Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Responses of Cassiope tetragona, a high Arctic evergreen dwarf shrub, to variations in growing season… Johnstone, Jill F. 1995

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

Item Metadata

Download

Media
831-ubc_1995-0364.pdf [ 9.08MB ]
Metadata
JSON: 831-1.0099012.json
JSON-LD: 831-1.0099012-ld.json
RDF/XML (Pretty): 831-1.0099012-rdf.xml
RDF/JSON: 831-1.0099012-rdf.json
Turtle: 831-1.0099012-turtle.txt
N-Triples: 831-1.0099012-rdf-ntriples.txt
Original Record: 831-1.0099012-source.json
Full Text
831-1.0099012-fulltext.txt
Citation
831-1.0099012.ris

Full Text

RESPONSES OF CASSIOPE TETRAGONA, A HIGH ARCTIC EVERGREEN DWARF SHRUB, TO VARIATIONS IN GROWING SEASON TEMPERATURE AND GROWING SEASON LENGTH AT ALEXANDRA FIORD, ELLESMERE ISLAND. by JILL F. JOHNSTONE B.A., Middlebury College, 1992 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Department of Geography We accept this thesis as conforming to th/required standard THE UNIVERSITY OF BRITISH COLUMBIA May, 1995 © Jill F. Johnstone, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make ft freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of &COG-Rrt-pH V* The University of British Columbia Vancouver, Canada Date 23 tn DE-6 (2/88) Abstract The short-term responses of Cassiope tetragona, a high arctic evergreen shrub, to variations in growing season climate were examined using experimental manipulations of temperature and growing season length at Alexandra Fiord, Ellesmere Island. Surface temperatures in the field were increased an average of 1-2 °C in two communities using open-top greenhouses. Growing season length was altered in a snowbed community by using manual snow manipulations to change the date of snowmelt. Growth and reproductive responses of Cassiope tetragona to these manipulations were observed over two field seasons following treatment establishment. Natural variations in vegetative and reproductive characteristics of Cassiope tetragona were also monitored in unmanipulated communities selected to represent a range of environmental conditions at the study site. Retrospective analysis of past Cassiope growth and reproduction was used to provide a record of variations in productivity spanning 25-35 years which could be related to climate records from Ellesmere Island. For the retrospective analysis, patterns of internode lengths were used to delimit sections of annual growth and chronologies of annual stem elongation, leaf number and flower number were then analyzed using methods similar to those applied to tree-ring studies. In general, the reproductive parameters of Cassiope tetragona were observed to be highly responsive to short-term variations in growing season climate, while vegetative production exhibited a much more conservative response. Flower production and rates of reproductive development were significantly stimulated by experimental warming. Retrospective analysis of flower production support field observations indicating that flower production is highly sensitive to annual variations in growing season temperatures. In contrast, shoot growth showed moderate responses to experimental warming. Records of past growth indicate that although vegetative production appears to be sensitive to annual variations in summer temperatures, the degree of responsiveness is much lower than for ii reproductive parameters. Net growth and reproduction were not stronly affected by natural or experimental variations in snowmelt timing, although phenology timing was significantly altered. The conservative growth response of Cassiope tetragona to short-term variations in climate is suggested to be related to constraints on plant phenology which may restrict flexibility in the period utilized by plants for aboveground growth. Preferential allocation of within-plant resources to reproductive structures during periods of ameliorated growing season climate may account for the observed strong reproductive responses to climate variations. Trade-offs betweeen growth and reproduction have important implications for predicting the long-term response of Cassiope tetragona to climate change. An understanding of within-plant allocation strategies is also important to the interpretation of past variations in growth and reproduction. Retrospective analysis of past Cassiope production is likely to be a very useful tool for investigating ecological relationships and past climate change. iii Table of Contents Abstract ii Table of Contents iv List of Tables vii List of Figures ix Acknowledgement xiii 1. Introduction 1 2. Response of Cassiope tetragona to experimental manipulations of surface temperature and snowmelt timing 4 2.1. Introduction 4 2.2. Methods 7 2.2.1. Study area 7 2.2.2. Surface warming experiment 9 2.2.2.1. Study sites 9 2.2.2.2. Experimental manipulations 9 2.2.3. Snowmelt experiment 10 2.2.3.1. Study site 10 2.2.3.2. Experimental manipulations 11 2.2.4. Environmental measurements 11 2.2.4.1. Snowmelt progression 11 2.2.4.2. Active layer thaw 12 2.2.4.3. Soil Moisture 12 2.2.5. Measurement of growth and reproduction 13 2.2.5.1. Sampling scheme 13 2.2.5.2. Vegetative phenology 14 2.2.5.3. Shoot elongation 14 iv 2.2.5.4. Reproductive phenology 14 2.2.5.5. Flower production 16 2.2.5.6. Seed germination 16 2.2.6. Data Analysis 17 2.2.6.1. Surface warming experiments 17 2.2.6.2. Snowmelt experiment 18 2.3. Results 20 2.3.1. Surface temperature manipulations 20 2.3.1.1. Environmental factors 20 2.3.1.2. Response of Cassiope tetragona: vegetative phenology and growth 23 2.3.1.3. Response of Cassiope tetragona: reproduction 34 2.3.2. Beach ridge snowbed manipulations 37 2.3.2.1. Environmental factors 37 2.3.2.2. Response of Cassiope tetragona: vegetative phenology and growth 42 2.3.2.3. Response of Cassiope tetragona: reproduction 44 2.4. Discussion 48 2.4.1. Experimental manipulations 48 2.4.2. Response of Cassiope tetragona 50 3. Natural patterns of variation in growth and reproduction of Cassiope tetragona observed at Alexandra Fiord 56 3.1. Introduction 56 3.2. Methods 57 3.2.1. Study sites 57 3.2.2. Measurement of environmental variables 59 3.2.3. Measurement of growth and reproductive output 62 v 3.2.4. Data analysis 64 3.3. Results 65 3.3.1. Environmental measurements 65 3.3.2. Cassiope growth and reproduction 69 3.4. Discussion 80 4. Retrospective analysis of growth and reproduction of Cassiope tetragona 83 4.1. Introduction 83 4.2. Methods 86 4.2.1. Species characteristics 86 4.2.2. Study site 86 4.2.3. Climate data 88 4.2.4. Data sampling and measurement 90 4.2.5. Retrospective growth analysis 91 4.3. Results 96 4.3.1. Chronology characteristics 96 4.3.2. Modeling the response of Cassiope tetragona to variations in climate 100 4.4. Discussion 107 4.4.1. Critique of Methods 107 4.4.2. Chronology characteristics 109 4.4.3. Statistical relations between climate and production of Cassiope tetragona 112 5. Conclusions 118 6. References cited 121 7. Appendices 128 vi List of Tables Table 2.1: Definitions of the phenological stages recorded for Cassiope tetragona 15 Table 2.2: Effects of site and treatment on environmental characteristics measured at the Cassiope and Dryas ITEX sites in 1993. Effects were analyzed using a repeated measures analysis of variance 21 Table 2.3: Effects of year, site and treatment on the timing (day number in calendar year) of vegetative bud break and initiation of active growth for Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1992 and 1993. Values were rank-transformed and analyzed using a three-way ANOVA 21 Table 2.4: Effects of year and treatment on the timing (day number in calendar year) of growth cessation, the number of days from bud break to growth cessation and from growth initiation to growth cessation, and total shoot elongation of Cassiope tetragona observed at the Cassiope ITEX site in 1992 and 1993 28 Table 2.5: Treatment effects (1993 only) on the timing (day number in calendar year) of growth cessation and the number of days from bud break to growth cessation and from growth initiation to growth cessation, and year and treatment effects (1992 and 1993) on total shoot elongation of Cassiope tetragona observed at the Dryas ITEX site 28 Table 2.6: Effects of year, site and treatment on shoot elongation of Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1992 and 1993 30 Table 2.7: Timing of reproductive phenophases in warming and control plots at the Cassiope and Dryas ITEX sites in 1992 and 1993. Mean day number, standard deviation (s.d.) and sample sizes (n=number of plots) are given 35 Table 2.8: Effects of year, site and treatment on the timing of reproductive phenophases of Cassiope tetragona and percent of tagged shoots producing mature flowers at the Cassiope and Dryas ITEX sites in 1992 and 1993. Data were rank-transformed prior to analysis 36 Table 2.9: Snowfree dates (day number in calendar year) of the Beach Ridge snow manipulation plots in 1992 and 1993 39 Table 2.10: Relations between the timing of vegetative phenology and snow release for Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed 39 Table 2.11: Timing of reproductive phenophases of Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed in 1992. Mean day number, standard deviation (s.d.) and sample size (n=number of wi thin-plot samples) are given for each plot 45 Table 2.12: Timing of reproductive phenophases of Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed in 1993. Mean day number, standard deviation (s.d.) and sample size (n=number of within-plot samples) are given for each plot 46 vii Table 2.13: Relations between date of snow release and timing of reproductive phenology of Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed 47 Table 3.1: Sample size and sampling design for environmental, growth and reproductive variables measured at each site in 1993 63 Table 3.2: Timing of snowmelt (day number in calendar year) in nine Cassiope-dominated communities at Alexandra Fiord in 1992 and 1993 66 Table 3.3: Effects of site and year on vegetative phenology and growth of Cassiope tetragona in five communities at the River Slope site, Alexandra Fiord, in 1992 and 1993 66 Table 4.1: Descriptive statistics of three growth and reproductive chronologies of Cassiope tetragona (arctic white heather) at Alexandra Fiord, Ellesmere Island, N.W.T., Canada (78o 53'N, 75o 55'W) 97 Table 4.2: Partial correlation coefficients showing relations between current (year C) and previous year's (year C-l) values of three Cassiope chronologies 99 Table 4.3: Analysis of variance of three Cassiope tetragona chronologies from Alexandra Fiord, Ellesmere Island, N.W.T., Canada 101 Table 4.4: Partial correlation coefficients showing relations between average July temperatures and July melting degree days at Alexandra Fiord and annual growth and reproductive indices of Cassiope tetragona (n=9)..: 102 Table 4.5: Partial correlation coefficients showing relations between climate records from the Eureka High Arctic Weather Station and annual growth and reproductive indices of Cassiope tetragona at Alexandra Fiord 102 Table 4.6: Growth and reproduction response functions developed for Cassiope tetragona using climate data from Eureka, Ellesmere Island. Models were restricted to include only significantly correlated climate variables 104 Table 4.7: Growth and reproduction response functions developed for Cassiope tetragona using climate data from Eureka, Ellesmere Island, with all climate variables and previous growth eligible for inclusion in the model 104 Table 4.8: Two climate transfer functions for July melting degree days and July average temperature at Alexandra Fiord, Ellesmere Island based on chronologies of growth and reproduction of Cassiope tetragona 105 Table 4.9: Analysis of variance between observed and predicted average July melting degree days at Alexandra Fiord from 1980-1988. Predicted values are based on the transfer function, JDD = 26.397 + (11.773*LN) + (10.061*AF) 105 viii List of Figures Figure 2.1: Map of the circumpolar North showing locations of current ITEX sites. Point number five gives the location of Alexandra Fiord. Map prepared by G. M. Marion, U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire 8 Figure 2.2: Progression of snowpack and active layer thaw in treatment and control plots at the Cassiope ITEX site in 1993 (n=14). Error bars represent 1 standard error 22 Figure 2.3: Progression of snowpack and active layer thaw in treatment and control plots at the Dryas ITEX site in 1993 (n= 14). Error bars represent 1 standard error.; 22 Figure 2.4: Mean soil moisture levels in treatment and control plots at the Cassiope (=CI) and Dryas (=DI) ITEX sites, measured five times over the course of the summer in 1993 (n=8). Error bars represent 1 standard error. Asterix denotes a significant treatment effect on soil moisture levels at the Dryas ITEX site on day 188 24 Figure 2.5: Effective growing season length for Cassiope tetragona measured as the time from mean date of vegetative bud break to mean date of growth cessation in treatment (T) and control (C) plots at the Cassiope (=CI) and Dryas (=DI) ITEX sites in 1992 and 1993 (n=14 except in 1992 Dryas ITEX treatment means, n=13). Error bars are -1 and +1 standard error for mean date of bud break and growth cessation, respectively. Dashed lines at the Dryas ITEX site in 1992 reflect uncertainty in the mean date of growth cessation in that year 26 Figure 2.6: The period of active growth of Cassiope tetragona measured as the time from mean date of growth initiation to mean date of growth cessation in treatment (T) and control (C) plots at the Cassiope (=CI) and Dryas (=DI) ITEX sites in 1992 and 1993 (n=14 except in 1992 Dryas ITEX treatment means, n=13). Error bars are -1 and +1 standard error for mean date of growth initiation and growth cessation, respectively. Dashed lines at the Dryas ITEX site in 1992 reflect uncertainty in the mean date of growth cessation in that year 27 Figure 2.7: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Dryas ITEX site in 1992 (n=13 for OTCs, n=14 for controls). Error bars indicate 1 standard error. The arrow points to a period of cool overcast weather with an overnight snowstorm on day 197 31 Figure 2.8: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Cassiope ITEX site in 1992 (n=14). Error bars indicate 1 standard error. The arrow points to a period of cool overcast weather with an overnight snowstorm on day 197 31 Figure 2.9: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Dryas ITEX site in 1993 (n=14). Error bars indicate 1 standard error 32 ix Figure 2.10: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Cassiope ITEX site in 1993 (n=14). Error bars indicate 1 standard error 32 Figure 2.11: Maximum shoot elongation measured in treatment and control plots at the Cassiope (=CI) and Dryas ITEX (=DI) sites (n=14, except for 1992 Dryas ITEX treatment means, n=13). Error bars represent 1 standard error 33 Figure 2.12: Mean flower production per shoot of Cassiope tetragona in treatment and control plots at the Cassiope and Dryas ITEX sites in 1993 (n=12). Error bars represent 1 standard error 33 Figure 2.13: Germination of Cassiope tetragona seeds over a 34-day laboratory germination trial (n=5). Seeds were sampled from treatment and control plots at the Cassiope ITEX site in 1993. Error bars represent 1 standard error 38 Figure 2.14: Progression of snowmelt in snow manipulation plots at the Beach Ridge snowbed in 1992 (C = control, R = removal, A = addition). Snow manipulations were initiated on day 171, 1992 40 Figure 2.15: Progression of snowmelt in snow manipulation plots at the Beach Ridge snowbed in 1993 (C = control, R = removal, A = addition). Snow manipulations were initiated on day 157, 1993 41 Figure 2.16: Effective growing season length for Cassiope tetragona measured as the time from mean date of vegetative bud break to mean date of growth cessation in Beach Ridge snow manipulation plots in 1993. Error bars are -1 and +1 standard error for mean date of bud break and growth cessation, respectively 43 Figure 2.17: The period of active growth of Cassiope tetragona measured as the time from mean date of growth initiation to mean date of growth cessation in Beach Ridge snow manipulation plots in 1993. Error bars are -1 and +1 standard error for mean date of growth initiation and growth cessation, respectively 43 Figure 3.1: Map of the Alexandra Fiord lowland, showing locations of study communities at the River Slope (RS), Erratic (ER), Cassiope ITEX site (CI), Dryas ITEX site (Dl) and the Beach Ridge snowbed (BR). The map is based on Nams, 1982 and revised using 1993 air photos 58 Figure 3.2: Diagram of community zonation and study plots at the River Slope site. Revised from Nams, 1982 60 Figure 3.3: Active layer depths observed at the end of July, 1993 in eight communities at Alexandra Fiord. See text for key to site abbreviations 67 Figure 3.4: Soil moisture levels in nine Cassiope -dominated communities at Alexandra Fiord, measured four times from late July to early August in 1993. See text for key to site abbreviations 68 Figure 3.5: Near-surface air temperatures (+5 cm) measured over the course of the 1993 growing season in four communities at Alexandra Fiord. See text for key to site abbreviations 70 x Figure 3.6: Shallow soil temperatures (-3cm) measured over the course of the 1993 growing season in four communities at Alexandra Fiord. See text for key to site abbreviations 70 Figure 3.7: Effective growing season length for Cassiope tetragona, measured in nine communities at Alexandra Fiord in 1993. Points on the left represent the average date of vegetative bud break and points on the right represent the average date of growth cessation. Error bars are negative and positive standard error for bud break and growth cessation, respectively 71 Figure 3.8: The period of active growth utilized by Cassiope tetragona, measured in nine communities at Alexandra Fiord in 1993. Points on the left represent the average date of growth initiation and points on the right represent the average date of growth cessation. Error bars are negative and positive standard error for growth initiation and growth cessation, respectively 72 Figure 3.9: Total shoot elongation of Cassiope tetragona observed at the end of the 1993 growing season in nine communities at Alexandra Fiord. See text for key to site abbreviations 75 Figure 3.10: Logio values of total shoot elongation plotted against the number of days utilized for active growth by Cassiope tetragona in nine communities at Alexandra Fiord in 1993 76 Figure 3.11: The period of active growth utilized by Cassiope tetragona, measured in five community zones of the River Slope site (RS1-RS5) in 1992 and 1993. Points on the left represent the average date of growth initiation and points on the right represent the average date of growth cessation. Error bars are negative and positive standard error for growth initiation and growth cessation, respectively 77 Figure 3.12: Numbers of flowers produced per mature shoot of Cassiope tetragona, measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations 78 Figure 3.13: Numbers of mature, live shoots of Cassiope tetragona per 10 cm2 area, measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations 79 Figure 3.14: Dead:live shoot ratios of Cassiope tetragona measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations 79 Figure 4.1: A schematic diagram of leaf morphology and nodal growth patterns of a Cassiope tetragona shoot from Alexandra Fiord with a) leaves intact and b) leaves removed 87 Figure 4.2: Average July temperatures recorded at Eureka and Alexandra Fiord from 1980-1988. The two data sets have a correlation coefficient of R=0.35. Data for Eureka is from Atmosperic Environment Service (1980-1992) and from Labine (1994) for Alexandra Fiord 89 xi Figure 4.3: Internode lengths plotted against position along the stem of a sampled Cassiope shoot. Note the wave-like oscillations in internode lengths, with minimum lengths indicating the termination of each year's growth 92 Figure 4.4: Standardized leaf number and stem elongation indexes plotted with flowering index values of the previous year. 98 Figure 4.5: Observed July melting degree day values from Alexandra Fiord (Labine, 1994) plotted with predicted values calculated from the transfer function developed in Havstrom et al. (1995). Correlation between the observed and predicted values is 0.87 106 xii Acknowledgement This thesis is dedicated to William G. Howland, a teacher and friend who has had a profoud impact on my interest and understanding of northern ecosystems. To be a true teacher is a great gift -1 owe him many thanks. I would also like to several people who aided in the production of this thesis. Much appreciation goes to Greg Henry, who provided inspiration, guidance and assistance in all stages of this project. Several field assistants helped in gathering field data at Alexandra Fiord: my thanks to all of them for their hard work and enthusiasm. Special thanks also to Michael Jones, Adrian DeBruyn and Esther Livdsque for advice and support both in the field and on the internet. I am grateful to Brian Klinkenberg, Greg Henry and Val Lemay for their helpful comments on the first draft of this manuscript. I am also grateful to Jonathan Henkelman for his assistance with computer programming and the creation of some diagrams for this thesis. Thank you to Giles Marion for permission to use the ITEX map and to Barry Maxwell for providing a disk with climate data from the Atmospheric Environment Service. Additional thanks also go to the biogeography group at U.B.C., for support and good times along the way. Funding for my term of study at U.B.C. was provided by a graduate fellowship from the National Science Foundation, which I very gratefully acknowledge. Field work at Alexandra Fiord was supported through National Science and Engineering Research Coucil grants to G. Henry, with logistic support from the Polar Continental Shelf Project and the Royal Canadian Mounted Police. Many thanks to all these organizations for their support. Finally, I would like to extend my thanks to the most important player in this research, a wonderful plant called Cassiope tetragona. May you flourish in the arctic light for many, many years to come. xiii 1. Introduction Many classic studies of arctic tundra vegetation have suggested that plant growth is directly limited by the low temperatures characteristic of that region (e.g. Bliss, 1956; Bliss, 1962; Warren Wilson, 1966; Billings and Mooney, 1968). More recent experimental studies have indicated that plant growth on the tundra may be limited by a wide range of factors which are indirectly related to temperature (Haag, 1974; Chapin, 1983, 1987; Henry et al, 1986; Shaver et al, 1986; Billings, 1987; Shaver and Kummerow, 1992), and that arctic species may respond individually to changes in potentially limiting factors (Shaver and Chapin, 1980; Chapin and Shaver, 1985). In the Arctic, increases in mean annual temperatures of up to 5-10°C are expected due to changes in the concentration of greenhouse gases in the atmosphere (Etkin, 1990; Maxwell, 1992). Most of this warming is expected to take place in the winter, and July temperature increases with a doubling of C Q 2 are predicted to be around 1-3 °C for the Arctic Islands (Etkin, 1990; Maxwell, 1992). Such temperature^ changes could indirectly affect tundra vegetation through alterations in the length of the growing season, amount of precipitation and winter snow accumulation, active layer depth, and soil nutrient availability and aeration, as well as through the direct effects of increased temperature (Chapin, 1984). If the vegetation response to climate change is to be predicted and understood, present interactions between plant growth and climate must be studied. In an effort to fulfill this research goal, the International Tundra Experiment (ITEX) was established to organize research focused on predicting and monitoring arctic plant response to climate change in tundra environments (Molau, 1993a). The main core of the ITEX investigation is the implementation of a common experiment at arctic and alpine sites designed to investigate plant response to levels of surface temperature warming similar to those predicted by global circulation models for the Arctic region. A portion of the results presented in this study are based on the implementation of this common ITEX experiment. 1 This investigation has specifically focused on the response of one species of arctic evergreen dwarf shrub, Cassiope tetragona (L.) D. Don (arctic white heather), to natural and simulated variations in growing season climate. C. tetragona forms an important component of high and mid-Arctic vegetation throughout the circumpolar North, with extensions into alpine regions (Hulten, 1968). Established individuals of C. tetragona generally form loose mats or cushions of multiple-branching, monopodial shoots (S0rensen, 1941). Stems produce four rows of alternating, resinous leaves which remain green for 3-5 years and are retained on the stem for approximately 20 years (Murray and Miller, 1982; Callaghan et al., 1989). Reproduction is predominantly sexual (S0rensen, 1941), and plants produce small, white, bell-shaped flowers which develop into erect fruits with numerous seeds. Despite a reliance on sexual reproduction, seedlings of Cassiope tetragona have only rarely been noted in the field (Freedman et al., 1982; Havstrom et al., 1993). Cassiope tetragona shows many of the common characteristics of arctic evergreen shrubs, such as slow growth, low biomass turnover and a relatively long life-span for established individuals (Nams and Freedman, 1987b; Shaver and Kummerow, 1992). Previous studies of arctic evergreen shrubs indicate that such species are likely to show relatively conservative responses to ameliorated growing conditions, such as increases in soil nutrient levels, water availability and temperature (Henry et al., 1986; Chapin and Shaver, 1989; Welker et al., 1993). Recent investigations of Cassiope tetragona indicate that the growth of this species responds positively to increases in summer temperatures in the northern part of its range (Callaghan et al., 1989; Havstrom et al., 1993), suggesting that high arctic populations of C. tetragona may be sensitive to the variations in climate predicted with global climate change. This research has further investigated the responses of Cassiope tetragona to variations in climatic factors at Alexandra Fiord, Ellesmere Island. Observations of plant responses to experimental manipulations of growing season temperature and growing 2 season length were used to examine the short-term effects of these factors on Cassiope growth and reproduction (Chapter 2). Additional observations of Cassiope plants were made within communities representing a range of environmental characteristics, in order to provide information on how natural variations in climate and other factors may influence patterns of productivity within different communities (Chapter 3). Retrospective analysis of growth and reproduction was also used to provide a record of Cassiope productivity which could be related to climate records over a period of several decades (Chapter 4). The results of these studies provide insight into the response of Cassiope tetragona to annual variations in climate and short-term experimental changes in growing season climate. This information is useful for predicting future responses to climate change and interpreting past variations in growth and reproduction of Cassiope tetragona, as well as stimulating additional hypotheses about important factors controlling the population dynamics of this species. 3 2. Response of Cassiope tetragona to experimental manipulations of surface temperature and snowmelt timing 2.1. Introduction Hypotheses concerning limiting factors such as temperature, growing season length, and the availability of water and nutrients are generally used to explain patterns of variability in arctic plant productivity {e.g. Webber, 1978; Billings, 1987; Chapin, 1987). Recently, the prospect of rapid global climate change has increased our desire to understand the underlying causes of growth variability within populations, species and communities in order to predict the potential consequences of various climate change scenarios. The International Tundra Experiment (ITEX) was organized to provide the framework and initiative for investigations of plant population and community responses to climate change throughout the circumpolar North (Molau, 1993a). The objective of the ITEX research is to predict vegetation response to climate change through field experimentation and to monitor response to current climate patterns. Several widely distributed key plant species have been selected from arctic and alpine tundra environments for focused research (Molau, 1993a). This research project focuses on one of these species, Cassiope tetragona (L.) D. Don, or white arctic heather, an evergreen dwarf shrub which forms an important component of circumpolar arctic vegetation (Bliss and Matvayeva, 1992). Previous studies of High Arctic populations of Cassiope tetragona have found strong correlations between growth of this species and summer climate conditions (Callaghan et al., 1989; Havstrom et al., 1995) and flower production has also been observed to be highly sensitive to climate conditions in the previous year (Bliss et al., 1977; Nams and Freedman, 1987b). Experimental manipulations of temperature, nutrients and light at a Svalbard site caused a positive growth response of Cassiope tetragona to 4 increased temperatures in the field, suggesting that northern populations of this species are primarily temperature-limited and would respond positively to increases in average summer temperature (Havstrom et al, 1993). Preliminary investigations of Cassiope growth under experimental warming at Alexandra Fiord, Ellesmere Island also indicate a positive growth response of C. tetragona to increased summer temperatures (Nams, 1982). These investigations used relatively large temperature increases for their experimental manipulations (approximately 2-5° C), and may not be representative of the vegetative response to more moderate levels of summer warming (1-3°C) predicted by some General Circulation Models for the High Arctic (Etkin, 1990; Maxwell, 1992). The magnitude of short-term plant responses to variations in a single limiting factor, such as temperature, is also likely to depend on the importance of other potentially limiting factors (Chapin et al., 1987), which may vary among different communities (e.g. McGraw, 1985; Galen and Stanton, 1993). Timing of growth initiation for many tundra plants has been noted to be strongly dependent on the date of snowmelt (Bell and Bliss, 1979; Murray and Miller, 1982; Nams and Freedman, 1987b; Kudo, 1991; Larigauderie and Kummerow, 1991; Shaver and Kummerow, 1992). The length of the growing season may be an important factor limiting the productivity and sexual reproductive success of tundra species (Chapin, 1983, 1987; Kudo, 1991; Galen and Stanton, 1993; Molau, 1993b). Variations in snowmelt associated with precipitation and temperature changes may strongly affect the growth and reproduction of species such as Cassiope tetragona, which is often prominent in areas of late-lying snow (Porsild, 1920; Brassard and Beschel, 1968; Bliss et al, 1977; Miller, 1982; Bliss and Matvayeva, 1992). To date, no experiments have been conducted to examine the effects of snowmelt timing on Cassiope tetragona. This study was designed to investigate the responses of Cassiope tetragona to variations in temperature and growing season length at Alexandra Fiord, Ellesmere Island. Two years of field manipulations of surface temperature and snowmelt timing were used to 5 examine the short-term vegetative and reproductive responses of C. tetragona to variations in these factors. Manipulations of surface temperature were initiated in two different community types, in order to investigate possible between-site variations in the observed responses. A variety of vegetative and reproductive parameters were also examined to provide information on within-plant interactions that may be important in determining the net response of C. tetragona to variations in growing season climate. 6 2.2. Methods 2.2.1. Study area Field research was conducted at the Alexandra Fiord lowland (78°53'N, 75° 55'W), located on the eastern coast of central Ellesmere Island in the Canadian Arctic Archipelago (Figure 2.1). The lowland is a semi-triangular, gently sloping glacial outwash plain of approximately 12 km 2 in area. It is bounded by upland plateaus (approx. 500-700 m) to the east and west, the Prince of Wales Icecap to the south and the waters of Alexandra Fiord to the north. The mean summer temperature (June - August) of the lowland is approximately 5° C, with mean annual precipitation of approximately 10-20 cm (Freedman et al., 1983; Labine, 1994). Although summer precipitation is generally below 1 cm, the large influx of meltwater to the lowland from surrounding glaciers and uplands provides a steady supply of water throughout most of the growing season. This water supply is further augmented by liquid water generated by the annual thawing of the active layer. Summer temperatures within the topographically-protected lowland are often significantly warmer than surrounding upland areas (Labine, 1994). As a result, the Alexandra Fiord lowland forms a sort of hydrologic and climatic 'oasis' and its relatively lush vegetation contrasts strongly with the large areas of polar desert and semi-desert in the surrounding region (Freedman et al., 1983). Over 50% of the lowland is covered by heath communities dominated by Cassiope tetragona and Dryas integrifolia (Nams and Freedman, 1987a; Muc et al., 1989). Unlike many other arctic sites where C. tetragonais commonly restricted to areas of late-lying snow (Porsild, 1920; Brassard and Beschel, 1968; Bliss et al., 1977; Bliss and Matvayeva, 1992), C. tetragona occupies a wide range of habitats within the Alexandra Fiord lowland (Nams and Freedman, 1987a; Muc et al, 1989). 7 Figure 2.1: Map of the circumpolar North showing locations of current ITEX sites. Point number five gives the location of Alexandra Fiord. Map prepared by G. M. Marion, U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire. 8 2.2.2. Surface warming experiment 2.2.2.1. Study sites Two adjacent communities (approximately 300 m apart) in the central plain of the lowland were selected as sites for temperature manipulation experiments. These communities correspond to variants of the dwarf shrub - cushion plant community type described for the Alexandra Fiord lowland (Muc et al., 1989). The two selected communities have similar species composition, but differ in dominant species. One site is dominated by Dryas integrifolia, with Cassiope tetragona, Car ex misandra, Eriophorum angustifolium, Luzula confusa, Papaver radicatum and Arctagrostis latifolia also prominent in the community. The second site is strongly dominated by Cassiope tetragona, with Dryas integrifolia, Luzula arctica and Salix arctica as subdominants. At Alexandra Fiord, vegetation dominated by Dryas frequently overlaps with Ca^/ope-dominated communities, with Dryas communities generally occurring in drier habitats with earlier snowmelt (Muc et al., 1989). Soils in both community types at Alexandra Fiord are characterized by alluvial sands and gravels overlaid by a thin (0-6 cm) layer of humus and organic litter (Muc et al., 1994). The topography of the study sites is a gently sloping plain with a slightly northern exposure (<2°). Microtopographic variation is caused by numerous small earth hummocks and scattered boulders. 2.2.2.2. Experimental manipulations Experimental manipulations of near-surface temperatures were established at the Dryas-dominated community (=Dryas ITEX) and Cassiope-dominated community (=Cassiope ITEX) as part of the International Tundra Experiment (ITEX) program. At these sites, open-top hexagonal greenhouses were used to increase ground surface temperatures (Marion et al., 1993, 1995; Molau, 1993a). The open-top chambers (OTCs) 9 are constructed from transparent fiberglasst with high solar transmittance in the visible spectrum (86%), and are designed to stay in place in the field over several years. The chambers have a top diameter of 1.5 m and a height of approximately 60 cm. Sides which are angled in towards the center help trap heat within the greenhouse by reducing convection at the ground surface. The chambers are open in order to allow normal levels of precipitation and to prevent large changes in humidity and C 0 2 concentrations from developing inside. The open top and angled sides increase penetration of solar radiation relative to close-topped greenhouses with perpendicular sides. Field testing of the OTCs at the Cassiope ITEX site showed an average increase in ground surfaces temperatures of , 1.4°C within the OTCs over a week-long period in July of 1992 (Marion et al., 1993). Fourteen treatment and control replicates were established in mid-June, 1992 within visually homogeneous areas of the Cassiope and Dryas ITEX sites. Positioning of plots was done selectively in order to include a minimum number of individuals of each target species (four target species at the Cassiope ITEX site, five at the Dryas ITEX site). Control plots were established to be of approximately the same area as the OTCs (circa. 1.8 m2). OTCs and control plots were spaced a minimum of 2 meters apart from each other in order to prevent snow drifting patterns from affecting the snowpack conditions of adjacent plots. 2.2.3. Snowmelt experiment 2.2.3.1. Study site A separate snowpack manipulation was established in a nearby snowbed community on the lee side of a north-facing beach ridge. Snow accumulation in this area results in a persistent snowbed which is among the last to melt in the lowland (Nams and Freedman, 1987a; Woodley and Svoboda, 1994). Vascular plant cover at this site is t Sun-Lite HP (1.0 mm thickness), manufactured by Solar Components Corporation, 121 Valley St., Manchester, New Hampshire, 03103 U.S.A. 10 approximately 75%, with cushions of Cassiope tetragona accounting for the majority of vascular cover (Nams and Freedman, 1987a). Soil composition is a thin (0-3 cm) organic layer over coarse alluvium. The soil becomes saturated during snowmelt and remains moist for most of the summer (Nams and Freedman, 1987a). 2.2.3.2. Experimental manipulations The aim of the snow manipulations was to alter the timing of growing season initiation by changing the date of snowmelt. The experiment involves three treatments: snow removal, snow addition and control (ambient snow conditions). Snow removal treatments were established by manually removing snow to a level 5-10 cm above the vegetation surface. Excavated snow was added to adjacent snow addition plots in the corresponding treatment block, resulting in -1, +1, and 0 treatment levels. Snow manipulation plots were established on June 19, 1992 and June 6, 1993. Snowpack depth within the experimental blocks in 1992 was approximately 50-90 cm at the time of excavation; in 1993, average snow depth at time of excavation was between 60 and 110 cm. Each plot was approximately 3x3 m2 and spaced at least one meter apart from the other plots along a transect. Plots were arranged in three blocks along a gradient in snow accumulation caused by patterns of snow drift. Snow removal plots were placed adjacent to addition plots in order to facilitate the transfer of snow and minimize impact on the natural melt patterns within the surrounding snowbed. 2.2.4. Environmental measurements 2.2.4.1. Snowmelt progression Snowpack depth was measured as depth of penetration of a calibrated metal rod. As the ITEX sites were established after snowmelt in 1992, snowpack measurements were taken only in 1993. Measurements were taken from the north, east, south, west and center 11 areas of each plot. Snowpack depths were sampled every 2-3 days until all plots were free of snow. At the Beach Ridge site, snowmelt in the experimental plots was visually inspected every 1-2 days following treatment establishment in 1992 and 1993. Snowmelt date was recorded for each plot as the first day the plot was observed to be completely free of snow. Actual measurements of snowpack depth were taken only in the control plots, along the north and south edges. 2.2.4.2. Active layer thaw Active layer depth was measured in a manner similar to snowpack depth, using the penetration of a metal rod into the soil. Systematic active layer measurements were taken only in 1993. At the ITEX sites, active layer measurements were sampled following the same design as the snowpack measurements. For the first 8-10 days following snowmelt, sampling was done at short intervals (2-3 days). Active layer depths were sampled again in late July and early August to estimate maximum depth of thaw. Active layer depth was not measured at the Beach Ridge site because the high gravel content of the soil inhibited penetration of the ground with the metal probe. 2.2.4.3. Soil moisture Soil moisture was measured every 10-12 days from 15 June to 14 August, 1993 at the ITEX sites. Measurements of soil moisture were not made within the experimental plots at the Beach Ridge. Samples were taken from 4 pairs of destructive sampling plots at the Cassiope and Dryas ITEX sites. Soil cores were sampled from the upper 6 cm of soil. Vegetative matter above the ground surface was excluded from the core. Eight cores were sampled at each site. The cores were weighed immediately after sampling and then dried at 40°C for 24 hours and weighed again. Moisture levels were calculated as percentage of dry weight, 12 %DW = j *100 (2.1) where Wwis the weight of the fresh, wet sample and Wj is the weight of the sample after drying. 2.2.5. Measurement of growth and reproduction 2.2.5.1. Sampling scheme Shoots of C. tetragona were tagged for repeated observations of shoot elongation and phenology at the beginning of the 1992 growing season. In 1993, additional shoots were re-tagged to replace those that had not grown or had been lost in 1992. Where possible, individual shoots were randomly selected from different clones in the experimental plot. In a few instances, the number of Cassiope clones necessitated that more than one shoot was tagged on an individual clone in order to obtain the desired within-plot sample size. Observations were made on 10 shoots within each plot at the Cassiope ITEX site and 5 shoots per plot at the Dryas ITEX site. In the OTCs, shoots were only tagged in the area directly beneath the open top of the chamber (1.5 m diameter). Sample size in the Beach Ridge snow manipulation plots was 20 shoots per plot. Shoots were not tagged within a 10 cm border around the edge of the plot in the snow removal and addition treatments. Measurements of shoot elongation and phenology were recorded at intervals of 6-7 days in 1992 from snow release to 11 August, and every 4-5 days in 1993 from snow release to 14 August. Plant measurements at the ITEX sites were done with the aid of large saw-horses which could be placed over the OTCs to provide a support structure for making observations. The saw-horses were used to support one's upper body while the arms were being used for measurements. 13 2.2.5.2. Vegetative phenology Phenological stages were recorded as the date at which that stage was first observed for each shoot. Three stages of vegetative phenology were distinguished in this study: bud break, growth initiation, and cessation of growth. Definitions of the phenological stages are given in Table 2.1. The number of days between bud break and growth cessation and between growth initiation and growth cessation were used to provide estimates of effective growing season length and the period of active growth, respectively, for each shoot. 2.2.5.3. Shoot elongation Quantitative measurements of shoot elongation were obtained at the same time as phenological data were gathered. At the beginning of each growing season, the upper-most leaves of each tagged shoot were marked with dabs of non-water soluble liquid paper in order to delineate the border between the current year's and previous year's growth. Shoot elongation was then measured as the distance from the marked leaves to the tip of the shoot. Measurements were made to the nearest 0.05 mm using a hand-held caliper. Daily elongation rates were calculated as ER»day= -1—L (2.2) where // and (2 are stem lengths measured on sample days t] and ?2> respectively. 2.2.5.4. Reproductive phenology Reproductive phenology was recorded as stages of flower bud appearance, peduncle elongation, corolla opening, corolla drop and mature fruit appearance. Definitions of the phenological stages are given in Table 2.1. Fruit dehiscence did not occur by the end of field seasons in 1992 and 1993, and was not measured. 14 c o Q c C L o T J 3 X) 60 C •c B c > o <u X3 CO CU c O g O 3 _e co w .22 -S c3 ~ oo| c oo <u co <u OO o <u CL. C L 3 TD C & <U > CU CU X5 •t—» -a c o £ -T J £ oo S d CU <u > co X ! oo CU > CU c c <u x; CO T J O l- , oo i— cd 3 O > C J 3 a (U IB £ cu oo Xi O C oo I .1 00 « C i> O <U 00 X) O 00 _o o c <u 3 C U 8 XI oo s  s o x: ^ CU CO CU T J .2 CO CO <u <u J3 c "o & 00 T J C CO <U 'oo CO JU "o (U CO oo T J 3 X) u O T J <U C L _o <u > <u T J c 3 _co 15 i— o o <u X) > <u oo JU CJ C 3 T J <U C L C ct o J2 "o O O u. o 00 c 'a. _o > CU T J <U X i e p c <u a 00 CO xs JO I-I 8 u, o T J a X! OJD CO t/2 es "53D _© "o c <u ft* T J 3 Xi c o c X! O LH 00 c o CO 00 00 <U O O 00 <U CJ C <u oo T J 3 X) !_ (U o CO 00 c o 13 JU c 3 T J ct 00 c '2 ct o cO o o C L 2 T J CO O O DC _o "o c <u x; ft* 4* 00 > > "•5 s»J -5 ° © o u c Cl v X ft. 15 2.2.5.5. Flower production Estimates of flower production for 1992 and 1993 were obtained by determining the proportion of tagged shoots which produced mature flowers during the growing season. More thorough estimates of flower production were made in 1993, by counting numbers of live shoots and reproductive structures within randomly placed 10 cm 2 quadrats at each site. Elongating flower buds, mature flowers and immature fruits were included in the count of reproductive structures in the quadrat. Because only Cassiope shoots greater than 15 mm in length had been observed to produce flowers at Alexandra Fiord (personal observation; see also S0rensen, 1941) live shoots less than 15 mm long were not counted in these measurements. These measurements were used to produce an estimate of the average number of flowers produced per mature shoot at each site. Within the Beach Ridge snow manipulation plots, 12 quadrats were sampled for each plot. At the Cassiope and Dryas ITEX sites, 4 quadrats were sampled within 12 of the 14 sets of treatment and control plots. Sampling was done at or immediately following peak flowering times (Cassiope ITEX = 9-11 July, Dryas ITEX = 11-12 July, Beach Ridge = 16-22 July). 2.2.5.6. Seed germination Effects of the OTCs on seed germination were measured using seeds sampled from the Cassiope ITEX site on 18 August, 1993. It was not possible to perform a similar test on seed germination for the Dryas ITEX site because too few plants had produced mature fruits outside the treatment plots at the time of sampling. In order to obtain adequate sample sizes for the control population at the Cassiope ITEX site, additional seeds were sampled from the general area of the community, as well as from designated control plots. Seeds sampled from individual plots were pooled into treatment or control groups. The seeds were stored in a refrigerator (~2-4°C) for 8 months. Approximately 1000 seeds were sub-sampled from each group and divided into 5 petri dishes for germination. Seeds 16 were placed on two layers of sterile filter paper and kept under 24 hour fluorescent light for 34 days at approximately 20° C. The seeds were watered every 1-2 days with distilled water. Seeds were considered to be germinated when the cotyledons had visibly emerged from the seed coat. Germinated seeds and seeds with fungal or algal growth were counted every 2-4 days under a binocular microscope (lOx magnification) and removed from the petri dishes. 2.2.6. Data Analysis 2.2.6.1. Surface warming experiments Tagged shoots which did not grow, were lost during the field season or which exhibited growth anomalies were discarded from the data set, after testing for factor effects on the frequency of these occurrences (Appendix 1). When data complied with assumptions of parametric statistics, analysis of variance (ANOVA) was used to test for factor effects (ANOVA procedure, SAS, 1985). In the temperature manipulation experiments, sites, years and treatments were considered to be fixed. A modified ANOVA for unbalanced designs was used where there was an uneven number of samples within each factor (GLM procedure, SAS, 1985). Where 2 years of data were available, a three-way analysis of variance was used to examine main and interactive effects of year, site and temperature manipulation. A two-way ANOVA was used where data were only available for one year. Factor effects on snowpack depth, active layer depth and soilmoisture were analyzed using a repeated measures ANOVA (ANOVA procedure, SAS, 1985) with sample dates treated as blocks in the analysis (Neter et al., 1990). Analyses of snowpack, active layer, and vegetative parameters at the ITEX sites were based on plot means obtained from the within-plot samples. Germination results were analyzed using petri dishes as replicates. 17 Data distributions were examined using frequency histograms. Data which were clearly not normally-distributed or had prominent outliers were either transformed to normality or analyzed using non-parametric techniques. Rank-transformation was used for most of the non-parametric analysis. Using this technique, observed values were ranked across a data set (RANK procedure, SAS, 1985) and the resultant ranks were analyzed using the parametric analysis procedures described above. Rank transformation was chosen for non-parametric analysis because it is relatively easy to implement and can be used for multi-factor analysis, as was desired for this study (Conover and Iman, 1981; Zar, 1984; Potvin and Roff, 1993). Where non-parametric tests on levels of a single factor were required, a Wilcoxon rank sum test was used (NPAR1WAY procedure, SAS, 1985). 2.2.6.2. Snowmelt experiment Analysis of treatment effects for the snow manipulation experiment was performed using linear regression analysis (REG procedure, SAS, 1985), rather than an analysis of variance based on treatment levels. Although the treatment replicates were organized in blocks within the snowbed, application of a randomized block design analysis of variance to this data set is not appropriate because the positioning of treatment plots within blocks was not random. Because the aim of the experiment was to alter the date of snowmelt, and treatment plots of a given level did not become snow free simultaneously, it was desired to use an analysis which would assess the effects of snow release date directly, rather than effects of the general treatment levels. To achieve this goal, simple linear regression was performed on plot averages of the dependent variables with snowfree date, rather than treatment level, as the predictor variable. Analysis of snowmelt timing effects on vegetative phenology of Cassiope tetragona was performed using separate analyses of 1992 and 1993 data. Analysis of reproductive phenology was pooled between years in order to obtain a larger sample size. Because the majority of shoots sampled in 1992 had not stopped growing by the end of the observation 18 period, only 1993 data was used in the analysis of timing growth cessation. Flower production per shoot measured in 1993 was regressed with snowmelt dates in 1992 and 1993, as flower production of Cassiope is often correlated with growing season conditions in the previous year (Bliss et al., 1977; Nams and Freedman, 1987a). Where possible, data which did not conform to the requirements of parametric statistics was transformed to normality. In cases where transformations were not able to solve deviations from normality, and where rank transformations were not appropriate, parametric analysis was performed with the understanding that the probability levels associated with the tests were likely to be inaccurate. 19 2.3. Results 2.3.1. Surface temperature manipulations 2.3.1.1. Environmental factors In early June of 1993, both experimental sites were still covered with snow and there was no obvious indication that the OTCs had significantly affected winter snow accumulation in the treatment and control plots. The OTCs also do not appear to have significantly affected the progression of snowmelt at the two sites (Table 2.2; Figure 2.2, 2.3). Snowpack measurements taken at the two sites suggest a general trend of more rapid snowmelt within the treatment plots, particularly at the Dryas ITEX site, but this effect is not significant (Table 2.2). However, the OTCs did exhibit a moderately significant negative effect on snow release date in the treatment plots relative to controls at both sites (p<0.06, Appendix 3). This may be a result of differential melting between the center and edge areas of the OTCs. General observations during snowmelt suggested that the edges of the OTCs melted earlier than the centers. Because the definition of snow release used in this study is the date at which the entire plot was bare of snow, differential melting within the plot would be likely to produce differences in the patterns of snow release compared to average snow depth, which is based on measurements in the 4 quadrants plus a single center measurement in each plot. Snowpack depth in early June was less at the Dryas ITEX site, suggesting reduced winter snow accumulation at this site than at the Cassiope ITEX site. Snowmelt occurred more rapidly at Dryas ITEX, with plots becoming snow-free approximately 4 days before those at Cassiope ITEX (Table 2.2; Figure 2.2, 2.3). Depth of thaw in the active layer at Dryas ITEX was consistently greater that at Cassiope ITEX for the majority of the summer, although maximum depth of thaw recorded on the final sampling date was similar at the two sites (Table 2.2; Figure 2.2, 2.3). Early 20 Table 2.2: Effects of site and treatment on environmental characteristics measured at the Cassiope and Dryas ITEX sites in 1993. Effects were analyzed using a repeated measures analysis of variance. Effect Snowpack depth Active layer depth Soil moisture Site *#* #** ns Treatment ns ns ns Site x Treatment ns ns ns Date *** #** *** Date x Site *** *#* ns Date x Treatment ns * ns Date x Site x Treatment ns ns ns Asterix denotes significance of effects: *** = p <. 0.001, ** = p ^ 0.01, * = p <, 0.05, ns = not significant. Table 2.3: Effects of year, site and treatment on the timing (day number in calendar year) of vegetative bud break and initiation of active growth for Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1992 and 1993. Values were rank-transformed and analyzed using a three-way ANOVA. Effect Timing of vegetative bud break Timing of growth initiation Year *** *** Site ns ns Treatment ns ** Year x Site *** *** Year x Treatment ns ns Site x Treatment ns * Year x Site x Treatment ns ns Asterix denotes significance of effects: *** = p ^ 0.001, ** = p <; 0.01, * = p ^ 0.05, ns = not significant. 21 s mj Snowpack - OTCs S3 Active layer - OTCs Snowpack - Controls C3 Active layer - Controls •8 V •8 C •s •I 1 ~ n — 155 158 160 - 1 1— 162 164 Day number 166 205 221 Figure 2.2: Progression of snowpack and active layer thaw in treatment and control plots at the Cassiope ITEX site in 1993 (n=14). Error bars represent 1 standard error. -40 H Snowpack - OTCs 3 Active layer - OTCs Snowpack - Controls H Active layer - Controls 162 164 Day number Figure 2.3: Progression of snowpack and active layer thaw in treatment and control plots at the Dryas ITEX site in 1993 (n=14). Error bars represent 1 standard error. 22 and mid-season differences in active layer depth between the Dryas and Cassiope ITEX sites probably reflect a delay in thaw initiation due to the lag in snowmelt at Cassiope ITEX. Active layer depths at the two sites show a significant date* treatment effect associated with the OTCs (Table 2.2). This significant interaction appears to be related to a delay in the initiation of active layer thaw in warming plots at the beginning of the summer and increased active layer depth in warming plots at the end of the summer. Early season delays in active layer thaw may be associated with a similar delay in the timing of snow release, discussed above. Positive effects of the OTCs on maximum active layer depth appears to be largely the result of strong differences at the Cassiope ITEX site (Figure 2.2; treatment effect p<0.05, Appendix 4). Measurements of soil moisture show high levels of variability at both sites (Figure 2.4), probably due to a combination of small sample sizes and large microsite moisture differences. Few patterns are evident in these measurements except a shift to dryer soils over the course of the summer, most likely associated with decreases in snowpack and permafrost meltwater. Although it was expected that Dryas ITEX would exhibit lower soil moisture by the end of the summer (see methods, above), significant differences in soil moisture between the Dryas and Cassiope sites are not apparent in these measurements (Table 2.2). Consistent treatment effects on soil moisture over the course of the summer are also not apparent. Soil moisture measurements made in mid-summer at Dryas ITEX show a significant negative response to surface warming, with moisture levels in the OTCs 5 5 % lower than in the controls (p<0.05, Appendix 5; Figure 2.4). Soil moisture measurements on this date were taken following a two-day period of warm, clear weather, and may reflect increases in evapotranspiration within the chambers. 2.3.1.2. Response of Cassiope tetragona: vegetative phenology and growth Experimental warming did not significantly affect the date of bud break at either site, but did result in earlier growth initiation at the Cassiope ITEX site (Table 2.3; 23 CO CO "o g u e u o o o u o U u u Q Q a El H ///////////////////// r//V tvss/ss/s/ss/sssss '/SSSVSSSSSSSSSSSSSSSs mm 'SSSSS-S/SS/SS/SSSSSSSSSSSS/SSJ' o o CN * h-\SSS/SSSSSS/S//// \ssssssss//ssssss wmmm 00 00 / / • / / / / , ' / / / / / / ' / • / / / / / ' / / / / / / / / • / / ' / / / / / / / / / / / ' ^ / / / / / / / / / / / / / / / / / / / / / / / / / / / rrft't t Issssssss/ssssssssss/ssssss/ ssssss>sss/ss/s/s/sss/sss//sssssss 3 Q VO / / / / / / / , ss/ss/s sssssssss/sssssssssssssssssssssss ssss/s//ssssssss/ss/sssssss/sssssssss / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / VO (iqSi3M Ajp jo jiraojad) ajnjsioiu jios CO o . dl ! i ! m «3 O ca co Q X J G (U -+-» So u CO -IP CO pH ai G *j co W U 2^ If ^ 00 a CO o CO 3 GO °o s § o -a a <D O s » _ i f s » o <-> CU ,<u i ° = s § 5 S > rs _, o c •ci co ao o> eg CN co Ul 00 E o •c 24 Figure 2.5, 2.6). Plants at both sites showed a significant reduction in the number of days from snow release to the start of shoot elongation associated with the OTCs in 1993 (p<0.001, Appendix 6). The OTCs did not significantly affect the calendar date of growth cessation, but there was a significant decrease in the number of days from snowmelt to growth cessation within the chambers at both sites in 1993 (p<0.05, Appendix 6). Vegetative bud break, growth initiation and growth cessation occurred earlier in 1993 than in 1992 (Table 2.3, 2.4), with plants at the Dryas ITEX site showing a significantly stronger response to annual variations than those at the Cassiope ITEX site (Table 2.3; Figure 2.5, 2.6). Despite this difference in the degree of sensitivity to annual variations, dates of bud break and growth initiation did not differ significantly between the two sites (Table 2.3). However, the number of days from snowmelt to bud break and growth cessation in 1993 do show significant site effects (p<0.05, Appendix 6). When standardized by snowmelt date, the timing of these phenophases was advanced by approximately 2 to 5 days (bud break and growth cessation, respectively) at the Cassiope ITEX site compared to the Dryas ITEX site. The timing of vegetative phenophases was used as an indication of the portion of the growing season used by the plants for growth. The number of days from bud break to growth cessation for Cassiope tetragona was shorter in 1992 than 1993 at the Cassiope ITEX site (Table 2.4, Figure 2.4), but the number of days from growth initiation to growth cessation did not differ between the two years (Figure 2.5). Because 1992 data on growth cessation at the Dryas ITEX site is not available, it is not possible to make a similar comparison between years at that site. The period from growth initiation to growth cessation may have been slightly lengthened in the OTCs at the Cassiope ITEX site (Table 2.4). A similar effect was not apparent at the Dryas ITEX site in 1993 (Table 2.5). Growth cessation at Dryas ITEX in 1992 was excluded from the analysis discussed above. Phenological observations at this site were terminated early in 1992, and measurements of shoot elongation indicate that shoot growth had not ceased for the 25 u I cn ON 5 H ro 5 CJ I cn ON V ro 0\ O U I CN ON H CN ON 5 U CN ON o H I CN ON u o cs lO cs o CN o r-i ON O 00 o 00 t o r-D C co Q 11 CO o CD £ X fc X ) T J t "2 -Si" * S <— ir T3 '—1 OO a c CO • -_ , CU Si T J co c -a CO " CJ II <u 2 c l - 0 ~ 2 o CO <— '53 H ±r _C ->-' r-1 co ii 1 / 1 3 ? C cu a TJ o d = 3 ^ CU CU CD 1 • £3 >-> CU g o » • i CO i- CO cN 05 ° S o 5 c 6 ° 5 T J s § •£3 C C C CU O S8 8f 2 00 O co cu o M X ! 00 c £ 'S P CO « oo CJ s * - 00 ° * T J £ «J S 45 n p <u o obro T J <u oo o ON X) ~ .8' 4—1 C in CN CU 00 S cu 8 <a s.s s 26 <0 <N O CN <N c i o IT) O CN 8 ON o ON 0 0 o 0 0 CD X) 3 C cO Q U I co ON 5 H co O N 5 U I CO O N U H I CO O N U O I CN ON Q <N ON Q U CN O N U l (N O N U § 8\ s CD ON a* B~ c 5-2 s I ' - O G XJ o L -oo^ s o ' O ra Q II P5 — »_e w a u « <D O o to £ '53 fc « <- CD ^ T 3 s -s 2 jy "O c5 C c§ § a » Q §n8x-8 5 o rt H d o u u 5 O £ Q I • S «* CM i - u_ > f J ^ MO > I "£.§ § « S OJ « D y » o 2 c CO tg xi'•= C C M ( J U g J ) ° ^ •= - S x : — 1 x : _ •c £ c £ " ^ O f l C O « c XJ H 2 oo 13 : ON VO CO CN " a 3 oo E £ T 3 o CD c co CO T3 CD O C 3 27 Table 2.4: Effects of year and treatment on the timing of growth cessation, the number of days from bud break to growth cessation and from growth initiation to growth cessation, and total shoot elongation of Cassiope tetragona observed at the Cassiope ITEX site in 1992 and 1993. Days from bud Days from Total shoot Date of growth break to growth growth initiation elongation Effect cessation0 cessation^ to cessation (mm)^ Year *** ** ns ns Treatment ns ns + ** Year x Trmt ns ns ns ns Asterix denotes significance of effects: ***=ps 0.001, ** = p <; 0.01, * = p =s 0.05, + = p^ 0.1, ns = not significant. M> Indicates values were log-transformed for analysis;Q Indicates values were rank-transformed for analysis. Table 2.5: Treatment effects (1993 only) on the timing of growth cessation and the number of days from bud break to growth cessation and from growth initiation to growth cessation, and year and treatment effects (1992 and 1993) on total shoot elongation of Cassiope tetragona observed at the Dryas ITEX site. Days from bud Days from Total shoot Date of growth break to growth growth initiation elongation Effect cessation^2 cessation^ to cessation^ (mm)H1 Year ** Treatment ns ns ns ns Year x Trmt ns Asterix denotes significance of effects: *** = p £ 0.001, ** = p <; 0.01, * = p <. 0.05, ns = not significant. ^ Indicates values were log-transfonned for analysis; ° Indicates values were rank-transformed for analysis. 28 majority of shoots on the last date of observations (Figure 2.7). Values of maximum shoot elongation for Dryas ITEX in 1992 were retained in the analysis, although it is likely that shoot lengths are under-estimated for this group. Peak elongation rates were significantly higher in the OTCs in 1992 but not in 1993 (Table 2.6; Figure 2.7 - 2.10). The timing of peak elongation occurred earlier in the OTCs in both years. General patterns in shoot elongation observed in 1992 and 1993 show that surface warming by the OTCs may cause accelerated elongation at the beginning of the growth period, but may be associated with a slightly more rapid decline in elongation rates at the end of the growth period (Figure 2.7 - 2.10). Peak shoot elongation rates differed significantly between sites and years (Table 2.6), with the most rapid growth occurring in 1993 at the Dryas ITEX site (Figure 2.9). The timing of peak shoot elongation at the Dryas ITEX site differed strongly between years, with a smaller difference in timing between years at the Cassiope ITEX site (Table 2.6). A noticeable dip in shoot elongation growth trends occurred simultaneously in 1992 at the Cassiope and Dryas ITEX sites in both treatment and control plots (Figure 2.7, 2.8). This disturbance corresponds to a 2-day period of cool, overcast weather which included a storm that deposited approximately 1-2 cm of snow on the lowland. The concurrence of these events may indicate that vegetative growth of C. tetragona is responsive to short-term environmental fluctuations during the growing season, as well as to larger scale differences between years. Total shoot elongation observed for Cassiope tetragona at the Dryas and Cassiope ITEX sites shows similar patterns to those observed for vegetative phenology, with plants at Dryas ITEX exhibiting significantly greater sensitivity to annual differences than those at Cassiope ITEX (Figure 2.11; Table 2.6). As shoot elongation clearly was not completed at the Dryas ITEX site at the last measurement date in 1992 (Figure 2.7), this difference may be exaggerated. Values of total shoot elongation for Cassiope tetragona also show a moderate year*treatment effect associated with the open-top chambers when analyzed in a 29 Table 2.6: Effects of year, site and treatment on shoot elongation of Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1992 and 1993. Peak elongation Date of peak Total shoot Effect rate (mm/day)^ elongation52 elongation (mm)^ Year *** *** * Site *#* ns ns Treatment ns * ns Year x Site ns *** * Year x Treatment * ns + Site x Treatment ns ns ns Year x Site x Treatment ns ns ns Asterix denotes significance of effects: *** = p <. 0.001, ** = p^0.01, * = p <. 0.05, + = p<; 0.1, ns = not significant. ^ Indicates values were log-transformed for analysis;a Indicates values were rank-transformed for analysis. 30 Figure 2.7: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Dryas ITEX site in 1992 (n=13 for OTCs, n=14 for controls). Error bars indicate 1 standard error. The arrow points to a period of cool overcast weather with an overnight snowstorm on day 197. 0.15 % 0.10 0.05 i 0.00 T 198 204 Day number Figure 2.8: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Cassiope ITEX site in 1992 (n=14). Error bars indicate 1 standard error. The arrow points to a period of cool overcast weather with an overnight snowstorm on day 197. 31 •3 § •a ca oo a o ox) cd > 0.30 0.25 B 0.20 H 0.15 « 0.10h 0.05 H 0.00 Day number Figure 2.9: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Dryas I T E X site in 1993 (n=14). Error bars indicate 1 standard error. •8 a o •a £3 oo a o o 00 ca 4) > < 0.20 0.15H 0.10 0.05 H 0.00 191 195 199 Day number Figure 2.10: A comparison of daily shoot elongation rates of Cassiope tetragona in treatment and control plots at the Cassiope I T E X site in 1993 (n=14). Error bars indicate 1 standard error 32 Dl OTCs Dl Controls • CI OTCs CI Controls TTTTTTTTH fSSSSSSSS fSSSSSSSS. fSSSSSSSSA fSsssssssA fssssssss fssssssss fSSSSSSSS fSSSSSSSS fSSSSSSSS fSSSSSSSS fSSSSSSSS fSSSSSSSS fSSSSSSSS fSSSSSSSS, fSSSSSSSS* fSSSSSSSS. fSSSSSSSS* fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. fSSSSSSSS. \* * * S* / * / r sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss sssssssss tit 1992 1993 Figure 2.11: Maximum shoot elongation measured in treatment and control plots at the Cassiope (=CI) and Dryas ITEX (=DI) sites (n=14, except in 1992 Dryas ITEX treatment means, n=13). Error bars represent 1 standard error. sx u & o a 3 1.00 0 . 7 5 H 0 . 5 0 H 0 . 2 5 0 . 0 0 Cassiope ITEX Dryas ITEX Figure 2.12: Mean flower production per shoot of Cassiope tetragona in treatment and control plots at the Cassiope and Dryas ITEX sites in 1993 (n=12). Error bars represent 1 standard error. 33 combined-site analysis of variance (Table 2.6). Individual ANOVAs performed for each site separately show a significant treatment effect on shoot elongation at Cassiope ITEX but not at Dryas ITEX (Table 2.4, 2.5). The positive effect at Cassiope ITEX was greatest in 1992, with an 11% increase in shoot elongation in treatment plots relative to controls (Figure 2.11). A similar trend is apparent at the Dryas ITEX site in 1992, but not in 1993, where average shoot elongation was generally less than in control plots (Figure 2.11). 2.3.1.3. Response of Cassiope tetragona: reproduction Flower bud development, peduncle elongation, flower corolla opening and corolla drop phenophases occurred significantly earlier in the OTCs than the control plots in 1992 and 1993 at both sites (Table 2.7, 2.8). This effect was strongest in 1993, where timing of the phenophases was offset by up to 13 days (Table 2.7). Timing of fruit maturation did not show a significant treatment response (Table 2.8); however, the sample sizes obtained were too small to develop robust tests of treatment effects. Except for flower bud development, reproductive phenophases generally occurred earlier at Dryas ITEX than Cassiope ITEX in both 1992 and 1993 (Table 2.7). The number of flowers produced per mature shoot in 1993 showed a strong positive response to surface warming at both sites (Figure 2.12), indicating a significant effect of the OTCs on floral bud production in the previous year. The OTCs also had a significant positive effect on the percentage of tagged shoots which developed mature flowers in 1992 and 1993 (Table 2.8). Because flower buds for this species are formed in the year prior to flowering (S0rensen, 1941), positive treatment effects on flower production in 1992 must represent differential development of reproductive primordia. Similar effects in the year following treatment establishment may represent increased production of flower buds in the previous year and/or differential development in the current year. Flower production did not differ significantly between sites in 1992 or 1993 34 5 cd i n < N 00 CI iri ON oo ,—i i c i c i — 9 - H ON s s CN ON <N CO r-l O c i c i c i c i c i D-2 T3 >2 "o Ui o u in cn C N r> r> oo ON V} co ^r' N O od NO r- r-<N v i N O r - N o r - N o - ^ - r - ^ - ' - i O ^ O O O N O O N O ( N C N C 4 ( N - - i ( N ^ C N C O-O cd O u c o i w i c o r - - ' - ^ c O ' - | T < t • ^ o o c M N o i n T t r o o o O N O N O N O N O O O N O O O O "o a 3 •a c _o c o •a CO C s E 0 ( N ^ - i o r o q ^ o q o ^ i n o N O N r ^ O N ' - ' O \ 0 \ o o o o r - o o r - o o T3 3 ,0 1— o E <N NO ^ 00 2 <^  c s S c e c/5 r-oo 00 a cd X w H U c i ON ON O 00 00 00 f - CN CO co o o i —< NOi C N v i co '—< co o oo oo oo oo c-» r- NO r» c o o 00 c cd X W H Q c o o 00 c u CO ON ON P • = C o o 00 c cd UJ H Q 35 T3 CD 00 00 8 c CD CD & T 3 C « K O oo 5 o •c o-<D 00 C 2 i c co CD CO Q 8 _ 8^  O oo T 3 C CO "I S ^ C •*3 '53 o 3 "8 i— D-CD O oo c CD » c o •4—i c <D S CO CO &i 00 CO O CD G .tn CD oo i _ S *~ to PJ H a c? Q c CO <D Cu O '55 00 CO u CD J 3 CD i2 O PJ do CN CD x> cO H oo c 'o 3 "8 CL o o oo oo CO C CO 8 oo T 3 CD 00 00 a 3 l— _cO 2 o U o c 3 OH T3 3 X> u <D O E o 3 CO £ c o -4—1 CO 3 to s o. 2 T J _c0 ~5 o u oo c 'c a c CO oo c _o S O C <u 00 CD oo C •X-•X X 00 C 00 C 00 oo oo C oo C X X 00 X X X 00 C 00 C X X X OO a oo C X 00 C X X X oo C c CD C Si £ H c CD £ •4—» H x l-c s C CD I Si H Si CO 36 (Table 2.8; Figure 2.12), and annual differences in the proportion of tagged shoots producing flowers also were not apparent (Table 2.8). Treatment effects on seed germination were examined using seeds sampled from the Cassiope ITEX site in 1993. Total seed germination of Cassiope ITEX seeds was significantly greater for seeds obtained from OTCs compared to controls (p<0.001, Appendix 12; Figure 2.13). Peak germination also was significantly affected by the OTC manipulation, with peak germination occurring an average of 8 days earlier in the treatment group (p<0.01, Appendix 12). The proportion of seeds in the control population succumbing to algae and fungal infection was approximately 25 times greater than for the treatment population (treatment effect p<0.001, Appendix 12). 2.3.2. Beach ridge snowbed manipulations 2.3.2.1. Environmental factors Snowmelt in 1992 was generally late throughout the Alexandra Fiord lowland. Snowmelt in the Beach Ridge snowbed control plots occurred an average of 5 days earlier in 1993 than in 1992 (Table 2.9). This difference was greater for removal plots (7 days earlier in 1993) and less for addition plots (4 days earlier in 1993). The observed sequence of snowmelt in the experimental plots also differed between 1992 and 1993 (Table 2.9; Figure 2.14, 2.15), probably due to changes in snowdrift patterns. Following manual snow removal, the last 10 cm of snow in the removal plots melted out concurrently within three days in 1992 and within 7 days in 1993. Patterns of snowmelt within the other treatment plots were less uniform (Figure 2.14, 2.15). Within the control plots, natural variations in snowpack depth resulted in some portions of the plots becoming snowfree as much as 4 days earlier than the rest of the plot. The transfer of snow to the addition plots caused the formation of snow mounds, which became 37 Table 2.9: Snowfree dates (day number in calendar year) of the Beach Ridge snow manipulation plots in 1992 and 1993. Year Block Snow removal Control Snow addition 1 172 184 182 1992 2 173 181 187 3 173 184 188 1 164 176 181 1993 2 166 180 182 3 164 178 184 Table 2.10: Relations between the timing of vegetative phenology and snow release for Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed. 1992 1993 Date of vegetative bud break R 2 = 0.8475 bi = 0.18 + 0.03 R 2 = 0.9742 bi = 0.79 ± 0.05 Date of active growth initiation R 2 = 0.6147 b t = 0.35 ±0.10 R 2 = 0.9012 bi = 0.56 ± 0.07 Date of growth cessation n/a R 2 = 0.7162 bi =0.31 ±0.07 Days from snow release to bud break R 2 = 0.9915 bi = -0.82 ± 0.03 R 2 = 0.7305 bi = -0.21 ± 0.05 Days from snow release to active growth initiation R 2 = 0.8508 bi = -0.65 ±0.10 R 2 = 0.8513 bi = -0.44 ± 0.07 Days from snow release to growth cessation n/a R 2 = 0.9257 bi = -0.69 ± 0.07 Days from bud break to growth cessation n/a R 2 = 0.7784 bi = -0.48 ±0.10 Days from growth initiation to growth cessation n/a R 2 = 0.3896 • hi =-0.25 ±0.12 Asterix denotes significance of the regression equation: *** = p <. 0.001, **=ps 0.01, *=ps 0.05, ns = not significant. R 2 is the coefficient of determination obtained for the regression between snow release date and the dependent variable; bi is the slope of the regression line, ± 1 standard error. 39 II oi CN O N ON CO >» ITS Q CN ON ON O 00 Q CN ON ON Q VU 40 c C •o CU CO a X CO c o B <u 00 > T3 CU co a X <u CU > o o o c CO M 8 CQ <N O o s CQ co ON ON ON VO >> Q CO ON ON co ON ON 00 r--Q co ON ON a" Q s c o o u SO 8} T J O a CO 00 S\ T J ON 2 rt„ xi o cn CQ & (U T J JS c ~ o - 1 O ca u| c o m > ca u 3 £ OH CO "!•! £ "3 c c co c j s s cu o £ c "I II oo< i i CU u 3 00 (U 41 exaggerated as snowmelt progressed. Because of this, the central areas of the addition plots generally melted out 1-5 days later than the edge portions. 2.3.2.2. Response of Cassiope tetragona: vegetative phenology and growth The timing of vegetative phenophases in the Beach Ridge snow manipulation plots were strongly affected by the timing of snow release (Table 2.10). Dates of bud break, initiation of active growth and growth cessation were positively related to snowmelt date (Table 2.10), indicating that the timing of snow release had a direct effect on phenology timing. The strength of this direct effect may have been partially counteracted by negative effects on rates of phenological development, as plants in early snowmelt plots required more days following snow release to reach bud break, active shoot elongation and growth cessation phenophases than plants in later snowmelt plots (Table 2.10). Based on a comparison of regression line slopes, it appears that the direct effects of snowmelt timing on vegetative phenology were greatest in 1993, and most strongly affected bud break timing. Similarly, snowmelt timing appears to have had a stronger effect on rates of phenological development in 1992. Although vegetative bud break and the start of shoot elongation showed strong correlations with snowmelt timing, the total number of days used by Cassiope tetragona for active growth was only moderately related to snowmelt (Table 2.10). The total number of days from bud break to growth cessation decreased for plots with later snowmelt (Figure 2.16), but the total number of days from growth initiation to growth cessation was not strongly affected (Figure 2.17). It appears that the impact of snowmelt timing on growth initiation was counterbalanced by symmetrical effects on growth cessation, resulting in little net change in the total length of the period utilized by C. tetragona for active growth. Timing of snow release did not significantly affect maximum shoot elongation at the Beach Ridge site. Mean values of shoot elongation in the snow manipulation control plots 42 170 175 180 185 190 195 200 205 210 215 220 225 230 Day number Figure 2.16: Effective growing season length for Cassiope tetragona measured as the time from mean date of vegetative bud break to mean date of growth cessation in Beach Ridge snow manipulation plots in 1993. Error bars are -1 and +1 standard error for mean date of bud break and growth cessation, respectively. removal 1 I — B • — 1 removal 2 ——1 — • removal 3 g 1 • control 1 • • — — control 2 1 • — • 1 control 3 1 • > — i addition 1 1 •-• • 1 addition 2 — • • addition 3 1 B 1 • — i | I l 85 1 190 1 — 'l95 1 200 1 1 205 210 215 220 225 230 Day number Figure 2.17: The period of active growth of Cassiope tetragona measured as the time from mean date of growth initiation to mean date of growth cessation in Beach Ridge snow manipulation plots in 1993. Error bars are -1 and +1 standard error for mean date of growth initiation and growth cessation, respectively. 43 were significanly greater in 1993 than 1992 (p<0.05, Appendix 14), however, indicating that this parameter is sensitive to annual changes in growing season conditions. 2.3.2.3. Response of Cassiope tetragona: reproduction Rower production in the Beach Ridge snowbed community is generally low compared to other communities at Alexandra Fiord (Nams, 1982; also see Chapter 3). The number of tagged shoots producing mature flowers within the snow manipulation plots was generally lower in 1993 than in 1992, reflecting the growing season conditions of the previous year (Table 2.11, 2.12). Estimates of flower production per shoot made in 1993 show no relationship to snowmelt date in either the current or previous years (Appendix 16). The timing of reproductive phenophases in 1992 and 1993 was positively related to snowmelt timing in the Beach Ridge plots (Table 2.13). This response was strongest for flower bud emergence and progressively weaker for later phenophases of peduncle elongation and corolla opening. Timing of snowmelt had no effect on the mean number of days from snow release to reproductive phenophases of flower bud emergence, peduncle elongation and corolla opening (Table 2.13). In 1992, only two of the monitored shoots produced flowers which reached the corolla drop phenophase (Table 2.11). In contrast, most of the shoots producing mature flowers in 1993 reached corolla drop (Table 2.12). None of the tagged shoots produced mature fruits in 1992 or 1993, and no mature fruits were observed on un-tagged shoots in either year. 44 <N TJ X ) o e CO CL, cu o E da T 3 2 CO o u C c^ CJ E If c °->. 8H x. 8 o u c 2 t«. o/j| c _o CL) _ C D O c 3 T J CL, o c cu £ T J 3 X ) 8 CQ c E a T J oo O O O - H O O — i o o I I I I I o -sr i i i CN i i CN C l •<t r> co ~+ co o I o o d d C l CO vd co in vo o CN vd o - ^ - ^ i n v o o o i n C N C N C N C N C N C N C N C N ^ r ^ r ^ r r - i n c N ' t ^ r o co in >n co o t oo d co O ON O d vd d - M ^ O N r - v p o ^ - H r - -- - I O O O ^ ^ O O C N C N C N C N C N C N C N C N ^ • ^ t ^ t r ^ i n c N i n i n o i n i n i n r - o c N o o o o C N C N C N ' i n v d ^ C O C O ON ON ON CO h h ON ON 00 00 ^ c N c o ^ c N r o ^ H C N r o c o c o T J T J < 13 > 0 E <u 01 45 CO o c X3 3 o E O O C N O ^ C O O C N - — i Q. e -o I o u -a GO in CO CI in CO in >n CN —i Cl Ol 00 c 'S <u a o 2 o u GO C s o d oo ON ON CN C o cd oo c _o o c 3 •O O O CO O o d o d ' Oi o o c 00 T3 GO c s E CQ c E CO oi co co c i oi c o U a o -a 13 > o E Pi «n C O NO CO o o CI CN O O C O O ^ H C O O C N O I 00 ON 01 ON co *n oo —< O O i 00 o O) OJ i—i Ol CO O CN Ol oo co oo go ^ o o r - H - - H C O O C O C O O C N C O 1> o in NO NO t m m —< oo NO 00 O 00 ' ON ON I r- 00 C N C O ^ - I O I C O ^ C N C O 46 Table 2.13: Relations between date of snow release and timing of reproductive phenology of Cassiope tetragona in snow manipulation plots at the Beach Ridge snowbed. Date of flower bud emergence R 2 = 0.7965 bi = 0.55 ± 0.11 Date of flower peduncle elongation R 2 = 0.5872 bi =0.55 ±0.19 Date of corolla opening R 2 = 0.4930 bi =0.41 ±0.17 Days from snow release to flower bud emergence R 2 = 0.0850 Days from snow release to flower peduncle elongation R 2 = 0.0693 Days from snow release to corolla opening R 2 = 0.1519. Asterix denotes significance of the regression equation: *** = p <, 0.001, **=ps 0.01, * = p <, 0.05, ns = not significant. R 2 is the coefficient of determination obtained for the regression between snow release date and the dependent variable; bj is the slope of the regression line, ± 1 standard error. 47 2.4. Discussion 2.4.1. Experimental manipulations Environmental manipulation experiments carried out in the field often have difficulties in achieving the desired treatment effect without introducing unwanted, secondary effects. The open-top chamber design used in this study was chosen to simulate the effects of moderate increases in average surface temperatures (1-2 °C) while maintaining close-to-ambient levels of humidity, precipitation, CO 2 concentrations, and light quality. Evaluation of the effects of these chambers at Alexandra Fiord suggests that the OTCs warm near-surface soil temperatures by an average of 1.3 °C during the growing season (Marion et al., 1995). Because the warming effects vary diurnally, with occasional night-time cooling in the chambers (Marion et al., 1995), plants in the experimental plots experience an increased range of temperatures as well as warmer average temperatures during the growing season. It is also likely that there are strong effects on temperature and light quality in the edge areas of the OTCs; however, as plants were monitored only in the center regions of the experimental plots, these edge effects should not have played a substantial role in determining the observed plant response. Snowpack measurements taken in the beginning of June do not indicate that the chambers affected net snow accumulation during the winter. The moderate delay of snow release in treatment plots compared to controls appears to be the result of small mounds of snow persisting in the center of the OTCs 1-2 days after the surrounding snow had melted (personal observation). This persistence may be due to shading and decreased wind in the open-top chambers resulting in slower ablation of snow in the center of the plots. The open-top chambers had a significant positive effect on maximum active layer depth at one site, but not at the other. Increased active layer depth is likely to have occurred as a direct effect of increased surface temperatures. The strong environmental similarities 48 between the two sites and their close proximity suggests that the direct environmental response to the OTCs is likely to be consistent between the sites, but net environmental effects may have been influenced by characteristics of the vegetation. The significant negative effect of the OTCs on soil moisture at the Dryas ITEX site in midsummer of 1993 may have resulted from a greater potential of that vegetative community to increase evapotranspiration in response to warm temperatures and high levels of solar radiation, as compared to the Cassiope-domimted vegetation at the Cassiope ITEX site. Differential environmental responses to the open-top chambers at the two sites indicate that habitat type may influence the magnitude of environmental effects caused by the OTC manipulation. Although many observational studies have focused on the effects of snowmelt timing on arctic and alpine vegetation (e.g. Billings and Bliss, 1959; Holway and Ward, 1963; Miller, 1982; Inouye and McGuire, 1991; Kudo, 1991), few experimental studies of snowmelt timing have been carried out in tundra environments (but see Benedict, 1990; Galen and Stanton, 1993). The manual snow manipulation technique used in this study has been previously used in a study of alpine vegetation (Galen and Stanton, 1993) and has some advantages over alternative techniques for snowmelt experiments such as the use of permanent snow fences or the transplantation of plant subjects (e.g. Benedict, 1990). Manual snow manipulation does not introduce secondary effects common to snow fences such as shading or wind protection during the snowfree period and also does not alter winter conditions, allowing the manipulations to represent only the effects of changing snowmelt timing, rather than changing snow accumulation patterns. With this techniques, plants are manipulated within their normal environment, and do not need to be disturbed beyond the experimental manipulation. In addition, manual snow manipulation is inexpensive and relatively easy to establish. However, use of this technique does present some problems. One important difficulty is that plots must be relatively small in order to provide adequate replication within a community and efficiently manipulate snow levels. At the same time, use of a small plot size may result in large proportions of the plots being 49 compromised by edge effects. Edge effects in the snow removal plots at the Beach Ridge snowbed were manifested as a slower rate of thaw in the parts of the plot adjacent to the surrounding snowbed and drainage of snowmelt water into the plot, which may also reduce soil temperatures. Snow addition treatments exhibited important edge effects, becoming snowfree at the plot borders earlier than in the center. Manual snow removal may also have negative effects on plants within snow removal plots, as plants may be compacted and damaged beneath the thin layer of snow during treatment establishment. Such negative effects are likely to be particularly important for long term studies. In general, the snow manipulations were successful in altering the date of snowmelt in the Beach Ridge snowbed community. The magnitude of the treatment effect differed among the treatment blocks, with plots located in areas of deeper snow showing the strongest differences in snowmelt date between snow removal and addition plots relative to controls. If this experiment were to be repeated, selection of a larger area for snowmelt manipulations is recommended in order to provide greater replication of treatment levels. In addition, recording the snow release date for individual monitored plants would allow for some calibration of edge effects within a plot. 4.2. Response of Cassiope tetragona Responses of the timing of vegetative phenophases to surface warming in the OTCs indicates that increased temperatures may cause a slight shift in the period of active growth for Cassiope tetragona, with growth initiation occurring earlier after snowmelt in warmed environments. In the OTCs, this advance in growth initiation was counter-balanced by similar effects on growth cessation, such that the total period of active growth was only moderately affected by the OTCs. Shoot elongation rates at the Dryas and Cassiope ITEX sites show general trends of more rapid growth in the warming plots early in the season, followed by a switch to more rapid growth in the control plots at the end of the growing 50 season. The shift in phenophase timing, without strongly affecting the net length of the growth period, may explain the conservative response of total shoot elongation to the effects of the OTCs. Treatment effects on total shoot elongation in 1992 and 1993 appear to be related to changes in peak elongation rates within the OTCs, rather than effects on the length of the period during which plants were observed to be growing. A shift in the timing of the growth period was also observed between years in the ITEX control plots. At Alexandra Fiord, 1992 was a relatively cool summer characterized by late snowmelt, while 1993 exhibited warm growing season temperatures with early snowmelt. Observations of Cassiope shoot elongation show that both growth initiation and cessation occurred substantially earlier in control plots in 1993 than in 1992. Despite this large difference, the total period of active growth changed only slightly between the two years, and was accompanied by little to no difference in annual growth between the years. Other studies of Cassiope tetragona have found that, although timing of vegetative phenophases shift in response to changes in growing season conditions, the total period of active growth remains relatively stable (Svoboda, 1977; Nams and Freedman, 1987b). This inflexibility in the growth period may serve to stabilize annual variability in the productivity of this species within a community (e.g. S0rensen, 1941; Shaver and Kummerow, 1992). Manipulations of snowmelt timing at the Beach Ridge snowbed also had symmetric effects on the timing of different vegetative phenolophases. Cassiope plants growing in early snowmelt plots initiated growth earlier and generally stopped growing earlier, relative to late snowmelt plots, and there was little change in the total period of active growth associated with the snowmelt treatments. The symmetrical effect of snow release date on timing of both growth initiation and cessation provides additional support for the hypothesis that the total growth period utilized by Cassiope tetragona is relatively insensitive to short-term variations in growing season climate, such as variations in the timing of snow release. Likewise, observations of total shoot elongation showed little 51 response to snowmelt manipulations, in contrast to the strong effects of the manipulations on the timing of individual vegetative phenophases. Although the timing of vegetative phenophases was positively related to the timing of snowmelt at the Beach Ridge community, the number of days from snow release to individual phenophases increased with earlier snowmelt. Early snow release plots at the Beach Ridge snowbed experienced cooler air and soil temperatures, and wetter soil conditions immediately following snowmelt than did late snow release plots. Pre-growth light levels are also likely to have changed with snowmelt timing, due to the rapid seasonal changes in solar radiation input at the latitude of the study site. These differences in environmental conditions may have stalled vegetative development in early melting plots. Phenological patterns observed over two years in the Beach Ridge snow manipulation plots indicate that the effects of earlier snowmelt on vegetative production of Cassiope in snowbed communities is likely to be dependent on early season environmental conditions in addition to the actual timing of snow release. Other studies of tundra vegetation have also found that environmental conditions following snowmelt play an important role in determining the timing of plant growth (Inouye and McGuire, 1991; Larigauderie and Kummerow, 1991). The timing of Cassiope reproductive phenology was highly responsive to both temperature and snowmelt manipulations. Reproductive phenology was accelerated in the open-top chambers, indicating some degree of temperature dependence in the time period required for flower and fruit development. The amount of acceleration between treatment and control plots was greatest at the Cassiope ITEX site, which suggests that site characteristics may influence the sensitivity of reproductive development to the level of warming induced by the OTCs. Snowmelt manipulations also altered the timing of reproductive phenology, but did not change the total number of days required from snow release to development of individual phenophases. This contrasts with patterns of vegetative development in the 52 Beach Ridge snowbed, which occurred more slowly in early-melting plots. The production of Cassiope flowers in the snow manipulation plots also showed no relationship to snowmelt timing in either the current or previous year. In general, it appears that reproductive development and production of Cassiope tetragona is not strongly affected by snowmelt timing except through direct effects of snowmelt on the date at which reproductive development may begin. However, flower production at the Beach Ridge snowbed site is generally low compared with other Cassiope- dominated communities in the Alexandra Fiord lowland (Chapter 3), and patterns of flowering may be less responsive to snowmelt timing than in other, more productive, communities. Flower production at the ITEX sites exhibited a strong positive response to the changed environmental conditions within the OTCs. Significant treatment effects on mature flower production in 1992 indicate that the OTCs may have positively affected the maturation of pre-developed flower buds. The large increase in mature flowers produced in the treatment plots in 1993 is likely to reflect both increased production of floral buds in the previous year and successful reproductive development in the current year. The quality of seeds produced was also significantly improved in the warming chambers relative to controls in the year following treatment establishment at the Cassiope ITEX site. Large differences in seed viability, germination rate and vulnerability to fungal and algal infestation between warmed and control populations indicate that sexual reproduction of C. tetragona is strongly constrained by low temperatures during seed production. Observed improvements in seed quality may be due to a longer period for seed maturation resulting from accelerated reproductive development in the OTCs and/or increased allocation of within-plant resources to seed production. Increases in the production of viable seed in the OTCs may also be related to effects on pollination of Cassiope flowers (primarily insect pollination), which was not investigated in this study. The strong reproductive response exhibited by Cassiope tetragona to increased surface temperatures in the open-top chambers suggests that this species is adapted to 53 capitalize on occasional periods of ameliorated growing season conditions for successful sexual reproduction. Increases in Cassiope flower production in years following favorable growing seasons have been noted by other authors, while growth remains relatively stable (Bliss et al., 1977; Nams and Freedman, 1987). Flower production of this species is also stimulated by moderate levels of fertilization, again with little observed growth response (Henry et al., 1986). In general, it appears that reproductive development and production of Cassiope tetragona is significantly more responsive to short-term variations in environmental conditions, such as growing season temperatures and soil nutrient levels, than vegetative development and production. Investigations using temperature manipulations at a high arctic site on Svalbard found similar patterns in the vegetative and reproductive response of Dryas octopetala (Welker et al, 1993; Wookey et al, 1993), indicating that this may be a common response of high arctic evergreen and semi-evergreen shrubs to periods of summer warming. Differential reproductive and vegetative responses to environmental variation may reflect specific resource allocation strategies within a plant, where 'surplus' resources are allocated preferentially to reproductive output, rather than a unform allocation to all the different components of plant production. Preferential allocation strategies and/or limitation of the physiological processes of growth and reproduction by different factors have important implications for the manner in which plants are likely to respond to short-term changes in climate. Strong reproductive responses of species such as Cassiope and Dryas to climate warming could greatly increase seed production, possibly resulting in increased colonization of bare areas and an expansion or shift in plant distributions (Molau, 1993; Wookey et al, 1993). Variations in snowmelt do not appear to strongly affect either vegetative growth or reproduction of C. tetragona in isolation from other environmental factors. Changes in snowmelt may alter the timing of growth initiation, but do not have a strong impact on the total portion of the growing season utilized by the plants for growth. Snowmelt variations 54 also do not appear to have a important impacts on either reproductive output or development when isolated from concurrent changes in growing season climate. This study has examined short-term responses to isolated experimental manipulations. If productivity of this species is limited by multiple environmental factors, it is unlikely that the short-term responses observed here would be representative of long-term responses that could occur as a result of climate change in the Arctic (Chapin et al, 1987; Shaver and Kummerow, 1992). Changes in growing season climate are likely to affect a wide variety of factors such as soil moisture, nutrient availability and cycling and between-plant interactions within a community (Chapin, 1984; Maxwell, 1992). Changes in any of these factors can be expected to affect the long-term response of Cassiope tetragona to increased temperatures or changes in snowmelt timing. Further study is required to elucidate interactions among these factors and growth and productivity of long-lived tundra species such as Cassiope tetragona. 55 3. Natural patterns of variation in growth and reproduction of Cassiope tetragona observed at Alexandra Fiord 3.1. Introduction Although field manipulations and experiments are necessary to investigate the effects on organisms of different factors in isolation or in conjunction with other factors, descriptive studies of biological systems are essential in generating hypothesis about biological functions and mechanisms. Observations of natural biological patterns are also useful in interpreting experimental responses and in determining whether such responses are realistic. Comparison with natural patterns of variation in vegetative and environmental characteristics may confirm or refute experimental responses, strengthen conclusions and help generate new hypotheses. This study was designed to provide observations of Cassiope growth and reproduction within both experimental and natural variations in environmental conditions related to climate change. The results of experimental observations were presented in the previous chapter. This section presents some of the observations made in unmanipulated environments within the study area, with the aim that examination of natural patterns of variation may aid in the interpretation of short-term experimental responses and stimulate additional hypotheses about factors controlling growth and reproduction of Cassiope tetragona. 56 3.2. Methods 3.2.1. Study sites The study area for this research was locate at Alexandra Fiord, Ellesmere Island (78°53'N, 75°55'W), and is described in Chapter 2. Field observations were made within 9 plant communities of the Alexandra Fiord lowland. These sites were selected from among the Cassiope-dominated areas of the lowland to represent a wide range of community types and microenvironmental conditions. Of the nine communities examined in this study, six have been previously studied (Nams and Freedman, 1987a,b). This earlier work documents a number of environmental measurements as well as observations of community composition and patterns of growth, phenology, and resource allocation of Cassiope tetragona. Three of the communities examined in this study were also used as sites for experimental manipulations of surface temperature and snowmelt timing, described in Chapter 2. Observations made in control plots at the experimental sites are used in comparison with unmanipulated sites. The two communities used for temperature manipulations are located in adjacent areas of the central plain of the lowland (Figure 3.1). One of the communities is dominated by Cassiope tetragona and the other by Dryas integrifolia. Following the nomenclature used in the previous chapter, they will be referred to in this discussion as the Cassiope ITEX and Dryas ITEX sites, respectively. The third community, used for snowmelt manipulations, is located in a late-lying snowbed area in the lee of a beach ridge (Figure 3.1). This community will be referred to as the Beach Ridge snowbed. A more complete description of these communities may be found in Chapter 2. C. tetragona was also intensively monitored in five communities along a river bank slope chosen to represent natural gradients in snowmelt, soil moisture, and depth of active layer thaw. The site is located on an east-facing slope (5-13°) on the banks of the principle 57 Figure 3.1: Map of the Alexandra Fiord lowland, showing locations of study communities at the River Slope (RS), Erratic (ER), Cassiope ITEX site (CI), Dryas ITEX site (Dl) and the Beach Ridge snowbed (BR). The map is based on Nams, 1982 and revised using 1993 air photos. 58 river in the lowland (RS=River Slope; Figure 3.1). This site has been previously investigated by Nams and Freedman (1987a,b), who divided it into five zones based on changes in plant community composition (Figure 3.2). These community zones appear to correspond to changes in microenvironmental conditions, such as snowmelt regime, ground temperature, and soil moisture, as well as variations in Cassiope growth (Nams and Freedman, 1987a). The communities are referred to in the text as RSI - RS5, in order of their position from the top to the bottom of the slope (Figure 3.2). Snowmelt progression along the slope generally progresses from the top and bottom in towards the middle of the slope. Soil moisture levels increase from mesic/xeric at the top of the slope to mesic/hydric at the bottom, with generally warmer temperatures at the top and cooler temperatures at the bottom of the slope (Nams and Freedman, 1987a). An additional study site was established on June 5, 1993 at the head of the lowland near several large glacial erratics (ER=Erratic Site; Figure 3.1). The community is located on gently sloping (2-3°, eastern aspect) ground composed of sandy alluvium with little organic content. Community composition is almost entirely C. tetragona with approximately 85% vascular plant cover. In this area, snowmelt is early and soils generally mesic, although occasionally the area is flooded by a nearby glacial meltwater stream. 3.2.2. Measurement of environmental variables The field work for this project was conducted from June 14 to August 13, 1992 and from June 3 to August 22, 1993. Measurements of selected environmental variables were made during the 1993 growing season to characterize the physical differences among the communities studied. The following variables were measured: snowpack depth, date of snowmelt, active layer depth, soil moisture and ground surface temperature. Snowmelt date was recorded as the date when an area became completely free of snow. Prior to snowmelt, snowpack depth was measured at each site using a calibrated 59 Figure 3.2: Diagram of community zonation and study plots at the River Slope site. Revised from Nams, 1982. 60 metal rod. Snowpack depth was systematically sampled along permanent transects at the River Slope, Erratic and Beach Ridge sites (Table 3.1). At the ITEX sites, depth of snow was recorded for each plot based on 5 samples from the north, east, south, west and center areas of each plot. Snow depth was sampled at 2-3 day intervals from June 4 to snowmelt at each site. Following snowmelt, active layer depth was sampled every 2 days at the ITEX sites and every 4 days at the other sites until 7-10 days after snowmelt. Additional active layer measurements were taken on 25 July at eight of the sites. Depth of thaw was measured as the depth of penetration of a metal rod into the soil. Soil thaw was not measured at the Beach Ridge site because the high gravel content of the soil frequently inhibited penetration of the soil with the permafrost probe. Soil moisture was measured every 10-12 days at all sites from 24 June to 14 August, 1993. Moisture levels were calculated as percentage of dry weight based on fresh and dry weights of 5-8 soil cores for each site (Equation 2.1). The soil cores were taken from the upper 6 cm of soil and vegetative matter above the ground surface was excluded from the core. Sampling of soil cores was done using a haphazard technique; at the ITEX sites, samples were taken only from the 4 destructive sampling plots. Near-surface temperatures were measured at four sites using 1-4 thermocouples buried in the soil to a depth of -3 cm and 1-2 thermocouples placed 5 cm above the soil surface. Thermocouples above the surface were shielded from solar radiation using ventilated aluminum foil tents. Temperature measurements were manually recorded every 4-5 days at each site within a 2 hour time period (usually 11:00 - 13:00) using a hand-held thermocouple reader (microvolt meter). Any changes in weather conditions during this time period were noted. 61 3.2.3. Measurement of growth and reproductive output Measurements of shoot elongation and phenology were recorded for a minimum of 50 tagged Cassiope shoots within each site or treatment (Table 3.1). These measurements were made at intervals of 6-7 days in 1992 and 4-5 days in 1993. Where individual clones of Cassiope were distinguishable, only one shoot was tagged from a single clone, in order to maintain a minimum degree of genetic and structural independence among the samples. In areas were clones were not easily distinguished, a minimum distance (~20 cm) between tagged shoots was used to prevent multiple shoots being monitored from a single clone. Exceptions to this procedure occurred at the Cassiope and Dryas ITEX sites, where the small size of the plots occasionally required tagging more than one shoot per clone in order to achieve the desired within-plot sample size. Phenological stages were recorded as the date at which that stage was first observed for each shoot. The phenological stages measured in this study were vegetative bud break, growth initiation and growth cessation. The definitions of these terms as used in this study are given in Table 2.1 of the previous chapter. Quantitative measurements of shoot elongation were obtained at the same time as phenological data were gathered. At the beginning of the growing season, the upper-most leaves of each tagged shoot were marked with small dabs of water-insoluble liquid paper in order to delineate the border between the current year's and previous year's growth. Shoot elongation was then measured as the distance from the marked leaves to the tip of the shoot. Measurements were made to the nearest 0.05 mm using a hand-held caliper. In 1993, counts of live shoots, dead shoots and reproductive structures (flowers or fruits) were made within 10-40 randomly placed 10 cm 2 quadrats at each site. Only mature shoots greater than 15 mm in total length were counted. Sampling was done at or immediately following peak flowering times (Table 3.1). Using these measurements, 62 Table 3.1: Sample size and sampling design for environmental, growth and reproductive variables measured at each site in 1993. CassiOpe Dryas Beach Ridge River Slope Erratic ITEX ITEX snowbed communities site Snowpack 5 /plot 5 /plot 9 (ambient approx. 12/zone 15 stratified- stratified- conditions) systematic haphazard haphazard systematic systematic Active layer 5 /plot stratified-haphazard 5 /plot stratified-haphazard approx. 12/zone systematic 15 systematic Soil moisture 2x4 plots haphazard 2x4 plots haphazard 8 (ambient conditions) systematic 5 /zone systematic 5 systematic Near-surface 2 air + 3 soil 1 air + 1 soil 2 air + 4 soil temperture stratified (RSI, RS5) stratified stratified Shoot 5/plot 5/plot 15-20/plot 60/plot 55 elongation stratified-random stratified-haphazard haphazard haphazard random Phenology 5/plot 5/plot 15-20/plot 60/plot stratified-random 55 stratified-haphazard haphazard haphazard random Flower 4/plot 4/plot 12/plot 20/zone 20 density (40 in Zone 1) random random random random random 9-11 July 11-12 July 16-22 July 7-15 July 2 July 63 estimates of live shoot density, flower production per mature shoot, and dead:live shoot ratios were calculated for each community. 3.2.4. Data analysis A one-way analysis of variance (ANOVA) was used to test for significant differences between communities for those parameters measured only in one year. Year and community differences were tested using a two-way ANOVA for parameters measured in both 1992 and 1993 (GLM procedure, SAS, 1985). A repeated measures ANOVA was used to analyze changes in soil moisture levels among sites across four sampling dates (GLM procedure, SAS, 1985). In these tests, samples within a community were treated as replicates, except for measurements obtained from the Cassiope and Dryas ITEX sites, where control plot means were used as replicates. Total shoot elongation values were not normally distributed and analysis was performed on login-transformed values of this parameter. Summary tables of the analyses are given in Appendicies 17-22. A regression model was developed to predict login-transformed values of total shoot elongation from measurements of vegetative phenology. Variables used in the model were selected using forward stepwise regression (REG procedure, SAS, 1985), with a significance level of a=0.05 required for the inclusion of variables in the model. Data means and standard errors were plotted across sites and years to facilitate comparisons. Patterns of Cassiope production among the communities were compared with environmental variables in order to generate hypotheses of plant response to different environmental conditions. Multivariate analysis was not used for these comparisons, because it was felt that the small number environmental parameters measured did not fully represent the range of environmental conditions characterizing the different sites, and replication of different community types appeared insufficient for a robust multivariate analysis. 64 3.3. Results 3.3.1. Environmental measurements Snow release occurred from 5 to 13 days earlier in 1993 than in 1992 among the communities examined in this study (Table 3.2). This difference is representative of the strong contrast in growing season climate between the two years at Alexandra Fiord. 1992 was an unusually cool and overcast summer, while 1993 was comparatively warm and sunny. The progression of snowmelt also differed slightly between the two years in the River Slope communities (Table 3.2), probably as a result of changing snow drift patterns. The upper and lower zones of the River Slope (RSI and RS5) had become completely free of snow before the beginning of the field seasons in both 1992 and 1993. Active layer depths measured in late July of 1993 are only moderately related to snowmelt timing within the eight communities sampled. Average depth of thaw for sites with early snowmelt ranged from 55 cm at the Dryas ITEX site to 28 cm at the Erratic site (Figure 3.3). RS3, the community with the latest snowmelt in 1993, showed a depth of thaw only 5 cm shallower than that observed at the Erratic site. The weak relationship between snowmelt timing and active layer depth suggests that other factors, such as soil temperatures, differ among these communities. Soil moisture levels measured five times from mid-June to late August in 1993 consistently show the highest amounts of soil moisture at RS5, located at the bottom of the river bank slope (Figure 3.4). This is probably the result of snow and active layer meltwater draining down slope and collecting in the relatively flat and peaty soil of the bottom zone. Soil moisture levels in the River Slope communities consistently decreased with height on the slope, with the upper-most zone remaining dry throughout the summer. Within a community, moisture levels generally decreased over the summer, as snowmelt sources dried up and the active layer deepened. This is most apparent in sites 65 Table 3.2: Timing of snowmelt (day number in calendar year) in nine Cassiope-dominated communities at Alexandra Fiord in 1992 and 1993. Site 1992 1993 River Slope 1 168 155 River Slope 2 175 170 River Slope 3 182 176 River Slope 4 177 161 River Slope 5 168 155 Erratic site n/a 155 Beach Ridge snowbed 183 178 Cassiope ITEX n/a 164 Dryas ITEX n/a 161 Table 3.3: Effects of site and year on vegetative phenology and growth of Cassiope tetragona in five communities at the River Slope site, Alexandra Fiord, in 1992 and 1993. Source of variation year site year x site Date of bud break *** *** ** Date of growth initiation *** *** * Date of growth cessation *** *** ns Days from bud break to *** *** ns growth cessation Days from growth initiation * *** n s to growth cessation Maximum shoot ns *** ns elongation^ Asterix denotes significance of effects: *** = p <, 0.001, ** = p £ 0.01, * = p £ 0.05, ns = not significant. ^ Indicates values were login-transformed for analysis. 66 60 50A 40-1 30 H 20 4 104 0 ~RS4 CI i r Dl Erratic RSI RS2 RS3 RS5 Figure 3.3: Active layer depths observed at the end of July, 1993 in eight communities at Alexandra Fiord. See text for key to site abbreviations. 67 . 0 0 t >o OO CO 00 OS 05 • • 0 3 «—i UJ u El 00 CO 01 oi E 0 s / s s s s s s / s / s s s s s s s s s s s s s s s . H s s s s s s s / s s s s s s / s * * * s * t~x-,.j—<:—*:—^—t t * J J: B/U 1 . ' S . N . X ' S . N S . N . X N ' V . ' S . N . X s s s s s s s s s s s s s s ^ ^ ^ >• ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ o o CN ~r~ 5? O o o in vo <N 00 00 & a c >> cd Q c a •—i & _cd e p CO 3 2 0 E e -a a CO < •4—» cd co .2 1 s S.o 81 -a o & ^ •2 fo u . s l .S w CO s.s 3 -.2 3 O 00 £ 3 -< 'S >i oo 13 co < 3 oo E (iq3l9A\ Ajp JO 1U33J3CJ) ajnisioiu jios 68 that were mesic/hydric at the beginning of the summer. Sites that were already dry in June generally maintained similar amounts of soil moisture over the summer, while wetter sites lost from 50 to 80% of their early-season moisture (date*site effect p<0.01, Appendix 18; Figure 3.4). Soil moisture levels did not decline in the Beach Ridge snowbed over the measurement period. The prolonged moisture at this site may be due to the long duration of meltwater input associated with deep snow accumulation and possibly an ice-rich active layer, combined with poor drainage of meltwater from the community. Near-surface (+5 cm) air temperature measurements made at 4 sites from late June to mid-August generally show similar temperatures among the sites (Figure 3.5). Air temperatures observed at RSI tended to be higher than at the other sites, but because the measurements were poorly replicated it is not possible to determine whether this is a significant difference. Shallow soil temperatures (-3 cm) show greater differences among the four sites, with temperatures at RSI consistently higher than those observed in the other communities (Figure 3.6). As with the air temperatures, however, there was insufficient replication in the measurements to test the significance of this difference. The general pattern of warmer near-surface soil temperatures in RSI compared to RS5 agrees with similar temperature differences reported for these sites by Nams and Freedman (1987a). 3.3.2. Cassiope growth and reproduction The timing of vegetative phenology observed among the communities in 1993 appears to be related to general patterns in the time of snowmelt. Bud break in the different communities occurred sequentially according to the order of snow release (Figure 3.7). Timing of the start of shoot elongation is less directly related to the date of snow release, but still shows a general pattern of earlier growth initiation in early-melting communities compared to late-melting communities (Figure 3.8). Observations of growth cessation 69 • RS1 (n=1) • RS5 (n=1) • Erratic (n=2) O BR (n=2) 20.0 T 4.0 6/28 7/2 7/6 7/7 7/10 7/14 7/18 7/23 8/2 8/8 8/14 Figure 3.5: Near-surface air temperatures (+5 cm) measured over the course of the 1993 growing season in four communities at Alexandra Fiord. See text for key to site abbreviations. • RS1 (n=1) • RS5 (n=1) • Erratic (n=4) © BR (n=2) 17.0 r 6/28 7/2 7/6 7/7 7/10 7/14 7/18 7/23 8/2 8/8 8/14 Figure 3.6: Shallow soil temperatures (-3cm) measured over the course of the 1993 growing season in four communities at Alexandra Fiord. See text for key to site abbreviations. 70 71 & CO . 3 <u o 00 UL, CO U, <D (0 > co (5 co ro CO . . <u cj 5 c 5 < a 0 0 O . T 3 2£ S .Six; c .t; oo o c 'C '-3 3 03 e -a S " S o n x: cj ° £ GJ 2 £ c c o S '5 ob G i—i T 3 ,0 T 3 C <—' CL) 03 u = s i B:|l g c o £ 8 ac > co 2 §J £) 00-3 ^ 8 . CO CO V J 00 <L) • — ^2.2; 55 X) <u T 3 CO X) ooS JJ cO c <u CJ *-> 3 £ 03 ' S 4 3 CJ co Q co Ja t " cu a >> M CD — x; 72 indicate that plants in later-melting communities extend the growth period later into the end of the summer (Figure 3.8). The total length of the period utilized by Cassiope tetragona for active growth (growth initiation to growth cessation) was longest at RSI and Dryas ITEX in 1993 (Figure 3.8), the two sites which also exhibited the greatest average total shoot elongation in that year (Figure 3.9). Regression of log values of total shoot elongation against vegetative phenology parameters found that a large portion of variability in shoot elongation is related to the length of the active growth period (Figure 3.10; p<0.001). A comparison of vegetative phenology in the River Slope communities between 1992 and 1993 indicates that Cassiope plants shift timing of growth according to annual variations in growing season conditions. Shoot elongation started and ended earlier in 1993 than in 1992 in each of the River Slope communities, although the magnitude of this effect differed among communities (Table 3.3; Figure 3.11). Symmetrical changes in both the timing of growth initiation and growth cessation between the two years resulted in only an average of a 1.7 day difference in the total length of the growth period between the two years (Table 3.3). Log values of total shoot elongation at the River Slope differ only marginally between 1992 and 1993 (Table 3.3). Primary productivity of Cassiope tetragona, as measured by annual shoot elongation, appears to be greatest at the RSI and Dryas ITEX sites (Figure 3.9). Flower production in these communities is also high compared to most of the other sites in 1993 (Figure 3.12). Flower production may also be high in communities where Cassiope exhibits relatively low vegetative growth, however, as indicated by observations at the Erratic site in 1993 (Figure 3.9, 3.12). Live shoot densities among the observed sites were greatest in the Erratic and Beach Ridge communities and lowest at the RSI and Dryas ITEX communities (Figure 3.13). Cassiope dead:live shoot ratios were similar among most of the communities, except for 73 RSI and Dryas ITEX, which exhibited much higher ratios of dead:live shoots (Figi 3.14). 74 5.0 4.0 H 3.0 H 2.0 H 1.0H o.o-X X X X X -1 1—1—•—i—•—•—i—•—•—i 1 1—•—•—i—•—1—r CI Dl BR Erratic RSI RS2 RS3 RS4 RS5 Figure 3.9: Total shoot elongation of Cassiope tetragona observed at the end of the 1993 growing season in nine communities at Alexandra Fiord. See text for key to site abbreviations. 75 76 in CN CN O CN CN in 1—I CN o CN in o CN O o CN in O as o o o oo in o r--CN cn CN cn CN cn CM cn CN cn O N O N O N as as as Os as as as • 1 T 1 CN CN cn cn 4- 4- in in cn cn cn cn cn cn cn cn cn cn Pi Pi Pi Pi Pi Pi Pi Pi Pi Pi 8. o a 55 ~ -S <- s £ £ ° o ° S § §•21 •a'a s O U , GO O 00 Q U i * . * > O T3 <~-4 D C 8 £ C <D > c c fl u ^ S *-• co ? C S <3 " v_ * S ^ P a a t C y u- M _ | •IS 8 o « s ^ § S>> •"3 ob 8. 3 , oo y - • « O CN fi « o 1 c5 1 53 u rvi -t—» C3 -r-P H ct< c -a . . s— y +-» '55 -C .5 CO 3 00 77 0.4-0.3 4 0.2 4 0.14 0.0-CI ~r T T 3_F: Dl BR Erratic RSI RS2 RS3 RS4 RS5 Figure 3.12: Numbers of flowers produced per mature shoot of Cassiope tetragona, measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations. 78 c3 CD C3 o o <D 8 00 CD > £ 3 z 30.0-20.0-10.0-X X x 0.0 • •••| i i 1 r CI Dl BR Erratic RSI X 1 X T T RS2 RS3 RS4 RS5 Figure 3.13: Numbers of mature, live shoots of Cassiope tetragona per 10 cm 2 area, measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations. o c3 o o > T3 <D Q 2.0-1.5-1.0-0.5-00 1 1 i 1 1 i 1 1 i 1 1 i 1 1 i CI Dl BR Erratic RSI RS2 _J_ —i 1 r RS3 RS4 RS5 Figure 3.14: Dead:live shoot ratios of Cassiope tetragona measured in nine communities at Alexandra Fiord in 1993. See text for key to site abbreviations. 79 3.4. Discussion The patterns of environmental and vegetative variations observed in this study do not indicate any simple relations between environmental conditions and vegetative growth and reproduction of Cassiope tetragona. However, comparisons of the observed patterns do suggest some factors which may be of biophysical importance to this species. The timing of snowmelt appears to play an important role in determining the potential onset of growth of Cassiope tetragona at Alexandra Fiord, as observed in the experimental snow manipulations (Chapter 2). However, other factors appear to interact with snowmelt timing to determine the period of active shoot growth. Cassiope plants in communities with snowmelt occurring at similar times did not necessarily start commence shoot elongation at the same time. Evidence from this study suggests that these differences may be related to variations in soil temperatures among the sites. Plants growing in communities with warmer near-surface temperatures seem to have initiated growth earlier after snowmelt that in communities with cooler temperatures. This pattern supports observations made in temperature manipulation experiments at Alexandra Fiord, which show positive effects of increased temperature on the number of days from snow release to growth initiation for C. tetragona (Chapter 2). Likewise, vegetative development occurred more slowly in early-melting plots of the snow manipulations, which was attributed to cooler and wetter soil conditions (Chapter 2). The importance of post-snowmelt temperatures on vegetative and reproductive development has been noted in other studies (Inouye and McGuire, 1991; Larigauderie and Kummerow, 1991). Observations made in the same communities over 2 consecutive growing seasons indicate that, while the timing of individual phenophases may be highly sensitive to annual variations in growing season conditions, the total period of active growth is relatively constant within a site. The stability of the length the active growth period among years has been noted previously for Cassiope tetragona (Svoboda, 1977; Nams and Freedman, 80 1987b). Similar patterns were also found in temperature manipulations at the Cassiope ITEX site and within the snow manipulation plots at the Beach Ridge snowbed (Chapter 2). Given that the period of active growth is strongly related to total annual shoot elongation, as was observed in this study, this may be a factor which stabilizes the vegetative production of C. tetragona in response to annual variability. The factors which initiate growth cessation in arctic plants are poorly understood (Shaver and Kummerow, 1992). Symmetrical effects of snowmelt timing and/or growing season climate on both growth initiation and growth cessation of Cassiope tetragona may indicate that this species exhibits autonomous cessation of growth after a set period of activity (S0rensen, 1941). Annual variations in timing of senescence indicate that the control of autonomous cessation of growth for C. tetragona is not likely to be related to photoperiod signals. Physical mechanisms for autonomous growth cessation may be the depletion of soil nutrients or within-plant resources. An understanding of these controls is important for the prediction of how species such as C. tetragona would respond to long-term climate change, as greenhouse warming may. increase the length of the potential growing season in the Arctic by as much as a month or more (Maxwell, 1992). Environmental measurements made in a variety of Cassiope-dominated communities at Alexandra Fiord indicate that Cassiope tetragona is capable of flourishing under a wide range of environmental conditions. Snowmelt timing, active layer depth and soil moisture levels all varied strongly among the observed communities. Other factors such as temperature and soil nutrients are also likely to have differed significantly among the selected communities, although sufficient measurements were not made to explicitly examine variations among these variables. The large range of environmental conditions under which Cassiope tetragona is found at Alexandra Fiord indicate that this species is relatively flexible in its habitat requirements, as is consistent with its wide, circumpolar distribution (Hulten, 1968; Bliss and Matvayeva, 1992). The abundance of Cassiope tetragona at Alexandra Fiord is uncharacteristic for much of the Arctic, however (Nams and 81 Freedman, 1987a), indicating that this site is particularly suitable for the establishment and maintenance of Cassiope populations. The most productive populations of Cassiope tetragona observed in this study at Alexandra Fiord occurred in areas with early snowmelt and deep active layer thaw, indicating comparatively warm soil temperatures. These sites also had the lowest densities of live Cassiope shoots, reflecting a lower percentage cover of C. tetragona in those habitats, and high ratios of dead:live shoots, suggesting relatively high rates of biomass turnover. In contrast, the highest densities of Cassiope were observed in sites where annual production appeared to be low, and which were characterized by lower soil temperatures and, in some cases, late snowmelt. These patterns suggest that, at Alexandra Fiord, the greatest cover of Cassiope occurs in areas which may not be optimal for the growth of this species. Such patterns may be an indication of a lack of tolerance to some other, co-varying environmental factors, or may reflect competitive exclusion by more productive species. Vulnerability to competitive effects may also partially explain the common restriction of Cassiope tetragona to late-lying snowbeds in many areas (Porsild, 1920; Brassard and Beschel, 1968; Bliss et al, 1977; Miller, 1982; Bliss and Matvayeva, 1992), despite the habitat-flexibility exhibited by this species at Alexandra Fiord. Competition-induced habitat restrictions have been suggested for other arctic plants, such as Dryas integrifolia, on a larger latitudinal scale (Svoboda and Henry, 1987). If competition with other species more capable of rapid growth is a important factor controlling micro- and meso-scale distributions of Cassiope tetragona, changes to more ameliorated climate conditions may cause a shift in the distribution of this species due to competitive exclusion from more optimal habitats (e.g. Svoboda and Henry, 1987; Shaver and Kummerow, 1992). 82 4. Retrospective analysis of growth and reproduction of Cassiope tetragona 4 .1 . Introduction One of the most useful tools for studying forest ecology and environments has been dendrochronology, the analysis of tree-ring variations over time. The use of dendrochronological techniques allows the growth patterns of trees to be reconstructed and precisely dated back in time for as long as a clear tree-ring chronology is available. Reconstructions of growth variations may be used to extract information on small- and large-scale fluctuations in climate, ecological relationships, environmental disturbances and other areas of interest, depending on the type of chronology and the nature of the study (Fritts, 1976; Cook and Kariukstis, 1992). Such analysis has provided useful information on past climate and ecological conditions for many environments of the terrestrial world (Fritts, 1976; Cook and Kairiukstis, 1992), including the boreal arctic (for example, Kay, 1978; Cropper, 1982; Kuivinen and Lawson, 1982; Briffa et al., 1988). North of the coniferous treeline, however, use of dendrochronology has been hindered by the absence of erect tree species. Although several species of prostrate shrubs are found throughout the High Arctic, the extremely small ring widths and numerous growth anomalies characteristic of these species strongly restricts their usefulness for dendroclimatic analysis (Beschel and Webb, 1962; Warren Wilson, 1964; Walker, 1987; Kolishchuk, 1992). In theory, any species may be used in dendrochronological-type analysis, provided that it possess a few essential characteristics: a) annual increments that are distinguishable for most years, b) tree-rings or other features that can be cross-dated, and c) the attainment of sufficient age to provide adequate time control for a given investigation (Fritts, 1976; Schweingruber et al., 1992). The study of tree rings has already greatly expanded to include studies based on radioisotopic, cell structure and nodal growth patterns in trees. In 83 the Arctic, there are many tundra species which possess distinguishable annual growth increments due to the marked seasonality of these environments (Kershaw, 1962; Callaghan and Collins, 1976; Callaghan, 1988). However, many of these species are short-lived or slowly lose the definition of annual growth increments over time, restricting their usefulness for more long-term retrospective growth analysis (Callaghan, 1988; Callaghan etal, 1989). One relatively long-lived arctic species, Cassiope tetragona, may have great potential for use in retrospective growth analysis studies in tundra environments. Average age for a high arctic population of Cassiope on Devon Island was estimated to be 30-60 years (Bliss et al., 1977), and it has been suggested that individual genets may live up to several hundred years (Havstrom et al, 1995). In many environments, Cassiope tetragona exhibits wave-length patterns in leaf-length (Warming, 1908; S0rensen, 1941; Callaghan, 1973). These patterns have been used to delimit annual growth increments along individual stems (Callaghan et al, 1989; Havstrom et al, 1993, 1995). Leaves are retained on the stem for as much as 20-40 years, making a relatively long record of growth possible through the measurement of annual increments based on leaf-length patterns (Callaghan et al, 1989; Havstrom et al, 1995). Using patterns in leaf-length to delimit annual growth, Callaghan et al. (1989) were able to develop several Cassiope growth chronologies covering a period of up to 20 years. Significant correlations among different growth parameters were used to generate hypotheses about the growth dynamics of this species. The relationship between growth and climate was also examined, based on techniques commonly used in dendroclimatology. Significant climate response functions were developed for the majority of the Cassiope growth chronologies and were able to account for up to 74% of the variation in different growth parameters (Callaghan et al, 1989). More recently, Havstrom et al (1995) used samples of modern Cassiope from Alexandra Fiord, Ellesmere Island to develop a predictive model relating July degree-days 84 to Cassiope growth and reproductive chronologies. This model was applied to preserved specimens of Cassiope released from beneath a retreating glacier at Alexandra Fiord (Bergsma et al, 1988). Based on their predictive model, Havstrom et al. (1995) estimated a difference of 0.7 ± 1.2 °C between the modern and pre-Little Ice Age climate at the study site. These investigations indicate that Cassiope tetragona has great potential for use in retrospective growth analysis studies similar to those used in dedrochronology. However, further investigation is necessary to understand how this species may be most effectively used as a record of past environmental and ecological fluctuations. For example, little information is currently available on growth and reproductive variability of Cassiope individuals within and between populations. Such information may have important implications for sample selection and the development of chronologies which maximize information on the desired signal (Fritts, 1976; Schweingruber et al, 1992). A better understanding of which parameters are likely to provide the greatest sensitivity to different environmental factors is also important for the design of efficient and effective studies. It is to be hoped that continued development of techniques for analyzing past growth and reproduction of Cassiope tetragona will provide a useful tool for investigating environmental and ecological variations in the recent past for arctic regions. The purpose of this study is to a) provide some additional information on the use of Cassiope tetragona to generate proxy environmental data and historical records of growth, and b) to use such analysis to investigate ecological relationships for this species at a high arctic site on Ellesmere Island. A different technique is used for delimiting annual growth increments than that employed in previous studies (Callaghan, 1988; Callaghan et al., 1989; Havstrom et al., 1993, 1995), with the hope that this will provide a method for extending the length of future Cassiope chronologies. 85 4.2. Methods 4.2.1. Species characteristics Cassiope tetragona (L.) D. Don is a long-lived, high arctic ericaceous shrub. Its distribution is circumpolar with extensions into the subarctic and alpine tundra (Hulten, 1968). It is commonly found in areas with moderate to high accumulation of winter snow and mesic, nutrient-poor soils (Bliss et al, 1977; Nams and Freedman, 1987a; Bliss and Matvayeva, 1992). The growth habit of the species is prostrate, forming loose cushions of stems branching out from a central root mass. The production of secondary compounds in the leaves makes the leaves un-palatable and the plant is rarely grazed except for the flowers (Nams, 1982; Callaghan etal., 1989; personal observation). Genets of C. tetragona are generally long-lived, with individual stems dying out and being replaced by new branchings of auxiliary shoots. Cassiope tetragona produces two alternating sets of opposite leaves along a stem which form four distinct rows of leaves. Patterns in the positioning of nodes in adjacent leaf rows appear to be analogous to patterns in leaf length (Figure 4.1). Because leaf scars are retained on a stem over the full length of a shoot, use of this pattern to date annual increments would give potential for generating longer growth chronologies than the leaf length method used in Callaghan et al. (1989) and Havstrom et al. (1993, 1995). Testing this methodology is one of the objectives of this research. 4.2.2. Study Site The study area is the Alexandra Fiord lowland, located on the eastern coast of central Ellesmere Island in the Canadian Arctic Archipelago (78°53'N, 75°55'W). A description of the study area is given in Chapter 2. Samples of Cassiope tetragona were 86 a 87 collected from a single community located in the central plain of the Alexandra Fiord lowland. This community is also the site for a temperature manipulation experiment, described in Chapter 2. The site is a Cassiope-dominated community positioned on a very slight slope (1-2°) with a northerly aspect. Other prominent species in the community are Salix arctica, Dryas integrifolia and Luzula arctica. Vascular plant cover is approximately 75-80%. The micro-topography of the site is characterized by vegetated hummocks and hollows with occasional boulders. Soils are composed of alluvial sands and gravels with a thin (0-6 cm) layer of organic material at the surface. The site is mesic-hydric and generally experiences relatively late snowmelt compared to surrounding areas (early to mid-June). 4.2.3. Climate data Weather records from 2 sites on Ellesmere Island were used in this study to investigate the relation between climate and growth and reproduction of Cassiope tetragona at Alexandra Fiord. Records of July average temperatures and July degree days for Alexandra Fiord from 1980-1988 are taken from Labine (1994). A longer and more detailed weather record was obtained from the nearest High Arctic Weather Station, located approximately 300 km from the study site on the east coast of Ellesmere Island at Eureka. Monthly temperature and precipitation records are available for this station from 1949 to present (Atmospheric Environment Service,1982, 1980-1992). The relation between weather patterns observed at Alexandra Fiord and Eureka is discussed by Labine (1994). Although annual mean temperatures are higher at Alexandra Fiord (5-6° C; Labine, 1994), this difference is primarily the result of milder winter temperatures at Alexandra Fiord. July temperatures and July degree days are often greater at Alexandra Fiord, but this difference is not consistent and varies from year to year, occasionally with Eureka experiencing higher temperatures (Labine, 1994; Figure 4.2). 88 Figure 4.2: Average July temperatures recorded at Eureka and Alexandra Fiord from 1980-1988. The two data sets have a correlation coefficient of R=0.35. Data for Eureka is from Atmosperic Environment Service (1980-1992) and from Labine (1994) for Alexandra Fiord. 89 4.2.4. Data sampling and measurement Samples of Cassiope tetragona were collected on 18 August, 1993. Collections of 5-10 shoots were taken from 15 plants within a visually homogeneous area of the study site. Samples were selected in an effort to collect long shoots with few branchings along the main stem. The collections were air-dried for several days after sampling and stored in loose bags. In the laboratory, 10 plants from the sample were randomly selected for measurement. Five shoots were measured from each plant. Plants with several shoots exhibiting growth deformities were excluded from the sample set. Prior to measurement, bends in the stems that had developed during drying and storage were removed by briefly soaking the shoots in water (1-2 minutes) and then inserting them into 1-1.2 cm diameter glass tubes to dry. Once dry, 2 adjacent rows of leaves were removed from each shoot by hand, leaving in place any flower buds or peduncle bases (Figure 4.1). Branches from the main stem were removed and their previous position on the stem was marked using color-coded paints. Growth measurements were taken in the form of internode lengths, or the distance between adjacent leaf scars (Figure 4.1). This was done under a dissecting microscope (lOx to 30x magnification) using a manually operated caliper system (designed by J. Svoboda, 1992). The measuring device consists of a digital caliper and location marker mounted on a solid base. Samples are placed in narrow glass tubes (1-1.2 cm diameter) that can be precisely positioned relative to the location marker with distances from some reference point recorded on the digital caliper. Internode lengths were measured from the base to the tip of the shoot. The positions of auxiliary branches, flower buds and flower peduncles were recorded during these measurements. 90 4.2.5. Retrospective growth analysis The delimitation of annual growth along a stem was accomplished using patterns in internode length (Figure 4.3). Internode measurements show a strong bimodal distribution which is caused by oscillations in internode lengths similar to those exhibited by leaf-lengths (Callaghan et al., 1989). The shortest internode length within each wave series was used to indicate the end of each year's growth. In order to test this interpretation, the terminal leaves of the previous year's growth were marked on living Cassiope shoots at the study site in early June, 1992, using non-toxic liquid paper (n=60). The marked stems were sampled at the end of the 1993 growing season. Subsequent internode measurements showed that the shortest internode lengths consistently corresponded to the terminus of annual growth as indicated by the marked leaves. Three parameters were measured to represent annual growth and reproduction for the sampled Cassiope population: stem elongation, leaf production and flower production. Annual stem elongation values were calculated as the sum of internode lengths within the year's growth. Leaf production was measured by counting the total number of leaves produced in each year within the two rows of leaves measured on each shoot. Flower production was estimated by counting the number of peduncles scars within each section of annual growth. Data series for each shoot were crossdated by using skeleton plots of internode lengths and flower production. Each chronology was then standardized to remove low frequency variation such as growth trends. Trends in the leaf and flowering chronologies were estimated using a linear fit and then standardized as It=Ot/Et (4.1) where Ot is the observed value for year t, Et is the expected value generated by the linear fit and It is the resulting index value for that year (Fritts, 1976). Because the stem elongation chronologies exhibited many different patterns of low-frequency variance and could not easily be fit with a linear trend, expected values were generated for each chronology using a 91 o o o o o o o o o o ^ ^ ^ ^ k o o d o o ( L U L U ) m6u9| apoujaiui 92 weighted moving average, (4.2) / n > where wi = N 0, — a o- standard deviation n= array of data points (after Cook et al., 1992). A standard deviation value of 1 was used for the normal distribution and an array of 5 data points on either side of the observed value was used to generate expected values for each observed values in the series. The array of data points included in the weighted moving average was truncated at the beginning and end of each data series. After calculating the weighted moving average values for a time series, a standardized growth index was produced by dividing each observed value by its predicted value as in equation (4.1), above. Following standardization and cross-dating, individual shoot chronologies within each plant were averaged together to form a mean whole plant chronology. The resulting plant chronologies were then averaged to form a master chronology for each of the measured variables. Descriptive statistics were calculated for the three Cassiope indices using methods outlined in Fritts (1976; Table 4.1). Mean sensitivity values were calculated as where xt is the chronology value for year t (after Fritts, 1976). A nested analysis of variance (NESTED procedure, SAS, 1985) was used to partition variance components within each chronology over a 10 year time period with continuous overlap of all individual stem data series (Fritts, 1976). Because the flower frequency data for individual shoots was non-normally distributed, the flower data was ranked and the ranks transformed to approximate a normal distribution using a Blom transformation (RANK procedure, SAS, 1985). The nested analysis of variance was then performed on the transformed data set, msx- n-l 1 (4.3) 93 analogous to an extension of the Kruskal-Wallace test for a single-factor analysis (Conover and Iman, 1981; Zar, 1984). The relations between Cassiope growth and reproduction and climate patterns were examined using correlation and regression analysis. Frequency histograms of the mean standardized chronologies for growth and reproduction did not show strong deviations from normality in the data distributions, and parametric regression analysis was used. Correlation matrices for Cassiope indices and climate records from Alexandra Fiord and Eureka were generated using the full overlap period of each data set (REG procedure, SAS, 1985). July average temperature and degree day records were used from Alexandra Fiord (Labine, 1994). From the Eureka data, monthly temperature and precipitation records were used for May through September (Atmospheric Environment Service, 1980 - 1992, 1982). Winter temperature and precipitation records were input to the analysis as totals or averages of monthly records from September to May. Weather records for both the current and previous year were included in the analysis. Two sets of climate response functions were generated for each chronology using backward stepwise multiple linear regression with Eureka climate variables (REG procedure, SAS, 1985). In one set of response functions, only those variables which were significantly correlated with the dependent variable were entered into the model selection procedure. The second set was created using all climate variables and the previous year's growth as candidates for selection for the model. A significance level of a=0.05 was used for the retention of variables in the models. The Cassiope leaf and flower production data sets was also used to test the climate/growth model developed in Havstrom et a/.(1995). This model relates July degree days to Cassiope leaf number and flower number (=actual flowering, AF; Havstrom et al, 1995). Because the values used in developing this model were not standardized (Havstrom et al, 1995), unstandardized chronologies of leaf and peduncle number were used as input to the model. Alternative July degree day and July average temperature 94 transfer functions for Alexandra Fiord was generated using the standardized chronologi developed in this study. Model selection was performed using backward stepwise regression (REG procedure, SAS, 1985). 95 4.3. Results 4.3.1. Chronology characteristics A total of 50 shoots representing 10 plants were measured to produce each of the 3 chronologies. This sample size was reduced to 9 plants for the two growth chronologies and 8 plants for the reproductive chronology (Table 4.1). One plant set of five stems was removed from the flower production data set because no stems within that plant produced mature flowers over the 40-year period of measurement. Because the lack of flowering was uniform and thus, contained no useful climate information, the measurements for that plant were not included in the flowering chronology. An additional plant set of 5 shoots was removed from the chronology data sets because of difficulties in crossdating the stems within that plant (see discussion, below). The average total stem length of the shoots collected in this study was 152 ± 45 mm, with an average age of approximately 33 ± 8.9 years. The oldest shoot measured was 75 years old with a total stem length of 225 mm. Estimates of mean shoot elongation and leaf and flower production for the sampled population are given in Table 4.1. Mean sensitivity and standard deviation values may be used as an indication of the sensitivity of the measured parameter to annual variations in environmental conditions such as climate (Fritts, 1976). Of the three Cassiope parameters measured in this study, leaf numbers showed the least amount of annual variation and lowest mean sensitivity (Table 4.1). Shoot elongation exhibited slightly more sensitivity, and flower production showed the greatest amount of annual variation, with a mean sensitivity 6 times greater than that exhibited by the leaf chronology. These differences in sensitivity are clearly apparent when the three chronologies are plotted together (Figure 4.4). Correlations between the three averaged chronologies are given in Table 4.2. Leaf number and stem elongation values in the same year exhibited a strong positive correlation, 96 Table 4.1: Descriptive statistics of three growth and reproductive chronologies of Cassiope tetragona (arctic white heather) at Alexandra Fiord, Ellesmere Island, N.W.T., Canada (78°53'N, 75°55'W). Shoot elongation Leaf number Flower number Sample size Interval years, A.D. Mean Standard deviation Between-plant mean correlation coefficient 1st order auto-correlation Mean sensitivity 9 plants (5 stems/plant) 1957 - 1992 4.78 mm/yr 1.32 mm/yr 0.55 -0.32 0.18 9 plants (5 stems/plant) 1957 - 1992 10.36 lvs/yr 2.54 lvs/yr 0.22 0.07 0.11 8 plants (5 stems/plant) 1967 - 1992 1.28 flrs/yr 0.96 flrs/yr 0.24 -0.16 0.66 97 98 Table 4.2: Partial correlation coefficients showing relations between current (year C) and previous year's (year C-l) values of three Cassiope chronologies. Shoot elongation year C Shoot elongation year C-l Leaf number year C Leaf number year C-l Flower number year C Shoot elongation year C-l -0.32 ns Leaf number year C 072*** 0.01 ns Leaf number year C-l -0.06 ns 0.07 ns Flower number year C -0.48* 0.75*** -0.24 ns 0.72** Rower number year C-l -0.12 ns -0.05 ns -0.22 ns Asterix denotes significance level: * = jx0.05, ** = p<0.01, ns = not significant. Year C and C-l denote current and previous year's values of the climate variables, respectively. 99 and both parameters were positively correlated with flower number in the following year (Table 4.2, Figure 4.4). Rower number and stem elongation were negatively correlated in the same year. The partitioned variance components of each chronology show that all of the three parameters exhibited high levels of variation associated with differences between stems within the same plant (Table 4.3). Rower production and leaf production show the highest portions of within-plant variability (74-80%). The stem elongation chronology exhibits the least amount of within-plant variance, with a consequent increase in variance common to all plants in the data set. 4.3.2. Modeling the response of Cassiope tetragona to variations in climate Correlation coefficients between the measured Cassiope parameters and climate records from Alexandra Fiord are given in Table 4.4. July degree days were significantly correlated with leaf number in the current year and flower number in the following year. Average July temperature was similarly correlated with leaf number and flower number, and also with shoot elongation in the current year. Significant correlations between the Cassiope chronologies and climate records from Eureka are presented in Table 4.5. All but one of the significant correlations (p<0.05) were with monthly temperature averages from May to September. Correlations with June and July temperatures accounted for 7 of the 12 significant correlations. Stem elongation was positively correlated with June, July and September temperatures and May precipitation of the same year and negatively correlated with June, July and September temperatures of the previous year. Rower production was positively correlated with May, June and September temperatures of the previous year, but was not significantly correlated with any weather variables in the same year. This pattern of correlations corresponds to the pattern of between-chronology correlations (Table 4.3), 100 T J C C3 CD Ui <U a H T J O E 03 T J CU s o '5b _o o c 0 u , 43 CJ § 2 1 CU <U CU o £3 CO O co 'oo 13 C < CU x> c3 H c3 T J CCS C c3 u c o ° xi E ^ c o '-C3 o 3 3 o o X ! GO S3 Oty c o w — .2 00 tU (3 CU 2 Ct, c <u OH c CU Ct, o xi 8 •g > ~ .2 o c o s o x: cj cu cj c <u I 8 00 cu CJ c C3 X i CJ CJ U c c 00 <u CJ c c3 C CU I (U 2 3 O 00 ON o c5 CO CO o o CO VO •—I o d 2 co 2 c o S S 8 ON CO o CO CO o o oo ON O O d CU cj c e .CU O 00 CO ON VO vo o d vo CN O <u 101 Table 4.4: Partial correlation coefficients showing relations between average July temperatures and July melting degree days at Alexandra Fiord and annual growth and reproductive indices of Cassiope tetragona (n=9). Parameter July average July melting temperature degree days Shoot elongation 0.733 * 0.547 ns (current year) Leaf number 0.898 ** 0.720 * (current year) Flower number 0.740 * 0.891 ** (following year) Asterix denotes significance level: * = jxO.05, ** = rxfj.01, ns = not significant. Table 4.5: Partial correlation coefficients showing relations between climate records from the Eureka High Arctic Weather Station and annual growth and reproductive indices of Cassiope tetragona at Alexandra Fiord. Climate variable Shoot elongation Leaf number Flower number (n=35) (n=35) (n=25) June temperature 0.5034 ** 0.3761 * 0.0324 ns yearC July temperature 0.5971 ** 0.3340 * -0.0793 ns yearC September temperature 0.3900 * 0.2215 ns 0.1275 ns year C May precipitation 0.3281 * 0.2442 ns -0.1135 ns yearC May temperature -0.3128 ns -0.2037 ns 0.5264** year C-l June temperature -0.4827** -0.2121 ns 0.6264** year C-l July temperature -0.3313 * 0.0965 ns 0.2670 ns year C-l September temperature -0.4223 * 0.0338 ns 0.4660* year C-l Asterix denotes significance level: * = p<0.05, ** = p<0.01, ns = not significant. Year C and C-l denote current and previous year's values of the climate variables, respectively. 102 where flower production is negatively correlated with stem elongation in the same year and positively correlated with stem elongation in the previous year. Climate response functions were developed for each chronology using backward stepwise regression with Eureka weather variables as predictors. Response functions using only weather variables that were significantly correlated with the predicand are able to explain 14% of the variation in leaf production, 52% of the variation in flower production, and 56% of the variation in shoot elongation as represented by the mean standardized chronologies for each parameter (Table 4.6). By allowing input of additional variables, including previous growth, the response functions were able to explain an additional 9% of the variation in shoot elongation and 32% of the variation in flower production (Table 4.7). Predictive ability was not improved for the leaf number response function, as no additional variables achieved the minimum significance level required for entry to the model. Unstandardized leaf number and flower number chronologies obtained from the sample population were used to test the July degree day prediction model generated by Havstrom et al. (1995; Table 4.8). Although the model tended to under-estimate observed July degree days (Figure 4.5), the differences between the actual and calculated degree day means were not significant at ct=0.05 (Table 4.8). Alternative July degree day and July average temperature transfer functions were developed for Alexandra Fiord using the standardized chronologies obtained in this study (Table 4.9). The July degree day transfer function uses flower production in the current and previous years to predict degree days, and is able to provide very close estimates of observed degree days over the calibration period (R 2 = 0.92). The model developed to predict July average temperatures is based on only current year values of leaf production, and appears to be less accurate than the July degree day model (R 2 = 0.80). Verification of these model would require additional climate records from Alexandra Fiord or independent Cassiope chronologies, which are not currently available. 103 2? CI ,—1 0 ha hi 0 0 VO ha hi 0 0 CN ha hi 0 0 O o 0 0 O IT) CN co 1 PL, ci d CN in CO 3 C8 CN VO lO lO O O O 04 , , u E— u >, H 03 CO c 3 w f—' 13 t-0 r-co "S 3 0 0 0 0 co c CT CO + 0 >—> c , , CN 0 u u .—1 4—» O 0 H O c 3 co c uly d ponse1 3 •—> + ponse1 56 (J VO 0 0 0 .9805 GO 0 .9805 Re = 1.5214 + 0.0 = 0.5992 + 0. 0 11 c 0 c redican roducti ongatic xluctioi a a 13 "8 O. we: loo \* •S Ho Sh J3 > •a 3 o 0. -a o u •s 3 H > o g T 3 u u •5 X ) 3 a 3 ca 8. a a u CO u > II H a a p co "8 a - .a c o 3 c8 T3 CO *-» 08 a 00 c ' G O 3 O c o GO _ 3 13 a co 2 2 00 to 13 > co GO a o 4—» 0 c 3 CO Vl C a GO a c 0 3 1 cx co u, T3 C CvS -C o O r^ c3 H GO 3 O '> CO •a c 08 GO co IS cd 'S > c3 a T 3 C 03 CO i _ co a GO s 0 1 P L _ 3 > OS O O O 5 CN 0 0 u c3 2 II 08 3 O l CO a o CO c a CO GO c a GO CO a! C O o d + u PQ OT O CN co 0 0 CN + CO ci II 1 a 13 "8 o o o d 10 vq co m V O d u H J8 a. CO 0 0 CO CN o d + u H 3 CN t^ 8 u H 3 O N DH CN CN C O ' -3 O 3 1 Q. u. CO & p E o VO CN o CO in r—I d o + i-H U UJ 0 0 ^ 1 u H •4—> CO I < O N CN O o d + CN vO CN 0 0 co CO r-d 11 c o 03 00 c o 8 0 0 CO I 3 8 o u H co c 3 •—> CN O + cn o 0 0 Ov d 11 p O 3 1 3 104 X ) cO x> O vo o o o o o 13 > •a 2 CM fi <u 3 "cO > OS c CO 3 8-c o '•u> CJ c .3 GO T J CJ 1 -8 co co CN d + u vo vd oo CN 00 CN + CO d oo co CO T J CJ <U u oo cj T J in ci O N C I o 00 d cO ll Q u Z •J o •—( in i — i + vo ON SM 3 fi <u D . S cu -4—» CU 00 fi <u CO 3 0 CM T3 a CJ o u •£3 a o 3 > o a u T3 u -a U x 3 0 0 B 3 a 1 -8 o c II CO Q CO „ -CD <U 00 CU T J oo e B >> 3 CU 00 fi cu > CO T J CU -M-» CJ '•a <u o c 3 co C CO 1 « O w •o CU x> § &o cu ^ 3 c l 3 B : 0 T J c CO T J <U > u CU CO X ) 0 c CJ cu 1 8 T J CJ > CO X B Q .CJ cE ? a: OM < ; $ 8 T1 d o 00 ON S O T J U i O fi S § ON CJ x> cO H o .a CO X (U + Z CO r> r-+ ON C O vd c t X ) .g o 00 00 c CO •c CO > CJ H 3 o oo CO o O N VO CO CN C l CN t—I CN CO DM 3 a o c CU cu » CU CQ 00 O N CN 00 «n in VO CN CO ON 00 C O ci CL, 3 o u O c -3 o 105 Figure 4.5: Observed July melting degree day values from Alexandra Fiord (Labine, 1994) plotted with predicted values calculated from the transfer function developed in Havstrom et al. (1995). Correlation between the observed and predicted values is 0.87. 106 4.4. Discussion 4.4.1. Critique of Methods The results of this study indicate that internode length patterns can be successfully used to delimit annual growth increments for Cassiope tetragona. Because individuals from only one community were examined, this does not provide direct evidence that the technique may be applied to other populations. However, unpublished observations indicate that the observed pattern in internode lengths is common among populations of Cassiope tetragona throughout the High Arctic (Ferguson, 1992; Svoboda, J., Henry, G. H. R., and Havstrom, M., personal observations). Two of the Cassiope chronologies presented here cover a time span of 35 years, 10-15 years longer than chronologies developed by Callaghan et al. (1989) and Havstrom et al. (1995) using patterns in leaf length. Chronology length in this study was limited primarily by the total stem length of the sampled shoots, as it was generally possible to distinguish leaf scars and, hence, annual increments, over the full length of a stem. It is likely that chronologies of greater length may be developed by sampling at sites where Cassiope produces long, unbranching stems, such as on steep slopes or in areas of well-developed hummock and hollow topography. Continuous stems of up to 350 mm in length have been observed on rock outcrops at Alexandra Fiord (personal observation). The procedure of cross-dating stem sequences within plants was problematic for this population, due to the level of inter- and intra-plant variability contained within the chronologies. It was initially expected that cross-dating would be easily accomplished, as each sampled stem possessed green leaves at the tip and was assumed to be growing at the time of collection. Later examination of individual growth sequences indicated that several shoots may have stopped growing one to three years prior to collection. Some of these shoots possessed a withered and darkened appearance of the undeveloped leaves at the tip 107 of the stem. Among other shoots, the only indication of growth cessation was a reduced number of photosynthetic leaves on the stem and dryer, more brittle leaf tissue. In a few cases, significant pointer years could not be aligned, suggesting the possibility of missing annual growth increments. It is currently unknown whether Cassiope tetragona can maintain dormancy in growth over a period of a year or more (equivalent to a missing ring within a tree ring series) and continue to grow normally following the dormant period. On one of the sampled stems, a growth sequence was found which strongly indicated that stem had ceased growth for some period and then resumed elongation. Leaf numbers and stem elongation steadily decreased over 5-6 years reaching an anomalous low minimum. In the following year of the growth sequence, leaf number and stem elongation values tripled from the previous year's values and maintained a similar level of growth for the rest of the stem sequence. Crossdating of this stem suggested that the shoot had been dormant over a period of three years. Aside from this example, no other direct evidence for missing growth years was found among any of the sampled plants. Difficulties in crossdating may have caused the portion of within-plant variation in the chronologies to be overestimated in cases where stem sequences were improperly aligned. As the time-series of all three variables were crossdated and aligned in the same pattern, any bias in variance estimates due to misaligned sequences should be constant among the three chronologies. Use of a different standardization technique for the stem elongation chronology may have introduced differences in variance and autocorrelation estimates that are not constant among chronologies. The numeric filters used in standardization of the three data sets were selected on the basis of three criteria: a) ability to reasonably estimate average low frequency trends in the data set, where b) low frequency trends of greater than approximately 5 years were classified as noise, and c) ease of calculation. A linear trend was selected for standardization of the leaf and flower number time series because the majority of the series exhibited linear or nearly-linear trends over the 108 time period of interest and more flexible filters appeared to be overly sensitive to small variations due to the discrete nature of the unstandardized data. The weighted moving average used to detrend the stem elongation time series is much more sensitive to low frequency trends in the data than the linear filter used for the other data sets. Different standardization techniques may have important effects on the signal-to-noise ratio of the mean chronology (Cook and Briffa, 1992). The increased portion of common variance in the stem elongation index compared to the linear-detrended indices is likely to be at least partially the result of using a more flexible filter for standardization within that chronology. 4.4.2. Chronology characteristics The estimates of mean annual growth and reproduction presented here are similar to those obtained elsewhere in the High Arctic for Cassiope tetragona. Estimates of flower production for 1980-1981 obtained from the unstandardized chronology agree with flower production estimates observed during the same time period for Cassiope tetragona in a similar community at Alexandra Fiord (Nams and Freedman, 1987b). Average annual leaf production values are very close to those obtained by Callaghan et al. (1989) for two populations on Svalbard, and by Havstrom et al. (1995) for a different population at Alexandra Fiord. However, the variability in leaf numbers within individual stem sequences in this study appears to be much lower than for the other sampled populations listed above. Variations in flower production also appear to be smaller than those in the Alexandra Fiord population sampled by Havstrom et al. (1995). This reduction in variation may be due to differences in the sample sizes used in the different studies or may reflect a more consistent annual growth and reproductive rate for individuals of Cassiope tetragona within the community sampled for this investigation. Chronology statistics such as standard deviation, mean sensitivity, and partitioned variance components are often used by dendroclimatologists as an indication of the 109 sensitivity of ring-width chronologies to a macroclimatic signal (Fritts, 1976; Briffa and Jones, 1992). Estimates of mean sensitivity for the leaf length and stem elongation indices obtained here are moderately low, but are close to average sensitivity values obtained for many northern tree-ring chronologies (e.g. Cropper and Fritts, 1981; Cropper, 1982). In contrast, the flower production chronology exhibits a much higher mean sensitivity, which may be an indication that this variable is more sensitive to high-frequency environmental fluctuations than leaf number or stem elongation. A l l three Cassiope chronologies presented in this paper exhibit large amounts of variation attributed to within-plant factors. This variation is relatively high compared with variance levels common in dendroclimatological studies (Fritts, 1976; Briffa and Jones, 1992). As discussed above, the type of filters used in standardization are likely to have had some influence on the portion of common variance contained in the chronologies. Estimates of variance components have not previously been calculated for C. tetragona and it is not possible to infer whether the estimates obtained in this study are representative. The large amounts of variation observed between individual shoots in the cross-dating procedure supports the low estimates of common variance calculated for the chronologies. Qualitative observations of the three data sets also supports the suggestion of greater common variance in the stem elongation index than the other indices. Large within-plant variations might a priori be expected for this species. Although the stems within a plant are connected to the same central root system and share nutrient resources, it is likely that each stem may experience a slightly different microenvironment depending on its position within the plant cushion. Large variations in temperature within plant cushions and rosettes have been noted for many tundra species (Warren Wilson, 1957; Molgaard, 1982) and differential shading effects may be important in areas where the plants grow in dense mats, as in this study. Resource partitioning among stems within a plant may also be an important factor affecting annual growth. Internal resource partitioning among individual stems of a Cassiope plant has previously been suggested by 110 Havstrom et al. (1993) to explain differences between main and juvenile stem growth responses to field manipulations at a highly vegetated low arctic site. Intra- and inter- plant variations associated with competition and resource partitioning are likely to obscure the strength of the environmental response signal contained within individual growth or reproductive sequences. The effects of competition are predicted to be greatest in areas where nutrient and light resources strongly limit plant growth, such as in nutrient poor soils and in areas of high vegetation cover (e.g. Tilman, 1988). Although little is known about the effects of competition in high arctic environments (Bliss and Peterson, 1992), it is reasonable to suspect that it may play an important role in the plant growth of some of the relatively lush plant communities of Alexandra Fiord and other high arctic 'oases'. The importance of site-selection to control for the effects of with-in community interactions has already been well-acknowledged in dedrochronology (Fritts, 1976; Schweingruber et al., 1992). Indirect evidence from this study suggests that community dynamics may also play a strong role in growth and reproductive variations in populations of Cassiope tetragona. High levels of variation among shoots sampled within a population have important implications for estimates of chronology error (Fritts, 1976; Briffa and Jones, 1992). Given increasing amounts of within-population variability, larger samples must be obtained in order to estimate population parameters within a desired level of accuracy. Because the decrease in accuracy is non-linearly related to the portion of with-in population variance (Briffa and Jones, 1992), sample size requirements can grow rapidly as variance increases from moderate to high levels. If the estimates of within population variance reported in this study are representative of intrinsic levels of variation within the species, then large samples are required for the generation of any Cassiope chronology. However, as mentioned above, it is likely that some portion of this variability is the result of specific community dynamics and can be expected to vary among communities. I l l 4.4.3. Statistical relations between climate and production of Cassiope tetragona Despite high levels of intra-plant variability within the chronologies, the averaged growth and reproductive indices obtained in this study showed strong correlations with climate records from both Alexandra Fiord and Eureka. The strongest correlations were found between the Cassiope chronologies and July average temperatures and melting degree days recorded at Alexandra Fiord. This is to be expected, as weather records from Alexandra Fiord provide a much better representation of growing conditions for the sampled population than those measured 300 km away. Stem elongation was not significantly correlated with July melting degree days in this study, but was significantly correlated with average July temperatures at both Eureka and Alexandra Fiord. The lack of correlation between stem length and July degree days recorded at Alexandra Fiord may be due to residual effects of the method of degree day calculation (see Labine, 1994), rather than insensitivity of that parameter to July temperature sums. Correlations with July average temperatures at Eureka show a similar pattern as with July temperatures from Alexandra Fiord, i.e., July temperatures are significantly correlated with leaf number and shoot elongation in the current year and flower number in the following year. This similarity provides general support for the use of climate records from Eureka as a proxy indicator of weather conditions at Alexandra Fiord. The majority of the significant correlations found between the mean chronologies and climate variables were with temperatures in June and July. The importance of early-and mid-growing season temperatures for Cassiope tetragona has been noted by other authors (Nams and Freedman, 1987b; Callaghan et al., 1989; Havstrom et al., 1993). In the High Arctic, the most intense periods of above-ground plant growth and development of reproductive organs occur during the first two months of the growing season (Shaver and Kummerow, 1992; others). Many arctic plants show positive correlations between growth and summer temperatures due to direct effects on photosynthesis and respiration or 112 indirect effects on soil nutrient availability and uptake rates (reviewed in Chapin, 1983; Shaver and Kummerow, 1992). Shoot elongation was positively correlated with average temperature at Eureka in June, July, and September of the current year and negatively correlated with temperature in the same months of the previous year. Leaf number was also positively correlated with temperatures in June and July of the current year. Positive correlations with current year summer temperatures are likely to represent a direct or indirect positive growth response to warmer temperatures. Significant increases in stem elongation of Cassiope tetragona have been observed under conditions of experimental warming at a high arctic site on Svalbard (Havstrom et al., 1993). Experimental manipulations of temperature at the same sample site used in this study also resulted in significant increases in shoot elongation (Chapter 2). Positive correlations between shoot elongation and September temeratures in the same year are unlikely to reflect direct effects of warmer temperatures in September on growth, as shoot elongation of Cassiope plants at Alexandra Fiord has generally ceased by mid- to late-August (Nams and Freedman, 1987b; Chapter 3). High September temperatures at Eureka may represent an extension of the growing season by mild temperatures at the onset of winter and may allow more comple translocation of nutrients within a plant, possibly resulting in some biomass changes. Negative correlations between growth and summer temperatures in the preceding year could be the result of within-plant resource depletion following periods of increased production. Negative correlations between flowering and growth in the same year suggest the hypothesis that production of flower buds in the year prior to flowering and subsequent development of those tissues reduces the potential for vegetative growth during the year of flower maturation (the 'current year' of the flowering chronology). Like many arctic species, Cassiope tetragona commonly initiates the development of flower buds in the year prior to when flowering actually occurs (S0rensen, 1941). A large portion of flower bud development takes place during later parts of the summer before flower maturation, after 113 growth has slowed or ceased (Nams and Freedman, 1987b). At the end of the summer, undeveloped flower buds contain the highest concentrations of nitrogen and phosphorus of all the above-ground tissues of a Cassiope plant (Nams and Freedman, 1987b). In addition to using the products of current year growth for flower bud production, the formation of flower buds creates a sink for next year's resources, some portion of which must go into flower maturation and fruit development if the investment in flower bud production is to be optimized. This pattern of resource allocation would result in the observed positive correlations between growth and flower bud production in the following year and negative correlations between growth and flowering in the same year. Indirect effects of such allocation strategies would also be likely to cause a reversal in the sign of correlations between growth and climate variables between years. Similar 'costs' of flower production for growth have been hypothesized for Dryas octopetala (McGraw and Antonovics, 1983), a species which shares many life history characteristics with Cassiope tetragona (Nams and Freedman, 1987b; Molau, 1993; personal observations). In years where growth is favorable because of warm summer temperatures, high levels of primary production may also increase nutrient uptake from the soil, depleting the nutrient reserves available at the beginning of the growing season in the following year. Although high summer temperatures may also increase the availability of nutrient resources in the soil through increased mineralization, this effect may be counterbalanced by greater nutrient immobilization (Nadelhoffere/a/., 1992). Significant climate response functions relating variation in the mean chronologies to Eureka climate records were developed for each of the three chronologies. Because the response functions were, of necessity, based on climate data from a station 300 km away, conservative methods were used in developing the models. Two techniques were used in the initial selection of predictor variables for the backward stepwise selection procedure in order to provide different levels of conservativeness in the resultant response functions. The first technique allowed only those climate variables that were significantly correlated 114 with the mean chronology to be eligible for inclusion in the model, thus ensuring that no variables were included that exhibit a weak relationship with the dependent variable. Because this technique might exclude variables which possess an important biological relationship with the dependent variable but do not exhibit a statistically significant correlation, an additional set of response functions was created with all variables, including previous growth, as eligible predictors. This second set of models represents a less conservative, but potentially more biologically realistic, group of response functions. The relations between climate and growth and reproduction of C. tetragona expressed in the more conservative response functions generally reflect the patterns of correlations discussed above. Climate response functions developed for the two growth chronologies using the unrestricted data were similar to the more conservative models. Although two additional variables were included in the second shoot elongation model, the general relations with the variables from the previous model were retained, suggesting that equivalent underlying biological relations are implied by both models. In contrast, use of less conservative techniques for development of the flower production climate response function resulted in a very different model than the one obtained by the conservative method. The less conservative model includes an additional four variables, with only one variable retained from the previous model. It is likely that the main difference between the models is the inclusion of previous growth in the model, which would be expected to change the relations with climate variables, as vegetative growth may be an important intermediary between flower production and climate. The response functions developed using both the restricted and unrestricted set of variables were able to explain over half the variation in the shoot elongation and flower production indexes, but only a small portion of the leaf production index. Because of the uncertainty in how well climate records from at Eureka represent weather conditions at Alexandra Fiord, interpretation of these models is difficult. Small differences in the correlation between the chronologies and climate variables and in the covariance of the 115 climate variables could result in the production of very different models. However, some general conclusions may be suggested. The relative portion of variability explained in the transfer functions of the three chronologies may be used as an indication of the climate sensitivity exhibited by the Cassiope parameters. Based on the climate response functions presented here, it appears that the leaf number chronology contains significantly less climate information than the other chronologies. This interpretation is also supported by a comparison of the mean sensitivity values of the chronologies, which are particularly low for the leaf number chronology. Climate response functions developed by Callaghan et al. (1989) for leaf number chronologies from two high arctic communities show a similar pattern, with variance in leaf numbers being relatively poorly explained by climate variables compared to other growth indexes. In contrast, Eureka climate response functions for the flower production chronology were able to explain large portions of the variability in flower number, suggesting that flower production may contain large amounts of climate information. Experimental increases in growing temperatures at Alexandra Fiord also showed a strong response of Cassiope flower production to warming (Chapter 2), providing additional evidence that flower production for this species is highly sensitive to variations in climate. However, significant relations between flower production and growth indicate that records of previous growth may be needed to calibrate the response of flower production to variations in climate. A l l of the variables measured show strong correlations with July temperatures at Alexandra Fiord, indicating that Cassiope tetragona may exhibit a stronger response to climate variations than the relationships exhibited with Eureka climate records would suggest. However, the limited extent of climate records from Alexandra Fiord prevents a more explicit test of that hypothesis at this time. Data obtained from this study was, however, able to provide an independent test of the climate transfer function developed by Havstrom et al. (1995) using measurement of leaf and flower numbers of C. tetragona at Alexandra Fiord. Verification of the Havstrom et al. model supports the use of 116 chronologies of Cassiope growth and reproduction to generate proxy climate data. Discrepancies between observed and predicted melting degree days over the 9-year test period appear to be largely the result of differences in standard deviations of the populations used in calibration and verification of the model, and the use of unstandardized chronologies. The general trend towards under-prediction of observed temperatures suggests caution must be used when interpreting predictions based on unstandardized chronologies. Two additional climate transfer functions based on standardized indices of Cassiope growth and reproduction were developed here for Alexandra Fiord. Because the three parameters measured for Cassiope tetragona in this study were strongly inter-correlated, the transfer functions each use data from only one chronology. Use of flowering index values from both the current and following year provides a very strong predictive model for July melting degree days at Alexandra Fiord. A weaker relationship is apparent in the July average temperature transfer function, which is based only on leaf number index values for the current year. It is hoped that future study of the application of dendroclimatology-type analysis to Cassiope tetragona will result in the verification and application of these models. 117 5. Conclusions Observations of vegetative growth and phenology of Cassiope tetragona under natural and experimental variations in growing season temperature and growing season length indicate that C. tetragona exhibits a conservative growth strategy in response to short-term variations in climate. Vegetative production shows little response to changes in growing season length resulting from variations in snowmelt timing, and only moderate responses to variations in growing season temperatures. This relative insensitivity to changes in growing season climate may be partially caused by constraints on the portion of the growing season utilized for active growth. Although vegetative growth began earlier under conditions of warmer temperatures and earlier snowmelt, symmetrical changes in growth cessation resulted in a shift in the period of active growth, rather than a change in the total growth period. Similar patterns have been observed for Cassiope tetragona in previous studies at Devon Island (Svoboda, 1977) and Alexandra Fiord (Nams and Freedman, 1987b). This phenotypic rigidity may limit the sensitivity of vegetative production to short-term variations in climate by restricting the degree to which Cassiope plants may capitalize on periods of ameliorated growing season conditions (Nams and Freedman, 1987b; Shaver and Kummerow, 1992). Thus, moderate increases in growth observed in response to warmer summer temperatures appear to be the result of increases in growth rate and not changes in the length of the growing period. In contrast to this conservative growth strategy, sexual reproduction of C. tetragona appears to be highly responsive to increases in growing season temperatures. Warmer temperatures resulted in increased production and accelerated development of reproductive structures, and improved seed quality. Timing of snowmelt appears to be less important than summer temperatures in determining successful seed production, although the short 118 duration of the growing season may be an important factor limiting fruit maturation in late-lying snowbed communities (Nams and Freedman, 1987b; Kudo, 1991; Molau, 1993). The contrasts between patterns of growth and reproduction of C. tetragona in response to variations in climate may indicate important internal resource allocation strategies for this species. In general, it appears that the potential for increased productivity during favorable growing seasons is directed towards sexual reproduction, rather than vegetative growth. Increased production of reproductive buds in one year also appears to decrease the potential for vegetative growth in the following year, possibly because of a further need for resource allocation to the development of previously initiated flower buds. Internal resource partitioning may explain the high levels of variation observed in growth and reproduction between individual shoots of the same plant; such patterns of resource allocation may be important in allowing long-lived individuals a flexible response to changes in micro-environmental conditions. The allocation strategies outlined above have important implications for predicting the short- and long-term response of C. tetragona to changes in temperature and growing season length expected with current rapid changes in greenhouse gases. If climate change results in warmer summer temperatures in the High Arctic, observations in this study indicate that sexual reproduction of Cassiope tetragona to is likely increase substantially over the short term. Increases in seed production and seed quality may result in greater seedling establishment and colonization of new areas. The conservative growth strategy exhibited by Cassiope tetragona indicates that primary production is unlikely to change substantially in response to ameliorated growing season conditions. Such a response could increase the vulnerability of Cassiope individuals to the effects of competition in communities with species more capable of rapid growth, and C. tetragona may eventually be excluded from competitive habitats. Thus, it is possible that the long-term result of increases in growing season temperatures may be a shift in species distribution for Cassiope tetragona Long-term changes in productivity and community dynamics are likely 119 to be sensitive to indirect effects of climate change (e.g. Chapin, 1984; Shaver and Kummerow, 1992), however, and are difficult to predict from the results of this study. In addition to predictions of future responses, the results of this research may be important to interpreting past responses of C. tetragona to climate variations. Use of internode lengths for delimiting annual growth may allow for the development of comparatively long chronologies of Cassiope growth and reproduction. Statistical relations between climate and growth and reproductive chronologies of Cassiope tetragona at Alexandra Fiord indicate this species may be very useful for generating proxy climate data, as well as investigating ecological relationships. Additional research on the effects of community dynamics, within-plant resource partitioning and allocation strategies is likely to greatly improve our ability to interpret past variations in the production of Cassiope tetragona. 120 6. References cited Atmospheric Environment Service, 1982: Canadian Climate Normals, 1951-1980, Vol.1 and 2. Downsview, Ontario. Atmospheric Environment Service, 1980-1992: Monthly weather record (continuing periodical). Downsview, Ontario. Bell, K. L. and Bliss, L. C, 1979: Autecology of Kobresia bellardii: why winter snow limits local distribution. Ecological Monographs, 49: 377-402. Benedict, J. B., 1990: Lichen mortality due to late-lying snow: results of a transplant study. Arctic and Alpine Research, 22: 81-89. Bergsma, B. M, Svoboda, J. and Freedman, B., 1984: Entombed plant communities released by a retreating glacier at Central Ellesmere Island, Canada. Arctic, 37:48-52. Beschel, R. E. and Webb, D., 1962: Growth ring studies on arctic willows. In: Muller, F. (ed.), Axel Heiberg Island Research Reports: preliminary report. Montreal: McGill University, 189-198. Billings, W. D., 1987: Constraints to plant growth, reproduction, and establishment in arctic environments. Arctic and Alpine Research, 19: 357-365. Billings W. D. and Bliss, L. C, 1959: An alpine snowbank environment and its effects on vegetation, plant development, and productivity. Ecology, 40: 388-397. Billings, W. D. and Mooney, H. A., 1968: The ecology of arctic and alpine plants. Biological Review, 43: 481-529. Bliss, L. C, 1956: Comparison of plant development in microenvironments of arctic and alpine tundras. Ecological Monographs, 26: 303-337. Bliss, L. C, 1962: Adaptations of arctic and alpine plants to environmental conditions. Arctic, 15: 117-144. Bliss, L. C, Kerik, J. and Peterson, W, 1977: Primary production of dwarf shrub heath communities, Truelove Lowland. In: Bliss, L.C. (ed.), Truelove Lowland, Devon Island, Canada: A High Arctic Ecosystem. Edmonton: University of Alberta Press, 217-224. Bliss, L.C. and Peterson, K.M., 1992: Plant succession, competition, and the physiological constraints of species in the Arctic. In: Chapin, F. S., Jefferies, R.L., Reynolds, J.F., Shaver, G. R. and Svoboda, J. (eds.), Arctic Ecosystems in a Changing Climate: an ecophysiological perspective. Toronto: Academic Press, 111-136. Bliss, L.C. and Matvayeva, N. V., 1992: Circumpolar arctic vegetation. In: Chapin, F. S., Jefferies, R.L., Reynolds, J.F., Shaver, G. R. and Svoboda, J. (eds.), Arctic Ecosystems in a Changing Climate: an ecophysiological perspective. Toronto: Academic Press, 59-89. 121 Brassard, G. R. and Beschel, R. E., 1968: The vascular flora of Tanquary Fiord, Northern Ellesmere Island, N.W.T. Canadian Field Naturalist, 84: 357-364. Briffa, K. and Jones, P. D., 1992: Basic chronology statistics and assessment. In: Cook, E. R. and Kariukstis, L. A. (eds.). Metlvods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic, 137-152. Briffa, K. R., Jones, P. D., Pilcher, J. R. and Hughes, M. K., 1988: Reconstructing summer temperatures in northern Fennoscandinavia back to A.D. 1700 using tree-ring data from Scots pine. Arctic and Alpine Research, 20: 385-394. Callaghan, T. V., 1973: A comparison of the growth of tundra plant species at several widely spaced sites. Institute for Terrestrial Ecology, Merlewood Research and Development Paper, No. 53. Callaghan, T. V., 1988: Physiological and demographic implications of modular construction in cold environments. In: Davy, A. J., M. J. Hutchings and A. R. Watkinson (eds.), Plant Population Ecology. Oxford: Blackwell Scientific, 111-135. Callaghan, T. V., Carlsson, B. A. and Tyler, N. J. C, 1989: Historical records of climate-related growth in Cassiope tetragona from the Arctic. Journal of Ecology, 77:823-837. Callaghan, T. V. and Collins, N. J., 1976: Strategies of growth and population dynamics of tundra plants. I. Introduction. Oikos, 27: 383-388. Chapin, F. S., I l l , 1983: Direct and indirect effects of temperature on arctic plants. Polar Biology, 2: 47-52. Chapin, F. S., I l l , 1984: The impact of increased air temperature on tundra plant communities. In: McBeath, J. H. (ed.). The Potential Effects of Carbon Dioxide-Induced Climatic Changes in Alaska. Fairbanks: University of Alaska, 143-148. Chapin, F. S., I l l , 1987: Environmental controls over growth of tundra plants. Ecological Bulletins, 38: 69-76. Chapin, F. S., I l l , Bloom, A. J., Field, C. B. and Waring, R. H., 1987: Plant responses to multiple environmental factors. Bioscience, 37: 49-57. Chapin, F. S., Il l and Shaver, G. R., 1985: Individualistic growth response of tundra plant species to environmental manipulations in the field. Ecology, 66:564-576. Chapin, F. S., Il l and Shaver, G. R., 1989: Differences in growth and nutrient use among arctic plant growth forms. Functional Ecology, 3: 73-80. Conover, W. J. and Iman, R. L., 1981: Rank transformations as a bridge between parametric and nonparametric statistics. American Statistician, 35: 124-129. Cook, E. and Briffa, K., 1992: A comparison of some tree-ring standardization methods. In: Cook, E. R. and Kariukstis, L. A. (eds.). Methods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic, 153-162. Cook, E. R. and Kariukstis, L. A. (eds.), 1992: Methods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic. 122 Cook, E., Briffa, K., Shiyatov, S. and Mazepa, V., 1992: Tree-ring standardization and growth trend estimation. In: Cook, E. R. and Kariukstis, L. A. (eds.). Methods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic, 104-123. Cropper, J. P., 1982: Climate reconstructions (1801 to 1938) inferred from tree-ring width chronologies of the North American Arctic. Arctic and Alpine Research, 14: 223-241. Cropper, J. P. and Fritts, H. C, 1981: Tree-ring width chronologies from the North American Arctic. Arctic and Alpine Research, 13: 245-260. Etkin, D., 1989: Greenhouse warming: consequences for arctic climate. Journal of Cold Regions Engineering, 4:54-66. Ferguson, A., 1992: Growth rates of Cassiope tetragona and climate pattern in central Ellesmere Island (79° N). Unpublished manuscript, Botany Department, Erindale College, University of Toronto. Freedman, B., Hill, N., Svoboda, J. and Henry, G., 1982: Seedbanks and seedling occurrence in a high Arctic oasis at Alexandra Fiord, Ellesmere Island, Canada. Canadian Journal of Botany, 60: 2112-2118. Freedman, B., Svoboda, J., Labine, C, Muc, M., Henry, G., Nams, M., Stewart, J. and Woodley, E., 1983: Physical and ecological characteristics of Alexandra Fiord, a high arctic oasis on Ellesmere Island, Canada. In: Permafrost: Fourth International Conference Proceedings, July 17-22, 1983. Washington, D.C.: National Academic Press, 301-306. Fritts, H. C, 1976: Tree Rings and Climate. London, New York and San Francisco: Academic Press. Galen, C. and Stanton, M. L., 1993: Short-term responses of alpine buttercups to experimental manipulations of growing season length. Ecology, 74: 1052-1058. Haag, R. W., 1974: Nutrient limitations to plant production in two tundra communities. Canadian Journal of Botany, 52: 103-116. Henry, G. H. R., Freedman, B. and Svoboda, J., 1986: Effects of fertilization on three tunra plant communities of a polar desert oasis. Canadian Journal of Botany, 64: 2502-2507. Havstrom, M., Callaghan, T. V. and Jonasson, S., 1993: Differential growth responses of Cassiope tetragona, an arctic dwarf-shrub, to environmental perturbations among three contrasting high- and subarctic sites. Oikos, 66: 389-402. Havstrom, M., Callaghan, T. V., Jonasson, S. and Svoboda, J., 1995: Little Ice Age temperature reduction measured by reduced growth of an arctic heather. Functional Ecology, in press. Holway, J. G. and Ward, R. T., 1963: Snow and meltwater effects in an area of Colorado alpine. American Midland Naturalist, 69: 198-197. 123 Hulten, E., 1968: Flora of Alaska and Neighboring Territories. Stanford: Stanford University Press. Inouye, D. W. and McGuire, A. D., 1991: Effects of snowpack on timing and abundance of flowering in Delphinium nelsonii (Ranunculaceae): Implications for climate change. American Journal of Botany, 78: 997-1001. Kay, P., 1978: Dendroecology in Canada's forest tundra transition zone. Arctic and Alpine Research, 10: 133-138. Kershaw, K. A., 1962: Quantitative ecological studies from Landmannahellir, Iceland. II. The rhizome behaviour of Carex bigelowii and Calamagrostis neglecta. Journal of Ecology, 50: 171-179. Kolishchuk, V. G., 1992: Dendroclimatological study of prostrate woody plants. In: Cook, E. R. and Kariukstis, L. A. (eds.). Methods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic, 51-55. Kudo, G., 1991: Effects of snow-free period on the phenology of alpine plants inhabiting snow patches. Arctic and Alpine Research, 23: 436-443. Kuivinen, K. C. and Lawson, M. P., 1982: Dendroclimatic analysis of birch in south Greenland. Arctic and Alpine Research, 14: 243-250. Labine, C, 1994: Meteorology and climatology of the Alexandra Fiord lowland. In: Svoboda, J. and Freedman, B. (eds.), Ecology of a Polar Oasis: Alexandra Fiord, Ellesmere Island, Canada. Toronto: Captus Press, 23-39. Larigauderie, A. and Kummerow, J., 1991: The sensitivity of phenological events to changes in nutrient availability for several plant growth forms in the arctic. Holarctic Ecology, 14: 38-44. Marion, G. M., Henry, G. H. R., Molgaard, P., Oechel, W. C, Jones, M. H. and Vourlitis, G. L., 1993: Open-top devices for manipulating field temperatures in tundra ecosystems. In: Lunardini, V. J. and Bowen, S. L. (eds.). Proceedings of the Fourth International Symposium on Tfiermal Engineering and Science for Cold Regions, 28 Sept.-1 October 1993. CRREL Special Report 93-22. Hanover, N. H.: U.S. Army Cold Regions Research and Engineering Laboratory, 205-210. Marion, G. M., Henry, G. H. R., Freckman, D. W., Johnstone, J., Jones, G., Jones, M. H., Molau, U., Molgaard, P., Parsons, A. N., Svoboda, J., and Virginia, R. A., 1995, In preparation: Open-top designs for manipulating field temperature in high-latitude ecosystems. Maxwell, B., 1992: Arctic climate: potential for change under global warming. In: Chapin, F. S., I l l , Jefferies, R. L., Reyolds, J. F., Shaver, G. R. and Svoboda, J. (eds.). Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective. San Diego: Academic Press, 11-34. McGraw, J. B., 1985: Experimental ecology of Dryas octopetala ecotypes. III. Environmental factors and plant growth. Arctic and Alpine Research, 17: 229-239. 124 McGraw, J. B., and Antonovics, J., 1983: Experimental ecology of Dryas octopetala ecotypes. Ecotypic differntiation and life-cycle stages of selection. Journal of Ecology, 71: 879-897. Miller, P. C, 1982: Environmental and vegetational variation across a snow accumulation area in montane tundra in central Alaska. Holarctic Ecology, 5: 85-98. M0lgaard, P., 1982: Temperature observations in high arctic plants in relation to microclimate in the vegetation of Peary Land, North Greenland. Arctic and Alpine Research, 14: 105-115. Molau, U. (ed.), 1993a: International Tundra Experiment Manual. Copenhagen: Danish Polar Center. Molau, U., 1993b: Relationships between flowering phenology and life history strategies in tundra plants. Arctic and Alpine Research, 25: 391-402. Muc, M., Freedman, B. and Svoboda, J., 1989: Vascular plant communities of a polar oasis at Alexandra Fiord (79°N), Ellesmere Island, Canada. Canadian Journal of Botany, 67: 1126-1136. Muc, M., Svoboda, J. and Freedman, B., 1994: Soils of an extensively vegetated polar desert oasis, Alexandra Fiord, Ellesmere Island. In: Svoboda, J. and Freedman, B. (eds.) Ecology of a Polar Oasis: Alexandra Fiord, Ellesmere Island, Canada. Toronto: Captus Press, 41-50. Murray, C. and Miller, P. C, 1982: Phenological observations of major plant growth forms and species in montane and Eriophorum vaginatum tussock tundra in central Alaska. Holarctic Ecology, 5: 109-116. Nadelhoffer, K. J., Giblin, A. E., Shaver, G. R. and Linkins, A. E., 1992: Microbial processes and plant nutrient availability in arctic soils. In: Chapin, F. S., I l l , Jefferies, R. L., Reyolds, J. F., Shaver, G. R. and Svoboda, J. (eds.). Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective. San Diego: Academic Press, 281-319. Nams, M. L. N., 1982: Ecology of Cassiope tetragona at a High Arctic lowland, Alexandra Fjord, Ellesmere Island, N.W.T. M.Sc. Thesis, Dalhousie University. Nams, M. L. N. and Freedman, B., 1987a: Ecology of heath communities dominated by Cassiope tetragona at Alexandra Fiord, Ellesmere Island, Canada. Holarctic Ecology, 10: 22-32. Nams, M. L. N. and Freedman, B., 1987b: Phenology and resource allocation in a high arctic evergreen dwarf shrub, Cassiope tetragona. Holarctic Ecology, 10: 128-136. Neter, J., Wasserman, W. and Kutner, M. H., 1990: Applied Linear Statistical Models (3rd ed.). Boston: Irwin. Oberbauer, S. and Miller, P. C, 1979: Plant water relations in montane and tussock tundra vegetation types in Alaska. Arctic and Alpine Research, 11: 69-81. 125 Porsild, M. P., 1920: Flora of Disko Island and West Greenland from 66°-71°N latitude. With remarks on phytogeography, ecology, flowering, fruitfication, and hibernation. Meddelelser om Gr0nland, 58: 1-156. Potvin, C. and Roff, D. A., 1993: Distribution-free and robust statistical methods: viable alternatives to parametric statistics? Ecology, 74: 1617-1628. SAS, 1985: User's Guide: Statistics. Version 5 edition. Cary, North Carolina: SAS Institute. Schweingruber, F. H., Kairiukstis, L. and Shiyatov, S., 1992: Sample selection. In: Cook, E. R. and Kariukstis, L. A. (eds.). Methods of Dendrochronology: Applications in the Environmental Sciences. Boston: Kluwer Academic, 23-35. Shaver, G. R. and Chapin, F. S., I l l , 1980: Response to fertilization by various plant growth forms in an Alaskan tundra: nutrient accumulation and growth. Ecology, 61: 662-675. Shaver, G. R., Chapin, F. S., Ill and Gartner, B. L., 1986: Factors limiting seasonal growth and peak biomass accumulation of Eriophorum vaginatum in Alaskan tussock tundra. Journal of Ecology, 74: 257-278. Shaver, G. R. and Kummerow, J., 1992: Phenology, resource allocation and growth of arctic vascular plants. In: Chapin, F. S., I l l , Jefferies, R. L., Reyolds, J. F., Shaver, G. R. and Svoboda, J. (eds.). Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective. San Diego: Academic Press, 193-221. S0rensen, T., 1941: Temperature relations and phenology of the Northeast Greenland flowering plants. Meddelelser om Gr0nlatid, 125(9): 1-305. Svoboda, J., 1992: Personal communication, August, 1992. Department of Botany, University of Toronto, Ontario, Canada. Svoboda, J., 1977: Ecology and primary production of Raised Beach communities, Truelove Lowland. In: Bliss, L. C. (ed.). Truelove Lowland, Devon Island, Canada: A High Arctic Ecosystem. Edmonton: University of Alberta. Svoboda, J. and Henry, G. H. R., 1987: Succession in marginal arctic environments. Arctic and Alpine Research, 19: 373-384. Tilman, D., 1988: Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton, New Jersey, USA. Walker, D. A., 1987: Height and growth rings of Salix lanata ssp. richardsonii along the coastal temperature gradient of northern Alaska. Canadian Journal of Botany, 65: 988-993. Welker, J. M., Wookey, P. A., Parsons, A. N, Press, M. C, Callaghan, T. V. and Lee, J. A., 1993: Leaf cabon isotope discrimination and vegetative responses of Dryas octopetala to temperature and water manipulations in a High Arctic polar semi-desert, Svalbard. Oecologia, 95: 463-469. 126 Warming, E., 1908: The structure and biology of arctic flowering plants. I. Ericacinae. Morphology and biology. Meddelelser om Grinland, 36:1-71. Warren Wilson, J., 1957: Observations on the temperatures of arctic plants and their environment. Journal of Ecology, 45:499-531. Warren Wilson, J., 1964: Annual growth of Salix arctica in the High-Arctic. Annals of Botany, 28:71-76. Warren Wilson, J., 1966: An analysis of plant growth and its control in Arctic environments. Annals of Botany, 30: 383-402. Webber, P. J., 1978: Spatial and temporal variation of the vegetation and its production, Barrow, Alaska. In: Tieszen, L. L. (ed.), Vegetation and Production Ecology of an Alaskan Arctic Tundra. New York: Springer-Verlag, 37-112. Woodley, E. J. and Svoboda, J. 1994. Effects of habitat on variations of phenology and nutrient concentration among four common plant species of the Alexandra Fiord lowland. In: Svoboda, J. and Freedman, B. (eds.), Ecology of a Polar Oasis: Alexandra Fiord, Ellesmere Island, Canada. Toronto: Captus Press, 157-175. Wookey, P. A., Parsons, A N., Welker, J. M., Potter, J. A., Callaghan, T. V., Lee, J. A. and Press, M. C, 1993: Comparative responses of phenology and reproductive development to simulated environmental change in sub-arctic and high arctic plants. Oikos, 67: 490-502. Zar, J. H., 1984: Biostatistical Analysis. Englewood Cliffs, New Jersey: Prentice Hall. 127 7. Appendices Appendix 1: Within-plot sample sizes for observations on tagged shoots of Cassiope tetragona at the Cassiope and Dryas ITEX sites and Beach Ridge snow manipulation plots in 1992 and 1993. Appendix Table 1.1: Within-plot sample sizes for observations on tagged shoots of Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1992 and 1993. A total of 5 shoots were originally tagged within each plot at the Dryas ITEX site at the beginning of each field season, and 10 shoots were tagged within plots at the Cassiope ITEX site, except for plots 13 and 14, which each had 20 tagged shoots. Cassiope ITEX Dryas ITEX Cassiope ITEX Dryas ITEX 1992 1992 1993 1993 Plot ore Control ore Control ore Control ore Control 1 9 8 4 4 9 5 4 4 2 9 7 4 1 9 9 5 2 3 6 9 3 5 10 10 5 6 4 9 10 2 2 10 9 5 3 5 6 9 3 2 8 8 4 4 6 10 10 3 1 10 8 3 3 7 9 7 2 • 2 9 4 2 4 8 9 7 2 3 8 8 5 3 9 8 10 0 3 7 9 3 5 10 8 8 5 4 6 9 2 4 11 5 7 3 3 9 6 4 3 12 4 10 3 4 9 9 4 3 13 19 17 2 3 19 15 3 3 14 16 18 5 2 19 16 3 1 Appendix Table 1.2: Within-plot sample sizes for observations on tagged shoots of Cassiope tetragona at the Beach Ridge snowbed in 1992 and 1993. Treatment Block 1992 1993 Control 1 17 12 2 15 14 3 10 11 Snow addition 1 14 14 2 16 10 3 9 11 Snow removal 1 14 15 2 18 18 3 7 18 128 Appendix Table 1.3: Summary of analysis of variance to test for site, year and treatment effects on the proportion of tagged shoots removed from the data set at the Cassiope and Dryas ITEX sites in 1992 and 1993. Source df Type III SS MS F p > F year 1 4.723 4.723 2.47 0.1191 site 1 0.438 0.438 0.23 0.6335 treatment 1 1.509 1.509 0.79 0.3765 year x site 1 2.580 2.580 1.35 0.2481 year x trmt 1 8.580 8.580 4.49 0.0366* site x trmt 1 0.009 0.009 0.00 0.9457 year x site x trmt 1 4.723 4.723 2.47 0.1191 error 104 198.92 1.913 total 111 221.49 129 Appendix 2: Progression of snow melt in OTC and control plots at the Cassiope and Dryas ITEX sites in 1993. Appendix Table 2.1: Mean snow pack depth (cm) ± 1 standard deviation in plots at the Cassiope and Dryas ITEX sites, by date of measurement (n = 14). Site Treatment 155 158 160 162 C. I. OTC 17.2 ±11.4 9.7 ±10.2 4.7 ± 7.5 1.2 ±3.1 Control 15.6 ±11.8 8.4 ± 10.2 4.2 ± 7.7 1.3 ± 3.3 D. I. OTC 13.3 ±7.1 3.9 ±4.6 0.2 ± 0.6 0.0 ± 0.0 Control 9.6 ±11.7 4.5 ± 8.3 1.8 ±5.1 0.1+0.5 Appendix Table 2.2: Summary of repeated measures analysis of variance to test for site, treatment and date effects on snow pack depth (cm) at the Cassiope and Dryas ITEX sites. Source df SS MS F p>F* site 1 2923.15 2923.15 22.01 0.0001*** treatment 1 88.37 88.37 0.67 0.4184 site x trmt 1 135.11 135.11 1.02 0.3178 error 52 6906.17 132.91 date 4 10943.22 2735.80 119.09 0.0001*** date x site 4 1392.73 348.18 15.16 0.0001*** date x trmt 4 156.39 39.10 1.70 0.1889 date x site x trmt 4 48.38 12.09 0.53 0.5841 error (date) 208 4778.38 22.97 For within-subject effects (date), probabilities are Huynh-Feldt estimates. 130 Appendix 3: Timing of snow release in OTC and control plots at the Cassiope and Dryas ITEX sites in 1993. Appendix Table 3.1: Mean dates (day number in calendar year) ± 1 standard deviation of snow release in plots at the Cassiope and Dryas ITEX sites (n = 14). Site Treatment Date of snow release Cassiope ITEX OTC 162 ±1.8 Control 162 ± 3.5 Dryas ITEX OTC 160 ± 0.8 Control 157 ±3.1 Appendix Table 3.2: Summary of analysis of variance to test for site and treatment effects on date of snow release at the Cassiope and Dryas ITEX sites. Source df SS MS F p > F site 1 24.45 24.45 3.76 0.0001*** treatment 1 154.45 154.45 23.78 0.0578 site x treatment 1 11.16 11.16 1.72 0.1957 error 52 337.79 6.50 total 55 527.84 131 Appendix 4; Progression of active layer thaw in OTC and control plots at the Cassiope and Dryas ITEX sites in 1993. Appendix Table 4.1: Mean active layer depth (cm) ± 1 standard deviation in plots at the Cassiope and Dryas ITEX sites, by date of measurement (n = 14). Site Treatment 158 160 162 164 166 205 221 C.I. OTC 0.0 ± 0.5 ± 4.7 ± 7.8 ± 12.4 ± 48.8 ± 54.9 ± 0.0 1.4 2.8 4.1 4.2 3.8 3.2 Control 0.8 ± 2.3 ± 4.2 ± 8.5 ± 12.7 ± 47.0 ± 50.1 ± 2.1 4.1 5.2 5.7 1.5 5.6 6.0 D.I. OTC 0.5 ± 8.0 ± 13.2 ± 16.8 ± 22.4 ± 54.5 ± 59.7 ± 2.0 3.4 5.9 3.4 2.6 4.1 5.2 Control 5.1 ± 9.3 ± 12.7 ± 17.4 ± 22.6 ± 54.7 ± 60.9 ± 4.7 5.4 5.2 5.2 4.4 2.9 3.5 Appendix Table 4.2: Summary of repeated measures analysis of variance to test for site, treatment and date effects on active layer depth (cm) at the Cassiope and Dryas ITEX sites. Source df SS MS F p>F* site 1 5304.35 5304.35 88.02 0.0001*** treatment 1 7.43 7.43 0.12 0.7266 site x trmt 1 59.26 59.26 0.99 0.3255 error 52 3127.29 60.14 date 6 169608.16 28268.03 2554.73 0.0001*** date x site 6 498.17 83.03 7.50 0.0003*** date x trmt 6 187.78 31.30 2.83 0.0510* date x site x trmt 6 128.11 21.35 1.93 0.1387 error (date) 312 3452.27 11.06 * For within-subject effects (date), probabilities are Huynh-Feldt estimates. Appendix Table 4.3: Summary of analysis of variance to test for site and treatment effects on maximum active layer depth (sampling date 221) at the Cassiope and Dryas ITEX sites. Source df SS MS F p > F site 1 46.45 46.45 2.17 0.1466 treatment 1 840.88 840.88 39.31 0.0001*** site x treatment 1 123.02 123.02 5.75 0.0201* error 52 1112.21 21.39 total 55 2122.55 132 Appendix 5: Growing season soil moisture levels (measured as percent of dry weight) in OTC and control plots at the Cassiope and Dryas ITEX sites in 1993. Appendix Table 5.1: Mean soil moisture (% dry weight) ± 1 standard deviation in plots at the Cassiope and Dryas ITEX sites, by date of measurement (n = 8). Site Treatment 166 176 188 200 216 C.I. OTC 132 ±112 60 ±40 112 ±40 74 ±56 40 ±24 Control 176 ± 87 77 ±63 77 ±63 67 ±38 48 ± 19 D.I. OTC 168 ± 73 104 ±57 50 ±29 91 ±50 66 ±43 Control 171 ± 107 113 ±57 109 ± 65 84 ±36 42 ±24 Appendix Table 5.2: Summary of repeated measures analysis of variance to test for site, treatment and date effects on soil moisture levels (% dry weight) at the Cassiope and Dryas ITEX sites. Source df SS MS F p>F* site 1 7418.13 7418.13 1.72 0.2004 treatment 1 1792.18 1792.18 0.42 0.5245 site x trmt 1 60.12 60.12 0.01 0.9068 error 28 120785.37 4313.76 date 4 221824.39 55456.10 15.56 0.0001*** date x site 4 12556.11 3139.03 0.88 0.4540 date x trmt 4 6350.03 1587.51 0.45 0.7200 date x site x trmt 4 23251.21 5812.80 1.63 0.1887 error (date) 112 399212.02 3564.39 * For within-subject effects (date), probabilities are Huynh-Feldt estimates. Appendix Table 5.3: Summary of analysis of variance to test for site and treatment effects on soil moisture levels (% dry weight) at the Cassiope and Dryas ITEX site on day 188. Source df SS MS F p > F site 1 1852.27 1852.27 0.70 0.4099 treatment 1 1176.85 1176.85 0.44 0.5103 site x treatment 1 17610.01 17610.01 6.65 0.0154* error 28 74102.69 2646.52 total 31 94741.83 133 Appendix 6: Timing of vegetative phenophases of Cassiope tetragona, observed in OTC and control plots at the Cassiope and Dryas ITEX sites during the 1992 and 1993 growing seasons. Appendix Table 6.1: Mean date (day number in calendar year) ± 1 standard deviation of bud break, growth initiation and growth cessation in temperature manipulation plots at the Cassiope and Dryas ITEX sites in 1992 and 1993. Growth Growth Site Treatment Year Bud break initiation cessation Cassiope ITEX OTC 1992 183 ± 0.4 191 ± 0.7 220 ± 0.6 1993 174 ± 0.5 185 + 0.5 213 ±0.9 Control 1992 184 ± 0.4 194 ± 0.9 220 ± 0.4 1993 173 ± 0.7 188 + 0.6 215 ±0.9 Dryas ITEX OTC 1992 184 ± 0.5 197 ± 1.4 n/a 1993 172 ± 0.8 184 ± 1.0 212 ± 1.1 Control 1992 186 ± 0.8 198 ± 1.4 n/a 1993 171 ± 0.7 184 ± 0.9 215 ± 1.0 Appendix Table 6.2: Summary of analysis of variance to test for site, year and treatment effects on the timing of vegetative bud break of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Data were-rank transformed prior to analysis. Source df Type III SS MS F p > F year 1 85427.71 85427.71 387.61 0.0001*** site 1 274.06 274.06 1.24 0.2674 treatment 1 488.22 488.22 2.22 0.1397 year x site 1 3588.87 3588.87 16.28 0.0001*** year x trmt 1 663.75 663.75 3.01 0.0857 site x trmt 1 4.27 4.27 0.02 0.8896 year x site x trmt 1 246.52 246.52 1.12 0.2927 error 103 22700.75 220.40 total 110 113430.50 Appendix Table 6.3: Summary of analysis of variance to test for site, year and treatment effects on the timing of growth initiation of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 77999.88 77999.88 311.21 0.0001*** site 1 118.68 118.68 0.47 0.4929 treatment 1 1728.91 1728.91 6.90 0.0099** year x site 1 6846.04 6846.04 27.31 0.0001*** year x trmt 1 0.03 0.03 0.00 0.9913 site x trmt 1 1070.04 1070.04 4.27 0.0413* year x site x trmt 1 119.31 119.31 0.48 0.4918 error 103 25815.40 250.64 total 110 113408.00 134 Appendix Table 6.4: Summary of analysis of variance to test for year and treatment effects on the timing of growth cessation of Cassiope tetragona at the Cassiope ITEX site. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 8184.45 8184.45 70.10 0.0001*** treatment 1 31.50 31.50 0.27 0.6057 year x treatment 1 189.45 189.45 1.62 0.2084 error 52 6071.61 116.76 total 55 14477.00 Appendix Table 6.5: Summary of analysis of variance to test for treatment effects on the timing of growth cessation of Cassiope tetragona at the Dryas ITEX site in 1993. Data were rank-transformed prior to analysis. Source df SS MS F p > F treatment 1 155.57 155.57 2.46 0.1291 error 26 1646.43 63.32 total 27 1802.00 Appendix Table 6.6: Summary of analysis of variance to test for year and treatment effects on the number of days from bud break to growth cessation for Cassiope tetragona at the Cassiope ITEX site in 1992 and 1993. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year treatment year x treatment error total 1 1 1 52 55 2511.16 80.16 216.07 11671.11 14478.50 2511.16 80.16 216.07 224.44 11.19 0.36 0.96 0.0015* 0.5527 0.3311 Appendix Table 6.7: Summary of analysis of variance to test for treatment effects on the number of days from bud break to growth cessation for Cassiope tetragona at the Dryas ITEX site in 1993. Data were rank-transformed prior to analysis. Source df SS MS F p > F treatment 1 • 175.00 175.00 2.77 0.1078 error 26 1640.50 63.10 total 27 1815.50 135 Appendix Table 6.8: Summary of analysis of variance to test for year and treatment effects on the number of days from growth initiation to growth cessation for Cassiope tetragona at the Cassiope ITEX site in 1992 and 1993. Source df Type III SS MS F p > F year 1 0.87 0.87 0.06 0.8090 treatment 1 54.02 54.02 3.64 0.0618 year x treatment 1 15.02 15.02 1.01 0.3188 error 52 770.64 14.82 total 55 840.55 Appendix Table 6.9: Summary of analysis of variance to test for treatment effects on the number of days from growth initiation to growth cessation for Cassiope tetragona at the Dryas ITEX site in 1993. Data were rank-transformed prior to analysis. Source df SS MS F p > F treatment 1 141.75 141.75 2.21 0.1492 error 26 1668.25 64.16 total 27 1810.00 Appendix Table 6.10: Summary of analysis of variance to test for site and treatment effects on the number of days from snow release to bud break for Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1993. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F site 1 1410.02 1410.02 5.92 0.0184* treatment 1 565.79 565.79 2.38 0.1293 site x treatment 1 33.02 33.02 0.14 0.7112 error 52 12384.68 238.17 total 55 14393.50 Appendix Table 6.11: Summary of analysis of variance to test for site and treatment effects on the number of days from snow release to initiation of active growth for Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1993. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F site 1 228.02 228.02 1.06 0.3083 treatment 1 2871.45 2871.45 13.33 0.0006*** site x treatment 1 126.00 126.00 0.59 0.4478 error 52 11199.04 215.37 total 55 14424.50 136 Appendix Table 6.12: Summary of analysis of variance to test for site and treatment effects on the number of days from snow release to growth cessation for Cassiope tetragona at the Cassiope and Dryas ITEX sites in 1993. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F site 1 1564.57 1564.57 7.09 0.0103** treatment 1 1244.57 1244.57 5.64 0.0213* site x treatment 1 257.14 257.14 1.17 0.2854 error 52 11477.21 220.72 total 55 14543.50 137 Appendix 7: Maximum shoot elongation of Cassiope tetragona observed at the end of the 1992 and 1993 growing seasons in OTC and control plots at the Cassiope and Dryas ITEX sites. Appendix Table 7.1: Mean maximum elongation (mm) ± 1 standard deviation observed in temperature manipulation plots at the Cassiope and Dryas ITEX sites, in 1992 and 1993. The dates of final measurements at each site are also given. Site Treatment Year Maximum shoot elongation Date of final measurement Cassiope ITEX OTC 1992 3.70 ± 0.21 224 1993 3.48 ±0.15 226 Control 1992 3.03 ± 0.15 224 1993 3.17 ±0.15 226 Dryas ITEX OTC 1992 3.24 ± 0.39 218 1993 3.68 ± 0.32 226 Control 1992 2.87 ± 0.25 218 1993 4.06 ± 0.32 226 Appendix Table 7.2: Summary of analysis of variance to test for site, year and treatment effects on maximum shoot elongation (mm) of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Data were logio-transformed prior to analysis. Source df Type III SS MS F p > F year 1 0.0006 0.0006 0.04 0.8352 site 1 0.0777 0.0777 2.29 0.0234* treatment 1 0.0224 0.0224 1.53 0.2196 year x site 1 0.0891 0.0891 6.07 0.0154* year x trmt 1 0.0406 0.0406 2.77 0.0993 site x trmt 1 0.0320 0.0320 2.18 0.1431 year x site x trmt 1 0.0058 0.0058 0.39 0.5324 error 103 1.5122 0.0147 total 110 1.7816 Appendix Table 7.3: Summary of analysis of variance to test for year and treatment effects on maximum shoot elongation (mm) of Cassiope tetragona at the Cassiope ITEX site only. Data were logio-transformed prior to analysis. Source df Type III SS MS F p > F year 1 0.0002 0.0002 0.03 0.8672 treatment 1 0.0622 0.0622 8.92 0.0043** year x treatment 1 0.0053 0.0053 0.77 0.3855 error 52 0.3628 0.0070 total 55 0.4306 138 Appendix Table 7.4: Summary of analysis of variance to test for year and treatment effects on maximum shoot elongation (mm) of Cassiope tetragona at the Dryas ITEX site only. Data were logio-transformed prior to analysis. Source df Type III SS MS F p > F year 1 0.1651 0.1651 7.33 0.0092** treatment 1 0.0013 0.0013 0.06 0.8089 year x treatment 1 0.0321 0.0321 1.43 0.2380 error 52 1.1494 0.0225 total 55 1.3505 139 Appendix 8: Timing and rates of peak shoot elongation (mm/day) observed for Cassiope tetragona in OTC and control plots at the Cassiope and Dryas ITEX sites in 1992 and 1993. Appendix Table 8.1: Timing and rates of peak shoot elongation (mm/day) of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Values are means ± 1 standard deviation. Site Treatment Year Date of peak elongation Peak elongation rate (mm/day) Cassiope ITEX OTC 1992 197 ± 7.9 0.15 ± 0.03 1993 192 ± 3.8 0.18 ± 0.03 Control 1992 202 + 8.1 0.13 ± 0.02 1993 193 + 4.4 0.29 ± 0.44 Dryas ITEX OTC 1992 197 ± 30.6 0.35 ± 0.48 1993 187 ± 1.9 0.25 + 0.09 Control 1992 207 ± 6.4 0.19 ± 0.06 1993 188 ± 3.0 0.27 ± 0.06 Appendix Table 8.2: Summary of analysis of variance to test for year, site and treatment effects on the timing of peak shoot elongation (day number in calendar year) of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 60207.56 60207.56 186.87 0.0001*** site 1 13.99 13.99 0.04 0.8354 treatment 1 1427.51 1427.51 4.43 0.0377* year x site 1 16507.23 16507.23 51.24 0.0001*** year x trmt 1 285.80 285.80 0.89 0.3485 site x trmt 1 109.41 109.41 0.34 0.5613 year x site x trmt 1 440.69 440.69 1.37 0.2449 error 103 33185.20 322.19 total 110 111842.50 Appendix Table 8.3: Summary of analysis of variance to test for year, site and treatment effects on rates of peak shoot elongation (mm/day) of Cassiope tetragona at the Cassiope and Dryas ITEX sites. Data were logio-transformed prior to analysis. Source df Type III SS MS F p > F year 1 0.6324 0.6324 55.47 0.0001*** site 1 0.2685 0.2685 23.55 0.0001*** treatment 1 0.0163 0.0163 1.43 0.2344 year x site 1 0.0006 0.0006 0.05 0.8153 year x trmt 1 0.0073 0.0073 0.64 0.4267 site x trmt 1 0.0532 0.0532 4.66 0.0331* year x site x trmt 1 0.0039 0.0039 0.34 0.5622 error 103 1.1784 0.0114 total 110 2.1625 140 Appendix 9: Timing of reproductive phenophases (day number in calendar year) of Cassiope tetragona, observed in OTC and control plots at the Cassiope and Dryas ITEX sites during the 1992 and 1993 growing seasons1. Appendix Table 9.1: Summary of analysis of variance to test for site, year and treatment effects on timing of flower bud emergence at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 14548.39 14548.39 104.13 0.0001*** site 1 28.48 28.48 0.20 0.6531 treatment 1 3212.96 3212.96 23.00 0.0001*** year x site 1 977.89 977.89 7.00 0.0102** year x trmt 1 564.28 564.28 4.04 0.0487* site x trmt 1 160.84 160.84 1.15 0.2873 year x site x trmt 1 4.72 4.72 0.03 0.8547 error 64 8941.38 139.71 total 71 30669.50 Appendix Table 9.2: Summary of analysis of variance to test for site, year and treatment effects on timing of peduncle elongation at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 11396.68 11396.68 95.91 0.0001*** site 1 2149.13 2149.13 18.09 0.0001*** treatment 1 3486.10 3486.10 29.34 0.0001*** year x site 1 0.003 0.003 0.00 0.9956 year x trmt 1 308.10 308.10 2.59 0.1123 site x trmt 1 27.96 27.96 0.24 0.6293 year x site x trmt 1 1009.93 1009.93 8.50 0.0049** error 64 7604.61 118.82 total 71 30869.50 Appendix Table 9.3: Summary of analysis of variance to test for site, year and treatment effects on timing of corolla opening at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 9746.67 9746.67 67.64 0.0001*** site 1 1598.95 1598.95 11.10 0.0014** treatment 1 3966.38 3966.38 27.52 0.0001*** year x site 1 392.76 392.76 2.73 0.1037 year x trmt 1 953.10 953.10 6.61 0.0125* site x trmt 1 367.15 367.15 2.55 0.1154 year x site x trmt 1 222.32 222.32 1.54 0.2187 error 64 9222.84 144.11 total 71 30649.50 1 The table of reproductive phenophase dates is included in the main text as Table 2.7. 141 Appendix Table 9.4: Summary of analysis of variance to test for site, year and treatment effects on timing of corolla drop at the Cassiope and Dryas ITEX sites. Data were rank-transformed prior to analysis. Source df Type III SS MS F p > F year 1 5101.69 5101.69 37.30 0.0001*** site 1 612.68 612.68 4.48 0.0388* treatment 1 2254.46 2254.46 16.48 0.0002*** year x site 1 2.56 2.56 0.02 0.8916 year x trmt 1 685.36 685.36 5.01 0.0293* site x trmt 1 704.07 704.07 5.15 0.0272* year x site x trmt 1 13.74 13.74 0.10 0.7525 error 55 7522.83 136.78 total 62 20727.50 Appendix Table 9.5: Summary of a Kruskal-Wallace test for treatment effects on the timing of fruit maturation at the Cassiope ITEX site in 1992. Sum of Expected Stan. Dev. Treatment n Scores under H 0 under Ho Mean Score OTC 5 19 20 2T07 3~8 Control 2 9 8 2XX7 4.5 Kruskal-Wallace test on Wilcoxon scores (rank sums): X 2 d f p>X2 0.23 1 0.6291 Appendix Table 9.6: Summary of a Kruskal-Wallace test for treatment effects on the timing of fruit maturation at the Dryas ITEX site in 1993. Sum of Expected Stan. Dev. Treatment n Scores under H G under HQ Mean Score OTC 4 1 2 3 14 2TT0 3 1 Control 2 8_15 7 2A0 4.3 Kruskal-Wallace test on Wilcoxon scores (rank sums): 0.51 1 0.4745 142 Appendix 10: Estimates of total number of flowers and live shoots of Cassiope tetragona per 10 cm 2 area and flower production per shoot in OTC and control plots at the Cassiope and Dryas ITEX sites in 1993. Appendix Table 10.1: Mean values ± 1 standard deviation of flower and live shoot numbers per 10 cm 2 area and flower production per shoot at the Cassiope and Dryas ITEX sites. Number of Number of Number of live flowers per live Site Treatment flowers shoots shoot Cassiope ITEX OTC 7.85 ± 3.20 15.81 +3.67 0.490 ±0.178 Control 2.31 ± 2.07 17.52 ± 5.88 0.144 ± 0.114 Dryas ITEX OTC 2.58 ± 1.81 7.75 ± 4.95 0.550 ± 0.670 Control 3.06 ± 4.06 11.33 ±6.18 0.213 ± 0.217 Appendix Table 10.2: Summary of analysis of variance of site and treatment effects on the number of flowers produced per shoot at the Cassiope and Dryas ITEX sites in 1993. Source df SS MS F p > F site 1 0.0502 0.0502 0.37 0.5454 treatment 1 1.4002 1.4002 10.36 0.0024** site x treatment 1 0.0002 0.0002 0.00 0.9676 error 44 5.9464 0.1351 total 47 7.3969 143 Appendix 11: Percentage of tagged shoots of Cassiope tetragona producing mature flowers (open corolla) in OTC and control plots at the Cassiope and Dryas ITEX sites in 1992 and 1993. Appendix Table 11.1: Percentage ± 1 standard deviation of tagged shoots producing mature flowers at the Cassiope and Dryas ITEX sites. Percentage of Year Site Treatment flowering shoots 1992 Cassiope ITEX OTC 16 ± 10.4 Control 16 ± 14.0 Dryas ITEX OTC 30 ± 39.3 Control 13 ± 23.8 1993 Cassiope ITEX OTC 27 ± 19.6 Control 12 ± 11.9 Dryas ITEX OTC 31 ±25.9 Control 17 ± 13.9 Appendix Table 11.2: Summary of analysis of variance for year, site and treatment on the percentage of tagged shoots producing mature flowers at the Cassiope and Dryas ITEX sites in 1992 and 1993. Source df Type III SS MS F p > F year 1 1837.96 1837.96 1.90 0.1706 site 1 6.88 6.88 0.01 0.9329 treatment 1 5109.58 5109.58 5.29 0.0234* year x site 1 769.01 769.01 0.80 0.3742 year x trmt 1 479.63 479.63 0.50 0.4825 site x trmt 1 10.90 10.90 0.01 0.9156 year x site x trmt 1 1023.01 1023.01 1.06 0.3057 error 103 99436.45 965.40 total 110 108761.00 144 Appendix 12: Germination and susceptibility to algal and fungal infestation of Cassiope tetragona seeds sampled from OTC and control plots at the Cassiope ITEX site in 1993. Appendix Table 12.1: Total percent of germinated and infested seeds and mean days to peak germination ± 1 standard deviation of seeds sampled from the Cassiope ITEX site in 1993 (n=5). Percent of Percent of infested Days to peak Treatment germinated seeds seeds germination OTC 58.84 ±2.32 0.78 ± 0.45 9.4 ± 0.4 Control 32.71 ± 11.51 41.75 ±13.11 17.6 ± 0.6 Appendix Table 12.2: Summary of analysis of variance to test for treatment effects on total percentage of germinating seeds. Source df SS MS F p > F treatment 1 1707.20 1707.20 24.77 0.0011** error 8 551.27 68.91 total 9 2258.47 Appendix Table 12.3: Summary of a Kruskal-Wallace test for treatment effects on the timing of peak seed germination. Sum of Expected Stan. Dev. Treatment n Scores under H 0 under Ho Mean Score OTC 5 TSTO 273 4A9 3~70 Control 5 _40 27.5 4.49 8.0 Kruskal-Wallace test on Wilcoxon scores (rank sums): x2 <|f p> x 2 7.7586 1 0.0053 Appendix Table 12.2: Summary of analysis of variance to test for treatment effects on total percentage of fungal- and algal-infected seeds. Source df SS MS F p > F treatment 1 4077.98 4077.98 47.39 0.0001*** error 8 688.40 86.05 total 9 4766.37 145 Appendix 13: Timing of vegetative phenophases (day number in calendar year) in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993. Appendix Table 13.1: Average date ± 1 standard deviation of vegetative bud break and initiation of growth in snow manipulation plots at the Beach Ridge snowbed in 1992. Date of bud Date of growth Treatment Block break initiation Control 1 192.5 ± 3.9 210.7 ± 3.6 2 193.1 ± 4.0 210.7 ± 3.6 3 193.7 ± 4.5 208.4 ± 3.0 Addition 1 192.7 ± 2.1 210.5 + 6.1 2 193.9 ± 3.9 214.6 ± 4.0 3 193.4 ± 3.4 213.0 ±5.1 Removal 1 190.9 ± 3.2 209.6 ± 5.5 2 190.6 ± 1.6 205.8 ±4.1 3 191.6 ± 4.2 207.1 ±3.8 Appendix Table 13.2: Average date ± 1 standard deviation of vegetative bud break, initiation of growth and growth cessation in snow manipulation plots at the Beach Ridge snowbed in 1993. Treatment Block Date of bud Date of growth Date of growth break initiation cessation Control 1 183.7 ± 2.7 196.2 ±4.1 216.1 ±4.6 2 186.6 ± 3.5 198.0 ± 4.2 219.0 ± 5.8 3 184.9 ± 4.0 197.0 ± 2.8 215.4 ± 4.5 Addition 1 185.7 ± 3.7 199.0 ± 4.2 220.3 ± 4.2 2 190.0 ± 4.6 199.7 + 4.2 218.4 ±3.4 3 191.1 ±3.1 201.6 ±3.9 221.7 ±4.1 Removal 1 175.2 ± 2.6 192.9 ± 5.9 212.3 ± 6.2 2 175.4 ± 2.6 189.6 ± 5.3 215.3 ± 6.4 3 174.2 ± 2.4 187.6 ± 4.4 215.2 ± 6.5 Appendix Table 13.3: Summary of regression analysis of the mean date of vegetative bud break observed in Beach Ridge snow manipulation plots in 1992 with date of snow release. Source df SS MS F-value p > F R 2 model 1 9.98 9.98 38.89 0.0004 0.8475 error 7 1.80 0.26 total 8 11.78 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 160.12 5.19 30.829 0.0001 snowfree date 1 0.18 0.03 6.236 0.0004 146 Appendix Table 13.4: Summary of regression analysis of the mean date of vegetative bud break observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R2 model 1 325.85 325.85 263.96 0.0001 0.9742 error 7 8.64 1.23 total 8 334.49 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 44.97 8.50 5.290 0.0011 snowfree date 1 0.79 0.05 16.247 0.0001 Appendix Table 13.5: Summary of regression analysis of the mean date of initiation of shoot elongation observed in Beach Ridge snow manipulation plots in 1992 with date of snow release. Source df SS MS F-value p > F R 2 model 1 37.12 37.12 11.17 0.0124 0.6147 error 7 23.27 3.32 total 8 60.39 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 147.63 18.69 7.900 0.0001 snowfree date 1 0.35 0.10 3.342 0.0124 Appendix Table 13.6: Summary of regression analysis of the mean date of initiation of shoot elongation observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 163.18 163.18 63.88 0.0001 0.9012 error 7 17.88 2.55 total 8 181.06 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 98.07 12.23 8.019 0.0001 snowfree date 1 0.56 0.07 7.993 0.0001 147 Appendix Table 13.7: Summary of regression analysis of mean date of growth cessation observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 50.47 50.47 17.663 0.0040 0.7162 error 7 20.00 2.86 total 8 70.47 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 162.77 12.93 12.583 0.0001 snowfree date 1 -0.69 0.07 -9.339 0.0001 Appendix Table 13.8: Summary of regression analysis of the mean number of days from snow release to vegetative bud break observed in Beach Ridge snow manipulation plots in 1992 with date of snow release. Source df SS MS F-value p > F R2 model 1 208.90 208.90 813.70 0.0001 0.9915 error 7 1.90 0.26 total 8 210.70 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 160.12 5.19 30.829 0.0001 snowfree date 1 -0.82 0.03 -28.525 0.0001 Appendix Table 13.9: Summary of regression analysis of the mean number of days from snow release to vegetative bud break observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 23.42 23.42 18.98 0.0033 0.7305 error 7 8.64 1.23 total 8 32.07 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 44.97 8.50 5.290 0.0011 snowfree date 1 -0.21 0.05 -4.356 0.0033 148 Appendix Table 13.10: Summary of regression analysis of the mean number of days from snow release to the start of shoot elongation observed in Beach Ridge snow manipulation plots in 1992 against date on snow release. Source df SS MS F-value p > F R 2 model 1 132.72 132.72 39.93 0.0004 0.8508 error 7 23.27 3.32 total 8 155.98 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 147.63 18.69 7.900 0.0001 snowfree date 1 -0.65 0.10 -6.319 0.0004 Appendix Table 13.11: Summary of regression analysis of the mean number of days from snow release to the start of shoot elongation observed in Beach Ridge snow manipulation plots in 1993 against date on snow release. Source df SS MS F-value p > F R 2 model 1 102.35 102.35 40.07 0.0004 0.8513 error 7 17.88 2.55 total 8 120.23 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 98.07 12.23 8.019 0.0001 snowfree date 1 -0.44 0.07 -6.330 0.0004 Appendix Table 13.12: Summary of regression analysis of the number of days from snow release to growth cessation observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 249.22 249.22 87.22 0.0001 0.9257 error 7 20.00 2.86 total 8 269.23 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 162.77 12.94 12.583 0.0001 snowfree date 1 -0.69 0.07 -9.339 0.0001 149 Appendix Table 13.13: Summary of regression analysis of the number of days from vegetative bud break to growth cessation observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 119.84 119.84 24.59 0.0016 0.7784 error 7 34.11 4.87 total 8 153.95 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 117.79 16.89 6.973 0.0001 snowfree date 1 -0.48 0.10 -4.959 0.0016 Appendix Table 13.14: Summary of regression analysis of the number of days from growth initiation to growth cessation observed in Beach Ridge snow manipulation plots in 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 32.15 32.15 4.467 0.0724 0.3896 error 7 50.38 7.20 total 8 82.83 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 64.70 20.53 3.152 0.0161 snowfree date 1 -0.25 0.12 -2.114 0.0724 150 Appendix 14: Maximum shoot elongation (mm) observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993. Appendix Table 14.1: Mean values ± 1 standard deviation of maximum shoot elongation (mm) observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993. Treatment Block 1992 maximum shoot elongation 1993 maximum shoot elongation Control 1 1.49 ± 0.43 2.50 ± 0.85 2 1.88 ± 0.76 3.16 ± 1.61 3 2.08 ± 0.63 2.73 ± 1.55 Addition 1 1.65 ± 0.65 2.51 ± 1.12 . 2 1.45 ± 0.70 2.88 ± 1.59 3 1.62 ± 0.65 3.16 ± 1.25 Removal 1 1.31 ± 0.38 2.42 ± 1.63 2 2.01 ± 0.45 3.82 ± 1.90 3 2.14 ± 0.75 3.96 ± 1.99 Appendix Table 14.2: Summary of regression analysis of logio values of maximum shoot elongation (mm) observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 against time of snow release. Source df SS MS F-value p > F R2 model 1 0.0596 0.0596 3.833 0.0679 0.1933 error 16 0.2490 0.0156 total 17 0.3087 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 1.7096 0.7134 2.396 0.0291 snowfree date 1 -0.0079 0.0040 -1.958 0.0679 Appendix Table 14.3: Summary of t-test between 1992 and 1993 values of maximum shoot elongation (mm) observed in snow manipulation control plots at the Beach Ridge snowbed. 1992 1993 n 3 3 mean 1.816 2.792 SS 0.181 0.228 calculated t t 0.05.2.4 -3.739 2.776 151 Appendix 15: Timing of reproductive phenophases in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993.2 Appendix Table 15.1: Summary of regression analysis of mean dates of flower bud appearance observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 239.06 239.06 24.04 0.0003 0.6490 error 13 129.27 9.94 total 14 368.33 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 87.41 20.86 4.190 0.0011 snowfree date 1 0.57 0.12 4.903 0.0003 Appendix Table 15.2: Summary of regression analysis of mean dates of flower peduncle elongation observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 582.49 582.49 14.66 0.0028 0.5714 error 11 436.94 39.72 total 12 1019.43 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 41.75 41.99 0.994 0.3415 snowfree date 1 0.90 0.23 3.829 0.0028 Appendix Table 15.3: Summary of regression analysis of mean dates of flower corolla drop observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 with date of snow release. Source df SS MS F-value p > F R 2 model 1 480.78 480.78 8.51 0.0140 0.4363 error 11 621.28 56.48 total 12 1102.06 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p>l t l intercept 1 63.16 50.08 1.261 0.2333 snowfree date 1 0.82 0.28 2.918 0.0140 2 Tables of mean dates of reproductive phenophases in the Beach Ridge snow manipulation plots are included in the main text. 152 Appendix Table 15.4: Summary of regression analysis of mean number of days from snow release to flower bud appearance observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 against date of snow release. Source df SS MS F-value p > F R2 model 1 92.17 92.17 4.432 0.0553 0.2543 error 13 270.32 20.79 total 14 362.48 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 74.33 30.17 2.464 0.0285 snowfree date 1 -0.36 0.17 -2.105 0.0553 Appendix Table 15.5: Summary of regression analysis of mean number of days from snow release to peduncle elongation observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 against date of snow release. Source df SS MS F-value p > F R 2 model 1 0.55 0.55 0.011 0.9173 0.0010 error 11 534.36 48.58 total 12 534.91 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 28.48 46.44 0.613 0.5522 snowfree date 1 -0.03 0.26 -0.106 0.9173 Appendix Table 15.6: Summary of regression analysis of mean number of days from snow release to flower corolla drop observed in snow manipulation plots at the Beach Ridge snowbed in 1992 and 1993 against date of snow release. Source df SS MS F-value p > F R 2 model 1 8.14 8.14 0.123 0.7324 0.0111 error 11 727.81 66.16 total 12 735.95 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 49.28 54.20 0.909 0.3827 snowfree date 1 -0.11 0.30 -0.351 0.7324 153 Appendix 16: Number of flowers and mature live shoots of Cassiope tetragona per 10x10 cm area and number of flowers produced per mature shoot, measured in snow manipulation plots at the Beach Ridge snowbed in 1993. Appendix Table 16.1: Mean values ± 1 standard deviation of flower and live shoot numbers per 10x10 cm quadrat and number of flowers produced per mature shoot. Number of Number of live Number of Treatment Block flowers shoots flowers per shoot Control 1 1.42 ± 2.0 29.5 ± 16 0.053 ± 0.07 2 1.58 ± 1.9 16.1 ± 12 0.101 ± 0.15 3 1.50 ± 2.4 25.0 ± 18 0.057 ± 0.08 Addition 1 0.00 ± 0.0 24.1 ±27 0.000 ± 0.0 2 1.08 ± 1.2 22.0 ± 8 0.060 ± 0.08 3 0.17± 0.4 10.8 ± 12 0.028 ± 0.05 Removal 1 0.00 ± 0.0 11.3 ±8 0.000 ± 0.0 2 0.67 ± 0.9 13.8 ± 14 0.201 ± 0.40 3 0.92 ± 1.6 8.9 ± 12 0.086 ± 0.06 Appendix Table 16.2: Summary of regression analysis of number of flowers produced per mature shoot in Beach Ridge snow manipulation plots in 1993 with date of snow release in the same year. Source df SS MS F-value p > F R2 model 1 0.0035 0.0035 0.907 0.3726 0.1147 error 7 0.0268 0.0038 total 8 0.0303 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 0.516 0.474 1.089 0.3122 snowfree date 1 -0.003 0.003 -0.952 0.3726 Appendix Table 16.3: Summary of regression analysis of number of flowers produced per mature shoot in Beach Ridge snow manipulation plots in 1993 with date of snow release in the previous year (1992). Source df SS MS F-value p > F R2 model 1 0.0038 0.0038 0.994 0.3520 0.1243 error 7 0.0265 0.0038 total 8 0.0303 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > l t l intercept 1 0.694 0.631 1.100 0.3079 snowfree date 1 -0.003 0.003 -0.997 0.3520 154 Appendix 17: Active layer depths (cm) measured in late July, 1993 in eight Cassiope-dominated communities at Alexandra Fiord. Appendix Table 17.1: Mean active layer depths (cm) ± 1 standard deviation sampled at eight Casswpe-dominated communities on July 25, 1993. Site Active layer depth RS zone 1 39.5 ±4.1 RS zone 2 30.3 ± 8.4 RS zone 3 23.4 ± 4.2 RS zone 4 31.6 ±2.2 RS zone 5 29.6 ± 2.6 Erratic site 28.1 ±8.4 Cassiope ITEX 47.0 ± 5.6 Dryas ITEX 54.7 ± 2.9 Appendix Table 2.17.2: Summary of analysis of variance of active layer depths in eight Cass/ope-dominated communities in 1993. Source df SS MS F p > F site 7 10944.30 1563.47 51.07 0.0001*** error 95 2908.64 30.62 total 102 13852.95 155 Appendix 18: Soil moisture levels (percentage of dry weight) measured over the course of the 1993 growing season in nine communities at Alexandra Fiord. Appendix Table 18.1: Mean soil moisture levels (% dry weight) ± 1 standard deviation, measured on four sampling dates in 1993 in nine communities. Site Day 176 Day 188 Day 200 Day 216 RS zone 1 28.9 ± 9.7 23.3 ± 4.7 15.9 ± 5.5 19.8 ± 6.6 RS zone 2 24.1 ±4.6 30.6 ± 18.9 27.3 ± 5.9 28.0 ± 14.3 RS zone 3 34.6 ± 4.2 37.4 ± 11.8 41.3 ± 22.5 28.6 ± 14.7 RS zone 4 60.3 ± 17.7 57.1 ± 25.2 47.2 ± 22.4 39.8 ± 21.2 RS zone 5 169.1 ±26.3 113.3 ± 41.9 85.5 ± 27.9 96.5 ± 42.0 Erratic site 23.4 ± 11.0 34.0 ± 14.2 23.2 ± 11.7 22.9 ± 15.0 Beach Ridge snowbed n/a 76.6 ± 48.9 80.9 ± 54.1 93.7 ± 60.3 Cassiope ITEX 79.4 ± 91 76.1 ±68.1 67.0 ± 41.0 47.8 ± 20.2 Dryas ITEX 113.3 ± 56.6 108.5 ± 64.9 84.3 ± 36.4 41.5 ± 23.7 Appendix Table 18.2: Summary of repeated measures analysis of variance of soil moisture levels in nine communities in 1993. Source df SS MS F p > F site 7 255016.9 36431.0 29.2 0.0001*** error 55 686341.3 1247.8 date 3 25228.6 8409.5 9.17 0.0001*** date x site 21 42506.6 2024.1 2.21 0.0045** error(date) 165 151276.7 916.8 156 Appendix 19: Vegetative phenology timing (day number in calendar year) and maximum shoot elongation (mm) of Cassiope tetragona observed in nine Cassiope-dominated communities at Alexandra Fiord in 1993. Appendix Table 19.1: Mean values ± 1 standard deviation of vegetative phenophase timing (day number) and maximum shoot elongation (mm) of Cassiope tetragona measured in nine Cass/ope-dominated communities in 1993. See main text for keys to site abbreviations. Date of growth Date of growth Maximum shoot Site Date of bud break initiation cessation elongation RSI 161 ±3.5 178 ± 6.0 209 ± 7.2 4.23 ± 1.99 RS2 176 ±3.3 191 ± 5.2 210 ± 5.9 2.48 ± 1.50 RS3 181 ±4.4 194 ±6.1 214 ±6.1 2.71 ± 1.56 RS4 171± 2.8 187 ±6.1 207 ±6.1 2.47 ± 1.37 RS5 166 ±5.0 187 ± 6.5 207 ± 7.5 2.26 ± 1.44 Erratic 162 ±4.1 181 ± 5.5 208 ± 5.6 2.48 ± 1.16 BR 185 ± 3.5 197 ± 3.8 217 ± 5.2 2.82 ± 1.38 CI 173 ± 4.5 188 ± 6.2 215 ± 0.6 3.22 ± 1.51 Dl 171 ± 4.4 184 ± 6.8 215 ± 7.6 4.06 ± 2.15 Appendix Table 19.2: Summary of analysis of variance to test for site effects on timing of vegetative bud break of Cassiope tetragona in 1993. Source df_ SS MS F p > F site 8 22286.35 2785.89 168.84 0.0001*** error 485 8002.55 16.5 total 493 30288.90 Appendix Table 19.3: Summary of analysis of variance to test for site effects on timing of growth initiation of Cassiope tetragona in 1993. Source df SS MS F p > F site 8 12777.29 1597.16 45.13 0.0001*** error 485 17163.57 35.39 total 493 29940.87 Appendix Table 19.4: Summary of analysis of variance to test for site effects on timing of growth cessation of Cassiope tetragona in 1993. Source df SS MS F p > F site 8 6176.63 772.08 17.95 0.0001*** error 485 20865.48 43.02 total 493 27043.11 157 Appendix Table 19.5: Summary of analysis of variance to test for site effects on the number of days from bud break to growth cessation observed for Cassiope tetragona in 1993. Source df SS MS If p > F site 8 12697.67 1587.21 26.94 0.0001*** error 485 28571.94 58.91 total 493 41269.60 Appendix Table 19.6: Summary of analysis of variance to test for site effects on the number of days from growth initiation to growth cessation observed for Cassiope tetragona in 1993. Source df SS MS F p > F site 8 9856.79 1232.10 13.94 0.0001*** error 485 42856.36 88.36 total 493 52713.15 Appendix Table 19.7: Summary of analysis of variance to test for site effects on logio-transformed values of maximum shoot elongation (mm) of Cassiope tetragona in 1993. Source df SS MS F p > F site 8 4.33 0.5413 8.86 0.0001*** error 485 29.62 0.0611 total 493 33.95 158 Appendix 20: Vegetative phenology timing (day number in calendar year) and maximum shoot elongation (mm) of Cassiope tetragona observed in five community zones of the River Slope site in 1992 and 1993. Appendix Table 20.1: Mean values ± 1 standard deviation of vegetative phenophase timing (day number) and maximum shoot elongation (mm) of Cassiope tetragona measured in five zones of the River Slope in 1992. See Appendix Table 2.19 for 1993 values. Date of growth Date of growth Maximum shoot Site Date of bud break initiation cessation elongation "RSI 178 ± 2.9 189 ± 5.6 217 ± 6.2 3.31 ± 1.27 RS2 184 ±2.9 199 ±8.1 217 ± 4.9 2.36 ±1.15 RS3 190 ±3.6 206 ±8.4 222 ± 4.7 2.04 ± 0.94 RS4 185 ±3.2 200 ±8.3 218 ± 5.2 2.18 ±1.17 RS5 184 ±3.5 196 ±8.0 217 ± 4.9 1.97 ± 0.90 Appendix Table 20.2: Summary of analysis of variance of year and treatment effects on timing of vegetative bud break of Cassiope tetragona observed in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year site year x site error total 1 4 4 435 444 19224.90 7270.62 2234.62 55075.06 86178.35 19224.90 1817.65 558.65 126.61 151.84 14.36 4.41 0.0001*** 0.0001*** 0.0017** Appendix Table 20.3: Summary of analysis of variance of year and treatment effects on timing of growth initiation of Cassiope tetragona observed in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year 1 12170.17 12170.17 243.53 0.0001*** site 4 11135.87 2783.97 55.71 0.0001*** year x site 4 565.05 141.26 2.83 0.0245 error 435 21738.31 49.97 total 444 48547.65 Appendix Table 20.4: Summary of analysis of variance of year and treatment effects on timing of growth cessation of Cassiope tetragona observed in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year 1 8312.66 8312.66 240.32 0.0001*** site 4 2432.06 608.02 17.58 0.0001*** year x site 4 260.19 65.05 1.88 0.1128 error 435 15046.52 1294.74 total 444 26699.18 34.59 159 Appendix Table 20.5: Summary of analysis of variance of year and treatment effects on the number of days from vegetative bud break to growth cessation observed for Cassiope tetragona in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year 1 2254.36 2254.36 14.28 0.0001*** site 4 3885.51 971.38 6.15 0.0002*** year x site 4 1419.41 354.85 2.25 0.0631 error 435 68669.48 157.86 total 444 77144.11 Appendix Table 20.6: Summary of analysis of variance of year and treatment effects on the number of days from growth initiation to growth cessation observed for Cassiope tetragona in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year site year x site error total 1 4 4 435 444 366.52 6062.15 327.19 38105.57 45691.52 366.52 1515.54 81.80 87.60 4.18 17.30 0.93 0.0414* 0.0001*** 0.4441 Appendix Table 20.7: Summary of analysis of variance of year and treatment effects on login-transformed values of maximum shoot elongation of Cassiope tetragona observed in five zones of the River Slope site in 1992 and 1993. Source df Type III SS MS F p > F year site year x site error total 1 4 4 435 444 0.1668 2.9050 0.1242 26.0408 29.6164 0.1668 0.7263 0.0310 0.0598 2.79 12.15 0.52 0.0956 0.0001*** 0.7217 160 Appendix 21: Development of a regression model to predict annual shoot elongation (mm) of Cassiope tetragonahom patterns of vegetative phenology using observations made in nine Cassiope-dominated communities at Alexandra Fiord in 1993. Phenological parameters eligible for input in the model were: date of vegetative bud break, date of growth initiation, date of growth cessation, days from bud break to growth cessation and days from growth initiation to growth cessation. Appendix Table 21.1: Summary of stepwise regression analysis of log in-transformed values of maximum shoot elongation (mm) with vegetative phenology parameters measured in 1993: step one. The number of days from growth initiation to growth cessation was selected as the first predictor variable (Xi) to be included in the model. Source df SS MS . F-value p > F R2 model error total 1 492 493 21.36 12.60 33.95 21.36 0.03 834.14 0.0001*** 0.6290 Variable df Parameter estimate Standard error Type II SS F p > F intercept Xi 1 1 -0.0885 0.0201 0.0184 0.0007 0.589 21.357 22.99 834.14 0.0001*** 0.0001*** Appendix Table 21.2: Summary of the full model developed using stepwise regression analysis of log in- transformed values of maximum shoot elongation (mm) with vegetative phenology parameters measured in 1993. The number of days from vegetative bud break to growth cessation and date of growth initiation were selected as the second (X2) third (X3) predictor variables, respectively, to be included in the model. No other variables were significant at cc=0.05. Source df SS MS F-value p > F R2 model 3 23.91 7.971 388.93 0.0001*** 0.7042 error 490 10.04 0.020 total 493 33.95 Parameter Standard Variable df estimate error Type II SS F p>F intercept 1 1.2724 0.2342 0.605 29.51 0.0001*** X i 1 -0.0053 0.0012 0.430 20.99 0.0001*** x 2 1 -0.0140 0.0014 2.117 103.30 0.0001*** x 3 1 0.0280 0.0014 8.586 418.97 0.0001*** 161 Appendix Table 22: Numbers of mature, live shoots per 10x10 cm area, numbers of flowers produced per mature shoot, and numbers of dead shoots per live shoots of Cassiope tetragona observed in nine Cassiope-dominated communities at Alexandra Fiord in 1993. Appendix Table 22.1: Mean values ± 1 standard deviation of numbers of mature, live shoots per 10x10 cm area, numbers of flowers produced per mature shoot, and numbers of dead shoots per live shoots of Cassiope tetragona in 1993. Number of live Number of flowers Number of dead Site shoots per 10 cm 2 per mature shoot shoots per live shoot RS zone 1 7.5 ± 8.8 0.249 ± 0.301 0.780 ± 0.829 RS zone 2 14.7 ± 6.8 0.068 ± 0.138 0.260 ± 0.249 RS zone 3 16.4 ± 10.0 0.008 ± 0.020 0.281 ± 0.182 RS zone 4 16.1 ± 12.3 0.010 ± 0.033 0.274 ± 0.249 RS zone 5 14.9 ± 7.3 0.022 ± 0.051 0.250 ± 0.232 Erratic site 24.4 ± 13.0 0.305 ± 0.301 0.307 ± 0.149 BR snowbed 23.5 ± 16.1 0.071 ± 0.106 0.190 ± 0.308 Cassiope ITEX 17.5 ± 9.5 0.144 ± 0.176 0.355 ± 0.319 Dryas ITEX 11.3 ± 9.5 0.220 ± 0.372 1.291 ± 1.826 Appendix Table 22.2: Summary of analysis of variance to test for site effects on numbers of mature, live shoots of Cassiope tetragona per 10x10 cm area. Source df; SS MS F p > F site 8 7375.15 921.894 7.99 0.0001*** error 261 30098.79 115.32 total 269 37473.94 Appendix Table 22.3: Summary of analysis of variance to test for site effects on numbers of flowers produced per mature shoot of Cassiope tetragona. Source df SS MS F p > F site 8 2.176 0.272 5.45 0.0001*** error 228 11.378 0.050 total 236 13.553 Appendix Table 22.4: Summary of analysis of variance to test for site effects on numbers of dead shoots per live shoot of Cassiope tetragona. Source df SS MS F p > F site 8 37.00 4.625 6.45 0.0001*** error 228 163.51 0.717 total 236 200.52 162 Appendix 23: Summary of climate response functions developed for mean standardized shoot elongation, leaf number and flower number chronologies of Cassiope tetragona using multiple linear regression with current and preceding year's monthly climate data from Eureka, Ellesmere Island. Models were developed using backward stepwise selection of predictor variables. Only climate variables which exhibited significant correlations (p<0.05) with the dependent variable were eligible for inclusion in the model. A significance level of a=0.05 was required for retention of each variable in the backward stepwise procedure. Appendix Table 23.1: Summary of the response function developed for the shoot elongation chronology. Chronology period is from 1957-1992. Source df SS MS F-value p > F R2 model 2 0.4015 0.2007 20.245 0.0001*** 0.5586 error 32 0.3173 0.0099 total 34 0.7189 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > T intercept 1 0.5992 0.1019 5.880 0.0001*** July T (year C) 1 0.0086 0.0018 4.858 0.0001*** June T (year C) 1 -0.0037 0.0010 -3.827 0.0006*** T = temperature, P = precipitation, C = current year's values, C-l = preceding year's values. Appendix Table 23.2: Summary of the response function developed for the leaf number chronology. Chronology period is from 1957-1992. Source df SS MS F-value p > F R2 model error total 1 33 34 0.0152 0.0152 0.0922 0.0028 0.1074 5.437 0.0260* 0.1414 Variable df Parameter Standard estimate error T for Ho: Parameter=0 p>T intercept June T (year C) 1 1 0.9805 0.0012 0.0134 0.0005 73.351 2.332 0.0001 0.0260 T = temperature, P = precipitation, C = current year's values, C-l = preceding year's values. 163 Appendix Table 23.3: Summary of the response function developed for the flower number chronology. Chronology period is from 1967-1992. Source df SS MS F-value p > F R2 model 2 4.1448 2.0724 12.542 0.0002*** 0.5217 error 23 3.8005 0.1652 total 25 7.9453 Parameter Standard T for Ho: Variable df estimate error Parameter=0 p > T intercept 1 1.5214 0.3890 3.911 0.0007 June T (year C-l) 1 0.0156 0.0047 3.322 0.0030 May T (year C-l) 1 0.0074 0.0031 2.397 0.0250 T = temperature, P = precipitation, C = current year's values, C-l = preceding year's values. 164 Appendix 24: Summary of climate response functions developed for mean standardized shoot elongation and flower number chronologies of Cassiope tetragona using multiple linear regression with current and preceding year's monthly climate data from Eureka, Ellesmere Island. Models were developed using backward stepwise selection of predictor variables. Al l climate variables and previous growth were eligible for inclusion in the model. A significance level of a=0.05 was required for retention of each variable in the backward stepwise procedure. The model for the leaf number chronology developed using this procedure is the same as the one summarized in Appendix Table 23.2. Appendix Table 24.1: Summary of the response function developed for the shoot elongation chronology. Chronology period is from 1957-1992. Source df SS MS F-value p > F R2 model 4 0.4639 0.1160 13.651 0.0001*** 0.6454 error 30 0.2549 0.0085 total 34 0.7189 Parameter Standard T for Ho: Variable df estimate error Parameter=0 P>T intercept 1 0.7338 0.1387 5.291 0.0001 SE (year C-l) 1 -0.2632 0.1255 -2.097 0.0445 July T (year C) 1 0.0091 0.0017 5.473 0.0001 JuneT (year C-l) 1 -0.0031 0.0010 -3.078 0.0044 Aug. T (year C-l) 1 0.0029 0.0013 2.142 0.0404 SE = shoot elongation chronology, T = temperature, P = precipitation, C = current year's values, C-l = preceding year's values. Appendix Table 24.2: Summary of the response function developed for the flower number chronology. Chronology period is from 1967-1992. Source df SS MS F-value p > F R2 model 5 error 20 total 25 6.6978 1.3396 1.2475 0:0624 7.9453 21.475 0.0001*** 0.8430 Variable Parameter df estimate Standard error T for Ho: 3arameter=0 p>T intercept SE(year C-l) May T (year C-l) July T (year C-l) Sept. T (year C-l) July P (year C-l) 1 2.3174 1 2.8320 1 0.0071 1 -0.0342 1 0.0153 1 -0.0022 0.9473 0.4672 0.0019 0.0091 0.0035 0.0007 2.446 6.062 3.752 -3.744 4.423 -2.928 0.0238 0.0001 0.0013 0.0013 0.0003 0.0083 SE = shoot elongation chronology, T = temperature, P = precipitation, C = current year's values, C-l = preceding year's values. 165 Appendix 25: Summary of climate transfer functions developed for July melting degree days and July average temperatures at Alexandra Fiord, Ellesmere Island using multiple linear regression with current and preceding year's values of three Cassiope chronologies. Models were developed using backward stepwise selection of predictor variables. A significance level of a=0.05 was required for retention of each variable in the backward stepwise procedure. Appendix Table 25.1: Summary of the transfer function developed for July melting degree days at Alexandra Fiord. Calibration period is from 1980-1988. Source df SS MS F-value p > F R2 model error total 2 6 8 8164.00 4082.00 33.51 730.89 121.81 8894.89 0.0006*** 0.9178 Variable Parameter df estimate Standard error F for Ho: Parameter=0 p > F intercept FN (year C) FN (year C-l) 1 80.3442 1 76.6336 1 28.1767 15.1601 9.4032 9.3607 28.09 66.42 9.06 0.0018 0.0002 0.0237 FN = flower number chronology, C = current year's values, C-l = preceding year's values. Appendix Table 24.2: Summary of the response function developed for July average temperatures at Alexandra Fiord. Calibration period is from 1980-1988. Source df SS MS F-value p > F R2 model error total 1 7 8 4.1869 4.1869 1.002 0.1431 5.1889 29.25 0.0010 0.8069 Variable Parameter df estimate Standard error F for Ho: Parameter=0 p > F intercept L N (year C) 1 -9.1603 1 15.1030 2.8450 2.793 1.4840 4.1869 0.0147 0.0010 LN = leaf number chronology, C = current year's values. 166 

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.831.1-0099012/manifest

Comment

Related Items