change in response to varying environmental and biological conditions. Clements (1916), who first defined the term succession, viewed the vegetation system as a superorganism in terms of growth, development and reproduction. Succession, in his interpretation, was solely sequential and directional, culminating in an inevitable fixed community assemblage he called climax. This holistic view was later criticised by Gleason (1926) and others (such as Margalef, 1963; and Odum, 1971), who emphasised the individualistic nature of plants, suggesting that ecosystems were no greater then the sum of their parts. Over the last 50 years, many studies of succession have been published. Picket et al. (1987) established a good set of working definitions separating the study of succession into 3 categories: pathways, mechanisms, and models of succession. Successional pathways correspond to the temporal patterns of vegetation change, while mechanisms of succession refer to the processes or interactions that contribute to the successional change. A model of 5 succession is "a conceptual construct to explain successional pathways by combining various mechanism and specifying the relationship among mechanisms and the various stages of pathway" (Picket et al., 1987). Most studies and theories address one or two of these categories, and criticisms often refer to the confusion resulting from the lack of proper terminology. Egler (1954) summarised patterns of succession as being either Relay Floristic (RE), corresponding to the classical model of succession or, alternatively, resulting from Initial Floristic Composition (IFC). In this alternative pattern, he argues that all species are present within the initial flora, and that vegetation development advances through a sequence of changes in species dominance determined by the their life-history characteristics such as growth rates. Following this trend toward individual-based models, Huston and Smith (1987) simulated patterns of succession based on species interactions occurring at the individual level, and discovered five possible outcomes: patterns of replacement, divergence, convergence, suppression, and pseudo-cyclic replacement. Numerous mechanisms have been proposed to explain observed patterns of succession, including competition (Grime, 1977; 1979; Tilman, 1985), facilitation / inhibition (Connell and Slatyer, 1977), life history characteristics (Egler, 1954; Drury and Nisbet, 1973; Grime, 1977; Huston and Smith, 1987), resource allocation ratios (Tilman 1985), vital attributes (Noble and Slatyer, 1980), and stochastic interactions between invasion, maintenance and senescence of species (Johnstone, 1986). The relative importance of these mechanisms will change relative to one-another depending on the degree of environmental severity (Svoboda and Henry, 1987; Walker and Chapin, 1987; Matthews, 1992; Chapine? al., 1994). Although competition is considered a chief agent of classical succession in low stress environments, Grime's model (1977) suggests that in marginal habitats the path of 6 succession goes almost directly from the ruderal to the stress-tolerant type of life-form, with much reduced evidence of competitor plant strategies. Under severe conditions with extremely low productivity, he even postulates that the path of succession would lead from small stress tolerators to larger ones, in such a way that the ruderal phase essentially disappears. Potential biological interactions are important in all habitats, yet in severe environments where species survive at the extreme extent of their habitat range, inherent environmental conditions may be more critical than any other successional controls. Consequently, other models and hypotheses have been proposed which integrate various degrees of environmental severity, such as those found in marginal habitats. 1.4 Succession models for marginal environments Svoboda and Henry (1987) first proposed an alternative model of succession, which integrated levels of environmental severity. They support the life-history approach, believing that species in severe environmental conditions depend largely on their individualistic strategies for survival, since the sparse and intermittent distribution of individuals would rarely cause any "ecological warm-up". Although most classical succession theories assume that the environment is passive and open to invasion, in their perspective it can be the limiting and resistant factor. Factors of environmental resistance (ER), such as low nutrient availability, drought, hard soil, low temperatures, and wind or water removal of seeds, discourage the invasion and establishment of individuals, or obstruct population expansion. Opposing these factors are the biological driving forces (BDF), which they define as the intrinsic (biological) agents, including species' life-history traits such as seed production and germination, which would allow species to establish and survive in intermittently adverse 7 conditions. It is the balance between these two forces that determines the rate and the direction of succession. Successional change will occur when the sum of the biological driving forces exceeds the sum of environmental resistance. Based on this, they proposed three alternative models and patterns of succession. The Directional - replacement model occurs in low resistance environments where BDF is much greater than ER, typical of temperate regions. This corresponds to the classical models of succession, similar to that of Egler's (1954) Relay Floristics where species replacement occurs due to competition or facilitation by the previous set of species. The Directional - nonreplacement model characterises high resistance environments (BDF > ER), typically of semi-polar deserts. Here, slow expansion of invading species occurs without displacement of earlier established species. And finally the Nondirectional - nonreplacement model is observed in extremely resistant (marginal) environments (BDF<ER) such as the polar desert. Under such conditions, relatively few species manage to establish even in favourable years, yet those that do are generally rewarded with long survival, with increasing or decreasing biomass in good or bad climatic years. In extremely adverse environments, invasions often fail to result in establishment and successional processes can come to a standstill or the community is subjected to what they termed "retrogression". Matthews (1992) presented similar concepts in his geoecological model where he integrates physical environmental processes with biological processes. He also emphasises that most models provide an unsatisfactory treatment of role of physical environmental processes in succession, pointing out that the environment can not be describe solely in terms of an unchanging set of initial site conditions. Succession, therefore, can not be regarded 8 purely as a biological phenomenon. Similar to Svoboda and Henry's (1987) work, his hypothesis is based on the general agreement that succession is determined by two driving processes: allogenic changes in the physical environment (externally driven processes) and autogenic changes in the biological processes (internally or community driven) (Muller, 1952; Billings and Mooney, 1968; Whittaker, 1991; Whittaker, 1993; Helm and Alle, 1995). Matthews' proposal differs from Svoboda and Henry's (1987) in that the physical allogenic processes are not strictly resistant to the intrinsic biological forces, but can also reinforce or facilitate the vegetation changes driven by the biological processes. In his model the relative importance of allogenesis (physical) and autogenesis (biological) changes during succession. Allogenesis is more important early in succession and decreases with time, while autogenenis increases with vegetation cover and biomass, thus with late successional stages. He also models the relative importance of these with different degrees of severity. In favourable environments (high resources and low stress) there is a rapid increasing importance of autogenic processes with time, and, therefore, a rapid accumulation of biomass. In unfavourable or severe environments there are more constraints on biological production and the allogenic processes are more important at any stage of succession. In the most severe environments, autogenic processes may be so weak that allogenic processes remain more important indefinitely. These three scenarios correspond to Svoboda and Henry's (1987) pattern of directional-replacement, directional-nonreplacement and nondirectional-nonreplacement, respectively. Although these models were concerned largely with the form rather than the mechanisms of succession, they contribute towards an explanation of the range of possible successional patterns. 9 Walker and Chapin (1987), while also incorporating levels of environmental severity in their succession models, focused on individual successional processes. The value of their work was to produce a framework of hypotheses that predicts the relative importance of the processes at different stages of succession, in different type of succession (primary vs secondary), and in favourable or severe environments. They illustrated the pathways of succession at two levels of environmental severity for the following processes: stochastic events, facilitation, competition, life history (growth rate, longevity, mycorrhizae) and herbivory. In severe environments, stochastic events wil l predominantly affect the colonisation stage of succession. Facilitation appears most important in severe environments during the stages of colonisation and early community development. Competition is not as critical in marginal environments, and potential growth rates will have the largest impact where abundant resources are available to support rapid growth. Longevity is comparable in both favourable and severe environments. Mycorrhizae are important in all stages of succession developing in severe environments. Walker and Chapin (1987) concluded, however, that many processes act simultaneously at any given moment, and consequently the successional pathway may be unpredictable or even appear stochastic. Despite the results of these studies, there is still a paucity of research regarding succession in high resistance environments. Further testing of these theoretical models is required in order to better understand patterns and processes in marginal environments and ultimately to better understand the concept of succession. 10 Chapter 2: Study sites, measurements, and methods 2.1 Study area This research was conducted on the Fosheim Peninsula on the western coast of Ellesmere Island during the summer of 1994. The Fosheim Peninsula frequently experiences unusually high summer temperatures reaching up to 20°C (+5.4°C mean July temperature), due to its central position between the mountain ranges of Axel Heiber Island and Ellesmere Ilsland. These act as a barrier, sheltering the region from the prevailing cold northwesterly winds from the Arctic Ocean (Edlund and Alt, 1989; Harris and Lewkowicz, 1993). The mature vegetation community is characterised by a Salix-Dryas hummucky tundra covering often less than 20% of the land surface (Edlund et al, 1989). The dominant flowering plants include Papaver radicatum, Alopercurus alpinus, Poa and Puccinelia species, Cerastium arcticwn, Luzula confusa and L. nivalis, and numerous species of Draba and Saxifraga. The woody species of Salix arctica and Dryas integrifolia are typically restricted to the warmest sites. The area falls within the continuous permafrost zone, and the soils are characterised by fine grain silty/clay sediments. The relatively high summer temperatures, the low plasticity of the sediments, and the elevated soil water content, due to melting permafrost, make the surface susceptible to failures. Active layer detachment slides have been recorded in numerous high arctic locations (French, 1976; Hodgson, 1977; Stangl et al. 1982; Mathewson and Mayer-Cole, 1984; Edlund and Alt, 1989; Lewkowicz, 1990; 1992; Harris and Lewkowich, 1993) and are an important geomorphological process which modifies and shapes the terrain, and thus influences the plant communities. 11 Figure 2.1: Study area of the Fosheim peninsula on Ellesmere Island locating the three study sites of 1) Black Top Creek, 2) Hot Weather Creek and 3) Big Slide Creek. 12 There have been few studies of succession in high arctic tundra partly due to the paucity of datable disturbances (Bliss and Peterson, 1992). However, active layer detachment slides in three valleys of the Fosheim Peninsula have been mapped and grouped into age classes (Lewkowicz, 1992), creating an opportunity to study high arctic succession. For this research, two valleys were chosen for detailed investigation: 1) Hot Weather Creek (HWC) (79°42'N, 84°23'W) and 2) Big Slide Creek (BSC - unofficial name) (79°42'N, 84°23'W). A third site just outside of Eureka, named Black Top Creek (79°58'N, 85°40'W), was visited for complimentary data but not investigated in detail (Figure 2.1b). 2.2 Active layer detachment slides 2.2.1 Definition Active layer detachment slides can be defined as rapid mass movement of unfrozen surficial material that occurs on permafrost slopes following a period of continuous warm temperatures. The unfrozen mass detaches from the underlying frozen substrate due to increased pore water pressure, and slides downslope on top of the underlying ice-rich material, the upper part of the permafrost (Lewkowicz, 1990; 1992). Within a typical slide different sections (or zones) can be categorized: 1) a scar zone, exposed when the material moves downslope, 2) a track zone along which the material travels, and 3) a toe zone where the transported material accumulates (Figure 2.2). In this study, both the scar and track zones were grouped into one unit and were collectively termed the scar zone. 13 Figure 2.2: Sections within a typical active layer detachment slide, and schematic example of the transects used for sampling. The distinction between these zones is necessary for two reasons. First, because the material moves by sliding and not flowing, the newly exposed area in the track zone is depleted of soil, seeds and other propagules. Therefore, the scar zone provides surfaces to study primary succession. Revegetation proceeding in the toe environment, on the other hand, will reflect secondary succession. Though the vegetation is disturbed, the soil, with its seeds, still remains. The second interesting component of these disturbances and the two zones (scar and toe) is that they produce different topographic changes. The scar creates depressions in the landscape and consequently concentrates drainage toward them, creating moist habitats relative to the surrounding tundra. The accumulation of material in the toe, on the other hand, results in slightly more elevated terrain, which favours drainage and creates moisture-depleted environments over time. 14 2.2.2 Time-space substitution As it is common in ecological studies, a time-space substitution was performed in order to study succession. Thus, the succession pattern was reconstructed from disturbances of various ages rather than from the observation of one disturbance over time. This is common practice in plant ecology and geomorphology where description of long term changes are based on observations of spatial sequences (Drury and Nisbet, 1973; Barbour et al. 1987). The slides investigated were therefore grouped by age class. However, conclusions on succession patterns must be drawn with caution, as the observed changes in vegetation community may be the result of spatial differences rather than temporal differences. 2.2.3 Age categories of the active layer detachment slides The slides were grouped into relative age classes based on Lewkowicz's (1992) work. He defined three age classes (young, intermediate and old) using a combination of ground surveys and sequential vertical and oblique aerial photography (Figure 2.3 and 2.4). Actual rates of succession were not evaluated in this study, since these classes represent a range in time, although changes in species and community composition can be monitored within and between age classes. Specific dating of individual slides could not be done, as was hoped, with internode counts of Salix arctica, Cassiope tetragona, or Saxifraga oppositifolia, since none of these species were found on young or intermediate aged slides, and only occasionally on old slides. The young age category includes exclusively those slides corresponding to a unique episode which occurred in 1988. Consequently, all slides of the young category were 6 years 15 Figure 2.3: Active layer detachment slides at Hot Weather Creek. The maps were originally drawn using air photos at 1:14,500 by Lewkowicz (1992) and further refined during the 1994 summer field work. This map does not depict an exhaustive list of all of the slides, but only shows the failures that could be positively identified on the ground. Owing to map scale, shapes are approximate. Numbers on the map refer to the slide ID-code. These were the slides used in the analysis. 16 Figure 2.3: Active layer detachment slides at Big Slide Creek. The map originally drawn using air photos at 1:14,500 by Lewkowicz (1992) and further refined during the 1994 summer field work. This map does not depict an exhaustive list of all of the slides, but only shows the failures that could be positively identified on the ground. Owing to map scale, shapes are approximate. Numbers on the map refer to the slide ID-code. These were the slides used in the analysis. . 17 old. The summer of 1988 was characterised by an absence of precipitation and continuous above normal temperature for over a two-week period in July. This caused a rapid thaw rate and resulted in unusual slide activity. The study sites were mapped in the field in 1988 priorto the failures that took place that year and remapped in the following 2 years. No other slides were observed between 1988 and 1994. Unique climactic circumstances seemed to be a primary factor controlling the slope failures. The 1988 slides made up the largest portion of all slides investigated, for two reasons. Firstly, the exact age of the disturbance was known, which meant that all sites would have very similar initial conditions in term of climate. Secondly, a major focus of this thesis is to look at the invasion component in the processes of succession, of which the 1988 slides provided ample material. The intermediate age category corresponds to a time period between 1975 and 1988 (terrain ages between 6 to 19 years of age). Lewkowicz (1990) determined this category from examination of the Black Top Creek (BTC) area, which had the best time reference with five sets of vertical aerial photographs (taken 1950, 1959, 1974, 1982, 1986). Aerial photographs for BSC and HWC are available only in 1950 and 1959, and both sets of photos were at such a small scale that identification of distinct slides was difficult. Although no time controls were available, Lewkowicz (1992) divided the slides of these 2 sites into two subjective categories (pre- and post-1975) based on visual comparison of morphological features of the slides from BTC area where ages were know. Slides in this intermediate age category still had well-defined lateral berms and transported blocks with near-vertical sides. The old category includes slides prior to 1975, aged 20 years old or more. Of the slides that could be mapped from the 1950 and 1959 air photos, most were completely imperceptible in 1990 when Lewkowicz conducted his mapping survey. This suggested to 18 Lewkowicz that none of the mapped slides could be older than 50 to 100 years, as weathering and erosional processes such as solufluction and freeze/thaw appear to dissolve any outlines or marks on the terrain. The surface expression of the failures in this category have lateral berms and blocks that have been weathered and eroded into less distinct forms. Only one slide identified from the 1950 air photos was selected for this research, creating a very old category (> 50 years old). Again a specific date cannot be given, but this slide was included for the purpose of describing the community assemblage that can be expected in older disturbances and to assess the associated environmental characteristics of such sites. Based on the 1988 episode, it appears that slide activities occurs during episodes of unique conditions of high temperatures, which produces rapid advancement of the thaw plane, and of soil conditions with segregated ice in the basal part of the active layer (Lewkowicz, 1990, 1992). The likely episodic nature of the sliding activity is important because it increases the validity of comparing vegetation changes between age categories. More faith can be placed in comparisons conducted between age categories consisting of one or two possible slide episodes, than from categories representing a series of slides which occurred throughout the time period. 2.2.4 Specific study sites Both BSC and HWC valleys had a north/south orientation and thus, most slides had an easterly or westerly aspect (Figures 2.3 and 2.4). As well, both sites are underlain by tertiary sandstones , siltstones, shales and coal of the Eureka sound formation and the fine grained soils were typically derived from the weathering of bedrock. Unlike BSC which is found at ~200 m in elevation, HWC, at 70 m in elevation, lies below the Holocene marine 19 limit of 140 m (Hodgson, 1977). This results in the soils possessing a higher composition of fine-grained marine clay sediments at HWC. The underlying rock beds at BSC dip to the east at 10° and failures appeared more common on east-facing slopes parallel to the local dip. However, significant numbers were still counted on west facing slopes where they cut across the rock structure (Lewkowicz, 1992). The dip angle of the beds at H W C were closer to zero, such that the failures either cut across the bedrock or occurred exclusively in the overlying marine sediments. Although an attempt was made to minimise terrain variability and to chose slides of similar magnitude, the selection was limited by what was available from a reasonable walking distance around each camp. Figure 2.3 and Figure 2.4, adapted from Lewkowicz's mapping survey conducted in 1990 (personal communication, 1994), shows the distribution and selection of slides for both sites of HWC and BSC, as well as the associated age categories. A total of 42 slides were used for detail vegetation surveys: 12 from BSC and 30 from HWC (Table 2.1 and 2.2). Due to the limited field season and other time constraints, only scars were surveyed at BSC. For each slide, length and width were recorded in the field and a proportional sketch was produced. As well, the transects used from sampling were surveyed using distance and angles (measured with a clinometer), in order to draw a cross-section of the slide. This made it possible to calculate the depth or the height of the disturbance in comparison to the undisturbed terrain. The average slide dimensions were 42.8 m (+/- 18.7 m) in length and 18.7 m (+/- 7.8 m) in width for HWC location while BSC averaged 111.6 m (+/- 125.7 m) in length and 20.5 m (+/- 12.7 m) in width. Typically BSC had longer slides on shallower slopes (12.5° +/- 3°) compared to HWC. One slide at BSC, in the young category was just under 1 km in length. HWC however is steeply incised in the 20 plateau and typically had shorter failures with greater scar slope (15.8° +/- 6.9°). The thickness of material removed in the scar ranged between 70 cm to lm, and was slightly deeper at BSC. However actual measurement were recorded only for HWC, comparing the height difference between the adjacent control and the centre of the scar or toe. Scar depth averaged 165 cm (+/- 63 cm), while the average toe mound height averaged 90 cm (+/- 58 cm). The morphological features of the slides used for this study for both HWC and BSC. Table 2.1: Morphological features of the surveyed slides for BSC. Measurement units for slide length, width and depth are in meters, and in degrees for slope angle. Slide Age Aspect Slide Scar/track Name Length Width Slope Cntr. Slope BS 1 Young ENE 950 120 15 17 BS 65 Young W 70 25 12 5 BS 2 mod. ENE 400 45 10 15 BS 44 mod. NE 150 25 13 16 BS51.5 mod. E 22 12 17 17 BS 66 mod. NW 300 40 18 6 BS 69 mod. W 45 14 13 12 BS 70 mod. w 70 20 8 10 BS 49 Old ESE 20 6 10 13 BS50 Old E 16 8 14 18 BS51 Old E 85 10 10 17 BS 64 Old WNW 50 20 10 10 Average with BS 1 181.5 28.6 12.5 13.0 (+1 Standard deviation) (+258.6) (+29.9) (±3.0) (+4.3) Average without BS 1 111.6 20.5 12.3 12.6 (+ Standard deviation) (±125.7) (±12.7) (+3.1) (+4.5) 21 B O U C73 01 a. 35 JS '33 X N O O N CN CN o CN i n o OO CN co O CN >n CN CN rt r o N O rt CN O o o t ON ( S r o rt r o m i-i o M rH N O N O OO CN 2 S 2 °o CN O N CN OO OO — i o ro © i n I - H r o oo oo rt C N O C N r o - H oo oo r~-CN O rt N O N O "d- i-i N O rt O O N Ti- CN CN r o T J - O O N — i © © - H ' ^ — ; — ; O O O CN t O CN CN rt CN rt rt OO t~- NO OO CN O c Si U 35 a. 35 a. CU Q WD S • * © © , . - , 0 0 0 0 0 0 0 rtrtrt^rtrtCNCN " " l O o o o o C N C N i n c N r H r t W M r N | ( S C N | r n rt m oo O N O O N rt , _ < N C N O N O rt rt C N rt rt rt o o v O M O h M m v i e o c N C N c o r o c N r t r t O N O N O o o c C T f - r o r o C N C N T j - C N r t C s l r t r t m C N rt rt ( N N O cn © rt rt C N cn t~~ m r o O N rt rt © rt NO O O rt r o r o "d- i n rt rt C N CN rt rt CN i n N O r o C N Tf o r o O r o CN CN CN OO r o T J -rt rt C N O N O m r o r o CN rt rt CN rt CN N O r o rt <n N O r o rt CN rt a u Si, a i d S U 35 4) -M •a en 35 « r o r o © , _ r ~ - i > O N i n rtrtrt^rtrtrtCN c N i n T f o r - o o r t t - ^ o r - ^ J - m c o i n r o c o c N c N a> a. W p5 w w w w £ £ z 6 0 0 0 6 0 6 0 6 0 6 0 6 0 6 0 6 0 c a a c a a a a 3 3 3 3 3 3 3 3 o o o O o O o o o >< ^ J " CN CN rt rt CN N O o m rt T J - r-. W W W W W T 3 T 3 O O O s e a •<t o N O N O o o o CN CN CN 0 0 CN ^ O m rt o CD -t-» 0 ) 35 rt ro m VO O O CJ U O rt ^ - i n N O £ £ £ ffi w w rt r>. oo O U (J £ £ £ ffi ffi K *0 *^  *0 *0 ""O "3 "3 "3 "3 "3 "3 "3 i n i n i n O rt CN C O -ct U U CJ U u u u £ £ £ £ £ £ £ PH PP PH PH PH PH PH 22 Chapter 3: Plant Succession - Vegetation Patterns 3.1 Introduction and Rational Following a disturbance, vegetation development is initiated as plants invade the stabilised land surface. In the High Arctic, little is known about vegetation succession, its spatial expression, and the controlling mechanisms or processes. In these environments, patterns of succession as proposed by classical models of succession may not be applicable. Svoboda and Henry (1987) suggested that, as environmental stress increases, the expression of succession may shift from species replacement to a fluctuation of species establishment and survival without replacement. As well, in these landscapes where vegetation cover is sparse and intermittent, the mechanism of competition may not be an important process structuring the communities, suggesting that the main driving force of community change may shift from intra-species interactions to environment-species interactions (Svoboda and Henry, 1987; Walker and Chapin, 1987; Mathews, 1992). The main objective of this chapter is to describe the patterns of plant colonisation and community succession in a high arctic landscape after a natural disturbance: active layer detachment slides (ALDS) and, ultimately, to compare the observed pattern with existing models of succession. Patterns are deduced from the changes in total vegetation cover, species composition and diversity, and ultimately through the identification of successional stages. For this purpose, the following questions were addressed: Ql ) Are plant species organised into distinguishable plant communities? Q2) What are the dominant environmental factors associated with plant community assemblages? 23 Q3) To what extent does terrain age affect the community structure (vegetation cover, species diversity) and are successional stages identifiable? Q4) Is the pattern of succession directional with evidence of species replacement proposed by classical models, or do other models proposed for high stress environments better describe the patterns observed? Q5) Are there differences between primary (in scars) and secondary (in toes) succession, and between geographic locations? 24 3.2 Material and methods 3.2.1 Study site Two geographic locations were sampled: Hot Weather Creek (HWC) at 70 m above sea level, and Big Slide Creek (BSC) at 200 m above sea level (Figure 2.1). Two separate locations were sampled to provide a better representation of the systems on the Fosheim peninsula. The vegetation at Big Slide Creek, being at slightly higher elevation, could be affected by different climatic conditions. Due to isostatic rebound since deglaciation, BSC has also been exposed to vegetation development for a longer period of time, compared to HWC (Bell, 1996). For each location, active layer detachment slides (referred to as A L D S , slides or failures in the remainder of the text) were classified into 4 age categories based on the criteria outlined in Chapter 2 (Figure 2.3, Figure 2.4, Table 2.1 and Table 2.2). 3.2.2 Vegetation and Environmental measurements For every slide, a stratified sampling technique was used to measure biotic and abiotic characteristics. Data were collected haphazardly by throwing a trowel every 2-3 meters along transects which were laid horizontally across the slide and the adjacent undisturbed terrain (Figure 2.2). Where the trowel fell, a quadrat was laid and/or a soil sample was collected. Percent vegetation cover was estimated visually using a 50 x 50 cm quadrat with 10 cm divisions, following standards techniques (Wein and Rencz, 1976; Levesque, 1996). Identification was made to the species level, when possible, for all vascular plants, following Porsild and Cody (1980). The vegetation sampling began early in the season (June 10th) which for many sites was before the development of the sexual parts, which made identification to the species level difficult. Therefore, generalisations to the genus level were 25 made when necessary (full species list in Appendix 1). Percent cover of mosses and lichens were also noted. Individual plant vigor and phenology was recorded but not included in this work. For the analysis, the quadrat data were averaged for zones of homogeneous environmental characteristics, determined by slope, aspect, micro-topography, soil compaction, relative moisture, presence or absence of sand and hummocks (Table 3.1). The number of quadrats averaged per zone ranged between 5 to 15 or more depending on the size of the slide and the size of the zone. A total of 1156 quadrats were averaged into 152 samples for HWC and 52 samples for BSC. Table 3.1: Environmental factors measured for each zone of the slides and control communities: Variable name Factors measured Description A G E Terrain age The terrain age was grouped into 4 classes: Young, Intermediate, Old and Very Old (Figure 2.2 and 2.3). EDGE Edge effect Effect of the proximity to vegetation. 1 = quadrat is within 1 m of undisturbed vegetation. 0 = quadrat is further than l m of undisturbed vegetation. SLOPE Slope angle Measure of the absolute slope in degrees using a clinometer. SLOPPOS Macro topography Values ranged from 1 to 10 indicating relative position of the sample along the slope, where: 1 = top of the slope 10 = the bottom of the slope. The range was calculated by dividing the distance from the top of slide by the total length of the slide, and multiplying by 10 DESS Density of A surface crack can trap seeds and water and therefore be 26 Variable name Factors measured Description desiccation cracks an important start of community development. The measurement is calculated as an average distance between cracks of the freeze-thaw polygons, measured within a zone. SAND Presence of sand Sand is also a better medium for trapping seeds, and for water drainage than the dry clay surface. This variable is a relative class measure of the presence of sand on the surface 0: No sand 1: Presence of sand 2: Dominant cover of sand H A R D Soil Compaction Soil compacting will affect how well the seeds can be trapped, as well as give an indication of the ability for the roots to grow through the soil. Relative values ranged from 1 to 3 The relative scale was established as follows: 1 : SOFT (can break the surface with a finger) 2 : HARD (can break the surface with back of pencil) 3 : VERY HARD (can break the surface with pointed end of pencil) MOIST Relative soil Subjective classes used to assess the surface moisture moisture 1 : Very dry and breakable earth 2 : Dry to the touch 3 : Moist to the touch 4 : Wet 5 : Standing water (see Appendix 2 for relation between absolute moisture and moisture index) ASP_NS ASP_EW Aspect Related to the solar energy input on the landscape due to solar radiation. South to south-west slopes will typically be warmer. A north-south and east-west relative index was derived from the sinuses and cosines respectively of the absolute aspect measured in degrees. North-South Index (ASP NS) ranged between 1 27 Variable name Factors measured Description (north) to -1 (South) East-West Index (ASP_EW) ranged between 1 (east) to -1 (west) DEPRLNDX Micro Depression Index 1 Measure of the relief of the micro-topography generally as a result of hummocks. A large value indicates high frequency of depressions in the topography while a small value indicates low microtopographic relief. LITTER Total litter Litter (dead unattached vegetation) can be important to retain moisture or to trap seeds. Litter was measured as a percent cover within each quadrat. Deprindx = [Depr I (Depr + Hum)] x [ 100 c m / (Depr + Hum)], where, Depr is the average length (cm) o f the depression between two hummocks and Hum is the average length (cm) o f the hummock bump 3.2.3 Additional Environmental Variables Other soil characteristics were measured for selected sites at HWC (Figure 2.3: Table 2.1). Two young slides (#s 1 and 3) and one old (# 2) were selected on north facing slopes while two young slides (# 4 and 5) were selected on south facing slopes for more intensive measurements. These slides were selected mainly on the basis of their proximity to the base camp. Additional soil measurements were absolute moisture, active layer depth, sediment composition, soil pH, salinity derived from conductivity measurements, and soil temperature. Soil samples were collected arbitrarily along the vegetation transects using a 5 cm x 7 cm (radius) soil auger. At each soil sample, a measure of the active layer depth was taken with a permafrost probe by pushing the probe (steel rod) into the sediment until it encountered frozen ground. These samples were collected at different times than the vegetation quadrats and therefore did not correspond to individual quadrat samples. Slope position and distance from the beginning of the transect were recorded to later match the soil 28 sample data with the vegetation and other environmental data. Samples were weighed fresh (0.0lg), air dried in the field in paper bags using a field oven and reweighed to determine the percent soil moisture content. For the 5 intensively studied slides, moisture sampling was repeated 4 times throughout the growing season (June 22, June 30, July 15, August 8) to assess seasonal soil moisture variation for different terrain ages. On July 15th, a full soil moisture survey was completed throughout the study area. Fifteen additional slide scars and toes (7 young, and 5 old) were surveyed in order to get a more complete representation of the valley and to further asses the effect of terrain age on soil moisture. Soil samples from two young, two intermediate aged and two old slides (#s 1,3; 6,7; and 2,9 respectively) were also used to measure soil pH, salinity and soil texture. A sample splitter was used to reduce the samples to approximately 150 g. Sediments were sieved to separate the sand fraction above 63 urn. Further separation of clay, silt and sand was done on the remaining soil fraction less then 63 um, using the Hydrometer method (UBC Department of Soil Science, Laboratory manual, 1994). The pH was measure using a mixture of 30 g of soil with 15 g of distilled water. Three pH readings (+0.1) were taken and averaged, for each samples after they had been left to equilibrate for a few hours. Temperatures were also recorded for 3 intensively studied slides: 2 young and 1 old. Unfortunately, the cables were damaged numerous times by fox, and data are available for only one of the young disturbances (# 1 = young and # 3 = old). For this purpose, thermistors attached to data loggers (CR21, Campbell Scientific) were installed in the scar and adjacent undisturbed area, at ground surface and at 10 cm above the ground. The air temperature thermistors were shielded from solar radiation by shade plates placed ca. 2 cm above the thermistors. Measurements were recorded every hour from June 16th to August 10th. 29 3.2.4 Statistical Analysis Two data sets were compiled to study succession on the Fosheim peninsula. The first data set combined the measurements on scars at Hot Weather Creek (HWC) and Big Slide Creek (BSC). This data set was used to assess vegetation patterns in primary succession and to compare the effect of different geographic locations. The HWC data set was then analysed separately because measurements were also collected for the toe environments. This permitted comparison of the patterns and processes occurring in primary succession (scar) versus secondary succession (toes). 3.2.4.1 Community Classification Classification and ordination techniques were used to analyse the vegetation patterns in the landscape. A l l samples were classified into community types by two-way indicator species analysis (TWlNSPAN)(Hill 1979). This divisive method produces 2-way ordered table which shows groups of species with similar distributions among the samples, and groups of samples with similar species composition (Hill 1979). The groups of samples represent the community types discussed in the remainder of the chapter. The cut level used for pseudo-species was 0-0.5%, 0.5-1%, 1-3%, 3-7%, 7-20%, >20%. These cut levels are lower than the suggested default or those in other analyses (e.g.Vetaas, 1994), since many species in such extreme environments have much lower mean cover values. Frequency distribution histograms were produced for each species, and confirmed the appropriateness of these classes. Four levels of division were used and groups smaller than 15 samples (10% of the total number of samples) were not divided further. For the HWC data set which was use to compare between scar and toe environments, the minimum group size was set to 12 with a 30 maximum of division of 3 due to the smaller number of samples compared to the combined data set. Each vegetation sample used in the TWTNSPAN analysis had a corresponding set of environmental measurements associated with it, as listed in Table 3.1. These environmental factors were summarised (mean and standard error) for each community type in order to characterise the environmental conditions of each community. 3.2.4.2 Ordination Analysis / Vegetation pattern and vegeation-environment relation Ordination techniques were used to further analyse the relation between environmental factors and vegetation, using the program Canoco (version 3.2, ter Braak, 1987). This type of multivariate analysis was used to reveal structure in the data by producing diagrams where samples or species are represented by points in a two dimensional space. Points are arranged such that those close together correspond to samples with similar species composition (Jongman et al., 1995). The choice of ordination techniques, unimodal or linear, differs based on the distribution of the data set. The length of the ordination axis (gradient length) from a Detrended Correspondence Analysis (DCA) gives an indication of the type of distribution found in the data set. Data sets with short D C A axes (< 2 standard deviation (SD) units) are said to be linear while data set with longer ordination axes are said to have non-linear (unimodal) distribution (Jongman et al, 1995). The combined data set had gradient lengths of 5.949 and 3.393 SD units for axis 1 and 2, respectively, confirming the use of non-linear models. The HWC data sets had gradient lengths of 5.425 and 3.252 SD units for the first and second axes. 31 Ordination can be performed as indirect or direct gradient analysis. With indirect gradient analysis, the data are organised based on the species composition, and can then be related to environmental variables with regression analysis. Canonical analysis (CA -indirect analysis) of the combined data set resulted in a strong arch effect, where the 2nd axis was a quadratic function of the first. Detrended correspondence analysis (DCA), which specifically attempts to remove the arch effect, was also unsuccessful. A function to the 3rd power was still visible in the biplots. The arch effect is usually present when an underlying gradient is too dominant and explains most of the species variation (Jongman et al, 1995; Pielou, 1984). Percent vegetation cover was the dominating factor, with a larger range of cover values among the species. The large percent cover of shrub species (Salix arctica, Dryas integrifolia) seemed to override the effect of many other smaller yet ecologically important species. However, downweighting the dominant species during the D C A did not alleviate the problem. With direct gradient analysis, also called canonical correspondence analysis(CCA), the site scores are restricted to be linear combinations of measured environmental variables (Jongman et al., 1995). The analysis was detrended because of the suspected arch effect (DCCA). Canonical correspondence was used over indirect gradient analysis because it removed the arch effect, and because it was felt that the most important environmental factors were included in the analysis. Therefore, D C C A ordination was applied to untransformed mean plant cover data for all of the data sets. Of the 42 species, 2 were made passive because of their rare or single occurrence in the plots (Cassiope tetragona and Aremaria maritima). These species were still given species scores in the ordination but have no influence on the extraction of the 32 ordination axes. Species scores were weighted averages of the sample scores. Monte Carlo permutation tests were performed on all data sets in order to test the significance of the overall ordination and the first axis in explaining the species-environment relation (terBraak, 1987). 3.2.4.3 Environmental conditions over time Spearman correlation coefficients were calculated to test the strength of the association between environmental variables. A l l environmental variables were included in the D C C A except litter and bare ground. These two variables were strongly and significantly correlated to vegetation cover and the other environmental variables, and so were not included to simplify interpretation of ordination analysis (Jongman et al, 1995). Although a few of the other variables were also significantly correlated to one another, they were included in the analysis as they provided additional important information. Edaphic differences between the age categories were interpreted from graphs. Analysis of variance (ANOVA) was performed to test for significant differences in soil moisture, active layer depth, soil pH and conductivity between the age categories and disturbance type. In order to not violate the assumption of sampling from a normal distribution and homogeneity of variance, non-parametric tests were used when needed, such as Kruskal-Wallis A N O V A by rank and Tukey-type mean comparison using rank sums (Zar 1984). 3.3 Results 3.3.1 Community classification - Vegetation association For the combined data set, the 157 samples with 42 species were classified into 10 community groups (Table 3.3). At the first level of division TWINSPAN separated the 33 shrubs and forbs from the grass-dominated sites (Figure 3.1). The second level of division is the most interesting in terms of succession. There is evidence of community separation based on functional type (sensu Chapin et al. 1996) and growth form of the dominant species, and on terrain age differences (Table 3.2). Divisions in community types at the 3rd level seem to result from local environmental differences. Table 3.2: Distribution of slides in each age category by community groups at the second level of division in TWINSPAN. Numbers in () indicate the percentage of slides for each age category of the total number of slides in that community type. Community Group Young Mod. Age Category Old V. Old Undist. Total Group A 22 6 2 7 37 (59%) (16%) (5%) (19%) (100%) Group B 13 14 10 31 68 (19%) (21%) (15%) (46%) (100%) Group C 7 26 33 (21%) (79%) (100%) Group D 1 5 14 20 (5%) (25%) (70%) (100%) 34 Q W ^ C U O H o XI 3 00 CN a a CO ' N 3 — i O CN - -7* 0 - O 2 0 8 r/3 H O J OH o a a a. a cn c/i O CO 3 O CU Cu •a c S3 c/3 — — CO C/N 60 a a CO o 3 CU NO o •> < _H co et) XI ° S ( N a " M S3 CO _ o ca cu <Z> co r s ^ ^ M a - g a § - a 0 O co 1- j - 73 QQ CO CO C CM CO .O 0 K c u Q "O CN -a-a ca Boo C/) PU CU CO _c 3 a ca ca co • eu J <5 «5 8Q 0 55 05 CQ c 3 „ S en CD s e O rt 35 Table 3.3: Summary of TWINSPAN sample groups. Numbers are the average of all samples in the TWINSPAN group and represent cover classes and also corresponds to the cut-level used: *=<0.1%, 1=0.1-0.5%, 2=0.5-1%, 3=1-3%, 4=3-7%, 5=7-20%. Species Species list TWINSPAN community groups Group A1 A2 B2b B1b B2 B1 C1 C2 D1 D2 1. Puccinellia spp 3 5 5 5 3 3 1 1 * Agropyron violacium 1 3 2 3 2 1 * * Bray purpurescens 4 1 1 1 * * 1 * Armeria maritima * II. Draba nivalis * Poa spp 1 3 3 4 4 1 2 1 Deschampsia brevifolia * 1 4 4 3 1 1 1 1 Alopercurus alpina 3 4 5 2 Taraxacum hyparctica * 1 2 1 4 * Oxyria digyna * * 3 * 3 3 1 * 1 1 Melandrium spp 1 2 1 3 * * * Potentilla spp * 1 3 2 3 1 * Ranunculus pedatifidus 1 1 * Cerastium alpina * * 1 * 1 * * Erysimum palasii * * Saxifraga flagellaris • Saxifraga nivalis * III. Arctagrostis latifolia 1 * * 1 * Festuca brachyphylla * * 1 * 4 * 2 * Carex spp 1 1 Stellana spp 1 1 * 3 * 1 1 Papaver radicatum 1 * 1 1 1 * 1 1 Draba corymbosa * Draba lactea * * * IV. Minuartia rubella * * 1 * * * Arnica alpina * 1 1 Lesquerella arctica * * 1 * Draba cinerea * * * * * 1 * * Moss 1 4 * 5 * 5 4 * V. Salix arctica * * 3 4 4 6 6 5 3 Dryas integrifolia * 4 5 5 6 Kobresia myosuroides * 1 4 4 4 Luzula spp 1 1 1 3 Polygonum viviparum * 1 4 4 Pedicularis spp * 3 1 1 Saxifraga tricuspidata * 1 Saxifraga oppositifolia * * 1 2 Cassiope tetragona 4 Erisymum compositus * * Draba alpina * * * * * * Draba subcapitata * * Lichens * 1 3 3 36 a ) Age cla SS t [ i 1 J i i i i i i i i i : 5 , ^ CD 7 . 4 X 0) X ) _c 3 c o act 2 C L E o o 1 o to 0 b) Soil compaction Index (l=hard, 5=soft) x - C M . O . a C M T - C M T - ' < - C M < < C N T - C Q C Q O O Q Q 03 CO r - < M . O . - a C N T - C N j T - T - C N < < C N l r - C D C Q O O Q Q CD CD c) Density of dessication cracks • > < < • > • > —'—I 1 i i I 1 £ 1 1 • T - C N J 3 J 3 C N T - C \ J T - ^ C N < < f t j r : c D c o O O Q D 0 .75 X CD TJ C 0.5 c o co co a. 0 .25 cu Q d) Depression index (0-1) < • < > - • 1 1 « r -I < < N j m a i o o D Q e) Moisture Index (l=dry, 5=wet) CN < < . Q . Q CM T— CM T - CQ CQ CQ CQ 8 O 5 Q Vegetation Classes o c ^1 5 co ' - 3 _ro "35 2 1 CD </) 9? C L •O0.5 c CO CO 0 f) Presence/Absence of Sand (0-1) I < < CM T - CQ CQ CQ CQ 8 o 5 a Figure 3.2 a-f: Mean (+SE) environmental conditions for TWINSPAN community groups for the HWC and BSC combined data set. Refer to Table 3.2 for detailed description of each environmental factor, a) Age classes. Values do not refer to actual ages but relative representation of age category from young (1) to very old (4) and undisturbed (5); b) Soil compaction (subjectively determined); c) Desiccation crack density measured as an average distance between cracks; d) Depression index is a measure of density of hummock terrain; e) Moisture index (subjectively determined); f) Presence / absence of sand, g) total vegetation cover measure (%); h) Litter cover (%); i) absolute slope measured in degrees; and j) relative position on the slope. 37 100 80 2 60 ¥ 840 20 g) Total Vegetation cover -,i—<> s -—I 1 1 1 1 l l h -— CN Q Q 40 30 § 20 o o 10 h) Litter cover — CN J 3 J 2 < < £ - OJ — C*l — — OJ CO CO O O O O i) Absolute slope — CM CN r -< < O l r - CD CD CD CD a 5 s a Vegetataion Classes Q . O CO <D C o c o * J 03 O Q . 10 8 6 o 4 2 j) Relative position on the slope (l=top, 10= bottom) I i 5 3 5 T - CN J 3 _ Q CN T -< < CN r - CD CD CQ CO Figure 3.2 (continued) 38 CO <D u c (0 J3 I— 3 tt *#-o CD E z 100 90 80 70 60 50 40 30 20 10 0 • Young ED Moderate • Old • Very Old M Undisturbed Combined data set of HWC and BS A1 A2 B2b B1b B2 B1 C2 C1 D1 D2 120 100 ,~ 8 0 \ 6 0 I 40 .« 20 "O "5 J ° E Z 140 120 100 80 60 40 20 0 H W C data set A1 A 2 B2b B1b B2 B1 C 2 C1 D1 D2 BS data set • 1 1 — twinspan group Figure 3.3: Distribution of the terrain ages in each TWINSPAN community group. Each box in the histogram represents, for each terrain age class, the percent of the class found in each community group. It is the sum of young disturbances (for example) in each community group, divided by the total number of young disturbances x 100. Figure a) is the summary for the combined data set, while b) gives the summary for BSC and HWC, separately. 39 3.3.2 Community types Group A : Ruderal Grass - forbs: Puccinellia - Braya association Age category1 average: class 2 + 0.6 In this group, the gramonoid Puccinellia spp. and the rudreal forb Braya purpurescens were the dominant species. Most plots (61%) in this community type were in the young terrain age (category=l) (Table 3.4). At the 3rd level of division, TWINSPAN separated samples with higher occurrence of Braya purpurescens (Al ) from the other samples (A2) (Figure 3.1). Terrain age, again, played a role in separating these communities, with A l samples associated exclusively with young terrain while A2 samples included some intermediate age and a even few old slumps (Young:53%, Intermediate:20%, Old: 10%). However, at this level of division, other environmental factors seemed more important. In particular, the strongest distinction between these 2 groups was the presence of surface sand in A l (Figure 3.2f). A l l sites with significant surface sand accumulations also had Braya purpurescens and a large number of new seedlings. Group B: Mid to late Sere Grass and Forbs: Poa - Taraxacum-Oxyria-Melandrium-Potentilla Age category average: class 3.3+ 0.4 This community type is very diverse (Table 3.3). It is characterised by the dominance of Poa species, and the absence of woody shrubs. Numerous grasses appeared in this group including Deschampsia brevifolia and Alopercurus alpina and Puccinellia spp. A. alpina was found almost exclusively in this community type. Many of the species are termed late to 40 mid sere species. A l l three of these grasses are also salt tolerant species (Porsild and Cody, 1980). The dominant forbs included: Taraxacum hyparctica, Oxyria digina, Melandrium spp, Potentilla spp, and Cerastium alpinum. In the rare plots where woody species were found, the cover was very low (Salix arctica: 1-5%). There was a range of terrain age in this group. Although the highest number of samples fell under the undisturbed category (46%), the next most significant classes were intermediate aged terrain (21%) followed by young and old (19% and 15%, respectively). Group B was further divided at the 3rd and 4th level. The 3rd level TWINSPAN division separated geographic locations of BSC (Bl) from HWC (B2) (Figure 3.3). Although both locations have roughly the same species, plots at BSC had some species that were unique such as Ranunculus pedatifidus, Saxifraga flagellar is, and Saxifraga nivalis. These species are typically associated with wetter environments (Chung 1989; Porsild and Cody, 1980). BSC typically had a higher moisture index and total percent vegetation cover (Figure 3.2g). Other species characterising BSC and the B l community type include the later sere grass Festuca brachyphylla and the forbs Taraxacum hyparticum, Melandrium spp, Potentilla spp, and Cerastium alpinum. Further divisions at the 4th level, within the geographic locations, showed similar trends. Differences between groups Bla and Bib or B2a and B2b were expressed by a lower cover of the colonising grass Puccinellia spp. (shifting from 7-20% to 1-3% cover) along with higher cover of the later sere grass species Poa spp., from 1-3% to 3-7% in groups Bla and B2a. Terrain age played a role at differentiating these sub-communities. 1 Age categories / classes: l=young, 2=intermediate, 3=old, 4=very old, 5=undisturbed communities. The values refe only to a category and not an acual age in years 41 Sites in groups B l a and B2a were typically characterised by old, undisturbed and a few intermediate sites, while groups B i b and B2b had mostly young and intermediate sites with few old ones (Figure 3.3). Group C: Shrub - grass - forb: Salix arctica dominated assemblage Age category average 4.5 + 0.3 The main characteristic of this community was the high percent cover of Salix arctica (>20%). A late succession species assemblage of Dryas integrifolia (7-20%), Kobresia myosuroides (3-7%), and Polygonum viviparum (1-7%) were associated with Salix (Table 3.3). Most species in the area were found in this community group, but at very low cover values ranging from a trace to 0.5-1%. This community type occurred on either old (21%) or undisturbed sites (79%). For both Groups C and D, further divisions by TWINSPAN are explained by differences in local environmental condition predominantly of moisture, total vegetation, litter, and microtopography (Figure 3.2 e,g,h,d). A l l of these factors are somewhat inter-related, as they all influence moisture content in the soil. In the wetter environment of group C2, a higher total vegetation cover was apparent with species of Salix, Dryas, Kobresia, Polygonum, and Pedicularis present with covers ranging from 3% to >20%. Moss was predominant in this community group. The drier environment of community group C I had the same species composition but with much lower total cover, and many species were found only as a trace. 42 Group D: Shrub / cushion plant: Dryas- Salix - Cassiope assemblage Age category average: 4.7 + 0.3 This community differs from the previous one in that the dominant shrub species was Dryas integrifolia (7->20%), followed by Salix arctica (1-20%). Also, Cassiope tetragona was present only in this community type. However, many other species were not found in this group or were very rare: Puccinellia and Braya were observed only as traces. Other species found were Melandrium spp, Stellaria spp, Papaver radicatum, Saxifraga oppositifolia and various species of lichens. Very old disturbances are found uniquely in this community group. At the 3rd level of division, moisture was again the main factor separating the community types. Group D l had wetter soils and high percent cover, with Dryas, Salix, Casiope and Kobresia as the dominant species, along with Polygonum, and mosses. The dry community type D2 had a higher cover of Dryas, Luzula spp. and Saxifraga oppositifolia and no occurrence of Cassiope tetragona or Polygonum viviparum. 3.3.3 Patterns of primary succession Ordination of the vegetation sites largely reinforced the classification results. A summary of ordination axes is presented in Table 3.4. For the combined data set, the first D C C A axis (X\ = 0.591) explained most of the species distribution across the study area. The variables most closely related to this axis included the terrain age (AGE) and the depression index (DEPRTNDX) at one end of the axis, and soil compaction (HARD) and desiccation cracks density (DESS) at the other end (Table 3.5, Figure 3.4). As seen from the Spearman's correlation coefficients (Table 3.6), these variables were not completely 43 Table 3.4: Summary of ordination statistics for a) the data set combining HWC and BSC quadrats of the scars and b) HWC data set with the quadrats for the scars and toes of the slides a) Combined data set of HWC and BSC: b) Dataset Combined data set: Total inertia = 4.776 Ordination method Ordination axis DCA axl ax2 DCCA axl ax2 DCCA age as covariate axl Ax2 Eigenvalues Gradient length Species-environment correlations Cumulative % variance of species data Cumulative % variance of species-environment Sum of all unconstrained eigenvalues Sum of all canonical eigenvalues 0.807 0.525 5.466 3.393 0.902 0.584 14.6 24.0 44.0 53.2 5.540 1.583 0.591 0.159 0.860 0.638 12.4 15.7 53.9 68.4 4.776 1.097 0.306 0.149 0.762 0.627 7.0 10,4 44.7 66.6 4.362 0.683 HWC data set: Dataset HWC data set: Total inertia = 4.674 Ordination method Ordination axis DCA axl ax2 DCCA axl ax2 DCCA Age as covariate axl Ax2 Eigenvalues Gradient length Species-environment correlations Cumulative % variance of species data Cumulative % variance of species-environment Sum of all unconstrained eigenvalues Sum of all canonical eigenvalues 0.802 0.520 4.873 2.776 0.819 0.573 17.1 28.3 60.4 65.3 4.674 0.980 0.459 0.135 0.800 0.733 10.4 13.4 49.8 64.5 4.425 0.923 0.335 0.121 0.752 0.691 8.0 10.8 47.2 64.2 4.213 0.711 44 Table 3.5: Weighted correlations between the environmental variables and D C C A axes 1 and 2 for the combined data set of HWC and BSC (weight = sample total). The values in bold highlights high correlation values. Environmental Variables Axis 1 Axis 2 AGE -0.671*** 0.205* EDGE 0.364*** -0.160* SLOPE 0.103*** 0.051 SLOPPOS -0.324*** 0.067 DESS 0.402*** 0.079 SAND -0.301*** 0.139 HARD 0.546*** -0.109 MOIST -0.101** 0.500** ASP_EW -0.030* -0.000* ASPNS 0.301*** 0.255** DEPPJNDX -0.434*** -0.040 LITTER -0.577*** -0.148 TOTVEG -0.476*** -0.389* Bartlett Chi-Square Statistic has been used to test the significance of the correlations. * : P=0.05;**: P=0.01; ***: P< 0.001 Table 3.6: Weighted correlation between the environmental variables for the combined data set of HWC and BSC. The values in bold highlights high correlation values. Litter and Total vegetation variables show Spearmean's correlation coefficients. A G E E D G E A G E 1.000 SLOPE E D G E -0.467" 1.000 SLOPOS SLOPE -0.092 -0.001 1.000 DESS SLOPPOS -0.223* -0.151 -0.175* 1.000 S A N D DESS -0.201" 0.046 0.031 -0.101 1.000 H A R D S A N D 0.121 0.045 -0.117" 0.043 -0.114 1.000 MOIST H A R D -0.545" 0.238'*' 0.002 -0.045 0.234" -0.223* 1.000 A S P _ E W MOIST -0.091* 0.001 -0.165* 0.047 -0.394' 0.023 0.096* "1.000 ASP_NS A S P _ E W 0.097 0.055 -0.003 0.037 0.064 -0.230' -0.100 -0.031 1.000 DEPIND ASP_NS -0.044 0.060 0.070 -0.111 0.137 -0.090 0.095 0.041 -0.028 1.000 LITTER D E P R I N D X 0.276" -0.194" 0.072 0.153 -0.266*"0.144 -0.244 0.192""-0.023 -0.024 1.000 T O T V E G LITTER 0.350"" -0.057 0.066 -0.007 -0.305"*0.099 -0.135 0.456*"*0.007 -0.161 0.400"" 1.000 T O T V E G 0.279"' -0.009 -0.105* 0.017 -0.317***0.051 -0.117 0.592""-0.047 -0.152* 0.391""* 0.710"""1.000 Bartlett Chi-Square Statistic was used to test the significance of the correlations. * : P=0.05; ** : P=0.01; ***: P< 0.001 45 independent of one another. Age was significantly negatively correlated with desiccation crack density and soil compaction (R 2 = -0.201 and -0.545 respectively, PO.001) and positively correlated to total vegetation cover (TOTVEG) and depression index (R 2 = 0.279, PO.001 and 0.276, P=0.01, respectively). As well desiccation cracks density and soil compaction were positively correlated (R2=0.234, P<0.001). Desiccation cracks were typically found on bare and compacted soils such as in young scars. Old terrain, however, typically had lower soil compaction as a result of erosional processes or increased vegetation cover. Although old sites were not always associated with high vegetation cover, the contrary tended to be true, that high vegetation cover was generally representative of wetter and older terrain. These denser vegetation covers were often associated with hummocky terrain measured by high depression index values, which explains its association with the A G E variable. Due to these above associations, the first axis could be considered as a terrain age factor complex (sensu Whittaker, 1991). Most of the measured environmental variables were significantly related to axis 1, although many of the correlations were weak (Table 3.5). Thus, a second underlying environmental factor complex could also be associated with the first axis. Macro-topography (SLOPOS), slope angle (SLOPE), and presence of sand (SAND) were significantly related to axisl but with a lower correlation strength. This combination may represent an erosional factor complex separate from the terrain age factor since none of these variables were correlated with age (Table 3.6). The positive end of the first environmental gradient (Figure 3.4) was characterised by steeper slopes, little to no presence of surface sand and sites located on crests or at the top end of the slope. Sites at the negative end of the gradient had gentler slopes, higher sand accumulation and were found in the valley bottoms or at the lower end of 46 the slopes. Erosion would therefore be higher in the first environment described while the other environment would characterise areas of sediment accumulation. Although a much smaller proportion of variance is explained by the second D C C A axis (A.2 = 0.159) it is best represented by moisture content in the soil (Table 3.5, Figure 3.4). When age was fitted as a covariable to remove its effect on the other environmental variables, the first and second axis had eigenvalues of 0.306 and 0.149 (Table 3.4). The covariable age explained approximately 8.7% of the total inertia while the remaining environmental variables explained 14.3%. This suggests that terrain age on its own is not the most important variable explaining the variation in species composition. Species most influential in defining the age factor complex were a) the ruderal species of Puccinellia and Braya at the positive end of the D C C A axis 1 (Figure 3.4) associated with young scars and b) the shrub species of Dryas integrifolia, Salix arctica, along with the herbaceous species of Pedicular is spp., Kobresia myosuroides, Polygonum viviparum, at the negative end of the axis associated with the older aged scars (Figure 3.4 and Figure 3.6). Although undisturbed communities reflected the oldest terrain, they were found around the centre and upper part of the ordination diagram. These were characterised by an assemblage of forbs and grasses which are also shared with many of the other community types and terrain ages, such as Poa species, Agropyron violacium, Alopercurus alpina, Potentilla spp, Taraxacum spp, Melandrium spp, Oxyria digina. These species were typically associated with drier and/or more exposed environments. The correspondence between the Twinspan and the ordination approaches was clear when the community groups were superimposed on the ordination scatter plots (Figure 3.5). 47 o J> B & 3 > •H X3 <D <D &H « GO <D § u T 3 " s i-i O H U 00 CD . P H-» o p o CO 0) > <H-H CD o p - § <D CO S .22 3 '3 5 u PH 6 2 f * > <a ~ . p co o £ PH •O g O <D o g ON § a O c3 <D CO rj <D fcn OS cO ^ <=! S ° O O VP1 <U as co .3 ^ O H—* <» .a « P H - S 2 ~ < <£ o y g T3 <D P • i—< PH X <D CD J D T3 p O O 0> CO !D •a on PH a o o o <H-H CD on S o u 2 <+H p X3 t £ o P cn <50 CD 3 * ° p 2 1 3 o o p CD <D H <D a o • i-H 3 H-» CO <D PH PH H T3 **H CD ^ > CD CD ^ T3 60 co P ccj 2 CuO <U ' p C a ID 2 Q 5 cd > p <D tao (D • — S-I <D S 3 <L) S P (D a 3 £ P W — C/j c/3 . r t <D <D 'o "5 ^ CD <D C3 PH PH S &o oo ra 48 49 <u 6 bo £ 03 -4-> CD CD O O 50 Outlines were drawn on the ordination diagram to delimit the spread of these communities, at the second level of division in TWINSPAN. Communities B l and B2 which were seperated at the 3rd level of TWINSPAN division, showed the differences between HWC (B2) and BSC (Bl) locations. These two communities did not play a large role in defining axis one (age factor complex), as can be deduced from their general position close to the centre of the first axis, yet they were clearly separated on the second axis. BSC data set were characterised by wetter soils, while those at HWC were comparatively drier. In general, each community type occupied a discrete portion of the ordination space, yet the change in community type was rather progressive and continuous, as seen from the overlap between community types. This overlap is even more apparent at the third level of community distinction, illustrated by different symbols in Figure 3.5. As well, the overlap between communities was more prominent between groups A1-A2 and B1-B2 (ruderal grass and late sere grass/forbs) and between group C1-C2 and D1-D2 (Shrub/forb/grass and Shrub/cushion plant). However, sites in community types A1-A2 were quite distinct from the community types C and D, suggesting that community composition and structure typical of early stages of succession is quite distinct from those of late stages of succession. Figure 3.6 shows the TWINSPAN sample group centroids with 95% confidence intervals, as well as the age category centroid of HWC and BSC data sets, with ellipses drawn to include 95% confidence intervals. Dominant functional groups defined by the TWINSPAN community types are also shown along the terrain age factor axis. Thus, over time community change appears to progresses through four stages: 1) Ruderal grass —> 2) Later sere grasses and forbs --> 3) Shrub with late sere forbs and grasses —> 4) Shrub with moss and cushion plants. 51 The progression through these 4 stages also appeared to be associated with moisture availability over time. Although BSC tended to be wetter in comparison to HWC, the change in the vegetation community displayed a similar pattern as seen from Figure 3.6. Older disturbances appeared to have high soil moisture content and softer ground compared to younger disturbances. Although old and undisturbed communities were comparable in species composition, old disturbances typically had higher concentrations of moisture, in comparison to undisturbed communities. The position of very old slides in the 2 dimensional ordination space re-enforced this trend, showing how these are wetter than the other environments. These slides were also characterised by a very distinct community composition, with the presence of Cassiope tetragona and a dense cover of Dryas integrifolia. Changes in percent vegetation cover over time is illustrated for each species in Figure 3.7. The pattern suggests again that succession proceed with a gradual shift of species density and dominance, without strong replacement, except in the very old disturbances, where many of the ruderal species are absent. 52 B R A P U R Ruderal Forb o o.s _ •— ^sliiS8ili™B W3_ . ^3R___ " "Iff: —. -, P U C S P P D E S B R E P O T S P P F E S B R A ^ 1 S T E S P P S A L A R C Ruderal grass (sal tolerant) Grass: (salt tolerant wet habitat) Ruderal forb Grasses Late sere forb (dry habitat) Grass Grass Late sere forb (dry habitat) Late sere forb (dry habitat) Late sere forb (dry habitat) Late sere forb (dry habitat) Ruderal forb (drier habitat) Shrub Bryophyte Late sere forb (wet habitat) Late sere graminoid Late sere forb (wet habitat) 1 I Young Intermediate Old Very Old Undisturbed Age Categories Figure 3.7: changes in vegetation cover (%) over time, for individual species. Species abbreviation list in Appendix 3.1. The succession group associated with each specie was derived from the Ordination and TWINSPAN classification (Figure 3.4 and 3.6) 53 3.3.3.1 Species richness: At HWC, the total number of species increased from the young to the old disturbances, followed by a sharp decrease in very old disturbances, and undisturbed sites (Figure 3.8). BSC displayed a different pattern, where species diversity increased continuously as the terrain aged. Analysis of variance (Appendix Table 3.1) indicated that there is a significant effect of age on species diversity for both HWC and BSC (P=0.001 and P=0.000, respectively), the significant differences being between young and older terrain, including undisturbed communities. 0.00 Young Mod. Old Very Old Disturbance Age Undist. Figure 3.8 Species richness comparing HWC and BSC location. 54 3.3.4 Primary vs. Secondary succession The HWC data set was used to assess the difference between primary (scars) and secondary (toes) succession. Although scars represented the biggest portion of the sites sampled (Table 3.7), patterns and trends could still be derived from the combined data set of toes and scars. Table 3.7: Number of samples for each disturbance type at HWC by age category. Each sample represents the average of quadrats, for zones of homogeneous environmental conditions within the slide of control community. Age Categories Scars Toes Young 25 9 Intermediate 4 2 Old 11 6 Very Old 5 — Undisturbed 76 Undisturbed extreme 9 environment TWINSPAN divided the 127 sample data set into 8 sample groups (Figure 3.9, Appendix Table 3.8). As in the combined data set, the 2nd level of division separated communities of similar life history and functional types (sensu Chapin et al, 1996). Although these communities were associated with different terrain ages, the distinction is not as clear, as seen from the ordination analysis (Figure 3.11). As well the distinction between the 8 community groups resulting from the 3 r d level of division was not very sharp, as seen from community position and overlap in the ordination space on Figure 3.11 55 ND CN o <D > or. •£> 3 o SB so 5 M 2 "3 i -<u •e s PS J5 3 © Tt tN a. u co O. O cd 3 L-cucn IT) c^ *-> co In 3 P cd u. cd P - o >*5 CN - ( N c s 1 t N cd 6J)__ >N PL, < GO O oo H o OH CN C/3 Tt CN (N " Cd " — cd cd oo Q o C N co Vi « o cu o NO c s y .S3 c fa c/i > o « jo Q ^ o n S i -o £>. Xi s cc CL — « r d ^ O cd « PH OO Q ro CN ON CO — o >™ E w ^ ° « l S OH 2 « ft™ Q oo J > CN o3 OH 2 •Si .a > C HC CL) e/3 CU O OS 56 Table 3.8: Distribution of slides by community group at the second level of division and terrain age classes for the HWC data set. Numbers in () indicate the number of slides for each age category as a percentage of the total number of slides in that community type. Community Group Age Category Total Young Intermediat e Old V. Old Undist. Group A 12 (67%) 1 (6%) 5 (28%) 18 (100%) Group B 17 (37%) 4 (9%) 6 (13%) 19 (41%) 46 (100%) Group C 5 (12%) 1 (2%) 10 (23%) 27 (63%) 43 (100%) Group D 1 (6%) 5 (31%) 10 (63%) 16 (100%) Table 3.9: Distribution of slides by community group at the third level of division and terrain age classes for the HWC data set. Numbers in () indicate the number of slides for each age category as a percentage of the total number of slides in that community type. Community Group Age category Total Young Intermediate Old Very Old Undist. Scar Toe Scar Toe Scar Toe Scar Toe A l 5 (100%) 5 (100%) A2 7 (54%) 1 (8%) 5 (38%) 13 (100%) Bl 10 (43%) 1 (4%) 1 (4%) 1 (4%) 10 (43%) 23 (100%) B2 3 (13%) 3 (13%) 2 (9%) 1 (4%) 2 (9%) 3 (13%) 9 (39%) 23 (100%) CI 5 (20%) 1 (4%) 1 (4%) 1 (4%) 1 (4%) 16 (64%) 25 (100%) C2 4 (17%) 1 (4%) 7 (30%) 11 (48%) 23 (100%) Dl 1 (11%) 3 (33%) 5 (56%) 9 (100%) D2 2 (29%) 5 (71%) 7 (100%) 57 a) Age class >,5 O a> o 4 o m >2 C?1 20 ~J5 E CD g10 rs +J (A C S 5 b) )ensity o f dessication cracks 1 1 1 f i i I x 1 1 1 • • — i — m — - — - — A1 A2 B1 B2 C1 C2 D1 D2 A1 A2 B1 B2 C1 C2 D1 D2 10 c .2 8 •*J at o °- 6 CD Q . o cn 4 CD > c) Relative position on slope (0=top, 10=bottom) -+- -+-30 S 2 5 CD | 20 13 215 I 10 5 CD Q . O w 0 d) Absolute slope T" 1 i * i I T • A1 A2 B1 B2 C1 C2 D1 D2 A1 A2 B1 B2 C1 C2 D1 D2 0.75 8.50 •a c c o (0 S-25 0.00 e) Depression index (0-1) 1 1 • 2.00 CD O S 1.50 CD O CD CO CD i _ Q . "D C re in 1.00 0.50 0.00 f) Presence / absence o f sand • 1 i i 1 1 • — ™ 1 1 1 1 A1 A2 B1 B2 C1 C2 D1 D2 A1 A2 B1 B2 C1 C2 D1 D2 Figure 3.10a-j): Mean (+ SE) environmental conditions for each TWINSPAN community group. Refer to Table 3.2 for detailed description of each environmental factor, a) Age classes. Values does not refer to actual ages but relative representation of age category from young(l) to very old (4) and undisturbed (5); b) Soil compaction (subjectively determined); c) Desiccation crack is measured as an average distance between cracks; d) Depression index is a measure of density of hummocky terrain; e) Moisture index (subjectively determined); f) Presence / absence of sand; g) Total vegetation cover measure (%); h) Litter cover (%); i) Absolute slope measured in degrees; and j) Relative position on the slope. 58 g) Compaction Index (l=hard, 5=soft) e • 30 -r 25 --20 --CD 15 -> o o 10 -5 -0 -I) To ta l litter cover 4 I i —\— A1 A 2 B1 B2 C1 C2 D1 D2 A1 A 2 B1 B2 C1 C2 D1 D2 Figure 3.10 (continued) g-j): 5.0 in 4.0 T -X O! 3.0 •a c CD &_ 2.0 3 in o E 1.0 0.0 h) Moisture Index (l=dry, 5=wet) A1 A 2 B1 B2 C1 C2 D1 D2 80 " P j ) To ta l vegetation cover-_ 60 > 40 o 20 e A1 A 2 B1 B2 C1 C2 D1 D2 59 Further division in community types to the 3rd level appears to be the result of local environmental differences. Distinctions between hCl and hC2 (Grass/shrub community types), as well as hDl and hD2 (Shrub/forb community types) most likely resulted from differences in soil moisture (Figure 3.1 Oh). For the Grass/forb and small shrub group, the higher vegetation cover in hB2 communities distinguished it from hBl. hB2 communities were associated with lower position closer to the valley bottom with gentle slope angle, while hBl communities were also found topographically higher on the slope, on steeper sloping terrain. Finally, hAl communities were only associated with young disturbed scars, while hA2 included as well a few intermediate aged and an old disturbance (Table 3.9). Interestingly, toe disturbances were found predominantly in community types hB2, hCl and hC2. A l l of these communities were characterised by abundance of Salix arctica. hB2 communities had species more typical of drier environments, including Poa spp., Deschampsia, Oxyria, Melandrium, Potentilla, as well as some Agropyron, and the ruderal grasses of Puccinellia spp. HC2 communities on the other hand included species of wetter environments such as Polygonum, Kobresia, Dryas, Pedicularis, Stellaria, and Mosses. Finally, like hC2 communities, hCl communities had a high species diversity, yet with most species occurring with very low cover (0.5% or less). The species associated with this community include Lesquerella, Papaver, Stellaria and Draba species (Appendix Table 3.8). Community type hC2, typically characterised by rich communities with wetter soils and higher vegetation cover, was more often associated with old scars or with young toes. Communities found on the old toes, on the other hand, were more often classified into community hB2, which reflected drier landscapes (b). 60 Ordination techniques were used to further analyse the two different succession types proceeding in the scar and/or toe areas. Age and soil compaction (HARD) were, as in the combined data set, the most important variables representing the first D C C A axis in the ordination (A,j =0.459) (Table 3.5Table 3.4, Table 3.10, Figure 3.12). Although not as strong as for the whole data set, desiccation cracks density and depression index were also important and significant in defining the axis. Moisture was again the dominant variable explaining the variation along the second axis. When terrain age was fitted as a covariable, the percent of the variance accounted for by the first axis (age factor axis) was reduced from 45.9% (DCCA with age) to 33.5%. Although age explained the largest portion of the total inertia (9.8%) of a single variable, it was still not as influential as the other environmental variables combined (15.2%) (Table 3.4Table 3.5). This was also illustrated in Figure 3.13 where sample scores were averaged by disturbance type (scar/toe) and age categories. In the scars, disturbances initially produced terrain that was hard and dry which then became moist and softer with higher sand content and increased micro-topograpy. This improved moisture condition was in contrast to undisturbed communities (c5 and cx5) which remain exposed to wind erosion and desiccation. Initial conditions in the toes were similar to old scars with wetter soils, and higher vegetation density. Both disturbance type appeared to be associated with community type hC2 (Figure 3.13). When a slide occurs, the toe is composed of a vegetation mat that has been churned and broken up, exposing new surfaces for the germination of seeds invading the 61 Table 3.10: Weighted correlations between the environmental variables and D C C A axes 1 and 2 for the HWC data set only (weight = sample total). The values in bold highlight high correlation values. Environmental Axis 1 Axis 2 Variables AGE -0.4642*** -0.2715 SLOPE 0.3495 *** 0.1459 POS.ON SLOPE -0.1275** -0.0082 DESS 0.3896*** -0.2711* SAND -0.2482* -0.1596 HARD 0.4191*** 0.2068 MOIST -0.2170* 0.6429*** ASP_EW 0.0814 -0.0842 ASP_NS 0.3761** -0.1630 DEPRTNDX -0.2312*** -0.1783*** LITTER -0.608*** -0.508*** TOTAL VEG -0.701*** -0.564*** ENVI AX1 0.7942 0.0000 ENVIAX2 0.0000 0.7402 Bartlett Chi-Square Statistic has been used to test the significance of the correlations * P = 0.05 ** P =0.01 *** P<0.01 Table 3.11: Weighted correlation between the environmental variable for the HWC data set only. The values in bold highlights high correlation values. Litter and Total Vegetation variable show Spearman's correlation coefficients. AGE SLOPOS AGE 1.000 SLOPE SLOPOS -0.118 1.000 DESS SLOPE -0.147 -0.211*** 1.000 SAND DESS -0.101*** 0.054 0.092 1.000 HARD SAND 0.101 -0.202 -0.147** -0.079 1.000 MOIST HARD -0.375*** -0.033 0.104 0.183*** -0.235 1.000 ASPEW MOIST -0.265*** 0.114 -0.147 -0.241 0.027 0.190*** 1.000 ASPNS ASPEW 0.157* 0.080 -0.106 0.153 -0.362*** 0.007 -0.134 1.000 DEPINDX ASPNS -0.113 -0.014 0.099 0.062 -0.081 0.114 0.090 0.020 1.000 LITTER DEPINDX 0.021 0.038 0.031 -0.257*** -0.041 0.186 0.298*** -0.050 -0.007 1. TOTVEG LITTER 0.175* 0.053 0.049 -0.255*** 0.069 -0.071 0.407*** 0.078 -0.134 0.513*" 1.000 TOTVEG 0.208* 0.202* -0.068 -0.386*** 0.097 -0.153 0.486*" -0.013 -0.133* 0.494*** 0.765*** Bartlett Chi-Square Statistic has been used to test the significance of the correlations * ^ = 0.05 ** P = 0.01 *** P<0.0\ 62 CD OH ^ °^  f ~ P CD H CO p CD O O PH CD SI! H 2 o 2 £ -3 « •5 ^ • t? o £ fa CD <D 13 "S CO CD 8 & g P « .2 £> rt 2 a o CD H-H CD CD w a s ^ 3 O OH CD O ^ 03 5b ^ s -P ^ 03 _rH OH Cd O S ^ 03 1 C/3 | > CD o t o -p c/3 CD C/3 & -P c3 r <2 ,o P Tu o Sj <fcl C/i CD b o m •+H CD r p c5 rt 3 ^ N=> CD CD CD vi P rrS rt . O CD o ^ -5 £ * * •+3 .03 <+H H O x .2 o rt &, .3 OH ^ < .3 ° 12 r^< 3 i2 U CN o 03 f--P a ^ CD <+H CD O _ _P P 03 CL -H H—» . LH CA rd +-> CD ,-H' P ro CD CD Is 5 a « t o c £ •£ " •3 <D w ^ p ' <u P -P T_3 O VJ ^ '22 JH P > o 0 _ 3 <-> CD T 3 00 63 cd o <D cd CO CD O OO 5 > 13 -o <u CD CO <U M OH <D rt on <D )-l o o oo 5 -g co C o <D UJ 2 £ T 3 « o o CD oo '8 <D _ X + - <D CO OH s < &.S s i CO _, rp t> cd • cd Q . <D ° I C CO "o w > "o CD O OH CD CO OH H-> CD cD CO CD S OH . O CD % O 2 to" M 2 3 s rtj c1 *5 - cd o <u cd c/3 cd td o -T3 0> £5 ~ £ cd ~ ^ O s> o cd >• CO .2 « o o co *j3 'x .2 cd > § 5 o c3 cD co co •a -2 o CD .!-. CO rt rt CD (TJ T 3 ^ &,£ o s° *H 5 00 J O 3 ^ O CD 3 « o O CN CD CO (D '5 CD OH CO < o u Q CN m CD bp cd CD OH CO L § 2 M H rt3 CD H f « M2 >? - H O ID 3 _CD > ! & « - H <D 3 ' S o £ § CD CD as CD (SO -a CD o "OH x (D (D rt cd co a O CO 'co Id •> ^ 2 H cd ^3 64 2 u a a o a O 13 II CD X 2 cd CD > 3 3 o X) > CD P-l OH • t—< • H—» V} CD X O ^> CD c o O II O VI CD <? 3 « I o C/3 US £ H3 C S3 CD cd g CO ^ cd C/3 T 3 >-H CD H2 H O 2 £ l-i <D 60 S H ^ 33 -3 II cd ^ 3 o OH CD > CD cd E3 H .2 53 2 cs C3 ti. 3 « o xJ cd 2 " o CD CD O cd O CD M CD .2 8P <; 9 <-> o Q -5 rn § "—'. <•> 2 » •SP & t i , CD Vi 77i 12 C . 8 ^ . CD ^ -=3 J> H 3 • r r t cn f—i u. I f C -(-) > " H *-H CD T J . 3 33 <D c2 c CD T J «a a o o ON OH O CD ,03 t H CH-H CD ,£> a cd CD C/3 C3 O a CD T J 65 area or found in the seed bank. A higher concentration of water accumulates in the young toes, in the depressions between the churned material, and from the surface run off from the upper slope. With aging terrain, toes appear to change toward drier and more desiccated soils (Figure 3.13). Toe environment resemble more the environments of undisturbed communities and even of exposed undisturbed communities more typical of severe landscapes. The toes form raised bumps in the lower valley. As the natural erosional processes modify the landscape, the initial churned bumps tend to smooth out and it appears that water drains away from them, creating a less favourable plant environment. The communities of the older toes generally had low vegetation cover with a dominance of Salix arctica associated with very low cover of forbs adapted to dry environments, such as those found in community types h C l . Once established, Salix appearred to survive even the harshest environmental conditions. Community change over time in the toes of the slides is characterised by a shift in dominance of the Salix species. Evidence of replacement occur at herbaceous community level while the dominant shrub persists over time. Further investigation of Figure 3.13 shows that community change progresses through TWINSPAN groups hC2, hB2 and h C l and can thus be described as follows: 1) Salix-Polygonum-Kobresia-Dryas --> 2) Salix-Poa-Oxyria-Melandrium —> 3) Salix -Draba-Lesquerella, as illustrated in Figure 3.12 and Figure 3.13. 3.3.5 Soil characteristics on terrain of increasing age 3.3.5.1 Soil moisture Moisture appeared to be important in explaining species composition and in separating community types in the ordinations (Figure 3.6 and Figure 3.13). 66 3.3.5.1.1 Effect of growing season To evaluate the change in soil moisture conditions throughout the growing season, slumps were chosen based on their proximity to camp, and only young or old slumps were selected to simplify comparison. Unfortunately, since only one out of five slides was classified as old, no statistical tests were applied to assess the effect of terrain age on soil moisture. Trends and patterns were suggested from the graphs, with reservation given the number of samples in the data set and the unique case of the old disturbance. For these five slumps, only samples that had repeated measurements four times throughout the growing season were used for the analyses. To test for differences among sites (slumps) and times during the season, the old slump was included. The trends across the season were very similar regardless of the terrain age (young or old) or disturbance type (scar or toe). High moisture levels occurred in the spring followed by a sharp reduction as the season progressed (Figure 3.14). The largest differences between young and old terrain were mainly apparent in the spring (June 22), where the old disturbance had much greater soil moisture. This suggests that a major moisture source for the season is snow melt, especially for the older scar. A repeated measure analysis of variance confirmed significant differences in soil moisture in the scar area between sites (slumps) and for the four different collection dates (Appendix Table 3.2). Single factor A N O V A s were repeated for the 4 separate dates, to further test the effect of site on soil moisture. These confirmed the observed trend and showed that only on June 22 were there significant differences between sites (P=0.000) and the most important one being between slump 2 (old) and slumps 1, 4, and 5 (young) (P=0.000 for all 3 cases, Appendix Table 3.3). 67 Figure 3.14: Changes in soil moisture for selected active layer detachment slides (ALDS) throughout the growing season. The data set was compiled from 3 young disturbances and 1 old disturbance for which 5 to 8 moisture samples were averaged by zone (scar, track, toe), for 4 separate occasions during the growing season. The data are mean (+/- SE). 68 80 70 60 Si 50 > o u c 40 o re "g 30 Ul o > 20 10 •+- -f- -+- -+-1.5 2 2.5 3 moisture index (1=dry, 5=wet) 3.5 Figure 3.15: Relation between soil moisture index classes and total vegetation cover. 69 Scar Toe Position on ALDS Scar Toe Position on ALDS Figure 3.16: Mean (+ SE) soil moisture (a) and permafrost depth (b) for disturbances of 2 age categories (young and old) and 2 succession types (primary = scars, secondary = toes) derived from the larger number of A L D S sampled on July 15th. n indicates the number of A L D S per category. 70 The position along the slope (scar, track or toe) did not have a significant effect on soil moisture for either young or old disturbances, or for the different measurement dates of the season (Appendix Table 3.4). Although significant differences were not measured in the old disturbances, probably due to the small number of samples, interesting patterns can be highlighted from the graphs (Figure 3.14). With the exception of June 22, associate with spring melt, the track of old disturbances were wetter than the scar, which can be expected as water moves downslope. However, the toes appeared to be drier than the track. Toes formed a small mound in the landscape from which water would tend to drain away. The old disturbance appeared to have wetter soils than the young disturbances, although differences were not large following spring snowmelt (June 22). Percent vegetation cover was high and continuous for this old disturbance (hwc2), in all parts of the slump, including the toe. The Pearson's correlation matrix (Appendix Table 3.5) shows a significant positive correlation between vegetation cover and soil moisture early in the season (R =0.624, P=0.000). Figure 3.15 also illustrates this trend towards higher vegetation cover with high soil moisture content. Site with moisture index >= 3 had vegetation cover ranging from 40 to 65%, while indexes of 1 to 2 varied mainly between 16 to 25% cover. 3.3.5.1.2 Effect of terrain age and succession type (scar or toe) In order to statistically test the significance of the moisture differences for terrain of different age, and of different succession type (scar, toe), the sample size was increased to include sites from the whole region. Due to resource and time restrictions, moisture samples on many slumps were collected only once in the growing season. The largest collection was completed on July 15th, which coincides with the driest soil conditions of the whole growing 71 season (Figure 3.14). Observable differences are therefore possibly more important, since this was a period of limited water availability. Soil moisture was not affected by the disturbance type or the terrain age alone, but by the interaction of these two factors (P=0.011, Appendix Table 3.6a). Young scars were generally drier compared to old scars, while young toes were typically wetter than old toes (Figure 3.16a). This again confirms that, over time, scars would tend to see increases in soil moisture, while toes would experience a decrease in soil moisture. 3.3.5.2 Depth to permafrost In the scars, the active layer depth increased progressively throughout the growing season ranging from 16 - 32 cm in June, to 45 - 75 cm by Aug 8th (Figure 3.17). Old disturbances had shallower active layer depths compared young ones. In the toes, no differences were apparent between young and old disturbances throughout the growing season. For July 15th, when numerous slides were surveyed, no significant differences in permafrost depth were observed as a result of terrain age or disturbance type (Appendix Table 3.6b). However, the pattern illustrated in Figure 3.16b suggest that in the scars, young disturbances tended to have a slightly deeper active layer than old ones. The relation between soil moisture and active layer depth for four different dates during the growing season is shown in Figure 3.18. There were significant negative correlation between soil moisture and depth to permafrost, yet the strength of the relationship was rather week (R2=0.217, i>=0.000; R2=0.131, P=0.000; R2=0.023, P=0.096 for Jun22, Jun30 and July 15, respectively, Appendix Table 3.7). Patterns indicate that shallow active layer depths, as early in the season, are associated with wetter soil conditions. However, the 72 Figure 3.17: Changes in active layer depth throughout the growing season for the two succession types (primary = scars, secondary = toes). The dataset was compiled from 3 young disturbances and 1 old disturbance, for which 5 to 8 samples were averaged for the 4 separate dates during the growing season. The data are mean (+/- SE) 73 100 90 80 70 -^•60 u.50 5 40 Q. a> 30 TJ •- 20 9 0 2 o O < June 22 R2 = -0.217 (P=0.000) Moisture = -0.50 perm + 30.233 10 20 30 40 50 100 90 80 -f 70 60 50 40 30 20 --10 0 July 15 Moisture = -0.053 perm*+ 12.128 R2 =-,0,023 (P=0.096) , 0 10 15 20 25 30 100 90 80 70 60 50 40 30 20 10 0 100 90 80 70 60 50 40 30 20 10 0 June 30 R2 =-0.131 (P=0.000) Moisture = -0.206 perm + 19.537 • 10 15 20 25 Aug 8 • . * • • . • • • Moisture = -0.082 perm +17.770 R2 = -p.073 ([P=Q.Q07J| , 0 5 10 Soil moisture (%) 15 20 25 30 30 Figure 3.18: Relation between active layer depth and absolute soil moisture at 5 cm depth, for 4 dates during the growing season. The R 2 is derived from Pearson's correlation statistics, and the significance of the correlation is from the Bartlett's statistic. For illustration purposes the x-axis scale varies between graphics to best represent the points in the diagram. The model, derived from a linear regression, identifies the strength of the relationship (R ) and whether the slope of the line is different from 0. 74 slopes in the regressions differed very little from 0 indicating again the weak affect of permafrost depth on soil moisture. 3.3.5.3 Soil texture Only soil samples from scars were used to differentiate sediment characteristics of young versus old disturbances. In a fresh slide, the scars were characterised by higher clay content, while old disturbances had a higher sand fraction (Figure 3.19). Soils in young scars were clay-loam with 30% (+ 1%) clay, 36% (+ 4%) sand. Intermediate aged disturbance had loams or sandy clay loams, while old scars were typically classed as sandy loams, with 18% (+ 5%) clay and 60% (+ 13%) sand. Young scars tended to have much higher clay contents than the nearby undisturbed environments, while old scars tended to have a much higher sand content then the associated undisturbed environments. 3.3.5.4 Soil pH and conductivity For this analysis, samples were averaged for each age category of the scar and the associated undisturbed controls. The soils from undisturbed communities appeared to have higher pH values (8.0 + 0.2) compared to disturbed communities. Disturbed communities had an average soil pH of 7.8 + 0.1 for young scars and 7.5 + 0.3 for old scars (Figure 3.20). Although these differences were statistically significant, these pH differences will not likely affect plant growth and development. Soil conductivity, which gives indications of salt content in the soil, was also clearly affected by disturbances. Young scars had extremely high conductivities and thus salt i content, in comparison to the intermediate and old scars (Figure 3.21). The post hoc Tukey comparison test indicated that young disturbances had significantly higher soil salinity 75 (P=0.000) compared to the undisturbed communities, while older disturbances have significantly lower salt content (P=0.009) compared to the associated undisturbed community. A single factor A N O V A revealed no significant differences (P>0.05) in conductivity between undisturbed sites (Appendix Table 3.9). 3.3.5.5 Soil temperature Table 3.12 shows the average temperatures recorded in the scar and control environments for both locations. As expected, air temperatures were cooler than ground temperatures (Table 3.12) with differences as high as 6°C in the scar of the young disturbance (Appendix Table 4.2a). Comparison of surface temperatures between young and old scars indicated that, typically, young scars had higher maximums (17.2 + 1.2 and 14.7 + 1.0 for young and old respectively), while the minimum temperatures were quite similar (Appendix Figure 4.1). A new scar in the terrain did not appear to have any effect on the temperature, as differences between disturbed and undisturbed terrain were negligible, except for air temperature which was occasionally higher in the scar (Appendix Figure 4.2a). The old disturbance showed a different pattern (Appendix Figure 4.2b), where surface temperatures inside the scar appear to be consistently, i f not significantly, higher than those in the undisturbed terrain, over the whole day. 76 Table 3.12: Mean (SD) air and soil surface temperature differences between young and old disturbances and between scar and control environments. Averages represent hourly temperature measurements recorded from June 26 t h to July 5 t h, 1994. Air Temperature Soil Surface Temperature HWC1 (young) Scar 10.0 (+/- 3.3) 11.5 (+/-4.2) Control 9.6 (+/-3.2) 12.0 (+/- 4.5) HWC2 (old) Scar 9.7 (+/- 3.1) 10.7 (+/- 3.4) Control 9.7 (+/- 3.2) 8.8 (+/- 2.9) 77 a) © Young scar O Control of young sc A Intermediate scar A Control of intermediate S C • Old scar • Control of old s c sand content (%) Figure 3.19: Soil texture for the scars and associated undisturbed community for three age categories: a) all samples plotted; b) Averaged by disturbance type (S=Scar, Cntr=Associated undisturbed controls) and age categories (l=young, 2=intermediate, 3=old). For illustration purposes, the ellipses represent 68% instead of 95% confidence interval. 78 • Scar fa Undisturbed T X TT •-! 1 Young Moderate Old Age Category Figure 3.20: Soil pH for young, intermediate and old scar, compared to their respective undisturbed communities. 3500 -r 3000 2500 J 2000 u 3 C 1500 o O 1000 500 0 16238.8 ( + . 2 5 5 8 S E ) Y o u n g • S c a r • Undisturbed Moderate Age Category Old Figure 3.21: Soil conductivity for young, intermediate and old scar, compared to their respective undisturbed communities. Soil conductivity was used as an indicator of soil salinity. 79 3.4 Discussion 3.4.1 Patterns of succession The vegetation on the Fosheim Peninsular was organised into distinguishable plant communities, and appeared to be associated with different terrain age and/or a combination of environmental factors. The patterns of community change suggested that primary succession in the scar zones of slides was directional. At a general level, the community change over time was expressed by a shift in growth forms of the dominant species, without strong replacement, and followed the following successional stages: 1) Ruderal grasses --> 2) Late sere grasses and forbs --> 3) Shrubs, forbs and grasses --> 4) Shrubs and cushion plants. However, at a more detailed level, community assemblages and distinctions reflected changes in local environmental conditions, especially moisture conditions. A similar pattern in community change was found at both locations (HWC and BSC), where vegetation cover and species richness increased over time. Environmental conditions appeared to improve in aging terrain, with the soil surface becoming softer, soil moisture increasing, and salinity and clay content decreasing. There were, however, community differences between BSC and HWC. Species richness continued to increase even in undisturbed communities at BSC, while the richness was lower in very old and undisturbed communities at HWC relative to the old A L D S . BSC communities tended to have a higher diversity of forbs, with some occurring exclusively at BSC, such as Saxifraga nivalis, and S. flagellaris while the shrub species of Dryas integrifolia and Cassiope tetragona, occurring at HWC, were absent or rare at BSC. The results suggests that communities at BSC did not 80 develop beyond the Late sere grass/forb stage of succession, even in the oldest or undisturbed communities. It could be postulated that i f shrub species (Dryas or Cassiope) establish in the later 3rd or 4th stages of succession, they may inhibit the establishment of other late successional forbs or grasses, consequently resulting in lower species richness, as in HWC. This possible effect of biological interaction would be more typical of lower resistance environments [sensu Svoboda and Henry, 1987). In true marginal environments, the difference in composition between the pioneer and the remaining later stages of succession is not obvious, according to Svoboda and Henry (1987). On the Fosheim Peninsula, the pioneer community group was quite distinct from the oldest community groups found in the scar. Although there is a continuous but shifting presence of the ruderal grass Puccinellia spp in the first 3 stages of succession expressed mainly by a decrease in dominance, it was none the less absent in the oldest community. The large gradient lengths of the first ordination axis (4.8 SD) confirms that sites at opposite ends of the axis had few species in common (Jongman et al., 1995). Hence, primary succession in these sites was not typical of highly marginal environments. Unlike other primary succession environments in the Arctic where mosses are prominent colonisers, the pioneer species in my sites were grasses and forbs. Jones (1997) also observed the simultaneous colonisation of forbs along with mosses, in glacial forelands in the High Arctic, and attributed this to a favourable moisture conditions. However in my study area, high moisture availability occurred during such a short period in the spring that mosses may not establish successfully. Mosses only appeared in old and very old disturbances where a constant higher moisture content was available. 81 The transition to later stages of succession was progressive and continuous, as was illustrated by the large overlap between the communities in the ordination. Although there was a shift in species dominance over time, there was generally not an obvious replacement, except for very old sites in which many species were excluded. Typically, in undisturbed terrain most species were found with moderate cover. At a large scale, Heikkinen (1996) suggests that species diversity is most significantly correlated with abiotic and energy-related factors, such as mean annual heat and humidity variables. This would agree with the generally lower diversity of species over the Fosheim Peninsula relative to subarctic or temperate environments. In succession, peak species diversity is believed to occur between the pioneer and intermediate stages of succession (Matthews, 1992), followed by a decline in later stages as a result of increased competition (Elver and Ryvander, 1975). Theories predict that plant species diversity wil l be maximised at intermediate levels of vegetation disturbance (Grime, 1979; Denslow, 1980; Fox, 1983). Only at HWC was this tendency observed, where highest diversity was noted in old disturbances, while the very old disturbance and undisturbed communities had lower diversity. This may be a reflection of how slow succession progresses in high arctic ecosystems, such that intermediate stages as described by Matthews (1992) are associated with old disturbances ranging between 20 to 50 years of age. It is indeed possible that competition may have been partially responsible for this reduction in total numbers of species, since very old disturbances were characterised by more favourable conditions, and dense plant cover compared to undisturbed tundra. Overall, it can be suggested that the presence of scars in the landscape of the Fosheim Peninsula provides small oases of 82 favourable conditions with higher soil surface temperatures and more soil moisture, which contributes to greater overall species and community diversity in the landscape. 3.4.2 Application to succession models In the scar environment, which a typified primary successional environment, succession progressed almost as in the classical models directional yet with only weak replacement. It would not be classified as a Relay Floristic (RF) model, but rather as a shift in species dominance due to different growth rates and life history of the species, such as proposed by the Initial Floristic Composition (IFC) model (Egler, 1954). However, young disturbances and very old disturbances shared very few species, suggesting a slow yet eventual species replacement over time. The patterns observed suggest a combination of the first 2 models proposed by Svoboda and Henry (1987) for low and high resistance environments, with an initial directional non-replacement succession, followed eventually by replacement. However, this pattern of eventual replacement may simply reflect how slowly succession progresses in the High Arctic. Secondary succession occurring on the toes of these slides is expressed by a different pattern. Although initial environmental conditions following a disturbance were favourable (intact but churned soil with good soil moisture, and seed pool from parent plants and intact seed bank), the plant community nonetheless changed with reduced species richness and total vegetation cover over time. The plant succession appeared to be retrogressive (sensu Svoboda and Henry, 1987). Environmental conditions overwhelmed the effects of biological interactions, as proposed in the models of increased severity (Svoboda and Henry, 1987; Matthews, 1992). Because of the micro-topographic characteristics of the toe, the environment became drier as the water drained away from the raised mound in the terrain. 83 The control of the physical environment on the species composition was well illustrated the D C C A diagram (Figure 3.13), where the separation between the age categories of the toe, in the ordination space, occurs along axis 2 which was defined as a moisture gradient. On the first axis representing an age factor complex, all toe-related categories occur close to the origin emphasising the negligible effect of these sites in defining this axis. This shows that successional patterns cannot strictly be predicted by large scale environmental factors such as degree days, temperature or length of growing conditions. Local or micro-scale environments may produce conditions such that, in the same geographic location, succession can be directional in one area and can then regress in another area. 3.4.2.1 Processes controlling succession Many models propose that species life-history could suffice to explain successional change in plant communities (Egler, 1954; Noble and Slatyer 1980; Walker et al. 1986; Huston and Thomas, 1987; Walker and Chapin, 1987). The pattern of species change without strong replacement observed on the Fosheim Peninsula would appear to agree with this theory. However, as suggested by Walker and Chapin (1987), succession most likely results from a combination of processes that will vary in relative importance depending on the stage of succession, the type of succession (primary or secondary) and the level of environmental severity measured by the availability of resources such as water or soil nutrients. In the early stages of succession, processes influencing colonisation will be most important, while later stages will be affected by processes affecting senescence and mortality. Matthews (1992) proposed that processes in the early successional stages are controlled by 84 environmental conditions (allogenesis) while biological processes (autogenesis) will direct later successional stages. This agrees with my results on the Fosheim Peninsula. Environmental conditions were most severe in the early stages (e.g., scars) and establishment was limited to species which germinate rapidly in the spring after snow melt and adapted to the high salt content in the soil. In the later successional stages, moisture was not as limiting and vegetation cover was higher, and thus biological processes were likely more important. Competition is said to be less important in severe habitats (Savile, 1960; Svoboda and Henry 1987; Bliss and Peterson, 1992; Grime, 1979). Yet i f evidence of competition existed, it would be predominate at the maturation or senescence stages of the succession as plant density increases, since competition is generally associated with high biomass accumulation where resources or space may become limited (Grime, 1979). In mature undisturbed communities such as those found on the flat or rolling plateau of the Fosheim Peninsula, one could postulate that competition is not significant, as most species are present and coexist, and bare spaces are still frequent. In these environments soil moisture is low, the plants are subjected to desiccating winds and are less protected from extremely low temperatures in the winter due to the thin snow cover. Plants establishing would need to be high stress tolerators (sensu Grime 1979), rather than adept competitors, to survive these more limiting environmental conditions. In the valleys that dissect these plateaus where active layer detachment slides occur, and particularly in old and very old disturbances, environmental conditions tend to become more favourable. This, combined with increasing vegetation cover and species density, likely increase competitive interactions, although no experiments were designed to test this hypothesis. Circumstantial evidence was found in very old scars where vegetation cover was highest, and numerous species found in younger slides were not 85 present. A n unique community assemblage occupied this site, which included important moss cover, the shrub species Cassiope tetragona, and a denser cover of Dryas integrifolia. The primary succession sequence in my study area also seemed to support the hypothesis that competition increases with maturation of the community. The pattern (illustrated in the Figure 3.15) of increased resources (moisture) associated with an eventual decrease in productivity (total vegetation cover) would, according to Tilman (1985) result from competitive interactions. However, in the secondary succession, changes in the community composition at the maturation stage appeared to be driven by changes in the physical environmental severity rather than by competitive exclusion. Facilitation processes are believed to be important in severe environments and, primarily, during the stage of colonisation and early community development (Lawrence et al. 1967; Connell and Slatyer 1977). However, my results suggest that the existing or previously established plants in a disturbance did not seem to aid in the establishment of others. It was rather a combination of physical environmental conditions, independent of the plant community changes, that appeared to facilitate the establishment of new individuals, such as surface cracks, localised moisture pockets, leaching of salt cover or added surface sand. In addition, the timing of seed dispersal with respect to favourable soil conditions can be critical for successful germination and seedling survival (e.g. Walker et al, 1986), and thus play a strong role in defining the species found in pioneer communities. Changes in species dominance in the mid to late succession stages could be explained by differential longevity or growth rates, as the importance of these processes are believed to increase with aging terrain (Walker et al, 1986; Huston and Smith, 1987; Chapin et al, 1994). The shrub community on the Fosheim Peninsula is characterised by slower growing 86 species that are long lived. Within this group Salix arctica is known to have the fastest growth rates, especially female plants, in comparison to Dryas integrifolia and Cassiope tetragona which are slower growing, and very long lived (Johnstone and Henry 1997; Jones et al. 1999). Salix is the dominant shrub to occur in old disturbances, while Dryas and Cassiope occur mainly in the oldest disturbances. In contrast, herbaceous plants are quick growing and short lived, such as the grasses, occur early in the succession. In the secondary succession characterised by a retrogression, the community changes from an amalgamation of numerous species to a community where only the long lived species remain. These sites were dominated by stress tolerators, such as those with well establish root systems such as Salix, which permitted survival of the harshest conditions and herbivory by musk oxen. Although the ruderal species populations decreased over time following a disturbance, while the later sere populations increased, both types of species coexist in undisturbed communities. Thus, the decline of the population of ruderal species in older disturbances did not necessarily result from biological processes such as competition, but rather may have been the consequence of the physio-chemical changes associated with the leaching which is independent of the plant community. Although this hypothesis was not experimentally tested, the patterns of environmental changes seem to support this proposal. 87 Chapter 4: The role of sexual reproduction in plant succession in high arctic ecosystems 4.1 Introduction and rational In high arctic ecosystems, establishment of new individuals from seeds is not believed to be as important as vegetative reproduction. The cool, short growing season and resource-poor conditions of these habitats is said to limit all phases of sexual reproduction by reducing flowering or seed maturation (Billings and Mooney, 1968; Bliss, 1962; 1971; Savile, 1972; Bell, 1975; Grime, 1979; Bell and Bliss, 1980; Archibold, 1984). Successful reproduction, whether from seed or vegetative means, is often infrequent for the long lived tundra plants in the High Arctic (Bell and Bliss, 1980; Levesque and Svoboda, 1995). However, these stress tolerator plants (sensu Grime, 1979) have been found to invest resources towards the production of flowers and viable seeds, particularly during favourable climatic years. (Urbanska and Schutz, 1986; Svoboda and Henry, 1987; Wookey etal. 1993; 1995; Krannitz 1996; Henry and Molau, 1997). It is believed that many of these arctic stress tolerators or long lived clonal plants show phenotypic plasticity in response to environmental change. In difficult years these plants rely on vegetative reproduction, while in good years a successful seed set insures genetic diversity of the population (McGraw, 1993; Wookey et al. 1993; 1995; Khodackek, 1997; Bliss and Gold, 1999). Good climatic years with increased temperature can result in invasion and colonisation from seeds of the high proportion of bare ground. Compared to the sub-arctic where clonal proliferation may be more important than recruitment from seeds due to the closed vegetation cover, favourable years in the High Arctic may influence community change depending on the successful germination of seeds available. The seeds originate 88 either from recent or on going introduction via the seed rain produced from the growing community or from the in situ reservoir of the soil seed bank. The contribution of sexually produced diaspores to the maintenance of a population, or to the initiation of community change, has not been treated adequately in the High Arctic. Yet it is an important variable in succession, where vegetation change is the result of interactions between invasion, establishment, maintenance and senescence of species in time and space. Although there has been a number of descriptive studies in the High Arctic on seed bank composition (Freedman et al., 1994; Levesque and Svoboda, 1995; Jones, 1997) there is still a need for more data to better understand, the seed dynamics in these ecosystem and how it differentiates from temperate environments from which many models of succession have been developed. The sexual reproduction processes, such as seed production, seed rain and dispersal, seed bank and storage, dormancy, germination and seedling establishment, are all important biological forces that can direct the course of vegetation succession (Svoboda and Henry, 1987). Some ecologists believe that life history traits such as seed dynamics, growth rate and longevity can suffice to explain succession (Drury and Nisbet, 1973; Huston and Smith, 1987). These processes therefore need to be understood in order to better understand succession. The main objective of this chapter is to examine sexual reproduction processes and assess how they relate to vegetation patterns and community changes throughout the successional sequence described the previous chapter. For this purpose, the potential for establishment from seeds throughout the successional sequence was evaluated through measurements of a) successful production of viable seed from various species at different stages of the succession following one growing season and b) changes in species 89 composition, density and diversity of the seed stored in the soil seed bank through time and in comparison to the vegetation community. More specific questions include: Successful seed production (Seed Harvested): 1) In an aging community, it is expected that population of ruderal species will decrease. Is this trend matched by reduced production of viable seeds for the same species group throughout the successional sequence? Is the pattern different for late stage species? Does the number of viable seeds change for different successional types (primary succession in the scar and secondary succession in the toe)? 2) Does the production of viable seeds change with the density of the extant vegetation? Although competition is not believe to be an important mechanism in high arctic ecosystems, it most certainly occurs in the individual pockets of more productive communities. When competing for limited resources, allocation to seed production may be reduced. 3) Is there a difference in germination rate between species? Given that the High Arctic experiences only short growing season, the speed of germination can be critical. Seed storage (Soil seed bank): 4) Throughout the successional sequence, does the seed bank density and composition change over time and is there a consistency in the pattern between the two geographic locations on the Fosheim Peninsula? 5) How do the seed bank density, composition and diversity patterns compare to the extant vegetation? 6) Which factors (biotic and abiotic) appear to explain most of the seed bank density? 90 4.2 Study area Research was conducted on the Fosheim peninsula (80°N, 84°W), on the west coast of Ellesmere Island. Two locations were chosen for sampling: Hot Weather Creek (HWC), and Big Slide Creek (BSC). Both location are characterised by fine silts and clay sediments, which are conducive to active layer detachment (Lewkowicz, 1992). Although both locations have similar environmental conditions, BSC, being at a slightly higher elevation (200 m above sea level compared to 70 m for HWC), would likely have been exposed earlier during the isostatic rebound following deglaciation (Bell, 1996). Vegetation of Big Slide Creek was typically characterised by a larger diversity of non-woody forbs compared to Hot Weather Creek (see Chapter 3). 4.3 Material and methods 4.3.1 Seed harvest The seed harvesting was done only at HWC for 8 slides (sites): 2 young, 2 intermediate, and 2 old on the west side of Hot Weather Creek, and 2 young on the East side. Collections were made for the two disturbance types (the scars and toes of the slides), as late as possible in the season (August 13th) in order to ensure maximum maturation of the seeds. Species harvested were Braya purpurescens, Papaver radicatum, Puccinellia angustata, and Poa glauca. The focus was placed on ruderal or invader species as these are most likely to depend on seed production for establishment. The grass species Poa glauca is a late sere species, and was included for comparison with the ruderal species. For each species, 5 individuals per slide and disturbance type were harvested when possible. Although an attempt was to set up a balanced design, it was not always possible since certain species were simply not present at some of the sites or disturbance types. 91 Seed viability was measured as a percentage of successful germination. The seeds were collected in paper envelopes and stored in cool dry environment (~7-12°C) until germination trials began on 16 June 1995. For each sample, 100 seeds were counted and spread evenly over a triple thickness of Whatman No. l filter paper, which lined 50 mm plastic petri dishes. Germination proceeded in a growth chamber at the University of British Columbia under 24 hour light and controlled temperature (20°C for 12 hours and 10°C for 12 hours) to roughly simulate arctic conditions. Seeds of most arctic species germinate best between 12°C to 20°C (Bliss 1971, Bell and Bliss, 1980). The dishes were monitored for seedling emergence every 2 days for 20 days, kept moist and rotated within the growth chamber to reduce any potential chamber effect. As soon as the first cotyledon broke the seed coat and measured 1 mm, it was counted as germinated and removed from the petri dish. Statistical analyses were performed using SYSTAT (1992). Since the selected ruderal species were not found at all sites and disturbance types, the effect of time on the viability of seed was analysed by pooling the species together into two life history strategy groups (sensu Grime, 1979): ruderal species (r-strategists) and late sere species (K-strategists). The ruderal species included Braya, Papaver, and Puccinellia while the late sere species was Poa glauca. Although this is not the ideal technique, given that these species do not necessarily behave in similar ways, it was the best option given the data. However, even pooling by functional group of grasses and forbs (Chapin et al, 1996) within the ruderal strategy group did not produce data that lent itself to analysis. Graphs by individual species were produced, but analysis of variance across site and disturbance type was not possible. A l l data were tested for normality using a Kolmorgorov-Smirnov goodness of fit (Lilliefors) test and for homogeneity of variances using the Fm a x-test. Non-parametric tests 92 were used where the data did not meet the appropriate assumptions, or could not be transformed to normality. To evaluate the effect of terrain age and disturbance type (scar/toe) on seed viability (question 1), a two factor analysis of variance (ANOVA) was performed on the ruderal data. For the Poa data set, which was not normally distributed due to the large number of unsuccessful germinations, a Kruskall-Wallis one-way analysis of variance was repeated by disturbance type, to test the effect of terrain age. A Tukey's multiple comparison test was then applied to the ranked data (Conover and Iman, 1981). Pearson's correlation coefficients were used to test the relation between the overlying vegetation cover and the seed viability per species (question 2). Averages by site (slides) were used in this analysis, and only sites which had seeds available were used for the correlation. 4.3.2 Seed bank For the evaluation of seed bank dynamics over the succession sequence, a total of 24 active layer detachment slides in all age categories were identified for sample collection. The number of sites in each age category and at each geographic location is shown in Table 4.1. Table 4.1: Number of sites (slides) by age category and geographic location from which seed bank samples were taken. Hot Weather Big Slide Creek Creek Young 7 2 Intermediate 3 6 Old 3 4 Very Old 1 93 Samples were taken following the same transects used for the vegetation quadrat data collection (figure 2.2). Depending of the size of the slide, a range of 3 to 8 samples were collected haphazardly along the transect, for both the scars and the adjacent undisturbed area of each slide. Toe samples were not included in this analysis due to limited time and space available in the U B C greenhouses. Position of the samples along the transect was recorded as well as proximity to undisturbed vegetation. Seed bank samples (10 cm x 10 cm x 5 cm) were collected with a trowel on August 6th.and 7th for Hot Weather Creek, and on July 31st for Big Slide Creek. The samples were packed in paper bags and kept cool until they were shipped back to the lab at the University of British Columbia, where they were air dried and stored at room temperature (~20°C) until germination trials began (February 16, 1995). Each soil sample was passed through a 2 mm sieve, mainly for the purpose of breaking down the hardened silt-clay material. The samples were thoroughly mixed and the total volume of the sample was measured (ml) using graduated beakers. A splitter was then use to reduce the initial volume, and a 130 ml subsample was used for the germination trial. Because the soil material was very rich in clay, the standard arctic method using 100 ml petri dishes could not be used (Levesque and Svoboda, 1995; Levesque et al, 1996), for fear of the sediments drying too quickly. Instead, a method used at the U B C Botanical Gardens Nursery was adapted for this purpose. For each sample, sterile potting soil (peat, dirt and vermiculite) covered the bottom 4 cm of a 20 x 12 cm pot. The 130 ml seed bank sample was then sprinkled on top with a thickness of approximately 0.5 cm. A fine layer of pebbles was then sprinkled over top to help retain moisture. Numerous tests performed before 94 initiating the germination trial showed that the pebble layer did not limit the growth of the seedlings but did help to maintain moist conditions for at least 2 days (Levesque et al, 1996). The samples were thereafter watered and monitored for seedling emergence every 2 days for 6 weeks in a greenhouse of the U B C Botanical Gardens Nursery. The seedlings were left to grow until identification could be made. After the 6-week period, unidentified seedlings were transferred to a standard potting soil where they continued to grow. Identification was predominantly based on vegetative characteristics, since flowering was rare. For the grasses, identification could be made only to the genus level because of the lack of flowers. Unidentified seedlings were classified as unknown monocotyledons or dicotyledons. Although detail is available to the species level, for the purpose of analysis of the change of seed bank composition over time, the data were combined into 4 groups based on functional types (Chapin et al. 1996) and life history (Table 4.2). Table 4.2: Seed bank species included in the four groups Life History Group Species included Ruderal graminoids: Ruderal forbs: Late sere graminoid: Late sere forbs and shrubs: Deschampsia brevifolia, Puccinellia spp Brayapurpurascens, Papaver radicatum, Stellaria spp., Poa spp., Agropyron violacium, Alopercurus alpina, Arctorgrastis latifolia Kobresia myosuroides, Luzula spp. Dryas integrifolia, Cerastium alpina, Draba alpina, Draba nivalis, Erysimum palasii, Melandrium spp, Oxyria digyna, Potentilla spp, Ranunculus, spp, Taraxacum spp, 95 The number of viable seeds per m 2 was calculated using an adapted version of the formula from Levesque and Svoboda (1995), G = g Vta* 100 where, G = germinable seeds/m , g = number of germinated seedlings in a subsample, Vta = Volume of the < 2 mm portion, Vtb = volume of subsample. Each 10 cm x 10 cm sample was converted to 1 m by multiplying the formula by 100. In general, the recorded number 2 2 of seeds per m were very high, averaging between 700 to 2000 seeds/m per sample and reaching up to 15000 seeds/m in a few unusual cases. These high values result most likely from two factors: 1) the samples were taken late in the season, and there was a risk of including the seeds set in the current year's crop, and 2) the samples were collected to a depth of 5 cm. This creates a large total volume (Vta) and therefore an inflated Vta I Vtb ratio which is multiplied by the number of germinated seeds (g). Samples to 5 cm depth were probably not necessary since, in polar semi-desert, the majority of the seeds is expected to be found in the upper 2-3 cm (Freedman et al, 1982; Levesque and Svoboda, 1995; Jones 1997). SYSTAT (1992) was used for all statistical analyses. A l l data were tested for normality using probability plots and the Kolmorgorov-Smirnov goodness of fit (Lilliefors) test. The seed bank data did not meet the appropriate assumption of normal distribution and could not be transformed to normality. Therefore nonparametric tests were used. Although samples collected within 1 meter of undisturbed vegetation were labeled accordingly (scar edge), the analysis of seed bank change over time proceeded by averaging all samples, including those labelled scar edge, within a disturbed area of a slide. Nonetheless, the effect 96 of vegetation proximity (samples within 1 meter of scar edge) was evaluated for both BSC and HWC. At BSC this additional information was collected for all age categories while for HWC it was done only for young age terrain. Although the strongest apparent differences appeared in the young disturbances ( Figure 4.1), a Kruskal-Wallis one-way analysis of variance by age categories showed that, for BSC, there were no significant differences due to the effect of vegetation proximity (P=0.286, 0.781 and 0.972 for young, intermediate and old disturbances respectively) (Appendix Table 6.1a). For young disturbances at HWC, however, proximity to vegetation appeared to result in larger seed banks (Figure 4.2, P=0.018 Appendix Table 6.1b). In the control communities, samples taken in extremely dry conditions were distinguishable from other controls. However since there were no significant differences in the germinable seed bank between the 2 categories (Figure 4.2; Appendix Table 6.1b (P=0.069) ), the distinctions were not maintained for the remainder of the analysis. The effect of aging terrain on seed bank density (question 4) was evaluated for each geographic location using a Kruskal-Wallis one way analysis of variance. This analysis was repeated for each life history group. Differences between individual slides were evaluated independent of age to see the effect of local environmental conditions on seed bank numbers. The relation between the germinable seed bank and the vegetation cover was examined using Pearson's and Spearman's correlation coefficients (question 5). For this purpose, seed bank values represent the average of 5 samples per site (individual slide). A regression analysis was also conducted to evaluate which biotic and abiotic variables had the largest influence on the seed bank total numbers (question 6). Variables included were: geographic location, presence or absence of surface sand, soil compaction, micro-topography, 97 age of the disturbance, density of desiccation cracks, absolute slope, topographic depression index, total vegetation cover and litter cover. The assumptions of a normal distribution, linear relationships, and equal variances were verified by examining the residuals of the regression analysis (Sokal and Rohlf, 1981; Steel et al, 1997). A square root transformation of the seed bank total numbers was used to comply with the assumptions of normality. An analysis of the residuals of the multiple regression showed that the variances were not linear but were relatively equal. Regression analysis is said to be fairly robust for non-linear relationship (Sokal and Rohlf, 1981; Steel et al, 1997). The 54 samples (out of 188) with zero germination made it relatively impossible to transform the data to fill all the assumptions. Four outliers were identified in the initial regression and removed from the final analysis. The number of seeds counted in these samples exceeded 3 standard deviation from the mean (1 391 + 2142 SD seeds/m2) and thus corresponded to unusually high total seed bank values (15 369, 15 365, 9 154, 8 538, seeds/m2). Both a regular multiple regression and a forward stepwise regression were used on the reduced data set. 98 a) Differences between Scar and Scar Edge 2000 •i 1500 to •o CD V CO o 1000 n re 500 Scar Scar Edge Disturbance type b) Differences between Control and Control Extreme 2000 CM | 1500 •a o a m a> 1000 a re c § 500 8 Undist Undist Dry Disturbance type Figure 4.1: Germinable seed bank at Big Slide Creek, comparing samples from a) scar and scar edge and b) from the exposed dry undisturbed controls and the more moist undisturbed controls. The data are the average of the combined life history group data set (+SE). CM E w •o o a> (A o> X} ro c CD 4500 4000 3500 3000 2500 2000 1500 1000 500 0 • Scar DScar Edge Young intermediate Age category Old Figure 4.2: Germinable seed bank at HWC, comparing samples from scar and scar edge over time. The data are the average of the combined life history group data set (+ SE). 99 4.4 Results 4.4.1 Seed Harvest Germination trials of the harvested seeds show a marked difference in germination success between the two life history groups (Figure 4.3). On average, ruderal species had a much higher germination success (53% + 2.1) than the late sere grass species (6.7% + 1.2). This was consistent for all the species in the ruderal group. Brayapurpurescence had the best success and was significantly different from the other 2 ruderal species Papaver radicatum and Puccinellia spp. (Table 4.3). Table 4.3: Percent germination of harvested seeds a) by life form group and b) by individual species within the ruderal group, n = number of petri dishes 21 Species Mean (+SE) SD n Late sere species group Poa glauca 6.7 (+ 1.2)* 23.3 120 Ruderals group 53.0 (+2.1) 13.2 123 * Significant difference (t = 18.96, DF = 241, p-value = 0.000) b) Individual species within the Ruderal group: Mean (+ SE) n Puccinellia sp. 46.1 (+2.7)b* 91 Papaver radicatum 51.6 (+3) b 114 Braya purpurescens 69.5 (+2.1) a 58 * means with same letter are not significantly different (ANOVA followed by Tukey HSD test, p=0.000). During the germination trials many of the Poa glauca samples rotted or suffered from mould, and often within the first few days of the germination period, which may explain some of the poor germination success. The lemmas that were left on the seeds might have been 100 harbouring the mould or fungus responsible for the rotting. However, this was infrequent for the other species, including the Puccinellia species, which also had the lemmas still enveloping the seeds. As observed in the field and in the seed bank, the lemmas remain on the seed following seed set and during germination in the natural environment. The two factor A N O V A indicated that for the ruderal species the production of viable seed appeared to be affected by terrain age, and more importantly, the terrain age and disturbance type multiple factor (Appendix Table 5.1). The subsequent Tukey multiple comparison test suggested that significant differences existed between old toes and all other environments including undisturbed terrain. Further analyses by disturbance type (scars/toes) were conducted in order to test for the effect of terrain age on seed production. In both the scar and the toe environments of young disturbances ruderal species appear to produce similar number of viable seeds (Figure 4.3b). In the scar environment the separate analyses indicated that there was no effect of terrain age on the seed viability for either ruderal or late sere group (p=0.624 and p=0.236 respectively; Appendix Table 5.2 and 5.3). Ruderal species seem to maintain a high resource allocation for seed production as seen from the apparent increase in seed viability over time. However, in the toe environment, the data suggests that the germination success of ruderal species decreases in older disturbed terrain (Figure 4.3b). Data for Papaver radicatum, the only species that was found in the 3 different age categories, showed this same pattern of decreased production of viable seeds over time, in the toe environments (Figure 4.4). The one way A N O V A on samples from the toe environment showed a significant effect of terrain age on seed germination for the ruderal species group (p=0.006, Appendix Table 5.2b). The Tukey multiple comparison test indicated that the significant effects were between the young and old toe communities and 101 80 60 a) Late sere species • scar • toe Disturbed Undisturbed 40 0> a 0. 20 Zero germination success Age category undist. b) Ruderal species undist. Age category Figure 4.3: Percent of germinated seeds from the seed head harvest by age categories life history groups. Age categories are: y = young; i = intermediate; o = old; undist.= undisturbed. 102 a) Papaver radicatum undist. b) Puccinellia spp. c o O) c 0) cu 0. undist. c) Braya purpurescens 80 No seeds found in age category i o Age categories undist. Figure 4.4: Germinated seed by individual species for the Ruderal life form group. Age categories are: y = young; i = intermediate; o = old; undist.= undisturbed. 103 between the old and undisturbed controls (p=0.044 and p.0.004 respectively). Finally, as expected, undisturbed communities found adjacent to both scars and toes had similar high seed viability levels, somewhat comparable to the young scar disturbances. Increased vegetation cover appeared to result in a lower production of viable seeds, as illustrated in the scatter plot of Figure 4.5a. Each point in the figure reflects an average of 5 samples per slide for the total number of seeds germinated from all four species. Pearson's correlation confirms this weak yet significant negative correlation (R = -0.404, P=0.036) (Appendix 5 Table 5.4). However, for the two separate life form groups, the data showed no significant correlation with total vegetation cover. Each species not only had a different successful germination but also a different rate of germination initiation (Figure 4.6) which may be a critical factor in the High Arctic where periods with favourable environmental conditions can be short. The ruderal species Braya has a high germination success and also germinated fairly quickly: 20 to 30% within the first 5 days, 40 to 60% of the seed had germinated within the first 10 days. At the other extreme, Poa glauca (a late sere species) not only had poor germination success, but also did not germinate rapidly. In many cases up to 16 days would elapse before any germination would occur. Puccinellia spp. and Papaver radicatum displayed a similar rate of germination initiation between 10 to 16 days, Papaver radicatum and Puccinellia angustata continued to germinate at a steady rate up to the 25th day. Therefore, the longer good environmental conditions prevail, the higher the potential establishment from seed would be for these species. There were no important differences in the rate of germination for the different disturbance types or terrain age categories, except for species on the old toes which had, overall, poor germination success. 104 b) 30.00 20.00 10.00 -f 0.00 i) Late Sere Group • R2 = 0.150 P=0.412 in TJ 0) o in •a o re c i 8 0 10 20 30 40 50 60 70 80 90 80.00 60.00 40.00 20.00 0.00 ii) Ruderal Group • R2 =-0.105 P=0.422 • * v • • • • • • 1 1 1 1 — 0 10 20 30 40 50 60 70 80 90 Total Vegetation cover (%) Figure 4.5: Relation between germinated seeds and vegetation cover, a) Total seeds harvested by site, b) Seeds of the two life groups. Samples were averaged by site and disturbance type (scar, toe and control) for a total of 27 separate cases. Cases where no seeds were found during the harvest were not included in the regression. 105 Scar Environment -Poa Puce -B- - Poppy X — Braya m •a a> cu u Q) re c E k_ CO 90 80 70 60 50 40 30 20 10 0 Young • 90 80 70 60 50 40 30 20 10 0 5 10 16 25 Intermediate 37 • • J r y r - * 1 r 1 <* -r B*-^ * • • <i 90 80 70 60 50 40 30 20 10 0 5 10 Old 16 25 37 ii^ 10 16 25 37 Days Toe Environment 90 80 70 60 50 40 30 20 10 0 Young -m 10 16 25 37 Intermediate 90 80 70 60 50 40 30 20 10 Old o m— 5 10 re 25 37 Days Figure 4.6: Rates of germination for individual species by age category and disturbance type (Scar,Toe) 106 4.4.2 Germinable seed bank A l l species found in the seed bank were also present in the above ground community. A total of 23 species (Table 4.4) were counted in the emergent seedlings compared to the 42 species found in the extant communities of HWC and BSC. Most of the species not present in the seed bank are typical of late sere species of mesic undisturbed tundra communities (e.g. Salix arctica, Polygonum viviparum, Pedicularis species, Cassiope tetragona, Festuca brachyphylla, Saxifraga species, etc.). The ruderal grass Puccinellia angustata made up more than a quarter of all the seedlings counted, followed by the ruderal forb Papaver radicatum (10%), Deschampsia brevifolia (8.6%) and then the later sere grass Poa glauca (8.4%). Many species of the late sere forb group comprised up to 23% of all the seedlings counted. Contrary to the germination trials of the harvested seeds, Poa species had good germination success in the seed bank comprising 17.5% and 11.5 % of all seeds counted for HWC and BSC, respectively. Overall, Big Slide Creek had larger seed banks than Hot Weather Creek for each age category and for the late sere forb category (Figure 4.7 and Table 4.4). 4.4.2.1 Seed bank change over time, in composition, density and diversity At HWC seed from the ruderal grass were found almost exclusively in young and intermediate aged disturbances. It is not however, the dominant species in the seed bank of these age categories. The late sere grasses and forbs made up more than 50% of emergent seedlings, contrary to the extant vegetation communities which were typically dominated by the ruderal grass Puccinellia spp. The late sere species represented an important proportion of the seed bank in all age categories except for the very old disturbances, where the ruderal 107 Table 4.4: List of species found in the germinable seed bank samples and total number of seedlings counted by species and for each geographic location. All 193 samples HWC: 108 samples BSC: 85 samples Species % of total viable % of HWC viable % of BSC viable seeds (984)1 seeds (417) seeds (567) Ruderal Grasses 35.5 26.9 36.3 Puccinellia angustata 26.8 14.4 36.0 Deschampsia brevifolia 8.6 12.5 0.4 Late Sere Grasses 25.1 29.7 21.7 Poa spp 14.0 17.5 11.5 Poa glauca 8.4 13.9 6.0 Poa arctica 1.7 2.6 1.4 Poa alpina 0.4 1.0 Agropyron violacium 4.6 5.5 3.9 Kobresia myosuroides 3.8 3.6 3.9 Luzula spp 1.6 1.9 1.4 Alopercurus alpina 0.6 1.1 Grass unknown 0.5 1.2 Ruderal Forbs 16.2 20.6 4.6 Papaver radicatum 10.0 19.4 4.0 Stelleria spp 5.8 2.2 8.5 Braya purpurascens 0.5 1.2 1.6 Late Sere Forbs/Shrub 23.2 19.1 37.4 Melandrium spp 8.23 3.4 11.8 Potentilla spp 3.8 3.6 3.9 Draba nivalis 3.7 3.4 3.9 Dryas integrifolia 2.4 5.8 Cerastium alpina 1.1 0.5 1.6 Draba alpina 0.7 1.0 0.5 Arctogrostis latifolia 0.5 0.9 Ranunculus spp 0.5 0.9 Oxyria digina 0.4 0.2 0.5 Eryserum palasii 0.1 0.2 Taraxacum spp 0.1 0.2 Dicots unknowns 1.6 2.6 0.6 108 forbs Papaver radicatum and Stelleria spp were the exclusive species. Undisturbed communities had a relatively equal distribution of all the life history groups in their seed bank, as in the extant community (Figure 4.7a). In the seed bank at Big Slide Creek however (Figure 4.7b), the ruderal grasses were distributed almost equally between all age categories, and there were no significant differences between categories (p=0.716, Appendix Table 6.2b). The younger disturbances had mostly grasses in their seed banks while forbs, both ruderal and late sere species, appeared more predominantly in the old and undisturbed communities. These patterns were also observed in the above ground vegetation communities of this location. As in HWC, the late sere grasses were present in all age categories with no significant differences between the categories (P=0.239 and P=0.144 for HWC and BSC respectively) (Appendix Table 6.2). Seed bank density and species diversity changed over time (Figure 4.7). If intermediate age slides are ignored, both HWC and BSC display a pattern of slightly increasing number of germinated seeds in the seed bank, between young and old disturbances, followed by a slightly lower density in the undisturbed control sites. Changes in species diversity followed the same pattern as the total germinable seed bank. There was a trend towards increasing number of species with older terrain, until a threshold is reached where the total diversity appears to decrease in undisturbed sites. This was even more apparent in the unique case of the very old disturbance at HWC where the species diversity dropped dramatically (Figure 4.7). However, these conclusions can only be inferred from the patterns as none of the seed bank differences due to age were statistically significant (Appendix Table 6.3 and Table 6.4). 109 Since there were no significant differences between age categories at HWC or BS, differences between sites was investigated. From Figure 4.8, it is clear that total number of germinated seeds was highly variable between sites, independent of age. At BSC, very high seed bank densities in the scar at site SB49 were accompanied by very high densities in the undisturbed community, emphasising once again, the strong influence of local conditions or site effects. In order to determine whether a disturbance in the landscape has an effect on the number of seeds in the soil, samples were averaged by scars and undisturbed community, independent of age categories (Table 4.5). At HWC the scars appeared to hold a higher number of total seeds compared to the undisturbed community, while a reverse pattern was noted for BSC. However, a Kruskal-Wallis analysis of variance showed that the differences between disturbed and undisturbed environments was again not significant for either locations (P=0.903 and P=0.631 for HWC and BS, respectively - Appendix Table 6.5). Table 4.5: Comparison of germinable seed bank for disturbed and undisturbed communities, at HWC and BSC. HWC BS Scar Undisturbed Scar Undisturbed Total seeds germinated: 1158 +215 848 +223 1658 + 360 2109 +543 Ruderal Grarninoids: 240 +95 42 + 16 * 654+ 163 649 + 301 Ruderal Forbs: 153 +63 332 + 203 182 + 102 238 + 87 Late sere Grarninoids 560 +134 269 + 88 551 +108 514+141 Late sere Forbs: 205 +59 205 +45 * 271 + 103 709 + 543 * * Significant differences between scar and undisturbed communities (P < 0.05, Appendix Table 6.2) 110 a) Hot Weather Creek: 1 1 rud gr 1 imrifnrh rrrnn ate qr lismwii #ate forh K species 500 Y I 0 Undist Age categories Figure 4.7: Mean (+SE) germinable seed bank and species diversity in relation to age category for a) Hot Weather Creek and b) Big Slide Creek locations. Age categories include Y=young; M=intermediate aged; 0=old; VO=very old; Undist=undisturbed control terrain. As well, for each age category, the relative importance of the 4 functional type is shown (rud gr = ruderal graminoids; late gr= late sere graminoids; rud forb = ruderal forbs; late forb = late sere forbs). I l l 112 4.4.2.2 Comparison of seed bank with above ground vegetation A vegetation quadrat was not available for each seed bank sample. Therefore, for the purpose of evaluating the relation between the seed bank and vegetation cover, the seed bank samples were averaged by slides. Data in Figure 4.9 and Figure 4.11 were compiled by first averaging the vegetation quadrat and seed bank samples by slides and then by age categories. This explains why the patterns of the seed bank change over time in this section, are slightly different then those of the previous section. For the data set combining both samples of HWC and BSC, both the seed bank and the vegetation cover showed a similar pattern where the total seed count and percent cover increase with aging terrain and then decreased in undisturbed communities. Again, differences in seed bank densities due to aging terrain were not significant as indicated in the previous analysis of variance (Figure 4.9). Species diversity was much higher in the vegetation than the seed bank, yet both communities had a similar pattern with increasing number of species over time (Figure 4.10). In the seed bank the total number of species decreased in undisturbed communities, while the vegetation had an greater number of species in undisturbed communities (BSC) or similar numbers as those of old communities (HWC). Once again, due to the large number of samples with no germination success, the differences between age categories are not significant (P=0.264 and P=0.344 for HWC and BSC, respectively). Even after removing the age factor there was no significant relation between the total vegetation cover and the seed bank density (Figure 4.11). Although the trend was towards a positive relation where the number of seeds increased with vegetation cover, the relationship 113 2 was weak and the Spearman correlation coefficient was not significant (R =0.019, P=0.372). HWC showed a negative coefficient (R2=-0.009, P=0.645, Appendix 6 Table 6.6). Therefore, other factors beyond the terrain age and the vegetation cover appear to control the seed bank composition and density. 4.4.2.3 Factors explaining the variation in seed bank density and diversity A regression analysis was done to see which other factor may influence the seed bank total number. Both the forward stepwise regression and the multiple regression using all independent variables showed similar results (Appendix 7 Table 7.1 and 7.2). The forward stepwise regression analysis retained, in order, the variables 'Location', 'Density of desiccation cracks', 'Litter', and 'Microtopography' in its final model with an R 2 of 0.147 and all four variables significant at the P=0.05 level. The regression combining all variables had a slightly higher correlation coefficient (R =0.190) and the significant variables corresponded to the ones in the final model of the stepwise regression. As well, the separate geographic locations appear to have different seed bank dynamics. Big Slide Creek, as seen in chapter 3 has a higher total vegetation cover, and also a greater seed bank density. It is interesting to note as well that the terrain age variable had no significant effect on the total number of seeds in the seed bank (P=0.941). 114 a) Combined data set of HWC and BSC: 3000 2500 E S. 2000 c ro S3 •a o a> V) T3 o> rs c 1500 1000 a co 500 M 0 age category Undist. b) Hot Weather Creek: 4500 4000 , 3500 I O age category Undist. c) Big Slide Creek: 4500 4000 <N 3500 - 3000 v> « 2500 a> ^ 2000 to 1500 c 'E 1000 <u o> 500 0 • Germinated seeds 0 Vegetative cover i 70 60 50 40 30 20 10 0 I Age category Undist. Figure 4.9: Mean (+/- SE) germinable seed bank and total vegetation cover by age category for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the separate data set of b) HWC only and c) BSC only. 115 a) Combined data set of HWC and BSC: 12 10 o <D Q. in co n E 3 C H§- — Germinated seed bank -0— Vegetation cover Undist. age category b) Hot Weather Creek: 16 "Is 12 o a> o. V) c) Big Slide Creek: in o o co a m CD o E 3 z 8 co 4 .a E 3 0 Germinable seed bank Vegetation to ! , , , I I Age Category O I O age category Undist. Undist. Figure 4.10 Species density (mean number of species +/- SE) in the germinable seed bank and the corresponding vegetation cover by age categories for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the separate dataset of b) HWC and c) BSC only. 116 a) Combined datasets of HWC and BSC: 2000 1750 1500 1250 1000 750 500 250 • • • • y = 3.205x + 660.83ft R2= 0.019 p=0372 0.00 20.00 40.00 60.00 vegetation cover (%) 80.00 100.00 b) Hot Weather Creek: 2000 g 1750 r 1500 1250 1000 750 500 250 y = -2.652 + 810.997 R2 = 0.009 p=0.64'l 0 0.00 c) Big Slide Creek: 20.00 40.00 60.00 vegetaion cover (%) 80.00 100.00 2000 g 1750 ~ 1500 | 1250 1000 » 750 .Q is c 500 250 y = 10.158 x + 357.622 R2 = 0.077 P=0.614 • 0.00 20.00 40.00 60.00 80.00 100.00 vegetation cover (%) Figure 4.11: Germinable seed bank in relation to plant cover for a) the combined data set of Hot Weather Creek (HWC) and Big Slide Creek (BSC) and for the two geographic locations separately: b) HWC and c) BSC only. 117 4.5 Discussion 4.5.1 Seed harvest and viability 4.5.1.1 Effects of time, life history groups and succession type (primary vs secondary) on seed viability As expected there was a clear difference between the two life history groups. The result that the ruderal species had consistently high germination success is consistent with other studies (Freedman et al. 1982) and agrees with Grime's theory that sexual reproduction is more important in the herbaceous or short-lived ruderal species. In a comparable study in the High Arctic, Freedman et al. (1982) found that the later successional species associated with the stress tolerator strategy (sensu Grime 1979) were more conservative in their reproductive effort, lived typically in closed, undisturbed habitats and propagated vegetatively. This was the case with Poa spp. which had very poor seed viability. Eurola (1972) also found that in controlled laboratory environment Poa glauca rarely had successful germination. It is possible that the conditions found in the environment, which will break the dormancy of Poa glauca, could not be achieved under laboratory settings. Although, at the time of the germination trials, it was felt that a vernalization period was not critical for arctic species, recent work suggests that repeated freezing of certain arctic seed have proven beneficial to increase and continue the germination success of the seeds (personal communication, Henry, 2000). The production of viable seeds can be influenced by resource availability such as light, moisture and nutrient supply, but also by environmental conditions such as available space, exposure of the plant to wind, winter conditions, frost disturbances, and other phenomena affecting the ground condition (Eurola, 1972). In my study area, the seed 118 viability decreased with successional time only in the toe environments of the active layer detachment slides (ALDS). As concluded in chapter 3, an important change over time in the secondary succession was the modification of the physical environment in term of soil moisture. In the toe, the soil was more exposed to wind action and became drier as water drained away from the mound created from the slide. Papaver, which typically has good seed production (Levesque and Svoboda, 1995; Levesque et al. 1996), produced a reduced number of viable seeds in the old toes, indicative of increased environmental stress in this habitat. As environmental severity increases, resource allocation may be concentrated toward survival rather than phenological efforts. Although most of my result were presented for the 2 life history strategist as defined by Grime (ruderals and stress tolerators) they could have been replaced by Molau's (1993) alternative model for arctic species of early- vs late-flowering species. In the scar environments favoured with increasing soil moisture over time, density of vegetation could reach up to 100% cover. With this increased productivity there is an increased potential for competitive interaction which could result in a decrease of resource allocation to reproductive effort. However, the seed production of ruderal species did not decrease and even showed a trend towards increased production, possibly as a result of improve moisture and other resources. Wookey et al. (1995) also noted the large phenotipic plasticity of arctic species in response to varying environmental conditions. Environmental constraints thus appear to be more limiting to sexual reproduction than biological interactions, as proposed by Svoboda and Henry (1987). 119 4.5.1.2 Rate of germination Appropriate moisture conditions needed for seed germination may be only available for periods of a couple weeks or less in the High Arctic. Species capable of rapid germination will have an advantage for colonizing a site and establishing themselves. Bliss and Gold (1999) found that species of polar deserts lack the ability to germinate quickly and thus may not be able to exploit favourable opportunities occurring in these stressful habitats. They suggest that these species are more adapted to surviving the rigorous environment rather than establishing from seeds. However, in the more favourable high arctic lowlands of the Fosheim Peninsula, germination rates were relatively rapid. The ruderal species sampled had successful germination within the first 10 to 15 days compared to an average of 36 to 48 days for similar species collected in the polar desert of Devon Island (Bliss and Gold, 1999). Braya purpurescens had the fastest germination rates and seedlings of this species were often observed in the field. Eurola (1972) also found that other species of the Bassicaceae family such as Cochlearia offwinallis and many of the Draba spp. were also quick to germinate compared to other species such as Papaver, Minuartia, Pedicularis, Stellaria, Saxifraga and Poa spp. Despite its rapid germination, Braya spp. was not predominant in the extant vegetation community. This may be a result of the high mortality of seedlings. Successful germination does not necessarily imply establishment in these high stress environments. Germination followed by juvenile mortality may result in a depletion of the seed bank, and may explain the under-representation of Braya spp. in the collected seed bank. The other two ruderal species collected {Papaver radicatum and Puccinellia spp.) not only produced large number of viable seeds, but also had a steady rate of germination over 25 days. This may be a good strategy by decreasing their vulnerability to adverse conditions. 120 Following a post germination frost, which may kil l most seedlings, there can still be more seeds available for germination. The success of this strategy may be evident in the extant vegetation where there were higher frequencies of these two species compared to Braya purpurascens. The low germinability of Poa glauca coincides with results from Eurola's (1972) germination experiment. He did however find that seeds of this long-lived species along with those of Poa arctica and Salix spp could persist in the seed bank for many years. Although Poa produced many seeds it may rely mainly on vegetative reproduction because of its slow germination rate. Although little is known on the inhibition of germination in the Arctic (McGraw et al, 1989), it is possible that repeated freezing and thawing of the Poa seeds, as would occur in the field and may have help germination. 4.5.2 Seed bank The germinable seed bank in the Fosheim Peninsula communities was larger than expected for high arctic tundra ecosystems. Seed bank values averaged 1859 seeds/m2 (+ 2797 SD) and 1029 seeds/m2 (+1621 SD) for Big Slide Creeks and Hot Weather Creek, respectively. Other studies for comparable latitude and climatic conditions had seed bank values ranging from 63 seeds/m for an undisturbed site at Alexandra Fjord (Freedman et al. 1982), 270 seeds/m2 for a site at Sverdrup Pass (Levesque and Svoboda, 1995) to 298 seeds/m2 on a glacial foreland at Alexandra Fjord (Jones, 1997). However, the seed bank densities from my study area varied enormously from site to site, and many sites had no germinable seed at all in the soil samples. 32.4% at HWC and 26.7% at BSC had no germinable seeds in the soil samples. On the other hand, six sites had total seed bank densities exceeding 7 000 seeds/m with two soil samples at BSC having more then 15 000 121 seeds/m2. Comparable seed bank densities for other cold climate environments have been found either in highly disturbed habitats such as a fox den at Alexandra Fjord (7 810 2 2 seeds/m , Freedman et al. 1982) or in alpine sites (3 335 to 6 179 seeds/m , Chambers 1993; 8 419 seeds/m2, Archibold, 1984). Alpine environments are also subjected to heavy soil disturbance and churning due to frost or other cyroturbations resulting from the large temperature fluctuations. Seed bank density results from the difference between seed input and seed output. Possible explanations for the higher seed count on the Fosheim Peninsula could include good seed production most years or long dormancy of the seeds (inputs), or low germination or predation of the seed pool (output). As in other studies (Jones, 1997; Levesque and Svoboda, 1995), all seed bank taxa were found in the surrounding extant vegetation, confirming that seed dispersal is from local sources. In the event of global warming, which potentially would improve germination conditions, the vegetative community would most likely resemble the current community composition but with denser cover, unless the climatic changes affect seed dispersal of southern species towards the north. Although the ruderal species represented an important component of the seed bank, they were not the dominant life form group, as postulated from other studies (Johnson, 1975; Freedman et al, 1982; Chamber, 1993, 1995; Levesque and Svoboda, 1995; Jones, 1997). Important contributions to the seed bank came from the late sere species such as Poa spp and Melandrium spp , typically encountered in well older or undisturbed communities. This production of a large number of germinable seeds is more typically associated with the ruderal strategy (Grime 1977) where short life span species manage to persist in the community as a result of frequent germination from the large pool of viable seed in the soil. 122 Even woody species of the later successional group, such as Potentilla spp and Dryas integrifolia, made up 3.8% and 2.4% of all species germinated. The absence of Salix in the seed bank could be explained by the limited longevity of the seeds of this species (Bell and Bliss, 1980). Other studies in the High Arctic reported that Salix as well as Dryas and Cassiope species did not have germinable seeds even from samples collected in lusher semi-desert (Levesque and Svoboda, 1995; Jones 1997), despite their dominating presence in the vegetation. As expected from the late sere species, the number of emerged seedlings of these woody species (Dryas and Potentilla spp.) was low in comparison to other species. Grime (1979) postulated that the strategy of the later successional dominants is to limit their reproductive effort in order to invest resources in surviving as adults and persisting in the community through long life span and slow growth rates. Certainly my results are consistent with this hypothesis, yet the successful germination of some seeds from the long-lived woody shrub and other late sere species coincides with the theory that in these high stressed environments, all species which normally reproduce vegetatively will adapt to produce viable seed crops during climatically favourable years to insure genetic diversity of the population (Svoboda and Henry 1987; Wookey et al, 1995). In these marginal environments, all species, even those categorised as ruderal, have adopted the slow growing, long-lived strategy. Dominant species of the later successional vegetation are often under-represented in the seed bank both in temperate, arctic and alpine regions (Thompson and Grime, 1978; 1979; Freedman et al, 1982; Fox 1983; Diemer and Prack, 1993; Chamber 1993;). Species not present in the seed bank were not only of the later sere communities but also typical of mesic sites (e.g.. Pedicularis spp., Polygonum viviparum). Other have also found that 123 length of seed viability can be affected by soil moisture (Murdock and Ellis, 1994) and by other soil characteristics such as soil composition and chemistry (Levesque and Svoboda, 1995). 4.5.2.1 Change over time and comparison with above ground vegetation Studies in the arctic and alpine areas show that throughout the successional sequence, the seed bank composition tends to become dissimilar from the above ground vegetation (Chambers 1993; Jones, 1997). The main argument suggested for this dissimilarity is the persistence of large number of ruderal species in the seed bank, compared to a marked reduction of these species in the older vegetation communities. M y observations on the Fosheim Peninsula partially support this hypothesis. Apart from the ruderal grasses at HWC, all other ruderal species maintained a large germinable seed population in the soil in both older and undisturbed terrain. However, my results differed from other studies in that dissimilarities with vegetation are also noted in young disturbances due to the important number of late sere species in the soil. This was mainly influenced by the large number of monocot seeds from the later sere group. Figure 4.12 illustrates the seed bank and vegetation patterns over time. As expected, in the vegetation the number of ruderal species decreases over time, while the proportion of late sere species increases. This pattern was not reflected in the seed bank composition. The average seed density in the soil was high 124 Seedbank Vegetation community y i o undist y i o undist Age Categories Age Categories Figure 4.12: Distribution patterns of the two life form strategies (ruderal vs late sere) over time for both the seed bank and the extant vegetation. This is the result of the combined dataset for the two geographic location and the two functional types (monocotolydons and dicotolydons). Data are means (+ SE). regardless of the life form group. Interestingly the proportion of ruderal species is often lower than the late sere species group and particularly in the young and old age categories. There appeared to be an influence of the local vegetation on the seed bank density, where patterns of increased vegetation cover was paralleled by increased seed bank density. In a similar study at Alexandra Fjord, Ellesmere Island, Jones (1997) found a positive relation between seed bank density and above-ground vegetation, concluding that colonisation was largely constrained by spatial distribution of viable seeds in the soil. However, my data rather suggest that there were germination constraints associated with the young terrain for both ruderal and later successional species, and the older terrain for ruderal species, likely due to some particular aspects of the biotic (e.g. competition) but more likely the abiotic (e.g. soil moisture, temperature) environment. The slow germination rate of Poa spp., for example, may explain its absence in the vegetation community of young disturbances where moisture is high only in the early season after snowmelt. In addition, the 125 high saline level of this environment may be acceptable for salt tolerant ruderal species such as Deschampsia brevedifolia and Puccinellia angustata. Further experimental studies would be needed to determine whether competition limited ruderal species in the older communities. 4.5.3 Factors controlling seed bank densities One of the most interesting results from my study was that mechanisms affecting seed bank dynamics appear to be completely independent of terrain age. Rather, large scale landscape features (different geographic locations) or, at the other extreme, micro features (desiccation cracks, small depression) appear to control the seed densities in the soil. Large scale factors such as precipitation, winter snow and wind patterns, elevation, distance from ocean and mountains, could have an affect on global pattern of seed dispersal. The multiple regression identified the different geographic location of the HWC and BSC as the first most important factor influencing seed bank densities. BSC, with a 70 meters difference in elevation, was located further inland, closer to the Sawtooth mountain range while HWC, closer to the coast (see Chapter 2), potentially affected by different elements of large scale seed dispersal factors. Further studies on seed dispersal would be needed to quantify these observations. BSC also had a larger species diversity in the surveyed extant communities compared to HWC (Chapter 3). At the micro-scale level, important factors included density of desiccation cracks, micro-topography, presence of litter and proximity to vegetation, which all influence the capacity to intercept or trap seeds in the soil. Other studies have also found terrain roughness to strongly influence seed bank densities (Chamber, 1993). Soils with a larger particle size (e.g. greater proportion of sand) can normally trap a greater number of seeds than finer soils (Harper, 1977; Chambers et al, 1991). Medium scale factors in my 126 study, such as overall terrain slope, presence of a slide in the terrain or even vegetation density, had little to no effect on the seed bank densities. Much of the seed dispersal is believed to occur throughout the winter over the snow (Grulke and Bliss, 1983). However, Jones (1997) observed that there was no correlation between autumn-winter seed rain and the seed bank, suggesting that although the seed can be dispersed across the landscape in the winter, i f there is no microtopgraphic roughness to trap the seed there wil l be accumulation in the seed bank. The role of the seed bank in succession is comparable to latent energy where stored reserves are available to modify the community in the event of a disturbance, such as ALDS or even global warming. From my results it would appear that following global change, the community would not necessarily change in composition but rather in relative density of the species present. Otherwise, the seed bank will affect current succession only in good seasons when establishment from seed is more frequent and seedling survival is higher. Although these seasons are said to be rare in the High Arctic, they appear to occur periodically in my study area. A critical point, however, may be the life span of seeds in comparison to the length of the periods between favourable years for germination and establishment, when there is only maintenance and/or expansion of individuals in the current community. M y observations confirm the previously presented hypothesis that good years insure genetic diversity (e.g. Wookey et al., 1995), as many of the arctic species do invest resources in seed production, as indicated by both the large seed bank and the germinability of the harvested seed crop. 127 Chapter 5: Conclusions The objective of this study was to examine patterns and mechanisms of succession in a High Arctic ecosystem. Active layer detachment slides grouped in 5 age categories for two valleys of the Fosheim Peninsula were used to build a surrogate revegetation sequence. Scar sections of the slides depleted of its vegetation and soil was used to evaluate primary succession while toes with the accumulated material were used to evaluate secondary succession. In the scar environment, distinct plant communities were described and appeared to be associated with different terrain age and/or environmental factors. The patterns of community changes suggests that primary succession proceeds through four main stages of dominance following a directional pattern: ruderal grasses -> late sere grasses and forbs -> shrubs, forbs and grasses -> shrubs and cushion plants. Succession within the first 50 years appeared to be directional without replacement, similar to Egler's (1954) proposed model of Initial Floristic Composition and the models of low to intermediate environmental resistance as proposed by Svoboda and Henry (1987). However, an additional very old slide (between 50 to 100 years of age) suggested eventual species replacement as many of the species found in the previous stage were excluded while other species such as Cassiope tetragona occurred only in this age category. Scars appear to result in improved environmental conditions in soil moisture and texture compared to the undisturbed communities. In such an environment processes proposed in classical succession models may play a larger roles in defining the community composition. 128 In secondary succession occurring in the toe environments, terrain age on its own explained little of the variation in species composition. Rather, changes in environmental conditions and, in particular, in soil moisture, resulted in a retrogressive succession pattern with a decrease in species density and cover. Only species better adapted to drier soil conditions survived over time. These two different succesional pathways, observed in the scars and toes, suggests that patterns cannot be predicted strictly by large-scale environmental factors such as degree days, temperature or length of the growing season. Local or micro-scale conditions may produce different successional patterns within a common geographic location. Although this is likely the case in other environments it may be more critical in the resource deprived High Arctic, where small amelioration of physical condition may result in a significant expansion of the vegetation. Some insight into the biological processes driving succession were gained by examining the production of viable seeds and the germinable seed banks. This allowed to better understand the invasion and colonisation potential of various species throughout the successional sequence. The changes in dominance of two life history strategies (ruderal vs stress tolerator) vegetation were not matched in the seed dynamics. The large production of viable seeds from the ruderal species in all stages of succession suggest that their reduced number or exclusion in the older extant communities is not explained by the lack of available seeds. The large number of seeds of the late sere grass Poa in the seed bank, yet low in number in vegetation communities of younger terrain, suggest environmental constraints to germination or establishment of these species. Early community composition may thus not 129 be so much affected by seed migration and availability but rather by germination and survival of species with trails adapted for these types of environments. The facilitation processes believed to be more important in the severe environments, and primarily during the early stages of colonisation (Lawrence, 1967; Connell and Slatyer, 1977), involved the amelioration of environmental conditions in the A L D S over time. Natural processes of leaching improved the conditions by reducing salt and clay concentration. Changes in the mid to late successional stages could be explained by differential longevity or growth rates of the late sere shrub species such as Dryas and Salix. Although competition is not believed to be important in the High Arctic, there was probable evidence of this type of inter-species interaction in the more productive communities. In the less productive environments, as in the undisturbed communities, many open spaces remain, suggesting low competitive interactions. However, at that level competition may occur at a micro scale within desiccation cracks or small pockets of increased moisture. Scars of A L D S thus contribute small oases of improved environmental conditions where, even in these extreme environments, it would appear that the pattern of vegetation changes are governed by that same sort of processes described in more temperate environments and classical models. Scars, thus harbour a potential for increased species composition and genetic diversity. However, with the eventual elimination of the traces of the slide in the terrain due to physical erosional processes, it could be postulated that the vegetation community would, in the absence of the scar protection, be subjected to environmental contrainsts and return to a less dense of lush vegetation cover as in the undisturbed control communities. This effect of the environmental condition, superseding 130 biological interaction was observed in the toe environment. Regardless of the initial condition of denser vegetation cover, the succession was characterised by a reduction of vegetation density and diversity. Conditions in the toes characterised environments of increasing severity, such that allogenic processes or externally driven changes in the physical environment superseded the autogenic processes (community driven), as proposed in Matthews, (1992) and Svoboda and Henry's (1987) model of increased severity. Results from this study thus agree with succession model which integrate levels of environmental severity as a factor controlling pattern of succession, yet further investigation of the mechanisms of succession are still needed in these marginal environment. Further experimental study could help to better understand the processes governing community composition and change over time. Sowing of seeds form both ruderal and late sere species as well as transplanting of adult plants in the scar environment could help decipher whether exclusion of these species results from unsuccessful germination and establishment in the early phases, or from the lack of appropriate environmental conditions for their survival. As well, removal of neighbouring plants and transplanting of ruderal species into denser vegetation could shed some light on competitive interactions. Finally, as the results in seed bank density and vegetation composition often differed between the two geographic location of BSC and HWC, increasing the vegetation sampling to include more locations on the Fosheim Peninsula would help to better understand the effect of meso-scale differences such as elevation, climatic condition, etc, and eventually to better characterise patterns of vegetation succession in polar semi-desert habitats. 131 References cited Archibold, W. 1984. A comparison of seed reserves in arctic, subarctic and alpine soils. Canadian Field-Naturalist 98(3): 337-344. Arft, A . M . , Walker, M.D. , Gurevitch, J., Alatalo, J.M., Bret-Harte, M . S.,Dale, M . , Diemer, M . , Gugerli, F., Henry, G. H. R., Jones, M . H . , Hollister, R.D., Jonsdottir, I.S., Laine, K. , Levesque, E., Marion, G.M. , Molau, U . Molgaard, P., Nordenhall, U . , Raszhivin, V . , Robinson, C.H., Starr, G., Stenstrom, A. , Stenstrom, M . , Totland, 0 . , Turner, P. L. , Walker, L. J., Webber, P. J., Welker, J. M . and Wookey, P. A . 1999. Responses of tundra plants to experimental warming: meta-analysis of the International Tundra Experiment. Ecological Monographs 69:491-511. Babb, T. A . and Bliss, L .C. 1974. Effects of the physical disturbance on arctic vegetation in the Queen Elizabeth Islands. Journal of Applied Ecology 11: 549-562. Barrett, P. and Shulten, R. 1975. Disturbance and the successional response of Arctic plants on polar desert habitats. Arctic 28: 70-73. Bell, K . L . 1975. Aspects of seed production and germination in some high arctic plants, in Plant and Surface Responses to Environmental Conditions in the Western High Arctic. In: Bliss, L .C. (ed.) ALUR Program Dept. of Indian and Northern Affairs, pp.62-72. Ottawa. Bell, K . L . and Bliss, L .C. 1980. Plant reproduction in a high arctic environment. Arctic and Alpine Research 12(1): 1-10. Bell, T. 1996. The last glaciation and sea level history of Fosheim Peninsula, Ellesmere Island, Canadian High Arctic. Canadian Journal of Earth Sciences 33: 1075-1086. Bergeron, J.-F. and Svoboda, J. 1989. Plant communities of Sverdrup Pass, Ellesmere Island, N.W.T. Musk-Ox Special Publication 37: 86-92. Billings W.D. and Mooney, H . A . 1968. The ecology of arctic and alpine plants. Biological Review 43: 481-529. Bliss, L .C. 1962. Adaptations of arctic and alpine plants to environmental conditions. Arctic 15:17-144. Bliss, L .C. 1971. Arctic and alpine plant life cycles. Annual Review of Ecology and Systematics 2: 405-438. Bliss, L .C. 1979. Vegetation and revegetation within permafrost terrain. Third International Conference of Permafrost: Proceedings 2: 31-46. 132 Bliss, L .C. , Henry, G.H.R., Svoboda, J. and Bliss, D.I. 1994. Patterns of plant distribution within two polar desert landscapes. Arctic and Alpine Research 26(1): 46-55. Bliss L .C. and Gold, W.G. 1994. The patterning of plant communities and edaphic factors along a high arctic coastline: Implication for succession. Canadian Journal of Botany 72(8): 1095-1107. Bliss, L .C. and Gold, W.G. 1999. Vascular plant reproduction, establishment and growth and the effect of cryptogamic crust within a polar desert ecosystem, Devon Island, N.W.T., Canada. Canadian Journal of Botany 77(5): 623-636. Bliss, L . C , Svoboda, J. and Bliss, D.I. 1984. Polar deserts, their plant cover and plant production in the Canadian High Arctic. Holarctic Ecology 7: 305-324. Bliss, L . C , and Peterson, K . M . 1992. Plant succession, competition, and the physiological constraints of species in the Arctic. In: Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.) Arctic Ecosystem in a Changing Climate, an Ecological Perspective, pp. 111-136 Academic Press, San Diego. Bliss L .C. and Wein, R.W. 1972. Plant community responses to disturbances in the western Canadian arctic. Canadian Journal of Botany 50: 1097-1109. Cargill, S.M. and Chapin III, F.S. 1987 Application of successional theory to tundra restoration: a review 19(4): 366-372. Chambers, J.C. 1993. Seed and vegetation dynamics in an alpine herb field: effects of disturbance type. Canadian Journal of Botany 71:471-485. Chambers, J.C. 1995. Disturbance, life history strategies, and seed fates in alpine herbfield communities. American Journal of Botany 82(3): 421-433. Chambers, J .C , MacMahon, J.A. and Haefher, J.H. 1991. Seed entrapment in alpine ecosystems: effects of soil size and diaspore morphology. Ecology 72(5): 1668-1677. Chapin III, F.S. 1987. Environmental controls over growth of tundra plants. Ecological Bulletins 38:69-76. Chapin III, F.S., Bret-Harte, M.S., Hobbie, S.E. and Zhong, H . 1996. Plant functional types as predictors of transient responses of arctic vegetation to global change. Journal of Vegetation Science 7: 347-358. Chapin III, F.S. and Chapin M.C. 1980. Revegetation of an arctic disturbed site by native tundra species. Journal of Applied Ecology 17: 449-456. 133 Chapin III, F.S. and Shaver G.R. 1985. Individualistic growth response of tundra plant species to environmental manipulation in the field. Ecology 66(2): 564-576 Chapin III, F.S., Walker, L.R., Fastie, C L . and Sharman, L .C. 1994. Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska. Ecological Monographs 64(2):149-175 Clements, F.E. 1916. Plant succession. Carnagie Institute of Washington Publication, No. 242. Connell, J.H. and Slatyer, R.O. 1977. Mechanisms of succession in natural communities and their role in community stability and organization, The American Naturalist, 111: 1119-1144. Conover, W.J. and Iman, R. 1981. Rank transformation as a bridge between parametric and nonparametric statistics. The American Statistician 35(3): 124-129. Denslow, J.S. 1980.Patterns of plant species diversity during succession under different disturbance regimes. Oecologia (Berl.) 46:18-21. Diemer, M . , and Prack, S. 1993. Estimates of alpine seed bank size in two central european and one scadinavian subarctic plant communities, Arctic and Alpine Research, 25(3): 194-200. Drury, W.H. andNisbet, I.C.T. 1973. Succession. Journal of the Arnold Arboretum 54(3): 331-368. Edlund. S.A. Alt, B.T. and Young, K . L . 1989. Interaction of climate, vegetation, and soil hydrology at Hot Weather Creek, Fosheim Peninsula, Ellesmere Island, Northwest Territory. Current Research, Part D, Geological Survey of Canada, Paper 89-1D: 125-133. Edlund, S.A. and Alt, B.T. 1989. Regional congruence of vegetation and summer climate patterns in the Queen Elizabeth Islands, Northwest Territories, Canada. Arctic 42(1): 3-23. Egler, F.E. 1954. Vegetation science concepts I: Initial floristic composition, a factor in old-field vegetation development. Vegetatio 4: 412-417. Elver, R and Ryvarden, L . 1975. Dispersal and primary establishment of vegetation. In: Wielgolaski, F.E. (ed.) Fennoscandian tundra ecosystems Part 1, pp. 81-85. Springer-Verlag, Berlin. Eurola, S. 1972. Germination of seeds collected in Spitsbergen. Ann. Bot. Fennici 9:149-159. 134 Forbes, B.C. 1993. Aspects of natural recovery of soils, hydrology and vegetation at an abandoned High Arctic settlement, Baffin Island. Sixth International Conference on Permafrost: Proceedings 1: 176-181. Forbes, B.C. 1993. Anthropogenic Tundra Disturbance and Patterns of Response in the Eastern Canadian Arctic. Ph.D. thesis, Department of Geography, McGi l l University, Montreal, Quebec. Fox, J.F. 1983. Germinable seed banks of interior Alaskan tundra. Arctic and Alpine Research 15(3): 405-411. Freedman, B., Hi l l , N . , Svoboda, J. and Henry, G.H.R. 1982. Seed banks and seedling occurrence in a high Arctic oasis at Alexandra Fjord, Ellesmere Island, Canada. Canadian Journal of Botany 60:2112-2118. Freedman, B., Svoboda, J. and Henry, G.H.R. 1994. Alexandra Fjord - an ecological oasis in the polar desert. In: Svoboda, J. and Freedman, B . (eds.) Ecology of a Polar Oasis, Alexandra Fjord, Ellesmere Island, Canada, pp. 1-9 Captus University Publications, Toronto. French, H . M . , 1976. The Periglacial Environment. Longmans, London. Gleason, H.A. 1926. The individualistic concept of the plant association, Bulletin of the Torrey Botanical Club 53: 7-26. Grime, LP. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. The American Naturalist 111: 1169-1194. Grime, J.P. 1979. Plant Strategies and Vegetation Processes. Wiley and Sons, New York. Grulke, N.E. and Bliss, L .C. 1983. A note on winter seed rain in the High Arctic. Arctic and Alpine Research 15(2): 261-265. Harper, J.L. 1977. Population biology of plants. Academic press, London. Harris, C. and Lewkowich, A . G . 1993. Permafrost. Proceedings of the sixth international conference of Permafrost, Beijing 1: 232-237. Heikkinen, R.K.I 996 Predicting patterns of vascular plant species richness with composite variables: ameso-scale study in Finnish Lapland Vegetation 126: 151-165. Helm, D.J. and Alle, E.B. 1995. Vegetation chronosequence near exit glacier, Kenai Fjords National park, Alaska, U.S.A. Arctic and Alpine Research 27(3): 246-257. 135 Henry, G.H.G. and Molau, U . 1997. Tundra plants and climate change: the International Tundra Experiment (ITEX). Global Change Biology 3(Suppl. 1): 1-9. Hernandez, H. 1973. Natural plant recolonization of surficial disturbances, Tuktoyaktuk Peninsula Region, Northwest Territories. Canadian Journal of Botany 51:2177-2194. Hi l l , M.O. 1979. TWINSPAN - FORTRAN Program for Arranging Multivariate Data in an Ordered Two-Way Table by Classification of the Individuals and Attributes. Cornell University, Ithaca, N Y . Hodgson, D.A., St-Onge, D.A. and Edlund, S.A. 1991. Surficial materials of Hot Weather Creek basin, Ellesmere Island, Northwest Territories. Current Research, Part E: Geological Survey of Canada, Paper 91-1E: 157-163. Hodgson, D.A. 1977. A preliminary account of surficial materials, geomorphological processes, terrain sensitivity, and Quaternary history of King Christian and Southern Ellef Ringness Islands, District of Franklin. Report of Activities Part A Geological Survey of Canada Paper, 77-1 A : 485-493. Huston, M . and Smith, T. 1987. Plant succession: life history and competition. The American naturalist 130(2): 168-198. Johnson, E.A.I 975. Buried seed populations in the subarctic forest east of Great Slave Lake, Northwest Territories. Canadian Journal of Botany 53: 2933-2941. Johnstone, I.M. 1986. Plant invasion windows: a time-based classification of invasion potential. Biological Review 61: 369-394. Johnstone, J.F. and Henry, G.H.R. 1997. Retrospective analysis of growth and reproduction in Cassiope tetragona and relations to climate in the Canadian High Arctic. Arctic and Alpine Research. 29(4): 459-469. Jones, G. 1997. Primary plant succession on the Twin Glacier foreland, Alexandra Fjord, Ellesmere Island, Canadian High Arctic. MSc thesis, Department of Geography, University of British Columbia, Vancouver, British Columbia. Jones, M . H. , S. E. MacDonald and G. H . R. Henry. 1999. Sex- and habitat-specific responses of a high arctic willow, Salix arctica, to experimental climate change. Oikos 87: 129-138. Jongman, R.H.G., ter Braak, C.J.F. and van Tongeren, O.J.R. 1995. Data Analysis in Community and Landscape Ecology, Pudoc, Wageningen. Khodachek, E.A. 1997. Seed reproduction in arctic environments. Opera Botanica, 0(132): 129-135. 136 Krannitz. 1996. Reproductive ecology of Dryas integrifolia in the high arctic semi-desert. Canadian Journal of Botany 74(9): 1451-1460. Levesque, E. 1996. Minimum area and cover-abundance scales as applied to polar desert vegetation. Arctic and Alpine Research 28(2): 156-162. Levesque, E. and Svoboda, J. 1995. Germinable seed bank from polar desert stands, Central Ellesmere Island, Canada. In: Callaghan, T.V., Molau, U . , Tyson, M.J. , Holten, J.L, Oechel, W.C., Gilmanov, T., Maxwell, B. , Sveinbojornson, B. , (eds.) Global Change and Arctic Terrestrial Ecosystems, Proceedings of papers contributed to the International Conference, 21-26 August 1993, Oppdal, Norway. Ecosystems research report 10, European Commission, pp. 97-107. Levesque, E., Desforges, M . N . , Jone, G. A. , and Henry, G. H . R. 1996. Germinable seed / propagule bank monitoring at ITEX sites. In: ITEX Manual (Second Edition), U . Molau and P. Molgaard (eds) Danish Polar Centre, Copenhagen, pp. 43-45. Lewkowich, A . G . 1990. Morphology, frequency and magnitude of active-layer detachment slides, Fosheim peninsula, Ellesmere Island, N.W.T. Permafrost: Proceedings of the Fifth Canadian Permafrost Conference, Quebec. Collection Nordicana, 54: 111-117. Lewkowich, A . G . 1992. Factors influencing the distribution and initiation of active layer detachment slides of ellesmere Island, Arctic Canada, In: Dixon, J.C. and Abrahams, A.D. (eds.) Periglacial Geomorphology. John Wiley and Sons Ltd. Margalef, R. 1963. On certain unifying principles in ecology. American Naturalist 97: 357-374. Mathewson, C C . and Mayer-Cole, T.A. 1984. Development and runout of a detachment slide, Bracebridge Inlet, Bathurst Island, Northwest Territories, Canada. Bulletin of the Association of Engineering Geologists 21: 407-424. Matthews, J.A. 1992. The Ecology of Recently-Deglaciated Terrain: a Geoecological Approach to Glacier Forelands and Primary Succession. Cambridge University Press, Cambridge, Great Britain. McGraw, J.B. 1993. Ecological genetic variation in seed bank: Differentiation of extant and seed bank derived population of Eriophorum vaginatum. Arctic and Alpine Research 25(1): 45-49. McGraw, J.B. and Vavrek, M.C. 1989. The role of buried viable seeds in Arctic and alpine plant communities. In: Leek, M.A . , Parker, V.T., and Simpson, R.L. (eds.) Ecology of Soil Seed Banks. Academic press, Inc., Toronto. 137 Molau, U . 1993. Relationship between flowering phenology and life history traits in tundra plants. Arctic and Alpine Research 25: 391-402. Muc, M . , Svoboda, J. and Feedman, B. 1994. Aboveground standing crop in plant communities of a polar desert oasis. In: Svoboda, J. and Freedman, B. (eds.) . Ecology of a Polar Oasis, Alexandra Fjord, Ellesmere Island, Canada, pp 65-74. Captus University Publications, Toronto. Muller, C H . 1952. Plant succession in arctic heath and tundra in northern Scandinavia. Bulletin of the Torrey Botanical Club 79:296-309. Noble I.R. and Slatyer, P.O. 1980. The use of vital attributes to predict successional change in plant communities subject to recurrent disturbances. Vegetatio 43: 5-21. Odum, E.P. 1969. The Strategy of Ecosystem Development. Science 164: 262-270. Pickett, S.T.A., Collins, S.L. and Armesto, J.J. 1987. Models, mechanisms and pathways of succession. The Botanical Review 53: 335-371. Pickett, S.T.A., Simpson R.L. and Leek, M.T. 1989. Pattern and process in dynamics of seed bank. In: Leek, M.A. , Parker, V.T. and Simpson R.L. (eds.) Ecology of soil seedbanks, Academic Press, New York. Pielou E . C 1984. The interpretation of ecological data: a primer on classification and ordination. Wiley-Interscience publication, New York. Porsild, A .E . and Cody. W.J. 1980. Vascular plants of continental Northwest Territories, Canada. National Museum of Natural Sciences, Ottawa. Prach, K. 1993. On the rate of succession. Oikos 66(2): 343-346. Robinson, C.H., Wookey, P.A., Lee, J.A. and Callaghan, T.V. 1998. Plant community responses to simulated environmental change at a high arctic polar semi-desert. Ecology 79: 856-866. Sokal R.R. and Rohlf, J.F. 1981. Biometry. Freeman, San Francisco. Savile, D.B.O. 1960. Limitations of the competitive exclusion principle. Science 132: 17-61. Savile, D.B.O. 1972. Arctic Adaptations in Plants, Monograph No 6, Canada Department of Agriculture, Information Canada, Ottawa. Shaver, G.R. and Kummerow, J. 1992. Phenology, resource allocation and growth of arctic vascular plants. In: Chapin III, F.S., Jefferies, R.L. , Reynolds, J.F., Shaver, G.R., and Svoboda, J. (eds.) Arctic Ecosystems in a Changing Climate, an Ecological Perspective, pp.193-211. Academic Press, San Diego, California. 138 Sohlberg E.H. and Bliss L.C. 1984. Microscale pattern of vascular plant distribution in two high arctic plant communities. Canadian Journal of Botany 62: 2033-2042. Stangl, K.O. , Roggensack, W.D. and Hayley, D.W. 1982. Engineering geology of surficial soils, eastern Melville Island. In: French, H . M . (ed.) Proceedings of the Fourth Canadian Permafrost Conference, ppl36-147. National Research Council of Canada, Ottawa. Svoboda, J. and Henry, G.H.R. 1987. Succession in marginal arctic environments. Arctic and Alpine Research 19(4): 373-384. SYSTAT for Windows: Statistics, Version 5. 1992 Edition, SYSTAT, Inc., Evanston, IL. ter Braak, CJ .F . 1987. CANOCO - a FORTRAN Program for Canonical Community Ordination by [Partial][Detrended][Canonical] Correspondence Analysis, Principal Components Analysis and Redundancy Analysis (Version 2.1). TNO Institute of Applied computer Science, Wageningen. Thompson, K . and Grime, J.P. 1978. The occurrence of buried viable seeds in relation to environmental gradients. Journal of Biogeography 5: 425-430. Thompson, K . and Grime, J.P. 1979. Seasonal variation in the seed banks of herbaceous species in ten contrasting habitats. Journal of Ecology 67: 893-921. Tilman, D. 1985. The resource-ratio hypothesis of plant succession. The American Naturalist 125(6): 827-852. Steel, R.G.D, Torrie, J.H and Dickey, D.A. 1997. Principles and procedures of statistics: a biometrical approach. Mc Graw-Hill, 3 r d ed., New York. U B C Department of Soil Science, Soil Science 200 Laboratory Manual, 1994. Urbanska, K . M . and Schultz, M . 1986. Reproduction by seed in alpine plants and revegetation research above timberline. Botanica Helvetica 96(1): 43-60. Walker L.R. and Chapin III, F.S. 1987. Interactions among processes controlling successional change. Oikos 50: 131-155. Walker, L.R., Zasada, J.C. and Chapin III, F.S. 1986. The role of life history processes in primary succession on an Alaskan floodplain. Ecology 67(5): 1243-1253. Walker, D.A. and Walker, M.D. 1991. History and pattern of disturbance in Alaskan arctic terrestrial ecosystems: a hierarchical approach to analysing landscape change. Journal of Applied Ecology 28:244-276. 139 Wein, R.W. and Rencz, A . N . 1976. Plant cover and standing crop sampling procedures for the Canadian High Arctic. Arctic and Alpine Research 8(2): 139-150. Whittaker, R.J. 1991. The vegetation of the Storbreen gletschervorfeld, Jotunheimen, Norway, IV Short-term vegetation changes. Journal of Biogeography 18: 41-52. Whittaker, R.J. 1993. Plant population patterns in a glacier foreland succession: pioneer herbs and later-colonizing shrubs. Ecography 16: 117-136. Wookey, P.A., Parson, A . N . , Welker, J .M., Potter, J.A., Callaghan, T.V., Lee, J.A , and Press, M.C. 1993. Comparative response of phenology and reproductive development to simulated environmental change in sub-arctic and high arctic plants, Oikos 67: 490-502. Wookey, P.A., Robinson, C H . Parsons, A . N . , Welker, J .M. Press, M . C , Callaghan, T.V., Lee, J.A. 1995. Environmental constraints on the growth photosysthesis and reproductive development of Dryas Octopetala at a High Arctic polar semi-desert, Svalbard. Oecologia (Berlin) 102(4): 478-489. Zar, J.H.I974. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, N.J. 140 Appendices Appendix 1: List of vascular plants species found in the study sites of Hot Weather Creek and Big Slide Creek on the Fosheim peninsula, Ellesmere Island, Canada. This list represents only the species sampled in our quadrats, as the complete species list for the Fosheim penisnula is much more extensive (Edlund, 1989). Nomenclature follows Porsild and Cody (1980). Because flowering parts were not available at the beginning of the season when the samplin started, accurate identification of individuals to the species level was sometimes difficult. For this reason, some species had to be combined into genus group (as outlined below), for the analysis. For the ordination, six-letter species codes were used, and are listed below. Most dominant species within a genus group. Species name: Species code Species code for Species code for for combined HWC dataset BS dataset datset Agropyron violaceum spp. violaceum AGR VIO " " Alopercurus alpinus ALO ALP " it Arctagrostis latifolia ARC LAT " Armeria maritima ARE MAR " Arnica alpina ARNALP " Braya humilis BRAY SP Braya purpurascens * BRAY SP " BRA PUR Braya Thorild-Wilffii BRAY SP " Calamagrostis purpurascens CAL PUR Cassiope tetragona CAS TET " Cerastium alpinum CER ALP " " Deschampsia brevifolia DESBRE " Draba alpina DRA ALP " " Draba cinerea DRA CIN " Draba corymbosa DRA COR " " Draba lactea DRA LAC " Draba nivalis DRA NIV Draba oblongata DRA OBL Draba subcapitata DRA SUB " Dryas integrifolia DRY INT " Erigeron compositus var. discoideus ERICOM " Erysimum Pallasii ERY PAL " " Festuca brachyphylla FES BRA " Hierochloe alpina HIE ALP Kobresia myosuroides KOB MYO " " Lesquerella arctica LES ARC " " Luzula nivalis LUZ NIV Melandrium affine MEL SP " MELAFF Melandrium apetalum ssp. Arcticum MEL SP Minuartia rubella MIN RUB " Oxyria digyna OXY DYG " ti Papaver radicatum PAP RAD " Pedicularis arctica PED SP " Pedicularis capitata PED SP " 141 Species name: Species code Species code for Species code for for combined datset HWC dataset BS dataset Pedicularis hirsuta PED SP " Pedicularis lanata PED SP Poa abbreviata POAS " Poa alpigena POA SP POA ALP Poa arctica POA SP POA.SP Poa glauca * POA SP " POA. SP Polygonum viviparum POL VIV " Potent ill a rubricaulis POT SP " POT RUB Potent ilia hyp arctica POT SP " Potentilla Vahliana POT SP POT VAH Puccinellia agrostidea PUCC SP " Puccinellia Andersonii PUCC SP t i Puccinellia angustata PUCC SP " Ranunculus nivalis RAN NIV " Ranunculus pedatifidus var. leiocarpus RAN PED RAN PED Salix arctica SAL ARC " Saxifraga flagellaris ssp. platysepala SAX FLA SAX FLA Saxifraga nivalis SAX NIV SAX NIV Saxifraga oppositifolia SAX OPP Saxifraga tricupidata SAXTRI SAXTRI Stelleria edwardsii STE SP Stelleria laeta STESP " STEL Taraxacum hyparcticum TAR HYP a Trisetum spicatum var. spicatum TRISPI " 142 Appendix 2 The relation between absolute soil moisture (%) and soil moisture index (1-5) is illustrated in the following table and figure. The index was determined in the spring and appears to be more representative of absolute soil moisture of June 22 n d . Later in the season, the range in soil moisture is much smaller, and thus the index differentiation is not as apparent. Overall there is a trend indicating that both measurements increased in the same direction. Yet because there was a large variation, even for moisture samples taken 10cm from each other, the correlation are not very strong. The following table give an approximation of the absolute soil moisture equivalents for the index values: Moisture Index Absolute moisture June 22 Combined dates 1-2 5% 2-10% 2-3 11-20% 10-12% 3-4 20-30% 14-17% 4-5 27-32% 14-18% Appendix Figure 2.1: Relation between absolute moisture (%) and relative moisture index (1-5) for a) the four separate collection dates and b) the average for all dates combined. The trend line is only for illustration purposes and does not represent the result of a regression analysis. e I 1 0 5 E I ° s June 22 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 ° 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 relative moisture (1-5) 143 Appendix 3: Vegetation pattern analyses: Appendix Table 3.1: Summary of one-way analysis of variance to test for a terrain age effect on species richness for the 2 location of a) Hot Weather Creek and b) Big Slide Creek. This test was conducted only for the scar disturbances (primary succession environment). Age group categories: l=Young, 2=Intermediate age, 3=01d, 4=Very old (only for HWC) and 5=Undisturbed terrain. Bold character indicate significance level, a = 0.05. a) Hot Weather Creek location (R2= 0.159) Tukey matrix of pairwise probabilities: Source SS DF MS F-ratio P 1 2 3 4 Age 336.056 4 84.014 4.788 0.001 1 1.000 Error 1772.218 101 17.547 2 0.906 1.000 3 0.001 0.382 1.000 4 0.895 1.000 0.199 1.000 5 0.020 0.985 0.146 0.939 1.000 b) Big Slide Creek location (R - 0.473) Tukey matrix of pairwise probabilities: Source SS DF MS F-ratio P 1 2 3 5 Age Error 730.107 813.226 3 47 243.369 17.303 14.065 0.000 1 2 1.000 0.264 1.000 3 5 0.008 0.000 0.220 0.000 1.000 0.186 1.000 Appendix Table 3.2: Summary of repeated measurements analysis of variance testing for a site and season date effect on species richness. Sites include 5 young scars ( sites: 1, 3, 5, 6) and 1 old scar (site 2). Measurement dates: a) June 22, b) June 30 t h, c) July 15 t h, d) August 08 t h. Between subjects Source SS DF MS F-ratio P slidelD 1395.794 4 318.948 11.296 0.000 Error 1081.150 35 30.890 Within subj ects Source SS DF MS F-ratio P G-G* H-F** season 1148.822 3 482.941 50.844 0.000 0.000 0.000 season* slidelD 1250.950 12 104.246 10.975 0.000 0.000 0.000 Error 997.332 105 9.498 G-G: Greenhouse-Geisser Epsilon H-F: Huynh-Feldt Epsilon 144 Appendix Table 3.3: Summary of one-way analysis of variance to test for a site effect on soil moiture. These measurements were conducted for HWC location only. The repeated measure A N O V A (Appendix Table 3.1) indicated a significant site and season effect on soil moisture. Single factor A N O V A s were thus repeated for the 4 separate dates to further investigate the relation between sites. Measurement dates: a) June 22, b) June 30 t h, c) July 15 t h, d) August 08 t h. Sites include 5 young scars ( sites: 1, 3, 5, 6) and 1 old scar (site 2). a) June 22, 1994 (R2= 0.740) Tukey matrix of pairwise probabilities: Source SS DF MS F-ratio P 1 2 3 5 6 Site 2354.808 4 588.702 24.958 0.000 1 1.000 Error 825.567 35 23.588 2 0.000 1.000 3 0.000 0.911 1.000 5 1.000 0.000 0.000 1.000 6 0.967 0.000 0.000 0.989 0 June 30, 1994 (R2= 0.194) Source SS DF MS F-ratio p Site 142.821 4 35.705 2.110 0.100 Error 592.279 35 16.922 a) July 15, 1994 (R2= 0.243) Tukey matrix of pairwise probabilities: Source SS DF MS F-ratio P 1 2 3 5 6 Age 72.546 4 18.137 2.816 0.040 1 1.000 Error 225.429 35 6.441 2 0.954 1.000 3 0.878 0.999 1.000 5 0.419 0.143 0.099 1.000 6 0.508 0.230 0.171 1.000 1.000 a) June 22, 1994 (R2= 0.150) Source SS DF MS F-ratio P Age 76.568 4 19.142 1.539 0.212 Error 435.207 35 12.434 145 Appendix Table 3.4: Summary of one-way analysis of variance testing for a slope position effect (top, mid or bottom of slope) on soil moisture for 2 different terrain ages: a) Young scars, and b) Old scars). The analysis was repeated for the four measurement dates of the season a) Young Scars: i. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 186.267 1684.433 2 27 93.133 62.386 1.493 0.100 0.243 ii. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 10.552 445.314 2 27 5.276 16.493 0.320 0.023 0.729 iii. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 20.452 171.048 2 27 10.076 6.335 1.0591 0.105 0.222 iv. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 21.943 280.757 2 27 10.971 10.398 1.055 0.072 0.362 b) i. Old Scars: June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 15.517 232.083 2 7 7.758 33.155 0.234 0.053 0.797 ii. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 77.517 140.083 2 7 38.758 20.012 1.937 0.355 0.214 iii. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 25.767 68.333 2 7 12.883 9.762 1.320 0.274 0.326 iv. June 22, 1994 : Source SS DF MS F-ratio R 2 P Age Error 18.583 163.417 2 7 9.292 23.345 0.398 0.102 0.686 146 Appendix Table 3.5: Pearson's correlation values comparing total vegetation cover with absolute soil moisture for the four measurement dates of the season. The correlation was conducted with 38 samples. Soil Moisture Total vegetation Measurement dates (R2) June 22n d 0.624"* June 30th 0.321 July 15th 0.194 August 8 th 0.121 *** Significance at 0.000, according to Bonferroni probability post hoc test. Appendix Table 3.6: Summary of two factor analysis of variance testing for an effect of terrain age (young / old) and disturbance type (scar / toe) on a) soil moisture and b) permafrost depth. a) Soil Moisture: Source Sum of Squares DF Mean Square F-ratio P Age 7.061 1 7.061 0.611 0.444 Disturb. Type 11.107 1 11.107 0.960 0.339 Age*Disturb.Type 91.297 1 91.297 7.894 0.011 Error 219.739 19 11.565 b) Permafrost depth: Source Sum of Squares DF Mean Square F-ratio P Age 81.789 1 81.789 1.001 0.330 Disturb. Type 2.100 1 2.100 0.026 0.874 Age*Disturb.Type 33.401 1 33.401 0.109 0.530 Error 1553.057 19 81.740 147 Appendix Table 3.7: Summary of regression analysis of absolute soil moisture to depth of active layer (m) repeated for the four measurement dates of the season (a) June 22 n d , b) June 30 t h, c) July 15 t h, d ) August 08 t h ) for a total of 98 samples. a) June 22' Source SS DF MS F-ratio P R2 Model Residual 2442.413 8836.067 1 91 2442.413 97.100 25.154 0.000 0.217 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Permafrost 30.233 -0.500 2.891 0.100 0.000 -0.465 1.000 10.457 -2.015 0.000 0.000 b) June 30th: Source SS DF MS F-ratio P R2 Model Residual 419.063 2775.628 1 111 419.063 25.006 16.759 0.000 0.131 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Permafrost 19.537 -0.206 2.165 0.050 0.000 -0.362 1.000 9.022 -4.094 0.000 0.000 a) July 15' Source SS DF MS F-ratio P R2 Model Residual 45.677 1914.665 1 118 45.677 16.226 2.815 0.096 0.023 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Permafrost 12.128 -0.053 1.689 0.032 0.000 -0.153 1.000 7.182 -1.678 0.000 0.096 a) August 08' Source SS DF MS F-ratio P R2 Model Residual 139.269 1741.613 1 96 136.269 18.142 7.511 0.007 0.0.73 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Permafrost 17.770 -0.082 1.923 0.030 0.000 -0.269 1.000 9.243 -2.741 0.000 0.007 148 T3 -4—* J-H O o OO i — i H o < 42) H o n ^ O CM Ch I T ) H « O j H CO H H « ) 3 r H I f ) r H £ £ J r H O r -^ r- if) Ch vo r-m co in r-in u> in in m o *T C T t i-t r n oo cn r-r n r n r o CM m CM o r H H H CM r-H CM ^ H n ,_, O r H Oi CM cn o CO 00 oo r-CO *7 oo cn r~ m r- CM r- o \o \o m in as O CM «J3 r H ^ m r H W cn cn o\ r-CTi \o co oi CO o CO U3 OD m 00 CM i CM cn I rH m on m in I T i i m i rn in I LO ID m CD m m i un m i m •«* i m in i m i T H i n I Lf) " 3 * i m rn I CM CM rn m i ^ ID CM I m CD H i in i n rH i un un i i m ix) i—i i m ID H i i n « in I tr ^ CM i i n i n i—i i i—i m I I i—I LD m ID m in CM i n i—i T i n CD I I I D m cn m m m i co i co M3 i ^ H m ^ un l I m H H CM CM CM I rH rH CM CM T I rn CM m I I I i on . I i—I CM CM CM rH m CM C-J m m i IN H •w m O C O T J • H "~H T J C D C O to T J - H o O T J ' H *a T J C D U Ti T J T J to ,o t0 ""H E q 4 J •u •H in - rH E C O £ 0 T J > , -U C D C O 4 H T J q -u 3 O 'H 0 M U T J T J T J •H •H -u E C D O ••H ( 3 1 C O in O ( T J C O C D C L , O ~" & 4 J a • M U - H ^ T J •-i U J H U a T J O C O 3 ^H C O O Ch •<H *H T J U •H •H C O C D U C D -H -u t0 T J T J a ' H Q, C D C O ' H O Q, T J C / l > T J T J C J Q) -Q ^ T J Q. rrj C L T J to T J a, a T J lH • H s & C O •H frj S q • H C D 3 C J C J T T ~ H •-i - H . co a q -Q to S H O cu cy O ^ b i q 0 'H C O 4 J T J M T J •M U S H T J C O —i T J C O ••H C O T J 3 .q T J 4 J C D 0 ( V H a -H T J in - Q C D T J C \ T J b i T J • U C D E -•H rrj a •u M lH T J -H. U T J T J C J E I H ^ - H E " H to • H ^ E q C D b> C O u q O q T J T J - H C D C O E ~< =1 T3 C O T J U T J o " H C L T J •H "H 0 T J 0 q is >H in •U ( T J T J C J •H U In C D C D q O \ * H ( C J C O in T J 3 p T J C D C D T J > H O T J 4 J > H M -u (tj (D q Tj - H w b . a SH T J q C D <u to T J T J C D C J O b T J 3 X -Q T J C O C D T J T J O C O C D m K •-i •C C O " H _ M ( T J C O - Q X} b •-H -r^  T J cy T J M 3 4 J •H T J • Q T J a - Q C O C D T J ^ C J •u E > , T J 0 * C J C J C O > < N -Q >, C O T J T J q TJ in C O Q, ( D q co -H T J > H o T J ^ 4 J ~ i ^ C O O T J C D T J in » H o •<H ( T J ( T J p I H 0 > H in in in Oj O C ) C D ( T J - U • H C D T J Q in 0 ^ O oj X C D O in "H T J bi ^H U C O K Q 2 Q Q &q C L , Cu to s; ^ C O Q o «c Q KI ^ O Q CQ E H 149 T 3 H OCO I d O lO I H O ^ ro 10 (N O l o o P. CJ to X • H U H ) ^ CJ to 3 O ^ ^ K Q . O Q, . O M q ft) fO U its O ~1 q *H rrj D O ID IQ U) U U t j CO -Q .CJ Ol 'H >, to its its q TJ -H O 1H In in in (D O 5; q q ti] ^ raj rrj u Q) S H V H in +j ^ CO Q QJ Q, <D q if3 ^ "-d &i to 5; ft) ra CO -H CD trj fii to q Q, •H M trj CX, crj *H (TJ CO a) 3 3 3 3 to U -H U rrj rrj to (D -Q -Q ft) rrj fO in O L, (D - i CJ Q U «C M o O CO CD CJ « trj a CD • H M oo a (TJ C/j rrj Q. trj • H ^ 3 C | q 3 - H - H Is 4 J in rrj rrj CD q TJ - H X} co CD US rrj 4 . J r -H » H ^ O QJ Q Cq ^ 5; ft) * H tn rrj to rrj o Q, *n (D a Ef tr, a in O u N co CJ rrj X (D ^ O O Q «C Qi rrj q -u - H D Sj i> rrj *H t; L-; ' rrj 3 3 O 1 • H a. in CD ft) Q) ft) in In ro q rrj Cj CJ U U U (Q tji 3 h «C Qi 150 Appendix Table 3.9: Summary of analysis of variance to test for effect of the terrain age disturbance type on soil conductivity. Tukey's multiple comparisons matrix shows the significance of the pairwise factors. Terrain age categories: l=young, 2=moderage age, 3 old. Disturbance type categories: s= scars, t=toes Source Sum of Squares DF Mean Square F-ratio P Age 485.588 2 242.794 6.083 0.006 Disturb Type 0.252 1 0.252 0.006 0.937 Age*DisturbType 2177.092 2 1058.546 26.520 0.000 Error 1237.370 31 39.915 Tukey HSD Multiple Comparison matrix of pariwise probabilities: Is lt 2s 2t 3s 3t Is 1.000 lt 0.000 1.000 2s 0.195 0.027 1.000 2t 0.991 0.000 0.378 1.000 3s 0.250 0.009 1.000 0.471 1.000 3t 0.777 0.000 0.008 0.339 0.009 1.000 151 Appendix 4: Hourly temperature measurements in the scar of a young and old disturbance. £ £ 5 £ c » c * E u E o o> co 0) co ~ t ~ t < CO < CO a j n i e j a d w a j i & *c> i *z CO ra (0 cu o c (0 CM s-i EE o h U oo B 153 Appendix 5 Havested seed analyses Appendix Table 5.1: Summary of analysis of variance to test for effect of the terrain age and disturbance type on seed viability of the ruderal life history group. This composite group includes the species Braya purpurescens, Papaver radicatum, and Puccinellia spp. Tukey's multiple comparisons matrix shows the significance of the pairwise factors. Terrain age categories: l=young, 2=intermediate age, 3= old, 4=undisturbed terrain. Disturbance type categories: s= scars, t=toes Source Sum of Squares DF Mean Square F-ratio p_ Age 3035.578 3 1011.859 2.191 0.093 DisturbType 3460.963 1 3460.963 7.494 0.007 Age*DisturbType 5626.887 3 1875.629 4.062 0.009 Error 53107.145 115 461.801 Tukey HSD Multiple Comparison matrix of pariwise probabilities: Is lt 2s 2t 3s 3t 4s 4t Is 1.000 lt 1.000 1.000 2s 1.000 1.000 1.000 2t 0.929 0.993 0.967 1.000 3s 0.956 0.940 0.978 0.605 1.000 3t 0.004 0.050 0.025 0.312 0.008 1.000 4s 0.995 0.991 1.000 0.585 0.995 0.000 1.000 4t 0.973 0.963 0.991 0.574 1.000 0.002 0.999 1.000 154 Appendix Table 5.2: Summary of one-way analysis of variance to test for a terrain age effect on seed viability of the ruderal life history group for 2 disturbance types: a) scar disturbances hosting a primary succession and b) toe disturbance hosting a secondary succession. Age group categories: l=Young, 2=Intermediate age 3= Old and 4=Undisturbed terrain. Significance level, a = 0.05. a) Scar disturbances (Primary succession): Source SS DF MS F-ratio P Age 717.154 3 Error 32074.558 79 239.051 406.007 0.589 0.624 b) Toe disturbances (Secondary succession): Tukey matrix of pairwise probabilities: Source SS DF MS F-ratio P 1 2 3 4 Age 8609.575 3 Error 21032.587 36 2869.858 584.239 4.912 0.006 1 2 1.000 0.890 1.000 3 4 0.044 0.780 0.196 0.358 1.000 0.004 1.000 Appendix Table 5.3: Summary of Kruskall-Wallis one-way analysis of variance to test for a terrain age effect on seed viability of Poa glauca, (the late sere life history group) for 2 disturbance type: a) scar disturbances hosting a primary succession and b) toe disturbance hosting a secondary succession. Age group categories: l=Young, 2=Intermediate age 3= Old and 4=Undisturbed terrain. Significance level, a = 0.05. a) Scar disturbances (Primary succession): Age Group n Rank sum 1 16 525.000 2 10 322.500 3 8 374.000 4 43 1781.500 K-W statistic = 4.244; p =0.236; df3 b) Toe disturbance (Secondary succession): Age Group n Rank sum 1 5 57.500 2. 10 283.000 3 10 161.500 4 18 444.000 K-W statistic = 10.396; p=0.015; df3 155 Appendix Table 5.4: Summary of regression analysis of total vegetation cover to germination success (%) of harvested seeds for a) all species combined, b) Poa spp. only and c) ruderal species only. Samples were averaged by site (individual disturbances) and disturbance type (scar, toe and control) for a total of 27 separate cases. Cases where no seeds were found during the harvest were not included in the regression. a) All species combined: Source SS DF MS F-ratio P R2 Model Residual 1993.521 10204.739 1 25 1993.521 408.190 4.884 0.036 0.404 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 45.321 -0.437 7.115 0.198 0.000 -0.101 1.000 6.369 -2.210 0.000 0.036 b) Late sere species (Poa spp.): Source SS DF MS F-ratio P R 2 Model Residual 22.423 1096.122 1 22 22.423 49.824 0.450 0.509 0.142 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 4.576 0.052 2.640 0.077 0.000 0.142 1.000 1.733 0.671 0.097 0.509 c) Ruderal species group: Source SS DF MS F-ratio P R2 Model Residual 96.872 6329.966 1 20 96.872 316.498 0.306 0.586 0.123 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 48.906 0.133 7.440 0.240 0.000 0.123 1.000 6.574 0.553 0.000 0.586 156 Appendix 6: Seed Bank Analysis Appendix Table 6.1: Summary of Kruskall-Wallis one-way analysis of variance to test for the effect of vegetation proximity, on the germinable seed bank. Samples found within 1 meter of vegetation was label as follows: se = scar edge for distances < l m from vegetation, s= scar for distances > lm. Analysis was repeated for a) Big Slide Creek with 3 age categories (young, intermediate and old and for b) Hot Weather Creek for the young scars only. At HWC, difference between the 2 types of control (cs = control of the scar and cx = controls in more exposed environments) was also tested (b-ii) a) Big Slide Creek (BSC): i) Young scars: Vegetation proximity N Rank sum s 40 1085.00 se 16 511.00 Mann-Whitney U test statistic = 265.000; p=0.286; chi-square approx. 1.137; df 1 ii) Intermediate aged scars: Vegetation proximity N Rank sum s 68 2910.000 se 16 660.000 Mann-Whitney U test statistic = 564.000; p=0.781; chi-square approx. 0.077; df 1 iii) Old scars: Vegetation proximity n Rank sum s 44 1344.000 se 16 468.000 Mann-Whitney U test statistic = 354.000; p=0.972; chi-square approx. 0.001; df 1 b) Hot Weather Creek (HWC) i) Young scars only ii) Controls Vegetation N Rank sum Vegetation N Rank sum proximity proximity s 84 6240.000 cs 25 653.500 se 76 6640.000 cx 20 381.500 Mann-Whitney U statistic = 3714.000; p=0.018; Mann-Whitney U statistic = 328.500; p=0.072; chi-square approx. 5.606; df 1 chi-square approx. 3.298; df 1 157 Appendix Table 6.2: Summary of Kruskall-Wallis one-way analysis of variance to test for the effect of terrain age, on the germinable seed bank. Analysis was repeated for the 2 geographic location: a) H W C and b) B S C and for the 4 functional types: i) Ruderal grarninoids, ii) ruderal forbs, iii) late sere grarninoids, iv) late sere forbs. Age group categories: l=Young, 2=Intermediate age, 3=01d, 4=Very Old and 5=Undisturbed terrain. B S C had no Very Old terrain age and thus category 4 corresponds to undisturbed terrain. Significance level, a = 0.05. a) Hot Weather Creek (HWC): b) Age Group N Rank sum Ruderal Grasses Ruderal Forbs Late Sere Grasses Late Sere Forbs 1 40 273.500 2008.500 2319.000 1884.000 2 10 803.000 562.00 610.000 629.500 3 11 556.500 713.500 716.500 703.500 4 5 227.500 350.000 152.500 190.000 5 (undist.) 45 2355.500 2582.00 2418.000 2809.000 K-W Statistic 15.312 6.514 5.511 10.922 P 0.004 0.164 0.239 0.027 Df 4 4 4 4 Big Slide Creek (BSC): Age Group N Rank sum Ruderal Grasses Ruderal Forbs Late Sere Grasses Late Sere Forbs 1 14 672.500 532.000 679.000 634.000 2 21 886.000 841.000 791.000 672.500 3 15 683.000 715.000 799.500 613.500 4 (undist.) 35 1413.500 1567.000 1386.500 1735.000 K-W Statistic 1.358 2.879 5.411 8.370 P 0.716 0.411 0.144 0.039 Df 3 3 3 3 158 Appendix Table 6.3: Summary of Kruskal-Wallis one-way analysis of variance to test for a terrain age effect on germinable seed bank density, for 2 geographic location: a) H W C and b) B S C . Age group categories: l=Young, 2=Intermediate age 3=01d, 4=Very Old and 5=Undisturbed terrain. B S C had no Very Old terrain age and thus category 4 corresponds to undisturbed terrain. Significance level, a = 0.05. a) Hot Weather Creek: Age Group N Rank sum 1 2 3 4 1 40 1970.500 1 1.000 2 10 709.000 2 0.035 1.000 3 10 695.500 3 0.115 0.664 1.000 4 5 219.500 4 0.150 0.333 0.333 1.000 5 44 2400.500 5 0.423 0.157 0.390 0.140 1.000 K-W statistic = 6.800; p=0.147; df4 b) Big Slide Creek Age Group N Rank sum 1 14 634.500 2 21 755.000 3 15 707.000 4 35 1558.500 K-W statistic = 2.423; p=0.489; df3 Appendix Table 6.4: Summary of Kruskall-Wallis one-way analysis of variance to test for a terrain age effect on germinable seed bank diversity (number of species) for two geographic location a) H W C and b) B S C . Age group categories: l=Young, 2=Intermediate age 3=01d, 4=Very Old and 5=Undisturbed terrain. B S C had no Very Old terrain age and thus category 4 corresponds to undisturbed terrain. Significance level, a = 0.05. a) HWC: Age Group N Rank sum 1 40 2087.000 2 10 659.500 3 11 395.000 4 5 162.500 5 45 2612.000 K-W statistic = 5.240; p=0.264; df4 b) BSC: Age Group n Rank sum 1 14 628.500 2. 21 769.500 3 15 771.000 4 35 1486.000 K-W statistic = 5.240; p=0.264; df4 159 Appendix Table 6.5: Summary of Kruskall-Wallis one-way analysis of variance to test significant differences between individual slides for both a) HWC and b) BSC locations. a) HWC: Slide ID N Rank sum 1 5 109.000 2 7 248.500 3 6 177.000 4 5 128.000 5 5 99.500 6 5 163.000 7 3 129.000 8 4 193.000 9 3 177.500 10 3 142.000 11 8 250.500 12 6 161.500 16 5 226.500 K-W statistic = 14.009; p=0.300; dfl2 b) BSC: Slide ID n Rank sum 1 10 320.500 2 3 31.000 44 3 144.000 49 3 117.000 50 3 98.500 51 4 73.500 52 6 139.000 64 3 78.000 65 4 63.000 66 3 66.500 69 4 104.500 70 4 39.500 K-W statistic = 23.741; p=0.014; d f l l Appendix Table 6.6: Summary of Kruskall-Wallis one-way analysis of variance to test for the effect of a disturbance in the landscape, independent of age, on the germinable seed bank for both geographic location: a) HWC and b) BSC. a) HWC: b) BSC: Disturbance Group N Rank sum Age Group n Rank sum Scar 65 3594.500 scar 50 96.500 Control 44 2400.500 control 35 558.500 Mann Whitney U Statistic = 1449.5; p=0.903; Chi-square approx. 0.015; df 1 Mann Whitney U Statistic = 928.500; p=0.631; Chi-square approx. 0.231; df 1 160 Appendix Table 6.7: Summary of regression analysis of total vegetation cover to germinable seed bank (# / m2) for a) the combined data sets of HWC and BSC and for the two geographic location seperately (b) HWC and (c) BSC. Samples were averaged by site (individual disturbances) and disturbance type (scar or control). The first regression identified 6 outliers from the residuals, which were removed (Seed bank values: 2251, 3397, 5915, 6090, 6712, 8085). The following regession is for seed bank values < 2000 seeds/m2. a) Combined dataset of the two geographic location: Source SS DF MS F-ratio P R2 Model Residual 202018.868 0.101812E+8 1 41 202018.868 248321.264 0.814 0.372 0.019 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 660.834 3.205 144.290 3.554 0.000 0.139 1.000 4.580 0.902 0.000 0.372 a) Hot Weather Creek dataset only: Source SS DF MS F-ratio P R2 Model Residual 63721.785 672131.975 1 23 63721.785 292231.390 0.218 0.645 0.009 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 810.997 -2.652 187.192 5.680 0.000 -0.097 1.000 4.332 -0.467 0.000 0.645 a) Big Slide Creek dataset only: Source SS DF MS F-ratio P R2 Model Residual 737281.592 2800376.408 1 16 737281.592 175023.526 4.212 0.057 0.208 Variable Coefficient Std error Std coef Tolerance T P(2 tail) Intercept Total veg 357.622 10.158 243.986 4.949 0.000 0.457 1.000 1.466 2.052 0.162 0.057 161 Appendix 7: Development of a regression model to predict germinable seed bank from a number of environmental characteristics. Parameters eligible for input in the model were: Geographic location (LOC = HWC or BSC), Terrain age (AGE = young, intermediate, old, very old, undisturbed), Presence or absence of surface sand (SAND), Soil compaction (HARD = relative scale of 1-4), Micro-topography (MICROTOP = bump, depression, flat), Density of desiccation cracks, Slope angle, Topographic depression index, Total vegetation cover (TOTVEG), Litter cover (LITTER). Bold character indicate significance level, a = 0.05. Appendix table 7.1: Summary of a linear regression analysis of square root transformed values of seed bank densities using all environmental variables: Dependent variable: Seed bank n: 188 R2: 0.189 Source SS DF MS F-ratio P LOC 1627.297 1 1627.297 3.914 0.049 DESS 2290.184 1 2290.184 5.509 0.020 SAND012 970.474 1 970.474 2.334 0.128 SLOPE 264.643 1 264.643 0.637 0.426 HARD 2208.805 4 552.201 1.328 0.261 DEPINDX 102.301 1 102.301 0.246 0.620 TOTVEG 171.220 1 171.220 0.412 0.522 LITTER 2293.517 1 2293.517 5.517 0.020 MICROTOPS 3360.157 2 1680.078 4.041 0.019 AGE 403.402 4 100.851 0.243 0.914 ERROR 70676.228 170 415.743 Appendix Table 7.2: Summary of a forward stepwise regression analysis of square root transformed values of seed bank densities. Minimum tolerance for entry into the model was 0.01, with an alpha to enter and remove or 0.150. The variables in the model are presented in the order they were retained during the stepwise regression. Dependent variable: Seed bank n: 188 R2: 0.147 Source SS DF MS F-ratio P LITTER 4184.923 1 4184.923 10.242 0.002 DESS 3665.138 1 3665.138 8.970 0.003 LOC 3752.177 1 3752.177 9.183 0.003 MICROTOP 3098.780 2 1549.390 3.792 0.024 ERROR 74366.952 182 408.610 162 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089829/manifest

Comment

Related Items