Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The spatial relationships among vegetation phenology, plant community composition and environment at… Bean, David 2002

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

Item Metadata

Download

Media
831-ubc_2002-0336.pdf [ 18.68MB ]
Metadata
JSON: 831-1.0090344.json
JSON-LD: 831-1.0090344-ld.json
RDF/XML (Pretty): 831-1.0090344-rdf.xml
RDF/JSON: 831-1.0090344-rdf.json
Turtle: 831-1.0090344-turtle.txt
N-Triples: 831-1.0090344-rdf-ntriples.txt
Original Record: 831-1.0090344-source.json
Full Text
831-1.0090344-fulltext.txt
Citation
831-1.0090344.ris

Full Text

THE SPATIAL RELATIONSHIPS AMONG VEGETATION PHENOLOGY, PLANT COMMUNITY COMPOSITION AND ENVIRONMENT AT A HIGH ARCTIC OASIS  by DAVID BEAN B.A., University of Ottawa, 1999  A THESIS SUBMITTED IN PARTIAL FULFILMENT 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 the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 2002 © David Bean, 2002  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  Department of Geography The University of British Columbia Vancouver, Canada  June IT, 2002  Abstract Environmental changes resulting from global warming are predicted to be most intense at high latitudes and this has considerable implications for the vegetation in the High Arctic. The relationships among plant community structure, diversity, phenology, and abiotic factors including snowmelt pattern, temperature, soil moisture and soil nutrients were studied at the Alexandra Fiord lowland (78° 53' N, 75° 55" W), a high arctic oasis on the east coast of Ellesmere Island. At each of 28 sampling points, vegetation was surveyed, soil was sampled, temperature was recorded by dataloggers and phenological observations were made on four dominant plant species throughout one growing season. Digital aerial photographs were used to study the pattern of snowmelt across the lowland. A geographic information system Was used to analyze the data from the discrete sampling points and relate them to the observed distribution of plant communities. Plant communities were analyzed using single linkage cluster analysis, principal components analysis and redundancy analysis including spatial information in the direct gradient analysis. Phenological development of Cassiope tetragona and Dryas integrifolia was strongly correlated to the temperature gradient across the lowland but Saxifraga oppositifolia  and Salix arctica  were not. The two former species flowered later in the season while the other two flowered shortly after snowmelt. The data were compared to an 8 year record of phenological observations at the site and Dryas was found to have a more pronounced response to temporal variability, whereas Cassiope had much more fixed timing for phenological development at a given place over time, while varying considerably across spatial gradients. Five major plant communities and two subtypes were defined. Moisture was found to be the most important environmental factor separating plant communities as is common these studies in the Arctic. Temperature was also an important factor in the indirect gradient analysis but this variable was highly spatially autocorrelated and much of the variation explained by temperature could be explained by spatial location information alone. The pattern of snowmelt did not vary at the same scale as the vegetation according to these results, and is thus not likely a major determinant of landscape-scale plant community distribution. The distribution seems more like a successional sequence resulting from differing times since the retreat of both Pleistocene and Little Ice Age glacier advances.  ii  Table of Contents Abstract Table of Contents List of Tables List of Figures Acknowledgements  ii iii v vi  viii  CHAPTER 1: Introduction 1.1 Introduction 1.2 Review of the Literature 1.2.1 Controls on the Distributions of Arctic Plants 1.2.2 Arctic Plant Phenology 1.2.3 Polar Oases 1.2.4 Classifying Arctic Vegetation 1.2.5 Remote Sensing and Geographic Information Systems 1.2.6 Landscape Ecology and Spatial Analysis 1.2.7 Climate Change in the Arctic 1.2.8 Vegetation Response to Climate Change 1.2.9 The International Tundra Experiment (ITEX) 1.3 Research Aims and Objectives  1 2 2 5 6 8 9 10 12 12 14 14  CHAPTER 2: Methods 2.1 Research Site 2.2 Sampling Pilot-Study 2.3 Field and Laboratory Methods 2.3.1 Temperature Measurements 2.3.2 Plant Phenology and Growth 2.3.3 Plant Cover 2.3.4 Soil Moisture and Nutrients 2.3.5 Aerial Photography : 2.3.6 Mosaicking Digital Aerial Photographs 2.4 Data Analysis 2.4.1 Plant Cover and Communities 2.4.2 Vegetation Community and Snowmelt Mapping 2.4.3 Temperature and ^Phenology Data 2.4.4 Spatial Analysis 2.4.5 Comparison with ITEX Data 2.5 Summary  16 18 20 20 -22 22 22 24 24 24 24 26 26 27 28 28  CHAPTER 3: Sampling Pilot-Study 3.1 Introduction 3.2 Sampling Strategies 3.3 Methods 3.4 Results 3.5 Discussion  29 29 31 34 34  iii  CHAPTER 4: Spatial Relationships Among Plant Phenology. Growth, and Abiotic Variables 4.1 Introduction 4.2 Environmental Variables 4.2.1 Soil Nutrients 4.2.2 Soil Moisture 4.2.3 Climate / Temperature 4.3 Plant Phenology 4.3.1 4.3.2 4.3.3 4.3.4  Cassiope tetragona Dryas integrifolia Saxifraga oppositifolia Salix anctica  4.4 Comparison with ITEX Observations 4.5 Discussion  3 8 3 8 3 8 3 8  40 4  5  45 49 54 56  59 61  CHAPTER 5: Vegetation Community Description and Mapping 5.1 Introduction 5.2 Environmental Variables 5.3 Plant Cover 5.4 Classification and Ordination 5.4.1 Cluster Analysis 5.4.2 Principal Components Analysis 5.4.3 Redundancy Analysis 5.5 Plant Communities 5.6 Relationship Between Vegetation Communities and Snowmelt Date 5.7 Correspondence to Existing Vegetation Map 5.8 Interpretation of Digital Aerial Photographs 5.9 Discussion  70 70 70 72 72 72 72 78 81 81 85 89  CHAPTER 6: Conclusions Conclusions  9 6  REFERENCES  9 8  APPENDIX A: Descriptions of Phenological Stages  109  List of Tables Table 2.1 Six plant communities of the Alexandra Fiord lowland identified by Muc et al (1989) and some cover characteristics 18 Table 3.1 The five snowmelt dates and associated areas, and numbers of sampling points per date for each strategy 33 Table 3.2 The interpolation error associated with each of the sampling strategies  34  Table 4.1 Summary of the temperature data from Alexandra Fiord from 1990-2000  40  Table 4.2 Comparison between range of dates (day numbers) at Alexandra Fiord for phenological events of Dryas integrifolia in the current study and from ITEX monitoring plots at Alexandra Fiord 61 Table 4.3 Comparison between range of dates (day numbers) at Alexandra Fiord for phenological events of Cassiope tetragona in the current study and from ITEX monitoring plots at Alexandra Fiord 61 Table 5.1 Results from the PCA of vascular plant cover  72  Table 5.2 Eigenvalues from the RDA of vascular plants with five selected environmental variables 74 Table 5.3 Correlation matrix from the RDA of vascular plants with the five selected environmental variables 74 Table 5.4 Results from the RDA on vascular plants with five selected environmental variables and geographic coordinates 77 Table 5.5 Correlation matrix from the RDA of vascular plants with five selected environmental variables and geographic coordinates 77 Table 5.6 Comparison matrix between Muc et a/.'s (1989) vascular plant communities and those identified in this study 85  v  List of Figures Figure 2.1 Maps showing the location of Alexandra Fiord  17  Figure 2.2 Plant communities of the Alexandra Fiord lowland (from Muc et al. 1989)  19  Figure 2.3 Locations of sampling sites  21  Figure 2.4 Locations where phenology was monitored for each plant species  23  Figure 3.1 The pattern of snowmelt in the Alexandra Fiord lowland  32  Figure 3.2 Location of sampling points for each of the three sampling strategies  32  Figure 3.3 Maps and graphs showing the interpolation error in snowmelt dates for each of the three sampling schemes 35 Figure 4.1 Soil carbon and nitrogen content, soil C:N and soil moisture in the early and mid growing season at each sample site in the Alexandra Fiord lowland 39 Figure 4.2 Mean, maximum and minimum near-surface (+10 cm) temperature for the 2000 growing season 41 Figure 4.3 Near surface temperature distributions for the Alexandra Fiord lowland in 2000 42 Figure 4.4 Variogram for temperature variables. Moran's I correlogram for temperature variables. Gearey's C correlogram for temperature variables 43 Figure 4.5 The accumulation of near-surface thawing degree-days (TDD) across the Alexandra Fiord lowland over the course of the growing season 44 Figure 4.6 Phenological dates for Cassiope tetragona and numbers of flowers and fruits.  46  Figure 4.7 Residuals from regressions between phenological stages and TDD for Dryas integrifolia and Cassiope tetragona that showed a spatial pattern 47 Figure 4.8 Variogram for Cassiope tetragona phenology. Moran's I correlogram for Cassiope tetragona phenology. Gearey's C correlogram for Cassiope tetragona phenology 48 Figure 4.9 Variogram for Cassiope tetragona flower and fruit numbers. Moran's I correlogram for Cassiope tetragona flower and fruit numbers. Gearey's C correlogram for Cassiope tetragona flower and fruit numbers 50 Figure 4.10 Phenological observations and growth measurements for Dryas integrifolia..  51  Figure 4.11 Variogram, Moran's I correlogram, and Gearey's C correlogram for Dryas integrifolia phenology, flower heights and percentage of flowers fertilized 53 Figure 4.12 Phenological observations for Saxifraga oppositifolia  55  vi  Figure 4.13 Variogram, Moran's I correlogram, and Gearey's C correlogram for Saxifraga oppositifolia phenology  57  Figure 4.14 Phenological observations and growth measurements for Salix arctica  58  Figure 4.15 Variogram, Moran's I correlogram, and Gearey's C correlogram for Salix arctica phenology and fascicle lengths 60 Figure 5.1 Distributions of vegetation cover variables  71  Figure 5.2 Complete linkage cluster analysis on percent cover of live vascular plants (data unstandardized). 73 Figure 5.3 Distributions of vascular plant communities on the Alexandra Fiord lowland.. 74 Figure 5.4 PCA ordination on live vascular plants, scaled to maximize site differences, with 95% confidence intervals 75 Figure 5.5 RDA on vascular plant cover without and with geographic coordinates  76  Figure 5.6 Partitioning of variance explained by RDA into exclusively environmental, spatially structured environmental, exclusively spatial and unknown components 79 Figure 5.7 Oblique photos of the lowland during the snowmelt period, looking west  82  Figure 5.8 Airphoto mosaic of the Alexandra Fiord lowland on June 12, 2000 (day 164). . 83 Figure 5.9 Airphoto mosaic of the Alexandra Fiord lowland on June 16, 2000 (day 168). . 84 Figure 5.10 Boundaries from Muc et a/.'s (1989) map of vegetation communities with 5 m buffer on either side of boundary and 10 m buffer 86 Figure 5.11 Airphoto mosaic of the Alexandra Fiord lowland on July 14, 2000 (day 196). .87 Figure 5.12 Representative patches of airphotos from attempted photo-interpretation of plant communities I  Acknowledgements Firstly, I would like to acknowledge the support I received in so many forms from Greg Henry. From developing the initial concept of the research, to working out glitches in the field, to finding me jobs, to reviews of drafts of this thesis, I thank you wholeheartedly. Cathy Chan could hardly be called my field assistant, more of a field partner and now friend for life. Plenty of your thought and hard work are contained in these pages. So too for the rest of the field crew who were the best a guy could ask for: Katie Breen, Kevin O'Dea, and Pippa Hett, all of whom suffered through point-framing in a cloud of mosquitoes for my benefit. I would also like to thank my lab-mates back here for day-to-day support and companionship: Heather Nicholson, Sandra Rolph, Shelly Rayback and Adam T. Young. Thanks to Brian Klinkenberg as well for advice along the way and a thorough review of the draft. Many thanks to all the agencies that provided funding and support for this project: The Natural Sciences and Engineering Research Council of Canada, The Northern Scientific Training Program of the Department of Indian and Northern Affairs, and the Polar Continental Shelf project. Thanks also to the Royal Canadian Mounted Police for allowing researchers to use the facilities at the Alexandra Fiord detachment for all these years. I am most fortunate to have an amazing family who have supported me and shown nothing but keen interest and enthusiasm. Mom, I know you would be proud. And finally, to Lara, my partner in all things, love and gratitude.  viii  Chapter One Introduction 1.1  Introduction  The vegetation of the High Arctic grows under some of the most extreme conditions present on the Earth. Low temperatures, light levels, soil nutrients and precipitation, short growing seasons and high winds limit the ability of plants to establish and grow in this region with environmental conditions often constituting the absolute limit of the ecological tolerance of plant species (Svoboda and Henry 1987). Models designed to simulate global climate consistently predict that the effects of climate change will be most intense at high latitudes (Houghton et al. 1995, Maxwell 1992). These predictions have significant implications for vegetation including potentially changing phenology, growth rates and reproductive success in individual plants, and the species composition of plant communities (Chapin et al. 1992a, Oechel et al. 1997). Evidence of climate change and its effects are mounting in the Arctic (e.g. Sereeze et al. 2000, Sturm et al. 2001, Zeeberg and Forman 2001) and in other ecosystems (Walther et al. 2002). The distribution of vegetation in the High Arctic is governed largely by the availability of moisture, above-freezing temperatures and suitable soils (Murray 1997, Rannie 1986, Walker 1995). An understanding of the current distribution of arctic plants is essential if accurate predictions are to be made regarding the response of vegetation to changing environmental conditions. Our knowledge of the ways in which the spatial distributions of growth and diversity in high arctic vegetation are influenced by environmental factors is incomplete. There is tremendous opportunity to apply new methods of spatial analysis and to employ new technologies in remote sensing and GIS in the study of the spatial distributions of high arctic vegetation and interrelationships with abiotic parameters. To date, much of the development of methods of spatial analysis has relied on the use of artificial datasets and there is a need for rigorous field-testing of these methods. This thesis aims to elucidate the spatial relationships among vegetation and environmental variables in the Alexandra Fiord lowland, a high arctic oasis. In this chapter I present a review of the relevant literature in the fields related to this research followed by a description of the specific aims and objectives of this study. Chapter 2 includes an outline of the field and analytical methods used to carry out this  1  research followed by Chapter 3 in which are described the methods and results of a pilot study that was undertaken to evaluate the effectiveness of different sampling schemes. The results and discussion regarding the spatial relationships among vegetation phenology, growth and abiotic variables are described in Chapter 4. In Chapter 5 I provide the results and discussion for the vegetation classification and mapping study, and Chapter 6 is a summary of the key findings of the research. 1.2 Review of the  Literature  Since the pioneering work of Sorensen (1941), Polunin (1948), Porsild (1951) and Saville (1961), interest in arctic plant ecology has grown steadily with an ever-increasing understanding of the significance of this delicate ecosystem. The vegetation of the High Arctic is distinct in term of both its composition and structure (Archibold 1995). The extremes of climate and resulting limited resources restrict the number of species that are able to survive in the Arctic and place constraints on the viable growth forms of arctic plants. The majority of the High Arctic that is free of glaciers is characterized by polar desert landscapes with a vascular plant cover of less than 5% resulting, in part, from the minimal supply of moisture (Bliss 1997). Despite the significant aerial extent of the polar deserts, the plant dynamics in this region have not been well studied (Bliss et al. 1994, Levesque, 1997). Until recently, very few detailed plant community studies have been carried out within the Queen Elizabeth Islands (see section 1.2.4). 1.2.1 Controls on the Distribution of Arctic Plants  The distribution of the arctic flora is determined by many factors, and these increase and decrease in importance at different scales and interact to produce many specific microenvironments. Among these factors, the most important are temperature, availability of moisture, snow cover, soil nutrient status and biological interactions (see below). Each of the abiotic variables interact with one another and become more or less important in plant physiological, population and community processes at different scales. Temperature  The most significant control on high arctic plant diversity at a large spatial scale has been shown to be temperature (Edlund and Alt 1989, Rannie 1986, Young 1971). Rannie (1986) demonstrated a very strong correlation between mean July temperature and vascular plant diversity for the Arctic as a whole. Optimum shoot growth  2  temperatures for arctic plants range from 10-20°C, compared to 25-30°C for temperate species (Kummerow etal. 1980, Tieszen etal. 1980). Though ambient air temperature is almost always below optimum, the steepest temperature gradient tends to be within 1015 cm of the ground surface and under the right conditions the soil surface may be as much as 30°C above the temperature at 50 cm above the ground (Bliss 1997), and this allows higher metabolic activity in plants. Given that temperatures are almost always below optimum for physiological processes (Chapin and Shaver 1985a), morphological adaptations to increase plant temperature, like dark colour, low stature and the maintenance of attached dead tissues, are very common. Plant temperatures in many arctic plant growth forms can be as much as 8-10°C above ambient air temperature (Chapin and Shaver 1985a). Soil temperatures are very low and soil thaw depth increases gradually to a maximum near the end of the growing season. During the period of maximum radiation the soil may, in fact, still be completely frozen. Moisture  The Arctic is also characterized by low but variable availability of moisture. In low-lying areas soils are often saturated due to accumulation of meltwater above the impermeable permafrost table. Bliss et al. (1994) found that species richness and total vascular plant cover in polar desert landscapes depend primarily on the duration of soil surface moisture and the development of a cryptogamic crust. Moisture status has been shown to be the most important factor affecting vegetation community structure in many studies (e.g. Muc et al. 1989, Sheard and Geale 1983a,b). Moisture tends to be positively correlated with soil organic matter (Broil et al. 1999) and diversity in moisture status associated with microtopography has been found to strongly influence plant community diversity (Sohlberg and Bliss 1984, Webber 1978).  Snowcover  Snowcover is widely accepted as being an important determinant of many ecosystem processes in alpine (Kudo 1991, Walker et al. 1993) and arctic ecosystems (Walker et  al. 2001), particularly in polar deserts. Snowcover affects soil moisture, depth of freezing, soil temperatures and soil heat flux, chemical budgets, and provides protection from strong winds, winter temperatures and desiccation (Walker et al. 2001). Snowmelt is the primary determinant of the timing of bud break in arctic plant species (Kudo 1991, Sorensen 1941 among others). In some cases, growth can begin beneath the snowpack  3  in 'greenhouses' as melting snow turns to ice at the snow surface (Walker et al. 2001). The presence of persistent snowbanks also provides a source of moisture in otherwise dry polar desert sites and gives rise to relatively lush snowflush communities. Seeds have been found to be smaller in the centre of late-lying snowbeds where the growing season is shortened (Galen and Stanton 1991). The distribution of snow cover tends to be relatively constant from year to year (Molau 1997) due to the dependence on microtopography. With the exception of persistent snowbanks, Woo et al. (1991) and Edlund et al. (2000) found little difference in community composition among early and late melting sites near Eureka on the Fosheim Peninsula, Ellesmere Island. Webber (1978) found little effect of snowmelt at Barrow, Alaska due to a lack of mesotopographic features, and Miller (1982) identified that differences in nutrient status were more important than date of snowmelt in determining vegetation composition. Nutrients  Arctic soils are characterized by low nutrient levels. Soils are poorly developed due to low temperatures, short growing seasons, surface instability, low plant cover and the relatively young age of arctic soils (Bliss 1997). Plants have adapted nutrient absorption capacities to low temperatures but remain limited by the low soil nutrient concentrations and, therefore, actual absorption rates are low. Nitrogen fixation is at a maximum between 15 - 20°C and is therefore higher where dark mosses and lichens warm the surface (Alexander et al. 1978, Henry and Svoboda 1986). Differences in substrate lithology and soil nutrient availability can also give rise to variations in plant community development in polar deserts (Levesque 1997). McKane et al. (2002) found that tundra plants partitioned nitrogen dependent on the preferred chemical form, with the most productive species preferring the most available forms. Many authors have stressed the importance of soil pH in tundra ecosystems (Edlund and Alt 1989, Gough et al. 2000, Walker 2000). Some species are more common on alkaline soils such as Carex rupestris, Dryas octopetala, Saxifraga oppositifolia and some on acidic soils, such as Hierochloe alpina, Luzula confusa and Potentilla hyperarctica (Elvebakk 1997). There are similar differences in the composition of bryophyte communities along pH gradients (Steere 1978), and vascular plant diversity was found to increase with pH (Gough et al. 2000). Many nutrient addition experiments have been undertaken, with a lag generally found between the time of nutrient addition and plant response, and plant responses  4  being different among growth forms (e.g. Chapin and Shaver 1985b, 1996, Henry et al. 1986b, Parsons etal. 1994). Biological  Interactions  There is considerable debate in ecology about the intensity of competition in areas of high environmental stress (e.g., Grime 1979, Tilman 1985). Edaphic parameters are thought to have more influence on productivity in the Arctic than biological factors such as competition and predation (Muc et al. 1989, Svoboda and Henry 1987). However, biotic interactions and particularly grazing were found to have a significant effect in the community structure of high arctic sedge meadows (Henry 1998) and elsewhere (Jefferies et al. 1992). Succession has also not received much attention in the Arctic due largely to the shortage of r-selected species adapted to colonize disturbed sites (Bliss and Gold 1994, Bliss and Peterson 1992, Jones 1997, Svoboda and Henry 1987). Increasingly, the positive biological relationships between arctic plants and mycorrhizal fungi are being studied in reference to nutrient uptake by plants (Aitken er al. 1993, Dalpe and Aiken 1998, Kohn and Stasovski 1990) 1.2.2 Arctic Plant Phenology  Phenology refers to the timing of life-cycle events. S0rensen (1941) completed what is to date the most exhaustive study and treatment of arctic plant phenology. Arctic plant survival and reproduction are severely constrained by the short length of the growing season, as plants must complete their life cycle in the brief window of time in which temperatures are above freezing. However, most arctic plant species are able to produce overwintering buds in the previous growing season (Bell and Bliss 1980, Bliss 1971) or otherwise extend their life cycle over more than one season. Molau (1993, 1997) has elaborated on Bliss' (1956) categorization of tundra plants into pheno-classes, with early flowering species being known as vernal, then early aestival, and late flowering species called late aestival. Vernal species (like Saxifraga  oppositifolia)  overwinter with highly differentiated flower buds and are able to adjust the timing of flowering in response to climate (i.e. snowmelt) but not the number of flowers produced, whereas aestival species (e.g. Cassiope tetragona) are better able to make adjustments in the number of flowers as the growing season progresses (Molau 1997). Molau (1993, 1997) suggests that flowering phenology is the most important difference in reproductive strategy in arctic plants.  5  Thorhallsdottir (1998) found tremendous interannual variability in flowering phenology of 75 plant species monitored over 11 years in central Iceland. In a couple of species, the onset of flowering varied by as much as four weeks, which could account for up to 50% of the duration of the growing season. Her study showed that grasses tend to flower later and less consistently than cushion plants and that warmer growing season temperatures tend to be associated with earlier flowering and more species flowering. On the Fosheim Peninsula on Ellesmere Island, Edlund and Gameau (2000) found that in warm years many plant species would complete their cycle earlier and senesce despite air temperatures remaining warm late in the growing season. They also noted 2  n d  and 3  rd  waves of flowering in some species, which may increase pollen production and successful fertilization, thus enhancing seed germination success. Marsden (1992) found global solar radiation to be the most important abiotic variable determining the length of pre-floration time, but as the growing season progresses, radiation decreases and the ambient air temperature becomes increasingly important (Molau 1997). If higher air temperatures occur eariier in the season it may eliminate the division of flowering times between species that can take advantage of the early peak in insolation despite colder temperatures and those that can not. Bean and Henry (2002) summarized the phenological responses of vascular plants to climate variability at eight sites throughout the Canadian Arctic.  Phenological  development was found to be a function of the date of snowmelt, the accumulated warmth in the current growing season and accumulated warmth in the previous growing season, particularly at high arctic sites. 1.2.3 Polar Oases  Though glaciers and polar desert dominate most of the High Arctic, in localized pockets environmental conditions can be much more favourable for plant growth and relatively lush vegetation can develop. These oases represent perhaps 6% of the area of Canada's Arctic Archipelago (Freedman et al. 1994.) and support vegetation similar to the tundra of lower latitudes. Increased vegetation cover and species diversity results from the increased availability of moisture and warmer mesoclimate (Freedman et al. 1994). High arctic oases support most of the local plant species diversity (Walker 1995) and are essential sources of food for local animal population like muskox, lemming and  6  waterfowl (Bliss 1977b, Henry and Svoboda 1994). The Arctic is inherently nutrient limited, so the greatest changes in growth, reproduction and phenology under climate warming will most likely occur in these nutrient rich areas (Henry and Molau 1997). Truelove lowland on Devon Island was the first polar oasis to be comprehensively examined, being one of 14 tundra sites studied under the International Biological Programme's (IBP) Tundra Biome Programme (Bliss 1977a). The objectives were to: determine populations numbers and standing crop, determine the system's efficiency in making use of available energy, study the factors that limit the growth and development of plant and animal species, and to develop models of high arctic ecosystem function (Bliss 1977b). The plant communities of Truelove Lowland were mapped and their structure and floristic composition described by Muc and Bliss (1977), revealing a division into 11 distinct groupings. This area was found to be more diverse partly due to heterogeneous meso-topography that included rock outcrops, raised beach ridges, wet meadows, ponds and lakes (Bliss 1997). Alexandra Fiord is located on the central eastern coast of Ellesmere Island, the largest and most northerly of the Queen Elizabeth Islands. The 8 km lowland is an oasis 2  relative to the surrounding polar desert and has been the subject of considerable ecological research over the last 20 years (Henry 1987, Svoboda and Freedman 1994). The aims of the research have been to identify and describe plant communities, and relate diversity and production to environmental factors. A total of 96 species of vascular plant are found at Alexandra Fiord (Ball and Hill 1994) compared with 145 at Hot Weather Creek on the Fosheim peninsula (Edlund et al. 2000), 93 at Truelove Lowland (Barett and Teeri 1973) and 75 at Sverdrup Pass (Bergeron and Svoboda 1989). At Alexandra Fiord there is considerable variability in the vascular standing crop among the plant communities defined by Muc et al. (1989), with the largest being in mesic communities and lower values at xeric and hydric sites Muc et al. (1994b). Nams and Freedman (1987a) focused on the Cassiope fefAagona-dominated heaths of Alexandra Fiord, finding that the ordination pattern of eight stands showed snowmelt and site moisture to have a primary role in separating communities. Alexandra Fiord and Truelove Lowland were among five arctic oases in which the structure of sedgedominated wet-meadow communities was studied in terms of controlling environmental influences (Henry 1998). Grazing by muskox was found to increase production by  7  stimulating new growth and there was a decrease in production with increasing latitude. Microtopography and site moisture were also identified as factors with a significant influence on production and diversity. 1.2.4 Classifying Arctic Vegetation  It has proven difficult to differentiate plant communities on a floristic basis in the High Arctic due to the limited number of species and their generally wide ecological tolerance (Bliss and Svoboda 1984, Walker et al. 1994b). Thus, topography, moisture status, characteristics of the soil and growth form of the vegetation have been substituted as the criteria by which plant communities are defined (Muc et al. 1989) and often named. The classification of the vegetation of the Arctic as a whole has taken many forms. Polunin's (1951) classification set out high, mid and low arctic vegetation types. Porsild (1951) divided the Arctic into ice desert, rock desert and tundra. Young (1971) set out a classification based on floristics with four categories while Tedrow (1977) developed a classification based on soils including tundra, sub-polar desert, and polar desert. Yurtsev (1994) divided the circumpolar Arctic into six provinces. Walker (2000) describes a 4 category classification based on growth form, as is Edlund et al.'s (2000) slightly more extensive classification and map. Site-specific studies have derived various sets of plant community types based on the nature of the site, the purpose of the study and the intensity and methods of data collection (e.g. Bay 1992, Bergeron and Svoboda 1989, Shaefer and Messier 1994, Sheard and Geale 1983a, Muc et al. 1989, Webber 1978). Approaches to naming communities have varied considerably with the European tradition tending towards the development of standard syntaxa and North American approaches being more site specific (Walker et al. 1994b). Vegetation classification is often described in relation to a toposequence (de Molenaar 1987, Edlund and Alt 1989, Miller and Alpert 1984, Jonsdottir et al. 1999), with wet meadows being found in low-lying areas and polar deserts on the uplands. In general, cushion plants and rosettes tend to be found on wind exposed sites with well drained soils, deciduous shrubs in areas with higher soil nutrient levels, and evergreen shrubs in more moist and less nutrient rich sites (Bliss 1997). Anderson and Bliss (1998) and Gould and Walker (1999) have stressed the importance of landscape heterogeneity in determining the diversity of plant communities found in a given area, with patterned ground being an important contributor in many areas (Webber  8  1978). Sheard and Geale (1983a) list the following communities as being common in the High Arctic: 1. polar desert consisting of widely scattered Saxifraga oppositifolia individuals on sand-gravel substrate; 2. snow-bed communities dominated by heaths, primarily Cassiope  tetragona;  3. Luzula steppe with Luzula spp. and others in a physiognomic grassland; 4. earth hummocks and vegetated hummocks on mesic slopes; 5. sedge meadows dominated by Carex stans and other species of Cyperaceae; 6. raised beach ridges distinguished on the basis of gravel substrate and abundant Dryas integh folia. :  1.2.5 Remote Sensing and Geographic Information  Systems  Instruments capable of recording a wide range of wavelengths reflected by vegetation were in use by the early 1960s and this technology was being used in satellite applications by the early 1970s (Craighead et al. 1988). The rapid growth in the interest in, and availability of, remotely sensed data have been mirrored by advances in the GIS technology with which these data are analysed. Remote sensing and GIS have been embraced by plant ecologists and phytogeographers (Goodchild 1994, Goodchild et al. 1996), though many of these researchers have limited training in environmental optics, image processing, computer cartography, and quantitative spatial analysis (Walsh and Davis 1994). The use of high-resolution data in a GIS has the advantage of being, statistically speaking, the population as opposed to a sample (Evans et al. 1989). The use of remote sensing in vegetation mapping in the Arctic is limited (Nilsen et al., 1999), and most of these studies have involved a ground resolution of 20 m or larger (Craighead et al., 1988; Ostendorf and Reynolds, 1993). Bay (1992) used the normalized difference vegetation index (NDVI) to map vegetation in northeastern Greenland. Hope et al. (1993) studied the spectral reflectance properties of tussock tundra on the north slope of Alaska using hand held radiometric data to calculate NDVI, and this work was extended by Vierling et al. (1997). Mosbech and Hansen (1994) compared infrared photo interpretation to SPOT satellite classification, finding the resolution of air photos to be much higher but satellite classifications to be more costeffective. Spjekavic (1995) found that a satellite-based classification identified more land cover classes than a traditional vegetation map on Svalbard. Nilsen et al. (1999) used  9  scanned infrared photographs and incorporated them into a GIS in conjunction with a digital elevation model (DEM) to classify vegetation in northwestern Spitzbergen. Marken  el al. (1995) used AVHRR data to produce maps of phenology for all of Alaska. The use of digital elevation models (DEMs) is an excellent example of the advantages of using GIS in ecological analysis and modeling (Brassard and Joly 1994, Nilsen et al. 1999, Ostendorf and Reynolds 1998). The well-documented relationship between topography and soil-moisture status (Ostendorf and Reynolds 1998) is evidence for the value of including topographic data when studying spatial pattern in vegetation. From the data contained in a DEM it is also possible to calculate other factors such as slope, aspect, potential radiation (Gottfried et al. 1998, Nilsen et al. 1998), as well as hydrological flows along the surface (Ostendorf and Reynolds 1998). The use of electronic surveying equipment and global positioning systems (GPS) greatly improves accuracy when mapping topographic features and facilitates the integration of these data into a GIS (Gooding et al. 1997). Evans et al. (1989) used a GIS to study the spatial interrelations between terrain, snowdepths and vegetation distribution at a site in northern Alaska. Extensive ground surveys were performed to aid in the interpretation and classification of colour-infrared aerial photographs. The layered GIS database was then used to produce two-way contingency tables, which show correlations between pairs of variables. They found that the vegetation types were highly correlated with snowdepths and these were, in turn, well explained by topographic data. Walker and Walker (1991) developed a hierarchical GIS database to evaluate disturbance regimes on the north slope of the Brooks Range in Alaska. Walker (1999) incorporated remotely sensed data and many other layers in a GIS to arrive at a 20 class vegetation map of northern Alaska. Levesque (1997) used GIS to study the distribution of plants in the polar deserts of Ellesmere Island. She identified 'safe sites' by comparing microtopography to the microscale distribution of plants and found significant clustering of plants in close proximity to boulders and adult plants of the same species.  1.2.6 Landscape Ecology and Spatial Analysis  Interest in spatial relationships and spatial patterns has gradually become one of the central foci in ecology (Forman and Godron 1986, Legendre and Fortin 1989, Wiens  10  1993). Landscape ecology is the study of the origin and characteristics of spatiotemporal heterogeneity within and between ecosystems. Advances in remote sensing and GIS have made the collection and analysis of large-scale high-resolution data possible, facilitating the testing of landscape ecological theory with empirical data. The study of spatial relationships and spatial distributions has centred on the degree and nature of spatial autocorrelation in the distribution of a variable (Legendre 1993). The term geostatistics is used in varying ways by different authors, but is ultimately related to mathematical operations that deal explicitly with the spatial locations of observations. These are generally based on the correlogram and variogram, which are designed to characterize spatial autocorrelation. Much of this effort has been directed at spatial interpolation in which the value of a variable at a given location is predicted using the value at adjacent locations and knowledge of the nature of spatial dependence of the variable. Considerable effort has also been directed at the study of spatial pattern in ecological variables (Dale 1999). This too is based on a quantification of spatial autocorrelation but in this case looking for characteristic scales of patches and gaps. The only known study of spatial pattern in the Arctic is that of Young et al. (1999), who compared the spatial pattern of vegetation in sedge meadows that were grazed with those that were not grazed and found spatial pattern difficult to discern in both communities. The presence of spatial autocorrelation in a sample is contrary to one of the fundamental assumptions of inferential statistics, that is, the independence of observations (Legendre and Fortin 1989). As a result, the effective degrees of freedom in any statistical analysis are reduced due to the violation of this assumption. The presence of spatial autocorrelation must be accounted for when looking for correlations between spatially distributed variables. Using classical statistics there is a greater chance of making a type I error, that is, finding a significant relationship when one does not exist. Several methods have been proposed to account for this spatial autocorrelation in bivariate regression (Clifford et al. 1989, Dutilleul et al. 1993, Haining 1991), but the methods have not yet been developed to the point where they are practical for general use, although some software packages do include them.  11  1.2.7 Climate Change in the Arctic  There is a general consensus that future global climate change will be amplified in polar regions (Houghton et al. 1995, Maxwell 1992) due to the albedo feedback, the effect of the arctic inversion and the transport of latent energy from the tropics (Maxwell 1992). Much evidence is already showing the results of these changes (Chapin et al. 1995, Myeni et al. 1997, Sereeze et al. 2000, Sturm et al. 2001) These predictions have significant implications for vegetation at high latitudes (Chapin et al. 1992a, Oechel et al. 1997). The implications of this warming are multi-faceted and characterized by a complex system of lags, responses, thresholds and feedbacks (Shaver et al. 2000) that are difficult to understand and simplify for the purposes of making accurate predictions about the consequences of climate change. Arctic vegetation is strongly influenced by climatic controls and is likely to be highly sensitive to changes in climatic conditions (Callaghan and Jonasson 1995). Sereeze et al. (2000) reported on a suite of biotic and abiotic changes that have occurred in the Arctic in recent history. Sturm et al. (2001) describe the increase in shrub cover in Alaska that has been observed. It is foreseeable that the now cold dry polar desert could come under temperature and moisture regimes that are more similar to the current conditions in arctic oases like Alexandra Fiord. 1.2.8 Vegetation Response to Climate Change  Vegetation response to climate change has been studied in several ways. The first has been to make observations and measurements of vegetation properties in different environments. The second has been to use experimental manipulations to change temperature, moisture and/or nutrient regimes and record the response (see section 1.2.9). The third method is the use of models, which predict plant responses in terms of different variables and at different scales. Different species and functional groups are expected to respond in different ways to environmental change (Chapin er al. 1996), and this is expected to bring about changes in plant community composition (Chapin et al. 1995, Henry et al. 1986b). In studying vegetation response to climate change, researchers have tried to separate the responses to changes in a suite of environmental variables but must accept a degree of uncertainty in predictions related to lack of understanding of interactions between variables (Jonasson 1997). Chapin et al. (1995) found an increase in shrub dominance in Alaska over 9 years of monitoring. Corneilesson et al. (2001) found that in the low and mid-Arctic lichen cover  12  decreased as vascular plant cover increased in response to climatic amelioration. This is consistent with Chapin and Shaver's (1985b) and Chapin et al.'s (1995) observation that changes in one plant species or growth form are often compensated for by changes in another, with the result being that production or biomass of an ecosystem remains roughly constant despite changes in species composition. The ways in which vegetation will respond to the alteration of several aspects of climate is of particular importance and is also particularly complex (Crawford et al. 1993, Kittel et  al. 2000). Differences in scale and the level of biological organisation considered (organs, plants, populations, ecosystems) complicate the study and prediction of ecosystem dynamics (Reynolds and Leadley 1992). The distribution of vegetation, rates of plant physiological processes and climate are inextricably linked at several scales and must be examined simultaneously if reasonable predictions are to be made about the response of vegetation to directional environmental change. "Vegetation type and distribution have large impacts on regional and global climate through effects on terrestrial carbon storage and on water and energy exchange" (Chapin and Starfield 1997, p.449). Climate-vegetation interactions are expected to be most important at biome boundaries (i.e. a shift of considerable land area from tundra to boreal forest could significantly alter regional climate patterns) and thus these areas are particularly sensitive to climate change (Chapin and Starfield 1997). Arctic oases could be important seed sources for the colonization of new areas if an amelioration of climate were to occur (Walker 1995). Large-scale equilibrium models predict that under forecasted climatic changes there will be a northward expansion of forested ecosystems and consequent reduction in the range of arctic tundra (Epstein et al. 2000, Lenihan and Neilson 1995, Prentice et al. 1992). But these new equilibrium conditions would take centuries to achieve and this timescale exceeds the immediate interests of humans and grazing animals (Epstein et  al. 2000). "In the face of climate change, vegetation has a certain inertia, comprising several components: (1) the time needed for individual patches of vegetation and associated labile soil components to adjust to climatic change when the available flora is held constant; (2) time for the spatial mosaic of vegetation to adjust to climatically induced changes in frequencies of fire and other natural hazards; and (3) time for ecesis" (Prentice et al. 1993, p. 236). Models with long-lived species that assume  13  equilibrium with climate will overestimate rates of change (Chapin and Starfield, 1997). As a result of different species responding in different ways to changes in climate, there are ensuing changes in composition, biomass, leaf area and competition. If time lags are not accounted for, species and functional types that migrate slowly will be under represented in the model predictions (Chapin et al. 1996). 1.2.9 The International Tundra Experiment (ITEX)  Small-scale experiments designed to simulate the conditions of predicted climate change and record the response of vegetation are numerous (Callaghan and Jonasson 1995). The International Tundra Experiment (ITEX) is a circumpolar network of 15 sites in which the responses of ten key circumarctic species to simulated warming are being monitored (Henry and Molau 1997). The ITEX approach is based on passively warming tundra plots using open top greenhouses (Marion et al. 1997). Species at all ITEX sites have shown measurable responses in terms of growth and phenology to the climatic treatments (Arft et al. 1999, Henry 1997). ITEX warming experiments were established at Alexandra Fiord in 1992. 1.3 Research  Aims and  Objectives  The prospect of significant climate change occurring in the High Arctic has made clear the need to study the unique ecological relations in this region. As such, a detailed understanding of the distribution of vegetation in terms of structure and function is vital if predictions are to be made regarding the possible response of vegetation to climatic warming. It is regularly assumed that the results of experiments and models can be scaled up to landscapes and ecosystems often without an explicit understanding of spatial or temporal variability (Arft et al. 1999). The intermediate landscape scale is appropriate for the purposes of developing functional models of ecosystem response to climate change (Chapin et al. 1995). Detailed data on current conditions are required if any long-term comparative analyses are to be carried out in the future. The examination of the vegetation patterns in a high arctic oasis is an opportunity to study an environment that approximates the predicted conditions for the rest of the High Arctic under climate warming, and the extent of the existing body of research at Alexandra Fiord makes it an ideal site to study vegetation patterns at the landscape scale. The detailed information on small-scale patterns and species-specific relationships with environmental factors will  14  be immensely valuable in interpreting the physical explanations for observed patterns and relationships. The purpose of this research was to study the diversity and response of plants in terms of phenology, growth and reproductive effort to natural gradients in snowmelt, temperature, soil moisture and soil nutrients in a high arctic oasis. Understanding the spatial distribution of the responses will help to place their variability into a habitat perspective and will also provide context for the responses measured in warming and snow manipulation experiments, such as those established as part of ITEX (Henry and Molau 1997). Also, the information gathered in this research constitutes detailed current baseline data that can be used to monitor future changes. This aim of this research has been to address the following questions: 1. What are the landscape-scale patterns of plant growth and phenological development in the lowland and what environmental parameters are most strongly correlated to those patterns? 2. How does the variability in vascular plant growth and phenology across the lowland compare to the variability observed in the ITEX warming experiments? 3. What are the broad-scale patterns of plant community distribution in the lowland and what environmental parameters are most strongly correlated to these patterns? 4.  How accurately can the tundra plant communities of Alexandra Fiord be classified using colour digital aerial photographs and detailed ground-truthing?  15  Chapter Two Methods 2.1 Research Site The research was carried out at the Alexandra Fiord lowland (78° 53' N, 75° 55' W) located on the eastern coast of Ellesmere Island (Fig. 2.1). The 8 km lowland is an oasis relative to the 2  surrounding polar desert and has been the subject of considerable ecological research over the last 20 years (Henry 1987, Svoboda and Freedman 1994) and hosts an International Tundra Experiment (ITEX) site. The lowland is a roughly triangular outwash plain, sloping gently (1-3%) upwards from the shoreline at its northern border. In the north end of the lowland the topography is dominated by a series of well-defined beach ridges. At the southern end there are two outlet glaciers, and 500-700 m cliffs and talus slopes border the east and west sides. The lowland is drained by a glacier-fed river and three smaller streams. Alexandra Fiord is protected from grazing by caribou and muskoxen due to the local topography (Henry et al. 1986a). Local geology is composed of gneiss, pegmatite, granite and dolostone (Sterenberg and Stone 1994). The soils are rich in organic matter relative to most high arctic sites and include Regosolic, Gleysolic and Brunisolic Cryosols (Muc et al. 1994a). Soil pH is generally acidic, indicating that most parent material is granitic (Muc er al. 1994a). Alexandra Fiord is situated within the Nares Strait sub-region of the Northern region (V) in Maxwell's (1981) climatic classification. Temperatures are higher in the lowland due to protection from the meso scale wind, and reflection and re-radiation of insolation, from the surrounding cliffs and ice (Labine 1994). Precipitation is minimal with less than 1 cm generally falling during the growing season. Average annual temperature is -14.8°C, with the warmest month being July with an average temperature of 6.7 °C. Since 1989 there have been on average 428 thawing degree-days (TDD) in the growing season with a minimum of 315 in 1996 and a maximum of 492 in 1998. The 2000 growing season was slightly warmer than average with 481 TDD. Maximum winter snowdepth averages 135 cm at the central climate station (Fig. 2.3) but is highly variable across the lowland with snowbanks forming in the troughs of the beach ridges and exposed areas blown free of snow by wind.  Vegetation  Roughly 90% of the lowland is covered by closed or semi-closed vegetation (Muc et al. 1989), which is significantly greater than the 5% vegetation cover of the surrounding polar deserts (Bliss et al. 1994). There have been 96 species of vascular plants identified (Ball and Hill 1994), 104 taxa of mosses and liverworts (Maass et al. 1994) and at least 119 taxa of lichens 16  Alexandra Fiord Lowland Fig. 2.1 Maps showing the location of the Alexandra Fiord Lowland. 17  (Mass and Nams 1994). Thirty-seven vascular plant species are found in the adjacent uplands (Batten and Svoboda 1994). Approximately 50% of the lowland is covered by a xeric-mesic dwarf-shrub heath consisting primarily of Cassiope  tetragona,  Dryas integrifolia,  and Salix  arctica (Table 1). Wet-mesic soils support sedge meadows on 20% of the lowland and the remainder is covered by various other plant communities and some rock outcrops and barren ground (Figure 2.2). Table 2.1.Six plant communities of the Alexandra Fiord lowland identified by Muc et al. (1989) and some cover characteristics. S-CP-DS - sedge-cushion-plant-dwarf-shrub, DS-CP - dwarf shrub cushion plant, L-CP-DS - lichen-cushion-plant-dwarf-shrub, HERB - herbaceous, DDS-G - deciduous-dwarf-shrubgraminoid, SM - salt marsh. % of land area % plant cover % lichen cover  S-CP-DS  DS-CP  L-CP-DS  HERB  DDS-G  SM  28.4 109 12  18.7 108 31  37 73 41  5.3 13 1  3.8 89 23  0.1 16 0  2.2 Sampling Methodology Pilot Study It has been shown that, when interpolating spatial data, it is the relative location of the sampling points and not the total number that is more important (Legendre and Fortin 1989). A pilot study was conducted to determine the best method of sampling the lowland for the purpose of characterizing the spatial pattern of variables. The study is summarized below and a thorough description of the methods and results are presented in Chapter 3. The number of sampling sites for recording temperature was limited by availability of 30 dataloggers and it was judged that 30 sampling sites was as many as was feasible for sampling and monitoring all of the proposed variables during a single growing season. An existing map of the pattern of snowmelt for the Alexandra Fiord lowland in 1982 created by Woodley and Svoboda (1994) (Fig 3.1) was used to test the accuracy of different sampling schemes because snowmelt is considered to be a surrogate for temperature and micro-topography. Using a GIS, systematic, stratified, and random point patterns were established and the date of snowmelt was assigned to each point based on the value from Woodley and Svoboda's map. A surface representing the snowmelt date was interpolated using the inverse distance weighted method and this surface was then compared to the original map for accuracy of representation of the original distribution. The observed values were then subtracted from the interpolated values to arrive at a measure of the error associated with each interpolated pixel in the GIS. All three interpolated surfaces showed a good correspondence to the original map though they revealed different patterns in the interpolation error associated with the predicted snowmelt  18  Figure 2.2 Plant communities of the Alexandra Fiord lowland (from: Muc ef al. 1989) 19  dates. As expected, the snowmelt date for the late-lying snowbeds was underestimated by all three sampling strategies. Early melting patches adjacent to the western and central melt-water streams were consistently overestimated. The systematic sample interpolation estimated many more pixels within two days of their true value but also classified a considerable number at greater than 14 days from the observed date. The overall Sum of Squared Deviations was lower for the stratified sample interpolation, yet it was evident from the map of the differences (see Chapter 3) that a much larger area was overestimated. Based on the results and the greater ease of implementation in the field the systematic sampling scheme was chosen. 2.3 Field and Laboratory Methods Data Collection took place between June 5 and August 15, 2000, which incorporated the majority of the growing season. For the remainder of this thesis dates are expressed as day number (day of the year with January 1 being 1). Upon arrival the lowland was still snowst  covered with the exception of a few snow-free patches where plants and insects were already active. A grid of 28 sampling points was established across the lowland at a distance of roughly 485 m between points (Fig. 2.3) and the geographic coordinates of each of the sampling points were recorded using a handheld GPS receiver. The arm of the lowland that extends towards the northwest (Fig. 2.2) was excluded to facilitate the collection of field data. Due to warm temperatures, there was extensive flooding in the lowland and one of the sampling points was inundated following a high rain event on day 188 which affected to plant community composition observations at that site.  2.3.1 Temperature  Measurements  At each of the sampling points a temperature logger (Hobo Pro, Onset Computer Corp., Pocaset, MA) was installed. The data loggers were attached to a plastic stake and installed in the ground so that the temperature sensor was 10 cm above the ground surface recording the air temperature every ten minutes. White plastic plates were attached to the stake above the datalogger to shade it from direct sunlight. During high wind events some of the sunshades were blown-off and had to be replaced. Four dataloggers were in place by day 165, and all but 6 installed by day 167. Site 3 was installed on day 174, corresponding to the late date of snowmelt at this site. For the first half of the season the data logger at site 21 failed to record any data. The data were downloaded to a portable computer twice during the course of the study period.  20  07 02  03  10  O *  01  4  °.  S  06  5  11 M- "  W  9  1.3  V  1  4  D  16  *  17  08  18  2,0  21  23  27  29 climate station sampling sites  •  ITEX sites  river channels ^^^^  25  28  N  30  2km  0 lowland  22  24  26  it •  19  m m  ^^^~.  Figure 2.3 Locations of sampling sites. Red squares are ITEX sites: M: Meadow, W: Willow, C: Cassiope, D: Dryas, and S: Saxifraga. 21  2.3.2 Plant Phenology and Growth Four of the most common vascular plant species were selected for studying the patterns in phenology across the lowland. These were: i. Cassiope tetragona (L.) D. Don. (Ericaceae), an evergreen shrub; ii. Dryas integrifolia M. Vahl. (Rosaceae), a wintergreen shrub; iii. Saxifraga oppositifolia L. (Saxifragaceae), a wintergreen perennial herb; and iv.  Salix arctica Pall. S. Lat. (Saliaceae), a dioecious deciduous shrub.  At the sampling points, five individual plants of each of these species, if present, were chosen at random and labelled with a metal tag. Phenological observations were recorded every 4 to 6 days throughout the study period following the ITEX protocols (Molau and Edlund 1996). At the peak of the growing season measurements were made on the heights of the Dryas pedicels and the leaf fascicle length in Salix. Descriptions of each of the phenological stages and growth measurements are provided in Appendix A.  Dryas was the most common species, being found at 26 of the 28 sites, and Saxifraga was the least common being present at 21 sites (Fig. 2.4). At site 03 Dryas plants did not produce any flowers so the actual sample size was 25 and at site 03 and site 25 Saxifraga did not produce any flowers so the results are based on 19 sites. Salix arctica was present at 25 of the 28 sites where four sites had only vegetative shoots tagged, two sites only reproductive shoots were tagged, and 9 sites had no male reproductive shoot tagged. 60% of the tagged shoots were vegetative (without flowers), 20% were male and 20% were female. 2.3.3 Plant Cover  Plant cover at each site was recorded using the point-quadrat method (Walker 1996). A 1 m X 1 m metal frame with a grid of string at 10 cm intervals was placed at a random location within 10 m of each sampling point. At each of the 100 intersections of the grid a metal pin was held between the string and the ground surface and each plant species that touched the pin was recorded. Nomenclature follows Porsild and Cody (1980). This was done at four random locations at each site and the data were combined for analysis as percentage cover for each species and ground cover type. 2.3.4 Soil Moisture and Nutrients  Soil samples were collected using a 2 cm wide soil corer to a depth of 10 cm where sufficient soil was present. Samples were collected twice during the study period: the first set between  22  A. Cassiope  B. Dryas  • • •  tetragona  integrifolia  #  _  #  • • -  C. Salix arctica  - Saxifraga  oppositifolia  D  • •  »  •  »  •  Figure 2.4 Locations where phenology was monitored for each plant species. A. Cassiope tetragona, B. Dryas integrifolia, C. Salix arctica, and D. Saxifraga  June 26 (178) and June 28 (180), and the second between July 24 (194) and July 26 (196). The samples were weighed, dried for 48 hours in a field lab oven (~40°C), and then weighed again. The difference between these weights is the soil moisture content. The second set of samples was brought back for nutrient analysis. Soils were sieved and the fraction that was less than 2 mm were analyzed for total carbon and nitrogen in a C:N analyzer.  2.3.5 Aerial  Photography  Digital aerial photographs of the entire lowland were taken from a helicopter at three times over the course of the study period. The photographs were taken manually at an average altitude of 1600 m a.s.l. using a Canon Powershot S20. Two sets of photos were taken to capture the pattern of snowmelt in the lowland on June 12 (164), June 16 (168), and one set was taken during the middle of the growing season July 14 (196) for the purpose of mapping the vegetation. Before the third set of photographs was taken a yellow garbage bag was laid out adjacent to each of the sampling sites to aid the mosaicking and rectification of the images based on the geographic locations of each of the points determined using the GPS. The GPS was also used to record the locations of prominent features in the lowland such as large erratic boulders for use in mosaicking and rectification. The aerial photographs of snowmelt were supplemented by oblique photographs taken from the East ridge on June 10 (162), June 11 (163), June 14 (166) and June 21 (173). 2.3.6 Mosaicking Digital Aerial  Photographs  Mosaicking the digital photographs began with the set taken in the middle of the growing season. Most images contained only two of the ground control points established at the sampling points and marked with the yellow garbage bags. The combination of the low number of ground control points per photograph and the level of precision of the handheld GPS did not allow for the geo-referencing of each individual photograph, so photo mosaicking was performed manually in CorelDraw (V. 9, Corel Corp.) using common features in adjacent photographs. The resulting mosaic was rectified using Warp in ArcView (ESRI Corp.) using the location of all 28 ground control points. The same procedure was followed to mosaic and georeference the other two sets of photographs.  2.4 Data Analysis 2.4.1 Plant Cover and Communities  The cover data from the four point-quadrat samples at each site were combined and these values were subject to a classification using complete linkage cluster analysis on unstandardized data using JMP-ln (V. 3.1.5) and ordination using principal components 24  analysis (PCA). Unstandardized data takes total abundance of species into account and clustering is based more on the presence of the dominant species rather than giving minor species equal weight. Several classification methods were tested but the complete linkage approach  produced  the  most  clear  division  into  interpretable  groups.  Detrended  correspondence analysis (DCA) was chosen initially assuming a unimodal response in the vegetation and to maintain consistency and comparability with Muc et a/.'s (1989) study. The DCA ordination explained 40% of the variance of the species data with the first two axes which was less than with PCA. Also, the lengths of the gradients were only 2.9 and 2.3 standard deviations for the first and second axes, respectively, and this does not indicate a strong unimodal response. Ter Braak and Prentice (1988) state that a gradient length below 3 standard deviations represents only a weakly unimodal response. For this reason, PCA, which assumes a linear response to gradients, was used. All ordination analysis was done using CANOCO (V. 4.0, 1998). Direct gradient analysis was also undertaken using environmental variables to constrain ordination axes. Redundancy analysis (RDA) is a constrained linear ordination technique and this was used to analyse the relationship between vegetation community composition and measured environmental variables. RDA explained more of the species variance than the unimodal constrained correspondence analysis (CCA). Forward selection was used to eliminate some environmental variables that were not contributing significantly to the ordination. The resulting list included mid-season soil moisture, early-season temperature, mean temperature, average daily minimum temperature and soil carbon content. Snowmelt date would have been a useful variable to add to the RDA. It was hoped that this would be derived from the snowmelt maps given that the sampling grid was established after snowmelt. However, due to inadequate spatial control and the fine scale of the pattern of snowmelt, these dates could not be derived from the aerial photographs. Using Borcard et a/.'s (1992) technique, the purely spatial component of vegetation community composition was partialed-out. They recommend using all of the terms for a cubic trend surface in the spatial coordinates matrix to allow for non-linear trends, but using forward selection it was determined that only the X and Y coordinates were applicable in this case. In CANOCO, RDA was run on species data and environmental data while using the geographic coordinates as co-variables and then again switching environmental data and geographic coordinates. Using the results it was possible to separate species variance that was purely spatial, purely environmental, inseparably correlated spatial and environmental, and unexplained.  25  2.4.2 Vegetation Community and Snowmelt Mapping  The purpose of collecting the aerial photographs in mid-season was to use the vegetation community composition data at the sampling points to ground-truth a photo interpretation of the distribution of plant communities. For each of the sampling points a representative area was selected and these were grouped by community type as defined by the cluster analysis in order to act as a reference for interpreting the photo mosaic. However, the similarity between community types in terms of dominant species and ground cover types made the differentiation of vegetation types on the mosaic impossible. Some patches are readily identifiable but the majority of the plant communities are not distinguishable. The intention was to perform quantitative analyses on the relationship between snowmelt date and vegetation community distribution. The inaccuracies of the ground control points for proper rectification of the images resulted in inadequate spatial control for exactly overlaying the three photo mosaics. Interpretation of the significance of the pattern of snowmelt in community composition was instead done visually by comparing the snowmelt pattern to the distribution of community types according to the plant community analysis described above and compared to Muc's (1989) vegetation map (Fig. 2.2). 2.4.3 Temperature and Phenology Data  Temperature data from June 15 (167) to August 10 (223) were used. During some wind th  th  events some sunshades were blown-off the data loggers and these data gaps were filled by linear regression with most highly correlated site. All R values were greater than 0.9 for the 2  data filling regressions. The dataloggers were calibrated by leaving them all in the same room for 24 hours and calculating the deviation from the mean for each logger and the data were adjusted accordingly (-0.4°C to+0.4°C). The phenological dates for the five monitored plants at each site were averaged. For some species not all plants flowered at each site, so the mean might be based on as little as one value. In some cases plants produced buds or flowers late in the season. These events represent a 2  nd  wave of flowering, which is not very common (S0rensen 1941, Edlund et al.  2000) and were, therefore, not used in the calculation of mean dates for phenological events. Phenological dates were compared to thawing degree-days (TDD: the sum of daily average temperature values above 0°C). These were summarised on a 5-day basis such that TDD195  26  is the sum of average daily temperatures above 0°C from the start of the growing season until day 195. This does introduce some error as the record for all of the dataloggers begins on the same date but sites were snow-free for differing lengths of time before the record begins. Site 03 was located within a snowbed and did not melt until more than 7 days after the majority of other sites had melted. The relationship between average temperatures and TDD was almost perfectly linear with the exception of site 03, where late snowmelt was not reflected in average temperatures although it had much lower TDD. Spearman's Rho was used as the correlation coefficient to describe all relationships. It is not as powerful as a parametric test but many variables did not satisfy the necessary assumptions, so the non-parametric test was used for all cases to maintain consistency. 2.4.4 Spatial Analysis  The first step in analysing the spatial relationships among variables was to characterize any spatial autocorrelation in each variable. There are several methods for describing the nature of spatial autocorrelation and several of these were considered. These methods generally evaluate the variance of points within a given lag distance of each other relative to the overall variance of the dataset. Variograms have been found to be sensitive to outliers and Rossi et al. (1992) suggest using correlograms instead or in addition to variograms. Variogram and correlogram analysis is generally considered to be meaningful at distances less than half the greatest distance between points in the data. Omni-directional variograms and correlograms using both Moran's I and Geary's C were produced for four lags with intervals of 485 m (the distance between points in the sampling grid). The semi-variance was calculated using Geoeas (V. 1.2.1,.U.S.-E.P.A.) and the Moran's I and Geary's C were obtained using Rook Case (Sawada 1999). Most authors suggest 30 as the absolute smallest number of samples for the production of a meaningful variogram (e.g. Rossi et al. 1992). The number of samples for phenological dates ranged from 21-28 sites making the variogram and correlogram analysis weak. Moran's I ranges from - 1 to 1, and has an expected value of 0 representing a random distribution with positive values indicating positive spatial autocorrelation and negative values indicating negative autocorrelation. Geary's C ranges from 0 to 2 with values from 0 to 1 corresponding to positive spatial autocorrelation, 1 is the expected value for a random distribution, and values from 1 to 2 represent negative spatial autocorrelation. An Arc View (ESRI Corp.) coverage was created using the geographic coordinates of each of the sampling sites with the data on the biotic and abiotic variables linked to this file. Surfaces 27  representing the data collected at each of the sampling sites were generated using a 2  nd  order  polynomial inverse distance weighted spatial interpolation with six nearest neighbours. This means that the value assigned to each pixel was a function of the values of the six closest points weighted according to how close they are to the pixel being estimated. Interpolated values are, therefore, based on points within 748 m of the point being estimated. The final GIS database included the following layers: 1. Mosaic of digital aerial photographs June 12; 2. Mosaic of digital aerial photographs June 16; 3. Mosaic of digital aerial photographs July 14; 4. Temperature surfaces interpolated from point values; 5. Phenology surfaces interpolated from point values; 6. Point values of soil moisture, soil nutrients and plant cover characteristics; 7. Digitized map of Muc etal.'s (1989) vegetation communities; and 8. Digitized map of Woodley and Svoboda's (1994) map of the snowmelt pattern in 1982. 2.4.5 Comparison with ITEX data  Data on plant phenology and climate have been collected at Alexandra Fiord since 1992 as part of the International Tundra Experiment (ITEX). These data were used to compare the range of phenological dates observed over time at a single location with the range of dates for phenological events found in the current study. All the phenological dates for Dryas integrifolia and Cassiope  tetragona in un-manipulated control plots at all ITEX sites (Fig. 2.3) were  averaged for each year and the earliest and latest dates for each stage were compared to the earliest and latest dates at the sampling points in the current study. ITEX temperature data were measured at screen height at a single location in the central part of the lowland. In analysing temperature, the season length was held constant to compare to the data in the current study (day 167-224). In fact, differences in the time of the beginning of the growing season and growing season duration are important components of temporal variability.  2.5 Summary  Sampling was carried out at the Alexandra Fiord lowland on the east coast of Ellesmere Island. At 28 sampling sites arranged in a regular grid data were collected on vegetation composition, phenology and environmental attributes. Data were analysed within a GIS environment to characterize spatial patterns and spatial relationships among biotic and abiotic variables. The results of a pilot study to evaluate the effectiveness of different sampling schemes are described in the next chapter. 28  Chapter Three Using GIS to Compare Spatial Sampling Strategies 3.1 Introduction Until relatively recently, the majority of sampling strategies have been designed to estimate the population parameters of a variable based on a representative sample. However, the fact that the collected data are spatially distributed violates the rule of independence of the observations due to the effects of spatial autocorrelation (Legendre 1993). Analysis of spatially distributed data requires methods that account for spatial autocorrelation and which provide information on this characteristic of the data. With the development of geostatistics, the analysis of spatial data is becoming increasingly sophisticated, and the way in which these data are sampled should be designed to optimize the utility of geostatistical methods. Using simulated sampling in a geographic information system (GIS), the effectiveness of different sampling strategies in representing the spatial distribution of a variable can be evaluated prior to undertaking a field sampling program (Atkinson 1996, Fortin et al. 1989, Neldner et al. 1995). This chapter presents the results of a preliminary investigation into the effect of sampling strategy on the accuracy of spatial interpolation of the date of snowmelt at the Alexandra Fiord lowland. The accuracy of interpolations based on random, stratified and systematic sampling of an existing dataset provided an evaluation of the representativeness of each of the sampling strategies.  3.2 Sampling Strategies Sampling which aims to estimate the parameters of a population depends on two conditions being met. Each point must have an equal chance of being included in the sample and each point may only be sampled once (Maling 1988). Increasingly, data are being collected for the purpose of studying their spatial pattern and not for estimating population parameters (Fortin et al. 1989). Collecting data for the purpose of studying its spatial distribution has different aims and requirements but, for the most part, the approaches to sampling have been the same. Sampling strategies generally fall into one of three categories: systematic, random and stratified. There are also methods that combine these approaches as well as transect sampling for specific applications.  29  Random sampling, as the name implies, involves distributing the sampling points at random, ensuring that each point in the area of study has an equal chance of being sampled. Theoretically, the number of samples per class or category should be proportional to the area of each class (Wolcott and Church 1991). However, some classes will be over represented and some underrepresented unless the sample is very large (Maling 1988). Though the selection of each sample location is independent, the presence of spatial autocorrelation in the data still limits the use of inferential statistics. Stratified sampling involves classifying the area based on prior information on the variable of interest, or another related variable, and distributing the number of sampling points evenly or in some weighted fashion among the classes. This is designed to ensure the representativeness of the sample (Cochran 1977). A good example would be delineating vegetation classes visible in air photographs and stratifying sampling sites based on those regions. GIS has been used as a tool for stratifying study areas based on several variables (Del Barrio etal. 1997, Goedickemeir et al. 1997). Stratified sampling has the advantage of providing data on each sub-population that corresponds to each stratum (Cochran 1977). More advanced approaches to stratification include the regression tree approach used by Michaelsen et al. (1994) to stratify multivariate satellite data for ground-truthing. This is one of the most important applications of spatial sampling (Congalton 1991). A systematic sampling network is a grid of sampling points that are evenly spaced across the area of study. Systematic sampling has three main weaknesses. First, if the sampling interval corresponds to periodicity in the data, the sample will not be representative. Second, if the sample spacing is larger than the smallest patches they will tend to be underrepresented. Finally, it is often stated that inferential statistics cannot be used with data collected through systematic sampling because the location of each sample is determined by, and therefore not independent of, the first sample. Systematic sampling is, however, often the easiest approach to implement in the field and may be ideal for presampling pilot studies. There are also hybrid approaches such as cluster sampling in which several points are sampled within randomly placed quadrants. Another method is systematic unaligned sampling in which a random point or several points are selected within each quadrant of a grid established over the study area. This method thus ensures that the entire study area  30  is sampled evenly without the possibility that periodicity in the data is being sampled. Transect sampling can be random or systematic, depending on how the sample points are selected along the transect and whether the location and orientation of the transect are random. Though many sampling programs are designed to estimate population parameters, others aim to identify natural groupings or classes within the data. This is used particularly in studies of vegetation in which the identification of distinct vegetation communities is the goal. In this case, though the samples are located in space, the aim is not to characterise the spatial pattern, but to gain insight into the number and separability of groupings in the data. A further sampling strategy for vegetation is that of the Braun-Blanquet school in which sample locations are subjectively judged to be representative of the surrounding vegetation. The inherent bias and lack of reproducibility are considerable weaknesses of this approach. Computer simulated sampling has been used since the early 1960s (Podani 1984). A number of studies have used field and simulated data to evaluate the utility of different sampling schemes and sample sizes for vegetation (Atkinson 1996, Fortin et al. 1989, Goedickemeier et al. 1997, Legendre and Fortin 1989, Lo and Watson 1998, Neldner 1995, Podani 1984, Smartt and Grainger 1974), land use (Harrison and Dunn 1993) and river gravels (Wolcott and Church 1991). The ineffectiveness of random sampling is evident in virtually all of these studies. Atkinson (1996) showed how GIS could be used to optimize sampling efficiency using the variance of kriging estimates. It has been shown that, when interpolating spatial data, it is the relative location of the sampling points and not the total number that is more important (Fortin et al. 1989).  3.3 Methods The original map of snowmelt patterns on the Alexandra Fiord lowland from Woodley and Svoboda (1994) was digitized and imported into the GIS (ArcView, ESRI Corp.). The map has classified the area into five snowmelt dates, which are expressed in day numbers (Fig. 3.1). Snowmelt was used to define the accuracy of the sampling schemes because it is considered to be a surrogate for temperature and micro-topography. It was determined that a sample of 30 points was the maximum possible in terms of equipment costs and the feasibility of carrying out all of the proposed sampling in the field (see Ch. 2). This is,  31  Fig. 3.1 The pattern of snowmelt in the Alexandra Fiord lowland. Dates are in day numbers. Modified from Woodley and Svoboda (1994).  Stratified  • •  • '  Mtttk  « • • • • • *.* •  • *.  *  w  • •  * •  Random  * * * * * . . . .  * •  Systematic  "Si  * ; ;  •*  Fig. 3 . 2 Location of sampling points for each of the three sampling strategies.  32  however, cited by many authors as the minimum number of samples that should be used when undertaking spatial statistical analysis (Fortin et al. 1989). The three sampling schemes were established (Fig. 3.2) and the snowmelt date for each of the points was assigned according to the value from the digitized map. The streams that drain the valley have a very different snowmelt pattern and were therefore not included in the analysis. This resulted in not using one of the points in the southern part of the systematic sampling grid. Both the systematic and random sampling schemes did not have any sample points in the class with the smallest area (Table 3.1). Table 3.1 The five snowmelt dates and associated areas, and number of sampling points per date for each of the sampling strategies. Snowmelt (days) 150 157 163 166 175 streams  Area (m ) 2  224,948 117,904 2,763,154 3,240,710 227,859 827,066  Points per snowmelt date Random Systematic Stratified 2 1 3 0 0 2 8 11 12 13 18 11 3 3 3 0 0 0  To gauge the effectiveness of each sample design in characterizing the spatial pattern of snowmelt across the lowland the values at the points were used in a spatial interpolation to produce a cover of the entire lowland. Interpolation was based on the inverse distance weighted approach using a minimum sample of five nearest neighbours. An interpolation using the spline method was also calculated but proved to have a much larger error than the inverse distance weighted approach, and so was not used. The observed values were then subtracted from the interpolated values to arrive at a measure of the error associated with each interpolated pixel in the GIS. If the interpolated date was later than the observed date, this is referred to as overestimating the snowmelt date in the discussion. The date was underestimated if the predicted value was less than observed. The overall accuracy of each interpolated surface was based on the sum of squared deviations (SSD) used by Smartt and Grainger (1974) and is defined by:  SSD  = E(  P i  -Xi)  2  [eq.  1]  where p is the interpolated value of each pixel and Xj is the observed value. {  33  3.4 Results All three interpolated surfaces showed good correspondence to the original map, although they reveal different patterns in the interpolation error associated with the predicted snowmelt dates. The snowmelt date for the late-lying snowbeds was underestimated by all three sampling strategies. Early melting patches adjacent to the western and central meltwater streams were consistently overestimated. The random and systematic interpolations revealed a slight tendency to overestimate the snowmelt date, whereas the stratified interpolation tended to underestimate the date (Fig. 3.3). The random interpolation estimated 55% of the pixels within two days of the observed values while it was 58% and 66% for the stratified and systematic interpolations, respectively (Table 3.2). The systematic interpolation classified over 1000 pixels at greater than 14 days, compared to 78 and 129 for the random and systematic interpolations, respectively. Despite estimating the largest number of pixels within two days of the observed value, the overall error for the systematic sample, as determined by the SSD, was 14% larger than the SSD for the stratified sample interpolation. The SSD for the random sample interpolation was greater than twice that of the systematic sample interpolation. Table 3.2 The interpolation error associated with each of the sampling strategies. Values are the percentage of the total number of pixels. SSD is the sum of squared deviations. Interpolation error -18 - -12 -12 - -7 -7 - -2 - 2 - 2 2 - 6 6 - 1 0 10 - 16 SSD  Systematic  Stratified  (%)  (%)  0.0 1.7 6.3 67.0 9.0 3.2 2.9 936,875  0.5 1.6 27.1 57.0 10.5 2.3 1.1 810,227  Random  (%)  0.2 1.6 15.6 54.5 19.9 6.6 1.6 2,067,204  3.5 Discussion These results support previous research that demonstrates random sampling to be the least effective of the sampling approaches. It is not immediately apparent whether the stratified or systematic approach produced the most accurate interpolation. The systematic sample interpolation estimated many more pixels within two days of their true value but  34  Fig. 3 . 3 . Maps and graphs showing the interpolation error in snowmelt dates for each of the three sampling schemes. Interpolation error was calculated as interpolated minus observed snowmelt dates (day numbers).  35  also classified a considerable number at greater than 14 days from the observed date than the stratified sample. The overall SSD is lower for the stratified sample interpolation, yet a much larger area (10%) was overestimated. The patchiness of the snowmelt distribution would require a much larger number of sampling points to characterise its spatial pattern with a high degree of accuracy. The purpose here is simply to identify which sampling strategy is best able to reproduce the overall trends. Considerable sampling effort would be required given, for example, the proximity of the late-lying snowbeds to the early-melting patch near the western meltwater stream. It is also important to consider finer scales of variability that exist within each of these patches that could significantly affect the results once this sampling program is established in the field. The late-lying snowbeds of the Alexandra Fiord lowland are a perfect example of periodicity in a dataset. If the sampling interval in the systematic grid corresponded to the distance between paleobeach ridges, it is conceivable that this could result in these late snowmelt dates being under- or over represented. The patches associated with these beach ridges are quite small and, given the limited number of points in this study, were underrepresented in the data. Stratified sampling is designed to collect data on each stratum in proportion to its total area. Though this is adequate for estimating population parameters, when characterising the spatial distribution of a variable, the extreme classes may be small and widely dispersed, as is the case with the late-lying snowbeds of the Alexandra Fiord lowland. In order to fully capture the spatial pattern, a sampling point would have to be located within each of the snowbeds and this is simply not practical, especially in a single season study. When sampling to estimate population parameters, it is assumed that if the extremes on either side of the mean are not sampled, that these errors of omission balance each other out. However, when the spatial distribution is being studied, the location and extent of extreme values will often be of central importance. Based on these results, the systematic grid was deemed to provide the best representation of the pattern of snowmelt with the added advantage of being relatively easy to implement. Further refinement of the systematic sampling strategy could be  36  achieved by modifying the position and orientation of the grid. The current grid was placed randomly over the area and included several points that were on the margin of the lowland and may not be representative of the patches within which they fell. Field measurements would be required to determine the nature of the boundaries between patches that may, in reality, not be patches at all but areas of early and late snowmelt that grade into one another. Shifting the starting position of the grid or changing its orientation could result in a distribution of points that is more representative of the snowmelt pattern and does not contain points that fall within the meltwater streams, but this was not done in this study. The only known comparable study to this one was undertaken by Fortin et al. (1989) and involved sub-sampling an initial set of 200 observations using random, systematic and systematic-cluster methods. They found that the non-systematic methods were more successful in reconstructing the spatial distribution of variables using kriging interpolation while acknowledging that these are generally the most difficult to implement in the field. The shortcoming of systematic sampling is that the nature of spatial autocorrelation at distances less than the sampling interval remains unknown. Sampling is always a balance between precision of observation and necessary effort (Maling 1988, Michaelsen et al. 1994). Pre-sampling pilot studies provide valuable information on the spatial distribution of a given variable and can lead to great improvements in the resulting sampling program (Fortin et al. 1989). Evaluating the effectiveness of a sampling strategy before it is undertaken will maximize the degree of precision for a given amount of effort. The best sampling strategy will vary among locations and variables but, with some knowledge of the spatial distribution of the variable under investigation, the sampling strategy that best suits the given circumstance and the purpose of the study can be determined.  37  Chapter Four Spatial Relationships Among Plant Phenology, Growth and Abiotic Variables 4.1 Introduction In this chapter are described the results of analyses aimed at exploring the spatial relations among plant phenology, growth and abiotic variables. First, the climatic context of the 2000 growing season is presented and compared to the overall climate record at Alexandra Fiord. Then, the spatial pattern of the abiotic variables is described. The spatial and temporal patterns of phenological development and growth and reproductive measures for each of the four plant species are presented with reference to the environmental parameters which are correlated to those patterns. The discussion is centred on the biotic-abiotic relationships and issues related to spatial analysis methodology.  4.2 Environmental Variables 4.2.1 Soil  Nutrients  The raw percentage of carbon in the soil samples ranged from 1 to17%, averaging 4%, and raw nitrogen content ranged from 0.04 to 1.28%, averaging 0.31%. Once adjusted by the soil bulk density the values ranged from 0.014 g/cm to 0.067 g/cm averaging 0.035 g/cm of 3  3,  3  carbon, and ranged from 0.0005 g/cm to 0.005 g/cm of nitrogen, averaging 0.002 g/cm . The 3  3  3  C:N ratio ranged from 12 to 29 with an average of 16. There was a very strong positive relationship between the abundance of carbon and nitrogen and a somewhat negative relationship between nutrient amounts and the C:N ratio. Hence, soils with higher nitrogen concentrations had a lower C:N ratio, which translates into higher quality soil. There was a significant relationship between all nutrient variables and mid-season soil moisture (carbon: Rho=0.67, p<.001; nitrogen: Rho=0.71, p<0.001; C:N: Rho=-0.72, p<0.001). Most of the higher quality soils were found in a band across the lowland beginning in the northwest corner and crossing southeast to the centre of the study area (Fig. 4.1A-C). There were also sites with higher quality soils in the northeast and southeast corners of the lowland.  4.2.2 Soil  Moisture  Early season soil moisture averaged 26.7% with a maximum of 66.4%, a minimum of 5.4% and a standard deviation of 14.7%. There was not a distinct spatial pattern in the distribution of early season soil moisture, though the southwest tended to be drier as were two adjacent sites in the north-central part of the lowland (Fig 4.1 D). The lack of a visible pattern in the point data is consistent with the variability in moisture status associated with microtopography. For this  38  A. Soil Nitrogen content  q/cm 0.000-0.001 • 0.001 - 0.002 • 0.002 - 0.002 • 0.002 - 0.003 • 0.003 - 0.005 3  B. Soil Carbon content  D. Early Season Soil Moisture  % 5.4 16.6 . 16.6 24.4 • 24.4 37.5 • 37.5 49.4 SI 49.4 66.4  E. Mid Season Soil Moisture •  g/cm 0.014-0.018 • 0.018-0.025 • 0.025 - 0.039 • 0.039 - 0.044 • 0.044 - 0.067  9  %  3  °A 4.6 -10.2 10.2 -18.3 18.3-31.9 31.9-42.8 42.8-61.2  C. Carbon:Nitrogen Ratio  ratio 12.0-13.0 13.0 15.0 15.0 16.7 16.7 20.2 20.2 28.9  Figure 4.1 Soil carbon (A) and nitrogen (B) content, soil C:N (C), and soil moisture in the early (D) and mid (E) growing season at each sample site in the Alexandra Fiord lowland.  39  reason, an interpolation of soil moisture data is not presented, as the surface would be meaningless in its generalization. Mid-season soil moisture had an average of 24.7% a maximum of 61.2%, a minimum of 4.6% and a standard deviation of 15.5%. The north-central and southwest comers of the lowland were drier and the west central and northeast areas had higher soil moisture levels (Fig 4.1 E).  4.2.3 Climate I  Temperature  The 2000 growing season was warmer than the long-term average at Alexandra Fiord (Table 4.1). The average near-surface (+10cm) temperature across the lowland during the growing season of 2000 was 8.5°C, with the highest temperature recorded being 20.8°C (day 176) and the lowest was -2.0°C on the last day data were recorded (day 223, Fig. 4.2). There was a strong zonation in most of the temperature data, with colder temperatures occurring near the shoreline in the north and particularly northwest portion of the lowland, and warmer temperature to the south in the area protected by the cliffs on the east and west sides (Fig 4.3). Average temperatures showed a pattern of increase moving from northeast to southwest, though the far northeast corner tended to be warmer than its surroundings (Fig. 4.3A). Average daily minimum temperatures were lowest along the eastern side of the lowland and become warmer towards the west (Fig 4.3B). Points along the eastern side are shaded throughout the morning whereas the rest of the points are sunlit throughout the day. Average daily maxima followed a similar pattern to average daily temperature, increasing from north to south with the northeast being warmer than the northwest portion of the lowland (Fig 4.3C). The accumulation of thawing degree-days (TDD) had a very similar spatial distribution to mean temperature (Fig. 4.5) Table 4.1 Summary of temperature data from Alexandra Fiord from 1990-2002. Mean annual temperature (°C) Mean growing season temperature (°C) Thawing degree-days (°C) Snowmelt date at meteorological station  Mean  1999  2000  -14.8 .5.0 435 166  -16.3 5.8 467 n/a  -14.8 6.0 481 167  The nature of spatial dependence was consistent for all temperature variables studied (Fig. 4.4). The Variogram shows a steady increase in variance with increasing spatial lag and both correlograms show positive spatial autocorrelation tended to occur at lag distances less than 1455 m. Early-season temperature had the highest degree of spatial autocorrelation and lateseason temperature had the lowest.  40  25  -5  T  I i i i ii i i i i i i 11 i i i i i i i ii i ii i i i i i i i i i i i i i i i ii i i i i ii i i i ii i i i _ * . _ » _ v _ A _ » . _ » . _ » . _ » . _ i . _ k _ » . r v j r v j N 3 ( O N 3 N J N 3 l > 0 0  -  ^  -  v  l  ^  ^  a  O  O  C  O  <  0  <  0  <  0  0  0  0  0  -  *  -  »  -  *  N  3  ^ I O O J O > C D N ) C J l O O - ^ J i k - > I O C O O > C O I V J C J l C D - *  Day Number minimum  average  maximum  Figure 4.2 Mean, maximum and minimum near-surface (+10 cm) temperature for the 2000 growing season recorded as sampling sites across the Alexandra Fiord lowland.  41  A. mean growing-season temperatun  D. early season mean temperature  Temperature °C  Temperature °C 7.7 8.2 8.7 9.2 9.7 10.3 10.8 11.3 11.8  6.8 - 7.0 I |7.0 - 7.3 •7.3 - 7.6 17.6 - 7.9 |7.9 - 8.1 |8.1 • 8.5 |8.5 - 8.8 |8.8 - 9.1 19.1 • 9.4 |  B. mean daily minimum temperature  Temperature °C 2.6 - 3.0 3.0 - 3.4 3.4 - 3.8  3.8-4.2 4.2 - 4.6 4.6 - 5.0  5.0 - 5.4 5.4 - 5.8 5.8 - 6.2  C. mean daily maximum temperature  Temperature °C 9.4 - 9.9 ]9.9 -10.5  ]10.5-11.0 111.0-11.6 11.6-12.1 J12.1 -12.7 J12.7-13.2 |13.2-13.8 113.8-14.3  8.2 8.7 9.2 9.7 10.3 10.8 11.3 11.8 12.4  E. mid season mean temperature  Temperature °C 5.9 6.1  • 6.1-6.4 • 6.4-6.6  fi 6.6-6.8 6.8-7.1 • 7.1-7.3 • 7.3-7.5 • 7.5-7.8 • 7.8-8.0 •  F. late season mean temperature  Temperature °C 6.9 7.1 ]7.1 7.4 17.4 7.6 17.6 7.9 17.9. 8.1 I8.L 8.4 18.4. 8.6 18.6 8.9 18.9-9.1  Figure 4.3 Near-surface (+10 cm) temperature distributions for the Alexandra Fiord Lowland in 2000.  42  Figure 4.4 A. Variogram for temperature variables. B. Moran's I correlogram for temperature variables. C. Gearey's C correlogram for temperature variables.  A. TDD on day 180  TDD 1330 - 3 8 9 • 389 - 4 4 8 • 448 - 5 0 7 — 5 0 7 -566 • 1 5 6 6 -624 • 624 - 6 8 3 • [ 6 8 3 -742 • 7 4 2 -801 •1801 - 8 6 0  I  B. TDD on day 190  TDD 190 | | 795 - 858 | | 858 - 922 • 922-958 — 985-1048 ••1048-1112 • 1112-1175 11175-1238 11238-1302 1302-1365  D. TDD on day 210  TDD 11625-1702 11702-1779 • 1779-1857 ••1857-1934 • • 1 9 3 4 - 2011 • 2011 -2088 • 2088-2166 • 2166-2243 • 2243-2320  I I  E. TDD on day 220  I  TDD 11890 - 2064  HH2064 - 2148 r=n2148-2232 -2232-2316 2316 - 2399 2399 - 2483 2483 - 2567 2567 - 2651 2651 - 2735  C. TDD on day 200  TDD | 11130 -1197 | 11197-1264 • 1264-1332 •11332-1399 ••1399-1466 • 1466-1533 • 1533-1601 • 1601 -1688 • 1688-1735  Figure 4 5 The accumulation of near-surface thawing degree-days (TDD) Alexandra Fiord lowland over the course of the growing season.  4.3 Plant Phenology 4.3.1 Cassiope  tetragona  Flower Bud Break:  Flower bud break occurred earlier in the central and southwestern part of the lowland and later below the eastern cliffs and along the shoreline in the north (Fig. 4.6A). The range was from day 170 to 186 (16 days), with a standard deviation of 3.7 days. This variable has a negative relationship, with TDD180 (Rho = -0.58, p < .003), with higher temperatures associated with earlier dates of flower bud break. Mature Flowers:  Mature flowers were first observed on day 171 and last on day 197 (26 days), with a standard deviation of 4.6 days. The time between flower bud break and the development of mature flowers ranged from 0 (both stages occurred within one sampling interval, therefore <4 days) to 9, with an average of 4.6 days and no spatial trend evident. The date of the development of mature flowers was earliest in the southwest and latest along the northeastern cliff base and northwestern shoreline (Fig. 4.6B). The site in the northwest at which bud break was observed at an early date was not equally early for the mature flowers stage. The appearance of mature flowers had a negative relationship with TDD180 (Rho = -0.69, p<0.0002). Immature Fruit:  A similar spatial pattern to the development of other phenophases was evident in the appearance of immature fruit (Fig. 4.6C). The range was from day 180 to 213 (33 days), with a standard deviation of 6.6 days. The time required to reach immature fruit varied from 8 to 24 days, with a mean of 13 days. The appearance of immature fruit had a negative relationship with TDD180 (Rho = -0.63, p<0.001). When the residuals of a regression of the date of immature fruit development against TDD180 were plotted on the map a pattern was observed (Fig. 4.7C). The regression tended to predict later dates in the north and south of the lowland and earlier dates across the centre.  Mature Fruit:  At only 19 of the 24 sites did any tagged plants reach the mature fruit stage. The spatial pattern visible in earlier stages was much less clear as was evident in the variogram and correlograms (Fig. 4.8) as well as the interpolated map (Fig. 4.6D). The correlograms, in fact, show negative spatial autocorrelation at the shortest lag distance. The earliest date remains in the rocky sites in the southwest but the range was from 186-224 (38 days), with a standard deviation of 8.5 45  A. flower bud break  Julian Days  mm 171 • 173 j173-174  |174-176 |176-178 • 178-179 "179-181 181 -183 I 1183-184 J184-186  B. mature flowers  D. mature fruit  Julian Davs 191 -195 195 -198 198-202 202 - 206 206 - 209 209 - 213 213 - 217 217 - 220 220 - 224  '111  E. flowers per branch  Julian Days  • 173-176 176-178 178-180 180 -183 183-185 185 -187 187 -189 189 -192 192-194  Figure 4.6 Phenological dates for Cassiope fruits.  # of flowers 1.5 2.0 2.0 2.7 2.7 3.2 3.2 3.8 3.8 4.6  tetragona  and numbers of flowers and  Figure 4.7 Residuals from regressions between phenological stages and TDD for Dryas integrifolia and Cassiope tetragona that showed a spatial pattern.  485  970  1455  1940  lag distance (m) flower bud break -*- m a t u r e f l o w e r s  -x- i m m a t u r e fruit —  m a t u r e fruit  Figure 4.8 A. Variogram for Cassiope tetragona phenology. B. Moran's I correlogram for Cassiope tetragona phenology. C. Gearey's C correlogram for Cassiope tetragona phenology.  days. The average time between the immature and mature fruit stages was 12 days ranging from 5 to 22 days. There were no significant correlations with abiotic variables. Numbers of Reproductive  Structures:  On average 3.6 of the 5 tagged branches (72%) produced flowers and virtually all of these developed to the point of being immature fruits. Of the branches that did produce flowers the average was 3.2 flowers per branch ranging from 1 to 8. Only 1.6 of the 5 branches (32%) produced mature fruit on average. Of those that did produce mature fruit, the average number was 2 per branch ranging from 1 to 7. Only 20% of immature fruits developed into mature fruits, with 49% aborted and 3 1 % still immature at the end of the growing season. The average number of mature fruit per branch was highest in the central part of the lowland and lowest in the northwest and southeast (Fig. 4.6F). There was no consistent pattern in the distribution of the ratio of immature fruits that developed into mature fruits and no significant correlations with any of the environmental variables. Spatial  autocorrelation  With the exception of the mature fruit stage, the variogram and Geary's C correlogram show a distinct trend of a decrease in spatial dependence as lag distances increase for phenology (Fig. 4.8). The Moran's I correlogram does not present as clear a relationship, with the mature fruit stage not conforming to the expected pattern. Based on the Geary's C correlogram, positive spatial autocorrelation was present until a distance of 1455 m for the first three phenological stages. The variograms and correlograms for variables relating to the numbers of reproductive structures did not show any consistent trend in spatial autocorrelation (Fig. 4.9). The presence of spatial autocorrelation was not detected by the variogram and correlogram largely due to the proximity of some of the highest and lowest values. 4.3.2 Dryas  integrifolia  Flower Bud Break  Flower bud break occurred between day 179 and 192 (13 days), with a standard deviation of 2.83 days. The earliest dates were in the southwest of the lowland and the latest dates were observed at the base of the eastern cliffs (Fig. 4.10A). The date of bud break was negatively correlated with TDD185 (Rho = -0.51, p<0.009). When the residuals from a regression of the date of flower bud break on TDD 185 were mapped a pattern similar to that found for Cassiope was observed (Fig. 4.7A). Again, the regression tended to predict earlier dates across the centre of the lowland and later dates in the north and south.  49  Figure 4.9. A. Variogram for Cassiope tetragona flower a n d fruit numbers. B. Moran's I correlogram for Cassiope tetragona flower and fruit n u m b e r s . C. Geary's C correlogram for Cassiope tetragona flower and fruit numbers. 50  A. flower bud break  D. seed dispersal  Julian Day  ••180-181 |181-182 •1182-183 J183-184 1184-186 • 186 -187 ] 187 -188 H I 188 -189 I 1189-190  B. mature flowers  Julian D a y  —  217 - 218 1218-219 • 1 2 1 9 - 220 • 220-221 1221 -222 222 -223 223 -224 224-225 1225-226  ^9  E. percentage of flowers fertilized  Julian D a y  180-182 182 -183 183-185  Percent  • B 1 8 5 -187  187- 188 188- 190 190 -192 192 -193 193-195  C. capsules forming  • • • •  20-33 33-53 53 - 67 67 - 78 78-100  F. mean height of fertilized flowers  Julian Day  1191-193 1193 -195 1195 -197 1197 -198 ] 198 - 200 ] 200 - 202 202 - 204 204 - 206 206 - 208  height (cm) 0.0-2.0 • 2.0 - 4.0 • 4.0 - 5.1 * 5.1 - 5.9 15.9-7.3  Figure 4.10 Phenological observations and growth measurments for Dryas  integrifolia. 51  Mature Flowers  Mature flowers were first observed on day 180 and last on day 195 (15 days), with a standard deviation of 4 days. On average 3.8 days elapsed between flower bud break and mature flower development ranging from 3 to 7 days with no apparent spatial pattern. This phenological stage followed a similar spatial pattern to flower bud break but with several late values in the northeast corner of the lowland (Fig. 4.10B). This stage was negatively correlated with TDD185 (Rho = -0.68, p<0.0002). Capsules  Forming  The general pattern of the date of capsule formation was similar to previous phenological stages but there was one particularly late date in the south-central part of the lowland and late dates in the north-central part of the lowland (Fig. 4.10C). The first date capsule formation was observed was day 191 and the last date was 211 (20 days later), with a standard deviation of 5.2 days. The time between the emergence of mature flowers and the capsules forming stage ranged from 10 to 18 days, with an average of 13.6 days. Capsule formation was negatively correlated with TDD190 (Rho = -0.54, p<0.005). Dispersal  The first day of seed dispersal was observed between day 216 and 226 (10 days), with a standard deviation of 3.85 days. Between 18 and 34 days elapsed between the capsules forming stage and dispersal with an average of 25.8 days. There remained a general pattern of earlier dates in the southwest and later dates in the northeast, but the pattern deviates considerably from previous stages as evidenced by the low correlation coefficients with other phenological stages (Fig. 4.10D). The date of seed dispersal was negatively correlated to TDD220 (Rho = -0.56, p<0.008). Percentage of Flowers Fertilized and Flower Heights  There was no clear spatial pattern in the distribution of the percentage of flowers fertilized (Fig. 4.10E) although there was a slight tendency for them to be low in the northwest part of the lowland and higher in the central part of the lowland. There was no significant correlation to any other variable. The average height of fertilized flowers also did not produce a clear spatial pattern and was not significantly correlated to any other variable.  Spatial  autocorrelation  The variograms and correlograms for Dryas  showed only a weak trend in spatial  autocorrelation for some variables and no relationship for others (Fig. 4.11). Again, there was a 52  485  -0.3  485  485  970  970  1455  1455  970  1455  1940  1940  1940  Lag Distance (m) -flower bud break  -capsules forming  - f l o w e r height  -mature flowers  -seed dispersal  - % f l o w e r s fertilized  Figure 4.11 A. Variogram B. Moran's I correlogram and C. Geary's C correlogram Dryas integrifolia phenology, flower heights and percentage of flowers fertilized.  discrepancy between the Moran's I correlogram and the Geary's C correlogram but neither had values far from the expected value for a random distribution. 4.3.3 Saxifraga  oppositifolia  Flower Bud Break  Flower bud break occurred earliest in the southeast part of the lowland and at two sites in the north of the lowland: one on the coast adjacent to the west side of the delta and one in the northwest lying between the two latest dates for this stage (Fig. 4.12A). The range of dates was from day 168 to 180 (12 days), with a standard deviation of 2.6 days. There were no significant correlations with environmental variables, which is not surprising given that this species flowered earlier than most environmental data were collected. Though weak, relationships with temperature were consistently negative (earlier flowering in warmer sites) and may reflect the general pattern of temperatures across the lowland. Mature Flowers  Mature flowers were seen very early at site 21 on day 172 and this value dominates the map of the spatial distribution (Fig. 4.12B). The earliest date this stage was recorded was day 172 and the latest was day 183 (12 days), with a standard deviation of 3 days. The average time between bud break and mature flowers was 6.5 days, with a minimum of 4 days and maximum of 9.6 days. No significant correlations with environmental variables were found.  Fruit Visible  The fruit visible stage was first observed in a patch in the northwest part of the lowland and at the rocky site in the southwest (Fig 4.12C). The range was from day 191 to 217 (26 days), with a standard deviation of 4.6 days. On average 16.9 days elapsed between the mature flower and fruit visible stages ranging from 13 to 27 days. There were no significant correlations to environmental variables. Note that site 21, which had the earliest date for mature flower formation, was the latest site to have fruit visible.  Capsules Open  The sites in which Saxifraga produced fruit early expanded eastward at the capsules open stage (Fig. 4.12D). Another site in the south-central part of the lowland also reached this stage early despite not being early in previous stage. The earliest date was day 197 and the latest 226 (a range of 29 days), with a standard deviation of 8.5 days. Between 6 and 29 days passed between the fruit visible and capsules opening, with a mean of 16.3 days. There was a  54  A. flower bud break  Julian Day 170.0 - 170.6 170.6 - 171.1 171.1 - 171.7 171.7- 172.2 172.2 - 172.8 172.8 - 173.3 173.3 - 173.9 173.9 - 174.4 174.4 - 175.0  8. mature flowers  Julian Day 172- 173 173- 174 174- 175 175- 176 176- 177 177- 178 178- 179 178- 180 180- 181  C. fruit visible  Julian Day 191 -182 192 -193 193 - 194 1194- 195 1195- 196 1196 -197 1197 -198 1198 -199  D. capsules open  Julian Day |201 -203 203 -206 1206 - 208 208-210 210-213 213-215 215 - 217 217 - 220 220 - 222  Figure 4.12 Phenological observations for Saxifraga  oppositifolia.  negative correlation with soil moisture (Rho = -0.54, p < .02), meaning that at wetter sites capsules opened earlier, and a weak negative relationship with temperature variables. Spatial  autocorrelation  The variograms tend to show a decrease in spatial dependence with increasing lag distance except for the mature flowers stage (4.13). However, both sets of correlograms remain close to the expected random value. This is not unexpected given the distribution of values, with eariy and late dates often being observed at adjacent sites (Fig 4.12). 4.3.4 Salix  arctica  Flower Bud Break  The average date of flower bud break was day 171 ranging from day 168 to 175 (7 days), with a standard deviation of 1 day. Flower bud break occurred at the same time for male and female flowers. The spatial distribution of the timing of flower bud break was not clear (Fig. 4.14A). The earliest dates were observed at one site in the west-central part of the lowland, and no clear pattern was observed for the rest of the lowland. There was a slight tendency towards earlier dates at the base of the eastern cliffs, but all of this was within a very short period of time. There were no correlations with environmental variables. Mature Flowers  The first date of mature flower stage ranged from day 170 to day 182 (12 days), with a standard deviation of 3.2 days. The average time from bud break to mature flowers was 6 days, ranging from 3 to 12 days. The spatial distribution of observed dates for the appearance of mature flowers is very different from that of the date of flower bud break. The earliest dates were observed in the southeast part of the lowland and later dates in the northeast (Fig. 4.14B). There was a weak negative relationship with TDD180 (Rho = -0.57, p<0.009). Flowers  Senesced  There was a difference of 8 days in the mean date of flower senescence between male and female flowers, but this was based on only four female flowers. The pattern of dates for flowers falling off was dissimilar from each of the previous stages (Fig. 4.14C). Early dates were found in the southeast and northwest corners, with late dates occurring in the centre of the lowland. The mean date was day 201, ranging from day 183 to 220 (37 days), with a standard deviation of 13.1 days. The presence or absence of strong winds blowing flower heads off adds considerable variability to this stage.  56  Figure 4.13 A. Variogram B. Moran's I correlogram, and C. Geary's C correlogram Saxifraga  oppositifolia  phenology.  A. flower bud break  Julian D a y  1168.0-168.7 168.7 169.3 69.3 170.0 170.0 170.7 170.7 171.3 171.3 172.0 172.0 172.7 172.7 173.3 [173.3 174.0  B. mature flowers  Julian Day  |170-171 J171-173 1173 -174 174 -175 175 -177 177-178 178 -179 179-181 • 181-182  C. flower dropped  D. leaf-out  Julian D a v  • • 1 7 0 - 172 | 1 7 2 - 173 | 1 7 3 - 175 175- 177 177 - 179 rni79- 180 I 1180- 182 ! 1182- 184 ! 1184- 185  3  E. leaf senescence  Julian D a y  |208-209 1209-210 |210-212 3212-213 J213-214 ]214-216 J216-217 1)217-218 1218-219  F. fascicle length  Julian D a y  183- 187 187- 191 191 -195 195- 199 199- 204 204- 208 208- 212 212- 216 216- 220  length (cm)  1.7-2.3 • 2.3-3.1 • 3.1-3.8 • 3.8 - 4.7 m 4.7 - 5.9  Figure 4.14 Phenological observations and growth measurements for Salix  arctica.  Leaf Phenology  Mean date of leaf-out was day 175 ranging from day 170 to 186 (16 days), with a standard deviation of 3.2 days. A negative relationship with TDD175 (Rho = -0.66, p<0.0003) was observed but this is highly dependent on leverage from the snowbed site (SITE 03) (Fig. 4.14D). Leaf senescence date averaged day 214 ranging from day 202 to 226 (24 days), with a standard deviation of 5.4 days (Fig. 4.14E). There was no correlation between time of leaf-out and leaf senescence or with environmental variables, and no clear spatial pattern in either variable. The time from leaf-out to leaf senescence ranged from 31 to 46 days, with an average of 39.1 days. Fascicle Length  An analysis of variance revealed a significant difference in the mean fascicle length for flowering and vegetative shoots (data log transformed to meet normality assumptions, p<0.0001). Mean fascicle length for flowering shoots was 4.0 cm, with a standard deviation of 1.3 cm and the mean for vegetative shoots was 3.0 cm, with a standard deviation of 1.0 cm. There was no spatial pattern in the fascicle length of either the flowering or vegetative shoots and no significant relationships were found with environmental variables (Fig. 4.14F). Spatial  autocorrelation  There was no clear spatial pattern in the distribution of sites with particular ratios of vegetative : male : female plants. The graphs of spatial autocorrelation do not show any consistent trend, with both correlograms showing all values close to the expected value for a random distribution (Fig. 4.15).  4.4 Comparison with ITEX Observations Average daily near-surface temperatures during the 2000 growing season across the lowland ranged from 6.8°C to 9.4°C, a span of 2.6°C. From 21 years of climate data at Alexandra Fiord, average growing season temperatures at 1.3 m above the ground were from 2.9°C to 6.0°C, a range of 3.1°C. The range is likely higher nearer to the ground as was found in the control plots at the ITEX Cassiope site where near-surface air temperatures from 1996-2001 ranged from 6.1 Cto9.7 Cor3.6°C. 0  0  Table 4.2 shows that the range of dates for phenological events was consistently much wider in  Dryas  temporally than spatially. This means that for a roughly equivalent range of  temperatures, Dryas responds more to inter-annual temperature variability at a given site than it does to the established temperature gradients across the lowland. Thus, this species is able 59  1.6  ud bre  1.8  1.4 SD  GO  1.2  3  0.8  nd asci  1  — K  0.6 CD CD 0.4 =J  CD OUBI BAI  7T  —t  CD  CO  0.2  =J  "  0  B. 0.3 0.2 </>  C0.1  co o  5= o -0.1 g  =  -0.2  485  970  1455  1940  Lag Distance (m) -flower bud break mature flowers  leaf-out -leaf s e n e s c e n c e  fascicle length  Figure 4.15 A. Variogram B. Moran's I correlogram, and C. Geary's C correlogram for Salix arctica phenology and fascicle lengths.  60  to take advantage of warm seasons by flowering and producing seeds earlier or delaying flowering in cold years. Table 4.2 Comparison between range of dates (day numbers) for phenological events of Dryas integrifolia in the current study and from ITEX monitoring plots at Alexandra Fiord. The ITEX values are based on average of all control plots at all sites for each year (1993-2000). A positive difference means that the range for ITEX plots is larger. Phenology Spatial (2000) Temporal (ITEX) Difference  Mature flowers Min Max 180 195 180 203 8 days  Bud break Min Max 179 192 176 196 7 days  Capsules Min Max 191 211 184 215 11 days  Seed dispersal Min Max 216 226 208 225 9 days  Cassiope tetragona responded quite differently from Dryas though it varied by stage (Table 4.3). The earliest stage had an equivalent range for spatial and temporal variability. Later stages, however, occurred within a very narrow time envelope at a given site over time relative to the variability across the lowland observed in 2000. This means that for later stages, the timing is more constant at a given location, despite being quite variable across environmental and spatial gradients. However, the mature fruit stage did not have a distinct spatial pattern or any clear relationship with any of the measured environmental variables. Table 4.3 Comparison between range of dates (day numbers) for phenological events of Cassiope in the current study and for ITEX monitoring plots at Alexandra Fiord. The ITEX values are based on average of all control plots at all sites for each year (1992-2000). A positive difference means that the range for ITEX plots is larger. tetragona  Phenology Spatial (2000) Temporal (ITEX) Difference  Buds Visible Min Max 170 183 170 185 2 days  Mature flowers Min Max 171 197 185 198 -13 days  Immature fruit Min Max 180 213 200 213 -20 days  Mature fruit Min Max 186 224 205 219 -24 days  4.5 Discussion Flowering phenology in Cassiope tetragona was correlated to temperature in early stages and thus had a similar spatial distribution, but the final mature fruit stage was randomly distributed across the lowland. Cassiope flowered after Saxifraga and Salix but before Dryas. Woodley and Svoboda found that Cassiope  consistently flowered 10-15 days after snowmelt at  Alexandra Fiord but the lack of snowmelt dates in the current study made a comparison impossible. However, the range between the earliest and latest dates increased over the growing season and this might be a result of a near synchronous start of the growing season for all plants followed by increasing divergence as the effects of other variables such as temperature exert influence on flowering phenology.  61  At the ITEX sites at Alexandra Fiord, Johnstone (1995) found that there was no difference in the timing of vegetative bud-break in Cassiope tetragona  between sites or by warming  treatments, but warming treatments accelerated the initiation of growth. The colder of the two years had later phenological dates for reproductive stages, which agrees with the relationships found in this study. OTCs hastened early reproductive stages in both years and at both sites, up to 13 days earlier in 1993, while the fruit maturation stage showed no response to warming. This is similar to the observations made in this study in which early stages were well correlated to temperature patterns but later stages had more random distributions. Johnstone (1995) also observed a strong correlation between date of snowmelt and vegetative phenology, but the length of time Cassiope spent growing was relatively constant such that an earlier start to the growing season resulted in earlier cessation of growth despite environmental conditions.  Cassiope would thus be classified as a periodic species according to Sorensen (1941). This limitation might be controlled by soil or within-plant nutrient reserves. Havestrom et al. (1993) found that experimental warming increased the growth of Cassiope on Svalbard while lower latitude sites responded more to shading and nutrient addition. Some species are known to have periodic growth patterns in which resources are accumulated in one season and growth and reproductive output are increased in subsequent years. Nams and Freedman (1987b) found that Cassiope tetragona produced on average 0.3 flowers per branch in 1981 and 2.4 flowers per branch in 1982. The number of flowers per shoot was found to be highest at early melting sites across the lowland (Johnstone 1995). In 2000, there were on average 2.3 flowers per tagged branch, which if we assume that Nams and Freedman recorded a high and low year, corresponds to a season with high flower output. Both 1999 and 2000 were warmer than average at Alexandra Fiord and this likely has some bearing on flower output. However, there was no relationship found between temperature and flower production in this study. Experimental warming treatments at Alexandra Fiord produced considerably more flowers per shoot than control plots and increased growth of Cassiope  tetragona  at the  Cassiope ITEX site, but not at the Dryas ITEX site (Johnstone 1995). Havstrdm et al. (1995) found that increasing temperature had no effect on the frequency of flowering in Cassiope  tetragona but flowering frequency was increased by the addition of nutrients and decreased by shading. Bell and Bliss' (1980) research on King Christian Island revealed that only in exceptionally warm years do most high arctic plants successfully produce viable seeds. They suggest that frost hardiness in reproductive structures is sacrificed for rapid development.  62  Dryas integrifolia was the latest flowering species monitored in this research. All phenological stages were strongly correlated to temperature. Bean and Henry (1998) found that snowmelt was the principle trigger of bud break in this species. Dryas was previously found to flower 15 to 20 days after snowmelt (Woodley and Svoboda 1994), which is consistent with the results presented here. Dryas prefloration time was advanced in ITEX warming chambers (Welker et  al. 1997, Woodley and Svoboda 1994) and a delay in flower senescence in response to nutrient addition has also been found (Woodley and Svoboda 1994, Wookey et al. 1995). The strong correlation between temperature and phenology was also shown at several sites in the Canadian Arctic by Bean and Henry (2002). At three of the four ITEX sites described by Welker  et al. (1997) warming increased the heights of Dryas flowers by 10-15 mm on average and increased seed mass as well. In contrast, Bean and Henry (2002) found that warm years tended to result in lower flower heights. In this study, flower heights were not correlated to any of the measured environmental variables. The range of dates for Cassiope phenophases increased as the season progressed, with flower bud break occurring everywhere in the lowland within 16 days, and 38 days elapsing between the first and last site to produce mature fruit. Dryas had a similar trend with flower bud break occurring over 13 days and capsules forming over 20 days, but the dispersal stage showed a contraction with only 10 days between the first and last site. This synchronicity could be a result of some cue other than temperature or overall heat accumulation. At Truelove Lowland Cassiope was found to complete its life cycle in 50-60 days regardless of the length of the growing season (Svoboda 1977), and this observation has been made for other plant species. Sorensen (1941) distinguished between periodic and aperiodic species, the former having fixed time intervals for active growth and the latter being more sensitive to varying environmental conditions. In Alaska, Murray and Miller (1982) found that Cassiope  tetragona  needed at least 105 days from the date of snowmelt to produce and disperse seeds but this is shorter than the entire growing season at Alexandra Fiord. Growth in Cassiope was observed to be more correlated to global radiation than to temperature based on ITEX results (Molau 1997). The comparison between the spatial variability and temporal variability provided an interesting contrast in the response of Cassiope and Dryas. Both are evergreen dwarf shrubs, yet Dryas was more sensitive to inter-annual temperature variability at the ITEX sites than to the climatic gradient across the lowland, whereas Cassiope was less responsive to inter-annual variability than it was to the established temperature gradient across the lowland. This implies that  Cassiope adapts to a site but not to inter-annual variability in terms of phenology. This is  63  consistent with other research that has found that Cassiope phenology is not affected by experimental warming or fertilizer addition (Woodley and Svoboda 1994, Molau 1997). Johnstone and Henry (1997) have successfully used retrospective analysis of leaf and flower scars along branches of Cassiope tetragona to correlate patterns of growth and flower production to climate conditions. It would appear that Cassiope responds to variability in climate in growth and reproductive output but not phenology. They found that flower production was more sensitive to climate than shoot elongation. However, Hollister and Webber (2000) attributed the large interannual differences in plant phenology to snowmelt date and the smaller differences between warmed and control plots to temperature. Unfortunately, snowmelt could not be taken into account in the spatial data for the comparison between Cassiope and Dryas. A study of other growth forms would be a valuable addition to this question of spatial versus temporal response to climatic variability.  Saxifraga oppositifolia overwinters with highly developed flower buds (Sorensen 1941) and bud break occurs 3 to 5 days after snowmelt (Svoboda 1977), as was found here. The spatial distributions of phenological variables did not demonstrate any consistent pattern. There were frequently adjacent sites with extremely early and late dates. The only significant correlation with an environmental variable was a tendency toward earlier dates for the opening of capsules at sites with higher soil moisture. Stenstrom et al. (1997) reviewed the responses of Saxifraga  oppositifolia after 2 to 3 years of experimental manipulations at three ITEX sites. At Alexandra Fiord, they found that warming produced an earlier first pollination, a slightly longer flowering period and lower flower densities in the warming chambers.  Salix arctica was the earliest species to flower. It too overwinters with flower priomordia in an advanced state of development (Sarensen 1941). Woodley and Svoboda (1994) observed  Salix flowering to occur between 7 and 10 days after snowmelt, which corresponds to the results presented here. Like Saxifraga, the sequence of flowering phenology did not maintain a consistent spatial distribution. The timing of leaf emergence was well correlated to temperature but neither leaf senescence nor fascicle length was correlated to any other variables. Jones et  al. (1997) evaluated the responses of three Salix species to ITEX warming. They found that warming advanced phenology in Salix rotundifolia and Salix herbacea but not in Salix arctica.  They did find that the length and weight of new growth was increased in the warming chambers and at Alexandra Fiord warming decreased the fruit:flower ratio. Jones (1995) found shorter  Salix arctica fascicle lengths at the wet site than at the dry site and warming treatments produced longer fascicles at the dry site but not at the wet. Warming also increased the number  64  of capsules per catkin and the number of ovules per capsule. In the current study, no correlations were found between fascicle length and the measured environmental variables. Overall, phenological development tended to correlate well to the pattern of surface temperatures across the lowland for Cassiope and Dryas but not for Salix and  Saxifraga.  Where relationships were found the warmer temperatures in southwest corresponded to earlier phenological dates. This is consistent with a study examining the temporal trends in Dryas  integrifolia and Papaver radicatum phenology over 10 years at Alexandra Fiord (Bean and Henry 2002), though the relationships were largely dependent on late phenology in one extremely cold year, which is analogous to the late melting snowbed site in this study. Phenology did not correlate to the spatial pattern of soil moisture or soil nutrients. Soil moisture is determined primarily by topography and varies on a scale that is too fine to be captured by the resolution of the sampling grid in this study. Soil nutrients are well correlated to moisture and therefore also not captured adequately using only 28 point samples across the 800 ha lowland. Henry et al. (1986b) found that fertilizer addition advanced the phenology in forb species by 5 to 7 days but there was no response in graminoids or shrubs, and Miller (1982) found a delay in plant senescence in nutrient rich sites. Fertilizer also had little effect on phenology in a study in Alaska by Larigauderie and Kummerow (1991). Soil moisture and nutrients are more important in determining plant community development (Robinson et al. 1998), as described in Chapter 5. The relationship between temperature and phenology has also been demonstrated for other species using experimental warming. ITEX warming experiments caused earlier phenology and faster development in warming chambers for Silene acaulis (Alatolo and Totland 1997), Carex  spp. (Stenstrom and Jonsdottir 1997), Papaver radicatum and Eriophorum vaginatum (Molau and Shaver 1997), and Vaccinium uliginosum (Suzuki and Kudo 2000). Polygonum  bistorta  released from dormancy earlier in ITEX warming treatments, but this response was only significant in 1 of 2 years with no noticeable effect of flower phenology or production (Starr et  al. 2000). Several interesting results arose from a meta-analysis of the short-term (4 years) response to ITEX experimental warming (Arft et al. 1999). Vegetative growth was found to increase initially and then return roughly to its original magnitude, though variability increased throughout, while reproductive effort was not significantly affected by warming. Phenological dates were consistently earlier in warmed plots with many statistically significant responses, particularly at high arctic sites. Senescence tended to be later in warming chambers but few of these results were significant. The lack of late-season phenological response may indicate that these stages are more responsive to photoperiod than temperature (Arft etal. 1999). 65  Arctic plants are known to initiate bud development late in the year prior to flowering (Sorensen 1941, Bell and Bliss 1980). Molau (1997) describes flowering in arctic plants to be highly variable and dependent on climate 1 to 2 years prior. Thus, the previous growing season's temperature is also important in determining the timing and success of flower production. Bean and Henry (2002) found a significant positive relationship between Dryas and  Papaver  phenology and accumulated warmth in the previous growing season at Alexandra Fiord, though the same relationship for these species was reversed at Tanquary Fiord on northern Ellesmere Island. No relationship was evident between flowering phenology and the previous growing season temperature in a study on Iceland (Thorhallsdottir 1998). However, given that the current study is focused on the spatial distribution of temperatures, we can assume that the general pattern of temperatures is relatively constant and, therefore, areas that were warmer in 2000 were likely warmer in 1999 and previous years. The growing season of 1999 was only slightly warmer than normal at Alexandra Fiord. There are also sources of variability that were not or could not be sampled in this study. Most authors agree that the breaking of dormancy in arctic plants is primarily determined by the snowmelt date (Chapin and Shaver 1985a, Shaver and Kummerow 1992, S0rensen 1941, Woodley and Svoboda 1994). A multiple regression using the date of snowmelt, current season's TDD and the previous season's TDD to predict bud break in Dryas integrifolia and  Papaver radicatum at Alexandra Fiord had an R of 0.92 with p<.004 (Bean and Henry 2002). 2  Accurate snowmelt dates for each of the sample locations would have been quite useful but the sampling grid was set up after the majority of the snow had melted. Most points had patches of early melting and late melting snow around them when the sampling locations were established. The exact date of snowmelt at each sample location might explain some of the unaccounted for variance in phenological dates, but the pattern of late-lying snowbeds was very patchy and would require either intensive sampling or highly accurate remote sensing to determine. There was no landscape-scale trend in snowmelt date (See Ch. 5), so the growing season can be considered to be roughly the same length for the lowland as a whole. Given the landscape-scale pattern that was observed in many of the phenological variables and the relative patchiness of the snowmelt pattern, it is unlikely that it is a major determinant of this pattern at the landscape scale. For those variables that did not demonstrate a clear spatial pattern, the date of snowmelt might be a more important factor. Examinations of the differences in phenological dates between areas within snowbeds vs. areas outside of snowbeds have been done in other areas (Kudo 1991, 1993, Murray and 66  Miller 1982). Woodley and Svoboda (1994) found that plants broke dormancy earlier in early melting areas in the Alexandra Fiord lowland and in experimental snow removal plots. They also observed, however, that early melting sites tended to have lower snowcover and were more exposed, and thus dried out earlier in the growing season. This affected later phenological stages with drought tending to cause eariy dormancy. Within snowbeds, plants rarely produced fruit at Truelove Lowland (Svoboda 1977), and in Alaska shrubs were found to produce more leaves per branch (Murray and Miller 1982). Bud break in Betula nana and Salix  pulchra did not necessarily follow snowmelt date in 3 years of monitoring in Alaska, nor was there a consistent accumulation of TDD before bud break (Pop et al. 2000). In another Alaskan study, deciduous shrubs were observed to vary up to 10 days in the onset of leaf expansion due to differences in snowmelt, but maturation and senescence were synchronous (Murray and Miller 1982). Site 03, the one snowbed site in this study, had much different timing of phenological stages than the rest of the sampling sites. This site was a strong contributor to the overall spatial pattern observed in the lowland, and was often crucial in achieving statistical significance in the correlation analyses, having considerably lower TDD. However, the overall area covered by snowbeds is low (see Chapter 5). Snowmelt dates at Alexandra Fiord can differ by up to 25 days in warm vs. cold years and this has a considerable effect on temporal variability in phenology (Bean and Henry 2002). When linear regressions were performed on some of the correlated variables, the residuals of the regressions had a somewhat distinct spatial distribution. This implies that some spatially distributed variable other than temperature is at least in part responsible for establishing the date of these stages. However, given that the pattern is only evident for a few stages in two species it is clearly not a very important variable. It does, methodologically speaking, provide some insight into the source of the error in the correlation. Some phenological stages are also prone to being affected by somewhat unrelated phenomena. An example would be the 'flower drop' stage in Salix, which could be brought about by high wind, rain, or a passing animal. Phenology is also subject to random genetic variability in the timing of plant development (Molau 1997). Thawing degree-days (TDD) or growing degree-days (>5°C, 10°C) have been used in many ecological studies to characterize the warmth of a given time period. In the current study many phenological variables were significantly correlated to TDD up to the date of the event. However, the pattern of surface temperatures was relatively constant throughout the season. 67  This temporal autocorrelation results in the inability to determine if the development of a particular stage in a plant species is a function of the temperature in the early part of the growing season, the time just before the stage is observed, or the entire growing season up to the time when the stage is observed. There is the additional problem of covariation among variables such as TDD and snowmelt date. In the case of SITE 03, the late date of snowmelt resulted in much lower TDD than other sites and teasing-out the relative contribution of each of these factors to the late phenological dates at that site would require a focused experimental approach. It should also be noted that TDD recorded near the ground surface, as was done in this study, are much higher than at screen height (~1.5 m) and more accurately reflect the environmental conditions being experienced by the low stature arctic plants. Several spatial analytical issues have arisen during the course of this research. The choice of a systematic sampling scheme offered some advantages and some disadvantages. With the number of sites that could realistically be monitored, it appears that it was still likely the best way to capture landscape scale patterns of many of the variables of interest. Others, like soil moisture, would have required a much denser sampling grid to hope to characterize the distribution. In variogram and correlogram analysis, however, the shortest distance class was limited to the shortest distance between points, which was 485 m. A multi-scale sampling approach would have provided more information on the nature of spatial autocorrelation over short distances. Fortin (1999) has shown that the quadrat size and the number of distance classes can affect estimates of spatial autocorrelation in ecological data. Variables that did not show a consistent pattern of spatial autocorrelation may vary over smaller scales and would have to be sampled more intensively. In some cases, adjacent values that were extremely different strongly affected measures of spatial autocorrelation despite the presence of a subjectively observable spatial trend in the dataset. Again, a denser sampling grid or multiscale sampling approach would likely have smoothed out some of these anomalies.  Many of the results were presented as spatial interpolations of point data in order to create a surface covering the entire lowland. In some cases, however, these were based on data that showed weak or no spatial autocorrelation in variogram and correlogram analysis at the same scale upon which the interpolation is based. In fact, interpolations should properly be used only when spatial autocorrelation is demonstrated. The characteristics of the variogram are the basis for interpolation by krigging. However, the interpolation used here was sensitive enough to local values not to obscure them by averaging with adjacent points, in addition, interpolation supplies values for estimated phenological dates where the species does not even exist. Therefore, interpolations must be used with additional information, such as plant community 68  mapping data, in order to provide a truly informative description of the landscape structure and function. Predictions of earlier dates of phenological events under a warmer ambient climate are supported by this research. This was by no means universal among species or stages but in all cases where a significant relationship was found it was consistently negative. Shaver et al. (2000) suggest that the indirect effects of a future temperature increase, such as an increase in nutrient availability, could be more important than direct effects. A correlation was not found between temperature and nutrient content of the soils, but this was largely due to confounding factors including soil moisture, degree of soil development, and vegetation cover.  69  Chapter Five Vegetation Community Description and Mapping 5.1 Introduction This chapter describes the results from the plant community description and mapping portion of the research. Firstly, the classification of plant communities using cluster analysis is presented followed by ordination of the stands using principal components analysis (PCA). The communities are mapped and compared to the existing map of vegetation at Alexandra Fiord by Muc et al. (1989). Communities are compared to environmental variables using redundancy analysis (RDA) and the spatial component of variability is extracted using partial RDA according to the technique described by Borcard etal. (1992). Finally, an unsuccessful attempt to map the vegetation of the lowland by interpreting digital aerial photographs is briefly described and the relationship between the snowmelt pattern in the lowland and the distribution of vegetation communities is discussed.  5.2 Environmental Variables The characteristics of the environmental variables were described thoroughly in the previous chapter (Section 4.2). Briefly, the general trend for temperature was that colder temperatures were found along the shoreline in the north of the lowland and warmer temperatures were generally found in the south which is protected and warmed by radiation reflecting off the adjacent cliffs (see Fig. 4.3). Mid-season soil moisture tended to be higher in the west-central and northeast parts of the lowland and lower in the north-central and southwest (see fig. 4.1). The presence of high amounts of soil C and N were positively correlated to mid-season soil moisture and thus had a similar spatial pattern.  5.3 Plant Cover A total of 34 vascular plant species were recorded in the sampling. The average number of vascular plant species recorded at the 28 sites was 8.4, with the highest being 12 and the lowest 3 species. The average percentage vascular plant cover was 50%, ranging between 16% in the southwest comer of the lowland to 80% in the northwest corner (Fig. 5.1A). There was a general trend of decreasing vascular plant cover from north to south. Average cover of graminoids was 13%, ranging from 1% to 68%, with a spatial distribution similar to that of overall vascular cover (Fig. 5.1B). Litter accumulation also follows the same spatial trend (Fig. 5.1C). Average cover of lichens was 9%, ranging from 0% to 34% with a nearly opposite distribution to that of vascular plants (Fig. 5.1 D). Bryophytes had an average cover of 20% with low values along the western margin of the lowland and central shoreline (Fig. 5.1E). Vascular 70  A. cover of live vascular plants  % cover 16-17 • 17-38 • 38 - 57 • 57 - 68 • 68 - 80  B. cover of graminoid species  % cover 0- 2 • 2- 7 • 7-12 •12-25 • 25 - 68  D. lichen cover  % cover 0- 2 2 5 5 11 >11 •18 •18 34  E. bryophyte cover  % cover 0- 8 • 8-14 • 14 - 28 • 28-48 •48-62  C. litter cover  % cover • • • •  4-13 13-24 24 - 30 30-47 47 - 82  Figure 5.1 Distributions of vegetation cover variables.  plant diversity paralleled vascular plant cover, generally decreasing from north to south (Fig. 5.1F). 5.4 Classification 5.4.1  and  Ordination  Cluster Analysis  Complete linkage cluster analysis on all live vascular plants sampled at each site produced a division into 5 plant communities and two sub-types (Fig. 5.2). Clusters 3 and 4 were joined at a lower level than some other linkages because this division represents a difference in overall cover rather than species composition. The separation of site 03 into the snowbed sub-type is made clear by the PCA. When the location of the samples classified into each of the groups was mapped, the spatial aggregation of samples within each group indicated that the classification had identified an overall pattern in the vegetation of the lowland (Fig. 5.3). 5.4.2  Principal Components  Analysis  The ordination of the stands using principal components analysis (PCA) demonstrated both the distinctiveness of some of the vascular plant communities described below and the similarities among some communities (Fig. 5.4). The first two PCA axes explained 76% of the total variance in the species data (Table 5.1) and are dominated by the most abundant species in the lowland (Cassiope  oppositifolia,  tetragona,  Dryas integrifolia,  Eriophorum  angustifolium,  Saxifraga  Salix arctica and Vaccinium uliginosum). Though the division of communities was  clear, 95% confidence ellipses showed that all clusters overiap with at least one other cluster. Table 5.1 Results from the PCA of vascular plant cover. Axes Eigenvalues Cumulative % variance explained  5.4.3  Redundancy  1 .535 53.5  2 .224 75.9  3 .153 91.3  4 .029 94.2  Analysis  Redundancy analysis (RDA) was performed using mean, minimum and early-season temperature, mid-season soil moisture and soil C content (Fig. 5.5A), which were identified by forward selection as being the most important variables. The first two axes accounted for 36% of the variance in the vascular plant species data (Table 5.2). The first axis was most strongly influenced by the temperature variables and the second axis by soil moisture and C content.  72  Cluster 1 Wet Meadows  Cluster 2 Dryas dominated * Salix dominated  O  Cluster 3 Rocky lichen / Cassiope snowbed Cluster 4 Cassiope  I Dryas heath  A  Cluster 5 Cassiope / Vaccinium  heath  4-  Figure 5.2 Complete linkage cluster analysis on %cover of live vascular plants (data unstandardized). Symbols for each major cluster (vegetation community) are used in subsequent figures. 73  0  1  2km  Vascular plant communities • Cluster 1 - Wet Meadows * — Cluster 2 - Dryas dominated <^— Salix dominated • Cluster 3 - Rocky lichen / Cassiope • — Snowbed A — Cluster 4 - Cassiope I Dryas heath Cluster 5 - Cassiope I Vaccinium heath  Figure 5.3 Distributions of vascular plant communities on the Alexandra Fiord lowland. 74  PCA Axis 1  Vascular plant communities © Cluster 1 - W e t M e a d o w s * — Cluster 2 - Dryas dominated #• — Salix d o m i n a t e d • — C l u s t e r 3 - Rocky lichen / Cassiope • — Snowbed A — Cluster 4 - Cassiope I Dryas heath -={=Cluster 5 - Cassiope I Vaccinium heath  Figure 5.4 P C A ordination on live vascular plants, scaled to maximize site differences, with 9 5 % confidence intervals.Data w e r e not transformed, standardized or centred. 75  Figure 5.5 R e d u n d a n c y Analysis (RDA) on vascular plant cover without (A) and with (B) geographic coordinates. See Fig. 5.5 for key to symbols. 76  Table 5.2 Eigenvalues from the RDA of vascular plants with five selected environmental variables. Axes Eigenvalues Cumulative % variance explained  2 .128 36.4  1 .236 23.6  3 .015 37.9  4 .010 38.9  Table 5.3 Correlation matrix from the RDA of vascular plants with the five selected environmental variables. SPP are species axes, ENV are environment axes and T stands for temperature.  SPP AX1 SPP AX2 ENV AX1 ENVAX2 early T mean T min T moisture carbon  ~SPPAX1 1.000 -0.123 0.760 0.000 -0.502 -0.281 -0.350 -0.039 0.090  SPPAX2  ENVAX1  ENVAX2  early T  mean T  min T  1.000 0.000 0.664 -0.110 -0.057 0.148 0.624 0.325  1.000 0.000 -0.661 -0.370 -0.461 -0.052 0.118  1.000 -0.166 -0.086 0.223 0.940 0.490  1.000 0.852 0.470 0.005 0.044  1.000 0.735 0.071 0.245  1.000 0.287 0.377  moisture  -  The RDA on vascular plant species cover with the addition of geographic coordinates to the environmental variables revealed a strong influence of spatial location on the distribution of vegetation communities (Fig. 5.5B). An additional 6% of the species variance was explained by the first two ordination axes (Table 5.3),  with the latitude axis being the longest of the  environmental axes. Spatial locations and temperatures dominated the first axis and the second axis remained primarily a reflection of soil moisture and carbon concentration.  Table 5.4 Results from the RDA on vascular plants with five selected environmental variables and geographic coordinates. Axes Eigenvalues Cumulative % variance explained  2 .135 42.4  1 .289 28.9  3  .034 45.8  4 .011 46.9  Table 5.5 Correlation matrix from the RDA of vascular plants with five selected environmental variables and geographic coordinates. SPP are species axes, ENV are environment axes and T stands for temperature.  SPP AX1 SPP A X 2 ENV AX1 ENVAX2 early T mean T min T moisture long lat Carbon  SPP AX1 1.000 0.071 0.829 0.000 0.478 0.267 0.356 0.105 0.355 -0.731 -0.052  SPP AX2  ENVAX1  ENV AX2  early T  mean T  min T  moisture  1.000 0.000 0.687 -0.187 -0.096 0.100 0.618 0.095 0.124 0.343  1.000 0.000 0.577 0.322 0.429 0.126 0.428 -0.881 -0.063  1.000 -0.272 -0.140 0.146 0.899 0.139 0.181 0.499  1.000 0.852 0.470 0.005 -0.022 -0.687 0.044  1.000 0.735 0.071 -0.060 -0.492 0.245  1.000 0.287 0.097 -0.370 0.377  1.000 0.115 0.094 0.726  long  lat  1.000 -0.234 -0.064  1.000 0.099  77  The RDA results with and without including the geographic coordinates emphasise the influence of spatial location on the presence or absence of plant species. Using Borcard et a/.'s (1992) technique, the spatial component of species variance was partialed-out (Fig 5.6). Despite the latitude axis being the longest in the RDA ordination, environmental variables still accounted for more of the species variance than the spatial component alone. 5.5 Plant Communities 1. Wet Meadows The wet meadow sites are located on the east and west margins of the sampling area close to the shoreline (Fig. 5.3). They are dominated by the sedges Eriophorum angustifolium, stans, C. misandra,  C. membranacea  and also contain Vaccinium  uliginosum,  Carex  and Dryas  integrifolia. The percentage cover and diversity of vascular plants and cover of brybphytes was considerably higher than among the other groups and the cover of lichens much lower. Both the cluster analysis and the ordination separated the meadow sites on the east side of the lowland from those on the west. Western meadows had lower moss cover, higher overall percentage of graminoids, and more prominent Carex membranacea, whereas in the eastern meadows Carex stans was more prominent. Vaccinium was only present in the eastern meadow sites. The meadow sites on the west side actually fell within the 95% confidence ellipse for the Dn/as-dominated community (Fig. 5.4). Interestingly, in the PCA ordination, the wet meadows and rocky sites are close together on both axes despite having seemingly quite different vegetation composition and environment (Fig. 5.4). In the RDA ordination, the wet meadows are clearly associated with high soil moisture and nutrient content and low temperature. The constrained ordination clearly distinguished the meadow sites from the rocky lichen/Cass/ope sites, particularly with the addition of geographic coordinates (Fig. 5.5).  2. Dn/as-dominated sites The six sites that comprised the Dn/as-dominated community defined by the vascular plant classification (Fig. 5.2) were found in the north central portion of the study area within 1km of the shoreline (Fig. 5.3). The dominant vascular plant species were Dryas integrifolia,  Saxifraga  oppositifolia, Cassiope tetragona and several sedge species. Vascular plant cover and diversity were highly variable and moss cover was moderate while the cover of lichens tended to be low. Most of the Dn/as-dominated sites correlated negatively with soil moisture, carbon concentration and temperature in the constrained ordination, but site 12 was an exception due to the extremely high soil moisture found at this site, which was a consequence of flooding in mid-season by an adjacent stream channel.  78  T3 <D  100  C  90  'fs  80  Q. X  d>  <D  O  unknown  70  spatial  60  c  .55 50 40  >  "o Q)  Q. to  ^° o N  •  30  spatial / environment environment  20 10 0  vascular plants  Figure 5.6 Partitioning of variance explained by RDA into exclusively environmental, spatially structured environmental, exclusively spatial and unknown components.  79  Salix-dominated  site  Site 11 had some similarities in species composition to the Dn/as-dominated sites but was dominated by Salix arctica, had a higher vascular plant cover and diversity, higher moss cover, less lichen and greater soil moisture than the other sites in this community. This site was located adjacent to a creek bank. 3. Rocky lichen / Cassiope zone All but one of the six sites included in the rocky lichen/Cass/ope zone of the vascular plant classification were in the southern end of the study area adjacent to the glacier and cliffs (Fig. 5.3). The exception was site 03 which was located in a snowbed north of one of the beach ridges near the shoreline and is described below. Site 23 was also included in this zone but this site was severely disturbed by a flood before point-framing was carried out, so it might not accurately reflect the vegetation present in the vicinity of the site. These sites had a much lower percentage vascular plant cover and lower diversity than the other groups and were in fact primarily exposed soil and rock. The dominant vascular plant species were Cassiope  Dryas integrifolia,  tetragona,  and Salix arctica, with the lowest moss cover and highest lichen cover  among the plant communities. Rocky sites had higher temperatures, and moderate to low soil moisture and nutrient concentrations.  Snowbed site  In the classification of plant communities site 03 falls within the rocky lichen / Cassiope cluster based on a relatively similar species composition and similarly low overall vascular plant cover. However, in all ordination diagrams and in the map it was clearly distinct in terms of its composition and location. This site was representative of a separate plant community present beneath the large late-lying snowbeds. It had greater vascular plant cover and diversity than the other rocky sites despite the similar composition of dominant species and higher bryophyte cover. The snow at site 03 melted at least one week later than all of the other sites.  4. Cassiope / Dryas heath  The Cassiope/Dryas  zone is located across the centre of the study area (Fig. 5.3). This group  was dominated by Cassiope tetragona,  Dryas integrifolia,  Vaccinium uliginosum,  and Salix  arctica with a moderate percentage of vascular plant cover, diversity, moss cover and lichen cover. Despite having similar composition to the Cassiope  /  Vaccinium  heath, the  Cassiope/Dryas  heath was quite distinct in the PCA ordination (Fig. 5.4). In RDA the  Cassiope/Dryas  community was confined to the mid-range latitudes but widely scattered with  respect to the measured environmental variables (Fig. 5.5).  80  5. Cassiope / Vaccinium heath  The Cassiope/Vaccinium  zone lies in the southeast region of the study area and includes two  sites at the base of the eastern talus slopes (Fig. 5.3). Cassiope  tetragona,  Vaccinium  uliginosum, Dryas integrifolia, and Salix arctica dominate these sites and they tended to have a lower than average vascular plant diversity and lichen cover but a higher cover of mosses. The  Cassiope/Vaccinium  community was found to correlate positively to temperature and tended to  be present where soil moisture and nutrients were low. This community was quite distinct in the PCA ordination, scoring very high on the first axis and low on the second (Fig. 5.4). In the RDA the Cassiope/Vaccinium  community was found at lower latitudes where temperatures are  higher (Fig. 5.5).  5.6 Relationship Between Vegetation Communities and Snowmelt Date The first observation made regarding the pattern of snowmelt across the lowland is the lack of correspondence to the map of Woodley and Svoboda (1994). Their representation of snowmelt showed a trend to earlier dates of snowmelt in the west and a few large late-lying snowbeds concentrated along a north-south axis in the centre of the lowland (Fig 3.1). Contrarily, the airphoto and oblique mosaics of snowmelt pattern taken in 2000 (Figs. 5.7, 5.8 and 5.9) show the landscape-scale pattern of snowmelt to be much more even across the area. The snowbanks are shown to be smaller, more numerous, and more evenly distributed. As is visible in the two sets of air photographs taken only four days apart, the majority of the land area melted-out at roughly the same time. The range of dates for snowmelt were, however, similar to those observed by Woodley and Svoboda (1994). The association that Woodley and Svoboda (1994) infer between snowmelt date and landscape-scale plant community distribution is also not supported. This is true for both the description of community composition described here and the plant communities described by Muc et al. (1989) (Fig. 2.2). As evidenced by site 03 being described as a snowbed community, snowmelt pattern does influence plant community development, but only in the few late-lying snowbeds that can be seen in the oblique mosaic from June 21 (173) and would constitute only a small proportion of the land area.  5.7 Correspondence to Existing Vegetation Map In addition to comparing the descriptions of these plant communities with Muc's (1989) (Fig. 2.2), comparing the spatial distributions is also informative. A comparison matrix was created by determining the correspondence between the community to which each sampling point had 81  B. June 11 (day 163)  C. June 14 (day 166)  D . J u n e 21 (day 173)  Figure 5.7 Oblique photos of the lowland during the snowmelt period looking west.  Figure 5.9 Airphoto mosaic of the Alexandra Fiord Lowland on June 16 (168) 2000. 84  b e e n a s s i g n e d in the current classification a n d the c o m m u n i t y in M u c ' s m a p in w h i c h e a c h point w a s located (Table 5.6). T h e w e t m e a d o w is the only c o m m u n i t y in w h i c h t h e majority of points are located in c o r r e s p o n d i n g communities according to the m a p overlay. T h e rest a p p e a r t o be r a n d o m l y distributed. S o m e  possible reasons f o r lack of c o r r e s p o n d e n c e  include:  differences in t h e scale of sampling a n d spatial location errors (from original m a p , digitizing, rectifying, a n d associated G P S error).  A n o t h e r perspective o n spatial c o r r e s p o n d e n c e is to c o n s i d e r t h e b o u n d a r y b e t w e e n M u c ' s plant c o m m u n i t y p o l y g o n s as transition zones b e c a u s e plant c o m m u n i t i e s d o not c h a n g e o n a discreet line. W i t h a 5 m buffer on either side of the b o u n d a r y (Fig. 5.1 OA), 6 of the s a m p l i n g points in this study a n d 2 1 % of the entire lowland are within a transition z o n e . If the buffer is i n c r e a s e d to 10 m o n either side of the boundaries (Fig. 5.1 OB), 11 of t h e 2 8 s a m p l i n g sites fall within a transition z o n e , w h i c h e n c o m p a s s e s 3 9 % of the surface a r e a . T h e s e buffers c o u l d also b e c o n s i d e r e d m a r g i n s of error for spatial locations g i v e n all of t h e possible s o u r c e s of uncertainty. Table 5.6 Comparison matrix between Muc et a/.'s (1989) vascular plant communities and those identified in this study. The number of sample points from the current study located in each community type are shown. Shaded cells represent corresponding plant communities. In the rocky lichen Cassiope column the number in brackets is the snowbed site. Muc et al.'s C o m m u n i t i e s  Sedge-Cushion PlantDwarf Shrub Deciduous Dwarf ShrubGraminoid Lichen-Cushion PlantDwarf Shrub Dwarf S h r u b Cushion Plant Salt Marsh Herb (Rivers) Total  Vascular Plant Communities (this s t u d y ) Cassiope/ Dn/asRocky lichen Wet SalixDryas Cassiope dominated Meadow dominated  Cassiope/ Vaccinium  3  0  2  0  1  2  0  0  0  0  0  0  1  n  <*  1  1  1  1 (1)  0  0  0  0  0  0  0  0  1  1  0  1  5  1  6  5  5  6  *x o  u  l i i i l i l l l l t ijiiiiiiiiii  2  5.8 Interpretation of Digital Aerial Photographs It w a s h o p e d that a visual interpretation of the airphoto m o s a i c p r o d u c e d f r o m t h e m i d - s e a s o n aerial p h o t o g r a p h s (Fig. 5.11) w o u l d produce a vegetation m a p that c o u l d be c o m p a r e d directly t o M u c et ars  (1989) m a p (Fig. 2.2). However, t h e five plant c o m m u n i t i e s d e s c r i b e d a b o v e  c o u l d n o t be distinguished reliably w h e n the portions of t h e p h o t o g r a p h s s u r r o u n d i n g e a c h s a m p l i n g point w e r e g r o u p e d by plant c o m m u n i t y type (Fig. 5.12). T h e u s e of o t h e r spectral  85  Figure 5.10 Boundaries from Muc et al.'s (1989) map of vegetation c o m m u n i t i e s with 5 m buffer on either side of boundary (A) and 10 m buffer (B). 86  Rocky Dryas Dominated  Cassiope/ vaccinium  Cassiope/ Dryas  lichen/ Cassiope  W  e  t  Meadows  Figure 5.12 Representative patches of airphotos from attempted photo-interpretation of plant communities. White squares are locations of sampling points.  88  wavelengths that can be recorded with other remote sensing devices would have to be used to separate the different plant communities. For example, obtaining infrared data for the calculation of a vegetation index such as the normalized difference vegetation index (NDVI).  5.9 Discussion The classification and ordination of vascular plants provided a meaningful representation of the plant communities of the Alexandra Fiord lowland. As in other studies of arctic plant community composition, the presence of near-ubiquitous species blurs the line between plant communities and can obscure the importance of rarer species in distinguishing between recognizable assemblages of plant species. In the Arctic, the presence of these rarer species are often more reflective of the environment present at each site, given the wide ecological tolerance of dominants. The techniques used in plant community analysis are geared towards identifying discreet groups where what actually exists is a series of gradients along which there are similarities in some key characteristics but there are rarely if ever distinct boundaries to plant communities. However, the description of common assemblages of plant species elucidates many important relationships and processes within the landscape. Of the five communities identified here, the wet meadows (also known as tundra mires) were very distinct as has been noted by other authors (e.g. Bergeron and Svoboda 1989, Bliss and Gold 1993, Henry 1987, 1998, Henry etal. 1990, Muc 1977). Sedge-dominated wet meadows cover less than 1 % of the Queen Elizabeth Islands in the Canadian High Arctic (Babb and Bliss 1974). Webber (1978) classified half of the tundra in the vicinity of Barrow, Alaska as sedgedominated tundra. Within the meadow community there was an array of different assemblages (Webber 1978). In Schaefer and Messier"s (1994) classification on Victoria Island 4 of 8 vegetation classes were dominated by sedge species and almost all of the plant communities were dominated by graminoids. The lowland vegetation of Polar Bear Pass on Bathurst Island included 4 meadow communities (Sheard and Geale 1983a). At Truelove Lowland 38% of the land area was classified in two (Muc and Bliss 1977) or three (Muc 1977) meadow communities compared to 25% at Alexandra Fiord (Muc et al. 1989). In the current study 18% of the sites were classified as wet meadows. Species composition was very similar to Alexandra Fiord at Truelove Lowland (Muc 1977) and Sverdrup Pass (Bergeron and Svoboda 1989), with the distinction between meadow types at those locations primarily determined by soil moisture status. These are one of the most important communities as they tend to be more productive (Henry 1987, Henry etal. 1990, Muc  et al. 1994b) and are an important food source for grazing animals (Henry et al. 1986a). At 89  Alexandra Fiord, Henry et al. (1990) found that graminoid species made up 30-80% of above ground standing crop in wet meadow communities and 87% of below ground standing crop. There were some sites that were drier and contained the deciduous shrub  Vaccinium  uliginosum among the sedges, while wetter sites were overwhelmingly comprised of Carex species. Dwarf shrubs (i.e. Dryas integrifolia) and cushion plants were generally found only on dry hummock tops within wet meadows (Henry 1998, Muc et al. 1989). Meadow forbs require wet conditions to germinate (Bell and Bliss 1980). The increase in vascular cover and moss cover in meadow communities is well established (Miller and Alpert 1984, Muc and Bliss 1977) and was found here as well. Bryophyte cover was associated with higher plant diversity as has been previously noted (Bliss and Svoboda 1984, Miller and Alpert 1984) and the opposite relationship being found with lichen cover. In many arctic plant communities moss cover can be considerably greater than vascular cover (Chapin and Shaver 1985a), but mosses and lichens are rarely studied to the same extent as vascular plants despite having quite different responses to environmental changes (Chapin et al. 1995, Chapin and Shaver 1996, Cornelissen et al. 2001). The rocky lichen / Cassiope community was found on well-drained coarse soils and among boulder fields, most of which have been relatively recently deglaciated or are at the base of talus slopes. This community covers 37% of the lowland according to Muc et al. (1989) and is shown to be widely distributed. Of the sample sites in this study, 18% were identified as belonging to this class, or 2 1 % if the snowbed site is included. This community type is often found on raised beaches (Schaefer and Messier 1994, Svoboda 1977) where soils are well drained and plant cover is sparse. Heath communities were divided into two types - Cassiope /Dryas  and Cassiope / Vaccinium  comprising 2 1 % and 18% of the sampling sites, respectively. Together these amounted to 11 of the 28 sampling sites. Muc et al. (1989) found the dwarf shrub-cushion plant community to cover 19% of the lowland and Muc et al. (1994b) showed this community to have the largest standing crop. Nams and Freedman (1987a) found that in an ordination of stands within the heath communities, snowmelt and the availability of moisture throughout the growing season largely determined vascular plant species composition. They found that Cassiope  tetragona  and Dryas intergrifolia alone accounted for 96% of above ground standing crop in this plant community. Similar community types have been described at nearby Sverdrup Pass (Bergeron and Svoboda 1989), and in Greenland (Bay 1992) with very similar dominant species.  90  The Sa//x-dominated site, which corresponds to Muc et a/.'s (1989) deciduous dwarf shrubgraminoid community, makes up only a small proportion of the surface area of the lowland and was only found at one of the sampling locations. It is similar in composition to the Salix heath at Barrow, Alaska described by Webber (1978) and the Sa//x-dominated community described at Sverdrup Pass (Bergeron and Svoboda 1989). Muc et al. (1989) found this community to cover only 4% of the surface area at Alexandra Fiord and, excluding the herb and salt marsh communities, have the lowest standing crop, less than half that of the lichen-cushion plantdwarf shrub community (Muc et al. 1994b). These communities are restricted to well-drained and frequently disturbed riverbanks and flood plains and are rare in the Arctic (Edlund 1983). Direct gradient analysis using RDA showed, as found in many other studies, that the availability of soil moisture is of primary importance in plant community distribution. Temperature was also shown to have some influence, but this variable also had high spatial autocorrelation (see Ch. 4) and, thus, much of the variance in the plant species data could be accounted for simply by including the spatial locations. Covariance in environmental and spatial variables explained 12% of the variance in plant community composition. Productivity and growth rate measures have generally been found to correlate more closely to soil moisture, aeration and nutrient availability than temperature (Chapin and Shaver 1985a). Though temperature may determine the structure of the arctic plant composition at the biome scale, smaller scale differences in plant community composition are more likely a consequence of differences in resources within the soil. Soil moisture has been found to be the most important environmental factor determining vegetation community composition in many studies (e.g. Muc et al. 1989, Webber 1978), although there are few studies where moisture availability was experimentally manipulated (e.g. Henry et al. 1986b, Robinson et al. 1998). Henry et al. (1986b) found significant responses to fertilization but no response to watering at Alexandra Fiord. Few studies have evaluated plant community response to environmental manipulations. Chapin et  al. (1995) did find an increase in shrub cover resulting from both experimental and natural increases in air temperature while Robinson et al. (1998) found significant changes with the addition of nutrients and water but not from warming.  The RDA and subsequent partialing of the spatial component of variation have demonstrated the importance of including spatial location in the analysis of vegetation community composition. An additional 6% of species variance was explained by adding geographic coordinates, which may account for up to 20% of species variance overall. It is important, however, to keep in mind that space itself does not shape vegetation distribution but that it is a surrogate for biological controls on species distribution like competition, predation and  91  dispersal, and unsampled environmental parameters. As such, the 15-20% of variance in species data that is potentially explained by spatial location remains unexplained in terms of the specific processes that are operating. A large portion of the variance in the species data were explained by the environmental variables considered, but there are clearly unsampled environmental variables that are important in differentiating these sites. Snowmelt did not appear to vary at the same scale as the overall distribution of vegetation communities. Snowmelt has been shown to be an important determinant of alpine (Walker et  al. 1993) and arctic vegetation (Evans et al. 1989) by some, while others have found its influence overwhelmed by other factors such as soil moisture and nutrient content (Edlund et  al. 2000, Miller 1982, Webber 1978). In the literature, snowbed or snowflush sites usually refer to the relatively lush communities that develop on the margins and downslope from late-lying snowbeds in otherwise sparsely vegetated areas (Bliss et al. 1994). The snowbed community identified in this work was more akin to the sparsely vegetated areas due to the extreme shortness of the growing season. This is consistent with the snowbed plant community described by others (Billings and Bliss 1959, Evans et al. 1989). As described in Chapter 4, the phenological characteristics of the single snowbed site were also quite different from local trends as a result of the lateness of snowmelt. These patches are not differentiated in Muc et  al.'s (1989) map of the vegetation. Treberg (2000) found that vascular plant species richness decreased in manipulated snowbeds at Alexandra Fiord, but there was no pattern in diversity across natural snowbeds. Biomass, however, did vary with snowmelt date with a decrease in the centre of the snowbed (Treberg 2000) as was found by Billings and Bliss (1959). Molau (1993) emphasizes that, though snowbed vegetation is under considerable stress, these communities are not disturbed sites, which would imply vegetation composed of ruderal species. Instead, the plants able to survive the extremely short growing season are stress tolerators because the stressful conditions are constant from year to year. The sequence of plant communities from north to south also suggests a successional sequence. This could be another factor that is being represented by spatial location variables in the RDA. Jones (1997) found succession with directional species replacement on the glacial foreland at Alexandra Fiord with four stages: moss  graminoid-forb —• deciduous shrub-  moss—* evergreen dwarf-shrub moss tundra. Succession has not been widely considered in the Arctic due to a perceived lack of disturbances and a lack of r-selected plant species well suited for colonizing disturbed sites (Bliss and Gold 1994, Svoboda and Henry 1987). Bliss and Gold (1994) also found succession by species replacement at Truelove Lowland where Svoboda and Henry (1987) found a successional pattern in plant diversity and soil development on  92  beach ridges that emerged over time. Succession is slow in the Arctic due to the cold short growing seasons and the relatively infrequent occurrence of plant reproduction by seeds (Svoboda and Henry 1987). The pattern of plant community distribution found in this research suggests a successional pattern superimposed on the constraints of environmental heterogeneity. The rocky lichen /  Cassiope zone corresponds to the earliest stage of development with the least plant cover and diversity and is located at the sites that have most recently been deglaciated and are most prone to disturbances. The lowland has been deglaciated for 7500 to 8000 years (England et al. 2000). Between 1959 and 1995, the Twin Glacier has retreated approximately 210 m (Jones 1997). The sample sites at the back of the lowland (sites 29 and 30) were within 450 m of the glacier in 2000. The most diverse and productive communities are located in the north, which has been deglaciated for the longest time period on top of having more favourable moisture conditions. A comparison between the current description of the composition and distribution of plant communities and the description by Muc et al. (1989) is made difficult by several factors. The first is the scale and method of sampling. The previous study employed a sampling technique whereby areas were subjectively deemed to belong to a distinct vegetation class and sampling was undertaken to confirm or refute this. This technique introduces the bias of the researcher, but the much higher number and density of sampling points resulted in a robust analysis of relationships and distributions, though the delineation of polygons remains subjective. The current sampling scheme was based on the assignment of sample locations independent of subjectivity. The considerably smaller number of sampling points were intended to be used primarily for ground-truthing aerial photographs which was not, in the end, successful. The result is that the scale of the results reflects the scale of sampling, in which the current study uncovered larger-scale patterns in plant community distributions than did the fine-scale sampling done by Muc et al. (1989). The map they produced is a patchy mosaic of communities without evidence of the general patterns detected here. Such a wide sampling grid is bound to undersample or miss some plant communities (e.g. salt marshes). Sampling is always a compromise between the feasibility of data collection and the precision of the results. In the absence of the existing map of vegetation at Alexandra Fiord, the current description of plant distributions provides an informative summary of the vegetation communities and their general distribution without the intensive sampling required for the production of a detailed vegetation map. Despite this, most of the communities described by  93  Muc et al. (1989) were identified and some even sub-divided to further elucidate differences in community types. There is, therefore, ample evidence to be confident in this classification of the dominant vegetation types at this site, though there are not sufficient data to estimate the proportion of the lowland covered by each of the vegetation classes. Determining the spatial correspondence between the plant communities defined at each of the 28 sites and the map by Muc et al. (1989) is very difficult, due primarily to the many sources of spatial error and the relatively small size of most of the vegetation polygons in their map. With only a 10 m margin of error on either side of each plant community boundary, 39% of the lowland falls within this zone. Locations of sampling sites were recorded with a hand-held GPS which introduces roughly 3-10 m of error. The existing vegetation map was made by hand, digitized by hand, and rectified using the approximate locations of common features in the aerial photographs. All of these factors combined virtually eliminate any meaningful comparison between the specific locations of the plant communities described herein and those of Muc et  al. (1989). In any study of spatially distributed features the issue of scale arises in various aspects, from sampling to the interpretation of results. In this case, there are generally multiple scales at which given processes are operating and corresponding characteristic scales of the responses to those processes. For example, many authors describe the pattern of vegetation that is readily visible across hummocks as the vegetation of the moist inter-hummock areas differs from dry hummock tops (Sohlberg and Bliss 1984, Young et al. 1999) and this variability in vegetation composition reflects small scale microtopography. Snowmelt could be considered a mid-scale feature as late-lying snowbeds are primarily governed by a larger scale of variability in topography (e.g. beach ridges). At the landscape scale, features such as temperature gradients or variability in substrate lithology will become influential in determining the observed distribution of vegetation. In many cases the use of a multi-scale sampling and analysis approach is the only way to determine at what scale each process is operating. Spatial generalization occurs at all scales. In the classifications of vegetation for the Arctic as a whole, Alexandra Fiord falls under the classes described as simply High Arctic (Polunin 1951), cushion-forb (Walker 2000), and enriched prostrate shrub (Edlund et al. 2000) among others. The High Arctic or polar desert designations clearly are not representative of the vegetation at oases like the Alexandra Fiord lowland. Walker's (2000) and Edlund et a/.'s (2000) classifications are more applicable yet do not reflect the vegetation on the surrounding uplands (Batten and Svoboda 1994). As is invariably found, when the resolution of sampling is  94  increased the distribution of vegetation becomes increasingly patchy (Mosbech and Hansen 1994, Spjekavik 1995). Using spatial data collected with differing methods and standards results in uncertainty that will, in many cases, not be possible to resolve. In this study, inadequate spatial control for mosaicking photographs resulted from both the shortage of ground control points in each photograph and the error associated with uncorrected GPS coordinates. This made impossible the accurate comparison of vegetation data between the current study and the existing vegetation map and accurate overlaying of the airphoto mosaics of snowmelt. However, visual comparisons between the photos and maps have provided interesting insights into the relationships among snowmelt, other environmental variables, and vegetation community distribution.  95  Chapter Six Conclusions Examining the interrelationships among the spatial distributions of several biotic and abiotic variables has provided some interesting insights into the way environmental variability structures plant communities and governs plant phenology. Phenological development of  Cassiope tetragona and Dryas integrifolia was strongly correlated to the temperature gradient across the lowland but Saxifraga oppositifolia and Salix arctica were not. Cassiope and Dryas  flowered later in the season while Saxifraga and Sa//x flowered shortly after snowmelt. Unfortunately, snowmelt dates were not available in each of the monitored plots in order to compare to phenological stages, but a landscape-scale trend in snowmelt date was not detected. Given the rapidity with which Saxifraga and Salix develop following snowmelt, this factor is likely more important in determining phenology in these species than in Cassiope and Dryas.  Dryas and Cassiope differed in the magnitude of their responses to spatial environmental gradients as compared to environmental variability over time. Dryas had a more pronounced response to temporal variability, with the range of dates for each phenological stage being 7 to 11 days longer among years than over the spatial gradients in the 2000 growing season.  Cassiope, on the other hand, had much more fixed timing for phenological development at a given place over time, while varying considerably across spatial gradients. This supports the notion that species will respond individualistically to environmental change. It would appear that  Dryas is more likely to adapt to changing conditions, but more research is required to evaluate this assertion. These comparisons provide much needed context for experiments such as ITEX, as efforts are made to scale-up plot-level results to landscapes in order to make predictions about ecosystem response to climate change. Five major plant communities and two subtypes were defined: wet meadows, rocky lichen-  Cassiope and snowbed, Dryas dominated and Salix dominated, Cassiope-Dryas Cassiope-Vaccinium  heath, and  heath. Similar communities have been described in many other parts of  the Arctic. The distribution of plant communities was highly aggregated, with the general pattern showing the rocky lichen-Cass/ope and Cassiope-Vaccinium  heath in the south  adjacent to the glaciers, the Cassiope-Dryas heath dominating the centre of the lowland, and the meadows and Dryas dominated sites on wet and dry areas, respectively, in the north. Moisture was found to be the most important environmental factor as is commonly found in plant community studies in the Arctic. Temperature was also an important factor in the indirect 96  gradient analysis, but this variable was highly spatially autocorrelated and much of the variation explained by temperature could be explained by spatial location information alone. A snowbed community was defined, but the pattern of snowmelt did not vary at the same scale as the vegetation according to these results, and is thus not likely a major determinant of landscapescale plant community distribution. The overall distribution is most likely a successional sequence, resulting from the retreat of both Pleistocene and Little Ice Age glacier advances. Variograms and correlograms were useful in characterizing the nature of spatial autocorrelation in the variables. However, the regular sampling grid and small sample size resulted in some limitations, primarily not having pairs of points at short lag-distance classes. Appropriate correlation analyses that account for spatial autocorrelation are highly desirable, but these methods are still in the early stages of development and not widely used or accessible. Comparing the plant communities described in this research with those of Muc et al. (1989) was made difficult by differences in the scale of sampling and by the many sources of error in spatial location data. However, a visual comparison between the maps was interesting, as the landscape-scale trend observed in the current study was not apparent in the patchy mosaic they described. Colour digital aerial photographs were found to be inadequate for plant community mapping purposes at this site. The use of infrared wavelengths would be required for a remotely sensed map of vegetation. Future work suggested by this research includes a more exhaustive comparison between responses to ITEX manipulations and natural gradients in temperature, moisture and snowmelt. With the differences in phenological development in Saxifraga and Salix an evaluation of their response to temporal variation will also provide valuable information. Also, studying the responses of other growth forms, notably graminoid species, to similar spatial gradients would improve our ability to scale-up experimental responses. An aircraft or satellite-based remotely sensed map of the vegetation would also be desirable to compare in order to the communities and patterns described herein.  97  References Alatolo, J.M. and Totland, 0 . 1997. Response to simulated climatic change in an alpine and subarctic pollen-risk strategist, Silene acaulis. Global Change Biology, 3 (Suppl. 1): 74-79. Alexander, V., Billington, M. and Schell, D.M. 1978. Nitrogen fixation in arctic and alpine tundra, in: Vegetation and Production Ecology of an Alaskan Arctic Tundra, Tieszen, L.L. (ed.), New York: Springer, pp. 539-558. Anderson, D.G. and Bliss, L.C. 1998. Association of Plant Distribution Patterns and Microenvironments on Patterned Ground in a Polar Desert, Devon Island, N.W.T., Canada. Arctic, Antarctic and Alpine Research, 30: 97-107 Archibold, O.W. 1995. Ecology of World Vegetation. London: Chapman and Hall. Arft, A.M., Walker, M.D., Gurevitch, J., Alatalo, J.M., Bret-Harte, M.S., Dale, M., Diemer, M., Gugerii, F., Henry, G.H.R., Jones, M.H., Hollister, R.D., Jonsdottir, I.S., Laine, K., Levesque, E., Marion, G.M., Molau, U., Molgaard, P., Nordenhall, U., Raszhivin, V., Robinson, C.H., Starr, G., Stenstrom, A., Stenstrom, M., Totland, O., Turner, P.L., Walker, L.J., Webber, P.J., Welker, J.M. and Wookey, P.A. 1999. Responses of tundra plants to experimental warming: Meta-analysis of the International Tundra Experiment. Ecological Monographs. 69: 491-511 Atkin, O.K., Villar, R. and Cummins, W.R. 1993. The ability of several high arctic plant species to utilize nitrate-nitrogen under field conditions. Oecologia, 96: 239-245. Atkinson, P.M. 1996. Optimal sampling strategies for raster-based geographical information systems. Global Ecology and Biogeography Letters, 5: 271-280. Babb, T.A. and Bliss, L.C. 1974. Effects of physical disturbance on arctic vegetation in the Queen Elizabeth Islands. Journal of Applied Ecology, 11: 549-562. Ball, P. and Hill, N. 1994. Vascular plants at Alexandra Fiord. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 255-256. Barrett, P.E. and Teeri, J.A. 1973. Vascular plants of the Truelove Inlet region, Devon Island. Arctic, 26: 58-67. Batten, D.S. and Svoboda, J. 1994. Plant communities on the uplands in the vicinity of the Alexandra Fiord Lowland, In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 97-111. Bay, C. 1992. A phytogeographical study of the vascular plants of northern Greenland, north of 74 northern latitude. Meddelelser om Grenland. Bioscience, 36. Bean D. and Henry, G.H.R. 2002. Climate Change Impacts on Tundra Ecosystems. EMAN-North Report, Environment Canada. Bell, K.L. and Bliss, L.C. 1980. Plant repoduction in a high arctic environment. Arctic and Alpine Research, 12: 1-10. Bergeron, J.-F. and Svoboda, J. 1989. Plant communities of Sverdrup Pass, Ellesmere Island, N.W.T. The Musk-Ox, 37: 76-85. 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. Bliss, L.C. 1956. A comparison of plant development in microenvironments of arctic and alpine tundras. Ecological Monographs, 26: 303-337.  98  Bliss, L.C. 1971. Arctic and alpine plant life cycles. Annual Review of Ecology and Systematica, 2: 405438. Bliss, L.C. (ed.) 1977a. Truelove Lowland, Canada: A High Arctic Ecosystem. Edmonton: University of Alberta Press. Bliss, L.C. 1977b. Introduction. In: Truelove Lowland, Canada: A High Arctic Ecosystem. Bliss, L.C. (ed.). Edmonton: University of Alberta Press, pp. 1-11. Bliss, L.C. 1997. Arctic ecosystems of North America. In: Wielgolaski, F.E. (ed.) Polar and Alpine Tundra. Amsterdam: Elsevier, pp. 551-684. Bliss L.C. and Gold. 1994. The patterning of plant communities and edaphic factors along a high arctic coastline - implications for succession. Canadian Journal of Botany. 72:1095-1107. Bliss, L.C, Henry, G.H.R., Svoboda, J., and Bliss, D.I. 1994. Patterns of plant distribution within two polar desert landscapes. Arctic and Alpine Research, 26: 46-55. Bliss L.C. and Peterson, K.M. 1992. Plant succession, competition, and the physiological constrains of species in the Arctic, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego, Academic Press, Inc. pp. 111-138. Bliss, L.C. and Svoboda, J. 1984. Plant communities and plant production in the western Queen Elizabeth Islands. Holarctic Ecology. 7: 325-344. Borcard, D., Legendre, P. and Drapeau, P. 1992. Partialling out the spatial component of ecological variation. Ecology, 73:1045-1055. Broil, G., Tarnocai, C , and Mueller, G. 1999. Interactions between vegetation, nutrients and moisture in soils in the Pangnirtung Pass area, Baffin Island, Canada. Permafrost Periglacial Process. (10)3: 265-277. Brosard, T. and Joly, D. 1994. Probability models, remote sensing and field observation: test for mapping some plant distributions in the Kongsfjord area, Svalbard. Polar Research. 13:153-161. Callaghan, T.V. and Jonasson, S. 1995. Implications for changes in arctic plant biodiversity from environmental manipulation experiments. In: Arctic and Alpine Biodiversity. Chapin III, F.S. and Korner, C. (eds.). New York: Springer-Verlag. pp. 151-166. Chapin III, F.S., BretHarte, M.S., Hobbie, S.E. and Zhong, H.L. 1996. Plant functional types as predictors of transient responses of arctic vegetation to global change. Journal of Vegetation Science, 7:347358. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.) 1992a. Arctic Ecosystems in a Changing Climate, San Diego, Academic Press. Chapin III, F.S., Jefferies, R.L, Reynolds, J.F., Shaver, G.R. and Svoboda, J. 1992b. Arctic plant physiological ecology: A challenge for the future, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L, Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego, Academic Press, Inc. pp. 3-7. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. 1992c. Arctic plant physiological ecology in an ecosystem context, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego, Academic Press, Inc. pp. 441-452. Chapin III, F.S. and Shaver, G.R. 1985a. Arctic, in: Physiological Ecology of North American Plant Communitites. Chabot, B.F. and Mooney, H.A. (eds.), Chapman and Hall. pp. 16-40.  99  Chapin III, F.S. and Shaver, G.R. 1985b. Individualistic growth responses of tundra plant species to environmental manipulations in the field. Ecology, 66: 564-576. Chapin III, F.S. and Shaver, G.R. 1996. Physiological and growth responses of arctic plants to a field experiment simulating climate change. Ecology, 77: 822-840. Chapin III, F.S., Shaver, G.R., Giblin, A.E., Nadelhoffer, K.J. and Laundre, J.A. 1995. Responses of Arctic tundra to experimental and observed changes in climate. Ecology. 76: 694-711. Chapin III, F.S. and Starfield, A.M. 1997. Time lags and novel ecosystems in response to transient climatic change in arctic Alaska. Climatic Change, 35:449-461. Clifford, P., Richardson, S. and Hemon, D. 1989. Assessing the significance of the correlation between two spatial processes. Biometrics, 45:123-134. Cochran, W.G. 1977. Sampling Techniques. 3 ed. New York: John Wiley and Sons. rd  Congalton, R.G. 1991. A review of assessing the accuracy of classifications of remotely sensed data. Remote Sensing of Environment, 37: 35-46. Cornelissen, J.H.C., Callaghan, T.V., Alatalo, J.M. Michelsen, A., Graglia, E., Hartley, A.E., Hik, D.S., Hobbie, S.E., Press, M.C., Robinson, C.H., Henry, G.H.R., Shaver, G.R., Phoenix, G.K., Jones, D.G., Jonasson, S., Chapin, F.S., Molau, U., Neill, C , Lee, J.A., Melillo, J.M., Sveinbjornsson, B. and Aerts, R. 2001. Global change and arctic ecosystems: is lichen decline a function of increases in vascular plant biomass? Journal of Ecology, 89: 984-994. Craighead, J.J., Craighead, F.L., Craighead, D.J. and Redmond, R.L. 1988. Mapping arctic vegetation in northwest Alaska using Landsat MSS imagery. National Geographic Research. 4: 496-527. Crawford, R.M.M., Chapman, H.M., Abbott, R.J. and Balfour, J. 1993. Potential impact of climatic warming on arctic vegetation. Flora. 188: 367-381. Dale, M.R.T. 1999. Spatial Pattern Analysis in Plant Ecology. Cambridge University Press. Dalpe, Y., Aiken, S.G. 1998. Arbuscular mycorrhizal fungi associated with Festuca species in the Canadian High Arctic. Canadian Journal of Botany. 76:1930-1938. de Molenaar, J.G. 1987. An ecohydrological approach to floral and vegetational patterns in arctic landscape ecology. Arctic and Alpine Research. 19: 414-424. Del Barrio, G., Alvera, B., Puigdefabregas, J. and Diez, C. 1997. Response of high mountain landscapes to topographic variables: Central Pyrenees. Landscape Ecology, 12: 95-115. Dutilleul, P., Clifford, P., Richardson, S. and Hemon, D. 1993. Modifying the t test for assessing the correlation between two spatial processes. Biometrics, 49: 305-314. Edlund, S.A. 1983. Bioclimatic zonation in a high arctic region: central Queen Elizabeth Islands, in: Current Research, Part A. Paper 83-1A. Geological Survey of Canada, pp. 381-390. Edlund, S.A. and Alt, B.T. 1989. Regional congruence of vegetation and summer climate patterns in the Queen Elizabeth Islands, Northwest Territories, Canada. Arctic, 42: 3-23. Edlund, S.A. and Garneau, M. 2000. Overview of vegetation zonation in the Arctic. In Garneau, M. and Alt, B T. Environmental response to climate change in the Canadian High Arctic. Geological Survey of Canada Bulletin, 529:113-127.  100  Elvebakk, A. 1997. Tundra diversity and ecological characteristics of Svalbard. In: Wielgolaski, F.E. (ed.) Polar and Alpine Tundra. Amsterdam: Elsevier, pp. 347-360. England, J., Smith, I.R. and Evans, D.J.A. 2000. The last glaciation of east-central Ellesmere Island, Nunavut: ice dynamics, deglacial chrolology, and sea level change. Canadian Journal of Earth Sciences, 37: 1355-1371. Epstein, H.E., Walker, M.D., Chapin, F.S. and Starfield, A.M. 2000. A transient nutrient-based model of Arctic plant community response to climatic warming, Ecological Applications, 10(3): 824-841. Evans, B.M., Walker, D.A., Benson, C.C., Nordstrand, E.A. and Peterson, G.W. 1989. Spatial interrelationships between terrain, snow distribution and vegetation patterns at an arctic foothills site in Alaska. Holarctic Ecology. 12: 270-278. Forman, R.T.T. and Godron, M. 1986. Landscape Ecology. New York: John Wiley and Sons. Fortin, M.-J. 1999. Effects of sampling unit resolution on the estimation of spatial autocorrelation. Ecoscience. 6: 636-641. Fortin, M.-J., Drapeau, P. and Legendre, P. 1989. Spatial autocorrelation and sampling design in plant ecology. Vegetatio, 83: 209-222. Freedman, B., Svoboda, J. and Henry, G.H.R. 1994. Alexandra Fiord - An ecological oasis in the polar desert. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 1-9. Galen, C. and Stanton, M.L. 1991. Consequences of emergence phenology for reproductive success in Ranunculus adoneus (Ranunculaceae). Americal Jounal of Botany, 78: 978-988. Goedickemeier, I., Wildi, O. and Kiensast, F. 1997. Sampling for vegetation survey: Some properties of a GIS-based stratification compared to other statistical methods. Coenoses, 12: 43-50. Goodchild, M.F. 1994. Integrating GIS and remote sensing for vegetation analysis and modeling: methodological issues. Journal of Vegetation Science. 5: 615-626. Goodchild, M.F., Steyaert, L.T. Parks, B.O., Johnston, C , Maidment, D., Crane, M., and Glendinning, S. (eds.) 1996. GIS and Environmental Modeling: Progress and Research Issues. Fort Collins: GIS World Books. Gooding, R.F., Rackham, D., Holland, J.P. and Robertson, D. 1997. Detailed surveying and mapping of plant communities on featureless terrain. Grass and Forage Science. 52: 439-444. Gottfried, M., Pauli, H. and Grabher, G. 1998. Prediction of vegetation patterns at the limits of plant life: a new view of the alpine-nival ecotone. Arctic and Alpine Research. 30: 207-221. Gough, L., Shaver, G.R., Carroll, J., Royer, D.L. and Laundre, J.A. 2000. Vascular plant species richness in Alaskan arctic tundra: the importance of soil pH. Journal of Ecology. 88: 54-66. Gould, W.A., and Walker, M.D. 1999. Plant communities and landscape diversity along a Canadian arctic river. Journal of Vegetation Science. 10: 537-548. Grime, J.P. 1979. Plant Strategies and Vegetation Processes. Toronto: Wiley. Haining, R. 1991. Bivariate Correlation with Spatial Data. Geographical analysis, 23: 210. Harrison, A.R. and Dunn, R. 1993. Problems of sampling the landscape. In: Haines-Young, R., Green, D.R., and Cousins, S.H. (eds.). Landscape Ecology and Geographic Information Systems. New York: Taylor and Francis, pp. 101-109.  101  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. and Jonasson, S. 1995. Effects of simulated climate change on the sexual reproductive effort of Cassiope tetragona. In: Global Change and Arctic Terrestrial Ecosystems. European Commission, Ecosystems Research Report 10, pp. 109-114. Henry, G.H.R. 1987. Ecology of Sedge Meadows of a Polar Desert Oasis, PhD Thesis, Department of Botany, University of Toronto. Henry, G.H.R. (ed.). 1997. Global Change Biology, V. 3 (Suppl. 1). Henry, G.H.R. 1998. Environmental influences on the structure of sedge meadows in the Canadian High Arctic. Plant Ecology. 134:119-129. Henry, G.H.R., Freedman, B. and Svoboda, J. 1986a. Survey of vegetated areas and muskox populations in east-central Ellesmere Island. Arctic, 39: 78-81. Henry, G.H.R, Freedman, B. and Svoboda, J. 1986b. Effects of fertilization on three tundra plant communities of a polar desert oasis. Canadian Journal of Botany, 64: 2502-2507. Henry, G.H.R. and Molau, U. 1997. Tundra plants and climate change: the International Tundra Experiment (ITEX). Global Change Biology. 3(suppl.): 1-9. Henry, G.H.R. and Svoboda, J. 1986. Dinitrogen fixation (acetylene reduction) in high arctic sedge meadow communities. Arctic and Alpine Research, 18:181-187. Henry, G.H.R., Svoboda, J. and Freedman, B. 1990. Standing crop and net production of sedge meadows of an ungrazed polar desert oasis. Canadian Journal of Botany. 68: 2660-2667. Hollister R.D. and Webber, P.J. 2000. Biotic validation of small open-top chambers in a tundra ecosystem. Global Change Biology, 6: 835-842. Hope, A.S., Kimball, J.S. and Stow, D.A. 1993. The relationship between tussock tundra spectral reflectance properties and biomass and vegetation composition. International Journal of Remote Sensing, 14:1861-1874. Houghton, J.T., Meira Filho, L.G., Callender, B.A., Harris, N., Kattenberg, A. and Maskell, K. (eds.) 1995. The Science of Climate Change - Contribution of Working Group I to the Second Assessment of the Intergovernmental Panel on Climate Change. Cambridge University Press. Jefferies, R.L., Svoboda, J., Henry, G.H.R., Raillard, M. and Ruess, R. 1992. Tundra grazing systems and climate change, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego, Academic Press, Inc. pp. 391-412. Johnstone, J.F. 1995. 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. MSc Thesis, Department of Geography, University of British Columbia. Johnstone, J.F. and Henry, G.H.R. 1997. Retrospective analysis of growth and reproduction in Cassiope tetragona and relations to climate in the Canadian High Arctic. Arctic and Alpine Research, 29: 459469. Jonasson, S. 1997. Buffering of arctic plant responses in a changing climate. In: Global Change and Arctic Terrestrial Ecosystems. Oechel, W.C., Callaghan, T.J. and Gilmanov, T., Holten, J.I., Maxwell, B., Molau, U. and Sveinbjornsson, B. (eds.), New York: Springer, pp. 365-380.  102  Jones, G.A. 1997. Primary Plant Succession on the Twin Glacier Foreland, Alexandra Fjord, Ellesmere Island, Canadian High Arctic. M.Sc. Thesis, Department of Geography, University of British Columbia. Jones, M.H. 1995. Experimental Investigations into Effects of Climate Change on High Arctic Plants. PhD thesis. Department of Renewable Resources, University of Alberta. Jones, M.H.J., Bay, C. and Nordenhall, U. 1997. Effectrs of experimental warming on arctic willows (Salix spp.): a comparison of responses from the Canadian High Arctic, Alaskan Arctic, and Swedish Subarctic. Global Change Biology, 3(Suppl. 1): 125-145. Jonsdottir, I.S., Virtanen, R., and Karnefelt, I. 1999. Large-scale differentiation and dynamics in tundra plant populations and vegetation. Ambio. (28)3: 230-238. Kohn, L.M. and Stasovski, E. 1990. The mycorrhizal status of plants at Alexandra Fiord, Ellesmere Island, Canada, a high arctic site. Mycologia, 82: 23-35. Kudo, G. 1991. Effects of snow-free period on the phenology of alpine plants. Arctic and Alpine Research. 23:436-443. Kudo, G. 1993. Relationship between flowering time and fruit-set of the entomophilous alpine shrub, Rhododendron aureum (Ericaceae), inhabiting snow patches. American Journal of Botany, 80: 13001304. Kummerow, J., McMaster, G.S. and Drause, D.A. 1980. Temperature effects on growth and nutrient contents in Ehophorum vaginatum under controlled environmental conditions. Arctic and Alpine Research, 12: 335-341. Labine, C.L. 1994. Meteorology and climatology of the Alexandra Fiord Lowland. In: Ecology of a Polar Oasis: Alexandra Fiord, Ellesmere Island, Canada. Svoboda, J. and Freedman, B. (eds.), Toronto: Captus University Press, pp. 23-39. Larigauderie, A. and Kummerow, J. 1991. The sensistivity of phenological events to changes in nutrient availability for several plant growth forms in the Arctic. Holarctic Ecology, 14: 38-44. Legendre, P. 1993. Spatial autocorrelation: Trouble or a new paradigm? Ecology, 74:1659-1673. Legendre, P. and Fortin, M. 1989. Spatial pattern and ecological analysis. Vegetatio. 80:107-138. Lenihan, J.M. and Neilson, R.P. 1995. Canadian vegetation sensitivity to projected climatic change at three organizational levels. Climatic Change, 30:27-56. Levesque, E. 1997. Plant Distribution and Colonization in Extreme Polar Deserts. Ph.D. Thesis. University of Toronto. Levis, S., Foley, J.A., Brovkin, V., and Pollard, D. 1999. On the stability of the high-latitude climatevegetation system in a coupled atmosphere-biosphere model. Global Ecology and Biogeography. 8(6): 489-500 . Lo, C P . and Watson, L.J. 1998. The influence of geographic sampling methods on vegetation map accuracy evaluation in a swampy environment. Photogrammetric Engineering and Remote Sensing, 64:1189-1200. Maass, W. and Nams, M.L.N. 1994. List of Lichens at Alexandra Fiord. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 251-252. Maass, W., Hoisington, B. and Nams, M.L.N. 1994. List of Bryophytes of Alexandra Fiord. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 253-254.  103  Maling, D.H. 1988. Measurements From Maps. New York: Pergamon Press. Marion, G.M., Henry, G.H.R., Freckman, D.W., Johnstone, J., Jones, G., Jones, M.H., Levesque. E., Molau, U., Molgaard, P., Parsons, A.N., Svoboda, J., Virginia, R.A. 1997 Open-top designs for manipulating field temperature in high-latitude ecosystems, Global Change Biology, 3 (Suppl. 1): 2032. Marsden, R. 1992. Snow melt, flowering phenology and reproductive success in tundra plants. - M.Sc. Thesis, University of Wales, Bangor. Maxwell, B. 1981. Climatic regions of the Canadian Arctic Islands. Arctic, 34: 225-240. Maxwell, B. 1992. Arctic climate: potential for change under global warming, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego: Academic Press, pp. 11-34. McKane, R.B., Johnson, L.C, Shaver, G.R., Nadelhoffer, K.J., Rastetter, E.B., Fry, B., Giblin, A.E., Kielland, K., Kwiatkowski, B.L., Laundre, J.A. and Murray, G. 2002. Resource-based niches provide a basis for plant species diversity and dominance in arctic tundra. Nature, 415: 68-71. Michaelsen, J., Schimel, D.S., Friedl, M.A., Davis, F.W. and Dubayah, R.C. 1994. Regression tree analysis of satellite and terrain data to guide vegetation sampling and surveys. Journal of Vegetation Science, 5: 673-686. Miller, N.G. and Alpert, P. 1984. Plant associations and edaphic reatures of a high arctic mesotopographic setting. Arctic and Alpine Research. 16: 11-24. Miller, P.C. 1982. Environmental and vegetational variation across a snow accumulation area in montane tundra in central Alaska. Holarctic Ecology. 5: 85-98. Molau, U. 1993. Relationships between flowering phenology and life history strategies in tundra plants. Arctic and Alpine Research. 25: 391-402. Molau, U. 1997. Phenology and reproductive success in arctic plants. In: Global Change and Arctic Terrestrial Ecosystems. Oechel, W.C., Callaghan, T.J. and Gilmanov, T., Holten, J.I., Maxwell, B., Molau, U. and Sveinbjornsson, B. (eds.) New York: Springer, pp. 153-170. Molau, U. and Edlund, S. 1996. Plant Response Variables. In; Molau, U. and Molgaard, P. (eds.) ITEX Manual, Second Edition, Danish Polar Centre Molau, U. and Shaver, G.R. 1997. Controls on seed production and seed germinability in Eriophorum vaginatum. Global Change Biology, 3 (Suppl. 1): 80-88. Mosbech, A. and Hansen, B.U. 1994. Comparison of satellite imagery and infrared aerial photography as vegetation mapping methods in an arctic study area; Jameson Land, East Greenland. Polar Research. 13:139-152. Muc, M. 1977. Ecology and primary production of sedge-moss meadow communities, Truelove lowland. In: Tuelove Lowland, Canada: A High Arctic Ecosystem. Bliss, L.C. (ed.). Edmonton: University of Alberta Press, pp. 157-184. Muc, M. and Bliss, L.C. 1977. Plant communities of Truelove Lowland. In: Tuelove Lowland, Canada: A High Arctic Ecosystem. Bliss, L.C. (ed.). Edmonton: University of Alberta Press, pp. 143-154. Muc, M., Freedman, B. and Svoboda, J. 1989. Vascular plant communities of a polar oasis at Alexandra Fiord, Ellesmere Island. Canadian Journal of Botany. 67:1126-1136.  104  Muc, M., Svoboda, J. and Freedman, B. 1994a. Soils of an extensively vegetated polar desert oasis, Alexandra Fiord, Ellesmere Island. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 41-51. Muc, M., Svoboda, J. and Freedman, B. 1994b. Aboveground standing crop in plant communities of a polar desert oasis, Alexandra Fiord, Ellesmere Island. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 65-74. Murray, C. and Miller, C.F. 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. Murray, D.F. 1997. Regional and local vascular plant diversity in the Arctic. Opera Botanica. 132: 9-18. Myneni, R.B., Keeling, CD. Tucker, C.J., Asrar, G., and Nemani, R.R. 1997. Increased Plant Growth in the Northern High Latitudes from 1981 to 1991. Nature 386: 698-702. Nams, M.L.N, and Freedman, B. 1987a. Ecology of heath communities dominated by Cassiope tetragona at Alexandra Fiord, Ellesmere Island. Holarctic Ecology, 10: 22-32. Nams, M.LN. and Freedman, B. 1987b. Phenolgy and resource allocation in a high arctic evergreen dwarf shrub, Cassiope tetragona. Holarctic Ecology. 10:128-136. Neldner, V.J., Crossley, D.C. and Cofinas, M. 1995. Using geographic information systems (GIS) to determine the adequacy of sampling in vegetation sureys. Biological Conservation, 73:1-17. Nilsen, L., Brossard, T. and Joly, D. 1999. Mapping plant communities in a local arctic landscape applying a scanned infrared aerial photograph in a geographic information system. International Journal of Remote Sensing. 20: 463-480. Oechel, W.C, Callaghan, T.J. and Gilmanov, T., Holten, J.I., Maxwell, B., Molau, U. and Sveinbjornsson, B. (eds.) 1997. Global Change and Arctic Terrestrial Ecosystems. New York: Springer. Ostendorf, B. and Reynolds, J.F. 1993. Relationships between a terrain-based hydrologic model and patch-scale vegetation patterns in an arctic tundra landscape. Landscape Ecology, 8: 229-237. Parsons, A.N., Welker, J.M. Wookey, P.A., Press, M.C, Callaghan, T.V. and Lee, J.A. 1994. Growth responses of four sub-Arctic dwarf shrubs to simulated environmental change. Journal of Ecology. 82: 307-318. Podani, J. 1984. Analysis of mapped and simulated vegetation patterns by means of computerized sampling techniques. Acta Botanica Hungarica, 30:403-425. Polunin. N. 1948. Botany of the Canadian Eastern Arctic. Part III. Vegetation and Ecology. National Museum of Canada Bulletin 104. Ottawa. Polunin, N. 1951. The real Arctic: suggestions for its delimitation, subdivision, and characterization. Journal of Ecology, 39: 308-315. Pop, E.W., Oberbauer, S. and Starr, G. 2000. Predicting vegetative bud break in two arctic deciduous shrub species, Salix pulchra and Betula nana. Oecologia, 124:176-184. Porsild, A.E. 1951. Plant life in the Arctic. Canadian Geographical Journal. 42:120-145. Porsild A. E. and Cody, W. J. 1980. Vascular plants of the Continental Northwest Territories, Canada. National Museum of Canada, Ottawa. Prentice, I.C, Cramer, W., Harrison, S.P., Leemans, R., Monserud, R.A., and Solomon, A.M. 1992. A global biome model based on plant physiology and dominance, soil properties and climate. Journal of Biogeography, 19:117-134.  105  Prentice, I.C., Monserud, R A , Smith, T.M. and Emanuel, W.R. 1993. Modeling large-scale vegetation dynamics, in: Vegetation Dynamics and Global Change. New York: Chapman and Hall, pp. 235-250. Oechel, W.C., Callaghan, T., Gilmanov, T., Holten, J.I. Maxwell, B., Molau, U. and Sveinbjornsson (eds.). 1997. Global Change and Arctic Terrestrial Ecosystems. New York: Springer. Rannie, W.F. 1986. Summer air temperture and number of vascular species in arctic Canada. Arctic. 39: 133-137. Reynolds, J.F. and Leadley P.W. 1992. Modeling the response of Arctic plants to changing climate, in: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego: Academic Press, pp. 413-440. Robinson, C.H., Wookey, P.A., Lee, J.A., Callaghan, T.V. and Press, M.C. 1998. Plant community responses to simulated environmental change at a high arctic polar semi-desert. Ecology, 79: 856866. Rossi, R.E. Mulla, D.J., Journel, A.G. and Franz, E.H. 1992. Geostatistical tools for modeling and interpreting ecological spatial dependence. Ecological Monographs, 62: 277-314. Saville, D.B.O. 1961. The botany of the Northwestern Queen Elizabeth Islands. Canadian Journal of Botany. 39: 909-942. Sawada, M. 1999. ROOKCASE: An Excel 97/2000 Visual Basic (VB) add-in for exploring global and local spatial autocorrelation. Bulletin of the Ecological Society of America, 80(4):231-234. Schaefer, J.A. and Messier, F. 1994. Composition and spatial structure of plant communities on southeastern Victoria Island, arctic Canada. Canadian Journal of Botany. 72: 1264-1272. Serreze, M.C, Walsh, J.E., Chapin, F.S., Osterkamp, T., Dyurgerov, M., Romanovsky, V., Oechel, W.C, Morison, J., Zhang, T. and Barry, R.G. 2000. Observational evidence of recent change in the northern high-latitude environment. Climatic Change. 46:159-207. Shaver, G.R. and Kummerow, J. 1992. Phenology, resource allocation, and growth of arctic vascular plants. In: Arctic Ecosystems in a Changing Climate. Chapin III, F.S., Jefferies, R.L., Reynolds, J.F., Shaver, G.R. and Svoboda, J. (eds.), San Diego, Academic Press, Inc. pp. 193-211. Shaver, G.R., Canadell, J., Chapin, F.S., Gurevitch, J., Harte, J., Henry, G., Ineson, P., Jonasson S., Melillo, J., Pitelka, L., Rustad, L. 2000. Global warming and terrestrial ecosystems: A conceptual framework for analysis. Bioscience, 50: 871-882. Sheard, J.W. and Geale, D.W. 1983a. Vegetation studies at Polar Bear Pass, Bathurst Island, N.W.T. I. Classification of plant communities. Canadian Journal of Botany. 61:1618-1636. Sheard, J.W. and Geale, D.W. 1983b. Vegetation studies at Polar Bear Pass, Bathurst Island, N.W.T. II. Vegetation-environment relationships. Canadian Journal of Botany. 61: 1637-1646. Smartt, P.F.M. and Grainger, J.E.A. 1974. Sampling for vegetation survey: some aspects of the behavior of unrestricted, restricted, and stratified techniques. Journal of Biogeography, 1:193-206. Sohlberg, E.H. and Bliss, L.C. 1984. Microscale pattern of vascular plant distribution in two high arctic plant communities. Canadian Journal of Botany, 62: 2033-2042. Sorensen, T. 1941 Temperature relations and phenology of northeast Greenland flowering plants. Meddelesler Om Gronland, Bioscience. 125:1-315. Spjelkavic, S. 1995. A satellite-based map compared to a traditional vegetation map of arctic vegetation in the Ny-Alesund area, Svalbard. Polar Record. 31: 257-269.  106  Starr, G., Oberbauer, S.F. and Pop, E.W. 2000. Effects of lengthened growing season and soil warming on the phenology and physiology of Polygonum bistorta. Global Change Biology, 6: 357-369. Steere, W.C. 1978. Floristics, phytogeography and ecology of arctic Alaskan bryophytes. in: Vegetation and Production Ecology of an Alaskan Arctic Tundra. Ecological Studies Series 29. Tiezen, L.L. (ed.) Berlin: Springer, pp.141-168. StenstrOm, M., Gugerli, F. and Henry, G.H.R. 1997. Response of Saxifraga oppositifolia L. to simulated climate change at three contrasting latitudes. Global Change Biology 3 (Suppl. 1): 44-54. StenstrOm, M. and J6nsd6ttir, I.S. 1997. Responses of the clonal sedge, Carex bigelowii, to two seasons of simulated climate change. Global Change Biology, 3 (Suppl. 1): 89-96. Sterenberg, V.Z. and Stone, W.E. 1994. Geology of the Alexandra Fiord lowland area, Ellesmere Island, N.W.T. In: Ecology of a Polar Oasis. Svoboda, J. and Freedman, B. (eds.). Toronto: Captus Press, pp. 13-22. Sturm, M., Racine, C. and Tape, K. 2001. Increasing shrub abundance in the Arctic. Nature, 411: 546547. Suzuki, S., Kudo, G. 2000 Responses of alpine shrubs to simulated environmental change during three years in the mid-latitude mountain, northern Japan. Ecography, 23: 553-564. 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 Press, pp. 185-215. Svoboda, J. and Freedman, B. (eds.). 1994. Ecology of a Polar Oasis. Toronto: Captus Press. Svoboda, J. and Henry, G.H.R. 1987. Succession in marginal arctic environments. Arctic and Alpine Research, 19: 373-384. Tedrow. J.C.F. 1977. Soils of the Polar Landscapes. Rutgers Press. ter Braak, C. J. F., and I. C. Prentice. 1988. A theory of gradient analysis. Advances in Ecological Research, 18:271-313. Th6rhallsd6ttir. 1998. Flowering phenology in the central highland of Iceland and implications for climate warming. Oecologia, 114: 43-49. Tieszen, L.L., Miller, P.C., and Oechel, W.C. 1980. Photosynthesis, in: An Arctic Ecosystem. Brown, J., Miller, P.C., Tieszen, L.L. and Bunnell, F.L. (eds.).Stroudsburg, Dowden, Hutchinson and Ross. pp. 102-139. Tilman, D. 1985. The resource-ratio theory of plant succession. American Naturalist, 125: 827-852. Treberg, M.A. 2000. Patterns of Diversity in High Arctic Snowbed Plant Communities at Alexandra Fiord, Ellesmere Island, Canada. MSc thesis, Department of Geography, University of British Columbia. Vierling, L.A. and Deering, D.W. 1997. Differences in arctic tundra vegetation type and phenology as seen using bidirectional radiometry in the early growing season. Remote Sensing of Environment, 60: 71-82. Walker, D A 1999. An integrated vegetation mapping approach for northern Alaska (1 : 4 M scale). International Journal of Remote Sensing 20: 2895-2920.. Walker, D.A. 2000. Hierarchical subdivision of Arctic tundra based on vegetation response to climate, parent material and topography. Global Change Biology. 6(Suppl. 1): 19-34. 107  Walker, D.A., Billings, W.D. and Molenaar, J.G.D. 2001. Snow-vegetation interactions in tundra environments. In: Jones, H.G., Pomeroy, J.W., Walker, D.A. and Hoham, R.W. (eds.). Snow Ecology. Cambridge: Cambridge University Press. Walker, D.A., Halfpenny, J.C. Walker, M.D. and Wessman, CA. 1993. Long-term studies of snowvegetation interactions. Bioscience, 43: 287-301. Walker, D.A. and Walker, M.D. 1991. History and pattern of disturbance in Alaskan arctic terrestrial ecosystems: A hierarchical approach to analyzing landscape change. Journal of Applied Ecology. 28: 244-276. Walker, M.D. 1995. Patterns and causes of arctic plant community diversity. In: Arctic and Alpine Biodiversity. Chapin III, F.S. and Komer, C. (eds.). New York: Springer-Verlag. pp. 3-20. Walker, M.D. 1996. Community baseline measurements for ITEX studies. In: Molau, U. and Melgaard, P. (eds.) ITEX Manual, Second Edition, Danish Polar Centre. Walker, M.D., Daniels, F.J.A. and van der Maarel, E. 1994b. Circumpolar arctic vegetation: Introduction and perspectives. Jornal of Vegetation Science, 5: 758-764. Walsh, S.J. and Davis, F.W. 1994. Applications of remote sensing and geographic information systems in vegetation science: Introduction. Journal of Vegetation Science. 5: 610-613. Walther, G-R., Post, E., Convey, P., Menzel, A., Parmesan, C , Beebee, T.J.C, Fromentin, J-M., HoeghGuldberg, O and Bairlein, F. 2002. Ecological responses to recent climate change. Nature, 416: 389395. Webber, P.J. 1978. Spatial and temporal variation of the vegetation and its productivity, in: Vegetation and Production Ecology of an Alaskan Arctic Tundra. Ecological Studies Series 29. Tiezen, L.L. (ed.) Berlin: Springer, pp. 37-112. Welker, JM., Molau, U., Parsons, AN., Robinson, CH., Wookey, PA. 1997. Responses of Dryas octopetala to ITEX environmental manipulations: a synthesis with circumpolar comparisons. Global Change Biology 3 (Suppl. 1): 61-73. Wiens, J.A., Stenseth, N.C, Vanhorne, B. and Ims, R.A. Ecological mechanisms and landscape ecology. Oikos, 66: 369-380. Wolcott, J. and Church, M. 1991. Strategies for sampling spatially heterogeneous phenomena: the example of river gravels. Journal of Sedimentary Petrology, 61: 534-543. Woo, M.-K., Edlund, S.A., and Young, K.L. 1991: Occurrence of early snow-free zones on Fosheim Peninsula, Ellesmere Island, Northwest Territories; Current Research, Part B, Geological Survey of Canada, Paper 91-1B, p. 9-14. Woodley, E. and Svoboda, J. 1994. Effects of habitat on variations in 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. Toronto: Captus Press, pp. 157-175. Young, C.G., Dale, M.R.T., and Henry, G.H.R. 1999. Spatial pattern of vegetation in high arctic sedge meadows. Ecoscience. 6: 556-564. Young, S.B. 1971. The vascular flora of St. Lawrence Island with special reference to floristic zonation in the Arctic regions. Contributions to Gray Herbarium at Harvard University. 201:11-115. Yurtsev, B.A. 1994. Floristic division of the Arctic. Journal of Vegetation Science. 5: 765-776. Zeeberg, J. and Forman, S.L. 2001. Changes in glacier extent on north Novaya Zemlya in the twentieth century. Holocene. 11:161-175.  108  Appendix A Phenological Observations and Growth Measurements A1. Cassiope  tetragona  (L.) D. Don  Phenology: 1. Flower Bud Break: as soon as any of the red pedicel is visible as it emerges from between the leaf axils; 2. Full Flower, the date that the flowers are fully formed and have opened; 3. Immature Fruit when the corolla has dropped off and an immature fruit is visible on the down-turned pedicel. 4. Mature Fruit when the fruit is mature the pedicel which straightens and holds the fruit up in the air. Reproductive output: • Number of Flowers: The total number of mature flowers per branch; • Number of Capsules: The total number of developing capsules on each tagged branch.  •  Number of Mature Fruit: The total number of mature fruit on each tagged branch.  A2. Dryas  integrifolia  M. Vahl.  Phenology: 1. Flower Bud Break: as soon as any of the white petal is visible as it emerges from the flower bud; 2. Mature Flowers: the date that the flowers are completely open; 3. Capsules Forming: after the petals fall off the flower produces a swirl or twist of featherlike seeds; 4. Dispersal: once the swirl has expanded and the seeds are beginning to be released. Growth Measurements: • Flower Heights: Flower heights were measured once the flowers were in the capsules forming or dispersal stage towards the end of the growing season. Measurements were made from the point where the pedicel emerged from the cushion to the base of the flower on all flowers for all tagged plants. These were divided into fertilized flowers, or those that did produce the swirl of long feather-like seeds, and non-fertiziled which did not. A3. Saxifraga  oppositifolia  L.  Phenology: 1. Flower Bud Break: as soon as any of the purple petal is visible as it emerges from the green bud; 2. Mature Flowers: the date that the flowers are completely open and bright orange pollen is visible on the stamens inside the flower; 3. Fruit Visible: when the flowers have dropped off and the fruit are visible being held up on the extended pedicels; 4. Capsules Open: When the fruits are mature the tops open releasing the seeds inside. Phenology  A4. 1. Salix arcticaBud Pall. s. lat.as soon as the soft hairy catkins are released from the protective Flower Break: brown sheath; 109  2. Leaf-out: when the first leaf is visible as it emerges from the catkin; 3. Mature Flowers: when the flowers of the male plants have yellow pollen on the stamens and the red stigmata are visible on the female catkins; 4. Flowers Dropped: when the flower has senesced and falls off; 5. Leaf senescence: when the leaves show the first signs of changing colour at the end of the growing season. Growth Measurement • Fascicle Length: The length of new growth on each tagged branch. This measurement is made from the start of the current year's woody growth to the tip of the apical leaf.  110  

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.831.1-0090344/manifest

Comment

Related Items